recanati, françois. truth-conditional pragmatics

331

Click here to load reader

Upload: ran-omelete

Post on 24-Dec-2015

413 views

Category:

Documents


148 download

DESCRIPTION

Ensaio filosófico.

TRANSCRIPT

Page 1: Recanati, François. Truth-Conditional Pragmatics
Page 2: Recanati, François. Truth-Conditional Pragmatics
Page 3: Recanati, François. Truth-Conditional Pragmatics

Omom. . ------ -- . .

UNIVERSITY PRESS

Great Claicruioo Skeet. Oxford oxz bup

C)xford CJnivrrs~ty E'rers s 3 depaxtmenr of the Univenny of OxLtrd. Lt further: d ~ e Unrve~srty'a ot?lecrivc of exceflcnce 111 rese21-ch, scholarship, and educauttn by puhlish~ng wodilwde in

Oxford New Yvrk

AucWar~d (;ape i'onn I l w es Salaan~ tiorq: Kottg K~rscttl Kuala Lunrp~rr Mzdnd Melboiune Mexico City Nmrobl ?Jew Delh~ Slmriglku Tdrpck Tort,nu,

Wttfr oii;cc.; 111

Argerttina Aiisma Urmi Chilc Czech Kepuhhc F~ancu Crcecr Gu;lternai:i liungvy Itdy Japan Poland Ltonugd Su~gapore South Korea Swr~erlanct 'Thailarid t'urkev I.1kr;une Viernan~

Oxfi;l;,nl 1s a r c g s c r ~ d bade m a k of-O&ni Uruvcna.y I'ress in h e UK and m ccrrar~r other couritne~

Publ ied m the liilrtcd States i>v 0,dorci ilruvcruty Rers Inc., New York

S) in tlrls volume F:mlqors Kccanah m t o

Lhe moral ngllts of tltc jtitthor havc been asserted Database nght f)xford Uritversrty Press (nlaker)

Fwr puhlisbcd ~ o r o

All nghb resetved. No pan of d ~ a p~iblicat~oir m y be repmduced, stored In a reutevsi sy$tcrn, or tmmnuned, 1n my form or by any meam. vvrtl~tlut tlrt pnor perrnrs>ion 111 wntrng of Oxtbrd Unlveni~. Ftcss, ~7r as expressly pertnincd bv law, or under terrns agreed with tlir. appropriate

reprof:raphrcs itgt~rs orgmlmoorr. Enqulnzs i,nncemlng rcpmduchori iirraicir tlie $'ope ~ C t h t niiovc ihosrlii be selrt to tile R.rghb r)epmment, O&td iJt~ive~~icy I'ress, nt (he address nbove

Yon lrlurt n<,t ~lrcul:~rc ~ I I S book nl m v other bmdiilg or <.over aid Vulr must impose the slrlle ionliltiuli on any acquircl.

t)nnsh L 1br:u.v C:swloj;uiiig in f'ublrc&non ibta

I f;lu av~~iable

L~b:ru.y o i ( :itli!~e',c C:at~logl:lli~: in Pubbcaoon {bra

D a c r $v:rllabii:

I yptbsct b y Sl" ll'ublist~cr Serv~ces, f'oiiii~ihcrry, 1i1Ba Pnlired nr Great Uriiat~i on acld-fice p a p a hy the MI'(; Book., gro:rp, Uollrnrn and King's Lynn

Page 4: Recanati, François. Truth-Conditional Pragmatics

Con tents

Introduction

I . The old picture and the nevi

2. Free pragmatic processes 3. Alleged a r g ~ r ~ ~ e n t s against 'T<:P: {I) Coir~rnunicatio~~ 4. Alleged arhwrnerits agirist 'l'CPr (2) S?i.;tertiaticity 5 . 'TCP ra Mirulrialisrrr 6. Pre-propositional prragrl~atiis 7. TCP as a forri~ of Cnntexttrdisrn 8. Defirurlg 'Radicd Contextsrdism" y. Una~ticudated co~tstituents

10. Overview and aclcnowtedgrients

r . Two types of' rxde L. Serxin tic flexibility 3. Standing niemit~g 11s oicx\loil xxacnr~ing

4. Coi~nterexarnples 5 . Contest-dependence 6. Satr~ratiorr and modr~latirtn 7. Conipositiondity and nlodrrlatiun

8. Is Contextudistn :i threat to c.or~r~x~)itionditv?

I . Two dogmas 2. The intersectivity of relative arijecrives

. . ... TjlC ic12F rf Gf-iilt crsc. zzjc;<t; >-cs

4. Colour adjectives: saturacioil or ~nodnlrction? 5 . Gradable adjectives: modulariorl or saturation! 6. Absolute graclable adjectives 7. Coerced modulation: back to tlrt+irrtexuct.dvity issirc

8. Conclusion

Page 5: Recanati, François. Truth-Conditional Pragmatics

1. 'i'n~tki --cun~iirrorld p r ~ g ~ ~ r ~ t i c s

1. Free pr:igriwrrc processt:.;: two iriterpretatioris

3. X,og~cal 6.x-11 rn rzicvanc.c theory 4. ilow rrlany systtbnis, m c i llow ciif$1;.rcnt?

i . inuotiuc. tron

2. Q';enc.i-;aiilrd ionversabond rrnplic.atrt~res: mvo ( : o r~cep t l~ )~~s

3. Fvoiir pragiri,rtics to ictiraritics +. L%ragrn,lr;rc iix~piicacrtrars ofsrrh- locut i (~~~;~~y ~ L T S ?

5. .I 'c>c:ll. sp<?c*<:ll :tc-t!,?

6. F.~nhcbdiE~>J Y I T ~ ~ ) ~ ~ ( J ~ I I * . C * S :md ti-ee ennchnierit 7 . C:oi~clrisior>: ileikilr irrlplic:lt~lres or free rr~ricluxroart (or I>od~)?

o . P~eL11n111arics

z. jiillegi;tvd) c-oritcx L - s l i x f i s ~ r n co~~iplert serrueuces

4. Pretending thlit tiit' c:orrtcxt i s <di&erent hunr wi~,rt rt i s

.*. I 'wo types ui'c.ontext a-c~d two types of

contest .s!~rftriig pretence

-;. 19~rrspcsc~vr j ror r r t z?s uttorarrie poiiir

0. SElrft:~I~le mdt.~x~;rc;~ls

Page 6: Recanati, François. Truth-Conditional Pragmatics

.j. Mixed quotation 5 . irtterpreting quotatioils: the pragnatic view

6. In ciefcknce of the pr~gnlatic view

7. (-:onclusion

8. Open Quotation Rev~sited

I . 'The yt:agmtic view

2. Sub-clausal open quotation: a multi-ievel an:dysis 3. Interaction with ~e~liantic content: (I) free eilrichnlevlt 4. Interaction with senlantic content: (2) t~otttext-S/U& j. Building the context-shift into the rrrrlantics

6. Echoicity 7. Mixed quotation: the callcellability issue

8. Cancellation or dsambigtlation?

R tfireiri cs lrz~kex of' ,V~lrnt>s

St.t!;elt Grdiih:

Page 7: Recanati, François. Truth-Conditional Pragmatics
Page 8: Recanati, François. Truth-Conditional Pragmatics

Introduction

I . The old p ic t~~re arlcr2 the rlcw

'Truth-Condlnonal Pragmatics, tire ~heoreticsl fixmework ,~clvcrtneJ nrrtV

ltlustrated 1x1 tl~is book, ct'ilrds ~n conhast to (what 11.lc.d to be) the tradrtroi3si way of looklng at the ~ e r ~ ~ ~ ~ t ~ t ~ c r / p r , i g n a t ~ c c d ~ \ t ~ n c t ~ o n

Accordmg to the oltl conccI7tlorl. hilc)wlcdge of tlie l~~l ,vir ,~c~--ar l i I 111

pamcula, selrlannc kliowledge or cevnaiztic cc~npe~elzc~~--e~i~~InJt :c I,mguage users to ascnbe trutlz-corld~tior~s to ssl?~rra~v tentencec of rhm lairgiiage X c-, be sure, wIler~ a sentence I \ (oiltr\t-\cn\~tlve (a\ rno\t \CII~CIII(-~\ ~t onl\

carney tnjth co~iditloris 'wr t l~ rc\pcAi t to cor~text'. $0 knowleilge c"oC'tl ie itrrltclit

~s retlu~red In addrtiorz to XIloM/lcds;e oftlrc language But tktc ct>ntc\t at r"r\ur lrlvolves or lly hmited aspects of the srtrl'rt-ton of utterance whcr jpe%ik\, wlle~r where, to wllo~ii, a id so ic>rth Giveri a cor*tex.t tllus unclerstood, ~ h c niic\ oi-

the Lmguag~--for exantple r l~c mlc that ' I 9 retrn to the speaker - - ~ u W I ( c cu

deternxlne the tmth-cor~&t~oxld i onh-rbution ofcontext senrrtive explessltrrrc

There a no need to apped, in ddditlon, to pr~kq~zat11 tompetcrrri By 'pragmitlc colnpeten~e', 1 ITII~,~JI the dbrjlty to unitc'rrmcl what the sl>eakcr

rrleans by h.ts or her unemice Ac C;nc e cnzpha~zccl, spe,~Ler'q nxcmng rs ,r

matter ofmtendons: what someorkc rlieani 1s W ~ I A C he (11. she overdy u~tetrdq->r, as Grice says, 'M-intends'---to get dcrofO\s CILI:C>U~~I hl\ or her utterarrce <'oiritnu- nicatlon succeeds w1ie11 the kt mtelrtloxrs the speaker are recoplred by the hearer. Yrapatlc colnpetenrce I r needed to detennune W ~ I L Z ~ M ~ c y)~aker titeu~ts on

the basis of whut size ssizp; but wliat tlic spealiler Fays a supyused t o be ammo- rnou~ly dcterrnmed by the sernalxacs (wrth respect to context), lnespectlve of the speaker's behe6 and uiterlnoxrs So the hx&t~o~~al story goes

On this conception, semantics ard pr:lgxriaacs arc: ursulated from tach other. Pragnatlcs takes as Input the output of ccrnantics, but tlrvy do 1104 rmx, and 111 parhcular, pragtnatrc proccs\es <it> not mtedere with the process

Page 9: Recanati, François. Truth-Conditional Pragmatics

2 I ' i i i : i >Li; Mlc liii<F: Kivir I M[: NEW

( p i ' ~ C ' I X L : I P ~ ~ ~ C ~ t > l l j p ( ~ ) ~ j t i ~ ~ O ~ ~ ~ ~ L I L S t j l ~ t1-:llltk~-i'0~1~fiti~f11~.' This ntakes

sc;gl"; ollr: ce,nstrrles senlarktic cc3mpetc:nci. rlrrci pl"g11latlC cOIIlpetertCe tkS

hc.lc,n*ng tcr twc, distmc-t ' ~ I O C ~ U L C S ' . :IS Rmrarrr V3ot-!: rioes in her attempt to tjt.iilrrj fjre rcldicic3nitl cc3nceptitrr) ( h r g 3004). Scrr~;~rrtic competence be- Inngs to rllc P,lrrgl.iagr faculty, she says; i t is an :isprct of our 'knowledge of jmgLlagc'. I'r:rgrniatic competcrlce l't3s ~nort" to do wit11 the so-called 'Theo- ry of Mirrd', that ficulty in virtue o f w11icX1 hurllan subjects are able to explnirr o t l ~ i ~ people's behaviour- by ascnbirlg intentiol~s to them.

Tlac (>Ed pi(-ettru I huve just ricsc-ribetj-- -an essentially modul~zr picture, .ic.ct )a-ding to whicl~ sen~uritics and pcagn latics do not mix-has stnrteti to lose grip irr rccent years. Nrarly evch~body oo\ri;;itlays ackno\vledges tile fact thde the rcft:rerrze of irldexicats and, more gt-rreralty, the semmtic value of ctrrrrcxr -st-rrs~l jve r.xpressions culrrlot be drterrnined w itho~rt appealing to firily-Aeilgrd pragxrr:~tic fi~stors (c.g. spr.akt;r's intentions). The serr~antic value of a context-ct'lisitive exyressiorr varies from occurrence to occur-

rence, yet i t ofien varies riot as a f~~nt:rrior.i ofsorntx ol2jective leahire of the contest hrrt 3s :I firnction of what the spenker r-rteiins. Pragi~iatic conipe- tcnci., therefirre, i c rcquired not otrly tt.? clt:rzrrr*inc: wliar the speaker nleans m i the ii~isis oi'kvhat she says, hrtr dso to determisie what is said, insofar as this i s ~listiriit I ' r t r r r r tlrc rne:aning of the serlterice type.

C:.rpl~"clera ivoritiers why sorrre tl-rersrisrs, inc.luding n~yself, n ~ a k e a fixss 'rborat rlirs isstic:

St>rric. ,rutF~or.?~ rilakc :i great (le;"latir oftht. cilstirrcttori beccveeri those indesicals that i-ia~-e p.ht?lr. scirwntJ.li- v,~!-\rr.i cletenrlrried by r l r c spe":Ll;ci-~s intentions :md those tha t are iieicrr~~irieti soieiv hy tile rittn-irrrerrt~tinaI i;iatures of"tE>e cc>iitcst ~Tutrer-arrce (such ,IS tilt \pf:akcr, tirne, ant i piace). . . . [Yet] ilo cribst;riitrve o r rven retriotely interest- irig issue depcrids upon this distinction. jC:~ppeleir suo?: 8 )

intleed t17t' alli\t~~~ct~ori does xtot 11ixtter 111tu-h fro111 .t techrr~c,il poult of mew, irni c. one. c m iallow K'I~~AII 's ~ t l g g t ~ t t o ~ ~ rid t~c'it tlw spealrer'~ Intenllon (or tlir relr~drrt pr'lgrnJtx f~c tor ) 2s tllc conte\ttral 111dex wwblch serves as

agrrirrent to (I~rrdcter timctrc?rr, alosig\rcle the speaker and the hme,

pi~ace, and w~>x~cI fc'rturcs of the cor~tcxt To trx the reference of, for ca;iuiplc", G* dcx r~ t - r r~~r r~~~rw orlc [ I ~ L I ~ C tdke t l~ to ~ C C O I I I I ~ who o r what the

Page 10: Recanati, François. Truth-Conditional Pragmatics

speaker ~tltencis to refer to, p s t as. to fix the refi.rellce of 'I9 , one nlwir take

Into account who the cpe'lfi-er 1s f-locvever, f3rantioni pornts out, ‘tills ternptmg ass~lndat~o~? el~des a very rrnport'lnt cftstrnctiorr'.

What I wcmt to call 'gerruine' se~rlantic indicrs are ft~atr~rcs ol'lttterances tlrar call be read off-without knowing anything d ~ u t what tlrr rttteuancr means. 'Titnc, place, speaker, and possible world are propertirs of tokertings that can be seccled and specified beiijre one alms one's attelltion to the content exprcssecl by tllose

tokeriings. . . . IThey] can be itrtenninccl ir~dependrntlv of [the context-.sensitive expressiun's~ sem;iiltic value 2nd tlleii appealed to as input &om wllicfrl thc v;ilue could then be computed by a character-firnction. OSrancioni zoo8: $8)

Arguably plragmlttrc t j c to~s such a speaker's rekreutul rritcntlon\ do r ro t

have tlus mdepel~dence property ,inti do not qualrtj, as gentilne contest~tal

~ndlces 1x1 thl\ sencz.l

Hut agatn, Cappelen nught ask, bvli): is the dl\tlnttiorr between thr two

types ofrndlces so important? r'here are scver,il reason\, but tlre ordy o11e

, that matters here I\ thn. d we let cvhat tile speaker riiedns be orle of the

factors on wlncki tmth-cond~t~onal content ctcpertcis, that ~ i te l tu~ t h ~ t ir c rLt vvt

tip the nlodzr/anty zaczil. We accept that prxgnat1z3 and \estlantics do mix In fixing tmth-cctncirt~or~al t orltenr -3 q ~ u t z t u n d ~ m e n t ~ l ( oncesslon to the

oppctnetlt of rno~iulantv O f course, d o n e wmts. to marntciln a \ cn~a r~ t~cs

pure of p~agmatlc intrtn~on, one cdn, but tllcrl ctne ~IJ\ to (orlstrire the godl of sernanacl tliEerentIy tharl it 1s on the t ~ ~ ~ c t l t ~ o i ~ a l cctnccytion Pure

sernarrtlcs \wll rro longer dehver truth tondltiorr\, bnt ~t will cti~lrvtr, say,

corlditiorial truth contiitrons, or sctleni~t,~, or ~har,~cters, or .~vlldtevcr To get fid-blown trtttli-conc3ltlondtor ontent, prdg~il,it~cj ~ 1 1 1 t 7 e r ~ e ~ f e t f

a. Free pragri~atic processes

H o w far car1 we go in letting pr'lgmdtlcr inlpinge upon the ~,tctlt~orral

dolndrn of \elnnntic\VPrettv h r , 1 take rt (see m y L,~tcr~il ,%leairrrz!,r (aooq)).

Rut the trad~tlori~d plcturc 1s still ~riff~rentlal, A I I ~ rtlo3t tllcor~sts do~l't \ ~ ~ ~ n t to

go beyond the corlcesslon that pragndhcs and s.pe,iher'c Ine'lnlng I l a ~ e A rctle

to pIay in as9igning values to irrdcxicJ$ 'tnd fiee vansiblei, in log~c,~l fonn

see ' i lcran,it~ ( L O O I ~ ' : 193 9, z,c)<) arid Sii,irnzc [fbrtlri olrlincr. r i l l ii,r ni<,rc r l r i t >i'>xr

Page 11: Recanati, François. Truth-Conditional Pragmatics

(That 15 the reason why I sad t h ~ t the traditional corlcephon has 'ctarted to' loce grip I t ham't lo\t ‘grip entirely-yet.)

On the currently dorniriar~t p~cture, prap~ahcs comes into play m die determmatlo~l of trirth-~ondttiorid content but dom so only when the sernantrc d e 5 of the language pre~cnbe it (as when a11 indelncal demands a contextual vdue) Innfar AS ~t plays a role m content detemvnatlon tlrerefore, pragrxlatlcs IS tubordinated to renxxrtlcr. S e m a ~ ~ ~ c s marks the phces where pra,matrcs n to rntervene, it rets up 'slob' that pragnattrs is to f i . O n this view the only tmth- conctttton~l role oj. pragniaclcs correspolids to what I have called 'saturanon' 5atiir~(ion is a pragnlatlc procmc of cot~testud value-asslgmtxzent that 1s tng- gcred {anti lllade obl~g~tory) by sonx t l~ng in the 5entence ~tself, nanlely the hngu~snc expression to whxh a value is contextually 'fis~g~red. For example, I_f

the spelrker use\ J detnonstrative pronoun and says 'She xs cute', the hearer

rllust detemiuic ~111) the speaker means bv 'shc' ln order to fix the utterance's t r u t - t i content 'The exprcasion ~tself- a ~ t s as a vmable in need of contt.xtud mmctantiahot~ So pr&agnatrcs coxnc\ Into play, but it does so under the g~ucbnc e of the hrigltistlc nldtenal

1 hiwe often dcrcr~bed ratt~ratlon a\ a 'bottom-up' process in the beme tliat it I\ agnal ctnvtn. not ri)nte.ct-clnvm A 'top-down' or c oritelit-dnven proces5 ic 2 ~ ~ ' I ~ I I Z ~ ~ I C process wh1c11 IS 110t tnggercd by an e~upremon in the senteiicc hut ~ t k e i pLce tor purely p r ~ g r i ~ ~ t i ~ re~ons-111 order to u r~ke sence of

what the spe~ker t i saying Such processes I refer to as '6ee' przipaac processes free heiaure tiley are not n~andated by the lin~untic rnatenal but respond to wllolly pragxnatrc. cotislderat~ora. f-or exatnple, the pragnatlc proce\\ though whrch ,tn rxpress~o~~ a gtven a non-fiteral (e.g. a metaphoncd or nietonvli~~cai) itrteq3rcLihou ~r context-dnven: we interpret an expression t~on-hterally ui order to mnke sense of the speech act, not because th s 1s d ~ c ~ t e d b y the hng11c;tic niaterrals 1x1 vutcle of the rule5 of the language

The doi~urlant vlew 1s that the only pragmatic process that can afftct truth-co~id~tional content n satur~t~on. No 'top-down' or free pra=matlc procef5 Carl affect t r ud~-~ond~ t i t - t u r~s~uch processes can only affect what the 5peaker meam (but not what slre say?) Or, to put ~t m King's and Stmley's

-+ . -.. tealunology there La11 only be 'y~ak-" pragmatic effects on truth- condltlorld content They define a weak pragrnatlc eEect as follows

A weak pragn:rti~ effect 0x1 what 1.i commt~rucated by ail utterance is a case in whch context (i~i~Iu<Lt~g speaker lrjtenttons) determes mterpretatlon ofa lemcal Itern In

Page 12: Recanati, François. Truth-Conditional Pragmatics

accord w'th the standing nzcrunir?,y dfthat lexical it(*m. A strong pragn~atic eNkct 01.1 what is communicated is a contextual efi?-& C)II what is communicat~d tha t i s not merely prampatic in the weak sense. (King and StarlPey 2005: rig-15.; emphasis mine)

The vlew that ordy weak p rapa t i c efl>cts clrrl aEec t what r\ sad I c ~ l l 'Mimm&sm' (Kecm,rtl 1989, 1993, zoo~d , 3004) The ratron,ilc for thrr, term is that the role of pragnatlcs lr r corlterit dctcni~inatrorr u Acpr to A

mnlmurw ordy saturattorl is dowed " Truth-Conci~ltloonal Prag~rrdtrcs 1s thr* opposte mew, haced on tlrc reJcc-

hon of Minlm~dism It says that tmth-tor1d1~1on.11 content be dYected not only by saturabon (as when an ~ndexrcal IS asslgrlect a contextual v~luc) but also byfrer pragmatic processes For exan~ple, a cerlterlcc like

(I) There 1s a llun In the mntddc of tllr piazza

h a several reacfings C)n one rea&rlg %ony 1s gvea a noil-llterd mteqrctabon and means soniethmg hke 'sstatue of a honYOi diat re~cfing (I) n-tay he true

even 6 I~terdy, there 1s no hori iri tile ~~xildle ofil-rr plarw (hut ordy iy atatue of

a hon). The non-hteral rcndlrig of %horlhrguably results .from A p ~ ~ ~ ~ a t u . " operattori of 'modulation' &,lt 1s IIOC ~n~riidatr~l by the lcx~c~.l item 'ljlcon' I kcre s no slot to be filled or anytiling of the rort The priygn'itlc. efiect tierr look\ lrke a strong prdgrriatlc e f ik t, yet ~t &&cis tnrth-condrbonal conterrt That rs

the sort of~exaniple kviuch nlotivates 'I njth-('onthtlond Prnginabc s

%%ether or not it 1s on tlic nghr track, 'Tnttt1-t:ontl1t1o11,d Pr,~grr-r,xat s

i;hodd at least be cois~ciered A\ a tlaeoret~c a1 alternanvc to Mnurr~~ltrrrr ura>~-tir

explomig. After all, ovlce pragiiatits is ,~llowed to plav role rrr the Aeter-trurr,t

non of truth-ct)nditlonal content, rt 1s \c)~~icwll,it arbrtr'q to sol I~tmts to a L \

In their influential Insen.qiiii~e Svnzunain C,ippeien and Lcpori: bot-row my term 'Mu~i~~mdisiir' (and the asoclated plmse 'n~mmial proposition') but tliry ilse it I I I C ) ~ ~ resmcnvely tIu1-i I do . For ilrrn, a nurunidist does not mereiy rule out \trting pragmatic c&e(:t.< on sernantio cctrrterlr, bur at tht* saint. hiale mes to keep weak praglriatiz eii'ec~s to a cnlriinium (by dtnwng that sorr~r esj>resslotrs ger~erally rrea~rd :is

context-sel~sttive are really contest-sens~tive). 'This, f i i v me, IS :I I-atirt:~.specral----and pai?icuiariy &ZI~--; form of @+~~r!rsrn: it 1s not llil~irimalisrn srrr-rl)Iicrtcr. So. ii)r rxarriple, I coilcider Jnsorl Stanlev .a

mini~nalist, since he einphaticdly rejects stlorig j>rdgni;iUt. ei-Gets on tnrth-cont%tioi~ai corlteilt; but lie is not a radical nurtlmalit i la Cappelen artd Xepore. for lie achowlrdges a lot ofconre.~t-scnsinvr~y (of the 'sat~iration' variety). The sarne son of tcrartinol~,gcd d~llerence is to he fo~ltid %it the ottiev end of the spectrum: whoever accepts a good deal of context-sensruvfry (1.e. everybody riowa&ys) 1s a 'contexvu- alist' by Cappelen's and Lepore's standards. I-or rile a coniextwalist is someone who gcni?mlhn currtelri- sensitivity (see Section 7 below). So Stanley is ncit a co~itextualist by itiy Lgtru, but he IS, ar.c.rrcd~ng to Cappelen and Lepore. (What I call 'CC:ontextu:hsrn' they cd1 'Radical C:ontextudism'. Strunley, for theixi, is a 'nioderate contextualist'.)

Page 13: Recanati, François. Truth-Conditional Pragmatics
Page 14: Recanati, François. Truth-Conditional Pragmatics

IN 1 RC)l)lU( T ION 7 - -

prdgma~tlcdIy ~nfencd, the cluebtlorl anses, wf l~t s;lidrdalteeb that the hearer

5vla be able to latch 0x1 to the exact sarxre serndrttlc v a i ~ ~ e '1s the ,pe,~ke~-2 qtlether the pragrnahc proces\ at stake is saturation or moduIatlor1 u rn-levant as fir w t l ~ s Issue u concer?ied, so the 'm~posstbil~ty of comll~inrca-- eon' argument cannot be dduced m favour of Mirtrrnillsn~ as aganst TCf'.'

Now, l ~ o w can we account I;>r content sfrnnng in comrrinm~dt~on? 'Thxce

&tors seem to be e\pec~ally irnpo~td~lt. First, spe,tker) and a~ktre\seej \h,~rc the same psychoiogcd make up. In I,ttural Meuttrn<q I that 'the mterper- sonal s t k h t y of trut11-cond~t~onal cotrterlt. . . can be accoimted f'or on rycholo~cal rather thari hngulstrc grounds' (Kecanatr 2004: 152). And X uoted Bolinger who notes that

eaken riot only share the ~ m r e cock but also sft,ue thc abdty to see tlie .;me

rsen~blances betweeti wh& thelr code already deugnates ard whit tttev worrld lrke ~t

deagnate, md so to xn&r tlie old tbnns rexh out to new nlea1mp rhat IS how guage bre& h e of its ng&ty (33ollngtrr 1968 230)

Note that tho so called rrur'tcle of corrlsjlllrn~cahon 1s sllrular to arrother rmrac ulous a~lt~evenier~t, well documented by psyt ilolog~rts: carlv word it%dr~i~r~g

(Clark r 993, Bloon~r?ooo). I tow do l,u~gxage lean.~ers tln~i out what the wot ck they hear denote? l'hey have to rely on context arld pr<qqnatlc Lictorr, even more thm the regular comiurucator does ~ C C ordriig to ri31y(rr~ice tile regxl,tr

mrrrunlcator a n at least rely on her mastery of the Imguage) kven tf dle guagc learner 1s exphcltly taught a gven word N---the mother point\ to

sometlung and says 'ttbs sn Ltn N- -we know, srnie W~ctgen.iteln and Qzime, that ax1 ostensive gesture IS mdefk~tely ambiguous. What ~nake\ ~t nevc~thele\s

See Maitra ~ ~ 7 0 7 : 120--8 ibr a s~rnrtar yolnt. Note that one of S d r ~ ' s early xgitnents against t7CP for the sane reason. In 'Context and I,og~cd :<inn'. Statdey obje~ts to 'ICY on the grou~ids that ~t

cs trilth-col~rlitional lntcrprecauon s~rnlhr to 'the kind irrflnterpretation/ involved in iliterpreung under tile table and taps 011 the stloulder' (Swniey zooo: 396). However, Staniey ackr~owledges tliat [context-rensiuve] element brings with it mles gownling wiiat context can and cannot assign to it,

yirg dtyrea qfiiwzty' (Stanley zooo: 336; elrqhasis mine). E'reumably, when the degree oCIarlty of e ingulsucdly encoded constrant IS f1igi1 (.a ~t 1s for .x great deal of context-seus~tlve expresslorrs). thr

rc ofthe expression, hence the truth-co~ldit~ons of the utterance containmg it, can be ckternnned by appedilig ti, pragmat~c conlpetelrce and figunrlg out what the Intentlorn of the speaker mght be

see (Zbapter 6, Sect1011 I ) . Tius, u~ ~csc:lf, 1s suificlent to niake the intexyretanorl of uttera~ces similar to hat of'kicks in~der tile c~ble (to some extent), e>en if we leave free pztgrnaric ynxesses aside. (I Tay: 'co ome extent', hecause truth-conditlond intepretanotr 1s granul~ai-dnrcn. cvcn if ~t rlceth to rely on

gna t~c competence. This jifferettce--betweer1 nurh-cond~nond 1~tCrpTeh?tlOIl a r ~ J the Intrrpreta- n ofk~cks under the t.ibie - mav be wliat S~znley has In lnlnd when he ol?lrcts to '1-CP. Set- Sectlrm 4

helclw lor a rcupimse tci the objection thus undcntuod.,

Page 15: Recanati, François. Truth-Conditional Pragmatics

posslhle for the language lexner to latch on to the correct senlanbc value n not a mracle but, p\ychologists telius, a ncb systenl of p~ycholo~cal bases and tacit assmpbou'S, w f ~ c h are amenable to empmcal study. There a no reason why the sdrne sort oi t l lu i~ \hould not dso be true in the other case.

Second key factor we keep nlonitonng each other's understand~ng of what we me sAylng, makmg repairs when neLesTary, negotlatmg meanmg, etc jree Clark 1996 for a s~~stalried defen~e of the thes~s that 'language use 1s really a fbrm of joilit xtron ' ) As Ernesto Pernni pornts out In response to Cappelen's s31id 1 epore's mirarle-of-coimi~u~iication Argument, the interac- trve nAture of rnutud miderstmdlng 111 comniunlcation is what ensures convergetic e and conterrt-sharing

Corl~~~l t r~vcat~on exl.ubits syitems ot rlegatrve and posltive feedback that help to assure rrirrtual understartclulg Tha n the case of [the] acceptance cvcles proposed by t-Ierbert C'l'trk x r t i LO-workers partlclpann tn a conversatron make efforts to

ecthh\h the n~trtuA behef that hsteners have understood what 1s meant by the speaker If ctie lrste~ier doesrr't see what object 1s amed at by the speaker, ?he wrll ~ndrcate ~ t , xtid the speaker 1.i expected to propose a new presentation, untd the h\tencr gzve5 air ~ckr~owlccfgerrteiit srgt, followed by a confsmatlonby the 5peaker {Perrr n~ forthi o~iung)

1 be third bctor that 011ght to be ~r~entioned 1s the s~gnlficant. degree of vaguerreqs drxd toler;~r~ce for r ~ u s ~ n d e r n n g that characten7es l-ium,m comrnulllwtlort I r i rn,lkmg that point, I am not 'biting the b~~llet ' and

" .l he t~asrc process, wlr,clr may be called the Lzciepcancc ~yile , consists of a presentation plus ~ t s verdict. L.er x. y, arid s stand fix nouli phrdses or tllelr en~mdatiorrs. A presents x and then f3 evdua~es ~ t . I f the verdict is riot positive, ttien A or B rrhust rehsb~orr ch:u presentation. That penon can offer: a repair x', an expansrun y, or a replacen~mt 1. The retishioned presentation, whether x', x+y, or r, is evaluated, and io or]. Accepsptailcc cycles apply iteratively, wltlr one repair, expansion, or replacement &er anolber, until n noun phrase 1s rnunrdiy accepted. With that, A and B take clle process to be complete' (Clark and Wilkes-Gibtx 1986: 24).

Elere is a11 exarnple:

:I. . . . weU I w3s the only one odier than than the uhul tch the f-ords? l l lt Mrs. Holmes Ford? Yorl know uh= = tire cellrst?

ti . Oh yes. She's she's the cellist.

A. Yes. Well she dnd her husband were there.

(Clark ad Wiikes-Gibhs, 1986: 5)

in t11rs dialobwe, the noun phrase 'the Fords' is introduced with a rising intonation, sf~owii~g that A was expectrrig an answer Srom B. As this answer doesn't come, A oKen the expanded noun phrase 'kfrs. klolrtzes Ford', and then 'rife cellist', ulrtil B identities the referent, ~ t d A confirms it.

Page 16: Recanati, François. Truth-Conditional Pragmatics

accepting that conxiiumcatrotl Ir ~ n ~ p o s \ ~ b l c or t f i ~ t co~lte'ilh are not shared (Cappelen and Lepore zoo ja: 126-7) Appe~r~nce.r rtottmthstar~ding, vdguc-

ness and tolerance fix n~~wnzferrtarrd~rrgg art. comp,xt.lble wrth the ~ d e a that

we do comunlca t e arid share content (ilecluirc~lty, we c,in use super- valuatiori techniques to define shared < orlterrgt whde allowug for sig~~lficanr: divergences between speaker and hexer )

4. AUeged arglrnerits agair-rst '1'C:P: (a) Systen~atic~v

The scconcl objectron to TCP 1s that It n-rakes sys t~~~ la t l< se~l~an~lcs ~~npo\- slble. Elere is the argument:

In contrast to the contextual asslgrl:r~(~rrt ~)fvalucjs to mdexicds, nloduh-

tion u not dnveri by the I i~lgi is t i~ rneamng uf bvords. Notllrng m the

lingulshc rneamng of the words who\c scn.ie 1s xnodulated tell$ us t h e

modul~tron ought to take place Arlod~ilatro~l takes place purely a\ a rndtter of context, of ' p r ~ g m ~ ~ t ~ t s ' , \V~I,II. h v e s IC I$ the urgc to rri~ke

sense of what the speaker is uvlng SC) i~~t )dula t~on IS U I I S ~ \ T P X I ~ L ~ I C If wc

'illow ~t as a detenrunant of scinantrc content, we mmbe ~t ~rnpoii;rbie to

construct A systeix~~trc theory trf Serrrantrc content.

I gr?unt the objector that mocjtllatron 1s ~tnivstematir. StlU, 1 t h u k rt Ir e;i\). to

rnCike room for it with~n a ~ ~ r t c ~ l l a t ~ c \enmntl~, " 1 x 1 gerler'il, nothi~rg pre

vents unsysteirlatic h c tors "i'ronr bt~rlg harldled svster~wtrcally, b y bcrrlg assigned therr proper place 1x1 the tEizorv

As far as modulation 1s concrrncd, X kvtll show how to 'Iccourit Tor it

system~ttc~lly ;I, the very first chaptzi of this h ~ ) o h 111 a r^rlrtshrll 1 will define a furtctroii rviod tahng ,IS arpjmertt an expression e and the rttrrtcxt r an

wlilch it occurs: the value of rnod is the part~c~llar modulatron futu. trong t h ~ t IS contex~rrdly salrent/relevant/'appre~pr~dte f i ~ x the ~nterprrtdtron of thdt

expression 111 that context (If rr t , rrloii~ilahon is co1ltextud)ly appropndte and the expression receives 1t.s literd ir~terpretano~~, the value of mod wll be

'Acknowiedging rnoduiations as prrrnnry pragmauc procerscs isn't in conflict LVIEEL a(-ccphzig 3

central role ior cornpositional semantics', Papti and Pelierier nghrly say (2007: 50; OK) p. '$4, ~OWCVCI,

they wrongly ascribe to mne the contrary view). Like Pagin arid Pelletier, Westerr;dhi ( f o ~ ~ h c < > n ~ i ~ ~ & sketches a compositional treatment urfmddatition.

Page 17: Recanati, François. Truth-Conditional Pragmatics
Page 18: Recanati, François. Truth-Conditional Pragmatics

~ ~ r i ~ o i ) i i ( 7 I O N T r

In the same way, he clalrm, 'Evervorse loves Sdly' carrtnctt be usrd to egxess the content oE'Evenione loves SaUv ale1 kus rnotlrer'

1 agree w ~ t h S~mley t h ~ t certxn dxng (lon't lrappet~, that would hapyen if

modula~oxl were totally uncor~srr~ilned. 13ut who cLlxned tlut ~i*c~cfuLnor~ was

totally uncortstr~~ned?j Work ut tfxs ate2 precisely neecis to '~cicircu the i\sue of udlat 1s poss~ble and whiii- n rrot (strip one), ln order to amve ,kt suttahle g e n e r ~ h ~ a a n s (step two), wkl~'tl tt wdl then be tnturrihex~t uporl praa crln~tlc theory to denve (.itcp three) (Elbourne 2 ~ ~ 0 8 ycp~oo) 5wdev's eff:~)rts to iden* '~mpossib~e modtllatlons' Are d welcome corrtn~tuhon to the first ctep

Those eZfbru have ordy been t111lcUy surces\ful, rhouir;lt 1 dke Staxtlq 's t~vour- ~ t e example 'Every Frenchnan w swtcd' ultnch, xcor&trg to h m , ~ ; m exprcs\ the content of 'Lverv brenchmml ztz the cltrss 1s \erttecY, btit riot t h ~ of 'Every Frcnclulran or IJt~tt-hmarz I\ sedted' I don't thnk Swdey h* got thC &tit nght here. When t-iercule Pollot uys 'S~n~dyspeahmg, I .mt not Erellcl~', I-tc concede\ tltatlc30se1yspec~, he rc French, &]dm the Ltter use 'Frelich' mem\ rotrletlur-tg lrke 'Fren~h or Nelgm' (see Hall zooS. 4 i t ) So 5t;tnley bw bcerl too qt11ck 111

ndmg out this sort of moctulntlon O n thr otl~er ~ L L I I ~ , ~iefrnciers of I C:P sstlc 11 rc\ Elbourne (2003) and I-XnU (2008) 'igree wlth hirn th,tt 'kvervune k)ve\ Sallv' carmot express the contertt ofLf.,ver).one love\ Sdly arltl h ~ c mother' ilssulrung that Stanley, Flbotlme md 1-1,ill Jre ndit w~tl3 re5pett to that c ~ ~ u ~ i p l r , rhe proper thirlg to do s to explalil why the uttrraicc cannot have the reletevaxrt re'icil~rg

Stmley's sweeprng response s be~arlse ttlere w xto ?LILA thmg a 'nlodnla- i - tion', free prAbmwtlc procesres', or 'stlong pr,gnwtlc effect\' ort sern,ltlhc

content. Mternahve reprises, conslsteilt with 'Rzlr', h.we bren put: forw3sd by E l b a ~ ~ i i e (who proposes ii ggmeral17~~~on and thereby contributes to step z) md by FI&. One may well remaul ~nicon\u~ced bv these altert~at~ve responscc. but even IC o,ze rqects them, there rs a good leason to keep loolung h r explai~tlons instead ot gcting dong w ~ t h S ~ a t i l e ~ in d~srlzirsing the whole field of enquiry If, followmg Ctanley, oxte takes tlre 1,rtrt.r c ourse, it will not be poss~ble to account for the dfieret~cc betweein the two exa~~iple\ cited t ~ y Stalley-the ' h e r t c h m ~ ' ~xluriple and the 'Sdly'exarlple. I l l kppropnaie contexts, ~t n porsrble to underswnd '1-rertdlnwn' loosely, so ttrat Polrot coirnts w French. Rut (rf Stanley, Elbourne, arxl Hall ale nght) there 1s rto way 111

whrch 'Everyone loves Sdly' can he understood ,as 'iSveryone love\ Sdlv and

One obv~c>us camtra1ilt, irnplicit in my w n t l n p on the topic, IS \h.lt mr>dul.ltit)ri iht)uiil prrcerve ,rrnarltic types, just as adjunciron priserses \yiltaiil.lc type\.

Page 19: Recanati, François. Truth-Conditional Pragmatics

I~2---T-c-~~v~xmmnw~------ --- ---- -

his mother'. Wholesale rcjectlon of-mod~daaon, as advocated by Stanley, does riot allow one to accoi~rt &,r the contrast between the two types of case.

111 ,tddltl~rl to the alleged arguments agamt TCP I have just discussed, there w a postme argament 111 f av~ur of Mirrimahsm It n a conceptual argument ,uld c,m be put as fi)llows. What n sad 1\ what a lzlerally sad, and that-by definitiorz- has to be determned by the conventrtons of the language. 1)ragnat"cs can enter the picture, provlded tts role 1s to asqign a contextrlal value to a lexcal item ill a bottorn-up manner, that is, in accord with (and txncler the gtnciar~ce o f ) the ~onventlonal meamng of that context-senntrtve Iten1 In contr.ist, strong prdgniatlc etfects achieved~n order to make sense of the speech act w~thout being llngulstlc&y inandated take us into the realm of speaker's mc,lmng, away from llterjl meaning.

Insofar t l ~ s argrrnent &r M u l d s m IS based upon a certlln undentand- 1% 01 the pllrdce "tvhat n \ad' (or 'wlrdt a llteraliy sad'), one -cvon~ters whether the q u . ~ ~ c I between TC'P and Miriuxr~lrsm may not be verbal rather than substdrltive There 1s no doubt that one car1 dejirze 'what 1s said' in such a way that or~ly weak pmgnntlc etfects c,ui affect wllat 1s satd. Ifwe do 50, f i n u n d - Ism corrlcs ctut true by delklhon On suc h m understandmg, however, there ir no s~bst~mt~ve diqq-eement between Mrrlunal~srn and TC13. For TCP doe4 not tnr 'what 15 \ad' 111 that ~ e ~ l s e - i f it d ~ d , it c o ~ ~ l d 1x0 longer holcl tliat what a \nd tun be &ec ted bv strong prqrnatxc proce\\es

What the trutl-1-1 ondrnonal pragrlatrst means by 'what n said' corres- ponds to the InflJrtlvt. tvz~th-~onrlztronul cc~ntent of the u t t e rm~e . '~ According to ?(A" the lntutt~ve trut11-condit~ons of an utterance are dected by frce prag~natic proccs\e~ A\\umlng thls 1s true, thls does not prevent us from defining ancttller notion of what 1s sad, contbrrning to Mlrumahsn~~, as uiggesteil above I n earller \;vribngs (Kecanat~ 1999, zoord, 2004) I used i;ubscr~pts to distirlgulclr tile two notion\, and I will do so here agaln Let 'wllat 15 \'~ld,,,,,' be the proposition expresjed by a11 utterance h e n strong pragnlatrc eifects have beer) cllscounted, 111 accordance wlth MinlmJLsm,

'O See my 'availability principle' (Kecanati 1989, 1993, 2004).

Page 20: Recanati, François. Truth-Conditional Pragmatics

and let 'what is said,,,,' ~onesponii to the lrlturrive tmt11-cond~t~csns of the utterance According to TCI" whit IS s\a~d,,, may be xflected by top-cjown I'ragnatic processes. '%IS 1s cornpatrble with Mm~malisrai underrtixd the claim that only weak pragnlatli et&ch car1 affect wwhat 1s .i;ud,,,.

Is the debate between TCP and M1ti11~jallsn-i fvierely vc~rbal, tl-reri? It depends, There are severdl vanants .i,l M ~ n r n l ~ l ~ s r ~ ~ , one of whrcl~ r \ ,I< rilallv incon~pat~ble w ~ t h 'rC:P The drhate I-~ctwren that form crf i \ n ~ ~ l r n ~ , ~ l r < i - r ~ n d TCP is subrtatltive. and ~t is to tliat riebate that I intend to t nntnbirtc.

The first vanant of' Mi~rinlalrsxn 1 t all 'Strptilatrve M1nun~hsn.n' (S- M~~l~mahsin) . It uses the al~xtmralr\t c omtr,rint as '1 cliterlun tor dcrllar~atrrlg 'what is bald' What is llterallv sad rs dr,fmed, m acclvrd.unce wlth Mlrlrxnahsm, as a truth-evduable ~ I O E ) O S I ~ I U ~ ~ ~ Y I _ ) ~ C C S C C / by the utterance in vlrtrac of It.;

cor~ventlonal nxearung and tire corlitext but withotit the hclp of arrv ' t o p -

dowti' pragmatic pro~ess Si~ice t h ~ t 1s 3. stlpulatit>n cont erntlig the usc of tkc pl-irase 'what 1s literally saicl', tl~crc rs 110 wav to cl15agree with S-M*nrmahsr~~, except on terminological q p u n d s S-M~rumahsrxl is vacuou\ly true We ( drr,

however, make ~t 'L little l e s rtlpiii.ihve by aug~liex-itr~zg rt w t h a n rxisrerrce

claim. The resillt~rig v x n , S*-h41ui1nallrrr1, al\o my\ tlie m~nmn~ri~st rcjsii;r ramt

to define 'what is said'. but it J J& to the ~iefiri~tion tile IOIIC)LLIIL~ clrl~ni tlui the nonon so detined has a rrrorlclnpn~ extcris~on, t11,xt is. that therc r 5 ,i ievci oi

rnearnng that 1s both n~irnmd ant1 prc>yositiorul Accorclit~g to tlitc vrew . a11

utterance does expess a cornpiecr. propo,~wori (hence c3creurlrnc.s a hncncllrr

horn pojssible world to truth-valr~c\) tn vlrnte of its coriveritlot~dl mcarung ad the context but wthont the 11elp of any '~op-down' p~~gn , i t l ( pro( C"\S

When S-Minimalam a strerlgtllened rrrto S*-Mmimahin~. non term- nologcal disagreement beconlrs pos5ible. I~iclecd some raiLc.~i to r l~~s of

Contextualis,.n deny t l ~ e existence i-I,um wliicli ctlstmhanhcs S* N~ii~mxl- ism from S-Mm~xndlis~n 'They hold that no proporition (.an he euyrr\setl b y a natural language sentence w~ttlout the lriterventlctn of top down prawiat

ic processes. (See below. S e ~ ~ c m 8 )

Though it conIficts w t h sonle h r m of C o n t e x ~ d i s n ~ , S*-Mmralrsm n stdl compahble with TGP; for tire level of nleatm~g it pos~ t~ , which satl~fks the m m & t constrmt by defir-iiaon, need irot be the scxme level of meanlry;: as

that whch concerns TCp, namely the ~nturhvc truth-conditions of the utterance (what n sac$,J, hence there need be IIO cotkba~~l~tlt~rl between TCP's n o n m i m a h t ch,~rat-ten7ation of dza t a said,, mulci 5*-Mmrrilnlanl The mew accordulg to w h ~ c h tfrrre are two cy u d y Iegitullate r-rotlorts of '\clhas:

Page 21: Recanati, François. Truth-Conditional Pragmatics
Page 22: Recanati, François. Truth-Conditional Pragmatics

I N 1 t i O 1 , U t I I O N 15

The debate l>ro;vcen TCP mci f-Mir~~ni;rli\m (hcrealier Minrmali\ni) docs not merely be,lr upon the lntu~tlve trutll-c ondl t~o~lal contclat of lrttcmsice\. It dso bears upor1 the proper cfi,lractenz,ttlctrl of top dtlwn prrlgnlatlc processes (those which are respo~lsible -f;,r \trong pragrttatlc eff?cts) TCP holds that such plttcesses may, m some c ~ ~ s , rnt~rjere unfh senzantrr comf)osztrun ~ i n d take plarr loc~zlly Ninin~altsrn i~~amt,llns that wh,it 15 a rd r i llldepcntlerlt of such processe\, whlch take place orlly dfter the truth cond l t~on~ l ~onterxt of die utterdnce has been compo\itron,lJly detcrrnl~jcxi.

Accordlrlg to both MulunAl\rn m d TCP, '\what 1s s'11ci' 1s \upposed to be the input, o r one of tlie Inputs, to tlie pl'lgnatic ~tlference tlut enables an mterpl-eter to d c t e m n e what the yeaker rilearrs (1.e what \he rncam by savlrtg what she says) That n one of the job\ trulh-cond1t1oitn1 corltcnt 1s

supposed to do for us. The ~nference in cluestion 15 'post propositron,d': it

preruppose\ the ~dentlficatlon of cvti,rt r \ said. \lrrce ~t uses ,n J premlss thc hc t that the speaker has sad what che has s ad (That I? one reason tvhy what is said need? to be available to the I'tnguage uwrs ) C;nce,ln 'ccrrner\atlox~d ~ q h c a t u r e \ ' are clearly p o s t - p r o p o s ~ t i o i r i t h ~ t sense. they ,Lre generated through an ~riference whlch u\es as a prerrliss the fact t i r d t tire spcAer ha\ s'rld t h ~ t p. For exmiple, if, in rcportlrig about d ti-lend who haijl~st hdti A car accident, I say- 'Fle h,u a broken finger', I ttiereby rrnply that out friend 1s sull alive, that he is not pardlysed, etc These \quantlty\~~rpl~c;itnres cdrl be tlenvecl througl> the follotmng type of irdfren"c:

I Thc speaker hay5 that our f iend hx5 ,I t>rt>l\en fixigtzr, .uic1 \ ~ y \ only that, 1 If our &end liad anyttring more \er-lotts, the ipe,iker shoultl h,we sad so; 3 . 7 herefbre our frlend does riot Ii;l.ve dnyth~ng Ittore ienorls tllan ,I

broken tinger (so lie 1s riot dead, pdralysed, ett .)

IIere, ur~cioubtedlv, the speaker7\ llavlng sard what be h a >did is one ot the prernlsces tlsed m denv~rlg tlic ~rnplrcatures

Now, on the uunrnla&st plcture, what n sac3 I\ only ;lffected by \aalr;lhon top dowrl prwnahc procese, come rnto the plctrirc at the ulext ~vherl ~ v c get horn what IS \ a d to what i s xneant In other wordy CdtLlr~ihOll IS pre-propos~ b o r ~ d or 'pnxnary' prapmti(. proce\\-~& pro( ei5 tll'zt cofltnbutes to dctenrrimrtg

Page 23: Recanati, François. Truth-Conditional Pragmatics

- '1 able 1 I. 1 he nun~xnathst picture

Sentence nieaning (character) 4-

pre-propositional pragr~tatics (saturation)

= what the sentence says i-

post-propositiortd prwatics (moclulation, ixnplicatures, etc.)

= what tile speaker tueaxts

the propozitlotl expressed-wlille n~odulatlon, miphcat~ires, and the B e are all

port-prop~~"h~"jltod ((\ccond;vy') So we get the picture showri m Table 1.1 TC'P's man objcct~on to that picture 1s that 111 modubt~on cares, the

ortly tl~rrig that can be i~iferreci $oballv Gom the speaker'? sayng that p (wilere p 1s the h n i m a l propout~on' espresed by the utterance. for exarriple the propos~tion that there 1s a [reall hon in thc courtyard) 17 the fact that the speaker does not rriean that. 1Vhat the speaker actually means cannot be ~i i fer r~d &om the tjct that the spe'lker rc raying that p (paw 5orr~c ext14 prcrrusses) 'She glob,~t-intlereiice ~nechanlsrn which works m the case of ~oovers~~tlon~tl iinplrcature.;, 4,s In (I)-(3) above, does not work here To get to w l i ~ t thc \pe;ikcr nledxis, there n no otlier opt1011 but to backtrack, ,rpplt local pragiiiatic fun< tioris to tlie literal sense of tile words (e.g map 'hon' t t r

'repre5exitatlon ofhoii'), and recalculate the nleanirig of the wllole m order to deterr~iirrr A difTereilt propos~non, more appropnate as the content of the speech act. Mod~rlatron processes thus construed are local, and that means that they rnterkre wit11 semantlc composition, despite thelr 'top-down' n,lture. l h e y take place I n the course of sei1iar1t1.t cornposltior~, so, for them to t d e place, wmantic corrtposition does not have to be complete; In other words, the ni l r~rnA proposition (whcb n the output of semmtlc cornpositlori when there is no pragrrtafic mterference) does not have to be computed In the process of understandmg what the speaker means.'" 'l'tie top-down prqnatic pro( esses responsible for alodulation are pnmay

'Z One rrug11t argue that the minimal proposition needs to be ronlputed in order to trigger the search tor a different proposiuon if n rums out to be pragmatically umcceptable as candidate content of the speech act (Bach 1994: 158). But I deny this (Recanati 1993: 265-6, zoo4: 33-4). Modulador~ functions may be sdicnt enough in context co come into operation directly, without tjrst considering the minimal proposition, rejecting it and then backnachg (see Kecar~au iqggb).

Page 24: Recanati, François. Truth-Conditional Pragmatics

pragnatlc processes, rather t f ~ m ~ecnnd~sy procesres presuppo\lrlg the prior ~ d e n ~ c a t i o n of what 1s sad

7. TCP as a form of C:oxltextuaIisn

In the case of mdex~cals, the prol3<c"slt1ori"l corttr~b~ttlcr~i niadc by an eupre*, SIOII depends upon the context and is nor iirlly deterrrxned by t l~e jcoii~text-

mdependent) nieaning wlllch the explesuon (type) possesse\ In virtue of the

senlatltli. rules of the latlgwagt- Conte~tunh~m 1s the phllosupl~rcal posaran whlch gerieralizes that feature ti) \orchnnry>xpress~oris It holdb that, ln general (1.e. not onlv rn the specrd caw of ~ndemcak), the propo\~honJ-l contnbutlon of an esprecslc,rl is IIC)C: fully deterrr~rrrecl bv the rrlc*ailanrt

rneanmg converltlonalv assoc~atcd ~vlth the expressloft type 1711~ depcrlds upon the context

There are two (main) versions of C:ontextuahsm 'ICP 15 the wc,kLer of the two It holds thdt the hngtiirc~c lnc~ri~ng of an (r)rdtrr~ry, not)- i r~r iexrc ,d) expresslor1 izecd p~ot he whdr the explrssron t oncribures to p r o ~ ~ o \ ~ t ~ c > n ~ l corttent. Raci~cal Contextli,~li.irn (RC) Isolii\ that ~t ritnrritt hc w11,rt the expresson contributes to propc,srt~onal (ontent. Although, rn thl\ boob, I make a caw for 7 CP, 1 dr11 symp,itiretic al\o Lo thr stronger porihon, wh~c'l-r 1s the reason why I inentloii It here I '

TC:P and KC can be c11aracten~e-d 111 term of a certain d\u1n113rxcrrr whrch they both reject ( tho~~gh fix d~firel l t re,L\om), 'itid whch ~ V C can call tilc 'Fregean assumption'((FA). 1 ct us as.;ttnie that uttcranres eup1er.i ]propo\r- tlons' or 'tlioughts', arid that thrle propout~o~is/thuugt~t~ ;Ire m,xrde out of,

or can he anallrsed mto, certam bt~~ldr~rg blocks oi cuxatlttlerlth, ro be i crlled sensa. 'The standard assumption 111 tlie hceratrirc \teniIri.ing 6-oiii Frege rs that

(FA) what an expression corrtr~butet, when rt rr used (together m t h other expressloris) rrr rnjklng a ~onlplete utterdnce, 1s the sense wlr~c h rt

Independently possesses m wrtxe of the coriventioris of the 1an~:ige

l 3 TCP and RC are both Instawres nf wh:~t Cappelerr atrd Lrpore cdl 'K~,~dical Contextudisrr' (see footnote 3 above). Irt Recanari (tyy3) arid Brziridrlihout (mozl>). the phrabe 'T~uL~I-.Coridi~ional Pragxnatics' is used more broadly than I do here and serves as a cover tenrl for ~oncestrzalrsr posinnrls swh as TCP and RC, without distinguishing bzorircrn ~ixem.

Page 25: Recanati, François. Truth-Conditional Pragmatics

This yresiryposes that the C ~ I T Z Z ~ ~ Y ~ ~ Z O F ~ S qftlrr krrgti~ge t~ssaciute expressiorzs ivith .ymrt;s. tlri.; the 'Fregcaai przsrippo"stitrn7 (PP). Is FP tenable? RC, the more ritdicni tri"thc two forms of C:oritextualisr~r, rejects it. TGI' accepts it brrt still rcjccti, ttic Fregtr:tn asstinlptjoil

Tcr inti~tclut e KC,, con\xdt.r the n m t 6-ecj~1rntly used expressions, those

cvl~iclr cxlrrlrrc ,r l ~ ~ g l i C1eg1-ec of polysel~~v 'iXjj.l.it (lo eiiey mean? A number of sc L-ioi,u\ bcxlrcswe therr tliearirrig 1s. sctrenlatnc arzcl lras t o be fleshed out on m y

piu t ~ t uLlci iuze rlui \ug<est~ t h ~ t , perilsip\, thcrr cunvcntional meArung is not a tirilv fiut'ig(~l scriw Cnn we Argue tlrdt r&ey &re unrb;frlclu, between a ntrrr~i-rer o f i j r i t u r i t \rn,esi Tirat 15 rrot c)bvior~s because r t does not xern t I~at thti-c ILI LIICIY~TL' hst ot S U L ~ senbc\ a\~~~lahk,je bur, rather, a cont~nuurn of poss"ib1e senses to which onc can creatively add in arr opcn-ended manner.

-T'Fme is rrot to say t lu t the nlc:~rlil~g tri' srrclr a11 expression reci~lces to an

,ibstr:rc.r sciicxn,n: tile rspressio~r is rxrrdoubrt:iily '11-so associated in memory w~cli c"oa~verrticirlal ways (of- using it in C C ) / I O C ~ T ~ O ~ S with (nmre or less)

t$etcr-rnr,natc~ senses. AU tliis--tile abstract s c l ~ e ~ ~ r a or scherriata, the calloca- tior.;, t1.o~ seiises- ~rgu;ibly goes i r ~ t o the Iix~gtlistic rrieaning of the expres- \ion, wlricPt st,trtj locthlng r'~t1ser nztX\sy 0 1 1 such ,L vrcw, the rmlemnlg of an cup~c\von doc\ rrc~t have tire tight Ytcirrr~~it' to be w h x the expresslor1 i oncrrbeitc.\ t t r ~~ropo\itrc>nal c oilterrt In orllr~ \vord\, l ln~i~s t rc . nleanrng a ~ c liot \cr~\e\ jrkltrtig11 they rrliiy IIIVOILC, rrltcr &Ira, \enses): tlle Fregcan

p~ri"strl~o~tlt"or~rrlet\t therefore be rejec red 1 hdt I " ~ L ~ S O the ( OIIC~LISIOZ~ one can 191,rw from I'~rl~rl~uti'\ c ansrclteratiorrs reglrriirrrg r1atnr;rl k ~ n d terms arld

Icwic al wrriarrtit \ Inore ger1cr,3Ily At io~clrrrg to I'utndrrl ( I V ~ , ) , the context- sntiep,rriticrri inearl~irlg of r r m t nouns ri .r ~\ccco~' ctPns1stlng o f a bunch of c l ~ rncrlil, i r l c 111~irng "~CIII,~IICIC II IJTLC~' JTICI '1 'itereotype' whicli Itself

l ' l i e two v r c w oflt:xic;il scrnniltics X 1x:tve *just ~~ien t ioned nray cveU he

wrcrrrg, oi'rourse. When it corncs to lexc:;r.L st~lrx;tx~tics, nearly evrrytlrring is 1113 ilsr. gr:ibs. 'TIIJL, however. is pri:cisel?/ li;C:'s point. As theorists, we have ;in rdc'r \;6 l i l t "ir~ize\ &re, that I\, ~vlrdt WOI iP\ con tr~hutc when we speak W e L ~ ~ c i w , anorc or Itass, how to tilocicl t i l d t llut we k~~o\ i \ r very llttle about what cvorii\ rX-rt~rrlwivt.s lirearr xiid wt~it: r t ' ld t~~tl ~ I I C ~ C I\ bet\*'ee~r word meaning

,rrrcE c-clntrrbutihtl seilse I n mew o f tlre litruts of our knc,wledge, rt 15 reason- alAe t i r grkc up rile slri~plr@~rbg ,l\sunlp&lon FI%e rriher~ted frorn Frege, In

o r d t ~ i t Iv'i\t. t o \ ~ i i t ~ r l ~ i h ~ r ~ g \ c r ~ ~ > t ~ s e r ~ q ~ ~ t r ~ t , ~ III that 'Irea

Page 26: Recanati, François. Truth-Conditional Pragmatics

As we have just cecn, the f regem asrtmlptlon presuppose\ that 11ngulf;tlc

m e a a a g are aemes, that 15, propos~non,al constituents If we grvc up that presuppOslttOn, tve are left with the ~dea that lexcal Ine,rrrmg plays sorne

role in deterrn~nlrlg the \en\e whlc h IS an expression'\ c-ontnhutmn to the

thoright exprecsed Thrs idea car] he e-cpresd by \<iyIng that thc sense of nn expression 1s a fuac tion of the temcal rne,inlng of thdt expression and sorne Gctor x, where 'x' 1s whatever, in attdlt~orl to lexical ~rlle;lnrxlg, 15 needed to

d e t e m n e sense If, AS seems very hkely, 'x' ~ n c h ~ d e \ the ~o i f tm f in which the expression w tokerted (mil m p'lrtrct~iar the most 1rnport;lnt ~xnorig c otitex- tual factors: what one is, taiklllg about), then wc get ,t tomr of (. 'or~textu'~li\n) that 'generahzec ~ndexrtc~llty' That is RC.

How detnmentd wwotlld ;~ccept.inc-e of KC: thus underctood be to tile project of systcrnAj.ltc sem'~rltics? Very Iirtle, 1 tllxnk tndesical~ty does not

prevent us from systen~atically mdpprrig serndrlric v,ilue\ tbr the pdrtr to

senlairtic vducs for the wholes ~n wl1icl1 they occur Tile only yu A 1 1 f lcatlon mduced by ~ndex~cal~ty 1s t11'rt the senaantlc kal~ter for tlre gats x-r~iy not be gven In a context-irlclepencierit rruriner; they rndy not be cietenllined solely by the conventxons of7tIre hrlguage. Tl-nxc, In A tntth-the~rret~~ ii,lrnework,

the theorem\ of one's semantic theory have to be ~o~tdztzorlill T-~erftence~ wl~enever the rentellee at mue contam rnctex~cnls (Burge 1974; Wcin~teirr

1974; F-hggnbotham 1988) Gcneralim~g index~c~llcy, a\ KC: does, \~mply rnearlr that the theorems \mil altliizys be condl t~onal T-sentencec

In corltr&$t to KC, TCP, the rrlorc cor~serv~it~ve posltlon, accepts that the conventions oi the l,mgu,tge a\soc-late euprcsslons with senre\ I t accept5 the Fregean Presupposrnon 'TC'P rreverthele5s rgec ts the rrege'zn a\ummphori

that the senses wlr~ch 'ire the meanlngs ot expreSsron\ ale ,tlso what tilese espres\rons co,~tnbute when they are ured (togetller with otl-ler exprcssrons) in rnaklng a coniyletc utterance. TCP Iloldc t h ~ t nil exyresclorl may, but need not, contribute lt\ \erl\e--tll,tt 1s to s ~ y , the remc ~t ~~rdepcr~delitly

posres\es UI vlrtue of the c onverttlons of'tile Innguage, it I r r q ,iL\o contribute an indetin~te rltrrnber of otCicr senses resulting fio111 ~nod~~la t lon operations (e g free ennchmenr, illetonyiillc transfer, scilse-extem~on, ctc ) applied t o

the proprietary sense 7 hi\ 1s 3 rbnn of Contestuailsnl, bec'~ri.;e rrroclulatlort itself 1s corrtext-sensit~ve. whethc~ or not modulation corrtes ~ n t o play, ,rnd II ~t doe\, whlch modulation operatron take? place, is ,t matter of corttext It

Gtllows that what a11 expresson actudy contribute\ to t l ~ e tlltxtght ex pre\\ed by the IittcXr,rtice In wl11clt ~t ot c ~ u s ir ~Iw,iv5 rrl'lrtei of'cortteut

Page 27: Recanati, François. Truth-Conditional Pragmatics

8. Llefinixlg 'IXadical CC=ontextualism7

In earlier wntmgs, 1 described Radtcal Contextuallsm as the view that

strong pragm~tic etfec-ts are cnrltminahle

Wlthvut pragridtrc Intrunons of the optional (top-down) vanety, no detemnate propotrtlon wotdd hi. expressed (Recanati zoo4 96, see also Recarlati rgy; 239-60)

It w not enyy to makc serse of t11,tt characteri7ation, however. The very nohon

of A top-down prqp~atxc process of 'niodulahon' rests upon the ldea of

o p t ~ o i ~ ~ h t v In sdtur'itlon cases the pragnidtrc process of ~~11ue assig~iment n rn~mddted bv tile expremon ur vlrtue of its s t a n h g meaning Not so wlth

nt~odulat~orr orte rn~y , but necd not, tliodulate the sense of an expression.

t.leytendrug ort the context. Iri contrast to saturanon, whicti 1s ln~posed bv tbe

senrutlc s (Ilv dre st~nclmg niedxung of context-rensitlve words), modulation

stLlj15 under i)ragn1ntic col~tr01. lt 'is apragmatzcully contuolled prdgmatic process,

rather thai A l i ~ ? q ~ t ~ t l c u ~ l y cotilrollrd pragnlahc process @ke saturatlon)' (1Xecanat1

~ 0 0 4 I jG) I OW, then, could it be thdt moduiahon IS n1~1d2toni7 Ifmodulat~on

1s rrct esrary to got A co~iiplete proposition, what difference 1s left between

snturation ~ n i I moduhtiori;

An answer to that cluestiori was skctchcd rn Dtrect R<ferefzce and elaborated

In I21te~m1 M P L ~ B Z ~ Z ~ ~ Illere I dncussed three radically contextuahst positions,

one of which (the '1)rapnatic (:ompoc;it~on' v~ew) 1 descnbed as t?,llo.;vs.

111e Y r r ~ g r i ~ ~ ~ i i C:ornposihort mew (PC) construes tl~e pragmnc process ofmodu- lntiort as optronal But it cotlstnles ~t a5 optlonal only tollll r a p e ~ i to the tvordwho~e 5en~e 15

modtiluted Ifwt, < otx~cler riot word? In nolatlort, but the complex expressions in w h c l ~ they occ ur, we see t h t the pnnktry pragtnatlc processes of niodulanon are not always Lonungent and dnpe~~salrle, but okea essenhd Even though the lxngulstic meamg of a given word (or tire senuntic value that results from saturatlon) could be the expressed smse, rhU the process of \emnmc composltlon, 1 e the puttlrlg together of that sense wtth t h ~ \rt?t2nhf V ~ ~ I * T h r nthpr P y r P c S i i > n S , r?~?nnt r ) rn t -~~~I tlnlpsc. 7pprnpn3t~

acijust~ncnts ~xkeplac e \o as to nuke the paxt\ fit together within an appropnate wl~ole On thts new worck have meamrigs whch could go cfirectly Into the mterpretatlon, wxthout rnoddaho~i, but it IS the colnposition process that forces modulatlor~ to take

pL~ce, or at leact mv~tes rt" oken the meamiw of indmdual words do not cohere by thmlsdves, xid can he fined together only by urdergol~ig a process of mutual adjustnlent (Kecmatl ~ o o q 138)

Page 28: Recanati, François. Truth-Conditional Pragmatics

There 1s a sense In which modulatroi~ is nec rsqnry, but that i\ not quite thib .;rrise la whch saturation 1s With satura~on there rs a st'nlanhc gap and ax1 ~IISLIUCLIOTI to fjll the gap-both the gap and the Instnrcoon Iremg part of t l~e I i n ~ a ~ l c nrean~t~g oi

the expression Wit11 modulatiori, therc need be no gap nnci there a i ~ c t lrkswucborr

to search for sorile contextud filler Skre exp~cssion means sornethrilg, a1u.i thxr mean~ng could go into the inleqret3bon- \O rnodu1~1t1011 1s opt~orral--bt~t tcr determne 3 slutable semc for tonlplert exy-tresuons, we need to go bwond the meaning of i~i&v~d~ial words creahvelv enr~rh or otherwise adjust wh<it w e are

given in vlrtue purely of. IIngmst~c rneamlrg We nlust go hevond hrlgrrls~lc meaning, w~thout be~ng 11ngrrlstrcsUy rrrstrrts ted to do r o , d we nr e to rr~ikc sense

of the utterance. (Recariatr 2004 139)

On th s view, rnodubcion, droug1:tl opt~ol~d, otten tunlr out to be n~andatory

As M a r h s Egg puts ~ t , the proces of \err~<inac corz\truc~oxl ~ t i~I f - \o r r~e t~nes

generates a form of semntrc unclenpec~fic atlon w h t h caririot be tra~ed to any

lemcal rterr~ rri partrcufar soi-rie elc~zient wlmh o riot requlred 112 vlrtrxe ot the

lemcd semdncics of ,my expresuon In the renterrce stdl ha\ to h~ L O Y I L P \ ~ I U / ~ ~

provicfed m order to niake the rnteqjretatlorl of tlre wntencc cofrere I n 4uc h Lacs, 'rernterpretatlon augments the sclriantlcc <>Can eupres";c)n wit11 r~itzble

estralcngulst.lc ~xzfomxinon ti1;it btrfi.cn the gotentl,rllv confl~c tmg rn,xr-clrr~l rn the semantics of the expresuon' (Lgg 200, s~l) Che uonta tud prvvi\ron ot the rrucslng element I\ a \trorrg pw~m~dtrt ofl'ec t since it n not ,rcIucvcci rn

ac~ordance w t h the stand~ng rne,rrlnlg of ariv lemc~l itern, yet rt n uken

tnaiidatoy-wthout ~ t , there ivottld be some foml of senr'lriac ~~usmatclr

The problenl i s wlth the "otte~r' 'OOl-I' IS not "~1.cvavs' ICtl~cre <ire m e s ,

however rare, rn which the rerltence rxpiesser a plnyosttlorr slrnplv rn vrrtsae

of the standrng nleanmg of wt)rd\ md ti le context jhr~t w~tfrorrt conteut~rai

niodulatron), then the rlotron of a ~ x u ~ ~ r r r i a l proyosltlori l a no t iricc~hcrcnt

after all So tlie Pragnat~c Cornporlrrorr view doer rtot wccecd 111 c\tablnl~

mg tlie ~t-tnchisrot~ that strong psaglrlatrc eKectf are rnelrnul:able

Thc other fomis of rad~cal conrcxcnal~sm I dcscnbed 1x1 I,tfertzl ,@IeantnLq, n,mely the Wrong Forinat vretv (VVF), and Medn~ng r lilnunatlv~vn (Mr), are varlants of KC '1s 1 ltwe ciitrcnbed ~t 'above gvmg up the 'rregear~

presupposition, they hold that word- type\ are not assocl,iteii mrtth scrr\es-

senses have to be cotlstnicted 111 conteur pragtndt~c'~iIy It 15 ikr CIZI",'~ZTII~V of

mews that 1 now wish to recerve the Lbel 'Ra&cal Coiicextuahsnr' It 19,

however, rmsleading to drscnbc sucli views a\ holhng tliat strong przrgnatrr

processes are ~nehmmable, for m RC', as 1 nleilcioned In li2terul ILlci~ri~u~q

Page 29: Recanati, François. Truth-Conditional Pragmatics

2 2 C i h A K i Ef i lLAi i-i) COiuil iT1ji.N 1%

(2004)~ the C I ~ S ~ I ~ I L ~ I O I I between \&tux- rtiorl ,tnd n~oclulat~on conapser, and wit11 11: rlhc dntinccron be.tweerl wc,th and srrcmg pragxnzltlc efikts.

7 (l'lr~re L~IJLCII i c)rlct~ments' are drtitled il,y IQerry) m aspect^ or elements of the irIterI?rcwtloxi trl. A \eriteuc r W ~ I L ~ die not tbc 111teq>!-ct~t1011 ofmythmg In the sentcrrc is dxcy are corwtlhient.. of the rntcrl,retaQorz correspondmg to no i I)II\CILU~'XXI, ~'il tElc \rxltencc berug ~nterpreteti " It tc customaxy to equate free c*a,ni f ~ ~ i c r i t /A q)cA(-tr\ of'1110d~lla~o11) to tllc ~'FTCIVISIOZI of 'unnrtlculated con-

?;hcuc.rra7 rrx Pew':, scant. I l ~ a I \ not quite nght, bowcver.'Wodudnt~on c,perdtc\ on tllc \er>ses of e\pressloris, s o , even tl~ough mocfulat~or~ is a j k e

~ u x g ~ ~ t u q p r ~ c a s , that 1s ,I proces;:, that xs not triggered by ~mgi~st lc nteaillng, r r l coal"tr,r\t rtr the iontextual ~ss~glnrrxrnt of'v~iues to mdemcals, stlli the senses

tia,it dec proviiicd ~hrckugll x~lo~lulatloxi JJe not 'tl~~art~culated constituents' of tlic iritcrpirt,inorl, ttecd~xse tlzey < c > m i ~ i > o ~ ~ i j t o \owie con\tlcuerit tn the scrrtciitc. ( v ~ r the C X ~ I ~ S S I O I I of W ~ U C ~ I CIICV arc the rnod~~lated sense)

I l.iet-c, AT(" C M ~ O ~ p e i ~ t - i .isc $11 W~ICIX we inlght \ t~ll be wllllng to tdk of uri,~rtrciil,tred ~on\tltucnts. E-rrst, rnc>clul,rtlon rra'w operate at anv level, so tlvesc rs, rr-r pnlr~ ilde, pos~~I?il~ty of ~ I O C ~ U ~ J ~ I O I ~ 'it the topmost level as well.

Jr i \ t '1s thc \cn,c:, of'cor-rshtuerzt exprcc.iiuri\ 111 the sentence may he modu- lated, tlre rcrrw of the ccorripltte srrlterlce rtjelf m.ty he nlodulated. 111 p.z~t~c"d~,ir, it 1s p~ssrblc to 'tkreIy erlrrci12rrut only tilie sense crf a canstltuent,

but also tire scrisc of the globdl S C ' I I ~ ~ I I I C ctich 6ee ennchnient at the topin(>\t icvel wu~lid corrcsyo~lci to the p ~ o \ r s ~ o r ~ of u~urticulated constltu- erm ('oxrsltic-r, 'swr eu'~srrple, the mctelrrcllogcal \entente 'It IS rasntng'. In C:liap;prer 3, i: dr gue tfi'lt. the phce whrc In the mctcorolog~cal event concerns IS

~n asprct of. the ixitepretntlon oii' the scntenr e wllich IS not prov~decl bv 'sdtiiliit~ng' 21) d~gl l~r le l~ t plitce I I ~ tire lex~c~rl iemantlcs of the weather

verb, but, ~dther , tllt-o~~gh tier ermt hri~c.txr There arc sever~l options, tllorrgh Fr-cr c~rrncX>~ncnt rnay ~)pera~e at thc" I C X I C . ~ lcvel, so that the vense

'" i,&r iil~ury contmiporaw philosopircrs, icken-y construes the inirrprrtariot-1 of an utterance as a .rtnrct urcd prt>pvsitirm' On tlits rtnrcruscd j)ropr)sitlotl ii:ur~c~v*>rk. see King (2007). On 'uilart,cidatrd con~uturiii .~' , rce lkcrxy (LOOO: 172-88 .trirl 2 . ~ 2 - h )

'" in t.driit:r wrrtiilgs, tC.fi. Ktrr,:t~~;rri ( I V ~ ) J , AC)UL). I ~ l i d r oilti-ibute to the ~onfis ion, by cirscrib~n~ h e cnn~ltiiieiie :L bri11 ( i f IIIOCIIIIIEIOII- 3s the provfi~otr I ,f 'irrr:uricrri,itri!i cumtiturtirs'

Page 30: Recanati, François. Truth-Conditional Pragmatics

-- riu I t c c l r ) u t r i t f N ~j

of' the verb 'to rain' is morl~~lated: 'rain' cotilc\ to riiean 'rean ~t - l ' , wlirre '/' Is the contcxtually provrded locatxon. ('Tila1 is die 'var~~dtc Furrctiorr' anYilyai &&red tn Recallati 2002 and iurn.rnan;red below in C:l.wptei 3, @.3 ) h o t h e r optton 1s to treat the proc uss office enr~chrueiit as tdl\mg placc at the topmost level: the verb contributes lts regilla semc, but the iilterpreta- tloa of the $obd sentence is e~lr~ched through the provolotl of ,112 extr,l

conjunct m ttle sc ope of the event quaxrttfier As I br~efly rxlexztion m Chapter 3 , there 1s vet another option for 'It 1s

ra~ning', and ~t correspond to the second type ot'case tn .cvl~lch we srr'xy be wrUlng to talk of unxtlculated constituerzts. In l'er?pccctivi~l '17zj"lot<q/lt (2007), I argue that the situation w ~ t h respect to wh~ch thc rontellt of-dtl utterance 1s evaluated-the 'circ~irrl+t.ance of ev~1uation'--rer~iains un,rrtzcul,tted, =lid 1 advocate a liberal vlew of what counts ds the c~rcuinst~r~ce of'evnlu~tioll. Xn parh~ular, tlie placc wh~c h the nreteurologlcjl event ionc crns i n ~ y be treated zs (an dspect ot) the ~ ~ r c ~ ~ n ~ s t a n c e of e v a l ~ ~ a t l ~ ~ n fbr the stdtc~~Ielit

'It a raining'. In thlr franiework, the lotatrorr of the ralnlrig event IS ail unartlculated constituent of the irlterpretdtiori

In Prrspt.ctztla1 ?7zozi",,ul1t, I corqecture that otzly aspects of thc utu.inott of

evaluat~on remim ur~aalculated. I d~st~ngu~sh between the cenr,ultlc- Iriterprc- tanon oftbe sentence wldi re5pect to context- -the loktorz, x I dl it -,mci the Atlshn~dn propo5lhon consrstlng of the lekton together with the situ,*tlon In \vhich ~t ~s to be evaluated (here, a cermn place wkch the speaker tac~dy refers to). The situation 1s cor~textud.ly detenmned w~tltout correspulicfing to any- thmg In the sentence whtxc content n to bc evaluateti w ~ t l ~ respect to that situdtton. So unmculated corlshttlerlts there are, but they are not constituetits of the lektc~n: they are conshtuents ofthe Aust~nlan proposzt~on. 111 contrast tct the Austlman propos~non, che Iekfc7n 1s 'fblly amctrlated'. 7 111s me,m\, basicnlly, that I take ti-ee ennchrnerlt ('mti modulatton generally) to operate only loc,dly,

on the senses of\entenhal constttuents. Global or topmost modula~oa 1s ruled out: there 'Ire no unaa~culdted constituents in the lektor-all u~r~mculated conshtcrents belong to the sltudnon o f r v d u a ~ o n . ~ ~

'" Whde writing this, I realize &at I clon't have aproper argument to rule out topnlost rrrodrdatron; so I should remain agriostsc ori this issue. 11rde'd sortle ofthe procrsses dnebrribeci tn the 1)orik :ls arnenablc to a pragmatic treatment within TCP (e.g. exhausnfic;~tron or contcxt-shift) can certainly take piacc. at the topnlost level. The arguments prr>vrded in A~specrival 'licorrght a p n s t 'unanicul;rreJ ionstiturIir\ In dtc

Irktor,' are dl negatlvc. algunrerlts, dtte~npung to sliitw that the reason rriidufcd zt) i;rl for wsig~mpecern

Page 31: Recanati, François. Truth-Conditional Pragmatics

- - - - - - - -- Because, 1x1 the I-'erq?eckvnl If'hcttr~ltt (2007) f%arnework, the lektotz is fully

ara~ulated, ~t h a beell cldrrned that ~t cannot be the same tfung as the r ~ c h 'what 1s said' ctf Ott7rnl Meantug (2004) Thus, In a d~scussion of Perspectival 7hot@t, Max Kolbel cla~nis t 1 1 ~ what 1 cdecf 'what is said' In my earher

book ~11o~ild not be equated to the lekton, despite the Gct that Yekton' means 'wliat 1% sard' 112 Greek. The reason why the Irkton cainot be equated to what is s a d is that, actording to 1i1y earher account, what is said is non- mmirnal-rt 15 affkcted by pnrnary pragnmatlc processes such as free ennch- ment--while, according to I)erspecttval Thotlght, the lekton Larinot contaln any 'urlart~c~rlated conrtltuerit' Whatever unamculated consatuents there ,%re are corlstlmerits of the ( omplete conterst, or Austinian proposition: they 3f-e ar;perts of the clrcurristance o-t evaiuation. So, Kolbel concludes, it seems that my earlier 'what i\ said' sliould be equated to the Austman proposition

rather than to the lekton (Kolbel 2008). Rut 1 marr1thni t h t tlie lekton of Pmspecttval Z%ozqht is the nc1z 'what is said'

of lzteral Mearzittq Whdt is sad is, indeed, affected by niodulatlon processes such '1.5 fkce ennchrnent. Ths, however, 1s not a matter of providmg unartlcu- lacd conshtuents, at le'zst rnso&r as rrlod~ilatlori uke5 place locally Through free e~rrichtz~e~it, we malie the contextud rneamng ofa lexlcd item or phrase more speclfic t h ~ n its literal (convetl-t~onal) rneamng The output rense, r e ~ ~ ~ l t ~ l g ti-orn tree e~mclnlent, is k r tx ulated, because ~t correspond? to some- tlung ui the centrnce, r~ariiely the lcxi~al item or phrase whose meamng has beell rr~ide more speciiic SO the lektott cart be &illy artrculated, w~thout being 'rnu~mrdl' arrd tr~xirfectrtl by tl-ee pr,agmatxc processes In m y frafarnework, the 1i.kton 15 verv f;ar ti-on1 ri~~~imial, imce mocluhtion (-an come into play at every mge ~n the proc css of sctn~ntit cocilpocitlon from which it ongulates

I o. Overview and acknotvledgxr~ents

lhic book-& sequel to I,ltera.uE Menntny (zoo4)-consuts of vaious case

\tr~dics intencied to Illtrstratc the TCP framework. Sorrie of the studies lravc already appeared In pnnt and are reprinted here in revised forni. All of tlren~ give prick of place to 'modulation', a farnrly of pnrnary pragmatic

urlarhcuiated coristituena to the lektorr (rarller than LO the situation of evaluation) are no good and should carry no weight.

Page 32: Recanati, François. Truth-Conditional Pragmatics

processes that make it possible to ai$mt the rneaning of words dnd phrase5 Irz

response to conversatlond needs, by endow~ng &em wlth contextud senses dishnet from thelr hteral meaxlings

The relevant nohons and the t h e o r e ~ c d framework are mtrotiuced rxl

Chapter I , which tnes to slio~v that contextud inodt-ilatlon 1s comp,~t~ble unth (a suitably weak fonll o f ) conlpos~trorlalrty (Some tlicoreti~d back- ground on context-sensitivity 1s also to be found xn Chaptt~r 6 ) Tllen cotne

two case sixidles Chapter 2 deals n t h ad1ectlves, I argue t h a ~ the apparent lack. oi irrter-

sectivlty of c ertdn (predicatlve) ;~r?)ec trvec 1s an ~ l l ~ t s ~ o r ~ due to con textual vanatlons In sense, some of w h ~ c t ~ are tr~ceahlc to n~odulat~on

Chapter 3, which ha;, alre<icdy btcn mentlnr~ed, deals ~ 7 t h weatlrer

sentences. I criacize the mew that the> cxry an arwnxent slot fix a ioc~uon, and argue that the taclt reference to ,I lot atlnn which 15 tyyicd of \critences like 'It is rmmg ' (aid w h i ~ h afi~cts their trut11-concfrt~onr) n, agalri, ,I

matter of pragmatics. In Chapter 4,X pause to ~orlsrder aj/terui~tive ways of construing the T(,P

haalework, conesponding to wl~dt 1 ~ d l e d the syntactic and rile cerilmtri 1riterl)retahoil 111 Recan~tr (aam), and I nzentlon ari rritngLiirlg version of the svntactic Interpretation which would riiakt &c~e pr'ipsatsc pzoctLssei 11ngtlrs tic phenorrlena in a hv1y narrow scnct

Chapter j cleals with errzhtdded rmyhcatureb, a pl~cnornelion wtut 11 h,n received a good deal of attenhon reccntiy and .ivhcll has let4 m,mv rc.iearclz- err to cast doubt on the tradlnors,tl cilvl\ion oflabour between wrliclrltlcs and pragmatics I dwcuss the reldtior~ between vanou\ porr~ble apilroac1z;c.i r t ) tire phenomenon, including TCP

Chq te r 6 lntrod~lces the ide,l of ~orrtexr \h~Ct The phcnorncl~a~rl of contest-sisili is of conceni to 1X,P becdr~sc, like rnodulntlori, it: intcrfi,res

locally with seinairhc conlposrtlorl and yel& strong pragrriatic eirects; Anlong these effects are certaln quorat~on'tl ~~henomeri~r (irr partrcular,

rn~xed cluotation) to whicl~ Chapter5 7 and 8 arc dthvotecl Tl-tew pl-rerionie- na. I argue, can be accounted h r liragnratrcallv, i r l terms ot lwtlr rcrntexl-

shfi and Gee enrichment The research leadulg to tlus book has received filr~dlng fiom the Luro-

pean Research Counc~l under the European Corninun1t)i's Seventlr Frame. work Programme (F1'7/zoo7-2at~) / LKG gra~it agreement no zac) 441-

CCC, and also from die Center ik)r the Study o-FMind in Ndmre (TSMN)

Page 33: Recanati, François. Truth-Conditional Pragmatics

in 0 4 0 X mi mdebted to the cdrtoz\ of the- jourri*~Is ,tncl volurnes In whlch caaircr \wr\ro*rs af sonre of the c h a p ~ e ~ s prrvnou~ly appe~red for thelr 1x"rxlfissrorr to repnrrt thrrxl, to the rc-v.newer.; corrvrs~sc~onecf by Peter Moanci&rr3062 ot OAbrJ Lfr rwer~~q krcss G3r t he~r i o m l e n t s and sugges- r toxls , d a d t tr tlie many colle~~,rues J X I ~ f r ~ r n d ~ wltli whom I have been lortiixi,iec rl~oiijiln to Inrer~c t rrtirtitllly over thc y e m or1 the matters dealt wntt~ rrr tl^rr\ book. Kent Bach, I Tcrrrxarl CL1ppelen, Kobyri Carston, Gennaro ("lue~c kir;r, Xienoil tle Con~ulier, t'lllhppe I)c. Br~banter, Jkr6me I-Ioklc, Pat11 1 gre, 1 5 ~ 1 l,il-otrrtre, Mar~i~ei Cnar~ra-C'~ryrrrtera, Ceurts, Lulsa M'LJL~, AtlPlr Mer-c rer, Stephen Ne,tlc, C;ec>EN~\rnberg, Peter P a p , Stefano Plrdeli~, Plrrlrpire Sclrlexilrer. BcryCunlxl Spcctor. D m Sperber, Jason Stan- Icj, $owf Stern, Is~dura Stol.~tlovic, C11.1,~des fravrs. Marcel Vu~llautne, Dag Westcestal-d, 1 Icrrdre Wil\on, 'rnd '1 i iurex~ ot11en who wlU furglve me for tlyurg to Lc'rp the i ~ s t of narxres re~soxr'al~ly short. I &st but not ledst, 1 ZITI

grarefbl tc:, the rxterrrhen of dle I cvcrl'rulnle rx~tcrrl~rtional network "xplol- tntiorr ob ( "oritcxt lrl C:ommuxncatro~x', I t k c 1 by I Aurence Goldstc~n, for marly

prrod dr,c rrsilnris, rid tcj my collc,igue\ and studcxlts, both at Instlttit Jean Nniod jP~rr5) . i d Archi: (7t Rrxdrexw), i i ~ r rilrrr jrrptit 111 sen~ln'irs ,is well AS

m ccrrrvex\,xnorr

Page 34: Recanati, François. Truth-Conditional Pragmatics

Compositionality, Flexibility, and Context-Dependence

r . Two types of rule

'The corrtpositionrdity idea is the idea that sei-t~airtic ir~terpretation proceeds in two steps. Simple espressiorls are interpreted by means of Ic?x.ic~zl rulrr, which assign rneanings to them directly. Com.plex expressions are inter- preted by means of covrapositional rules, which assign meanins to them indirectly. as a f ~ ~ ~ i c t i o n of the nleanings of ttreir pats.

For any sirnple expression a, tlle associated lcxicnl nile says that the

interpretation of a is a certain entity vn:

There wd be a many ruler of ttr~s sort as thcre ace sixnple exprc5sicsrLs (or, rather, reading of s~rnple expressions1) In the language. Smce the number of slrnplc expressions and the number of rcnd~rrgs wl~rclt ,In nmbrguouc expres5ion h a are both finlte, it 15, in pnnc~ple, poss~ble for a fitlitc rnmd to get to know the mearungs of dl simple expressrorss of' the Iatlpage by learning zach of the lex~cal rules that are assoc~ated w ~ t h thein ut t h~s way.

The syntdx of natural language 1.; such that (because of recurav~ty) the n u ~ d e r ofconrplex eqresslons is not fin~te: for any expression of whatever con~plemty it 1s always poss~ble to construct a more coxnples expressxotl. So ~t would not be powble for a fimte mmtl to get to know the nieatllng of dl1

For the puxpose~ of applying rlle iotcrprrtatiun iunct~on, an ambliguous exprrssictn a etld4)wed witfi n reachrigs counts as several hornot~yrnorrs expressio~ls nl, n:. . . . u,, r:~cl~ of ~vhicb is r t~ tc~rck"d)y rnrxra ofa rille like (1).

Page 35: Recanati, François. Truth-Conditional Pragmatics

expressions of the lartguage, simple or complex, by learning that meaning dlrectly If we only had rdes hke (I) to interpret a hrlguwt~c expression, there would have to be ail ~nf~illte n~rmber of tbeni, and we could not learn them So we need A ctlKerent type of nrle than ( I ) for intefpretlrlg complex expression\

Just 35 tlle nu~nber of smiple expressions 1s fmte, the number of ways m whch &\tinct expresslora can be put together so JS to yleld a complex expres- sloil of the I r n g ~ ~ ~ g e 1s fltute In other bvorck, there is d f i t e riumber ofsynt;tchc r~de,, tllrougl~ :l?.v111ch an t d i t e tlurtlber of conlplex eqress~ons can be gener- ated The soluhon, hen, rs to p w each syn~ictrc nde tylth d cemanhc rule of a new sort--a conipos~t~ondj. nde A cornpouhond rule 1s somethmg hke

bvhere '*' \~%nds fior at1 arbrtrary mode of combination The rule says that the interprctdtron of the cornplex express~on a*P 1s the value of a certam functron f when it t&es as argtrrnerrts the mterpretatlon of a and the lnterpretdhorl of /3

A cornpos~t~orjd rtile ~ ~ s o c ~ a t e s with a paaiculdr wdv of combining two expre\slons (A 3 r d /3 furiction whose arguriients arc tlie nleanmgs (mter- pret,~t~trton\) of u snd 8, ~ n d whose value 17 the resultrng mearuiig (interpre- tatlour) for the corr~plex expression a*/3 1 h'mks to rules of this sort, ~t 17

possible to compute the meaning of m expresslon of whatever degree of ~omplexlt-v on the basis of tlie rnelmrngs of its parts If the parts are slmple, tlierr nlcdniilg l(u) arid L(P) will he glven drectly by lemcal rules such as (I) If tlrz put\ are thern\elves complex, their iileanings will therrlselves be de~~vxblc via co~tipos~tional rules such ~5 (2)

1x1 this GametvctrL, rrlterestmgly, the rnearnrig of a complex expresslon only depend\ upon two things. the meari~iigs of its inlmediate constituents

(the wnpler expres\iori\ into whrch ~t cdn be an'tlysed), and the way they are put together Nothing else c onnts 111 pdrticular, the n~eaivng ofan expres- siori does not depend upon the medliings of other expressions that are not its cctnstrtuent?, eveii if they occur iii the sarlie sentence or discourse Nor can tlie nieariing of a glven expression deperrd upon the meaning of a more complex elrpressron lri \vhlclt zt ocCLir) ds a constituent O r at least, this 1s

staridartily con\iderecl to be a consequerice of composit~onality In .+I com- posltlonal langtxage, we are told, the meaning of an expression depends upon the n~eanlngs of itc parts, In a bottom-up fashion, but ~t does not

Page 36: Recanati, François. Truth-Conditional Pragmatics

depend upon the rnearung o f the \v11ol~ to whrck? ~t belongs, nor upori the

rnemngs of the other p ~ m of that s~mcr whole '2"op-cfo-tv11' or "Xatcx,d'

rnfluences on akeanrrig are ruled out h y tile compoanonal pzclredure Vat, accordrng to some author,, such r116ucncrs ale precisely cvlrat we observe

2. Sernan tic Aexibili t y

A language eA11b1ts sewratztlcJc..czbtlztlt, rf tlltt follow~rtg c ondlt~ott 15 \~trsi?ri i

in that language, the inearling 05 a. W<XC~ may vaiy honr o c t rrrrrncc to

occurrence, and ~t nidy vary, 111 p,irtlculrrr, 3% A fi ln~ttc~n OI the other worik rt

cornbriies wltli Tlirougl~ scmAritlc Het~b~lity, tllc rnedltir-1g of exprcssrorr

may well depend upon the meatllrig of t l i t corriplex r r r w h ~ ~ l r r t occurs (top down influence), and ~t inay ASO depenc! tiport the mtdrnxtg of tint other

words that occur in the sarne complex (latcral mfluente).

O n e of the authors who has 111\1,tcrci thdt naturdj l ~ n g ~ i ~ g e ~ exh~bxt semantic flexlblllty is Jonathan C oherr, r r l A scnes c t f paptxr\ rn which lrc cntici7es mainstream ~pprt>clcI~v\ 111 ~ ~ I I I ~ ~ L I C S IIe give\ exarxrplcs lAe the

follo wlllg

Consider 'drop' III tile scnttxnt r\ (3) Most stitdents here h o p g ~ ~ g r ~ ~ p l i v 111 thcr~r find vear

(where 'drop' mems 'drop jtudylng'), (4) Most students here drop grograpl-~v iat tulcs I r i tlierr final vear

(where 'drop' ineans 'drop atte~~durg'), (,) M o ~ t students here dmp geography lectures rra&nga\5ig1meilti ul t l ~ e ~ r find v e x

(where 'drop'mea~ls 'drop execut~ng') (6) Most students here drop geogaplly lecnlrcu, re'td~ng ass~plr~lt~r~ts llbrani kes

m their h a 1 year

(where 'drop' rrream 'drop pavlrlp'). and 50 oil lndefiniteiy If wc a c c w t thar a

sentence can be as 1011g as we please, thcri rlrere seem7 no ptedrctable end to the vanety of expressions that we can put t~~eaim~@x~LIy afier 'drop', so 3s to i~npubc~ a

serles of hfferent mearnns on the latter word (Cohe~l ry86 227 8)

Accorhng to Coben, the verb %rop\takes on a difikrent mear~rllg m edch of (3) to (6), and one of the thing5 that deternr~nes the rneamng ~t tdItier o n 1s the noun phrase ~t cornblrles vat11

Page 37: Recanati, François. Truth-Conditional Pragmatics

IFle iort t-ri tIrorg that constltxte. ccrrttlng the gx'is5 l r cluite dlil^erent from, e.g , the

ion ~ i j - t j l l l l ~ t11,it (:~n$ritute'I cotnng n kc oiics t v ~ v to bee tiuh is tct IrnaWle wk~ac consntrrrc, cibt~v~ng tile orckt to ( u l sorxletl-iing I f iorl8eone tells rile to cut

g~ '~s i 1 rrlriil otit &nunti stah ~t wirh ;f kx11/(", or i f 1 ilnl oiclered to cut the cake 2nd B rirn tn7cr I T wldr 3 Idtvnrrto\ver, m e'xc /I c,rsc I wsjl have faded to obey the ortier (Sts,nie rc)Xo '23 3 )

Ac-tor-clirig tc> Seadc, ' C L I ~ ' means sonlerhiog itiiferent---bas digerent satis- Fat-titm coriditiorls-----in ' C L I ~ the grass1 ;rl11Z1 it1 'ctlt the &e'; arld tllat is hecaiisc: ttii: mcanirrg which the verb "cut' takes c'rn in a particular occur- rerric depexltis, ~xtter ili;l, upon what. i s said to bc. cut. Sirnihrly, tile verb "iikt:' t:lki:s c j t ~ .I tfiffer-cnt tncnrting in (7) and (8):

(7) 1ir likes rrry sister

(8) i-rr likes roastcti jrc>rk

The tint serrtcl1r.e tdks ; I ~ O L I C 'afPet-tivt.' Lihrxy ~ I I C ' ~ the sci:onLf L L ~ O L I ~ 'culinary' Fining. 'Ttrere is :rs I-rrtrch difjfere~~ce b e t ~ v e r r ~ the two Pitrids of state as there is bctwcc-rr dria processes ofcuait~g irlvolvirig grass ~ i r i d cakes respectively.

'T'l~c cxasiipies 1 have given st) Eir :rlJ iiivolve cr, transitive verb the (exact) nremirtg of wl~icl, ciepends upon the nt-rurr plrrase tlr:lt serves as its cornpie- r*>errt . Arr evr:rr nwrc i-irotitiitivc class ot'zsarx1p1t.s involves a4ectives the ((~xxct) rrreasring o f which deperrds rrg:ctrx the ~ O L ~ I I tbcy n~ociifir. A good car is n t r t good in exactly the s ~ ~ u e scrlse jn which n goo~i h011se is; a piece of Pugg:igi. is r r o c liglit 112 exactly tire siirne sense in which :i round is light: a big

uioirsc's \v:ry oi'iieirrg big differs, to sorrle cxtcrpt, lxrorr~ the cv;~y in whicli a

big eXt.ph;~rrc is big, ;r pink grapefruit i s not pink rr l thc sartlt: way----under the

sLirne aspect --as a pink raincoat; :r ~;LSI- typist's way of being Gst is not the s:mjc: as ,a ikst rurrrrer'r way ~ f t ) ~ i r ~ g I;ist; n17d SO 0x1 ;rnd so fbrtl,. In ;111 cases tire. basic ~nlc,-ixling of"ebc a~tjec.tivcx is iXcslleJ out diti1srerttly arcorrlilig to the no~111 i t rrrirtiiiies.

Scnnarrcic- Heuibifity arlct compositiondicv. 16 1 have characterizeti them, citbcxir ttr he rrl~rr\rrilly exclusive propert-irs. As Jerry Fodar p ~ ~ t s it,

. - I he cornrj)trsrtiibn;iiily rliesis says that corrlplex repri.sentaQcrlis inherit their content &on1 srnipic. orre\, #lot via. t)f:~rs.u ~ Z L I ~ tile [Besrbilityl thesis says that the content of a

srrilple [rci~7rcs=ritrcr(~r,j depertds (inter d i a l ) ( j i~ wliici-r consl,les [representation] it's

Page 38: Recanati, François. Truth-Conditional Pragmatics

-- C O M P 0 5 t T I C ) N A I 11 Y, F L b X l B l L i T Y , A N D ( O N I t.X I -1)i-J'bNLU NC'k 31

embedded In. CJearty, it can't bc that both are tnie. Something's gort,i givc. (1 odor '003 06- 7)

So, ~ f w e take r ~ ~ ~ t u r a l langizges to be colnpo\rt~t>nal, tor the rc~sonc < i J d ~ ~ ~ e d

above, ~t seemi th'it we mu\t re-,ir*,~lv,e the Aleged cx'xm plcs of seln'intlc

fleulbllity, so 2s to make ttrcrr* cornpat~ble wlth tire c ornyoslhon~l l~ the\%.

I will pu~\i tc t h ~ t line below But we rriay also, follo.vvrng C:ol.len, give up

tire \tar~clard. 'msulatronirt' approacli to senzailtrc coiriposltion auurne(l by I-odor in favo~tr of an alternatrve, 'rnteractronnt' approach:

According to the irls~tlationist account the meaning of-any one word that occurs 1x1

a particular senterlce is insulated against interference fruni the nle:iriiilg of any other

word in the satire sentence. O n this view the composition of;t sentence resembles tile c-onsrruction of A wd1 fi.i.0111 bricks of difirent siiapes. l'he result depends on the properties of the parts anti the pattern of their combination. 1-3ut j i l s i as each brick has exacrly tlrc sarntx sl~ape in every wall or part of 3 w& to wliicl, it is rrroveci. so too each standard sense of a word or phrase is exactly the sarrtc iil cvery sentence or part oia sentence in which it oct:i)rs. . .

Intertictionisni r~rakes the contradictory assertion: in come sentences in some

languages the meaning ot'a word in a sentence may be detcnnincr'i irt part by the

word's verbal corutcxt in that sentence. . . . O n tliis vicw tlie corrrposition o i a

sentencc is more like the construction of'a wall &on1 sand- bags of'ciifirrcnt kinds. 'Riough the size, stntcture, texture and ccx1tent.s of 3 sand-[rag restrict the mlge of shapes it can take on, the actual shape it dopts in a p:krtjcnlar sitttat-ion clcpcnc;is to a

greater ctr lesser extent on the shapes adopted by otl~er sand--bag ii~ clle wall, and tile same sand--bag 111igl:llt take on a sornewi~at difi'erent shape in anotl-lcr wall or i r l :x

~Sf&:ret~t position it1 tlie sxne wad. (thhen r ySft: 223)

According to ('ohen (1986 A?(>), 'we callnot construct 'I 'lelriant~t \ h r m y

natur,tl 1mgunLre doug tlre ranre lmes ar a semantics for n fornral sy5tern of any currently farrrrliar ktrd Projects 11ke r)avldson7s or Montagie', t'introt

succeed ' They cannot sut ceed pseclsely becaucr 'art~fic id Lnguagei, s,~tisf;r

a1 imulatlonnt aCCouIit wherc.,ls rtntural lang~~agcr recliilre a11 ~nterac-tlonlst

one' (Cohei~ 1986 224).

Whatever Cohejl r n , w have had ni nrind 111 fils talk of ' ~ n t e r a ~ t ~ a i ~ ~ ~ t

~ei~r'rntlcs' it is not clear to nrc that we Ir,lve to depart frorrr the standard

c o n r p o ~ t i o n ~ l trarlrework irdlented hoi11 Uavtdwn 2nd Montague if we are

to account for \ernmtlc flexrb~Ltv In thtj chapter, I wlll ilueitlon the

a~sun~ptiou t h ~ t sem,mtlc Aexrbillty i\ mcoml>u~ble wtth cornl)os~t~ortahty,

Page 39: Recanati, François. Truth-Conditional Pragmatics

3, Fodor cla~ms n~ the above p~rssage. 1 thlnk it IS not. I t 15 true that. m 3

t ompo~itional language, the 1~1eaning ofa complex expression ordy depends upoil the rmeanixlg of its irrunediate constituents and the way they are put together: nothirig else counts. 'This seen= to rule out top-down and lateral influences of the sort the interactionist describes, but, 1 will a r p e , it does not really. When Fodor (2003: 967) writes, 'The cornpositionality thesis says that con~plex reyresentatio~is inherit their content from simple ones, not virr vetra', he overstates his case. It nxry well be that complex representations inherit their cc:)ntertts hoot simple ones, irl a strictly bottom-up fashion, while ul the samc7 time sixllple representations l~ave their contertts determined, in part, by the complex expressiorrs in which they occur. O r so I will argue. I{ this is true. then a latlgrrage can exhibit both cornpositionality and sert~antic flexibility.

3. Standing mealling 21s occasion meaning

1x1 all the exanlples of setriai~tic flexibility I have given, it is possible and tfesirable to draw a distirlctiort between the standing tneanirlg of the expres- sion (verb or adjective) xi fired by t l ~ t : semar~tic conventior~s of the lan- guage, and tire amrsiorr rtzeaniug which the expression assumes on a particular occurrence. Tllus 'cut' has a standing meaning in English, and that stmding n~eanirlg is canie~f by all non-ictiornatic occurrences of the word; yet we need riot deny that 'cut' takes on a different occasion meaning it3 'cut the grass'aaxid in 'cut the cake'. Likewise for all the otller examples: in all cases we can draw a distinction between standrtg meariirlg and occasion mean- irtg. Note that the dist-irrction does not apply to tndy ambiguous expres- sions: in the case of, for example, 'bank', there is no standing meaning which tlie word--type itselfcarries, whether it is taken irr the financial or in rl;e other S ~ F S I _ L . Rather, here sw= disdnct wclr&-tjTe-, c;cll "r'h irs own (standing) n~eaning.~

.' What ai3out 'pulysemous' expn-ssmns like 'Iigttt'? Here, I would argue, &ere is a standing nleanmg whrzlr the word 'light' carries 111 tlie language, even d the vanous senses the word can take m vanous cnv~rotnrietits ("Iiglit sound', 'bght luggage', etc.) happen to be conventlctrlalized and sotnehow pre- cornplied in the Lex~con. It ~votlid be a mistake, in the case of 'light', to treat the multiplicity ofreadings a a mulnpliuty of words (honlon>my).

Page 40: Recanati, François. Truth-Conditional Pragmatics

The standlng nlearung a tf-tc "iremmg which the word (type) lras m wolat~on, In vrrttre of the t c ,rverrtrons of L ~ C l a ~ i g ~ ~ ~ g c I kt- o( CdSICI17

niexlmg 1.5 the rilearirllig lvhrih a11 ociurrence of the word t ~ k c ) on LXI ,I

particular linguistic context. 'What varies as 3 function of the other words iii the sentence is the occasion nle~riing, nitt the standing ~llea~liltg. DOCS this variation, and the cxisterice OF both top-ttowx~ and 1ater;rl ii-tflrrerrces on occasion meaning, corrflicr with ~or~1positiox1:ility? Arguabiy r l o ~ .

Corrsider the 'cut' exarrq3le. 7 '11~ word 'cut' has a certain n~earrirtg in the language. It also takes or1 a ci-rtaiil otcasiorr inearling in t11r pfrrasr ' c u t tl~c' grass'. I.et us assume, wit11 Searlc, tli~it tile rltcarlirrg of "cut the gr:ss' i s

sonletlung like M<)W 1-r.iri ~:RASS. Thus the oicasiorr rncaiiiirg of 'clrt' is the sense MOW, and ~t tcrkcs 031 t i l l s oca-;isioti tncaning as ;1 rtsuit ot.'1~cer:~l/rop- down influences. But this is co r~~~~a t ih le witli cor~~positriorrnlity, brc:inise- -

otie niay arg~ze-the occ;rsiorl oicanirlg oi"cur' is ~ ~ o t h i n g l x i t an i~yjc-c.i qfthc

rneurziq of tdzo contplcx verb-pizrnse "cut tlzc Rruss'. Now the ri~eaxui-r;i, of the verb-phrase depends uporr the n~catiing ofits various constituerrts, rnc.l~iiiity the cttrttplcvrrrizt 'thegras'. i lerrcc it is xlcr surprise that the occasiort xncariirrg of ' C L I ~ ' ( ~ U L I aspect of the tnr:~~iing of the verb- phrae) dcper~ds upon ttrv

i~ieanirrg of tlie deterilliner p11r;rsc that conlpleres the vf:rl?, ,just as i t ikpends upon the (staiillingj t~rcanirxg of tllc itcrh i~seli; since Itotl~ the verb : i r d its conrplernent are ~:unstiruerrrs of ~l re wrh-plirase.

Following a suggestion \vliiclr Scarle traces t o Etf Keerian,' 1t.t us ascit~~le that the s&nding rrleariing of "cut' is a hrrceiori from oljcct-s oi'cutdrjg (the soas of things co11e certs: cakes, 9-ass, etc.) LO specific eut t i~qj operatioi~s

relevant to those ot3ject-s: nlowirrg, slicing, etc. Let trs assume, tilrchc~r. tl>:rt

the argument ~7f the fuirction is deternli~led by the granl~~~atic;ri object of 'cut', and tlrat the value of' rlie i~~~ictiorz (thc specific ctitting ol3cratiuri at

stake) is the occasion menrlinl; of' the verb in thc verk--phrase. Si1ri.c the value of the furlctiorr depends both upon the funztiorl and its argclrnent, i t is

no surprise that the occasioxi nieartirii: OF 'cut' depends. m pan. upon the object that is s i d to be cut. On this analysis tlre illearrir~g ofthe ccjmplex "cut

According to Keenan's srr~csu~rii . as riatrd Ivy Seirrlc, :just : I S . . . sonip nwrhenranc.ai h,nnii>rrr take driil.rent inteqxetahons depending on wlrerher they t i k c acr ever1 or an odd ni~xnher as argurircnr. so the word "cut" has Merent mterprcratlans.. hut ihrsc difiprent interyrewaom art* <ietennoleii hy tlrr hfferent argume~~b---grass, hair, take, skin and ciorh.. . . O n ttrls accoirrlt r t IS rhe wnrd "coi", together with clie literal ~iieaning of "pass", that drtormineb that it11 '"cut the grass" "cut" h a a difierent mtrrprewtiot~ fiom die literal nieaiung of '''cut''"'cut" rn “nit thc cake"' (Searle rp8o: 2,z4)

Page 41: Recanati, François. Truth-Conditional Pragmatics

i 4

tlre i? ; r~ \ \ ' iJepc.nld tq)ox"he (~t.milt11g) II IC 'L~II I~S ot I & part? u1 3 stnctly

bottaan, -tip nILtnner 7 he phr'w kcnt dllr gra.rc' rcprewxrts a certain process (mowuig) opczrrmng on 3 ~ertarlr object (the gr<a\) The \ern'lnh~ contr~bu-

tlon of ttrc L erh 2 (lit' n l iot directly the proc tcs of rr-rowmg but iornetlnng more ,&,tl,tcr, nnrnclv a ~ ~ X I I C ~ I O I I wh lc l~ tdkcr that p1oces.r as a vdue for a

gvcrx .rr gunlent (tErc WLSS) The cc3rlrplenrent 'the grass' contnt>utes both the .ir gunlcrlt of tlte funcnon .u~d the object the rnowxxig process operates ox1 Clo wll,rr I c ailed rllc occLt\~on rnednlng c l f ' c u t q r \ not ntliEy tile rnearllng of rl-rc word 'cut', or1 anrlly\a. rt 15 ' ~n aspect of the rneanmg of the c orrrplrx plr r.rsc, contr~birted jointly by the verb "cut' ,u?d ~ t s cornplernent. 7 hr later,ii/top dttwrr dependence of the occzr\lox~ nrean~ng of 'cut' 0 x 1 the rneLirrir)g of 'ihr" g~'i\'r' ii nothlng brrr a ude effect: of tllc con~poslt~orld, bortom ""1; dependiericc oftlrc Ixle,mrrlg oftlic cc>rriylcx 'cut the gr'iss' upon tine (\t,srrc?irrlg) xricarrrrtg:, of its p.irt\

0 1 7 ~ w,ii of I - X C ~ I I I I I ~ out the itiggt"\tt11 ~11,11y\2c xvc>~~ld L)c to .isslgn the 1bIIowillg \i,i~iil;lr~p, nwdning to '~ci i t ' -

The occ;asiorl rrre:iiling is what wc. get \vl~ei* the garnrnaticd object of 'cut" p~wV"d11:s ;1 V:LJUC fiir the 11igfier ortier varix1)ie "S'. If the object of cuttir~g is $:lid to be grass, we gct:

PIIIIS ' c ~ i t ' me;lals c:ira. I N ,rrrh i\irnNNrrn or: GRASS wl-tert its object is a y such r l u t l;utnsii (y), ant i it x t i ~ ~ n ~ i s (:~J.I. I N . I . E ~ E M A N N ~ . R OF (:AKES L V ~ C I I its object is 3

y sitcir tla;ri < : ~ i u i ( y ) . If1 v ~ i i r r soincones t i ) cut rhe gms, I order 11irrr to cut tile gr,iss in a specihi. rrl;rnricr, rrarrlely, in tire inatrner orlc cuts @ass (by rl~ocvirlg i t ) . 'TIIF order will 11ot be satisfit-iect rf, as Searle imagines, my ai"itlrt.ssett rushes out . i r ~ c i st;lt)s the lawn wit11 a h i ! ; .

? . b lit- slurit: sort of dn:dysis applies to tlae other rsanlples. Consider 'big

I I I O ~ I S L ~ ' : ;I big IIIO~ISC is not big irt the same serrse io wbic11 a big elepba~tt i s ; Grr arr cle.ph;rrrrj~~sr ;L\ big as a hig o1c)use wcrrricl not: count as a big elephant. Tbc occasicttl rrrraning of 'big' clearly depetrcis tipon the noun it modifies. But this car] ki th ac.cotrrrced tor by asslxrriing t11at tllc cctnskxr;lnt mearur~g of 'big' is a firrictio*~ ~ l ~ c v:rlt~r of' which is the c>cc:asicrri xlrealling of 'big'. .That i:onstar*? xlrc:rning cnrj be represet~ted as ar) oiler1 predicate, BIG F ~ J K L ~ N X, \*.liere tile f izc Eligi1c.r onicr \;:triable statlds I<,r the ;irgur.r~cnrt ~ E t h c fixnition;

Page 42: Recanati, François. Truth-Conditional Pragmatics

the occaslort meamng will be the pr ecilcate we get whcrr the frce v,ln~hle 1s ~s s~gned a p~mcular value, wllich value wdl be ( i e t emr~ed by tlre rloirn wblcb the adjective modifies.

Standing vneunitig of "bg' hx hx [X(X) v) (BIG FOR AN S) (x)]

O n dxs sndysn the occas~ori nieanmg of ‘big' (e.g. B I ~ , IOR A Morrir: ~ r t %btg morise'. oor 1rIG FOR m FIEIIWNI 111 'big eleph,ult') 1s nothrng brrt .trJ aspect of the (stardmg) nmeaung of the comnplcx n a ~ u ~ - p l j r a ~ 'big ilio~tsc' or 'big rlepli~t~~t':

That exphins why, llhe the stnrlcl~ng nieanmg ofthe conrplck r~oun-phrci\c, tlie c>ccasronal meanlng ot the adjtctlve tlzper~dc, III p ~ r t , upon rhe ~ilein~rrg of tlle rlorlil it modifies.

As usuAtl, time devxl a ~n the de~rlc, '~rld a lot of detark wotlld 1 i . t ~ ~ to be provlded to rrlake t l ~ e suggested ~nalvsn worthv 0-1' $enoils c orl\~der~~tlon (especially when ~t come\ to the verb-olgcct construchon) Mli t the rffoi-t can be spated beeawe, evidently, the aridysli, does ilot work Fverl IC we f1.i the lmguntlc envuourrt~ent (I e. the modified nottrr, or the object of the verb), the occasiori meanmg of 'brg' or 'cut' rndv shll mry: thl\ silo\;\/\ that the v,m~tron 1s not due merely to the hrig~~istlc envuonnient, contm-y to what the a r ~ ~ l y i ~ \ elalms Ifwe rn~nrtam 2 f u ~ ~ ~ t i o t l d an'dyuc ofthe st~nding lrre,mllig of 't~lg' anti '( nt', drld represent tfm,~t niemulg by rrle,mh of an operqrrcfic,rtc &rc: FOR AN X or 111 cur 1N 1HL: WNFH OF a\ \ugge\ted, we mufr a~kr~owledgr: that the value of the Gee variable is not fully dcternllned hy d ~ e hngma~t contexr, but may be overndexi by exhrallrlgulstlc (or extr~serttennnl) ~rrfonrl~~troti

i?erm and Krat~er glve the foUowing exarnplc to dlow that a sniall elephant 1s not necessarily an e l e ~ l l ~ ~ i i t tirat is srrzall fix an elephdnt

Irrlag~~ie we had iint ~ntroduced a ,ien,mo poprdated w ~ t t l C I ~ I army ot rtkonsten lrjse

K~ng Kong We in~ght then have s,nd sorni~h~t~g ithe 'j~riribo doc-sn't ha-ve ,I t lr'i~ice,

Page 43: Recanati, François. Truth-Conditional Pragmatics

i. be's only a srn& elephant', and ths could have been tnie even lfjunibo were as large as

or even larger tlun most other eleph,mt\. (Helm and Klratzer 1998: 71)

111 tflis context the ~mphclt co~npanson class for ' smd' a the class of

xlionstcrc of the relevatit sort, not the class of elephan~5. even lf the noun

whuh ksxnall' modifies is 'elephant'. Tlia shows that the doman of the

f~~nct lon rs not the clms of objects denoted by the niodlfied noun, but a

coiiipariwn class cleternuned by the context In central cases that corrlpan-

son cl~sc WIU bc the class of objects denoted by the xnodified noun, but there

are other posslble cases in d i ~ h it is not

Searle also provides couilterexarnples to the analvs~s

I t 1% easv ro uriagine cv~trxnrtances 111 wl~tch 'cut' m ' ~ u t the grass' \voilld have the sarne ~riteq>tc.ution rt hns m 'cut the cake', evtm though none of the sernarttlc contents of the word has clt~ugeci Suppose vou and I run a bod tmn where we sell smps of p u turf to lreuple wlio want a Lzwn in a hurry Suppose I sav to you, 'Cut halt'an Acre of gr,r5s tctr du$ cttst'oll~er', I nught meal rlot that you bhould mmu~t, but that vou shodd $kt e It ~ n t o \ta~p:, as vou ~ould cut a cake or a loaf ofbread (Searle 1980 224-5)

Y('lrt tllc g~-AS)' here doe\ not niem cu r l I r E GILA~S IN IISE MANNER OF GMSS,

A C ~t would ii~lder the suggested analysls l h i s shows that the value of the

frtc vlrnable Irr tile open pred~cate czur rcrt: GRASS IN TZIE WNER OF S (or.

more str~~~l~tfonniarcily perhaps, cur THE CJXASS IN MANNER wi) need not he

detcrnlineci by tlie hnguistic object of the verb. ~t is detenmned p rapa t l -

cally ,ind may but need not correspoiid to tlie hnguistlc object of the verb.

Searle gves a parallel example 1n whlch 'cut the cake' would mean some-

thlrrg I~ki. M t ~ w 7 I ~ F CAKE

Slippose we run a bakery where due to our super yeast i;tran our cakes grow upw,rrds url~ontrollably 'Keep cuttlrlg those cakes!', I shout to the foreman, meaning rlot that he should sl~ce them up but that he should keep tnnmng the tops okf (Sedrle 1980: 224-5)

Similar counterexaniples can be constructed for all the cases of semantic

flexibility I have mentioned. 'l'hus, in an appropriate context, 'He likes niy

sister' alight be interpreted in the culinary sense of 'like' (if the referent of

'he' is a cansiibal); or 'good car' in the sense of a car that is a good place to

inhabit (if the speaker is a lionieless person).

X coticlude that semantic flexibihty is ultimately a matter of context- depmdence. The tinpistic context plays the role it does because it is a

Page 44: Recanati, François. Truth-Conditional Pragmatics

prominent aspect of the context 111 wluch tlie expressron OGCLI~S, but nonhngulstsc aspects of the context are also relevant, a, the exa~n~>lrs show

7'0 deal w ~ t h mciex~cals hnd other co~ltext-cen\lhve zxprciractr?\, we need to

revlse, or rather enricll, the &arne*vork set up at the hcgnnrlig o f tl~rs chapu r In son~e cases rile cor3tetrt of an expression t annot be assrgned drrectly by ineans of a lcxrcd rule such as ( I ) , repeated helow-

The content of a context rensltlve exprecsion depends o n the context o t

utterance, SO wlrrat we need n a co~stcxt-.;er~sslt~ve lexlcd rule such a$ (I*)

The constant nietuimg, or 'iil-tracter', of the espresnon cx. detcrrll~nec a functionjwh~ch, giver1 A iontext L, returns a certain conterrtAi-;i a, senzar~tl~ value for the eupresslon Tor exarrxple, the. character of 'I' xlrapi, the ccmtnht

of utterance to what Kaplan cal1s the Agent of tlie c-olltcut, n,~nlely the speaker. Sim~larly, the character of a deiisonirtrative d detemz~nc\ A i~rnctron

which, gven a (nondefectrve) contest c, returns the object demonrtmtcdi Intended by the user of tf in L a \ seniantlc vrllue In all such C,AC\, w e need (something hke) the l < a p l a ~ ~ ~ t ~ Cfl~bn~tlon between 't b d ~ ~ c t c r * CIIICP LO^ tmt'. The character i r the constaint rnearrillg of the expr-rcslon, rcpre\entrd 3s a function tronl Lonteuts to cctiitentf, ,lnd the content, rcprc\errted irm (I*), 1s i!se value wlsrch the < haracter itcternrlnec giver1 ii partrt ular i orite:t.rt "

in acldiaon to ledc-d context-serainvity, we IIUV also rreed to imkz ruum for rnrsinartu,tiiil context- sensirivitv, 1.e. for cxes iri w h ~ c l ~ the nrnclt\ oi con~binati~?ri nlaps rile cunteru.5 oft-he parts to rlrc ct,nrcni ofthe whole only with respect to d ~ e Lontexr oi'ntrerance. Noun -noun conipotrind, in FEligiish provide a pnma Iicie case of that son. Evcrr if we know what a burglar is aird what a rnyhrrnarr is, we don't yet know, out of the blue, wh,~t a husdar rrightmarc is, ibr die value of the co~ripicx phrase %trrglar nightmare' depends upon the conrest r n additLon to the concerrts of-its para. A burglar tiiglitrnare is a ~ghtnmre that bears a cercairl rclat~on K to burglar;. Willat the selevmt relauor~ R IS depends upon the context and dle speaker's intentions. '1'0 keep tlurrgs sornple, I will ignore constri~cnond context- sensitivity in what follows. See Weiskopf (2007) ii,r a treat'nlefir ofcompoiuid nomirials ism tire spirit oC the present chapter.

Page 45: Recanati, François. Truth-Conditional Pragmatics

(I*), covering the cases which are

LilPCI tllcasc whiclr xc riot context-se~~sitive. we can use a sirrgle type of rule, liarrlrly ( r *), by co~~siciering .ihc -furlctiori,f'as constant w ~ ~ e r ~ e v e r tile ex-

prcs~ic.,rr n at isslic i s riot cor~text-sencitive. 'Thus if 1no;xlls p f j in ;t context- itrdepcridcxrt rnaa~ncr, I(u}, -J(c) := rrl, fC>r corl tex~ c.

If, A\ i \~rggi=c\rcd JL tflt' cr~d of Srction 4, srm'tntlc Nex1b1ht~7 rc x rnatter of coritr*xt-tl~~>~r~d~r~c c, the dat~nt tlun brtwcen iundtng lrlearrlrig ,md occa- 5ioj1 X I I ~ J I ~ I ) ~ tu*lt~ out to be ,i pdrt~cuidl c '15e of the I\=.~planla~~ dlst~nctton i>c*twcx%ii cl r ,~r~ctur antJ tzti~tcnt. C3rr tlus ~pprcrarh. we can still treat the

s~ul t l iu t ; i~xcnrlirrg of an expressiori such as 1big"or 'cut' as fuxlctiorlal, as it

w:rs iri the prcvioris ,licvlrnt, Imt the argiilkrexlt oi'the htnction no loriger c"orresporids ttr thr linguistic expression with which the expression at iss~le

cumk~ixics; rritl~cr, elze li:ncriot~ takes thc conrext (or some spect of the

c.cj~rtcxt) as :~ rg i i t l i r~~ t . 111 tho case of'ssrnall' or 'big', the argament to the I-iin~.tiii.n) is a c:c?niparisor~ class prtrvltleci: by the corttext. The standing

111e:iniarg of "h is ' c;krr still bc repmsenteci as BIG FOR AN X, but now 'X' wiU bc ;issigrieii a val~rc. in c-ontest nluch as a clcrnonstrative or LL kee pronoun is

:\ssigilzd :I V;IILIC i r i coiltext. 1 cc i t \ c,JX cElis iicrib ~ppxo.it h tl~e L ontcxtli-d theorv. 1 low ddterent a it from

* I X ~ c a irr\c rhi.or) sws tlut the occ ASIOII xne,irimg 1s not (really) the

~ I I Y , ~ X I ~ ; ~ I ~ of I ~ C ' t x ~ p r e ~ ~ ~ ~ ~ ~ ~t 15we, but ir~drer '~11 d F p C C t of the meanlng id tllr inrxnpit*u phmw in wlrlc-h d la t expresston occurs. Thus the prec%sc ,rtc HI<. E OR A MOIJsL rs nut corrrtrrbuted by tile word 'big' in 7)rg rarlorrsc' but by the ctltnpl~~x phraw l?ig ntoux' ~tself. In ' b ~ g I ~ O I I W C 'hg' i olltnb~~tes I Z I ~ ~ ~t )K AN rY and 'rnouse' ~ o t l t ~ ~ b u t e s both

rlic v iliac ' X' . I P I ~ the prccil~~te MOTI\L, I X L S I I C ~ I a way that the 6 oaiipit'~ pllt i\t' ( ~ l l t n b ~ t f i ~ I C ~ O n j ~ ~ r l ~ t l v e pred~cate MOU\E & BIG

L i (1 .I\SUIZIC tli,~t . I C ~ ~ L L I V ~ rliodlfi~.~t~ori I C ~rtteqrcted by illr,lni id prrdlc Jte colyitnc tlon-we CIhapter 2 Cbr a ctefen~e of thc il ' irti l rhae prci31c,ltr\ic ac$ectlve\ are Intersec tn-e ) 'nlc predicate src, koxi. A MOrJs t hcrc is m aspect 01 j7311t of the rncanlrtg of the complex ptir ise %rg I I ~ O I ~ \ C ' , dctenr~rrled by the aleanllzgs of its vanous con-

~.tltucncs, tnclr~tlmg thc e,vpresslon with W~IJCII the achecave 'big' conrbiarc:, in ttlc phrx~e

Page 46: Recanati, François. Truth-Conditional Pragmatics

e 7 he contextu,d theory sces the occ asrori nxemu~g as tile context ctepen- dent content of the expres\ion, dctennvled by (I) tlie stdnmxig memrlg (character) of the cxpre?\ion ,md (u) the context of utterar~tc liere the prec3ic'lte BIG FOR A MOLJSL is truly c ontnbuted by tlrc ac$ec nve 'lj~g' in

' b~g mouse', but ~t 1s contributed in ii context dependent Iixxnricr The s ~ n d ~ n g niemmg or cli~tacter ot 'hg' n J filnct~oi, nmppmg a contextu-

ally provded conipanson class to the property ofbelrlg b ~ g b r that class, t h t is bigget than most member;; oftllc clds I hat property 1s the ( ontent wllrch the a+xtlve cames 111 context. Now ttie relevait c on~p~ulroxl clas

nay be contevtually provldecl by Irt;quzstlr medns. The noun "rno~xse' denotes the clms of nuce and nxaker that c h s hlgl~ly sahent Unless an

altemanve Lorripanson class 1s made rriore sallerzt hy extr~hrrl;~ux\~c or ex trasententrd means (as rn the 1 31~nx-Kr~~tter \cerl,~no), %la' m %brg ~ncxse' will be contcxtudly ~~~~~~~eted as contnbutlrrg the precficate BIG

I-OR A MCIIJSZ~ Since 'xliouse' cclntnbutes M o r s r t and adjectlvd rnc~ci~fica-

tlon 1s ~xiterpreted by rneam of predicate t otiluiictlon, the coa~plcx phrase 'big rnoure' contrtbutes the t orgunc hve pred~c~ate ~ o u ' i ~ e RIG

FOR A MOrJSk

Ln this theom the content of the cort~plex phr~se 15 a fi~nc tlort of the contents of it\ p,wts, ~ r i a stnctly b~fttoni-up nlanriel, but thc conter~i oT tlir p,irt\ 15, or

may be, context-dependent, xnd the I~xi'$u~st~c context rn whrch 'tn cxprec- 5ron occur? 1s an aqect o f the context whic11 m;ly ictfiuetlce its content I ateral ~ n d top-down ~rlAuerices .ire therefore posq~ble--the coiltent camed by a partlc~~lar eupresaon riuy depel~d upon the otlxer expcprersion., wltl~

which it co~t~bine\-but this 1s compatible wrtb the f ~ c t that the content of the whole depends upon the contenu of ~ t s parts In J. stnctly bottom-up rnanner: mdeid. on the plctllre I have sketc llecl, the content of the wtlole depends upon ttie ctxitents of its parts (and thcir mode o f cotnhmat~on) a t ~ d

rtoll.lzn<q else

6. Saturation and modulation

Although ~t u on the right track, the contextual theory a rtdtecf almve rufl'en horn a senow hnutatlon. It unduly rcstncts tlie ptienonxerion of

\ernantic Aes~bllrty to a smdl range of eqre?r~ons tlidt are ~nciexjcal-lrke In

Page 47: Recanati, François. Truth-Conditional Pragmatics

~rlg is 'gappy' and stands in need of contextud cornipletlon. Indexlcals need to be contextually assigned a value, arid s o do under-specified expressions such as 'burglar rughtrnare'. the intended relatton R needs to be contextually specrfied In all such cases the standing nieaning of the express~on niay be represented as ~nvolving a kee v,tr~able to tvluch a value must be contextually assi_gned, and the express~on carrlcs a definlte content only with respect to such a contextual asrlgnment It is pla~xsible t h ~ t ad~ectives hlie 'srnall' fall into that category and rxtvolve covert refkrerrce to a cornpanson class or standard, but what about 'cut7-) L>o we really want to say that the meaning of 'cut' ~n Enghsh is gappy ,ind ~r~volvec an irnphcit reference to a contextually gven manner of cuttxilgi 1s the word 'cut' 111 1:nglish covealy mdexcal?

I do not thnk ~t is I assume that the standma meanlng of 'cut' 1s rornethlrlg like EFFECT A IINEAR SEPARAlION AFFFCIWG THC INTEGRITY OF

(SOME onj~cr) BY MEANS OF AN EI3GED INSTRIJMENT. There 1s no hee vanable here i o be sure, &lie context may spec~ljr a11 Torts of aspects of the cuthng operatton and flesh it out iii vanous ways (a in the Searle examples), yet I doubt t11at the les l r~l nmeaiung of the exprecsior~ converitionally singles out a p,~m~ular ii~rnension (a "manner of cuthng') such that the context mu5'

provlde a defimte v~lue on that dimenuon (3onslde1, a\ an analogy, the Rurnelhart example 1 discuss In L~terul

Vlttanzvg (1Xecat1,rtl mo4 73, 105-6)

19) I he poltcenlan btoppcct the car

We natur~llv ~nterpret this as rneanlng that the pohcenian stopped the car bv dddreumg an appropriate \rgnal to the dnver, just AS tve ndturdy Interpret 'Jollli cut the c,~ke' xi nieamng that John sllceci ~ t . As Rumelhart (1979 78)

pointr; out, however, a cl-rfferertt mterpretahon emerges if we lrnagme a

context irr which tlte polit enian i s the dnver of the car slxch a context provldes for a totally different 'manner of stopping the car' on the police- ~nan's part Do we want to say that the transittve verb 'stop' in Enghsh tovertlv refers to a rrlanner of stopping which the context is to specify? Of cotir\e not Trala~tive '$top7 means CAUSE TO STOP, and this can be fleqhed out in aB 1~ort\ o f ways, yet the fleshing-out procesa 1s drfferent &on1 the \aturahtrn process mandated by indexlcals and other expressioris whose starld~ng meanlng 1s gappy and requlres contextual conlpletlon Indeed we can construct a context in which (9) would mean that the

Page 48: Recanati, François. Truth-Conditional Pragmatics

polsceman stopped the car trz some tvdry or othet, mdiEerendy No srrch opnor~ exlsts for mdexlcals or under-speclt~ed expressionc, which do not carry ;i definlte content unless the free variable is ass~gned a defiir~te v~ lue m context (see Chapter 3, Sect~on I )

In the Rumelhart exaiqle the cantext suggests a parh<ular niariner crf stopping on the agent's p a If the contextual suggestion is corzveyed by linguishc means-& ~t is 1x1 (g), \vhere the phrase "the pohcenian' Ir what evokes the trafic-regllatlon frame and thereby nidkes the relev.lnt inanner of stopping cars dent-we luve a case of iern,inhc iles~bility the interpre- tanon of 'stop' a agected by the sil&]ect of tire verb That 11 is is ecrhhshtci by the contrast between (9) and ( r o ) tvllen both are taken out of context

(lo) The driver stopped the car

So there is semanhc flexlblhty In there exa nples; yet I do not want to Lredt transitsve -stop7 as ~ n d e x i ~ d or serlla~ltically under-spec~fied And the same thng holds for 'cut' Abctract tlst3ugh rt rc, the lingusshc rrxemlng of these verbs is not gappy in the way in which the meallng of an ulc i ie~ icd or

under-specified expression 1s

Of course, I may be wrong about "top' oi "cilt'E31tnt 111y p a n t ii 1~10re generd. I t l n k there may be scrndtlhc f~enbility even 1f the cxpreislon whose occasion rrleanlng 1s al5ected b\ the iieighbour-~ng tvortls n not context-serislhve in the way iym willc11 ~nr%t'xicdl~ dlld s e m d r ~ h ~ , ~ ~ ~ ullder- specified expressionc are Conc~der another ex.rrnple I c4iscuss 111 l i t e r d l

-lileunzng (Recanah zoo4 34 6)

( I I) The c i t y IS asleep

Because of the apparent c'itegory vrolation (a c ~ t y is not the cort of thrng t l ~ t

sleeps) either 'asleep' mnust be ~rrteqreted 1x1 a rnetaphoricd or extended sense as meaning QUIEI'ANU SII(fCYr1NG LI1 rLL A< IlVIl'Y, or 'the cltyy~las to be Interpreted metonynucdly ac refemrrg to the rnh~brt~nts ofthe clty Eltfier way, how we interpret one exprrcsiorl depencls upon how we interpret the other. Tfus a selnantlc flexibility onc c J~AIII, hilt of course we do not want

to account for that type of example m tenns of context-seasitlvi~ md the character/content distmcaorr. Rather, we take tfirs case to znvolve A dcpx- ture from hteral meanmg, result~ilrg from some form of coercion Let us

assume that (11) is Interpreted by gvmg to 'asleep' the extended serrse c)trrFr AND SHOWING LITTLE ACTIVITY. That IS not the literd sense of %sleepy The

Page 49: Recanati, François. Truth-Conditional Pragmatics

ilreral scrr\ta of *;riirey7 is A S I ~ L Y , and drcre 1s; nothmg £mcy bout ~t (no

Ii~dJcm~ uri lcx~c~d~ no free vx,trrable, etc.). 1x1 tlrrs partlc~llar case, the proper w,tu of ~ d i l i ~ i ; ~ ~ out the d~\t~nctioxx betweerr standmg xneaung and occaslon rneankrrg n iwr by I ~ ~ , U I S of the d~s~rrc tnoi~ between the expression's

~Eldrnctes and m context-cieyuxsdent on tent, hut, rather, by Inearis of the

Jlstmc taoui bc.twc.t.xr the expression's niea~ix~rg in tire language and the non- I~tihraE. 5C117e rt takes o n ttrrough coerclorr m tfsr corktext at hand. In I,zttml l/fi*ian~fiz~q a~ci elsewirerc 1 drew a systematic ci~stinctlori between

~ W O type3 of C O I I ~ C X ~ ~ I ~ proc es\ posihly &octrng trutts-corrcl~tions: the (rxraltilatoxy} prot r\\ of '\atur~tion' througfr wl.trc1.1 mcle;ucds and hee van-

&les rxr loped Ccrnlt .ire ass1g11c.d a corrtextllal value, and the (optional) prow\\ c>/ ' I I I O C ~ ~ I ~ ~ ~ I O ~ ' thr0~1g11 W ~ ~ I C h illc nteallll~s rYn of an expresson 1s

anaytped to ;t clr\txnct rneanlxlg ~{EI) , whcrc :g'is a pragmatic hnctlon.

Met<iplrronc .rnd metonyrn~c'd ~nt-eq>retatroxls re\!& fi-orn the operahon

of scrc h prag1:rrr;ieic- fuiichons, anti tllc drgurr-ient to the tu~rctlon nray be the

rne,rntlag of rrny r*pressiori, wllcther 01 rrcrt ~t rs 'contest-semihve' In the \t,irltl,ir tl seriic- rn wlllt 11 ~ncicxrcals and scli~~~rit~c ,diy under-\pet rf~ed expres- Sroxxs ,ins Ar~othcr type of pragmatic lurictro~l, u~volved In so-cnUeci 'Gee erirrclrrxient', ruqs the r-rtcanmg o f 311 ~ X ~ X ~ S S I O I X to '1 nioxe ?pec~fic mean- arlg One w'iv o f ,at countlrlg for the *\top' 'cllt' ca\es wtruld be to argue that the \tarrdxng rne.innsg 01 the verb (c"~\rist xo s lop, EISFFCT A LINLAR

4EPARRI I( )N, id I r 1 li, ~irltdrrsto~fd 111 ~ o r ~ t e x t m a nxore spccrfic sense, through tikc proviuon of& prt~cular m'llmer ot \t<3pp111g or of ctlttmg On &IS mew the context rs rrideed what make\ the relevant xn~mner of stopp~rlg or cuttsng \,~bc.sxr arrd force\ Ir rrlto the mterpretatrcrrr, but the contextu~l process at r%ut ui rlnr geikrrdtion of t h ~ t occA5ion rlledrllxsg is rlot saturatlon, but modnl,rtlcrrr Agakr*, 1 lruy be wroi~g about "cut' and 'stop', but my point rs rrlore ger-icr=d ard ran he put ;u tallows- we m ~ y get the sort of c~nte~xtual rr~fluenr e trrr the irlterI3retdtlon of a I C ? U C ~ Itern \.vtz~~h gves r-rse to the phex-~onlenon of selxlantlc fiexll~~lxty cvext rf the exprewvn who\e ixiterpre- tatloll COIIC~Y~UAUJ~ v'lrles 1x1 tlm way 15 rrot rrldexrclrl or context-sensitive m thC iitarxclarcl aerise Ifsernant~c tlcx~bli~t-y 1s to be accoullted for by appedlng

to contcxmd proicsse..,, 'I\ 1 have suggested (Sect~on s), therc. is no rearon to rcsmci tlrc type lot ctw~textu,rl proce\s at iss1.1~" to saturatlon. Modulation play cxai tlv tlic 5,uxle ole j u t AS the vnluc t or~textually assigned to an rndex~cal or free v tn'lble rxlay be xnf-luenc ctl by the Lmpntlc context, the

modulated v'ilue uhrcll ,i gvcrl expcessrun take\ ln contest map also be

Page 50: Recanati, François. Truth-Conditional Pragmatics

irifiuenced by the linguistic et~vironment, that is, by the other words with

which. die expression combines, as in (9).

7. Cor~lyositio~~ality and ~noclufation

Where doe, thrs leave Lrr with respect to the coxllposltsonalrtv ~s\rxe? if the folegomg is conec t, we taunot nrairltalxt t h ~ t the liledtling of .I coraplex phrase a (wholly) determined by the mearimg\ ot I& pelr.rsts dncl their lnode of combmahon. As Searle polritt out, the ?atls6~ct1or1 c U I I ~ I I I O ~ S of the rrtlpcr-

ative Tenteni e 'Cur the gmssl' rrlay v,aay evcrj tl~('i~tgi:ll the \t.xnctmg nle'mlrrp and Kaplan1a1-1 contents of ,111 the word\ m t h ~ t cextterlre .ire fiacd, A well a\ their mode of conlblrldt~on So the lritelpret,~tlon ot the cotnyleu- -1moL2r as it detennme< cat15factlon conclrtions--15 r~ot A f i i t l ( tlot~ of the 'rileal~~nji"i'

(in one of the jtandxd rervies. tbai.~ct-er or content) o t rts p~rts and their mock of ~ o l i ~ b i n a t ~ o r ~

At thn po~nt there are twTo m~1n opntms, conesp,l>~)ntI111g to TWO posit~~tfi 1n the philosophy of language. Mmmthsx11 azd t,onteut~i,dvn for the 11irn1- rnafist (in the sei~se of C:appelexi and I epore zoa>a), there n &r sliarp cil\tmc tron

between \enr&rrhc me.ming, incltui~ng Kaplarlran \o~ite~rt', and speaker':, n~e'rrung. Xnsohr AS rnodul'lt~on-hence \peakcr'\ nreax11ng -rrxter\ lxlto tllc deterrninatlor~ of sansfactlon ton&t~om, it e not i n ~ u ~ i ~ b e ~ i t up011 selx*anhc\ to account for ratisfactwn cortd~tinn~ and dle coxltent of ~peech aLts ]nore

generally the c o n ~ p o i ~ t i o t ~ ~ l nmcllmery IS or11y supposed to (led1 with sel~~intlc rriearung. For the contextualrst, 0x1 the conmy, we should do oirr bect to account for the illtulhve truth- and 5dtahctlon-concfitloro of utterance$, arld to that eirect we rrlay have to hber,~lize the notlor1 of rric,m~ng/cox~te~~t to the pomt o f b l u m g the scrn~rrrtlcs/pragrr~tlcs drstlnct~ort

The c o x n p o ~ t ~ o i ~ ~ l Gamework a\ pre\rltted \o fir ordy make\ room for those forms of relncmnc. flex~bil~ty which m\e fsorli (leu\ ,rl or Lollstrut- tiond) context-seni~t~v~ty Since tlitqe are trot the only fctr~r~s: of ~exnantlc flex~bll~ty we have to dccourtt h r , the nght thing to do, tt seems to me, 15 to revise/ennch the frarneworli oxit e a p ~ r i , in tile spir~t of the conteutudist posstlon Examples like Searle7\ dekat compo,laon~hty (I.)? sllo\?nrtg that the interpretstion of the colnplex ts not ,I f i m c tlori of the xnrariing, of I&

parts and then mode ofcor~~blxl,t.tlon) provded we tahr "tile meanlxig? ofthe

Page 51: Recanati, François. Truth-Conditional Pragmatics

-- 44 T m l ~ m ~ m A ~ m ~

pam' In one of the standard sences (character or content). Why not, then, take the mnear~mgs ofthe parts to be the= modulated meamngs, and attempt to prewrve cornpoc~rion&ty m ths manner?

Let u\ define d futi~tion rriod taking as argument an expresslor1 e and the context c 1x1 which ~t occurs: the value of mod is the particular rnod~llaaon funchon g that rs contextudly sahent/relevant/appropnate for the Interpre- tahon of that expresston i r i that context. For e d m p l e , m (II), we can easlly irnagne that the context c m wI11ih the expresslon 'the city' occurs renders a c ertmn metoiiyrmc furlchon gjU trorri clues to their inhabitaxlts d e n t and relc*,uit to the ~literpretatron of that expresslon With respect to such a context we get:

mod ('the c~ty', c) = gjt3 rnod ('the c~ty" c) (I ('the city'),) = g,,, (TEFE CITY) = m~ INHABITANTS OF

TEE CCflV

The suggestzon, then, n that we should take the niodulated ineanmg of an exprewon u 111 ( ontext C, VIZ. ttlod (a, C) (1 (a)J, ;IS the bullding block whlch our cornpoc;itronal rrl,uil~nery recjuires to dehver the correct interpretatlons tor coniplcx expressions Accordmgly. we can keep the type of lexlcd rule we 'tzave worked w t h so far, VIL

:.i *) I(ajL - fbj (where "f 15 the ctlaracter of exprrsslnn u)

but we must change the fonnat of conipo\itroriJI rules so as to inake room lor ~~~oclul~t to i i . Inste~d of

cve slloulci use sonlet11 jng llkr

Here'c" and 'cr" correspond to sub-parts of the context c in whrch the corr.~plex expression tl*p 1s used (I acsume that ifa complex expresslon a*/3

is used in 3 context C, e ~ c l i of its constituents is used m a sub-part ofc , for exanple u 111 c' and /3 in c2) The gs correspond to pragmatic modulation tilnc.tlons which the context rnakes salient. If no modulation is contextually

'This StIggeShOn can be traced back to Sag (1981) (and to Nunberg's ideas, which Sag attempted to formalize).

Page 52: Recanati, François. Truth-Conditional Pragmatics

- - -

C OMPUSIIIONALIIY, ELEXIBIX I 1 Y, & N i l ( O N I E X I U t P F N L j C N C I 45

appropnate and the expression receives ~ t s Iiterd tllteqretatlon, the value of mod d be the tdenhty funct~on hterainess 1s treated as a spec~d ~ca5c. 01

(zero) niodulatlon. Thus understood, the foml~lla says that the trrterpreta- tlon (content) of a colnplex express.mn a*fi is a rirnct~on of the moclulated meanings of ~ t s paas and the way they are put together (and t~othing eke)

For sinzple expresslorls there Ir a clear d~s t~nc t~on between their content I(a), detenxuned by lexlcal rules ruch as (I*) and then mutiulated meilr-r~ng mod(n, c)(l(a),). What about complex exprescions such m n*/33 I>tre\ the distlnctlon between content aid n~udulated inearling apply to the111 as we112

O f course: a complex expres\lon e c,m be a coxlstltuent In ,I more complex expresslon e+, and we need the nloiiuldted I-neanmg of r to rervc a:, a

bmllng block in constructing 3 coritent for e 4 We must therefore geiler- allre the rlotlon of modulated niednltrg to ,dl exprt.ssions 'I o reach the rlgtlt level ofgenerakty, Pagn suggests that w e define a Fuxict~on o i rnodlxlatcd interpretation M recursively Thn c'rn be clone as follows

Since the content I(?), of a conmplex exprtsssrorl (' = *((,, , r,) r \ A hinctrot~

of the modulated mean~ngs of ltac part\, r h l ~ defr i~t~oo oi niodnl~trd rncarr lng e n d s the recurslve c l a ~

As before, * n a syntactic rnode of ~ o m b i ~ ~ ~ t i o : ~ , driil f is tlie corlxposrtrorl funchon of the ordinary sort concsponding to that xnodc of ioxurbrrratron The recurslve clause says that the modulated ~ncanlrig ot-a cornplex cxyrcs- sion results horn applylng a coxitexru&lly appropriate modul~il.tlon h n r tiorr

to the result of cornposing the nlotfulated rneanlng of- ~ t s part\ 50 the

distinctlor1 between content ,md rnothl~ted rncanln1g appher to cotnplex expressions as well as to simple ones the cosite~~t of a ~o~i ip lex expressrout 1s a i;clcct.an ~f the madu!ztei! m e z ~ ~ x i ~ p x t ~ , id the x~i3di11zted ~r,ecln~x?(* t-,

of the expresslon results Groni n~odulatltlg the content tlius dctervrr~rlecl In ths h e w o r k , do we rrafly aclzieve conipoatlon&ty' Nut m d ~ e

strong sense m whicli cr>mpositlonabty n s~~ndardly understood As we have seen, the cotztent of a complex is a fiirlction of the rnodtila'clted rnPaveaTs of ~ t s

See Pagin and Prllener (2007: j(i-30) f i r ari el~boratiun.

Page 53: Recanati, François. Truth-Conditional Pragmatics
Page 54: Recanati, François. Truth-Conditional Pragmatics

C O M P O S I ? I O N A L I v I Y , FI FXIBIL I T Y , AN!) C O N T L X I - 1 ) t - P h N l ) b N C t : 47

'eeyu~vocatlorl that can't be resolved", ,~nd IL 'unctenrzinr[s) tlre conlposl tmn- aliw of knghsh' (Fodor 2003. 99).

lnsohr ac I undererand the asgimrent, it doec not go throng11 Context~1,rl rnoclulatzori provtdes fc)r yotentrally urrenctmg nle,inlng vdnALiorr, but never gves rlse to any n ~ t u ~ l f unending meaning vanatlor1 Me,zinng everltu.~lly stabihzes, niak~ng ~orrlpo\~tionality possll,le, because the (Itngurst~c as well extrahngp~\tlc) context, however blg, 1s 3way\ ftriite

The contextlt~lxit empliasrzec the unc.nding potential fix vJnatlon In order to point out that the (modulated) nie,lnmg of-an expre,norl slw~ys depends upon the context arltll cannot be fixed slnrply by cot~rplexlCytrig the expression and 'nmakmg everytfllng espllclt' Thui, the ~ontextu'~lr\t giver

the followrr~g sort of exaniy3lc m cupport of tlte ~rredrrcrbly corltextr~d ~i~aracter of tlle tnterprerahon process. Ijobn took ont h ~ s key and opened the door' is interpreted in u ~ c h a way that John 1s tl~iderstood to Ii'lve operied the door wtth the key: ttr~s we get through ~nc,dul,ition of 'open the door' whlch 1s understood via the t ontexhid provtslon uf a specif c

manner of oyen~ng Cdn we nrake that expl~cit 111 the sentcrrce, so as to get nd of the context-dependence? Not qulte If we say 'tie opened the door w t h the key' the new rnatenal pves nre to new uriderdetern~~n~clee l~ecause it, too, can hr v,lnourly ritodulatrd. The key may tlave been used as an axe to break the door open ac well a mserted into the keyhole (Sc,rrle 1993: 182) Artd if we make the way of rrrrng the key explicrt, filrtf~er rniletemnac~es wril anse, and dlllerr~lt meanli~g5 \vrU errlerge tl~rough rnodulatlon Iiowever, wlteli language rr actually used .irttX cctnlethlrli; 1s sad, there I< d cfefinitt context botli Intgux\ac and extraln~gt:urc;tit) .tad ~t I\ f i~~r te Iri vlrtue of the context, tile vanow, espresslons used ln st get a definite mewing No umtablhty 1s to be feared dnd, puilir Fodor, Conterrtu- ahsnl is perfectly cornpat~ble wtth the clelndnd~ of'cc>rrt~>o\itr~ndl~ty '

Since 3004 1 have had oppertunlbes to present tire rrrarrnds in thrs t.tlapter at co~~f>rences, workshops, or colloquia in nlarvy places. lncludirig Gal-gnano, !'art. Montreal, Kingston, (Oxford, Geneva, Lisbon, Stanibrd, Cadiz, St Andrews, I.und, .uld Stockholm. I an1 indebted to the organizers c t i tltose events for lnvinng me. to those who atteilded for tflcir c~i tc~t~orn arld nl?lecttons, and to W. tlinren, E. Maclrery, and hit. Wemlng, tlte editors of tile tiurtdbook 4 Ctmqositionaiity (Oxford LJnlvenity Press), where thls chapter is dso due to qlyear. 1 an1 especially gracefit1 to Gennal-o Cliierchia. Peter P a p , Josl~ Arnrstrorlg. Mariln Jonsson, and 1)ag Wzsterstihl fix coitrnlents 2nd ditcuss~ons which insp~reci mi..

Page 55: Recanati, François. Truth-Conditional Pragmatics
Page 56: Recanati, François. Truth-Conditional Pragmatics

Adjectives: A Case Study

Two dognsas

In l~ght of our ciiscuss~ori of.;eiiidr~tlc ilexibrlity, we should he suqxcrous of some of the clairns that are qtal~dardiy n l d e aborxt ac?jc.ct~ves W;'lr,rt kdiows

a a pauage reflect~ng textbook wisdorrr in tlus area

There n a cr~lciai cfiKere1lce brtween "absc>Xutr' adjectives like npii m

( I ) There was a red book on tiir cdl)lr

and 'relnt~ve' dctjectlves l ~ k c smdl 111

( 2 ) There ur,x, a sinall eicpharlt ~n the LOO

Absolute adjecuves drcc~~be propertre5 of ir~ii~vidtuls indc~>cndc~~~dv o t tl-ie Lonrri

bution of tlie h a d noun ILela~vr ~djectwe\ are ilependent on rlle xiorln the\

niocfify 'l'he set of red book e a st3bsc.t of t l~e set ofbooks, whlc1.1 wt: can find hc takmg the intenechoxi of the set ofbook, md the set of red ob~rcts

In the same way, the iet of small elephants is d whet of the \et ok elgh-qnt~ But u&Le ul die case of 'led hoctL', wc tarir~ot constme tlvs \et fronr the intersection of the set ofelepi~atits atid the set of s n d ob~ect5 Aftel all, 2 \rl~,xll elephant IS st13 a rather b ~ g oblect Re1at;tve acl;lec txvcs do not drsc nbe propertics of m&v~duals mdependently of the co~itr~b~ibon ot the head noun (l)e Swart

1998: 158-9)

In tlus passage two cla~ms are made

r . 'SnxaU' is a relative ad~ectrve, ijepencient for its mteq3retation rlporl the noun it modifies, whlle 'red' uri't

2. In contrast to absolute a4echVes, relative adjectives hke ' s nd ' arc not Intersectlve.

Page 57: Recanati, François. Truth-Conditional Pragmatics

Wrdespreatl dxotigll they ,Ire, l7otl-r clalrn\ seeit1 to nre to be -tjlse I st'irt m t h

t ! ~ 1.161rn ,~\x\iii. 11011 i r l t e r \ e i h v ~ ~ , to] we b,lve ,icre,i~ly dedt wit11 that

ni,iretwr r r r pa\slng In our prev1011s d~s ( LISSIOXI ot' the \emantlcs of phr~nes like \1nA1 o"l~p11~~11t' J I I ~ 'Lxg mouse' (C IrL+ptrr 1, \cc tloxls 4->).

,a\l,\olrrtz ~tlject~vcs lihc 'red" are w d LO be rtlterseitIve becduse some-

rblna2 14 (c # ) ,I red b ~ ~ o k af.'md only l i l t rs red ~ n c l ~i 15 a book: the set of red h ~ o k b ~ r \ h e IIIICISC( ti011 ot'the set of'red thrtlgs and the set ofbooks. Let me intr adrrc c \orn~. j i r o r ~ r t a i i d , ~ ) ter~r~lrrology that -cvrll be usefkl irt 'tpproach- 1119 L ~ I I S is\iie It1 '1 ( 0111171ex phr,t\c At3 < c>rlsl%trng; oftwo terrris A and R, I s q ellst there r* s r n l i r c , ~ f i r ~ ~ r y rii'one of-the terrns divriics ti-e estelalon ofthe other terr-rr, tlldt I \ , ifff rh r e.;trrl\ion of the coozpieu plrrac APS is a subset of the

euterisiorn ok'tlrc* other tcrnr. 1 Irere 15 l t~fl-subst~tzi~rry rff the left tenn's (A's) r\tcrr\ic>rr x i ~irvitletl by the otlrz-r temr 13, clrrti n~2~t-rtih.t~~rzozty iti'B's esteri-

<,roxi r t iiiviidcti bv iL Irr the for~rler type clfc i-\c, lion1 x's beuig an AIZ ~t EolTo'iv~ tlr'ci: .I; ri, d r i A, tn thc Lttter type 05 r,tse, Ci-orn a'\ being a11 A13 r I t i 15 CVak ~ntcrscifzz*~ry 15 the c,ue lo whlctl there IS both k t t r n t l rrg1~r iulaett~vity, l r r stacll \;v'~v tll,rc \onletflmg's belng an AN ent,111\ botli c h r IL a ,111 A. ;tnd thdt ~t I\ H Irlrer\ecr;rvrty proper or strorzg

inrrr\i.(trzrlty 15 rlic c .i'ic In whlcI1 the ent,irImrnt goez both ways: something 15 in API rflrnd only r,f ~t r \ arl A , ~ n d rr I S 'I 15 7 c r w i n up

le*i'i- srii~scrtivrey: A is the tern1 wl-rosc exterrsion is divided. trrference p'.Itlcnr: A13 ;?. A llightcuhstctivity: I$ i s the t.enn whostb ;.exrcnsitrn i s divided. Inf'erence p:itr;crr~: AX5 -2 t3 We:A iiltcrsec.trvit~r = leit--suhset-t,ivie -1- rigtzt-suh.;ectivity. Inferer~cc p2tt~1.11: Al l :=:. A & 13

@ St.rorrg inressectivity: AB i* A & 13

I t r\ c;i\y to i Iicc h that, rrl sc,rricl<trit achectrve rlorrr* corrlbi~latrons strcli a\ 'red hook' , t l l i ~ c 1s 130th lthfi-subsix tivlty nlld nght subsrct~vrty. 1 here w left-

\ r ib~rct ivr tv ~ C G ~ ~ L ~ S C " ,* red L>oob ts reti (A]% =/ A), .tnd there 1s nglit- iu l rwi t rv r tv b c c ~ u s e J red book IS 2 hcx~k (AN -i R) In verb phraes, on t l ~ c c~thrr h ~ ~ d , tlierr r\ u ~ ~ l y left s l~bscct iv~~), tile exteI1'ilt)n of'cut the grass'

I\ pluper subscat o t r - l ~ c x cxtencrur? of 'cut'

Witlr ~c$cc at e4, nrtrrbcctlvlty n \d~d to Eul. l i t L ~ V O ty77e:, of case. Rel'ihve ,rcijei tatcrc cirsyl~v rr~:t-rt-rtt-sul-trecbvrty (A cmd t*le~~t~,u*t kc ,ul elephmt) wthotrt

left-ztrbsccrivrry (A \ ~ i l d l l C I C X ~ I I ~ U I C t\ r j o t ,I srndll dirarg) 13nv,ltrve .id~ec tlves belong

Page 58: Recanati, François. Truth-Conditional Pragmatics

Al.)Jl.i(.:'f'fVfi.S: A (:ASM S T U D Y jl

to a class of adjC?ctlve\ th2t chphy the oppoutc pattern. dlrre 15 left-sul>\echvic) (a hke dollar ts f&e) vnthout ngllt-subsect~vity (a Ldse dollar 1s not a dollx)

3. Tlle intercectivity of relative acljcctivcs

We havc seen th,rt art adjective hke 'sa~all' 1s context-setlsltrve; it rrreans sornethlng lrbe SMALL FOR AN ,Y, wfrcrc the valrre of T' tleperrcts upo~ i the context. S o the (occasional) rne,1111ng of 's~n;lll' i r liable tct bar): 111 cl~ffererlt cclntelits the ac!jec~ve 'srnall' is understclod dlfirently and contnbtttes cilffererlt predicates smufl,, ~nrull,, snzall,, etc Thr \ \uggc\t\ 31-1 ~l tenlat~ve expldnat~orl ot the ~rrfelentrdl behaviour ot relat~ve aiijectrves It I\ true t h ~ t 'a cnlall elephant i s strll a rather big object' ( I k Cvrrart 1998 159). he11ce no t A

sniall ttl~ng, but that does not necerranly \how that rhc left-~uh\r.rt-rvrty patteln (AB =+ A) n vloiated 'The pattern would be vlolated tf 'cnmll' were unlvocdl and cll~dy!, contr~bured the sane prcdlc Ate A. Rut it n not and the serrtence % ~ n ~ a l l elephant ic not srn,iU' (or 'not ,I ?l?i,rll tllrng') choulci he andyccd as follows:

( 3 ) A s l n d , elephant 1s riot s n i ~ U ?

Ilere the ccn~irit~c value of 'sma1llI' sh-rfts troxrr onc occurrence ctf the djectrve to the next, I-recause two dlst~nct vrilltec arc contextually ;r\slg~:r.trd to the vdnable. 'A sr*ld elephant 1s not snldl' Ineans that an elepir'u~t that n

trtzcillfor urz elqlzant i s not snzall loniptlreri to orditzizry oblcltlc. *T h,tt tlus expl~natron ot the alleged farltire uf left-\t-\ub\cctivity I \ the

correct one can ekc11y he es~blzcheli i fwe Leep the v,rlue of '\niaJl' cctltstant, the left-s~xbsec xv~ty rnferenc-e pattern 1s scrtlsfied. a srrtall, eleyhartr 1s ~ndeerl. small,, tllat is, irr~dl for an elephant So there 1s no firlure oflcfi-sub?ectrv~~ <ifter dl, and no reAon to deny that ~tdject~ves llhc 'rrn~dl' are elrtcnectrve

O f course the left-nrbset ttvitv rnferen~e pattern IS cupe$~telly vrolated, hut such a suyerfic~ai vloldhon does not counc &om ;t logcal poirtt of v ~ e w Qu111e has insisted that

The trustworthiness of logical andysisaalld inference depends on our not gwing one and the same expression ditferent intccpretarions 111 the course of tire reasoning. Viohtion of t h s principle was known traditionally :s tlic GdIacy <>f eyuivocatiort. (Cruine 1962: 42--3)

Page 59: Recanati, François. Truth-Conditional Pragmatics

- - - -33 - - ~ ^ ~ N . I . ~ E R ~ T * C W W ~ ~ ~ Z Y H W ~ B J E i i W 5

In the argunient Goni 'A small elephant 1s not small' there is ecpvocatlon (srilce t11e Intrqrctatlcln of 'small' shrfts), arid ~t a only because of that eyurvocatlorz that the left-suh~ectlvlty mn&rence pattern seems v~olated In the s'trne p,k$sagr frorn Mrthnlts oflogic C>uine gives an analogous example of the fiLl,~cy of equivocation

Tllr cwo conjunction\

(4) t ie went to I'dwcdtuck and 1 went dong ( 5 ) f i e tverlt to \,rugatuck but L did not go along

mav both be tnie, vet IC we tepresent them as of the fbnn 'p8q' and 'r8-q', ac

srerlts superiicrdly to fit the. caw, we come out innth an inconsister~t ~ombinaaon 'p&q&l&-~l ' Actursllv of course the 'I went dong' in (4) must he d~stingu~shed koni the 'I went alorlg' wliose ncgatIor1 apprars In (s), tlie one is 'I went along to I)~wcatuch' al~d the other IS 'I \vent along to Saugaturk' When (4) and ( 5 ) are

t ornpirtect 111 rhis ~ ; ? ~ L J C ) Z I they can no longer be represented as related m the rnatlrter of 'pKq' dnd r&-~q' , but otdy in the manner of 'p&q' and 'r&-s', and the apparent irlcot~s~\rency drsappedn

llle faUacv of equlvocAtlon xnses when the lnterpretatlon of an amb~guotrs expresstorr 1s 1t3lltlcnrc.d 1r1 vm7ing wavr by ~ m ~ c d i a t e contexts, as 111 (15) arid (rti),

so that tiit: vxprec\irrrr t~ndcrgoes ~1.rangrs of meaning w~thin the hrriits of the drgu~"1~1~' l r l S U C ~ lairs rrv havt icl r~pphrcxse before pro~eedlnq (QUI~P 1962 42 3.

rlotdriotl drld cniplrasr~ nurtel

li, according to (&me's rmtnrctlctm, we rephrase the sentence 'A small elepl~ant w not small' before s~ibmttlng ~t to logcal analysis, we get (3), and all appedr.rric cs of A vrc>l,rtrotl of the left-subsecnv~ty mferertxnce patterns varusli.

irotvc'illy, Qurne hiimelf cctmnltted the fallacy 1n lus well-known ds- cus?~vn of the 'cC;lor$;~one' sentence. Fro111

(6) (;lorpone WAS so- c d e d becaure of lus size

Quine says thdt we cailtlot infer

(7) Barbarell] was so-called because of h ~ s size

everr though Glorg~orie = Barbarelk. I agree. But Qume concludes that the pnnc~ple of cubsbtut~v~ty falls 111 such cases. Thls 1s fdlacious If we keep the contrxtud vdue of 'so-called' constant, the Inference holds. The pnnc~ple of subst~tut~vity seems to be violated only because the sense of 'so-called' sliifts Goxn (6) to (71, blocking the tnference, just as the sense of 'I went

Page 60: Recanati, François. Truth-Conditional Pragmatics

along' shfk &om (4) to ( 5 ) If we rephrase before proceed~ng, m accorcbmce m t h Qurne's mstruchons, we find that '61orgorie waq ro-cdled, [; c died Giorg~one] because of h ~ s w e ' tndeed e n ~ a l s 'Harb,*reU~ was so-~~llecl, because of his srze', juct as 'Junzbo 1s a brnall, elephant' itidccd ent;l~is 'Jumbo is .imaU, [=L snidl for all elepllallt]' Wut 'Giorgone tvdr co-called, because of his me ' does not e n t d rs 'Barbarc111 WAS $0-called, [= caileil Burbarellzl because of hls size', hut tberc i\ no reawn why it ~ h o ~ i l d "

In Recanat1 (zoooil, zooob), I drew tile co? icluvctn tliat TC;iorg~~)tre\is purdy referentlal in (ti), appearartceb r~ot \ l~? t tn tx~ brig (5ee Kecarlatl zooo'~, r;ouish

for a defence of thzs clam) Lrlrewne, 1 conclude that relative ,~djectives iclch as 'small' are mtersechve, appe'rrmci-s to the contrary norv~.lth:hstandrng' Below, 1 w~ll argue that the fallaty oi equtvocahori xc alst, ~ v f ~ a t ,iccourln

b r the appeaa1ice.i of non-mterrcitlv~ty In the caqe of pnvAtlve '~ctJectlver; (Sectlon 7). Before turnmg to that r\sue, however, L n~ur t deal with the second

of the two dogmas I mcn~oried the clarnl tliat, urillke ddjectwes such ds

'small', lntersechve ddject~ve~ hkr " r e ~ d " are nctt relahve, that 1s cut h that the~r Interpretation depends upon the noun they rnodllj.

3 . The relativity of ultcncctive ndjectives

There n a sense irk which "led' i? no le\s relat~vc th,iti ?\m,xll' For a "red book' e not red m m y old way: ~t I S red tn rlze spcczfir uiny zrl ~crhirk books (<n opprlscti to Lars or newspapers) are red In tlm re\pvc t;, 'red hoolr' a like 'pmk grq>efmrtq or 'red pen'. A puik gr~pefru~t 1.i ,I grdpehnt that is pt~lh 1?1.5ld~, axrd A red per] r\ n

pen tvho5e ink IS red In those cilies tho predicate does rlot ~ppIy .h?c>c'luteJy but rn

a specific way ,hat depends uporr the cjbje~t to which the ~olotrr is arcrrbrtl Tliese are, adrmttedly, radrer speci,rl calcs, but ever] orclrndry cases s~rclr as "rcc hoop or 'red bur& ittsplav <a srgndlc~nt meAsure of relttt~vity, M ~ ~ I C ~ I has betm emphasized in the contextuaL~st htcraare I'hus Rori Lallav wrrtrs-

Tor 2 bird to be red (m the rror~rral c*rsul, ~t siiotild have most of the cttda~e ot ~ t s

body red, though rlot its beak, leg, eves. and rrt'~ourse ~ t s liltler orgdns Fnrther-

' Peter Pagin thinks I arrr uriijir to Quine, be~ause, he says "we can weU fitmi t l ~ c urzboc'rl one-piacr predicate " . . . was so-called because ofhis ss~e", where [tire] argalnent place 1s not transparent'. Birr X fid to see how silch a prehcate can be both one-place and uriivc>cal.

See Kamp (1975), Heim and Kratzer (rgg8), arid Sz~b6 (2001) for sinular concli~rrons.

Page 61: Recanati, François. Truth-Conditional Pragmatics
Page 62: Recanati, François. Truth-Conditional Pragmatics

Indefinite variety of tlnmgs, atuatrons and cxpencnc es, arid we nl'mage to do that by 'ac%wtmg' the nieanlng ot wnrci~ to the vdnety ot tlur~gs ,znd sltu,ztlons we talk about.

Cruc~d to the secoi~d posiclon is the ~de,i of modulation, for rnod~datlc,n potentially all ,r<lJe~trves, irldeetf dl opext-~las lterrls W i ~ ~ l c mekxlcihty or g'tppynes or senlmhc under-spe~~flcahon 1s a lexcd property of certixtn expressions and nor others, the su\cgtlbrllty to n~odulatiori I \ U I I I V C ~ S ~ ~ ~t 15 '1

general feature of the way we use lmguage So to argue 61tx~1 h e ob~ervat~~r t

~oncemlng 'red' to the conclus~on that all adjechves are relanve 15, I ~ I eVec t, to

treat the relahvlty of 'red' as a ~narufertatlon of~nodul,rtlon. I'he iswe, whether or not d adjectwe~ are relative, 1s thcrcfore Lqely ter~nlnoIogc,d All ,ic)cc-

aves are relative 111 the sense that, through filc)dul,it~on, the meanmg r t t XI

adjective can alwayc bo adjusted to the thmg being talked abotrt, but in more restricted seme, otdy cermn adJechves are relahve 111 the way m wblch, sav,

'small' 1s What datmgmhes the adjective? that are telatxve 131 thC tllznozrr Fence Is that tjzar ltrzgursts~ menntng ltceg indexes thelr Irtterpret'rtloit to the~r corite\t,

through the rneclia~usn~ of inclesrcalrty/s,tm ratrctxt

When ~t corm\ to 'red' and colour atilechve\, the \ub\tanh,il issue 15

therefore the foliowrng a the flexlbllity ot theqe acljecaves a mitter of s,~tur~h- tion or a matter of rnodulattori~ 's/ab<j un;lrnb1gwt>11\1v treats it a rllattcr trt

saturation, wfule Charles Travn-who, for decades. h a been mlng the euml ple of colour predicates UI arguu~g for Contextu&sil~- treat5 ~t a matter of modulaclon For 7 raws what 1s true ofcolour adJcctives 'holds of 'my Engl~sh pred~cate' (Travw 19'34. 172), it is a getled property ofthe ~lsc of language For

S72h6, ~t is a lexlcal property ofthew adjectives wkrch doec not extend to other ad~ectrves and (therefore) does not suppolt coi~tcxtrral~st concluuora

4. Colour adjectives: s;rturatio~i or n~odulation?

A c ~ o r d ~ r ~ g to S~ab6, there are many w~yys of bemg @ern, aid ax1 Ascriptiort of greenneqs tacitly refers to one way or another 'The wdys 111 clue\tion correspond to parts or aspects ot- the object belrig tdked about

There are at least two ways In wh~cf~ a11 dpple (-a111 be grcert. from the c-tuts~cic, or

&OITL the ~ n a d e In the forn~er casc, ~t can be 11pe, In the l~tter it cannot 'l'herr ,ire at

Page 63: Recanati, François. Truth-Conditional Pragmatics

- --56 - r ~ t t c m r m ~ e m m ~ x m ~ m , i t l ~ ~ t x % i i f iok?

least three way5 m whch a book can be greerr due to havmg a green dust jacket, a green cover, or green pages, Axld a corridor can be gee11 tn many way>. bav~ng green WAILS, or green cetlmg, or green carpet, or green doors, etc. jln general] an

object 1% greert lfsorrie corltextudy c;pecfiabIe (and presumably sufficlendy large) part of it n green (SzabG 2001: 137-8)

$zabb goei further and ~Llirns that a colour adjective such as 'geen' involves a covert vm,lble to w h ~ c h a vdue murt be contextually assigned. Just as 'sniall' requlres conipletion by means of a corltextually prov~ded companson class, a colour adjective require\ cornpleaon by means of a corltextuaUy provlded part of the object Actually Szabb thlnks that colour adjectives requ~re both tho ont textual provision of some relevant part and of a cotnpmson class, so for him dle 1og.lcd form of 'green' is

(Green (C, P)) x

where '(2 IS A van,tble standing for a cornpaason class a i d 'P\s a vanable sta~idirlg for a certar] part of the object

In contrast, the standard contextual~st posltlon regarding colotlr predi- cates i s tkut an adje~trve bke 'green' or 're~fdenotes a detenmnate property or contriblxtes a d e t m n ~ n ~ ~ t e pred~cate. the pred~cate GWEN or HAVING lm COLUIJR GIZ~FN Tliir is a complete predicate, riot something that stands In need of contextual completion, hut m context the property that 1s ascnbecq n rnade Inore specrfic through spec~ficat~on of(1nter &a) the parts or dim :n-

\ion under vv111tl.1 tlte property ~ppher to the object wUed &bout This ~rgaablv n a CXP ot 6ee enrichment, sirnilar to the 'cut' and 'stop' ex~rnples dncru~ed h o v e

'The Lollo*vmg c on\ideratlon\ lend ~ I - I ~ I J . facie support to the contextu- d i ~ t dppr~ac h In ternis of modulahon.

It is not prr~narily the sernaritlcs of the word 'red', but sonle extrrnslc urage iac tor, naniely tizejart that tt is used lo predzcate ~orncthipag d u n object

posqanng r ~ i z ~ l t ~ l e ~ part$, that ~nt r~)d~ices dn ~ndetenn~nacy ~n need of contevtual resolution I'flere 1s no pruticular reason why that uidcter- run,lcy h u l d be reflected in tlie seniaritiLs of the expression (though, of course, r t might) Thrs make5 Szabit's account vulnerable to the Gncean sort of objechon h p k e proposed to semantlc accounh of the referena,~l use of definite descript~onci. There a no need, Kripke polrited out, to endow referenha1 descnpt~ons w ~ t h a specla1 semantics

Page 64: Recanati, François. Truth-Conditional Pragmatics

to account for their referenad use, for even ~f we strck to tl-re rtarrdard, Kussehan semaltlcs wl-uch tredts descnptlons as cp'xnhfren, we c,in still account for their referenha1 use In pragrnatrt terms. i fwe rmdglle a

language m whzch descrip~ons have, by stlpulatlon, the R ~ ~ s s r l h a ~ ~ semantics, and mraglne such a la~igtlage to be used, ~t 15 pretty clear chat its descnptlons would Lome to senye '1 referenhid Sur~chon just as

ord~nary descnptlons do. S i~da r ly , let 11% rm,Lgne 3 larrguage rn whlch (bv ftipulation) 'red'rxlrans RED, where m,n 1% a ~ o q l e t r preclr cate. N o fiee van~b le 1s mvul\red But assunre t h ~ t this Imguag~" 1j used

to ascnbe the colour red to varioui, olyects, and that the objects rrr

question are complex and posress varrc,u+ parts or ~udicei, wl-rocc colour need not be honioger~eo~~s Tlle yilestlon \vlU anse, wEilc,Il part of the object is relevant to the colour axnptlorr And the i orrtcxt

WIU often mswer the queshon implicitly, In \uch a way t l~at a 5rrnpXe

ascnptton, ' a ir red" will be mlderstuod as a\cnbirig rednev, to (2 wrth

respect to some contextu-illy 5,J ent part P So whether or not there 1s ,L

overt vandble, there ~7111 be tjclt rekrence to the relevd~lt p ~ r t s of the object; it follows that pocrtrrlg 3, covert varrable to azcoutlt fbr he L ~ C it

reference in qttestlon ic s ~ ~ p e d - l t ~ o i ~ ~ lnme wIi1ch coi~lpllcate'a the semantics without buving ui, anythlng " The clrrlple fict t h ~ t the torn1

ofwords 'a 1s red' 1s uled to a5inl)e redness to a under some corltcxtu

AIJ relevant pdrt or din-rerlsi~~~ P promdes no evidence that the x J j i ~ c

tlve 'red' Itself ~r~volves A frec van,ible sra13d1ng for sorrle r ont cxtrrailv

relevmt part of the object So the onlv reason we have to dci rpt tlrr mdexlcal~st mew asglled l.hl bv S ~ a h 6 is that. I)v SO doing, we (a11 S ~ V C

the compositional &arriewolk in rr+ unrevised forn~ (1 e \v~tlioiit nl0t1 ~llabon 3 r d Gee pragmatic fimr tlcms) But this a a reclson n ~ l i ~ i h Sr 'd l?~)

cannot invoke m the pre\erlt debate without beggng rhc cltre\tlon (s~nce his opponent 1s a corrtextct-u&~t mvukzilrlg ~modulatlon 111 Cavclrrr i r k

a revlsion of the ~ t ~ n d a r d frarr~e*vork) (Ophonahty cntenon) Berng a top down, context-dnven process, rnodulatlon takes plate in some contexts mt.1 not otl~elr, cvhtlr

saturation, being Iiiigu~strcdly nraliciated ill vmue of lcvical propcares of the expression type, is bouncf to tale pkt e in all felleltous mrs of the

See Hall (zuc)X: 440) for a sirniiar poirit.

Page 65: Recanati, François. Truth-Conditional Pragmatics

cxprcszion Mociul'ltroa 1s ognorlai, saturdtxon rq rnanclatory In the c olcaux kbcr~ptlotl case, 1s the ~ontzxru,il detcmunatlon of the relevant part(\) ~ridfidc~tory or ol)tloxbdl? Ac- ord ding to the i ~ ~ d e x i c ~ ~ l ~ ~ t ~t is n\,irrd,ltory bcc,au\e dlerc. n A free vnriable, anti frec vrt~~nbles need to

bib ~ s s i g ~ ~ r d v,iiuc\ or1 pam of. serraarltic in<-orr~pletencr\ (whatever the

c v n t r ~ t j Atcoading tc, the rorxtc~tu~~list, the deterxrun~tlo*~ of the reiev~rxt p a t is rcqulred lri come cor.itexts but not others; for what irrdmcc.\ the rcqurrcrxserrt 1s (~rltcr ,rIr.ij the c olxlplerztv of-the object to wlri~h rlic i olotrr 1\ ,~scrrbcd d i d thc c ocx\lstence oi scvcr,d p ~ r t s or

suri;\ics tirxt ilt~ght bc. drf-Cerentiy colourecl, rdther thdn s lex~cal prop- rrtv o i rhc expresslori. To put the ~ p t ~ o ~ s ~ ~ l i t y cnterlorl to use. we luve to arnagmrc ,I c ‘rat 11) mrb~ch the c ornplex~ty oi the entltv to L V ~ I C ~ the it-riour r \ ,~\cribeci 1s 3ultably rcd~rcccl I'he c trntextual~st prec.tict\ that, m s u i l ~ A c,r$e, the preci~catz K r r > v,rU ~ppXy 'c~bsol~~tel\i' to the object wittrout 'txry iirrther s ~ ~ c c ~ f l c ~ i t ~ o r ~ Irrsuf-ir 35 OIIC c a r tell, th,a seenis to

hc t l ~ cxce " I f , rristeaci of "'l'hc. boob rr red', we say something llke

'hilagme ,L rcci \uri ,~i e' or 'I~rn~lgxr~c .I reti \pot', tlrere 1s no need hr the ( i)rlter\t to sptbc ii; Jrlv releb,irrt p,lrt o r Ci i l l ~ t "~~ \~o t l ofthe cirl+>c-e or the \pot cdki.cI ahour.

s Xir ,iriv c 4i\e, elre riiciexrcalrsr propod r\ ~rn~idecjuate a\ ~tated, for thele

,Ire otkrc.~ ~irarrcr1rrorr3 of ~ikdi.te~rrrtrrac v wkut 11 the no ti or^ of 'part' c arrnc >t prcsyrczrly 11;lxlcilo. Sce tlae 1 dlrav yirtrt'3tron ,~bove. bec~des the \devaxit paat ' iwrr, there xc Aro tlrc riwr itbcjut ~vllether the c o l o ~ ~ r I\

riat~rr,iE/trr~g~~l,~i/grntl~rle or, o r 1 rhr contrnr)i, a-t~ticiai and super-

'iddeli I'hlt clirner~slorz 1s callerlt rn the I'ravn c-x~z~~ple tvll~ch S73b6 ( l ist u\\i \ 111 \ C > I I I ~ C O ~ ~ C Y ~ S 1~di-t XI I)c \ ad LO be greet1 oldy rf it 1s so rr~turaly, rn other contexts rt 1s dcc epti"ule rf ~ l i z Iedrs panted peen. (Sz'rbO ah~r~ks tlrzr vanatlon car1 bc ttdndlccl r n tenns of 'parts'. but I arll

e p i c A/ ) 01, tc) take another 'Tr'rvxc exanxpic, what rridtters 117 certd~rl i3~t" 'r IS OW t i l e object tenrporanly ioak5, In other cuutext5 how ~t IS.

i r a ti nlcrn: e~ldtrnng senre (r e how rt look\ ln ~ ? o r t $ z ~ ~ l ~or td i t~on~) .

Max buy\ A 1t.d 2nd red p,untixrg i-rc-rrn ,r cut-reirtiy popular abttmct painter He 1i.tngs ut 0x1 the w ~ l l o t - h b1.u hgh t room Willen he tllnr, on the blacWlg11t.

i say \iniofzi ~li; onc c,rri ~eli', bcca~lsr the ~ n d c ~ ~ c ~ l l s t c r r i dwlrv\ q p i e that the "an' vnnable m tn\~i:+iIy ,uiciicd a cot?textir;d v.riuz in \u< h a casc (11hn1ciy the only surf:~ce/pan uf the uiisiri~ctured e>lgi:r.~ 1~11keci .ri)c>iirl (hi ilie i i t i i i t idiv oC,tp~pisrr~g the ojruvndriv criLeSnon, sce Rec:tnati (aori4: l o r - 2) .

Page 66: Recanati, François. Truth-Conditional Pragmatics

--- A I ) J E ( , L ~ L ~ ~ A ( ASL 5 1 LJI)Y -jj)

the panting looks puqile 'Now ~ t ' q p~uple', he says. Did he trU the trtlthp We can't tell. 111 coimnentmg on the effects ofbl.zck_lrglits oo tlurrg, orlc xlzigttt say (or utter) cvktat MAX LYd and say something true ('When I turn c31l thc I~glit, tins rock dows green, that poster sparkles, the hra o r l my parits gl~lows, ~ n d tile pant~ng IS p t~ple ') But if the politt a that Max \enorisly bc*hcves that ~dtrrtvlolet ravs char~ge the color of tf-ung ('lt', red now, but exposrwe to tdtrav~olet mll turn rt purple '), then, assurfung that we are i l r~f~ng w1t11 a

~ ~ o r n ~ a l red on red pG~mt~ng, what M.LY sad IS, of~ourye, i;l~e (7 I , W ~ 1975 )I)

Onc e agaln we find that the ianle forni of word 'the pA1nting rs purple' can be used to sexy true thmgc in some context\ and f&e thing I n

others, even tllough the cond3tron ot the object ~ t lked '~ltotrt dcxs i ~ o t chalge &om orte context to the next; yet t m t appeal to the pLkrb ofthe object does not seem to explain ar~vthlng In t h s part~cular case ( ~ n d in ir~definitely marly other<, arguably)

5 . Cradable acijectives: nlociuPatio11 or saturation?

So far I hdve ~5sumed that the ~nterpretat~on of gGida171~ a4e( ~ I V C \ s u ~ h 'I\

'small' or 'tall' does ~nvolve the Fatilratlon of '1 free v ,~n ,~bl r jstandlng for cornparlion dass or standard of comp~r i~on) , but tlldt too rs tleb~rable. (Sec Cappelen ~ n d Leporc 200ja for doubts about the ctandad mdysu.) What reasctn 1s there to hold that tlxelr nieanlng IS gappy whde rhat of colour ad~ectlves zs noti

Adjrcnves hke ' t d ' are a subclass of the class of graddle ddjt.ctlve\ Gradable adlectrves have a tlumber of vlterestlxlg piopeales (Kleln 1991

673-4) they &nke nlo&lrf-~ers such as 'very'; they t y p ~ c d v come In pam of polar antonym (tall/short, young/old, brg/srriall); they enter into compx- atrve constnlctrons (a< tall AS, taller thm, tallest) ,md, rnore f i l nda~ i~en~ l l j , ~r~ipose a1 ordenng on objects accorchrrg to whether they possess the property to a greater or lesser extent 'To account for tttetr behavtctur, it a corlmon to ascnbe two argunients to the pred~catei corrmbuted by gr~ctable adject~ves: besldes the object to whlch the predicate applies, there 1s the degree to which that object possesses the property In 'a little happy', ' ~ c r v happy', 'extremely happy' the degcee is located as low, high, or very hlgli on the. scale of happmess (I e the ordered set of degrees along the happlnesc

Page 67: Recanati, François. Truth-Conditional Pragmatics

d ~ ~ i ~ e n s i o ~ i ) In comparatrves (John IS happ~er than Sue) the degree 1s said to be supenor to the degree--rfar~y--to ud~.~ch S L I ~ n happy

FoUovvlng a liumber of authors, let us assume that, arsoclated w ~ t h a gadable ad~cctlve, &ere 1s a 'measure function' from objects to degrees on tlie relevaxit scale The vduc of the tunchon for a certan object as argunlent

* * I ir the degree to which that object: possesses the gradable property. Intuitwe- lv, jf 1 say tjotin lr b~ppy', 1 say that there is a (nonzero) degree d on the scale of happ~nes such thdt % w ~ ~ ( J o h r i ) = d. 'Joh11 IS happy', therefore, means 'John rs happy to seine degee', and 'Johri n very happy' means that for sonie d, ~ m ~ u ( J o I i n ) = d ,ind d n I.11gh on the scale 'John 1s happier than Sue' says that, for \onie ptzsitlve d, LULAI'PY(JO~~) = d and for every d' 6 m e u ( S u e ) =

d' then d > d' So fir I have not introduced any free vanable fbr a conipanson class or

standarc1 of Lonipanson. Indeed to say that sonreone a happy a not to \ay tiidt he or \he IS lrappler thai average or happler than some contextual st~ridard It rs jmt to say that the perron 1s happy to some (unspecified) extent (Pragrna~cdly, of cotlrse, the degree of happiness wdl be speclfied at least ' 8 % betng slgr~fiiant enougli tct be worth riientionlng ) But what about 'all'; To 5ay t h ~ t sorrzeone 1s tall 13 to say that he or she 1s taller than some >t~indard, thdt 15, that tlre degree to w h ~ c h he or she rs tail 1s greater than some degee wrvlng ar the standard of conipan\on. Here, m contrast to the case of "happy', it \eernc, that the positlve 'ritr~xi is coztettly covnparirtlve The evlderice Ibr tlils 15 twofold t mt, ,ls we Iiave seen, wmeone can be tall wlth respe t to a certaln courrpanc,on class (e g the class of zo-ve~r old bovs) but not wtl i respect to another (e g tall for a basketball player), this relativity makes sense ~f ' d l ' IS understood as 'tdler than some contextually provided standard d,' Set onti, h r n the f x t that A IS taller than B, it does not tbllow that A is tall Again, tiirs nldkes ~ n \ c ~f 'A a tall' 1s analysed as 'A IS taller than dC', since R's s17e xilav be LVC~I below R, ~ n d A's size rnay be Internledlate between B's size ' l ~ d Lf,

If we \A> tliat, i r ~ the case of 'tall', the poqitlve form 1s covertly compara- cive ~ r r d sl10~11~1 be dti~lv~ed m terns ot the more bas~c 'taller tlian', ~t seernls thdt we Inwt draw a dist~nct~on between two senses tbr 'tall' and slnular

yohn is lvapp~er than Sue' entails that John is (to some degree) but does riot entail that Sue is. tiSue IS urdnppy, John is happier than she is. So we cannot analyse 'John i s happier tlmt~ Sue' as 'for some d, for soirle d', happy (John) = d & happy (Sue) = d' and d > d", for that would entail 'for some 8 happy (Sue) = d" whiclr is our rendering of 'Sue is happy'.

Page 68: Recanati, François. Truth-Conditional Pragmatics

adjectives. the sense t id, ~t takes 1x1 the ~)os~tivc cor~stn~ctlolr (yoha r i dl') and the sense tull, ~t takes m the more hasic cornparatlve constx'~ict1on rr,

terns ofwhch the positlve colistnlctloxr n to be axraly\ed ('Johrt IS tallcr than Bill'). The pnma ljcie diEererice betweerr the two 3eli5~c t d l l and lull, I\

wliat the fact I have already ~~ierrtroned porrlts to: TVhile 'A 1s tlapplrr tlrarr B' entails that A is happy (tcr corxle extent), % 1s ~ l l e r than B' does not e r r t ~ ~ l that A is tall; For A may bc small (hence not tall) while st111 beirri: faller than 13 'Vhat IS presumably due to the &silrlc~or~ bettveeri tcvo sensei 1x1 thr case of 'tall' (but not ~n the cJse of '11~qqy') 111 'A 1) taller than I)', 'tall' 1s takcrl m

the baslc sense ttill, ('A is tdll, to a hrghsr deg~ee t h ~ n B')-- i r , what "A 15 t d e r than B' entails 1s that A rc LA, (~'121 to wrnc degree). Rrrt X AS t,ll19 doer

not mean that A w tall,, ~t n1e~11s that R ir tall, " We can define the prlrn't Iac re cerlcc t d l , a, follocvi

To be tall,, tllereforc, is to have sot~ie (n011zer(3) degree ofherght L.,tcr~ ven ihort persons are tall, la that yemr Wllexi 1 qav "John 15 tall', cnl c otzrqt 1 do not rr1ea that he 15 tau, What t me~trl~s that hrs degree ofhelgllt 1s \upmot to some contextual standard (s,u), the average hrrght atllong 20-vear oliic) I-lcre 'tall' is used u~ the serotid wnrc, tiill,

where 'd,' is the contextual staardard Intu~tlvely, the bxic mearlmg ,:(all, i z f-oilnd r ~ o t only 111 explxclt cornpara-

t~ves ('John 1s txller tlran Bill') 1)ut ~ l s o ln questions ('Flow tall n Jolrr~i', meanulg: Wh-tch degree d 1s such thi~t I ~ ~ L G I I T (John) - d 7 ) and m expl~crt statements ordegree ('Jot111 1s six feet tall') Neltller the qilestxon '1 fow td31 I\

John' nor the degree statenlent 'I lle IS srx feet t,d' entail or presuppose that John is tall. Thn 1s presumably due to the f i t t that 111 all ~he.ie cnnstrrrcnorls,

" In his conm~ex~ts on an earlier version oftlus chapter, I%a\d Egri expressecl c2isratistachon with ti-re

claim that 'John is happier than Sue' e n d s tlrat John is happy to some degret;. f f r g1;rvc-s an emrxiyle in wllich John is definitely unhappy bur 'still happier than those guys on the street', :md cottt:iuclo that there i s no significant difference between 'i~appv' and 'titt31 ':in nritlter crcse does the cornyardtivr Qrin eritaii the positive t o m . However, in tile case be cites one could respond by correcutig the speaker: 'you nrran, he ic less rinhappy than they are'. EIilr ill pointrng ti) orrr nf two very shon penons, 1 say ' h a t one a taller', it would be not only pedantic but strange {or humor~)us) to correct me by raymg: 'you i-rrea~r, 11e IS

less short'. (If you share Eg&'s inturtlons, rcphcr 21r3pp\i"/'unIu"~~~' wit11 'sa(t'/'gayyl tliror~glmut.)

Page 69: Recanati, François. Truth-Conditional Pragmatics

62 CrKiii rAl;i E. ALgR EVE\ . h/lUt>t/~P\l !<XU < ) I < \ATUIL;\ ION?

't3U' t,ikc\ the seriit, ~ 2 1 1 , wbrle rt ukec the w z \ e ttrll, in the posltxve fi>rrrr

*]<)llll I', 1~11' \c:, tlrc p(751tlvc" GITXI~ act\ .IS 2 ( crvcrt conlp*isatr\e Why is that so? Orre

possrblc c~j>l,il;i '~tli)~~, mggestcd by votl Stecl~ow (rgC)qj, rs thdt there 1s 3 null cit~gjec rrtorl>i~ernr PO5 In 'J>tln i s t.d1', whwe sern,lntlc contrlbutmn ~ c c oirrlh tor the srunlse tidl, cv'li~cl-t the ad;)et t-ive takes lrr the poslttve con- ltrut n o r l i/i thl\ I \ nght, tile11 one rnny corgecrure that the lexlcal Itern 'tall' hi, tE~c s , , l ~ ~ ~ ~ I I I C ~ A I I I I I ~ ,XU r l~e corupardtr.ve (% 1r\ taller than B') d i d the poutsve. tor111 ('A I\ tau') niter ,111, htrt thzt rl: rtrnlpo\es cvlth the cornpAranve ~norphenie nr rrone case ancl the pltorrologcally null yos~tlve morpheme m IIIC trrller the scrrse tull, I \ the rewit: of corrrposlrlg the baslc sense of the ,idle( CI'SC" jwherl~rr i t rs f i r l l , or \osr~et\tmg eke)'' wrrtl the sense ofthe poslhve rr lorpk~crr icb X I w e opt f o r ;1*1 nr~alysls +*long tl~ctie I I I ~ ~ ~ , we st111 have to

rspl,rin t i le tl~iiercilt e hetwecn 'h.~ppy' and "tall' c\liy does POS iriduce a T ~ ~ T W I ~ J I I I I ' , L X I I I ) ~ I I I one c,ne but nctt i r l the iljtherj I. lcre we cannot but ~ppeal to, tlrc Icar c,d prc )pertics of 'tdll' zjs 'Ir6ippyi to e ~ p l a ~ r l tlre ii~kGerence.

Ancjtircr rlicorc.tic,k) O ~ ~ I O I I , C V J I C ~ L t , ~ ( cij w ~ t b the pnina ficle &fikren~e between the \err\c\ tialt, &ind tull,, 1s to selcgate tlre recor~d seme to pr~glatrcs h-c a ppivsng ( ;rrci-'s nniodlfie J C )c can?" l<a./nr ' 3 he fbllowmg account mggests it4e1f ' 1 ~ a l i ' ,lctu,rlly ha\ (only) thc \erlsr ~ L J E I , To 5ny t11'1t romeone I\ t.rli 14 to 5,1v (licer,ill.t) that r h ~ t per\on hm wnze posrtive ilegrec of height. ~ Z U L of ccrrirsc, rlrn r \ ,tbrolttteIy unrrlkorrn,rt~cc, zlrlce ~verybody 1s d l , . The

;W<ici. i i ~ r i tiii\ri I, ,a (liiicirricc bctweerl 'tali' anii 'Jiurt' xri t l r t 5 r rg~rd . We canrrot say .John 1s three trct r l r n k ; r r i l i if' fie .I& '/ IOW s110st 15J01111', t11:tt entail?, o r ~ ) r e s ~ ~ p p o s ~ ~ tll:ir be is ~ h n r t , 1.2. shorter t h . ~ ~ .ivzragi.. I'l\,it d~flcrriicr 1, pre*umrtlilg due t o the Let th.1~ in the pair o ipuhr oppra,trs. 'shtm' is tlre ni.irkcti terrii wliilr -t'~ii' 1s t i le iinrnarkzd one. 1 1 niryirt be tlro~rgin that IJUS IS ail there is to ttte (listlnctlon beiweeil roil, J I I ~ tall, it I$ well- biuwrr tl-~it rlrc u~~niarkrci Lcnrr nC a pzllr of xutonyrrri ten& to be piilyicri~i>iz~ a i i ( i can t.\I;i: cithct- the c(>rr.~plenrentzry sense whrcil m tius i-ast corresponrl~ to the upper part ai' t i i c iiisgltt c,i.ale w i h 'si~on' correspondrny to the lower p:rrt- trr itre n e u n l sense that i.orreiptxfiiii to tiic entire \(..zii,. thit the riisrinctiot~ herweer1 tbc two \rrtses t~dl, ,md /nil2 goes beyond tIi'at. it IS !lot t i ~ e t o tFie i,iri rirat 't:ilt' is the unmarked tern1 of the pair and rx lub~n the well--known i~dyscni\- . Ili(iec3, 11ic aLo need to .poiit u ba i t ~ertseli,r 'short>in 0rJr.r LJ r r u o w r ~ t ji.)r co~nparirtivrc: for 'Rill is cl-iovtcr ~ l i '~ r ! f i~h i~ ' doc\ iiot entali rtist i31UI 1s shon. I te n ~ : ~ y di-rul/jy be ~.r~t l~er ~ 1 1 . The conmst becween iiie pOSlhVe b r r r ~ ,liid tilt. t.ornpardtivc fbrrr] ho/ds, ~vhethrr rjlr term at issue 1s the marked or the urr111.trkri1 rr~rnrhi;~ td'tiie 11ar.

\.2ccordrng to n10.i I!fig~u\ts who halve dc.veioperi iorrrpclrtrorid .rccounw ;tlorrg those lines, the liasi< zr r i se oi thr. ~ c i j c i tvve i? more I-maic tltxl tail,. "Tdl' denotes eiilrcr the measure tunction &om cii)jem to ~lcp,rw~c,rr !tie ireiylir zc;rSe, or a rrlauon iietwern inti~vidu:ds and degrees on th;lt scale. Degree iiiorl)iroiog/ finc.litcliiig i 'CIS) is wh.rt converts n~krtevcr a wken to be tiic adjechvc's derlocano~~ into 3

[vropcriv o r individ~i.ds Cln I \ ~ I \ i crrcinlrrn vlew. tile exrstentrai qrar~aftc-arrcir~ over the degree asgurncnt is xiot p a t o,j the 11351( rnc:iii irig o/' I he n<?jecrive

Page 70: Recanati, François. Truth-Conditional Pragmatics

n-r'lrrm of qumbty therefore leads us to systerrr'ittcal1y rirlclerstal~d the

~~~ lq t~a l r t i ed staternertt 'John 1s tall' as meariirig not orily t h ~ t he has conkc:

degree ot height, but that he ha\ sonie degree of hcrght quffic~eutly om-

5tand1ng to be worth nrentronllig S~rnilarly. Spct her ,lnd VVil\ori pt~inted

out that dn utterance 11ke

I t w~ll take us sonie time to go there

hter,*lly rnearls something quite triv~al tlidt the tnp wrll not be t r x i t a u t

aneotrs I%r,~grnatlcally, however, tills 1s stengthmed so 35 to lr~njie the

st,rtement mfomianve: the utter.xnce 1s urrderstood ,kc \ay*)ililg t t ~ t ~t wrll

taLe 11s a s ip l f i ta t r t tlrne to go there

In t h ~ s w,iy we can attempt to exyl,~~n the ,illegeJ d~ffcrence I>tatwectl 'tall' ~ n d 'happy' For not evelybocty I\ h p p y . Smne per\ons are unhappy 7'0 be

told that sorr-reone n happy (to some extent) 1s itlfonrl~tive; there no nced

fbr stiengtientrrg B l ~ t given that dl the okyecti, In tire Jonia~n ofthe f - l f . 1 ~ ~ 1

fu'urli tlon helve height, to be toki that ,lo o b j e ~ t lrt that ~las,s I\ t'ti!l I\ totally

unlnl'ornrat~ve 5trengtherllng is requuect

A 1o11g t~trie ago, Jerry Sddock put forwacJ a prLLglllattc t1lec)ry nlong tliose

line\ to ~ c c o t ~ ~ i t for the polycemy of 'till'

Pile reason that Art!tur is lall so strongly indicates that Artltirr's hciglrt is at>ovc- averagc [can] ht. tracetf to a cortversatio~r,d ir-r~plicature. S~rying orlly that Artlriir 11% height to solile extent is not a partic~ilarly erllightrling contributio~?. In ;lhnost all

circiimstances, we would be 61rceci to search for a way ro m:ikc this trivid

observation re1ev;tnt. Tlic relevance cannot be sougi~t in the speakcr's desire to corivey that Arthur's height i s suq>risingly s1n;~ll since if that werc. tile c~sc he coulti

(a~iil tbrrefctre shoulci) l~ave said Arthur is slzort. So all that rem:uns is tilt- prt-stimp- tion tirat the spraker nrieant to convey h i s feelings that Arrli~~r's heiglkrt is urrusually gc:zt. (Sadoclc 1981: 261n.)

Can we, In this way, ac count fkr the sense t(t11, by s~ylng thdt ~t 1s a

,t~c~tgtlrenmg of the bast( \en\e tczll,, hence rrlst'ir~cc of ~rtodul'~tlt>n?

il-rzrc ,Ire A couple of prlmd £iCie objeitions t o sue t~ a move Let us, see

whether they can be met.

On the piagnratzc story, nottr~ng preterit\ 'Jotln 1s t,ll17 (wrtlr the literal sense 'John ~c tall,') i'rorn being strengthened-- rendered rnore nifortn~txve--

111 the opyo~~tt" d~rectiotl ni~d understood as ~ileaning t l~at Jolln's height 1s

surpnslngly low (rather than st~rpnurtgly h~gh). To rille out \ ~ l i h an rritcrprc

Page 71: Recanati, François. Truth-Conditional Pragmatics

- -64- - G ~ ~ ~ F .kt?* W%s;.MiOmA"Floi.u OR ~ A T ~ ~ R A T I O N ?

tation, Sadock invokes the existence of the word 'short': if what the speaker meant was that John'r degree of height is surprisingly low (rather t h a ~ surpnsmgly h~gh), he should Ii,lve said that John 1s chort, since that is what the word \,hurt' 1s for. For Sadock, 'tall' means tall, and acquires the sense tall, pra~m~atically, but 'short', being the marked term of the antonym par, lexically codes for the seme dmrt,. In other words, Sadock's pra_grnatlc story orily applies to the ~inrnarked tern1 of the antonym par. But if we want to generalize that story to all the ac-ijectlves wh~ch, in the posltive forn~, intuit~v~.ly irrvolvc 'i corirpar~son to a contextudly provided standard, then we must qply it to 'short' as well as to 't~ll'. (Indeed, as I mentmned in hotnote 7, we also need a di\t~nctlon between two senses in the case of 'short': 'A is shorter d ~ a t ~ l3' does not enwl that A lr short It only rneans that A's degree of height 1s .irnaller than R'r. There 1s a.i rliuch difference between the .iei~se slrorl, we tirid in 'John u shorter than Bd ' and the sense short, we find in tjol~r? 1.5 short' AS &ere was between t d , and tail,.) Now, rf 'short' hterally rnedtis slzort, ratlrer than sftort,, then we can no longer rule out the unu~rlcotne \trenhshening of 'John is tall' by arguing that the speaker should 11,tve uwd the htcr,ll fimn 'John is short' instead, for thdt ~ r p ~ i ~ e n t makes sense ortiv d :)oh11 1\ short' hterallv mems that John 1s short,. If its literal 111eai11rlg 19 as ~111n10rn1ativt' as the literal meaning of 'John 1s taU' is supposed to be, then the argttrncnt collapse.,

Let u\ *r.isrmle that thts problem can be solved. How would the generalifed pragri"lrxtic story go in the case of 'short'? We have seen that 'A 1s rhorter than 13' doer not cnml that A is short. It only means th t A's degree of her& 1s sndler than U's. IHere the ac)echve 'short' seems to coritribute ($molt) the sane property as ' t d ' does in 'taller than', namely the property of havirlg sonle degree of height. So perliaps we can say that '~11' and 'short' are associnted vvltlii the same measure function from objects to dcgrees on the he~ght scale. Of course, there is significant diEerence: mth 'short' the dzrttctzorz of the st d e (the ordenng relation) is reversed. ro that the cornparatxve of supenonty 'more' (or '-er') means lower orz the scale in the 'short' c'ue arid htyCter in the ' t d ' case. Given dl thts, the pragmatic story mris as followr: 'John is short' literally means that he xs short,, and that rneans that lie h a some degree of height. This is pragmancdly strengthened: the degree to whlch he 1s short, 1s understood as 'niore' than normal or expected. Because the direction of the scale is reversed, the comparative rnorphenie 'more' rs trnderstood in the opposite direction than it IS in the

Page 72: Recanati, François. Truth-Conditional Pragmatics

A 3 V <; I j .E.cTiVGS: A C A S E S ' l ' i i l jY 65

case of W: it now presents John's degree of height as lou~er than rtor111alJ expected. This gives m the sense sCtnrt,.

Note that the lexlcd er~codusg of the dlrcctlon of tfrc ccalc solve\ the problem we first ralsed. If the \peaker, througii ctrcrlgherr~tig, wanrccl to

*n conmunlcate that John's Itelght is ccruzparntlvely low (rather than cotnpar-

; atively high), he should 11ave saicl llohrr rs short' That does not htera1Lll-y xnean that John is short, hut nevertheles., tunvcys that nleaning througls strengthening. With 'John 15 t~ l l ' , streixghcnlng gves the wrong result, gtven the dlrectlon of tlie scale So sol~~ctlxing llke S~ctock's argtmleiit goci through, even rf we general~~e the pmgxlatrc- c to t~

W ~ d t about the optiondt) cntrnvxl7 Eor d lo~lg trlne 1 r1lougllrt thdt ~t

ruled out the moduia~on optlon &I gr~~dabie ad~ect~vcc Irke "tall' I i t i le

baslc sense tall, was freely er~rlched jitrertgl~enecl) ~rlto t(xJ1, in order t t r awke

the statenlent mfor~natxve, then rt oiqqCet a) he posithie to corzcrtrtii ti ionluxt w z

zi~htch no such rtettgtj~enrn~ IS vt7ytured jlxx,niir even nnrnmal v a l ~ ~ e would bc mforniat~ve). BLI~-I thouglil~t -that I\ not poss~hle. :John 1\ t ~ l l I I C V ~ ~

n~eatls that Jolm IS tall to wnle cfcgree or otlter, t11xt IS to s q , t h ~ r Ire h,14

sorne degree of height, evcn rf we Irxlag>rri. a ~ i c r r ~ t context 111 wlut h not

everybody has height I an1 no longer so sure Dan Sperber (rn dricucsron) I1.n ~oxlrc tip wrth 2

cor~vrr~clng exmiple: that of ,r ultlvrr\cX wrtklr both two-cl~mcncron'~l 'and

three-dlmem~onal objects, wilere r t would be naturdl to rcSer to the thee- Jlrnenslonal objects &, the 'thtck' o \ ; l j c~~s TI'fricli91cre nieau~ iiorxsetl~mg hke 'having some non-~ero de'gree of tlllcL;rle.is' So I am no longer SO sure

what the verdict of the optronal~ty ( r-lcmox~ n ur tho case. I3e that a\ rt rrray, the modula~on approach s tdes xttc as, 0x1 t i le whole. llot very plaucrblc It says that we h a ~ c to strengtlnen 111 the case of 't,lll' because the ~~rmrm~xl degree-of-helgl~t property 1s trivial arid rb scnption unlrrforn~atxve Rrxt what about 'expengive'? Some objech are Gee, hence tlney axe not expenclvc to any depee--acqumng them ~rlvolve\ rzo cost wlt,it,oevcr. So rt is xlut that

hrvial to be told that sornethrng IS c.vpeuiAtt?t*,, that n, thdt IS has a co\t St~ll, 'tha is expensive' IS-always-undentt>od I ~ I the sense expensaslvc7, (m_volvu~g 3 hlgher cost than the contextual sta~~dard) TI113 13 ltndccounted for, on the pragmatic story.

Page 73: Recanati, François. Truth-Conditional Pragmatics

Kt:rinedv riotcs tlrlit tlrc cascts ~ r l i i i . 1 1 IIPVOIC,~ ;L co~ltrxtunl standard teird to be the c;tst.s in which the scale ; k t issur is 'open'. that is, hxs neither a minimnt~rn nor l l laxbillti i lt (/e"gree (Ker~ne~iy .xrtcl iblcN;~lly hooj; Kenrredy 2007)." I-Ie otfi=r.c rile &>lltr-c\i~ng ggcner:iliz:rtic,n: vvivhcrr the scale i~ssttciated with an ;tc~ectivc. is (partly or t:ot;dly) closed, the positive form is typically under-- stooil :is stlying t113i the s~~bject 's degree on thi: scale is at least as great as the

rninu.trrla1 C)B. I I I ~ X ~ J I I ; ~ ~ degree (2s the case xrray ire); wberr the scale is open, the posflive fi>rrrr i s iliicdcr-stood as saying t h t the sutject's dogee 0x1 the scalc i s at least ;rs gr:rrLrt a s . . . S0311e cor1testu~1 sr;in&~rd. So context-sensitivity is iinlceti I ( r ~lnt)oundecixless of tire itssociatecl scale. That (for Ke~~nedy)

is cvl~err ~ ixc cdifjkrcxrcc between 'l~.la~>py' and ".call' comes Gom. 'Happy' is

aswciatcd with a ic\\r\iur-boundeci scalc (2s rlxr i'c1icit.y of 'absolutely/totally r inl~~ppy' skiows); sts rct say that sorlleone is happy is to say that he or she has

solno. rrotwero ciegrre c,fltappitiess (or c-h:xr his/l~cv- degree of happiness is at

Fcasr w l;ri.,lt ;LS r:l;rr rninirnuxn tlegree o n tlrr sc;~lej. Kcrt 'tall' is associated wit11 a11 opein ~ i l c : ~ I ~ e r c is rlo rt~iriinrtrnl (rxor~zeso) degree of height, nor is tbicue ;r ari,rxiriirrrr~ ticgree. Thc sc-ale is i~rtboalndeci at both ends.

ZCenirr~iy's tlet:riie(l discnssic>xl corrvinc~ixlg1y establislres that there are two t y p t a s o f gradable ircijectives: rhose cllat arc relncive arlci context-depentfesrt

('sslrort', "srn:dl', 'expprsive'. etc.) nnci t.l.lcae tll;tt are absolute ('happy', 'wet', ope^'‘, 'ct.rr~pty'~ 'srraigllt'). Wittrirl tht: class of' absolute gadattle acjicctives, t-hc:rc; is :I tiarkher srlhctivision, cleprnding on whether the scale has :r maxi- r - r i r r r i~ . Aai :idjcctivc whose nssociatc.~! sc:iic 913s $1 I I ~ ; I S ~ ~ I I L ~ ~ I L o111y applies (in tllc positrvr fi~rrri: "a is F ' ) to tltose objr:crs that :ire ~ i z ( ~ ~ i ~ r ~ ~ i l l y f? SO to be i l ~ p p y i s to ht" h;~ppy to SOXIIC degce, hrrt to be rrrxpty is to be rn:tximdly t:rnpty: ;i Ir;ili:-c:rnpty contairler is not empty.

1 s a i d t h a t ari ;~pj?roach in terrrrs of' rrrodrrlat-iorr is n o t very plausible for relative gradable actjectives, but lor absolute gradahle adjectives of ttre 'rnaxirnal' tvpe such :.it1 approach looks rr~oxc pronrising. Let us corrsider kerrrpt-y'. Sirce "c-rrlpty' is a gratiable adjthctive (it accepts degree moclifiers

'"~r: know r.hat .&ri .ici]ecu\rc X" 1s ass~(.~aicd wit11 .% 3 ~ 3 1 ~ thrlt h.1~ d r~~axinltim &gee w h e ~ l ~t makes icrisr t c i rncid~ii, the id~ectivc> with 'absoil~tiiy' (itl~solzctcly rrrrpry, ,il?iciir~r~~ly sfruiyht, absnilcicly ~ r i~? iy las , vs.

o b s i l l t i t e l y d ~ o r t ob~~ltdtely I~.~~C~I(SILV)). W h ~ r l ,tsi rll / leci~~e J; IS sszot~saie~d wlrb a sc'rie thai Etas J. minirnum degree, it mkri iriin. to mo~ii+ t i l t . polar arltunyili n'iih "ail~roiirti . ly ' (sri~ct: ntg.it.ion reverses the scde '~iiti iii.ilrr\ ,i !osvt*t hotcritit~d ic'ile tipper- Iroiirttirii)

Page 74: Recanati, François. Truth-Conditional Pragmatics

A I > j t < r i V 1 \. A C A \ E \ I U I ) Y 67

such ,-is 'very' ,lnd the coniparanve coitstruct~t.an), tl^icrc 15 ,in a\\oc~'xt~d rrleasure htlctlon, EMPTY, rrlapplng objects of the rekevai~t sort (vrx. corltarn ers) to the degrec to wh~cll they are errlpty As wxth "short' ~trlci relG1~ve ac!ject~ve\, but m contr;rst to 'Il,lppy', we find that the t onlpdrat~ve fc)rnl cioes not entail the posttive torrn 'A is elnptler than U' docs riot ent,ul tll,~t A n ernpty. So it seems, once again, t h ~ t there are two seascj, elrfpfy, (itelng ernpty to some degree or li.lvtng Tome d e ~ ~ e e of ernpt~ne\~) arid errzpty, (being erzrpty sztripltnt~~v) 111 contrs t to relative acijccttves, ho.i;vever, there n a r n ~ ~ ~ m ~ l degree on the scale, cor respo~~d~ng to the expres\rorr 'ahsol~ltely empty'; 2nd there 1s ari obviou\ llnk between the property of'betrlg cnlpty cimplznter (e~npty,) and the property o f b e ~ n g absolutely c~rlpty When wc \,ly that 'A n enlptrer than 13' docs liot errtall that A 1\ crrilTty, ~vhdt we 1r1earl r\

that it does n o t enhi1 that A a ~ h ~ o l ~ ~ t e l y emptb. Given all thr\, wl'tich property does the adjective 'etrlpty' c'rpresbj I f-wt

laulme that whatever property the adjective c*xpreuc\ rrlrlst be cvhat tieter mines rts extension, ttlen we carirlot s q that the property Iri questior~ 1s the property of be~rlg empty to sorrle degree or liavllzg wine degree of rrzrptl- nesc (cwipty,) For a t oritdiner tlut is not ernpty S ~ T ~ I ~ ~ I L ~ ~ P Y , ti12t IS, dbsolzitely emyty, doec not belong to the estenslon of 'empty' A haltltrx~pty glas, IS

not ernpty ever1 though it has sollle degree of emptine\\, 4 0 ~t does rrot fall wtt-nn the extensron of 'empty'.

As a matter offact, we know perfkctly well which property the acijccrive 'empty' exprecses. It 1s the property (tor '1 t uritalrrer) of inot t or1t;ltnlng anything, of beirig devoid ofcontent\. 7 hat 1s hc>w we define 'ectnpty' Note that this is a 1 a b ~ 0 1 1 i t ~ property, a p~uperty wtrtch A (o1ltnini.r h,n or doe3 not hme E~ther rt corltnlns something, or ~t doc\ not coxzt'xr~~ anvth~ng 50 the property wb1cE-1 the ,rcfjective expreszes drld wh~ckr deter~ixrtcs its extellsxctn a not A property that adrri~ts of degrees. Plow, then, can we exp1arl-I the grttdability of the adjecnve'r

Let LIC, grant that the adjective 'enttpty' eupxe\ws the property ot being (abrolutely) empty So tlie pnni'xry sense of 'empty' 1s ernply, Now we can construct a scde the m,lxiinurn degree ot w h ~ c h corresponds to thc property In question. The me~isure f-uncnort r-n/IPn WIU thetk 111,ip corltai~~ers t o

degrees on that scde, depending oil the extent to wtlrcf~ they apyrowmate ~mE>hfi"SS. Every degree on the fcak wlth the esceptlon of thtb rrl,lxirnllnl clegree wrll be cucl~ th,tt tvtrerl d corltalner rnnyc to that degree, the predicate 'empty' does rlaf applv to ~ t , heiati\e a (otlt,tincr ttrdt 15 not ~r~,t\i~rnallv errtpty

Page 75: Recanati, François. Truth-Conditional Pragmatics

- - 7% 7YFfSOLUTF

rs not erilpty \o the property whicl~ acimrts of degrees, and which the rneawrt: ttlnction measures, IS not the basic property of emptiness wluch the 'idjective 'eenlpty' pnn~anly expresses, but a dlsnnct property that can be defiried I ~ I terrtls of ~ t . the property of (as 1 sad) apro~rmntlvg ~~~pt tnes - . A half-eriipty g1:lnss possesses that property to a ineasurable extent. So it is

empty, If tins is nglit then there are ltuo properties assoc~ated w t h an adjective

sucli ,IS 'el~~pty' rhere is the basic property of ernptlnesc, corresponding to the pnlnary rerlse (ewtpty,) I t 1s absolute and does not adnut of degrees In tcriris of that property, however, we can define another predicate and generate scde ~ i>r respond~~g to the degrees to which thdt other predlcate apphes The adlei tive 1s polysenious because in additlor1 to ~ t s haslc sense, winch rs al>solute, it also takes A \econct, gradable sense that secondarv sense is what we get 1vhzr2 the ildjecti~e ukes degrce mod~fiers as in 'very enlph' or the conil~araavc torlbtructiori 'eniptier than' 1x1 'very empty', the ad~ec- tive 'emply' does riot contr~f~ute dle basic predlcate which applres only to enipty containers, b u ~ a der~ved predicate ~vhicli apphes to afl contaners th,rt ~pp10~11natv enzptmers to ronie extent

The scc ondary \en\e is ar~uablv denved from the baslc seme VIA 'broad- eiilng', wh~t h i< A fonn of nlodtdation (the converse of strenghening) Bro;tdtni~~g, or rcrne exter~sic>r~, is resporlsible for the rnetaphoncal under- standing of 'asleep' in exaniple (rr) of Chapter I (p. qr) The exteiision ofa pred~cate that liac tiridergone broadenmg a a ruperset of its orignal exten- sio~i (Carston r9(~7,zouz) I11 Chapter r s example (11) 'asleep' cxpres es the property qurx-I ANI) saowrivc, LITTLE ACTIVICY, a property which sleeping people (or ar~lrnkls) do have but ~vhich 1s also yo~sessed by many other objects. 1x1 the case of kn~pty', broaderiing gves us the seme we have when the word I\ used loosely and applies, say, to a theatre with only a handhi of spectators, or to a glass of beer that only c ontans a few drops of liquid The theatre or the g1:~d~s are not literally enipty (of beer or of spectators) but they are empty lo a loose, extended sense (Sperber and Wilson rgX6b)

L asenohn (1999) has formahzed the mtuitive idea of loose use through h ~ s notlon of 'pragrna~c tido', and I wll follow h m here. The pragmatic halo of an expression (relative to a @veil context) is a set of enaties of the same logcd type as the denotatlori of the expression Each entity in the set is

undentood to di8t.r ha1 the denotaaon orily in some respect that is

prajgmtically ignorable in the context Let us focus on the phrase 'emptv

Page 76: Recanati, François. Truth-Conditional Pragmatics

theatre', assurmrtg that the word 'theat~e' rn t l ~ l t phrase I\ urldet \toad stslctly and hterdly, so that the pragmatic hido of the conlplex phrasc only depends upon the pragmatic halo of tlie aj,lectlve "nipty' The tfenotanon of rhc phrase 'empty theatre' n dze \et of eltipry theatres (1 e theatre., mith rro

spectators) In the sort of context In whlctl one nz~glit describe theatre wrdi only a handful of spectators as empty, the halo of 'einpty tlleatre\n a i e t

including, in addltion to the set of hter~lly errzpry theatres. dze srt of tt~eatres

m t h only one spectator, the set oi tlzeatrer with only two spectatoa, up to a certain number n of spectators dny llilnlber of spectntors below n IS co

low that it is pragnlatic,illy ignor~hle m the context we mlag112r 'Thr example dlustratec an iniportant feature of tire pragllatx 11sltr of drr exprcs sion according to Lasersohn: ' ~ t may he ordercd, either totilly or pxt~dJy, rn

buch a way that the deno~zaon fbrrr1c a n ~ t u r ~ l endpoint for the ordering' (Lasersohn 1999: gab). I. et us ';ay that nrr object helorrg? to the exten\lorr of

'empty theatre' broadly urzderrtoud (in 2 context c), It a~ld only ~f rl: rr a

member of one of the rets In the halo of that expressrorl m L So a thcatre w t h SLX spectators is ~n d ~ e extenslotl of "er-ripty theatre' broadly understood, even ~ f l t 1s not In the extentslorr ol'enlpty thcatre' understood \ t r ~ ~ t l y ~r l i l literally.

The secondarv property w'tlrcb "mpty' expresbes through broaden~r,g 15,

1 sad, the property of apprttxlr~~atirzg emptiness In cirritr,ist to the j3nrnarv

property of emptiness, it ,zdrmts of degrees. '17hu\ we can say that the t l ~ e ~ t r e a 'very empty' tonight, or ' empt~er ' t l~ai~ it wxs yertcrday T11r exten\ion of the secondary property PruLpty, IS the ullrori of the \eh In the halo Kt rruy be ordered because tlle halo itselfis orde~ed hv the relat~on ok c-lorerlers to the

denotahoii: the set of theatres with e x a ~ t l ~ one spect'ttor 1, tlo\er to the denorahon ( V ~ Z the set oftlzeatres w ~ t h no spectator) than the wt of tlle~trrs with two spectators, and so fort11 So art object A possesses the pzopcrLy of dpprommahng emptmesr to a h~gher degree tharl anotlier ihject B IRA i b L

member of one of the sets in the ha30 and tIiat set n c-loser to the cler~o'iat~orr of 'empty' than any set of-c;vli~ch. B IS A n ~ e ~ l ~ b e r in tile izalo

It important to realize t11,lt the gradahlr seme i r k adjective\ lrkc 'cmptv', though derived by means of s pr,lgliatic rneclltlrrlsnz @ro'~denrng-~ form

o f modulahon), is neverthelesn part of rile convcntlonal nze'~r~ing of the

adjective. It is notjust that the cjer~vatioli ofthe gradable sense rc 't oerced" by degree modifiers s ~ ~ c h as "very' it 1% also converiho~iali;.ed an11 "pre- compiled in the lexcon' (as is <ad p m a footnote about ,motl~rr

Page 77: Recanati, François. Truth-Conditional Pragmatics

c aic of poh\r.aity) b lras to be atRnowleclgecl1f we x e to accoLrnt For the dri-It.~eria r bc*tweerr .il,sc>lute rrr,txlm;rl ,rcijc~t~vc$ fIkLe "wind', which are

gritd;rllie Ilc.iLrrnc tlrcv are cc>nvex~tioo,lily Auor lared w ~ t b the secondary

p ~ q w r t y tnd ';djrc tivrs Irkc "syber-rc & I 7 w i i ~ c h ,ire riot gadable (we cannot \,tv 'vverv y9hm1~dj' w~ l f~ t )u t odct~ty -sche 1 'zwrstrl~n [got). jq), even though thc.\i r o o i , l x l uridergtr broacferrttig rrt p'crt~c~rl-ir ~oalttxts. In the 6,unework

I propo*, 1 he yr r\s ctfrnoc!ul.lt~ori 14 c O I I V C " ~ J ~ I C ) ~ ~ ~ ~ * / C L 111 s(3111e (ayes and nc r t a>~lai'r\, l a r d wn/hct~ ~t 15, the ,iClject~vr IS ~ U ~ ~ ? I ~ W I I O L I S l t i

1 ?ihuiigEi rrrocl.iri,rtrc~rr rs optmo~rl, ~t t dn bc coerceii, a t d tbdt IS what h~ppens

wcoliea~ C ~ C " iisc tlegrcae rrlodrficr'l rn c-onnect~r)ri w ~ t h aci,lcct~v~% tlut are not

p r ~ n ~ ~ n l y 9L~clarblc i I-lc degree mod~fietier reqmrcbs i t g d a b l e sense to nrodifv. W/ircw i h i x pnln,ilv krthe oi'ttlc dc~jcctlve b riot gradable, itre rnodllier can be p u t t o c l or-k trlrlc Irv sirblectrrlg tlte pnnr"'in/ ~ I I S ~ ~ to moci~~lahor), 50 as to get an

dp/yrty~i i ~ t t 4 v gr,iiiahle :err\e 'I'hus Yi-rtxnc lr' I \ not gr~tl~rldc, brrt one cdrl maLe i t gr~d~iirlc ~ I I T I P U ~ I I 1110dnl~tlf011. by taklrlg the ~~ijectwe to express not the ~ - ~ ~ q v r t v of bruig I rcrlih. but, for exY~~r1p1e, the property ol-h,~v~i~g the fe~tuies type .illy ,asoc rdtetl wrttm I'retic 11 people 'I t i ~ t property 'tdrritts of degree (the triore I xcixc 11 tr,rtrirr.\ 011~ &A, thcr rrlore Y rcrrclr' orre 1s). Note that to be rreric Ir rr l ellr s c c o n d ~ n ~ senre, one doe^ not have to he Eren~h In the pnrnary \crl\e ( )ni3 117Xy, i t t i - c ~ . ~ ~ ~ p l e , he an Amtancaal ,mil shU he 'veq French'.

A xxicrrc cirdrlr~trc t ,lse> dlrc~r5sed by Kerrr~edy ('ritrr Unger), 1s the phrase 'pretty str algXlt7 "Irer-ty' ms a degree n~ucirt-re~, sttrrrlar to the ixlter~s~fier 'very'. i t ( &~II o i ih I I I O C L I ~ V I gr:l d(iitble ~ d j e t t ~ v e . VVherx the .idJective it 1nodrtie4 1s a1

'" Accorciirry t i t licr~ven~sti. (rg74: 127), pcdyscmy u c~?rivcnrwri:ri~zeii n~oduiauot~. Indeed the same ji,itit.i.irs of ~~CIIV.IIUOII LCC ,it work 181 ~x~~tdu~at tor i and (diachn>niraiiy) in polpse~ily (kgg 2005: xiu). Thus ,in exanil~lr t i t piiiwsrn~v which Austrn borrow h n ~ Ani to~ le IS .ti the w n e tur~e a good illustration of iiir,torr?;iiric 'trarlsicr , .4 third ibrin oi'mnciu1.itrori bzsrdes streridrtrri~rlg .~nd liroaderung.

i'iierr :s L lnrnialy tr~rcleiir sense of 'iledtliy': the st-nw in wtlich '11e.iItlty' 1s used of 2 bcalt11~ hod\, Ei is 'cozi~une-ci as a pan' in the other two senses whxch xrvay be 5et out as 'productive of hc:dtiry i xd iec jliealrizy rxernsr] and "resulturg fi.onr a he:dthy irocly' [heulrizy tompfexionl. (Austul 197r. 11)

I.vcn ti~uugli. i n pillvwrti~~us cxprussiotls. the serrse tlut 11s hzerr denved hotrr a pnnlary serue throud, tnt~di i i~t io~i 1i.a i>t.ic coi~v~~ncion~i1i7edd tire detrvauutl I( .it111 dive irr the r~xntis ofthe Iairhwage rrsen, who *tic JW:UTC thx t l i ~ dRi11iti SCV~SCS of &C exprt5stc>zi h r n ~ n firxrrly. That &s~~nguislies p u l y s e n ~ ~ &om Iiortluriy~ri\

Page 78: Recanati, François. Truth-Conditional Pragmatics

absolute adjective of the xnaxiriral type, 'pretty' torcrs the acljcctive to take the secondary, gmdablc serlse. So, in 'pretty sst.raigi!7t7, 'straight' has tllc secoridary sense upproxinautirlg xtrukhtness. Nocv Ungcr (1975) lsas made the fdlowing observation, which nlust be accourlted for. Consider (8) and (9):

(8) The rod IS pretty rtl-sight (9) ?'he rod n straight

A\ expected, (8) does not erltarl (g)." 1 h e rexion why it doec not 1s obvto~rr In

(9) 'straight' is takcn in the pnmxy sence, wl-r~ch rs the dcC~f;rldt. But 111 (8) the inodder 'pretty' coerces the acijectlve Into la se~o~rdary sexlce upprox~r~z~itzt?q stta;q/ztrtes, Exaniple ( 8 ) does not entad (9) bccduse 1 rod c ~ r i approrrnl'ite str~~glltncu, eten to a h~gh degree, wrthorit bemg cictuLrlly \traght. linger's c>bservatlon goeq beyond thxt, however What he note, 1s that (8) dctt~klly entails the llqyirtron of (9). A rod that 1s pretty stralgl~t' ~urirzof be rcrxlght (its

degce on the strarglitiiesc; scale earxrrot be ix~axm~tl) T hrc suggests the follctw- iilg constrau~ts for m andyu? of 'prettv'. An ddlec Qve A ~11~)Cilfied by 'pretty' tntnt expresc J property I; that ~dmlt\ of degree, dnd the corl~ples phr,~re 'prettv A' ascribes a Izyh but rzotzmuzmnl degrre of iL-t icus l 7

I'h~s inkes 'pretty' sri1111,tr to the \o-c~Lled 'priv~t~vr* ac?Jec tltes \ueh as '&Le7 111 'fake dollar' For (ro) dl\(> entculs the neg,ltlori of (I I )

(ro) Ths 1s a b k e dollar

(11) 'flus 1s a dollar

Thls leads us back to the Issue 1 rarsed edrl~er about alleged Sai1x1rc.s of rtght- cubsectrvity . In both 'pretty straight' and 'fake clollar', tlre tlght-subsec tlvlty tnference schema Ai3 -+ B seerns to be violdted

Whert left-subsectivity s~lperfic~dly fails (as In A srn~ll elcpfrarlc IS tlot smJU') I haid thdt th~s 1s because the sense of the xdlectlve sh~fts

" Thls n m contrast to what we find lfirlstcad of 'strruglrt' we use a relativc grtt~lable adjective or an absolute gradable adjrctlvc of 1l1e rmtlirnili type:

( i ) The rod i s pretty longwet (ii) The rod is long/ wet

F-lere (I) elit~iis (ii)

' W o r e prec~sely: it ascnbes a high degree of F-ness, such that there exlsts ;u-2 eve11 lugher degree. Tlzis entzils nonntavin~ality witfmm presupposing that the relevailt scale has a n r ~ x i n ~ ~ u ~ > . (This primp- poa~tion would be ~n~welcomne, since 'pretty' cart, modify re ta~ve gradxbie adjectives. as irt 'thc rod n pretty long'.)

Page 79: Recanati, François. Truth-Conditional Pragmatics

elephant, not smd ,

Because of the sh~ft, tlras is not a genume vlolatlon of the lefi-subsectlvity ts~ference cchern't AF3 + A Smldarly, in the Unger case, we find that nght- iubsectlv~ty only wenis to be violated If we pay attention to the sense

which the adjettlve takes on each of ~ t s occurrences, we find that the ctructure of the example ic as follows:

A rod t h ~ t is pretty stralght, is not stralght,

So tlie nght subsechvity ii6erence sclien~a AB =+ B is not violated elther. It would be v~olated only d the sense of 'strarght' were constant 6-orn one oLcurrerlce to the next; which it IS not.

Can we extend this idea to pnvative adjectives, as m (lo)? Tins n what Barbara Partee (li>rtbcoming) has suggested. Accordmg to her, words like 'real' or 'like' divlde the estenslon of the head noun under a broadened nlterprctatlon. Real arid counterfeit dollars are all 'dollars' In the extended scilse That we rleed to xppe,d to a 'broad' sense of 'dollar' in certrun cases IS

z\t;lbl~shed by exdnlple like.

(12) Is t h ~ s doll~r ieai or Gke! (13) There x e uu tikv-dollx notes m &s box, two ,re red, md four are countert2lt.

tlere, V~rtee p0111b out, 'dollar' canrlot take the smct sense; it must take an extended scme. To be sure, the stnct sense 1s the primary sense, and it is the defiult. So ( r r ) rr ~~ntlcrrtood as follows

'Ibis is a dollar,

where A dollar, is ,I geliulne b~nknote. But m (10) the adjechve forces the sense of tile noun to be broactened. It follows that the apparent hilure of nght- subsectrvity exhib~ted by the example in (10)-(11) is, once +gun, an illuslori due to a sh~ti m the meaning of the ac3echve. If we make the shift exphcit, we get:

A counterkit dollnr, is not a dollar,

I'lie right-sub~ectrvity inference schema AU =+ B is no more vlohted here than ~t W ~ S ~n the 'pretty stralgl1t7 case. Indeed, dwe keep the semantlc value of 'dollar' constmt, the nght-subsectwe inf5rence holds:

A counterfeit dollar, is a doliar,

Page 80: Recanati, François. Truth-Conditional Pragmatics

R!)JECI ' IVES: A C A S E SST'iii.iY 73

O n this account, alleged Eduses of boeb lee-sul~sectjvity and right-subsec- tivity rest on an equivocation; arrd there is no reason to do~tbt the iriter- sectivity of any of the adjectives tfsat carr occur in predicative position.

S. Conclusion

As I argued m Chapter I , sernastlc ile-irbllxty I\ coriipabble w r h compasr-

no~iahty prov~ded compositional~ty dself- ~c ~~rzderstood 111 a suklicrentlv flexible manner. The meaning o i tire whole rs a fixncho~l of the 1rleAiling of the paas, but (I) the rtemulgt ofthe parts are thctr ocrduond mednmgr, (11) the occasional rneanu~~. of word? 1s ,iil>ctcd by context m a w,~)i w I ~ ~ c I L goes beyond lndeslcallty and extends more or less ~rruversally, and ( l r l )

among the contextual factors which shape oecdslonal niezlnlng the lrngtrrrt~c contest plays A central role. Thlr gves LIS A mechamsm t1:llrough whlct~ d ~ e meaning of the parts depends upon the rriear~lng of the whole and of tllc other parts, In a wav that is nevertheless compaat~le wl t t~ coniposr~onahtv

Adjective.; provlde a s t r ~ k ~ n g exar~~ple '1 he t.rtltl>-condrtro~~i oful utter

ance contairung an adjective niodsfi~~lg a rloun depend upon the occamn,il meanings of the ad~ective and r11r rlouxi, mdi the occasiori~l meb~riing ot- either, resulting from sacurat1ot.r ar~tlior mociulatior~, depcnd5 upon dle other words in the sentence the occasional nlc,xmrig of the ad~ecttvz depends upon the noun (2% \\.ell ac orher clerrtei~tr in the sentence, ulclh as, for exantpie, degree asod~firrq), tlie o~c.141una1 rnezlxiliig af the noun ltseif

depends upon the ddje~hvt , and so 012 arid c o forth Ais lrrunedlate corlsecluence of that rrtterdctlorl is t h ~ t dle c\c-cas~r?n,rl

nleanmg of words 1s li~glily chihable one ~~~~~~~~~~~e of a no11 - ,u~l\~~i , . r ro~~\ espressron need not possesss ;she s z ~ i ~ e (oc-cxsronal) nleming '1s mother occurrence of the expresaoxl rn tire senterrcc Urllers case 1s t,rkcn to track the vanom senses which a gveri C Y I _ ~ ~ ~ \ ~ I C ) ~ dvurnef OLI difl'esenc occur rences, the shfi~b11xt-y ofoccaslon,il trtcanir-ig gves the illusrou that there are hrmts to the vahd~t): of standard patterili, of i~?t't.rence In the c'rw o f

adjectives one gets the lrt~presslon tliat some &re not lrltersechve ever1 though they can occur rn predicatxvc posrrltrri That IS argudbly an rilusron, due to the mistaken assumptior> that the t l u t ~ ~ - c o n d r t . ~ o n ~ ~ cont r~hrxt~o~~ of non-ambiguous word niust be the canre on Any occtlrrence of the word rrr

Page 81: Recanati, François. Truth-Conditional Pragmatics

the \c.rrti%rir c* Xt,ither t h , ~ ~ g ~ v e LIP the P ~ I S I L I ~ I C that a11 ad~ectlve'i that can go rai p~cdicativc* povaon are mtc~cect~vc, 1 suggest that we give up that as\unlptrou ,rnJ srvne the rrirethodulo~y irn accordance with Qurxle's In-

~trut troll" ~ r n c licc Ling the v~lreilty of an r r derrxrt e \illenla, ensure that all oc c tirrrxlirs ( i t J g;r\wn wold *ire tdkerr In the s,rn~e iense. m e n that I$ done, the cv~cjcriccl. for rwix intcrsectlvrty m~orrg prmia G ~ l e predlcahve a d j ~ ~ t l v " ~ 1s xiu Eoi~gc-r ( trrrxpclhng

A I,i*,t txarlrlde WIIL serve t o rlltrstrate tlre gctierdl pomt. Conslder a scerirano rrr wlalcfr john, '1 retlrcd ~LI~LJ ICX, beconie A world-farnous sclrrltIS1: dftcr 111s dl\r overy of X Y L . In tlus i c e n ~ n o the fofollowng proposl-

0 1 1 the as\rrrnptBorr 111'tt 'ret~red' n txitcrscctxve, (14) sho~ild be analysed as

e i~unqhsr r t to "Jtrlrrr n r-et~red .u~dJolur is J brrtc her'. Tliat entads that Jolrn n J. butt hcr Iri I o r l j i i r i c trcm wtlr ( I$) , 1701.111 IS rvtircJ 'in4 john w a butcher' ent,~il\ 'J>hn 15 ri-tirrd 'mtl Johrl 14 ,I tx~t~ht ' r a ~ l d Iofln 1s a sc~eritlst', a i d that crit,rils rbcs 1,royositrorl whlclr, o r 1 the unteriectxvltv aswniptlon, 'Jolt11 n a rctlreii \CICII~Z"I' ex~7rtXs,e5, ~iarr~taly the propo\rtlorx t1ia.t John 1s rctrred and chat he n 'i .icI~rrti\t 50, 011 the rriterscctivrty asstirllptron, (rq) and (rs) ought to e r ~ r ~ l i ch 41 Jcrlrr l i s A butcher arrd tk~tt t1c. 1s a ret~red ~ c ~ c n t ~ s t . But nelther of thew enra~lrncrrtr holds. l6ro~lr (14) 3 rd ( k i ) ~t doe\ not tallow thatJohn 1s a

butcher (he r \ no lortgcr one, th(>ugh he was), nor that he 1s a retired sclentlst

(11c I \ onIy .it the i.rcg;iran~ng of hi\ new c dreer) SIXIC~ tlfese erltalments do not Iroid, \rrtntuii9 rs cc>nc~dercii ai, noit tntersectrve Its nun-lntersectlv~q u ~ciountccl for by ,inaly\~ng rl: AS ,,r predicate n~odrfier (11ke 'fornler' or 'p.ieudo7) r,rr ilcr rllan cs a precbc dte OII ~t"iown right, tiespite the Gct that, unllhe ' ~ O ~ X P I C I ' ($I. 'p\t '~tlo', it (drl occtlr 111 " ren~~ig ly pred~cative poslt~on (tjohrn 1s retlrcl"i')

I hxve ~tE,err thas cx.tx~ll;\le bec,rusca ketxed' u ukcn cot1\1dered as a paraQgn of nou ru-rwnect~v~cy I vcn 1x1 that c&e, flowever, nort-mte~ect~vlty rs drgu-

ably ,rrr xllirsrori strnirrlrng Gorn the" xrrearlmg \tu-fts wluch vaflouc t e r n (Yretarcd', .laid 'l~iirc~~cr') undergo Irr tkioe st~temer~ts. m a t fouows IS a

pos51bic <~rrdysrs along those lixlcs F ~ n t , ca>rr\rder 'tetucd' 111 (14) 'rrtlrrd' a ur~clcr\toocl ~n a more speclfic

wnse t h ~ n r r k Irrcral kcxrse 'Clrl5 ts an rnstdrit e trf strenghenixlg The hteral

Page 82: Recanati, François. Truth-Conditional Pragmatics

sell%? of ?retired' (retzred,) 1s F W t N C KF L 11~1 I) 1 ROM ONE'? J O B 7 I i ~ t sen% car]

be nzade rnore s p e ~ i f ~ ~ by spec~fvltlg the 5017 m quebtlc?~~ tn (14) ~t ~ix~derstood thatjohn !tax retired fro~ti i ~ ~ t f i h t ~ t h o ~ d 1 h ~ s \pet 1l.i~- tlriderstnndrng of 'retrred' in (rq) 15 an lnctaxlce ofnrotfu1~tror-r. not s~tu~dtioxi, for ~t IS not rnanddtory. I t I:, qulte poss~blc to use 'rctlred' wrtlloitt espl~c~tly or r~rlplit ~ t l v

5peclfj.mg the job 1x1 cluestron, as 1x1 'John is retired' Even d 'retired' n followed by 3. noun, tills n pou~ble, as 112

(16) John a a rettred Frenchrrran

Pretty clearly, tha doe:, not mean tirat John fi.ir retired horn Fren~Iintanslttp,

hut only t h ~ t Ire 1s a I-renctrrnan who 111'~x retisect from 111s job. 'I lle same sort of readlng is dva~I,ible for (rq), even i f ~ t 1s very hard to m ~ k e s,iller~t because the ct~cngthened I tadlng (rertwil, = KF I I H I I) IWOM BXI I ( HI I ~ T < ) o c ) ) mtpctsec

~tscli-ln such a case. I colrclude t h ~ t In (14) 'rrft~recl' take? the ~pec~t-rc selzse

rettred, through rnodulat~on. So (14) S~OLIIJ be ~ n ~ l y s e d as

John is a retlred, butcher

Now 'butcher' ltcelf 1s undc~utooci 11ere 111 'i 11ivdtil~1t~C1. setlse b u l c / z ~ ? ~

through broadenmg. Solneone 1s a butt her, in th'tt sense, rtT thdt perjorr is or t ~ ~ u s a butcller." In thdt exteridect \elise, lrtitcher:, d~vlcie Into thore kvho are

still actlve and those who lzave retired I his n ,malogo~i\ to the 'ilollar' c,ae d~scussed by Partee A5s~lining such a11 ar~alysls to be corrc-ct, the logcal form of (14) a

FoHow~ng Quine's ~nstn~ctrons, we slrould ctleck the val~drty of inference

patterns .fterre.~hraslxig the selrterlce r o '1s tct track the various senses w h ~ c h a given expresuon may take; so ~t ir tlre ~trzequxvocd (17)~ r ~ t h c r than (14),

whlch we should use as input. Now when we cor-tsldrr (KT), a i l appe'lrances of nor*-~ntersectrvity turn out to be I U L P , ~ ~ ~ 'Ret~red' was sad to be rron- intersective bec'luse (14) and (IS) lack two erlt~ldrneut:, w11ich they should

Partee (fonht-oming) g i v t ~ the following exarnple of a rloun wlrosr terrrpoml exter~sivn vanes Colltextlli&y:

(GI) How many poets are there in Anherst? (iv) How itlany poeu a-e huncd In Alnltcrcr

tn the first statcinent bur not In tlie irrond only I~vlr ig poets arc I" the extinslon of 'porrr'. S~mil.uly, the extension of butcher, n thc ret of <urrrnt burchers, b t ~ i that oi'burtlirr, is thc set uf ( urrcrrt orpiril ht~tcl~ers.

Page 83: Recanati, François. Truth-Conditional Pragmatics

have if 'retrred' was intersect~ve. But are thece e n t h e n t s really mssing? (14)~ analysed AS (IY), doe5 entall that John is a butcher,-so nght-subsec- tlvrty ir preserved. It seems to fail only because ofan equivocation which we

can make euplic~t as follows:

A retlred, butciler, 1c not (or no Ionger) a butcher,

If we keep the sense of 'butcher' constant, the problern disappears. What about the other inference (&om (14) and (I j) to 'Jofin is a retired

scientist')? it is also preserved, appearances notwithstanding: &om (14)- analysed as (17)---and (I j) it follows that John is a retired, scientist, that is, a scientist who bas retired froin butcherhood. It also follows that be is a retired, scientist, that is, a scientist who has retired &om his (unspecified) job. What does not follow is tliat he is a retired, scientist, that is, a scientist who has rc.tired,from his job us a scientist. That reading of 'retired scientist' (where the sense of 'retired' is strengthened by speciwng the relevant job as

being that which corresponds to the head noun) is the most salient one, so we hzve the fkelirrg that it is what 'retired scientist' means; whence the conclusion that (~4) - ( IS) together do not entail that Jobn is a retired scientist. But this is hllacious. There are different possible senses for the phrase 'retired scientist', depending on wllich modulation processes take place; but: the serae that is relevlrrit to evaluating the intersectivity of 'retired' is cdefir~irely not the most salient sense retired3 ~cient is t . '~

o n g ~ n ol'rlus <.liaptcrls a plenary iccture on 'Context-sensrti~~ip: the Case ofAdjectiv .' I gave :n Llsbori r r l Augusr aoo.j fix the Vth European Corlgress of Arlalytic Philosophv, and a rela& talk on 'S.itur:nlon arid Modulv~ol~ In the [~lterpretahorl of Adjectives' I gave at the L,exlcal Pragmatics conference in C:unlherland Lodge 111 Septerni>er zoos. I am indebted to the yanici~ancs at these meet- mgs, and to the p~nrclpants at the 2005-zoo6 Indexicality seminar at Il~stitlrt Jeal-Nicod which w s devoted to the salne toptc. 5ppec1:d thanks arc due to Chris Kennedy, who aroused rl~y interest in dtis Issue during ray vislt to Nortl~westenl Ihdversity in 2004, to Darl Sperber who participated in both the Lexical Pragmatics corrfererlr:e arrd the subseqtlerit ccnunar, and to Paul EgC and Peter Pagin for their comrt~erlts on a first drafi.

Page 84: Recanati, François. Truth-Conditional Pragmatics

Weather Reports

In tlils chapter, f WLU exanune another type ofc-ase 111 which Get- praginatlc

proceses systeniatlcalky ai6et-ec-r rht tnrth corldlhond co~itnbut~ort wllrcll part~culnr lexical Item rn~l\er, or seenis t o mike. The lexical Items 1 w111 hc conccnled with are 111iperso11,LJ weather verbs crrcll a5 'ram' or "snow', anti the aspect of their trutll-conctrt~ori,il cotrrrtbut~orl I tv11l ht o n IS t l ~ e

alleged 'implicit argument' they n r t snld to cnxry I wxll argue tllat there IS,

in fact, no such ~mphclt argumerrt u~ thr Itts~cdl sen~diitics ot weather verbs, and that the 111iphc1t reference LO a Io<at~oii t;>urld in bseattlcr reports without an overt locatlo~tal compr>nCnt IMS '1 purely prAgnratlc I3,isrs

I b F ". I . The unavailability of ?rnciehnite' reaciings 1 for implicit; argun~ents

r . I hilefc.orolugzru1 ytt'dzcate~ the stundizrrl vmv

The received mew about metcorologcal prec%cattc llrke 'rau~' I\ that they c , ~ m an arprnent slot for a locattort whch LAXLX) be filled erdier exphcitly Ity xrlea-i-, oi

an adverbial k\hrase, as ~n (I), ox ul-lpl~crtly by A conteau,ally deterrxmetl locatton, as m (2) T h s m e a r ~ drat A locatlort has to be contextually prowded when none is eqhcitiy menboned

i (I) It's r a m p here/in Pans

t (2) It's ram-.

More prec~sely, the standard mew asunle, that 'rnin', 111 the absex-tce of AII

exphcit location, denlarids that tile tontest ltrovide a ~pecif i~ locatron The possibhty that the sentence 'It tn ~;unrxlg\ni~gflt express 3 2on~tzoit-tre~~e1;ntc~e content is considered as ruled out Thls intri~duces an intercstlrl-g ayrmlletiy

Page 85: Recanati, François. Truth-Conditional Pragmatics

bc>tweerr ~ i l c inrplicit :lrrd the explicit-; for-, on the side of the explicit, we find

rtvo sorts trf uses: the 'rielinice' c.>r >siitgui;rr' c;ws in which a particular ia>c-ar~oii i s i,icxltionecl, as in ( I ) , :~r l t i thc. 'initeii~lite' or 'general' cx!es it1

whit-1-i t1)er.c is cjn;rnrii;c:xtion ot7cr Jocat:iori\, I r l (3a) and (3b):

T'lxere Arc tl~ltt-rerit w:iys of irniilerneotir~g the s~~nt-tnrct view. Sonic theorists lioEci t l u t i r r logical f;)trri 'rairi' is associact.cI with a covert prorloilunal ele-

r-rrer it ---3: lo t2 tinti \~;1~-i~ible--~vl~ic11 ~ 2 1 1 elti~er /)ts s~tt~i~it'etf by means of 311 overt loc;itive plnr;lse, or be lek -t:uns;~tur~tec?, (irr w!~ich c a e the free vxiable rcr.eive.:, '1 \pc.cific V ~ L I ~ fro111 the e..ctr;rli~tg~t~stic. contest). One 111~1y 31s~) treat "ntirr' '1s ,I prt.dii.;itc eslrihiting ~~~ri~i i lc~pc?i~t~z ld in l ) i : tvIxn rlne Location is ex~licit,

~s irr ( I ) , clie prc:tiicate tikes ;r locatiot~ nrgrrneitt; but wlirr-I it is left implicit, as rxi ( L } , the loc~txor~ ns ptut o f t l ~ c i.orlfexk ( r ; l tkr than p a t o f t h e content) and

kr,rir-i' Ci~ni~iorrs :LS arr hrcltr~ic-;il: i t expresscLi; a ilctinite c.irrttcnt only with respect tc, a context c-c"sltrct1 on 3 p:lrt-iz~ilar k~c;rtiori. i'This, in eil'etrt, amounts to tlrc.:~tirrg kr;3irr7 ;LS I pa-tirrze indc.xiiii1, in chc terrrnilloloh.). c-tf-lX ec;lnnti ;loo rc) .

I f ; like Staricy (;?ooo), wc take '~erirl' ro hc ,fisc\ci~t~ii with a covert prcjnonlinjl idcnierrt i r i Xogic:d ii>srrl, w e tvill iwt be sii~"yriscrtl bv tile asynnrjletrq. I mentiorlecl 3130e.e. F(>r p r o ~ i o ~ ~ ~ i s C ~ O have t'lic. prol-rt:l.ty t l ~ i t , itrltcrr urlbowd, t h q nlust be ioiitrxtrr-rib ,issigned ,i spccilic v~ilrre: '1 i c is kdl' c.311 never mean that sorrie 111de or other rs tali. 7'he view tIi;tt 'mili' is a part-time indexical also prcciicts elre jsyrrzrnli*tly: fix i~iilcxicds tlu rcclnlrc ilr;ir tlic cotitext provide a specific valtre hl- tile C ' E ~ I I C C ~ Y ~ ~ I ~ L J rl:il.:1111~ter 011 whi~-11 ~ I I C I ~ S C I X I ~ I K ~ C value tiepcricis.

I . i . C h t ~ i i l 'nd iy qf iariplicir [aTztrvti'r?i.s

'!'tact asyrulri1uil-y hctwcen the explicit , u ~ i tile jrrrplicit whcrl it coriles to tile .~v;rilab?elit): trf iiiciciirritc. reatlirtgr i s 3 vcry gcltcrnl plrcitctrt~enorl. If I say t11:~t

tlri: stool IS o r , the Icfi oT' tllr table, X dci not explicitly rtlentiorl the perspcbitivc ('1s 1 wo~rid if 1 sirid '*I'he s t i ~ o i i s on tile left of the table frorn 111)' pc.r~pr-"c~~vc~.'I), b111 ,i perspective flas to be ccmtextually provided. W-har Bi~s to be ~~rovidcc;! is 3 dt;finitc pcrsyertivc: riiy uttemnce cxr~~tot mean that tire stool is or) ti-rc i e f i of thc trthle fi-otrl sonlr pe~spective o r other. Yet if we rri,lke {-Ire pcuspecrivc explicit, tvc z;trr introduce this 'indefinite' reading:

Page 86: Recanati, François. Truth-Conditional Pragmatics

W ti\ 1 ! ikK I \ r ii0i\ I \ 7L)

noth~ng prevent, me froni sayu~g 'The stool n on the left oi'tfle t ~ b l e from some perspective or other'.

In a justly fixrious paper, U,lrbar,r I'nrtec 'rrgued tltat lrriplicit vmable\, ltke esphc~t variables, run be hountf and cio llot need to he a\slg~ted \pecrfit values (Partee lySg: see ~ l s o Mitcfrell ry%7 and Nunberg rygz) Iloes tlnn contrad~ct my clam r e g a r h ~ g tlie ~ ~ ~ ~ i ~ ~ a l l a h d ~ t y of'mdefuute re,rciings h r inlpli~it argi- merits? I cto riot thlnk so Pastee bvas concen~ed wrth c,ws lrhe (4)-(6), m wWh1c11 the expresslo11 caryrng the rnlpl~clt argunlertt occurs in '4 qu,altificd context

(4) For rnany arab countne5, Arnenca rc tl-ic- rneu-ry [of tIlo\r couritne\]

( 5 ) Wllerevcr I go, it ram\ [tlleru]

(6) Whenever secretdry ni-~de & rrust~ki. tlie others dlli not t_totlce [the ntt5takel

In those cases, the bi~ider is exp l l~~ t , even if tlie b~~iciee ir m~phclt argument. The rnlplicit arprncnt is ovenly qtl'lnnfied. I3ut rxiy c1,lirrl regard- ing the unavtlrlablhty of indefinltc readings coneerris argu*xlel.rts that ,ire 'munphc~t' 111 the (strong) wise that they are neltlzvr overtly tnrntzunril nor clvcrtly

ytrutltlficd. In (I) the location argurneut 15 tsvertly mentlorled (by means of the phr,ue 'hcrc/ln Pan\') 111 (3) and 0) tht. 1oi'~tion arguiment 15 cr~ertly y~rantlfied (by Inearis of the phrase? '\oriie~vhe~e'. '111 all n q o r cit~es', and 'wherever I go') In (L), however, tire l o c a ~ o n argtinreirt wliich the context provldes 1s nelther overtly rner~tior~ed. nor overtjy c~~~xi i t~ l ied Accordlrrg to the standard mew, no indefin~te rendilig I\ possible t%r the implrcit location argunlent m (2) .

In t l ~ u chaptel I will argue that the stand'ird viest IS nctlidlv rnl\taken, AIIC~

thdt a11 ~ndetinite reachng u a v ~ i l ~ b l e f o ~ (2). f%ttt 1 rna~r~td~n t l ~ t implicit arguments, when they exnt, caririot bt. given litdefinite rextmg, (What I w d deny, therefore, is that 'ran' Larries art m-tphc~t loc atxorl ~rgurnent.) Thus conslder the expression ' I t romeVl~en the I-rer\on wtrosr tlonle 1s rn cluestlon 1s not cxphcxtly cpecitied, ~t a corttextually provrdrd qua lrl~plicit argttinent Titius 'Jool~n 1.i home' c,m mean that John i\ at john's home or that he 1q nt clie speaker" ((or rather: the ~pe~tker'c and' hearer's) horne, dependlx~g on the context.' It is also pos.ilble for the implrcit argurncrtr to be overtly bound, ~r in 'Evelybody's honle' on one reading, that means that for dl1 x, x

' As Polly Jdcobson po~ntrd utIr to me, ctle vaiue for the iznyiiclt argunlzot c.in only be either Jolm (the prcfined reading, act:ording to her) or the purr corlsi'hng acthe y7eaker and the liearer. See F~llt~rorc- (1997: (to-0) for ii~scossion of the ct,rrq)lcx L>eil~v~our of 'home'.

Page 87: Recanati, François. Truth-Conditional Pragmatics

0

8@ -&#-IS U N A Y A I & & ~ ~ R + ~ ~ M X U - ~ . ; ~ T - E ' READINGS FOR IMPLiCiT ARGUMENTS D - ---

15 at X'S borne Wiia 1s 1101 pos5ible is to use 'John 1s home' to alelui that he is at sonreotze or otller's Izon;~? S~t ru ldy , 'Jolm is faraway' cannot mean that he IS

far,rway l-i-orn some place or other; the locatlon he is Graway fionr (typically but not riec essanly the place of utterartce) mu5t be contextually spc cified

J q 'lioo type-$ o_fimpllnt syc;runzent2

Accord~ng to FiUniore (1969, rgS6), there are two sorts of irnphcit argu- ments, and the feature 1 have mentronecl (the unava~lab~fity of indefir~ite readings) ~1ia1acten~e.i only one of them. the 'defm~te' intplicit argr~ments 1 he otlier categorv a that of 'ir~definite' implicxt argurnentr intransitive 'eat' is a case lri pourt it carries an mdefinite impllcit argument Thus 'John 1s edting' 1ne'ms t h ~ t be is eatlng sornetl~trzg or other

l here are rrllpl~c~t ~rgmlent5 that dlow for both poss~blhtie, Relatiorla1 rlotxms siic k~ ;ts 'rnother', 'L~ther', and 'husband', in the absence of an expliclt coniplen~ent, can be uncterstood e~ther way In (7) the implicit argument of 'rnother' 15 ctefirlrte, wh~le ln (8) ~t i r mdefirute.

(7) A baby s t~r te t I cnii~lg 772~ mother went to corrifort rt

(8) 'I he toyscurc* ~ ~ 3 5 t'UU ofparerits I n search of g~ilks. 7'he motken were espec~allv

ilitervstrd ni edu~atronal games

S~nularly, even though 'local' w often (Indeed, Jbvays) heated 4s requirlsig a

spec rfic trnpli~~t argument wheri unbound, a\ ~t does in ( c ~ j , it alro accepts rndrfinlte reati~ngs, tu ul (10):'

(1)) I spent the >umijer vacatioti In N i ~ e and I enjoyed reahng + e local

ocrwqtaper eve rvd ,~~

( 1 o) Mary t ollecrs locd newspapers

I do not mttxnd to deny the facts on whch F h o r e ' s disanctlon 15 based- there are ~r~deed two type5 of case, as he pomts out. But I use 'mphc~t ~rf iun~et~t ' 111 such a way that onlv one of the two types deserve$ that name My reason for so dorng w that, whenever an alleged mphcit argument can be understood uidefui~tely, an altendhve malysls is avdable, whch cfispenses w~th inlphcit argrnients altogether. 'Thus l~itransltive 'eat' arguably denotes a propcq , whch we car] define by exlstentldly quantl@mg one argiment of the

I owe this obsewation to C)rin Percus, who gave a11 example like (to) in a &scussioxr dur~ng the zoo4 Milan Meeung orr coven variables.

Page 88: Recanati, François. Truth-Conditional Pragmatics

two-place relatlon denoted by trdrlsitive 'eat'. That IS, in eEect, what Qn~rie'c

Der oper~tor does (Cfuine 1960 239 31). applred to any n-place predrc ate, re yel& a n-'-place predicate by exlstenhdiy qu,ma$ng the last xgunlent-rc>le

(Deu P) x, . . . xh.,-l iE tllere is sorrlething x,, suc:h that P;v, . . . :Y,,

If '17' is a two-place predicate, hlie trarlc~t~vc 'eat', tller~ 'l)cr I)' will bc a

genume one-place p red~c~ tc , deilot~r~g a propertv rather thar~ a relatrtrn O n t h ~ s mew, uitransltlve 'e'lt' does ~ o t deiiote a r e l~aon betwceri the edter anti

an 'imphclt argument' (tile food).3 $mularly, rn thls frail~ework, "mother9 will be construed ac a polysetnous predic~tc, denotlng ettlz~r A two place relation (whose cecond .irgumtr?t (-An bc exphc~t or ~ n i p l i c ~ t ) t?v a propert\ that can be defined by applying Q U I I I ~ ' ~ DPI operator to the homo~~hr>nli two-place pred~cate No ~mplicrt argunlelrt rs involved oil the '[3ropc>rt\"

readmg. Likewise, 'local' i r i ( r o j wlll be treated ~b denoting A propertv on ,l par with the propertlec expressed by "regtorral', \nahorixl', or 'ix~ternatxonai'

2. Implicit argu111~ll ts vs free e11Ticllni~eilt

L. 1 . 'i 'he ~ueatlirrvntzn txcltnpli2

As I have sad, 1 thvlk axi ~ n c k t i ~ ~ ~ t c mte~ret3tlcrn is , ~ v ~ ~ ~ l ~ b l e i o ~ A rrrrrpie mcteorologcal selltenre lAe (a), 221 ~riterpretat~on whcli rn'lkes 11 ccjtr~valet~t to 'It'r rairurlg somewhere' fn X < C C ~ ~ (2002) 1 gdve the followmg exd~zlpXe

I can Irnagne 2 ~lf l ldtlntl 111 w h ~ h lain become extremelv rare 'arid nupoltnnt,

and ran detectors liave been d ~ s p ( ~ w d all over tile tenttory (whatever the terrlton--

possibly the vhole Earth) In the 111lagtled \cennno, each detr~ror tnf:gel\ at1 damn bell in the Morutor~r~g tinoin when 1t detects r a n There rs srxlglc

bell; the locahon of the tnggenng detector 1s mcbcated by a hght on 4 bo~rd rrr the Monltonng Koorn After weeks of total drought, the bell evcntudllv angl rn thC

Of course, it is possible to ~ ~ l l r > w ~ n that intraisitive 'eat' denota a reiatlon betweerr the eatpr and the food, the dsermce between the ir~transitrve a i d the transitive use being that on the lnaarlritivr LIS~, the: second argument of the relation undergoes exrstential ge~~eralimtion. But die 'property' ,mdysis has an advantage over tlle 'existetltial generaIi~a;rr~on' analysis: it correctly predicrs tlxat the eslsterttiai quantifier a n only take narrow scope jsrslce it 1s birilt into the kxicd envy of the verb on it$ irrtransicixrr: reading). On the 'existential genzrdizarlon' aoaiysm, the obligatory narrow scope of the existerltid quantifier has to be stipulated. Sce Section 5.2 for snore on this Issue.

Page 89: Recanati, François. Truth-Conditional Pragmatics

8.2 Ih\l?'r,ii:I? hfiC;Ub1Et4TS iS FREE- ENKICi-iMEi.j'ii

Moi~iccr~.rirg lioiirli, i le:rring it, elre weatlicrmari on duty in the :~.cljacent room shouts:

'le's runaiigi' 1 Lu iltt~rairc-e i s true, iifk i s ~'~nzing (at. the drrrc. of ~ltternnce) in some

pkici. or orircr. (I.?ec.anrrti 2002: 317)

I t,rhc 11x1.. L ~ U J I I I ~ I L to cdst doubt oil thc stand'~rt3 k1e.v~. ficcord~ng to which '1.,1111' carlner, i n . ~ ~ g u t t l c ~ t t t slot ibr A I O L ' I ~ L ~ I ) . "1x1 the has15 of that type of' coirr1ci~rr~.\itn1pIc~, I Iwvt. put ii,rward '1x1 .ilterriatlve p r o p o d regarchng 'ram'

aid other rlrctc.orologcd pred~c~ttcr The g~rop(>s,il 112s two s~des:

e (hi i l r c~ <,crrl,rrrt~i- s~cfc, 'ram' 15 tt-e,lteci 3s ,r lerct-plnt e predicate (just as

iriEr,irltrCaL cz "e,rt' 15 orrc phcc precbcate m d tr Anslave 'eat' a two-&?lace pru"iic ,rte) No loc,+tron ,rrginterst rs rrn.otved rn the argun~ent structure

of the pr ctirc ,rtc

a ()i1 the p i ~ p ~ ~ ~ t f i - d c , '3 proccss ot tree errrrcllnlerlt (otten) take? plarc, r r i vlr tire o f G V ~ I I L ~ ~ the nle'til~tlg of arr rltter;ince irlvolving the 'ran'

p r ~ ~ ( l l 'itc I\ rri,ide corltcwt~l~~Uy nrore specrkic tha l~ the renldntlc content 6 t ( l j l n ~ i l r ~ ~ c ! I?v rhe Ilterd 111~;1111ng o/- the S C I ~ ~ ~ K ~ C C . More precisely,

tlrrougl) tlidt: pro< esuof 6-ee exlr~chrrte~rt t i le rllcanlng oftlle utterance 1s

n1,1cich rtlon ipec~fic, ever1 thotiglj tire sentence ttself ~nvolves no ( o b t ~ r 01 ( overt) reference to a yl,rce

z..?. ,Tj~ellir<q o t t l f h ~ ~)rol)t).s~tE

WIlzri L fay tha t * I - : L ~ ' is 3 zero--place preciic.:atc, I do not mean to rule out a I.~nvitistrr?ir,an arulysis, tllat i s , an a11;ilysis ,licorc!ir~g ta wlticEt 'action verbs' (as I );iviJson calls thein) take ;tn extra event- argurnent in addition to their 'ctanc/'lrii' arg:iirlrcXrrt.;. Wllat I nlenrt, r-:itlxr:r, is that nietcorolo@cai verbs take no othzr iiv2rtninit lhrin {he event ~zcq~imcfzt. And h-y this I do not meail tlut the stai~dartl nt-gurricrits oi'tlle verb are not vt>trl/y argurrrents but are introduced irldirectly ilr logicrc~l ibrm via their relatiorls to the event arglinlent (the only 'irgrrrrient clr t verb really takes). This is the knzu-Llaviclsonian' analysis. g~tcordirig to whiili 3 sentence like 'Mary dar1c.e~'----which I wiU pr-etenct to bc rc.rmszless" -- has the fitllotvirrg lcrgicnl k>r-rrl:

~ L ) l ' c i i i i r ~ c . tiic'rc Jrr L+,IVS ~>i'deicndirig the s t an ihd vrc-iv irr the i i ce o i such es.mrpIes. They Mnll be iiiiroiiui ed. arid diu-riueil, III Sccnvns );A.

' Icsrres jiriwinixig to tens(. aiid aspect ~ 1 1 1 he kept mdr: 1x1 this c.hapter. utiless they have a direct bcanrig 4)n tlrc tiisclrssioii. So, firr cxnrriple, 1 will systctnaocaliy Ignore the sernantlc contnbation of tile prtigressrve in I t 14 r.1111ir1g' (dnd I will often ig~~ort: die coritnhvrtri>n oithe present tetlsc). Of course, a w,illv ttroniilgii i.tli,rt to irivrsngste the scriisnncs oi'we.atiler ceritc:xiies \\rouid habe to t&e bott~ tense .iiid ~ $ / ) i ( - t i i i i i ) .I(.( o i ~ r l l . ~ ) L I ( rrly d r ~ ~ b i t i o r i 1ri t l i i s ch3pie1- is (iei~iierateiv iimicd.

---*,--- - -

Page 90: Recanati, François. Truth-Conditional Pragmatics

3 r [DANCING (e) A AC,PNI (M'lry, tv ) ]

Here, the agent, Mary, ts nttroduced rrrtilrectly, a\ heanrlg the rc!,it~ott 'AC,ENY'

to die event e of danctng wh~ch w the orily ~frrect argurncrit of'dcf,)nce' (1 e the ordy argument of the pred~cdte bun~c,~' whrch the verb 'd,ulce' contnbures to logclil forn~) 1 wdl nlyself use thc neo Uavtdson~~ul f ia~~iewolk 1x1 whdt follo.ivs, b ~ ~ t when 1 say that 'r,lrn' n a ~cro-pLxe predrcatc., wl-tat I rrneatl 1s not that tt hag no other dzrect drgumetrt thai the event .krgtxnent, vrti e that n the ~ a e for d l actlort verbr or1 the neo-l)avrdsonr~n nn~lysw What I rnem, r'ktf~er, 1s that it h x ?to '~tl~ntfa?d' aztlmenf t r r die firsf plar r In contr,~st to 'dmce', wlucfi has a st,uldard agent argLurrelit (rntrodnt eti rttdrrec tly 1x1 lopcd form, ,is

m the dbove repre\entntlon), 'r'un' has 110 &gent or thel-tlc- -the pronoun "t' IS

A d~mxny subject, with a syr~ t~~chc role birt no \enlcLntic contrrbiltron (TesniGre

1959: 239-40) 'Ran' ir /era-place, whrfe intransltlve vertts Itkc 'cf'gr~cc' ,ire

one-place. transitwe verb5 lrke 'klbs' are bvo-place, and ditrx~srtlve vcrbs hkc 'gve' alre three -place.

Slnce 'ram' hru 7ero argurnent, on m y accotint, ~t does not ltnve '3 l o c ~ t ~ o n argu~nent. That is not to deny that .\onre verbs 1l:ive 'I lacatron ,Lrgurnent. 'Arnve' IS a case ~n yomt. The logcal hrrxl of (tenscle\s) 701111 amve5' IS

sometlung hke:

3 e [ARKIVING (e) A A G ~ N ~ (jolm, (.) A L0Cr);lION (I, e)]

The locat~on of an x r~va l event is the goal lacat~on--the dcstm3tlc>n--t>f the nwnon whrc11 culnunates tn that evenr 711,~t loc,rtior~ need r t o t be overtly specltied, but at le,lst tt mu\t be con t ex t~~~~ l ly untlentocld ' T ~ L I s , 111

tile f'ullowmg ciialogue, A's secorrc-t Euternent 15 tnfel~c~tou\.

( 11) A: John has arrived.

R: Where has he a~nved l

A: *I Iuve no idea.

Tfnc is the bas~c test I use for deterrrunrng whether there 1s an argtrrncnt slot u~ the lex~cal senianttc~ of a v r r h r f tltere w one, the slot has t o be (ille~i,

" There are spec~d uses of.'rrrin' wl~ere i t ukec .I cuniplernenr, as In 'it tarns catti atid clogs' (or 'rr, rams liogs'). I an1 nor concenied w ~ t h chore uses here-- even tt>ougt~ I w ~ l l occa\ion?illy rnc11t1011 thcnl s e.uc~nipiec whcri the s t ~ t t t % ~ i t t ~ e ~ c ~ i t ~ p l e ~ i l e ~ ~ t IS r i o t wtiu 15 a i i s ~ c

Page 91: Recanati, François. Truth-Conditional Pragmatics

whettier exphc.idy or contextually. (In the above logcal fom, the slot is reprc\ented b y the free variable I , to wluch a v:due must be co~~textually

assrgtled.)

In cootrayt to 'drr~ve', whlch cames an argument slot for a lot atlon, 'darrce' does [lot, as the following d~alogue shows.

(rz) A- Johri bas danced.

13: Where h'cs lie danced?

A: 1 llave no tdea.

'I'lre lotar1or1 of the ctanclng event rieed not be specified, expircltly or even contextuallv 'io 'dance' does not carry an ,ugument slot for a location. Whet1 the locl t~o~j of the dat~c~rig event 1s espf~cidy prov~ded, ~t has the

stmi\ of 'adjullct'. S L I ~ C C it I\ semantic~~llv optional As 1 Ieanlt from a re&ree for I ,~ t { yu t~ t i r$ urzd I'i~llosophy, this way ofdrawing

the dl\t~nc tior] 'iletwecrr arguments 3 r d adjuncts is due to the Prague school,

and spec~lic~Uy to fanrirla Panevova {see Sg& rt al. 1986).~ I have r~lyselfput forward a sirxlilar test (which 1 call tlie 'Opt~onallty Cntenon') 111 order to dl.itmguistt bcrweeri the corltcxtual plvvislon of An implicit argument arid the pragmatic process of 'tree eimchnieet'. Free ennchnient as I construe it rs J pragndtl< process through fivh~c h the actual Interprctatlon of 'm utter-

atrcc 1s lir.~dr ~ontexttiaIly r11~)re specific than the l~teral nieaning of the uttered sentence Whde iiidemcal resolution or the assigiirnent of contex- tual values to tree vanables 1s a 'bottorn up' process mnandatecf, by the Imgutstxt matend, free ertncfviient IS a 'top-down' process ivhich IS not mciridated bv the lii~~wistic matend-~t 1s not mandatory, horn a 111 yuistlc polrrt of view---but takes place for purely pragmatic reasons, that is, 1 t order to nuke sense ofwhat the speaker IS say~ng The speaker utters a sentence

which Etas a ccrtaxt xneanlng, but, owlng to hee ennchment, the utterance is

undentood (md expected to be understood) In a more specific rense than its hteral memxxy hcence\. The gap between .ientence rnedning and utterance meanrng IS badged by world knowledge and contextud expectations For

exxnple, 'rabba' ill 'Mary hkes to wear rabbit' IS understood In the (specific)

' Actually Panenova's classificatiori is cornplex 2nd reiies on a double distinction: beetween free adverbials/n~odifiers arid inner participma, and, within the clm ofmodifiefs, between those that are 'valeticy members' (a feature which 111;ikes them similar to inner paaicipana) and those that are not. Vdency nierrihers themselves may be obligatory or optional with a particular lexical Item. See Panevova (2001).

Page 92: Recanati, François. Truth-Conditional Pragmatics

sense of rabbrt fur, even though 'rabblt'qylku nlas\ temz llterdly nxearls somethmg more general Ilke fizbhzt .stti8 The recacork ~vlly 'rabbit' UI 'Mary hkes to wear rabbit' is mterpreted differentljr tharl ~t rs In, for es'inlple, 'Mary hkes to eat rabbit' IS a matter ot world knowledge and contextual expec- tations. W e know that people eat rabbit rneat, whlle they use rabhrt fur for thelr clothes, and we expect the spe'ilner to say plausible tbzngs

The h h a r k of the pragmatic plocess of Gree ennclzrrzcnt, on rxy acl-

count, is that ~t 1s opaon,d it rnxy or may inot take place, depenrilng on the contest. Nothlng prevents ‘rabbit', y r d i ~ m~cc term, from be~ng intei-pr eted rn

the general sense rubbzt stuff("ttfter the ,~ccident, there LVA\ r'xbbrt '111 over the highway'). St&, In many context&, it ~ m l l be 1nteq3reted mole S I ) ~ C I ~ I L Z I I I ~ owlng to free ennchlnent Sinularly lor \ddnce7. ar (12) shows, a 'dan~e'- statement may be understood m A locat1or1-lilcleiinite rslarliler SLIU, a

locatlon for the event m,jy also he coi~textu;llly provided a\ plrt of the Interpretation of the utterance, ac iri (13)

(13) A: Was John present at the ball? B: Yes. I-Ie danced 1111 rt~glit.

Here B's statement 'Fie danced ;ill night-' me,rns tl-i,~tJohn danc etl all rlrght at

the ball The contrAst betweerr (12) and (13) shu.cvs tllxt the iontextual specification of the locat~on nfthe dltrrcnlg e3errt Ir option,il, ~ n d thrs feature dlstmp~shes t h ~ t speclficatic)n in (13) frorn tlte coi~textu,~l provl\rt>rr t,t an lmpllclt argunlent. The context~ral prcrvl\lon of an rrz~pl~c~t argtlnrent r i

urzand~ziltory process, since ~t rs requirtad In \ilrtue of tlie cenrantlc\ OX rorne expremon in the sentence.

Armed with these dlstmctiorls, let. i r i return to 'r,un' ar~rX ri~eteorologcal premcates. It ceer~~s that "raiii' patterrlr with '~rnve ' and ~nvol\.e\ A location argument. If someone tells me ' l t ' ~ ralnmgkand I ask 'TVl~ere?~, lt wtr~xld be infelicltom for nzy conversational pdtt13er t ~ ) reply 'I have no ~cie,~' Thrc n the intcllhve bass for the stas~dard V I ~ W regarding ' ran ' arid meteorologcah predlcates. But the \ i tuat io~~ ir not A\ rrrnple as t b ~ t . In the weathenriarl scenario, we can have tlie full~~tvrrig ilr;tlogue

The 'rabbit' example is discussed by Nvmbcrg ,\rrd Zaenetr (r9g.z) who I>ort.uw it l b r r i Cilpesiakc: and Briscoc (1992).

Page 93: Recanati, François. Truth-Conditional Pragmatics

(14) A (the cvi~;,tircnnaxi): Xt rs raining!

8: Wllcre?

A: 1 liave xlo rait:;t---let's check.

i prec ~sciv ( onjrtut red the we\fathilnl~'~x~ cxd~riple 1x1 ortier to show that, ~ p y e,tr an i c.\ rrotw~th\t~nclrrrg, the c c~rltextu~llv spec~tled locatron 1s not a gcxniilnc rrripliclr cirgl:rmre~lt m dre 2-an' c,iw It 15 tlot A genutne zrnpl~clt .~rgil~iie~it IXT au\e, tf rt u)rrtJ, zt i i~cx~ld huz~r TO be provldrd lt l every context, iviiludarzq /hc, n)rrfr I! oftlze t~~e~ith(,ntzatl cr~r~myle (where no locatlox1 is actually \ptxc~iied) Irr X<ci,lnatr zoo^) 1 used tile wtatllern~~in example to argue that the lot cktlon rjf salrl, .rvl-reu t ontextrially \peclfiecl, rs specified through a pnx et".\ O ~ L I ~ I T X ~ ~ ~ I C erlrtctrrnen t , 11 ptotr\j f1'ztit m z y take plilcc En conttectlon with iiiry V L ~ L I Z I ~ M ~ I T ( I I P ~i~hitf~ocuer. ~ ~ ~ ~ l r i d l l l g ' ~ l ~ ~ t l i e ' (+I\ 111 (13)). and which casts

no dl\trr*~. trve Ilgl-lt or1 the 4erllantlcs of ~neteurologicki predic'ites ns opposed to othtr cvtXrat prctirc ate5

Chic rriay objcc t t l ~ i t tilere i~ J d~l&rc.r;ic c between "ram' dttd 'hnce ' as far as lc t i P"LOXI\ ,Ire coriicrrxcd W i t l r '&tnce', rt's ~ " K V to conle up with exa~~iples In vclricE~ r l r r l a t mori doe\ not nl<ittc,r G~r~cl rs lcfi ~insyec~hed. No so wit11 'ran':

~t L , ~ ~ c " s work to (onstruct an examplc lrbe the wedtherrnan e ~ a n i p l e . ~ T'lxs ilrilercnc w rrilrst bc dccocrr~ted for

1 A ~ I P C 5~111, 1 xr*arrltalll, the slreer puss~bdrty of the weatiien~ian example S~IC>W~ 111~1t the yrovnron of,% locatlori for tlie r'irrlrtig event is not mandatory (scn~,irrt~c~~lly trigg;crccl), but upt~c~nkl (pragxn~trcally triggered). If that is ngPtr, tfrc-sxr thib cilrSti.rence. l)etween 'rrnlo'aild "\fatice' will ~uel f have to be e*pl.trxred o r r a prcrgl:r-rlatlc. It<rsls. S ~ t c 11 an expian,ctiott can proceed ~ O I I ~

tile fi>foElowi~rg 11xrc.s rn gener~l, we c are dt-ro~rt ~~reteor ologcal events to the extent ~!r,rt wc* < dre dbout the locatrons wllcrc they take place. ?'he exrstence 01-3 ra~lnlng e v ~ n t J I P P S P (AS (oppwed to the esrstex1c.e of ,r r,Lmlng event nl such uwd iuih LI: ~ I ~ I C , ) 1\ tldrclly ofsutiic~rrrt mtibrest to he woah mentionmg. 'Che \amc thrrrg doc.r ]lor lroltl fez the otller typcr of cvcrlt. It rliay be mterestmg

ilic f a c t that ~t takes w ~ r k 10 LC~IISLTUC~ t.hc weatliertli;li3 cxarxrplc su.ggests that, by invesung as ~nucti work ill iire (.oiisLrtxction of' ~verrti areriarios, one might prrlwp.i ctnnc up will> examples in wlucb predrcatec ~viric!i (accorciir~~r e c i me) do carry a covert rlgurr~rnt x e IICVCTTI~~I~SS gvei l a11 indrfirute iradiiig I~i i~: , t~l>jci. i i<>i~ wiii be cprllrd our i i i Sectxoxl o .:

Page 94: Recanati, François. Truth-Conditional Pragmatics

to hear that Jobrl kissed Mary e v e ~ though we are not told wilere that event I~appened

To check that tlm explanatron n correct, we only have to Imdgne .I

meteorolog~cal event hkely to arouie o11e75 interest qnite lxldrpe~~derit of 1t.s location. The prdgIlXhc account I favour pred~cts thtt satelncut descnb mg such a, event wodct not necessarily be tirxder>tood 'I\ lo( 'iaort specific. The follow~ng 1s an example.

(15) Ortce, m Antiquity, ~t rznrd blood. Since then, no 5ucli tluxlg ever hjippened agar]

The precfict~or~ seems to be borne out. In (151, the statement ' ~ t r'tlned blood' 15 under$tood in a location-indefinite nianncr. (:ompare tfiris with art 'ordinary' type of example hke (16).

(16) Yesterday, it rallied

In contrast to ( I j), (16) will-typically-be Interpreted I n 2 loc.it~on-cpeclfic

rnanner But thls coi~trast c:m be n~aniy~~lated by rx~~rrl~pttlatl~~g the context, a 1 d ~ d m tonshuctlng the weatherman example AH thi+ iugge\ts t h t w'ilatevcr ~mphcit location-pec we flsrcl Irr 'ram7-\tatrlncnts 1s cf~le

to pragmanc factors

3. Four levels of analysis

3.1. 'lhc metaphysical lvvel

According to my proposal, the sentence 'It 1s ra~ning' rrieans soilretlung very snnple and vei J close to the surfdce &at a rarn everit is oc c rtrnrig. No pl~cc

15 mentioned by the sentence as the place where thdt eveltt occurs <>I: coune there rnust be such a place: ~ 1 1 event call ordy occur at a particular place (and A pna~cular m e ) This n a 111e~physic a1 hct about events, tvhicb holds wliether the event is a ralnlng event or a dartclng event or a krss~ng event Those event-types d~Eer In xnmy re\pects, mci one of the\e d~ffer- ences is reflected m the difference between the arguxnrnt stntchrres of tllc correspondrng pred~cates 'ram' is a i.ero-place predicate, 'd~r~c e' 14 a one-

p la~e pred~cate, and 'kns' 1s a two-place preilicate. 7 he plxe anci tlrrle of the described event do not autoniat~c,dly count as argzments of the predlc~te,

Page 95: Recanati, François. Truth-Conditional Pragmatics
Page 96: Recanati, François. Truth-Conditional Pragmatics
Page 97: Recanati, François. Truth-Conditional Pragmatics

oc i rrrnrq; \ 11xr j7lLtct~ 1s L.LI\CII tale ot by the ri1e~iplly51~5, and. does not have

to J P ~ I I I P " 111 the \rt"iz~~nt~( \ . I 1

Notr tllar wliat Nealc \.qs ~borl t 'r,rru' utru1c.l apply to 'd,ince2 ,o well, or to

JIIV otlicr i c t l o r t vt'rb It 'the iliiestron wlrcti~rr r t n r2112mg (now) prrnkt has no awwra. bi.cari\i. rt rr r'ccat J gerkulne yue\tlon7, what alw~lt the questtori wllether M'rrv ri iiarrcix,g (rrow) puttkt2 Is dri\ a gelltllxie cluest~on' As In the other case, we c ,lrr n~al.,c scrlw of the questlor1 orily ri w c xrrterprct ~t AS asiung whether NXLV I\ d = i ~ ~ r - ~ r i g ~ t vrmz past~cnl~rrpl~cc, or GI\ .~\kmi;: whether she tr tlmcing in

wrrrr ~LIJ.LC or otlat'r l>oes ~t follo\w tllat tllc rr.l,-tt~on exl?reueJ by 'dance' rr~colvc*s arr ,rrgurixztlr role tor n lo~ahori: No, be( ailse, agan, the locatlon 1s taken c x i : u f b y the mer;~pllvslcr I 'vc~y rverlt ~ i h e s place sorne\vhere- ttzat I\

tlae reA\oxl wlrv ct r c'ni irlfkr % I x y daricrtl \orrr?cwilere7 fronl Mary d~nted ' , cwrr ti-xowgb ' M ~ f v tiulccci' bays rlutllir~g A*i,o~rt pIwe5 5trawrsoa rnade that

pomt to] c ~ ~ u ~ I c IIIXI~Y yenn go. I Ie wtotz

C ' h r yrabp oi- tiic JCC th3t these qu;~tltii-ien [c.g. 'ssoinewhere'j can be added to . . . sinrldlc :rscnprloi.is of iluxxlaii ,li:tior~s witlrout ir~odificat~oii of tmth-value rests on i~otiri i lg i~ lo i t : ret.o,ltii~e t lun trLrr g;i.sp ~ C t h t " ger-xertri coticept oi';rcrion. (Strawson

10'17: 74)

i nkv I)av~il\ori, Ztxa.vvsor~ spe'ihs ot ' ~ c t r t r ~ a ' Ilut I think the polnt extends

to c ~ c x l a i r i gericral I he 'ger~er'al tomccpt' t r l ,111 event IS the con( ept- of sojnc.tluxrg that )lappens, ~ n d tvl-latirevrr Ir,ippen\ Ilaypcxss at a place arid A

tlnjrc I bcii. rrx,ry bc dif-hcrrlties 1x1 ~dentrtyrrig tlie tune or place olan event d the t~nic. or ~>l,ice r r l qrtestron h~pperls to be corrrplex or disconnected. Wihat i.s the C W C ~ tlmr ,tl~d pl~(-e for t i r r event ot cucunmavlgatlng the gjohe? 1s

re6\l!jy 1x1 everlij IS c o l l e c t ~ o ~ ~ m f C V L X I ~ C everzt-" What ' h u t the Fr cxlc h li cvoltitiotr; All wit:, ol questlorn ,m\e In t-111> area, but I WIU not p ~ u s c to < I)~ISICL~I. thct~l, as they 'rre rrrclcvarrt to the maul Iscue at stake

I'1kena irrav rho ht2 rr-iore p~ob le r~ l~ t r c c&es lu \vhlc!a we use d ~ h ( P f 1 verbs to talk ~ t ~ o i l t ttrrixgj t l - i~t ate not rc~Uy evtants arid therehre 130 not saas& the

xuet,iphv\ic coilstr;n~?ts on events. In \LIC I I a ( ASC event tiilk 15 a n~ercJu~otr de

/ j ' i iic i~irnc rrnsoriing appixe~ tcl tht.inl;lrrnct f ron~ 'Sue 1s 'I good (1:rnccr' to ‘'1'ht:r.e IS a way 111 which

Sue IS g<m(d. t'oitan Siatx'i,i, :irir;ues that, to at courit Ibr 1111s ~nt;.rrrrc', we t~ced to p o w o vnriable .R' fur a 'way ofherrig goil(i' irr tile Ic1~1c-al fbrtn oFCSut- w a go:td d:rrlccr' (Srabb ~ o u l : 132 - 5 ) . B u t we don't need iirch A v:inahit: rri log(..rl fi)m~, 'it Iri~st if'we t : i k ~ .I<)gc-aJ &'rix?' in t i le synt:rc:txc sense. -The :Jleged .way of

be~ng g<)<~d' it presirrrr'~iii.r- -taker) care ofby d-rr r ~ l r e p h y r i c s ofgociciries and dues not have to fib=re in t i le se:.inaircrc ... 5ec ('appc:lerr :anti I epore zooja (axtii Uorg ;rtioj) fix rciated points about the 5emaritlcs iriet.ipii)sii i iiislrr~ctii~rr. wiri(,lr tliry iirli~>rnrxlatciy irverir\r

Page 98: Recanati, François. Truth-Conditional Pragmatics

pavlcr: we tdk of soxiie thmgs that Are not events .as ifthey were event\, and

tbore tl-~mgs, mso*ar as they ,Ire not gerluine event\, ~ e e d not h'tve a tlrne or a location coorct~nate. 1 wll ledve that Issue a ide because 1 ~ ~ k e ~t AS O ~ V ~ O L I S

that metcorolog~cal events are g e ~ ~ u ~ n e events j 2

j. r. The pru~i~zat ir level

In a cae hke ' c h ~ e ' , the 1ocdhr)n ofthe event tnay bc contextually undentood

even tkiough t h s IS not irnpo.iec1 by t l~e a r ~ ~ n ~ e n t structure oftbe verb. T t i ~ t llappens whenever the location of the event s so relevmt to tllc convcrsatronal

purposes at h a d that ~t carmot be left out of the p~ctrire Tllus m (13) thc

cor~venahonal j~artr~en nrc &mg about the bdl, md 111 that contest I)':, utterance of 'Fie ct~nced n~ght' 1~ 11ahlrdIy uildentood .IS sayi11j: t l ~ ~ t Jolin heed all m&t at tlze bull. Ttlis 15 kee ennclinieiit tllere 15 tlo lulgu~stlc reference to dle ball, whether overt ur covert, m the sentellce 7 Ie &11ccd a.U ruglt', yet the

speaker ~ c i d y refen to the b,d (dnd s urtclerstood as so referring).)." r ? I he pragmatic enncllrrlent refultlng from the tncrt refererice to A l o c ~ t ~ o n

can be reprecented AE; the contextual piovl\ion of an extra conjuncr, e g

'' One last issue worth corrs~dcniig 1s the dstriictlon hetween evenu and states w~diili the overall category of eventualitlrs. evzntualitlrs. t ad tliat the possesion of spano-temporal location m a gerrrral char:uteristic of events, rather t,han a distlnct~ve propttav that son~e events have and otliers lack; arid that was orie ofthe reasons for not considering die locauon as autonratically pan of the ;ugunxrit stntcture of event verbs. But what is true of events is not uue ofstatts. States need riot be Located in the sense in wbirh everm r~red to be located So if we work with a unified rxegory of eventuality (as sortie peopie do, who use tlx variable 'P' as an evetituahty variable rather than specifically as an event vanable) tltcn we irxy have a reason for it~troducing locatiors In the logical fonn of event sentrr~cea. A verb v denoung an event of ~ y p e CJ will be diyllnguislred Germ n verb v1 denoting 21 state of type Y' by the fact that the verb's contributioi~ is bound to be a located evmtuality only in the forn~er case:

[[v]J = Xe [3 f ?l [TIME ( t , e) A 1.o~-ATION ( I , e) A @ e / J

[[dl] = Xe 1 @el

In that sort of framework, we could roll represe~~t the ditii-rence betweeri eveilt-verbs which, like 'amve' or 'le:~ve', do carry an argument slot for a location. and tl~ose whicli do noc. For tliose wtuch do carry an argument slot, we would sin~ply add a conjunct ' 1 = 1". where 'y' is a 6ee variable to wluch A vdue must be ~ontextualiy assigred (unless it is bound htglier up in the sentence). 'I'hus the lextcal errtry for 'amve' would be:

h- Xy he [3t 31 ~ I M E (t, t') A I,O(>YIION (I, e) A ARRIVING (4 A AGENT (x, e) A 1 -- y ] ]

In this chapter, h o w m r , 1 assume tbat 'e' is specifically ;UI event variable, w the introductroxi of umws arld places iri the Iogicd fomi of everit sentences n not necessary (udess they are conn~bi~ted ai sotlit. level other thdri the metaphysical)

' T h e r e are various ways of iiiipleinenung the idea of free erlnchnient .IS .ippiicd to such cases: see below, Section 6. See also Gardent (zoos) for a fonlid analy.;a involvirrg a genrr:d '

r ~ u i ~ i d ~ i y ' .

Page 99: Recanati, François. Truth-Conditional Pragmatics

- - TWUK ZWaT-OF-XmSTSL "-

LOCAIION (tile-ball, e), in the scope of the event quantifier." Thus enriched, the 1ogxa.I for111 of 'I le danced all night' 1s somethrng 11ke'~

(21) 3~" 3t [~'AYI ( t ) A ~ I M E , ( t , e) A DAN< ING (e) A AGENT (John, e) A ALL-NIGHT

(4 -A I C ) L ~ ~ I O N (the-baH, eN

More pr~grxlahr enrrcliment 1s perh~pb needed to account for the hct that the past tense here ir understood &~cttc,~Uy, a, refemng to a spec& penod ofttme.l6

The fc)llowing ex.alnple of tree ennchrneiit, due to Robyn C x s t o ~ i (ry88). c m be handled s~inilarly through the addition of an extra conjunct about location

(22) John went to Austria and (he) ran into Hans

In (z.~), the event of rutmmg into Ikns 1s understood as liavlng taken place zui itzt\rna, srnceJoltn'5 move to Au5tr1a Izas just been mentioned SlmplrfSi- mg sornetvbat, the literal meaning of (32) can be represented as follows.

(2 3) 3 e 3t /PA\] jt) A ~ L M L (t, e) A GOING (e) A AGENT (John, e) A TO (Austna, e)] A

3e' 3t ' [ ~ A \ T ( t ' ) f lhE (t', e') A ENCOUNTERING (el) A 5UBJ (John, 6'') A OBJ

(1 lam, e')J

That I\, there ic .I p,ist cvent e whlch rs a travel by John to Austria, and there 1s a p \ t e v e ~ t e' wlilcll 15 xn encountenng of Hans by John (I gloss over rubtle drstin~tlor~s hxv~ng to do wltll the exact roles played by John and I lam ~n the encoLlntenng event-hence the use of ' ~ u s ~ h n d 'OBJ' lnstead of 'AGLN I ) , 'EXPLRILNCLR', or 'II-IEME')

Tliroi~gh free ertricl~nient, two extra pietes ok ln fomt ion c n ~ cerning the tirlte nrtd locxtron of e' are I~udt mto the logcal forni FA st, lt IS

conteutuallv understood that Jolin ran 111to H ~ n s after going to Austna More precisely, e, qua tehc event, lnvolveg a transition to a new state (here, the st'lte s ofJohn" bbe~ng ln Austna), and the tlrne t' ofJohn's nlnnlng Into I lar~s 1s ~tncterstood incl~~ded in the tlme spar] of s I wdl represent t h s piece of information by mearis of an extra conjunct L (t ' , 1.1 (where 'L'

stantls tor tlie relation '1s zndlau'ed tn the tzme >pan of the state brozght about by')

' V n dternatrve represetitation will be introduced in $6.1. " 1 lere as dsewhere 111 this chapter, I disregard tile contribution ofthe pedective (resp. iniperEective)

aspect. ft could be represented by us111g Parsons' 'cui.' and 'IIOLD' predicates and substituting 'CUL (e, 1)'

or .HOLD (e, tj' for the aspectudy neutrd "I'IME (t, e)' iri the logical forms. l6 Tlius we could acid an extra conjunct ' t = x', as suggested in footnote 12 for a different type of case.

Page 100: Recanati, François. Truth-Conditional Pragmatics

Second, e' IS understood ar takrrlg phce m Amtna, that is , in the ioc~tlorr where John finds h~rnself as a result ofe Tll~s second pere of lrri.orrl~atrt,n budt Into the logical ibnn thrc3ugkl Gee ennchmexrt caxt also be repre\exited by rrteans of another extra coquntt, I,OCA~ION (Aurtn~,c'J The rnc~d~f ld logcat form we get E therefore

(24) 3 e 3t [PAST ( t ) A TIME ( t , e) GOING { P ) A A ( J N ~ (John, P ) n) (Ai~\tna, c)

A 3e' 3t1 [PAST ( f l ) A TIM1 (it, c') I? I-NCOIJNIERINC (1") A St1111 ( J ~ h i i , P' )

A OBJ (Hans, d ) A L it1. <y A LO(.A1ION (e', Austnn)]]

Jwt as I have done for (13) ~ r l d (LA), 1 d ~ \ n ~ r g u ~ \ h the hart* 1o~lritlS;rrnt oC,i

statement B e 'It 1s ramulg' . v 1 ~ 125) bdow, 6-ntn ~ t s v w c ~ d $ r r f kc;qlailforun ( 2 6 ) '

What u responsible for the &-tier encc between the bare logcal hrrn ( 2 0 and the modfied logcal foi-m (26) r\ free ersr~cfmlenc S~nce free rnncbment 15 optlonal and coiltext-dnven, chert* wll be contexts in u ~ h l i h r lo rnod~flc a-

tlon will be ~ntrodut ed dnd the llltuitrve truth-cond~nons of' I t r> ralnrng'

w~l l be deternuled by ~ t s bare J o g t x l hrrn I'he weatherin~rl context is

arguably one such context Ih

3.;. Thc lex~cul level and 1 l t r 7 phrtzsclli'~enterztitx1 ictvl

Note that, m the part of (2;) tl3'1t represents the bare loglc~t tor113 of tlre secorld clause of (32). we find rnfc~rr~~,rt~on about the tlnle of the encottar- tenng event el. Infi)m~atioll about tllc locabon of P' 011ly Cotnes LEILO the picture through pragniahc exx~chmerrt, but telnporal rnhrl,*t~orl rs dlrcdclv present at the hteral level. Incieed, ,I\ x L,lngutsttts und l'/zalosophy reffree pointed out, the hme 1s treated very rx1trcl-r llke the agqnt. Botll <ire rrrtro- duced into the logcal form v ~ i thelr rrlat~ons to the event argurrlerlt X)oes this not rontr2tl1ct whnt 1 h3ve ~ 1 1 4 1 stld 217011t the hmt' r io t herrrg i t ,

argument, contrary to, for exviiplc, the agetit?

l7 The distinction between bare logical fbmi and rric&~ed logical Ibnn will he ehbor~ted 111 Chapter 4. This is actually debatable. If, as srserai corr~nietltatorr luve pointed out, rain on Mars woilld tie

irrelevant to the truth of the wacl~emran utterarlcc, tlus stiggests that ssonle fi)m~ of ertnchn~rnt okrs place in ths exarnple as weU.

Page 101: Recanati, François. Truth-Conditional Pragmatics

94. i ( ri irc c r vr a \ < )r- ANAt ?St\

It docbrl't, brt JLIW the hntC 1s ~ntrotluced mto the J o g c d h n n through tlic \exiraniut conmlslatlorz of the past tc115e li IS not irtlroduced t/lro.~tgh the ioirfral~iiiiotr OF thr lic.ril z ~ ~ l j While sorrle vert>i, nuy have a temporal A I ~ L X -

xalcupt, V C ~ P S like .(Imie' <>I- ' ~ L I I ~ ~rito' do IIUC Illdeed the ilexrcal exrtry of most vvrbz l a both tenrpordly 2nd lot ,rr-icrr~.rlly neutrdl The Afference Arrwccn tlnw 'ilxil F(-tt aaorr iorrrc\ about ,it ti-re level of finite clauses: ter~ipiir~l ixrfi>rmat~arl 1s t<>ntrrl~~ttecl by the tenses, so there 1s a ternpor~l

I T ~ U I I I ( , X I ~ 111 C I ~ C ~ e \ ~ ~ l t i n g I(>@cLtl hrni, b c k w enrichment. In contrdst, a IOICJL~V~ argti t~it~r~i. 1s i r ~ t r o ~ l t ~ ~ t ' d into the 1og~-a! ibnn ot- a .;tztternent hke (13)

c>r l ly thlro~~gir tier errnchrricnt: i t doe., riot belong to the lrteral medning of 1 hc wrrreric c -

1 ( orrc iirde thit wc \11ould ~i~st~ngtllsh ii)i~r Ievei~ the rnt~tapl-tys~cal level,

the jex~cdl ~ C V C ~ , thr \yrlt'lg~n.>bc (I c. phris~sal ~ n J / o r serrtent~~l) level, and rile pktgxn.itlr level TI-rcre i i no iert~por,rl or tcx-ntrbc argtrnlent In the lewcd crltrv iiirr ~ c r l ~ i , 11l\c 'd ' in~e' Still, \ U ~ I v c r k ~tcriote events, ,ind an event lms. I)c>tlw a rrrnc , i r d 3 Ioca~otl coordtnatr at tlvc rrtet*aphys~cal level Further- rrrorc, ~1nie4 ,ir ix ixltrocftrc ed IIIKP the l o g ~ c ~ l hrnl at the selltea~al level VIA

tire tclrw t r f tlie verb thrr dntirlgual~rs tirr3c.s frorri Lo~citlons, srrice tliere a 110 ( c ) u t ) t e l p ~ ~ t t i ) ~ ~ K I S L " 1x1 the c p ~ t ~ , t l dolridtil I O C L ~ ~ O I I S car^ be optionally i r ~ t r c r c i ~ i c eci i x r t i ) the (bare) litg~c'd toml at the \)int,lgmatrc level by Incnns of i t l c ~ t l \ c $)rLt~c\, or they can br ~ntrocirrcetl r l r to tho ir~~ocilfrecl) logl~al f?)~rn ~t thc pr igl:rli,rtac Ict.c.1, via fret. crrarrclznler~t

Rcr-~trrrrrig to rriy rtiarxl po~rrt. I l~olcf c l ~ a s 'rnrr~' 1s lrke 'cinnce'. There 1s no itrcii~atrr~ ;nsgti~ilcxit 1x1 t l ~ lexrcai critlry fbr tlrc verb '~.-trtl', whrrh can be I i~prcrenrt~cj i~ r r~p iy ,I \

Met'lphj ~ r c 'tlly, Ixowevcr, thcru's bourrtl to he 21 lcic,lt~on tb1 the event, and pr,ii";m<ttic rlly A c r , ,r iocat~on I\ most o-t the elirlc cctntextu,dy understood ~f not e\l>irir~I) \pet rlic.cJ. The rtz5or.r. aig41~11, 11 tl-~it we cale abotrt meteoro- io#~c.mi isvuxrr\ Itr the extct~t th&t we t-,ire .~hoilt the13 loc~tlorls. I nlay be intihre~rtr'd rir Flrar-lrng tliat it r,tlns fi-og\ ctr th,xt rr ralrrs blood, m whichever plaice t h ~ t h ,rppen~ bbrtt fbr orclrn~ry 1 ~ 1 ~ r or \rlom, i'rn rnterested in heanrlg

~bo i l t it c~rrly rf' ~t ioricernr sorrle p l ,m tlrdt I'rxr inttre\ted m* I-lence a ic>c~trun-ar-rdeiit~~te "t'c r,ilrung' 1% uxllrkeiy j~rrllru we dev~st. a specral

( ontext, lrhr dle wc'atllcmlan \t-cr~..~r 10)

Page 102: Recanati, François. Truth-Conditional Pragmatics

4. Rein terprcting ttlf PC tveathermarl cx,t~rlplc

kven though the reference ro tlle b'tU is Irng~t~strcttlly ulrart-ltulated In

exnr~~ple ( r j ) , it Akctc the tritLlltlve tr11tl.l-condltlonl conter-rt of the utter- ance That 15 wiry 1 talk of ‘pragmatic e~trtchrnerrt' lather than ofCconvcrs~- tiondl in~phc.ttule', 5rrtc e the riotion of lnrpllcahlre 1s ofteri taken rn a ~l,trrow

sense wlvch entds truth-concflt~onal ~rrelevarlcc Yet prxgrn,itlc ermch- merit has sornethirlg xrnportarit in coilitrlori wlth cor~versat~on~~l rrnplicatlrre: in both cases the aspects of rneanmg tll,tt relult fro111 the pr~g~:lltatlfc prttc ess

are optlonal Notlxrlg 111 the rentence ~tself- - nothing hqquuttr--req~iii es the praginatlt ennciiinent to take pl'lce, hence ~t nay or ntay not take pl,tce, cfepetrcl~ng on the coriteut In ti115 re,pcct prahm,lnl,ttii e ~ ~ r ~ c h l n e ~ l t CiifEr\ from tndcxic'tl rcsolr~t~orl dncl fro111 the z>slgrrtment of v~1tlt.s to tree v'in~bles/ ~~ronornrnd t.lex~rents Whcri ,t v,ilile I \ contcxtnally dss~glecl to arr rr~dex~cal

or a bee vari'lble 111 Ioglcdl fo~rn, the contextual piocc\r (-rCv,,~lx~e Jssignmexit zs tnggered bv son~ettluig Itnpistlc. Since ~t ongnates In wrnc property ot' the expression type, the pragrrl'&tlc process m q~~es t ror~ P V r r t ~ t &kc place, that ts, ~t taker place In every context In which the rexitent c 1s &l~citourIv ilttered.

Tfus provrcics n\ with a tert For dec~drng whetlrer ,i c oritextu'xlly ~-rrovlded

elctrient of utterdnce nlealilng results tforr~ prapntrc enncklrncnt, or whetl-rer ~t rejults troni 'snturatlon'", ttrdc is fio-oltl ii r-rr,ir~d,itory pro( esq of contevtual assignment.

1x1 tlie tale ot 'rain', we find that the contextual ~ p e c ~ h < dttorl oftile place

of rairl r e d h korrr free erinchnlexlt, rather t h ~ n lion1 s,LtLtratron (a\ thc' starldard view 1125 tt), because it 15 not ni'ind;~tory Irr the wc,itherrii,rri ex~rtiyle, no ?lace of r a n 15 spec~fied Thi\ \hawc rlx~t ri2eteorologiral predic'~tes do Trot carry an arg~trnent slot for ,I location, contr,rry to the rtandard view Thev no more carrv an argurrrent clot tor .i 1ocat1011 tllarl event predicates l ~ k e 'dance' do. In all cases, a process of fi ee ertrlchrnerit

rlidy take place, EL tlrtue of wh~ch the speaker tacltly referc to a particular place as the place of the described evelit, b t ~ t tbls process 1s cr~tirclv prAgmatlc and 15 therefore irrelevant to the selnantlcls of the event predicate

1 take the we~therrridn exanrple to cast tloubt on the stmdard view Sukt e 1 pubhslled ~ t , howevel, attempts hake been nlacic to account for ~t wrthout depart~rrg &on1 the latter. Two W A ~ C of reinterpretlllg the we,ttlrcr111,an

Page 103: Recanati, François. Truth-Conditional Pragmatics

exlunple lldve been suggested so as to avord my concl~lslon that the locatlon c t f rain n not lrnguutrcdy represented m sentences hke (2).19 Accordu-tg to I

C -

the f i t relnteryreLatror1, the weatherman example IS compatible w t h the cla~rn that 'rain7 I \ assoelated with 2 kee location vanable (or IS a part-time irldewcjl whose content depends upon a contextually provlded locatron) because, appearaxes nohvrsthstartding, a specific location contextually

provrdeil ever1 In thdt exaxiiple The second reinterpretation concedes that the locatior~ ctf rain n not specified in the weatheman example, but rejects 1 the conclm~ctn that a is not Iingulstically represented. Rather, ~t holds that there is mnqd~crt existennal quantificat~on of the locatlon vanable.

4 2 Ku~t.2 on f,urth

The fint re- ixtteqretdtlon has been put forward by L,ulsa Marti (2,006) and, ~ndependrntly y. by Paul Elbourne (p.c.). They argue that, in the weathemtan ex'irrtple, dre lot atloii varl~ble 1s assigned the wl.lole terrztary as contextual value. "t's ra~ning' therefore means something llke 'It's ralnlng on Ear&'.2o

Whoever puts t'orwxd such a proposal trust explan whv we get an enisteril~cll leathng 'It 1s r,unirrg (somewhere) a n Earth'. Norrn~lly, when we sky, for example, 'lt's raining in Pars', we mean son-teth~ng nearly ~n~ tve ru l that it is ralrilrig over Pans (I e. at most sub-locanons in the Pans are,^). But clearly, 1x1 the we at he,.:--^ example, tlre rentence 'It's rammg' doe5 not riledn that it's raining over the Earth (1.e. nearly everywhere). The weathe~n~ar~'\ LttLerartce orily means that it's rznlng sorn~~vhert

It IS not hard to find an esplalat~on Sor &I\ fac t, however. One may argue that the un~versal readmg, though wiciespread m d poss~bly standard, i s not lmguisnt dty rrrclrzciated but itself results fro~n a pragrnatlc process-a pragnuuc

lV Actrtally, four ways of reinterpreting the wcathennan exatnple have been suggested. The third tC~s i t t ip~eLX~<,~ i ~ p ~ c d ~ t" d pt&@tixbC #iCC"tAit "fdic2 &a t i i a ~ L5 sig&~&C.isidy Metc2ti~ &.Oil1 the 2 C C " ~ i j i t

I have sketched; I defer &scussion of that alternative account until 56.2. The t ~urth reinterpretation, due to Andrea lacotla, invokes a distinction between epistemological content and semantic content which I mm unwiliirlg to accept I wiIl not diqcuss lacona's solution in this cllapter, but refer the interested reader to his anide (lacona 2OU5).

"O Elbourne says he does not actualiy hold the view, but nierely pu-* ward for consideration. Besides Marti ar~d Elbourne, otlter persorn have argued along the same Lines. The first to have done so in prirtt ;ue Sandro Capone and Jason Stanley in their respective reviews of my Literal Meaning (Capone zooj: 46, Stanley zoogb). Stephen Neale (zoo7) defends a similar position hut, like Perry, he refirams Crorn positing an 'argurxrent place' or a fiee variable in logicd form and commits himself only to the existence of an 'argument role' in the lexical semantics of 'rain' and other meteorological predicates.

Page 104: Recanati, François. Truth-Conditional Pragmatics

process that does riot take place, for prmcipled reasons, nl the \v~dthel~ldn example. The explanahon proceeds roughly ns follow5

I The sentence 'it rans at ? rs hterally true ~f 2nd oalv if rt ram\ at. some

sub-locahon I' of 1 2. In many cases, ~t 1s relevaxit to melrtlon run In coiincct~on wrth ,I

specific place I ordy 11'tbc sub -1ocatlorrs of 1 where rain actualilly orcucl

are representative7 of 1.

3 If it rams over 1, then ralii occun at rliost sub locaaor15 of- I, a i d th5; I\

sufficient to guarmtee represeotativrq

4 'The hearer assuxnes (ancl n e~pectecl to d~sulne) that the utteralce r\ relevant. hence m rllarrv caw:, he or shc wrll be led to as\nn\c that ~t

rams over 1 5 . In some cases, however, o i ~ e of the f~)llowxig t or~ct~tinx~s lruv I.rc

sahsfied: it is relevant to inentior1 rail1 1x1 connection w ~ t h iorrre

locatron 1 even d the suh-loc'it~on> of 1 where ram actu~lly occ un are not representatrvc o f / , o r the .irib-locations rrt cluc\tlon re repre- serltatlve of I even ~f lt doe\ not r ~ i n over. 1 If e~ther o i these iorltlrt~oni,

is sattsfied tt will be relevdtlt to mention r.iin 111 corlnectloli wrth pYacc i even though rt docs not rain over 1 I or cx~mplc. if I am rold t h a t ' r t \

rairxing Gags 111 Dostorrr', X tlo not (neceswnlvi tonclutfc t h ~ t rt'c rarxb-

ing f rog ni most pots of- tile Morton Area It i z relevant exioragirll to

know that in some spot in the Bostoti area, ~ t ' s ratrirng frop 2 i

6. In the weatherman exanlple, argt~ably, one of the defkattng condriiunr

1s satisfied, just as im the ralrzng frog exainple. For chat reawn, tlrc pragrnatlc step frorri ex~sterltzd to u~~rversal ir not taken

According ti, the propoi.cd explanation, the exrstentld (non\pec~fic)

read~ng of the weather~mx~ exnmple I< cornpattble 1~1th the E ~ c t that a

speclfic locatlor1 is contextually provided, for the specrfic l o c ~ t ~ o n r n que\ t~on-the E,artli-1s the place of ran only m the sense that ~t I I ~ C I U ~ C S (4s j:

sub-locahon) the place of ram I t turn5 out that tliere are two sensedor the phrase 'the place of ran': in the nat~ot t~ rense, the place of ran is the place which rain actuaUy fds; 111 the broad >enre, the place of r a n car1 be dcly piale

I owe the 'raitlmg frog' exanlple ti) I h n Sperk~er. (The sanle exairlpie has independently conie up in discussions with Pranav hiand, Eric Swansorr, arrd Sarah Moss at MIT.) W k c h of'tlre two dcfeam~g conditions obtains in such a case is an issue I will llot go into here

Page 105: Recanati, François. Truth-Conditional Pragmatics

?r8 i.irifu I kiir-KF 2 iht* iHt- 'WFAf HFAMAh kXAWi.Ik

t h t e lzrc irrt'ic,\ the narrc~~v-pLu e-ot-ram .n a srrh- loc-~tlon. If ~t rams in some

piat e (iar tlrr irxrow sense) then (in the brad WIISC) ~t ra111~ 111 ally place ulc-lrr~\~ng r t II it rxn i ~n Nexrc o ('~ty (or rn sornc s~rbirrl) ofSMevlco G~tv) , thcaai 11 rain<, ori I * ~ r t l i T'lre wesrlrern~~rr ex,arrrpic. 15 t~onspecrfic (exr\teiltlal) WIL~I re\pect tc9 tlic rlarrow p l ~ c c of rain, bur it rrtxtertllelen contexttially s~gecikicis the p \ : ~ e ol rail] ill the brwr e l . s ensc'.

C ~ I this view tlri* lexical en0-y h,r the verb 'rraiir' is soil~ething like:

ar~tl, :as the sul3scnpt itrciic:~tcs, the 1oc:itiorr re1:itiorl itself i s understood in the irroad senscb ' 3 r d axialyscd as fidlc>ws:

(Vt9 (Vl) 11 o c ;\r ( E , (9 ri-f 31' ( rocn~ro~ ( I f , t) A 1' 2 1 ) )

wlirsc ' 15 the suli locat~on reldtxorr .lrrtE ' IO(AI~ON' iorresporlds to the

( > R ~ ~ X I , I ~ . r\;rrro\c notron oi l o c n t ~ c r r r l i r ~ the wcatftermn ex~mplc, the

v~nablc 7' i\ ,rssrgl~cci the vdlue t " ~ r ~ h , , i r d tire cvcrrt vnr~able rs ex~stcntially cj tunt i~f ie i - l , \o thrlt (ciinreg<~rdrng tcrlw :inti .r+pec t OIIC~" dgarl ) we get the

e x p ~ l e i j rea t i i~ ig

" 3 ~ ININ( IN(^ (i") r o c A ~ I O N , , (the f 'trtll, c)j

t h t n,

3 e 3 S ~ i a n i ~ r r j c , (c) r c ~ t AI LON ( / I , e) A i f <-- the Earth j

/ s o p t l l 9 t i c r / tJc*r~alili'$

The icc oncl r e i r~iurpri . ta t~o~~ aylw~15 to the f ~ c t that, ,n tllc f'artec example\

\I-row, An rrarpirc 11 t an'ible L ~ I Z bc 't)our~(.I Sint e thdt is 50, why not CAY that the i r xz ld~ i rr ~0idt101k v d t l ~ b k 15 1>01111~1 by c ~ ~ ~ t t " r ~ t l d q~tant~fter In the wcatlrt'rr~l~\lr ~ ' l d ! q ? k , t l l l~j g v m g rise ru the ~ndefinite readlr~g! Why not arl,dvse "t'i rg~rrlxrrg3, m t h ~ t example A\ 'Tl~herc rs .r I (xa t~on 1 sucli that ~ t ' s sarxnng a t 7 "")I , t%~lmvxlc~-rrly, wiry lot s,ry that tlrc ,~rgunrcnt-slot for a

Iocaticlrr e\ t~lletl by trit%arls of a c ovcrr rndtt-urlte ('sso~rrcwhere')?

The rcJ\c)n 1 t~xvc ctfferetf for resistrng thlr sort of ,ul~dyus I\ that overt

va~7dlsics (e g proootrns) canrlut be coverrly hourzcl. 'Thus, ,is I polrited out,

"He rs tnII\ i,ini~ot ritcLGut that \otneone 0t2lel 1s tall. Why should covert v.tn.tblcs bch,ivc tliifcrently froin overt var~ablcsj

Page 106: Recanati, François. Truth-Conditional Pragmatics

Afier preseutlng this argument In 'Unartlcu1,tred C:orircrtt~ex~ts', I J I I ~ I C I

pdted a posslble response. There are, one nugilt atgl:ue, two sort\ of var~&!tlc<. those wblch (I~ke pronouns) must be ~ontextu,~lly ,iu~gltecJ a value wlien unbound, and thaw wh~ch, when unbouxid, cart t ~ t l ~ e r be tonte'rttrnlly assigned 3 value or undergo existentrd closure. The v ~ r l ~ h l e , whit h rlred not be assigned a speclfic value but rnay unciergo existerrtxal closure I dubbed 'opaond vmabler' T o actourit for tile weatfrierman example. then, one only h a to tlaim t l l ~ t the locatiort vanable belongs tto the ‘optional' t ategory

Thi5 response w~l l be convmcing only I€ the iategory of opt~onal vartahlc\ 1s lr~deptjndently needed, that IS, orlly 11 we car) tillci clear iiistnnt e5 ot the

category. It n temptlng t o argue at this poltit that rel,ihoi\dl noulrc hlie a noth her' prec~stly cany such ,m optlorial va~l,ible; a \an,aI>le wtri~h, ,I\ examples (7) arid (8) demonstrate, ~ a r ~ be eitl~er bound by ,I covert eur\terz tid cjuanhfiel or &\igned a definite vaille. Yet we cdnriot 30 .ugitr wittiout beggmg the yueshon, for what 1s at s;take a the c o n e ~ t analysrs c,f'\erlterlcc\ in which a putative inrpl~clt argumcrit 13 undetstc>oci as euatrrittally q~i, lntl-

tred 'To jusnft- the analysis in ternis of oytlor~al v*inahle\ both ol'ex'i~~iplcs hke (7)-(8) und of the weatlieman example, we r-tlust ti11t2 tnstarlt es of opnonlrl vanables among the overl vanable-like elemet.lt\

Tlie orily candrdate 1 tart thiiih o i here 15 tellct. Accorct~rlg t o 1-''1r tee (19731, tliere 1s a stnklng parallel between terism a i d pronourls, 't prt~allrl that justlfies treating tense\ AS vLtnai)1es. Like pronouns, tcnses ftave dcict~c, , ~ n a p l ~ ( ~ r ~ c and b o ~ m d uses In Partee's 6lnious es,lltrple, '1 dlcl not tLrrn otl' the stove', the pa\t tense 1s urliterstood de~ctic ~ l l v the 5pt~~rkrl ~cker\ to '1

\peclfic hnle in the pnst As Partee wntes,

When utteled, fbr Inbtance, halhvc~y down thr t ~ l ~ ~ ~ p ~ k e , S I I C ~ I \cntcLtIie ciea~iy doe5 no t xnean eltlirr that there eylsts some tune In the past d t W ~ I C ~ I I dl~f not C L J ~ I I

off the stove 01 tkut theie emts no tnnr in tile past ar wh~c h I rurrtetl oti the stove"

The sentence cledrlv refers to a parbctllnr orlie (1)arrtcr ro73/zooq j ~ )

Pxtee provides other examples In w h ~ c h teriscf ,tte l~sec! ,~lid~>l~ori~ally 01 a\

bound vanables. BLI~ what she fads to notlce 11-1 Izei payer Ir that teriscs 11ave a reading w h c h pronouiis do not lave

T11e extra readsr~g I have r n mxrrd lr the very leading whrch I-'altce xlys is unllkeiy sn the context *he descr~bes 111 .* drfh ent corltext, ~ T C S U I ~ I ~ ~ ~ V , the past tense renrcnce '1 did r ~ o t turn off the stove' ~ o n j d t)c. givert ,t11 '~ridet~rlrte'

Page 107: Recanati, François. Truth-Conditional Pragmatics

-- -- I 03 - & ~ F F R P K F T I N ~ T X T - V ~ ~ R M N XXAMPLE

or 'exlstent~al' mterpretatlon, rather than a deichc Interpretation One week after the event reported by her utterance 'I did not turn offthe stove', Partee ~lirght have told the story ar; follows 'Last week, I &d not turn oEthe stove and I risked a serious accident' In this senter~ce the slniple p a t has an

existetitlal re,~dirrg the sentence says that there u an event e and a past hrne t ,uclr that e w nry rtot-turnlrig-ogthe stove at tirne t, w h c h tlrne t rs located within the tune-span tndic ated by the te~nporal adverbial ' la t week'. Even though the serrterlce t oilta~ns ,I reference to the tmre-span withrn whicli the event talked about 15 said to occur, there 1s no deictlc reference to the nme of the event Itself, cvhicli rernans ~ndefinite.?~ Tlvs 1s in contrast to Partee's ongmal example, in w h i ~ h ~t \ eem that there 1s reference to the time of the described evertt If that is right, artd ifParteelr axlalysxs oftenses a5 variables is correct, tlit.11 teoipor~l v~nables may not only be as\lgncd specific values, a? m Partee'., ongxi,rl euarnple, but may also undergo existennal closure, as xn the revised verzlorr of the example.

1x1 the revised verslon of thc Partee exarnple, we need a temporal ,idverbla1 srlch as list week', referring to a penod m the past, because wlthot~t such .I reference to the past we would have to rise dlc present perkct to iotivev the cxtstential zlriterpretation In French, thm, 0s are s ~ m - pler w~t l io~t t ~r rv tt~niyoral adverbial, the pass^ nttrzpose can be gven either a clei~tic or an exl\tcnt~,~l ~nteq)retation The same thing 1s true of the future, both in t rerich and in English 'Tile sentence

nsay be uriderstctod in two ways: either the speaker is referring to some specific hture time, niade salient in the context, and she says that she will go to C:lzina at that time, or she is rxierely sayrng that there is a hrture tirile at whicfi she will go to Cllina.

What matters to us is that the second, existential reading is not available tvit!~ pronowns. 2 Tf-2 is h:rld' cannot mean that snmp n~2le or 0 t h ~ ~ i s h3ld 23

Uecause of that extra reading, if tenses are to be treated as variables (as Partee 1973 suggests), the variables in question must be &Eerent &om pronominal

"L I an ~ ~ d e b t e d 10 Phshppe Sct~lelzher for cfisctss~on of ttus type of exarnple Z3 Paflee ( 1 9 7 3 / ~ 0 4 jz) mcrltlons the existence of what she calls a 'nonspec~fic' de~coc use oftenses,

1x1 LJOIIII went to a private ccllool', and she clans diat pronouns too have such a use ('They haven t rnstdlled my telephone yet') So she mlght be tempted to deny the asymmetry tlxat I am pollltlng out- chc m~gll~ atgue that the ailegrd e,astennal reacfings are all nonspec~fic d e ~ c t ~ c uses

Page 108: Recanati, François. Truth-Conditional Pragmatics

W E A r I I F R REllC)IIII \, 401

vanables. they must be opttonal vanables An optional vcmalde, when un-

bound, may be contextuallv x,sssg~ed a specifi~ value or undergo ex~stentlal closure

We can, however, reject I-"rteee"s elmre dpproacb and maintarn that tensed sentences-r a t least, tile senterlcec xn the slmple p a t \he user as exa~nples~~-quavatlfj, over trmes, even or1 the Aeged d e r c t ~ rcahng. Tlxe desc~c rea&ng arguably results fron~ restnc ting the donlain of yri~ntificatjorr 1n a manner that rnssrlvcs singular refkrencle '"'Thm 'I clidrr't turn off the

stove' means that, m tlre qet ofpast evelzts rnuliedi~tely fbllo\.vmg rny lact trse of the stove, there 1s no tunnng oi3 ot thc itove by mc I11 thrs way wc account for the coexrstent e of eslsterrtlal arztl of (alleged) deictlc uses rtrltlxottt

having to posit ophonal varrables Which theory are we to ~hoose l E~erytli~lig being equal, we \t~oirld

prefer the most pamixnonlour theory, tllat I\, the theory that dotc rzot posrt optsonal vanables (in add~tion to s ~ ~ n d a r d pronr.rrruna1 elements) But ev-

erything may not be equal T h e a n s l ~ ~ s ~ s of teme Ir a complex dfrjrr, ;md Yartee's type of approach using van'ables IS gener,llty conrldered JS qulte suc~essfkl,"~ it rnay be, tl-ierefi)re, t h ~ t cve shaU have to swallow opttonal vanables in the package If so -11 opt~trr~~il v~rrdblcs are ~ndepel~de~rtly needed to account for terns then we rnav [eel f h e to use the111 to accourrt

for the weatherman example If we take t h s hne, the Iexrc~l entw &>I "tarn' wll be. again,

,\l he [RAINING (e) A LocAr roN (1, rjj

but t h~s time the locatlon relac~orr wrll he undecitood i rz the stdndarcii.

narrow sense (as the absen~e of subscript nl&< ate\) 011 t i l l s theorv, \vXl~t happens in the u-athenrim esrtnrplr 1s that the l o ~ a t ~ o n vemdble I I ~ bouxrd

" Thss qualification n needed in view {.if' cE~r fact that. as Schie~ihr~. poirtted ixit tn rric, L:rcr~ch 'impahit' does not accept emsterlod rcadi~lgs.

This alternative analysis was &st 111~11tinned by I'mre lrerseii(ser t'ooniote 3 113 her xy;i payrcr). See Lasersohn (1999: 537) for a more recent proposd riorrg those lines.

" d o t by Partee herself though. In Partee (rg8qb) she givcs up her earlier a~iaiysis and trcats he tenses as establishing relations betweeti event tillre anci ref5rrrrcc tirne (as m Wolfg;atlg Kleiri's Kerch- enbachian &mework). She not oiliy gives up the treatrt~rnt oftenses 2s variables, but also refra~ns ken> positing reference-time vanables: die reference tulle, she polnts out, 'does not cnrrespo~~d i~rutbnnly to any single constituent ofthe sentence' and it resists irluoductiori in a direct model-theorcnc mntcrpretn- tiorl of the syntax (Partee rg84b: 266).

Page 109: Recanati, François. Truth-Conditional Pragmatics

I hrce TI~C~~I."', drc 111 ~~I~lpetl t lOll LO dCCOLlnl to1 the we&t~iernl&tl example. A< ( or(ling to t f l ~ fir\t theory, xlietcr~rc>log~lcal pred~cates do not czm- an

.irgurilcxrit dot for d loc-~ t~on (excepe rrr the ger~er,~l sense m wtzicli ever).

cvcxrri p ~ c t l ~ i ~ t c C,trne\ SIIC-11 3 slot), it fOuowr\ that mrtearologpcal sentences I r Le " i t 14 T , I I E I I P I ~ ' i l e c ~ ] not be under\tood as ioc,it~ors-spec-~fic. Puttmg tense

'iaii J \ ~ c ( t 151dc, the iellldlttic c01itc1lt of (3) 15 ~l1np1~:

( ) ~ i t h i ~ vit"'i.'i, tlii' j)o\~~L~lfity of arl urdefixrrtu rcachng of (z), ai in the weatl~cr- l i l ; i i ~ c.i,iii~pic .i:ril crrtra ely eq?et tcd. Ex~iliplc (2) 4ay4 that s ram event 1s taking p1.iic. 'iritl rLiln cver~t, hke m y event, IS b o ~ m ~ i to t,&e place somewhere. I ierlc r_. "I>-'\ x a~i rng" ts cqin~~dent to ' h'5 rAxrlmg \orrlewhere", just as 'Mary 1s

t i a r c ln;< i s cc ju ivder l~ to M~,rry IS da~lcmg S U ~ I I C ' L I . ' ~ ~ " ~ ~ ' . I3crtj-i h a * \e'coxlcl mtl the third ttxeu~y Irx'arntam tha t tneteorologcal

pn-d9r(-,~~(~ 1liL.c ' z , I I ~ ' c J I T ~ AII A P ~ U I I L C I ~ ~ dot Jbr n location:

f\( I 01 IGLI$: LO the s t - ~ ~ ~ ~ t t f thcory, mrlucl~ n m Instance of the standard wrw, tlie slot rnu\t be cor-rttbxtu,lUjr filled WI th a specdc location; and a

kpcuif ii Itrcdhoti ir ~ndced prcw~detl ~n tllc. weathen-llan cxanple, namely t i 1 I 1 Ile taxrstent~~~l tii,rce of thc w catbc~~nan exanple 1s accounted Gir iry rrrccrl)retrirg the locatlon rel,le-ion ~ r r tl-te broad sense.

e A i i o r cling LO the tlnrd tlieory, t l~e argrrtnerlt slot need not be filled

i + r t I r \ p ~ ~ l i i ( - loc~tion, rt rn.ly bc bound by a covert cxlstential cl~~iilrrl-rcr. ,rud that IS w l l ~ t bitppen) rtr the weatherinari exaniple.

I Eow are we to ~d~uthc'lte betm~cen the three tllisoncs? In this secbon, I argue thdt both t l ~ c \cc o t ~ 3 xlld the rbuld theory Llc e plwtrlerns. 'I'he tlxrrd theory does

nor rtrEc. wit ,I i C T ~ J U I r r ~ ~ b q r , which, J. matter of tact, does not emst, or 1s

Page 110: Recanati, François. Truth-Conditional Pragmatics

h a d to get; cvtule the second theory weaker~s the notion of 1ocatx)rt to the porirt where ~t can no longer do tlre jol) it w ~ s ~r~tel~deti to cfo.

j.2. &~ziti~!~f the tibird theory

Cor~siiier the sentence:

(27) It is not raining

Can we run (A vanant of) the weatliermaa ex'tmple tvrtlr that sentence, so as to get the foilow~ng readlng: In sorne ylau or other, t t ' ~ rtot rtazntut@ Let us try, by adjustmg the o n ~ n a l scenano

Inlagme a ~ l tua t~or~ 1x1 wh~ch the absence of rdln has becortre cxtreinelv rare and mlportmt ( ~ t rasns allrlost ecerywhere and everytmlej. nil over the terntory detectors have been ciaposed, whlcl~ trigger 'in alami bell in the Mo111tonng Room cvben they detect the dt>\ence of ran 'l'here rs a curglc bell, the locd~lon ofthe tnggerrng tletector I\ irldlratect by 'i hght oti a hoard In the Mon~torulg Room. After wcekv of flood, rhe bell cvcntu- ally r ing m the M o n ~ tonng Room. HeAnn# it, the wedtherman otr duty In the adjacent room shouts. 'It's not r<i~nmg~'

Can we say that the wedthemian's tlkter&rlce 1s true ~fi; 1x1 some plat e or other, r t

a not rauang (at the tlme of utterance): I tlnd ~t rather liazcl to t~ndentand the utterance that way, deyrte tlie colitext 'I'hc weatilernran ought to \,av sorne- thrng hLe "'fhe rain has stopped' tlia could be uilderstood 3s a ~ e a r ~ ~ n g that the ran has stopped somewhere Rut rt IS very h , ~ d tc3 sugn to 'It's rtot rarr~ng' the (w& scope) mdetunrte readmg-rrn~tch harder than rt w to unclmtand the poutive sentence ~ndefmtely, as In the onpal weathernlan exm~ple

The unavai~=bil~ty of the wlde-scope ~~rdefimte reacting of (27), In con- trx~t to the avallabdlq of rhe i~rdefintte redding 111 the ongi11'll cveatlremran exanlple, rnust be ,tccounted ior . I wll argw that tlrc asynmetry n unex- pected on the thsid theory, whale ~t 15 expectect on both the fint dnd the second tl~eoty 27

Accordmg to the first theory, 'It7\ runlrrg' s~mply says that a r~iming event is takmg place, and 'It's not raln~ng' say? tl~at it 1s not the case that a rdinlng event a talilng place In both cases tile htrral rneaning of. the sentence can be

1 am indebted to Pcmav h a n d for suggcs:esung tltat way of test~ng the theorits, and to Paul t(1boume for poznong ou t that the iecotrd theory l).c;se, tirc te\t, i.ontisry to what 1 it~~t~rrilv tllought.

Page 111: Recanati, François. Truth-Conditional Pragmatics

cnrlched through some klnd of tacit reference to a place; thus both 'It's ra~ning' and 'It7\ not rarnmg' can be understood as saylng that a 1s r m m g (or not) in Berlin, if Berlln 1s the contextually understood location But the tndefirute reading of the ongnal weathemian sentence does not result &om such a process of enncl~ment, on the first theory: the indefinite reading IS

what we get when we doyz't ermch the rneamng of sentence but take lt at face vdue (I e as rnemmg that there w a raintng event, penod) If we \inul~sly take stllternent (27) at hce value, ~t says that there n no rain (i.e.

there 1s no ran urtywhere). It does not mean that there 1s no raln somewhere. Accormng ttt the second theory, the Earth 1s contextually asslgned to the

locahori vanable in the weatherlnm example. Presumably, thls also happens In the neganve vanmt of the example; (27) is therefore analysed as saylng that on Emh, IC n riot ralrnng. Is thls the unavadable readmg, and is the second theory gurlty of prentctlng that readmg? No. Accordng to the second theory, the Earth not understood as the narrow place of ram (the location filled by ram) m the weatherman emnple, but as the broad locabon, tvlrere the broad Iocatlon 1s defined as a locaaon that corrtarns the narrow place of rain In the broad sense, to say that i t rams at a gven place E n to snv that there IS a sub-location I' of 1 wh~ch IS f i e d bv ran; and to say that it doec not ram ( ~ t I ) 15 to bay that there IS no sub-locat~on r of 1 which IS

filled bv ram bcntence (27), ln the negahve vanant of the wedtheman exanple, 1s therehre ,~rralyrcd ,w sdylng that on Earth, it IS not ralrtmg, ln the ienre that titer< zc no ruztlirtg spot (1.c lt's not ralnmg a~inywhere). T o asslgn the Cnrtti to the covelt lot atron varr,tble ul Eentenee (37) theretbre resu1t;s m a readrng qulte drfftxrerrt from the unavailable reahng SornewI.tere on Earth, it's

not tnivzrng Being bu11t 11lto the lexttal entry for 'ram', the exlstentlal quantlfler over (narrow) locahons takes narrow scope, whereas lt ukes wlde scope on the ~~navallable reachng.

In contr,ut, the thlrd theory h ~ s trouble accounting for the unavallablhty of the lrrdefirirte rending of (27). According to that theory, 'ram' c m e s A

IOC atlon vnn~ble, \vhic11 IS opnonal and can be bound by 2 covert ex~stentlal quantifier That ir wliat happens in the weatherman exanlple. In the negatrvc vanant of tlie example, therefore, the exrstentral quantifier 1s expected to interact wlth negation, In such a way that two readmgs ought to he generated, depending on the scope of negation: tile sentence wdl say eltl~er that at some locatlon 1, there IS no ram, or that it 1s not the case

Page 112: Recanati, François. Truth-Conditional Pragmatics

that, at some locanon I, there E, ram But the fmt seachg li not actually ~ v a l a b i e . ~ ~

To be sure, as Paul Elboiirrir 111\1\ted in 111s CuImnents on a11 3nc C S ~ O T of thu ch~pter, ~t is not the Lase &at any scopal ordcnng of scope b e ~ n n g elements 1s always possible I agree. Bur d some pdrtic~llar st ope urderrrlg turns out to be mposslble, tfserc rruust I x an esplanatiorl &>r that fact.

Whoever pos~ts two scope-bearmg elements t.vh~cb turn out not to interact owes us such an explandtion For eua~nl-rle, C:hierch~a and McConnell- (&net notlce that 'there do not appear to be a~~lh~s t r t i e \ rzsillhng frorir

mteracaon of the negahve and tense tnoq~hemes~(C1nerclzla md McCoiu~dl- Glnet 1990: 232) . They oEer a tteIltdhVe euplmanon for that &ct: there no ~nteracnon between tense and ncgdhori, hence no scope arxrbrgulty, thev SAY, because the correspondtng syntact~c elements (NE,(; and TN\) are both parts of the INFL node, arrd 'the elerrrents of INrL are ur te~mted In a fmed order, w t h NEG always hmvlng mder scope tlrarr TNS' (Chierchia and McConrleE-Cinet rr)r$o 232) Another ex.imple, C ~ O L C to

our present concerns, regards the Lack of rnter<ictlon between rlegatlon 'in J event quantlficatlon: negitlon alw,ivs, or Ano5t al~jvayr, tdk-e\ wadel scope than the dehult quartitlficat~oj~ over events '" Addressiilg thls ~sslre, P,~rsc>ns

wntes:

111 . C ; ~ m p ~ ~ i h ~ n ~ ~ t f , Pmee notes that mclefii~rce rrnplicit .&rgumenu, ~f tieared 2s exisccninailv quantified variables, are such that the eslstenti:~! qumnfier cur c~~liy take Ilatlolv scope relxrivs ro any rrit~er scope-beanris dement in the senterll-e (Rinee i gS~ai2ooq: roK-i)). As a i l axioriyrnois I .i"rf)rt:f&rc.c pwnred out, this fact was kni3wn since tlie iy7os (and I t w.1~ one of LIIC argmnens XL tivour of a Lrxii:ri approachto dlesed indetimte implicit argunlents). in a related veln. Bier\vach otti-red fire foiIo\ving rule iii the c.~.rrly 1980s: whenever a lexically spcc~fied argmient 1s sy~~iactrc,~ilv oniitted, i t is bound by an existeii~~al q~~antlfier whlch rakes varro\v scope ('Unspeciiied Argui~r~llt Rde'---see Bicrwach 1982, igX<j: 76).

" 4 %say 'almost always' because negaoon c ~ n he inteqrrered n~ predicate negaaon, ni whii.!~ case, uf course. it does not scope over the event cpantlfier. Ln rlw G)llow~t~g example, dre second clause conrairn a pronoun anaphorically (or quasi-anapl-lvricauy) rehrrlng to an event ~ntrtduced in the first ciause, a11d &at event is die event ofrlot-doing sonlettlmg:

Brutus did riot greet me. i t happened yesterthy.

In standard event renwntics, diat ex:~rriple, wirrch resenibles the dassicai "Unitus stalrbeed Ca~sar. It happened at noon', should be anltlvsed in the wrlc wdv: ?here rs iz p u t aacnt P ~uhich i j u r~t~rz-.~r<:i:tir<f and

whose agent i s Btutu.c. ' h l r rwnt r taok pt~tic yestmdrday. Another exairiple on tire Ealirr patrcrn mvolves adverbial mo&ficaaon rather thzm event aiiapiiora:

With contenipc, Brutus did not greet nie.

Again, in standard event senlaritics, &at ougftt to be attalvwd AS: There IS an everit e siich thm e is a rion- greeting, e 1s located at some orne t in the pact, e's %rut is Bmtus, and e is conturnpdiii (i.e. urvoives contempt on the part of-its agent towar& is patient). Of course, not any old event that happens to not be a greeting event will count as a non-greeting eveizt. To coirllt :~r a nori-&~eeu~~g event, an evrnt rnusi Irave

Page 113: Recanati, François. Truth-Conditional Pragmatics

Too WHO 14. RII,C-I 14

ii theri. i z 4 iicg,itlo~i 111 the ccrltence, rt mubt have wrder scope than the defat~lt c1t~u"i"iii;i atrori, tlxrr a I t c ~ ~ ~ u i e .r 5eutrnce Ilkti "Maru did not rurl' says that ~t 1s not tlic ~2.r tiiiit t ! i t ~ r ~ w a s a nlrltlrng wllow agent IC Marla; 11 the negatlon occurred ~ns~de tlae t-leiault t juar~t t t~cat~on, r t would irrdke the sentence say that there was an

e v r i i t nlix~h wd\ riot a runnlrlg wllose agent rs Mar~d, 311d t h ~ s would be true even if Mxi in d r i l r uil (Y anons zt)na r 9)

1 ,1117 r i o t 9urt' this 1% 1rlte11ciet1 '1s ~n cxpjdndt~o~~ for the fact that negatmn and c v e r ~ c i I ~ i d l * t i f i ( <it~on d ~ 1 rtot keely Intvrdct The pdssage m;iy he read as a

starearlent c i i w l ~ t I'rlrqor~s takes to be a (I~rute) fact n u t ~t may also be read a*; pi~w~dtng .L ~cason for the lack ot ~fdno'uv-s( ope renchng for negdtlon. 'The nad,trn wotrici be thn (1) 'Mma rd11' aid 'MMi.nn d ~ d not rtut' are supposed to %,c co~rtr,ic~rctorv, but (11) if jrcgntrorl a~ ld event quartt~iic~khon can 6eely irrtcr-ac t, tiilcr c. wlil be il reading ai-the negative rentenct. whlch will rnake IB

trutlt c o~~~l ;u~ t~ ! ) l c w ~ t h t h ~ t o f the ilosrtlve rerlteilcc That, ho-cvever, cdnnot be the cxpianatron for the oblrgatu~y wide st. ope of negahon. when we say

tlrlit 'Man& r ari' and M n r ~ a ci~d not run' are supposed to be contracllctory, wc irc d r c ~ d y exc1udir1g the pos\rlrhixty of J n'rnow-scope reading for

ncg'itroxl I.C,ithcr t h ~ n accountrng tor t h a r rxiiposs~l,rhcy, we are plesup- ~w"" ,BI"~ IC Ve t t h ~ pntdtive ArgLiIIIeIlt C < L ~ I>r rrnproved rf \ve take the sC:la~erc 1113. M~C,ox-ineil-C111ct polnt &bout rlegatiori and tense for granted Ihalrg~trcm u opri, over tense, tbcrr, if cve grve the event qu~nttlier xope ovci xicgati(>rr, tiad- rerultrr~g wntersce ' R / t ~ l ~ a C I I ~ not run' wdl say that there is (rcn\elc\r) a n cvcXrit whlcb 1s IIOC an event loc ntecl m the past and conslstrng ~ 7 f M ~ X U nrnrxirig I lrat IS obvruusly too tnvr'rl to be worth sbiyrng, or even wor tf-r ckpr-e\irrrg

Wlrdrcvc* wc thiitk of these elcplataaorw. they ,*re (tentatlvr) explanahorxi. My p m r ~ t ~ t ~ I V C I S C ~ Y that the lack of Interactton between two scope-bearing rle~rrctrts recp~rrr~ sc) inr krrd ofexl)lan,zhon A propol7ent of the tlmd theory otrght to prcrv~de iucEi an t l xp l r l~~~ t~o l~ , irlnce s/he po\rts two scope-beanng elcrrrerks ,?.(rreg~t~c,ri, ~ n d thc exlsterlhaj qualt&>er over loc~hons) between wlrtc 11 tllcrr xs no rtrteractlon Urilesr ,r convlnclrig esp lmat~ot~ IS promded. tlie hrst m c i the ct'cc)nd theory fare better rhar the thzd theory bec-ause they do

~xot have ~111s proljlern

brio esjrcr t ~ ~ i to j?<, .I grcerlrrg. ~ I L C 11terat~1.0 o n 11egativi events i s $carre. Set' Iliw,?nbotl~ar~1 zotx~: 73- j &)r J few rcrr~.~r-h on thc ~ o ~ s i )

Page 114: Recanati, François. Truth-Conditional Pragmatics

A post~ble eAxpl'inat~oii for the la& of ~rrter'act~on between the ex~sterlhcll cjuxit~fier a id negatlon ~nvcdvcr sk sr~iajl rcvtslorl of the tliu-d theorv. rherc 1s

x~o ixiteractlorr, otre ~mght argue, becau\c tl-tcre n lrr) euxrtentiil quaritrfier 1x1

die fint place StrU, in the spirit of tlic sccoild theory, we (an ri~aiilr~tcurr &kt sorneth~ng hke exstentldl qualtliicahor~ tkes piace Instead of rdylng that the locatlon vanhle 1\ bound by 'I covelt exrsteritrcll qt~x~tlfier, trt the weathernun esaniple, we car1 say that the algur-t~ent slot tar ,r lacation is tilled t7y xnems of a

covert pronoun oftlre rlglit sort AH trtlperst-sonal pronoun, IAe French \on7, h a exstentliil force, m d it n ch~acten/cd by the tjct tl-tat ~t can orlly tilie rtAn-o\v

scope. (Thus ' O n tle roizne pas d lapoufe' can c~nly rxlem th,tt it 1s 11ot the case tlidt

sorlieone is nngng at tlic door, on the lndefin~tc use of 'on' It c,ulnot mean that there ss someone who 1s not rlnglrrg ,it tlic- door ) If we Js\unle t h ~ t die xlrgument dot is filled by the (covert) conrtteqmt of st~c h ,z pronotrn m the

locatlon domam, we ,mount ior the tact h i t the negat~on can only take wrclr scope. (Of course, an ex~l'inatlon h c ~ to be provided for tlir bellamour of

~riipenonal pronouns, brtt that IS a general p~ohlern that canaot he bl,med specrfically or1 the tllvd tl-reory )

Aga~nst thts vdnant of the thlrd theory, I ,rclvmce a rr~etliociol<>g(d principle. for obv lo~~s reasons of pcllslmorly, one rhttnllt liewet. postt rol*crr syrztactzc eltmefzts thctl (it7 tzo Certi~rttzc fuotk, ~i iz le~c there cue ~rrdepmtfetzt syntuctzc gr(~undsjor yosrtlrrg tJzetri 30 In the 'ra111' cme, I cl,~rrrl, tire 'dIeged 1mpersor2al

pronoun doer no sernantlc work. f i e sheer exrsterrce of a ralnrng evellt already entzls the existence of 2 locnt~ctn wlze~e rh,~t event t&cs place, hence the acldinon of an inyersond pronoun starldlng for .t locatron coil- tributes notlrl~ng 31 One jbouJd therefixe refrarn front pos~ t~ng such a covert pronoun, ~~lrless there are tndependent syntacttc redsons Lor so dorng

j.3. Aguinst the second thleorjt

Like the first theory, the second tltcorv lids no proble~~i with the negahve venlon of the weathrnennan exarrrple What mdko ~t posslble to dccotrnt

for the t~n,iva~l~bdtty of tlie wlde scope exi\teritlal rcad~ng, otl the secoslcl tlieory, 15 tbc fact that the lociltlon rclation rs construed in tile hroad sense

30 An arlaiogous pnnc~ple was 1nvr:ked by Ir'tie kleinr in o13c of lrcr icrtutes at Ecole norlrtde sup6neure (Heinr zooj). See dao Jacobsvn (tc)<)o).

3' 11.1 contrast, the ~mpersonal pronoim 'an' in Fren~ 11 'Ovr sonne' alicl? somcthlng to what 1s expressed by the inqlcrsorlal fnnns 'il rorzne' or '<a conne' Clt's nngng'), nmlely tllr ~liipl~taiion that the rllliyng everit tws a human or h~mlan .Ilk? .igetri.

Page 115: Recanati, François. Truth-Conditional Pragmatics

The problem w ~ d i the second theory 1s that lt cannot both have ~ t s cake and

eat rt. the ~dea &at 'ram' uivolves an argunient slot for a locauon no longer pemrts one to &stingunh meteoroloacd predcaes &om other event pre- mcates, when the notion of locatlon n mterpreted m the broad sense.

Accordmg to the standard view, whch the second theory n meant to protect, nieteorological predxcate.; carry a r ~ argument slot for a loca~on, as pan of therr lexacd semanhcs, rather than sxrnply m virtue of the generd fact that event? take place sometvhere. A contrast xs thus drawn between meteorologid pred~cates, hke 'ran', and other event: predcates, like 'cld,mce' (Tdylor 2001. 53-4). Even though 'dance' is an event prelcate md an event i s bound to happen at some locahon, 'dance' does not carry an arpn'ent slot for a 1o~atlo11; 'ram' does. If we elaborate the standard vlew in the nranner of the second theory, however, the contrast between 'ran' m d 'dance' van~shes. For nothing prevents us &om analyslng 'dance' the way we have dndysed 'rdrn', that is, as involvxng a covert reference to some locatlon, possibly the Earth, understood the broad locanon of the dancing event To say that M ~ r y danced, on that analysis, is to say that there w (on Earth) stme s~lb Iocatlon I / where she danced Thls captures the standad, Io~atlon rndefixute resdlng of 'dance'. In other words, that defence of the stanclard clew weakens ~t so niuch that the ongxnd intuinon ir lost. The 011g111al mtultl(>li WAS that 'rain' sentences mvolve some torn1 of reference to some rpet 1f6c iocahon of ran, m the nuwow sense of 'locatlon' By conceding tliat t h l ~ need ni)t be the case, one accepts my point that the contrast between "an' ,lnd 'dance' 1s 1ll founded. or at least exaggerated.

Pollv J a c o b ~ n has objected to dils arg.mlent against the second tlleorv b t ~t 1s too \IVC<~L xt ~tavs that the second tlleory opens tlie door to analysmg 'dance' the w ~ y 'sr,un3 is dnalysed (1.e as ~ ~ v o l ~ ~ r l g covert reference to a broad loc~tionj--thereby losxng the dlshrlcbon which the theory is meant to protect; but nohmg, she points out, woulcl legzitre t11c second theory to make t)us leap Z msa Marti, who advocates the second theory, also hlls to see the force of the xg~nietlt (I7 c ) Those reacnonr mdke me thmk that I should spell out the argument a hale more, rn order at least to explain why I find it conipellmg.

The bds~c rssue IS tllxs: r>o meteorologcal predicates (hke 'ram') pattern w t h 'arnve', or do they pattern wlth "rice'? In the case of 'arnve', there 1s

a locabon slot m log~cd fomx; m the case of 'dance', there isn't. That 1s what the contrast between (11) arid (rz), repeated below ar (28) and (29), is supposed to show.

Page 116: Recanati, François. Truth-Conditional Pragmatics

(28) A John has m v e d B Where has he arrived; A *I have no idea

(29) A John has danced I3 Whe~e has he danced; A I have no idea

The standard view 1s that 'raln' yrttterxxs wltl* 'amve', not ur~ t l l \Liarrcr3 I cla~~rrr that, appearances notw~thstand~ng, the oppnslte IS true, and X offer t l ~ c weatherman example as a coutlterex3niple to the stmdard view I'hc second theory 1s meant to protect the \tanctdsd vrrw firon1 the countcrex~nlplr, a r x l 1 argue that ~t Lids. It fads bzcau,e, ervt2 tf we tanc7pt the secorld theory. we i anrlot

Inaintain that 'rain' patterns wltl? 'arnvr' rather tl~zti 'dance', c>n the con trary, we must accept nly vlew, that 'rain' pdttenls w t h 'datlr e'. not cvrtfi

'arnve' That 1s the point of 1uy aarguntt.xlr rllc point 1\ that the monii ~ht~ovy cannot do the job zt ts trteant to do So X have nothing ~garnst the re~orrd theon:

per se-indeed, 1 am wtllirig to accept it r f there 15 ,my good resoil to do so

But that reason cannot be that, bv arceptlng tl.11~ theory. we protect tlic standard view from the countcri~~altip1e- - Ior that IS rtot tmt:

The se~ond theorv savs tfut r n the Inglc al fixmi of (2) tllerc 1s a slot for

location, ul~dmtood in the broad certst. Let'\ accept th1.r I et irc 'Lccept that, puttlng tense and acpect c~slde, the Itrgtcal t o i ~ n ok (2) is

(30) 3 e [RAINING (c) A WCATION~ (1, e)j

where 'I' IS a free vanable tt) tvhicli '1 vdue 1s cox~textudy a~ lgned 1 ct ur compare thls to the logml forn~ ctf 'Johrr arnves' 111 the CJSC of brnve',

what the ont text n ~ u ~ t provide rs the nurrt7Lv locat~uri of tlie event, si:, rr wall not do to anaiyse 'John arnves' along the hnec of

(31) 3~ [ARRIVING (I?) A A W N T (Jolm, C) A IOCArXONjj ( I , C)J

ThJs a ~ Q O --.ie& Lnd does n ~ t I~setj!i t Xi16&C1k3r t2f (23) XTfh~t xx7f: 1x7 k i l t ,

rather, 1s something hke thn

(32) 3 e [ARRIVING (e) A AGFNT (Johli, c") A IUCAIION (1, t7)J

where 'locahon' is understoocl rn the orchnary, narrow sellre A spe,lher uttenng ‘Job arnves', thus an,ilysed, a res~tot~slble l i ~ r the contextual assignment of a particular value to the &re variable 'I7, anti thrc act o ~ ~ n t s

Page 117: Recanati, François. Truth-Conditional Pragmatics
Page 118: Recanati, François. Truth-Conditional Pragmatics

W L A I H k K I( Ll 'OR I 5 I l l

acknowledge the fac t ellat ~t 1% not nlanctatory to ill1 that slot by \pecrfymg the It>cat.lor~ of the event Either w ~ y , the startdard vrew regardin, - rrteteo- rological predicate., must be gven up rt rs not true that a loc~trorr h,lr to be

contextually provided when none 15 esphc~tly mentioned by a meteorotog- ~ca l sentence 'Kaln' a lust lllre 'dmce' rn thn respect. flre orrfy dllkrence between meteorologcal predic~tes and other event pred~c ate5 is p r ~ r n a t r ~ the locat~on ofthe event IS often releva~lt when the event being dcscnbed 13 'i meteorological event, hence ~t 1s qulte typical to fmct taclt refererice to ,I

place m meteorologcal i~tter,ln~es-~?iore t yp~cd tlt,tn for other cvent p red~ca tes .~~

Regarding the p r a p a t l c proces3 at issue when the iocc~t~on ofthe event 1s

tacitly lehrred to, I have not said nt~irll m d r r i this section 1 want to say ntorc. 1 et us start with what 1 ac tudy sdrd Elrst, I \ a d that the pragrr_tat~c prclcess rn q ~ ~ s t r o n Ir an illstance of free ennchix~ent What cf1dractcn;res tfee cnncltn-icnt, hke nlodulation ~n general, 1s that it rs optionnl kn c otttr.kst to sahlir'rtlon (the contextual xslgnnrcrit of' valites to ~ridex~c,tl\ drld free van,hIes), Gee enriclpnlent rrtay or nay not take pldce, ctepending c x r tile context. 'That 1s the reason why I treat the sperrticat~on of ,L locattorl a\ arl litstarice of ennchrnent for a place may cnr nuy not he co~ttcxtually spec1 - fieci, depentimg on the co~ttext--that 19 the lessoll of the wrath~,rrn,m esarnple Second, I suggested that we corrsme the p r apa t l c e11r1t hrrlent at lssue m our exanlples as the corrtextcial provlsjon ot an extra conjunct 1x1 the scope of the event quantifier. Now that co~~stru'il IS not forced 11po11 11s. W e may t h ~ n k of the pragrnatlc ernicim~t.xlt ,it work 111 e~~11irple (r3l2) (the 'dance' caw) 111 tc-nns of a co~ttestu,ll restnct~on on tile dornam c>f the eve13t cjuantlher, lather t h m m terrrw of an extra conjunct In the scope of theit yuarrtlfier. Irtktead of analysing ( r j B ) as ( a r ) , rcpc,~ted below '1% (3j), we

could analyse ~t ar (36).

( 3 5 ) 3 c 3t 1 1 ' ~ s ~ (t) A T I M E (t, €) A DAN< 1NG (t) A M,FN1 (Jcdin, 1.1 A h I I - N I ( X l 1

( r ) A ~~~~(the-ball,] (36) ( 3 e g ) c A r l o N ( the-bulL4) (3t) [hs r ( t) A (t, tt) A i3mcr~c; (e) A

ACFNT (Jolm, e) A ar I - ~ r c a r (r)]

" I n Lialrig this ciiffcrci~cc to argue for the standdrd virw, C:orazza (aoo't: 7.7 -8) Talk to nuinta~ri the disi-~ilctiorl between sernarlt~cs arid pnpmztncs.

Page 119: Recanati, François. Truth-Conditional Pragmatics

- -- - - - - TI?, - h L T F I C N K T I P B ~ % ~ i ~ ~ S .*----

l'hus analysed (r3U) says that, among the evenh that took place at the ball, there -1 - was a danclng event whose agent was John and \-vluch lasted all iught. What is contextudly provldect here is not an extra con~ullct rn the scope of the 1 event quant~fier, but a restnct~on fbr that quantifier. 1 -

A slnular anily\is s&aighfortvardly apphes to the 'rram' case. Instead of I

andysmg ( A ) a, (26), repeated below as (37), we would analy,e ~t as (38):

(37) 3e 3t [ P K E ~ L N r (t) A rxm (t , e) A RAINING (e) A LOCATION (Parrs, ell (38) (32 r o c a r i o ~ (Parts, e)j 1st) [PRESENT { t ) A ILME (t , e) A RAWING (is)]

Here the taclt reference to a place 1s viewed as an instance of a familiw phcnorncmon. rontextual dornam restnction 'It's rxning' hterally savs that there 15 a ranlng event, but may be contextually understood as savlng that tliere ts ct~ch an event arttong the e11eu.1~ that fake place at u certutn locatton. just AS

'kverybody came' can be understood as saying that every member of ourgroup carne O n tt~a con\trual, the reference to a place in weather sentences 1s a by-product crt the contextu,il restnction of the event yuantlfier

i nlyl;elf hdvc nothing. ag31nst the contextual domain restnction account. wl-lrch I take to be ,I varr'mt ofrrly proposal (~nsofar as ~t preserves the idea t h ~ t 'ra111' Carrie5 110 ixiiphcit Iocatlon drgmtrlent) Deep down, that 1\ the accocrrrt I favour I-or I hold, with Barwne, Kiat~er, and others, that every utterance 15 evalu~ted agalnqt 2 'toptc situatiorl' which the utterance con- cerns ~ i l d the context yrovlcfes (Recanah 1996, 2007). 111 simple cases, the topic situation is what restricts the donlam of quant~ficat-lon So the seman- t ~ c conterrt o f 'It n raining' 1s the proposition that there is a r m m g event,

but tfus 1s uncicrctood as a comnient about a parti~ular utuanon, for example the present 5ituatlon it1 Pan,. O n th~s vlew both contextual donlam relitric- tlon and the bcit reference to a l~catron in weather sentences are Jspecb of a very generd phcno~nenorr, v i ~ the tacit reference to a topic situation aganst whlch the utterance is meant to be evaluated.

Re thnt a c it may, as coon as we construe the relevant form of ennchrnent as related to contextulif dornaui restnction, we face a sigxificant challenge. Quanhfier donrain restnchon, in general, can be understood rn two ways (Kecanati 2004: 87-8, 12s) On one analysis, it is a matter of free ennch- rnent W c (optiondly) make the meanlng of the sentence more speclfic by rertnct~rtg the domaln of quantificatlon, on pragrnatrc grounds O n t h ~ s vicw, argued for by I(cjnt Bach (zooo), the l~teral reading of a quantifica- tional sentence is the coritextually unrestricted readmg But contextual

Page 120: Recanati, François. Truth-Conditional Pragmatics

- -- WFATI-1I.K R F I J O R l S IT{

doman restriction rnay also be treated w all uibtajrce of saturatlon Manv theonsts beheve that quaxztlfier phraws carry a dornatn variatde to whrcl~ a

value must be contextually asngned. If- that 1s so, then treatmg the tacrt reference to a locatlon as an r~rst~trrce (or by-product) of contextndl c l t o r ~ ~ ~ r r l

restncbon would not be a trcatrrtent ra terms of fiee ennchnienr, but In terms of saturatlon; and tlzls approacl~ arguablv Leads us to sorilethmg v e q

close to the Marti-Elbourne view Let us assume that quant~tier phr~yes are indeed assoc~ated w~tl-r do~x-r~~n

vanables whch ~xiust be contextually assigned valtles, a\ sortie cem,rlltrcr\ts

believe. O n th s view the alleged rrnrestr~cted readmg-s tvheil 'every

body' a interpreted as medi-i~ng ez1er)lbody ut tk~ ~.~ i l (~ l t? tgorld---results froxn assigning the niaximal domain to the donlam vannblc. 50 that rzad~ng 1, not

the 'literal' readlng m contrdst to the prap~~ltic-aUy enncl~etl, rccrr~c ted

rcadlngs- ~t 1s only the parhcular cjkre 111 which the contextual rt'stncbori turns out to be empty becduse the rl~asin~dl doina~ri is selecteii If wc take

this hne, what w& prevent us horrl savrlzg sonrethrng sinirlar m tlrc krsrri" case? The tdcit refererice to a place wl~itli we 111 typ~cdl ~ ~ ~ e t e o r o l ~ ) g l c ~ l utterances wlU be ra~cl to result from t h ~ c~xitextl~al ~jsig~unexlt o f C ~ value a v,anable In logical f<xrn, narnelv tlie donraitl vanable wh~ch the event

quantifier carrles. The dornaln vanable rs ,~rs~gme~l the sct of c*vc~rzts tlli~u tnkf

plae at such urzd ~11~12 Q lo(atz(itt. Tho rxldeiilr~ite ctr esrstelrtld reading c.)bservrii in the weathemian case w~l l then he ~ccounted h r a\ follows In thr5 parbcular tax, die domarn variable 1s assigned fhr sut o_f event.; that hkr pirice. . . aflywhere on Euvtlz

O f course, the suggestion that the c ovcrt ovetit qrlant~fier bei if carries ,1

covert domaiil vanable seem\ strange Che is tempted to argue "thnt onlv overt expressiorls can carry covert elerrients But we don't have to thublc trl the contextual restriction of the event q~xantifier in ternic of htd~er-~)rdcr covertness. If we hold, as seerrls reasonable, that whenever there n quan~ifa- cabon the context must provide a relevant ifomam. then we wdl .iay t h t

whenever covert event quarit~ficat~on occurs, the context rtiust pmvlde a

domain of events over w h c h to And that means not that the

covert clualltifier carnes a covert vanable, but merely dzdt the covert quantifier that lias to t)e provided 111 the course oflnterpretmg the sentctrcc is itself a restricted quantifier, where the restncbon depends upi>rl the

context. So tfie covert event quar~t~tier 1s a\soc~ated wtlt 3 vanable ~1otn3~11, rather than with a doman vanable 'I'lris makes no digerence to the present

Page 121: Recanati, François. Truth-Conditional Pragmatics

argur'ricxrk l hc \\~eatlrclm,ln esdalp1e c'lr l snll be haridie~i by saylng that h e cltrr,larn t l u t 14 selected rn thrs partrc~~l~iir c , ~ 1s the "rt of c\entj cones-

pomhlrq to {lie rx1.x i i ~ n d l 10c~it.ion. f riictlt~arncci CXIC ,itLiatrc,tl tfreorctri vrzw, accord~ng to whrch contesh i~ l

iic,rn,,iri rihstl-lc Iron r \ by-pro<'Iu~t of L ~ C " trlc~t rcfe'rcnce to A topic \ltxjatlon. I >oe\ tiits parlrc rd.sr t ~ k e ctm carlte~tual iiurrz,srn rcrtn~tron avoid tlle prob- icni I h ~ v c i t l e t wr\ed tor rnv dicoiiirtj 0 1 1 ~ I Y U ~ ,UI;;UC: that ~t d o a not, t'Or tilt= fi~ilouxnp, rc'norn Evcri ~ f w e hold, '1.5 1 (la, chat the toprc \rtuatlon Ir nut r t p ~ \ e n r e d i r i the wnlantlc cc7ntent of-the rrtter,mce, hrrt only come\ into

d ie pic ~ L ~ I C i~ ttl(- 'I IT ( ~mi\t~llilce' rcl'rtrvc to tvl1lcl-t the cor~tent IS ev,tluntetf, \rill r t rs ,I net r e s a r y t onrcrtuent of the glollal, 'Austwl'~n' proposltton e:*pi-c\\cii b\i the utterance, whlcb AUSCII)I.L~Z proposlhC)il goes beyond ser,r,i~rtlc- cc,ntc"rlt pnjpcr (the lektorr) and mct)vor,ite\ the contextudy

proviiivd \r~uxtiorl \ ~ t ~ c e ,I tctplc \rtri,ltron hdzs to hc jrnnpz~it'ti in the course

ot'c"t iju,itli>g the t i t t t ~ , ~ ~ ~ c - t ' . lt IS yos11hle tt) 1~1'1t11tain thdt, 111 certan \erne, t i i t . L O I E ~ ~ ~ X L L ~ J ~ ~ S O V I ~ I Q ~ ~ 01' tlidt s~tri,thon I C i t r ~ ~ ~ t t e r of (mandatory) satura

iron r,itI~cr t i i ~ l i o t (i>l>tro~~al) eni-lchni~ilt A I I ~ if we say t h t , then the piciblcrrl isnic-, in die \axlie terixl\ '3s J>cii,r c.

f o \urn lrp, ,n soon ,a wr treat the eorrtextirnl provlstotl of a place tn W C ~ L ~ ~ I C I \CIICI 'XIC~Y 111 terrm of coiitext~a.i( cloma.irl re5tllctinri, we open the door to Jrr Lirialv.il$ of' r ontcxtl~il pl.~ce \j>cc~/i(-, i t~o~j i l k teim-is of baturahon r'1thc.r t b m crrxrchrr~ent Such arl ,u;lalysn contr,rdlctz rtmy own propoval and iicnicr tl-ic opl;ivil,d ci~nra~tl 'r of pldce spe~lfic~rhon O r r o lt seems.

P5ut dot,, ~t r c~ l ly ; I en nut 40 sure Even ~l-'.v\re t,~he the t'trrt reference to '1

p h i t- rn v+r",ttircr scrr?i-.nt e\ to be hy produc r of( ontc?ctn,rl tlorrlarrl restnc- tion, JIXC~ \ $ ~ I I L I ~ L , I I ~ C * O I I S ~ J ~ ~ o n \ ~ t n i ~ ( onLext~r,~l cic>tn,irn rc\tnrtlon as ,m 111-

ztbrncib of S,itLIrdrioli, I t h~nb rrly p~ i lpo~a l I \ [lot c'Ict~p1y ,~ffcctcd, appearances

irot~vil.hsr:lr~i!111&: I hat rc, \o fix tmo reawtns F~rrt, the (posslbiy empty) rc\tn~bori ir l tllc clorr1,un of the wen t q~r~ln t~ t lc r ha4 to take plnce 11, izEl

c'iws, ~ I I ~ ~ " t l l c r the prCcficdtc .lt ISSUC I\ ~ l l e t e ~ r o I ~ ) g ~ n l prehcate or ,my iverii preclr~,lr~ (c g Q~nzc'). \O wc czlrx nlaiiriLdlll that ' rd~n' pdttenls wtfl

' i ian~c' (rather tllarr wlth 'nmvr') Second, the v6tn,ibfe to w h ~ c h a value

" 1 h1-i ic ~ctn:)llv ki~batalrle, bccnt~se ricrihillg In ihe sc.nlcnc.e swritk for the sltuahoil of evaluation. i'ltc top,< sitiratrori rs t t n a r u c ~ ~ i ~ ~ ~ c d , cven rt it ha to be provided in order to y1e1d a complete Aust~nim p r ~ ~ ~ ~ o s ~ x c i r i ; s o ii., (orrtextu'tl irrovlaon m neltlrer a lriaLtcr c>Zs~n~r:li io~ i nor t~i'cnnchtnent. In general, I ttimk notltira bkr '\ilinratiori' and .erirr~,krtc~c.iil' only ,~pply to iltc irkitin; they do not apply at the :\ostini.~ii propocitiori !eve1

Page 122: Recanati, François. Truth-Conditional Pragmatics

rrmt be contexhldlly ass~gnetl 15 not a locatnotl vanxhie, but a tfonidin Only ln certzn cdces, detennzrred on pragrri,itlc g~our~dr, wrll

the contestxtal restnctlon of' the event cjumtlkrer take the form of the conte-ctnd spec~fication of '3 ~t l ,~ce It rr tlterefjre ~~ossrl-tle to rrr,llntarn thnt

rneteorologxcd pre&cate\ do not Larry an arguruerlt slot for a location, or no rimre carry such an argwlnent clot than other event pred~cdtes tto Whenever there IS tacit reference to a locdtion, ~t takes place for pragmatic reasorr\ ,xnd casts no hght on the semantics of meteorotoglca1 predicate\ T h ~ t is r~ry

polnt, and it rernaim.

I conclude that my proposal IS rtot threatened ~f we to an ~ccourrt In trnns of contextud domalri restncnon X wlll tllrrekx-e rel-1laln o f tc~,dly ag~~ostic abo~it the two usues I helve ralied In thlr section

* whettier the contextual \peclticatIon of,^ plxe is berr construed ar the

tacit provlslon o f ail extra conjunct irt the rc ope ot the evcrlt cju~nttfler, or as a by-product of the process of col~textirallv restrl~tlng the dorltarrl

of quan t~ficat~on; whether contextilal domaln restncnon itielf is best constl-tled a ,111

msance of fiee ennci~ment or ar an inst'tnce of ratur,%tlon (or as tlelther-see footnote 33 ) .

A$ A referee put ~t when I subnrltted ,m dncestor of- till\ ~ h ~ i p t e r to l d r r ~ I g t i ~ ~ t ~ ~ ~

(2nd Pi~llocoylsy, tt takes work to get a location-lndefirn~r rexcllllg fitr 'ral11' the weatlienri~rl sceriano n a '~levrl-iv t o~istruc ted rce11;lno'Thts rases ,in

objec tx~ '~~ I have heard n~ailv tlrl~es slnce the puhh( atlon of rny p q e r 'Unartlculatea Constltuerlts' (zoo?)

The objecholl lull\ a\ follow\ If whdt tr ,rt iscue rs the contrast between 'ralr-1' and ather predicatec whicll do carry '1 covert arg~tnient ~ r t d therefore cannot be given an mdefin~te reading, then rt I \ only f ~ t r to ,fik for \cznle

exernon of sirnllar cleverrless wltll respect to thern. Now ~f we do try 11,lrd enough, then we can provide ~ndefirutc readlngc tor those predicates wh1cl.1, according to me, requ~re the context~ial proklsron of 'I definlte ,irgument to fill a slot set tip at the lexical level My bvounte example\- those whit h

34 Or, l fwe eschew talk ofvanahles for lie reasons nremoned al>t,ve: what is cnntex~u~liv ~itvvicled 1s not 2 vari~ble i t ) ~ a n o ~ l , but a v.triahle dori~aui {or the evrt i t q\lanirfirr.

Page 123: Recanati, François. Truth-Conditional Pragmatics

t r i j - % 3 % f z S A ' - f T v E ? m m m nccrwmr s

ticulated Conshtuents'- are 'not~ce' and

'fin~dl'. If l cay :fohn nobced', there must be sonietlrulg on textually salient

and identfiable such that ~t a that thing w l c h John 1s s a ~ d to have n o ~ c e d .

'I'ile wxitence ~ ~ n r ~ o t be understood in the indefin~te scrlse: John noticed somrthirg or other (or so 1 clain~ed). h d the same thtng holds for 'finrstl':

yohn has fixrrshed' cannot be used to say that he has finished something or

other. Now, accttrding to the okgect~on, we call get such 'indefirute' sead-

Ings if we try I-tard enough atld exert as much cleverness as I d ~ d m cc~nstrlirtlng the we~therrnatl scenano

1 he first person who raised that objechon wa\ llma Fra~ia, then a student

of Sxniiro Lucclu. She came up w ~ t h the following sceximo for 'firush':

We are ill n factory. The i k t o ~ produces tinned food and is equipped with a specid il~arhrrir that carries out dl the activities concerned with the producnon. 'The spi~cial inraclG~re Inanages hundreds of activities, from the preparation of the fbod to the prilrtrng of the labels for the cans. In order to save time, all the activities are cnrrie~i out in parallel. There is a mile that says that the hctory can't be closed until tlic machine has finished at least one of i t s daily activities. Wher-1 the 111achiue finishea arr activity d bell rings. (Since the rnaclune does inany activities in pardel, it IS possible that lalore than one activity is finished at the same time.) NC~W, irnagine stirneorrr lieanrrg tile beU ring and uttering: 'Ok, the machine 1x1s finished. You car1 s~vitctl it oEand go hot~xe'. 1x1 this context [the sentexico 'the machine has finished'] rxtearls that tllc machine has finislied at least one ofits activities, but no activity in particrtlar. (Frana 2002: 18-19)

A while later 1 received two add~tlorial scenarios horn Sarn Wheeler 111, a

ph~losopller of language at tllc University of Connecticut:

Firlishinq: "llherapists are monitoring a large group of patients suffering from Fred's

Syjyrlclrcro~lle, a yatho1ogm.l tendency to start projects and never complete

thern. A new drug, Completirx, is being tested. Patients are monitored by

pdc$uate students, who push a button every time a patient finishes a project.

'Patient #271 has finished,' says the researcher, looking at the comole. 7ohn

has fi.nislred7 can mean 'John has finished something or other'.

Riottnnr

Another b n d of psycholog~cd dsorder, hyperconcentra~on, In which

pahents thirrking about niathernahcs far1 to regster shmulahons &om

Page 124: Recanati, François. Truth-Conditional Pragmatics

their environment (a kind of mathemn,~t~cd coma) Chce again, rewart In- ers x e teshng a drug, Whauthahn, tha t enhances ~ ~ o h c i n g A vanery ol

snmulattons are randornly ,~ppilet.i to a group of patients who llave becn given Whatsthatm. In a iew pattents, neur,xl s1gi:llals m&cate that they 11~1-ve notlced. 'Patient #271 ha\ notlc ed 'Johii ha5 nohced' Eecnls to he able to

mcan 'John has notlcecl sornethtrig or other '

Flndy, a reieree for 1,zvgut~ttn tznd f-lhz/~)roplty joined the cho~l

Consider a scenano w ~ t h a patierlt who h a beell In A \enn-coi~la, i~nd a

technlcilu~ 1x1 mother roorn rs reacflr~g the output of an EEt; or whatever it is that rnesures bran actlvlcy In vdnouc are&\ of the bram It secnls to

me that a trarned technx~an co~lld know when braln ~ctrmty signals 'nohcing', and sinLe for the \exllr-con.*a patrent, tlie tact that he" snotlclag (somedling) is dl that's srnportmt, one nligllt inlagme the techmc~ari being able to shout 'He'c not~cmgl' w~thoiit being in any y o s ~ ~ o n to

know or say what ~t that tlic pahcnt n noticulg

One may finci the exan~ples more 01 less convincing, but rt 1s not u ~ y intentron to quibble. 1 concetie that, by exerting enor~gh clevernesc jai the referee pi~ts ~ t ) , tt IS poruble to conre up wit11 occurrences of 'fimzh' or

'notice' for which the xgumenr slot rs not filled with A spccrfic value provided by the context, but IS exr~ter~t~al1v bourld Such oci urrenc e\ src

nlargmal, however, and I th~iili the\ can be handled bv saying tE~dt lrr ss~ch cases a I ' P I ~ N I Z Z Y ~ ~ ~ll!ft ocCurs- the word5 ;ire t , ~ l \ e ~ ~ ln A spec 1~11, ifcvli~~it mrse

In the ueurologtcal ex,.1111pLc.; provrcllerl by Wt~eeler and the reieree, Instead ot gtving to 'notice' its st~tlJ;irij vxlue, ndmelv:

Ax Xy l i e [N;)TIC,ING je) A AAGtNI jt, e) A TIIE.ME ( y , e)]

I think we grve it a dnfted value t~lXored to the context, r~mtcly

AX Xe [gy [ N ~ T I C I M G ( P ) A ACLNT ( A , e) A TI~LMF (y , e)]]

The cfi~fted v ~ l u e of hotlee' 1x3 t lmc example\, or of 'finlsh' In t11c other examples, 1s easy to account tor it c m be obt,uned bv applying Quine's Iler operator (or a close relat~ve) to the stancl~rd value. As I sad 1x1 %eectroil r 3

above, Der apphes to any n-p l~ce predlt ate and ylelds a n-l--plzice predrc,~te by existentially yuarltlfylng the l ~ t argunlent-role of the ongnal predicate Uer conmbutes what, m 'U~idt~ct~la ted C:ox~stltuellts~, 1 caUed a vanddlc

Page 125: Recanati, François. Truth-Conditional Pragmatics

frrr~ctrorr a ftirlc tron t11,it rrlcrr,i\e\ or detr eawr the acbcrty of a predicate. F.~lait trim\ wk.rlcl-r, ilke that ~oi lh lb~i tcd by I)fxr, dct sedsc the ad~crty of the prctlriatc. ii call (itrllo-cvir~g I c\r1rc21c) recesv-vc firnctions. The meaning >hilt

effec tc i t I.iv Ilvu n ~ l v itsell bc c ,lllcti ,I rcciS\\lvr 5I1itt 1x1 Ore c , n t ~C'cdt ' too, I ~ ~ ~ v o h c d thc I l ~ r tqx.x-tltor to st ount for the ,hilt

fioa.i~ clrc t r ~ ~ n \ ~ t l " v t v o the lntrdrlsrtrvc usc That h i t 1s leslcall~ed (conven- tiondJircd) 111 d l c cdse of lntrarlsttive 'cat9, but 1x1 the Lase of 'notice' and 'ki-lrnrsb' tire vex\ san-ic rries\ive s i z l f t t ; l i t be tnggctcd conteuttldly. I ikewrse, \C)IIEC I~ I~LOI IVEI I IC 01 rtietnphonr s h l h dre Iru~c-ahzed, but others aren't. I,vc.a, 11104~ tlidt .ITC ICXICA~IZC~ rnust have \tnrtcd tIie~r t areer as purely

cirntt u t ~ l d irlicanrtsg s t ~ ~ i i s A\ the I rcrrclr I I I I ~ L I I S . ~ E r~~ l l e Renverllste orlce put r t , pdrdphrnrrrg rile famous errlp*rrcl\t d ~ c t ~ i ~ r i , 'rttrhi/ c75t in irngttu quad nor1

p t ~ r ~ > )I;(L"YII 111 I I Y ( Z I I ( ~ I Z C ' So ii d~)n ' t t111nk we h,rvc iinv tiorlble accounttng for

+pecr,ii wrr of "tlrr~\h' . r r d 'ttsot~ce' s u ~ f i CIS thore t h ~ t occur In the exarnples abtrce, arid X \iorr91 ~lllnli that thev tt~re~irex~ rl ly c larns tllat sdturatron of the

relevdrit 'crgrnrnexlt \lots 11s nr;rrrct~tory 11% the case uf 'finish' and 'noace'. Wiilxcu 1 IAV t l u t i t ' \ ~ l s d ~ ~ d ~ ~ t o q ( ~ ~ t h e r thCl~i o p t ~ o ~ ~ a l ) 1 n1cdt1 tlut it I\ n~ade

1>13lr~irc>rv Plv rlac Iexical wrn,riltlcs of 'rrot~ic"' ~ ~ l c i 'fliusl't', thdt 1s. by the~r ( O I I V ~ I ~ ~ I O ~ ~ A ~ lnraisllig or ~t<ul(iard ~ e r ~ ~ ~ n t l c vdhe It is not my irrtention to cieiiy that those Ic*~c,~l ttrrrls C'LII , ~ l s o be gven nc)n\tand~rd srmantlc vjlucs rf tlar t ontcxt tr1gger.i A prapnatrc lirrlct~c-ir, tesnlt~ng m A rrieanrilg shlfi.

W3~le rr~arrdat<ln: from the statldporrlt of the aaxrdard value rriay no longer he 50 w h e n we c . i t l ~ f i 1 0 'L non\tandard vdtr t , (011 metlrling shil'w ln general, \ec Nurrberg ,in(\ A,iciiert 1992, Niinberg SO,, IGrtee 1098 343--51.)

Once I have 1x1 lde t h s t oricess~ol\, htrweve~ , I hdve to answer the crucial ( h j c c t ~ o ~ ~ wk~y 1101 \dy the 5'1nic thlrlg 'about 'ltq\ r c a ~ ~ i ~ n g ' 3 Why not say that It is ~rl;irlJ,rtor y to !?I-ovrde a vduc fh\r tlte locatlon parameter, unless a

I ~ C ~ S S I ~ J C drrf t . oc cur\ ,raid 'r'un' cc-tntnbutcs tlsc pred~cate Der (WIN)

ratl-rer tildri rlrc 5tatidarrci prcd~c.ite tcnr~; 1x1 vicw of the pos\lt)rllty of such ule,arrzrig .ihrhs, the \we~tl~err l~ai r ~ ~ u i l p l e docr iiot force u5 ro g v c up the

stL~ndart4 -clew, cortling to w h ~ t h 'rxrn' carr~es ,m argument slot for a

iocairtrc~t~ tie weatii~errn'~~~ example c,jrl be handletl by s,lying t h ~ t , In that qxud (ontext, 'r rrreartlng sltlfr t&cj place atid ' r m ' does not ~arunry I&

standard v,riut

I agree t t ~ t t thrr ic . J pctsslble triove to nlake -a poss~blc tlleow to tlold. So 1 graat tlrdr two d i ccxirxts of tlie weatklex-n~an example m ternzs of free

ys'rgnr'ittc ~ I O C c s ~ ~ ~ i i c 111 coXXIpetitluti ( ) r 1 ~ JCC ot111t I\ ~neo~llpatlhle w ~ t h

Page 126: Recanati, François. Truth-Conditional Pragmatics

W P A J i l k t i K F V O K I C X i < )

the stmclnrd view It r'iys that the lot atlon-\pecrfl~ rc,*ci~xlg> :!,of'lt's r,~inmg'

result from a process ofenr~chmrnt, cvhlch &es place moft oftlle tlnle hut riot dways (as the weatherman example shows) The other ~ccourrt rs compatible xwtb the standard view. It says that the ktcatioo-lridefifimte readmp of 'it's ra~niaig' (e.g. the weatherman c~,rxnple) rt.sult frorri ;L free

pragrnatlc proces5 of xneaxllng shlft, 1x3 v~rtue of wtvch 'ran' contnbnres tlre preCIIcate Der (RAJN) rather than IWN to the (mod~fied) log.~c'd fo ln~ of tire

serltence < I l ~ e two cvr-tipetmg accounts have \orrrethng In conrmon In both cases

we ac t otlnt for the lntultive truth-condltlox,~ of a r,irkge of euarr~ple, by dppeahng to free pragmatic processes that 111~1~ tllre barc 1ogc.d Ion11 of the cexitencc to ~ t \ modified lngcd form. 'I bus the first dcc ount appr'ih to fret' enrlchn~ent to go fro111

3~ 3 t [PRESENI (t) A ~ I M L ( t , e) A RAINING ((')I

3f 3t [PR~SFNT ( f ) A TIM) ( t , C) A IIAININC; (I,) A rOCAI1ON (l'arI\, ~ 1 1

or possibly to

(~P:SO<:ATION (Paris, e)) (31) ~ P R E ~ E N I (t) A ~ I M F ( I , iz) A RAININ(, (?) I

other accourit apped5 to a recessive &in( tltrn (correrl)ondulg to C)ume7s Drr operator) to t&e us frorn the <allegeti standard denot'ihol~ of 'r,i~rz', v r ~

A1 he {RAINING (e) A LOCATION ( i , L')]

to its alleged nonstanciard denotation m the context of the weattrerman

scendno, v17.

he (31 [ X ~ ~ I N I N G (e) A LOCATION (E, e) ) ]

Everr though one Account, but not the othex, 1s ior~ip~tll?le wrtilr the standard view regarding meteorolog~cal p~ed~c~ i t e j , 6-olna general method-

olugcal pomt of vlew the t ~ v o X C C O U I ~ ~ J 3re very siurr~l'lr Both ,ice cpt thc emstence of what f<mg dnd kinley (2005) cd1 'jtror~g pragrndttc cf3ectf7, that IS, effects of context on truth cond~tions that are 11ot tnggere~l by the

I i n p ~ \ t ~ c niatenal (e g by ~ndex~cals or free variables In iteed of contextual v~lues) but are pragmatic through ~ t ~ d through S i~ch prdgxtidtlr efjrects are

Page 127: Recanati, François. Truth-Conditional Pragmatics

responshble for the d~fkrence between wtr~at I called the bare logcal form of an utterance arid it^ pra@mtically modified lopcal fornl---a &fference which has been a recurnng therne in ths chapter and to which 1 will return in Chapter 4.

1 have jrlsr coriceded that the weatherman example could be accounted for in ttXnrx of vmadic silifi, and that such an account wodd be compatible with the sta~~Ct$Td view regxdmg meteorologcal pred~cates But ths sug- gests, wrorigly, titat an accourit in ten%= of vanad~c shift n an alternative to an account m terms of fiee e~mchment

Actually there are ttvo disanct issues One issue concerns the directlon of a n a l y ~ z ~ Whrch reading of meteorolog7cal verbs n the bare, hteral reading the locatrorz-specific reading or the uncornrnon locauon-mdetirute reading ~llu\trated by the weatheman example? O n the recessive account, the literal readng 1s the lo~atron-specific realng, and the other readng results from a free pragmatlc process ofrecessive shsft, through which the verb is assigned a nonstanddrd sem;rntic value Il'liis w just lske the neurological uses of'notlce' 111 the above exan~plcs On the kee-enncbment account I have orxttined, the terb does not rdny a Iocatlon slot OD I ~ S literal re'idrng, and it 1s only tllrougb the pr,lgrnanc process of fiee erlnchrlle~lt t h ~ t a locaaon is contex- tu;rlly 5pciitit.d tor the meteorologcal event So the direction of praginatrc ~ndlv\~s I\ reversed But this irsue pertaining to the drrect~on of analysis 1s

itlstrnct from, Indeed orthogonal to, another Issue, whrch concerns the 17ragvrzutzc nleihtlnrrvn tat stake or the tools to be used in the analysis T11us ~t I\ possitrtle to b u ~ l d an account that takes tile loca~on-less reading a baic ;rrd the locat~on speilfic reaciirig as pragnatlcally denved, wrh~le appedlng to the nutrorl of vdnddic qhift to account for the pragmatlc denvatlon of the latter from the hnrler Suclz an account would be a parac~ilar unplernei~ta- tlon of the free errnclinlent account I h ~ v e outlmed, rather thdn a coin- pen tor

Vanadl~ shifts can be erther recessive (1 e adicity-decreasiiig) 01 eupan- sive (adicity-mcreasi~~) 7'hus we can introduce a vanadic operator, h c ,

that dddc a locdtion argument-role to the argurnent structure of the lnput predzcatc If we take the t3asic argument structure of the predicate RAIN not to carry such an argument role, then the predicate Loc (RAIN) will c m y such

Page 128: Recanati, François. Truth-Conditional Pragmatics

an argument role. The variadic 5fufi esected by LLTC will take txs frfrani the. detzotation of 'rain', viz.

he [RAINING (e) 1

to its slzifted denotation under 'Lo(.'. viz.

Xi Xe [RAINING (e) A LOUTION (I!, r)]

Is t h s what we want? Not yulte The ord1i1a-v ures of 'It is rannng' tacrtly refer to a specrtic place, and &we w,igrt to L ~ c ~ o ~ i ~ l t for such tacit referenccvoo pragmatic grounds, we have to sorrlehow incorporate rt mto the nteatirng shlft The relevant shlfi must do two tlzmg\. (1) add rui arpument role to the argument structure of the input pred~catc, and (11) fill that role with 3

speclfic value provided hv the context Thn titxal role is cliar~cten\t~o ok adverbial and prepositional phrases on McCorrnell (;inet\ accr ouilt

(McConneU-Ginet 1982). So d I sdy that it 1s raining irr E'aans, the phs<~s,rse 'in Pans' does two thngr add a locatlion dmrc~isroa, and specify ,1 ~ a l t i z (Panr) on that dr~nenszon That can ALSO be done ~niphcitlj, ttuough coa- textual clues In such cases the pragxlittl( process wh1~I1 enncbes thc bmc rnean~rig of the yredlcate by specikrng tlie location of the evertt (3x1 be represented as a vanadic shrfi accon~yallted by tlie spc~lficatlori oi a ( nntex-

ma1 value tbr t l~e added argumer-rt iolr In "Uomt~c~ilateci C:onrncuunrs3 have, to that effect, decorated the PAX operrltor with a \ub"rr~pr ~orrespond~ng to the contest~~ally prc,v~.iled locanon

[[~.ocI~,, (RAIN)]] = he { I W ~ N ~ N G (e) A XOCAilON (Pans, e)]

V e see that. through that sort of meanlrlg slldt, uTe carr erinch the logxrnl form of an utteraxe thmugh die provlclon c>f &XI ~rnphcjt locatiorl fot the dercnbed event Far from belng an dtern~tlve to the hee eil~ichl~lrxlr account, a rneaiiing, stllft along those lirles c ollstlcutrs an impletrienwoon of the Gree enrichment ar count.

Thls mplementaaon o t h e r ennchrnrlit ( '~1 be reen a\ ~lilcernat~vt. to

tlie standard, Perry-style r~i~plei~ientat~on In remx\ of hunamculatcd onstl-

tucnts'. According to Perry arid ills followers, un~irnct~lated colishtuents defeat what Perry calls the pnnc~plc of "homomorph~c represerit~bon' (Perry 1986/2ooo: 174) and Cmnni~ns the panc~plc of 'full arhculatton"

Page 129: Recanati, François. Truth-Conditional Pragmatics

A ' i ~ l ~ l ; I X l t I C ~ ~ ~ S I I I I I C S t/13t 2 statriixent is iirily articiilatcd, whcn each item that it uses to gciicr,mrc thc content of thc sctltcncc (cacli "input' to the cornpositsorl nrles fbr serntrmces oi'rlrat kixrtl) is itsclf thc coilccrlt of'svrr~e expressloll within the sentence. For rxnir~ple, i t is piansrble that ~1 use of tht- sencrncc ?<ex is now scratcllitrg' is fully ~aiculaccd. 'l'irt. prcn17c)sition esprt:sweii rs gt.rier:tted tiorrs R.es, the property of \c.u:rrciini~g, anti t i le trrnc. of the st.rternrtrrt. Anti f i x each of these builtiii~g blocks' ofc-lri: yrojrr is i t ror~, there is ;ur cxpresslon 1x1 ihr seirtence with that building block as

rts c-viltrrit ( ' ~ K C Y ' * ' ~ i ~ r ~ t ~ h l n g ' . 2nd 'now'). A priric.aplc of c-o~tipositioxralicy satisfies clie c oilstll.;flilC of'fi11I :ii-tlcti13tiol1 ifit erlt~i1s that cve1-y stcrtcnrent it i-onccn~s is fully a~~ic i r lac~i l . ((:rrirurrlxis rc)yt: 10)

A scxlteircc s ~ i i l ~ as 'It's rai~iixrg' (siritl while t'lc-itly referrirlg to 3 particular loc;rtiou, r-g. laat-is) is said to vio1;itc firif :rrtici~lation bec:iust% the propositjot1 that i t cxprersvs in coxrtext involves ,i p:-articular p1;icc to which r~otbing i11

r l ~ r ccxrreracc. corresyronds. 'l'be place tacitly rvfkrrccl to i s an '~~narticulatcci i:c>rrstit~itsnt' of the propositiorl zox~textually c:xprcssed by the sentence, since

llotl~irlg i x i d-re serlterice specifically starltls f ix it, But on the variadic-shift irtiplerricaitrxtiori offrcc. t~nricllt~ielrt, then i s no violatioll of f~ill articulation.

WI-r,rt e i i v prc.q>osztion c:oxltains is efic pro~xrty of c:iiriing--in-I'afis (or t l ~ e property of E~eixig -a--raining-cvcnt--ili "Itiris). ;tad tilere is an expression in the selitcrlce wit11 chiit property as i-ontclir, rr;xxxleiy the verb 'rain' wtzich is assigrr cd that crirrt.hetl cotlterlr: 11s a resrilt of r.hc y r:igmatic shift.

Ofcorrrse, r;hc property ufbeing-a-rairring--cveilc-111-Paxis car1 be analysed into the silrrpler properties ofbeixig a rainirrg evcxlr and tahng place in Paris, bill. that iivc:xs not lalakc Pxis a constitueric of-the proposition, in violation of full ;irtic.aiP,itiori; li,r i f t ha t wtxe the case, tlrerr 7R.c~ is sc.ratchirrg7 would not be firliy ;rrttculatet.f eitlrcr, colrtraxy to w l ~ ~ t Cri r~~rnins asslimes. The property ofsiratclrix~g cnri icself'lte ar~alysed into 3 nirrnber of simpler properties, like the property o f t l s i ~ ~ g olke's claws; a t d these is ncitltirig in tlre sentexice "Rex is sciatclrirrg' tlcat syrcificaUy starids fix the chws. Ilocs it r-ilcrke Rex's claws an

rmanictabixts_.ci c.orrstitcrcrlt o f t h t j~fiq~(fi i tt ic>~l cxprcsscid by 'Rex is scratching'? Obvitr~rsly naot. I f i t did, hardly any statcnrvnr bvorild be fi&y articulated.

Ar thi. orld crl'Sccdoti 6.1 1 said that the coxrrcxtttal specification of a place rrrny E-)r a,onstnreti either as the tacit provisior~ of' an extra col~junct in the

scope c\k'the twerit ijilantiticr, or as LL 'I>y--produ~t ~f the process of coxltextu- ally rrst~<i.tixrg rkre d o ~ r a i n of ci~iantificacion (or of contextu;rlly specifiivtg the

circ~ia~ntane,r of ev~d~i~i t ia r~) . 1x1 the prcsrilt sectir~rl I have suggested a pzrticrrlar isrrpicnrer~tarion of dtc forrncr aptioli. 'X'l~r extra coqjunct is

Page 130: Recanati, François. Truth-Conditional Pragmatics

~ntroduc ed in the scope ofthe event quaritltier by lexrco-pragriant m e n \ it

is because, rrk context, the verb 'r'iin' corrics to mean nlirz-tn-l'drrs that tltc locatlon of the ralrilng event ends up bemg \pet rfied and part of the utter- mce's truth-concfitions Since the locat~orr gets 5pec1fied as part of the contextual meaning of cl partlculal lexical Itern, the pnxlcrplz of €dl artrcu1~-

tlon IS not molated. I fowever, it may be that the best a~ialysu 17 the situ~tlon tlxeoret~c analysn,

for whlch the locatlon of ram is prov~ded 111 the course of specifvmg tlte srtuatron wlth respect to whcfr the ntterance I\ lxlearlt to he evalri,ued On that alternatwc accourit it niakes sense to say that ttlc plat e IS an tlrtctrtlc ul,~tcd

constituent of the (Austiman) proposltlon expressed by the utter,uxe According to the Austmra~l account sketc lied In Sectlon 6 r , ,mtl elabo

rated m Kecar~atl (zoo7), the place of r an 15 an aspect of the coplc slttlation,

'md the topic situation 1s not ar t~cul~ted by anything in tire wntence, nor I\ i t

part of the content that 1s evaluated wltli respect to thdt sitil'mon. No place n nienhoned 1x1 the content, which is sil~iply the proposltlon that. there 1s a rairllrlg event, but thclt content IS evalrtdtccl with rebpzct to 2 pa-t~cular

sttuattorl (tnvolv~ng a partrcular place L'kns) ard the Lrtterdr)ce, theretote. IS true if and only ~f there IS a ranlng event in Panr 1 hc global, Aux~ni~sr

proposition ulvolveq a particular pkce, but the content w111ch ( ' ~ l o n ~ ~ i d e the ntuation) is a consntucnt in that Aust~ruan propo.ilaon does not lnvolve A

place. In thls framework the principle of full arttculat~otr holds w t h respect to the content irr the narrow sense, that rs, that w l ~ ~ c h I\ evnluatcd dgaln,t the

contextually prov~dcd sltuatlon 1 he content in qucst~on is fully artic dared, what 1s not fi~lly articrilated is the Aushnlan proposltlon rtself, 5vnce the situation coniporient In it ts not articulated.

If, In ttte "~1x1' case, there are two optlons avadahle, In other rases dn account b,tsed on contextud donlam restriction and/or toprc cltuatlon would be hr-fetched. In 5uch cases the nat~rral arlalysls 1s iu terrrx of 'tn

extra conjunct in the scope of the event quant~fier Let us look at one cuch

case, 112 order to motivate the type of mn~ple~~ierrtat~oxl I have advoc'tted Conslder the well-known example

I have eaten

Suppose tills e said to dechne an ukvitatloii to d~nncr. The utterance then ineans that the speaker has eaten (1) dznner (n) that evelung (Wrlson and

Page 131: Recanati, François. Truth-Conditional Pragmatics

hperber 2,002: 607-12). If the speaker has eaten a couple of peanuts, or if she h a eaten tlree days before, that would not be sufficient to rnake what she has ~a rd (~nttutively) true In tluis context So there IS impllcit reference both to the tcriipor,~l uiterval dunng whlch the eamg event h a taken place (vrz. the evenrng or1 whch the utterance is made) and to the therne of the eatlrig tverrt (vi, ciiniler). Now the lmphclt rekrence to the hrne mterval can plauubly be handled in tenrrs of contextual dornan restnchon, but the tmplrc~r refererice to tlue type of thuig eaten (a regular drnner rather than a few peanut" IS nsorc naturdly eoristr~ted as iritroducing all extra conjunct m tile scope of the everit yuanuntlfier. The utterance can therefore be gven the followmg analys~s

(3r:Ti~f~-EVENING (eu (34 [PAST (4 A 'I'IME (e, t) A EATING (e) A AGENT (the speaker, ci) A ~ E E E M E (dinner, ell

L>oes this rrtean tludt the therne of the eatlng event is an unart~culated conchtuent of the proposition, in vrolatlon of h l l arhcula~on? The appeal to ur1,irtuulatetl const~htertts \trikes me as impla~~sible and unnecessary m such a case The rnrplicir reference to the type of thmg eaten 1s best arlalysed 111 tern15 of a mean~rig shifi nlaklrig the sense of 'eat' more specific than it literally rz If the hearer s~lys, 111 tlre same context, 'I have not eaten' and the he,lrer replre, Yt,ut voli have-1 \aw you eating a peanut', the speaker wlI respond ' I bar I\ Trot wliat I call eattrg', and he wlll be nght. In this context 'eat' irlearus eat iitnner, just as, in certain contexts, 'to dnnk' means to d n ~ t k cll~ohol Agan, th~s IS toi~ventlonalrzed. at least in the 'dnnk' case; but what has got ronvenoonaltaed Is a rnearurtg slid' that 111ahes the sense conveyed by tlur expresslor1 more specific than the literal sense. Rather than ray that a new constituent has been contributed to the tnterpretatlon without cctrrespor-lding to aiiyth~ng in the sentence, in violahon of fi1U diculahon, rt is rrlore satr\fjctory to ray that tlue word 'eat' in this context has acquved a specific cert\e whlctl niakes i t eq~~~va len t to 'eat d~nt~er ' .

One ot nly goals 111 th s chapter has been to cast doubt upon the standard view regarding rr~eteorologcal predicates, and to argue for an alternative

Page 132: Recanati, François. Truth-Conditional Pragmatics

analys~s. Even though I have done my best, 1 have to admjt that thc l\sue rs far from settled. 1 have argued aga:nln\t a defence of the stand,krtl mew ka\ecl on optional variables (Secoon j 2) , but cl~e eviilence I have adduced rs not decnive. The rnarn difficulty I luve rased for tha t ctefence rs tlxrt there IS (SO far) no convlnclng explartabon fbr the lack of ~nt~ractton betwren negation arid the alleged emstentid quant~fier, but ~t may well be that such ax1

explanation u forthcormrig ALSO, Irr Secbort 6 2 , I have cant cded that, pend~ng finer-grained test5 h r Iltcr,drre'i\. ,111 aczoLlnt ot' the we:~therrn'~n example In terms of a recesuvr ,!lift wwtiks as smootl~ly ,LS rrly kavoured account in term ofhee ennchrnenr Tli~r r \ a significant Loncessloa srnc c an account based on the idea of- rrcesuvc. sl~lft ~vould be colnpatlble cvlcll tlae

standard v ~ c w

My main goal, however, h,rz bccri tcr rr\e t f r~r case \tndy (like thC c*lce

study in Chapter z) to support TCl' a\ ;I gcrie1~~1 ~riethociologrcal posrtlon I th~nk thls goal has beer1 AC ~ I C V ~ L ~ , fix tlze dccorrnt b a d orr tlic d e a uf recessive shtfi would SUPPOII TC:P ju\t as much ,IS tile account In terns of

free enr~chment That a wl-tv I taie rl~c p o ~ s ~ b ~ b t y (ofsuch an ar ~oillll: bjr;j~t heartedl)

At several potnts irr the chapter I h ~ v e dracvn a dist~n~trori b ~ t w e i ~ l thc 'bare' logcal forni ctf nn utteraclcc aad thc 'i~iud~tied' logtal lon~r t l ~ ~ r t results from procesjes of rrludirlatlort Irr the next c-haptcr, X ivill offier

some clxiiication for bod1 tths p,rxt~ciilar dlit~rrctron 2nd the f C P bdrxtc

work more gener&y.15

31 This chapter is an expanded vzrs~on of rriy paper ' k t i s r&irilng (somewhrrc.'), Lirzyuzstza arid I%ilosophy 30, 2007: 123-46 (itseKa sequel to my "Un:~rtlculatetl cortstituertts', Lirguist~rs and I-"hiiijsopizy 25, 2002: 299-345) Or rather: it is the L.GP paper which is a triircii reduced venlori of an e;rrlier manuscript, which consisted of Chapter 3 and an ancestor to Cllapcer 4. Tire nianlacnpt 1x1 tjitmtlon benefitted watly from the c o m e n u of a reCerec on ari even earlier dr&.

Page 133: Recanati, François. Truth-Conditional Pragmatics
Page 134: Recanati, François. Truth-Conditional Pragmatics

4

Pragmatics and Logical Form

I . Tnl th-conditional! prag~i~atics

Tnrtir-contlr~ond pr;tg~~atlcs (TCP) 1\ the vrew that the eftects of context on trvtth-ronditloriai content need not be rrzce;;hle to the Irngiist~c rn<~terldl~rl the ~itterect sentence Some e l k ts of-< orrteAt on mith condltlonnl corlterlt are due

to the hngu~stir 111atcnd (e g. to contest- sensltrve word, or rnorphex~res wltrcll trigger tllr search for contextual values), btit otliers recult 6<1rri 'top-clown' pr,Lglnatlc protl-e.;se\ tbdt t k e phcc not bec'~use d ~ e itrlgilistlc rlraterizl den-tandc ~ t , but hecawe the utter mce's c orlterlt n not tjltldr~lly or -cvl~olly cncodcd in the

uttered 5entenc c, whose mearung reclulres ddjl*strr~e~lt or elahoratlot1 m crrder to determine an ddrru,slble c o n t c ~ ~ t &r the speaker's utter'ince.

The extra step reciu~red to get 6orn tollventtorla1 me,xjrng to ad~i~lsslble

content a l~suallv tre,~ted as external to truth-condlt~or~d c ~ n t ~ r l t proper, becar.;e n-uth concilt~onal content proper 1% suppowi to he unafl2ctcd by prag~n~~t tc concldera~ons t r r z l ~ s s wch cor~sider'rtlom ,ire forced upon the mtcr- pretcr by the h~iguntlc ill'~ten,~l Itself Now, m I wcl Irk the Irrt7octuct~or1, we cdn cl~drdcterrre ,i notion of l l~crd tl-ctrrtent suc-h that Itteral coritent is, by defirrztlon, ~rrdcpendent of-pragnuti~ conslderatton< (unle\s w c h con\~der;it~orlr ,ire -emposed b\ the Irngui\tlc rnclter~d tltself), b ~ l t when ~t comes to the rntuzfzz~p

tiuth-con&tlons of .m uttetxnce, f C:P holds that they renitt, In pxt, from pmgrn~t~c processe5 that <ire nor triggered hy the h~lgulctic nwtend. Assitnung

that scilxinacs I \ to J C ~ ount for tlle r~~tu ihve truth-cortdit~ora o f .~~ t t e~~nces , ~t

must rrl,~ke room tor 'hce' (pragrldhc,r~y conrrolletf) pr,igrn*thc processw,j~ist: s, rt rndc5 room fix h~rguisticdly costtrolled pragrnatlc processes tn order to

secure contextudl valrler for the context S C I I S I ~ I V ~ elemertts 111 die ienteilce.

Free pragmrtic processe, take as inprtt the Illearling whlcfi IS the reinantic mterpretation of sctrric exprejr-tor~ and yield ~5 output the nlodulJ.ted rnem-

1119 tltat ~vd l undergo ser-rzdrrtlc corllposrllori 1~1th thr rnedr ~mgs of the other

Page 135: Recanati, François. Truth-Conditional Pragmatics

expressions m the sentence In other words, the compos~t~on rules deter- s~lrrie tllc v~ltxe of a con~iplex expression on the basls of the pragt~tatzcally

modz~lated vdues of tile parts, a c ~ o r d ~ n g to formula (F)

In that tonl~uln 'f st,?nd\ for the interpretation tunctxon, 'tz^P' stands fbr a c onlp1e.u exytrcsc~orz forrned from the parts 'a' and ", and the '2's are free lltglier-order v,rn,rbles rdngng over available praplatxc functions (~ncludlng rdentlty, w h ~ h pves US the 'hterG;17 cae).' The formula says that the senimtlr vdue of 'I c o n ~ p l e ~ pluase nAP 1s a function of the pragmatic valuec ofthe p ~ m , where the 'pragniatic values' m quegtlon are what we get when we subject the 11teral wrrrantxc values of the parts to prqgi2atic modrilation. Pragtn~tic mctdul~t~on covers optional processes such as Gee enrichment, Ioosexting. rnetitnvrrnc transfer, etc. processe\ which (arguably) affect the rntrxltive t~utli-cond~tions but wh~ch take place fur pragmdtxc reasons, without I,elrig triggered by the I m p i s t ~ c inatenal ln an obhgatory manner.

Orte way Oiutrden~tl~%ng tlic formula IS to cay that setnanttc ro~~zpocttton itself zr LE ~ ~ ~ r ~ t r ~ t - d ~ p ~ n d e ~ l t p ~ t t c ~ s ~ m the cotirce ofdenving the semantlc value ot ,I c ornpicxx expression, one optlonallv modulates the semmtic valuez of the parts, .ifid it ts tlze col~text which ctetemi~nes whlch pragmat~c fi~nction. rf any, c cttxiei intct p1,ky arid v~elds the modulated v;rlue that undergoes serllantic compo5itlnn 'f ha corresponcis to the mew ~vhlch, In my book lrrenzl n/l<~untn<q, I called 'Prapiat~c Conipo~itlon'.~ Another, even more radtt al w ~ y of understandliig tlie formula conespor~ds to a wew put tbrward b y Grnnaro C~lnercl-nd In cotx~ectlon with scalar implr~atures (Chlcrchia 2004) O n C'lucrchla's plcture, the InterpreQt-ron 'function' n no longer a fu~lctxon brrt a relatxoli Adaptlng Chlerchla's idea, we could say that each expresuort denotes a set of admtsslble values the came Itnguistlc form receives arn ~ndefinte nurrzber of dstinct, alternative denotations, depending on tyfi~ch ~pt ior~al prqpatlc processes (wl~lcb '$s distinct &om ~dentlty) come into play 'Chus 'hger', in the nght context, comes to rnean REPKESEN-

1A1ION OF I I G ~ R , 'stiagllt' conies to mean menoxImTrNG STRAIGHTNESS,

' I am indebted to (;enn~ro Cllierchia and Peter PMn for discussion ofthe over& picture which I an, preserrtlog here. (For details, see Chapter I , Sectlon 7.)

Recarlat~ (2004: 138-40). See Jackendoff (1997: 47-67 and 2002: 387-94) for a sknilx notion oi' 'enriched compositiori.

Page 136: Recanati, François. Truth-Conditional Pragmatics

and so on and so forth. Those niodulated iriear~lnp arc the bu~lctiarg bioc ks out of wh~ch the meaning of ~ornplex pbraser llkc '\tone hon' or 'prettv stratght' are budt A stone hori n not a {real) Iron, ~ i i d s o n ~ ~ t h i ~ i g that is

pretty straght IS not (redly) stralgtlt That suggests that in tlloce phrases, the words 'Iron' and 'straigllt'get a r~iodulated value, cllstsnct from tlielr stariclnrd semnhc value.

Whichever construd we favctnr, ~ r : 1s rmport3nt to re,d~ze tl~xt the van ables over pragnanc finct~on\ thdt occur In fonilula (F) are there only rrr the theonst's metalanguage They are not iiupposed to be present at nny level of'

s y ~ i t ~ c t i ~ structure sn the object-larlguage ' T h ~ t rs, indeed, what dctincs tice pragrnam processec. they are not triggered b y a t.;rr~able ni the syntax, or anythng of the sort, but take place to1 purely prngmatlc reasons -sn orciclx to adjust the corlverlbonal riieanlng of the urords to the utuatrorl at. fidnti Even though they have an lirlp~ct 011 troth c ~)t1~3ltl~>~1al content, thev Are a

matter of tise, not a rriatter of conventlond xlreatllrlg

'The clalrn ~vbsch T'CP ~nakes regxdxag the role of Gee pr;lgxnatrc processes 111 the detenn~natlorl of ttrtlrltivr tr~~tll-i on<i~tiur~,~l content 1s art

enlpu~cal conjecture about natural l a~ igu~g t~ C)ther phliosop1,eri ot larr p a g e (Stanley 2000. SrabO 2000. Klrrg and St,rl-rley 2005) have n i d c the opposite conlecture, Inore ln l u ~ e wltft tradit~or~al ways of tlilnbng about meanmg arid truth-conditrons Since ~ t . give\ up thox tlddit~oti~d A&\~III~~>-

tsons, TCP sounds revolutionary, and there are theonstc (e g IYretlelil zoo~a) who are suspicsou$ of ~t because tkev take ~t ti3 threater~ the vex?

enterpnsc of seniantlcs 7hss 5eems to me gros\lv euaggeratcd TCP nixv

cornplscate the task of sernantsi.i but cerrait~lv does not mike it nnpossihle 13e that as it may, I vvrll wiirtrne TCP ~ r r whdt ti~llows, arid cvrU onlv bz concerned w t h IS-ues coacernrrig 18 proper mterpretntlon What cxac t l b

does TCP say? How are \we to 1mder\t,~11d free pragmatic processes7

2. Free pragmatic processes: two interpretations

Although variables for pragrnatlc ftirrtnctio~~s nre confined to the metalan

p a g e and are not syntact~cauy projected, st111 lhere are two possrble wiry, of loohng at the role played by fiee pragmatic processes. C)nc con\tn~al rs 'syr~tact~c' and the other one 1s ' sesi~a~~t~c '(Recarl~jti 2002. 339-43, 5c,lnlry

Page 137: Recanati, François. Truth-Conditional Pragmatics

i i t r $ t . i X + IWAhir\rl4i iC, iJRri;CM,is55F%- i Wt> ih i bKfiKE I R TIOi\cS

loci .; tr 237) I st,rrt wlth the serrt,nnac c-ctnsrnldl, wbrch I have titken for ~ I - , I P I L C ' ( ~ 4 0 t i 1 111 l i ly preserltatlon oCvl CI"

A xrprescni.~trorr 15 n s y r ~ t , ~ ~ ~ c ohjet r- J. wqlrcnce of syrnhol\ in 'iome r ( l ~ ~ t - \ ~ ' i i t ~ i t ~ o ~ ) . i l : systeilr (povirhly tlrc "l,rrig~i,tge o f tilought' I € the represen- tatmrl at *\sue r i Lt ~ I C I I L ~ I I repre5elildt101l) C>II the renlanuc constmA, < t r r r , y o g to ft~miula (F) ,iltove, the t>utput oC&ee pragxnatlcpproces\es rs not i .rc.prstc.rit,itrcrn but a "~ritposxbcin', t h ~ t I\ '1 rernarltlc object. If the .cvtr~c! 6pri\po\rtitrrr' \otrnd% too rcpreser~tatron,~l, we call simply talk of the

rsttcralri t i, (irrt~rrtrve) t r u t h - ~ o i ~ d ~ t r o n ~ 'IS berxrg the output of semmtrco-

pragrn,rtic prcxewrig dong the he r , trf io~rrlula (F) However wc call it, the O L I ~ ~ L I I . 111 qrrzshon is the (rntutnvc) rlltcrprctattorr of the trtterance, and r t dc.lwtfi"i> iqwn two thmgs 112 a~fd1t1011 t t 1 tllc se~llcl~rtl~- Ides ofthe langpge: rile loq.rcirl fotrrt of the sentence 5ervci .ls rrlprlr to the iritcrpretation process,

\IL 1111~ ~ I I C praqmtlt~( C~)YIICXI Cieter~lll~les i30ih the (lelJld1ltlL values of c ~ n t e x t - jcrrsrircc cxjri-.~.\\iotl\ in lagcal fort11 ,rl-rd the pragInat1c klnchons whlch

optrorr~lh c orrlr nlto play 1x1 clcrrvrng the ierrr,i~.it~c \ALE of the \;ciholt. frcrrn rhc (pn\srblv rrrodul'lted) scniantic t ;lltie\ of tlre parti

811 ( I~xI IL*EI I~oI , ) IV gerlerdtlve Jrrigtri~trc\, dic Eog~cal fomr o fa sentence, or 1 I', i\ \ralid,lrJlv t orlsttued a\ a level t r f syritdi at represent<ltron that 1s the

pro17~1 lllpiit I O 6 ~ ~ ~ ~ ~ ' ~ ~ t ~ ~ ~ i ~ t e q > ~ e t , r ~ i o ~ ~ At tlirtt level, nrrportant logcnl propcrtlC\ trf ~ h r " \crlterice SLICII ,~flthe rclrltlve COPC o f q~~antlfiers and

rrraphoric dcpcx~dcrrctes arc forr-rially drsyl.ryc~l, Irr S U C ~ 3 way that that 1rvci of 5\ rrt'jctrc xcprcsentt.-ttrot, i ,rn bc sv;vctematicaliy mapped to logcal Ion~-iui,ie wlircl.1 c .y)turc t l ~ e irlierentrdl ptrrex~eral ot tlie senterrce (hence the rlLrrrrcb "Iog~:rc ii iorrxl' fbr tlinr Ievcl of \yrrtat trc rc13rescntation). Tlie coexl\- tent i- r)i wr l t . i c tit strut turcs oi"i cert,irn \art .lrlti the lopcal repie\ent,lt~ons ;i~soi~,itc(l w ltli tlwill under t l ~ r heati~ng "log~:rcril ii~ms' creates .L potentla1 rrnltigtriiy, wlirch d brcrch1,r anti McC'c~rrirell-(;rnct propow tto avoid by tlriiirlg~u"'t~~rrj: thc I O ~ C J I I'oriri qrtu syrlt,tcLi~ ttrrrcttlrc (L F) korn the logcai ~OXIIIII ' I ' I I I irr.lp\ to (if )

Now t l u c b ytopi>\rtrori whrcil rs tE*e outyrrt oi. sern~ritlco-pra-ratlc pro- i c\sirig r t ~ t ~ E l dt,terrrliric\ a 'logrc nl iirrri~' tor tlrt- uttcran~e, that t s a certaln

irrf2rericral prolile w l ~ l t the utterance* rlit;nls ant1 w11at it I$ enta~led by, ~n

(/Ir1ucL oftbe [)fopo\lt'lorl ~t ~ritirrt~velj cxprzscs Thckt 'iogcnl i-onn' too can

be rcprc\c.rrtcd :~t ,I forrrrrila In ,I logic,~l raicrrlus--pourbIv the sarlle calculu\ ~s t l ~ ~ t fsoni w i l r c h ! js are c f r a ~ ~ i 1 o tmt'rj?, o?ilt the c~rldogy, I wlll u'ie 'Ef*' AS

Page 138: Recanati, François. Truth-Conditional Pragmatics

an abbrevl,~nori for the modified loge a1 torri-r rh,rt resr~lts fmrn {ern'lntlco- pragmatic plocessmg. dird wkrch correspoirci\ to the Iltter<tnt e's n ~ t i ~ ~ t ~ v e tnitll- corld~tlorr\ (,ls oppowd to 1t5 rn111in1,ll or hterd trrttll-i ortdl trorzs, that w vvh~t we get if wc snbrnlt 1 to ceni,tntlc- mterprctatlon wit)~orrt m y recourse to Gee praginaQc proce55es)

011 the s e w t l c c-onst-ntd, the m o d ~ t ~ e d logical fornl * 1s only a prrcpic u- ous represen~~tlon, rn a well-behaved log~cal 1angudge, of the rtttcrar~cc's Lntuitrve tnr th-(-on& hons, wh~ch ti uth-cond~irtroui\ result trolls lrktel~-tretlng die uttermce's I E m a~cord~uice wlt1-1 fonnula (I") above. 1 call tlus ( o a r t n ~ d ' se~mnt~c ' rather than 'syntactic' because free p~ ,~grriatlc proce~ec coxxtc into play purely ar a matter of ~riterpretation They tit, rrot give nse to .I fur the1 level

o f reprocnt~ t~on in <iddltlori to the syntactic represrnktlctri\ (I Fs) whlch \cr\e as Input to the ~nterpreution procers In part~cul~~r, jr*s are not '\ern,~ntlc representattons' lrl the object-language In t f le tmcjlnon ol'reierent~~rl \errrarrirtrcs

(ds opposed to rrmslahon.tl sernmurhcs), Inteq7retanoil prot eel$\ b y rn'lpprng representattons (e g LFs) to wolldlj cntihzs or cornplexe\ ctf' \rith, n t ~ t by rnapplng them to firrther reprereritattorlc

I now turn to the 'syntactic' Interpretntlon, fivortred by rele.v.lnce theor- ists and nrost rere'rrchers m pragnahcs 'The startlrtg point a the snnre. there 15 a level of ryritat t ~ c reprecentahtlfri--1,F-that Incorporate\ 'wirate~rr

featutes of \cntcnces \tmcturc ( I ) enter directly into the cein.mtic Interpre tatlon ot-sentent es, and (3) x e ~ t n ~ t l y detemllrlccl bv propertie\ of sentence gdrnmdr' (Chomsky 1976: 30j) VVllnt dlsti11l;ulshes the s y ~ ~ t ~ c tie l'ronl the serrlaritic constlual 1s tire l i~ l low~ng cla~itl, also m ~ d c by Gl1t>ni4.y. Kepre- sentatlorn at 1.E; itre rt1,ippi.d to rrztlrf el~ihorc~tp reprfcc,tztiztrr7tr( 'whx h r-rray

lnbolve belief: expectations and so on m ;1Cidit1011 t o propert1e5 of I I deterrrllned by grgrdrrsrttatical rule' (zhid ) These, addihollgrdl I epresentatrorlr ,ire the rnocirfieti loggrcal tttrrris (2/'*s) on tile sy~s~a.hc construal Thu~ rcle.- vance theorists t,ihe pragn:nlatlc processlrig to operdte on reprcsent,ltlons, and to ouptrtfrrrther rr.r,reLentatlc>rr> The representation operated or r is the u t tcr- dnce's ' logcd fornl', as dehvereil by the llnguistrc ntodulc 1 hc rcpr c\cn ta t~on wli~ch results from pragmatic proccsGng 1s the t-rioclttied log1:lcal turrn (If*), sy~rtactically conctrued. Sperber arid Wrlson (1086d) cdl it the utter- ance's 'pro~-tosit~orial Corm'.

Is the synacbc con\trual corrslsteilt w ~ t b refercntldl semdrltlc% or doe5 ~t n e c e ~ s a n l ~ go togethe1 w ~ t h '1 tramlatiot~~l appro'ich to serliant~csi 7 111s I\ nrt

trrtrrcstllrg arlit tr~ckv t s c~~c Kottvu (:,\r\ton \12'gge\ts tjiut rrlc.v,it~ct~ tl-rcttrv 15

Page 139: Recanati, François. Truth-Conditional Pragmatics

----- - T-JJ - FfLE3? IXAGfiZKf 1 2 , ~ ~ ~ ~ 0 ~ l A E W ~ YAT?ONS

inconsistent with truth-con&tlond senianhcs but thnks t h s 1s acceptable clrrce 'work In other frameworks shows. . . that givuig an account of natural

ldnpage :ertrtant~cs in tern of Jconceptual representahons] is very inuch a

live optlox1 (see K a t ~ 1972, Jdckendoff 1983, 1990)' (Canton 3002: 8-0).

Llke Sperber anti Wilson, she opts for a translational approach, where

\emantics map\ \yntr*ct~c representations to conceptud representahons

(whch can thcrt be modliied or 'developed' tlwougl~ p r a p a b c processmg). I wlll discuss the relevance-theosehc vlew m some detarl below (Sectlon 3),

but for now I want to defend the follow~ng claim: the syntachc construal of TCP, by it!\elg 1s rtot Inconsistent with the project of pvmg a truth-theoretic

(= referentral) qelnaltics for naturd hguage . I-ollc)wrrg the rtlajonnr view in seimntics, let us assume that semantlc

l~lteqretatlon 15 referentla1 rather than translationd: it maps representations

to ztlhat tltcy reprcxnt, rather than rnapplng them to further representanons.

2'111, 1s 111 contrast to the 'tramlational" or 'copihve' view, accordirig to

which serxyantrcs maps syntactic representations to 'semantic representa- tions' con~rruc;.cl AS pxtial or schematic mental representations. Stdl, nodung prnlct t / r a ~n~fh-iarrdltr~?t~al cpirtantic~st f io~n br~ngl t~g t~tental representat~o~ts tnto the pzcture Incteed, 1he 'log~ckl fonrls' which undergo .semxntic Interpretation in

the tiuth-theorctlc &antework can themselves be constrned as mental repre-

sentatiorts As Chierc-hid wntes

The hypothesrs of a log~clcal fonn onto w111ch [s~~fface] syntactic structure 1s mapped fits well with tkt' idea that we are endowed with a language of thought, as our ma111 metflu~z~ tor 5tontlg and retrnevtrlg ~nfi,rmatson, reasomng, and so on The reason wtlv thlc 1s so ss f l~rly apparent Ennplr~cal feature5 of langxages lead hngu~sts to detect tile rxlctenLe of a covert level of representation w~th the properties that the proponat' c)f the Language of ttiought hypotl~es~s have argued for on the bas15 of 111de1)endertt conudrrat~ons It 1s h@y temptlng to \peculate that logcal fonn at-tually ts the lalipage {of thought. (Chiercbza 1y9y c-ci)

On this mew, win-u~h h,s the xnent ot umphcity, logical forms are conceptual

representation\, yet conceptual representiitions that arc stnctly detemuned

by the gnnxnlar (and a.; such belong to the language system). They belong

both to the hnplstic system and to the conceptual system, and serve as lm interlbce between the two systerns. This view 1s conustent with m u - stream referential semantics because such representation.; are construed as the tnput to selr~antic interpretahon, m the truth-theoretlc sense, rather than

Page 140: Recanati, François. Truth-Conditional Pragmatics

P R A G M A T I C S A N D 1OC;ICAX F O I I M 13;

rts output, as 111 translatio~idlicognltive remantlcs (where senlantic s rs uld to map syntactic structures to concrptud represexltahons)

In t h s kamework pragmatic processulg rnay be allowed to "laborate' the logcd forms, qua conceptud represenutlonc, mto &rtlier conceptual repre- sentations. Indeed, Irf*s can be construed as nieritd representananc re\ultmg &om pragmatic operations on lfs Thu ~c the sytitactic consrrud of Gee pragmatic processes: they are keen a, raiappmg the r~iental rrpresentatlc)rr\ associated wrth sentences in vrrtue of tl.ierr gralmlat~cd propertre.; o~~ro h a h e r mental reprecentmons resulting 11-1 part fronr prdgaattc proresslrrg

3. Logcal form in relevar~ce theory

Relevance theonsts think the project of gvmg a trucln-coriditro~d renlan t ~ c s for natural lmguage 1s doomed tc3 fa~lrrre, and they ~usume a tranclatioridl senlantzcs i la Jerry Katr. S e i ~ z a ~ ~ t l ~ 1xzterI)retatrorr is vlcwed as (part QJ ) J

'decodmg' process, tlirough which (;~r)tact_l~ representatlolls are svsteniatl- cdly rrlapped to senlanhc reprerentatictra. Tho\e semantic rcpre\rntdtions, resulting from decodmg, are what Sperber ,~nd Wrlso~i call 'log,vl~al f 0 rn1s7

rhey are conceptual structures cvid~ logcd propertie\ (hence thev c,~n undergo logcal opera~ons), but they are hstlnct ii-orn thoirghtr in thrrr they Are not 'complete' and trrxth-rval~xable The logrclal forms tllai: ,we

~ssociated wlth centencec AS thew Inearilerg .are partid (g~ppy) nwntt,"l rcpresentahons They are cotnpar;rtllc to ( a ~ d can be reprew~ted I>v rneanc of) prccficate cdculu\ forxnui,le contarrllng free vanables (p l~n ln- rtruchons on how to fill the vanables) "

Rccorchng tc relevance tl~eory, pragmatic procesririg take4 u\ fro111 the gappy mentd tlepresentahons dsrociated wrth sentences 111 vrrtue of the semantlc rules of the lanpage to fully fledged nlental represriltntlorlc, ~iantelv the thoughts whlcl1 the corxtext~r&zed utteraricer expres*~ Those thoughts are semant~c,~lly conlplrte-trrrth evalluaable- ,tnd constrcute tlre 'propontlond forms' of the utterarrcer wkch express then1 Frce prdbqxlanc- processes operate along the way 111 tlze cour5e ot'tvh~t Sporber a id Wikorl call the 'development'aof the logrcal hml, va11xec are &signed to variables In

See Cantor1 J ~ O O L : bu) for an rxanrple.

Page 141: Recanati, François. Truth-Conditional Pragmatics

r171 10git ,I] form, i>ut c~~nc!interzt or. looserti~~g t d ~ e p i a ~ e ,is well. The whole

~ I O N ~ S S 15 ~ y ~ l t ~ t . t ~ c 111 the fouowlng senw. xr operates on representations and y~clii\ b.\rrthcx rel)rc\crrtatlons But OIIIV t h ~ mput rcpresentatlolz IS 'Iingti~s- trc ' P Ixc output rtxpr e\elrtatlorl i c n ruientol rcproentdtloil, correspoi~dzng to the i ontv\tu;r/ IntciT)rt'tatxolX of'tbe uklcrailre It r\ ;l rerttence, but ,I \elite~?ce i r i i hc iat~ig~tcigc oi tbottght

O r i tlir 1C I vicw, loglc=il Cc)rrns, the outptrt v f l ~ n g u ~ s t ~ c Jecodlng, belong to elu: I;lr?g~ragc \vseibrrr (they 'Ire the 'ssenlar~t~c re[)resentations7 of transla- tlolr'il \errlxutrc s) b i ~ t they Are 'dso con< eptua2 they are partial or schematrc ccrucej~tri~l xejlre\rnt'rtlorls. So logrcd loall 1s the ~rlterface between the

1axrgii.igr avstcrli ,srlcl dlc tlloug11t system, on tluwvrew, a\ ~t w oil ( ' h l e r c ln~ '~ W h t , then, I\ L ~ C dityererlcc between C:lrlc rc krra'\ vlew &rid the relev'mce- tP~co~c:trc i icw j

O n c wpcxh IJP ci~fferclrcc 1.i th'rt 5perber arid Wllsori take l o g ~ a l forms to

\w tire output of ietndntrc inteqrct'rtroll ( ~ n the translaborlal sense), whde

( ir.renhr;i r,ide\ thein ro he the lrlptrt to serlzal~t~c ltlterpretatlon (ul the t i r x t l i the( 11 eric wnic) I c ,>I1 thrs .I \iql.icr i l c 131' ( I I ~ ~ ~ ~ c I I c ~ \)el-ause there 15 no ro,>fi c c > r ~ f l r i i : licre, i l r lc tL two ciif-ier~\~xt ~~t)ri i)n\ ~ f ' \ e r l u n t ~ ( lnterpretat~oii are

;it i;i,~hc Ntr~I-ring pri>verit\ J tbeon\t fiorrr Xlold~rtg &at lo,srcal fi lms are boih i l l e c*uljbul o f wrli'rtrtlc Iriterpreratrurr r r i a frnt sense, anti the mput to

~ i ; i i ~ ~ n t i ( IIILC"LJ)TC~,LTIOII 111 a scc 011d smsC Robyn Cdr~ton iras xketclrcd such dl1 i c lIlllelll( 'ti VXeW

'l'iri:: ~~~xSXt"oxAjiu I>et'.xr expressi:J ofitxn in rhe relevarrt-e-theoretic literature in taUc of ?WC) types ~ / ' s ~ " ~ ~ i a ~ i t i c s ' : (1) R ~r~lli&~tiu~i:%/ Ling~~istic semantics, which could be desirrhed i r r sc'ieerrzcnti of' the f&nn iliit.' pni3am i= t:rtcodc:s) 'ijk', where 'a&' is a

public-larigirage ibrrrr .irlcf. 'ilk' is a Mentaiesc lorn1 (most likely an incomplete, sclzrinratic Mc%r~rde*;c fi ~rrnj; (2) a 'red' ~ e ~ ~ m ~ t i c s , which eqilicates the relation betweerr triir rnent-il rcpresentatiom anti chiit which they represent (so it must be '~ i r s~ i~ rc , t~~ t io i i~ i i~ ) . ((:ILXP~L)II L(>OZ: 58)

But tllcrC rs '1 tiec'pcr cilifi!rence, arid rs tone erm the proper ~riput to 'real'

\exil,intri s or? (:hit.rchw'.; plctlrre, as rrldrristreali geilerahve hnguist~cs I l i c u r r yvrler'illy, chr \ogle ,rl forim Arc d h-del of cyntdchc repiexntahon wht~h rs ri~tii/lf~~{tjfy ~ n f t ' r p r ~ i r i l ~ l ~ (ITX the se115e of leal, truth-theoretic senranacs) J ogu *lB forrtrr ,Ire both syrlt'tcrlc r epre\etltanorrr delrvereci by the language ivsrrrn trnil' c o r t ~ rptu'd reprcsen~t\or,\ that can be setnanhcaUy evduated

clr~orlt i io \k~~trrdticrrr} I'liey are rrI>rerrnrdcrun\ endowed with 'a l o g c d

Page 142: Recanati, François. Truth-Conditional Pragmatics

syntax aappropnate fbr recurs~vely statlag the tiuth-corrcl~t~ons of ,i sentence'

(fiomstelri rc)yg -,) Now thls 15 sonlvthmg which relevance theonst5 dt:, not accept. For relevarice theonst?, the lu;lp;ul,trc repre5erit;ltlons which the linguwtlc module delivers are too indetem~inate to be the Input to selnaritlc tvaluatlon pro~edure5. I'hey rnust ttrst be elaborated pragnat"c"uy 1x1 other word$, the two a,pects whlch go together in the malnstreanl notion of logtcd tolm (bang deternvrleci by gr-alinzar, being ,ern'tr~ticaUy evaluable)

arc d~sjomed in relevance theory. what the granu~la~ dehver-i (the logteal f o m ~ in the xexlse of relevance theory) 15 not (yet) ccmantically cvalttahle

Apparently, [odor holds the same vrcw. I le ixlsrytc that what gets (ompct\i- tlonally mtctrpretcd by mean? of rec ur\ive truth-theoretlc proc eclures rs liot what 15 deterrruned s tn~t ly by the gram3i:rr but the nlod~fieci Ioglc a1 form

wfrictt. l a a syntactic representahon in the la~~guage of thotight, aud w h ~ ~ h 1s aEected by pragmatic procesce5 and world knowledge (Podor zoo1 12-13)

At tllis point, one ililght tlilnk that even ttlls 'deeper' diff;-rerlcc IS

terrmllologlcal Netther ~eferentidl values nor tnrth-values can be ass~gned

to llngtllstlc forins ~ridcpendent of context That tuut 11 1s (or should be) grhnted "o~ everybody So 1tlgl:lcal fcmns, q z ~ properher of \enttxnt c-~vpes, are not truth-ev~luable They are incornrtlete anct 'ggrtyy' It tirat 1s what relevance theonsts nleari when they say that logical 6)mrns callnot be gven a 'real' 5emalitiLs (arid th~s IS cleally purl of whxt they mean), tlzetz tllere 1s no

confl~ct with the tnith-concbt~onal mew, contrary to cvllat they assul.rte. From the star~dpoint of trutli-conditlo~lal scrnaritics, logical fonrls are se- rnanhcdly ~nterpretablr ordy 1n the followiog senre they can be dsslgned truth-condrttons relattve to yarkz~~~lav cusrgrimt7nts cfvulmr~s to tts context-~erzslttzte elenrents Rut they cannot be as5igne~2 ~ibsolute tnlth-c orldlhons (~ndepen-

dent of context) So, there 1s a sense in whlch they art, not sernantl~dl) mterpretable, hut there is also A (weaker) sense 1x1 wluch they are

StrU, the diragreement between truth-c ox~&trorial selnarltlcs and releva~rce theory is not merely terrtimologcal. Relevance theonstr (hhe other TC'P tlleonsts such as Kent Bachj deny that logcd toma 'ire sexliarlt~cally inter-

pretable t n the weak semzsu What they c d serna~axlbc under -dctenrurlacy goes 1,eyostd itidexcahty. For them, 1og1c;il properties hke cjuaitrtler smpe and maphonc deperrdencies are not f i e d at tlie level of IogcdJ form (qtatl

dehvered by the graxl~lrtar), but only at the hrther conceptual level of

'propositional form'. Ruth Kernpson cor~lucks that:

Page 143: Recanati, François. Truth-Conditional Pragmatics

fix. truth-theoretically) interpreted.

Natural lanpage expressions are provided an interpretation by processes of gram- mar orily in the sertse that they are associated with some construct in a system of

representations which is said to constitute the language of thought. . . . This system o f representations onto which natural language objects are mapped is iaelf a sernariticdllly transparent system, with a recursively definable truth-theoretic seman- tics fix all expressions of tile system, a principle of compositiondity applying strictly to iietennirie the se~rrantic properties of all complex constituents on the basis of their parts. It is in this Lantguage-of-thought system that inference is definable, not in :my natural-language gramrliar. (Kempson 1993: 72-3)

Kempson sornetinles makes tllis point by saying that I,F (as trachtionally con-

ceived) is not a naturd-languqe smcture but a language-of-thought structure.

The mapping fkorn (sudace) syntiictic structures to LF is therefbre not part of

the varn~ma--dtho~lgh it is constrained by the grammar. MI LF-bddmg

processes, she says, must be construed as f~lters on the pragmatic process of

constructing a hlly specxfied propositional representation (a pragmatic process

that may well ir~volve fiee pragmatic processes as sub-components).

4. How rnany systerns, and bow ciiEerent?

knee they are rrot strictly ctetenmned by the grammar, mocbfied lo,vicd forrns

do not belong to the languqe system, but to a drgerent 'system of represenra-

tion', a Chon~cky puts ~t the conceptud system {Fodor'c 'language of

thought') Now Jackendo5 e ~ n p h ~ > i ~ e r the ht.tooyer~etty between the two

ivstcrnc oirepresentatlon Acc ordu~g to Jackendoff, the language system and

the conceytnaj system do not ~ntersect in the m y suggesred by the mmtwarn

'logcd ltoni~' ~dea Ratlier, the two systems are d~gorned, and adchhonal 'rules

ot ~onrebp<pondence' are needed to bndge the g ~ p between the syntxtlc

strilcturcs oflanguage arid syntactic representaaons GI the language of thought

( Jackendoff 1 9 ~ 3 I ) So there is no reason why we should expect any level of

h n g u s t l c representahon to &splay the logcal properties (hke quanhfier scope)

wh~ch chClracten7e conc eptual representaaons (see Jackendoff zooz 270).

Other theorists have denied the alleged heterogeneity For Chierclua, as

we have seen, l og~c~d fonris are ulreudy representaaons in the lmguage of

thougltnt-they are conceptud representahons Logcal f o m are conceptual

represerltahons that r e stnctly deterrmned by the grammar (and as such

Page 144: Recanati, François. Truth-Conditional Pragmatics

belong to the language systeni), but as conceptual represent;iaons they iarz

also be elaborated or modified througl~ non-llngx=lllsac consrderatlons The bare l o ~ c a l fonn of a sentence 1s a conceptud represeritaaor~ that 1s deter- rmned gtnctly by the grammar, wll~le the nlodliied l o g ~ a l forrn a a ccmcep tud representahon -cvhlch has been chapecl, .m part. by extrahngirstic facton such as world knowledge and context~lal expectations

The T<T view seenss to me mtei~nediate betwccn tile two posrtlorrr Lrkr Chierch~a, relevance theonsts view logic," lorrns as the ulte&c e betweerr language arid thought. they are (p'irtld) conceptu~l represerltatlolls chat nrr detenmned solely by the hngi;t*age 5ystenl. But these logcal &mns ;Ire not endowed w t h the detenn~natenecs of hnlly-fledged concept~ral repre\exlta tions. They lack Important logcal propertie5 that LVIU only be detemrzried at

a further Level of conceptual elaboranon (the levei of propositiond forms) and, h r that reason, they cannot be gwen a truth-theoretic i11terlsrecahol.i -"

There is yet another pos~tion. So Gr 1 have arrunied that there are twc,

distlnct systems the language system and the concepnlal system Accctrtjmg to some phdosophers, however, there I\ a stglt~ 4yrtern (e g G'~rrutl.rer\ 1996; Ludlow 1999: 1644)) Thought 11 nothi~lg hut 'inner speech', or at

least, it, 15 underpinned by the I~~igwfit~c system, ~vhicli provlcle5 tlsr stnic-

tures h r the amcdat~on of tliouglits If that is so. then we c~1.n account L r modified logcd fonns without appenlrng to a sr~oiid svstern In ~~~l ld l i zor l tt)

the language system. If a $enten( e rs rrttered and a\signe~t, in ctrnteut, A

nio&fied logical form resr_~ltiiig ficfrri the operatloti ot free pr,lgnt,ktrc plo cesses, there are actually two wtitencei; at: pl,iy What I have cdlled tile

' Even within the '1i;UrLcmam generative trad~don there are authors who think that r:ertaiii iogxcal properties relevant to semnnc interpretatidrl are not fixed by tlie syrit= at LF and trlust tberefi>re he dealt witli at the 'conceptual' level. Tiins the L1: which May ( I 983) ascnhes a j a srrlcencc llke "'Wkr.~t dtd everyone bring!" is interpretivdy anlbiguous with ritlier iju:mtifier mpable of beanng wide rcupc' (I-Iorristein 1995: 20). This is noticeable, for that i~arplx~c that 'Mav (1985) drops the requlrnnenc tt~ai sentences be disambipated at LF' (ibid.). So tirere is a connnnnm [of positiorxs: if we dcfi11e 1-F ~v j that level of granunatical representntlon that incorpornres wl~arever features at' sentences irnrctilre enter directly Into the semanuc interpretation of selrterices. the question anses as to tiow tnucli prrrlry~~at~ci conceptud elaboration n needed to get to $, the logicd fonriuiae which are ir~put to the rmch-theoretic machinery which delivers trurh-coricltions. One possibie aliswer (favoured by reszarchen in the mainstream ndjtion) is: none-LF directly maps to II; witliout any p~gnatlc/conccpnral pi.octxsing. As we have just seen, someone like Mav tias to admt ttlar some pmginanciconceptual pmcessixig 1s rieeded in some cases, e.g. to disanibiguate the LF in order to get die pir-list reading k>r 'What did everyone bring?'. Relevance theorists (and 'TCls-tt~eoost.; more generally) think a lot of plsagi~anci' conceptual processing is needed. (For more or1 ambiguous or tinderspecified logcal forns, see trail Deemter and Peters (1996); see also Egg ((2005: 65--8) and the references thereiri.)

Page 145: Recanati, François. Truth-Conditional Pragmatics

rnodrhcd logrt ,d iorrrr 1s In fact the jbxe) logt al fonrr of mother sentence, ilrat wBrrt h Iran\ thrclugh the \peatcer% sandlor the Interpreter's mrrd.

Ihcra: are two powlble ver\lon\ trf rbrs -\new Clnr of t l~eni 1s rather F~irulzar, i t CJPI ' 1 ~ ~ O L I ~ I ~ IXI the wnclngs ot td-reur~sts such s Jeny Katz,

I<rrit H,lclr, ~ r ~ t i "stt.pl.ierr Nelrli. The IcLacl~rlg r d e ~ I\ this Sentences have Irter,~t ir-rtcryrrtat~oas, but they t-,tzl 4140 br uteci to convey sornrth~ng iiiflercxrt lrorr~ (c g rriorc detcrrrlitl.lte than) tllerr s t n c tly l~teral mterpreta- tlcrlr Wilrcn t h t L w ~ ~ c ~ I s , prdgllX,lhc I?iroce'rcil7g m ~ p s the sentence they uttcr to 101~1(* orhex srntcrlizl th'rt was riot uttorctl hrrt mlglzt have been. Let uc call thr wntrxrc c tlr.rr w a xtu3Uy utteted s, 2nd tile other sentence s,. 'The

nlotl~hrd iog1:rlal fortrr ofthe utterdncs rs the (bare) logcal fixn-r of^,.^ T ~ I S 1s how 1 untler\r*lud Bach's nohurr of 'eexpan\ron9 (B:~cll 1987, (ti. 4; Dacb rgt, I;) OXI 13,~ h's VICW the se~ltem es 011~ titters are oken elhptlc~ll tor rnore

c orlap~ltax herrtetrc e\ one 1 2 ~ 5 111 rrund, I r l thc rrctx~hnguisti~, \ellarsun sense of

keUtptla '11' 6> I-or exar-rlple, 1 can \ay * I llerc rs ,A I(lotl 1x1 tfee xmddle of the p ra t r~ ' uitl mean that there rc 'i \Iritr*e rt,bJtcist7nlitiy u 210n In the rruddle of the

I ) IA/ /<~ I - f i i h ~ I I T C X C L I ~ ~ I~ltcrI)retcillotl ilete171l1nt""rthe utter~nce's mod~fied log( al torrli, 2nd that i'r acrrr,\lly tllr logx ,rl lorrxi of ,i rnore complex WIIL~I)LI~ t11,2t wd\ not 11tterecI, but \\ids rrrxlrrrnk; tlrrouph my m n d ('7'here is rc statuv xcprc\cr,trxlg a lictn ~n the middt~ of the p ~ a r ~ a ' j . O n 13a~h's mew, h e f r cr pragmdtlc proces'i of "expamron' r rwp a ;cic*t~rd-l~npage sentence \, ttr ariutllrr n~tr1ral-l,trrguagc~ rcrrtence s, by adthrtg elenlents to s,

' f i e crtller pusrble uitell>xtatlon of the mew rs less farmliar, but ~t has recently i-bunti ,idvocates tn the Ilnguissrr~ ax~d ~yf~~losoplucal cornmurut-y. It

goes like this 111 tlhe t elevarit tx,rr,rplec, tlrere 'ire, ~nctcecl, two sentences c , and 42, tvhcrr ( 01 ~ C - Y I I ~ C ~ I ~ C ~ S to the artud lutcvretatiori of the ixttervlce; but ~t 1s a n2tsr;ike tc-, rlurik r t l ~ l : s, 15 uttered, wizrle a a only rrrentay tokerred. What 1s

uttcred .I< ttiilly 15 s,f Ort thi\ view the two it"rrtencc\ s, a ~ d s, are yhondicully ir~ldtst~nguahiaijlt., bzc,ttrw &at dr%ererztr,ite~ therlr are otdy cuvevt elzvrlenh m

the logic ,rl for-rr? of\,, wh~cb are tr~rs'rlrzg Irk dlr Log~cal form of s, 'lhece covert rlcnlcx~ts nr~~ri-tcsr tllerttselvtv m the sernrsntl~ r?tterl,reklaon ar~d reveal that the ~ltttxred wntenc e rs s,, not 5, O r r tttrc p~c turt-, co-cdled 4tr011g PrapIaatlC &CFC~( tl\ (the cliec ts trf fi-ce pragnatlc gr~ccs\cx) are nnotlmg but the effects of

Pi, KI~J puts it, 'Ow utteranir rrich;uning iif':i icntcriic S cut hi* i:xi,ressed 3s the gzuzmiliti~al m a m g oC arioiirzr sellien,nc e 5" (lidrr 1~177: 19).

"311 i i ie ~eiiancrri \t:libe ot 'telhprrcal', sir Ncale (LMIO: 286 -7) See dso Eibonrilc'~ cotluTients on I"..lr":de's ,Ipptna.ln l r l i:,ibounic (roo#)

Page 146: Recanati, François. Truth-Conditional Pragmatics

\rmlultlcdlly ulterpretrxig covert elcrxiextb ~n 1ogca.i torr~i 50, 111 a scrise, there I\ no &ee pngndtlc process In mother senst., however, there ,ire such processes, but they rnu5t be retie.icribed arrd at countctl to1 1x1 tcri~ls of the 6ce generatron of pragniatic vanaIil,les In the svntclx

T<&e the senter1c.e 'There 1s a liort ni the rnldclle of the pfaz~d' agairi, anti assume tllat as a re5~1lt ot'(what 1 t&e to be) art apt~orial yrocless of ~tlod~tla- tion, the word 'Lon' liere 1s uilder\tood a\ it IS rn tlse pllr~se 'stone Ison'. Then, atcordlng to the vlew ~inder dlscti\sioll, \vl~,lt I t,ike to be tfse

modified logcal b r m ot'the sentence r i its logical folni, ,uid what I c.lU Its bare logtcd foms 15 not its Iogi~al i'orrr~ a t all. st 15 the loi?;scal forrr~ of the

drstmct, lioniophonous sentence '7 liere is .r 11011 In the rxitdde of the pla?r;a7 \vhltIl memc that there i c a real hon it1 thC mitiifle ot the ptazra Whit

aucgedly c~i\tmgu~sties the two sentence\ is the oct urrenc c m the first one,

but not rn tlsr \econd one, of a covert, optzctna1 elerrlerlt, for chxample a tovert metoiiyrmc. operator (01 \vh~itcvt.~ accotrr~tr for the ~ I I O J ~ I ~ ~ L I O I ~ of 'lion' in this context) I he element 1x1 questloit 112s the iollom~tg proper-tie\.

i I t 13 ~ovcrt--tl.~at 1s wlsy there is no srrpt.rf~cldl difiercrtc e between tlic two sentences

z It 1s optional. hence ~t s\ dway, powble tor wh'lt loc~h\ supertic idly hke the same serrtcrlce ucjt to cdrry thdt covert elclut'rlt ~nci therclore not to have the nteancig that re\ultr train the addition c3f that element.

Elements that have those propertie, I dl 'covert optlonals'. By gc,s~t~iig the exrsterlce of s~tch elelrleists 111 the laligu,lge, orie can account fix thc et&cts

of free pragt~i~tlc processes while claiilt~ng that they 'Ire not prafjlnabr procesws at aU, but regular protessef of ~rnl~intic rntcr~rct~trion applied to covert elemem. Such a mew liah been p ~ ~ t fo'orward by IJur\,~ Marti (2oo6),

byJo\ef Sten1 (zooo, zoo6), and by X'olly J'icobsor~ (aooc) /

As an exanlple, tahe metaphor, dncusrcd by Stern An exprexss1ou I?

interpreted tnetapboncdly, aclcord11lg to Stern, ~t and only rf a covert 'Mthat'-operator apphcs to that exprcseon --a I rrntcst -sensltrve operator i'or whch Stem supphes a ICapl,ln-i~lsp~~ed sernantlc s. The 'Mehat' operator 1s opaonal: wl~erzever ~t occurs, lt 1s also polclblc to build a sentence

One may alao lnierpret in tills hght the 'rvntacnc' Z I I ~ ~ S I S 01 sc:Lj.lr i ~ ~ ~ p l i c a t ~ ~ r e s [)tit forwaril by Danny Fox and h ~ s coilaboraton. 011 llrat ,inAyrls cca1.u nnplrcat~~res result Crvnr the Cet: insr~~ic.ioa uf a covert eA&ustivity operator Exxh witil a Inrnnlng akii? to tlrat of 'only'. See FOX and I-Inckl (2006: 543). Fox (1007), Cl~rrchia, Fox. ;uld Spe~ro r (filrthccrniing).

Page 147: Recanati, François. Truth-Conditional Pragmatics

-- mdisanp~shable &om the metaphoncal sentence but wlth a dsfferent

rneanirlg (slnc e the altea~atsve seiitence does not carry the 'Mthat'-operator

that a responcible -fbr the rnetaphoncal interpretat~on) Stern suggests that

the same sort of ~ C C O L I K I ~ wljl work for metonymy. Slrmlxly, Luisa Marti, m her dl\cussron ctf rny vlewc, pos~ts covert optnonals to account for all the

Lases for whir h I ~ppea l to fiee ennche r i t . Whenever I invoke a fiee

pragmatic, furictron that makes the nieanlng of a11 expression more specfir,

Marti poslt~ cobert varsable g which 1s assigned, in context, that very

f;ilctlon AS I ~ S \ernantlc interpretahon. O n that p~cture the alleged cfigerence

bemeen kee ennclmient and saturaaon IS smply a drfference between two

mpes of covert elenients. those wluch, IIke the g variable in quesaon, are

o p t ~ o d and can be omtted mthout m&ng the sentence u n g r m a t i c a l or

ottiemwse devrant, 'and those which cannot be omtted. In 'John is short', a

ccrvert vanable ( h r a conlpanson class or whatever serves as imphcit param-

eter) n also rnvolved but it ss not ophonal: whenever what looks superfi-

cially like the seritence 'John a short' rs uttered, the coven element has to be

there But the covert elerrrents thdt account for metaphors, metonynues,

tree enncturrent, etc are charactenzed by therr opt~onahty. They may be

generated 1st the syntax but they need not be As Marti wntes:

The i mcial d~fferarirc between the tv~o proposals res~des in what bears the respon- sibrlrt-u tor opttorid~ty Iii Retanatx's svstem, that is the respons~bihty of the pragunltro, o i the properties of the context ot utterance In the system proposed here, the pragmcltrcs has the rarne responsibd~tv it has in the interpretatron of pronourrs, and only that That 15, gven a vanable in the svntax/at LI-, there has ro be a vmable-wstplmerit, volrich ot Lourse depends on the context of utterance, that j~roviddes vnirie\ tor thrs van,ible But the pragriiaucs does not trigger anything m elre sense ok Rcc,~nat~, tliere 15 no process ot 6ee e~incbment. Whether la] variable 1s gelrerated m the syntax or iiot is lef? coniplete1y bee.. . The system tnes out chfferent der~vanorts, and only those that comply wth all the pnnc~ples of gammr, rn~luchrlg Gncean pnn~ipfe\, are successful (Marti zoo6 14p.jo)

Elow itre we to ,iscount for the Ji"dkrence bettveen covert optianals and

otlier covert elenlents lacking tile optnond~ty feature? Marti insists that there

1s o111y one sort of covert vanable the cfifference between the two types of

case (alleged 'ssat~lratlon' cares and alleged 'modulation' cases) is smply that

Wheri dtey are genmted, they must be semantically interpreted and, if they are free variables, they must be assigned a cantest~ral value. What is optio~iat is their generation, not their interpretation.

Page 148: Recanati, François. Truth-Conditional Pragmatics

V R A G M A l l C I A N ] ) L O G I C A l TOKM 141

someth~ig m the sentence imposes the presence of the covert vanable ul

some cases (e.g 'short' docs) wl~ile m other i ascs the prwence o f the covert element 1s not lmposed by ar~ytl~mg m the seritexlce and could be oxnrtted w~tbout umigrammati~ality In cormilrtitlng on 'a firs drdft of Marti's paper, Polly Jacobron suggested that coven optionals are notlur~g but (overt t z t d J ~ r t ~ t \

It 1s of the essence of adjuncts to be optxmnd, she salci, slmlce their type IS aJa In contrast, standard saturatiorl v~rsables f l l a r ~ karguaient plxe and cannot

be omitted wthout ungaiimat~c ~ b t y (MA& 2006. 146-7) Whatever we thlrik of the line purcued by Marti, Jacobson, and Stern, 1

thlnk it lias to be counted as another- -,xdnilttedy deflatioriarrr---syi~tactri

corzstrual of free p rapanc processes Ifre man dlfirence wlth the othtr two syntactic accounts 1s that ot~erythmtq 13 rinw done tvithtn tlzc iur;qnuagr. syrtem

on t111s account, what I cdl prci@l3hc 1110d~latiot1 t;lkes place through {I) the

free generatior1 of xddltrional element^ xn thy (covert) syntzc, 2nd (11) the semanhc interpretanon of thobe elerrlrxlts alorig Lqnufiar he.; Tho resulting blew sounds diarnetr~caJly oppoied to TCP, but the appearaxes nlav be deceptive. As far as 1 can tell, the only .;uhstarrtld ciift;rence tliere 1s l9etween that account and the other svlttac tlr accolints n t h ~ t the level of avrrtactrc

representatiori to \vliich the addrtronal elerneritc belorig rernan, wlthrru the conkkies ofthe language system (rxther tkrarr i r ~ ~ u l c ~ n g a sh~fi to the ion( rp- tual system). Wlut this d~tin r~lce exac tiv dtt-1ounts to--tvliat ~ t s curi\c-

quences are-remans to be cteten~unctf ""

' In a rdated vein, Jacobson shows tltat quatlriiier dixmist rrcltriction (a phenorncr~on which Bach (2000) treau as an irAimce of 'expansion') can he lccourrtcd for ur t c m of coven: reluttve i - l d i ~ ~ ' : &rely adjoined to the nouns (Jacobson ~ o o j ) . -(bus "very girY can be eitl~cr:

every & girl] (contextuaUy unresrtlaed rradsng) or:

every jN *I kc PRO!] (contextrrd?y rrstrizted rrduilng)

Contextual donam restriction on t lus account is a illalter of crtntexnidy assigtiing a vdue to a siieni variable 'PRO', standing for a relative rlause. T>iEcrent rt%uictions ('every girl i r ~ the room', 'every girl urho's got an A'. . . ) correspond to different a%ii$l~inenu nfvalrte to titat varkble, whiie the unresmcted reading correspon& to the case in which rio variable is generated. As always with coven opdoxrals, the assignr~~ent is obligatory; what is opuorwl i s the p ~ ~ e r a a o n of the variable.

'" This chapter was originally the last sectiorl of (the long versiori of) my ydper 7t i s ra~nirrg (somewhere)'. After a short version of the paper, oumt~s~~o~~ding rougfdy to the tint iralf appeared in Linguistic and I-'hilosophy, I used the last section as a basrs for my concribunott to Expikit Gorctmrrni~iltiun: Xobyn Carston's Prag~natics (E. Ro~nero and B. Soria e d ~ ) . 'This chapter is a sliglrtly mnodiiifid vetvivti clf'rny paper for that volume. I am indebted to Philippe Schlenker f c ~ r conlrnertts on a first

Page 149: Recanati, François. Truth-Conditional Pragmatics

Embedded Implicatures

C:onversational implicatx~res are a species of pragrtlatic irr~plicatior~;' they are

implications of an act of 'saying'. 'The speaker's saying dlat 21 i~x~plics that q ( p e n the presuniptio~i that he or she respects the rlorrxls of conversatiorr

Crice's 'rriaxims'---or the overarching 'Cooperative l'rinciple'). 170r exanr- ple, the speaker's saying that some st~rdents came, together with the prei~iss

that s/he n well-informed arid tnci to be ,I\ mforrriatxve as possrble, r r ~ ~ p l ~ e s that nut all students cane . (If all \ t~ldenb hati come, the spedlicr ought to

have said so.) Insofar a\ the speakel overtly lnterlds the heartr to recover tllo,e pragniatlc ~r r~pl~canons of tile speech act, they are pdrt of vvl~at the spc'lker rrlealis, thotigll nut part ofwllat tlte sentent e rne,mj In thts w,iy we can accoul~t fhr certa111 J S P C C ~ S t>f utterance rrreanlng w~thrrl pragniatlcs, wltho~tt burctenlrlg semantlc theory.

Conversat~oll~l ~ntpllcattires thus understood have two Iniportant fea txres. Flrst, they recult frorli an 1nt2rence Now 'lrikreuce' car1 be used m

two w;tys: the broad and the narrow serl\e 111 the \tnct, xl,irlow scllse, ~ni?rences satisfy what I call the ava~l~~hri t ty contx'ztlan. whoever nrakes .ul iriference (111 tLlt r imow sense) IS uuure t h ~ t tlne jutlgen~ent he or she amvcs

at is mfereirtlany b a e d upon s o n ~ e prtvro~lr jrldgcnrent. No S L I C ~ I ~ o x ~ d ~ t ~ o n apphes to 1nf2ren~cs 111 tile broad xnse Imagrte \orneorie hcxnng the doorbell, and corrung to helleve, on that hasrs, that there 1s .ionleone at

the door. T h ~ s example irivolvec both an trifercnce In the broad sense mcl 'an

~r~fcrence ut the run ow serzse Identlljrir~g the souxid one h e m as that of the doorbell arguably involves an '~nfkrerice', rw \ o ~ x ~ e c ognrttve scientists tell 11s; but the subject 15 not a w x e t h ~ t that 1% so T11e avarlabll~ty c onditlorl 15 not

Page 150: Recanati, François. Truth-Conditional Pragmatics

though the sut3ect's irrference that there rs \orneone ~t the door 1s sporltaneous and unreflective, stlll ~t is 'ava~lable'to the subject, who Lr~owt, that the basis for his or herjudgement IS the fact that the doctrhell 1s rmgng T h ~ s is an irrferencc in the narrow sense, one that takes pl,lce ~t the person~l rather than rnerely at the sub-penonal level

Cor~verratroiid ~mphcatures, I take ~ t . are Inferences in the nxrow sense the subject hiows that 111s or her jtidgernent relatrve to what tlie speaker rti~pl~es I \ babeti upoil sorrte mdependent judgement regarding what the

speaker ~ r y s To take a standard exanlple: rt, wlien asked whether I can cook, X reply 'I air1 Prctic h', my utterance conversatlondly ~niphcatcs that I can cook, ar~d whoever u~iderst,inds ~t is aware that what I ~rnply (that I can cook) %ii,Uows fiorn' what I say or niv savlng of ~ t , that IS, whoever tiillv utider\tands the ritterance is aware of what IS said. ofwhat IS imphed, and of

the ~riferentr~d connectron betwceri what is rtnphed and (the say~ng of) what

rs sdlc3

1'kre second feat~~re of convenatrond miplrcatures I want to draw atterrtton to is tile post-proposrtrond character of miphcattlres. Imnphcatures are generated vi,r ax1 ulferencc whov z r p t t , the ha dzat r-It~ speaker ha ,aid ihut p I Ietlc e tio iniplrcature c m he c omputed uriless \omethmg h a been said. some propositmil expresred In ~ A I tlcular, no irnph~ature can be computed at a \ilk) locutionam level We luve to compute tlie tmth-con&tions first. so as to nbe ,I defui~te content to the rpeakcr's speecti act, before we can mfer anvthrng firon1 that 'ipeech act

Gnce's theory of conversational iniphc~~tures has eryoyed a tremendous success \iriLe 11e first put ~t hrward 111 tlre 1960s; but an iniportant change has oc~nrreci, ~ruriated by Gnce hiniself The notion of lrnplicature has been

extended to case\ ul whlch rieltiler avaslab&ty nor global~ty 1s present Those are, indeed, the cases that xnatter most to semantrcs

C:onsider the followmg exaniples

( I ) 15tU and Jane got nlxned and had marly chldren (2) fJ1U &lid Jane have three cldcfren

Sentence (I) urlphes that BrU and Jane got named before h a m g the chddren. T h s ts ~tarld~wdly accounted for by saylng that the speaker IS expected to respect the m ~ m ~ of rnmiier, whrcll enjoins one to be orderly and, in repomng

Page 151: Recanati, François. Truth-Conditional Pragmatics

- LMBE.I>I)TI> IRIPI I c A - i L r l l L r , 145

events, to report then1 in the order xn wlnch they o~ciurred Givm that assulxiption, tlie speaker's saymg (I) unpiles dlat the rrimlage took place behre the bmh ofthe chddren. Senterrce (a) 51,lr~idxIy unpfies that U d arttiJane have at

most three chlidren, for dthey h,id more tlrarl three chddren the speaker ought to have sad so (in m e of the nr&xml of quarlhty w h i ~ h he IS 1,resrrnred to

respect) So both the upper-bounded reaclrng of the tlurnerd m (a) arld the

temporal readulg of the con;lrrncnon ur j r ) ale qud tr, result &oil1 exl~tcErrrrg the core nleanlnp of the wntence with a corrvers~honal imnpllcaturr fhrs rs a

typical use of the nohon of corxvexr~t;ior\,il m~phcdturr ur coritciavorai-i &scusaons. m a t I find stnhng. Ino\vever, n tlie lack of the twtt fedhtrrs I menaoned earher. First. tile dncourri- p m ~ i l p m ~ x e not ,wale tls'rt the alleged mlplr~~tule 1s slot px t ofliteraX ctrrlterlt, the av,lrlal7ktv cotlclitron n not .jatisfied, m contrast to ivh,it happenc rri rXie 'I a111 Frenc11' type of i Jie

Seconct, tlle deged unpiicatute.-i seem to wcur loc'dly They fjfl w~tlrixr the scope of operators, AS us tlre follo~vmg e~~~rrlple\

( 3 ) Bill and J m e have three or b u r cillXdrrr~

(4) Every father-fkzb happv ~f 1115 ciauqhter grrs niarneil and gv's birth to a child much lebx ~f 'she glves birth to .i clzilii 3 n c i get5 aiarneri

In (3) the nun~ersls are given tlie upper-buurrdeci readirig (txtutly t!trcts of

txactl~~_fotarj, but this can tlardlv be Jerrved ~t~ferei~t~allv &om tIlr speaLer"~

sqing (3) in the mlmnial sense (at ledst r i l v c ~ or at lcusi (out) T11c 9trcngthen mg xems to occur locally, w~thln the scope of d~~junctlorr, rather t h ~ n giob~lly And In (4, the tern17orJ suggestloll IS lrltegral to the antecederrt ol

the condltlonal pr'i~~s contrasts w1tl-r the nornrzl beh~vlour of ~mpbcat~~rcr , whlch do not fall w~tbm the sctrpe o f operaton hecaiise 111ey arxse at. d ~ e speech act level. not at tlse level of sub-locut~ortarv constituents

Faced wlth th~s abnorrnd beh~vlour, one has to make room firr a special class of 1lnpLicatures (or pseudo-~lr~~~lri<~tures) vvkllch are intulhvely indrc- t.tngunbable 60x11 sernarlnc content md can arise locallv.j Some rheonsr?, have appealed to the xlotiovi of a 'generxlr~edkonver\at~ond rmljlzcature ~ c s

There 1s an dterriatrve analysis, which does nor rest on rriarrter-l~ased tntpticahlres but on rhe fict that tile tense features rliust be cotltextualiy assigned teurpord vduw. (And there are other dternauve andysrs in the sanie vein.) ' I assume that the two properties go together. ilut t t~u i s only a conjecnrre. In tlus chapter, however.

1 am concerned mainly with the second property (locality),

Page 152: Recanati, François. Truth-Conditional Pragmatics

11;ixiiXie silt-li i<fic:s. Clrttlcrs t1~1ve appralcti to the ~Jistiactiorl betwcen genuine ilripliii-;stuces, wlriclr arc disririct h r a r and 3ddPtiori31 to what is said, and what Kerrt 1-LiciFi h ~ s cnlletl 2conversaticrnai Irrrplicitr~res', that is things that .lire "irnplii-it in wlrnt is said' r,~t.lrcr- than ixrlplied by the act of saying it." My ,urn 111 tliis cl-i:ij)tcr is t o t..orr~pare these (and otl-tcr) approaches to the prolrlenr r;iised Iry what I will healcefi>rrh call "rnbedded implicatures': scc:rrxirig irrrplic:at~rrcv t h t arise loc-ally, dr a sub-krcutionnry Level, without rcsiilzing ~ronrr liar irrferetlce in thc narrow sense.

Act trrtling ttr (;r si.c, sortie c.c~r~verslrtioxr;il itriplicatr~rzs ,Ire 'gener:ilized', that is,

they tlo riot arise 'in v i i l ~ x of special features ofthe c:ontcxt7, but. are nonn:tlly cnrr-icti i ) y sdyirag :r i.c.rtair~ rfzing or- type of-dlirlg. The irnylicature arises 'in the ;thselic.c c>f'!jpc:i-i~l c-ir-c.~~ri~stant-es'. lie s;+ys ( C k e 1j1Rc): 37). The hc t that, in a

rr,lr-rative, ,i c.i>rrJiiirc(lori such ns 'They got x n ; ~ ~ c d : L I ~ C ~ had manv clzildrer~' is iritc>rp~etcil as mrin-wirrg tlie trrrrpord ilorcdcr of't l le reported events is seen by ( ' - ' . . r i r , $.; ., rsr ritirig liom a gencr;rlized cotivcrsatiorlab irrrplicature: such a11

irr~~plic,ituri i 4 lic)~-ni.dly ciu-ri~d hy 211 evtbnt--rt.portirlg cor~jnnctive utterance ccrci-r :rs (1) .

l'irc iil"e:t t h t \ o ~ n e i~llplic:it~xres gelieraiized goes some way toward erlrieiirilx*g wily, in such exes, tile :tv:4ilabiliey zclindit.ion is not satisfied. As I c v i i ~ x o r i putt" i t , gciier~lizcci c.orivers~tio~i;d irllp1ic:lturt.s are ']hard to distiu- giiis11 i l - o ~ i the ;ttniitlli(: content of'lir~gnistic cspressions, beczuse [they are]

ro t i t i~~cIy . ~ s \ ~ o c ~ ~ c c ~ with 1i1lgujst-i~ ~ X ~ T C S S ~ O I I S in aU ordin;~ly contexts' (1 c-viiicoal 1983: 127). *This cs1)lr;uu"lioar i s n t r t witl'rout its prublenis, hut Let It p a ~ i n 3 " 1 i,cirxsider !low, ilsil~g the noric?~~ o f a gex~cralizeci conversation;ll

iixiidic.:icuri., wc, c-,iri acrourtt (or tire secirrrc"of.'the two hcts noted above: the i i c t i l ~ u tire aliegccJ 11np1ic:iture c ; ~ arise at a. scih--se*~rerrtial level, as in (4).

Ac-c-ortEirlg to Mirc:hi:li Green, the ccrrr~rectiori is straightf&rward----a

gcric:raiizcd i~nl~lic;ltrrrt. is ipso cri~hediiahle:

8ldlicE1 (x90.$) I r l rciuvari< 2-iiieurrtrc tennirioiogy ((Spei'iwr . t r d Wilson rg8hu; Carrtorr zoo ) , sizcll 1h111q A I ~ ~ a ~ d ! I ) kw ~ r l ~ p h c - ~ t coristittzctkts I > ( t h - ~t:.xplz~~at~~re' r.ttiw t h r k ~ i~~p l i ca t t~ res .

Page 153: Recanati, François. Truth-Conditional Pragmatics

If assertiorl of a sentence S conveys the iltlplicatu~sl tl;m p with nca~ly -l\/inivcrs:cl repulxity, then when S is en~beddeti the content that is E ~ S I I Z ~ I Y ur~tierstood to hc enibecidcd for semantic pttrposrr i s the proposition (S & p). ((Greer* ~g<ji;: 77)

But wlry I\ that s o exactly? W l ~ v 1s a geucr'~lrred ~rnplrsatxtre--or nt leact, o r ~ e t h ~ t 15 'rscacly un~vers,ll'-*uppo\cd to be e~nl~eclciable?

Generah~ed ~onvenational ltlipl~eat~ires art st111 C O I I V ~ ~ S S ~ C I O I I , ~ ~ 11npI1ca-

tures, for CIILC. 7 0 ( n l c illate an rmpllcnture, cvhetlter gel~erallred or pdrtl- c~~larized, '1s to c'dculate wliat hac to be slippoced irt order to prewrve the

suppoution that the Coopcrnri~e l k n c ~ p l e i r b e ~ n g obierved' ( C h t e sg89. 39-30) The unphcatures are interred t ~ o r n the \pealrcr7s xsymg that p ~ n t l the presumpnou that he w obsc~v~rtg itx ('aopernt~vr 13rlll(-~p1e 'rht, O J L ~ V

c3rilfference het~vzcrr gelrerahred anel. particu1,anred m ~ p l ~ r attires Ire.; 111 the amouiit of contextual rnfolrnntlon rrcecietl to tlenvc tlie r~rspl~cature fic)nr the \pealrcr's 5peecl.i act When an irnpl~t attxre rs gener ' i l~~ed, i11e lnCerence goes through '~ndeperkclently of larforrn,ttlon ibout yartrt rll,ir corltext\ of utterani e' We brrow thnt, zn grttcral, \orrsconu who sdys t h ~ t p respcc ts the

Cooperatrve Pnric~yle ordy ii q- , ~ n d therefore we car1 ~nfer t11,lt q f'ronr hi, saylng that p. w l t h o ~ ~ t J I ~ L V I I I ~ to rely 011 specihc inft>rr~tatiort abo~st the

c ontest of utterance. Thu, understood t t i c h notloll o f a gcnrrahzed conver- satlorial miphiatnre (henc etorth to bc c al1ct.l .t 'C(;r~cc~m' germ ,d~~et f rmplr- cature or (;GI) IS a graded r~ot~orl : ail ~mpl~catnte IS rr-tore or lei$ getler~h7ed (or yartlculnnzec-1) depenci~ng oil die ; t n>o~~n t cti~nfonnatxon regarcixng the context of the utterance t l l ~ t rc r1ecess'lr-y to derive the ~niphc,~mre. But tile mech,inr\n~ of tlie den vat tor^ 1s tire rdn~c in dl1 case\: the ~n lph~a t~ i r e s are ~nfeured froill the pefirrxihllc c of rhe lo( 11tl011al-y act (I e trorrt tlse spe,iber's saylng that p), gven the prcsunlphon t h t [-re 1s ohservmg the C'oo~tcrative Prlrtcrple. (See Car4to11 3002. I I 1 , at id the hter,lture cited tliereln )

In thls Gr~cean trantework, I~o\vevcr, we a n n o t accolrllt for sub-se1rte11- ha1 case:, If c ~ n v e r s ~ ~ t ~ o ~ ~ d l irnplrcatitres , I I ~ the praqianc unphc ~hor i s of ;r

speech act, they cannot al-tse at 2 sirh-locut-ronary level. TIxs pomt was ntade most expl~cidy by I3ucrot. In the tatc 1960s I3ucrot haif lndcyer~derltly come up wrtb a theory of ~niphca t~~res very \~mnd,rr to (:n~e'\ (I3ucrot

rg6g), and 111 the early 1970s Ire got rnvolved 111 an ~n-deptlr study of

senlarltic \tales (e.g. Ruirot 1973, ch 1 3 ) E-lowcver, contrary to I Iolrr ,md Fauconnrer, who wcrc exylol-mg the snn~e remtory, he artd 111s co-

author, Anrc ornbre, r eslsted the ctrnlghttonvard appllcat~on crfGr~cea;lr~ dens

Page 154: Recanati, François. Truth-Conditional Pragmatics

348 G ~ F ~ P K P ~ E !&a11 c,GmLm-Amm*Wt7t~ I~RES: r WCY c t>Nt;EiFTIcfi%

ro scalar pllcno~llena T h e alleged scafx ~rnplicatures, tliey xguecf, cannot be genuine ~n~plicatures because they fill witlun the scope oflog.lcal opera- tors In 'Jolirl t ~ s either five or SIX chddren', the numerals take tile upper- hounded re~dlrtg ('John Iias elther e x a ~ t l y j i v e or ~xuc t l y szx chrldren') rather than the rrim~mal rea&ng ('Jolin has nt leustjzre or at Eruct six cliddren'), but thzs ~ ~ r m o r : be due to ~n ~rnpllcature smce the lniplicature In quectlon would tall wlthlrt tlie scope of the logcal connectwe 'either or'. Th~s , according to Anscornbre .and Durrot, 15 ~mposslble, in virtue presumably of the foU(7wlng dr@llTleIlt

(a) Converidt~orral ~nlphcatures are p r a p a h c consequencec of all act of sdyuig soriletlllng

(b) Art < r c t of wymg sonieth~ng can be perfornled only by means of a coiripletc tittcrarzce, not by rnealis of an unasserted clause such as a ciirjun~t or the antecedent of a condltlonal

(c) I Lcnce, no lllipl~cat~rre can be generated at the sub-locutlonary level. 1 e at tlte lcvel of 2x1 ~inaserted clause such A d~sjuiict or the

ante< eclertt of I ondtt~ctndl

(d) l it s'3v that an ~rnphcature 6x11s ~ v ~ t h i n the scope of a Iog~cal operator is to ~ d v t l i ~ t ~t 1s gerter'ited at the iub-locutlonary level, vlz. at the level of tlre dnuse on cvh~cli the logcal operator operates

(e) Ilence, no inlplicature car1 fall w~thln the $cope of a lopeal operator

It fbllows that in exantplec lllce ( 3 ) and (4, the alleged iniplicatures respon- slble for the tetnpor~l reading of the coryunctlori or for the upper-bounded read~ng of the rit~lilerals Are not genuine irnphcatures, for they are not rnferreci frorn the speder" sperch act but are corlshtutive of the proposition

that is the rontexlt of that act At t h s pnmt, how eve^, we niay be tempted to ch'lxlge the frdmework md

alter the charactensatton of gerierAzed conversaaonal uiiphcatures. From <;nce7s idea that genenlrzed mphcatures arise m the absence ofspecla1 cucum- stance$, tllere IS but a sliort step to the ~oncluslon that they are gepzernted 'by d<f;IuEt', 1.e. bltndly, us soon u? the relwantform ofwords 1s enrounteved. Accordmg to I-Iorn, Ga~ctur, and especially Lemson, who took that step, generAzed m- phcatxtres nse d&ncrlt ~ ~ ~ p l x ~ ~ t u r t ~ (DI). they are generated autornahcally (mthout

See their collectiort of papers (Amcombre and Ilucrot 1983), and especially their reply to Fauconriier (1976), Ansconbrr md Ilucrot (1978), reprinted therein.

Page 155: Recanati, François. Truth-Conditional Pragmatics

mference). They belong to die 'x r i rcro i~~-~p~a~chather than to the "rrucrc7- prapatxc'level, m Robin CanipbelYc tqo lom:

A rnacroprag~natic process is one constltutrd by a seq~lerice of explrcrt inferences governed by pnnclples ofratioild cooperdaorz A imcroprawatlc pmcesh develops a!, a crypnc [= unconsc~ous] and heurrs~c procedure wluch partidly reph~es some rnacropragnlatxc process arid wlllch dV6r.rlts ro a 111 the event of breakdown (Campbell rg8 r lor)

Ceneral~zed conter~ational implrczlnrrcs, this tlnder\tood, drc no longel mfened 6orn the speaker's saying that p together w t h tile presnnlptrolm that tlie Cooperative Prvzclyle IS helr~g observed. They arlse rl-rmngh a different rnechmarn. they a e generated by dchult whel, tlie relevar~t hnguistic trigger 1s encountered, utllr.c~ stlrxnetl~mg m tile ling~iist~c or extra- hngmstlc context blocks tbe gcnrr;rtron and %defeat\' the ~ n ~ p l i c a k ~ r r

O n thls view generahzed convcr5atrond lrmpllc~tures ire rloc rxlerelv generalzed, they are also c o l i v c ~ ~ t ~ o n ~ e d they are ussoctdfed tozth ccrkirrll

btzguistzc zteurt-7 s e r m g ds tnggerc fix the autornatrc procers of rrr~pi~canarc. geiieration. That general~red ~~~ ip l l~d tx~ r t " \ rend to becoxne cor~vent~onal~reti in t h ~ s way seems natural The corrvenrrorls 'xcsociatlng hngp~strc tnnxrs witla Dl argu.lbly belorig to the cdtegory oi 'ccorlventlnns of use'. a toppcaed tc)

stralgfitfonvard 'rneanlng conveot~ons' (Sedrle 1975, Mcnrg*~n 1078) ?Shcv are sl~nriar to the convelltlon\ rn xlrtae of wtxlch a n lrrstmce c,E rhe con-

struct~on 'Can vou VP;' is rexdrly 1oterj3rett.d ar a request, even though hterally ~t n '1 yuesnon The cleriv,it~on of the 1ndlrt.c t cpecch d ~ t ofrequest

&on1 tlie dlrect speech xct ofclue\tlorr ii b,rsed tipor1 ratlon~lrty r trmidera- hons o f the Gr~ceali cort, hut the rrlfer erlr e 1s stiort c~rcu~teci as A reiull 01

gene r~ l~ ra t~on anci, conventrorz~l i~atm (B~cli Z I I ~ ~ E ^L,zrnait~ r079)

Lcvlnson has expl~atly resirted ripatnoxi ofDI to 'sm~d~r&zed" or 'd~cjrt- c~rcurted' impltcatures The latter rely oo ~o11ipressrorl by precedezit and mse horn rotlhmzAQon, he polnts c)ut, wide deGult irrlphcaturts 'are gmcranvc, dnven by general he~xnshcs nrtcX are not dependent upon rc>um7atlonV (l evmson moo 14) ?'h~\ contrast ru1a-v pelhaps be interpreteiti A fo~\ourc

Cons~der scalar implrcatures (the p~racfrpn c.se of Dl). They are hrggered by a spec& forrr~ of words only hecatlsc (I) that forrn ofwords (e g 'sonlt.') rs

convent~ondly recohm7ed ~r he long^^^ to a rcale (e.g tile scale <sonic, mrsr, dl>), and (11) there ~5 a rule tliat &ternmie~, for every par <S, z> cotisishng of a scale S and an Itern i belonpg to that scde, h e L>I associated \vith 1 In the cast: ob

Page 156: Recanati, François. Truth-Conditional Pragmatics

i><> t#ENIW%.i I / l 1 ) t <7hVkR\Ali<tNf\! IiW-VN Aj OtZL\: i Wt.J < < I N < 1 p i l O N \

iPrcia-c rrr ~ l r ~ ~ i i I X I X ~ I I C 'xn~res, ' ~ r g ~ ~ ~ l > l y , the mtplu .Lture a d~rectly associ~ted bmth .i pamc lcur i trrnn oli wort% thc>ugh rroutx~x~~t'~orl, \vxd~ocrt there beurg my d e

Pi-crnr wlut h t81e corivcnt~otld a\focl,ttltrri Now\ i et 11s a s s i~ t~e that tit13 lr~akes

kcaise x-rd 1s wlut I C ~ V I L I S O I ~ II,~s 111 r~l~rld. 1 be5131te tlur &aged c11Eerenc e between

$boa ( i r c l ~ i t ~ d I U I I . F ) ~ C , ~ ~ ~ L ~ ~ S ,117d 1)1, Lev~?l";ori d~kxlodedge~ thdt both belong

r c i AIL ~ n t ~ n ~ l t ' ~ i ~ , ~ t e IWCT betwee~i w n t e ~ ~ ~ t " n~c*,filil~g ant1 bpe~ker's mezmg:

Accorciii-ig to the st;xrid;rrif line (rxlorr often presupposed than justified) therc are just

two kcek to rt riii:ory o f cortrx1~11r1ic.atior1: a icvei of scnt.cnce-meaning (to be c ~ p l i i c ~ ~ t e d by tinc t-iieo~y of grailulr'lr in t i le large sense) atid a level c a i speaker- rrleaniirg (to bc: esjrlicatetl by ;i theory of prahmratics, perhaps cerrtrdy e~nployiizg Grit-e's iitrtioit of r~leaoing~,,,). . . . Sjteakrr -irrrrrriiirg, or ixl~rr;u~ce-token-xneairirig,

will b>r rliatter oKthc actud rioixce ox. orrcr-off'iniirences rrlade in actt~al coatests

by di-rij,d rccipic:iits with ,111 of tiierr r-rc-Xi yaxd(:rrIaricies. Tirris vicw, although ~ : ; t r ~ ~ u w ~ " o u " ' , xs siirely rriadecirratc., 111t3ec.d pc1tcrrt~;illy pernicious, becarrse it under- cstir~r:itiss clxe ri~ir(~l,~r-ity, recrrrrerice, ,tnd systexrradcicy of rnany kinds of pragmatic

niiibrrxit-es. Wh;rt i t umrt-s is a tl.rird laycr . . . of iyste~iiatic pragmatic inference based vioi 0x1 tlirct-t i o ~ ~ l p ~ i t ~ l t i ~ ~ i s L L ~ O I I C ~pe3ker-iiitt.litiorili hut r;itlrcr on gecteral ctxpecta-- t~oiis about how iaxigr~agt: i s rrorrrl:rily I I S ~ ~ . (1 CVIIIS(JII .moo: 7 2 )

I hc. third idvcr i~rvcdc i s cor.avel inor L\ of use, ul vnR1r of wixch ceficul form of wotd\, IOI OIIC T ~ J S O I I or .mother, conic to Iw jdeikas~biy) .~hst,~r,rtcct \.nth ceaun

rrleaxlirlp trver .rrlc! dhtrvc dre meaxnng ,?hat ‘ire encoded at the ht level.

I o \alrri LI~ ' , wc I I ~ U S ~ tll~tw ~i I ~ I S ~ L I I I L ~ I U I I X I C ~ T ' V ~ C ~ two sore- of generalued

ilrrpiac atrare\ ( i r r c earl generalired rmplrc.lrraxes (< ;<;I) azc strll conversiinon;rl

lin11-illi , ~ ~ P I ~ C & L I Y I C ~ , AS 1)ncrot arid otlren jc g ('ohen 1971) pointed out, they C~IIIIIOC ~ i - 1 ~ .it the su1~-~cnte~it1al level I Pef,xult rxnphcamce~ (111) Are con- vention'div ,n\ot wtetl w ~ t f ~ cc-rtdiir Irrlgurstlc fimns, \ervmg as triggers. Smce

iliey amse , t ~ i t o~~ la t~ t ~ l l y latllcr th,m tllroilgh ( ;PIC~~II rcasc)r~lng, they can be rnrhcdtied

rX~c difkescrlce bcc~nicen L)1 ,rxlti (;<;I rs n t r t rrrc~ely the fact that UI are

( tanvc*risiorrsillv ~*s\oc rGkred with c ert'rln forms trl wol-cls GC;I thernselvcc rriay

gCt cr1r1ver1tic9xr,~lrlrcl, wrtho~it c e'1~111g t o be coavri~~trc>tial rmpl~catmres m

" iirc:r\rse [)I-iiieorists such as Lrvirrson use ttir Grit can k-itirl .C;er~erahzzd Gonversdtiunal Irnpll- r~t i i r r ' , iht- rivi.ii)ti of i s crx~urnnly iliistAeniy) liscnbetl LO (;me. Tltus B:1rt Geurts cvntes: 'C;nrrx'i. icid-a \<.ems io have hem iliat rf s convemtiorrd i n ~ p i h - d ~ r e d oi:run u&rz enough in the preseiiie oiari raxpreuioii i t , tlieli the irrrplrcatr~rr will icio~ei-iow hecor-rw conve~~tionully msociated with u ~tsell ' (Geiirts i:>!>X: u i 0). Note tl1:rt the C;nit,~iti rirraon of generalized c.oriversatio~~d inlpiicature n Irnmtirit: to rhr t nt ic i rz i (;i.iirts acidri-~srs to tlrt: xiui~cii i oC1)1 and its ow irr iiic tbrvry ofst-dnri.

Page 157: Recanati, François. Truth-Conditional Pragmatics

the stnct, Gncearl sense. At a certan pornt, however, they wdl rnev~~.rbiy become DI. The trarislhori &urn GCI to UI ~ ikcc pl'ice wllrn, '1s a result of convent1onali7ahon, a getierdrzed miphc,ittrre lows the property o f ' ~ n o r l -

detachabrl~ty' w h ~ c h Gr~cc use\ to char,tctrn7e iorivcrs'~tloudl u~lplrcature\ Wheri say~ng a certazrr tl i~ag canes a convcn;lt*o~ld u~~phc,~ture, (;lice

d a m , ~t IS not poss114e to fin<+ another way of sdylng the s.iltle t l~u~g- -~aot21er

form of words, with the same content -cvluch does riot ( a n y the lrn- phcature. r b s i s nordetdchab~l~ty Now the existence ofl~nguntli tnggcCn fox an inlphcature does not, by ztself; precnlrrdc 1t5 beu~g rrc>l~de~lc tl.ihlc If%\\ertmg a d~sj~~nctive proposltloct <-antes t l ~ e gmerxl~ed iri.ly,hcatuc that thc .ipe.tlrer a

not 111 a pos~t~on to wparately Assert my of the ~ % ~ L L I ) C ~ S , t h~ t 1111p11c dturr ~ 7 1 1 be caned whchever h n n of words a ?elected to ey?re$s the iiiqunc-tlve propoyr tlon m yuestlori. even & as t l~c result of tlic :;ereerLdz~at~on oftlie ~rt~pl~wture. the word 'or', whch a our pnlndsy IneLns for eq3resstug d~yurlctron, t l ~ s come

to be llssoc~ated w t b the ~rnphcatrrre a id ~ c t s 2s A 'ttnggcr' firr It.. SnU. onre iz

certutre degree ofront~rrztzo~zul~ziitzon has h e e ~ rmlzrd, A new po\\~lditv w~ll ,mse 'I he \n~plt~ature' wtll tend to be rout~~lely ge~rer~ted even 1r1 c orrfiprations m

wklrch ~ c , could not reiult fi-orn a globid iilnterence i 1'1 Gncc J Ir,lt I \ flow sub- scntenhd tmphcstures can be acc oua~trd for, ~n tire revncd ka~nework Whdt

starts hfe as a gerierahzed ~xnpl~c,rture bec~)lr~es c ~ ~ ~ v ~ ~ i t ~ o ~ ~ a l t / e d , and at B celmn point IS mggered even m contexts ul wknch ~t could not t>e gc:~eratetf w A ~ I

mrphcature vla the Gnccan ppoqt-propos~tlotrd n r e c h ~ ~ r t n . At tEu5 porxit cve no longer l-rave a GGI, b i ~ t a DI, chdracten~ed by the los\ ofthe nonde~cl.~ab~htv~tv feature. For consider the 'tmplrt ature' ~t anccs m hnguisuc cc'lnteut (sily, at the sub-sententral level) where it couM not 1 . i~ gencr~ted ma tbc C' J ~ C C , U I

mechatusm. In s~ich a context ~t mses only bec aise there rr t crnvenboo assoclatmg it t~ the f o ~ m of war& that happens to be. used. Were a not for the coxiventloli ofnse m nrtue of wh~ch the n~ll>lrcature i\ triggered by a rertars form of words, ~t would not be generated, In such a context. It follcrws t l ~ ~ t , tn such a context, the 'mn~phc~turc' c'ul be detx hed by cll,z~grrlg the foml of words ttrat It is no loirger 'nondctdd~able'

Nondetachability charactemes corrversational linplicatures k~ur tlirre 1s an exr.cptn~n: the ~xclplicn- tures which arise from the "maxim of maruler' are dctacl~~blr.

V s Gazda notes, 'to r e d off irn--glkatures C1.e. clcfiult lnrplicaturusl f~otn tllc terrlaxltic ~ntrrpret.t-- tion of the sentence (i.e. the proposrtlon it expresses) would be m~possiblc, rtrlce miriv chii;.rent sentences can express a glven proposition and 1-ruxiy crT tbcse will not rwntain rht- scalar llent atid tllus riot carry tile im-plicature' ((.;az&r ry,/y: j t i ~ .

Page 158: Recanati, François. Truth-Conditional Pragmatics

One might object that dehult impiicatures, thus understood, are nothing other than Grice's conventional irnpllcatures. But that is not so. What distinguislles cor~versational inlplicatures fiom conve~itional inlplicatures, according to Crice, are the two properties of noridetachabllity and cancell- ability. A conversationd implicature can always be denied, cancelled or defeated, wllile this is not the case for conventional implicatures arid other aspects of serxlantic content. Since default conversational implicatures are still carlcellable (defeasible), they remain distinct fiorn coliventiond impli- catnres, even ttlo~~gli tliey do not possess a high degree of nondetachability.

3. From pragnatics to seniantics

U\mg the notion of 'default unphcature' the hivo observahons we started from can be accounted for The detbult linplicatures are not conscrously avdable because they result from a 'cryphc and heunsttc procedure', not fiorri a riiacropragrntrc uiGrence conducted at the personal level; and tliey can arise locAly beca~ite they are not generated through a global inference usmg as prerru\s tlie Kxci t h t the speaker h a sad that p, I>ut Are automahcally triggered bv c ertatn e.iprr.\siom dumg the oilltrre processing of the utterance

In re~eltt work, several .jemaslticlsts-most prominentIy Fred Landman (aooo) and C;enri'iro Chlerch~a ((.-004)-have endorsed the rlotjon of a JeK~ult r~11pIic~~t~rre (as opposed to die clx,ucal, Cncean i~otion of an

rrnpllcuure) Thev l-rave put forward detailed proposal\ regdr~ling the para- digm cn\e ctal,u rrz~plicature\ Both Chltrchia and Lanclrlian reject what Landsnarl caUs 'the C;rice~n Root'. the idea that the scalar operation that denvcs the ilplrcattrre operatemon the o ~ ~ t p u t of the grammar, wlzere the output of thc g a n ~ r ~ i a r is the proposltron expres\ed by the ~ornplete uttcr- ance fristead, they hold that the default ~mpllcature (or at least, the "ore' of thc ~mplicatllre) \s derlved at the edrlrest level in the grammatical denvatton ot the sentence ciscerteci where an appropriate .icale 15 av'tdable' (Landmm 2000 379) As Ch~crch~a p~lts it, 'rr~~pIIcat~~res are not computed u j e r trtith- condrttonr ot (root) sentellies havt, been figured out, they are computed plxae by phrase ui tandem with tr~ith-cond~horis (or whatever coniposit~onal senantics computes)' (Ch~ercha 2004. 40) Thcy are 'mtroduced locally and

Page 159: Recanati, François. Truth-Conditional Pragmatics

projected upwards m a way that rmrron the sta~idarcl serllanhc recur\lor~' (ibzd.)

In Landman's fkriework, a rri~rrrer~~l (or driy otlrer scalx term) rs a "cdar trigger', that is, rt is associated wth a scde of altcrna~lves that 1.1 exploited rrl generating default irnphcatures. The imphcaturer (or rather, therr "~orc ' ) are

budt fro111 that scale, as soon ~oc\zble rn the grurrirrzntzral ilcnvtrtlon oj drc ,entenre, by negatmg the ]tern\ orr the scale that art rtror~ger ( 'oxlvdcr, Ccw example, sentence (ga)

(ja) Bill believes that there were fiwr boys at the party.

We first denve the inlyhc'iturc-ctrre (.ic) dt the levd of the e~nbcdtfed sentence (jb):

(jb) There were h u r buys at the p w t y

(jc) There weren't nlore than G~rr boys 3 t the party

From there on, while coaiposrt~ondly btrllctrrlg up the nrearlu~g of tlrc complex sentet~ce (ga), we budd up ,mult,meously its amplrcdturc frcm the mxnplrcature-core (jc), iidlowmg the semdtltlr composrtlort of (ra)

Thus, &om the level where the core oi' the ~r~ipl lc~tl l re IS derrved, cvr

success~vely budd up the follo~dlrig i)ai~\

that there were four boys at t"n emty that there weren't more than fiwr boy\ at the party

believe that there were four I>OY\ at the party believe that there weren't rncxc tbdn htrr boys at the p~rr.)i

Bill believes that there were ii)~lr boy.; at the party Bill behevec that therc wcrcri't nlore tl1~11 four boys at the party

The last sentence couesponcis to a deflatllt rnlpilcature of (ja) r-liar cdnrrot be generated in the clasnc'+l, (incearl karncworl; We cannot gerzerate the

scalar mlphcature 'Bill belleve, that there tveren't more thdn 1;)~tr boys ,it

The ~ d e a tliat xliphca~res may be cor~rp~ited at tile ptrm~d irvrl nukes 1ts hrrt expirclt appearance un Com&er 1984 663-4 (see also p 089)

Page 160: Recanati, François. Truth-Conditional Pragmatics

tE~c { ' ~ r t y ' I)) I I C S , ~ ~ I I ~ ~ A prece of miirrxn,ltrc>~~ stronger than (j'l) on some

\c,iiu I txe C.JGX~C ihiilg 11oLtIs for (oA), wb1~X-Y: by delrl~ilt. ~~npjlc &re\ (Gb):

As I;lnrciroiarr poirl tri a x r t , tk~c giobaI tnetlrc)tl le:rils m rocv vile re in a case like i l l i s . Wc i : i l l r i c > t gexicratc ( b l ~ ) by rregatix~g '3 piece ol'irlfornlation stronger

tlxari (Oa) X O X I I ~ s(:dc. '1'0 LICI'OLIII 6. h r the SC~J ; I I - iir~plicature (612) we rnust give r i p ihc ' ( ;E~CII , I I I K.oot' 2nd ;LSS~IIIIC" t1111t the (.ore of the il-uplicature is iierivcd bqjijrrr t i ic iiriivc~s:s:ll qti:rr~til.irr c.orrles irrto play. At m e:irly level in

the c:c>nlposition:tI p r o ~ ~ s s we clerive the pair

x,, b.isseid tirxec g r l s x,, kicsc~i xior r r iorcqthan three. girls

At: ,r iiihseyucrrt st;ige of the tlcnvaciolr e~ci'yy iwy i s i r~trod~ced and we get:

(oa) For every boy x,,: s,, kissed three girls (d) For every boy x,,: x,, kissc(.i 110 t i~*oi.i' ti,alx threc girls

C 'Kc-orrncA, t11c itiiplic.;ct.trras t h ~ ~ s gcx~cr.;rtccl by the ctrr~rputatiorld system of'

grLinmir,r;ir. rcri.i,riti i ~ ~ ~ j ) i i m t ~ ~ r e s ; they c.311 defe:;~;ut:J or c ~ ~ ~ c e l l e ~ l , by 311 sorts ot'nie:iris.

Iar C:hls-.rc:lii,n7s fklllle~Ork, sc'ihr tcr~xls ;ilnd the L . O I I ~ ~ ~ C X expressior~s t h t coritrial tlrrrrr :Ire b~ssociaied with two rnc;u111>gs: the pl;iin rrrcanirig of the expresson", wtlicli is 1-orllputed in thc rxsud way, at-td its strcnrgtl~ened

(irp~~er--.b~>~ari~luiE} rlic:uiing whit-li iricorpor;t~cs tbc scalar implicature. By t lv i l~ni r , rhc stn:~igrher~cti rllcaning is !>reli*rr-eai; but the irnplicatures tnay he carrc:ciic.cl by c i~c iirlgr~istic or csrralingrristir context, in kvhicl~ case one falls bac-L 011 rbc pPairi rrrearling. As irr Id~rr~llrr~an's kL;rilrc.\vork, the scalar i~r~pli- c.atlri'i:s, gi-r~c-r~ttetl by ncg:ttixtg t l ~ c iterrrs stroriger on the associated scale (or rather, the" \we;~.iLcst o f tllosc itcrxis), i1.r-ib nrltoail;rtic-lrlly iritroducecf by the

c-unrg>ratatioriLii systerrl of g)-;ulllll;ir, a ~ l t i their iniroductio~k cakes place as soon :IS possitrit akkr a sz&r ten i t enters the cur'ill?ut,rtic,n. As c-ontpositiori p r o ~ ~ ~ d i s , XIOWTVC*~, the i i ~ ~ p l i c : ~ t ~ ~ e s that have \leer* locajly irttroduced can

be filtcrecl out. 'l'flr origtrality of l:lxierrhia's position lies in his suggestion that :;caiar irnplic.:ltlrres arc riot ordy gerickratcd lry rielartlt, bur are also r~inoa~c,c/ hy dlgtiuiii in ctxrtairl lillg~iistic. i-o~ltcxts. .l'he (.ontexts ill question

:ire tlrosc rEi:ia- I;nric.rrnrrier origin;rily c-X~;rractc~ized : ~ s 'cbntailnient reversing':

Page 161: Recanati, François. Truth-Conditional Pragmatics

negative sentetrces, ariteccdeitts of c orldlt~oxlnls, drill atore gencr'rlly duwn- ward er-ttarllxtg envrronxnents. In \ucli euvlrorrixlellt\ he ltla~u rnearnxtg (w~tlrout the ~niplrcature) become\ infi,nrtatroxrall y strorrger t i~a r~ the

strengthened rneanirtg (wrth the u-tiplicnture) , \o that r i l a ~ l l t ~ ~ ~ l ~ ~ i l g tltc inl- phcatu~e would lead to a weaken~ng of rnforinatron Ltxiterrt I'he ~iefault generatmn and-sernoval of sc~l 'u ~mpircatures therefore moucs, wlthln grarrunar, the Grscealr search fbr trrmmal mfi,mzattvene\\ W e nt'iy perhaps t h k of the Gncean pat-proposltiond rriectiarnsn~ as hemg ttte evolu~oxl-

ary source of tlrc granunat~cal~~~ecliarr~srn wklrch Chrert 111a clc\ci~t~es It a as if a pragrnatlc rnecha~usnl lzad been irtc orporated lrito tEtc deagr~ of grdrrlrndr to mahe ~t more eificrerit."'

bverl though 11 presumably cvolved from n yragm,rtlc nxzc h,mlsnn itlvolv-

log the Grlcean rtraxm of quantity, tire cielult generation of sc Ja r 1mp11- catLires IS not ttsck a pragtnatr~ n~ecl~ldnr\~n 111 the hll-blooctctl seuse. ,is hotll

Idandmart arrd Chlerchi,~ mahc clear, ~t belongs to the cox~t~~~tt,rtion.iI system of garrrt~~ar. In t l ~ a respect Chlerchla's mci I aritlrrr#rxr's prc)po.i,rls are unrrldr

to that put forward byJon,ttllan Cohe i~ ln his c ~ s l y x,s'iult on (;nc c Shortly dftt"r (;rice dehvered the W1111;lxn J,rrnc\ I cc trrrta, ('ollen ~ n a ~ ~ ; . e t i his 'Conversntional l--lyytotlxcs~s\ctn the grouncfs that IC c'ii~not account for embedded ~mnplic,tturec; and he ofiijred h ~ s own 'S~r~~iant~cal Fiypotl~er~s' as a vlahle alterriatlve (Coheir 1971). 'l'hc Conversat~orb,ll t Iypothesls says that the rrnplr~atures assctclated wrth tlre log~cal corinct tlvcs result hoxn a convcnahond~ ~tifcrence, \vEtde Cohcxt'j alterriative 'Serrlarltlcal Flypothe- sw' says tliey are part of the mearung of the cotlnectlves (',ohen WAS swell- aware of Gnce's c ntenon, Mod~fied Occaxn'c K a ~ o r , wluckt cay$ tlrdt semes should not be multlplled wrthout rlccc\slty. Btlt he ttiolrdit i t was posslble to accribe ~~rlgie, unequivo~d tlrleanlng to the log~cal connectives, by treatlng certaxn aspect? of that nngle rneanlng as carrccllable I'lrus ttutb Chce and Colzert ascnhe a sxlide xneanlng to the connc~tlves, m cox~tornl~cy to Modifird Occ,~rn's Kamr According to Gnce, that medxlrng can be contextually erln~heci as a testlit of a praglil'ztlc ~nferencc, ,~c tordu~g to Cohen, ~t caxr be ~mpovenshecd by c~~ncelliiig out a ciefeas~l~le aspect of

'" Levmson si%cidatw that such a mnt:chail~srn wa\ 1ndct:ti rlerdcd t c ~ ovtxonie \&at he calls the 'encoding bottleneck'. 'The ,actual process of piloneat art~culauon', lic says, '1s a bottlenec:k III 3 system that can otherw~se nin about four tinles ijster' (Lcv~rr~on 2000: 6). 'I'ht: SC)~ULIOII to the bottleneck is this: 'tinid a way to piggyback lr~enning o n top of-the riir;~ning' [zhid.)

Page 162: Recanati, François. Truth-Conditional Pragmatics

hierclza as putting forward an ac~ount which, ltke Cohen's, 1s based on the acceptance of defeasrble smantic

features Such arl account IS sernantlc, riot pragnahc At this pollit ~t is wort11 reflecting on what, rn the overall process of

zriterpretlng an utterance, drsmpishes the mechaiums or processes that are 'pragnahc' from those tliat are 'sserriaitic'.

A pxadrg11atrc~.1ly pragmatrc process or mechanism such ds the Gncean generation of (clasncd) conversatronal r~nplican~res possesses the following tedtures

o It appeals to extrdmgmst~c inforxnat~on- facts regarding the s~tuatlon of utterance or the ortgolng conversation, background knowledge, etc It rs 'top down\at?ther than 'bottom up', that rs, it is not tnggered by something hrigmst~c-sonie aspect of the lmgmst-tc sign~l berng pro- cessed -but takes place m order to make sense of the cornmunicatfve act perfc)metl by the speaker {Note that thrs contrast a not the same as the prevrouz one h process rmy Invoke ex~alm~rstlc-contextud- rnfontiatron wh~le being hri~rst~cally tr~ggered m a typicdly bottom- up r r~~nncr Indemcal resolution 1s a cae u~ point. the procesq of corltext~xaliv as\igmiig a value to dn indemcd I\ tnggered by the occurrexrce of &at ~ndes~cal m the sentence, yet extrallnglllstlc rnfor- mahon 1s cleaxly and cruci,dly irivolved ) It r j globdl r~tliler t l i a ~ locd, 1 e , ~t n not pdrt ofthe stepmse process of cornposl~lortdly detemnlng a sernanrlc. ~rlterpretatlon tbr the sen- terice, but take3 place afier the global interpretation of the sentence has been c dc~~ldted

e It I\ trancpxrent (%v,~rl,ible') to tlie users ofthe lanwge because i t n a matter of 'spe,~ker's nieatung', and speaker's meamng IS essentldy overt ~n die iense glcfised by Gr~ce md 111s followers. the corivenatronal prot,%oiints r r w t be consciously aware of what the speaker means, vvhrle tliev t1et.d riot be cctnsc~ously aware of the gramrriat~cd meanrng oi tlie C ' X ~ ~ ~ S S I O I Z S used, nor of the processes through whch the memng or the whole n deterrmned on the basrs of the meamngs of the parts The output rt delivers ennches the Interpretation of the utterance rn an optror~dl nimrier, thnt 1s. there are contexts m whch the s ine form of words would carry t l~e plam, unennched mterpretation.

Page 163: Recanati, François. Truth-Conditional Pragmatics

The default generation of sc,d,lr mipltcacures described by C:li~rrchla d l ~ l

Landrnan (the 'DCSY', for short) possesce\ clnly the last of. tltzre featur cr,

Tbts 1s not suffic~ent to nialce ~t a prag11iatic process Ex&alixiguxst~c r&r- rnation plays no role--~t only comes Into play to defeat the deg~ult inl- phcature or to reinstate ~t (to 'freeze' ~ t ) m cdse of deGult rerxrov,d " fiche

DGSI 1s clearly bottorri up ~t n tnggerecl by the occurrence o f s c x ! ~ ~ renns

m the sentence. It takes plate locdlly arid .;nb-personally ,t\ p,~rt of t11e composit~ond procesc of deten~l~nlrlg the (default) truth-cont%t~oin of he sentence Those fedtures, and espec~nllv the fact that a IS l~ng~~?lr\tic.illcr triggered and autoniatic (context ~lldeperldent), put the L)(';hI ccjuarcly on

the semantic s~de, despite the optional jdcfea5lhle) ~ h a r ~ ~ t e r of thr o~lltput

In t h ~ r regard the DGSI 15 a b ~ t l ~ h e the process of i~~dexic~i l resolut~c>ra That process too pocseses onlv orre out oi LEI^ five teatures whit h c I I J L ' ~ ~ ~ te1-17~ paradigm~trcally pragnians: processes A\ we Iiavc. z~oticed 111 pdssrng, the process of lndelilcal resolut~on a hng~i~st~cally tr~ggered (bottom up) I t tskrs place locally and sub-personally m the denvatiori of the \entetice9\ truth corid~tions. It 1s nldnddtorv rather rharl optional (I e we h t ~ v ~ to arsrgrr 3

contextual value to the mdex~cal, xn vlrhri' oZ the rrder ofthe I,~i~gra~gc~) I hc only thing that 1s prapiatlc llere u the fact that contextual, extrdilmgtrr\tni infon~iatloo is appealed to u l abcrgIinlg 'a valrtc LO an rndcxrc,il

'Table 5.1 sumanyes the sn~~ilant-res &ind to~ltrd'it:, l )etwee~~ the L ~ X Z C '

processes we tisve bee11 t3Ikllig ,.;bout the C;rlcc,txl post p x o p < ~ s ~ t ~ c ) ~ ~ ~ l

Table 5.1. The similarities and corltrast-s between t l~e Gricem post-"-propnsitiorilxl>~~d mecharism, the default calcdatio1.l of scdar irnpl~cattrres, arid irldexlcal resolrrtron

"-

Extralinguistic l'ersirrrd- Levcl Global! 'Top- Optional? infonnauoo? Availabiiitv! down?

GPM yes Y a Yes ye\ ve9

" This ought to be q~ialitied. In Cbierchia's accomlr the extrAnpistic corrtrxt ertters the picture ax an earlier level: the relevant alremativer are 3 corriezru~lly sprafird iubsc.t of the alternauvcs induced by lexical scales. But this is only an addition to or rclinmerzt 0x1 a l):~sic rnecllstlism &at is islot &t~ciairient:~Uy context-sensitive (the way, e.g., irldexicd rclsolunon is). ltliiependerit of context, &re langtlage syscenr gives us both the (lexical) scales and the meails for xeneratitlg detjult imphcahires h r r l the scales. We we that system in a colltext-sensitive manner by ignoring ccrwi~r altenratives and focusrng on otht.r\.

Page 164: Recanati, François. Truth-Conditional Pragmatics

rxrc-t hdri~srri (( AWivl) i trnstmeii ns ~?ar,rilrg~l,~t~cally prtlgmatlc, the. default

t a ? t t ~ i ~ i t r c r r r ( 3 i SC.II,IL i~npi~catllrc\ desctrbrd b j LGarldrllz* md C h e r c b ~ ~ l (1 X 3 \ 1 ) , " U t d I I I L ~ C ~ ~ L ' L ~ rc501~1t1011 (IR)

4. I%rclgr~m~~rac irxlplicatit>n\ of 4~ll'r-locutlonary acts?

i",vcn choirgh ri-ir c.Xassica1 Griccan ;~ppro:tch csrlnot I-randlc t.rnbedded im- pEic":~tiircs, c v h k thr semantic. :rypro;tc.lr citxr, trrw rrray still itttelnpt to

acx:trunc iibr rIlervr i x j pragrl~atic (rvhcr t h r l in a st~mailtic.) franie.cvork; by givirrg try corrre ,~spc.ct: of dre Cinze'rn picturt~. For example, one may ccmstrirc: the r.c.Irv;rr~i Irr~plicaturcs as j>rlip;~il;itic irrlplicatiol~s of sornethillg

ot1rc.r tl.r,irr ;i st:irrd- done speecl-r ac t .

ILci-;iii tinc , ix r t i (;rii.c type ot';irgtltncrlt prit ii,r.\v:ird hy botli l>ucrot 2nd Cokrcr-I:

(.a) 4 :orrvcrsatioii,ll in~plicatures are pr'q<wiatic cc,rrseiluellces of arr act of s,rylrxg sc rroctl~ir~g.

(b) A n irc-t of i;,iyirlg sorsscttlirrg c,in 1-w pc~r-fc>n~rcd orlly by nleails o f a ccinr~dctcb iittcr,rrlce, riot by rrieans of 2x1 uriasscrtecf clause S Z I C ~ 2s a ijlsjurict or clie drlrecedent of a condition:d.

(c) 1-1 t-rlce., no iirrplrc-ature car1 be gear xvrzltecl :it tPlc sub--Eoc\ltiorl31?iti1: level,

r.e. ;it t l ~ c Icvel of' an ur~assertt.ti cb;rtsst. suc.11 as 'L di:jurlct o r the

'rriiccedrait oi',r iortditiori,il.

I t rs p m ~ \ . ~ i ~ l i ~ ti( rqci-t (I)) hy givirrg J xvc.,iker il~tcrpretatit,il of the notion of'

'"i;iylng9. 0 1 1 T ~ J C i~iie~~~rct:i t jort , ( ) l i c a 's:I~s' (tbottgh o ~ i e clues riot 'assert') strn,cthixri: by u~tc.rir~g a disjt~rlct or tllr :rniecc~dent of a corrditionnl, and

t in t : ' s:i\;ixlg tvlldt C I I I ~ says rn:ry carry c-orlvc"rs;itiorial irltp1icature.s. Alterria- tivc*iy, i C tint, $tic-hs to the strctlrg ititcrpr-etatiori c i i - tlic notion of 'saying', :I<-e:onirrrg 10 W I I I C ~ one floes riot "say' ,irrythiiig by ~ l t t e r i ~ ~ g a disj;_turict or the aa~tec-i:derrt oi'a corrilltion;tl----if; thereli)rr, crne accepts (b)-- -&elk orle may rqjcc-k (,I) :rrrci clirjirr tl.rat conversat i0~~1 irr~pjicaturcs need riot be praprrtic ~niidicatiorss of" ;iri ;rcr of siiyir!y (i11 tlti: strong sense) hilt ruay also be pr.~grri:itic i~irpllc;itloris of an act of \represerrtixrg' or- "lescr-ibing'--1vhere rcprescrxtir~g or ciescribir~g are things ttlr;~t wn I)c riorle by rrrearrs of an iin:lsscrtrd ciauce ~ c l i as a cli!jjtruct or t h e .rntec:edrnt of ;L coi~diti~~rlnl.

Page 165: Recanati, François. Truth-Conditional Pragmatics

T h e llne I llave juft de\crrhed (rejecting (.L) or (b), cieprmci~xlg o n the

lnteqretdtion t l i ~ t is grven of tile not1011 of 'saylr~g') 1s basically that taben by

Italp11 W,~lE\er 111 L I I \ reply to ('ohen'\ crltrcl\rrl o i ( ; r ~ t . (W,dkt.r 1 ~ 7 5 ) . W ~ l h e r drgries t h ~ t the wn,~<sertcci ,irltec edent of a coridrtloxuli5 u t v e ~ t h e l e z ~

uttered whcii the conci~tiorral I \ utteristl, and 'in t~x~yl~c~atl lre, he \ays, is 3.

pragrmtlc ~ ~ l ~ p l l c , t t ~ o n of drt tltteranc e .ict- -not IW es\arrly of a full bbclcted

~l iocut lor i~~rv o r eberl locut~ortary a< t l i e cvntcs

['l'hc (-:onversational t iyp~)tl~esisJ holds that by a pa-ticul-ir utterance or1 a p;trticul;tr occasion the spe:tker (-:\il coirvey rlrurt. than Iris utter;tncc strictly nrc;ms tlirortgh relying on a gcller:d recognition oi'<;nie's Co-c~perarivc I'rinc~ple. It i s tklcrefore concerneil wit11 litterances, whetller- tiley col~stit~tte self-standing cpcerh acts or nor; an utteralncr o ia suborciiriate clause, as in the :rnrrcedent oi;l cc>ndition:~i, i s still an Litterar1c.e. m d tirerefore I X I ~ ~ c~i ivey ~o~rvcrsatiorl;iU~ ITIOfC" t'liari it 1iter:illv rrrcal:s. I t rny corlvcy, fitr example, a fi~rtl-icr contlrtion on wlliclt the zoxiseiptcr~t is to be take-n c o drpcriti. (Walker 1975: 151)

Consider (:ctheir's origirral example:

(7) fi tile old Klr~g has (died of A i-icdrt ,rrt'kck 'uici ;I Krpubltc ha\ t-ict.11 det l.~recl, tllerr l or71 wdl be qrnte cot~terrt

It involve\ ~m rtllpll~'iture t ~ f te11'1por'iI orele~ 5t'trid~rtDy ,~cconllted ibr- by ;~ppealrng to the sub-max~in of orcte~lrnec\, lxrt oile th,rt, 111 tlrr\ p'brtic-ular

exanlple, ,%n\e\ ioi'rllv at the Level of tlte ,ultecedent of tlie condit~orral

(Torn wiU be contcrtt, or) (:ollcrt'\ st enilno, c>nly ~f the King tlit~d hejt~re a

fXeptrb11~ wa\ ticclarcci ) If W,llker I\ nght, t l~li , \IlouicJ rlot he a prohlcrn

Tile I I I J X I I E ~ of i)rcierIil~e\~ arg~,lltly cic~rl'~nd\ t t ~ t , 111 repre \ r r~t l r~g o r de

s inbing \eyuences of events ( w h e t h c ~ 'l~scrtivelv or riot), orte refr,itn\ ti-on1 reprecentlng tLorn ti? 3 dl1 fererlt cjrdc.1 thxn the order rn wl i~c h rhe \peaher

wants the ,tcirecwe t o tlirrlh o i tilent '1s h,~ving taker1 pldce (vvi~cther thc addres\ce'a 'tlirtthing' le\elf 15 a \ \c~t ivc t ' r l~nktt~g o r niere clttertarn~~ig) if the spe.&cr wants the dddre\see to tlunk of A ,xs 11.1~111g taken pbcc hc4?)rr U, he should. tty vlstue of the maxim, fl,trne h ~ s descrrptlon Irr such '3 wav tii,rt

the replesentatlon of A precedes the rcpresenratloii of I I , that IS, In sucl-t rl

w ~ y that, t t i the ~ J I S L O U T \ ~ , A IS ~lztrod~l(-ed beiore 23. In tllrs way tiit speaker

\pdres the hearer unnece\sar)i efforts Now tlir, c orlstrrtlnt 15 a cc~ri\trd~nt or1

how tei-tiporaliy orclered event\ are repre\er~tc.d or cleccr~becl; '1nt-l \rich

represerllatlctlr/des~ript~orl oT\eyuerlces of event5 lrray sitrely t,the place 111

Page 166: Recanati, François. Truth-Conditional Pragmatics

the arrtecedent of conditionals, or in unasserted clauses nlore generally. Thiu 111 (7) the axltecedciit contams a descnption of two events. the King's death, and the declar~tloa of a Kepubhc. The speaker does not nssert that tf~ose everits took place: he speaks conditionally. But the everits are noiietheless described, and the sub-maxm of orderliness therefore applies. In V I ~ U P of the prerumphon that the speaker respects the m&xIms, h ~ s descnb~ng the two eveiits In a certam order suggests that, when m a h g the ,uppo\itlorr c(rrresp(>nd~l~g to the antecedent of the con&tional. the hexer IS to tllirik of t l l o ~ events ds having taken place m that order (Or at least, tlnis sugges;tron wr11 be conveyed if tile tenkporal order of the events 1s communicatioxtdy relevnnt.) Uttenng the mtecedcrit therefore cames a conversatrrox~al i rnph~~turc whrch erlncfles the content of the supposition

the hearer is 11ivitcd to rrlbe, the ~mphcature prov~des, as Wrrlker puts ~ t , 'a

further cooht~on on wblch the corisequent is to be taken to depend'. Or cons~der a bel~et report such as 'Paul beheves that the King has died and that a K e p ~ ~ b l ~ c has been declared'. The speaker's describing the two events in

tlut order SuggesLs that, a( tording to Panl, they took place m that order I he speabcr docs not assert that they took place in that order (riot even that tl~cv took place). be reports Paul', beliefs Shll, his describirlg the events in a cert~rr~ order rn repcrrtlng Paul's behefi carries an imphcature relatlve to the tcrnpordl o~dcnug of the events In F'atll's belie&

In thi\ way, cvrth little e&rt (111 order to m&e the proposal more precise), we cat1 account for *ionie of the problemat~c cases. But i t is not Lertdltl that the \trdtegy I h,we outhned (foflowmg Walker) can be geuerrilized 'rrtd accourit for dl the cases. In particular, it is unclear flow it worllci apply to sc alar mlpl~cat~irec The scalar reasonmg appeals to the Idea thu the speaker respects the n lmni of quanhty, that is, giveb as much (relevnnt) ~nfom~atroxt as poss~bte; now ~t n f a from obvro~ls that the notion of 'glvlrig ixkbmlatlon' c,an be i~~vorced horn that of assertmg (or &om s d a n~)t?on~). as the strategy requlre5.

In sorne cases, adrnlttedly, the strategy can be ~nvoked in deahng m t h dleged scalar ~~nphcatures. For example, when the scalar term receives focal stress, the irriplxc.lture rrtay be construed as ans~ng as a pragmahc irnphcatlon i~o t of the act of ascertlng or gvmg mfomatlon, but of the sub-locutlonary act of stress~tg a partzctrlar word (something that may happeii in an embedded clause). By putting a heal stress on a word, one ~mphes that the alternatives to that word lri a context-t~aUy salient contrast set do not apply, that is, that

Page 167: Recanati, François. Truth-Conditional Pragmatics

the result of substituting them for the word beanrrg focal stress would rxot be acceptable.12 If tlie word is a scalar term, the sailerlt alternatives will typrcallv be the items on the associated scale, or perhnps the ~tems on the upper pm of the associated scale. The re~ulttrlg rmpl~rdture wdt therefore luob v e n

much hke a scalar ~mplicature, even 11: the rrrecharricm through WXIICJI 11: 1s

generated is quite dlgerent." Be that as ~t may, the strategy I have desc nbed, followmg w,&er, unprob~eri~dt~ca~~y applres to scdar unphcatures onlv rn that son of case. It is unclear tbat ~t can acconnt fbr the cnries rn w111sh 3

scdar imphcature 1s in the $cope of an operator wthout the sca1~r word belng ctre~sed.

5 . 'Local' speech acts?

Alternatively, we can mantain tbat c onv~n,tt~on;il uripllrature, a e prag- mahc implications of a full-blooded speech act, while rejectlrrg the ~ l , ~ r n i that no such speech act is performed by urtcbnng a d~yunct or the arltrccdexxt of a conditional On th:s vlew the sex~~entr~lj pdrts of a ~ o n ~ p l e x sentt,xrc.tb ,ire

used to perfon11 speech acts of clmr own Just 2s rile coxrrplex serrterric 15

built up horn its parts, tftt- speech act rt serves to pehnr r 1s 2140 h u d ~ LIP

from the local speech acts \s~l~icl.x the srntent~,li yarn of' the iorr~ple\ sentence are used to perfun11

The best example of tlrat sort oi ilrovc n prov~cled by t f iv cpeech act

theorehc malyses of corrd1t:on~l.i put forliv;trd 111 the early ~970s (cec e g Ducrot 1972, Mdckie r973). Accordmg to these ,in,zlyres, a cuizcf~aor~d sentence 'If P, then Q' may be coxi\trued 2s servmg to perform a cornpic*~ speech act conslstlag oftwo locd speech acts a ilnt speech dct wvl~ereby rlze

speaker makes a qupposlhon. 2nd a second speech act whereby, III the scope of that suppositton, the spedker assem sonleth~rtg I'he first speech act xc n p r C n m ? ~ A by azer,fig 'If I?', axid tlle second 59eecll act bY ~tt-mlsr th 1 Y""'"""'"" :C, """"

consequent In the context created by the fint speech act. In tlus frarrlewulk,

" Ths is a simplification. The expressIan ux thcc~s need riot he the word actudy bearing scr-ess but rnay be a larger constituent contain~lig it.

I think the inlplicatures generated by putnng h c a i srrcss on a -word--wlrether scdar or riot-- -arc. best heated as convenrional implicatures.

Page 168: Recanati, François. Truth-Conditional Pragmatics

16~2 ' i <)<;,8,; ' spi2jz4.'+1 ~c;l$

ncxhlng pri-vcrrts rlre firsr speech act, pedbrmed by rrreering the antecedent, fronri c-;irayirlg cc>rlvursation:d inlphiicanlrcs oftlie star~cla-d sort.

-l'lr~s str;iicgy 114: chat ivllich llatj Ireen used by Stalniiker to solve the ~vojccricsrr pn4"btcnr fix prcsirpposicit>r.rs witl~irl a pr:rgnracic Frameivork (%,if u;ki:x r 974: sc:c ,iisc, Kartt~~rirtt 1974). Consider a corijunctivc statement

'P aalii Q', wkierc- ( 2 prew~oe"". Rlr-horrgl~ titc g>resrtppositiorrs of the ~ ? ; ~ t s ;we norarznily irrhel-ited hy tlic whole, thcre art. cases in which the ctmlpica scritexlc-c 'I\:lncl Q' cvJII riot prrsrtpposc, 1<--- h ; , r esarllple if P itself e~xt;lils Vi (K:rrtta~rc:rr 1973). I Iwv can we account tbr that hct? St:d~l;lker

i . ~ f j bn t i i t " fi,IEowrng c.splanatic>x~. Tlrcb spe;rki.r who says '1' arrd Q'-first ~zsserts

I\c*rtd tlirn 1i5:s:scrrx (0. 1:or (.J tc:, IJresupptrsc. R is irr ~t tc:, be assertyable only in a contc'st i r r wllicll 1% is assirsnvd to hoW. Ntsw tlrc speaker's asserting tJ cla;lrrj:cs the context by atltling Ptro the conlrriur] grornrd, in snc-h a way

char thc coxisecjixcriirs of I>, irlcllidixlg I<, wJi themselves be part of the ccrrrlarroii grcrtrrld wlreri tile cortsecjrrcrrt, Q, i s nttt'sr(l. I t fol'ol/ows that 'P and (2' \\ill i>r :isscri:ri-iie riot urdy iri cz>ntcxrs in wlriclr I?, is antcc.edently assun~ed

io hoici, 1711t l r l :illy (.011tt"xt (silic-tx the prcsuppctsioai rionnn.Uy camed by (;! is \ir~tcrxi:rliy' s;irrsfiec.i hy the tirst sur!jirxli~t in tlrc corrlplex sentence 'F' aarlcl Q'): irci1t.c -1":uitl 6>', corrmry to 'Q', rdoes not presup17c~e that R (i.e. it is not ';asacrt:~bic crariy III CcrrlLCxts in svhich R 'is :r?;sirrrrccl to irold').

StdJli , ik~r ~ , " ~ v c s the same expl;i~latiorr k?)r coritiitii.i~~al sentences irr which a . i ~ ~ > ~ ~ ~ ~ ~ x s k i c r z ~ n of thc conseyuunt is irltc.rrially s,:~tistioii by tile antecredetlt, as in 'H"li'r~irr-i. EXXS A king, the king of Fraa~~t- is bait!'. X ierc too he describes the cliscoeirsc: a?; irlvolvllig two spe t~ l l acts: a first s p ~ ~ ( : h i1c.t ~ f ' s ~ l p p ~ ) s j t i o ~ ~ , and PI act of asscrtiori ptv-firrnted ir t thc tcrril)or:Iry ct.rtItcTlt creattd by the first

speech :;it t. /'hc S:II~I~" so1-1: oI'31iaIysis ~ ~ 1 1 1 e;~siiy be cxrcniictl to disjunctions sircii ; i s '1iitiie.r I.r:rrx.c.: i s a Kepulilic, or tlre XCrrig oTL:mnc-e is so shy tl~at one never \ecs h i ~ n in p?h1ic3.

Wc c-cruiii ciiJ4apt this ai~alysis t(.i our r~xaxtrjries inivolving embedded irtr-

piicaturci. 1:,rc-ccE wid) a ~lifijcillt c:ise S U C ~ I 2s 'Jotl~t 1'1;~s five or six children', wlrere rPrc i~vtritiy- iniplicattirrs Edi \vit11i~1 tlic scope of' tile di?junctjor~, we rriay rergrie tiiar, jrrsl :is '[I" :inti (2' i:orrsists oftwo such-cssive assertions in tile St~1r~;xlicnri~irr Iiarilework, ' P or (2' also cc,i>sists oi L~ZIO sticcessive assertions.

The ciiffercrace hctwccn 'P and (2' anti '.P or Q' is that the first sequence of .tssertitrr~s is coxyuni:.trve, while t11e secc~nd seqtrcrrc-e is disjunctive. 3'0 say

t1t;lt ;r sccjxrerrcc. of ttvc., assertions is disjur-rcrive is to s ;~y that the second ,tsrertioil c l~~ ; f i i i i e . ; tlrcs first, ; ~ r l c l is prvsentc-cl ;is hoidirlg just irl case the first

Page 169: Recanati, François. Truth-Conditional Pragmatics

EMUI:.I)l)EI) IMl'LlCA'I'Ullt-S 163

assrrtiori turns out. to be fabe. '1' or (2' tE~erelore lllearls sornethirig like: 'I>; but if not-P, then Q'.

It is ensy to check that, ort sad] ;m analysis, wi, c:rr~ 1i;lntIle etnbe~!iiiezl irnplicatures. The speaker first asserts that Jolrn II;~s five ct~ilclrerr, thereby

conveyrllg the ~nrphcature that he has no Ini>re t h ~ n five l; hen, hy \,~y~rrg; 'Or he hac six', he zset-ts tliat, lfJohn does rrot tl'ive ex,lc tly five chddrerr, Ire bas slx (thereby corrveylrtl; the ~mph~at t t re '110 lliore t l m ux')

'The problem wit11 tli~s arial.lyu\ w tltat it blur\ '8x1 lntuttlve cintlnctiorr between a disjur~ctive sequence of assertions, hence6jrth tc) be called :i

"disjunctive assertion', and the assertion of a diqunction; a cJistirtrtion analogous to that between a conmtionrrl ssertiorl (e.g. 'If you are hungry there are cookies in the sideboard') and the sssertiott o t ' ;~ corldiLiol~al. As a r ~ ewmple of disj~rncrive assertiox~, consider:

J o h n has tive children. O r he has six. John has five children; or he has six. John has five chiliireil, or six.

I-Icre, clc,rrly, first assertion to rile c'tfei-t tltat JOAII ~ , L S five ih~idreil is

followed by a second Acsertion, intro~lluc~d by 'or'. 'Or' rrtdlc,cte\ that the second assertloll is an alter~~~ittve to the lint assertroir. 'Tiin, C'ornuher

sngge5teci (1982 88- go), can be cached out by repre\cnt~rrg the content of

tire cecor~d assertion d~ a c~ridltio~ial w h ~ w mtt-c ederit 15 tllc rregntlorr oFtE~e first 'istertlon (1.e the negatlon of what ~t aueit\) '* Si, h r \o good 13ut when

I s,ty 'John has five or \IX chlldr~rl', ~t (toes not iectii rtidr 1 h r r Ascrt that John I-ian five ch~ldren and tlren a w r t that rrr the appo\ltc t L~sc Ilc 1 ~ 1 5 SIX.

This ex'unple Ir tnosr naturdly ~lnderstuod '1s the Ilsserhotl t r C ;I (\iisgle) tfisjunctivc prt-rposition--a re~idirlg wl~ich car1 be ~n:ttle explicit by llsir~g

'either. . . o r . . . ': 'EitherJohn bas five drildren or hc has six'. 'This (azlrmt 1)e interpreted as a disjunction of assertions; h r the speaker a t n o yc-iirtt :merts that John has five children: f ro~l .~ the vely start, the pr-oposition that Johrr 11;~s Gve ctlildrerl is preserrted as one of the disjtincts, only the tlisjunction bei~rl;

asserted. Yet this does not prevertt the e~nbeclded rxai f ly rrt~pllcdt~rre frotlr arising.

'4 Com~~ller's strggmon inriccms cliqunnctwc (or, as lie put tt, 'd\e~-~i~~tirvve') q~w.~tlo,zr. but li i \ propos~i easliv ~cnc~r~il~zrs, lie hlniseif points clut (C:oniciIicr rt18:: <iq 101)

Page 170: Recanati, François. Truth-Conditional Pragmatics

1x1 responw to thls abjecaon, one nvght grant the Intuitive dlstinctlon

between a digunct~ve ascernon and the assemon of a &sjunction (or be-

tween a conditional aser-tlon and the assertion of a conchtlonal), whde

holdu~g that diyunct~ons, l ~ k e cotldltlonals, can thenaelves be analysed 1x1

speech-ac t thrt)retlc terms It a true that, when we aswrt a disjunction, we

do not separately ascert either the disjuncts. But the consequent of a

concf~tlori;ll is not r ~ n l l y asserted e~ther, and that fact does not rule out a

speech-act-theoret~t andysrs accordxng to whlcl.1 the consequent 1s locally asserted, that 1s. as.s\erted 111 the local context set up by the suppositron of the

arite~edent The I;;une sort ofanalysw in tenils of local speech acts and local

contexts rn~ght be ~t tel~ipted for d ls j~~ic t~ons , a d for complex utterances

generall~

13ut tlm w~ll tlot do 1.t we treat atly ofthe d~qtlncts m 'FltherJohrr has five

chrldrerl ctr he 11'~s SIX' AS locally acserted, J.S suggected, then it is clear dlat the

riobon of iocd asserttori we use is 11ot tlir full-blooded speech-act-tl~eoret~c

notioir of assertion, but A sexmntlc surrogate T h ~ s 1s son~ethulg that has

been itldependently noted in conriect~on ~vi th the Stalnakenan dnalvsis of

rond~tionds I he Dut rot arid Mdckre, Stalilaker 2nd hls hllowers say that

wheli wc 'rssert a c ondltio~~al, we assert tile consequent in the local context

created bv uttcrrrig the a~ltecedelit As Lanhian eenlphas~zed. however, the

'local tonteut' In tvlllctl ttldt d~sertloll takes place is not d real context

7 he context nt wlvc h we evalurtte tlie assertton ofthe consequerlce 1s not the actual speeih context, hut a context wiuch denves from the actual speech context b\ xd&tlg the mtecederit rhrs context rs rdled die loccif ~onttxt But of course the cotlseyuent t~n' t u s ~ r t e r l m the local coiltext, and the local context tsn't an actual speech context 7 hr pre.iuppoution\ that denve from the actrral aqseruon of the serrcence In rhz rtctud speech context are chxactenred in t e r n of what the parts of that ';mteoce t~ottlif prewppose dthey tueur asserted m a local context The local context n derived from the actual speech context, followxng the semantlc compo- sztron of the sentence l'hr\ mean.; that the nohon of local context 1s a _oramvnattral

liotrron (LaxlJrr~u~ 2000 237)

I,ax~drrnri tllen goes on to stress the diEer-ence between the pragmatic notion

of context st~r~dardly used in implicature theory, and the grammatical

notion of local context used in presupposition theory:

" l'hc emphasis in d ~ s sentence is inine (FR)

Page 171: Recanati, François. Truth-Conditional Pragmatics

Whde presupposition theory has been rcga;~lrcied \incr the r,trly severlnes as bit\rc ally a theory of local context, the notson has been dl but rgnp;nored 1n ~rnph~a turc : theory . . The reason IS, I thnk, tint uril~ke Ibr presupposiDntls, there i s a coi~seii sus that rmplzcatures are dmved dzrritlyjorur tlzc rlctulrl spetth ~ituatlort I f locd context IS

relevant for ~mpllcature well, t l u mean\ rhar we cannot, follow~~ig tile Gnceal Root, let the theory of co-operatlve m(i)matlorr exchange denve thern from the actual assemon of the sentence in the a t h d context It nlcans that we have to let the grammar denve thern from nnplrcature', that would he der~vikd iflt5 pxts were

asserted m a local context, vvlilch itselt. IC dzxned &om the actual context Collotlr~ng the semnantsc compostson (1,andman ~ooo 237)

I conclude that, suitably weakened ro as to avoici die conf;s~on between A

&sjunctlve (or con&tlonal) assertion and the assertion of a &?juiicttor~ (or

con&tional), the alleged speech-act-dleoret~c analyus 1nz;plred by S t h ~ k e r

collapses lnto a semanac anxlyw of tl-rc \art advocated by <:h~erch~a ,lnd Landinan It is not a genume dtcrrkatrve to that sort of andyas

6. En~bedded implicatures and free e~~nchrxzcr~c

A last strateby is available to the thetrrist who \want$ to reslst the semdntrcr

zation of embedded mplicarurc\ I t relxcr on ttie clisnrtctlorr be&-een pnrnary and secondxy prngrnat~c processes (Rrcanat~ 1989, ry93, rsuoxd, 2004). Secondary pragrlat3c procei.;e\ ate posr-proposlt~orl nli2rcntc-s a la Gnce: In lnterpretulg 311 utteraxe, what 1s trnphed, m the lrltultlve senxi, IS

inferentially denvcd horn the ipeaher'i, s,lylng what $/he s ~ y s (in ,a way that

sansfies the avallabihty condition) In contrast, prtniary prai,mlatic procevscs

are pragmatic processes at work 111 the ven, deterrnlnatlor~ ofwlrat I \ r ~ r d

For example, we need to assign rndich-lc*d~ and other context-srnsrtuve

expressions a contextual valite m order to fix tmtll-concflt~crr-ral Goriterlt

Thu contextual process of vJue as\lgirnent, w h ~ c h I call "a~uratlon', 19

~rreducibly pragmatic. consider~tions sclat~ve to what the speaker uie,uls

plav a crucid role In that process Shll, ~t coirmbute\ to the cleten~~rn~tlon ot the utterance's semanac (tmt11-co~zdiaor~d) conterit, hence ~t counts AS a

'pnrnary' pragmatic process.

In contrast to secondbuy pragn'itlc processes, p n m q pr*itgmatic

processes operate locally rather than globdly, ulJ they do not satisfy the availability con&txon. In these respects tlrrey are lrke the dehult gcneratlorr

Page 172: Recanati, François. Truth-Conditional Pragmatics
Page 173: Recanati, François. Truth-Conditional Pragmatics

P M t l f Xj111 11 [MI"! I < A L i l R I . 4 157

back), but the property w111ch the expression actu'rlly contributes to the

truth conditton5 m this utterance 1s i ~ o t a property of cais but 'mother,

qrrternatrc aUy related property, rrtlrnely the propelt): A car owner has whet1

fns or lter car ha\ the fom~er property In co~itnst to \aturatxort. the proces\ oi preci~~ate transfer n ne~ther

maildatory 12or bottom up. I t rnay looh as ~ f , ~n nn Iittcrailce stlclr as '1 arxi parked out back', that prc-tc ess nttrst take p l ~ c e, becarrse there 1s a l~rt, "u1st1c mlsnlatch between tire predicate (which denctte? a property of- cars) ,md what ~t 1s applted to (,i person) B L I ~ such nrlslnatch 1s not neces\'try fitr predicate transfer Just w, througll transhr, "1 he ltan s'andwlch left cvrtttout

pdylng' IS uU"entoocj as saynig 4orxethlng about the ctistoiiler who ordered the smd\vich, " f i e hnln cdndwlch sttnks' can be $0 ~inderit.ood, m n. ~uitablr

context, even though the propclty of s t ~ r ~ k ~ n g poterttldly ;ipplles to smd- \v~ches as well as to customen.'* 'Tire proce\s ot tlansfer xr not a l~r~pt~stically

corztrolled but a pragnratrcally controlled pragmdtic process it I\ not tug- geled by sonreth~r~g ~~rzgjlmctlf--some acpect tlie ~mgulzhc stgi~al bung g3rot e>\ed-but takcs pldc e In order to mahe \crtsc o f the c ornniur~~cat~ve , ~ c t

perf'ornled by the spe~ker Moreover, ~t lr oljt~)x~"ltlzere ,ire C O I I ~ C X ~ Y I ~ I

wltlclr the sdltie t on r~ ofwords \vould c a ry the plam Interl,ret,rtiorr, wl t l lo~~t tramfer In some contexts, 'Fhe h ~ r n smclw~th stulks' t*iUc\ bout the sarrdwrch, r n other contexts, through tmnsfcr, rt tdlks abo~lt the cu.;tomcr Wl~et l~er or not predicate transfer t'lker pkcc 1s A wholly pragrnatlc rnattcr It IS not \on~ethulg t l i ~ t IS dlctatec! by h n p ~ s t l c conventzora

Even though the pragnianc process of preci~eate traxi\&r 1s optronal (rather than manc.larory) and top down (rather ttlarl bottorn up). it still takes place locally and ~nterferes wxth the proceTs of \ernantlc con~pos~tton

In artother cla.\ic example, 'There 19 a Lon 111 the courtyard', ' l ron 'a r~ be tlnderstood, throug11 transkr, 111 the repre\enkxt~oonl srxne the thulg that 1s s ad to be In the t o~rrtyard 1s riot a (real) hon but 3 reprtrs~nfatton (more spec lf~cdly, a statue) o j lion Now c onsider "I here is 2 stoxie lion tn the courtyard' What w sad to be tn;lde of stone here? (:le,ziy, rt 17 tlle statue, rather than the llon whlch the statue reprewnts 7'111s ririlplc fact shows that

the process of represent~tlunal transfer murt hke pl'rce befart, the cornpcl\l-

tlorl rule assocrated with the noctn-~~oun comtructroxl applres to the

'" I :un indebted to I3ar1 Sperbrx ius tbls example. The ongrtal h m .s:~t~dwiclt cxan~pie is, of course, N ~111t1erg.s.

Page 174: Recanati, François. Truth-Conditional Pragmatics

-- - - - - - - %S hMBEI>DEIl IMPLPLT<'ATUIC\ AN11 FREE ERRICHMLNT

-

seniantrc valuec of the nouw 'stone' and 'l~on'.'~ If predicate transfer apphed globally, aker the g a n m a t ~ c d y tnggcred corrrposxtlon rules have appked, the ~nterpret-atron we would get for the noun-phrase 'a stone hori' would be sometlslilg hke: a rcpraentatzon o f (a lzon t h t L mu& $stone). But the correct mterpretabon Ir (a rqrecentatzon of u lion) that 1s made of stone. We must therefore g~ve up the Cncean ~ d e a that pragnratic processes operate globally on the output of t11c garmar.18

Acrordlng to TCP, just as indewcal exprevlom are standardly asqigned two level\ ofsern&~ti~ vdue ("~hacter ' arid 'content'), we ~riust dwonigu~h two levels of serrlarttic values for orcfinary, nomdelncal expressiom. Semantics asi1gn5 lemcd values to iirriple expressions; pragmatic5 then optiondy comes mto play to detem~ule the comyositiunal vulcca which thore exprBsions assume 111 the iingugunt~r and extrdusgueuc context 1x1 which they occur. Composi- tiorial value\, not lexical values, are what undergo semmhc corrlposrtion. In other words, the cotiipoclaon rules detenmne tlie value of the whole on the bllris of the pqmuttr~illy rietemtvted conrposrtronal values of the parts.

Prehcatc tran.if.er IS only one among a fmdy of pragmanc processes that h ~ v e the properties 1 have I~sted: they are pragnatrcally controlled (top dowri) rather than ~inguichca~ly coritrolled (bottom up), they x e optional rather than rnarctdatow, ar~d they take place locally, thereby interacting tvlth the compocit~ond detenr~lilntlon of truth-conditional content. This famdy ofpnnlary pragm1,itlc prc3ces~es 1 call 'modul~t~on', a.i opposed to saturation (Rccanatl mod) Modu1,ition takes as input the ~nean~ng of some e~xpresslon [whether urnpie or complex) and realms as output a pragmatically denved nlearring serving a a\ orr-rpu41 tioaal v'ilue.

Air~otig dle proce\ses of modulaaon that a&ct the trutis-condrt~ons of trtternnccs, the most typl~al ~ r ~ d pervmrve is fi-ee enrichn~ent,'~ 1x1 virtue of w111ch an exprctrcssion is contextually gven a n~ore spec~fic interpretation than it lrterally eiscodes Through free enrichment an expression takes a

" Note that ltiis conipnsit~ctn rule itrelf 1s context-sens~trve (I'artee 1984: 294---5). The denotatioz~ of riie compouri~l results fiom ln~ersccti~~g the (literal, or praplatically derived) denota~on of tile head noun with the set of objects &at bear a certain relation K to the (literid, or pngmaticaliy derived) denotatnon ot'tbe modify~ng noun. T'iimt relatior1 can only be contextually determined. In 'stone lion', 1K ' is typically assrgrxed the relation bein2 made 6 but in less access~ble contexts a different relation will be ~qstgned to the variable.

'' See Sg (1981) AIJ Jackendoff (1!)97: $5 and 6 5 4 ) tor simdar poomts. '" fit the pragmatics literature this process is J o called 'narrowing', 'strengthening', or 'expansion'.

(There we a couple of other labels as welt.)

Page 175: Recanati, François. Truth-Conditional Pragmatics

prapatscally denved denotaaon that is a subset of the irutlal derlotatxon (Carston 1997). Thus, to take a clasrrc exaniple I hdve dready ~rrleilha~led (dacussed by Nunberg and Zaeneti 1992). we undentand the ~.n,bs tenn 'rabblt', w h c h hterdy means something llke rubbti .stgiJ ac tlaeaning rubhrtJtdr In 'She wears rabbit' and rtzhbzt mrut 1x1 "She eatr rabbit'. Or, to take example discussed by Searle and arralyced u~ Chapter 3, we rxncfer\.l;znd the word 'cut' differently m 'c~it the gr:r,uskalid ul 'cut the cake'

Though the occurrence of dlc word 'cur' I\ htera! 1n Potli] utterances , and though she word 1s not anzbtgroui, rt drtc-nn~nes d1Eermt sets oftruth ~i)nti~nons h r the dtfferent sentences The ion of tlz~ng that constitutes cuttnng the grm a

quite dttTerent Gom, e g , the sort of thmg that consatUte\ cutting a cake One wdv

to see &s n to iniagne what constlmtt>$ trheyrng the order to ~ t t r iomuthr~rg 11 iomeone tells me to cut the grass and 1 rurh imt arici \tab it wth a knrfc, or xi I an ordered to cut the cake and I run over ~t w t h a hwnrrxnwer, in cat h ca\e I \vrU I3ave

Wed to obev the order That r\ xior what the speaker meant by his lrtridl and serious utterance ofthe sentence jcie,ule rg80 222-1)

Both 'cut' a d 'rabb~t' are gverr cor~textually spcc~fic intc.vret,itron\ through f;ee ermchment. Of course, the hngni~tic context play\ arr ob.vrous role here, but tiee enrichment reratdrrlr a romteatually-drr.vcn (top down)

and o p ~ o n a l process. Nottnng prevent\ 'cut' in ' C Z X ~ tlie gas\"fronl helrrg contextually Interpreted m the scnsr crf s l t i ~ rnio rtrtp3, or 'rr,rbb1t9 rn '5he wears rabblt' horn being i n t e ~ r e t e d in dle senre ot' rabbit rncaf And notlirng prevents the rneaixng of erdlel esproslorz &orn rerriaiiiilig conteu~ually pla~rl and unennched ('Aiker the accrdexrt, tltert w ~ s rahh~t ail over the hgfrway'). These proyernes, wtlich tree rrlncllmcnt sh~res wrth the other pragrnatlc processes in the mori~tLtrorr fa~zm~ily, arc. not exi~ibited by yrocesber of the saturatlo3 faliuly {~ndex~cal ~esolucron, etc ). We n l u ~ t therefore complete orlr table and n i d e room for niod.tJation alongs~de the three type.; of process we have already de\cnbcd (the C n r r m pozt-proporrtrcrnd rnecl.~anum, the default genelatlon of sc~lar imph~~tures , 2nd r'3tura~orl)

Once we acknowledge the pn111~ry p";1.g111at1c pwi-e\ses 01: n~o($ujCal,lt~on

(Table 5.2), a new approach to errlbccided mlpficat~ires becomes dva~lable. V e can consme &em a a partic~~I,x ( a\e of lfee e~incl~merit, whereby the n ieamg of, for example, scalar term:, n ct>~itextx~aUy strengthened Tlus Ir, mdeed, the view whch r m i y advocates of T m t - n o Pmpatrcs (e.g. Bach 1994; Rezu~deirbout 2oo2u) acnctually hold. They take die dieged

Page 176: Recanati, François. Truth-Conditional Pragmatics

I 'a iAe j . ~ Atitlrrig iiloaiulatioxl -- - .- - -

Extra- Perso~~:xl (;lobal? 'f ap-- Optional? l~rrgiiist~c lzvcl down? i n f i x i ~ l ~ a t r ~ r ~ ? Avaaability?

- . ~ -. - -. . -. - -. .-. . . - .- . - . .. - .- . (.in< c's post- y cs ycs yes yes yes ~>Tc>~~>sIL~<>I?'+!

niecli,irrisiii

Motl~rl.itxc,ii yes no rr o ves v es .- - - -

inrplii-atlirc\ to be Yirrrplicltnrcs', &:it is, asprcts ofthe proposition expressed

whicEr ;ire provrtlrtl by Cccly erlnc-hixlg or cxp;uxclifi@the literal meaning of the <icritcrrc-a>. ' l ' h ~ fi1c.t that the alleged ir~~~>\icrctrrrcs fir11 widlin the scopc oC operacon is i:tkcrr t o be the Iitxlir~s test showing that they are not really ct>ilvcrs:itiora;ii i r r ~ p l ~ c a t ~ ~ r r s L~CI-~VCC~ through trbw C;ricean rnecfiar~istn, but

jlragn7;itic c.onrstitrients ot'.cvll;lt is s;iid (Rccanat i r o X t j : r r a---14, 1993: 269-74:

C:arston 2oo.t: I y x -7, roof).

7.. Cclnc-In\ r on: dehult irnylacatures or free enrichment (or both)?

Wc P-r~vc 5ccbra dr,u il-icre ,ire (at Icd~t) t w o VXA~SLC dpproat hes to embedded

x~~iyhcaturt~z ierr2,itlhc ~ppr0<1<11 UI teniis of- defdult ~rnphc~~tures, and a

pragmltlc s j p p i c ~ "l" 1111 tcrrnc of fi-ee crlnclrrrxle~lt. Wlttcfi one I\ to he prckixrcd; Kdi, I a i r IIOL sure tlut wc rc,rlly h,rvc to choose. 7-0 corsclude LIIIF ~Luptrr , I ~ 1 1 1 ,lrgtie th,rt the two vrcivc w e have discussed do not

i~ecr\\,anly \ t , i x ~ l rrr corrlpetrtron to C ~ C 11 other

It r \ rruc. th~t iiiotlr cl?eonc\ providc ,rn a( coun t of enlbcclded mphc~tures

bodr rrrAc roofrl for process of stre~lgttserung t h ~ t 1s optional arid takes pl<xce lordly I t I\ trtic also tItLit they o&r corii-hcat~g chamt taatlons of the process

xrr clutb\tii>r~ Ac cordrng to (me tlieory, thdt pmccs\ rs context-mdependetlt d l i d ~ C ~ U T I : ~ to !ht" c omptl t~l t l~r ld sy-strrn ofl;~x~inrrn,cr ALC or&11# to the other

Page 177: Recanati, François. Truth-Conditional Pragmatics

tlleory, ~t w a luUy p ~ a g n ~ t ~ c , c ontext-Jnveii prstceis. Stdl, I thlrtf, tilere n 1 1 0

ciowimght ~il~oinp.itib&ty behvee~f, these appro'u hes, .qq-wxra~~ccs x~otw~th standing 'They are mcct~t~pafible only ~f we auui-rle tliat there Ir ,I srny$e process ~t suke. But we carrrlot p r e q y p t e t h ~ t that 1s so, koj, it\ we shd1 we, that 15 one

ot the yuestlon~ ,kt I ~ ~ ~ L I C rn the cleb,~te DGSI-theonsts argue t h ~ t s~dldr strengthening is l~rrgursttc ally triggered

and context indeyerldellt B ~ i t the default ~~ltylic,itures they po\lt uzn be defe'ited, 111 a \uitable c or~teut ( ~ t they couldn't, they wonldu't be cnnver\d- tiorral ~mpl~c~ltures) I-ronl the poiut of vlew of D(l;\l tlneory, the extcilm- guist~c context plays a role ~t l ca t In the fixutlorl of tire mipl~c.ittire--a

cosltextual proccc\ wl~lch dctermrilcs whether or not the deGrllt lnipl~ca tur es are defeated or (rn the event of def.irllt rex~lctval) w l ~ e t l . i ~ ~ or not tlley are 'frozen' 'rnd mnrntalr~ed despite the do\vnward erttnlrng operator. It i-i)llows that tlrerc arc two proce\ses 'it work In the intcrprctatlon of scalar ~niphcatures, 311d two c o ~ i ~ p o t ~ e ~ ~ t s it1 the overall theory One conlponerrt

belongs to sernant1c5 ~t c oncems the def~ul t generatlor1 of ~~npltcaturrs I he other con~pot~eilt 1s pragln,ltlc anci c oricenls tvllut htz~?pt.rrr to thu dt*l;rzlltr udtctz

the wntorzie u z i t t e t~d zn tt teal /cwlcst krtc-e thrrt are ~LVO iompollents, AIIJ two proces\e\ jo~rltly at work 1i1 the 1rlterprct.ltion of scalar utterdrices, r t

may be that the tw(3 c ~ n t ~ ~ t l r ~ g ~ h ' ~ r : ~ c t c r ~ / a t l o ~ ~ \ of' 'ttie' lo( '11 proce\\ of

optlortal strerigthenlng 1 have mentic~rlcd are ui fact nonc onfl~cting ch,tr-

actenzat~ons of two d~itinct processes I t ntay be thlt there rs both a hngu~sticall\j tnggereti, context independent prates of dehult gcneratiorr of ycrtentzul ~mplic'mlre\, 'tnd a prapi,xtlc process of free e n n c l i n ~ ~ n t u h g 2s pr t la l lnput the o ~ ~ t p t ~ t of the prea~otrs pro< e u md lead~rlg to the ~rcrll( i i

strcnghening o f the nieattlrig of tlie scalar tern1

O f course, one need not Accept the existence of default unplsc<ttures 1n the first place. OIK m < ~ y deny thelr e ~ ~ s t c n c e and actourit fbr ernbedded

tn~plicntu~es .iolelv ul terms of flee enr~ctlirient Grantthd, one 13ecd not accept the eqstence 05 free erinchmcnt eltfler But nothwg preve~in a theonst from acceptlrtg both default ~mpbcatures and free ennt lxnent---

that n nly p ~ i n t . At l>ottonl, tilere &re two dlranct quest1oil.i II- we provide a posltlve answer to either of the two qucs t~o~~s , we h v e a solution to die problexn of eritbedded mrplicatures. Anyone who @ v e a poutlve answer to orie questLon may therefore safely give a rlegatsve answer to the other. But he or she does not have to pve a negat~ve answer to d ~ e other quest~on. In

otllcr word\, thc two issue\ ,ile (to soirie extcrrt) ortllogo11'11

Page 178: Recanati, François. Truth-Conditional Pragmatics

- 37.3 - <~3P+ttrJiC).id" -~Ei%+t5k+ ~~T~ O k FREE EbRICWMENT (OA BOTH)?

fthe two orthogonal lgsues concerns defaults Are there default pragriratli values (e.g. default scalar tmplicatures) whose calculahon w part of the coliiputattonal system of language? Follomng suggestions by Ga~dar and I,emnson, I. andmall ar1d Chierchta provlde an a~irmat~ve answer to this quest~oil Sperl~er and Wilson, the~r followers and a few other researchers (e.g Geurts 1998) promde a negatlve answer. The second issue concerns pragniatics, arid more specifically the context-dnven, optional processes of nloduldbon Are such procesres pnm~ry l I.10 they operate on the tneanings of the parts h e h e the rnearimg ot the whole is ~alculated? Do they take place lorally so as to affect sernantlc composlhon? Advocates of Truth- Cortditiond prap&~trcs provide ui &rnlahve answer to thrs question.

Defenden of the standard, Gnceai ptcture g v e a negahve answer: prag- ~xutic processes are essenhdy global and post-propos~t~onal, they say. Un- less they are ltngu~r~cally triggered and mandatory (as saturation is), they do not <&ect truth-conditiortal content.

To say that the two Issues are orthogonal 1s to say that there are four pocsible positsons, depending on one'b ,inswer to these yueshons. Let me revlew ea~l i of these pontlons in turn

The first posihon, N N (for 'No-No'), accepts netther default pragmanc values nor pnmary pragrnatx proc eaes over arid beyond saturation. It sticks to a ciasucal conception ot-both serllallttc content (whicli does not Include a itck6n~hle layer) arid prapiattcs (which does not mterkre wtth serrlantsc cornposltlon) How, tf one takes such a position, car1 one account fbr enxbeiided implicaturesi One has to say that they do not really exist. Thuc. ~ccord~ng to King and Suliley, who dehnd the clxcrtcal view, embedded tmpl~caturer arc an rll~ls~on (&rig 2nd Stanlev 2005, ~ection V). Constder the followiiig exarrlple.

( 8 ) Eating some of the cake is better t k n ectixlg aU of i t

1 Iere a scalar mpkc ature seeins to ennch the left-hand-slde of t l~e 'better than' relahon. (It rs better to eat some-but-not-all of the cake than to eat dl of tt ) Tfl~s IS a tvpical casc of e~~ibedded ~niphcatme. Krng and Sunley thlnk such examples can be analyred rn terns merely of: saturation A staternelit of the form 'Better P than Q' is true, they say, lfand only If-the most suntlar worlds In whch the lek-lt'xtid-side (P) holds are preferable (XI some contextually deterrrlined sense) to die most sulvlar worlds in whxh the nght-hand-side (Q) holds. T o be evaluated, such a statement requlres the contextual prowlon of a

Page 179: Recanati, François. Truth-Conditional Pragmatics

speclftc slnulanty relahon between w ~ r l i k . ~ ~ Ili'hrs, they thu~k, n an rrrsLTrlce o-t saturahon. So they offer the tbllowmg alidysr5, cvhch disper~ses kvrttl ernbed

ded unpl~cat~rra. In (8) the word 'some' c~lrveys I ~ S hterd content (crt least some), not ~b 5trengthened nleztmg (snrne but rzot all), but focnrsmg 'sso~~rc' Xe~ds

the interpreter to choose a s l n u l ~ n ~ relatlon such that tire most ~ulular

P-worlds are worlds ~n wkic11 tzot all the cake 11s been eaten In this way the truth-cond~t~ons of the utterancc are ~Eected m d'some' had beer1 g v m the

upper-bounded readmg (even chouglr rt has not) They conc lrrde

By focussing the relevant word, orxc a&ects the choice of the sitirilarrq relation

between worlds that is relevant lor t i le tnrtll-corlditiorrs of dre 71,ette-f-than' construction in that context. So ttic trrtth-condidom of these: c-omtructions are

affected by scalar facts, but itrdrpendrntly of processes such as expliratrrrc o r

irrlplicature 'iiltrusion'. Nor does the sc;~ia-f inicln~~ation "etrrich' the serrrantic content. Rather, the truth-conditinrw oi' 'better-.ttian\sentrncrs are sensitive ro

the choice of a similarity relation I>etweeri worliis, a~zd focru aEects the choice of that relation. (King and Stardev 2oo-j: 152--3)

Kmg ,md Stanley deal c.vltlt the ~r*lphratures enibrddrd la1 the a r ~ t u i riierlb ok

conciiaonalb iri tile sanle way, by exploiting ti le ont test-dependence of h e

slndanty relahon In tcnns of wblcll ionifrtional\ ase starrct&rc#y ,kn,dy\cd "" It1 replv to K111g and Stanley, Ict rrle note, ku~st, that Ioc,ll streu drm not

seem to be necessary to get the L I c ~ I ~ c c J . eEe< t 111 French At l e ~ ~ t 1 car1 sa\

(9) Meux vaut manger un peu de g;iteau que le g5teau tout eniler

Even though the words 'un pet,' tfo not bwr focal stress, stdl a contra% 11s ~r,xdc. (by xntjans of the comparative ccrx.lstnlctlon itsell-) betweerr 'uit pcu de 3att>att9 (sonie of the cake) on the left-hand side and Vcgitcrlu totkt enttc+J (all trl: the cake) on the nght-hand s ~ d e . ~ ? Now sllcln a curltract xnakes sense only r4 we strengh-

en 'un peu' so to get the upper-bounded redd~ng. The not1011 of ' ~ n ~ ~ h n g sense' that comes into pl:~y here n pr<wIzatrc ~n the &Hest pos5lrMe senw (the

2') In this respect, the 'better--tlrari' ro~~strucnon is similar to conifitionds, wllicX1 dso s v c nw to embedded iniplicatwes.

Z1 G d a r is the tirst theorist to have attenipted to esplJin away the einbediied irnphcatnrcs ooi'cohen- conditionals by appealing to the contest-sensinve sernatraGs of cond~tlonal seiitences in the S d m k e r Lewis hnework . See Gazdar (1979: 70). At the very end rrftlle same book, howcver, Gu&r seenls to revert to the view that there are genuine embedded ~xllpiic:atures in cxan~ples like (7)-(8). " In my original paper (Kecanan 2no3), 1 had used ur ur~srresscd vers~on of e ~ ~ s r i p i e (8) to make rriy

point, but Geurts (zoog: 73n) ha? respotlded that focal stre% is obligatory in Englist~ 1x1 ewxnplcs like (8). To be on tbe safe side, I now use a French cxmlplc..

Page 180: Recanati, François. Truth-Conditional Pragmatics

i74 t ,( i!Vc;t,Li>lr Xv: kpi:kAUk, k--lbtPkK;A i-tff<Eb <>t< 1314Ll!~P, bWt< ~~X+IV~P,N 1' (( >k B C . 7 KH)?

"top dowxr' scnsr): die process of s.ire~lgthenirig calics place not because it is iingtintrcdliy iri,uid,itt:d, but in order to marlrtr serlse oi'rile speaker's cortxrnurG- c:itivcb ,icr. liiarg , i t i d Stxrliey hide t h i s Gcr by tising, olrly examples in which silrr\c iitr rrla! i k : i t u r t x ofthe lirzg~iistic sign;ti (vix. the use ofrtrcal $tress) drjves the

str-cxigtlii:rrr~~g pn>ci:is. X3irt this fi::ttt~re is no t es5c"riti:ii t(:, the erbt~l.u_ilple. King arrtZ Stciriltcy ;irg~re that, in this pnr~icul:ir t":lse, the strel~gthening

ci1bc.t c-ari bc :ic.hicvctl intlirei-ily by r~lsriipirlirtin!; tile sirililarity rehtion irrvoked by t/7c 'I>e~ttci"-tlx;inf c(-)t~stl~lctielli., ~ - 3 t l l t ~ t:hnu i ~ y clirectly enriching t h t x roie:"trrrnrg of "sc?rlre'. (;r:ir,tetl; b11t tk~i";s, o i $ ~ t . ~ ;ij?;;lirz, a fe;~t i i r~ o f the

i.x:irrlpic t l ~ ; i r is ,rc~ciiis~rrtaI ;irltI i r r r leva~~t to the isstit*. Wr will have es:lctly the sararc: sr~-cngt~~t~xiix~i$ e t k t u~iliciltt'vrr iclrrstruifiotr wc IUSC, CIS long as it invoivcs a c-trntrast ht-tween ' so~nt" ;;lnd '~ l l ' . -X'i1;11 wilt he SO, in particular, ivc'n jf tire njnilnrchoft irl qr~r:~tic)n tSoe r iot itzsjoke (1 sirniltlrily reliltion betwecn

iin~rldc. For cx:irlrplc wc. Irl;ry s;ty:

jotitr ,st<: w r i i c of- h ~ h cake hut. j i r l ~ stt! ail oi'his

&-ierc the scr~lnr i.~~riihxare~it of 'ssorrie' takes pi;~('e witltin the tirst conjunct, in

tire scrspt. ol'"but7, yi:r i t be espl:iincti ,icvay i r r terrlls of some process of' satrrrnriorl L E I , ~ I ~ ~ ~ ~ I ~ p e n d e ~ ~ t l y takes plac-c i t r xntcq>reting rilis cortstruc- r roo. ("There is :i proc-rss ofs:lturstioax at work iir the irrtcri,ret:ition of'hut', hiit 1 tad<. I r ttr bcs irrclcv~rit t o the 11llttter a t st:ikt:.)

7'0 c.traic-luck., the cveak~~css of-tlrc NN position is rll:it it offi.rs no ge l~crd account r>t'enituciitlrd in~plic:~~tnres. I t orrly gives LIS s;~t~~r~tiorl-bnsed at~alyses i i r r - pariic.rrlar i z w s - - nrllrlyses wlilt91 c -a r~x - rc r t i w g:cni.ralizcd becarrsc they ~.xpEoic ~~i.c.e.~Jcilrai k-;.dturcs trt't.hc ciocs i r i clrrestitrri.

- 1 % ~ I I V X C positit~li is the Y N ('Yes - - No7) VIC'W, w11it.b p ~ ) > i t ~ ctehult i~irpiic,rtrires birt sticks to a ~13ssi~;lrt ~011ceptii)ii o f ' l ) r :{ j - ; : :~~~;~ t i~~ :is ~)pcr;ltiilg on

rlre OIILJ)LII ol ' thc g;r*lXrtrIl;tr. 1 this ti) I x ri:i?ii*r-c:11i,i's position. ?'his tmlay cou~ici par:nt3o?ric:rE, ur1c.c tlrcrse wtlo, like C:hicr.i-hi~, posit defliult itnplicn- ttrrcs cxl~fic.irly rgccr r i - r t h 'C;riccali !toot.' , ~ a l c I I~oltF insrc;rti th;it 'pr;t,vrll;ltic conlpiaa~ion"sariti g;r-;Lrrlrrlar driverl ones arc ""irltersper-sCtl"' (Chierchia ~ood.: 40). Xhor rlrc pragrostic computntiurra that are said to interfere with sem;ultic colripositicln are not gCnuiae 1>1-;1g11iati(. yrocesses---the sort of p ~ q y u ~ t i c prcxL'ess that 'T(-:l' talks thol lit. Wfmt I->C;S1 thcc~r-ists hold is th;~t rherc ns, i n the gwvrruntrr, a ~mrecilanisli~ t b ~ t c-:~icrrl;~tt~s ciefaltlt iniplicat~rres.

"That nlcibanisrri operates locrtlly. Rut it is riot a gerl~iir~e 111:1g1:1ntic rnectril- nrsni: it Is coriresr- ~ric.lcpc:ntient ;irrd bcltrrrgs tc) the ii~igrristic system. 11s Etr

Page 181: Recanati, François. Truth-Conditional Pragmatics

E . i ~ t i t r k ) ~ ~ E l ! IIVIPI I L A I ut<E\ 171

as genuine prdgrri.ltlcs 1s coricemed, a 1 IGSI tl~eornt nl,iy well as\unie ,I

conservative conceptltm ofpragrnatlcs & operahrly on t t ~ c output ovgranr niar (where 'gr'ruirnar' now rnrludrs the 1)C;SI ~rsecfr~irntn~)

Accordrng t o Cliierchl,l, ~vhen a defdrllt inlpl~cattlre thtlt ~ I ; L < bee11 ,,rtto- vl~nt~cdly factored m m cdlculatlng the tmtfi-coridttroxz\ of tlie \erttence I\

felt cor~textually Inapproynare, some backtr.lcking take\ p l , ~ e anti tile 1m- phcature rs cancelled. 'Tli~s c'ln only fr'ippen glob~lly---aim the detiult tmth-conditions of the utterance Iiave beer1 t i p red out As Chlerchr'~ wntes, 'ccanellat~on ,rmcturit\ to a slrnple kmd of backtracking' ((:t~rerthia 2004 66) The speaker computes the dchnit rrneamng (truth-i-ordstto~~\) of the \entente, ~nc re~ l i e r~n the context with the \eritencc thus interpreted, and d that Lils, b'lcktr.~cks and get\ nil of some cicCxult ~rlipllcat~trc. st> 2s to

get a nrore acceptable lnterpretaholr ( h t dteniatrve pr oceclrire rc to c-hnnge the context through acconrrnoctdt~orl- see below )

O n the YN vww, tfte plrenorrlenori of embedded ~rnplrc~ture\ I \ ,kc

counted for entlrely In tenxi\ of defa~llt3 Only defkllt ullpl~t nturcs; a n be enibed~ied xr-id drlct locally; nortce rmplrcat~rrc~ ca~inot, h r the pragnl'ttlc meclia~usm wblch yreid\ them is cl,rsc;~cal, tllclt n, giot)~llst. Tills view 14

eilipmcdly testable It d l be p n n u hcle dlrproved, ifwc c ~ n find n c'ue in wtiich an mnlphcaturc whlc h rcst_llt\ front a frilly-fledged prdgn~nbc proces, LS

embedded. Tor ex'irnple, let us conslder caes in bvh~th \ (me tleOult ~mpilcature wluch h ~ s been ~uppre5sed becatlie tt occut\ in a downward critnil~ng cnvrronrilent is corltcxtually rem\tatcd ('froreri') 111 such c,ise\ rt cledr th,lt the 11npI1czture IS .illnerlt fro111 the o u t p ~ ~ t of gr'tmrn'ar (siltc r it Li~s

het.11 supprefsed) itr procrlce 111 the ac tu~ l InteTrztatron of rlle utter,uicc I\

entirely clue to the yragrnatic. process cvhlch override\ the def2tllt rernoval It" that pragmatic orocess c,u! onlv he g l o t ~ ~ l dntI po\t propofit~onal, ,n t12r YN view holds, the ~rliplic~trrre 111 yue\tlon cnrlnot be errrbeddetf Br~t tt i c not d~fflcult to firid c,lse\ 111 wh~cl-I a nonce ~~~rplscature ts embedded (:l~icrchla hriricelf gxves the followung es'~mple:

(lo) It wa\ 'a two courbe 111e'il Z311t everyone wl-10 shpped tile frnt ar tile Tec orrd cause enjoyed a nlure, fctr he wasn't too €till to appxtscrdtc' it

The diquacnorl ('skipped tile firct or the second course') occurc in a do\vnwarci erttx~ling ertvlronrlrertt here, 5 0 the scalar rmplrc~tt~re cvlilch rrormally gven nse to the exclusive retldrrig ot' 'cor' rs suppressed 1x1 tlre

dehult InterpreMttorl ctehbereii bv tl-te i o~riputatlon,d \ysteni of gran~nla,

Page 182: Recanati, François. Truth-Conditional Pragmatics

- --- 176 ( ONCI riCION' llI5FALII T IMP1 ICATIJlLhS OR FREE E - ~ I C Z - ( ~ W (OR BOTH)?

- --- - -

therefore, 'our' taker the inclusrve ~nterpretat~on."~ S d , the context IS such that only the exclus~ve tnterpretatlon makes sense. As Cluerchla says, 'we don't mean to sn~lude m o n g the most satisfied customers, people who shppecl both couner' (Cluercl~la 2004. so). 'The renwved lalpltcature a ttierdore ~onteutu~dly re~nstated T h ~ s is a genulne pragmabc process, one chat, according to the UN view, cartnot take place locally Yet, as Chiercha notice\, the scalar rxnphcature wft~ch rest~lts rn the exclusive reading of the itrsjunctiort 111 (ro) cannot be accounted for by the glob~l method 7'1-le rtxipllcarure we would get by the globrrl method would be sornethng llkc 'Not everyone who shpped the first and the second cotme enjoyed the nleal ~ ~ i o r e ' ! ' ~ ' / l h ~ ~ IT not at dl -,hat we want. Chlerchia concludes that 'c~ses Iike (10) reern to ~onstrtute t'urther evlclence agamst globahsl~~ We seen1 to bc in preselxce here of an et~ibedded x d a r 1ip11t ature' (Ch~erch~a 3004. j ~ ) Itlcofar ' ~ 3 the miphcature in question is not prodrtced by the compu~~tronal systerrl of pimxiar but by a cuuy-fledged prapnahc mecha- msm, tllw type ot example seeins to show that the YN mew cannot be nght.

Yet Chier~lh~a cticks to the clnssrcal vsew iuf the relatlorl betmeen gm-

iftar axtd (genuule) przgmatlcs, so Ile hns to deny that there 1s an enlbedded in~pl~cature 111 (lo), appear,mcer notwltilstandirxg To that effrct be uses the \zme stratep <as Klrxg and Stanlet. anti a r g ~ e ~ that we can account tor tile strengd~ening etlect ~t~directly, by tnanlpulatlng the dorn.un of the yuantil-i- er tllrough accontmoctatlon

I'he n~teq~erawrt~r of (LO) rryiures a doman of people who don't shp both courses.

Thi\ yields the 5ame elleft that we would obtain by rtot removtng the locally added nnplir dtrtre (i;ltlcrchia 2004. 60)

Chlerchia .ilso 'ippejls to &at strategy to handle the scal~r imphcatures that are emt7eddc.d in t l~e atltecederit of cond~honals (smce the antecederlts of con&- tioxids ,ire downward e n t a h g envzonments m whch scalar lnlplicatures are rznioved bv default) 'Thus he a~lalvscs Idemson's exanlple (11) e-mctlv as fhzdar or Kmg ar~d Stanley would

"J 'Thus ln a standard example like 'Every studen1 wlio wrote a squib or rnaclr a dassroorn preentauon got extra. cretiit', ttre d&ult interpretatioi~ of the disjunction is clearly inclusive.

24 'The global metltod corlsists in fint replacing the scalar term by the weakest of its stronger :dternallves, arrtl theti tregating the restilt. ?'he (only) stronger alternative to 'the first or the second course' is 'the fint ar~d the second coune', so the 'global' implicature of 'Every orte who skipped the first or the second course er!joyed the meal more' ougt~t to be 'Not everyone who skipped the first and the second course enjoyed the meal t~iore'.

Page 183: Recanati, François. Truth-Conditional Pragmatics

(11) IfJolvl has two cars, the thlrd orre parked oitn~de must be sorneboilv elscr"s

'Here too', Cherchia says.

we want to restrict our collsiderabon to re& of worlds fkon* wfiicli people wrth Inore than two cars we excluded 'TEle eflect of tlnr ac conuzlodat~o~, 1% the same as

the computdtion of an mlplrc,iture Dirt d we are ctght, tllr rneci~anrsni ~hrou_& which this happens i s very &&rent fro111 how oori~x~lly hiplicntctres cornc about. In (11) the implicatuie is not adtled in LocaUy. I t is ~cconx~rrlodated at seine point to avoid a near contradiction. (Chierr:hia 2004: 67)

The last two pos~tlons are NU atid Y Y NY 1s the vrew supportecl b\ relevance thconsts. Deirdre Widson ha, perh,rps heen the first Irugtllst ro noocr

the importance of embedded ~mplrtaturcss t;u,lmple\ hke (8) ((I), \vhrrli pardel the Cohen-coi~d~t~onali, arc: orrgu,tliy due tcr her (Wilsorz 1075 r , I )

InsoGr a< they result 6oxn a ~>ra~m~arrc proc ess, enxl>edded imphcatnlre\ sl.ro\~ that such procetses can take place locally ,rnd ,&ect tnitl~-corx&h~~>+---~ tlrrxnc central to relevance theorv From rl-ic verv bcgi~lxvi~ng, relev~nc c. theorb ha$ n~adt: roo111 f o ~ pnrnary pragxnanc jrrorcssec \uch as free enncl~rnent, m i tesms of wlc f i enibeddcd mrnphcatxlres C ~ K I he ' ~ i coun t~d fbr On the other hand relevance theonst$ have t~e.ccrr been Il~ypy -vi l t l ~ the rlotlon ofdeLtult p r~g r l~ ,~ t - ic value They take the alleged \ileiaultsY to be not v'ilrtef cdcubred I?) tlnc

hnguage faculty mdependcx~t 01 context, but p r w s a c valtle5 that ,uc. cleter-

mned (m the usual, context-se*1\ItIve way) 111 the mact: e~s1l.y dcr curl)le contercts that come to rnil~ct when ntr q>ec*fic ~oritext is otherwise pro\qiieci For exanple, one niight argue that the phr~se 'ccu the gmu' I-rsrs a deiarrlt readtng (in wh~ch 'cut' is understc>od m the spec~fic sense ol mm\ --~n ennched re~drng w1ucl-t can be contexardy overnden and w1ncl.i s p l ~ n p to

rmnd d e n the phrase I S produced 'nut of the b l ~ ~ e ' Kelev,irrce theornsts

would reply that, when the words "cut the grass'are uttered out of tile blue, a \tereotyplc al scrnano lr evoked wlrr~c-h 1s. U \ P ~ .I\ t nnteu-t (tle-r;llli c n r l t c ~ ~ , we

riught say) and with respect to whrch thc ~i iean~ng of "cut' ntay undergo free

enrichment, thereby yleldmg the alleged dchll t readmg. The rnccbmsln at

work in prodwing that rca&txg 1% exactly the sane as (arrd 1s no Irs context- sensitive than) the mechanim at work ul producirig '& context-specfic ren&ng

such as slt~e ~hegrnss trzto strip 1x1 otie ofScarle3 t'm~y tcenanos ."S Accorc111~ to

2' See the quotation &om Scarie 1x1 Cizater r . Section 4, p. 35.

Page 184: Recanati, François. Truth-Conditional Pragmatics

rchi~hvnric-e thec~rrsts, the sarne thirlg goes fix scahr irn.piicatures: the rneclranisrn

t113~ i s r e ~ ~ p ~ ~ i s i b l c ibr the aUegeSi.c ''dc&ult i~l~plicatiires' is mactly tlie sarrle as the r-ricth,iriwxr t l l ~ t is rcspnnsi13le 6)r context- spczitic:, norrce irnplic-atures.

Strrlic work i s rreetlcri orr tllc part of r e l e ~ ~ ~ ~ ~ c z tfle(>i-ists if they want to

cul~ctnr~n~te dlerr. cEarm. For then: is a sigxritic:rnt diiik-ence betweer1 the

iiei;iiiit re'ldrarg o t ' ' c - ~ ~ t the gr;~ss' , ~ r l J the deC~lilt reading of an utterance of the fisrrnr "If i-' or ( 2 , dten I t ' . i::!lierc!ria cites ysyc.l-rologica1 experime~~tr by P4ovc.i-l, n ( I / . (2003)~ cst;~tflishi~~g that tri'tlw two inferences tflat f'ofiow, the first one rs irvcrwi~clririslgiy at-ceptccl, wl~ile. the icconci one is rejected:

'P'hosc expcrirricrits sttern to confirm that, by dcj;l~ilt:, '(IT' is interpreted I I I C ~ U S ~ V E " ~ ~ in EIIC ~ritc~(:cderi t o/- 3 c~ftlditio~l:*l, but C I Y C L \ I S ~ vely in ;I simple sti3tcllicrit like tErc coxrcliision of (r213). This hc t c-aliriot easily be acco~lrrted fix i r r t?rrlii of s~txeiit)rl)i~ scrrr;rrios. I"ls C:hcrclrin writcs:

Wii~;i( i s i i i t e r ~ s i i n g here 1s that w e :Ire de,~lirig W I C ~ ;~hstr,tit syllogistic ii-arnes wit11 lrbttc*r \'arr:rbichs, .iviirre r l i ~ trnly 'rr;rl' warc8.: are or ' ~n t l $. . . tlzcjn. i- l e ~ t i c tile rc:levant c ikc t i ; l r1not bib I I I ~ J I I C L Y ~ CCI aliyti1irri.:like scriprs, &i.,rlii ica~c-d or world krlowledge oE ~ i i y kiriti. I t rritrsr be i iuv to the rneaiiing oi*ti,c: r r r l l y ' I C ; ~ . ~ C ~ I I L S thitt occur in the expcr~n~~c"nc:"i rnGirchri:d. (( :i~ie~-ciii;t 2004; 03)

I tunx to t l i c 3 U U L I C W , wh~il ' i IIA IICVCS betarn ruplic~tiv defended in the i i i ~ r ~ r t i i r e '" 1 ?;iAc YN, WY poslh two rne~I~~~n~\ l l r :~ , . onc \znwnt~c (the nGSl) .ind dlc. otll4:r pr.kgrli.lrrc, but the pragnt,itit lncc hai-rr\rlr ri, t,rket~ to be pnnixy r itltrer i ~ k , l r l sci o rn t l ' ~n / i t rrrtcr,Lzts wrdt the prore%\\ o i - s cn~~~r t t~c ( ornposrhon,

rnther th,~rr t q~cr,ilrrlg or) it? o~itptlt -1 i) xrlahc bcnre of t h~s p c x ~ t ~ o n , let us

30 \tcjrhi.ri ie\insocr, tlic 1t::iding aivocate of Jel iui t pra$<rna~lc vniiic5, i s .&so ,m ;dvocate ctF'rnttb-

(Coiid~t~ozi.d ilr*ig~~iii:it~ii (he 'a-septs frei: t.nrirhrrlrnt, pr,qgnntlt~ intnisrun. i.tl-.), so ire is (i pc>renunl buyer t o r d ie Y Y vrcw

Page 185: Recanati, François. Truth-Conditional Pragmatics

consrder the proceulrig nlodei outline~f m Eiecu~at~ (rc)c)5h) and Tee lrlow

CherchlG3's ~deas ~ o u l d be accormnod,ited w~th311 \UCR 3 fr,inlework In that p,y,er 1 drgued tkiat the three coirtestu~~i procews ot clr\.~nsbrgua

tlon, s'ztaratlon, and alodu1,rtion have the followrng c hdracten\tlc\ In com-

mon In all c,tces there .uc seveial ~~z r l~ i zcd~z t e i for the st,itus of corr~pos~ttt>n,d

vnlue (the compoc~eional value of an expre\s~on-token berrlg w k ~ t ~t c c>n- trlbutcs to the lnterpret'lnon of thc sentcnte-tttlieil wl~ere ~t occur\) In ~tr\'imbrguatio~~ the carldlcldtes ,ire the di\t~nct me,irlings of the ar~~bigtio~rs expression In \aturatlon the c'it.ldid,ltes &re the v'lnous th~ngc whlt lr ca11 be

ass~gned, tn contest, to the lingmrtic elerrlent m r~eecl of 4drtlrntiort; h r ex'nnplr, d~Eerent referents/antecedents can be ,~i\rgned to the pronoun 'l-te\m the hscourse 'Johr~ w'is arre\tec? hy a pof~cctn~~tl ye~terd~tv; I . L ~ trad jn\t

stolerr a wallet' 111 modulation the c,mJi~Lates &re (1) t ltr lexrcal value encoded by the word, and (11) pragnatlcallv denvect vxlsluer thdt are freely $:enerate"iy ndppiyr rrg vanous prngnlatrc tilnchctrls to the Ic\l i 21 v,ilue 113 ail

caws, the cdnctidates receive ;tcttv,tt~oil honl ~ n ~ i l t ~ p l c \ou~ce\ (rncludtng, but not restrlc ted to. Ilngu~st~c. sotlrccs), and the csndidatc that ~v i r l s (that whrcll '~ctually goes Into the IntcrpreQtlon 2nd ,aruirle\ the status of on^

porit~ond value) is thrrt whlcll gets the hlghe\t '~ctivat~oti when thc pro< e\s of activation rpreadlng .;tablh/es

The in~portant potnt, h r cxir ple\ent purposes, rc that the nctlvnborl ~ v h ~ h a candidate ttltxrlrdtel~ recelves conies fro111 severxl \oilrce\ One source of

,ictivnhon u 1111g~:111stl~ For example, it IS well-kirown that M. t~elt dn anib~giuouc, word 1.; uaeted, ~ t \ dntulct nleanlrqy,, ilc)\vever c ot~ttutu~dly irr,tppro~x-~,ltc~, are Llutornatically acttv,ited This \~rnply ITIC~II \ thdt P,XI t 0 5 the activdtlon cvhi~h a

gven nlednmg receive\ collies frorn thc tact t h ~ t the W O I ~ tlidt C I I C O ~ C F that rne,iruilg i1as ~ W I I uttered. A~tlv~ihori ASO CC)IIIC\ trorll other jotrrcc\, of coune, and rf 1x0 actrvatlon conles fi-01x1 other sources the nic~nrag 111 que\txoxl I\ deact~vatcd and loset, the comnpeution O r ~oncrder r~iodisl~t*on thc lcxlc d vA11e a autorxlaacAy achvdtetl, 'wd i t ir tliro~tgh the I e x ~ c ~ l vdtie tl-rdt the other carrd~dater zre acccs\ert and call ther~lselves get ~ctrvated 13ut wlxtch val~le actn,illy gets Into the 1nterprel;tnon and be~onles the t umporit~onal value depends upon the aitivatlon level reac hrd Av the vanou\ candldatc\ when a11 sources ofacttv~lt~or~ h ~ v e been taken mto account. 7 he l e x ~ c ~ l v,dtle will get 111to the mterpletahon only i f ~t recelves rrlotigh acttvatitxi from other

sources to stay at the top of tile acbvatlon rarlkng 111 tli~s 11g11t we t ,UI we tire 1 )<is1 <L'I cio~rig two tltrnt?" Frnt, ir jx-ovr~ie\

I . - f -1 "

Page 186: Recanati, François. Truth-Conditional Pragmatics

-- rBi? (;C%~Ciii5itnV BWA7&-i-WEATwRES O R - F F ~ ~ F W R I C R M N T (OR BOTH)?

t

- -1

he p l m n~eanmgs) Second, it is a @nguistic) source of achvahori for the ~dn&&tes----it contributes a cemm degree of achvation to therri dong the luies of the Cherclna algontlm: the strengthexled value of scdaJirs gets a h~gher degree of actlvahon unlesc 2 downward erltarhng operator u encountered, m which case the rmkzkrlig n reversed and the p h value gets a higher rankutg Or perhaps, we sbottld construe the strengthened readlags as gettmg a hlgher actlvatlon thm the plan re~dlng3, and the downwxd entahng operators as

deactivating the strengtheried meamgs of the scalars m the~r scope, so as to reverse the rrilttd ranhg."' Whatever the details, the grammar w~ll be a hriigu~stlc source of JcbVabon for both the plan rnearung and the strengthened rriemzng Hut thn 1s onlv one source ofactlva~on, as m the other cases. On the over211 plcture, L~~~imidtieal %id contextual hctorc jou~tly contribute acbva-

trot1 to the candidates, 111 5url1 a wav that a sc~lar term 111 the scope of a

ito~vn\n~,ird e r ~ t ~ ~ l ~ i l g coperator m y be ~sslgned the strengthened meanmg ~f the ~ontextud activatloxi of that nlearung IS higlier than the deactlv~tlon resultmg fiol?l default removal C>n this 'conmlon currency model', we don't have to construe the ov~rncl~ng of deLu1ts '1s an lnstaiice ot backtrackmg or an)ithmg of-the sort. 0 1 1 the YY view, defaults are overnden locjuy, just as referents and rrtcle-acal vdue\ ,Ire aqs~gned loc ally '"

'rltls follriula~iorr is not quite sat~sfactory beca~se it dues not take the 'Hip-flop' effect Into account. What downward errtadirlg operators nlust dv 1s reverse the rariking previously esttblished berweetl plazn mcmirig azid rbengthelled me;mlrig. Thls :rmounts to deactivating the strengthened meanirlgs o11lv In certxn cctses.

l 'hu cliapter was conginally published (under the same titlej as an a t i d e in Wzilu.<upkicul Pcrspctivr.s 17, 2003: z y y . - ~ j ~ . 1 am ~ a t e i i l l to Robv11 Carston, Gennaro Chierchia, lferlo'lt de Comulier, Ira Noveck, Ben~arnln Spector, &n Sperber, and Deirdre Wilson for discussloris, conmients; or biblio- papbi:al help. Sllrce the paper was published, a lot of work h a been done on tl~zs topic, in two dirc-ctioin. First, therc liave been attenlpts to defcnd the Gricean, globalist approach, by showing that (s~litably rtlodified or mricheti) rt can account for the data; see, ul panicular, Sauerland (zoo.+), Spector (2003; ~u t io ) , van lloolj and Schulz (2004), KusseU (zoo6), and Geurtr (zooc), 2010). Second, on the loc&t side, l)ariny Fox and his allies (including Ghierclua and Spector) have developed an inhentiat, . > y i ~ u t , ~ i ~ ...... . , dppu*~li .... wiiidi &LCW~~-L& h i si,d& k i p L ~ ~ t i . i ~ ? ~ by psit"ig A ':0\-~5< c ~ h ~ " ~ G ~ & y "ptiALui

to "only' (see Fox ~oog,zoo7; Cixierchia 206: Cherchia, Fox, and Spector forthcoming). From the TCP point of view, two things are worth noting in coriliection with these developments: (1) The staunchest defender of the neo-<;ricean, globalist approach, namely Bart Gems, acknowledges the existence of a class ofcac;es which can only be Iiantiled locally via modulation (which he refers to as 'recotistrid and characterizes as 'a local pragnatic process'). In the case of scalar terms, reconshud 'results in a special kind of meaning sk&, in that the lexical meaning of an expression is narrowed down; e.g.. . . the lexical n~eaning d t u a m i s restncted to "warn1 but not hot"' (Geurts 2009: 74; see also Geum 2010 38.sj. (2) The covert exhaustivity operator posited by Fox et al. looks strikingly similar LO the 'covert optionals' mentioned in Chapter 4 (see note 7, p. 139): so I m i not sure the 'syntactic' approach rs redly an rrlterr~ative to TCI'. See nly renlarks on covert optionals, pp. 139-41.

Page 187: Recanati, François. Truth-Conditional Pragmatics

6

Indexicality and Context-Shift

1 . 1 . Indexici~lity and setlzuntic uniieu-spec$cufior?

Indexlcals m the ctnct sense are expres\loris who\e senlantac value \y\tenrclt

~cally depends upon the context ol utter,mce, and whose llng~ustic naeanlng

somehow encodes this dependenr y upon the context of uttcrarrcc 7 hus I do not count as ~ndexical In the stnct ievtse those exprenionc vvllo4e

sernantlc value depend\ upon the r orrtext alerely because they are icrurnntj- cally under-5pe~lfied and &nct~on AS i-ree rr'~n,i>Ies to whlclr a v a l ~ ~ e I ~ I L L S ~ be

contextuallv asugned. Whenever an exprescxm n i irltfexxcal rrr the stnct

cense, ~ t s Ilngiistrc rnclmrrrig errct>rlc~i tclken-re.)ltrxfv~ nrlt whrclrr tcllr us

how, Eor each partitul~r token o-l the cxyreulon, we call detrnninc tlre content carr~ed bv that: token AS A filnctlori 01 the Lirrunrrtatlcec of utter-

ance.' Thus the melmnirqz, of "I' rs the nrle that token ofthai word refen to the producer ofthat tokt.11, the me,inlng oiktoday' i \ the n~ l e t l u t ,i token of that word refers to tile day on wlllcil the toke11 IS prodnceci, tile rrredrrung of 'we' 15 a rule that a token of tlrat word refel$ to A g o ~ l t ) that corxtarr~, the speaker, and yo on and so forth

Note that itit being iridexlcal En the stinct seriic doec rrot prevent an

expression from also berng rerndllhc,dly under-sl~ec~fied Thus '\t?rc9 I* lroth

I Such exprcnlons nlay themselves be ci~ristrued AS fiee variables, d o n e w-isi~es, but then a token- reflexive presupposiholl constraining the ~.onccxruai assignment of value to the var~able has to be associated wlth them. On this view wl~at driarrguisbes irrdexrclrls from tlie ocher expresssons corisnzred as free variables is the natnre of the prrsup~)osiiiori ,~s.cociateci with them. T l ~ e cspresslon counts as rndexicai (in the stnct sense) o11ly if the presupposxion ii token-reflexive. Of course tins rs a terznmo- iogcal snpulatiori regarding the use of 'i~~~ci~xicai'; but the rrilpotcant polnt is that we r~eeci to draw a distincuon between two types of indexicals-m-the-i>rcrad-sense: iridexicak in the stnct serrse (token- reflexives) and se~nantically under-specified rxpressiors.

Page 188: Recanati, François. Truth-Conditional Pragmatics

~ndc.\rc,rl (tire crlcodeci rule tc tokcrt-rrf1eur.s~) and ~ ~ ~ ~ t i e r spe~ified (the

"group' whaiti n\ the r-eferc-ut of 'we' IUXO~ IIXII(~I~C!V determuled by the

11: ,I I I I ,L~~L"T of tlcb,~ti. wtrlt h wurtls art- i~rciex~i a1 (tokerl-rcNes~ve) arid v\rhiilr are rrrereiy corrtext--sertsitive. Strn~ta pc%rrple, for example, iiold that thin-i\ 1x""non "1-~""louns like '~Ic ' ;111(1 ' S ~ I C' are i ~ i ( i ex i~ ; t i ; o t h e ~ . ~ treat tlzern as

free v;iii:rlilirs rrssoci;rteci with ;I rlvn tokcxi - rc fkx ivc ~~resugposition (to the ~I l i :~t that illeir. scri~;rntic value has to bi: :I rrralc or fkrriale irrctividud, as the

i d\e may ire) 1 will trot go lrxo this debate X-ieli, I or cierr~on~tr~tives too, i t11c1-c arc" two tiptioils, b u t I cvill assuzrlc. the i ~ r d e x i c d ,111aiysis. according to \vhltrkl 1.1rc ~ l ~ c ~ a n i n g of 3 delxi~>~i"rtrative like 'this' or 'this car' encodes a

t o k r ~ y X.I:~~C'XIVC" rule jscc S e c t i ~ t ~ 1.2).

I ..?. ' I 7ri: role O ~ L " S ~ C L J ~ C ' ~ ' . ~ itltct1liot1.s

W h c c i ~ r ~ : j r l cbxpresslt.rxr is indcxii-aI or rnc:r-cly c.orrtest--sensitive. it.s coritent dcpcbaicli uporr s i l i~ i le kdtiirr o f the i.r)ntext oC' rrtlcrarxce. Merely context-

serrsi~ivr cxprcssiorrs arc stlctl t1131 their c:orlicrlt ~11lih17'111y depends upon tlzv ipetikcr's iirtrtzlioll (or ;it lust , tire iritentir,xr \?rhich it is reasonable, in the

conrcxt, to as~.rii,c r o the spe;~ker). 'X'llus wlrcrr I use ;x gcrlitivt as in "Jolin's car-', 1 rtrikr to t i ic c:ir that I)c:~xs '1 ccrt~irl relation I\" tojolrn, whicll relation is detc-s~xilrrcd ira ctrrrtcxl; as a furtctioxr of- the spe~ker's intentions. In contrast. irldexrc;l"iqyressii~rls are stich t11;rt alrc>ir corrtcal iri (txdz ctse dr.ycrt~ls lipon cr ciirsi,oncrtc.d Ji~crt~tnr (!/ ilzr tatztc7.xt qf t,ittenzni.P. ' I 'hilt fkat.~lre which, following Nirr-rlrerg, I c-all the ' i t ~ t ~ e x ' , ~ ~ specified by ttrt. token-reflexive rille asso- ciated wit11 tire irttlcxical. For- the first person pronourls 'I' a ~ i d 'we' the incicx i s circa pcrsor-r protlrrcinp dtc ntterancc. I k r 'you' ir is the addressee.

For thc tc.r-isrs ~ x i c i teinporal adverbs like 'today" ."tornorrow', etc., it is the einre ciCuttcratu:e. 1x1 i.:ich case, the rcfercllcc oi.tilt ir~dexjc;il is determined :w J. firrictiorr uf'tire c o n t e x t ~ ~ d index."

' '1 hi3 is rit)t to l ~ t ~ t.~~rfi~scti with the use of'indrx' to rrkr to i i i c r iriunismxcr of evnluauor~ (see 52.2).

Q ~ i o r d x t i g t o Nunixrg ((rgyj), ~rrdcxicds 'ricodc tiirec types cor~nfonriaaon: 'tii.sctic' ~nfbrt~mtion rtbiative ~o tlir iniXi'h. 'dc:iinpiwer' ~nfonnatrocl reiauvi: to thc rciircrit, :u~ci m(iint1aaor1 prrtaining to the relauaii bet-iveen rncics .urid rtfcxent. 'Thus 'tottrorrow tnkes rtrt: time. of ilttcraisc ;ls tndex (de~ctic rrlibt-nuiiorr). i t t,Ar, &v ~-~tPrent (~Iescs~ptwe ~ t ~ i i x ni~itio~i), and ~t alsu t ~ ~ c o d e s tlte snfornrntioti that rile 1-efiwni i iuirieih~tc+v fbllow~ (lie &iy t-~)~~crlli~in~: t i i t * index

Page 189: Recanati, François. Truth-Conditional Pragmatics

'I wo specla1 cases are worth ctnc ussing . ~ t thls polrtt X n-st, wllcri , ~ n

expression 1s both indexlcd a i d \ealaiitlcJlly nr~der-sl3ccrtieti (as 'WL' 15) ~t.; corltcnt depend\ both 11po11 the des~giiated ~ndex und the speaker'\ Illten-

tloris. Thus, as we have seen, the \ernant~c value o f 'we' a a group ~ox,L~irl~ng tlze sprakerar~~ong rts riienlbers. Eiere the speaker 1s thc ~ x l d ~ ~ (so 'we' 11,is tht same index as '1') b ~ l t we Aso need the speaLcr'\ interlticttl$ to ftx the relevant group whlcli rs not f~rUy deterrruned, but merely c orrstwlrred, by t'ric. hngurs- t ~ c mcanlng of 'we' In contrast, the rneartlng of '1' or 'tctrnorro.iiv' Lull) detenlvlies the content ofthe word a? a fitnctlon ofthe Index o~ic e I tie rrrctex 17 contert-tuaIly identified, the refetent IS eo zpfo rdentified

Another mterestlng case rs tbxt of demonstr~tlves like 'rhls' or 'tins car' If we treat tlieur as rndemc'il expresuon.i, as I Jrn tlorrig here, 'ivl~at wlll be tl-lc

contestud index? l he lide ex liere rs s~ndardly c orlsideretl to br the tiemon-

rtrtztitni. that IS, the ent~ty to wlurh the speaker Ll\lng the derilor~s~r~~tlve c t ~ ~ w s the hearer's attentlcln by means of a polntlrtg gesture or by arry otlier tileans

7 here 15 ,in ongolng debate regardmg the cletenuins~t~on o i the ctetrlor~str~l- tun1 ~t\elf 15 it deterx~~ilied by the ywaker's Intentron\, or 15 ~t ttetcr~ntrled I n object~ve t j~ to r s SUCII AS w111ch entity of the relevai>t w r t tir\t lntctsects tht

str~ight h11e errrAnatrng horn the speaker's po~ntlng t~ttgrr?' I he '~~ltteltt~orr 15s' tale the index to Ite whit the 5peaker den~orntratcs. ttut IS the elktrtv suclr

that the spealer m,skes rn,inlfest to the hearer hw or hcr iritention to brntg t c

to the he'uer's atterttion by rile'ins of thc hezer's recogrutiort oc tlllr Intenrron (where 'thrs' rctieuively refers to the whole, c otttplex tntentlon, as m standard (;nceaxl ,ln,tlyses) O n tills mew, the speaker'$ cfer~~on\tratrve mterltlorl 13 the cn~cl,d aspect of the context on which the refelerlce 05 a tlernonsnative depends--1t 1s fconstltut~ve of) the 11idc.s

'flie situatlert n comyhcated by tlie fact that den~onstr,rtl.iic\ too are \tArnnn- tlc& under spetrfrcd 1n their own tnshroa. Accorctlr~g to Nunberg, the rebtton between the ~ndex and the referent I\ not Irr~gilrst~c,~~~y \peclfied in the crtx o€denronshatlve\ (cosltrary to w h ~ t h~ypens wltll pure r~~dex~cals lrbt-

'I' or '~OITIO~EOW'). It1 derxlonstmte my c'rr keys whlle snyrng ttr tfic valet "Tlils 1s parked out back', the demonstratum (~nctex) n the cet of keys, nccc>rctirtg to

Altcmanveiy, orie eqLsxt". the ~ndex wit11 the iierrivtistr;ited pla(e (111 such a way t l ~ d t 'this car' re&n to the car a? the iniiicated plare) (Lyons 1973; l\rt.;mati .zoq: 302). lx7 wliat r0llows I assxit~ic the

startdard position. See Kaplatr (rg89h: 582 QO), Keimrr (19")), Bxlr (~y(tz), Koher??, (rvy7), Slegel (LO(IL). MOIIXIL

(2008). Inter alra.

Page 190: Recanati, François. Truth-Conditional Pragmatics

Nunbcrg, brtt tile referent is the car to which the demonstrated keys belong. In other cases, perliaps, the referent wdl be the ixrdex, and the relatxon wiil be that of-rdentrcy So we need the speedier's rntentlons to ftu the relation because ~t 15

sen~~ail~cally under-specrfred, but we dso need the speaker's dernonstratrve lntelihon since ~t is corvhtutive of the rndex In the fmt place (at least d the lrltertnorltsts are r1g11t) T h s suggests that the speaker's rntentlorzs are playng two roles here they ,we both d ~ e feature of the context whch the rneamng of the denror-rstr,lhve (I e the ~\sociated token-reSfexve rule) mv~tes the hearer to corzil~ier rr i order to Jetenintle the vdue of the dernonstratrve as a fun~non of the ~rdex, ar~d they art: what rnakes ~t possrble to overconle the under- specificanon of the token -reAewve rule, whcb its& involves somethng 1lke a free relat-ron v,mable On thrs pr~ture the nlearung of a demonstrative n the rule hat the demonitruttve r&rs to the enhay thut bear5 relatlun K to what the speaker zs Jemanstratzng The \pealier7r's intennons are requured both to deternvne what the \pe;ikcr is demo~istratxng (rndex), arnd to detemne the relevant relatlon R such that the referent 1s the object, or an object, which bears R to the index. (Ifthe referent rs ' ~ n ' object bearmg K to the rndex rather t h i 'the' object beanng R to tbe mdex, therl the speaker's lntenhom wlll also be needed to srngle out tire rclevsnt object )

1 3 1&e context ill trtti.runcr

So far 1 have talked of Yeatures of the context', includmg arnorlg such features both the spesker? intentlorn and 1nore obje~bve ,wpeLts of the situat~o~i of utlermue wcfi ,is who \peaks, when, wilere, and so forth Now what 1s a 'conte~t'i For a srtuanort to courit as a context, an uttermce rliust take place m that sxtuabon T 'h t rne,rrb t l i~ t tilere must be an agent u 2nd a language LA such t h t a utterc an expression e of I., thereby performing, or attemptmg to perf:i,m~, what Austirt cdls ,I 'loc~rtlonasy act', ,in act wtuch requxes on the pax of tllc agent certaul behe& and mtentlons A context, In that \ense, IS not an ab\tratt obje~t- a ceyumrc of features-as Icaplaruan 'contexts' are. It I\ a c oncrrte stustion wrth a pafircular ind~vid~~ai 111 ~t endowed t v ~ t l ~ cornplex nlentd \ute\ (e.g beliefs a ~ d mterrtlons). 'Inlpn~per\ontextc m Kaplan's ccme--e g context5 in w h l ~ h the '*gent d o e not exst at the m e of the t ontext--arc obviously ruled out, but so are 'proper' Kaplanr~n contexts In which no utterance n made or no language emits or tlie agent IS unable to thurk or talk SaU, nothing prevents us from analytically extracnng a set of

Page 191: Recanati, François. Truth-Conditional Pragmatics

'features' that all contexa rnust possesc, or -t).orn hshng the featrue of cor1tex.t~ (the 'm&ces7) on wluch the refereli~e of mdexcal expre.ssions depend (wheth- er or not such features are necessary features of contexa)

The issue I am spedcalty mterested r n 111 th~s chpter 1s t h ~ t ofa)rztc.xt-shtfi

Slnce the semanhc value of an indexxcd depends upon the contee, rhrfilr~g tlre context results m s M m g the value ot the rndexcal Rut to what extent ri i t possible to slift the context? Does the Larguage itself-provrde Inexis fol dc~ng so! 15 not the contextgiven to us pre-s"r~'~fiacdlly' A\ we sh~ll see, the arrncxver

to these questions depends both upox1 the rclev,mt feat~tre of the context some are more sluftable t1ia.n oo~cers--,urd ,rko on ttze nature of die ihrh My m n top~c In the chapter mll be tile vmetles of-contest-shrft ,mcl the rlzr outy o f d r a m g ctlst~nctions between them Xrr the tiela two cli~pten vvc wll look a1

5orne quotabonal phenoniena \which xguablj~ rest on context -shdt

3. (Alleged) cotitext-slrzifis in corliples senterlces

Let us first consider a group of c a w that have beet-1 tfzc I'ocus oC rrlmrlh

attenhon in the hnguishc hteraturre the rLses In wlilch dn ~rlcfeucai <q-tpe,~rs to be hound Ach~aUv there are two sorts ctt caws that one ought: to

dicnnguish: tlie t-lelrn caces aid I he IJmec cdses " I--Ieini's ongnal exan~ple w , ~ s "C)nl\ I cl~d n q horrreworb', Irut ttolLo\v~rrg

RuUnla~m (2004) I will charrge the eumq3lc \o ~s to de,rl w t l r t l ~ c hrit

person pronoun 'I' rather d ~ r n the posl;e\slve "rrrv'.

{ra) Only I got a questlor1 I ~~riderstcnocl (rb) I got a questrori I understood; ar~il so &d John

Lke ( ~ b ) , (la) has two readnlg O n one of the bvo r e ~ d ~ r i g , the cecorid occunence of'17 works 1k.e 3. b ~ u r l d ban,tl?lc (311 tlie reahig 111 qurstiors, (re$ savs that 1 am the only x sut h tliat x gut J ipre\aon x tmderstood (t )rr rl-re other redmg, (ra) says that I m i the ordy .x s ~ ~ c l r dlat u got a cjuestron I undelrtocrd ) Srdarly, on tlie so-called 'sloppy' reddurg of (i b), the first ( onjunct says that

), The fonner were introduced by Irerrc I-leim i n class rrotcs in rite early rggoc (see refera-nces in

g Rullrnam (2004) or Schlenker (fofthcomind); the latter are discussed in I'xtecl: (1989). $ B

Page 192: Recanati, François. Truth-Conditional Pragmatics
Page 193: Recanati, François. Truth-Conditional Pragmatics

How, tlxen, car1 we account for the fxct tbxt in e~cl t ex'~1111~le the ~x~dexlcal

a bouridl In the case of(r3, one may argue that the <ecolld o( ctureilce of '1' 1s a dumnzy rndexlc'd, just as ' ~ t ' ~n '~t 's r'i~n~rrg' I\ a dt~n~xny subjcc t Jlir;t as A

ciurnrny subject has to be used In Engl~sh fix syrxt,itt~~ Irdsolls, the d~trirriry mdexcal 'I' lras to be wed fbr xnorpttologpcal le'isoryi (bet-,*use thexe llrttct be 'agreerlient' m gcrzder, nunher and person betweer1 xn cmaphortc pronourl

and xts antecedent) f ronl a semarrtlc pomt of mew, the secolld occtrl~rnc e of 'I' 1s he\c construed AS a bourld v:tn~ble, ur1fizortt t llt token jeflexivc presupposrtton t l r~t charac terues tile gerlutrie lrrciexxcnl 'II' (5ce Schlei~her

forthcon1ing and the refererrces dlereiu.) 'The case of (3) is different. Here we do have a gerruine installee of the

indexicnl 'we', but, I would argue, no coiltext sliifi is required to account fbr the fact that it is bound. The value of the indexical depenrls LIPOII a fe:rture oi

the cot~tcst of uttertmce (the index, n;mtely tlxe speakcr) but ilzt~l.fi;atnrc. stays (-onstunt crvefz thortgj~ the value cf the indewicrd xysti~~?fotim~/y S ~ I ! $ S i ~ 1 thix qt~antific(i- tiotzal exatr~plu: all the groups in the course of'v311ies of the I > O I I X I ~ indexical

corltaln the speaker (the spexker of the context) '15 r?i~er-nl>er. as well ;uJolirl and a third person ( J o ~ I ~ ' \ friend A) who vnnth\ frctt~l C ~ S C tt) CI\C. 50 we t ,m

n ~ ~ n t a r n that tliere is .L ~zngle ont text ot utter,rrtc c here (the ont text tn wh~ch ( 2 ) 1s uttered) suclt th.a the value ofthe indexical at 'my polnr 111 ~ t s roulr\c of vdues 1s a function of one partrcular fkatrtre of tliczl (ontrxt. kor tllere to be a

context-shztl, there would have to be at least two contexts, and the context- u d index (1.e. the feature of the context of utterance on whiclr the value of the indexlid depends) wotdd ltself have to be shlfted.

2 2. (~t~~d@~~r(~nce-sh! fL vs co~teat-shift Accordulg to 3avrd Levas, tt is often the case that 'the tr~ltti of a wntexice ux 2

context depends upon die tn~th of some relxted sentcrice kvhel~ sc>me featur e of the o n p a l context 1s shftfd' (Inwis rySo/rggli 27) '[bat ~5 so because om langugudge conwls se~ltenhd operaton wth the foglowlrrg j)ropcrty the truth of

m y c ~ r ~ p l e x selrtexlce Op conx~htlg of such hn opeleitot 0 apltheci to \oxntb serltence p a system~trcally related to the mtll of p when \ornc fgature of the o t ~ ~ p a l context (the contcxt m whch ttle conxplex sentence 013 LS uttered) fras been slxfied ul a ~ c o r h ~ c e \ylth the rules govenisng O I erstpord, spatlal, <md nmdd operators have tlre property 1r1 cluestton: the tl-utli of ;illy conlplex

sentence Op corrslstlng of a temporak spatlal, or modal operator 0 apphcd to

Page 194: Recanati, François. Truth-Conditional Pragmatics

come sentencep ir systerruucdy rel~ted to the truth ofp when the mle, place, or world of the o n p d context has been sMted. Thus 'There Iuve beer1 dogs' jcvkrci.~ L ewa pmtu as: ' ~ t hm been the case ht/ there are dogs') a true now If 'there arc; dog,\s true at sonre hnle before now; "onlewhere the sun is s h g ' is true here d'the sun n shlmg' is true somewhere; and so on and so Cod.

I thlrrk ~t woul(i br quite nlirleaduig to tdk of context-stz$ In connection

with the pbenonlenoa drctussed by Lewls. Saymg wl~y wdl hopeftlily shed sr>tnc 11ght on the relevant notion of context-shfi.

I,ewrs descnbei A ieature-s11ihlng process that takes place in the course of evaluat~rig the caniplex sentence Op: startxlg from the lrutxdl sltuatlon s ~n wh~ch the complex centence itself is being evaluated, and \vhich Lews ecjl.iatesto the context of utterance, we shift some feature of s and evaluate the embedded sentence p with respect to the distinct situation s' resulting ho~mi the shifi induced by 0. Now, as I said above, a genuine context-shift requires t t ~ ~ contexts-the input context (the context before the shift) and the output context (the context afier the shift). But the situation s' resulting from the shifi is not a context; nor does Lewis claim that it is one. The output sltuatrort need not possess any of the constltuhve properties of a LC)IICCY~. ~t need not contan a speaker, An utterance, nor a language.'

There 1s another reason for de~~yr r~g that the shift here is a genuir~e context ihlfi. Nctt only 15 the output slhxabon s' riot d contest: contrary to Lew~s, 1 tfo n c ~ t tllir~k the Input srtuation s--that ul whlcln the complex sentence Op itselfis ev,ilnated--lid, to or can be equated with the 'contest' of Op. 1 grant t t ~ t d corltext aIwav\ includes '1 trme, A place, arid A world feature (slmply

Oi'cource, it, rnay posses such features: nothing prevents the sentence p In the scope csfd~e operator ttortt descnbmg a sirttaat)rr in which sot~teone say.; sornetIGng. Thus in the sentence 'Son~edzy, someone wdl \tanJ up and say comrthmg' the operator 'soznteday' (or more explicitly: "someday it will be the ca.e that') cakes rts to a s211iled slnratiot~ s' which is located in the future, and which--if the sentence is atre--- h;~pperis at he a simanon ofurteralce: a situation in which sonleone says sorr~ething. Even in that sort of. case, however, the output situation cannot serve to fix the value of the indexxcds that occur in the sentence p which is to be evaluated wid1 respect to that sittuation. The vdues ofthe indexicals that occur in the en~bedded sentence p are fixed by the context m which the cornpiex sentence Op is uttered. Thus :i'thr complex rc~lterlce is 'S~o~nehy, soirieooe will stand up and say something about the clothes f am wertnng today', the values d t h e lndexicals 'I' and 'today' (as well as that of the present progressive) will uot be determmed by the features of the future situation in wbicli someone stan& up and says something. 'I' will not refer to the person ~ i ~ o speaks in that situation, and 'today' will not ref& to the day of diat ntuatiorr; nor wiU rlre present progressive refer to the time o f t h t situation (even though the cime ofthe ongind utuaoon has been shifted). Rather, '1' wiU refer to the speaker in the origlnal context of utterance (that uul which the con~plex sentence ic uttered), and 'today' and the present tense wiU have their values determined as a fiinction of tile time of drat same context.

Page 195: Recanati, François. Truth-Conditional Pragmatics

INIII?X ICAl.X'l'3' A N D G O N T E X l . ~ ~ S I I / F'T 189

because any uttermce 1s bound to tdke place somewhere, at a pmcular tune, m d in a pmcular world) X dso pant that a ten~poral, .;patxal, or snodxl operator 0 sh~fts the tune, place, or world of the 1n1b,d sl&tahorl 5 in vvhrch the complex sentence O2p 11s bcu~g evaluated, arrd thereby detemllxres the (distinct) situation s' 1n whicl~ the sentence p ~t operates on has to Ile

evaluated in the course of evalu3611g ( J p What I deny I\ that the lxlrrrd situation s in which the cornples sentence 1s evaluatecl has to be identified w t h the crtrltext m which that ( ornplex mitence 1s uttered 'tnd 1nteq3rctcd

Or, to put it another \.vnj I tleny t h ~ t the time, place, 2nd world w t h respect to wlvch the complex sc-ntcrrc e 1s evaluate& -the tltue, plat c ,

and world whsch the operator \vcterrratrca_llv sh~fts -1s the m l e , pXxe, and world of the context ot rttrermcr It need riot be there 15 a pnncri3lcJ d~fference between, say, the place oi ttre context of utterance, and dle place with respect to whicls the uttered ientclrlce 1s evalllateci

Take 'It is raning'. T o eval~i~re that sexitence we need A pldce, but that need not be the pldce of the context (I c the place where 'It 1% raulrx-ng' rs uttered). It nlav be any p lxe w h i ~ h the spedlier 11s cnnetitly c o r ~ s t d r r ~ ~ r ~ arid taUmg about Si~mlarly, the tune ;rnd would wrth respect to \n/h~ch we

evaluate a sentence need not bt the tune d i d w0r1d at the corite*r 111

which that sentence 1s uttered rlr~s rs undoubtc~dlv tile rnosl conrrrlun

and the sirnplest case, but t h ~ \ ~c onlb J particular c ~ s c neverthele\i, l i l o

place, tune, ~ n d world with respecno wllrch we t!~dlldtl) ,L \ezitence x e

features of tlze situutrorz tulked libour i r t utterrn'g thut serafetzcf, and thc altnatlun

talked about need not be itfentlc~l to the \ittratson of utterance The point I have just made w quite tentdtive f-or I ddrrut that the theorv 1.r

sunpler ifwe car1 Identify the r i i i t l ~ ~ situnholk and the context ofnttrratlce, a\

Lew~s does. k i d in the c,ise OK f e t ~ ~ p o r d locations (in contrast to . ~ p a t ~ d locations) it 15 tenTtxng to say t l ~ i t ,t ]>recent telrse setitence--wluch I take to

be counterpart of 'It is raning' rn the tenxpor;rl order-descnbes '1 \rt.txalon

at the brne of utterance. rather tIiax1 any ";lrnat~on semurg a5 refererice hnre

(though see Secbon 5.3 for tlre oirposrte mew). Be tfsat as xt may, even ~f we accept Lewis's polnt rega&np the mput \ituatlori arsd equate xt to the

context, I would resist tahng the shrfi he iiescr~bes as 2 context shift, because the output situation is not a torrtext but, as he h imel l puts rt, only the 'mdex' or (as Kapla~i pub i t) the ctntrrrzstunre of wuluut~orr tor the embedded sentence. For that reaon, 1 wt>nld classify the sort o fsh~f i I.,ew~\ talks about as a curcumtmce-sh~ft ratlier t11u a context-shift

Page 196: Recanati, François. Truth-Conditional Pragmatics

l ' l r t . corxic~xt~r~ij I'caiures wl-riclh rrxay be slrihed by :rxl operntor are few h trrir~rbtbn-, I,r\siis s a y : i>esiiles tl~t: tinle, 197t pincc, and the world of the ctrnrcsxt, Irc r?irc.xrric~ra I Iw 'stnndnrcis of precisiorr' ;is the fourth (and ultimate)

sf~iti,abic i;*,itrirc of tdlc c.orltcxt irl wlut hcs acirrrits is a "short list'. NOW the starrdacc$s .c>Spr.cc-isiori Jrit the only case 0x1 1,twis'i; list in which it seems that

sonietlrrng i:kc ;I gelriiir~e context sh;f r t O C C E ~ T S .

e { b t ' i , d\ititlie t11~11 we mrt. wirb a ( cintcXat 1x1 wIuc11 certain stmdards of precisio~i 3rc I n i iLrrc.c -say, Ioosib starrd:rriis. 111 s~ich a context 'Ilcxagond' is

irjterjrsi.teil Foost*iy :irit-i tr111y applies to France. (llr a context. 111 which stricter

st~rrd;rrcl'; 01. prccisac~ri ;ire in ibrce, 'Ilexagorr:lE' tlt>es not appLy to France). Now, hv i rs i r rg:~~~ c.xj>rt"ssio~l l ike 'strjctjy speakirrg7. one ~ u ~ ~ ~ s ~ c o n t c l x t irk zuhiclz ioosi. .c icindiirtr"? (~j"'prcti.sioti trrc irz .f;~rcc, inrv u corili':xt ~ i i t k r stricter st'znd~zrds. As a

resrrbl:, '""C;fic-el;: spe.;ihin~g, £ ' m r ~ i c i s riot hexngor-ial'7s tnle even under low \ta~lJ:~rtis of' prccrsiori if]' '"1:mnc:e is r i o t htxagonni" is bxe uncier stticter st:rilvi;irtis' (I cwis rciXo/rc)c,S: 27). This Iotrks like {.he C>L~ICP. c;ises f l l ~ ~ ~ t i o n e d

i ~ v B.CWIS, but therc is an iinportarit ~31Akrcriie: in this case, tlie stritted citti'ii ion to \\lhic-li the operator "sric tly spc*ikirlg' t:ikes us (tlle o u p u t i ; i t r l ; l ~ ~ t r ~ i s i ) (10~s ilii:ili+ :IS :i ctxitext. c r f trttera~lc-e. The cvords tllat ColIow ' ~ t ~ i c : t l > ~ ~pc~kirig, ' t.rviv ttttered in ;i cczntcst t lut differs bonl tlre original context iri tDi,it sonic fcxlrrrrc. of thta origin;rl cantext has been shifted: wt,

wrrteij wit11 ior i tcx t in wliirll loost" s~ancJar<is ol'prCcisiotl were in force, 'irlii wir1~1 1113 EPI ~ ' O I ~ I - C X ~ with stricter ctarrclarck. LVc may constnre the si~r~ri;irii.; of' pri-c-isiori :LS 3.~1 11sr7ec-t of" the / c ~ Y ~ ~ ~ z ~ c ~ < c ~ p o k e ~ n in the contcxt: i r ~ tho fir\(. I -OIILCX~ the I-rrrgir:lgc s p o h i ~ is loost., in tire stbc-c>rld context it is strict. O n t h i s c oi)si:rii;ri, t11c tvorcis tilar ;ire rrncrec?l :ricer 'strricr.ly speakirrgq are ir~terpretecl :ici-ordirig to tl?c ~-ules oi-the: strict l a n g ~ ~ ; t ~ c , ill force in the siliftttcd ctrnteui. 1'111s rs dxr:~logoils trt wl~;it lraytl>e~ls ill rile fi,ilo\uing exanlplc:

1 fvre woa stax wit11 .i corrre'ct in wltic-h S;,r~$?;l~sl~ is spoken, hut, after the

p l r x x "'as tbc L;rcrrch say' (:and t>cc;suse of it), tlie ldii~gti~tge feature of tlke cuiirext sblifi.; l ' r i ~ r r r I:nglish to Frenc-11."-Flus is J gerliiirle co~ltext-shift.

V/ si( i i i i ( . o ~ i i t > i i i i i t o l r i r t i~i. "in~~g~i~ge' .dirii>n!: t i ie feaiure~ of a context (aiongsidc the nnle, tire place, i l ie ipeakcr, r.ti.,). E cio io f i ~ i LWO rc:rlsorls: firs(, I - ~ C A C I S ~ (if m y co~x~~n~t~xlezlr to ' r~d~stic ' nq~tioil of ( . i > i i t ( ' ~ t ac iiecc~blxiiv irivoiwrrig soriiconc: who uses a i ing ia ia t ic cxprc*s!on tu say soz~ietirir~g ( sr .3 ) ; ><.I voci. iiecrrr\e i i.EiiiiL c j r ro t~~cioxin l dcvii:cs bavc tire power EO sllifi the CCIIIIZX~ dnd, ~n p ~ ~ i c u l d r , LO

s I t ~ ( t the l ~ r ~ g ~ ~ . t g c f i & ~ t t i w t b f - f l ) t r ontext (50.3). & ~ o - L . t h s 111 < : l ~ . t p ~ < - r \ ztn~l 8.

Page 197: Recanati, François. Truth-Conditional Pragmatics

I" j Slzt(tln2 'inliritionitl' fc~ztz~rt'i (I/ t iz~ ranieat

Acc orclrrrg to I etzils. tllerix Are \cmlltrhl r?ulrc for s h ~ f i l ~ ~ g the ~ s p e ( f of the context

or) whrcll the ( or~ter~t of words hke 'heu<y,on~l' depe~-id\. ri,rrrlely thC rtdind,xd\ of I?wciclon "mctly \pe&ing' 1s governed by tlre brirlantlc I L I I ~ that 1t sll~i't,

the st;lnd,~ril-of precrctori feature of the ~orxtext (]nit l ~ k c trsod,~l .~rtcl tenryord oper'rton .Ire gocerne~t hy the sem<tntrc rille tltat they sh~Ct the worki or irrw of evdu,itlon) U L I ~ till\ con~t"1ltlon'il14t expI;~l>at~on ( ~uz be t lt,tllcnged, ;uld J

p~agmatir ~ c c ount oI the context sfnfi ftrov~dcle ci~latr,rtl 7 be prag111'1tlc A' C O L I I X ~ 1 hkvr in I I U I I ~ r~ht%E ort ,t dl\bnctlol~ f b'fve ,dre,itly

dhlded to l~ctween 1x0 cortJ of contextu'd teattcre $onre fc~turc\ of thr conteut es\et~twlly dcpenct upon the ~pt,tkci's intentic~m. rtrus wllonz the

\pe,tber rc ,t~iJressiiig, or to ~vildt fie IS refernrrg wllcrl Irt~ A ctenlo~s\tratrvc, or how 111s word\ :ue to be t'lkelt (scnctly or loosely, say), ,111 ti115 to a lxge extent depends u l x ) ~ ~ tflic \pc&er's lrlteritloiir Wherl ~ 1 1 ,~\pe( t O/ tile ~orltext

clepertilc upon the spe~her's irlttbritlc)rr irx tll~c w.ty, 11. rc poss~ble t o 5hlh th,u feature of the co~itext ilnrplv by ~ n a b ~ n g ::one's intention to c t o \o sukl-trcrently exp l~c~ t 7;h1\, I O'um. ts ul~at happen5 w~tlt 'strrc tly \pe,rkirli;' By It\lrig i h , ~ t

exprecsion one ~rtdlcate\ one's mteritlon to \pe,lk strt~tly 111 the hrt o f drsc-o11rc;e

that follows bspressrrrg that nit en trot^ I r eriorr gh to ac tu,xlly dctet-n~inc how one's words c>uglrt to he t'ibcn, fix the followtrig ue,xsoil how the spedcr's wor& arc to be t d e n 15 an aspect ill ~vtiat tire cpc,rkzr niedn\, <~rrd thc \peaktrr's rncatnng works by gcttirtg rtseifrccog~i~et~, t11at is by lettir~g tlie hearer krtow

what the spcaker'i cctintl~ultlc,rtlce lntentrons arc CIie speaker'\ co~rtrnunic-l

tlve lrltelltlon5 11,ive the cii~tirrct~v~ property tlrnt tflcir rcc og~lltl(>rt 1edci~ to, 01

po~gbly co~~ctrtute\, thc~r i~rltllmcrit, ,u many autltor5 111 t t l c x (;nc c,in tradrt~on hme c;uggesteJ (sce o g U ~ c h anti 1 I,lntlslt 1979) So, tct rrl,xlic John m y ,~dckessee, I have slntply to ~ m k e ilea- t l r~ t it I\ Ixrn 1 mtetld to ~clcirrsr 'n-iuc

I may cay, 'Yo~I , Jo11~1 ', or I rlidy looh .it l l~rx~ cvlz~le \ye,xkxrig, or tne whatever 111t%,lxu are kwdable for nr'1k1ng my riltentiori \uNiclerrdy ri>~~uii'st 111 tlus wav f rnay e'zsrly siltti tlre addrccsee I-e<~ttire of the c orltext Irr-i,~gxne I start by addre,ilng a crowci A t some point In rr-ty chscot~r\e t n ~ i y slufi the 'tddres\ee feitt~lre ot the context by nlaklng cvplrcrt that, f~or i~ now on, 1 <rrn ad&cs,~ng

Jotu-t Tlu\ 1s st~zular to the h i t m stand,lrds of prectqlorl dexr~bcd by I ews , ,snd 111 both Lass what nr<ikrs it pos\iMe to shrfi the (ontext tr not it sctuantrc rule a\rrgrung 2 \pecittc cuntc'ct-shli-t jtotcntr,~l to wme e.'cpresnon, but 5nrrply

Page 198: Recanati, François. Truth-Conditional Pragmatics

- - - -

the fact that orie is rnahng one's mtentxon mmdest, in an area where the speaker's iitteiltroxB 'ire the cnrciLil f a c t ~ r . ~

W11en an aspect of the context does riot depend upon the speaker's rrttemion, hut tr &xed by ~orx~e objective fact, one s~n~p ly cannot shft thdt feature of tlre context by rnakmg exphcit one's lntentlon to do so. Who the

sj-teakcr 15 or when the utterance takes place 1s an objectwe fact independent ofthe speiLer79 rxitentlons. Such features of the context of utterance cannot be chlfted at w~ l l 7 hu\ the word 'I', In the i~ iou t l~ of S, will of riecessity refer

to 5 , who hdppens to be the speaker, even ifthe speaker mtends to refer to Napoltoii, anti nlakcs rizanifest his tntentlori to do so @ ~ r m s e 2nd Pern 1983 148) 1 ha 15 dlB%rent k o n ~ 3 demonstrat~ve hke 'that country' whose reierence ~rgu'lbly depends upon what the speaker intends to refer to. Here tlre relevant ,Icpecr ot. c orrtext-the speaker's reference-is up to the speaker a id can tte t~xed by him at will ('tfldt country, I mean France. . . '). The speaker car1 s t i ~ ? t i l < a t ~ what his words ' t l r~t country' refer to; but the speaker t anltot rtlpulate t h ~ t X I S the speaker, or that t is the time of utterance. T h s 1s stnrply not ln 111s power

3. Prctc~~ciirlg that the context IS difikrent fioin w11,xt ~t is

-3 1 . Irn~zgti?ary contexls

1 have just ~ard tlut 0i11y feature, of the context wluch dre 'up to the speaker' c m be \hrfietl t ty exyressmg one's ~rtterltlon to do so. The other features of the

ln thi, lrght we ~;rn reconsider an eext~~ple of knyage shift briefly &cussed by Stdnaker in 'Assert~on' (Stalnaker ~ g y y . Xu). Stalnaker's example is the fanirltar riddle: 'If you call a hone's tail a leg, trow many legs does a hone have?' There are two possible interpreeations, one of which i s more sattent .,. u k a t tits OGLC,. <.?ti m:tttdiezi~ i e~&ig ikic spekcL &S iiie iieate~ t" hi%hc A si~aaticii Li wiii~ii .I hone's tail is called a leg, arid asks how m y legs a h a m 11% in such a hypothetical situation. As Sdnaker says, calling a tail a leg does not change the nun~ber of legs a hone has, so the answer has to be '4'. This rcading is nonsalierrt precisely because the antecedent is irrelevant to the consequent, in such a way that the questlnn (thus in~erpreted) hardy rrlakes sense. O n the other reading, the speaker asks the hearer to make a tcnlporary assm~~ption, to the effect chat the language feature of the context is changed as far as the interpretation of 'leg' 1s concerned. The qoesuon that follows ('how many legs does a horse bve?) IS

meant to be mcerpreted in the shiited coritext, in wkrch horse tails count as legs. The correct answer is now '5'. What maks the context sbifi possible here is slrr~ply the fact tlut the speaker's Intention In askmng the hearer to assunre that horse's tails are called legs is taken to be an intention to sh& the context cur the rnterpretation of'thc qtrestion that foUows.

Page 199: Recanati, François. Truth-Conditional Pragmatics

context are p e n as a m t t e r of objective hct a id cannot be \hked l'l-ru\ the

speaker has no way, m spekuig, to 5sl1ft the reference of'l', or of 'today' This conc~uslon n~us t be qualified, however The objechve featirres of

the context of utterance are i~idced 'ggrvexr' and, to that extent, they cmnot

be shlfted. But what the speaker can do rs pvctend that thp ~ont("xt u dgercfit

fvom what t t is. If the pretence IS mutually nran~test, st w1l1 be part of what the speaker meaiis that the sentence rs uttered 111 a context chffi-rent ii-or11 t11c actual context c. In such a slhlation a Coiltext shlfi does occur- tl~ere ,ire twco

contexts, the actual context c In wh1~11 the sentence 15 prodnc ed, and drc pretend Loxitext L' in which the ut'irrance present\ stself as being producetl

Such a dual contest situahorz, b,i\ccS on prettSnce, 1s very cornmoil ul tXxr literarv realm. Thus a novelnt ca1 wnte

It's been titree years slnce we lrlt tlre Earth A couple of weens direr the Last Day, we lost track of the other syace,t~lp\ 1 \tdl don't know what

happened to my twln brother I lcnr).. If- be w &ve, he probably thrnbi; I died in the collision

Let's Imagine t h t thls 15 the tmt paragr~pli o i ,i aiovt.1 LYlll,&t rc the c ontext

for those sentence\? Clearly, tcvo sorts of 't coritext' dre relevant her(% rrrsr, there ii the actual coiltext ofutterancz tire novrlnt wr~res those sentence JT

the begltnlng of ticr novel. But that 15 not the "corltext' 111 the o rd~~ ia r \

seme, that is, what deterrru~re\ the refercxrr e of txtdexicr,il\. ' I 111" word 'I", arr

the thrd sentence, does not dei~otc the yrrsor~ x~I;io, ~n the act11~1 ccontcxt,

issues the sentence (the ~ ~ o v e l l s ~ ) , rdther, it purports to derrote a ( harsctcr trsi

the novel: the narrator, d ~ s t n ~ ~ w ~ s h r r i &oxxi tile ac t~ial author 1x1 ,r perfcc-tl\ good sense, then, the context for t11ose sentence5 1s not the act~ral curltext, but an imagnwjT context Iri that iirnagnary ont text, the speaker n orr btrard a spaceshp, he or she has a 1 x 1 1 1 brother called 'Ilenry', there lur been ,r colhuon, etc.'"

As Ducrot pornted out many y e m ago, wc need sosnethmg Bske the author/narrator drstinctiori to deal 'tv1t11 a nuniber cnf~ases of lanptdge rrw m which the actrtal utterer I S not the person wlloin the utterance ~tselfpresencs

as the speaker; and we need sr~ndar d~rt~iictloris for the ot l~cr feat~lres o f t l ~ c context (Ducrot 1980: 35-6, 1984 193 - 2 0 3 ) A s~sn$e rase oi that \ort involves a spokespenon readulg a speech ior someone else, says the Presrtlexlt

lo T h ~ s exmiple and the paragraph .bout ~t are btrmcved krom Kec,~nat~ (.&oooir r7x)

Page 200: Recanati, François. Truth-Conditional Pragmatics
Page 201: Recanati, François. Truth-Conditional Pragmatics

functioi~, that of referring t o the spe<iker But the spe,lkcr 1s not 13111. To be sure, ~t is 13111 who utters 'I'm ted L I ~ wrtlt all tlus' rrt the courw of repctrtlng John'\ speech. But Irt trtterlng tlur scrlterlcc Bdl Ir overdy pbylrz(i:ft)hn i ~ l u r f :

he tempornnly pretends drat he 1s John at the tlrrre of the ~cportetl cpeccl-i,

and utters the sentence 'I'm fed up wrth all tl-11s' c~cc.or<lnlgly ((:l,srE, anci C;enlg 1990). The prerence n con5htrttive of the Illcaning of the utt;erancct, which presents ~tself as uttered by Jolt11 at tlt'it tlrxlcL. ?'lilr rr enc>ugh to c onfcr

to John ttre status of " ~ p c d e r ' , at-id to the tlrne of t11c reported speccb dlc status of 'tl~tie of utterance*. In this fi-,irrrework we (-XI rrlairlrain that, ul tllt, quoted sentence, 'I' referers to the spe,lkcr, n:ilnely-lohn, , r~id the prerent tcilsc refers t o the tmie of utterdnce Serltence (3) thcrehre diy~ldyr ,t contest chlfi: the complex serlterlce 1s rnterpl etcd ag,lrn\t a contest 1x1 wlrich 8111, the

uttrrcr, IS the spc;rLer a ~ i d t* the dct~,tl LII*LC of'utrcr'~rlt c, 1)ut the cltrated niaten,il witfi~r~ the cctmylex cerltetlc.e 15 interpreted ~ v ~ t t t respect to a stlrfrctl context m whrch John I S tile speaker ~ n d tlie tirnc of ~i t terai ic~~ 15 wrne t1111e t \uch that t t*, ~larnely the tlnie of tllr reported speech acr.

Thhn approac-h to quotatson and dlrcct speech h ~ s beer] punued by a n ~ ~ r r ~ b c r of authors, to wtlorn I refer the ~litcre\teif rcadcr (see In p,nt~t ul;vr Clark and (;ern% 1990; <:lark royf~; l?\er,ln,ttr 2 0 0 1 4 rcpnxitcd below ds C'h'ipte~ 7) Before c lomg t i l l s sujn-cectlotl atld turnlrlg to .I related nwr, let

ine mention ,I type of exaniple of poterrt~~ll Interest to lmgutsts

(4) '1'111 g a n g to scc tlic dean', John sari, ~ r l t i lie drct

Note, first, the ust of tile parenrhetlc-a1 'johrl sad. Accogdiilg to Dcrloit de C'ornul~er (<:or~~ulicr 1978 85-91 p'~rct'"~tIletlc,rl ~l,tusc\ \IIC 11 a \ 'Jolirt r~ i ( i ' (or, with charaiteristl( tnvcrsion. ' ~ t rd Jollrt') be ~ ~ s e d only rf tllr q ~ o t ~ ~ t ~ o ~ ~ t f r e ; are appended to i c ail autonitino~is pie( c of nurruc1-y Tllur

m (4) the quotatlo11 'I'nr going to sec thc ~fe;rn' 1s ctflered 2% plctuie or replrca of the reported uttcr,mce, '1 p i~ tu re W L L O \ ~ re1,lt1011 to the CIi~p1ctcc.1 target 1s I nlph~itly '~scerted' by rhe vem f ~ c t c>f ortenrrvely prodtrc~tlg that

pleLe of rnn~uc-ry Slr~ce the deprchve reldtlo~l to the ta-get 15 part o f the nre'mlng of the quotatlcri~ piece of lnlniirry, tt cannot he cieu~ed w~tl-to~lt mconsisten~y. Thu5, ('orrr~~her pomt\ out, we cannot sav

*'I'm gomg to see the dean', John d ~ d not rdy/d~ti 110 s ~ y fohn.

'The other intereamg thing In tlln example tr tlw 'znd he clr~I'. 1111s 15

ell~ptical a r d sliows that rlre e11dc.d nl,itcrt~l ('we tile tlt.,~~,') I n e~ctr~ally

Page 202: Recanati, François. Truth-Conditional Pragmatics

- - 396 FRTmmING T T E m F T E - m W I S I3TWElC&NI FKUW V H R I Ir fS

availal)le for copymg. 'Tlzis avzulabhty lends some support to the idea that the cluoted rnatend 1s used, and not merely mennorred, in the quoted sentence. We ~otrld i~ardly say:

?"see the dean' rr a verb-phrase; and Jolin did.

3.3. lZe~orrled unessa'qes

A well-known example possibly anrexlable to treatment in terms of contest- sl11fi 1\

( 5 ) f ,UJI not here riou

recorclcd on Art an\wemlg rrla~lllne By 'now' the speaker refers to a time lntev tlran the niortlelrt of utterance, rramely the tlme when someone will listerl LC) the rerordett rriessage 1 his type of ex~mple raises a pnnla facie d~&culty firr the ~ r p w thdt 'now' 1% an iiidemcd, referring tu the tinie of utterutte (or, rliore accurately, to a tlme mtesval including the time of uttcrdncej 7 hat vtew 13 iiideed corttrovers~nl, as we s h d see (Sections 3.1-

5 z below), but the cXiiXi~ultv 1 arrz trow talli~ng about 1s not speclfi~ to 'now ' we c a 1 ac111evc exdctlv tile \ame egect with the u\e of "today'. ~f we Irnaglnc that the q7eakc.r wrote 'I am not here todav' the day before the day re-fcned to by 'today' and left the note on 111s office door

7 c) hatidle thl\ type ot exaniple we have at least three opnons (Recanati rgyjirj hrqt c~ptmn we car1 appeal to the notion of a chzft t r t point qf vzav or d ~ z ~ t i c 5k$ In ~cstalrt, niode5 of conunurucat~or~ (deferential comrnunlcatron rri so111e langunges, ~vritteil cornniunication in niany others) one may \hift the kie~c t ~ c centre'ffor indelv~cds &om tlie utterance point to the reception point. I hus itr 1,atrii correspondence one fzact to use the past tense to refer to the bme of wnting We find sotnethmg s i d a r in the use of the past tense to refer to the utterar-tt-e nrne m recorded messages ( ~ n Frtmch we say thmg

ndvyg &e Lvr"25J&'t) Note that suc]2 ilke 'yc f'2i;r"'EJ1: m wze tu &YO jtlC . . ', YV"*' "' ""' shfts riiay be possible o~dy m the spec~d mode of cornurncation at issue In the riom~ijl mode the lmgwage may well forbid that sort of shift (See Fdl~nore 1997: 122 on deicc*~ shifts in Tha~.) So the fvst opclon consists in acknourledgzry the possibhty of deicnc shlfts in certain modes of communi- cation, In a stlpulat~ve manner and wrthout otherwise modi+ing the se- rnantrcs of rndexlcjls

Page 203: Recanati, François. Truth-Conditional Pragmatics

1NUSEXIC A l I T Y A N D C O N T E S I ' - \ l i lFT 197

Second optlon: we can reft~rmuiate the token-reArsrve rule govenll~lg the use of indexcals so as to allow a certitur &erdorn In the ulterpretabori of the rule 1x1 cases of delayed comrnunlcahon I'or exaniple we can refijrrnu- late the rule for 'now' as the rille &at "now' refen to the time at whlcIII the message is delivered. In a norrnd sltuatlon of (face to face) conlmtunt atrorr the tlme of dehvery 1s the tune of utterance H~i t In a atuahon of delayed communicat~on, the notlon of 'dehvery' become\ vague ,md equlvocd Ls tl-ie message dehvered when kt's issued or when it's rece~ved? Both mter- pretations are legtlmate. It follow5 that 'IIOW' can reter e~ther to the ame of utterance or to the hme of recephon (2nd 'ttod&y;ly", either to the d,ry of utterance or the day of recephon), dependmg on the spe~ker's mtentlc>ns 'I

The thurd ophon menhoned ul Kecaxlatr ( L C ) ~ S I I ) i~ivolves the noQon o i

context-shhgpretence I used earher in thrs sechori. To handle chlfied rr.irs of temporal indexlcals m recorded nreTwge\ one iliay argue that the speaker pretends that the utterance takes place at the m l e at wixch it is received

T h s last ophon rams an ~mmedrnte problem when ~t conlec to example\ 11ke (s), for what the speaker of (5) says ought to he-fulse on tlre rnterrdcrl mterpretatlo~i. If, in the pretence, the spedker of (5) 1s rupposed to br \aying what he is saylng (or vvntmg) at the time at which tile hearer 1s recevrng the message, then the speaker ought: ro be rrr the place of-utterance at t h ~ t tinre,

contrary to w h ~ t he n saying. Iiowever, in '3 paper about tluosc r#\ue\,

Stefano Predelh argue\ that the dlirjcultv c,tn he avorcfed. He cl~im\ that

an utterance such as (5) n mterpreted with respect to a11 zrngroper contmt Irr

which the agent of the context is ntot located 111 the p l~ce of. the zorrce~r at

the time of the context in the world trf t11t context (Predelli t99X 410)

Schlenker also says that we need rr~~pr~g>er contexts to hruitile certaari ~ ~ 1 s t . ~ of contest-shift, :nvolmng the hictorrt a1 present (Cchlenker 3004 208 9,5ce

also Schlenker 2003: 7311) This mow seem\ to tlie unwel~onre, Irr mew of

the 'reahstlc' ~iotior~ of co~itext I ~(lvocdte b~11, I thlrrk the general approaclr to recorded utterances based on the rdea of co~ltext-sh~fhng pretcric e ought to be pursued, and its corlseqrrencer explored, for the sort of ~hdUow pretence l t posits a cornrrio11pl~~e ,md info~111~ lot o f - o ~ r talk 111 (;1tuC11 10115

of delayed conlrnun~cahon Thus, rri a rrcordecl message rrieaat to he hstened to only after his death, a b~lhorrt~alre may say-

" One problenr wth tlxr soluuorl rs diar n &>& not exrend easily to cases oi cier~nc sY111.t in 'deferential language', as m tire W ~ u a hnguqc of Memco nierruoned by Pdlmore (1097 122)

Page 204: Recanati, François. Truth-Conditional Pragmatics

I slrr hrillc,rxrr6~rre ; ~ r p ~ a b l y pret~t13\ to bc spr,rlnng f'rom the grave; kt IS m t h t c y ~ c d t o rLki,lt lilr,aS?;xnary ~orttext tlut the iitrer<irrt r n n ~ e a r ~ t to be inter- puet:tl t ) r I c b u l i r i t c r tEsc h~llotrvraig cexaamrplc, arr,+seJ in Recmatl (xyg(u) .

C)ric.c ;i;Gilrl, this is ;i sitt~:itiotl ofcjelaye~i C L ) I I I I I X I I I I ~ C ~ ~ ; ~ : the spe-lker is writing h i s n"sponsch tc! thc ivrrcr in &oxit ofllinr r~~uc-h bcl-lrre thr ;~cldressee can read the ri>sysa>r1sc in q'ii(~licili~. Still, tile two CCIX~~LII ICIC; arc it1 tItr present tense, alld ifwe cLrlic :;crio~ldy rhia idc~ tXc1t die present tenst: re ten to the tinze of~merancc, tlus sugg.:sts tPl,rt d>c iittcrxncr is nzcalrt to be ir~tc-~rct-ed ~vitls respect to an 11ii;ig4ri~uy colit(*xt i r i wllich the ac.t oCcc>n*rrrr~~nic;ition is ir~stantaneous ratlier r9i;trr dclayeal. X l i other words, elre writer speaks as if he w s talking to the

;~ckdn.ssce in 3 xrorri-ial, face--to-fikc cori~rnrrnica~ori situatiorl.'"

4. 'I wo types of colt text and two types of

cont~hxt-finxft111g prctellcc.

A P I O L ~ ~ tcann trl. I oxrtext-sh~ftmg pretentc 1% .rt work ~n examples like the hillcrwerig

( i r ) Johri r o liiii: Okdy, 1 atn stlipid 'IJICI j: Jor1't U I I I ~ C ~ S ~ ~ X L ~ the inattrr. Why do vou ask. irie for advice, then?

S Bcri~, yrcsr~rxa'~bli/, Jol-iu a et horxrg B~li's ~vor ik- - ~ C L ~ I J ~ C exaggerdt~ng

tlrcrri 111 rhc f a r \ z pkrr of the utcer~nc-c l " l ~ c : rctllencc '1 Irn s t ~ l p ~ d ~ n t i 1 tlo~a't ora~ticrit,rxrd clac rruttt.r' n 110t asserted hy John, rrctt eveti 111 A

corlce\\ivcb nl'uirlc-r What tllc scnterlcc c\prcuc\ 1s stmlettzlrrg t f i~ t John put\ i n r tlii ~ ~ i o i i t l ~ of11ls ,itlcire\see, BIL~ It rs Bdl wkto IS supposcci to th~nk or

u y t l ~ i t Joliri itlmpd ~ t zd does not t t~( jfc~-st ,~r~d the ~ ~ ~ d t t e r , and h ~ s SO s;l)iing

oi ciri~Arrrg pstrvrdc\ Johrl w t h a reams Ior ~rLlrlg the questlor1 ~ r i the

' * As i PUII" C M ~ ~n ?Ke~.ixlat.i (rr~yya). the 0131 :ird third iipiiolis C:UI be integmted wlthlri a ~nii led .tt i ~ C > I l l I ~

Page 205: Recanati, François. Truth-Conditional Pragmatics

1 N I > k X l C A L 1 1 l' A N I J ( O N I F X I -1,1111 1 IQ'j

secorrd part of tlie Litter'mcc: 'Why clo you .ak me for ,~dv~cc , therr'? In k

nutshell, the f i i ~ t part of the rtttermce drrylnys Bill', cr~\crtton, m ax] etlrort manner Irl saying what he says, John 1s glaylr~g Bill's p,rrt

We m y descnbe t h ~ t rort of case by \ayrng t h ~ t the sp~%aker eugagcr ui a fomi ofpreterrce and acsunlrs M~il'l's point ofvlew-puts I~i tr~self~r~ Bdl's ~iroes.

But note how d~Ecrent tlus cart ofpretenc e 1s 6-0111 t h ~ t dustrated by drrec t speech reports such as (3). In (6) the tndesical 'I7 reten to Johrr, not to the persoxi whose c ~ e w 1s he~rrg exprerwd or ~ c n l n ~ e d (13111) Brll hiriself, rfhe were

to express the vtew In cjuestion, would r ~ o t s.iy 'I stupid ad I dott't uncierrtrind the matrer'. but 'You are stupid arrd do no t uxidcnt,u~d the narter'

So Jo1u1 does riot pretend that Bill rs uticrtr:q the sentcrrc e, 111 thl\ ex.u~lple (as opposed to the previous onc.) It bllows tIr~ltI31111s not .the \peakcr7, that 1% the person who a presented ,w. uttemig t l ~ c sentence John, the actu,il utterer, 1s tlre

speaker in th~s exarnple ShU ,r ton11 of pretence I\ ~t cvorli, for Joirn, c y z ~

cpeaker, does not express hn own yolnt ot mew hue- that ofB111 A s~rrular phenonzelioi~ occurs In Irony. In Irony the speaker rdyi some-

tlilng cmthout actuallv acscnulg w h ~ t she sdys or 'rn,ikt's ~5 lf ~ C I s s ~ ' (<;nee 1989). The point o f viem exprcsscci by the trttcrdrtce t s not t l ~ t oC the speaker, but that of mot l~er (.lctudd or potential) agctrt wlmrn tlie cpe,.tker dtternpb to ndlcule by dl\playxng her vlew lrt a c ontcxt Irk wh1cl.r ~t n lrlicly to seeill dramat~cally lnappr oprizte (Sperber and Wtlwn tgX r . 308- r o ) I-or

example, just die1 having silo wn great IngenuI ty Irr solvmg tlre clrfiic ult prot)lcrn at halid, John can say tto NIU: 'Rernernber, I nrn stupld and I don't uxlderstand the matter'. 111 sayrng th~s John attempts to ncircr~le Bdl. he expressec hls vlew (the view BIU 1s s~lpyosed to have volcetl at Jn c"~rhcr poult irl the co~lver\atron) at a tlme w11en lt IS pretty dear that dlat view

ilrdrnat~cdy c ~ n f l ~ c t s with the facts By showrngr how inept the vlew is, glveii the clrctrrnst;.?rices, the speaker oken rxlaslages to convcy the opposite

of tfrxt mew--but thst IS not cfefirutlve of Irony ~t 15 nlerely a ltkely conwqvrence of the basic rnecharnsni, involv~ng 'pretence' (Clark and

Cerng 1984) or 'ecbo~c rnent~on' (Sperber and Wllsolt 1931)

I ct LIS ttse the label -displayed asuertion' fix the type of case L h,rve rllustrated- the cses In whtch the utterer does not pretend that surrrcone else Ir tlttenng the .iet~terrce but where, nevertheleu, a hrnt ofpletence ts at work because the speaker expreusec the vlew of sonleone else rattler t l~dr~ h ~ s ow11 In such cases, lswe have seer), the reference of 'I' does not rhifi. In (5),

2' refen to JC~~III, wh<) tittel) the belitetlce, rather th'ir~ to Bill, wl>oac mew rc

Page 206: Recanati, François. Truth-Conditional Pragmatics

-- -- - 200 TWO TYPES OF C 0 m - m UF CcYMm-I-SI;IiMIP.IG liltETEhC&

- - - -

being exprecsed. Tenses also take thezr nomnal, unsbifted values in such cases. 30 iln,igIle John is reporting h ~ s exchange with Bdl, several months later £-Le cm say:

(7) We kept d~sparapg my conrnbuaons I was stup~d, I cfid not understand the mt te r Fie wctdd be better off if I stopped help~ng . ---I wasn't discour-

aged, axid 1 manqed to solvc rhe problern

in (7) we kind scnteiices cvluch express Jolui's point of view ('he kept drsparzprlg my contnbutrom', 'I wasn't dacouraged, and I managed to solve the problern') 311d ~enteiice~ whch express Bill's point ofvrew ('I was stupld, I did iiot underctatrd the niatter', 'he would be better ofEd1 stopped helpmg'). Eve11 En the latter, however, the pronouns and the tenses take features of the dctildl co r~ t~x t of utterance as ~ncirces: the referent of 'I' (the speaker) is John, the referent of'ltle' 1s BiU, and tlle m e of the exchange between John and Bdl 1s prese~lted AS past, that is, as anterior to the &me of utterance of (7). All ths suggem &at the context of utterance does not shdt in this type of case

Still, we cannot str.iightforwardty conclude that no context-shft takes pl,*ce, for thcre x t = plenty of iridex~cals (e.g the demonsbahvcs, 'today' and 'tornorrow', 'lrere' and 'now') whose value 1s hkely to shlft when they occur 1x1 a dlspl~yed ,nscrt~on l'lin sort of chit2 i s very conirnon In 'free ind~rect speech' {A form of dnplayed assertion to be found In certdln hterary nma- t~ve,) Whdt fc)Llinvs 1s a made up example-

(8) T h e butler c m t e back wrtki the arlswer Tomorrow, Ladv f3. wctuld see nle wtch plez~nre, brtt she w~ too btrsv now

I et us x,\irrrtc (8) t\ uttered In A context c, wlth John as speaker and t* as

t n c ot utterance Ifie first person pronoun 'me' 111 (8) refers to John, and the past teuwtto a rtnic antenor to t*, as expected. But 'tomorrow' rekrs to the dav bllowing thc day of dle reported speech act, rather than to the day follo\v~ag the dGiy or1 wllich (8) IS uttered. Smirlarly, 'now' refers to the trme of the reported speech act, not to the hrne of ~ltterdnce

4 2 1 ~ ) ~ ~ i t t o t t u r y vs ~llo~~itiifnu~y context

Sctdeder say, rhat what shlfts 1x1 s~ic11 cases n not the context of utterance but the ont text i f f tlioyqlzt (Scl~lenker 2004) Even though John 1s the speaker (the agent of the conteAxt of utterance), dte thnker (the agent of the context of

Page 207: Recanati, François. Truth-Conditional Pragmatics

- - -

I N U b X I L A L I T Y A N D C O N T E X T - \ H I 1 1 201

thought) 1s B&, and a s~mrlar dntvlcaon r;ur be nude wr& respect to the otlrer katures of the context- the tune of tllought i s cbstt~rct h r n t l ~ e trme of utterance, erc In our example, hotvever, DrU rteed not really tllmk, or ]lave thought, the thought that 1s In cluecho~"to the effect that John a stt~prci and does not understand the ~natter) Wr car unagple that IS111 was mssrlc. ere, and perhaps overtly so, when he sad, or nriphec-l, that Jolio was stlrprd mil

lncornpetent SOU John can use (7) to de\cnbe tlie sltuatlori So tire not1011

of 'contest of thought' 15 not quite appropriate Of course, there. ,*re nrarily cases m wlvch a 3entence m fiee m&rec t speech picture\ a tkotght a ~ t rather than a speech act. Rut in aU caces the act 011 ifispky 1s an ace of nss~rt.ion or judgmtent or more broadly an eqression of ~ t t ~ c u d c ' ~ (whether sincere or msincere, pubhc or pnvate). The act of a\crtion is precisely w h t tlre cpe,rkcr does not perform when she cays that p iron~cAly, rather, she play\ scr~neone else's part and rntnzza an act ofas\ealon 3~ con~phslied by that person \he doer so not by pretendmg that that penoil rs speaking-rk that were the c&e, '1' would refer to that persoti irncier the pretence-but by herseKenctonmg the hncttotl of speaker and sdylng h a t p, whde G) riot tak1~19 ~ ~ T J > O X I C I ~ I ~ ~ Y ji)r

what ~c being sad , and (u) uiiphatlv a s ~ n b ~ ~ r g diat resporarl>~l~w LO sortlcorie else, namely the person whose a ~ t ot ,rssernon 1s bang rnrrrucbecl

I conclude that the ctistmt-trcjn we need 1s a d1.itinckiot-1 between the loct~ttonary context (the context c d uttzlarice, wllvre dgent 1\ John), and the zllo~utlontr~y p on text (the context o i ,wcertion, whore zgerxc 1s R111) I"

In the tradihonal fiarxiework of spee~h act theory, rl~ere 1s no rormr tor such a ~ I S ~ I I ~ C ~ ~ O I S . An illocntronary ,*ct rs taken ro be perlonncd 233

pertbnnmg a locu~lonary act (Aust~ri 1 ~ 7 ~ ) . in s ~ i c h a w a y that thcre r i a angle context, and two poss~b~l~tler E,ltExe~ the agerit of-tlir locutlonaw aci

(the spe&er) pertornis tlie ~llocutionan act (e g \errou\ly ,rcsem the propo- sltlon he 1s expressing) or he doe., not If he doe\, the speaker Ir the agent of

the assertton. the tune ofspeech I \ tile tlme of tlre mscrtiozi, arid co ora icr

there 1s a s~ngle context, and two acts (the locutrorrarv a ~ t and tlrr I ~ ~ O C L I

aonary act) pedornled 111 that co~itext- If the speaker doe., riot perlorrt~ the

docuaonary act, then, dga~n, there r \ 'I sti~gle context, but t h i s trrr~e there rr, a

smgle act performed in that context the 1ocutlo1r;rry act

" Among the rxpreviioIls of attimile' I ~n i lude rxprrssiox~\ ni affecti~c amhidm (a? in rrilLrn anorrs CUISSS, etc )

" I use 'dlocutron' In an extended sense t<r cover Il~oueht ACB 4s well speech a< ts

Page 208: Recanati, François. Truth-Conditional Pragmatics
Page 209: Recanati, François. Truth-Conditional Pragmatics

the 1n1l1re5\10n that the scene descrrhect 1s presexitly Ll.y)pcrrlrrg before therrl Acc orchng to Schlcrrker, that e13.2~ t 1s a~luevecl by shiftl~lg the I ontext of utterance, t h ~ t I\. by jpcaklng ac ~f the ,rct of \pcccI~ w,is srmult~rieous wrtll

thc sc erle deccribed.

(9) Fifty yea]-s ago to tllis day, oil J:u~rrary 22, 1944, the C;C-X.II~;~~IS : ~ t t ~ c k Vercors

\chlcnker ionclucles that the ln\tor~cal prewnt tc the ' ~ lnno r ratage' of free indlrec t ycec h I11 tree indirect \pee( h only the '( onteut of thought' shrtts, In such a ivay thdt the ternes- -whose v~iues iieyerrcf upoil tfie context of

utterance-keep t h r nonnal v d ~ l e ~ , 111 the 111~tt)ric~~l PresrIit, only (the tune featwe of) the 'context ot nt te~a~lce ' \111Its, s o the v h e s of-the tentrcs ,hrft, brrr the vd*ii~e\ of the otlier 1nctexlcal.i (~nclrrcf~rig t~rnt. dvcrblak such A\

'fifty years go') rcrrlzllrl un\l~ifte<i 5cltlct1lter suggc.\tr; tt.l,lt, ~n tirrcc t speeclt

leport, ~s 1x1 (3) above, repeated below, both cor~tcxts shift

(3) . . . and rlletl Joiin said, 'I'III fed r ~ p ~ v i t l l 211 rliis!' A I I ~ Ile wdke~1 avvay

I agree wttlt Sclrierrlier t l i ~ t the two context\ (the I o c utrctxs~ry 2nd the

11loi lttlor-iaiy corlteut. 111 rsly fr'xrr~eworL) car1 h i l tardepeilciexltly, a11d 'tlsc,

that they i,rn s h ~ f i ~urtultctneon~ly, 111 dtrert sprccll reports Cjther carec rn whscfi the tcvo i ontexts slmultclr~eouslv d l i f t CAII easily bc tr.ll,~grlcd. 1 or exalnple, one inn put tlre example o f d l ~ p l d y ~ ~ i &isscrtton I clncur\ed above 111

the histoncd present, as Ir t the fo1le)wrrig il~alilople ('1 he pasrage\ In the bwtoncd pre5rist Are 111 1tahc5 u41tl~ the d~\pLyed asscrtrorls ,rrc prmted ut

bold type 'The ~ ~ l e v a r i t passage., rnvolvlng cl~spt;ryed Assertron (2nd the hl\toncjl present are la., bold italics )

(lo) - W ~ A L h'rppenecf rhen;

- Noth-ltg 6c)r a tveek U ~ t t ortr d ~ t y , I m e t hun ~ t z it bar I/K, talk, atul hr kctys dzy7~"~~::u;q tny ~ontnbtltron5 I a tn stupid, I do nut uttderstrr~zd the matter. He wurtld be better df iJ I stoyyed helyirzg 'Iliis ifld ttot ~ b x o u t a g e ~rsc, howe~er

I he dlffererlcr between tlus c o l t of case (d~sphyed aswrtion w1tl.1 ln\toncal

present) and < a\e of d~rec t speec 11 repolt like (3) 1s that the pronoma cfo not shlfi thew vdues ' I ' itdl refers to tlrr litterer rn (xu), wl111c rrt (3) '1' refers

to the person -~vtlosc speech 1s reported. This if due to the Lrc t tltae 1x1 (xo) , ~5

1r1 (9) ( t l~e other es,rnlple t r l t11e I~rstctnc a1 prercnt), the only t c a t~~re of the context of utteranic that \h~t t \ ~r the tlnw fe'itt~rc Its co~ r t r~~r t to what

Page 210: Recanati, François. Truth-Conditional Pragmatics

- - - - -- - - - - - happens rrt direct speech reports, the speaker m the shified context is the

sarne person 35 the utterer In the achla.1 coritext.

5. Perspective point vs utterance point

j . 1. ,.IIle'qcd context-shytt, involving 'fiere' and 'nou~'

In his paper 'Utter~nce, Interpre~tion, and the Logc of1ndexicals7 (Predelh

1998), lo wluch lie thscu\ses recorded-message examples, Stefano Predelh

gives further example\ rn which 'now' arid 'here' are used m a way which

seerns to motate the token-reflexwe constraint

(11) fn the simtlrlrr ot 1829, Aloysia Lange, ntte Weber, vlsits Mary Novello in her hotel roorrl m Vlerina Aloysia, the once celebrated singer, nozo an old ladv of slxty-\even gt5es Marv the lnlpresslon of a broke11 wonian

larller~txng her kte (12) If an cntrre ne~ghbortiood could q& as an outdoor museum, the Mount

Wahrnggton d~cmct would probably clurge aclmtruon Here, just northwest of downtown, are several p~cture-hook expressions of desert culture ~v~t lxn a few block\

As Predelh points out,

'now' In rllc passage on A1oysl.i refers ne~ther to the hnie of encod~ng (when the book was wntten) nor lo the tlrrie of decodirlg (when I read it), and 'here' in [tlte other pasrage] p~ckr up rterther the author's honxe nor tlhe locahorl of the reader. (Predellr 1998 407)

The orily Lxv;til,ihle ~olueion In such cases seem\ to be to appeal to the notion

of context-stllfi Incfeed, ifwe look at the fmt exaniple, we notir e the use of

the 'hl~torical present' which, on Schlenker's analysis, involves a context

shift. If; through loc ntroriary preten~e, the tmie of the context of utterance

1s shified and made ctrr~tetriporary wrth tlie described scene- lf therefore

cve pretend that the scene described is presently happening--~t is natural to

use 'now' to refer to the tlme ofthe scene.l5 Sinnlarly, in the other example,

" Natural, but incoilristent with Scltienker's claim regarding 'now': according to him, 'now' is coritrolled by ttre context of thought. not by the context of utterance. (At the end of 54.2, however, I suggested drat the indexicak which shift in &ee indirect speech can be controUed eitlier by the locutio~iaq or by the illocutionary correext.)

Page 211: Recanati, François. Truth-Conditional Pragmatics

the speaker speaks as if he was m tile place hc IF &ing about (~>oss~bl) rn the company of the addressee) 'Che place rs spdttaUy present~fied by the use of 'here' just as, 111 the other example, the eveat IS temporally prescntrfrcd b\ the use of 'now' (and the present)

Yet it a important to reahze th,~t sl~xrrrlar use.\ of 'rrrow' cart occur cvcrr rr,

the absence of the hlrtoncal present 7'11u\ we rnay rc-plat e the prcc;eilt terrse In PredeUl's example by the past tei~se tl~roudiout wltltout any probler~r

(13) In the sumner oflS29, Moysia Lauge, nee Weber, vrcltrd Mary Nokello ui

her hotel room 111 V~enna Alovsra, the ttrlc-e celebrated ailgiX:c.r, now ao

old ladv of sLy-uty-\eveli gave Maw the Implc\\lon of a brokcn wcnnral

larneritlng her tat?

The whole scene is reported wrng the 17;nrt terne, wlllch \mggc\ts t h ~ t the

locutioriary context, or at least tlie t11~1e fe<~ture 01 that L O I ~ ~ ~ X L . I:, not \hiked in contrast to what happens m (1 1) StlU, low" r\ ~tsed E-iow are cvc to

eqla1r1 this f ~ c ti Orie may argue tltat ~ r i G-ee lr~dlrcct q3cecIh \ve find o((-iinei~ct)s of

'now' in cu~yunr tior1 with the past tense, refer-rrrxg to whdt W ~ S the prescttt at the time 1x1 the pa t when the repor tccl epzsodc. took place Scntr~~ce (8) x i

an example. But it would be a 113ritakc to cur~rlcier an ex,rrnple Irbc. ( r 3 ) '4% ,211

Instance of tree indirect spcecl~ Maw Novello need riot have rliought ot Aloys~a as '(now) an old lady of s~xty seven' t h ~ t qu.sl~fit atlor1 rnay iefle~c entirely the pan t ofvlew oftbe wntrr, r~tlrer tlun the yolnr of view oforis" of the characters (Mary Novello) Ti w c un~ier\tand the pasage rn tlrts wa) ,

as rzot an mstarlce of 6-ee ~n&rect speec tr, "uotv' ctlll rrtakec sense rl h ~ s usc 06. 'now' therefore rases a ~ ~ g ~ i ~ f i ~ d ~ l t challenge t'ur the view we hdvc drvri- opped so far: for in this sentence, o t ~ tlie suggested mte.1-pre~tt-ton there seems to be no locutlorlary context \I-rtft (as sfiown by the belrdv~our of dre

tenses), and there seerns to he no rllocutronary cor~test s11ift erttler (as showr~ by the hct that the p a n t of view rs that of the rimator dl along)

There is another problem with tliose alleged examples of corltcxt slldt involvirig "ere' arid 'rlovv' In liree ~r~direct speech, a coiltext-sluft rc

operative so that orie can, for exa~nple, trse 'tomorrow7 to refer to the day followurg the reported scene (rather than the day follow~xlg the report) One

can hxdly find such shrfted uses ol koriiorrow' oub~de very specific hteraq settmgs. But-according to Marccl VulUaume (p c .)-- *&egedly .iirifted ules of 'here' and 'now' such as those m { ~ r ) arid (12) can he found all over the n l ~ c p 2nd this dsvnunetrv tnilst bt. acc-oltrlted for

Page 212: Recanati, François. Truth-Conditional Pragmatics

2,iir"r iJE.:i<4iLi:c i'iLJk iitiiiV!"j-i' VS -;liiti'i:KANck i-c riN E

To :it-count for the asynrtnetly hrttf?oci: 'today' ,[rid 'ixow' it1 terrrls of tkiesr potf:nrial slrilL:lb13;tv ( 1 ~ +~:UC"III jlliCt.~bility), wC\ltiiIe h o r ~ o ~ i n n g tlic fact

i113il j i 3 ) ~ P ~ V ~ ~ V C , S r )~*~ thc r the historii-:11 ~ C S C I I L xior trri, indirect spt~ech, we

rar;iy i.trrisit-ier ,lir altcrr>:rtivi, ;tppru;ii.h to cllc ;rili.getily shifird tlscs of'llere' a r d 'xiciw'. (h i tfiis .ippro,ic.h, tlre rciev;uxr; uses ok"llerc7 anti 'rtow' <.lo not

rcaliy irivajlvc: GL i.c?tibext-sbifi, whecht~r locrrtio~xsa-y or illoctltic>nar)i. I I L

.Tlic rixasor-i why cve iixl cotilpriled to say th;rt the iorit.est sllifis i r i ( 7 3 ) is

i / l , r t "artrcv' rcal i~n.c t o a tinrr wlticir i s clearly rrot &kc (actu:d) ti111e of utterance. .l'lac .~lri-r.arati-vc~ ~pprcxich which I/ tllillk oh(iulil bc cxpIorclct consists in

g c b t r i a ~ g I-iid c d tl-ic x c S c ~ i titar 'r~ow' i s : r ~ i ~i~il~xit . ; t l , rviirrr-ing to the time of

ilitc:nlrici* (c>r to J t inw iriterval irrc-itrciin+; ;:ekii. i r r i i r of rattcr:~nce). We should rather th111L trf i t as ,I ~~crspecti\~al>referrirr~: t o a l i ,rrbitmr?; refcrerlce tirrze tl-e,itc:d ;is :i pa".i~xcti"vc point. Karnp ;irr i l i<i.yle atiopt precisely this np- F)I.O:ICI x ~o"r'i:~ircEs ' I I O ~ ~ , '. l"

-l',iLt. t hy ! i n i y t i c ) i Britiivtni~~z or ariy siirrii;rr i i > r p u s . C::heck all occur- rcbirics o f 'IPCV-C* aaiif 'rxcrw'. M m y will kbc pi--inr\a fLic-ic. c ouriteres~nlples to rhe ( i t i i i l a l e l ~ ~ i r 78icrr' ~ e l ~ r - s to tile place ofrrttcr,inc.cb ,tr.~rj %ow' to the tirrxe of urrer:iiicc. 117 pd~-t i~. i i I; ir , ,is ~)oi~lte;I o~1"I.lil /< CC:IXI;IT~ (2001 bj, -here' atten

~-cxili-r.; t o ,r ldla-i. I ditirlct hoin tile ~ J : I C C O~'UI~C'I.,IIICC coiltexts ill -,\-hictl a

~-o~lti.;ii;t IS dr:iwl~ bccween 1 : i r r t l strn-ri. crrhcs pl:lcc col~sidcreci as rriore

rcriroie i i ir oi ic r-i.,i\ox~ or :ri~trtlrer. SiarrilarEy fix '~niocv': it otter] refkrs to a

a r r r r c (pcbr-iod. tiriic:- ~rrtc"r-\~;d) irir.lui!ir~g C ~ C rrrorrrcrit of ' utterarice iri ctrnlit*xrs Ini ~vdi i i~h rhrr trrrie ~t reft.rs to st:irltis iir r.trrltr,rst to :tr~othrr, nzore

ciist;riit i i r r i c . ' l ' l ~ i c sxrggcsts tlltrt 'liurc' and '~r i i rvv ' havt. readirlg that is rc&tive

r.;ttl~c\r rBr;iii :ihsol~iti". O r 1 t h a t re;rtiing "l~ere' (contrasts with 'thcre', alld 'rloiv' ~vltlu ' tBlt"11~. Wi~t:ii col~sidct-i~~g t\\rt3 tinlcs, 01. ~ C V O I ~ I ; I c ~ s , if one is tllought of :is c'iorcr 11r:ur t 1 1 ~ trrlrcr, cvr <.art rt.it.r to tlac- cloccr clot" :is \now' or 'ht>rt7--at Sab3hi in ~C"X.L ' I I I I C J S ~ S . T ~ L I S XI the ;i\)ovr t.x;irrrplc L V ~ ; I I : is relev31it i s the cork!-r.isr: 'Aic~ysiai, t b ~ OI?CL, celcbnlrcd srliger, trout ~r r l t ) l ~ I lady of sixty- icvcri . . . ' In I-%rrdclii's Mount W;lsltir~g~<-irl ciistri-4c.i- t"xan~plc, n place is

I * i'hw :iij:iiz th.ti iiw con~stu'~int O I I ttie nAierr.nt.c u~\I(,w' i, i l ~ i r th'it i l i i cii.sc,crrhc.d cventudity rriust

&fi,rri'ip witit i/ic iiti<:i;iiiie timc, liut t l u r 11 rxiii"i overI.qi h i t 1 1 t~/) . i i they c.d l the "ienlpiird prrspecnve ~ > O I I I ~ ' (K.iiiip .ii~ii i<i.ytt. 11103 (o, 0 )

Page 213: Recanati, François. Truth-Conditional Pragmatics

recen-ed to a< 'here' prerurnably i r l contrakt to tire other p1,lces that wcrc prevlotrsly rtreritloriecl in the g u d e before that one c,+riie to the fhre,

On t h ~ s vletv, the con\tr.~lnt c>rt the reference of 'here' ;inti 'now' IS

cvedkcr tljar~ the mn<brd rildruical a~~aly.its suggest3 The p14icr or hrne

referred to silould be close ('pro.rmia17) b t ~ t 1x1 a ceme whlcli need riot be

,~hsolute but may dso be rel,ttlve, as 111 tllc forego~rlg e.r'~rnplec The pl'ice of utterar~ce arid the tlllle of utrer'lnce l m e t h ~ s feature pur ex~r l len~r; they have tt 'absolutely' Wllenever we engage ln lrleiital snnulatton arld pretend that sorne place or trine 1% the place or time of utter.lnce, the phce or trrrle in

yuestlon thereby acqulrcs the relevant feati~re of- promm,ihty, si)solutcly tlrlderrtood. Hut my place or trme, ~f ~t r \ c h e r t h ~ n sonre other pl,ic-e or tune In ~1 relevalit contra\t par , cxrl be refimect to by 'here' or 'now'. I t w ~ l l

he 'close' 111 a relative sense, a r ~ 3 titat 1s enctugli. For eurnple, we can

contr~st two tlnler, c q one rn~llri~n years d g ~ &rid twro hrrndred ur~llrorr yexs ago 111 tbdt context, if iert'lin a5pectunl comtrarrrrs ,ire s'~n.;fieif, ~t will be poss~ble to refer to the fbrnler tirlie as 'now':

(i4) 7'wo huridreti ttiiilion years ago, sl~cll ~r-r i l snch was (c-)r: ltatl beer>) tEle L.asc.

Now the situation w:rs corr~plrtely diffkrerit.

At tills point ~t rnav be ciouhted that the n e w analvns we 'lrnve at is a getlulrle dtern.ltrve to ttre mcfexit ai ~n'aly~ir Wh,it Joes 'cclme' rnedn, i fnut 'close to us', that I\ , claw to the cpe.iLer (or, perhaps, to the hearer)? Is Itor

the p~osrnr,rlrty constr,ilrit a toke11 rtflex~ve cori\trturni? 13ut we do t~o t have to analyse yros1111al1c)r ~n tiiiis niarrner C:onstder thc

well known corltla\t betwcen 'cor~lcx ' ant1 'go' Lt I \ oitert t'iben to be part of the tne'rriiilg of 'come' that die klrget place (tlie de\tmat~ctri) rrlu\t be wllere the spe~E-er I.;, hut thir is r ~ o t clrl~tc nght. 0 1 l e docs tiot want to tlede as

S.~CLI,II or tiev~ant .in utterance of '1 wrll conle to your party' lr i whrt h tht. Ile,-rer's point of vlew prcdom~n;itos (Tltir notton of corztext-shirt IS tnvl,l- h7ed ~ f w e apply ~t here ) Nor w11l ~t be ruffitierlt to s,ly t l~at "corne' dcr~otcc

a nlovernent toward tlir pl'jce where the \pcaker or the Izearer 1s. fijr therc are cases In whit-11 the point of vlew ~lnposed by the use of'corne' 1s that of an arbitr~ry charscter One st~ould rdtller say ' c orne' tleriotes a moveirtent

to ,I place w h ~ c h 1s proxtrndlti~ the sense of 'close to the perrpectlvt: ponlt'

The perspettrve polilt in cluestloll rleeil not coincide wrth what Austitl tr~ed to ~ ~ 1 1 tlle 'tltterance point', that 1s the locatlcln of ttle spcalicr/hearer W e

h'~ve such Aex~bil~ty in vnrtlt ig tilt. ptrspc.cti\e ploirlt 111 speech t l ~ t it 1.;

Page 214: Recanati, François. Truth-Conditional Pragmatics

208 PERSPYCTIWPOTW VSmmB TOImT

better to treat the perspectlve point as an mdependent parameter, dstlnct 6-om the parameten relevant to the evaluahon ofindexllcals

h n d Idem.; gves the followmg example of a sh f t in perspectlve point involving tlie verb "orno'

When the beggars c'me to town, the nch folk went to the shore. But soon the beggars came after them, so they went home.

The penpectivc point IS first Located rn town, but in the second sentence the perspecave pornt h a fbllo~ved the rich folk out of town It 15, therefore, thelr perspectlve diat 1s expressed in this two-sentence p'lssage. As 1,ems polt~ts out, 'in tlnrd-person rtarrative, whether fact or fiction, the chosen point ofrefereri~e nay have r-rottur-rg to do w ~ t h the cpeaker's or the hearer's location' (1,ewrs 1983 243). Slmtlarly, the out of 'pros~mdlty' that u ~nvolved 1x1 the analvus of the paxs 'here'/'there' arid 'now7/'then' (arid possibly khls'/'that'j may be undentood as closeness to the perspective pornt, rather than as ~Iosexless to the utteraxice pornt Closeness to the utterartcr polrrt 1% ot~lv a specla1 case, that in which the utterance point and the ~terspe.i-t~vc point coinclde

-$ j 7?ztp yer~prctzva~ 5 y \ten$

1 hdve just cic\c nbed J. possible position regardrrrg the senlantlcs of 'here and 'now'--one that explain5 tlicrr problen~atic uses (as Jn (11)-(14) above) withctut po51ting A context-sh~fi But the fact? turn out to be vastly more complex th'l,tn tile poslhon wggestr. Even though apparently shifted uses of 'here' cucll as (12) Are more cotnn~on thari slufted uses of 'today' or konrorrow', strll 'here' IS fir fiorz flexrble 1n terms of the values it can talie For exartiple, as P i h o r e points out (1997. 83), 'here' d3Eers fiom 'come" in that 'tiere' cannot earlly refer to the place where the hearer is (unless the speaker IS also there) Thus we cannot say 'I'll con~e here'. Tlus clearly COilflJLt: Wltll the FrpaPeL~Y& t:eat:l:eng of 'here7. PUqother rezsol> r , ~ t

constnie 'here' as a perspectwal is that one cannot say thirigs such as

(JS) When I c a m Inck to Ronie &r niy svrt-month stay m Tokyo, I greatly enjoyed the food one could fmd *herdthere

I

Sentence (15) w mfehwtous when 'here' refers to Rome even thoug11,as the use of 'corne' indrcates, Rome IS close to the perspective point, in the

Page 215: Recanati, François. Truth-Conditional Pragmatics

l N D E X I C A 1 17 X A N D ClON I E X T-SIHIFT 209

relative sense (1.c. closer than 'Tokyo) " MI thn ~ g g e s t s that we mav rreeci to treat 'here' as an indemcal atter all. A~xd &IS mneatls that exmlyles Irk-r (12) will have to be handled by saynrg that, m \uch uses, the speaker pretends tn

be located at the very place of whrch he is spedkrrtg, thereby gelleratlrxg a spatlal versxon of the present~fi~dtxo~~ e$32ct

Even 'come' cannot be treated d~ U I T I ; U ~ I ~ I ~ I O U S ~ ~ persprctlvd. In h*s ciet;uled study of the toplc, FlUrnore concludes tix~t 'come' I.ia two d~rtuict uses, whlcln should he given two mstvrct seI1Laltlc tre:~ttnerlts (313 otle use "conlc\rs ar mdexlcal \vhch denotes a c%splacernent to rtrc place where either the spr&er or the hearer 1s at either utterance trrrlc or rcii-rence tlnle.'" O n ths 11w thvre

can be an mtrasentennd shig of die contrxhrd p'naxleter (spe.rker/l~t-arcx, utterance ixme/reference &me) hc>rr3 one occiilreni e of '~(xlle ' tc) the I X ~ X ~

On another use, lllustratcd by the ZX.cvx\; uxunplc. "come' is a penpectwd O n that use no mtrasenteritl~l shlft of. perspective n possible. One canrlot s ; i~

*After John canie to RrU's tlouse, Bob and WIU together cmre ovrx to

Mary's house.

What about 'n(xv'? Insofir JS I c,tn tell, therc zc 110 dcclrrvr oi-)jcctrorr ccr

treatmg that adverb as a perspecttval r~tllcr t l ~ m an mdcxlcd (at least 011 soxr~u

u\es) Since, by dolng so, we cdn ~ L C C ) L L I I ~ 1 0 2 e~'istlples 11he ( X 3) OX (1 41, IOI whch no exp1,matxon IS] tennrs ot corrtcxt-shtii: n rradllv dvarl~hlr, I reriatl.cdt

conclude thdt 'now' 1s a perspc~t~val, or A t led\t, rt h,u perspectlvnl uses

Be that a:, ~t Iriay, illy p o ~ ~ i t 1% lot that t h l ~ or that expre~\lon s l l~uld hc treated as a perspect~val, but ratl~er tlla~ the qtdotzt~n may antc, lor &xi)

expresslon that 1s stmdarclly Treated as an indexscd, wliett~er it n not, 111

fact, a pe~spectival My suggestion 1s t h ~ t there is, m the lang~~ige, perspectival system, by vsrtue of M~XIIGI~, when a scene 1s dc~cnbcct. r t IS

described &om a certaln pollit of vlr\v----a point of v ~ e w wh.1~11 need not be that of anyone in the atuatlon of utterdace 7 1 ~ perrpect~vdl system rs

d~shnct from the index id swtexn. bv vrrtrle oi wblch the reference of certam expressions is a ti~ncnon of tearules of the situation of utteralce Tredtlng an expression as a perryechvd rdtber than irldexlcal has ctrn-

sequences for the alleged inrtances of co~rtext-shlfi If 'ul expresslon e a

" This example is due to Pl&ppe Schienkcr @.c.) " "This sort of pretence was pewaswe in n i~re~eenth- -cer~ t~ i~ t~ovrls. See Vuillaiitne (xyr)~) for nn rn-

depth study. " By 'reference tinie' FrUmore means tile time of the ciesrribed eventuality.

Page 216: Recanati, François. Truth-Conditional Pragmatics

per\ptSi trv,rI r'itl-rer ~ h , i ~ t ,iri tr~tlcxic~d, tllvrr we cio n o t need to post a ( oriieut s f r r i ~ w < , I L L O I * I I ~ fix \ I \ c ~ L ~ I ~ \IV!IIC~ the relev,tnt perspec tlve pomt 1s

1

ijistirict fits~ir the ilttrr;rncc point.?') la) i l l i s i'r:rrirc.wtrrk, tlic cjtrestiori ;triscs \zrhetllcr rhr terrses tl~ernselves are

intii~xic :ii rrr- rrlercly ~rerspzctiv:rl. Schlerlkcr's nn,~lysiii ofthe historicd present E pwuq~-"p) i c~r l id t rZic prclscnt r \ xntlexrt dl m c l rcxfers to the time of the contest. E Tl7c 'pk-t"wnti/~< ~CXOXI ' effec t rc cxyl.iinrd a\ toliovv~: The speaker preterltls

tiiar c i r c x rrt1c.r r x l c c. act takes ~ I . L C C '11 thy \~anie t ~ ~ i ( i %LCI tlie descnheci event, he cbicrcttrrcb prcxeer i t i s i.ir~J gtvej t i le hcarcr ttrv feeling) to be wtxleu yo the

I i

event ,is ~t trriibitjs. 'rrentir~g the pr-i:scnt tcrrsc riot as an indexical properly speaking bur: :is .I i perspec-tid' worlld i~ lake ;in :Ijtert>ntive an;llysis of the

hibiorii-,3l pre'ic~it ~~ossiI71r.~~ CC)tl tll:-i(: analysis, we deny tirat the preser~t is ttrlcxi rci-lrsivc :ii*t-l rcfi21-s to the cin-ic oi-'u~enince, anti krold instead that it rei lm t r ~ ,irihitr,lr-y 'refbrer~ce tirrrc' scrvirrg as perspective point. That

rc~~~:rcxii-i. tirrlc typii.,iily is but riei.ti 1101 hc rhc tiar-re of ~itterartce. (Similarly, the J11t~1re will rc.'li"r t o ,1 tiine pos*iCt-t(ir t t r tlic" rcafercnce Ti~ne, and the past to ;i

tinit: ,uilcrros c<> tlse rrtkrertce rline.) Xr i s rri:ikes i t ~-rosslble to ilse the present icr e:llk ,iboi:t a past rvent, ifthe refkrelrct: ~ ~ P I D C is set to t l ~ r time ofthe event. Tkrc ~ ~ ~ w n " r i l - i c ~ r ~ i m cf-fCcr i s no\v cxp1ainr:tl by thcs hc t that the evtnt is vit.w~\:~tI fi0111 ,i I C I I P I P O C ; ~ ~ v;~rlt;igg~ poilit whir11 1s that of the event itself.

'" it i i i , iy /x. ~1i .u i h z j i~isp~i:t~v~:k iy~ i i . l i~ e ' r ~ ) l v ~ i I ii.<iii~ ttw iiidt,xicaf syste~n. IJriitorytxc~Ily illc pera,pc"~vc ~x~n""tu1~"itElr: i~ttt~rdr~ce p01111 c O ~ C I C I C I C . I )C'VI:*LIOLIS ( i ~ r l i the prototypi, t h . ~ IS, cases In wiuctr i hcv t i c , iioi a o~iit.iiii., i i i ~ y 1 ~ . expl~ui~erl iirni.hriinli .Aiy irr t t ~ t i i s ofcontext-shift. One can project or~eself r i i t i ! ,uiriroirr ci\i.'i \kii)c:,. I L ~ I I C U aiiopt c h i p ~ n p e c t ~ ~ e O ~ S O I I I ~ (it.her 'tgeni. But wbdt may have started ss ~ i l ' ~ v t e i ~ t c d ' i t s t (>I ' ( I I C IIII /~.XIC.LL wstcrn has ~r>:nai,iy \~ibiiizcd AS ;ui ii~tiepeildent jyqreni: the [~efipt.""n.ii 4v~ttciir. As a reiial:, t l lc~c is riottrirrg rpi.ci~1 (ir devi,trit &bout, 2.g., the use of 'now' to 1c.fi.s t i) L t i i i ~ r iri r l l ~ just, no rwe"do appeal LO pI;i+;[nl*tIc Y I I C C ~ I ~ I ~ I S I I ~ S S I I C ~ a contt:st--shiking to rlccottt~t liir s w c i i us<,\ 111 i orltrasi. ilrerc is atimrc.tcii\iiy u~rneittrng sprciai al>out irhirrg ' r t r i i a~~ or .tornorrow' to refer to a dav rn tire F)ASI. 1:

" i n i<ct , l~ i , ; r i i jxou>a), i qkvtt ht:d suclr An 'dterrrat~vr~ an:~iysis, tizseii or! the idea th~u the present tense rs ts:~i~por,"lir iirhil\t, i i ' "rnt r i i~ctt 1 iii~couer~:i? tliat hiit. J I I ~ ~ S I S h.id ~ ~ C C I I ~niicipzted by HCZI~Z~C, W ~ I O b fre,its the p~e\iwt :i\, 118 et1Y.t t , p~mpe( , r~ \~d (t3chai.~r6e t y q . +!-, 8).

Page 217: Recanati, François. Truth-Conditional Pragmatics

I N I ) k X I ( A1 1 I V AYL) ( O N T L X I ' i l l11 1 311

alla-tlre adctrerser leatu~e ot tlie context, the l ~ n p i , ~ e f%.ititilre ofthe ctlntext

(~nchtdnl~,: the stctndrtrd\ of prcc~\lott), or the tienroustrat<~ Other feature\ of the context car, he \h~ftctl through pretence f.ollo\nnng .r ntlrxtl7cr of;i~itllors;, I h ~ v e d~\t~ng~~isl-iecl bvo wpes of c ctr~text 5lt1tttng p retent c 7 Ire fir\t type of

context-{hifbng pretence illriatr'itecf by dlrecr speech rtLports, rercrrdcclt utterance5 (on orie ~rialysi~), tile hl\torrcal prescltt (ag~111, or] cxle an,rlysz\), 2nd the prescxl~ijiir~g uses ot 'here' whit 11 are the spaall counterpart (>fthr l~rstt>ncaI plesent 'I he secorid tvpe ot cor~text-\h~iilng prctcxlcc 1s rlht~tratcd hv v'inous iorts of (ilsplayeti 'Isceraon (nonq~lotational echots, Irorly, frec rrldrrrct speeclr) md, .ig~rn, b y direct (peech report< rrtsofar 2% they involvt. the two

types of ski& c r r ~ ~ u l t a n e o ~ y The f?fi,urth type of case I \ tllat of eylres\lc,n\ w h c h ax not redly ~ndexcnl, hut penpec tlvid, a11d W ~ O S P shifiy hehaviow

rases 110 pxticuliir problcrn. In rhl5 ategory I h'tve plac-ed rhe ,dvert> 'now', ,ind the verb "conic' 111 o11e of its u\c\, 1 I~avc aL\o I I ~ ~ I I ~ I C ) I I ~ C ~ R p~s \~ l ) l e

tl-eatrnent of t l ~ e English t.enses ;is perspectival rzther tll;srt incjexic;il. 'T11e ynestioi~ xises wt~ether dl tl-ris con~plc:xity is neeticd. M;lyhc this is too

m~ic-11 arid sonre category car1 be i%srrsisscd as strperf.J~~ous." 'jut drlztthcr possi-

bility b &:kt tlis is stiU not enough. Ir~cleed there is a rlcttiorl of cotltcxr;-shifi that

has tleen prorrnrtent 1x1 the rertant htcrature c-trt indtmc.,l)~ty and wlllch 1 11,lvc not

dealt ' ~ v ~ t l l yet. To m d e room for rr, ~t \earl\ that wt. need ii ftiiii ~~ltcgot-y

6 z. S/~r(tiible rrzdt>c.xlcrzis in attrtltde r p r t ~

Twenty ye.m ago, A~lcieric)~~ and Kecrun pornted ottt that In sortPe larrguagcs tile trldeulc<~lc occrtmng 1r1 the emheiidect portlot1 of a reimrt wxi have tlmr vdues detertll~rleii riot by t l ~ e context of nttttr,triic of tlrc ~epor t t r ~ i t by the cotjteut d t b e reported Lrtter,%ncc (Antier\orr arid I<c,et~an 1985: 260, 3 0 1 ) .

A \111ul,ir otxer rmon bvtu rcpovteci by E+rtec (rcjS4 and rg89), ailti crecirted to F m a ~ o ~ t Bac 11 In Alntrar~c, Partce notes, ' t l~e fint petron prorlotrrt { c An he] used as a logophonc pronoun ccrreferential w ~ t h the \rrltject of ,L velb of\avrrtg

or beltevtng' (1?1ltee 2mq.1 378 n ~ ) . " ~ In 'A Pled Monsrcx\', \.lthlerrkc* h a used tlm arid reLtted Gt ts to rrloavate a tlrastlc revxrrorr to the \~*tn.lttard, Kaplanldr-, approach to u~deslcals (Schlenker 2003) Arct~rtiii~g (c) S~hlcnkcr,

?Z it car) indrccl be argued t h ~ t ttir 'pmprctive pomt' (on whrcll tile remantic vdue of persyct-uais de~)alds) itself is an ' i~ i tent lond~ feature ofrt~e context, a~rcl ,LS suctl can be rlufteci at wlll.

'' Panee points out d i ~ t rlrr I<~go:ophon~ ilse of Amharic 'I ' inakcz ~i .i~tiriiax to C~.utaiir&'i ytuai-

incircator 'hr**' ili:~ritt, LOO<&: 17711. 1 3 ) .

Page 218: Recanati, François. Truth-Conditional Pragmatics

the value of art indexical depends upon tlte context, but rf the mdemcal is embedded 111 a report the relevant contest need not be the context of utterance ofthe x epon but may be the context ofthe reported utterance or Or

at least, come ~ridex~cds-those w h ~ h Schlenker calls 'shfiable mdex~cals', for exmplc the Russtan prfient tmre-dow for thus poss~bhty, wWe other rrldex~cals ('nnllltnx ~ r tdexrc~ ' ) can only be evaluated wlth rerrpect to the actual corrtext of utterance. Schlenker argues that logophonc pronouns belong to a

thrcl category of xndemcals that can only be evaluated wrth respect to the context of tilt& rrported utterarice/thou&t

I,ct 11s assurne that the typologjsts are nght and &at, m some languages. the rrldcs~cds (or some of them) are sh~fiable tn speech or thought reports. Are there slufiablc ~ x ~ d e x l ~ d s in Englr~lr or French? Scblerdcer savs that there are, while %plan fmously cldtrned that there are not: embedded In .i report, Kaplai arped, our fjnullar lndextcaIs autornatlcally scope out, in srrch way t h ~ t they can only be Interpreted wtt11 respect to the contest of utterall1 e of the report (KapI~n 19890. 510).

hi sup1i01-t of 111s drsenwng vrew, Scblenker grves as exarnples ter~~pord ,~dverbials surh ar 'TWO davs ago' or (French) ' Ja~2s Je~txjorirs' If, ISC week, John s a d to rrle 'I w a ~ sick two days ago', 1 car1 report h s utterance as follows

(16) 1 ast wrrii, John told ltir that he h.td been sick two &vs ago

1 Irre the indexlc a1 "two days ago' n evaluated wlth re$pect to the context of the reported speech ,ii.t.z5 T1xs 1s riot posd~le lf we replace 'two days ago' by 'tlie &y before yesterday'

(17) L,xst week, John told me tl.i,~t he had been s ~ c k the day before ye~terday

I11 contra5t to 'two days ago', die phrase 'the day before yesterday' cdn only be evaluated with respect to the context of the actual speech act. Schlenker coticlucles that 'two days ago7 IS a shfiable indexical, wli~le 'the day before yesterday' irllrents the ~rm'tllftablhty of the matnx mdemcal 'yerterdq'. But

2+iri a Kapianian Fwnework, this suggests that phrases like 3ohn said that' are, in effect, context- shifung opetacon ('morrstsn', m Kaphr's techilical sense). dufang the context for the evaluation of the indexicals In their scope. One cozlserluence of this analysis, anphasized in Anand and Nevins (2004) and Anand (zoo/)), is that all the indexicals in the scope of such a phrase ought to 'shift together'. " Mmy speaken of Er~$isf.i tand in particular, Robert Swlnaker) do not accept this interpretation of

(16) . But some! do, apparently.

Page 219: Recanati, François. Truth-Conditional Pragmatics

to so conclude one must first rule our the eventuahty that 'two days ago" mgh t be a perspecoval rather th,m an mdemcal. A sh i f~b le ~ndexlc. al shrfis only m speech or thought repom, w h ~ k a penpecnv,il may be cvaltrated vvlth respect to any s&ent penpeccrve pomt, even if tho perspective in

yuesnon is not that of a person whose speech or thought I \ being reported T o rule out the possrbllrty that "wo days ago' might be a per.ipect~vab,

Schlenker argues that (18) below rs 'tdegraded', while it ought to b t perfectly acceptable ~f 'two days ago' were a persl,ecrrval (Schlenker ~ o o j 65)

(18) I nict John lact week. 'Two days ago he wa"rich.

But the reason why (IS) rs degai-ied nzw have to do with the ~ h o r c e ot tense, which arguably atZects tile selcct~ort ofthe temporal perspvcbve pornt. The following seenis to nie (more) acceptable

(19) I met Jolm last January. Two years sgo be had been very sick, but nriiw he looked much better.

In (19) the temporal perspective point, referred to by 'now" 1s set to tlzc tlme of the January meetmg. '3 svu v e m ag0' refers to J. tune two yean

before that meetmg, a (one rrrsglrt cupect d'tvj-o yean ago' n a perspec taval

rather than an indexcal. Schlenker hinlself co~isrders ~11r fact\ rnvolvmg 'two days ago' ac ic \ i clear

thm the facts regarding French 'dcluvi3 deux joz.rtc7, but ttrerc &re also ~ X ~ I X L ~ J P S in which 'dans deux jntirs' seems to work hke a perspechval r ~ t h e r thari a

sh~kable indexical. Scldenker's case tor tlre exrstellce oish~ft,~ble ~rrduwcals m Enghsh or French is therefore not tullv corrvmi~ng.

t believe that a better c'ise can be nsa~iie fhr Scl-rlenker's cl~srn, bdsed r,n ti-re

En&sh verb 'cqme'. In 'Pragnrtr~cs .mii the L)e\criptsoxl of I)lscorrrsr' drrcl

elsewhere, Fillmore argues that kcorne' 1s dxl exception to the generabraaon that Enghsh indexcals are not \hlftab1e In rcported speech

In ;l sentence hke

izo) 1Ze uskd her to cornc' si hli p'irty

the word come can he undentood as the word EIE W O L L I ~ h ~ v e used, not

necessdy the word I would use d r h n were not an rns~tnce ofreported speech To show that dlls IS so, we can luxtayore to lait: selitencc one m ~vft~clr the

cho~ce of come or g o mu\t be 111tde 60111 the pomt of view oftllr speaker of the outer sentence

Page 220: Recanati, François. Truth-Conditional Pragmatics

214 SE1li:'I APIX b XNI)bXX('AIS

( 2 i j J le 'lskcd her t(i c0112f1 10 his party. Ski(* stticl that $ 1 2 ~ ~ u o ~ ~ l d , hztt it1 the end she decided ' 1 J , 6 (E:illn~ore l \lN I : f 50)

' I ' t r bt, ssrre, t1.o f a ~ - t that 'come' xwry bc i~ltcrl~rcteri froni the point ofview of the pcrcorr wlic,sc speech or ~fiolight i s bc-irrg reported is not sufticierlt to s l ~ v w t l ~ ~ l : rt i b :r sll~fiat~lc indexical. 1C t.hc pr-ohlcm with 'ONO days ago7 was that t h i s cxp-cssion rimy be ,r perspeitivai, or h,ive pcrsyectival uses, we clt.:rsiy Oci: rlrc sarire problerrl wit:fr \-or~~r', wknosc penpt.ctiva1 rlat.ure is tvell-

cstnlriisheci. &Zl.lt \VC have seer] that. ' co i l ley i s , in fact, hoih ;LIT inciexicnt 2nd n

pcr?iyrci.tivrui. Ac~cortEing to E'illn~ore (xc,q:j, 'ccori~e' h;~s :a perspectival use in tllird.-pcrsol, riarrauvcs, but c.>tlrerwisr i t is arr ix~dcxicd, governetl by n

tokihn -T~"J~;L%XIVC C L E P ~ " . 'l'lle cluestiorl tlist ~rnscs, tiiercfbre. is ~vllethcr 'come' i s ;in ixrilixic-a1 or. perl7cctival iii shiiieti tiscs of ~ h c type illristrated by (20).

'7'hi.s ci~~cstciuri zdrr be dcc-itltd, I>ec::luse Fillrxic.)re. I r a devised a test for distirrgaiisl.iirlg the persyrctivd uses c ~ f 'cuanc' f i o n ~ its indexical tlsrs.

'I'ht. test, whic-h P I>a\~e alrc3dy rrrentioxicd, irrvolvcs the possil>ility (or E~11possSii-rility) tlf ilit~aselltenti;l1 shifts. 1n ttiird-j>erso*i rl;lnative, tliere car1 be only oric pc"'q~ectnrt~ point ir, ;r i;ivetn szntcncc. '1-hris we carinor say

"Rlior Jtri ix~i c,rrllo rep I3ril's lac~u~e, Jolial arid Bril together c,mlc over to M3rv'i l ~ o l ~ ~ > e

0 1 1 thc orhrx Ir,rnd wc can s ~ p (or1 the 'irrilcxrcal' w e of come) I l i you i orrxe ttr ice KIC tolt)orrC)w, 1'11 L O I I ~ C to see you ~if t~r- ton~onow I

Now, i f -we apply t h i s test to uttcx~rlc-es in wlnckr 'collkeys ir~teyreted &om

the ptr iar t of view c.aftfle penon ~ ~ 1 0 ~ 6 speech 01- tilollg1:lit is rrf?ortecl. we see

tiixrc- irirrasur~tetiri;il sllifis are possible. Wc cxrr say things like:

John cold hfi~;dly that he wviild to~xzc to scc Ilrr tlie ricxt day, unless h e pefecnecl to ~ o n i ~ ~ t o itis PIACC I d t t ~ ~ 111 tlir wculi,

Insofir- ;is this is right, we <:an concluclr: that the sirified rise of 'come' in rrporteii spccc:b is the in~~exici! use: \vlicrice it Ibllows thnt 'conie' is a

shifkl Ae irrcicxic J.

(i.,j. Coralr.xl-slagis in mixed t juoi~ifi~kfi

'The rc;illy Jifkicult cl~restiox~ is drat oCtht* relatitm between the sort, of-shift that c-lranli.ic.niztr sl,ilt:ablc indcsicals iri ;ir-titricle reprorts and the context-shifts

Page 221: Recanati, François. Truth-Conditional Pragmatics

ment~oned earlier rn thn chapter. I 11ave ssld that cfrrect \pet.ch reports

(qur~tatlotrs) c m be arrxlysed as mvoirnng 2 6>1111 of co~rre~t -" r~l fa t~g g)rc

term Now i t m;ty be that, when ari rndexlcal sh~tis in report, a5 I r l the A d l a n c exanrplehs cil\cussed by Sc Idenker, the \,irtie form of currteut-sh~lt- Ing pretence operate\; 111 wl-nch c ~ s c the repons 1x1 ytlejtlon ,Ire 'i mrxtulc of ~ndlrect spcecll report 2nd clurtt,ltion 76

Su i h rnlxtures are riot unlteard of A good (-led t r ~ s been written ,rt>ctut

the pher~omenuri of 'nuxed quotabon' 111 tlle recent phtlosoplr~cdl L~ter~i~urc The standard exatriple rs J)civ~dst)r~'\ (1979)

(22) Quine said tlzat quotation ' lus a certaln a~~ot~xdons f5;rtltri.'

1'111\ lr a mlxtirre of ind~lcct \peech repoit . ~ r d dlrcct cjuot'ltlori (her~ce the

phrase 'mixed cfuotat~on', courcd by Cappelen dxrd 1,epore 111 their 1097

nrt~cle). In Kecanatl ( ~ o o r a , reprinted below w ~ t h rnoc!rl~rat~oxa a\ Chapter 7), 1 argue thdt EI ln~xed quota tloxt the speaker t i m , in reportutg the ( orltertt of an utteiance, the very words that xniere uied by the speaker o f the onginal

utterance, and bv rtressrng thole word\ (by ~inl?taving t h t ~ n u, J I I o5tel.r\ive manner) mabcs ~t rxr~r~~fcst (hat thore arc tlit very wcrrd\ tr\ecl by tlii.

ascnbee. In other trnnc, tire speakcr rcports the uttsr.xrlcc ttsitrg ~nrllrect speech 1n thc usud t~i,uirizt, alrd st the sailre tlrrle XmrnIC\ tilt" \pedkt-r 01- the ongnal utteranc c b y octextclvely ilclng h1.i ow11 welch Ori tlm vrcw (22)

.in.dysed 2s follows: the speaker cupres\e\ the propoc~tiotl that C&ilnc .;ad tfi<lt ctuotdtlon h'n ,L certalrl anorrzalous feature, and at h e \arnt. tirlie I ~ ~ I ~ I I I C ' ~

Qmme's use of tile words '1)~s a ( ertain *ulor rraloir\ feattlrc' 'I'll~s n sl~sx~lar to w h ~ t we find in tire fbllow~ng examples, whcrc, 1x1 rlie course of tepon111g the ascnbee's speech act, the speaker n m l c s tlrm or her hy pt~rctstng #lnd/or

prollouncrng the complement sentence In a certan~ way

( 2 3 ) 'Po wl~iclt Mr Bailey modestly replied titat hc hopcd hc knolovveci ~vo t

o'clock i r wos in @nerd. pickens, kfilrtirr ~ ~ ~ z ~ z x l e z ~ i t , cited in Clark 2nd Gerrig rc)c)o: ,791)

(24.) L!tze vieifleji.mmc. . . virzt at4 seuil rt me dtmllnda yrrr'. cjttc j'vaulais, d'mne voiu

trafnantc ~f hagneusc. (13arbe)i cl7AureviUy, l.'Ert,.orceli:e)

"" Were this so, Schlenker's a~tertlptcd refurarton niK;kplan would fall. For Kaylarr's cialr~t ts not that

mdexlcals cannot be sbified, biit that they canrlot be shifted 'wiiholtt cneakirig 1n 3 yuor:kuon cievrcr' (K.rplan ryKga: 5 I I i.

Page 222: Recanati, François. Truth-Conditional Pragmatics

Mere the cpeakcr reports the aqcnbee's tdk, using indrrect speech m the n o m d way, and at the same nme shows >vhat that talk w 3 ~ lrke. The speaker mght slrrrrlarly gesture n r a certan way to m c the a~nbee's own gestures.

In (23) and (241, arguably, the language feature of the context shifts through the pretence tlie speaker does not nierely use the ascnhee's own words, he u\es the d\cnbee's own language and rrranner of speakmg In Chapter 7 1 d~scrxts other exarnplcs of nuxed quotation in which the language fe~ture of the context sltifis.

(25) Paul says hc's due to present his work in the 'paper session' (26) Jarnes says that '(Qtrine' wants to speak to us (27) Nicold believes tlrar his father is a 'philtosopher'

In (2s) and (26) the expremon wlthin tlie quotatior1 rmrks IS not used m t h its standard n~eari~lig, but 1~1th the Inemlmg it has for the person whose use zs

being ehozrnlly ~1rntrIuted In (25) the speaker does not use the word 'paper' in its nornlal \ence, that IS, in the sense ~t has ln academic English (where it nreans urticle), but in ille seme it has m Paul's idiolect (where rt means what '~mster' nx3.1~ 111 at nderntc Englnh) 1 he sarne thing holds for tlie proper name 'Qulrle' I n (26) that name ~~ormallv refen to Quine, but in (26) it 15 urcci ~ronicxlly l r t he sense it has m jan~es' idiolect, wl-iere it refen to Tim McPI~erson {wl~orrt J~rnes nzlstakes for (Jume). So (zj) expresws the prop- o,itlon t h ~ t I'CLII~ uvs he's due to present his work in what he calls the 'paper res\lon'. rwniely the poster session, ~ n d (26) expresses the proposihon that Jlrne5 say\ that the rrlifiv~cit~d he call\ '(,)uine', riarnely McPher\on, wants to bpe~k to us Lxarrlplc (27), borrowed horn Gappelerl and Lepore (r(~c)7), is rltore co~riplex In a x rib~rig n certain belief- to the five-vear old N~cola, the spe<il\er user '3 word &oxn Nlcola'r idiolect That word does not e m t rn English, cvcn tliough ~t 15 etvrnologcdy denved liolrl the English word 'phtlosophei'. Example (27) purports to express the proposlt~on that Nicola bel~eves. 1113 fatl~er has the property he (Nicola) assocrate., with the word 'plultosopher', whatever that proper@ may be 27

WItrle in (25) m d (26) we krlo~v what the relevant words mean in the relevant idiolects, in (27) it's hxd to say exactly what tile sense of 'plriltosopherr' in Nicnla's idiolect is. %'hat makes this example complex (and interesrtng) is the kct chat both Nicola's own use and the speaker's echoic use can be said to be 'deferenrid' in their own ways. Nicola defers to rnature speaken of Enghsh irt his use of what he cakes to be an Englisll word, wl~ile the speaker of (27) defers to Nicola's own use which he mimics.

Page 223: Recanati, François. Truth-Conditional Pragmatics

Just as the language feature of the context can be shrfied 111 the cornce of nwru~l r ig the ascnbee, other tkatr*re, of the context c'in also he ~lnlfircd The follo~mng exa~nple co111es from Cappelen arid 1,elrore ( 1 997 429)

(28) Mr Greenbpan sad he a g e d vat11 Labor Sccr etary K B. Kelch 'on quite a lot of h g s ' . 1 her accord on thir mue, he 5ard, hits prctved 'q~~itc" A strrpnse to

both of us'

The word 'us' here refers to Mr C;reensparl anci Mr Re~ch . I t does nor refel to a group including the speaker of izS) The first person prctiioun '~1.i' nrr

(28) is interpreted \nth respect to the 5blftecl. i oritext of the reported iyccch episode. In that context. the speakel rs Mr C;reen\p~n himself (rather than

the person who reports IS utterarrt e), so '11s' refers to a grotip ~rrclu&ng Greenspm rather than to a g o u p rraclrrdrrlg t l ~ e reporter

The Greenspan example comes &om the New York Emes In thr same newspqer I h v e found other ex~xnple.i, where the context-shrh aN?cts the reference of an indexicd What hllowc n an exa~nple irivolvirrg the flrsr- person pronoun:

(zy) Lev1 Foster, in hct, n the grea grc<~t gr.~nclfather ot Gov Mlke I ostct ul

Loulu,ira, who s a d recerrtly orr a ra&o progr~rdnr t b ~ t it rlirouici be 'news to

nie' daxivone ul t u b farnily had owned .;ldves

The first percon pr0110~11 'rneY ~n (29) rcferb LC) the ~\cnbcc, Mikc rorter, rather than to tbe ascnber- --the speaker of (29)

Here we find an intere~tiiig dlklercrrce betwerri cvliat is pennlisibLe m the

ianguage of Engl~slr-cpeakrrlg nrwspJyerc drltl ~vhat 1\ permisable Iri the language of Erench-spe~k~ng rrewspdperc The r onven t~on~ governrng mxed quotation 111 newrpapzr-wntirrg turn out io be di8'cre11t in tile two

language~/culturts. In French, the wde;\rc,il\ are not dlowed to shrft 'tn vruxed cluot;ltlon. Ex'unples ( 2 8 ) a d (29) wollld have to be rephrased d i

folfows to comply with tlre French conventlorkc

(28') Mr Cleenspm sad he A & T C ~ ~ wldr Libor Sec I etary R H Kerch 'on qultr a Lor ot tllmngs' Theu: acc nrd on rlii\ rime, he sad, 11'~\ proced 'qtute a siirplrv to

both of then^] ' . (29') Lev1 Foster, In fact, is the geat-peat-gandfather of Gov Mike Fmter ol

Louisiana, who said recenrlv oil a r~dro p r o p a l that it woiljd be 'xlt"\v\ TO

[h~rn]' f anyone 113 lus Cjaidv 1i;lil owlled $laves

Page 224: Recanati, François. Truth-Conditional Pragmatics

Wc cxrr say ciclic~r t l ~ t t I:rcr~c:h iradexicnls are riot slzift:~ble in rnisccl cluo-t;t--

tion irl tiac iarigii,rgi. oi~xre.ivsp:cpe~-s, cir ( c i ~ u i ~ d ~ n r l y j that that iarlguage sets lirraits tcr the axlixing of tlisecc aritl intiirccc spc.ccir: the only fe;tture of the c o r i t c h t ~11at c a n sllilt i r l rrlixcd c~trotatic>rr i s tiic; 1:1rrguage feature.'"

l i r Ilg.lir or'ti~is dlikrerlit. Lwtwecrl lirlgiislr :rtrcl X:rcnc-h we might consider

tllr p)~""SEk-irlit\/ t l i ~ t S O X I I ~ t;'lrrg~idj_:~"i i;)r .;"x;il~lpk "unhasic----(lo not even iir-xw ,I slx;ix'p tiistixlition between irrclirc.c-t spccxch wports nrlci direct qilota- tion, .illi$l ,I(.CC~PI COIL~CT;( . stliiis in I ~ J ~ ~ X I : . I T ~ C ' ( I IIIST:*XICCS of inclisc'ct reports. Unloier tP l i4 kiypotliesis, d ~ ) wcstirX necci ,I ' f i fd~ catcgony', that of stlikable I I I C ~ ~ X J ( . ; L I S ? O r diws the phc~loxncnorl ciisccxrsscci I)y Scl->lunhcr reduce to that

of' rniscd i j L i o tG i t xo r i , I.rerlt:c, ~i t inr~t te iy , tc> ilic tbnrr uf contest-shitting pw't-nl i~ h~~nikr t l abo~tl: in cor~~lc*c-tiorl wit11 ciirt:i-t spccc:li reports?

I L~eircvc \r/c ilo rrCc.cf the kiltlr z;iccgory, I?c.caiisc (:ultl to the extent that) in

oric ~ r r d thc. r'iaruc E,~aigrr;ige, riot aU iridc:\-ic.als I?el~avc ir l the same way. 'Thus ~ I I

K irssi;rar. Si.11ic.11 ker cidirris, the prcscxrt t c ~ 3 ~ i s ;I slliftahl~ ir1dexica1, but the f i r s t pcrxori i i r r guLrr is not. A distinit~c.in bi"t\v~"en the iangu:rges wllich ~ l ~ a i l ~ t ~ ~ ~ r ~ '1 ~ I J I - p ( i ~ ~ i s i o i ~ ht:t \'vc!t~l inc1irt:ct rt:por~5 ~ r ~ d direc-t cj~tc>t;itio~l and

t l l c . l~~lp~iiaj_rc"s whl('/i CJC) t ic~t is t t ~ ~ r ~ . t i ) r i % 1101 "r\lfIicii~rit; {vitki respect to give11 larlgtr:ig,:sb, wc: riccti t o c!.s:s:lmv '1 tita-thcr tiisurrrc.iiosi ' r)ct~recn tht. irziie..l?c,zls wixich ctui 6 , l i i f i . iri i~iiirri%c.r iliscunr~e axrci t.i?o~e cvllic,ir c.;ixrrlr>t. Moreover, tlre hct that I:rc:iirh (or r:itlli:r, t i le J~iiig~~:tgc 06 Of ;rclizl\ iic\vs.jrapc.r-s) does nor ewily aUow ~ 1 7 ~ i~idcxic-;ds ( t i ihitt in rtiixetl i juocacior~ C;PII IPC iiii~d to lispr rove the Ilypodicsis t8a;lt t l ~c p,licr~cltrrer~c>n of 'sd~ifkirlc ilkcIcxicals' reduces to that of

nrixeii iltrotatiorr. a;Sic%re this liypoda":si~ig;hr, cl).crc r~udi t to be no sh i f~b l r

ir~tiex~c,ds in I'rcbrich. B u t thisre i s at least oxit*: fix wb;it I have saicl of the IE~~glish ve~rl~ 'c orric-' applics to !'av.~~clt " v c ~ ~ i r ' . i t &>!Lows that the shiftability of

Pnilc.xia,als I ~ I attitutle reports ciocs not TC"(ILICC to the possibility of shifting the i-orltcxt r r l rxiixarci ijiiotation. '['his iiocc xirr t il'ie:)li i h , ~ t ilit' two phenoxnena ;ue 11 r.rrck>t cad. \rco\*vt.-vt.r. 'I ' l~c ex;it.r ri-.lation r liLit hsjttls bctwc.cn them is a crucial :rrrd c I i t f i t : ~ ~ I t i p t ~ s t i o n , \vl,ich iirtl~re work oirgh to ailth-ess.l"

'VI ' l r i \ *Jxoiilii bra ( / i ~3h f1 t~c i . Eve11 in /:S'IIC~I I I C W S ~ + ~ I \ , ~t IS ptaz~bie to f ~ n d occilrrerlcrc of silifird xndtsxiz.ris \vitbiii ijict)r.ii.ivli\. So rlrr i1ifi;~rcnic Irerw.r<.ii I'iiglnii .md i7rencir rceins to he n matter of tkgi'"

LV I'llr. ongirl <)f !lit\ i.1i:ipic.r i s '3 t d k 1 ~ ; w i . t i tire i i.irvnd-MI X -ilronri ~vorl\Fhr>p on 'Index~cnls.

spe~.ci i XLS .ind I ~ ~ o ~ h o l v ' ( I ih i r ih~~i lge, iMA. 3.1) No>t"~nI>i.r ;0(,4) jlohert Sr,iln&er was tny respondent. ".;in<-e t i i e i i iiwrc lid** hwi i iriticlt work on in~icxi t , i i~i~~ mil cii ir ioxt-\i i i ft orr the prt ot'senrantic~sa. (For a

s i i i iTr,y, w e '.,i lilenkcr kwtlii c~ in i r~g . ) Ariarrci (2000). iir p~rttcu1.11, erpl~t x i l i dirci~\ies the relation between \ h ~ f i ~ t i d e I ~ I ~ ~ C : X I A '111cl 1311xg~d <jl6i>tdtIOZl

Page 225: Recanati, François. Truth-Conditional Pragmatics

Open Quotation

r . fritroduction

Philosoplncal works on quotation generally ;tttrlltpt to ailswer some ofthe

fbllocving, basic questions:

What is the reSerence of'a qtiotatiotl? IS it 2 lil~g~~istic o~pressio11 ;IS

such, or a bmte shape? Is it a type or a token? Is it possible to cpiote

meanings or contctxlh as well as t'orrns of w-ords? 0 Whdt doc\ the relcr~~ng in '1 qrrot~tlon; 1\ it the quoted matt.ri~l i t d f

which (self) ref'ers? Is ~t the complcl e\prc\slc-tn, c orr\i\tlng of t11,rt rnateri~l ~ i i t f the ~luotci, which rc-fvis to the quoted r~laten,d? O r 1s it

the quotatloll ~narl\\ whlcll do ~ l l t f~e rekri-111;: by tl~ertl+elvc+' 0 Wlr~tever plays the role of slngulr term, what wrt of-,I vx,gui~r tcnm 15

it! Is ~t dk.111 to a detixllte clescrryt~on, or to J gcxlultle vogul,u terx~lt? If the latter, shouid ~t be amrtulated to a nanre, or to a tlemo~rctlat~vc~

?he.ie, and other yltcchoizs in the \aole vein, 'irr Intcreitlng mci wo1tf1 anjwenrig (to the extent that their yresiippont~ort~ 'trc J( ceptable) U t ~ t clrcy Loncern only a part~ci~lar type of quotat~on, wilicl-1 1 call ?lo\cd quotation' I he otllez rnzm type, 'open quotatlctn', rs ~gnorod, and tins xteglcct lead5 to bad theon~~ng . Not only 15 J general tlteoq~ of q u o t ~ t ~ o n out of reach. the ipecrtic phellonlenor~ of closed quotatioil ltself carrrtot he ~'roperlv undcrctood d tt 15 not appropriately \~tu'*ted iv~thin tl~c. Eitrrd to wlrir h it

belorrgs. Most ttleonsts are dlrnly aware thdt here Are exotlc vanctles of quot.rtior1,

not encon~passed by their theones. ‘Start (1l~ot111g' 1s Jn esarnple But thw take this ft,n;lt of quot~t~oxl to be iiindanrontdlly ditferclrt froln w h ~ t they arc

Page 226: Recanati, François. Truth-Conditional Pragmatics

lryrng to deal w l d ~ , so &&relit that it 1s misleading, they think, to call lt by the same name. "Thus ~t IS a wdespread practlce among phdosophers to deem 'cjuotrt~on ~mrks\arnblbwous, and to attempt at &sanzbrguataon by usrng bstlnct ryn~bols for (closed) quotauon 1n the stnct sense, and for 'scxe quotmg' This suggests that the two form do not really belorig to the same kind I,et us cdl this suggestton the Homonymy Thens. It is imphcit m many ph11ocophica-l itlscusuons of quotatlon, but nowhere n an argument expllcldy addu~cd m I C ~ favour. Nor is a svsternatic cornpanson between (closed) quotahon and, for exdnple, scare quohng anywhere urrdemken- to my bowledge--m pl-tdosophtcal stu&e.s of quotanon.

111 recer~t yean, undoubtedy, some progress hw been made. Tbls is

manly due to tllr influence of Davsdson's iconoclasnc vlews of the matter (C>avrAon 1979) In their paper 'Vanenes of Quotatron' (1997) Cappelen dnd 1,epore 111\ist that any theory of cluotatlon should account for the phenonle~~or~ of 'nuxed quotat~ors' rllustrated by Davldson's firnous example

( I ) (2une say\ &at cfuotjltron ' . . . has a certain anomalous feature'

I w1L1 myselfctft'er an ~~ialysis of rilrxecj. quotatlon m ttus chapter It n, I w~l l argue, an lr~rtance of open quotatlon For lack of a disnnction betcveerr open drld closed yuotahon, L)avldson, Cappelen and Lepore, and other recent tfieorlrts \who h,rve attempted to deal with it have been unable to prok~de \ansf~ctory accounts But tc was a good thng to broaden the scope of ph~lctsophlc ,tl iiisctia~ons of quotation by focusmg on that phenomenon--- tile t - t~ of 311 iceberg, m mv vlew

(_)ix-e die dirtrrlcnon between open and closed quotanon has been i2ra.cvn dad properly apprec~~rted, ~t 1s temptmg to consider that only closed quota- tion is relevant to semantics Open quotabon is more a rnatter of pragmaclcs. it is a matter of w Isat people do w ~ t h words, rather than a matter of content and truth roncfitionc In th~s way one can provlde the beg~m-tng of a just~ficjlt~o~~ for the negect of open yuotatlon m current semantic theonz- irlg Tllere is sortte truth in ttus view. yet mmed quotatior1 1s interesting precl\ely because zt shows that thmgs are not so ssrnple M~xed quotanon, I will arbwe, provsdes pnIne evrdence In hvour of the pomt of view defended UI tbls book, tlrat pragmatics affects tnrth-cond~tions.

Page 227: Recanati, François. Truth-Conditional Pragmatics

2. Quotations as pictures

2 r i>lsptuylfzg itnd d~mouz~trat~tq

In quotation, as several arzthors enrl~h't.irzetl j( :hnstc-nsetl 1967, Searle rpriy,

$4.1; Read 19~7) , the quoted rriatct~al 1s disltluyed or pre~ented That ~i~crar~s that a token 1s produced and the attellhon ofthe audience rr drawn m that token. To be sure, tvhenever one r'lys sometlimg, one produces (token.; of) words, with the lntenboit that the aud~er~cc pcrcelve tl~enl. Yet one does not normdllv intend the addressee to pay roriscrouc dtteXlhUIl to the words oiie utters. In hnguishc cornm~lnit at10~1, the ~vords are automatlcaliv pro- cessed, and audience attentlor1 i\ dr'awil to what one says rather thd11 to the means by which one r'lvs ~t Wfren word.; are mentioned, liowcves, tXre

t?redttttrt Itself is brought to the Icrre&orrr of rrttentlon: thc words are da played, exhibited Thls 1s a form 01. ostcn~sion, as L3:~vidsurr and ot11e.n

pointed out, but it d~ffers &om ordrnary o\temrtrn m one re\pe~L tlic osterlded token docs not exist ~rictcpe~~der~tlv of' the dct OF ostel~drng r t

Wlien I polrlt to a bird, the btrd e\clc;tT iricte~>ertdertt of u-ry po~~l t lng i?;)4rat when I mention a word, the osrexlcled token 15 produced- rt corrlr\ into

being-throug1~ the verv act of O S ~ P ~ I S I V ~ (tsplav At tliis point the yuestlon drrses wrkat exac tlv IS displ,cveclV oLex-rs or

types? The proper answer to that question 15 both D'ivitfson cavs that tire quotation niarks 'help refer to .a shape [a type] by pomtrng out somctbnng

that has it' [a token of the type1 (Davrdson 1979 90) Thn ~orrld be understood m the hght of Ntinberg's dr\~ix~ctiort between incleu. dx~d rekcrent

(Nunberg 1993) W ~ e n e v e r our pornts to soa>ethmg, Nunberg S J ~ S

distmc~on should be drawn between drat wl~rcli 1.i pointecf tcr (the mdr.c), and that whlch 1s rekrred to tbrouglr that pomtrng. Even ~isorx~ehanes they happen to ~osncide, tiley do riot always do so Thm, hattdmg m y c ar key to

an attendant at A pdrkitig lot, I Isray 5,~y '7'ba is parked o ~ ~ t backy the rndex w '.t" 1 --.- -L ^ - *L_ u r c h f i y , ~ i l i iCuzIcut 13 ulc- L.*L Oi t 11147 PulEii to a pdir af \~I<JCS ( ~ I ~ c A ) utter, 'Those are no longer 11n fjshlo~r' In ro dorng I may well refer LO t ? ~

slcoes that are ofthut type rather tLxan to that part~culat palr of shoes Clmilurly, one nug-ht ague, the quoter pomb to a token (index) and thereby destg- nates a type (referent). For reaons tlut wdl become apparent, however, X do not want to bnng the lloaotl of referer~ce 111to the picture at &n stage Rather, I suggest that we &stixt_~~sh between the token w h ~ h rs dzrf~layed and what jr (thereby) &naorn&u&d W t ~ t the speaker ultm~ztely cdls aaei~tloxl

Page 228: Recanati, François. Truth-Conditional Pragmatics

io i s r a a r i thn* tlrspl,;;vrci coke11 itselfin ;dl its cing~il:lr?ry but ccrtain properties of it., that rs, strrnir type w k c h it ia~stnrlti,~tcs.'

-1'Eic cl,iarn? r$r,4t cjuc~t:rtiorr i s ;a rxllittc'r c a t ' \ i cmi~x~s t r : tg a type can be iili~len~ooij 111, two WAYS. ik%:lt i s C C ) I I ~ I I ~ O I E to t>t>th ixlterpretarions is the a>stCnsIvc I~:LIIITC.: ot'(.j~~o~:~tioll: the (Jell loll~tra ted type is exhil~ited by &splaying

a rokerl o/"didt vpc. S~)XIIC irlicrpri't tlris iri ten-rn o i r e fk ren~~ . For I>avi&on, as

wc hravc seen, trv denrorrstratc a type is t o rct'c'r to i t by tlisplayirlg a token ofit. In Sectiorr ;, I \viii cridci/c Ilt~vTclsorl's cor~t-eritior) t i lac rile yuc~trrtiorr marks have a rekrrirrg ii~nctiorn ,kin to that ctf n ciex-nons~r:itive. The othcr interpretation,

w/iii;11 1 :~<Jt)pt, i~ IXIOSI ~ ) m r ~ ~ i ~ l t : ~ ~ t in tJ~c ~17tli . igs oOI'CIIC. psychol~gist Herb Clark (C:l:ra-k &u ic i C;e.n-ig r9c;ro; Clark. xc)ydj. Ia] cc~ztr:s\t i o 1)aviclsoa. Clark tlses

"denroaatr~tc" "ri i t s cvcayifay serisc oi'"'illtritr-;xtc. hy cxc~npkficatio~~"' (Clark :rrid (;c.rrig rci:)o: 704 in.). In tllat. serrse, (:lark. ;mtlti Ger-rig say, 'you can ilcri~trrrstr;itc ,i tcrrrris serve, n ti-ie~ld's Lirrip, trr tl-ic nlt)venlcnt of a pend~rlunr' (B.:lar& ariti (;i.rng a()Oo: '764). To 110 SO yotri Pnust yorrrself produce 'in i~istmce

e~i'tbe sccr..i:e, rl.ie l i t r i p , or the r~~ovenlerlt. 111 ijr.icrt,lr;ion, wll:lt we cieriic>nstrate is 2 piece oi.vcr-l~ai I>ckxaviorrr- ii way ofspeakialg. Wc alcrrlonstnltc it by produc- rrlg ari i t rztallc-ih (pi' that hehavic>kir, that i. by sye,ibirrg irl tlkc r-devant way."

T l r u ~ ~ o ~ t ~ " . t i " s ofthi' iiispl:rycd toketi tvliich the. cirrotcr ~~111s attention to rrc,r~f riot Ile jrrirciy iirrg~tistic propertic~s----the sort ol'propcrties which are

c.irilstittrtlvc OF li~igdiiitic types (word-types ,11-1(1 si:~~rer~cc-types). AS Clark

poirxis trrrt, wc do rrot rrreroly derrronstr;tt<: tlrc .cvarrds, ixrt we also 'depict all rnarrrlc'r of' spcccla chrrr:rctcristi(.s -speed, gender, age, dialect, accent,

' (:.ippc:i,i~ .iiici i ~ : p o r ~ " ( d f r r ditlcrerit iiiierpicr:tl,ilon ui I)ai.icisorr's 17.w.3ge ( JUOC~UOII maukr, 1 ),iv~d\tiii uvs. 'IicIp i c k to .I +li.y~c lxy poirrc-in$; ou t \i~nii.tiio~g ti-31 11,s 11. . . . f'lre \inguiar ten11 is clte qi iot~j t~ot~ r n . ~ ~ k \ , \\11)( 11 t~i&v bi" re,ad A S . ' ~ I IL* t;xpr~ssion 'I cokt~i of '.vit~ct* I \ !here"' (!>.~VI&\~XI 1979: 90).

I'.ippcieii .irrd i cpoie r.iki i i i i \ .it ihcc v'riur 31111 (~o~i ( . I~ i Jc thrit i j i i o t d t ~ ~ i i inl:irks dre 'dtlfiliite dcscripnons c o r r t a r r i ~ i ~ ! ; deriiori\ii,iirvcc' "the derrwitsrr~iivc~ f ~ r i ks ("it ttirc rokcir wit111ri LIIZ q~~ot .%~iori IILAT~': and the (iclinirc. rlr.ci i rpiioir ii?ltorr.\ , i r ~ exprc',sroti, r.e., a s t r , r p c o r J pattern, rnst.urnated by tire tirn)orrstra~ed tohen' (( -ippt.icr~ .iriii L cbjii>rc. x007: -(;q). Ail~airgii piiisii>ii-. iiiis rnierprct'aticrri uf'l)aviiisoi~'s prssage is riot rn3xi(1:itiii \ 1 '11~ prii,ri~ve ccjulvaieir~c. I-ecwceil the q:iotaui,rr n~,ir.Ls ~ t i d ilic dcscnpuve pllrase .the

Cxpreul(1i1 '1 :.o/\r"il ofrrl./ll~/% iiere' 1 ,in be corlsiriicti rrrore ioo!,t'iv, o n tile pdltc.nI of tlrat benveea, r;zy, tiic pn)nour<'E' ar i r i i l l ? dis(-rrpnon 'the Irrociiicc.r cd~h!s iittes,mcc' t i ~ t t as i t may, 1 drri not cr)~lcerrred W)L/I F ) J V I ( ~ S ( > ~ I I . I ~ I CXC#C\II i i i tiiis paper. I I I~CI I I I ( I I IC(~ I ) r l v i dso i i ' ~ p.sbri~:e (itrly b e ~ ~ u s e 1 tliirlk i t poirlts in iirr riglit iiirccirfil~

Accorciiiij? to (.-dppeie~i (1' i.), the evi:rycby scnsc ui"d~rrronwi:ire' urily corrccrns evenu or acnorls: Y o n <ierr~r>115tr3~e Iinw t o iitr :~orneri~isig, or how sornrrbrog iisppeirs or Bappt-ned. You can't (in th is ierise) iicitionst~:iii d i i ~pl i le (tiiouglr you can dcriiclnstratr h o w to r:di x i apple).' 'This rnay he ngbt, b ~ t t the TCS~.IICIIOI~ 111 ( j13 i ' "~ i~~r i ~ ~ r t l e v d ~ l t 1 0 I I I ~ ~ ~ I I ~ U S C S . 111 wIlrtt COJI<ILNS 1 will use 'ift.r~l~>nstraie' 11% such a \.C.EI thdt t ~ i c iiiri dt:i~iou\tr:~tr r i r ~ t t type oCtl~i11g To de~~i(instr:ite ari apple, lri the relevarlt seme. 1s to ill~i\tr~it-. t l t e ',3{5!7ks tylw 1-,y ii1\p1~3yi11*, :$ toht~xi of t11,at type

Page 229: Recanati, François. Truth-Conditional Pragmatics

t f i ' i l N t ~ i i t Y i A ' i ' i ( ) N 2Jj

dnlnkermess, lisping, anger, slrrpl-ise, har, st~~pi(iity-, I~esitax~cy, ~ O C V C T '

(Clark ~996: 175). Nor art. the 'derrloaetr:~tr,tI properties cor~tirxed to the redm of 'shapes'. Any aspect of spcec-11 call t)e CIer~~o~istr:ttt:tj, r~ lear~ i~ lg _:as well as tbrtn. Thus tiit sentenc-e

i an be wed to report a ~ ~ ~ i t t e r a ~ c e 111 Itcd~&l, p~ov~cfec( the I t ' t f i ~ r i 5erlteuce hncl a rneamng s~ltlicret~tly 51nnlsr to that of the 1-ng1:lnh \cntencc We're going to

close exly ton~ght' (\yerL)tlr 'xiel W~lsor~ 1986'1- 228) W~LIC the clcrro~~str~~tor

rephcLtte\ here 1s only tilt rrle,m~ng of the reported uttcr,irlc e, riot the lt,llra~i wording (whlcl-, thr xdtlre\\et. woulcl perh,lp\ n o r ~~ttc-Ler\tar~d)

As t h ~ s es~rnple \how\, not ,111 tlir prol.>ertx\ ~li , l~~lt 'C~ted by the dlsp1ayc.d token are constltuti.ve of the deltion\tratetl type Marly proycrtlrs of the

cinplayect token are at CIC~CIILJI o r lrrelcv'i~~t and tllir\t be drs~rris\ed (C:l,.~rl\: &rid Grrng xc)cjo. 768-9 &inti 774 '(1 ) In (2) rile pt o1)erty ot being an hlg11'111

ientence 15 rrrclevant to the cie~~~onitr~ciul i . o ~ l l y the irtedrnng 01'tlle sen- tence matters. (711e denlct~~str,ttetl type. in tl~at ex,t~liple, t a11 be reprc5ented xs the clas, of sericer*rcs tvh~t h h,rve tile \ L i ~ t ~ z me,urlrlg ,u the f:ngI:ll\ll \entertee, wltlc-ht.vcr Lngtage they belorlg t o ) 0 3 1 th t ott~cxr har~cf, there Jsr propertie\ ~vhlc-"n 'ire corlrtrttltlve of the ticrnorr\tratcd typz but: tvlr~t-ii

are not dchl,~lly 'rnarl~kstetl', irl the seme t l i ~ t orle c-aalrtot le'nd t'ircrn ofi't11e daplaved token One re~lrzes t h t the ci~\pl'~vt.d tohen !]as thcxe ~ I - O ~ ~ I - C I ~ ~

cmly whet1 onc redue\ t h ~ t ~t is (~rrte~kdcci as) 3 ~oLe11 of the rcle\iant Type.'

'Iliis iz tiic c . ~ 111 />.~rticl~i,~r &)I. \orr~e ot the pnjpvi-tic\ lira1 aii. coi1ititutlve ot'lin,quiiiic types. Iri "Types 2nd tokens iir lingui\ti,.s', Syiv'ilrl I3rornbi.rgi.1-\&ys tirit 'tliere rs a st:t \if dm~ci~stons, rcprrscnt.rhie .a y t~es t t "~a , 1x1 whi* 1) t~c11 won1 type will finti a lo( atio~i, .t)rtf s.~cli t l ~ d l eac.11 word is fully udivrtiua~c.d bv the pos~t i~xi 11 O C C I I ~ I C S i l l tlidt spdci.' ( L ~ T C ) ~ I ~ ~ ) L ~ I ~ C I - 1989: $1). fixa~riple~ OCWC~I C / I ) C S L ~ O ~ ~ S arc.-

e How rn.iny ry1JaiAes does ttte word fi:or~t*~~n? e W h ~ t is IS unticrlgil~g represcixatioir? o Wi~a t a its sur&~ic represtiita~ic~riZ e What is the onset of its iirst svll~l~ie; e Wl~at 1:. ~ t s algurrleni stritcrure?

Similariy, 'there IS .I set of'qne,t,o~rs to whit ti e.ii11 \crltolze type h e m an .uIsc\,er. .irrd whost arraweiz frdiy uulividuatr sentence tvprs' (ibid.). i:oi rx~i;rpie:

0 How nuny words doe.$ it I - O ~ L ~ I I ?

e What is the ritat~nx verh? e VVllar 1s the 1) -struct~~rc?

What is the S- strucurrc? Wlrlch p h r ~ t c receiver what t11err1,~trc role fro111 wiiii:ti prcci~c.~tt,'

Page 230: Recanati, François. Truth-Conditional Pragmatics

Identie~ng the demorlstsated type is therefore like ident~fying the referent rn an act ofdemonrtrat~ve reference: it 1s a .Fully-fledged process of mterpre- tanon, possibly lrtvolvrng a11 assessment of the speaker's conmurucahve intexlt~onz

z 2. 7&c taqet(c) o j quofallon

T o aridyse the InterpretatIan process which underhes our ord~nary under- standing of yuotntiotls, tve need niore d u n tmo entitles-the &splayed token, and the properne5 of the displayed token to whch the speaker intends t o draw tile hearer'., ,ittentlor1 More often than not, the properties 111 question .ire demonstr.~tcd heiliuse they are properties ofsornething which one ,rttzrnpts to cleylct through the drmonstratlon. Let us call that thing the "target' In

(3) Axld the11 Greta Garb sa~ct, 'l want to be alonef'

a roken ot the sexitcnce 'I war~t to be alone' IS d~splayed. It n interided to deptct (ni~rruc, cirnrrl~te) the target of quotat~on Garbo's utterance. T l~e depli ttorl is citected tl-rrough properties shared bv the &splayed token and tile tdrgct The spe4ikcr tflerebre does three th~ngs at the sane time he iflcplayr a tokcn, ctcnionitratn certan propemes of that token (a type), and thercbv depir rs tlxe wrget

The target itself nlay be e~ther a token or a type. When the target 1s a token, I t 1s deprctcd by cJlsplayrrig another token of the same type Wherr the target 15 a type, IC werm that we have only two entrties: the &splayed token el~ld the type wki~clr it Instanhate\ But that need not be the case. The dettzorz~tratcd type n such that the d~splayed token 1s (by defimnon) a token of that type Wrrt orre may attempt to deplct a certa~n type (target) by &splayng a toketi which 1s tzot of that type. Conslder the followmg example

When a token i s cisplayed ccmm que\nons are nmned~ately answerable (by mpectltlg the token) Ottter? can bc ansuered only tfthe token o itlenut~ed ds a token of the relevant type For example, in 'Put 1s a tlrrer-letter word', the quoted word pirt' wean the number of 18 syUables on ~ t s sleeves, but a YUCShOIl concernmy ~ t s argument %trut cure can be answewd only rf the type ofthe &played token IS

rdentrfted

* Unt e %an, r l~e propwtm 111 question need riot be hnguistlc propemm, and the demonstrated type need riot be a brigulsnc type As Clark potnts out, 'manv demonstration5 comb~ne s~ghts and SOUR&, as when George denionstrates Greta Garbo's "I want to be alone" In a Swedish accent whde clutetung I s anns to isn chest in a C~arboesque pose' (Clark 1996 175)

Page 231: Recanati, François. Truth-Conditional Pragmatics

of non-hnguistsc dernonstrabon (C,lasli 1996 r 72 -4). 1 ca1 de~nonst~ate to a

&lend how niy sister Ehzabeth dnilks te,r To that efEect I tio sometlirng which resembles my sister's d n r h n g tea 1 holct an rmagnary saucer n? rn\ hand, lift it to my lips In a certarn way, ctc T h ~ t ~ ~ t d l my demonctratroz~, rrlr fixend 'has a partial experience ot &at rt lvoidd be like to see F:hzabeth herselfdruihrig ted' (Clark 1996 174) IHere tlir target of my derr~onstrat~orl undoubtedh a my stster's way ot dnnhng tea. That 1s a typo of aczlori mvolv~ng (I) a pamcula agent my slcter Ehzal?eth, (11) the two-pbce relatlon 'dnnkntg', (111) a type of beverage (tea) fillu~g tlre second arsurxrent role ot the relation, (iv) certaln gesture\, clraracteactlc of Ebzabeth wlien she dnnks tea, etc. Particulx irrlsslan~~s ot th,rt achon-type will ~rlvt)lvc spei rfic (temporal and spatid) locahons AS nsveQ m yarticiular *rnt&>ncei ot the type 01.

beverage ment~oned ~ r i (111) Titus Eli~abcth~s J r~nh~r ig a certatn cup of tea ai.

a Lertan piace at a certsin tinle will corrxit AS an irlstdnce of the art.1011-typc 'Eli~abeth dnnking tea' (ED 1 , h r short) Now wlien I cle~nonstrarc lknu Ehzabeth dnnks tea, I do not proclut e dti ~nrtllrlce of E117'. .fix the dctlon 1

produce ~nvolves riel tiler Ehzahech nr )r tecx, cilrce I ,*m the agcrit the c r r p I hold is ~ ~ n a g n s r y It hllows r h ~ r the actton-type EL)T c~ranor be the demonstrated type The dcmon\trated type, rattier, Ir I I O W E11~jlbetl-r dniiks ted. a certam patteln of bodtlv i?Ioverllents, wirlcll rs xnstnxrtratecl not only bv Elizabeth wllerl \he d t ~ r l k i ted but alio bv nivseli cvhc11 X prctend to hh a saucer to my bps m a I ertrllrl wav That act~ctli-type I\ \uperordrrlate to EDT m the sense that every llistnnce of I:U r 1s a150 ax1 rn\t,rrr~e of that

achon-type Let us call it Hcul. 711r action wtlich I procluce, durrng t i le

demonstration, 1s dso an Instant e of X hcxlcu ElEIjT call count A'S the ciernonstrated type. But the target, w h ~ t I ~ritend to tfepict, 15 another, mure speclfic type, namely EDT. And there 'ire vanants of the example r n \?r)vch the target of the demoni;tration w~ll be even more specific- no t a type, [ r a t 4

pmcular episode of tea-dnnkmg bekl~vrottr on the part of Eltzabettl I said above that there hfien' 15 2 target xliihlcb the de1r1oric;trator rntclrcls

to depict. 'Tbr Lrrlphes that, iri sonle cases at leait, no such target can be found. In 'Cut is a three-letter word', there n no txrget over arrd bevond the word-type 'cat' whsch IS denlonstratcd cil,ylay~ng a token oflt f l i ~ s rs 1x1

contrat to (3), where the speaker's s , ~ i n ~ , r*r demonstratlilg the words 'I ww,mt to be alone', is to prcture C;arbo9s srrttemnce

At this point, a caveat is In order rlrc sinlple fact that (3) 5dys somc,tbrng about a partlclllar speech eprscxie (C;~rbo\ utterance tl) is not sitBiclexrt to

Page 232: Recanati, François. Truth-Conditional Pragmatics
Page 233: Recanati, François. Truth-Conditional Pragmatics

(Srtch an ii1terpretatior.i wortlci be tbrced L I ~ O I I US if the spcakt:~ was, AS 111

Clark's exrul~pfe cited in footnote 4, ntil~tickir>g C;arlx)'s gestrrres arxl accent. This could rzot be :in irrstnncc of flat rner~tic.)r~.) So we see clearly, with this exarl-iple, that the: (internd) target ~:cecl riot be iclenticnl t o tlte utter-;kl,cr

\vct/b~ch the speaker rs reporting ctr attenrptrng to c l r a~ac t e r~~e (cxtenr,d target). Pirrt, there need not be a n ~n t e rnd hrget for the iluotat~t~n rr i onlbo

rt'ctir, even though, Irr (1rd120 rec t~ i , there ~llcvays 1s An extert~al t'rrget. \ccortd, ~f

there 1s an ~r~terndl target (sonietll~ng tlrat the cluot,ttton ~tself 15 uricfentoi,c3 as echoing), it ueed not be the uttelmc-e wl~lclr tire ipc,ihc~ lc, reporting or

atter~iptlng to charncten~e. The echoic ~11'zr;t~ter of A c1it(~tah011 I \ 0bv10us tvfie11 the \pcakc,r eugnges

ln overt rnltlvcry I have just rnen~orlecf (:l~rk's example In which '(<;c.orge

ciemonst~ates Gretd Garho'c, "1 want to be done," rrt .r \wcd~\h accertt whrle cl~ttcflrng hls anxr to hn ( he\t in a C~arboe\clile pow' (('1,irA t yon

175). In tllar exaniple. the echo]( cI-i,ir,rt ter ortlle cjtlc>tatit)rl 1s nlacie n~anrfefc\t

by the cpeakes'c gestures, but alro by the drscontruutty irltrctd~rctci by

tlie change of tone and accent 113 speahig the elrioted wort% fhe w n e sort of eftect c,in bc schreved r i l wnttcn yeec tt by ~yl,t>gr3pllrc,rl ~rlcdrls (Corntdler 1978: 79- 80). C:onsidcr. for exarriplc, the ioilowltig vnnmt

of ( 3 ) .

(3'1) A11d tlrell Garb(-, ~tlc3:

-I want to be Atlone'

f Iere the typograph~cd sepLiratlorl of the cluot,itlon t r o ~ r j the piehtory words 'Garbo said' \riggem tlut tire ipc6tkcr u t ~ r ~ i r g 'I wmt to bt. d l o n e '

i\ nllrnlchng Carbo, playrng her p x t In\of,lr ,is this IS c itr~veycd, t t le~c will be an ~nconsisfency ~t the spcLlker ~ t c u ~ c < , r,rther t11,trl d\,cr tb, tll'rt ( ;nr lw's

utterance 21 was of the clemon\tratecl type

( j a * ) ?And then Carbo chd 1rt)t u v - -I want to hz alorttJ

Tht. precxtory word\ rrow cor~rrad~ct sornethlrtg cct~~veyrd tile cluotat~orl, gven its ecllo~c chacracter. J~ t r t as the word.. '(hrt-to said', 111 (3,~). are to '10111c

extent supeduotls md s~nrply make lllorc e s p l ~ c ~ t what r\ ,tlre,idy tinplic ~t 111

the quotat1011 rtseIf yucr piece of rrunucr-y, ttie ~vords "Garbo d~ci not \a?', 111

Page 234: Recanati, François. Truth-Conditional Pragmatics

(3a*), are ~nconipatible with what the quotation itseKsuggests in virtue ofits echolc ~haracter .~

As (lornuher polxited out ul the paper from wluch these observnt~ons are borrowed (C~ornulier 1978: 85-y), and wh~ch I cited in Chapter 6, paren- thetlcal clauses such as 'he said' can be used only ifthe quotation they are appended to 1s an autonornous piece of mmicry. Thus 1n

Get oxit before I punch your nose, he said

tlie quotatlo" lr an atltonoaious piece of rmmrcry, whose r d a ~ o n to the depicted target n ~xrrphcrtly 'asserted' by the very Gct of ostensively produc- ing a plece of rtiixmcry. S i n ~ e the dep~cnve relatlon to the target IS part of the rne,irilng of the qilotatlox~ pi i t prece of mirrucry, lt cannot be denled w ~ t h o ~ i t mcoxlnstency 'l'l~us we cannot senslblv utter

*Get out behre I punch your nose, he did not

3. (2uotations as singular tenns

1 Open vs ~Eoccd quotarton

\uppore 11ry fnc~id je'm does cornething (sav, bnng us a bottle of wne) rather clumuly I n.to~k her by rn~rn~cking her clumsy behaviour. In so domg, do I refer to her behav~our? I anl reluctant to say so. 7'0 be sure, I call the attention of niy a~idierlce to that behavrour and possibly convey some- thrrig aborrt it There 1s no doubt that I thereby non-naturally mean \omett.~irig (f;nce 1957) Yet the 'mode of meanmg' at play here IS utterly d~fferent frorrz hnwwtic rneanlng. Not only are the means used to convey the nlessage nou-lirigua~c (I don't say anythng-I do something); the conveyed message itself I:, not proposittonAy articulated 1n the way it IS

(3a*) can be made acceptable by changing the interpretation. As I pointed out above, we c m ir~terpret tlie speaker a rnirnicking Carbo's ucuaf utteratice of 'I want to be done' while saying that this tinge she did not say it. That is sidgtcient to remove the inconsistency; for what is inconsistent with the proposition expressed by (3a*) is nor the echoic character of the quotation, conveyed by the typograph- leal Layout, but, more specifically, the assumption that it is Garbo's utterance u which the speaker is rnlmicki~tg. ' 1 reniember reding the same observation in sonic early piece of work by Tanya Reinhart, which I

have been unable to locate.

Page 235: Recanati, François. Truth-Conditional Pragmatics

O P E N (>LJO I A'I I O N 929

when hngulsclc conununica~on takes place lt 1s even msleadiag to u\e the s a l e term "message' in both cA\ec, @sen the heterogeneity of the re.;pecrlve modes of meanmg.

What the cornmumcator puts fonvard In the above example IS sornefhing llke a yzcture. Now picture\ n~ean ur a way quite different fro~n the way 111

which utterances mean. Uttzrancec \xv, p~ctures \how The diEerenr;e between symbohc amculation and icolirc tiisply is rlotoriously hard. to

pin down, and I do not liltend to contribute to that field of shrcly here But I want to reltnct the notlol~ ofsefcrence to the llngiatlc redm

mmicklng someone's behavrour r, rrot eo ~pso referring to that belia vrour. That 1s true even ifthe urger: of the de~~w)nstrahon 1s \omeonc'c ~perliul hehavlour. Suppose Jean sad romlethrrig uUy, and I want to rriock her hv namcking her slUy talk Adop~ng her ge4tures md manner of speech, 1 reproduce the sentence sfie'sj~lrt uttered Am I refe'emng to her utterarrcc' No more than I was rekmng to her hel.r,~vlour UI the previous exanrple I am picturing her speech and, throu$ that picture, conveying somet~rmg about it. I show how silly her speech vva5 That 1s nut the rarne thing ;:as sayztzg

that her speech was 51Uy No referelice rakes place beca~aw rcfereilcr, 2s

Aushn put it, is an ancdlanr act (Arrctm 1975 97). a ionlponcnc 0 C the

complex act of saylng soniethmg &out. 4omethii1y; -and nothing 1s sad isr this pxt~cular cze.

Of course, rekrencta ~ a 1 take place r12 i~(lditio~t tct the drrrlonstr~tron While p~ctunng Jean's hehav~our, I r r 1 q say sorneth~ng aborrt i t Or,

~rnmediately after the den~onstmnon, X rrtay utter7 'That wac ciurr~s+, wasn't it?' I-lere, undoubtecilv, I reier to tlie dctnorrstrated bellamour Ilre

demonstratlon provrdev the "muiic of preselr~;~tron' of the rekcrent, herrr c determlnec the contextu,zl "sense' ofthe 13er110n~trdtxvc. Rut i t I \ the d~11non strative, not the demonstratlon, wllrc h rel'cn 'The denionsrra~c,rr ~hows what the demonstrative refers to Were it not for the den~oi~str~hve, wllrc irr belongs to the hnpistic realm, therc worrld be no rekrerice but a mere d~splay .

The sarrle tlurrg holds. once <tgarn, ior the spec 1,&1 cJse In whlcll thr txgct. of the dernonstr<~tlori 1s hngursr~c X i the French teacher S,IY\.

This is a stipulation, not a substantive pcrirrt.. I want m corlfine 'refer' to cia of iarap~soc reference-reference by means oi'a si~~g~riar tern,. No sir~bdar cernr, rio rekrence (in that sense).

Page 236: Recanati, François. Truth-Conditional Pragmatics
Page 237: Recanati, François. Truth-Conditional Pragmatics

A quotatiou wlliill is not clostaci is (as one rrligflt expect) open.' l'trc

iol-ituast between opeti and closed quot.atiot1 is illustrated by the fc>llowing

pair of sentences:

(7) Stop thatJohn! 'Nobody likes me', 'I ;rill r~lis~raide' . . . l.)orr'r: yoti think you exagger-ate a bit?

(8) John keeps crying and saying 'Nohocty likes me'.

In (7) '1 token of 'Nobody hko ~ n c ' arad 'I a n ~~~-iser,~ble' IS C11~pL,iye11 for

de~iiomtratrve purpows, but ~t 1s riot used AS 2 C I I I ~ L I ~ J U tenn, m cootr,rst to c\it~at h,~ppens In (8), w-iiere tfte cluotaaon serves GI\ a s u ~ p l a r t a rn to cor~~plete die

sentence 'John keeps crymg and ~iyurg _-_'. Sciltertt c (7), tl~ere&tore, 15 An

m~t.lxlce ofopen yuotahon, cvhde (8) a dn irnt,irtce of c lo\t'tj. c]rrotLlhon

To sum u p follow~rlg Clark, 1 lioltl tlrdt yuotatlons u c Itngurstlt tlcrtloll

s t r~ t~ons . What the 'quotdtion ni,tsLs' c onvetrt~olialiy uldic,rtr I r t wrltlrig i s

rile f ~ c t that the entlosed niaten,rl 1s ihsplaycd for dernort\rr~t~ve purposes rather tSia13 u\cd ln tlie nornral wu)i. But ncitlier the tirsjd.rycd mAter-rnl nor thr deii-ionstrnted type (let ~ f o n e tl~c. t q c t of the denloristratto~t) I., referrcci

to, unless the quotatiorl I-tapprm to lie 'clo~cd', tlut 15, ~trllc\s ~t q111re1 the granuut~cal tur~ctton of a angular terrrl w~thrn J \critetlc c 1x1 wh1cl-1 ~t fills ,I slot. When that is the case, the cjuot'itton tr~il\iorn~ecl rrrto a .irrlguIar tern?

,rccjuue~ refere~rtd v ~ l ~ i c . Recause the dernolrstr~tioxl ac qulres a referr~ltl~l value in such cam, moct theon\ts hdve jlrrnped to the ~ o n ~ l u s i o n t h ~ t cluotatlons 111 genera1 refer to what the): p~cturc.. But that 15 not true. Orxi):

closed quotatrolls refer Opcri cp~otdt~oris merely plcture

In quotation. &ether open or c lojed, two nkodes of rrlcanlng 'ire sun~itta- neously dt p l q The *riatendl whiich I\ d~spiaycd f o ~ cfexnoilrtr,~tlve purgx)"ss h,is 3. c e ~ u l n hnplsnc medt~ing l'he de~11@n\txdtt01l ~ l f o l f d ~ inedritng. ~t

*I'lus should be qudlfted in vtew ot'the fact tlt:it a ciemotattation may be rcctuitcrl not unly ;a a sr~tgulnr ti:nn, but dso as, say, a comrnoir nout~. (1 ant grateful to Dick Carter for remitling nit. of tlus fact.) Such cases art. left aside in tlus chapter, bur if they were akctl illti> cocrsi~irrat~on olle would tmvr ck~aracteze open quotation by the lack of any form of 11ngt11stic secrwtntent.

'' Eveti though open quotatiortr are geiicrdly rchorc. ;uzti ~ r ~ t x l c a ofllnc srrcrrtti>tl ;xe ~ y p ~ ~ d l y cI(>secl, one ?liouid nor co~lhte the two distinctions (6:cirnic qU<>Llhons 11s &at rnennwa, on the orlc hand; open v$ c1ost:d quotaaorls on the other band). hs C:larEs C;& exampie $lows, clclscr i cjuotanvr1s can he ccho~c. SlrntIarl3-. tl~ough Ica obvtously, it seems that there are caxs of {tat mt:ntion tlrnt are open nth(-r titn~~ closed (.lc In

dctinruor~s sut:ft ;IS: A 61t111gi1c' is a peliud ofiirtmcttn drryr). Mow on thrs in Clwptt-r 8

Page 238: Recanati, François. Truth-Conditional Pragmatics

prcmres somethag AI; I pomted out 111 Sechon 3 r , the two fonns of meaning are utterly diEerent-tl~ey belong to dist;mct redms

In lox xed quotat~on, die slhxatlon rs even more comphcated Instead of two levels of irzeanlng (the 111ig~tlstic nxanlng of the d~qplayed token, and the picton'~1 meaning of tlze detnonstrahon) there are three The ebrrd level of mearlzrlg conies Into play when the denlonstrat~on is h ~ ~ p ~ s h c a l l y recruited and acsL1rne.i the rolc of a singu1,tr tern withzn a sentence (the 'merttlonmg senter~ce', a\ 1 wll hencefirth call ~ t ) I,ike the clrsplayed matenal, the \entente in L V I I I L ~ the denlonstrat~on till5 the positlon of a noun-phrase has a hngulrtl~ rneat~rng; and tlie dernoristratlon itseK, insof& as it 1s a conshtuent of t h ~ t .ientence, contributes to that meanmg @au singular tenn, therefore, the demonstrat~on acquires a hngulstic meaamg, distinct both from its level 2 pictonal meanirlg and from the hnguist~c meaning of

the displayed matend That Imguistlc meaning which the demonstration acyulres at level 1 II; a referential value.

Tu sort tlimg out I wggest that we dlstmwlsh the dernonstrahon itself, whch picture\ wlthout referring, and the demonstrat~on-qua-syn~ctlcdv- recnnted, wluch refers 1 wLU use the following notahon the Greek letter '6)' wlll stand ribr the c3ispIqed token, 'Dern' d stand for the demorutratton accomphshed by ii~splz~lrlg &at token; and '[DernlN,' vvlU stand for the den~onstrat~ort ijua syr~tac~caUy recrutted. The nleailng of both 8 and [Der~i]~, . 11, hgrat~t . whule the meamng of Den1 belong to the pictorial vxiety

To ~Uitstrate the three levelc, let us consitlex exanlple (3) once again:

(3) Arid then Garbo sad 3 want to be alonc'

The Enghsli sentent e 'I want to be alone', which h a J. certan memmg (level r ) , 1s &splayed for ciemonstrat~ve purposes. The demonstration ltself

" I agree &at ' d c p ~ c ~ o n st:enls to be very complex', as a referee for Mind wrote, b~tt not (or not rrecrssartlyj cl~at tr IS 'certa~~lly tior su~~ply a matter oCreseniblance'. There is no reason wily resentblance ~tselicould not he soniettnrrg coniplex and subtle. Tile referee points out that 'one mght depict the fact that sorr~cctne IS speaktny 111 bls rrative lanywzge which is &reign to one by speaking one's own language wrth a iore~gzl accent'. Brit it IS tiot ohvious to Ine that something like rcseniblance in relevant respects is not operanve In such a c:~qe. Nor an1 I convinced by Cappelen's objection that, since 'there is no iarereshng sense III wlnch writie~i langwage ptctures or resenlbles spoken language', one 'cannot appeal to sil~rilarity go exphln cl~e connection' between displays of written tokens and spoken utterances sewing

targets (persorial cot-rur~unlcatlon). Resemblances seem to nte to be involved in such cases. Be that as it may, as 1 said in $3.1, I intend ti> rely on our intuitive understanding ofthe contrast between saying and p~cturing, wirhout providing an analysis of tile contrast.

Page 239: Recanati, François. Truth-Conditional Pragmatics

O P E N t)I IC>TATIC)N 233

carries meanmg- it p~ctonaliy represents what Garbo sad (level 3) A5 it 1s

lingu~stxally recnl~ted and arsuines the Eunctlon of a ssngulx term 1.n dre rnent~onlng sentence, the dernonstra~or~ acquires a level 3 rneallrrlg qua smgular term, the derno~lbtr,itmti refers nlwh as a ndme or a dernoristrat~ve would

A fundamental property of clo5ec'i quotatron which nlust be mentioned at

this polnt is the semantrc tnertrd of the quoted rmtendl, ni~nlfected &rough several features For exarilple, it does not matter whether or not the yrloted matenal makes \ense on its own In the nietal~ng~iistic f3ranle 'John snrd " - "' I can insert a meaningless string \vithor-rt thereby rertcienug tile sentence meaningless. The inserted nratenal can even be ungamn~,~trcal -

mdeed, as (6) shows, it rieecf 11ot be l~ngu~shc rnatenjl at all." L ~ t l r e r feature wh~ch provldes evlderlce of the sernannc inertla of quoted words rc the irrelevance of the (irztm~rsrc) g,t~~lrnaticdi f i~nct~on of the c%sy1,yei3 matenal to the function of the qr*ot,ltlon wlthill the nielrtioning sentence Thus even lf what n chsplayect 15 rtsclt A sentence, ,t\ 111 yoha1 sad "kt"\ late7'', or an adjective, ar m 'Jolul \a1c3 "haXd" ', thc quotation firl?chorts as a 51ngular tenn tvlthlll the rnenhonlrrg sentroc e

Struck by that senlanhc Ineraa, rrlarly theon\& have derlrcci that the displayed words actually ouur , yraa sktrrik, m tile qxlctt;ttlurl, or that they occur with their norrnd Inednlr3gs :!,onre have trcdted tpot,ihons ar, r ~ ~ ~ r r t s whose internal cornplex~ty rrc3 serimhc s~gnificlirlce I'hc *vorch displayed vvlthsn CjuOtdhOn marks "fulfil the sa,lrrie fut'urrctton A\ the letters and complexes ofsuccessive letters 111 s111gle worrk', Tdr\ki s ~ y i , "hence thev can possess no i~ldependeiit ~liedriing' ('Iarsh 1933 rgg) O n t h ~ s ar~lvsn, the meaning of the whole cluot'at~or~ doe\ ricrt depend upon the rnedrunL% of

its parts, s lna chose parts ;Ire orll~l fi~1grncnt5 of the ndmc and \ (ourit for no niore than serifS or syllablei' (Qume r g r r 26) Others, more plausrbly, I-L,IVC rndintaned that the displayed words occ rrr m the quotation, wlnle ascnbmg them a new, autonyrrlous nleallrllg 1x1 thn context (Canlap 1937 I 56) But ~f we look at operz quotation, ~t 17 c ~ 5 v to cctnvln~c oiirselve\ that the dlsplaved words not only occur but ,dso c a m tllerr ~iorrnal rilearllrrgs In (7) rlie

seiiterlce 'I m i ~nwerahle' ohiously keeps its norind ~~leal-~lng Tlrc dr&r ence with ai or&nary use of that sertteilce 1s sin~ply that the quoter engages

' Z To use an example &om Searlc (rqhr) 76), an ornjrhoiogst nay say. 'The soulid n ~ a d r hv the

Cahfomra Jay 1s ', and what cor~~pletes the centerice 1s a sound, not a Lngttirttc expreslon

Page 240: Recanati, François. Truth-Conditional Pragmatics

isr J tcrraii ~>l'pl,b)r-;icti~~g: the cpoter sirn~~kitrs the person whose speech he is

I-cpol-tins, nn1i.h ,is a a r actor sirnulatzs tlir character whose part he is play- inrg.'.VOr iorlsitlcs ehc Ii~llowing ciiantple:

'l'lre story -teller c"1e;ired his ttrrozt and srartecl tilk-trlg. Qnce upon a time,

tEichrc- w;u a be;iumiPrl pri~lcrss nsnled Rra1x:'ila. Six loved snakes and atw;i)is hat i a io~rp ic oi'pytkrons ~rruur-rcl her.. . . '

l ' he disccxirss. ,rs ;i whole is, in past, about sn;ikcs arrtl 'kbout. a princess rialled

Ara/rc.il;i. I t is also, arid primarily, about ;I story-teller teHing a story. Indeed it m aboiat :I story-teller teUing a story about sridkrs and a princess named ArattciE,i. 'l'tlc rrrcariliirg of tllc sc-rntexlces wiririn tlte quotation marks is obviously r~:lt*v;alil: to the rlli*;*~~illg of thr whole disco-~rrse, to urhich it

urrtloi.d~ti.<iiy c.c)rrtnbritc.s.

In the: s;inic wiry iri which tlrc ilcrcltcd ~nnturiaI is sctr~ar~tically active in the cs,rmpie oC crlrrrx~ cjiioration I I2;ivc J L I ~ ~ given, it cat1 be recoplzed as scxrr;irrdr.aiiy active i t , a closeci van:rnt oftlrat cx;ur~pLe:

'X'II~~ seciu-y -tt"llt"r c-lenrecf t i i s tllscdac arid sixrti: Y Orrtre ripon ;r iinle, ttlerc was l x x ~ r i t i t i i l prirti.tbss l~a~ l fed Ar;rbeli;l, bvl-io itrveci srinkcs 2nd alw:rys had a

ctrsrpir of pyclitrns ,rrcjtlxlci her. . . '

11jc4t jraeirb ol tc.\r ~ C X F I\ dbout J sto~ytellcr leUxng ,L \tory dbolrt Pnncess A~abell't 'rran-l hvmr srx~kes. C:ledrly, the a,lca,uring ok the quotect rn~tenal 1s relc.vxr~r to rbe ri-ie'rrirrrg c,f'tl~e cvhoics I,et u\ chercforc stick to the v ~ e w that

tliercx ,ire three li-vcl\ of n~eznlng rrr ,I closed qrri>r;itlon, inlliddtrz~ the level i

nncanrrrg of tFlc \iwpl,wecl ~II~~~c"TI'I\ WWrh,~t tlic j~heiiomenon of sen1'1ntlc inertrii il.iows r \ only ~111%: 111 i lewd qtrcrc&rtron, the l a n g ~ ~ s t ~ c rneanlng of'

tirr dz\pl,ivctl nrdlienCd (level I ) rcrn,krrrs sei;.regiiteJjr~jun the I~ngint lc mealling oi tire wtiterrt c- r r i L V I I I L ~ the del~~otlstratic)~~ wrvc? ,LS .-L sillgj~1~r ten11 (level 3) Ni, Intc.gralirorr of d ~ e ii>t-nrer Into tht- i.ltter (110 "enlantic conipi>- srtrorl') tdAes phi, e 11. ~j tha t rrnrglrt vv).ucli u e xrlilrt nr>w try to spell out.

'' 'I'liu i d i i serve :is r i le basis (i)r :I general a( corrrit ot qirotatwri .IS s~mulatiun. Src Wier~bicka (~974)- I )i~crot (xvi;.r), Clark and ( kxng (199~3). Stkc11 ,an a( count grxs a Lciiig way towad expininir~g XVIIV 'I., irr *xarnples klir (71, does not iefex iu rht. i>crsiri? w h o (jut9~c.i. bur to the quoteti pt-rson--ar,ii also w11" the ~wo~~s~ouon~il ;\i<-~ntrrir d t l i e ,rcntcnt e (even iv i ih I . ~ * s ~ c c ~ to ~ l ~ e preieird i.trrltexr) IS riot senousl~ asserted. Acciinilrig rri t)errclrc Wrlsc~xi, hocvcver, a11 :u-i.i)urit in ternis of pretence Laxinot handle cases of fiat iuc:rx ion ( W I ~ S O ~ >t>oi))

Page 241: Recanati, François. Truth-Conditional Pragmatics

'To k~colint For the \rrnantli segreg,ltlon of tile d~spl.lryt-d rt~,ttcn.il, I)av~dson

i j m a ~ s l y suggested thit that maten~l 15 not part ot the rncotlon1tlg sentent c

at ,ili, serlidnt~caliy ~t lrcs outside the n~erttictn~ng smterrc e, wh~ch contams a

denio~ls t sa t~v~ (the quotxtlon ma%\) letl-nrr~g to tlre dlsplnyed materul or

sotni. typc \?;filLh 11 lnStdlltlSte5 ( I>~L . I~SOI I 1979) I his ,4pprodcfl t o quotatlo11

cva\ flrst tncrrt~orleci (\lvitb q p r o v ~ f ) by RrtL1til Pnor

Soxrke . . . would s a y that t i le quotatior]-marks arc cl'c.tnctns~r~~rir~c.r which poirtt to tticir interior, so tilat ""'l'lre cat s'it 0x1 the in:itM has nineteen letters' is r;irl-rer like ''['hc cat sat c i ~ t tlre xitat. ---- rl'ili\ I r ~ s *sninetc>er~ Ieitcrs'. I i11c line to this view 111yselJI : a i d

ctrtaixlly ii'it is the correct view it is easy to cInssifV the illusion iilvolvecl in tre;ttirlg ' . ', l'ite cat sat o n rllr nlat" has r~illrtren Iettthrs' 01- ' "Tlre tat sat or1 t i le m,xt" was

trttered by fohn' ns c-ornpourlcI sentences wit11 "rhc cat h a t on the rn:tt' as a

corqxxnenr. ?*l~is is siimply the illt~sion of seeing two senteric:es as one, bec-ause

they happen to ctarld i l l an intrrt~sting rclatiorl to ctne arrother. ("rinr 197r: ( i t . r ~ )

O n t1:lj cicw, tllcrc 15 crx11,rrrtlc mertid oitly to the extent tlrat, strlctlv

rpc~bla~g, the nklta'irilng of the quoted ~n,iten.tl rs not n g ~ s t of t t ~ c I I ~ P S I ~ I I I ~

of the sentence rrl mlirch the r r t a t c r ~ ~ l 1s qliotctl. But the tittrourse coilt~txls

more than t iw t scritence t t alto corttaln tire qnott~f nl,~tclid F-Ience the

quoted r?lLiter~.d n \~"rr l~int~~diiy lizert only in A rt.,iurtvf, sertsc It Icnldrrts, or

can rc111~1112, rcmdntlc~liy act~ve 'it the sepqu-<1te level to whlcll rt belollp. In

the 'hove esttrnplc (Arabelln and 1ic1 cn~kc) the quoted nratrnal I\ scmantl-

ta l ly ra~ert ti1 h e senleurc In wllich it.ie>rr>t~cc 1 % m a J ~ to that muterirrl-~t 15 not

s e ~ ~ i . ~ r ~ t ~ c ~ U y part oftheat \cnterlce, bnt rxint be wen ~ . i ly~xig outsrde r t Yet it

1s \em,mtrcall? actlve ln the chscourse a> 't wflole

i ~holehe'i~tecily criciorre the negmve patt of IJ~v1d5on's ar,*ilysl\. in

~loseci tltlotanon, the &\phyed matend 15 not scnl.lrztlt ally part of the

nlentionlng scnter~t e, ~t Ir dtcplrryed alongslde the serrtcnce nrld reierred to

by so~rretlnxlg (d \irlgular tcrrn) i l l that sentence Ac cordir~g to me, ~ C ~ C V C V C ~ T .

the srrrgtil~r term ln the ntentlonlng ,:errtence cannot bc tlre pair of cprota

tron m,at-l\s

If the qtiot;ction illarks were x stngular terxlt, AS lI)av~d\on cl,l~rn\, tltcrt,

gxven the nsstt~npt~ori t l ~ r t they 'ttrrtctron In tlte saxrle way and Etave the \arrle

se~naxlci~ vahle whatcvcr I~ i lgu~s t~c colitext they occ11r 111' (Gappclcn ~ I I C ~

I epure 1997 $3-c), tllrle wottlct be x dar;qbr(e slqul ln tcnfl Irr ,211 Irlstance5 of

Page 242: Recanati, François. Truth-Conditional Pragmatics

open yuotahon- a urigular term wthout a sentence fiarne in which to fit. Thus (7) wo~lld consist of four sentences ("Stop that, Johxz', 'Nobody hkes rne', 'I ~rn rrnserabte', 'Don't you &I& you exaggerate a b~t?') and two d~nghng s~ngular tenris (the quotation r m k s around 'Nobody hkes me' arid 'I arnx rnlserable')). To nuke sense ofthose singular terns, one could perhaps argue that (7) ts acttrdiy ellipt-rcal for somethlng hke

(7") Stop that John! 1 b i ~ say "Nobody ltke, me', ' I aril mserable' . . . Don't you thr lk you exaggerate a bit?

In (7")- indeed, there is a suitable hame, namely 'yo11 cay-', where the alleged \irlg~tlar terns cam fit But I denv that (7) and (7*) are synonymous. Nor are there any grounds for postulattng ellipsrc here except the desire to cdve the throrv in ttte f ~ c e of obvious counterexan~ples.

An acfdit~onal d~Uitulty arises m exmple5 of open quotat~on 111 wh~ch there rs a gentiule singular temi over and above the demonstratron. If Dav~dson were nght about quotatroil narks, the senteiice '(:ontment aliez vous' would be referred to tznce in example (4), once by rneans of the quotatior1 m~rkc (d danglmg angular tenn), another time by means of the dernon\tr~tlve 'tirat' Evidently, it rn~kcs rriore sense to say that the sentence

fint ci~\l\pl,iyrd, then referred to 7 he problen~ of ihe cianglu~g srngular term become\ p~mcularly embar-

ra\smg In corlrlcctloli w t h the plie~lomenon of 'rmxed quotahon', to be coit,~dered dt ler~gth below. l>avldson cla~rris that in ( I )

(1) Quinc savs that qucttatiou " . . . h:a a certain ailctmdous feature'

the words 'har a cemin arionralous feature' are quoted ,it the same nme as they are uwcl 13ut rf-there were a s ~ n ~ l a r term refernng to those words m

( I ) , the sentence would be as ung~mrrxatical a$

Qu111e \ay5 that quotanon these wordb has a certain anomalous feature.

(See I)avl&on 1979 81; Searle 1983: 183; Czappelen and Lepore 1907. 437-8).

Thrs shotvs that the quotatahon inark5 do not function AS a ~ngular tenn 111 (I) j4

" To be sure, chicre 1s a closed variant of (1) in which ttlr words me explicitly rehrred to, namely: 'Qrunr rays tirat yuotaaoa, rn his words, tws a cenaln anontalous f'ture'. 0112 may argue that (1) is to be undrrrtood on this pattern: the quotation marks in (I) furlction just as the description 'his words' in the above vxkatlr. I will discuss that view below ($4.1).

Page 243: Recanati, François. Truth-Conditional Pragmatics

St~fl, we Lan retam the ecsentids c~f D,w~dson\ theory, wlizle rcjectrng XIIS treatment of the quotahctrl rnarks as a s ~ r ~ g d ~ r ten11 The quotdtloxl x-narkr, merely ~nlllcate that the quoteci ~vords are belrzg dernoii~tr~~ted It rs, X suggest, the dentoiatral~on ~t\el i \lo1.11c l r assunres tire firn~trori of s~ng~i lx r term, in closed quoQtlons l5 The muvcal es~nlple (6) pmvrde5 evrdenc e thdt even a nord~r~guist~c demonstrrrtion can play the role ofa sutgalar tcrnr Thus we can agree that, in closed quoratlon,

(1) the quoted nlatenal is d~splaved or presented for cten~nnstrat~ve purposes, as ln operl quotdtrora,

(u) the demonstrat~on assllnlr.; a gra11111idtical filn~tlofi 111 the sentence that of a s inG~lar trrln rcferrrng to the demonstrated type,

(m) the quoted rnatcnal itself-, dntmct 6ont tlie prr"sertt(ltzon of th,rt rrlate-

rlal (the den~on\trat~ori), rs not scm,~rltlcnuy a part of the c:,entcltie nn

wluch it IS presented

A sentence bke 'Garbo s a d "'I warit to be alone"" c'in tIiereG>re be malyred, i Id D A V I ~ S O I ~ , AS

Carbo sald \ D e ~ n l ~ ~ , . I warlt to he alone

where the second sentence corrcsporlii.i to tfit ~f lspl~~ycd ~ I J E C ~ L L ~ , c~hrle the firrt sentence contalns [I )rntlNi, the pr erentatiorr of that matermi, iervllig =IS

s ~ n p l a r term and referring t o \orl-xe typc wlllcb the matendl 111 qqiiiAstion in~tantlates.'~

4. Mixed yrxotation

4.1. Clpen or closed?

By 'mixed quotation' Cappelell :~nd I,epore rxxean a tnixntrc: oforatio obliqua aiid oi.di;o ialt'i, c;iji;litsi"iztd by tjie ii'5e ,.,f quOtjtioil ili,ii-+s iii ;kte i ~ a j

complement of an indirect.--speech constructio~l. They use L>avjdson\ sex-

ample (I), repeated below, ;is a pxriidign.

l5 1 owe this point to Berroit cir Cornahel l6 1 assume that the %dying rekd~on hoids betweerr a prn-son and a Lype just in 1.3s~ person in

question utters a token of the uype.

Page 244: Recanati, François. Truth-Conditional Pragmatics

'l'iirr first thing t o notice i s r h r the ilispl;ryeii ral,~teri:li is not sentantically

1lrcr.i Iture. 'The words '1.1;is :i certain arzorir;rlotis fc;ituri2 work 3s. :i pr~t.liir-rrfc, in

(I) tircy kcep rixeir i~cisrrr;ii sex~l:lrrtic firrrctiox~ within thi: ruerltioning ieritc.rii.ib. I ' X i i x srik;gesr-s tlr;tt rrrixcti ilirtrt~3tiorl is riot an insta~tcc o f closed

~iuot~ition. In cioceti c~uot:rtion, the ctisplnyc:d rrratcri:~l i s scgregntrcl kern the

ra~cliticj,mi~ig sciitcxic.tr. l'lie nicntiorrirrg scrrrcrrctL orily coatnirls the ~ C I I ~ O X I -

S C ~ L I L ~ O P I , rc(:rii~t~"d asb 3 si1igu1;lr ~cr111.

Still, i)avi~Ison ;ind lris fidlirwers rrl:tiarrai~l that the cluotcd rrraterial is rt$xm'd to. ? 111s rises tile problcnr of tlrc il;~llglirlg sit~g~il:~r term, 2s we have

sccx~. Aftcr tlrc cvor-ds 'QCne says clr;rr ilriot:tric.irr' we (lo not expect, and

tr: i~irict t nrcor.r~xr:i>cEarc, a hi11gul;is tcrrli. WJr;it \via cupect is 'i pcdicntc-and

irldrctl wc i i ~ l c i trrle, sillci: the dispi;avcil 1rlntcri~i.1 p1;u;rs its ~sol-tnal ser~lalrtric

role. I 'hc :~llcgcd \irigul:lr terxli i s lor] extra cor~stitiserit ~vliictr does not fit

a r iy~vi~cr i ' 111 drc sclriterrc-c. What. i.ar-i wt: Jt, wr~ll ~ t l

?'o ;~cc-i-rr~imlicst.Eaie the allt~gccl sisigirl;ir tiSrlil, wi~c;lpl s:~y t k r r tllc mixed- i l t ~ ~ ~ u g suritcricc 1s clli[~lictrl Jrr a It-inycr scildenci' wl~crc it (the Jangling ~ingukir ierrii) c-,iri i i t . The sentericc in qr r r s r io~ i rruiit corlt;iir~ ;I rrletalirlguis-

tic. prc.iiicdtc, srric:tb ehc. ;rlleged sixrgrriar tru-rrr r t k s to \.vonls: tile sing~:txlar

terns ~ 1 1 1 thus hc c-orrstr~~ecI :is prclvidirig t o r w cd'thtr nrg,?ri-rierlts of the elided

1m"*iiia.:rtc. I >;rvriisi~rr l~inr~seifstrggcscs that (1) carr bi, ~ri,ttlr nlclre esp1ic:it as

Usiarg those very words [or, ;is i>;ividsorr prlts it: 'using words of which rll i5 i; ,i tokcai'], (&rirje says tlr'it ijuoc~ritrr~ 112s :i ci.rtaiirl ,ini~rii,iIotrs katxirc.

l'tuc cieirionsrx:itivc "hose very words' i s v sir~guinr tcrrrr refcrritrg to rlre wi)r-:Is '1-i:rs ' i ccr-i,Irrr '~rlonl:tlok~s feature'. It piayc rlrii s;.irxlc rcrie as ( a r d is nlorc or lcbss syrioaiyrutirls wirt-1) tlic ~ L L O L ; ~ L ~ O I ~ rlrarks :trcirrird those ivorcis, ir, i 3:iviciscirl's :~rialysis. If wc rcp1:rc.c i t by rhc tlcr~ic\rrstr~ciorr pir syrlt;lctic:i11y

re<.%-~~i t t A , wt.: d~:ill :in;dyse ( I ) :ah

where tlic" verb i n ix)Xdiice has been eliciecl.

C l r z this ;rri;ilyrxs, tlre speiker of (1) says two rirarrgs at the sarrle time: (i)

~ll;il Quilr(x~3y~ ihal qllcftati~rl h;ls c~rt:Gn ~ P I ( I I I L , ~ I O I I S fc?alllre; ~ ~ n t i (ii) that Qriirie vays str risirlg the words %;is a c:est;~irr a~rorrr;doi~s ikature'. In contrast

t t r the first itatexnciit, chc seconcl c>ne is c1liptic;rl: thc spcakcr refers to Quirle

Page 245: Recanati, François. Truth-Conditional Pragmatics

,nlci to the words 'h,i\ .t cermrl .irtoinalous h'tture', hut the rr~ctdllitg~~lstsc prcdlcate rerrr,urrs ~ ~ ~ p l n c r t The &\played rt~litenal lias 1t5 norrn'd serrldntrc

fuunc t ~ o n In the e x p l ~ ~ t \t,itemenr, the ricrrzon\trat~ort \crvt\ .) .~~rlgul,lr terlr~ tn the el11pt~c~~I \tliteilzelst

An dternative i i i ~ d y ~ s 111 the \ ~ I C (pint cfispen\es wltll tlie pos~~ildt~on of

Jn clrded predrc'ite, wiule maintalnlng that there are two overldpp~rtg \tatements, OIIC superimposed or1 the otller The 11.~et~~l1ng111\tr~ pledicate, ~t can be argued, 15 drexdy articulated in the seiltence. tt 1s the verb '~;rys' In (I) 'says' ~ A e s tzuo iufzthvict dttrct nf.ilccts trt the carvrc tirrtcp. the 'th3t7-clduse ~ i x j

the quotatlor] eac Lh prol/~de\ one ( I ) I\ therefore to he atla1y\etl as

that cluotdtlon I~,ls li c e r t a~ l d~ io r r l do~ \ fi"~t~11 t- Qrrrne says

Illrin JNtl

tvlicre the \ i r~gt~Id~ ten21 ( n ~ l t i ] ~ ~ > refer\ to the word\ 'has a ( crt,ur? ariolrl,r;

lous fefe,lture7 (Cdppien ,ulcl Lepore 1997: 447 fn). I find hot11 v,mnnls of the '\~~pennlpovhon' ,~rldys>r corlvc,hrtt.d drld pcltit-

tou us. The fonx ofcomnpcl\~t~on they appe,al to rs, to 111y liriowic~cige, u~ihcnrd ot-. The only lnotlVdh(3rl fbr offbnrrg baoque 'rccounts lrke these I\ ille tlcurc to \ave 3 dogma the mew th'tt yuotdtic,ns refer. Whert we t e a l ~ ~ c dt,it iluatatlom

car1 be upcn a \veil a\ closed, Iiowever, we rzo longer 11,rvc to worry ,tboitt the dat1dmg \~ngrlar tenn ,tnd bow i t fits In the sentence T11cre r t o lorlger 1s '1

r i r i ghg s~ngdar tenn, bzc.~ii~c. there no lo~lger 15 '1 \~rrguldr tern]. Oii the vlem I ;lcfv~cate, ~ I I Y C ~ I I CILIO~'I~IOII 15 (orrectly t i e ~ c ~ ~ t ) e d '21,

iofto-cvs: Tile very tvords wliirll nre uscd to expre\\ the corttcnt of tlzr

reported 'ttt~tslde ( o r cpeec l.1 act) are 'it the s d ~ r ~ e ttrlsc d1sp1~~y~c.i for ~ C I I I O I I

str'ttive pulpotec, brrt they .ire not iehrred to by rne'ux of A \lngll,lr tcrili. As f po~ntcd out In (:lr,~ptcr 6 , t l ~ c sltu,ltior, 15 s~nirl,tr to wlut wcvc ftrjci 111 the followrilg exdnlples, ~ l i c r e , in tlic Lourw of report~ng t11e Gi\cnhee'r \pc.ccll dct, the speaker riumsc\ him or her by phr~srllg arlit/or prctnourtt I I I ~ the cornplernent sentence In 3 certain way:

(9) 'To which nifr Bdev nrocicstly replied that Ilc hopeci be k n o w 4 wot o'clock it wos in gined. (l>ickerrs, :VI~zrtin C:kttrzzl~-urit, citcd 111 (:kirk and

Gerrig 1990: 791)

(10) C,ke vieil le,f imnfi~. . . vint ill6 seui l et nic d~tnurtdL2 qu6 quc jlvc.trdrus, d'tine voix

tuuiriat? tc el hatqrzensc. (kku-bey d'hurevilly. L,'Bzsouielt!~~)

Page 246: Recanati, François. Truth-Conditional Pragmatics

At the same time as he reports the ascnbee's talk, using indlrect speech m the nonnal way, the qpeaker shozi)~ wl~at that wlk w a llke.I7 'The speaker nught s~rmlarly gesture In a certam wav to m n u c the asenbee'., own gestures. In such caces linguittlc and icoriic meaning run In parallel somethmg IS $aid md, s~rnultaneousfy, solneth~ng 1s shown, concerning the tarne top~c. Yet there 1% no conversion of the plctonal into the linguntic, rio syntactic recrultnlcnt of the con ;15 a constituent in the sentence. In particular, the dernorictratioil does I IC)~ become a smgular term in the sentence (ar it does in closed quotzltiorl)

4.2. 'Wixed guot~ztior-1 (2s hybrid use

When Davtdson fmt rntroduced example. (I), he presented ~t as a 'mxed case

of use asiti rne11ao11~, that is, ; i ~ a hybnd case in wtuch the menaoned words are, at the smie tune, m active use (L>amdson 1979: 81). Now there are many

cases of that sort whch do not ~nvolve orutra obllqtka at A, for example

(11) Joliri a very 'cool' (12) 'The deinonstraaon prov~zfcs the 'mode of pre~entahun' of the referent.

hence dcternunes tlie contextual 'sense' of the demonstrative. (From Sec-

tlon j I hove)

In wckt cdscs, typic ally, one trscs words to say somethrng while at the rarne tinie rrkostzg sonle otllcr person's use of the came words '' Even though the words 111 quotat-on marks are u\ed in part demonstrahvely, to deplct some- one's urage, they d1w do their normd semantic work in the sentence. That characten~atiora doe, not take us very far because the quoted words gener- agy do their normal semallbc work (and, I would argue, are used to tay

something) 111 irntances of open quotat~on, such as (7) But the chxacten- ration can be rnadt. rnoxc preclse What disnnguishes hyblnd cascc hke ( r r ) and (12) &on1 other exanlples ofopen quotation llke (7) is the fact that m (7), thc derz~omtratec! ~vorcls are uttered for the sole purpose of the de~nonstra- tiori the speaker arguably says s o m e t h g by using them, but what 1s thereby

" See Stanton 1999: 273 -4 for a similar analysis ofmixed quotarion. Stainton and I agree that 'mixed quotation is cquivaient to indirect quotation-give or take some nkkcry ' (Stainton 19yr): 275). I Not aJI hybrid cases are echoic, though. The exanzple I gave in footnote ro (the dei-kition:

I A 'fonrdght' is a period of fourteen days) rs hybrid since the word 'fortnight' is both nleririoned and

S used, but it is not echoic because the derr~onstration lackr a carget. (Or so 1 d~ought; see Chapter 8 for

i q~lalitications.)

Page 247: Recanati, François. Truth-Conditional Pragmatics

O P E N Q U O ' 1 ' A T I O N 241

said is said as part of the act of demor~st~atrorx. When the speaker of (7) rays '1 am mserable', b?is locutlonary act rc raleaxlt as A repllca of the delr~or~strdted speech and has no lndepetldent xnt)tivatron 13ut when the sye'lker s&~yc jxx), he performs an independent locutmnary x t , to wllich the dernonstr~tcd words themselves contribute That ,let 1s not subservient to ttlc act ot denlonstratlve slmulatlorl wlncl~ thc speaker also pcrforrtrc t l ~ r ldtter rurlr In parallel to the act of sxy111g that John rs cool. It 1 j XI that sense that the quoted words can be said to be ured s~n~~~ltntteoiisly ~n 9,1y11ig soneth~xig and performlng a demonstration (That the locutrvnarv act is not ruhordmateii to the demonstration is shown by the tact th,.tt the Jemonstratlon takes 'narrow scope', as ~t were; rt is local and c o i ~ ~ e r n ~ only a lirnlted por~ioxs. irf the sentelice by means of whrch the locutlortarv act 1s pcrk)m~ecl Irl

Chapter 8, using Potts's terrn~r-tology, wch c , w s wlll hr reterred to .I\ instances of sub-daustzl open yuot'ltion j

By draw~ng the hearer'? attentlorr to the words he or she mes, the speaker typically suggests thdt those words are to be ascribed to some otl~er prrson (or group of persons): the 'internal target' of tlrc quotation But thc rug$gcs

h o ~ l in yuestlon d o e not become p ~ r t of the proposition exprc\,c.ti bv the words. In (12) the expressiotl*, "rrlode of preseritntion' ,mtt 'sense' are 1114

played, suggcstirig that thev are used r r l a x cctloic nrat112cr. A noritla1 ar~alvtii philosopher readuig sentence (12) rrl thri; article ~~n~l~edla te lv urrderstariid., the author as echolng Frege, ~;lr-rci as expectu~g hrnl or her to tahe the d~splayed words m the sense tliev have m the Fregean Irteraturr I b ~ t rnetahnguistic indication may help dctcmlule wlticll propo\it~or~ i s ex pressed, but it 1s not Itself part 01. tha t p~op~)s~tioxt 'The propt~iztror~ ex

pressed by (12) 1s slmply t11c propoatlon that- the denlolistration prov1de.i the mode ofprecentation of the ret^c.relzt, hence detemunes the coxltextrxlrl 5ense of the demonstrat~ve Stnlilarly ro ( K T ) the speaker echoes ,I ccrtalrl

group of people by uslrig the word 'cool' w111ch (we m,ky assume) helonp to

thelr vocabulary. He ostensively 5peah hke them. but he does n o t IAV that he speaks l ~ k e them.

Mmed quotation is only a part~cular case of that phenor~~enon of echolc use. The speaker uses certnln words r o ex1)ressing the content of the dttitudc or speech act he or she is reporting, wh~le at the same tune drawng the hearer's attention to then1 for demonstrat~ve purposes. On the tuost riatural lnterpretatlon of the dernonc~abon, the &splayed words Are ~xxlplrcicly

Page 248: Recanati, François. Truth-Conditional Pragmatics

24.2, biiXF,i> ~ ) l l i l T A l ? C 7 1 ' 3

,iwribeii t o the vcry perscm whirse attinxdr or speech ac:r is reporteci.'" Yet i.lie q)t-".iAn;r docs not .wy rlt~at tfre ascnbce useti rlaese wcrr~is. I ie merely sho~,us (iie~rl6rriC,~r,lti"*r) I i i e W O ~ I I S the :iscr&ec ~iscd. S'11is is like csarnple (9): the re.:iilcss risrtii.r\t,ind I fickerrs AS 1111nlickii~g B:t i j~y 's IPI : I I~~CI ;S of speech, but n r o \ r i k t ~ * o t ~ rs i t said rlwr 1I;aiiey sye~~lis in t h i s way.

I u t h l c i+;r rxrcw ork, it is siulply r i o t true &;re tl-rc proposition expressed Ity r l~e cor~iplerricx-ct si~iterice in ( J ) is "alitrtit wcurtl.;', as C:;~ppelen and Leyore ci~riiar (witlici~rt :ul?;rrrllentj. C'::rppclen arrti 1,eyorc use this trnsupported claim t o rirrdel-iriiire ,iii ti.ie stlir~dar(l ' ~ ~ ~ ( I L I T ~ ~ s of iildircct speech, based on the

k)Llci\vir>g prinirpiu:

(A) A pnq~'(-)""Ci'~~i"l 'ittit~idt. r-cpcirt is trr.~c.jusr in c-;~.;c :-in ngcnt s~mik irl ;i cerwin rc~l,iiic~ai, c.g. the s:ryiitg rcicttiorr, t o the coirteart iifthc co111l3lernent clause.

(C::~j)pelc~l ar~il 1,epore 1997: 435)

i r l ( IS )> L; :,ippc!cxi, arid L epoxv 'irguc .1.33), the cor~lple~nerrt clarrse contauns ~ ~ L I O I ; ~ ~ I O I I I I I L I ~ ~ S i"r1~1 is chcrek>rc ; ~ i > o ~ ~ t w-or&:

t I 3 ) A l ~ r \,31(1 t l~ :~ t hk 1% '~!iL'fic~ilt to L I I K ~ ~ I - s ~ : I ~ ~ '

Yut Aiic-i! did rlor say 'inytl-iirrg ai)out worth, lrencr she does not stand in the wyirig rci,ltioai to the propositit>r~ cxprt:sscil by the conipler~~erlt clause in (13). I-his is s~rpxwcd to slionr tti;it tllc st:iiid;lrd , ~ C C O L I ~ T S ~ E i ~ ~ d i r e c t speech

l-xiscil oni pnnciplc (A) art. ;riP ri~istaken. 0 x 1 the prcsiant proposal, )nowever, 1 8 7 ~ propo5iii011 cxprersetl by t-I-lc corr~pieri~errt sentence is the sarne wit11 or wiil-rcxri the. ila~trtatiorl rr~:rrks, ~ r r t i it is not :rhorrt worcls. T'he deirlonstr-ation (.of~i:eys ,III : ) ( k i i t ~ o ~ x ~ l r n e n ~ ~ i ~ ~ g , 1 3 ~ ~ t i~~r t p~ctoriid ~~~ t - ' i t ~ i r l g I-UIIS 1x1 p:ir;dIeI to

r i i ~ . p~-o,i~oulcio"n \vhit l i rs ii~lg~~istlc:~lly .ir~Erxrlatc.cl: i t is i ~ c t t ;1 part of i t . Mixecl cpo"ritba>ii ti-ribrcfi,rc does not c'o~rstitriee 'i i:trrrrrtcr-exalliple to Principle (A).""

At tlrxs point "ZCippel~ x t r i tepor-c car) JrgLrc that, surely, the li~lguistic nlrbairing tri-(13) is rrot tfrc saltlc :is tIl;st of'the scrrtence w e get when we drop dici cjiiot31:iorr rrr.irRs, vir. ( 1 4 ) .

''7 i i i rs r., oiiiv t i i c iiroii iiac~lrll ~ntr~iprct.itiuri. lro\vn;ever. Orrc r ,ui c.&tlv uilaglnc J c o ~ ~ t r x t in which a \ t . i l i c i~cc. c*.it i iy iikt. (i) VI'OLIICI he usel\ with wiiic.tiuiig oiibcr c11,ui tlic dst.rih~r's utterdncce :rs lnterrwi IAIget fill the- f~~' l~1~)1i \ t r~~lO6i 5t.c ~ < c ~ ~ d ~ l ~ + t l f A ) l > l > / > : 7.i4) ioi drl exA~1lpk:

'() S<Y \~ ' i111tOl l (l()O<j 2;! 3 ) fix \ l l l ~ ~ l ~ ~ ~ l-?rrl'irL:,

Page 249: Recanati, François. Truth-Conditional Pragmatics

( I 4) Abt e 1.1sd tkit [Lfe 15 dlfErult to urtde15t,md

Tile quot'rtron rrt'lrks are p x t of the lariguage They ,\rcx Itr~grrrscrc \ymbol, e r~do~vcd with C C ~ I I L C ~ ~ I O I I ' ~ S I ~ S Z ~ ~ C ~ K I C ~ W e c'trrrlot tortsrcicr t17~'1)1 J\

depnved of m y lulguistrc rrwarxxlg wliat\oevcr anc3 't\ opcr;lt~vc only rtl

the p~c t orla1 realm Iri reply, I should fir\t p a n t out that ~t Ir n o t obviotxr t l ~ x r tile cjuotatlon

xrlarhs arc p x t of 'the language' The cluotahoit ~n~t rL\ arc ,t puntfltiltioti s tg i l ,

and a\ ~ L I L ~ beloi~g to the ;cuto~ronious systein of zr~ntfrri Lr-rgu~ge (Nunberg 1090) 111 poke^^ I~ r~g i~age , there 'tri. nu qutitdtron m,lrhs proper the spedker 511nply draws the hearer'\ ,ittentloI1, by wh,itever ine3114, to the ~vor(i\ be or rhe utter\ " Be th,kt a\ i t 112ay, ler 115 asurnc thdt f ndi\l~ contarir cjtiotatloxl nl.irk\, as wntten f:nglnh clexrly doe\. Tlicn, ndlnlttedly, ( I 3) ~irci (r.y) 'Ire rlot the same 5entence (13) cont'tins a Lirlguistlc sign (thc cluc,tltiorr niark\) w h ~ h (14) doe\ not ( orrtarrl As a result, ( I j) ,rrici ir t ) (30 riclt Ir,ivc the sanlc Iin'gu~\t~c rnemrng What c ~ n we szy about th,rt tlrfieretlce?

I think the dlEerc~nce between ( r j) .rrlcf ("4) 1fiilmrl~~r to chat between

(I j) Flz 15 17ch and stup~d

(16) I-ie is 1%-h but stupid

l%oth utterances cxpress the proposthon that the yer.;ttn 111 ciilc\ttorl 1s hot11 n~11 aid \tuprd, btit (16) cotivrys '1 frtrtller indrc,ttlor~ that t?i\ it~1~71dity 15

ur-ir~pe~tw3 glvert lsrc nchi - tc~~ *f 1115 ~ndicahoil 1s t onkeyed r r l vli-tue of- the cc~nd~tiorzi of w e of the word 'but' RccortLng to O\w,tlct I)iicrot at?cl Iln

tcfiool, 'but' 15 to be used to t oryorn two ct;itcrrlet,t\ 1' '>rid C,) ortly rfthcrc 1s 'in c ~ ~ g u r n e n t ~ i ie eoritmct of rllc (i)llowi~~g sort hctlvccri tlrc'oi

1' provides evidencc irk &our of, i.e. 'snpports', a certaix) concl~tsion e

'' 7'hc tlispiay can be ut~deniood In vanoits ways-- not rrit-e~s.it-11y i s a qclitt.uiorr. (:i>ri-$ii$t.r h r esarnpic the kollowi~tg utterarlcr (where block iectcn ~rliL(:ate ttrac the words arc dispixyecl ul spciker~ speech):

(I) Tint said ibat John is HIC;HI Y datigerous

Ifre ikrplay of 'lngl~ly' call t)e u~ltit:rstood ill vnsloir\ wdys: cxpre~si~rg exnphws r.tr- A> ~rtd~catlitg '1

drrnolutr:itive lnkcrttion on rtie p:~st of'the rpzakrr. (Qrlot~txon nwrh i i r wxlticr] speecll c.1ri hcar- o~l ly tltc \i.(-onci iritrrprctant)n.) Gor~traitive t i> ius i s another ponlhli: irticrliret.ttion.

Page 250: Recanati, François. Truth-Conditional Pragmatics

- - - - - -

Q suppow the oppogite conclusion not-? Q IS stronger tharl p.

In virtue of t:hose condztions of use, the speaker saying 'P but Q' indicates that there is a cooclilsion r such that the first conjunct supports it whiie the second cor?juilct: provicfes a stronger argunzent in favour of its negation (see e.6 Anscomnl~re and Dtrcrot 1977). This indication comes in addtion to the conjunctive proposition expressed by the uttermce. The utterance there- fore means two things:

(I) that I? 8c (2 (11) that there is a conclusion r such that the first conjunct supports it

wtilte the 5ccond conjurrct prowdes a stronger arguinent In hvour of its t-re&nbon

Irt d sense, then, the utterance expresses not one, but two proposiaons (Bach 1999; Neale rpyy). In the case of (16) the two proposibons are

(1) F fe 11; both nrh and itupld (11) 'There Ir A ~ ~ n c l u ~ l o r n r (e.g. 'John is mtelligent') such that h ~ s being

nch support\ r to some degree, whlle his bcirig stupld refutes ip

'rl-rese two propositions are not on the same level, however. Only (i) is com~>ositiondly articulated. 'The coniplex met&iiguistic proposition (ii) is cxpressed holistically, in virtue of a pragmatic mechanism. By using 'but', one irnplies that the conditions of use of that expression are satisfied, hence one irnplies (ii). 'r'hat is a pragmatic implication, or 'imnplicature', to use Grice's cover tenn. Since the iniplicature arises rather directly from the coriventiofts governirlg the use of a pxticular expression, Grice called it a conveniiolral implicature, in order to disthguish it both Gom the conzposi- tionally articulated content of the utterance, and Gom the conversationul

i~nplicatures which :ire not directly tied to the conventional significance of words.

Quotation rnarks too have conditions of use: one should use quotation marks only if one is using the quoted words demonstratively. Using quota- tion rnarks therefore indicates that one is deri~onstrating: that is the hguistic meaning of quotation rnarks, \vIGcl.l differentiates (13) from (14). Sentence ( I 3) tl~ereforo means two things:

Page 251: Recanati, François. Truth-Conditional Pragmatics

--. O P E N QLJOTAl'lON 245

(I) that Alice said that hfe 11s dl-flic~ilt to underctaid (u) that the words 2iEicult to understand' are belng used itemot~\tra-

tlvely

XVhlle the proposition 111 (1) is con~poslt".onally amculated. the proposltlon in (11) is not. It 1s expressed hohstlcdiy and lras the c ~ t u s of ~t yragrndnc unplication By usrng quotation niarh, the \peaker inrplles that rlle ccsncfx- non governing their use obtains Since the cctridibun in yire5txon 15 t~xed by the conventions of the language, tbi4 sort of irllpllcature deserves to be called a conveu~tzonul mlphcature 7 hough rotiven~oxlal, it 1s not past of tire (composinondy artrtcixlated) prol>os~tional content: of the rltteraiicr The proposlhonal content of (13) n the rarne as that of(lq), ever1 r f (13) 2nd (14)

do not have the same hngulstlc rneanrrlg

5 . Interpreting cluotatioxls: the pragxratic view

j. I . Three levels o f meat~zn<q qqurrz (b1.46 not the santpi

Quotdtion marks tun1 out to belong to the clas\ of prqgniatlc rrrdrcatnn

expression\ which liave c e r t ~ ~ n ctrrrtlrtroti.i of use, and \vho\r usc ixtdrrates that tlle condltrtons 111 qilertlor~ ctbt~lxi (Recanatl 1998, $4) For all ~11th expresaons, we can d~stingur\h several layer; of rrlearlrlig

(a) The ineaning of a prapz,itlc inch. ator, p i . 1 exprcs'lon type. IS the convention govenlmg its use For example, the inlperative rlioocl rs governed by tile coiiverltlon th,rt 11. IS tir he used oidy ~f the speakcr usiilg it IS peTfomlixig a 'd~rectrve' i l locut~ot~~ry act "Bur' rs govcrlred by the convet-itiorr that it IS to be tised in a conjnrlcnve utterarrt e only if there 1s a certalrt concluaorr r such tlrar the fir\t cctryunct 4ilpportc r wMe the cecond co~qunct provlcies a atrotlger argurxrria 111 &?votrr ot m negahon. In all cases of &,bt cort the ccrrrventlon tikc~s the ionla of a

con&aonal, the ngl~t-h,md srde ofwhich is ~trlplrcltly or expla~~tlv "token- reflexrve". The convexibon s'iys drat fix every token x of clle rkpressron,

x passes muster only tff(z) Yiix)' Ir a conchaoil or1 the token--3 COI~S(S~.III~ whch the token must satx@ for the rrre to l:,hc &hcitou\

(b) At the next level the nlemmg of tlir exprecsion type is contemldly applied. When a specific token 7 of the expression IS produced, ~t (the

Page 252: Recanati, François. Truth-Conditional Pragmatics

toner-i) lrlrarls that Z(r) in virtue of tl-ic ri~echa~tisrn described ill the

yrcviorrs sceutrrt. '1'11~s ;in i1lxper~4tiari. iltte.r-;rnce i rriearls that i serves to pcrfi~rin ;r clire.c:dve illocutioriary 'ice. Ari niteruxrce u of 'Llle is rich but ~ t~rp id ' ~ ~ I C ~ I I S that tllcre i s a coiic.li*sitrxi rsrrc-11 that the first ic~i~j~irrct of u ? x ~ p ~ ~ ~ - t s r \vtrlie thc secorrc.l i:or!junc-t o?a~ pioviit~s a stronger arglxnent in Ilvotrr o f ~ ~ o t - r . ?'I'IIC ;tpl~lietl rne;rrrirrg of' die token is but an i~rstantia- tiora of tllc right -harid side of t t rc* corrvvtitiorl of irse. As Jolm I'eerry likes ti) j r o i r ~ c ( > p i t , 111eani11g ;it t h ~ scc:orli! level is rssenti:llly reflexive (Perry 200 1 ).

(t.) Next.. tlac ~pplliect ~~lear l i l~g ui- thc ttrkc.11 is cor~~cxtu:IUy fI~<sb~cri out. For rxzlr~ipic, t1w Iria;uer irnlst itlenldljr a specific iilr~cutio~lary act, within the itircctivc class, 3s heirrg tll;rt wllicll the speaker interrtfs to perfon-rr. Or, ivr tlic c:lsi* of 'but', she I ~ I L I S ~ identify tC11c c o n c l i ~ s i ~ ~ ~ r such that the fl~rst

c-oryriric.t siipyzorts r wl-rilc the sec.oiuit r.or~jirnztxq?po~t~ not-Y. Note that flcslrrrig out is ,ill ol)lig;rtory stcp, similar to tflc ob1igitoi-y itientificatiori critbrc rt~firc.rizc irl intcq7rt:dng n dircc:dy i.t:i'crential exc>rssion. Just as ycur cFio arcit prtrperly ~rniler-sca.ri~i 7 II~. i s Xxriti' tlrlless you itierltiFy tlre rc.icirerit trf illc j>roriourt, yorr ilo 13ot ~l~ltfentclriri 6it'im h3s ri~easlcs but

J i a l ~ i i ' i ; i)otk i s tong' ifytrri cIv 11oi cc~~\t~)itiially icientify the relev-ant r, in tibnx.rx i>i'wkiic).i 0111. can rn;ilie qcrsv of ~171: stlggested coi~trast: behvcen tilt two iarxqirxicts.

0 r l c . r rJnc ,rppilctl r~reanir~g oftltc ~okear has becx~ fiesheti out. the interpreter casr apprvcii:"~": ~ h c ixtterance's illoc~idorl:xv fli>rt-t. (i:.g. the 61ct that it is

intt.nrit.d AS ;i I C ~ I I C S ~ ) or its argi~llit~r~eltiv~" (e.g. the L~ct that the whole utter.irlc:e itself is ofired ;is an :Irgtixllerit Irr Kriro~lr of not-r, since the sccor-td r:crrljcr?;rct is ;irgcrriruirc,rti\ strorlger tfiarr the first oxlc). 'These dir~-rensioris of rlicLaalng-----ilPcicutitrn;uy k>rce ant3 :~rgi~nieritttti\re \rLiXtr~~--~re distinct from and extcrrral to tPrc i~itt*r:rl~cc's propositiorr~d contcrit, yet they frclvc converrtiond ~nmclic;ltor.s irr t l ic scrirence. Sitrrilal-ty, I IlolJ, the iluotatiori ITIIL~ELS are a pmg-- uratic. irltfic:ltor wl~ic.Xr aontsii>rxtes to tlre nic;uiixrg oi'tllc seritellc.e, withotit c~or~rril?rrci~rg to i t s propositiox~al coxiterrt. I-1rrlc.e we shoulcJ bc able tc~ distin- g~lislr t l r t ;ri,ovt= Ic~vc.1~ oflnetrrling in yiloe;ltiorr:d rlrtterances. Indeed we are: (a)

'X'hi. yuot:irsicrrr lrr~trks lzavt conditioxts of'iise: dley ;Ire to be used only if the

spcakcr is usirig tt>r quoted words ~clcxlrcinst~~tively. (11) In virtue of this c,c>rlvc,ndori:rl rcilirircrncx~t, using thi" yuo t .~ t i~ r~ I~~RL'F,s in 3 particular utter:tnce ir irrc1ic;ibcr; t h a t tlxe rc~keri 0 wic-hin rlrc ~jrrot:atiotr rnarks i l i u is clisplayctl tc~r

Page 253: Recanati, François. Truth-Conditional Pragmatics

dertiomtr,~txve purpses Th*s utchcxnot? (the ~pplrcd Lile'rrirllg of tire tokrr~)

has the s t~tus of "~ortventlondl impi~t,lt~rre', as we SAW 111 Itf< tton '4 3 It IS

drrectly getrerated by the c ori\7c=ntlon ~eg~ ld t tng the tiw of the cxprcsinrt l3irt

(c) it nlu\t be iievired out in ( ontest. r l ~ e intclpreter rltrist ~cicrrtri"j/ rhc u~kerxl,rl target of the dcrnorrstratlon, 11 there n one, ~ 1 r t 1 he rnust ~dcr~trfv tlie proptrtles

ofthe toberl w h ~ c l ~ at: 'deptctlve' (c onstltutive of tllc dc~-t~on,tr.lted type) knti tliuw wllicli are r~xerely ac~idcrrt~tf or 'supportivt~' ((:ldrk ,rnci C;erng I~(>o. 768). When the apphed meari~ng ofthe quot'ibon irtdrks b ~ s heen flc\I.rcd out m tlxs way, the lrlterpreter 1s able to apprecldtc tlre dcrnt>nstr,tt~o~l'\ plctor~,d v a l ~ ~ c L'he p ~ c t o ~ ~ a l value tlrt~s detenrl~tied L\ ,u extcrrlal ro tbc ~ i t t ~ t , ~ r t ~ c ' s propos~tlond cor~tei~t as 'rrgrrtrlcntdtlr e vilue or tlioc utror~im foxc t"

What 1s the relatton between the three l'ryerj we I-tave lust tlntlngri\hcci- (A), (?I), ,md (i)--G~~id the three levels of meauxig we t~lbecl , r i > t r ~ l t r,rrl~er 111

collne~tlon w ~ t h closed quo~t io i l , nx~nely levels r , 2, ~ r t c i 3 below2

I t'rke it that tile I ~ W I'tyer~ c ~rre\pond to J t c p Z F I t h ~ ~ ' t z e ~ t t l o ~ ~ ,if the tor1<11 rtleanrllg of the tfcnlo~ tstrat~orr. Level (A) 1s the Inrtgulstlt nle,rnrng of the cluotclhon m x k Tius IS dlshnc t from, drough on '4 par wxth, the Inednulg of the cirylayed m,itenal (Ievei I ) When ,L sentert~c wltlr (IUOLI~IOF~ 1xl;lrlhC IS

issued, hot11 the rnatend vatl;lu~ the ~j~otdt lon 111dri.s 31id the qutrtdoon r ~ r ~ l r k s

themseives have ,I lutgul\tic, r olivcrlt~onal nre,rnmg wlut 11 1s c o r~ tc~x t~ t~d lv

proce5sed. By (b) ,~pplying anti (c) Aedl~rrg out tlie lllealllxlg of tllC qt~ot,ltloll rnarks, one determines the pictor~d value of the clernor~sh-at1011 (level 2): r;vlrat ~t deplc~5 and w d e r whch a5pct ts. As for the thlril of tlie t l~rre levelc t,iIked 'llmi~t u~ Scctlon 3 2, it Ir relev'int only when the quotat1011 is t l o ~ c i and serve\ JS a \mgt~liir term 7 he levcls ('I), (b), and (c). 017 the O L ~ C ~ lr,mtl, can I)e fottnd

m all tr~staxlccs of quotation, whether open or closeit

Even III open q u o ~ t ~ o i l , the proce\s of nntelpletatlg-trt docs not stop wherl tlre mterpreter gasps the qr~otation'~ prctonai vdue 7'he mterpreter xtnv

well grasp the latter-lle rruy realwe tblt the spe,iker rs clcplctrrls so ,trzd \o's

speech In such and sucll re\pec ts - wltllour gettutg the point of tlre demon- i t r~t ion In closed quotatton, tfze dertlonstr,~tioi~ 15 u\cci rcftarnr~gl\ i h ~ t

Page 254: Recanati, François. Truth-Conditional Pragmatics

248 INTERPKE TIN(, QlJOTA TfONS: TllK PFu-GMATK %'I - - - - - - - -- -

mdy serve a riu~nber of purpose, whch c m only be deterrmned on a pragrnatlc baclr In Irony, one rakes fun of the person one is m.mckmg. The speaker thereby drvioclates herself honl the target She may do so jokingly or sarrxstlc~lly, depe~xd~ng on whether her intentson n to muse or to harm. At the other end, the polnt of demonseamg can be to appeal to autl~onty to Invest one's utterance ~vlth the authority ofthe penon whose speech is den~onstrated. Or tile speaker car1 express condescendent approval of that per\ion As Ijerrdre W~lson points out, 'the attitudes conveyed by echo~c utteralce, are very nch and vaned: the speaker may lnmcate that she agrees or th,tgrees mtt1 the orignal, is puzzled, angry, an~used, mtngued, cceptical, etc., or many comblnahox~ of these' (Wlkon 2000. 432) "

Note that the speaker's a m 111 ciepict~ng the target need not be unique or well-dehileated in (12), for example, the polat of the demonstration IS

multsple: the speaker's a m 1s (1) to rnake sure that the readers ~vill take the \peaker's words 'sense' and 'mode of presentatson' m thc proper technical sense he mtends diem to have m t l s conrext, and thereby to help them grasp the prc>poat~onaf content of t11e utterance; (11) to appeal to Fregean authcrnty, (111) to srtggest d 'context' (111 the sense of Sperber and Wilson ty86&), n~rnely the boQ oi' Fregeart theses about sense, in wh~ch to draw relevdnt crrmecluences ti-on1 the proposition expressed by the utterance Or con\~dcr the follow~ng, rather typical example of 'scare quotmg'.

(17) Koun Skour,~to~, general probecutor, wa suspended tn March by Bons Btsm. H i s suLcesi;or, Ivurl I chaka, was 'promoted' mmster ofJusttce UI August

The quotdtlon niark~ aro~lrlci 'promoted' convey rnany things to the mter- preter (1) that tills tern1 was oi-Frc~aUy used, or at least, that Ioun Tchaika's itd dirge of aEect,itlori w ~ / i presented a3 a promotion, (11) that the speaker (the newspaper colu~rm~st) doe\ not fully endorse that descnption, (111) that the reawn why he does not IS that ~t was not a real promotion, but rather a way of' getnr~g nd of loun T'chaika by 'lacking hirn upstalr5' There are no iledr Ii~nltr\ to ~IJliat ( an be corlttrstuallv suggested in th~s rnarlrier This is all in atld~tion to what the utterance d~rectly expresses, where 'what the utterance directlv e~cpresses' Includes not ordy ~ t s corrlpos~tionally arhculat- eci ( ontent but alco the plctonal meaning of the demonstration

""I tfiese atntudes can &o he conveyed in closed quotation: the referring functiorl of the quotation ~ioes not pt-everit it born dso s e ~ q a number of other purposes.

Page 255: Recanati, François. Truth-Conditional Pragmatics
Page 256: Recanati, François. Truth-Conditional Pragmatics

250 iN1 F-I<I"XIL;,'l-IN(; <)II(YTATIONS: 'i'iii I"<AGMATlC VIIIW

(i 8) It~iiii v~ys lie's iirre t i ) 1)reselrc his work iii tiae "xiper session'

wl1cr.e tixc spe;rkcr is rrxlilerstoud 35 ~~Gnlickirig: i?;>auI's deviant use of tlrc phr;ae "paper I C S S I ~ X I ' . LVlfat I ? ~ u i rrltx:irrf. \vircrr PIC said what (18) reports was thr t 81c w;rs ( i ~ 1 ~ to present his work rn tlrr poster. scssioti. 'l'lic speakcr in-or~ii-;illy c.c-Erlws it:rtri's tlrist:rl\crr iise. W11:1t iirc the tr-ritll-cotrditiorrs oi'

(1811 A r c tBat.\~ the rarrrc :;ts those ok'

( i 8 k ) i E , ~ ~ ~ I ~ ~ i y ~ he'\ clue (.I) pi-cst"iit 131s wol-k iii the paper x s s i o n

I (lo s i o r ililrik \ t r . I t scerxu to nrc t l ~ t (18) ,rscrihcs to I%ul refel-ence to tjrr postvr .s~.ssioti I I I I C E C ' P t l i t ' wroriir, r~crnlc-:. Xbrt i r n (18") rrc, reference t.ct the poster scssiorl is ,nsc~-ii)cd t c i I'aui.

B+i.r.haps :r rixisii'lr i:uarx~yle. wit11 L, proper ~ X I ; I I ~ I C ~ziill eliiit clearer intui- ri,911s. S I I ~ I N ~ ~ ~ C ilr ii ~ ~ ~ u t t i t l l l y linowri ttr till(: :;pcakcs and his addrcssec tlvat

'j:irnc.s rxtisiilc~m-iriiicd .i c.i-rtai~r oitl r r r n r r ('l'ini Mi.e"llersoi3) :LS the 6~nious

plriii~sopl~er Qr~irrc. '!'he spe:rkcr i-ari tl)c:rr irc>rrrc,~liy irsc thc rlarxle 'Q~rine' i r l ~ ~ u o " ~ s t ~ " n c i . i : r to Mcl"~crsorr. 'l'liirs lse <::rrr sa\-:

1x1 (19) d i e speaker xcfen to Mcfqirt:i.;c)n under. illc narxlc wlliclt J~tmes

s ~ i i s t a k ~ r i i y g~vlwcs hi~rr . Simiiarly ill (20): tlrc spe;rki:r rei'ers to McT"tlerson (riricler- tile Irasiie wlriih Ja i r~cs ~nist&c.lrly givt..; him) 2nd says thatJarnes says t l~a t hc (Mcl~licrsou) w;ukts to speak to ris. Now C O I I S I ~ C I . (20.~):

(a)*) J : ~ I Y ~ s ~ ; ~ ~ a I h , i t Q t ~ i i e w;rrits to speak to 11s

1 hr\ ticscr r i to t s .1~ irrytl~rng ,rl?o~it Mc l%c"x\trsxl 1 acrlce tl-rcre 1s an obv~otr \ t171111 c O X I ~ I ~ I O I I ~ I E d i f f t ~ c ~ ~ c L' ~ C S ~ W C C I I (LO) J B I ~ (io*) CIIIC s t~ temei~t IS ~ b o ~ r t

McPiicrsoxi, tilc cbsl-rcr r\ a l l o~ l t Q ~ I I I I ~ I. tlllnh WC" Ii~iil the SJIHC sort of

ilifierer~c-cs bcxtiikccr~ ( r 8) arid (I X * )

Arrotlrci- pc~cc.rltr;ll c ci~rx~tclcu,urlplc, drre YO ('appt'icr, .~rld I,rpore, 1s

.cvlrerc Nic-ul,r i s ri, fivsx--yeas-trlit boy. Serrrcrrc,e (21) c:tnnot have the same trrktlr - cox~eiit i~rxls : i s

Page 257: Recanati, François. Truth-Conditional Pragmatics

O l ' b N ()LJi-rI 'A I I O N 251

(21*) Nicula bdieves that h i s tither i s a pi~rltosopbrr

for tlic simp1e reL&on that ( I I * ) 15 rtlear~rngltss the scnteuc c cor1ta1115 't non- \vc)rd, hence ~t does not exptess ,, caniplete p~opositt"on Illut (21) ~rguably

does. In aU these p,urs--(I )-(I *), ( I 8)-(I X*) , (20) (LO*), (T: I) - (A I *) here is ,i

truth-conctlclond drffrreric e between the t w o menil,ers of the palr lllrrs \bows that tnlred quotatsot1 nEccts, trt~th-condlt~ctrl I)ocs rr tollow t h ~ t 1 w ~ s wrong when 1 cl.ams\~lled tnlxed cjuat,ltlon as ,III Irr\rciltt e of open

quotxt~on? No, for the problem I\ rrlore gener~l I l ~ e problcn~ ts that soinetlrxles open cltrotatro~lt seerny to affect tnltJr-cc?~rciitrt~~~\ Miard yuota-

Don n a case 1111 p o l ~ ~ t , but there ,ire otliers. T hi]\ 111 (I<>) the iranic 'C>tl~nr' 1s

dert~oristrated, at the sane tinlc '1s it IS usrti to refer to Ncl21etsctrl L'hat 1, a llybnd case, but trot an Iristmce ofrillxed cluotatron (\~rrc.e there 1ms\ no omtrc~

obliquai. Yet the ciernonstration c~ppe.zs to ~ktxect the truth cctrid~nom of the trttermce. &s ~t does t r i m5tances of rn~xeti! qtlotation such d5 (SO) 4rntentc (LO) cbes not have rhe same trutlt-corrdrt~ons .IS (19")

(19*) ldoo?+. who rc ( ornlng! (Jrurlc VV~~IZS to sj~rak to ti\

Xn contr'jst to (I()*), which SAY, \ornethlr~g about (>urnc, ( I 9) siys \olnetbmg dbotit McPl~erion

I11 tlrc last sechori of tfus chapter X wl1 Jedl cvlth tlrc3se countorexa~nplr", to

the vlew t h ~ t open quotation jli~clucilng 1111.zed cluotatson) 1s a prginxtlc

~>hei~(>menon. 1 w ~ l l Jsgue CoUo~vs. '1 he counterexriniples do not show that the v ~ c w rtself IS ttr-rtenxl?le, but only that ~1~~t7Cklrd uus,i,tnzptmn\ lancemlng tJlc t n t e f i e betuietn sc~nzurzttc-i and pru'qmatic-c f t tust he qutk~titl~tlcd. I '~LZS it IS in-

dardly ,zssunied thxt there n a s-rngle rlotlorl oflltcrxi content wl-uch IS both the ttzt~tftve tmtll-~3ndlt1olial content of tlrc utter,mce mind Its l i ~ ~ r ~ i c t u a l l y avr-~aifutcrl conteiat ( w h ~ t the autonomous rnecharrist-rl of- 'w~n;litt~c c olnporltion' dell- ver;) But m tlre prev-rous chapten, I have Vven up that a'r\\lilrptloli argued t b t the Ilituihve m~t11 -cond~~onal contei~t of 411 urtermce results rrt part b(3m

'pragrnabc ~ntruuons' of vmotts sorts-pr~grlatJtc uttruuons whtch, by uiter-

fenng wtd~ tlzc process o f c e ~ ~ m ~ n c conlposit~c-)n, cnnc h or modify w11,lt would otherw~se be the truth condltlonal cortte~lt o f the utterance. If I xrn nght, mxed qLrotatJorr pro\des A s a ~ k ~ n g rllumatlon of t h ~ t phenorncrtoxl

Page 258: Recanati, François. Truth-Conditional Pragmatics

6 . In dcfencc of the pragmatic view

6 r C~mufutrve hyhrrds arid pruyvnczlzc er~nctzment

We have just seen that in some insQnces, removmg the quotation marks ~n hybnd cases of open quotatlton somehow & e c ~ the content of the utter- ance Thus there 1s an lntuibve &f%erence 1n truth-con&tlons between the two members of each of the above pars (I)-(I*), (18)-(18*), (19)-(19*)* (20)-(LO*), &rid (21)-(21*) WOCV can this hc t be reconcrled wlth the pragmatic v~ewi

Before we mswer tllat cjuestion, we niust draw a dlstinctlton between two sons ot'caw Consrder the pa r (I)-(I*) once again.

(I) Quirte savs that quotation ' . . .has a ceaaix~ anomalous feature' ( I *) Quine says that quotation has a certain anox~~dous feature

As 111 all the counterexanlples mentioned in Sect~on 5.3, the first member of the pa r errt&il.\ sorneth~ng whch the second member does not entad. In thls partlculx case, however. that 15 the only seinanhc &Eerence between the two stnterrients The quotatron adds somethmg to the content of the utterance, hz4t L C does rzot st~btr~zct ~tnytizlvtqfrom at. As a result, (I) entads (I*). 7 hat i\ -I= Corn trivral for, m tfie ocher countcrewmples, the fvst member of the pa r does not e~lt,ill the second member. Sentence (19) does not entad (lo*), I I O ~ does (30) e r~ta~l (~o*)

(19) f1,ook who is conGng!] 'Quine' wants to speak to us.

jig*) [1.00k who 1s co~x~irlg!] Quine wants to speak to us.

(20) James says that 'Qtxinr.' wants to speak to us (lo*) James says tk3t Quine wants to speak to us

The diftereuce between the two sorts of case can be spelled out as follows. While trr (I) the contelit of (I*) 1s eilnched through the demonstratxon, ~n the other exarnples rt seerns t h ~ ~ the content of the utterance is transformed ini;tr,id of bemg nicrely enriched 111 the first tvpe of case I sav that the hvbnd IS 'c~trrrlu1,rtlve' fbr the dexnunstratior~ contnbutes sonlethrng in

udditzotz to the nonnal content of the utterance, whlch 1s preserved (although possibly ernchecf). The cunlulative nature of the hybnd 1s estabhshed by the tact thdt the \er~tence cor1tamiIig ~t eiita~ls the sentence obtained by

Page 259: Recanati, François. Truth-Conditional Pragmatics

O P E N Q ' C J O 1 A 1 I O N 253

removing tlie qilotatlon ni~rks. In the recolid type of care (to be dealt: wrth m Sections 6.2-6.3), the hybnd is 'i~t>n-cunlulat~ve' the dclno~rstrat~on results m the fact that the utterance ~zo laugcr e x p r f i ~ a 'SES n~)r~null ro~ltet~t, not

even as part of a ncher content The smple hybnds I merit~oned m Sectloxi 4 2 were clearly cuniulative

Thus (11) entarls (11").

(11) John is very 'cool'

(11') John a very cool

The differerlce between the curxrt~lat~ve hybrid 111 (xr) and the noa curnu- latlve liybnd in (19) u s ~ d a r to the d~fierence between (I) ant1 (LO) To be wre, (I) and (LO) are inrtancec ofmrued quotation. whle (11) and (10) arc

not. But that is irrelevant I he cun~u1,ttlveInon-c~lmiilative ciis~rict~on 15 a

drshnchon between two sorts of lryhrzd rise S~nce xll~xed qliotatlon a 3 variety of hybnd use (~nvolvmg orcltw ohlrq~ma), ~t 1s not suq7nsing tilac the dlst~nctlon between the two fi>rxns apylles to ~ n ~ x e d qizotat~ori as well as to suxipler cases hke (11) and (1"3 ( K'alde 7 r l

lfespite tlie fact that both (1) and (11) .ue cuniulatlve livht-id\, rn ivh~cll the normal conterit of tile utterailce rr preserved, onlv 1x1 (11 rs there ennctixnent of tlrat content z f r tire mrfh-iotid~tronal sttzsc. In (11 ), the dernoxi- stranon 'enriches' the content oi tlrc" rrtlerarrce rxi the rense th,~t rt ~ d d s to IC a

layer of pictonal meanlilg, as we have secrr, but the trutli-coiici~tior~s of the utterance are unagected. From a truth-r ortiirtiol~~rl purrit ofv~e.cv, there 1s no difference between (11) and (rr*) Bur therc 15 a tn~th-ct~ndrt~oxld ii~fiertxric e between (I) and (I*): (I) entats sometbr~ig whrcll {I*) does not entall 1 Iow can we accoiint for that Gct;

When, as in ths case, the hybnd is ciimuiatrve, there IS ari edsy expl,ma- tion for the demonstration's rnlljact upon the tnlth condruoni, of the

Table 7 . I . C:umulanve and ~~on-r-unlrxl,ittvc l~y.\>ncis - Cumulative Non-cumulative

SmpEe (IT) John n venq 'cool' (19) 'Qulne' wmts to speak hybnd LO u\

&Itxed ( I ) Quine says that quotatlorl 'ha\ a certairr (ro) Junes says t h ~ t 'Q111ne7

quotatxon anomalous feature' w.xr1t.s to tnlk to us

Page 260: Recanati, François. Truth-Conditional Pragmatics

2s 4 IN I>IITEN('F (>I: l'l-iE PR A<;Mi"rTlC mEW

uttrrawic-r. We c.;lri irivcjke t.lre noric,xl trf prirgniatic enrichlllent of.' truth-

corlci i t~trn~a~ crraltcnlt, u7hi(:h 113s ~ C X P I ~ C ~ kbl:ge i n thb: p rev jo~~s chapters. \Xilroixr ,I rpc";A.c~.issc:rts sc.rn~c~.tl~ing, tl-acre Arc 0tie11 aspects of what be

,~ssca.is rll,it , 2 r t 1101 i.xjrlic.irly sr:rte~I hrrt arc- pr-ovicied by the contcxt. This is different horn tx~dcxic:rlity irrst>firr- ;LS ~ l i c coritcxtrr~lly provided constirl~cnts rcrrrLlirl chrinrcly 'ixrrr,irt*c:iil;ltcit'. K goocS ex,~rr~pie is

(2.2) 1 i e i c~c ik o u r 111s k r : ~ and opei~etl t t r c - tioov

in viihic-j-1 i t 1s ~ i :~ t t r r ,~XJv ~lil(ierstoo~t tliat thc (loor mentio~ied in the sccorld

coryur1c.t wdc o j>~ '~ icd 1 4 i f l l tlrc key xrrerritiorlrd i ~ r the first cor;junct. Tlizt is a j>rag~r>;itic. >~i:ggi~stiorr convc~yrd by the I.lttc~r,mnc-e, r:ltht:r than an aspect ofits

c c . i ~ r i ~ ~ ~ ~ i r i c ~ n ~ i i i y ariic-ulatecf r:cTr>tc8rrt. (Judi j>r:"~gxiiatic suggestiori it can be

c;iiri-eiied, ;is r ii

( 2 3 ) I ili. took cirr t l i i i key ~irtd opcrieci the ciocrr, brrt he used my key instead ofbis i)c.i,itasr iir rc.,rlrieci his w:a hr-okcn

i )cspiti" itr pr:tg.:iri;itic: ixtturu, the suggestiorr corrvcyed l1y (23)---to the effect that ~ l r td i ioor w:a opcltec! with the key --docs not rernairi external to ~Lle irrcrirrivi: trtittl- i:c>riditiotls ofthe irtter-:irrc:cb. This is ixr contrast to st:mtfard *e~ouvcrs;itrors-Li ir-ripiii.ntures', whit-h rt-.rrl;rirr irlt~ritlveiy distinct from wlrat is ,auertc:ii. I'hr is i s ]

(24) 1 Wor~/c/ yo131 like rorrre bmi~tly?j 1 tion'r tivrrik alct.thol

rlae ,rirst\c.r rxripl-xc,irc:s i.tt:rt the sycaliix tit.^^^ not \v:mit the 'r,r;inily she is kindly trfi2rcti; b i i t c i i c s irtrplii::tture rt-.rri:~itis drsiiric-t tioln, ax~td cxten~al to, what is ~sser-tvd. b5y *WII:II is xsserted' lrcre t rrie;-inr .;tx-nc*tl.iing corresponclirlg to the

ixrttiiiivc tatat11 c-otiilitioris of ' dre uttcr;kric.c. Xlit~~itively, thc speaker who in,iki=s t i ic nniswcr ir l (24) :isscrts tlr;lt slxe i10c.s trot ilririk alcohol and 'implies' tllat she ciiic-s l i c i t w:trrt :trry br-antiy. Irr (221, Iruwcvcr, tllc flict that the ctoor wils cipt":wd \ w i t h ~.lrc key i s r l o t ixrt~rltivcly takexx ttr be 'iirriplied' as opposed

to nsscr-tc<j. El c;ikrs soirle rt~flectioxx to realize that that h c t was nut expiicitly clrtic-rri3i,ccJi iii tlrc" sciiterlcc. I:ronr 3 ~)syctli;)logic;~1 paint ot'view, the prag- rri:lric ~rri,ngc~".tiorr k Irrcorpor.atrd irxttr cvl~ ,~r is asserted: a sirrgle rnentai represc~itacirrxa is constructed i~sirlg Iroth ling1nsr-i~ arrct corltextual clues,

i - , r t f i c ~ r11,tri twtr i 1 i s t i l ~ t . r : rt:present.atic)r.~s ns i r r the case of the answer irt

Page 261: Recanati, François. Truth-Conditional Pragmatics

(24) 111 $11~11 ed\c\, I \ay thdt tlle tr~itir-c c)xlJ~trc>n~l cortterlt of tlze tlttt-railce 15

pt;ymcztz~aliy enncht~d Now rclxember LLEIAL I \aid ,ii>oitt ( I ) A( c trrding to urly '~rt.tly'irs (5c.i t~ort\

3 r - 5 L), the hr,>rer gr;l\pk the ptc ton'ri trlcanlrig ctf the cluc>tatiorl \vherr 4t1e

~~rlcferstands die spe'tker a\ tleylc ttng Cjulnc's own use of tile Cfexr~ori\tr~ltcrl \vords. ULI~ thdt IS ]lot >~ iM~c~e r~ t . tlte hearer rtrust ,tho i(+crttdy t11t porrtt of

the derllon\tlatlorl, that IS, she ruu\t r e t o g r ~ ~ ~ e the \~ ) e~Le l ' \ lnteirtion to let lie1 knom that (2urne used those very W C ) ~ ~ S 111 the speecl~ ep*\ode wtllch the utter'mce reports I Ilo ldst aspec t of-tilt Intc'rprctal~orl oi tl~tvittcr,rrtcc, relatlttg to the poxrrt of'tlrc: cte~rio~l\tr,itic)~, i\ ~ i t t e~ ly plagxildtrc, yet I r ,rffec ti; the lnturrive t~ uth conill tlonal contetit of tlw uttcrdnt e ( \ vh~ t ir &isserteJ) The h e ~ r e r unde~ct,tnci\ tfrdt Q L I I ~ I ~ exy r eswct die I cportect pi opc)+itlon ,trlct

c11d so uszr;? tllc delrzorastrutc~d ivorciJ, trrtich in ( 2 2 ) tlic lie'lrcr ul~dcrst.incfs that the t n d ~ ~ i d i ~ ~ i i lrfe~recl to took out hl\ kcy '111rf OI)C=IIC'J the (iot>r I V I I ~ I ~t

In both case\ wtl<rt the 11earcr trrirlc'r\t'irrds a\ berr~g 'i\ri.rted c orlt'lirri rlrore

than what i\ i~rrgul\t~callv drtlcul,tted, 35 ,1 ~c\til t of pwgnl'ittc cllrrc Jirrterit I r i (sr) , the praglnalrc rnealing oi tilt" dcliiorntr,~tior~ docs no1 ,i1Pect wJ~,lt

I \ r~rrderstooci AS d\\e~trd, hut only ~ a i i ~ ~ t 1) I I I I I ~ I C L I (111 the ~nt l i~ t lv t~ S C I I ~ )

Exnr-nple (11) theretore 1s rnore l ~ h c (24) tlt~~tn lrLe (22) 7 he cvndit~c)n\ ~tndcr wh~c-ii praglt1,ztrc enr~cllnicr~t ot tr~ltlr-co~tcl~tioj~,il c orlrerlt lil.iy or

illay not t'rbe place 1s a diilrcult rr\rre wtict\c, d~sc u\w)rt I wlll pctstpottc rt~ttll CIlaptet 8 (5e~tiorr 3) 13ut w l l ~ t 1s lnlpoltatlt f a r rllc I>re\crrt argllrnelit I\ the

iac t that soinchme\ the] c 15 p r ~ g ~ t ~ ~ t ~ t trl~lcjirrtc'rlt of tmth -c oncl~tron,~i content 511lc c t h ~ t IS \o we need to cl~\tzrtgit~\h L>t~twct~rl the c o~x~posltri,nally ,irt~c~ti,ttcd cortterlt of tire t i t t ~ r a ~ ~ i c ~ (C iotitcl~t), a l c i it\ lrltliltive tn~th-

cond~tlon,ii contcnt (I c o~rtent) Wlrcnelicr ~ I J ~ , I I I , L ~ I L er~ri~hrr l r~zt t~l.e\ place, as u~ (27,). the I-content coilt;ilxl\ irlole tli,~rl the c c orltctlt

As 5oon as orte draws tlsc ~iiu'eperidex~tlv needed cf~rtxtctron bct\veeli the two r ~ o t i o n ~ of content--the I -< ortterlt and tk~c c corttexlt -the problcltl raued fix the pr;igri~,~tlc vrcw by thc 1~~1 th co l~ ih t~or t~ l (31fkrc11ce berwceli ( I ) ,aid (I*) vdnlshe\ Wllen I sav that; open ipot,ltrori 111 generzll, .utd inlucd

quot&hort rn particular, n pragm,ruc ~ l l d cioe\ not ,Ifrect thc corrtcrlt of the utterarict., I mean illat ~t does 11i)t nfrer t 1t3 L-c oxtent. Tlr'tt I \ cousntcrit wt11 the Lct that ~t can affect IS I-content Irdeed, I 1lolt-l thxt I corltent in general IS yutte systervl~ttcallv aftei ted by pragnlanc Intru\rons o f v ~ n o n s sorts. So I ~ 1 1 1 not worrlecf by the i ounterexanrplc\ ~nvolvnig c~nlirllatlvc hybrid\ 51rlcc' tfre)~ car1 bc 11~1ldli.d ur telnn of pr,igrrieltlc crtr!tli~~rerlt, tlrc'y

Page 262: Recanati, François. Truth-Conditional Pragmatics

are not counterexa~nples to rny mew. They ram difficulties only for someone who holds both the p rapabc mew (argued for m thn chapter) and the tr'~31ttonal view which equates the I-content and the c-content.

6 2 La~guu~qe-sla$f~ ftir~ no~i-czfrntr~atzve hybvtds

When the hybrid IS non-cumulatlve, we cannot Invoke the notlon of prdgrnatlc emcbrxient of trt~th-condalonal content to dispose of the coun- terexamples 'I he Iilillr~urk of the process of content enr~chment is the fact that the ennched tontent (the output of the process) entails the ongnal content (the irlprrt to the process). Thus we have seen that ( I ) entalls (I*). But rn thc non ctmxulatlve cour~terexamples, repeated below, the first lnenlbcr of the pair does not cntxl the second mernber Instead of belng pr~servecl and riierelv eriri~hed, the o n g n ~ l content 1s transformed

(18) Paul savs he'\ due to present his work 1x1 the 'paper sess~on ( I X*) I'arrl ,ays hc'9 due to pre~erlt 111s work in the paper session

(LO) jarneb LJVS that 'Qrune' waxits to speak to us

(20") Jaxl i t '~ savi t h ~ t Qiur~c want\ to spesk to La

(21) NICOIA hcltrves thdt 11.1s fatl~er 1% a 'ph~ltosopher' (a*) NtctrL beltevc~s that his father is a philtosopher

I-Jere n wlrnt I t ~ k e the Inttxrnve cclritcnt of the first nlember of e a ~ h palr to be. Serltel~~e (18) cxpresres tlre proposrbon that Pml says he's due to preserlt hls work m what he calls the 'p;rper session', narnely the poster secuoa, serxtence (20) expresses the proposltlon &at James says that the ~ndlmdual he cdls "Qwne', namely McPherson, wmts to speak to us; sentence (2 I) expresses the proposstion that the five-year old N~cola beheves hrs fither Ilw the property he (N~cola) assoctates mth the word 'phlltoso- pller', whatever that property my be In each case the expression wlthn the quotation nlarkl\ IS not used wrth ~ t s statldxd medmng, but with the niedn~ng rt has )or tht* person whose use u betnq echutcally stmtrlated. The reason why the first rnenlber of the pax does not entail the second member can be traced to tlrc f a ~ t that the xneanmg of the quoted expresslor1 does not s ~ y invdn~nt Instead of supenmposlng an extra layer of meanlng upon an otherwise invanant content, the demonstrat~on changes the m e m g of the expresrion in quotes and thereby aEects ~ t s contr~bution to content.

Those cdses can be described III terms of a farguu~qe-shlfi In (18) the speaker does not use the word 'paper' sn its nor1na.l sense, that IS, m the

Page 263: Recanati, François. Truth-Conditional Pragmatics

sense a has m academc Eng11sh (where ~t alealrs urtde), but In the scr~sc ~ t :

has m Paul's idiolect (where ~t Iiieans what 'poster' means m clc-ademlc Enghsh). 711e sanie th~ng holds fix- the proper mine Qume' In (20) t t ~ t

narne norndly refers to Qurnc., hilt nI (20) ~t is used in the .ieir\e it t i ~ s 111

James' id~olect, where lt refers to Tmi MtPherson In (11) tlie speaker uses a

word from Nicola's ldlolect 'l I u t word does not emrt ln kngl1s21, even though ~t 1s etymolog~cdy denved &om the Engl~sli word 'yhrlosopher' L3 24

As Bar-HiUel panted out hail a century ago,

Anv token has to be understood to beloiig to a certalii language Whet] somcbodh

hears somebody else utter a sound wilrch \ounds to tum like the Ex~glsi~ "ilrne', he lrught sometimes have good reasom to believe that tlus sound doec not retcr to the nurnber rune, and th15 1s the case that he will have g~)cd reaoris to dsunle that t h

sound belongs to the Ceririarl Inngtlagc, r i l wlut h Lase it refer\ to the sarrrc as tire Englrsh 'rio' In this sense, no hngu~s t~~ eupresslon n corrlpletelv rrldeprxlderic oi t i le

pragmabc context (&r-kIfiel 1954 80)~'

In \onle cases the Interpreter mav have reasom to belleve tlidt ,I pdrt~cuXdr

portlon of a grven utter'ulce belongs to A d~Ee~en t 'language' t11~n the rest of the utterance 1 put 'language' 111 yrrutatlon inarks Rere bec:,tuse 1 arrl illslrrg

the word 1n a fairly ~rlclusivc sensc The shlfl ' ~ t lssttr tl-tcly he fioili Lagllsl-, ti) (say) French, AS ~n 'He say\ I i onlplctelv togrrh', or rn the exarriple iron1

Chapter 6 Sectlor1 2 2 ('as the tretxh sty, otz r?'e~t pas sortis dc l'a~thccqe'), but ~t niay also he fiom a &dec t of l-,ngIrsh to another, or frc~rn ,I 'level of language' to ailother.. 7h1u I xvould routlt exznp1c.j (9) arid {roj fronr Section 4.1 as Instances ofsenterrrce-lxltenlal Ia~~wage-sh~fi

Zj Wlde in (18) anu (20) we know what the relevarit words nleari in the relevant idiolects, in ( z r ) i t ' s

hard to say exactly what the serise of 'pitiltosopher' m Nicola's idiolect is. 'Thrs caw n actually qmre complex (and tnteresting), for both Nicola's own use and the speaker's echoic tse can he sad 10 be 'deferential' in dieir own ways (Recanati aooob, Part VI). Nicola d r k to [nature speakers of Er~giish m 1.'. .. C . - l . . , ...- I ..... tub uhe OL waut ,LC -m LU bc L&sii ~ o i d ~ w t d e die spiaLeiof[i i) d~fcib t<> ? < A c c ~ ' ~ O*U t i x *>tit:li

he nlirnics. Be that as it may, it is out oftbe yuesnori to undertake the andyrh of such a cornpiex exatnpie here.

Z4 Stefano Predelh @.c.) raises the following irl?lecuori to my treatment of these examples r n tesms of language sbtfi. 'The thesis is that, say. "Quine" in (30) "IS i~sc:d in the sense it lcls i n Janies' idiolect. wherr it refers to Tim McPherson", and that "philtosopher" in (21) is "a word froin Nicoh's idiolect". that a, a word referring to philosophers. The obvrous worry L\ that ~r would be slmpfy Use to say "'He rs not a 'philtosopher"' and contradictory to say "He 1s not a 'phdtosopher', he 1s a philosopher".' But the objection assumes that the same interpretatiotr 1s given ofthe qltntahon it, examples like (20)--(2r) and ur instances ofso-called metalinpistic negatiorr such ar dwse l'redetli cites. 1 reject cbac asstrmpooal, " See Kaplan (1978: 228-9) and Keciiratl (rg79, 16s) for sinrilar remarks and exarrrpi~s.

Page 264: Recanati, François. Truth-Conditional Pragmatics

258 IN I>I;I'FI".JG:E ('I17 "I'Tl)i i'l<AC;lbZKl.lt: VlLIW

ioj 1.0 \z r i i rc . i~ Mr I33ilt.y inotlcstly replied tir.it he hoped he kllowed wot

o'cXoc-l, ~t cvos 1x1 gi~lzr-d. (:ti) 1 :rrr-! vii~i!~r&?nr?ri". . . ?,sirzt 1124 i e u i j (!I v i e d ~ ~ ~ r a r l d i z ye14 i_ i i~~~. j 'vo~~iais, d'tirtr: rioi.x

avcl;ri,rnii. ~ i : irnrtrii.~tse.

1 t;rkc i t cP.);ie ( i X), ( L O ] , I I ~ I ~ (21) can be ~IKIX~SCCI ~ I I ~ ~ ~ o r e or less the sane terns. TItc ~xpres s io l~ / I ) (.III(ILCS i s ~~tlderstot>il as Oeloi~gi~ijr to s ~ n ~ c o t ~ e ' s idiolect (l';rilE7\, Jarncs" ox. Nit-(cia's), irr eoritrast to tllc rcst of tire sentence. As a restzlt r-lrxt vxprcssic,ri i s givcsrr :1 speci;+l ~ ~ i e , x ~ l i ~ ~ g , bkrl*icli detcnrlines n speiid content.

Let rr:, t , ~ k t ~ t o c R . in IIOII ~ u t ~ ~ i ~ l ~ t i v e il~st:\li~es of lnixed quotati011 as e/sewI\crc, the clrxc)t;it.ion rt~arhs i x l i l i ca~~b that tlrr worcis inside the quotes .ire l ised tleri~cjarstr;~trvely. 14s the irlcgc;itictrr Is Aecl~etl out , the henrer under-

S C I I J L J S t l x i t t1 i t .h ~l)tsak(:r L I S ~ S ttic \vorcls L"CIICIIC;\I/Y in tile (deviant) sense they Ir:ruc hi,i, srrcir h i r ~ r i .; l i t-!. i :t person. "I'his ;~fTi.crs tX~c t.ontent of the utterance, by cletei-~r~iriirri; tlic 1~1)gtt:tgc ~ v h i t - \ ~ is ri~levar~t CCP the i1tter1;retatiw of the i*xprrcs?;iorr i r i quottts. li Ierc, in i-orltrast to what happens ill cumulative c : ~ ~ e ~ , the i col l t~n t of the U~I[CTL?IIC.C i ~ ~ i l r ~ t e d . For the process ot'sersiarttic c-o~aiposii~icirr cvlmrclr outputs tire c-ci-~rrtrnt t;il\cs rlrr rt-icnning of the corrstit-

i ic"rxi wa)rc.ls ~s irrprrr, aritl the ri.rc:ir-riilg of'tlsc constitucrlt wcjrds itselfdepencls i i p ~ ~ l i c " I I I I I ~ U ~ I ~ V ( S ) ti) which the words i r r qricstion ;ire tnkerr to belong.

I:wesr clrvirgir i t ~ffi-cts tlie c -corriex*t of tfrr eittcr;lnc-e, the demonstration doc5 so only irrr~iructlv. T ~ T U L I ~ ~ I the ~ict~~~'i"ril):etii>~~ ofthe i r ~ p ~ t t to tlle sem;u~tic

~ ~ E " C C S S W ~ I C I I ot l tpl i l~ the c -(:otittX~it. l ' l~ ; i t PTOI.~~"; o f ir1l>ut:-~3etenrUn:~tiorl is, by dehnitiora, prc scrrl;u~tic. As K~pla r i ( 1 9 8 9 ~ 1 : 559) writes: 'C;ivcn an iitier,lfii e, scrr 1arrlic.s carrrrot tcii 11s what. crxj-.rcssio*i was ~itterecj c)r what iarigu;v,c i t \*,is uttered i r k . ' T l l i s is J 1mre~trrri~irrtii- task.' 1 angtiagc-selection is

itkc, W C " X I ~ I " . - S C ~ X C " ( ~ i o ~ i 311 that rcspei,t. WIlcrl ari. ari~il)ig~;llotrs serltertce is ~l t tere~i , wrrlnarri i :~ b y itscil' c:mxiot tell us in kvhic-XI senso il must he taken. 80th ili~~i~lrbigrratior and I:urgri:rge--detur~~iii~~iti~~~x arc prt~glilatic- processes wltich

arltisr takt: p1;wc. hr;/bri> ttxc sc~raa~rtic prtrcuss of c.ontent-deterr~~i~~tian can cvt-rr starl. I t i s only :rt that .~?r~->crir;rntic levci rllllr tlre ctcmonstratiorr has an

ilr~pac-t upor 1 tlrr. rarr erarlcc.'s s-coxltcnt, iri 1 1 7 ~ X I O L L - - C L I I ~ ~ L ~ ~ ; L ~ ~ V ~ cotinter- cxar-~iplrs. ii k r c rigairi, the, an:ilogy :yvit.h cfisaiixrbig~ratio~r is useful. As Kaplan

poiirit~ o t l t * 1f':t Il;rherJ~~sl~rr says ' I ::\rtr ou t of r:llecks7 to a banker, 'whether

tile uttcrancc t<rl;cs piact. ia t-he store or at idle bank help the bartkeu to cietrrrmir-re what 1I-i~ Ir3berciasher htis said' (Kapbn 1978: 229); yet the

Icrc:~titrai of the trtcc:rauce rxlakcs no tjireit contritrtltion to its content,

Page 265: Recanati, François. Truth-Conditional Pragmatics

whlc h ~t affects mcirrectly. ?he \nxxse thtllg c ,311 be \a111 ~ h o u t tlie role of the

demonstrat~on m the lion-c~~mulatrve c o~~rtterex'ritlyles It a f l i~ t 4 tit? truttl- cond~trorrs ot the utterance tlliilrec tly, i ty slgndllrrlg ,t sclitcrtc e-tr~tern,d language-shrft 'Tile \hifi in clut\tlori rlprtrc di~tlv ~ttZ"ct\ t l x ~ c-t oliteilt of

the utterarlce, but the dernotl\tratron melt rllxkes no dlrc'ct cotttnl~rrt~orl to t h~ t coritcnt. l l ~ e o111y tilrect ~on tnb~ l t ron the tfentc>n\tr,it~on lirlakes to tlie rneaxllng of the utterance tn the noti-cun~~ri;lnve cou~.rterex,inlple\ 15 the

p~ctortd rneariing oi the quotnhon. That, xs w c tl,*-Le reeti, rein,lius external to the c content

I conclude that the rlon-cunnrlat~vr counteres,unple r t o tnorc tllre,rten tlte pragimtrc, vww rhan the ct~rr~ulative c uirntcrexal-f~j~le~ (to rllc C I I I I L L I I ~ tibe counteresaa~ples are cilali~cten~eci by tlrc opcr'rtlorr of '1 pmgmatlc

pro( ess of content-er~r~chrtlertt which attzcti the I -content oi the uttetdrlce hut Lenves ~ t s c-content ~ ln to~t~ l lcc t NOTI ( I I I I I U ~ ~ ~ I V C ~ c o i i n t e r ~ x ~ ~ ~ ~ ~ j ~ l t ~ Arc harder to dlsposr o f -s lnc c tlle r cor1terlt of the ntt-erancc I \ ,~i\o dffec tcd lS~1t

cvc h ~ v e just \een &at ~t I\ nffe~ted oi~ly ~niiirectly. at ,A pre-\cttldntlc leccl 7 111s mlses no prol>lcm for the pragmatic vie-~v ""

There 'ire rtorl-crr~n~ilatlve hybrrcis that carr~lot be t~~ndlc t l 1st terms of lar-tguage- shzft. In Kecmatl (1987- h3) 1 g'rvth tlrc fifollow~ng e ~ 3 r t i ~ ~ l c C ~ I J ) - pow f i 15 ~nurual knowledge [setweerr john arid nryselt t11,ir PCtc'r wrongly hel~eves that Mary I\ my sister Spottlng M,try ,it 2 clrst;tt~cc,Johrl c;ry\ to nltX

(25) 1-ook! 'Yotir sister' is coliung over.

By u\lng the mcorrec t dercrrptlorl 'your \~ster' to ~e fe r to M'+r-v, J o h n

~ronlcally echa es Peter's ~ n ~ ~ t , t k e . Tltr~t t lr~i 15 J nori-c uxnulahve hylmct 1s

shown by the h r t that (25) cives rlot cnt'i~l

(26) Your srster IS cornrng over

Sentence (25) 1xle;Lns tlrat tl?eperson wi-tom f ' e f p r Bcsrrrbec ia m y srcrrr I\ corntrlg over. That does not elit;l~l that niy , i ~ t u d iirter (j;lne, say) 1s coxntilg o v c ~

2" Note clrai h e pragiianc psoces o i cczntrrit-.enr.ich~llenz aisrt wkrs place I r i x~o~~-c~irn~rlaove instances of mixed quotaeon. What sixgles out non-cuiuulative i~istanccs of niixcd quotation n ~tre Bct drat afunhrr process ofhtrguage-shifi (or, as we sh;tll rec, cutrtcxt-sliili-) trikes placc, wkuch affecis the c-conrcnt. It is because crftiiat fitrther proces, aiso trigscrid by the clcn~onstr,icro~l. that tlre first mernbrr 10Fthe p i r does not errall ttlc second rt?e~rit)cr.

Page 266: Recanati, François. Truth-Conditional Pragmatics

- - - - - 260 IN 1)TFFNCE 0 1 7 i-Ed ~ M ~ M ~ T ! ~ - % E W -

Even tllotigli the hybrid 1s non-cumulative, it cannot be handled in tern? oflanpage-sh~fi. For the phrase "our sister' is used wlth its nonnd meamng in (251, rt ir the itandard Enghsh phr~se 'your s~ster', with its standard scrnantlcs The ditTerence between (25) and (26) concerns the circumstance

wlth respect to witich the extensrctn of the descnptlon IS detemned. The descnptloil 'your slster', as uttered l)y John, denotes Jme, my actual. sister, tn

I ~ P u~tt1(11 tfl~rld, Lbut rn Peter's 'behef-world' (that is, in the set of worlds con~patrblc w ~ t h what Peter belreves). or rather 111 what the speaker takes to he Peter's brhcf-world, the descnpt~on denotes Mary. m a t 1s special wlth (25) is ttte fact that the relevmt crrcunistance of evaluation for the descnptron rs Peter's alleged behef-world rather than the actual world

She ~onterlt of an utterance 1s traditronaby equated with the conhtloris that m11rt hold in ~lrcurrwtatice of evaIu,ition for the utterance to be m e ~n or at tttat c~rcunistmce Wtien a sentence 1s uttered m a context k, the relevant circumstnxlce of evaluaaon IS normally the 'circumstance of the context' (Kaplan 1989u). A context comprises a specific situation of uttermce (~nvolving 3 speaker, a Ilearer, a time and a place of utterance, plus vanous other hctors) togvtlxer wit11 a ~ontplete po\yible world to wfilch that 5ltua- trorl helung Sslien a \entente IS uttered, the specific srtuation rerves to anchor the ~ n d c x ~ c ~ i r md prov~de ~enlaxltrc values fbr them, thus deternun- mg d tmtli-cvdlu~ble content frtr the utterance, that content Is then evaluated 1t1 the po\sible world of the context, thus detemunlng a truth-vdue. The rarne thulg holds for sub sentenhal constituents ruch as the description 'your sister' it a fint interpreted w t h respect to the s~tuatron ofutterance (in order to fix the value of the mclexlcd 'your') then evduated in the circumstance of the context so as to detenmlle the extension of the descnp~on m that c i rc~i~i~r~x~ce. ' / What is special m the case of (25) IS, agam, the fact &at the ctrcurz1st;lnce of evaluation for the descnphon 1s not the actual world (in whch tire titterarice takes place) but Peter's alleged beltef-world.

Thn case c a n be dercnbed ~5 follow^. The speaker temnporanly pretends that he is Peter: he uses the dtvcnptlon 'your snter' to refer to Mary, the way Peter himselfwould. In other words, even thou& (25) is uttered m a context k, the speaker pretentls, at a certarrt point, that the context is dtfferent &om what ~t act~rally w. 'That 1s an msmce of context-shlft. Slnce the speaker tempordy

'' Unless, of couae, the eqxcsslon is in the scope ofa circumstance-shiftirtg operator in thesmtence. See Chapter 6. $2.2.

Page 267: Recanati, François. Truth-Conditional Pragmatics

--

- - O P E N Q U O T A I LIIN 261

. . - pretends that the context is ddrerent &or11 what it actually ts, the relevant

Luccumstance of evaluahon 1s t e n ~ p o r d y s h ~ f i ~ d the world of the context r10

longer 1s the actud world-it 1s l"eterqs dlcged behef-world. The analysis 1 have just gven must be reaibly quhfied Even though, m a

sense, the speaker pretends that he IS 13eter, sttU tile context-shlfi a#ects only the czrcumsmce of evaluation. Iri reply to (251, 1 m.ly sdy

(27) You're nght; 'my .ilster' rc llidcccl corntng

Here I myself 'pretend that 1 am Peter' 111 ~ I l r senre that 1 urltmii lurri by my improper use of the description, but the fmt person in my rrtterarrce itdl refers to mysex-it does not refer ro Peter This suggests that the coritext whlch shlfis here 1s what m Cjrapter ! refened to as the zllo~tstlonary

ont text. The speaker tempormly ~dop t s Peter's poirlt of vaew ~ r t d erlgdgrs in a form of pretence, but he does not pretend thar Peter ii; uttenllg the words, as happens in locuhonaq corrtext-shifts

Be that as it may, other 11on-cnmula~ve Iq4~nds wlrlch cannot be hantlled in terms of language sh~fi urn be a~l~dysed a\ instances of locutlonary context- \hlft. That rc the case m particular for an ~ntngiamg exanlple of mxcd quotation menhoned by Cappelerl arrd L epore (1997. 429)

(28) Mr Greenspan said he agrezci wth Labor Secretary K. U. Iieich 'con qurce a jot of things'. Their accord on this issuc., he said, has proved 'quite a surprise "ro

both of u s ' .

The word 'us' here refers ro Mr Greerispaz and Mr Keicli. I t ~loec riot refer to a group includ~ng the spe~ker of (28) . AS A result, tlle hybr~d is non- currtulative. Sentence (38) does xrot eritail (29)

(29) Mr Greer,.ipan s a d iic agreed wit11 I ahor Srcreurv 1X B Rerch o n ilrute a lot

of things. Theu accord ori tlui nsur, I.te sluci, has proved q~trtc s surprtse to

both of us

That ir so because (29) entail5 that, 'accordlrrg to C;reensparz, t l~e agEement reached by Greenbpari &ind Retch has pt ovcd ;l 5urpme to thr rpcakcr (I e. to

the person who reports C;reensp,~tl's ~rrtera~lce). 'Tbat en~tilnreiit n absent &om (28), since 'us\n (28) 1s quoted and ~nterpreted wrth respect lo the shifted context of the reported speech ep~sode. So it IS, agan, tile pre- semantlc process of contest-select~on ~4ucl r is afEected by the quc~tdt~c)~~ h the sh~fted context relevant to the uiterpretat~ori of the cltloted worda, the

Page 268: Recanati, François. Truth-Conditional Pragmatics

262 (-oN~:I I~SION

sycahcr i s Mi (;rc.crlsparl ilimscldf (rather. than dre persoti who reports his

rrtec:ranit~)~ so "1s' re i t rs to n group irrcl~idiiag (;reerrsp;ul rather &an to a

qi~tlj"irxclt~tling rile reporter. Caeb irr wltic.11 r i iv laxrguagtx s l l i i i s can thers~sclves be andysed as irlstarlces

of Xocutitrrr;~.; coritcxt-slrilt, at I~:LsI. it' wc ;i(.ct:pt the idea of treating the larlg~i~gr :is a j i"atii I-c of- tile C O I I ~ C X ~ ('is 1 /I;IVC "~~ggestcd a cotlple of tirnes). So wc call say tli;it, in the l1011--~~1rltiI:itiv~ cmliples dealt wit11 ill the

pcviorrs icctiori, rc is the 1;ulguaiye feature oC the curltext tl~at shifts. illhe sye;~ktr-r spe,rks a I-ert<tin I;inplage wl~ic.h is c:unstitutive of the context in wl-iich the rittrr.;rrii:e tnkes place; yet ii pot-tiorr ol'tlic utterance is interpreted

wit11 rcspec.t to a tiistirict context in wlriclr anvtl-ier 1aitgu;ige is used.

7. (:o~rclasiro-r~

In tlln ( Ir,iptthr I IMW ,irgtred (Collo\v~ng i ; ' l ~ k ) rll,tc- cluotntlotl 1s ,I fonn of t ~ c r r ~ c ~ x i b t r , r t x o r l rliL0ivl;lig 11!1g~ll'.tlc ~ l l d l ~ r i i ! <)i-ld iie~~i~~ll\hdtion, ~t h a P I L ~ O -

i r e t i t I I'kli\ nukes ijerotit;rutl, at bottorm, A para-

i i r~ggu~c t l t ~rkicritrxilc~rion, l ike gesrirnxig or rntr)rl,itlorr iuc-h y l~enorlterra play a (c..rltr d xidc rrr ilr\c trurst., so to s'ry that cjrlotauon rs, d l I>ott0111, paralxngu~st~c m n o \.c ,1v rrrr(terpiii\fs i t s ~crle or- rrtlport,rrtc c' rar Ix~lgx~.lr_s~~-rc p r a c t ~ ~ c . ~ ~

I,vthir thcilig;h, clalci denaoxxstrat~ons, cluotaatrlcjni have ptctor1a1 rather than Irngwitrr- rncaxilxrg, they iio a r c t Lrt~gnirrc rlle*trring r n several w;ly5. I havc

ciescmbcd die anq>,ict of yilotAtron 011 In~ha~\trc rnc'uirng at tour IcveLs. Fmt, the C X ~ E O C ~ O O P I ~ i ~ ~ r h : , tlierilcelvtu have IUIL~~IISCIC i~~eall~ng------tllcry convetntiondy rntiic at r the t rhlt the rpe&er 1s cier xlor~t~ ating the enclosed worcls. 'I hat iratllc ,ltitrri, lrke lltc rxieaillllg of i>rccgtihrtic irxdrc'tton rn general, n 'use- t orrchtior~~.' rather dr'u~ "trrrth c-on&tlolld7 (XXccaii,icl rc)c)8), xn more tradlaon,d tcnm, i t la,r\ dic \t,il;iis of J. 'con.i.mrror~d rrxlplrccx~ri~' Second, tile clenionsm- tion ( ,rrr yrc "I I I I ,WII I I , IU~ ,dI;'c t the tn~th c<rxlcIrannd content o f m uttem~ce by 41ltt1ng tire torltclt rn wlucll it, or part of it, rr rntciyreted. 'Tb~rd, what the

derxrtrur\tx-itrcrtl (vxrvcys m vmue of'rts p~ctund rncmrlg c3l.i be Incorporated

"" Not oi i iy tic> circEi plirr~ixnt:rra play a ~:crrtr;il rok: in kt~iguage ~.nct!, &.hey ~xiterac.cl wlth the liripisuc \ystcnl iucil- iri i i r ~ t~ ' t r )y ways that t11c jattcr cdnrlot IJC ~>ri>prrig dcscrrlx(j witl~clout &ing them into < ur~~,rdirntioii. Par e:i:urrpiv, the r ~ o ~ l c l r ~ oi' a ttr>nlrngu~urshc 'painting' can ItartlLy be ilispr~ised with rn r ~ l t ~ ~ ~ r l z l # l g " t i > O ~ t I ~ ~ ~ ~ ~ l ~ ~ l l 5 ~ 1 , ~ I ~ V ~ ~ ~ .

Page 269: Recanati, François. Truth-Conditional Pragmatics

illto the titterance7> tnlt11-concj1tlon.31 cortterlt--or ,*t lest, Into tlit* iontc~lt of.

the spe'ker's d%erbon- through the pmcers of praii;"x'xac clmchlrzex~t. f l hat I\ only to be expec ted, s111ce both the pi~tox~d ixlemirlg ofthe denlonsh;ttzort md the hngwist~ cxeamg ofthe senterlcc c ontr~bure to detemtrrnng the noxl-xldhxr'll memng of the tlttewrice, by nlalikstrr~g the ipe.&er's corrullurrrc&xtxvc

~rttentlons.) The tlzrec contr~bt~tlcms to Lmg~ust~c me,inlrlg I luvc just descr~beti

are sonlewhat pcnyheral. They are not cillect cox~trtbut~orts to the lrtcrd truth-conclitions of tile utterarm--to wltat 1 referred to ,~bove '1s the 'c cotitent' But clnot.ltlons can d ~ o , and often do, contribute ~Ilrectly ro

tmth-co~~cfitlor*d content. 'X'h~t happens whetlever '1 qucltntlon IS clo.w./, t11,rt IS, lulgu~sticaUy recrtr~tetf 2s A ungulclr tern? In the nleritionrng \entent e. Bec'rnre it f~~nct~orts a,; a r~gwlar ungul;it- terrll, both 5yrttactIc ally

2nd senlant~cally, clo4ecI cj~ot~~troli undoubtedly n ;I genrrrrte I~ilglrst~c

phenor~ienon. C:ioscd quotation as I lrdve criescnbed l t 1s orriy A specrdl c~sc , driti riry airn

was preciselv to chow that i t 15 only ,r spec~al caw Illat \pecr,i! cd\c I\ w h ~ IS

ilstidly cbi.lUed 'cj~~otat~or~' rn the pbtlosophicnl 11terdtrrre C $ f i t c1tlot.ltitJn lids beer] ignorccf s o far by phdosoptler~. Oile m~ght tl-t~nk t h ~ t one ( . ~ n 'ififord t o

1g1wre it, pre~~sely because ~t is not directly selt~v~u~t t o tlrc \yntactrco- senraritlc .itudy of language, which rs tlze cerltral concern of fclrn~dly rxxnded

pb~losoplzers. f3ut that rs slot nglzt. By igr~onng ope11 qctotatron, phlloso- phers have been lecf to colzinvt mstakes and overgertcr,1h7atio1~

Quotation, plrtlosophea 1zme said, 1s A device li)r tAc~~rg about word\; 1t7s a

r<kr~atr~al device. But quotatlun r~liarks are ~ i s o used for ciomg other thing,, Ilke

dntanc~rig one5elf &on1 '1 word or errpressrorr whxch one r\ rising. Accorchng tc)

the standard plulosoptl~cat vrew, there i c not ni~lth 111 corxinlon between

the two usm, hence cpotatlorr rnarlis dapl~y a €ofonl~ ttfnnlblputv wh~rlr Is to he d~spefied Cor the sake of cl'irity Il1u.l phdosopben otter) dMCIllPt to w e

cl~fiercrrt sort.$ of quo~itlon tnarks accorc%ng to the purpose at 11,rrid.~"

My oblectlon to tlxs approach IS twofold. First, I tlurrk we rnos a gencral1- iatron rf we pos~t axrl dnx~bzgu~ty. Clored quotst~o~j is only x spec1:il cace of J

amre general phenomenon, and scae quotulg 1s ailother, even more specla1 case. (It's A special case of open yuotat~on.) Secorld 'md initre uxlpott~nt, the

29 Daviiison rrco&mizes that this a an 'absurd .%aid unworkable forrrtttla' (197~: XI), but fctr the wrung reason. Sce fooirrure jo beIu>+

Page 270: Recanati, François. Truth-Conditional Pragmatics

-- - -- - standard approach n methodologcdy confused The distancmg eEect char- actensac of scare CJUOQII~ is the pol& of the demonstration in an unportant group of cases But there are other cases urhch are quite srrmlar on syntac- aco-semantic groulids-they ,ire also cases of open quotanon-but where the point of the detnonstrat~on 1s riot to distance oneself &om the displayed words In pmcular, there are instances of quotanon, like (31) below, whose Iriarn poirit ~c to a s~nbe the displayed wcrrds to someone whose speech the speaker 1s reporting, but whtch oygjzt to be cluss$ed as open rather than closed on ~. \ynta~t~~cr-~er~zd~~t~cgrounds, even though i t 1s closed qtlotanon that one typl- tally uses 1x1 d l re~ t speech report Thls shows that one should carefully distulbwish two Issues. a bas~c syntactlco-senian~c issue (Is the quotation open or closed, that is, 1s rt, or n ~t not, linguls~cally recnnted as 2 smgular term'), d11d a pragnatlc IssLx (What is the polnt of the demonstratton;). Those issues are hopelessly conhced In the standard approach.

U\mg the sirnple-minded dlstmctlon between 'genu~ne' quotanon

(where one tdks about words, and t l e quotacron funchons as a smplar tenxi) and scare cluotmg (where one achreves a dlrtmcmg effect, w t h no referentlal mtent), urie cannot account for exan~ples such as the followmg.

130) A 'lortrirght' is a penod of touneen davs (3 I ) To begn wth, I endone Evam's (r 982, ch I I) &sancaoil between prodgrcea

'anti corrszrmzrs in a proper-narnc-using practlce Cons~der 'an orhnary proper--.llau~e-us~ng pramce, In whch the name "NN" is used to refer to the prrsor~ x' Ttre producer, are the members of the 'core group oispeakers who have been introduced to the practtce ma the~r acquaintance wth x', that n. on the bass of learm~zg 'a truth which they could then express as ' This a NN", where "I hrs" inakes a demonstrative reference to x', m the ~ o ~ ~ t e x i of a crrtarn capa~ity to recowre persons, or at least: that pamcdar peron, crier tlriie (p 376) [rroln B Brewer, Percepttori lrnd Rea~on , Oxford: Clarer~dorl l ' ~ ess 1999 411

In ( j 1) the po~nt of dle three succesr~ve demonstrations ic clearly to ascnbe to Evanc t l ~ c h particular words that are enclosed wclun quotation marks The page u~ Bwrrs'r book where those passages occur u even mdcated Those yuotanons are therefore @raplatic&y) smlar to those that occur m dxect speech reports such as "age 321, Evans says "-'". (Cappelen and Ixpore cLa'~l% such ~dse\ a caes of 'dlrect quota~on', where we quote someone by nlentiomrtg the words he or she used ) In (30), we gve the mearung of the

Page 271: Recanati, François. Truth-Conditional Pragmatics

word 'fomght' T h 1s (p-aacdy) su~ular to w h ~ t Cay>pelen and I .epore call 'pure quowon' , where we ascribe properties to words (as when we 'Foortn;lght is an ~danxhax kvord') Despite die pragn~lbc s m h t r e s vvrtlr mstarices of either pure or dxect qL1otaacm, however, the quobnons nr (30)

and (31) are open rather than closed The yuut3aons x e not lmiguwhc,&y recrurted as smgular tern15 in contrast, both in 'Page 321, Ev,irls says ' '---' ' '9

and 111 'Foenliyht rs an u d ~ m h a r word", the quot;lt3011 LS closed. For lack of the dist.mctton between open and c1i)sed quot,iaon, ~jhrloso-

phen hke Davidsori or Cappelerr arid Lepore, confronted wldr examnpler slm11s;r to (31), have bee11 msled k ~ y the prag11'ttJc slndmhes Into thlnklng that such cases are cases of genuine quotatrtorr In thelr serise." Tl-hrs n'~~ura1lv led to the mistaken assuniphon that he re must be a singular ten11 ret'enrmg to the displayed words somewhere rn the seliterice

What I have just sad may sourld urkfa~r to I)avldson and <:ai>peler?i arid Lepore, since they do not &scuss cxalrlples 11ke (30) or (31) But I take ~t that

(33) and (34) below present the came chararlenstlc teamre\ as ((32), wluch 1s

extracted fiom (31):

( 3 2 ) 1 endorse Evaris's &slinctiol~ betweer1 produ~ers 311d C O P Z S H ~ ~ Y S j.11 a proper- name-using practice. 'The produi;crs ;ire the rrlr:rtibers oftlie 'core group r-bf

speakers who have becar introdclced to the practice via their acqrraint2nc;e with x'.

(33) According to Evans, the yrodtccers are the ~ncrnbers of the 'core grozlj, of speakers who have been introducrd to the practice via their acquartitance with x'.

(34) Evans says that the producrrs Lure tile mmben oftlie 'core group ofspcakers who have been irltrodticed to tlie practice via their acquainr*a~~ce with x'.

It follows from whkt D~vrdson ~nci Cappelen anif Lepore s ~ y *,xborrt inned quotatrori that 111 (34) the word\ 'core group o l speakers who lzave beerr introduced to the practlce vla thew acquwltarlce w t h x' are referred lo h't means of a singular tenn occurring n~ (34) By par~ty, the ~an ie thmg S ~ _ I C ) L I L ~ be saci about (32) and (33). In dl sucll cases. bv Igtlonrlg open qtrotdtion,

one IS led to posit a smgulnr tenn where there is none.

The pragmatic simidaritlcs bemeerr cases oi'clobcd ilt~otdrion (tvlrere the 3re retkrred a)) a~rd cases ofopen quotation where rhe potrrt oftlre dcnronstzihon 1s to cotrvey sometlli~lg ahour tile words IS

what seems to have convinced Davidson that the simple-ntmn<tcd dntinctiorl hemeen genuine cases of quoration in which we niention words and case5 in which we rr~erely scare-quote &ern 1s smwrtrkable.

Page 272: Recanati, François. Truth-Conditional Pragmatics

My own 'xtl,xPyui. ()I- c-lo\cd ijut~t~~bcm rs vex? S U I I I I ~ I ~ to DJVI~\OII'S, except I

that kc)r rrre, ~fi, kwr plays the role 02. the 5inp1Lr t imr rs he i3emonstratro11 itself, rrcat the quoutrora rarark What i olyei-t ttr 1s the terldency, ~rs~phclt m I>,\\rckorx and cpriirih t.-\plic~t I n i (,,ipj?ii~Xr~n dx~d 1 crpcti-e, to extend &'it anJyss

I I

to i ,ire\ x j i optm cjtlc>c,ierorx. hke (14) In (34)* the \pe&er chplays certam ofthc w o n t Exc rs u w i 4 wit11 r-he oven rrleerrtron I X I , I ~ the i~earer urider'ltnnd these words ,I\ ~ P Z I I T ~ S ~juored vcrbatln~ Crk.orrr Ffv,rn\'s I>ooL. I? ' tz~t mtent~c>i~ unclerlyung

~llcvpzaL~r'i cicrrson~tratio~t r\ ~"ti.iLy rct r9gnwahie, lrctice the speaker \uiceeds rn iorirrnuxii~dtllig: to the fie,rrer that the cvords wrtl~lr:, yuotahori nsdrks come fsclrt~ l \TAXI\'\ b)oojl Yet t h ~ t dipcv t of wtldt the uttcran~e ionulstmlcate, 1s not

part id it\ ~ - ~ o ~ i t e " ~ l i I he ascrrptrri of tllc cluotcd words to the reportee IS part of rkre: ( oxrtclxt tiic actrrfrclrz \vllrc"l dlc. sprcker T X I ; L ~ L ~ S , but thdt px t of the I OII~C"IIL of t l l i \ d \ \ i t t ~ o ~ l l'i s l o t ~ o ~ ~ ~ o " r f i u n ~ ~ l J y CJC~~ZIIIIII~CI hv the worcfs wtlrch thc LIW~ ?i-s ftorrr expltcrtlv tdL1trg <il>oirt the quoted words ~ ~ r d 'iscnbrr~g thr~ix to ~ h c r rp (~r te~ , the \ p ~ l k c r ~ C W S 1101. even refer to them; he drsy~i;cv\ tlicrrr, . \ a d his illtcrttrorr to aicnl-tr thrnr to tire reporter I\ rccop7n17eJ xlrcacEv by gr ryrltrg i i ie parxlt tri rl-ica rirrno~x~t~~~t,rtrc~x.~

f o l i i h br ,rix~ton, cv,vl;lo 5in111ar ly ~ c l v trc , t t i ~ rd a pragrrutic zppr0~~11 to

t~irxcil cjinot,rtiori," ( .ippt'lcrl ,ind X epore rrpircd tXut iuch '1 vlew 'under- idl,ryjs] irrijaora;anr irng~ur,tic , r q ~ c . t t b of the prxtlre of ~ilciirect reportma' (('appelcn ,IIPC~ / i p ~ ~ r i ~ 1()(j9 282) l311t ~t does not, or need not. One need

not dcriv r-h,~r p a t of wlt.3~ oxre \ ~ ~ c l n r c i l y 3x4wert~ by t~tterlllg a anlxed- cluotiirl:; \crxterrcc ri; that tlrt. reportrc LESLX! the C J L ~ O ~ C C ~ words; the only thug one hi, to i";l.rry ts t h t rhs .upet~ oi'thi" l i j tk i l t lve trutll-co~~drnond c O I X ~ ( " I ~ ~ of tllc IIC~CT;IIICC 1\ i o n ~ ~ ~ o s ~ ~ ~ o ~ ~ d l l y determlncd by the uttered

cvortis O r i c sinrply hcsf to apply to arrrxrd qrlotdtion the strdteg9 wwhh s ~onla l ror~ly c cptrd i i ~ r serlterrc-es X~kc (21) or (35)? c ~ t e d e~rher:

(at) I tc5 rook orat i l l s key arlct c-tpet~ed the i icros

(:35) I%.ccbr ;axid M ~ r y got ~l~xrrreci a r ~ d hai l 1rr:iny c.hildre11

In ( 3 51% the tetlrpor'd sugestlorr C l ~ k 1 : Vettxr dxrcl Mslly fiad ch11clrt.n after gettmg xrt rrnrii (r,izlrcr dr,~n the other w ~ y ro~1nc3) i\ rtot c o n ~ o s ~ t ~ o r r d y deteimlned

" "iri riiixeil qriai;riiori o r i c b "sirows" the i i i ig~iist: ic tools vihlcli wrrc* used by the reported speaker; i i ~ i i l rhrse "'ah~,~.ri.ziiitig ~ h c tIi(~w" ncqiurr be1lt.k about the hmx of spret.li exrrployed. l h r , to pnr.~plir;ae ID:~i-riisorr'i thorigha i ) r i riietaptic,r, i t - s at error ti?. Cisten olr the (cnr~te~lts of the thoughts a nlised ~ ] u o t a L i c i c i pftividLt:s, , i i x d i t ) ic,aci titi:\r i(,riti.ritc iaco the rlriarci quularlrrri ilself' (Swrtlton rr)c)u: 273).

mw-"s---

Page 273: Recanati, François. Truth-Conditional Pragmatics

by rile littered word%, it IIOC part ofthe U ~ ~ C M ~ I C I " ' \ c < otttent '' Yot one neccl

n o t deny tlrat tlin ~ I ~ ~ I I X ~ I C ~ugLYtlon I:, p x t .tnd p'ui el of wlx'lt 1s dswrtcd by lrttenrig ( 3 5 ) in din 3ppn)pn~-te ( onlext I he salrii. t f t~ng ( d11 bo SJIC~ ~ h o a t the urggesnon, rn ( r z ) , thdt the door w'u opened w ~ t h t l ~ c Lcy r r ~ ~ ~ l t ~ o n e d 111 tiic ftrst col~junt c 'The fact t h ~ t those \ug.ge:ec;trtr~is arc lr~tu~tively part anct parcel of

the tnitll-corrclrtlotw of rile Irtterance no objec t l o x l to a. pragrlmtlc treatment. onc o ~t 1s ac ccptetl th,rt the lrialltlvc truth-condrtion\ of LltterdilCr ,Ire sh,tped, to n. large extent, by pugmatit f~ctors '"?'

" i'his 1.; ;~itit~lly controvcnid. As I a1re:iciy rirerrtlrttied, tl~ert- are ct>rnptLc,iirg andiyscs ofthe cetnporal \uggest~xi ~onveyed by (3 3).

'' At tlus I J O L I I ~ a dcfrntier of the swrtciard &:count tilay rrpiy that In ( L A ) o r (33). the pr:i:;lnatii cii:iractcr ot tbr srrggesttuti i:.m he e~kildished by rts~trp, the c.arr<:eilab~Irty tezc: rinr- cari lit-vxse corttcsts lo wiuci~ (21) or (35) do r i ~ t CO~IVCY the relevalit iuggestrons. 13u~ tti's~rnc thi~lg (211 bc d,tor. wlih ~X~II IPICS lrki (31): One has orrly L(> ric1ai:irie a context in wk~ctr the interrra? large' oi'tirc: deinutistranon 15 distlncc fklm the reported iirter:rrice (exterrial target). (See iiiutnote 19 fibove, arid Chyte r 8 beiow.)

'" Ti~is cliapter, which brnetittcd fro~ri thsc~issioris xmtii tieniia1.i (hppelen, i<oi>y~r (;ars~on, I)ick C:.rner. iicricrit dc Corn~~iier , David KapLin, Stefarro i'recielll, I"u1ippe SI-hlenker, I-\an). S~ruth, Darr Spi:rl>i.r. R.ob Scalritor~. -l-itrr Willijlrnron, .tnd Dctrdre Wii?,on, a a srvi>cd versiori c ~ f tlic p:q)rr 1 pubinhi-d uritlcr the s.iitlr titic irr :blind I 10, -our: '''$7 -X/; ~ v t u c b paycr t)fft,ied ,* \on~cwb,~t slrclpiifled ~ c r ~ i i i n of ~ l l e story pt-esmtrd one vear earlier i r i rrxy hook (tr~itro Ohllr(>ri i , Omtio ilnru. Sirtc c the j q r r

WAS yubiistizii. a lot oi' work i1a5 bei.ii doric on the tupir aiupeti quutatii)n, arid irr i7.rnlcuirc a spet:ini issxrc vi'tlle l3?[fiur?,~ctarnui i$ l . .~ f z~~urs t?~$ tias sheen tiev<)ted ro i t . trridrr the M-I\? ed~torship crfl'hilrppe I.)? f+rabaritt:r. -0pcii Qtxowtxori I<cvmltrd' ((:li:ijlter 8 I~rlrrw) is trly rcli t1*,11 to tile articles <hsctissitig open quotmon "1 hat joicrr~d, .bird at thc canie tirm A rrtursi to he ~lighliy Irrclre 1,rtrnpiex stcrcy of Oruriu Ol)iiquu, Omt io Rtciiz. ((hi tile more cotnplcx s t i q , drr: i.r)r)tc.i~ .shift W ~ C I I ~har:1ctrn2c% ~CIWIC y~ot.itaons inust lie reflc.c~ed 111 rile relriantics; so i t rs not A pfrrcl)' pragzxlatic ]~flc'rromenon dtcr all.)

Page 274: Recanati, François. Truth-Conditional Pragmatics
Page 275: Recanati, François. Truth-Conditional Pragmatics

Open Quotation Revisited

I . The pragmatic vlow

Sonietlmes, by putting wor& wrtliln quotatloll marks, we bltrld an expres- sion whicli refers to the words wrthln the quotahon markc Example., a e

prov~ded m (I) and (2).

( I ) 'Very' n a four-letter word

(2) He sad '1 ain fed u p wit11 AU tth~s'. and slarrui?ed the door

Even though 'very' n an adverb ,11111 '1 2111 fed up w t h aL1 thi\' a i i " l ~ t e l ~ ~ t ~ . the expresslorn constructed by putnxlg therri w~tlx~n ~lji1otat1orl n~arX\S pIav t l ~ ~ gra~mnaucal role of a 'smg~ila_r tcnn' in tlrrc sentcncec ~n .cvhlch thaw quotations occur Becmse they go where singular tenns can go, qllotdtrorls

have been andlysed ac singu1,~r ter rxxs home theorists \re then1 AS a v.irrct\~ ol. proper names, others ar cfernonstrat~ves, still otl~ers as deiin~te dcscnptloi~s

In 'Open Quotahon' (Recanatx zoola) I have argued that rt 3 % a ranntxke to

build a general theory of quotatlorx oil such exaniple\, srrice they are not

representative of the erltlre category of ~ases Exa~nples (I) a x ~ d ( 2 ) ,ire

instarlces of 'closed' quotahun Closed quotations are slr~gular team rekr- nng to the enclosed matend, but not all quorxtions are like that In '~operr' cluotatlon, just as in closed quot,raori, worifs are ostcns~vely drsplayed, but theri syntactic a i d sea~~mtic type zemams ~lnd3ected-tl1e quo~~tioix 1s not .g-ATulla~idy recrdteLi &a 'a ,,illgd~Ax tc~l- , <:xjrppre (2) ('1

(3) 'I am fed up with aU t h s ' Don't you have anything mcrre constmctlve to say'

h (2) tlie words 'I am fed up ~.vitll all this'5J.l a slot m the sentence frarrte 'He sad -, and slamlled the iit>or9 ' f l ie qrrrotahorz serves as a grdmnlatlcal

Page 276: Recanati, François. Truth-Conditional Pragmatics

totyec " I o r d i e Clril~~lln1e verb 'ay' No \ucll ~17113g l~~lpprrrs 111 ( 3 ) Yet (2) dncl

(3) i ~ l v c s lot ir i (oralrrmn. LT~. E?;ocIL LJ \C\ rRic wolilr '1 ~ 1 1 1 fed ~ i p wltb ,111 ihl5' i 1 I

art- o s i c x i s i v c ~ l v disphycd (r~tdrer tj-ilrn rlsc"a1 in] thc" norriral way). (311 illy view. clc>sc~l c~tltri~~tli~rxi <art like operr cprc,tciirtrns, hut they arc disting~~ishctl by an

lrddiriorrtrl i>alurc: !.heir pra~rlrri~~ticdl enro!nleail ;is siiigdar terms designating

thc: sxrlcloscil ur,ita.rr;rb. Intcrrxsrii~g ti-~t>r~gh ir. i s , this ndditiorlal fcat~rce is too I

spec:'tic ti) hc rl-ic rvnrrcpiec-e in the gcrrerrrl tlixcory of cluotatictn, ;rnd more clttrritic>ri slxcitilii be paid to open i{llclratiotr tliari h;is beer) the case s o -fnr (or than h:td 1ictt11 the I::XSC w h e . ~ I wrote "Opcn Q~io ta t io t~ ' ; tor tltc sitt~atjorr

E.r:ls cii,rrigcc3, :is wirricssect by tile specnrl rssire which tbc Bc[qi~trt Jotrrlztzl of

I.irt'qg"isrii:s tievotcd to &is topic ;r few yi>,le-s .lgo). " A iitrsirii ijrrtrt:lr.ion rt:&rs t o the enc.ic>sccl expressiorr, or, i~iori: precisely,

t o soaars: type wiriih t i l e displ:iyeti tokrt, irrst,~nti,ltes. l'li;it i s the corrtcnt, or

scrii:ixitic. cor~trrl>~ltitrn, of rhe ciosecl cluot:aiurr. TIrc major diiEicirlty which ciosctl quot,ititrlr r;riscs fbr c-c>r~lposititrx,ai sernaxiric.~ is tlr;lt the contc~rt ot'tllc

earciostci wijrtis tlors i ro t sctBrrxl to bc rvlev,irtr to the content of tlic ~ L X O ~ : L -

r i o r i : iriticcii, evcii if'tisc erlcloscd wr~rds arc rirmrringless a n ~ i latrk contertt, the q i i c ~ t , i t i c ~ r will hcep its refGrenti;rl val~ie (EL will still refer to tile cncioseci words). Ycr ;ic.ct>rili~ig to i-o~nposirir)rr:ri c;crr~rtritics, tljc ec.intrnt of at1 es- pwsior~ d"ilclirnlt1 iirc ,i ii.ixlc.tiorl of t l~c~c t rn t c~ l i i s o l ' i t s 1r;irts. This s~~ggcsts tl-tnt

the crlc.losc,.d cxpresslott does Trot really ccuirsr as "p;s~-t' of the. qucttatiort iliiilt by earcitrsirig i t within qnor;ttiorr rnirrks. Willat docs that rr lear l exactly? I will n t r t br ci~ric.tsrrtcc1 with t h i s issui' ticre, but \wit11 a tiistinct issue

I I I V O / V ~ P I ~ open c l ~ ~ ~ i i r i o n arid itr conicrrt. Wli;ie rs i lrc. coxrtcxtt, or sernaiitic- cc>rlrribtitrt:m, o f z r r open quot;itiorl? In

c:orii.r:ixi. to cdc~sccl cjriotntiotn, an oprri LIIIO~:- :L~~<)XI docs 11~12 refix. A lingttistic exyrcssialrr rs irsrrrisively displ.iyc.cl or, as I sli;iii s~iy Gclril now ozi, 'deriloti-

strarcii', .ivitlrorlt i-iuirlg ling~iisric.ally rc.fe1-1-cii t o . S o what is the content of,

fix- c.x:il-ii~rlc, thr. oyjcrr cjtrotaticrri i r r (3):" Well, it is uncle;u that iirl opel-j

cj~rot-ation I r~s corikrrrt, or r~r;rkcs n senrariiic- c-irnrtnbutiort, in the way in

which cioqcd ijuotations do. 'L&/h<ir i s spe<:i'il nborrr (3)- ---what makes it a crrsc o f ipmtlltiori rnthtv cEran it rtorrllal trse of thc worcis '1 ;tin fed up with ;dl this' - i s ~rgilal-rly not a rnartcr t>f'senaat~tic. corrierlt. Wlletl>er X say "I arls fi-d up wid1 all this' seriously, by wlty of u~,lking a r i ;issertlon, or--as in (3)-

c~c-lroic;diy, to reprotlirce tllc previc>irs spe;rkcr's iitteu;lrlce, is a rnattcr of

I \r,e 6)r l3raha1lir.r '005

Page 277: Recanati, François. Truth-Conditional Pragmatics

hrce r'tthcr thm ii matter of content. It h,i\ lo do w ~ t l t wll'it we do ivxth

words, rather tb,ir~ wlth WII,IC the cvords rrle,rrt l>ot.c ~t lollow t l u t open (1110t,lt1011\ have 1 1 0 rne~tling over dnd bcyorrd tfie nreallltrgs ofthe dernorl \tmted word\; Not cltlite 'fhc s17e,&krr74 ~t c of O\LCIISIVC d15pl~y cltsarly rncans sometfiing 111 (3) it rne.a15 tlrat the spe,~her is echoing the prctlotts speaker's utterance I h c spe~ker doe< \o 111 older t o iot~vey scrrrlethritg 'ibout that uttetdnce, nartxrly that IL n not very ( on\tritchvc Hut nonc ot thls is cornposihon.1~1~ articulated ln the way In whrch r~onxrai \eolalltic contellt n ~~olnpositioxlnlly artl~ulated What 1 h ~ \ i e just described n, 114 1 \did, the

meaning ot the spzaker's act of o\tcnuvtx d~splay 'Tint ~ne,rrtrng is pragnnhc: kt I\ the xneaning of an ACC perbrrnetf by the spe,iker. r,~tfler t l l n r ~ tfic

sernantlc cct~iterit of an e\pre\slorl ~ttttrcti by rhe speahct \~ni~larly, the speaker*\ act ofrarvr~g hrh voice will, rrt Apprctpnatc cirt L~rrtstaxlccs, convey ,1

t m a n meanlrig Thir t;llie\ u\ to the llr,art of the differcrrc e i ~ c t w e e ~ ~ open yuotatlorl and tlowtl qrrotntion wlltle ,4 clo.ie(f quordtlon 14 elevated to the

\t;~txrs ofllngutstlc c onstltuent 2nd l z ~ \ a sserrza~it~~ cortterrt oflts own, ,In open

quotxtion 17 not ancl does not It] O ~ ~ I I q ~ l o t ~ ~ t ~ o ~ ~ tile speaker t-loes ,erne

thlng w t h the worti\, .ind tliereby riie,m\ \orneclilng. Wh,it the \peakthr nieans by osrel~snely d~sptay~~rg the wold\ shotilct he ~leal ly ci~st~rlgt~j\l~eJ

tram what the words thern\el\ies mc,-rn W11,ir tile vbords trie'tti I$ pretry d e t c m ~ i n ~ t e and cdn he consputetl hv appealrng to else serrlalitlc rule\ of the Ixlguage What the \peaker nlearl\ Larl oldv he lnlened context ,itld I J L ~ S the detcrnur~a~y of \enlailtlc content

Exanyle (3) rs ,ui ur\tanLe of what I b t t s (2007) lefen t o '15 L ~ - I U S ( I / (OPCII)

c]tmo"tlo But the rlloct 111tere5hr1g L,r\es ,ire the 'ssub-cl,t~~\,d' rr-rsr,lnte\, where the dcmonstratcd words occur loc,illy wlttrtrl 3erltenc es that a e riot, ~s a whole, demorntrmed. (I), (j), arld (7) belnw ,KC cx,is~~ples of.irth-cl,tu\nl open cluotaaol~ drscusred 111 "pen Quot.itron', wlille (6) cortte\ Gon? a book about Cli, i te~~lbnar~d I wll concerlmte on mch cases fi-orn 11ow on

(4) John is very '(coctl'. (3) 'The de~xxoratration provides the '~tlode of presentation' of the reEerei~t,

hence deterrriirles the contextual 'sense' of tile dm~orrstraiivc.

Page 278: Recanati, François. Truth-Conditional Pragmatics

(6) Chateaubnand returned to France in 1800, 'wlth the century'. (7) A 'fortnight' n a penod of fourteen days.

Given the locjl chmcter of the quotation m such cam, we e,xpect some mter,iction w ~ d t the procas of s e m n c Loniposltmn whch outr ~ t s the content of the sentence LII wlxch tfir quotmon occvrs Such mteracnon d be docu- mented m Sectlorn 1- j, yet, I w11 argue, we need not ~i&icantly depart 6om die pragniatic mew of open qtiotatlon sketched m the previous recaon.

In ex~rriples (4)-(7) the words wrthin quotation marks contribute in the norn~jl wdy to the 5errlalrtrc content of the sentence they occur in. at the sxme time, however, they are demon,trated, and thls contnbutcs an extra level of rrlearrrtlg that 1s clearly prapiatic. The question that anses for the interpreter w whv is the speaker drawmg attention to those words he 1s

uslngi If the t.pe~ker 15 reporting someone', speech, or has just done so, expticltly or even mtplicitly, then a hkely explanation for the demonstranon r \ that the spe'xkcr, by Aagg~ng those words, makes rnamfest that he is echonig the pcrsoxi IU question--usmg his or her very words "n such Lases I sav t h ~ t tile "oint' ofthe quoQtlorl is ascnptlond the speaker wants to inCor1r1 tile he,krt.r t h ~ t the person m question used those words Exan~ple (6) beloxlgr to th~s categorv by putting the wordr 'vvltlr the centuly' within yuotxclorr m~rtrks, the spe,iker irrlpties that Chateaubnand hmiseli; when dcscnbitig the eteilt of h ~ r return to France, used that phrase. In (5) the cluoted words are fa~nou~ly associated wlth frege, and the speaker, by flagging those words, presunlably rnterlds to mdke man~fest that he is usrng rregea1.t tern~mology. I Iere the pornt 1s not ascnptlonal the speaker doe., not lnter~d to 11iform the hearer that Frege used the words 'mode of presentnt~rtnythat is coiiunon kiiowledge): what the speaker intends to do, rather, n to warn the hearer that the terms ought to be taken in the techmcal sense they liave 1x1 t21e Fregean hterature, to evoke the proper background for the mteqretatson of the clams he is mahng by using them, and more generaiiy perhaps to appeal to Eregean audionty. in exampie (71, cne reason why the speaker hi&hgh:hts the word 'fortnlght' a that the sentence in which thrt word oLLun rnay be taken as a defirutroti of that word. St~ll, sentence (7) 3s not fornrally 'about wor&', but about thngs (namely, fortnights).

f &eely bomw some ternnulogy (but not necessarily the associated doctnnm) &om other people 'Flaaflrig' ronres &om 1)redeth (2003, zoos&), 'echo~ng' 6orn Sperber and Wdson (1981, 1986a), and 'u.itnq otlrrr people's words' from Scnbaj~ (2004)

Page 279: Recanati, François. Truth-Conditional Pragmatics

There is a daerence between tfidt selitence, where the woni 'iortnrght' is

used and the quotation open, arlci a nretal~nwntlc sentence like (81, rivlierls

the word is nienhoned rather than used ,md the quotatlon belongs to die

closed vanety

(8) 'Fortnight' designates a period of fourtee11 davs

FinAy there is the Lase of '\care cjuotes', illustrated by (4) the spe~ker

indicates that the words she 1s urtrrg xrr not her words, that she tr ~ O L I O W I X I ~

them frorn other people withctut filly erlt3orsing them Why the spe,ilier 1s

dista~lc~ng herself horn these words I\ anod~cr questlon t11~1r h a to be answered 111 order to fully grasp the rneanulg of the iluotdt~on Another, related questiori that h ~ 5 to be answered b c m urt tile tl,lture of the attl-

tude-playful, derisive, cntrcal, o r whatever ---wli~ch the speaker contcutu -

ally expresses towards the source of the use ,he 1s echo~ng

These are only a few typical elidlnple~ There rr t i o lrii~it to the nuxnbrr of more or less fine-gamed e;lcpi~n~t~orrs that r axz be iitund in rctt~tost br t11c

speaker's denionctratlon of sonle ot d ~ e word\ lie 1s us~nij: Nor Art. t l ~ c euplandtlons UI questlon cx~lus~ve olz,zi Ir ortier Corisiclcr vx.~rrq~le (91, and what I sa~d about it in 'Open Qtiot;ltlon'

(9) Iouri Skouratov, general prosrcutnr, was suspendrtl in March by Bons Eltsin. His successor, lo~iri Tcl~jiic~, was "proinoted' r~~ilristcr of j ~ x s t i i r in

August.

The quotatlon nzarE-s around )pmnioted convey IIWIIY thlrtg co the ~~i t~ 'q>reter

(I) that thx7 term w35 oBi~~all\i used, oi at Irxst, CEIJC loi~n l'ch,uLa'\ cll,ix~ge oi

affectanon was prcscbtztcd a pronlntlou, (11) that the speaker ( d ~ c rrewspaper

coltimnist) does not f d v ertdorse that descr~ptlor~, (in) that the reA5orr wllv he does not IS that ~t was not a real p ro i~~ot lo~~, hut rather a way ofgetting nd of Lrmn

Tcl~aka by 'hchng tcm? ups~lm' (I<ec,ulat~ zoora. 666)

The quotat~on here belongs hoth to tlre "a\rrt~~trotral' n~rd to the "care-

quotes' vanety.

m a t , then, is the meaning of a rub- clausd open cluo~tiou' l ' i rw qucs hon h a no simple answer, and we r~eed to draw a threefold dist~rlctrort

between

A. the meaning/content of tllc quoted words, i.e. t11e1r c o ~ ~ t r l b t ~ t ~ ~ n

to the meaning/content of the sentence they occur 1x1,

Page 280: Recanati, François. Truth-Conditional Pragmatics

li. tirc ccrrrvcritit>n;~l rtit.:lr~ic~g trl'the quot;ition lnarlrs (or of their corrn- tca.px~( / ~ r orai speei-hj;

i '. t!ic ic>sitcxtir;i\ tncarlirlg r>l 'k) i~ cluotaticli~r.

As f i r ,i.i A 1s corrccr~~ed, not rrrixi.ii ayp,~rciltly neecls to be said: as I pointed out , i i i (,I) (7) tbr, wortis keep their ncrrrrlal corrt-exlts :triil rrlake their rlornlal coritrii~eltior~. I,accu, I~owevi.r, we sirall sce tl-r:it i n a t least sonlt cases of sub- i.i:iii";Jj ( P ~ L * I \ 11irot itiilii, tkr rncrinirig/cox~te~it clr'rl-ie quoted ~vords is a&cteJ.

Ac ibu 13. t1rir.c arc two irrrptrrrarlt issi.icc to he c~dcfresscd:

* I'lt~sti-irsi Issrie I9e;1r.s or1 1.11~ n~rti~e~iticrttcd ~~~ i i ~z r i i i ! y ot'the quotation r~r:ks . Wl><nt I\ it cx.tc-tly? So CAI X E-rave suggvsrcd that the quotation marks ia~crc\y t3;ig r l i t s \vorcis as spcr:ci:ti :inid express thc speaker's inter~tion to

~ i r , i -~v riiu a~riiit~tlce's ;ittt.r1tior1 t o t .he r r r . Alvcrrrativciy, one 1ni-u). take the ( i>rrvc.ii~iori;al rrle;*riing of the qrrcir:ie;ror~ rrrarks to hi. rrlore specific than

iliiit. P I I , B ~ \ > C they ill(iic;att* s lo t o rdy t l ~ i t there is sorilettiing specid 'L , 1 >out d i e i ,~ ic- iostbt i wc)rtls I m t :L/SCI ~~.pllill i s sj)ei-WJ ' rb t~ut thein, t~;inrely, tirnt the \pe:~Rt~r- I\ risrrig ttl~crrl c.il~.oii-(~llli, SO as to replicate the words used by strrrii: o oiirc.stri:lily itlcntifi:ibli. source. Tlris dt,c1~1:~tive is stlggested by t i i c t.1c.i i h t t i le ~1uot:rtic)rrs in (4)--(6) arc ali c1c;rriy echoic, despite their tirffi~rt:rii.e>i I will c.trrrlc I ~ c k to tl?:~t issuc irelow. For the tirt~e. being, in ctrtiet not ro pr-e judge t h a t issue, 1 :issrrrric th;it for sornc relntion K bcrsvc~en tlic 5pe:iker 'inif tile words, the iltrot;ttiurl nlarks indicate t$r,it tIlc \pc;ihcr is 12-irig thr~ erlclored wortis (i.e. eatl~er denlonstrating

i l r t b r l i , or> 011 the 111o1.e sp~(- i / ic . ~ ~ ~ t i ~ l . p r e t ; ~ t i ~ > i i ~ ~~s i i l g them ecl~oically). * ( ~ ~ C C O I L ~ lsiiric: \&3t i s the S ( C ~ ~ E ! S U / ' C ~ : I C I . O I I V C \ ~ ) L I O ~ ~ ~ / nleanirrg? Does it,

o r ~ C L C * W ~t r i o t , bt4ol1~: to ' ~ t i : i t i s siid', to. die rcgu1;lr tn1tl-i-c-onditiortai ciriitcrir tri'tlic' ~rttcr;uice? It st:erlls tEut i t docs not. T t ~ c speaker who

titters :IILY of'(4) - (7) tlocs rroc sc2y rXut s / I ~ r : Ks tlre rniLoseJ words. ?'frat, r : i i l rcbr , ii, :I '(.oilvi.rltjot~:rl irtiplrc;irlirc' (or, pt:rh.ips. a presttp~fositiori- \ isc ircEov/). I r r 1ny -Cr;rrrrework ccrrivmtior~,~l iariplicatures are use- c.c~niiiticlit/ri c,orlxl>ont:llts o/'lltC:1*lirrg (1Ccc:nrr:lri r c ) ~ S : $4.). An expression c.oric-ey\ tac~-i~oritiit,i(-rr~:~l r-~lrariir~g if :rritl t r r ~ i y if i t is governect by a

c.oiic.iitionl of' rrse i r i sirdl 3 way tll:tr: lrsirrg thct expr-essiol~ irrlylies that

riiv ioirtliriorr is s'itistied. NCW t ~ ~ l ~ t ; t t j o ~ ~ I I I ~ I - ~ S hxve conditions ofust: t l r cy arc to bc' used only ifthe spc~kcr i s M-irrg (or i ~ ~ t e ~ z d s to be R-irig) rlic i~rrc.icisc.ii worcls. In virtue trfthis rrorrll:~civc corltlitiorr on the use of

Page 281: Recanati, François. Truth-Conditional Pragmatics

~LIOL. I~IOI I rliarks, ~ctctally ustrlg them conver~trc)rlcllly ~nrylrc ~ t e s that the speiker 1s K-lr~g the errclo4ed words

l-lr-i,tlly, C' n what we get wherr we ~ l l s w c ~ the. i l ~ i t - s t ~ o ~ ~ ~ ~ l l y 1s tile >peakcr rtoli~g t h (R-lng the er~clu\cd cvotcis)? Wlx~tcvet rearor1 can be i ontextu~lly inferred fbr the speaker's quo t a t~on~~ l bchav~ot~r 15 part of tlir < clntestual rne,lriir~g of the qrloratlon Thr5 15 opal-ended tl~ere I.; n o ll~rllt to the explan,itiorlr one tan gve , I r l tile \ert\e that one c,ui gct a\ fine-

g r ~ ~ n e d 2s one tvishcs In titis respect the ~nterk~ret,it~on of open iprotclttort looks niorr i ~ k c the ~ntc-rpret,~tton o i ~ l~tcrdry tcxt thnrt the ( on*putatroll ot

sern'irttlc content AH this \uggcst\ ,r muit~-dl~~lenrlotlal ,in~ly\r\ C: 15 a verv dli-i:I.rcxtt sort ctt

h e ~ < t than A It 1s prdgrn,itrc. open-ended, artd vdgue It 15 ,t rmtter OF ~1.1- thr i~p-c cmsl Jered intercsrc e, cvblle 're11iant3c c oilteiit i,tri he c onlpirte~t o n

the ba\rs of rules NOW 13 EL A nutter of iules, like A 11 1s what tlie cjtiot~tlon m.irLs ~ndlc,lre 111 v ~ r t ~ t e of. the (c~riventlos~ rh'rt: go\iern\ t h r use Yet B 2nd ,L,uguably ren3,un separate 01) the \ t , ~ ~ l d , ~ r ~ t , ~ l l t ~ l t ~ - ~ l r ~ r ~ t ~ r l \ i o r ~ a l applo,~clr~ to corjventlond ;llrr~pilc ~ t ~ t ~ e s (013ginat~11g lri rrege) J grverl \cntenctX rti.ty

express several proposltmil\, hierarzl~lc,illy ordered 1r1 terms ot centrality fhr truth cva lu~t~on . So ,L sertteltt e cctrrry~cliztzotz~~li~~ exj)rc.sv\ ,I given I>roposrtlorr, w h ~ c h 1s what has to be conurcfered m ~ ' V J I L ~ A ~ I I I ~ thC t ~ t t c r ~ ~ ~ ~ c e ,I\ tilrc or t,llse 111 'tddicron. IL iriay ~ct~- tur~Jr l~otz i t l l y o?Lprcrs fll~tfler prol>o\ititlris \~vllc~,e hcnrir~g 011 the truth or falaty of the irttcrlnce 13 lei5 ohv~our \ttcir A

mult~gdle-propc"t1t1011 fr'rrrlework ha\ bceli put fonvarti lor deallrtg w ~ t l ~ , Inter ~11'1, ( i ) l l v e t ~ t ~ ~ ~ ~ , t l I I I I ~ ~ I L Z ~ U ~ C S , arrd <evc*r,~l autl~itrs, ~tr<lcidlng 1ny5eIl; have suggested unng ~t for cie,~lrng with (opcrr) qttotatron ' O n the vcr<Iorr o t tlrls view 1 put forw'zrd irr C,h,~ptet 7, any ot (4)-(7) coingx)~~"tx~u11y expresser a ccltaln propo4rtion -the sdnlt.. rt -\?roulil et\plcs'\ ~v l tho i~ t the yuotat~on ruarks-'uld us(.-cor~tl~troricliy exprev,~.; 2 firtiler propo\itlo~r to

the e&c t t k ~ t t the speaker 1s R-rng the e11t loscd wolds I11 addlt1011 the uttt-r~nce pragrrratit ,illy convey\ nrr array u f p r o p o ~ t ~ o n d ~ a v ~ r ~ g to do ~v1t11 the speaker7\ p o ~ ~ r t in fi-tng the ertclosrd wolds

j 011 the rrlultrple-propos~tioz~s Cmncwo~-k, see flactl (r:)g<)), Neale (rt/t /t)) , Putr?, (zoos), arrrt>ng ochcn; on irs use in dedlng ul th (open) quoranon. rze ILzwnati (~oocib, Lcwra), I'lrdrilr (nooj), (;:trcr~

( : ~ r p ~ ~ t c r o (hcroj), Fiott.\ (,oo-;)

Page 282: Recanati, François. Truth-Conditional Pragmatics

3 - 1nter;ictiorr with semantic cont :nt: (I) free enrichnzent

Bart Geurts and Emar Ma~er have ob~ected to a multi-dmenstond andysls, ort the ground$ that it is only 'suitable for representing dl-trerent hnds of content that don't Interact too rnuch'((Geurts and Maier 2005. 113) There is a lot of iiiteract~oa between the utterance's regular truth-con&tiond con- tertt ~ n d what the quotation coritnbutes, tiiey claim, co it is a m~stake to po\it 'J atnct cbv~de' between thern (Geurts and Maier 3005 115) Now, 1 agree thdt the rneaning of art open quotation can agect the trutli-con&tions ot the LltteranLe in wl~ich it occurs, but I dlsagree that ths necessarily goes apmst a rliultl dlrnens~onal a~zalys~\.~ X also d~sagree with Potts, Cappelell arid I epore, L I I C ~ xri,Ariy other authors that suc 11 Interaction goes dglaSt a

yragng'l'at~~ treatment s u d ~ as I advocate (Given the context-sensitivitv of tn~tfl-coritiiitiotlala Lontetlt, Interaction w~tti the pragmatic meanlng of the qu(3tation n pred~ctable, since the acbon of quotmg takes place in the very context with recpecr to which the sentence rs asslgned ~ t s truth condihons

I here are two broad types o f interaction between semantlc content and q~iotanond ~tlreaning la the .sort of c,uer at Issue. In th~r sectron I deal on15 filth tlie first onc, which ~nvolves the ~nechanism of '&ee ennchment' 7 hroug11 tllin rnrc hanism, a11 expression IS glven a more specific mterpret;r- t1011, 111 context, than its serndrltic content stnctly hcenser (Recmati 2004).

If, m (lest rrblng a traffic-regulation scene, I say 'The pohceinan stopped the car', we niitx~r~lly understand this as meaning that the pohceman caused the car to \top by rgnulbr~q to the ifnver t h ~ t Cze wus to stop, thereby rnaklrzg htrn or her do so Brit the senterl~e itself says nothlng about the means used by the pohcerr~art (it does not even Fay whether the stopping of the car was ;un

mtent~onal action or1 tlie part of the pohceman) The ~tahctzed part in the above cle\~nptioxi correspond:, to an aspect of the Interpretation that is ~ur~iirkiutrd by iLe l r e d ~ r ' s C X ~ C L L A O J ~ S gii.cii t i t s iii her bac'rigro~tiid

kr~owledge S ~ r ~ c e d1e speaker, who knows what those expectdtxons are (and knows that the hearer knows that he knows), has done noth~ng to

C>fro~me, it depetids or, what exactly one means by t h . G e m and Maier's main target is Potts (aoo7), and it 15 not nuy imentiorl to defend Porn's account against their criticisnl. it may be that Geutts and Maier, whose view is rrrnilar to niinc in rnany respects, w-ould have 110 objection to the 'multi-level' analysis I defend in this chapter. Yet they clainl that their criticistx~ applies to all versions of multi- dimensionalism, and rhey cite me as a would-be multi-dimertsionafist.

Page 283: Recanati, François. Truth-Conditional Pragmatics

prevent the hearer honl lriterplehtlg the utterance in thls way, he clearly mphes that t h ~ s n the way the ~>ohccrnm \topped the cdr Now thrs mlphcature is mtegrated with the serildrltlc content of tlie utterance rn such a way that they are ~ i o t tntult~vely disangtztched the InRntIve truth-

conditions of the utterance incorpordte the ~mphcatme Why 1s there yra'qmattc ~ntrwtort---Chdt I), ~ncoqjoraaon of the ~i i~pl~cature

rnto 'what is said"ir1 this type of case but not ln others?$ Prenm~ably, because the irnphcature further .iyec~fies the event which tlxe $entcncc

descnbes in virtue of I& serrl'iritic ont tent The l ~ x l p l ~ ~ a t ~ ~ r e provrdel an extra argument (the relevant 'way of ctoppmg') in additloll to the asgunicnt\ that are tinguistlcally articulated (the agent and the theme of the stopplrrg event). Whenever an anphcature overlap.; with the seniantlt conteat of the utterance m h s way by provldrrig Fbrtl~er s p e ~ d i ~ ~ a o ~ i s of the deqcnbed event, it tends to get incorporated i~ l iu the uttermce's intuxt~ve ~ ~ ~ i 1 1 - ~onditions, mstead of remalnlrig rntult~vely sepaate a t v h e ~ ~ ~ t s contnl-in tlon is orthogonal to semantic content

In some cases, sindarly, the n~eanlag (>i- an open yuot%tlon can ea\*ly bc mtegrated wth the sernantlc conterrt OF the utteratice, bec~usr thev overlap Con~ider (6). The quotation is e~horc , I iald, m d IL\ polrlt is ascr~ptror~al r t

contextually means that Cliateaubnzrnd, nl de\cnb~ng the event of 111s lerrrrre to France, used those very word\ jxvrth the ce~itury') Ths (Chatem- briand's usmg the words k t h tllc centurv') n a ci-r\tmct event trorn &at

wllich the main propositlor1 deccnbes (Chateaubnar~ J's retl*ri>iiIg to Frarrcc in 1800, w ~ t h the century), so the 'overldp' ~orldltlon 1s nor sntafird But let us change the example a bit

i (10) In A%f6inor~es d'outre-tomhe, (Il-latearrhriand wrvtc that he retririled to Crnncts n t< in 1800, 'mth the century' r

Here the rnasn proposition descnbes rhc cverst of Chate,~nbria~td's Aiaystg

~omet/zzng about his returrz to Frmzce The Iilezrning of the cluot~tiorx- - \d~srch arguably does not change--carr easily be mtegrxed wlth that propoarrorr unce it is about the same evmt, wh1~1l it iiirther specifies by p~ovrdrng an

~d&aonaI argument to the "sayzng" relrltaon (vi* the wordc that Ch'rte~u- bnand used in saying what he w d ) As ,I result of thls xrltegatlon, 'tn ordmary user of the language dskecl about the truth-corid~t~on\ of (rcs) rs

k The phrase 'pqmatlt- mtna\ion' cornrs iron1 Lrvmson (zoooi

t

Page 284: Recanati, François. Truth-Conditional Pragmatics

2 7~ IN i I 11: AC J ION WEFT 51'binlirir'tc c(-wrr Nr- ( a ) 1-11 FF r~tii(-t(~t. N i

jikclc 10 ~ c \ p i ~ a ~ t l tlr.ir t t l t= trttcr,rncc rr txiic tf axrci only II, 1x1 ,I.lirizorrtp~ d 'o~ t t r r -

ioniirt, (1) C Bl,rte,rrnl>r~arrd wrtrtc that hc rctorsried to Fr~rrce m [goo, and (11)

hc ti\uiI the" wtrnh 'with rlrr century' tar (1e"rarlhr the trrnlng of h14 return. i i r i s / rrcot , i t loi l , 111 euarnple (ioj, cledl-iy t ontribules t o ttle tmtll-concjlhons

of the iittt.rii-it c Iloes r t tctjlow thdt (1110t.xt1011,d ixied~~l~ig aiid sen?antlc iositrrit tlo ncr t E)c/ong to SCP,IT.~~C c i~tr iel~\~i)~~"r 1 tit> not thlrlk so. I thznk we c'irr rii,ilrut,iiaa tlr,rr r1-r~. cor~trxtu,ll xxrednlxlg of tiit. quotatxon (to the eEect that (;jiattc ilil)rr,~x-eii u.icii tX-losr vcry word?) helorrg to a clrgercnt cbn~ension than t i i t * *cail,isrtrc ioaitrirt of tllr crtlerdllce, rf w,vc <rc'cizpt t h t an utterance', rlrttritrvr iirrt1-a coiitJirtivnr cdn go htyrrxiil w h ~ r \emarrtlr content p~ovides A\ ckrc j i c t l i c carxi,ira e~arrrple h ~ 4 \Lzciwrl, tlie i r t e rd truth cundit~ons of an trttca,trrc u G a r I ~ c pr"rg~:z'it~cdly tsnr1checl with iurtlle~ spec~ficdtiolts of the tlr*i s rlrrJ cvexit ~ h i c l t the ori it cut, o r gent-rai world-hnocvledgc, makes l\iar lalde c c er i tbcttrglr thcy ,ire I K B C Irngr~r\~ic,illy crrcoded A11othi.r well- hr ior \ r i i.x,\onpEc of sfl,rt \tnt of thlrrg wlrrilr 1 h,we ~ l~en t~or i ed alreacty 15

1~ rr,iiirr 11 to rritcrprct thcs wionil i c r x l j r i r i l t ,is nie,ulirlg t h ~ t the pels011 acfcaaid i t r h)v "slit ' opcrlcd the ~Xoor wdx thr kcby ~ ~ s e t ~ t ~ o r r e d 1n the first coxrgiirit t, 'i ct nhe scbnterxce doth\ not 11tlit~~iIy 3dy dl~yt_bmj: &out dle tnstru- xriene uwd i l i opcnrrrg the door I l ~ c fiisthcx y)etdxcatloil of the opernng i.\it,rit rlrstrragh the < otltextual I ) ~ ~ V I $ I C ) I L of tlldr rxlrtrrrrxterlc 3e5ults fkorn free cnrlt hxxieirr 1 111s 15 lake an ordln'iry ctrrrverrsxtrnr~~l ~mpltc-ature, hut ~n t h t

4011. o f ( d'b~" the ~ r ~ ~ p h c ' t t ~ ~ r e ovt:r? ldp\ w ~ t h the IiidlIl $>ropo~itfoli expzened by tlrc iec irulcl r oxljurrt t (th'tt: she c.il)cbnctl tile door) since ~t ii~rthcr specrfies the csvc.xrt ot opcSniirly, tlic door hy provitl~rrg ,in .iddltronal argurlrent (the key

xncxrrtic>ricit lii tlic firit coxljurlt t) I'he rnxpb~c anrrc tlr~rel-ole tcnds to Intrude into t l ic E I ~ V C I J I I C C ' I ~ ~ i t t i ~ t i v e tnjah cuntlitiorlal content the httcr 1s p r q - n ~ , i t ~ c .ill\ CI r r i i heel t h ~ oilgh IIIL orpor~h( XI c>i the rir~pl~( ,lted conzportent.

Sor~rc~tlirrrg i~rnri.ir ,rrgunbly happer13 rn (10) P hc irlturtrvc truth-corrd~ tron,il i trrrrcirt of tllc uner'xrrt c n crsncl~ctl 111rolrgI:li the urt orporatlon of the anedrrrrrg of tlir cluot,iklorl. Uut tlus possrbrlity docr not Argue against 'i nlultl- dmlcraxcarr,~l ttcatniunt 'l11c hce t1-rat the rlrcarlrrrg of the cpwtatlon can be xrrc uq,ob,rtecl illto turtll-t ondrtronal c onteni tlrrough free enrrcllment does not \li,oc\ r8ictc i t I\ xlot '5ep;rrut.' froxlr \errl,tritct content If lt ~hd, then the f d c t that d [ m i e of (oritrxtudl rnfc>rt~l,thtrri ox \kcrrld kilo\.vlecige, or a conver- \nut r r x , ~ i i~npiii ,icu r e i ,r n be irliix,u;r 'itecl a r a t o tlre ~axt~iitrvc m t h -con~l~tion,d

Page 285: Recanati, François. Truth-Conditional Pragmatics

content ofall utternnce throirg1-r free crlncfu~icrlt wot~fct vrlrri,uly show t h ~ t

that plece ot ~ n i o m a t ~ o n 1s not sepalate from senl.~rttlc c trrltertt --,in ;rltsurci cortcluston One may deny that tree rnrrtl-rlllent exrsts, I ~ u t tf i t c\~cts, tl-reil ~t prov~des an eq>Lnatloir of the Inter'kitrotr of scrrlarrtlc content ,uid

quotat1oxr.d rncartlng ui (lo) that 13 fully contp~trble botll wrtb ~ r ~ ~ r l t r dimeliuon~lisrn knrf wit11 a pr'agrnatlc nppro.iclr to o p ~ ~ n qiiotatlori

4. Interaction with ser~~azitic cotlterlt: (2) context-sflik

I'fre scc ond fcjrril o i Interac t ~ o ~ betwitclt quotatlorlnl IiIeanm;: ,lnd ,crn.intlc c o ~ ~ t c n t 1 lilexltron 111 'Open Q i ~ o t ~ t l o t ~ ' rnvctlve\ the pl~t~nontct~oxl of ~ o n f c n t-sh$ W e l-rave sect1 rlrat 111 ril,tny c aw\, open qtiot,ltlon 15 rcl-rorc tl-re speakel's u\e 05 the clemon~tr'ttecl wor ch 15 Incartt to ct h o tllt wortis uwtl

by sonleone else (the lntrrrtnl target of tlrc i juot~t~ori) I ct 11s cIrLiw 4

i ln t~nc t~on between the r u ~ r ~ ~ ~ ' l c f i o ~ t t ~ ~ v t - - t h ~ ( ontext 111 w h ~ c h the tjuotatron,rl (iellr~onstr,rt~on OLC t~rs--~ir~ti the \ t ) ~ ~ i t ~ i c l f t t ~ ~ f (th(- c~r~t i t ' r t of the o i -~g l~~ i l L I W

wh~ch the yLlotatrwlr 17 zilednt t o echo) ~ ~ L I I O I I ~ I I the \ d ~ i i c * W O ~ C ~ S 'ire LIIC'C~ 111

the current context ;L\ tri the iourt e cctrtte.*t, they rnay carrv tilift-lent

contents with respect to the two c o n t e ~ ~ ~ I I C I C will be s ~ i ~ l r ~1 poti'l~t~al

dlvergerlce in coritcrlt wherte%er the word\ 111 cltiestlcln r r i c lude .HI 1nt1e.ctc ,ll expresnon whlch talncc ddkrent v-xlue\ m the two contexts, or t.vhcnever the worck nt muc happen to not nlem the sanie tlilrlg i r t the two contexts. ( ThlnL of the c s e m \vblch a Rnt~rlr spc"~Ler eclioes. an Axl-rerrc,~tt's me of an thglnh word ~vhic h means drkkrent. thing, Irr lirtt~sh IInghsh arlti A~xlcr~t ax1 I uglidr.) Now, rfthc speaker IS r ecagr~~aeil ,a mtentior-rally ec Ilorng tfle \ourte's usr of

the snrne word, ~t \vlU be rl,~tur,~l, 111 VIC i t CLSCS, to ~arl \ t rue the q)e~her a) deliberately urmg the dcrnonrt_mted Lvurdc wztfz thr rrrc~tzfflrg/rot1t1~1~t tlrty htrd t17

tizc sozsrcc conirtxr. r'tther than with the niemrng/c orlteat they wo-ulti L ' L I I ~ ~ I~,ld they bcerl used 'plalnly' (non echorcally) rn tlrc c urrexlt I oxitcxt

An example frorii Cltaptcr 7 I r

(12) f a ~ l l says he's due to present lus work 111 t.he "pi~per session'

where tllc speaker n untieritood as rnrn~icklng IQ111d7s deviant ubc of the phra~e

'paper session'. Wbat Paul mea~lt when he sa1c1 what ( I L ) repom w,ls t11at 11e was due to prewnt 111s work In the postPr cemor] ILOII IL~U~ echolog t'nnl's

Page 286: Recanati, François. Truth-Conditional Pragmatics

280 INTEILACTION WITH IEMANTIC CC)NTENTT-@J Z O m X F S R ' I r - - - -

nustaken use, the 5peaker ascnbes to h m referexice to the poster session under the wrong narrre. Here what the words 'paper session' contribute to semanhc content is not their regular semanhc value, but rather the semnhc value they have In Pa~h's i&olect (gven that he uses 'paper' to mean 'poster'). In such cases senlantlc content Ir af5ected by the quotdtiond demonstratson. S W l y , when the quoted words ir~clude an lncle~cal esprek3lon, that expression inay have to be rrtterpretcd with respect to the source context rather than the current context, as in this example fro~xi Chapter 6:

(13) Lev] Ectster, lrr fkt, I\ the @eat-@cat-gandfatflef of Gov. I k e Foster of Loruslam, whit sad recently on a ra&o program t at 1t would be 'news to me' if uiyone m h s f a d y h ~ d owned d me$

7 he first-person a~cusative 'me' w ~ t h n the quoted matenal refers to the speaker m the source context, namely M&e Foster, rather than to the speaker 1n the current context, namely the utterer of ( 1 3 ) . ~

In the standard, Kaplaruai~ sense of 'context', only (13) Involves a con- text-sh~fi Example (12) involves what I dubbed a lagquage-shift. the words with111 the yuot&~tlon rnarks are interpreted as b e l o n ~ n g to the 'language' (idiole~t) of the source, and thls aEect\ not only their content but also the~r lrngu~st~c mt.ming or character Yet, rts I pointed out ln several places, the rvvo p i i e n o r r ~ u ~ car1 be uiufied if-we let the language spoken m a context be one of tlte c oor&nates ofthe context m queqtlon. In this framework, which I adopt in what follows, both (12) md (13) involve a context-shft.

In (12) and (13) the seriwnhc ~ortterlt ofthe sentence rs not the same as what we get when we reniove the cluorat~or~ m a h . So it seems that the cfuotahon ( ~ n m b ~ l t e s to 5ema~tic content in such Lases Yet a &stuictmn can be drawn between srra?igi~tforwarcIly cuntnbutzn'q to semantlc content, and merely havlrzg art

gflect out seabrirttc ont tent. Open quotdhon has an obvious eLfect on sernanhc content m c35e5 l&e (12) and ( r j ) , but that e&ct can be decnbed aspre-semuntac. It need not be taken as 'i genume 'contribution to' senlanhc content.

1 et LE\ G ~ s ~ i ~ i ~ ~ e , w ~ t h Davd Kaplan, that semantIc3 maps sentence-coritext pars to co~itents tension^) Both 'sentences' and 'contexts' are tf~eoretrcal construi ts mtended to track relevant properties ofutterances and die sltuahons

The smlr bhfi & o r ~ ~ current context to source coritext m interprehng an lndextcal untlun an echorc yrromuon can be obsened wth both clausd open quotation and closed quotation In examples ( 2 ) and (3), arguahlv. die quoted seiiterrce 'I a m fed up wth all ths' n to be construed as echoing die source's use ot the unle word.;, and as remnirtg the corltent a had when it was uttered by the source

Page 287: Recanati, François. Truth-Conditional Pragmatics

-. 0l"N < > U 0 1 A? I O N R T V I S I T L I ) 281

they are produced m. As Predelll ( 2 0 0 j r ~ ) .forceMly pomtetl out, bebre semanocs can apply to dehver the content ot a pxtxcular unerance, a Y ~ ~ Y L >\enla-

tzonal problem anses one must cieclde whch (abqtract) "entente' wdl d e - yuately represent the concrete utteraxe that was mnade, and winch (absh';tract) 'context' IS relevant to the lxiterpretalclon of that sentence. These represents-

aonal problerm are not tnvld Moct h r ~ ~ g ~ ~ l ~ t ~ c S L ~ I ~ S are a n ~ b i g u o ~ ~ ~ and a

decaion has to be made, oil pragluhc j q o u ~ i ~ . to declde whclr sellten~e 1\

actually uttered arnong several competittjrs cornpat~ble w t h tlic perceptrble ugnd (As Bx-HiLIel(ir3 54) er~~phlls~zed, u1 ail cases a pragrnatlc tleclson ha& to

be made to d e t e m e the language to wluclz the uttered exprcssictn belong ) Sometrmes, PredeUi (20ojb) suggests, i t m ~ y ever1 be aclwsable to asslgz1 drc perceptrble signal to a sentence that 1s not tonmdly conipatrl-tle wrth. it, IE tho speaker's performance happens to be taulr-y but 111\ or licr llngulrt~c intentrcrn a

s ~ ~ c i e n t l y manifest to deten~i~ne whrch cuyresslori sihe mterded (but &led) to token.' Be that as it may, tlse repre\entatmnd problern bas to be solved hgove

semnhcs can apply to a gven utterance Sim111xly, behre \emantrcs can apply, one has to decide wh~ch context the \entetice count< a uttered ln 'Tlu< rn*iv sound tnmal but it is not. the relevant conte7ct need not b t the srtuahon m which the utterance is actuaflv nude Irr Chqter 0, I gmo dze following exmple (&om Kecanan 2ooob 271)

(14) It's been three vears smce w e lclr the Earth A couple ofweeks Bter the Last Dav, we lost track of the other \pacesli~ps 1 d l don't know whdt happened to nlv twln blother Wenrv If he r\ ~ l rve , he probahlv thtrtkb I cired in the

colllsloll

Let's lnlagine d ~ a t ~ h l s 15 the ftr\t parLag.y>l~ of a novel What rs the context

Ibr those sentences? A tint c,indidate n the .~ctrial\ltuatlon of utterance the novehst wntes those sentence? at the 1)eglnnlltg of her novel. Yer the word 'I', in the third sentence, doe., not denote tlse novel~ct; mthe-r, it purport.$ to denote a character tn the t-tovel-- the su ceded 'narrator' '%he novelist pretends that (14) is uttered bv that chmcter, and IS thdt pretend c o n t c ~ l ,

rather than the L I ~ Z ~ R ~ L O I Z ~ C X ~ , whicI115 relevant for ititerpretlng the serrterries in (14). In the pretend context, the speaker (I e the narrator, cirstmct fwnr

the novehst) is on board a spaceship, he or she has a twin brother called

k ' Thus when George Bush, wllose collu~l~mnd ol English 1% less dun pertec t, ir~cs the ~rorr-wold

rublunmable', r is clear rh i t he inten& to hr uvrig be inghh ward 's.obimunli.

Page 288: Recanati, François. Truth-Conditional Pragmatics
Page 289: Recanati, François. Truth-Conditional Pragmatics

1x1, we woil't get the nght content, sulce (becduse octfie sub-claus,d context

shrft) we need tlrio cotztexts rather thm one. 1 o uzterpret (12) or ( r j ) , we need both the Lunent context and the sorirce corltex* S o the represenc;lnoxlal p~oblern (In w h ~ h c ~ ~ i t e x t arc we to ~ntrrpret the sexltertce?) cannot be 5olved for tlie sentellce a n wltole. We m ~ ~ y , oof coune, revice the Irr*unewc>rl\

a id decide, for exx*nple, that o~dy sunple expre\slorir \ a l l he &\slgned ctlrtr- acters. for nture ~0111plex expresslon"r~lke sctltenc n, we wrll ti~rectly cor-trpose tlre co~~tents detcmur-ted by the ~hari~dcte17~ of the pait!! 1x1 the11 re\yechve contexts, ~rrste~d of fint cornyosmg tlte ch~rat ters of tllz \crltel~tl,tl pasts to get the clrrx~cter ot the centertee, arid then determlnlng the toirtent of ~tle

sencencc by applymg t h t character to 'the' context o f tile ser~terlce (See JSmg m d Srmky (zoos) for the sugqestion that we don't liec3 clwn~ten ti,r elmre sentences.) llut si~pyosc we WAI% to \t~ck to thc ~~rrzevrsed K~~?lalutu~ f~a.t~iework. what cm we do;

, ? I be ohvxous ,olutlort. cori\lsts in sernarltr~1;tn,o the context-\hltt and makrng ~ t , rts rt were, ~nternd to ttic c llarncter of the seriterlcc 1111s we can do by asslgiurlg to tllr sub cld~~s,*l cjuotdhon ,I naetc~livrqlczttzc L ~ L ~ M ~ ~ C V ,

wli~cb 111aps the context in W ~ I L ~ tlie st~b-cla~ii~f quot.it~o~l C U J S jvi/ the

current context) to rile content expre\ced Ity the criclosed cxpres\iorl when rnterpreted m the source context. C l i l thrs vtcw the cluotatlon m,rrb fuunct~o~r as '1 conttxt sfrlftlng operdtor cl rhat operator corrtbines w ~ t h an

expresslor1 n (the expresslor1 w ~ t h die cjuotatlon rn,rrhr) to yield An cxpressron of the \&me type, and ~ h ~ f t \ the context for the Interpretatton of cr horn the current context I to tl-te wurce context c' So, In the current context, the chxrL~cter of dn determ~rlcs the srtme corlteilt ,is the cl.tardcter of o In the source coxltcxt. 7 hrot~gh th r~ ~einautlcl7~hon of thc coiltext-sf~lft,

whicli 1s now hullt Into tlze ~ z ~ c t d ~ t ~ g u n t ~ c c h,mcter of' tlie yuotaaot-t, we cat1 Interpret the ceiltexlce with reyect to A slrlgle conteut (the cunent contest) axlet get the nght results: I'ur the expresuon dlr which occurs rrt

the sentence 17 cuch that ow11 Ixlteqretahorx In tjzc cttnetzt cm~lext yroc eeck

vla tire nltcrprctatlorl o f u an the sorrrte c o n l a t f have no objet tion to ruch arz analysis, wlrrch 1 aryseK put fimvarci 111

OIZJ~EO Ol~hqua, Cjrutto Krcfa (Re~artdt~ 2000b c-kt. 17) ' BLI~ I w ~ ~ l t to qtral~f\r

tire conclua~on that the contr~buhon of the quotittlon 1s tlxereby \emantr~rzed'

'The context-sh~fi~ng anaiys~s has its roots 111 Kecanat~ (1997) (wht~e I use the 'ilefirt?rrtd opemior'): bee also Benhzi (2004) ior a r ~ analogoi~b proporal.

Page 290: Recanati, François. Truth-Conditional Pragmatics

284 tc~iorcr-r Y

In one sense it 1s-the context-skrft IS, d e e d , bulk mto the character of the sub-clau\al quotabon-but m mother sense it need not be We cm accept that the sub-clausal quotation l~as a netalurgwsnc character and can be formally represented as da, where d is a context-shlfisng operator. But we need not c on\ider die quotahon marks ul the object-language as themselves such at1 operator Rather, we can mantan what I said earher about them: that thetr only semanhc coritnbuhon n a conventional imphcature, to the egect that the speakcr rs K-lng the enclosed words. In the relevant contexts, the 5pcaker's R-iiig the erlclosed worS ruggests that a context-shlfi occurs ,it the boundary marked by the quotes; and such an intra-clausal context- shrk has a rern'illtrc eBer t that 1s best captured (wthln the standard Kapla- nian kamework) by posrtsng a context-shlftlng operator in the abstract fonnuld whrch represents the uttermce ~ n d serves as one of the Inputs to the scrnarlhc ~naclilllery On this view, the context-shlf~ng operator need not have any syrltact~c reahzatlon In the hject-language. the theonst uses it to persp~cuously represent the context-shfing effect of the yuotatlonal demonstrahon 'The sugges~oi~ 1s that, whenever an intra-sentenha1 context shrfi occur\, whether or not t t zs converzt~orzally stgncdled by a speczalzzed hngui~txc dcvicc, ,i (o~ltext-sh~ftl~~g operator 113s to be pos~ted that 'represents'this context chi& arid makes it visible to the semantic^.^ T h ~ s leaves +AVO ophons open w ~ t h respect to the quotat~on marks* either we treat them as the object l~ngtrage realirat~on of the c ontext-shlfting operator and gve then1 the ~omspoiidmg semdntics, or me simply corlslder that they signal the fact t h ~ t the speaker 15 K-lng the enclosed words, which tact in turn suggests that A i013text sftiit occurs arlci n1ale.i insertlor1 of a covert operator d appropndte ln the fi)r111al represerltahorl of the utterance. In Oratto Obbyua, Ckutzo KL'L~U I took the latter posrtlon (Recanat] 2000b. 1 5 3 ) , but I adnxt this IS a c oi~trovenral issue

6. Echoicity

Should the quot'at~or~ marks be construed as a context-shifiing operator, or do tliry nterely convey a conventiortd ~mphcature (to the effect that the

' In many uses we need to posit the context-shitting operator in order to get a cohererlt interpreta- non. even if the context-shift is not explicitly signalled by quotation marks or anything else.

Page 291: Recanati, François. Truth-Conditional Pragmatics

-- O P E N Q U O I A T I O N KEVISITI .13 285

speaker is R-ing the enclosed words) which pre-se~~iant~cally atTecti, the character of the utterance by suggesting a coritext-rhifi?'O rrfils qilestxon ic

directlv related to the ~ssue raised lar Cectlor~ 2 whdt 1s the conventrond meaning of the quotatson marks;

The quotabon marks, 1 said, slgr~d tlratthe rpe&er w R-lng the enclored words. What 1s the relatior1 K? I ~rrerrt~ox~ed twit opaom. either the speakor rnerely calls attention to tlie ericloseil \vorcl\, deninnrtr;ltes them, or, more specllically, the quotatlon marks srpal that the speaker u\es the enrlosecl words echozcally. Ifwe take the first ophorl, the echoic interpretatlol~ n utilv one contextual interpretation ofthe speaker's act of drawng atter~t~vrr to the enclosed words, among other possble lrzteqreuhonq. O n the second iap

tion, the echoic interpretatlor1 wlsat the quotatlon marks tkierr~selvci encode: puttmg words m t h ~ n ~ ~ U O ~ J ~ I O X I l~larlir i~ieanc that one 15 uclng

them echo~caUy, that 15, that oiie rs thereby cvok~ng some use of tlholc

same words m a source context r "~st~nr t from the c tirrerlt ccrntext 1 arn

going to argue that, &we clsoo\e tlus optlo", i r will be very liard to revst the claim that the quotatlo11 mdrh are an object-larigtiage re~hzatlor~ ol tire

contest-shifting operator I menuoned two cases of context sh.rCi I ~ I rub-clausal open quotatloll the

case in which the cluoted word\ 111clude ari rndevlcal expre\slorr wl-ruse semmtsc vslue varies across conteut\. and the c,lx\e irr which the quoted words iticlude an exprc\sion cvhose I~nguistsc tlientiulg vane\ acres L o r r -

texts. In both cases, the content ofthe cnclosed words changes a rerult o i

the shtft. Now, most echolc ures are crlcfn that: the content ot the quoted

words remains unals-ected Corislder exam~~ler (4)-(6) once aga:axrr

(4) John 1s very 'cool'. (5) The dern~~~strabon prov~de~ the 'mode of preser~tat~on' of the referent,

hence d e t e m e s the conte~tud "censehof tho dt~monstrative (6) Chateaubriand returned to Er-ancc ur 1800, bwlt-I~ the < entu1-y'

These quotations are echoic (the qpeaker t~cl t ly refers to sonle use of the

quoted words in a source context Cfrstrnct jjjroill the current context), yet the

' O That the cluracter of an utterance may bc aEecrcd by prayrtiatic facts, rather than l~cirig e n u ~ l y and exdusively determined by the convenbons of the iangt~age 1n a context-independeru: nramcr, rs something we know already koril the study of den.rorrscnhvc~. (A sentence in which a dem<>nstrat~ve occurs ha a determinate character only if the demorutr:ative rs completed, in context, by ati appropnatr 'demonswation'.)

Page 292: Recanati, François. Truth-Conditional Pragmatics

t-c~ritcrit ~ ) f the wtrtcrsce 15 tlre wrne ,1s i t ~ v o ~ ~ l d be ~f we re~rioved the

q u o ~ i t ~ o n H I ; K ~ S . '~LUS ~1.ifih~ be taken to show th:li not all echoic uses iuvoivc: ,r t .orirc.xt- slrifi. O n this v i e w , svt. ~ ~ 1 2 2 resist t l ~ e clnint that the y i i o t a i i t ~ ~ : rn , i rba :~r.i. iiri o i - t j e c . t - l ;~~~gu :~ reaiizatiori of rkre context-shiiting opcntor-, even r fwe tnkc tlrei*~ to encode rhc f i c t that the speaker is using chc eric:b.usct.l wcrriis ec:huicaily.

I J r r t the n.Lison.iaig &tit Ie,~ds ro that corultrsion is fi~nlty. A context-shift

ilx,iy trt-i rir u,tirl.rout ail;>c:ting tfre ccrnt:rrrl: cid'the wwt-ds it c:oncerns, so the fict that (:el-taxrr cxc-E.ic>ic lrst:s ilo r i o t ;lflk:ct' the coutcnr: c-rfthe cjtroted -~vords does

not sliow tliat: rlcl context--stlifi occurs. 1 et us define a, coiltest-shik as "ber~fges' tlrlrenevi.r tire content of tlre wc~rtis ;it iss~le is the same in the t.-rrrrerrt ci"lrltext ~ ~ r d the source corztext. 111 srrth cases the contest -shift has 110 dIi'ct ora ctriiterrt. Tlmt is. :arguably, what h;appens in (4)-(6): in virtue of

cErc coritt-xt- slnlir, thc w ~ r d 'c-001' witllilr quot:itiorr rrrarks denotes the

property wIlic.lir el ic wtrrci 'cool' derrores in] iirc sorirce ctmtest. Sirrce the

word kct~ni' derjotcs rhe same rhirlg ill the so~rrc-e cor3tcxt anc3 the current colxrcxi, tllc word 'ccilol' within q~itrtirtloxi ni:irks dcr~otes the same thing as

thc wonl ' C O O P ' w i t h c ~ u t the ijiio~;itio~~ lil~~lrlrs (F.e. witliix~t the shifi}. The co~rccxt--silitr is tticrc:iix-e beraigl: it does r r o t aKect the content of the scel~ttric-c-; l7rat i t :iffi:(-a i ~ q t:b;rracter, sirrc.c t i le t.haracteu of the expressiort withrri ij;riotation rxriirks is x~clw rner;llirlguistic. Ag:lin, 'cool' witl~iri quot:a- t ic i r~ ni , i rAs L ~ C I I C P L C " , ~ ~ ~ ~ ~ ~ ( C O ( ~ ' ~ P Y L O ~ L T Sir? the rcrurcc c~>niicxt. j7'liis fornula 'says' d r a t r t ~ e i ~ ~ ~ t t ~ i t ~ ~ / ' ~ ~ C X C X ~ ~ C ~ S S ~ O X ~ in ~ L I O I C S is and t ~ r th2 snrrle time 'shows' its ss~ct.:lli~rguishi- ~tl;il-:sc-ter.~')

Siirc~c~ i oiircvt-diifis c:tr1 'rx. benign, tlirre I:; rto rcnsorl to cle~ly that, by lislrig: worth c*c-iioizaIiy, one slrifis tire corrtcst f i r their irlteq3retation from

dii. currerri conrcxt to the source contcxt. 7 ' k s nlay, or trlay ricrt, critail n cjiiarigc~ i x i c.oril.cnt tbr the er~clased woriis, irrit in all c;~ses a change in

cliarnrrtcr rcstrlts. So, t o S ; I ~ that tiire c ~ n o t n t i w i i rnarks ror~ver~tiorr;~lly rnearl tha t orrr is iasirr;: r l i c i:riclused cvcjriis ecl;ralic:xliy is to say that they are an c~i~jc*c-tiaxigxclagc rc,ili~ation of'tite ctjritexr -shitiirrg opc:ratc)r d. On this view, ttrc cqnotatiorr rrlarbs :ifL.c.t the character of rhe wc)rcls tliey nttacli to, niaking i t rrietniialg~iist~c: the i-lmrttc:ter of' a sub--d;rrrs;d opcrl cluotahoxl maps the

I ' Vli!i.it 13 b c ~ i ~ g ~l t t i<~t : r i b y ~ I I C ~t~b-i : I~i i i \~t I open ~jIIcIt*ItIoru 111 f h ~ s SCII~CIICC is I>umrt~eir's use of the Wittgcrictc~ixu.trr ti~ow!c.q dl\troctroo 1.0 exphi11 t i k t ~iica th,rt A ilrcorp o f relkr-cr~cr can 'scrvc as' a theory of .ClI\C

Page 293: Recanati, François. Truth-Conditional Pragmatics

c)~).N ~ C I O i 11 I I O N I : ~ . V I S I r e v 287

current context (the ( ontext iii \ vh i~h the qtiotattor) 01-~'111->) t ~ ) the corltent expressecf by the cluotrci expression wLrtrl rrrterprctcd in the wtrrce torltext. That rriet~hrrg~ilnt~t char,icter Ir drfirlccl orl ly t o r i ctntcut\ in wlllch tlrc spe,lher (1 e. rile perion who cli~otei) tticidy rclcrr to \orrie ustA o f d ~ r cj~iot~cl expressio~t $11 '1 xwrCe C O I I L ~ X ~ i ' Cl~\t~llcr fi0111 tllc C L I I ~ C I I L ccmtext. SO the corrvent~oital ~~nplrcclttlre, to the effect tli,rt tile \pe&~ker n thc kxoirlg some use of e ci~rnnct horn ~ h r crrrrent we, cast be ~cdescnhed ,I\ A prrcriI)!Jo,tttotz of tlrc

y u o t i ~ t ~ o ~ i In corltcuts rri wlrrcll tlre p r c s u ~ ~ p o \ ~ r ~ x x \ not t \ , i t ~d i td (I c no use o i e ir echoecf by the speaker) ttlc r~rct~lrl,grrr>tic ~tur,i t tcr retklrns rro

c o l ~ trllt. We strll have to choose between tht. i-ir\t arlri the \et c l n d optrorr ,is far . ~ s

the corlventrondl s ~ g ~ ~ ~ f i t a n c e of quot~t ion nr rr ki I\ c ont crnc~J. l>o the

yuotatiori nrxik.4 inem t11,~t the er1c1ov.d words are usccf i~chol~,llIy (IT)

wtlich C ~ S C they are t>e\t ar1,1ly\ec3 .ts a (C)rittx\t '1h1fi111g operLxtor), or (10 they have a vaguer and morc gc.ncral rr-tcilr~i~rg, to the cfkc t tJ~,ct t l i t~ \pe,d,er

is deanonstrating the tilt losed .l.vo~-cji (tor some reawn or othcrli To answel-

this question we trlust t l e t en r~ i~~e tvlletlrcr ;ill uses of cluot,rtjo~l mzrks are echoic, or wixtllel- the echoic interj>ret:iriorr is oilly orrc c.o~ltt.xt~iai inter- pretation among otlica.

It is pretty clear that riot all uses o f tluot:ttion rr~arks ;we ecllctic. Many irlstances ofcii-jseit q~~ota t ion are not: the speaker t;~lks J I > O L I ~ WOI-ds urith(xit evoking any use of those wortls iri a source context ciistirlc.t frorrx the crtrrerit context. Example ( I ) belongs to t h i s r:~tc.gztry of Y1:it rtrentiorl'. (3rtc of the rrlerits of tlre idea that quotatiorl rrxrrks arc :I d:irlgc:r sig11:~1 by mc:ir,s of which the spr;xker draws ,itreritioil to the words hc is crsirtg attcl 'cicri~ott stmtes' tile111 is precisely tlrc fkct th3t it ~1pp1ie.s to botll open nrld closetl

ci~~otaticm (tvbat clistinguislles closeti cjtlotation being at! acitiitictixai katurc). In contrast, the rxlleged context-sllifiing filrtctiorl of iluotatio~t r~tarks cannot be generalized to all cases ctf clrttjtntion. l3rtt this is not a decisive ol?jcrtior~ to

tlie echoic view, tor the sernanric propertics of'ctosi:d tltiotation ;Ire, i r ~ any case, special, a i d it is by no mealls cle:tr that we c:in ,ic-hieve ;I ~ ~ r t i f i e r i

sernarltics for clpel~ and closed cluotntiolt. So closetl q~~ot:lC;ort i s not directly relevant. Nor is clausal open cjuotation diret:tly re1ev:iut. 'VVh;it we arc

concerned with is snb-rkznsul open iltaalution, ior that is the type of rase that. casts doubt on the cl~tln~ that open quotaticxl (whether clatxsal or sub chusal)

is a pragmatic phenoli~enon that (-ails hl- a rr~ulti cfinie~rsior~al analysis. Iideed, if the cjuot:;~tio~r nlarks i n st~b-c-iausnl opcrr qrrot;ltioxl can urrifbrrirly

Page 294: Recanati, François. Truth-Conditional Pragmatics

be atialysed ar the context-shfiing operator d, this would a r p e in favour of the type of rernanttr andysls whch Ceurts and N a e r provlde: a senianhc. one-ciirnensiooal analysis. So the questlon we must answer 1s: Are all instances of sub-clau\al open yuotdtlon eclioicj Or are there cases of f l ~ t

niennon anlong t i i e i ~ ~ ? ' ~ 112 previous work I argued t h ~ t there are non-echoic instmces of sub-

clausal open quotahon, met I provlded (7), repeated below, as an example:

(7) A 'fortn~ght' a a penod of fourteen davs

The speaker of (7) does not tacitly evoke sorne use of 'fortrught' In a source context dnhtict troni tlie current context, I clamed. t le or she means to corivey sontetlxng about the word 'fixtrught', wthout echoirlg anv parnc- ular itse of r t This is a care of 'Hat mention' very much llke (I), save for the fact that the qnoution 1s open rather tlzan closed

But 1 all no longer so sure. Maybe the speaker of (7) is tacltly refemng m some gerlenc rr~mner to the use of'fortnight' by whoever, In the lmguistic cornn~unity, conectlv uses that wctrd. Mdybe tile chdracter of 'fortrught' m (7) IS rrlade tilctallrtgulsnc by the cjuocation nzarkli and can be spelled out as thtng tnllrd 'Jortntght' or sometillng like that. If tills type of example a our onlv reasor1 ifor resist~ng tlie echoic analyss, that 1s clearlv not suficient. What a requtred, ~t thls stage, 1s a detxlled eniprn~d mvestlgatron of the phenomenon of sub-clausal open yuotation, in order to ariswer the ques- t ~ o n are there trnrnlstakable ~nsta~~ces of flat nieritlon among them, or more generally cases t f ~ t cannot I)e given an echoi~ ~nterpretation? For the time beuig, 1 want to remain agnostic a~ict wrll leave that Issue open.

Where does thir Leave us \.nth rerpect to Geurts and Maer's cntlcrsm of the nzulti dmenslonal malysls? Let us look at thelr own axlalys~s. In the relevant type of caw, they clcui~i, the quotation rnarks around an expresliion e presuppose that there 1s a speaker x, a use-event u dlstxnct fiom the current use. and d11 entlty % (of the same type AS the re_rmlrrr denota~on of e) such that u lnvolves x's expressirzg % by riieans of e. Besides th~s presupposition, whlch u to be resolved by eltber bindmg' x, u, and Z to su~table ante- cedents provlded by the coritext or by 'accommodatmg' them, the regular

" Gomez-Torrmte ciamms that there are also non-echoic uses that are not instances of Aat mention. Fie mentions cenairl 'scare-quotitig' use that he thinks are not echoic (zoos: 150, fin 16); but I am not convinced by his rxanlpie, which seerns amenab1e to a corrtext-shilnng analysis.

Page 295: Recanati, François. Truth-Conditional Pragmatics

(3PRN C?IJOTA,Tl<2N R E V I S I T E D 289

semanac content of the quotatlonjust n the entity Z, which it denotes 'The upshot is that, ~n the current context, the yuot;ltlon denote? whatever d ~ c quoted expression denotes in the source context. 7h1s is roughly equxv,*ler~r to the 'metahnguistic cl~aracter' analysi\ 1x1 terms of the context-shzftr~ig operator d. Note that the rrtetalnl~e~,t~c character theory 'dso appeals to presupposition (see above) But the bxndlng theory of presupposition ui

which Geurts and Maier couch their proposal enables them to r11d.e clear what distinguishes their account frorn trjvo-dlmenuond approac he\ srrlce they treat presuppositions as a fonn of amphora that has to be 'resolved' rn

the course of semantic ulterprekiticrorl (rather than a a separate colxiporrent that 1s 'passed along' as we nlovc ixp the sernantlc tree, as In Potts's niultr &mensional analysis), they are able to provide a one dnnenslol~ai treattnenc that integrates the convent~ond conrl lbut~ctn ot rluotdtloxl nldrB5 with regular semantic content

Should it turn out that sub-~lausd (open cjtrotahon 1s always cchorc. I would have no objection to thrs at~aly\ls Aga~n, ~t is roughlv equlvalexrr to tlle metaluiguisac cliaracter approach of <J)rut~o Ohltqw, Oriatzo Kccrn But that does not niean that I g v c up twn-dlnien\~on.~I~sn~ irr thc anal.ir;rs oh open quotation. I tliink we still need two dintensrotis, even d we at cept the

Geurts-Maier analysis. h r we st111 haverr't r apturecl the corztcxtr~al tneanal;?

of the quotation, wh~ch hds to do with the reasctll why tl-rc spe,~lier r q

echomg tlie source's use of the \anre words 1x1 ,oixxe caes, the \peaLerss a m is to u~forrn the hearer that the soulce used the yuoted \vol-dc, 111 other cases she wants to &stance herself kern tlrose c~urds or to ~niiicate thkt <he

does not fully endorse them, becau\e d.lev are lrladecluate or for whatever

reason the context nlay make n~an~Cest, ~ x l stdl other case$ she want\ to

appeal to autholty by using words recoguzably associated with the sour i c,

and so on and so forth if we leave as& wtlat I called 'the contextual

mearung of the quotat~on', we get onlv a tmncateii account I conclude that. even lf we take echoic-rt\i to bc the convenhonal ?~.gr~.lfi

cance of sub-clausal open quotation, w e st111 need a complex picture with several leveb, in the spmt of what I proposed We still need to thstlngurch

'3 I say 'roughly' because the C;eum- M:uer ,urxiysls :dlows tirrr scope jltTerer~ces ~ v t r ~ c h t r y analysis ln term? of cliaracter doesn't allow--an advanage for ~lteni, urrdoubtedy, if the rdevan readings can Ire attested.

Page 296: Recanati, François. Truth-Conditional Pragmatics

Erc:tr;rrcerr iss~rus A, li, 2nd C:. The alkswcrr u7c get ifcvc take this option are tl1c fi>l~o\v~~lg:

e j 13: d :or2~p~'rot1iirzcd nl t '~~ir1q if I ~ C q~iotlz(iot1 marks] 7'11c" C ~ I I O L : ~ ~ ~ O I I rnarks

signxi that tlii: sjic.;ikcr i s wing tlxe errc-Iused wor-ds echoic-ally arid tacitly rckrs t t r sclnxlc use o f ' the same words In a sotirce context c t (to he corrcexi~x;rliy itlentitieci}"

* 1 A: ~ t e ~ ~ ~ i r ~ q / ~ o r r t c n t of t11t: q~iot t ! ( / ~,t)ordsj S~llltlliti(.~lly, this has the effcct i r i ' a;liifiirig the context &.>r tl.ii. interpretation of- the etrclose~t words, r l ~ a k i r ~ g their c:Vlar:rcter rrl~talirrgrristic-. 'This rri;ty. or m:y trot, dfcct the

cor iccirr oi' thi= i.vords irr c i t r e s t i o n . * 1: C.'ontt~xtti'ti riz~(lniri~q qj' ~ I Z C i j t lot( i i io l~I I'rag~ri:ltically, this raises tile

iks ire tri' t l ~ c ' c~uota t io r~n l poirit'.--dre speaker's season f i r ecl~oing. I-llc ,rxrswe.r t o ih;it C~IICS~IOII provi~ies t l ~ c ~ o ~ ~ t e x t ~ i a l meaning of the

ipi i t ra: i r i t r rr , which rernains separate (Llthoi~gI~ it may ;~tlkct the tn~th- c . i ) i i ~ j i t ~ o ~ ~ s v13 t r ~ v ~ I I ~ ~ C ~ T I I C I I I , 2s i1111I~ate(l ir-r Section 3).

7. WIIXCCJ C X U O ~ ; I ~ ~ O I ~ : the t-d~~~'~"IldB>Xlity issue

Thr gist c s f tire j-)cq~?;nraf;c npprcx:lrfr I acivucatc is thttt if~ioration involves the $peak;.crqs c i i i i q aonlething ("R-ing' the qrrtri-ecl wcrrcts, wlintever that mrns orrt

to tic) drrci rhcrcby ~rl~ylyiltg :i ntrrr~ber ofthkrgs having to do with the reasopts wiiy c/hu docs so. 7'11~ contC..itxr;rl r~rranirrh: ~I'tXlc: (/~~otdtioil is sometlGl-7g we gct to tl-tro~iglr iritc,rpr-ctir~g thc spc;rkcr's ,tr.tioxi, r i o t st~rlietl~ing t h ~ t rcsul~s

kiie>rri xlic~-ir,lrric.;iiiy '~pl7lyitlg ;L sct ot'rriles. Whcbi:frer or riot, scr~~aiitic;dly, the

cl t~~i"i if i>n xn":~-Xcs i'iirrctiot~ *u c~)rltex~-sl~ikix)g; O~C*;IIOI- i l i t ~ not change tl-zat b3sk' p i s i t i o i ~ . It ~ l n y be th;it thit ~ja1ot~tio1> I I W C ~ S s ~ s c < : I T ~ ~ I . ~ c ; ~ ~ ni&ct the rrit.:inirig oi'cEic scbxrtcrlce they occlrr irr , wielrotit the conrextu;il n~enrring ctf'the cv~fi- '~t~xi" k s d f resnhrrg &onr the applicrrtior~ ot'sernarrtic rules.

Ntrr i s talc pr~g~:rn:xtic: appro:~"h -under~.rrirlcd Ily the Gct that the contextu~l nrcarririg of' a-he c.~rrotation rr~nkes a dif1l.rer-r~~ t o the utterar~ce's truth-

colrt-licior-is- Ttrar it tioes i s shoxmr by the pherlorrxe~lotr known as 'mixed c111u"tatitrrl'. Mixed cluotatio~l is the c s e , il2nstrated by (10) ; L I ~ (12). where a*) cittcfiuicr wXkli-11 reports n loc:r~t*ioxl:lry a r l p e r f o n r i e d by some agertt x spec-itivs LIE cc,r*tcr.rt oKrh,tt act I-ty i.tsirlg a 'that'-clause as cctrm17lenient of tlre

Page 297: Recanati, François. Truth-Conditional Pragmatics

locuwon~iry \ielb I/', arid at the sane bllle spec~her (\ori~e of) the words ,ictudly used by tile dgerll t l~ro~igfi a ' ~ u ~ - c ~ ~ L L I s , A ~ ~ I ~ O ~ J O O ~ ) 111ude t f l ~ 'thdt' (-1~~1\e:

(LO) In :2.fii.lzoirt.s d'cllrtut,-totrzbr, C:hateaixbri:ir>cl wrote that Ite rrtlrrned to Fmiice

it1 1800, \\\nth the crr-ltury'. (12) Pat11 says he's due to present h i s work in tllr 'paper xss"ix~'

An utterdace hke (10) irlt~~itlvely enta~1.5 that ( :hdtc~itt~nand used the word\ 'with the century' In perhniirng the locuhor~~iry 'lit t h ~ t r \ k)emg reportetl In general, 3. mlsed-iluotat~oxid uttewnce 's i7-eJ that ' - ' ' eirt,irlr t113t the ,tgent x u$eJ the quoted wvrcis 111 perfi)l-r?i~~~lg the lo( ut1oi1ai-y ,tct Vth,it 1s be~xlg reportecl. NOLV I ~1k.t: the ' en~~~ lmcr l t ' 1x1 cjue\hon to ~csn l t f io~i l free

ennch~nerlt through iut ozpor<inorl of the contcxtu~~l rtre;irurrg of tile cluctta hon. More preclselv, ~t n ger~er~ited tl~rougli the fitllowg rx~ech~~r~lsrn

I L3)r ostens~vely qirot3rtg the wureis 111 tllc corttext of rc,portln;: C,lr,l

te~ubnanJ ' \ descnptlorl c>fhis returri to Fr,lnce, ilre \pr.aber nlanlfe\ts

h ~ s lntentlon to tnfonil the i~c ,~ re r that Ch,rtclr~ibr~anit uwd tllo5e very

wolds 'That 1s t l ~ e ' r o ~ ~ t e s t u ~ l rrieanlng' of the quoLttKotr 2 By thn5 ni&ing 111s rnto~indt~vt- ulterttrori rr~ttt~~ally n r~~u lc< t , tlic

<peaker co~linturucates that piece of ~rzi;>nndtrotl to itie hedrer j ~ r l

tile Griceai~ serlse oi'coxnrrriililt atTotr 35 ' I K ) x I - I I , ~ ~ I ~ ~ A ~ r~lean~ng')

3 The prece of~r i t i ) r l i ld t~~>~i tllt~\ prdpi,iticaily fn~pnrtetj overl,ip\ wlth the utter,ulcr7s sem,ruitrc ror~te~lt. 2nd fine\ w ~ t b it ttirct~lgh the rnezha

nism of free erzr~cl~lrlent l4

Many author5 reject the tree cl~rlcf~rricrlt dccoilitt, b e i d u ~ thev tltlni,

wh~tece l is clt.,trly pdrt of triltll ( o~idttrons ~ I I L I S ~ he. ‘tee oiirited Tor in ternr\ trf serrlalrtrc r~lles O f courw, thns c'tnnot be the ~ r g ~ ~ r n e i l t illvolied fix rgzctlrlg the at coiirlt, srnce that wori1r.l beg tile quc\tlctn So w11,rt is the qywnent7 Well, the o ~ d y arglrnlent X have Reen eupo.\cd to 1s the fc\lIow~rrg P he relevant 'cntxrl~~ient' (to the effect tb,~t C:h,rtedubrr~rtci llreci thctse words) cannot srrrlply be a prag111;ltlc suggeqt-on that ttns .tourid w'iy into \ernanhc content, fix l i l t were, l t $hotlld be tnrrn31fahicz- hut It 15 notf

'' fbe sane niecllailxsm 1s argual)ly re.;pornble f i r the "clr~riy conrrzistlve enu~lrnents' vvh~ct~ Poth

(zoo.-/) detecu m the following srntcrrcrs:

(i,r) Wltexr 111 Santa Cruz, t'etcr ordt-n '/er/p~lcots' a t the locd nnrkct.

(ib) Wlien in Ai~llrent, Peter iitiiers ' j,zj~iriiirt~' at tlic local rrilwkri.

Page 298: Recanati, François. Truth-Conditional Pragmatics

-- - --- - -

This 1s the Gncem test for tellulg apart the seniantlc &om the pragmatic: pragmAac suggesaons, by the~r very nattlre, can be cancelled, either con- textually or eqhcrtly I agree that ~t rs tbrs sort of consideration that should be appealed to for settluig the Issue, though, of course, 1 disagree that m the present c;r\e, the test argues aganst the Gee ennchment account.

Cappelen and Lepore have exphcitly appealed to car~cellabhty in arguing dg~~ns t the plagrriduc approach to m~xed quotation (Cappelen and Lepore zoojh, 3007) Their argument proceeds m two steps.

r If a corriponent of content expressed by a sentence S is not cancellable. then we hdve good reaon to thlnk that this feature is part of the sernantic content of S.

2 In XI Irn1,znce of livxed cluotation like (lo), or Cappelen and Lepore's own exlzlrrplc (IS), the a\cnptiori of the quoted words to the agent whose locutlonary act IS being reported (Chateaubnmd in (lo), Alice 1x1 (IS)) callrrot be cancelled.

(1,) Alice t a c i that BIB Climtoil is 'smooth'

juct a'r (10) entlzil% that Chateaubriand used the worck 'w& the century' in perfornlrtrg the lotuttorzarv act which (to) reports, (15) entals thdt Ahce used the exac t word 'smooth' when she described Chnton as smooth. These e~i tarhne~~b are not cancellable, C;appelen and Lepore say. That they are not '1s evidenced by the fact thdt there are no true utterances of (16)' (Cappelen and Lepore 2oo5b. 66):

(16) Alice cdd that Ntll Chtorr 15 "slriooth', but she never used 'smooth'.

Now X accept the first prerlliss in Cappelen and Lepore's argument, but I reject the second prerniss. Indeed, both in Orntio Obliquu, Orutio Recta and in 'Open ()uotation7, I pointed out that the ascription of the quoted words to the ngent of the locutiorlary act being reported is cancellable.

A mixed quotation is a sentence of the schematic form 'x Vthat p' where the "hat'-clause contains a sub-clausal open quotation. Let us grant that sub- clausal opeu quotations are (typically) understood as echoic: the speaker tacitly refers to sorne use of the quoted words by some agent y in the course of a speech event e. T o get the entailrne~lt that x used the quoted words in performing the act V which the sentence reports, it is necessary that the reportee x and the echoee y be one and the same person. Furth.er, it is

Page 299: Recanati, François. Truth-Conditional Pragmatics

oprr j QCJO ~ A I I C I N K X V I \ I I ~ ~ . I P 293

necessary that the locuhonary act Vand the speech event e be identrfied Even when those con&~ons are met, we get the relevmt en~llrnent oxdy 1f the speaker's '~~uotaaonjl pomt' I \ a\cnphon,ll l f t t rs not, we don't get the entalment. So there x e three ways to contextudly 'cancel' tile ilnplrcdtrnrr that x used the quoted words in perfor~mng act V We rndy conte~hrally equate the echoee wlth some person 11 # x, or tve nay equate x and y brtt take V to be perfbmied ui a speech evelit el dnt~nct from tile speech went e belng echoed, or the point d the yuotatlon m ~ y 11ot be ascnpaonal (ibr example, the speaker may slniply wish to clrstnrlce bllliself &om x's tr\tS of that word). In all such cases it wzU not bc part of die truth-mnrlltloris of the utterance that x used the quoted word\ 111 pe&>nnmp act Ti .

Take Cappelen and Lepore'r exaniple (I.;) To get nd oftlle eirtailnlerlt that Ahce used the exact word %n~ooth\vherr she described CXnton as rrnootlz, we only have to inlagme a context m whch, bv mmg \ii~ooth>choicaUy, the speaker tacitly refers to sollie use of the word \moot!>' by some agent y ~hstrnci

Jiom Ahce. That person y might be the bearer, as 1x1 th15 vanant

t C (17) Ahce sad that Chnton 15 'smooth" AS you worrld put rt (3f c oitrie tl~at'i, xlot i a rhe word h e used.

The 'as you wotild put 1t' is oyt~ond rt muty be ~ontextrrally clear that the speaker is echoing the hearer" uu\e of: \\mooth' rather thein Ahce's we have

to i m a p e a scenano In whch that wvrcl 1% saliently a\\oclated ivrtlz the hearer, perhaps because he keeps usrrig rt, while IC is kno\.vn that the word does riot belorig to Ahce's \iocahuinrv ) 50 1 tbrrlk ~t 1s quite poss~ble, ~n the

nght context, to Irnagme sorrleolie 5rnvhly uttering Cappelel1 and Leyore's 'imposs~ble' sentence (16) '"

In Orirtio Oblrqiln, Orutzo Kectgz 1 gave an example 1r1 wluch a thrd party's

use of-the quoted word rs belng e d ~ o c d

(18) Paul says that Q u ~ n e ' i s late f??r lus own ~ I I ~ P X .

In chapter 7 1 used the rrotio~l of'intenlai targec' co rel2.1- to the use being echoed. The target, rhas understood, involves botti a particular agent and a particular speech rverit. However ~t is better to clearly iiistinguisl~ bztwecn t21c age~ir and the everit. as Geum ,snd Maier do 11) tlleir o themse sjnuiar account.

Here the quotation marks around 'ur,possible' are to be interpreted hv rtriEiing the world coordinate of the cozitext. This e m ~ l p i e is anliog<ous to thr 'your aster' exanlple dicussed u~ C:hap~er

7, $5.3.

Page 300: Recanati, François. Truth-Conditional Pragmatics

WPrr ir r \ Lt*i~lg aiclnrc ,111~ echortl Irere Ir ( m o t IQul's, but). J~zrnes'r mrs~xkien uce

o f t l l c r8'iirrc. Xo;!lrunu' t o rchr to Mcl41ersorr the spc~ih~r o&(x8) xuo~k_mgl~ r n w i k ~ i t rldrirr LO refer to J\I1cl~llersi>r1 III liep(9rt111g I)Au~'c 5tntelnent t h ~ t he (McXP1~critrli) I \ Sate fhr 111s ow11 ~."V"S 1 /he LISC (91 the 1la111c '(Pulne' IS not

.~scnE~cd rci drr pznorl (Paill) wl~osc locirtrc>n,rry t 15 hc-mg reported. The p ~ ~ f i o n X ~ I W LI\~L" i ~ f the l1dlTIe '%)LIIVIC' irexng e~E locJ rs Lt t h d pdrty O,ieizc\j who ~ili'it,~ki'c P J I C X ~ I I ~ ~ ~ O I I ~ O I Q u ~ ~ c

I lie f r c t r11,at the use (3if the I I ~ ~ I V '(2ultlc' ~ ~ b i i l l ~ ~ bang echocst 111 (18) n ;i r r r r \ t ~ h t m ,nxA ~icvr,rr~it r~sc- -1rLe the ir\e ot pGiper7 ln (12)--1s LII trrrlevant

bcit~rre. of t111s PJITICIII,!~ exLul\ple- wlrlc\.r w,ls ong~rully ~rrtroduced 111

( c\ru\ci UOLI '1~1th the ph~r )o l l l eno~~ 01- jllltr~ c l ~ u s d j 1'~ngu~ge-ihlft. I t 1s

tX,t4y t o get ~ i i l 01 thdt fcdtrlre wl111t. rctxrrlxrig the strric trrre ofthe e x ~ ~ i ~ p l e B cr 114 I I X I L ~ I ~ C " th,it LIIC P C I ~ O I I \rl tlo t i \,t~d to ht41dti" fix h ~ s OWXI paper 15

C ) u ~ x l c l~ r~r r se i i , mil tlz,it the y c ~ k o r u'itixi (10)

(10) lDdrrl ~ y ' i i11,it '~iVlr (Zuinc-' i s I;rtc hrr h i s c.rwir paper"

Ag ' i r r r i t r m y be that the use o f the W O ~ I ~ S 'Mr Qzrine' which is being csa:iri>c:il EII (19) i s tlieii- use by :I third p;irry, Jarr~ihs, w11o always rekrs to

(,)iiilic as %lr i,)rrirrc9. Yarrl is riot :Iw:kre ot'tl~ac ii( . t , and he sinlplv refers to Q i ~ i r ~ r . as ' C , D I I ~ ~ P C ~ ' . Rut WI~CX:! rcporti~~g his I o ~ u t l ~ ~ ~ l l ~ y act, the speaker ironic-'diy srscs j:itrics's l-rtirrrase. 'l'his is a r t c . r?. report : there is no intentiou t r t i tRrrb s~rcaber's ~m-t to ascril,c"l~c rrsr of'thc plzmse 'Mr Quine'to Paul,

. . whose SPCCCII ;KIT Ixc is rcportulg. t 'he speaker heely chooses that pllrase to rcport Ititci's staternrzlt ( t h ~ t ' s what nxskcs the report de r ~ ) , a r ~ d the use of t-Bxit plicase i a i t.Lre spedxr's niorrth i s clearly ecl~oic: it irorlitrally echoes Jait~i:.;' use oi'it. XI? tlris context, the su:~c:stiorr that P5arxl (the agent of the repwtccl locirtior~:~ry act;) usccl the ijuc~ted ~ k m s e in perfbrnlkg the locu-

tloriiiry ;ri:t that i s Ireing repor-tccl is clearly ;ibse~it. ?'his sl.lc>\vs that the srigs:c.:\~ion in C ~ I X ~ * S L ~ O I I IS c~ntexz~~ally c,lncellab>ie. Accordiizg to Grice's clr:irac,tc.nxatir,rr, ara ir~~plicatuze ' i s corrtextu;rily carrceliable if one can find iiitirar-io~ls irx wlrii,h the utter;uicc"of tijrel fornl c~f'words woidd sixrlply not

c;irry r he ixirpircatrrre* ((-;rice r98c~: 4-41" In the CLW of Y I I ~ X C C ~ quotation, the reicvarnt 'fdi,rrrr r)f words' is a serttcncc "x 1* that y' where the 'that7-clause

" i*liii ii rcintriiscent irf t t x 'M. Arig~ntc' rsaoipic. wtiiclr 1 ~iiscrisseci at icrigth in my first piece of LVWI-~ i m i t t i t ~-1~tus~d optw q t i u ~ ~ t ~ o i ~ (b<e(-,ir~;tti IO/$J Ji. 4)

Page 301: Recanati, François. Truth-Conditional Pragmatics

roritJlra an open cluot'ltlrtn. Fi., (17)-(~9) sflow, notlnrrg 15 ea\~er tllaln to Ilnd

sltuatrons ul C V ~ L I C ~ such a Corirt of words doe!, not c axy the trrlpirc ~trorl that the Agent x u e d the cjuotcd wordi m performing the KC I;' whtclt r\ belrtg

reported. It IS surpn\mg that C~ppelcn arid Lepore u,e the ca~~cclI'lb111~ argurrleltt

dgalrl~t the prdgrlldtlc at count; for In both flrtltio obitquit, Orcrfro Rert~~ and

"Open Quot'itlon', I yotnted crut that the relevtux 111113bc ,>taon 1s c~ltcell~bic. t d ~ d so in titotrrc~tes, huwevcr, alxl thwe footnotes app,rreritly esc~ped Cappelen arid liepore's attel-ttloxl or] tkierr furrt readrxig. When they tooh notlce, they came up wlth d response (C:appelcil artd Leporc 3oo5b. 67-4))

whlch l la~ alio ternlmxl other colmxrent=itors (Golrrm-'Corrente zoo5 13.~-5) and whlc h 1 am rlow golng to d~sclrss. 'l'11e methodoIogtc'11 ilIszL1e.i they raiw cone zrtllng the senlantl~s/yr,~grn,~tlcs bouxrrj~ry provide 'in ap-

pn>pndte top" for the cc>n~l~tdrng sect~ox~ of thli ciraptcr, and oftlui book

('appetm mcI Ixporc arid Ck~mez-7 orrerite argue t'tl,lt we shouldn't \pet& ot 'canc ellabllrty' rf the form of worljs a t Issue IS, or c,tn he wker-1 to bc, '~t~lbsguou\ C:otnsitler an ar-r1blpou~ form oC word5 f+, for example the qentence 'I am going to take a bath close to the bank' ((GOIII~L-'1 orrertte', example), uttered r r l J. context c that rn&.es a c crt.iin re~lllitlg K sd~ent (e.g

the readmg ori wlz~ch 'bank' I:, taken ul the sense of ruvt.a BANK) S~nce F 1s

antlul@mus, it wll l-te posirblc to find a sltuatrcm (,L co~ltext) c' u1c11 tlrat, were F tokerled In c', an ctirernat~ve reading Ri (e.g the re~ding 0 x 1 wtniclr

'bank' u taker 1x1 thc fin~nci,~l sense) would be more sdllent than K. I>oec tlris dlnotrat to c~ntextctaily 'cancell~rig' whatever ttnpht ahorls cf~,tt~lgtru~\lt fS

from K'? Wo~~lci the add~tion ofthe worcii '1 don't rtzear~ cloita to tfar rlver bartk, I mca11 clo\c to die findrictal tllst~tutilo~l tell blc>cl\\ way' coutlt ,L\

exlyl~clt c aric eu,tt~or~ of 31) unplrcature? Of cottrse not. 't'hts example \bows that C;nce's clxasacten7atlon of canceEaI>lfity rl-roirld he used wrth cdre for a

nleanirig component in earned by an utterance of I: to count ac 'cancellable', hence as a candidate ~nlphtaturc. ~t n clearly not eizough tlut tone can

find sltuntrorks In W \ I I C ~ an utterance of F would not c a q that nlearllltlg co~nxponent. G n r c ' ~ cancdlablhty te5t ,~g~?~ficantly applies o~zly to iortxls of

Page 302: Recanati, François. Truth-Conditional Pragmatics

words that are not (relevantly) amb~guous; for ambiguous expressions pass the test for tnv~ai reasons chat have nothmg to do wlth ~mphcatures. So, rf the form of word5 we want to apply the test to n anlbiguous, we should disambiguate it (I e rpec~fy the rntended readtng) bi$ore we apply the test and niake sure that the expression is taken in the same sense m the ongmal context c and the alternative context c' in wfuch the alleged implicature dnappears

Now, precnely, all open quotatrorl in the complement clause of a locu- tionary report is ambiguous, accord~ng to Cappelen and Lepore. That arnblgu~ty IS due to the fact that quotahon marks are carelessly used both for quomg (their properly sernarlhc flncnon) and for 'scare-quotmg' (a Qstlnct Euilctlo~l wlth no SemanhC import). According to Cappelen and I,epore, \care-quotlng has nothing to do with quoting, semantically, so the quotaeon marks ~hould be treated as ambiguous (or rather, as they put it, as havlng a 'du,rl u\age7). Smce that 1s so, the alleged canceUahon effected by context rn my examples (18) and (19) s~mply reflects the fact that, by charqgng the context, we sh1f-t the (most sdent) readrng of the sentence horn the qtrotational reading to the scare-quote reading, j u t as, by chmg- mg the conteut, we can \hlft the most sahent rea&ng of 'bank' hom the nvcr reading to the financial readmg. If, befbre apply~rlg the test, we daaaibip,itr the sentence (as we should) and gve the quotation marks the quotahonal reding (both m c and c'), then the entdnlent turns out to be non~'~ncelIahle! That u Cappelen and Lepore's response to my examples Gorney-Torrcnte, who agrees wrth me (agamst Cappelen and 1,epore) that t l~e mixed-quotatlctnal ascnpncrtn of the quoted words to the reportee 15 cancellable, agree$ with them thdt my example (IS), because it

involves xare quotes, 'cannot convlnce someone who does not aLcept that uses of the q~lotatio~i mark hdve the same meanmg' (Gomez-Tonente

zooj r 35) Before ctrnsidenng the rrl~tiorl between cancellabibhty and ambigpity in

mine drtnrl, let rrle first point out that the use of 'scare quotes' 1x1 my examples 13 not e~aential What (IS) and (19) x e meant to dustrate are cases UI which the echoee is distinct horn the repoaee. That the speaker of ( I 8) and (19) 1s ~roii~caliy inochng the echoee, or distancrrig herself&om the eclloee, is an Irrelevant feature of the example, which cal be removed.

Page 303: Recanati, François. Truth-Conditional Pragmatics

O P E N QLJU I A L I O N K E V I S I T h l 3 297

In t h s respect example (17) appears quite difkrent, and I uonder what Cappelen and Lepore wortld say about - ~ t t8

m o e v e r holds that the use of quotlltiori nlarhs m scare-yuotmg exam- ples hke (18) and (19) has nothli~g to do w t h their quotllbond me in (so) and (IS) d have a very Iixd brne defenhrrg the same hne with respect to

the followmg examples.

(20) fice sad that Clmtor~ IS 'slnootlr', to u\e her awl word (21) fice sa~d that Clinton is 'ssmootll', a\ she put it (22) fice sad that Chnron 1s 'ssnooeh', as you woulcl put ~ t .

For it is pretty clear that (20) md (131) are very sirdar to Gappelex1 and Lepore's own example of mixed q~lot;ltloli (IS) the addit~onal phrase 'to me her own word' or 'as she put rt'mnply rr1ake.i explicit that tlre quoted wordc are ascnbed to the reportee. So I don't t hu~k Cappelen aid Lepore can denv

that the yuotahon marks ftinctiori in tfie \aim way in (20)-(21) and in (IS),

(ro), or the other exaniples of rn~xed quoutlon The problem, ii,r thern, 1s;

that (22) n extremely slnvlar to (21) the nmn ~ixfference rs dtat the addi- n o d phrase contalnr second penon pronoun in5te~ri of a thrrd penon pronoun (that is what inxkec tlre echoee distlnct horn the reportee) In a

nutshell (32) is extremely s~rnilar to (21) whlrh 1s exbenlely sinvlar to (rc;) No sharp demarcnhorr between two distmct ren~fmgs or "usdgel' crf the quotation marks can plausibly be invoked here Now (17), the exzxnple which is supposed to demonstrate the ( ancellabilrtv of tlie mixed-quot,i- t l on~ l implication of authorsll~~,'"- 2nd the po~s~bl l~ ty of Cappelen 2nd Lepore's 'irrzpossible' sentence (10) --IS nothing but (22) plrrs m ac3d1rronal sentence cancelling the iniplicatrox~ m qixest~on' In tliic Lace, 1 think, the

'mb~guity ' or 'dual usage' responce r\ hopeless Now I turn to the methociolog~cd irrue. when arid how can we apply the

cancellabhty test? If a meanmg conrponerit m is an unpliczture or jrllore

Reuncr gves an exanlpie hke (r7), Ibr whrclr she crecirts Ph~iippe De Ur~lxnrtcr (Relrncr 3005: 180). See also Gonlez-Torrente 200s: I j j ti21 a S U Z I ~ ~ e.unipie ~~nerriied m deinonsrratr cancell- ability \vlthout relying on scare-quonng. Benb~ji (2005: 35) also nlentlvos the possibility that clle i.ciloce might not be the reportee, but strangely shies awav froxri the i:onclusitm tha~ are nuxed-q~ocanorial implication is cancellable. on the grounds that the cunvenrrond inlphcature/presrlp~~0s1h~11l that somr<)nr used the words is not.

" I will henceforth use that phrase to refer to the 'en~~iimrnc' allegedly contributed I>y the iponoon marks in mixed-quotationd sentences (to the cfiixt &at dre reportee used the quoted words rn prrrforming the locutionary act V that is being repolzed).

Page 304: Recanati, François. Truth-Conditional Pragmatics
Page 305: Recanati, François. Truth-Conditional Pragmatics

Before proceeding, let me rlotc tlki~t the sitiration is exactly t11c slime in tlie present c;1se ;md in the classical cases iri tvlzicfrl the c;ricea~~ criteriz have been applied to est;tblish the pragrn:ltic nature of' an im1,lic:ttiorr. ( :c)usicfcr 'and' arid " o r ' . 'l'lrc colzuer:tive 'or' fi;ls two pnrna Gcie 'reacli~~gs': the inclrisive reading and the exclusive reading. It ntay be takerr to be axnbigu-

ous I7cmeext the two reaciings, but it is also possible to take the inclusive reading to be the only 'litel-rd' re;rciirtg oftile connective, arld to ilccount for the exclusive 'reading' by positing at] ixlrylit-ature that corrlbines wit11 the iiterd readir~g in sor.ne contexts. On this pragmatic ;~nalysis, the two 'read iags' are simply: the reading wit11 the ir~tplicattlre (exclusive reading) :);~nd tile 'plain' reading wi t l~o~ i t the itnpliiature ijtrlclrtsive reattirig). <;sic(% sixrrilat-ly

;txguetl that 'arid' is not anlbiguous, that is, tlnt it llas c.)nly oxtc literal 'rer~cling' (die logical reading): the vanous otlter 'reactirlgs' (thc tcnrporal

reading. the caus:ll reading, etc.) nli result Si-om the additit111 ctt'all iniplica- turc. Now, of course, ca~icellability cannot Ire ~ ~ s e t l to tiisprove the ambi- guity ;mdysis; for i f 'or' were anibig~~ous hetweer~ the two I-eactittgs, h e

exclusivity itnp1ic:ttiolt would he prinra facie c.:lrlc-ellhie. Czaric-cllabiiity does not tell US wt~ich axidysis is right.'"

't'llr ilnair~ tool in (;rice's hirtrcis for sho~vixzg tlt;xt a nlcan jng conlponent is prag~~~:itic mtl-lcr than sel~i;irltic is the metllodologt-:tl prcc-ept llc callect 'Modified Occam's Razor': Do no1 mralllply s c ~ s r s b<:yond v~eressily ((;rice

1989: 47).21 According to tlxrt p~inciplc, if a given plreriomer~orl (e-g. t l~e two 'readings' of diqunction) can be accounted fix either in terms of

sernat~tic alrrbigrrity or in terms of implicantre, we sfnoulii choclsc the at-co~rnt in terms of in~plicaturc because it is rnorc pusinionious. ' I'iltls

Crice :~rguecf against the alnbiguity ctf logical particles like 'aud' arld 'or'

between several reading on the grounds t-hat a 1,ragtlatic story rc-ln be tolti tirat ilosits only an irrlplirature and a basic meaning. Since irnplicntttres :ire c:irrcellable altd appear ordy ixl some contexts, t11e p~agnatic story accounts for the existence of two apparent "readixtgs': the reading with the

"" T o esrxbirrl~ tlrr pr.rgdtlati( nature of trrtar~ing coiupor~tnt, we r r u y appeal to othcr i;.:tti,res of rrripiic~rures, s~rclr 'is G~ice's 'r~ondtt:~chabdtty' :uid 'cdculahiiity'. ITa rtrclrlttlg corxip<)ttent is r ~ r ~ t only

cancellahie but also noiidetachable atld calculd)lc, it is vcry ilnllkdy thrt tt is 'scri~.rnt~c' r,itht:r than ~""gn><tt".

21 SCC also tjiipke's ~ I F ~ U I I I : '1 fo tii)t pos~i .LII .uni>tgix~t~t iinlr.ss vou arr rwliy con-rd to' (Knpkr rs.17: 208 :

Page 306: Recanati, François. Truth-Conditional Pragmatics

300 LANCE1 I. --- -- -

impltcature and the 'p1~1 ' readmg without the unphcature Note that, zn all such caser, an uwrb\qurty theoon~t ~ n ~ g h r respond us Cappelen and h p o r e do. that cartceUablhty per se does not disprove the ambrgulty thesis. But Gnce and the Gaceaiis never clamed that ~t did What canceilabihty (together wrth the other feature\) does n show the pos~zbilrty ofa pragmatic explanation. T o &pose of t l ~ e arnbigu~ty thesls, Gnce appeals to Modified Occam's Razor. Accordllig to Gnr r, it is less econonucA to posit an ambiguity, d a prag- rnattc explanation 15 avarlable, slnce tlie p r a p h c story stipulates only one sertse (m5te.d of two) and appeals to gerteral and trzdependently mottvafed prmnp1i.s to account for the generation of the ~mpl~cature.

It IS mn\tnlctlve to look at b p k e ' s $veil-known ~pplicatlon of the Gncean strategy to tlie refererit~d/attnbunve distinenon. b p k e argues that the definite art~tle 1s not anibiguo~ls between two readings, on the grounds that a pragtilatic explmatlon n available The pragnatlc exphlatlon 1s niore parsunonlous, for tlie mgrecbents ~t uses are avalable wlzether or not one posz" an arnbgutty Vosinng an ainbigurty therehre achieves notbng: there are n , phenomerid that can be explaned m the anib~gu~ty kamework, w h c h could not be esrylalned rf we did not poslt the amnblguay T o show that, b p k e t~iviteb us to inugne a lnngwage In whch the defimte arucle IS, by sapulatlon, L I ~ I V O L ~ and tlds the Rmsekan semmtrcs. Detinlte descnptloix wlU s& have 'two rues\u ruch a language, slnce the refkrentxal use can be accounted fix on the basic a f the Russelhan senlantrcs plris general and Independently nlotivatcci pnxlc~ples of language use Thls tbougllt-experiment stnkmgly ciex~~omtr~~tes that pontlng an amb~guity adds nothlng In terms of explana- tow potential (but co%s more).

fhe didcctlc witli the nuxed-quo~it~oi~al i~nplicatiorl of autllorshlp 1s exactly the same Cappelen m d Lepore accept that the lmpllcatlon is pnma hcle caacdlablc, but clalln that this results &om an anibigulty (even lf they reC-aln f+om usmg that word) the quotation marks mean somethtng differ- ent when the rn~phcatlon 1s absent The utuatlon is as follows (I) There are two pnrrla tacie 'rradrngr' of sentences ser?tencesof the form 'x L7 that p' where the ~ o ~ ~ ~ p / e l n e x ~ t centeitce cotttdliir a sub-thusdl open c~uotat~on-one readlng with the mixed quotatlorid i r~ iph~ahon and another reading wlthout lt; (11) &ere are two Lumpeting accounts- a semantic account that posm an ambi- gulty (or 'dud usdge'), and a pragmauc account that posits only one basic

Page 307: Recanati, François. Truth-Conditional Pragmatics

meaning plus a pragniahc suggeshor:, (nl) the pr;igildhc account rs, always, more parsmloruous."'

That the pragrnatlc .~tox-y rs Inore p,irsiniornoils, and shoulci therefore be preferred, can be shown by fi,lluw~ng Ib~pke's proced~tre an~l I I I I ~ ~ ~ ~ I I ~ a

language m which, by stlpulatlorr, tkrere &re no 'yuotatlon marks' wrtlr a special sernantlcs, but the speaker cdn stdl ostei~srvelv derriorrstrate the words she mes whlle speakmg. In some contest the densor1str;ltlon \viH be undcr- stood as a plece of rnirnl~ry whereby tlxe speaker echoes some crtrrtextu~llj identifiable person. If the piece ot rnlrlucry occurs in the cornple~rrent c Xnuse of a sentence reportmg the content of ronieot~e's speech, the speaker *vrU naturally be understoocl as nunxtkmg t t ie reportee and speahng llkc h r ~ n (In other words, the echoee and the icportre- the rxrternal dnd exterr~ai targets, m the temmology of Clr'lpter 7- - u r ~ l l be idenhfieif ) Thr darnon-

strated words--or the toi~c of vorce, or the special proI1uIiiX&t1<)n, or whatever is ostensively dcrnorrgtr~ttd--~\*rU be ascnbetl to the rcportce It1

thls way it cat., be shctwii tbdt the rn1xed-qtl~3tdt1~t1dl 1nip11cdt1011 of author- shp can be accounted for on piGxgxll;rtlr grour~ds

Instead of iniagnmg a lazlpage 1% ~thotlt quotdtlon rrrarks. we carr ssnlph look fbr an utterance ofthe form h that p' contuning no quoc,ttrorl urlxbs

but such that the speaker, vvlirle rcpurtlxtg the content ol x'\ speech, ostensively displ~.vs some of the \vor-& ~lseri in the coxttent clarxrr, ,lnci does so with a mamfestly ecliorc rntrx~t~r)u " Ihe pragni~"b titheory pirc11~ ts

that, in such curcumstances, the mrxecl-quotdtlon,il iniplrcanorr that the reportee used those words (or this tone of vorce, or this s p e ~ l d ~ ~ O O I I Z I E I C I L L

tlon, or whatever) w~ll be gerier~tcd 'I hrlt thr:, pred~ct~on IS borne out rs

shown by thls exaiiple ti-om P)rrkens, whlcl~ I l ~ v e clred sevemi t r r ~ ~ e s

(23) To which Mr Bdey niodestlv replred that he hoped he Llowed wrtr

o'clock ~t wos m gnerd (I )~ckms, jMjrtil~t C:Iitizzl~w~t, C I C ~ L ~ 1x1 C'ILark and

Gerng 1990 791)

Here the speaker report:, the c?scnbte9s talk, uslllg ~ndlrect speech m the nonnal way, and at the same airre shotvc wl~at the tdk was hke. 'The Fentence

" Thus Saka rightly criticizes Cappcien and 1.r-pore's at:corrrit on the grounrls t h a ~ it 'pusitls] unnecessary ai~lbigliry' (Saka zoos: 203).

23 That there are insctnca of xnrxed iluotahot~ w ~ t l ~ o r t t quoation n~arks IS shown by rlrrs example, due to Philippe T)e Brabanter @.c.): 'At surne poitit, ~ o n ~ p r o ~ l ~ ~ e requires us to stop being hidxk~i servants of Christ. I t is at this point that we rrrusr, mrh Lilik~er, srace that here f stand, I can do no ocher ~

Page 308: Recanati, François. Truth-Conditional Pragmatics

L ~ O C ' \ not ( i ) ~ i t , i ~ ~ i L~I IO~ ,L~IC) I~ III,K"~S, vet dit. rxnt.ircatron thdt Mr Uarlcy \poke

rhuc WAY e,%ag e a d y be tlerrvcd, ~1113-147 by L S ~ P L I ~ the hliowmg cjueshon\:

Q r : 'Ullry i s the speaker cisterrsivrly speiik~ng in this iiinrly way?

l ~ l n n ~ x r : liiac;~usc 11e rrlirnics Mr t$ailey.

? : Wlay IS he tloir~g so?

d+lnc.ziiitr: I r i orcicr to irllirrrrl tlrc I1eart.r that tlrnt's the way M r Bailey rqrclkc wlrerr he rrpliec-l (while a t the s;rlr,c tililc 111;lki11(: fun of him).

Sirrcr r r ii; ailritu;~lly nr.rrrifest to tlre speaker ancl tlre lrearer tiiar the hearer c,m

iir l t l the .irlswi*r to ( 2 2 , ;mi since tlre spth:ikcr- has dorie nottlir~g to prevent

i-Cie ire;irex. frorri re:iclririg that CUIICI I ISI~FI , the spe3h;er overtly implies, hence corrii-ii~init ~i1c.s (in i l r ~ (;rice;rrl sense of" "rlorl--natirral nieaning'), rbat M r

I3:rilt.y cpcrkc t h i s way whet1 ire. replied. The picce of infifi,rriuticrr~ tllus pr;igfi"atic,rlly ixnl~:~rtcti overlaps wit11 the seirr:rntic. cilrrterit of'thc Lltterarrce jro tlii: i : i t i s i . t cli:~r Mr 'b3:aiic:y said ss~clr. ,inii sucll) ,lnd fiises wit11 it through the 11icii-r;airi51ri oi'frcc crinrl~ruwrrt. 'Phist is tlxc ~irag:?;nrntic explanation I offer for rilr jrhcilorsuclxoii .it issue ( d ~ c dscriptiotl ofl;,kle speckrl way ol'spetiking to Mr X5aiio:y r i l c h i s cx,irrrple, tile iiliseci-clr~ot~itrori~il i111j7licatioil of authorship ill rile oiEic:r c:jses). As crltv;rys wi th pGrgur~;lric i*spiariatior~s, tllc irrain reason tor

trc-i:rpririg r x r,irlrcxr tilr,rrr its serrl;ulric riv:tis i s i t i s n.i<ire p ~ r s i m o t ~ i o n s . ~ ~

Page 309: Recanati, François. Truth-Conditional Pragmatics

References

Anarid, 1). (2006) 1)e tic sc. Phi.> ifssertat.icitl, Dept o/l,ingt~istic:s, MI'X'. Anaild, P. arid Nevii is, A. (2ooq) 'Sl-ritty Indexic-als irr < :ll:tirging C:ontc,sts'. 111

K. Wntanabe and K. Young (eds), r"uori.cn'ittqs o$thc 14th (:orifirrnie on Stirn~zrzti~i

~ z d Li~gttistir 77re~)ry. Itliac;~. N Y : (:LC Pt~blications, 20 3.7.

Arlclerson, S. and Kcer-iart, E. (19Ss) 'I Icixis'. In M. Sllctperi ((~'1.). I~yquu~q?ire 'Iypology irrzd Syrztcu-tii- Dar.ripiion, vol. 3: Ct.unrmaiic-ui (','~ztqqories irt tlrr. !.exic;orz. (:srril3ritlgr: Carllbridge University Press, 259 308.

Anscornbre, J.4. tind l)ucrot, 0. ( I "37) '17eux rnais c11 francais?', 1,injitrcz 33 : 23 410.

Ailsconlbre, J.-C:. and L311crot. 0. (1078) 'Echelles itnplicatives, &circllts , (11 -~, ~ L I ~ I I ~ ~ I -

tatives ct lois cic tliscotirs', Srntuniikos 2 (2 -3): 4 3 -66.

Anstortll~re, J.-C. arrd I)tlcrot, (3. (1983) L'clyqttrnerrlatio~z licrt7.s l(i lirt?qtd~. Bnrssels:

Mardaga. Austin, I. ( r 97 I ) I~izi1c)sup~tircrl hpen , 2nd edrr, Oxl-brd: (:larcntioin l+ess.

A~istiln, J . (1975) ~ ~ O L I J to I lo 'ihirzgs i~ t i l i z !Wonk, 2nd cdn, Cfxfi~rd: Clartsndorl I+t-ss. 12nch. I<. (1987) 'Ikol!qht urtt2 Kefin,r?rtv, Cfxhrd: Clnrcniiolr l'ress. t-iach, K . (ic/(i~) ' Pav i~~g tlx lioail to IIeFererl~e', I'!tik~so!~hir~~l Sftdies 67: 295 300. Uacil, K. (it)c;+) 'Cor~versational tmpliciture', .blind utzd ~AZIWL(<?YI ' 0: 124-62. l k h , K. (rt)gg) "T'hc Myth of (:onvcrltionaI Ilnplicature', l,ir~~ui.iiii:s urtd ij/ktilnsu~llry

23: 327- 66. 15ach. IC. (zooo} 't2~tanrifitraticrrr. ~iraliflcatiorl and (:~otitcxt-', iLlin(1 mii I~rz~qr&u(re 15:

262- 83.

l2,1ch, I<. arid I l~rr i i sh , R . iC1. (1979) Z.it~,qttittir C:oint~itdrlkirtic,n ~ t d Spc*ah ,li.ts, Carnbridgr, MA: Mi7' 1"rt.s~.

IS=-l liIlc1, U . / i t/c.t) ' l n d e ~ i c ~ ~ l Exprt:ssions'. ILeprinteti 111 h i s /l.spr:if.s q/' l a i ? ~ q u ~ ! ~ r ,

Jenlsalzm: Mag,mtxs l'rress, 1970, 64, 88.

I ~ ~ T L V ~ S C , J . and J. Pcrry ( i 5183) S'iftiizti<rns arid .4tfifurlt.s, (-:alr~l>ridgc M A : M1'1' Press/ Rracift~rd Iiooks.

L2e:~uz(le, N. (1707) C;run~nruire C;irz&rrlr, 'l'oxlrz I , Paris: Mar1,ou. Bcrlh?ji, Y. (2004) WUsg i 3 t . h - s ' Words', :Touniul qfl"ilil1)soj)hiriJ Kcc~trniz 29: 9; I 12. Ilenbaji, V. (zoo5 j "Who Needs Sernairtics of Quotation Marks?', B(,kirzrt ,7otrmcII q'

1.ingttistics 17: 27 50.

Hrriveniste, E. ( r ~ 7 t ) '!>a forlne et le serls dms le larrgagc'. 111 tiis 1:)rohll.tncs dr, litz,orcistiqine gcrtiriiit, 11, Paris: Gallirnarct, a1 j - 38.

Page 310: Recanati, François. Truth-Conditional Pragmatics

304 REFERENCF% - -

Dezu~denhout, A (aooza) 'Generalized Conversational lrnphcatures and Default f'ragmtzc Inkrences' In Karm Campbell, J., O'Rourke, M. and Scher, D (ed?), Meanrng and ?kilz Inve3tlgattons In I-'htlosophtcul Sem'znttw, New York

Seven Bndges Press, 157-83 Be~mded~ou t , A (aaozb) 'Truth-Conditionat Prag~xiatrcs', P!ttlo~uphzcal Perspectttfes

16 103-34 Dxrrtvlrch, M (1982) Tornid and Lexical Sernanbcs', Dn,outsttsche Ben~hte So. 3-17 Ulemsth, M (15181)) 'The Sernanti~s of Gradanon' In M Bierwlsch and E Lang

(eik), Dtmrn~ronr~l AdJecftves, Berlln Springer, 71 -261 Bloom, P (aooo) H m f Chrldren iaurvi the h.Iu~zntra~q of W d , Carnbndge, MA. MI r

Press/Bradford Books Bohnger, D (1968) A ~ p e ~ h Ofhnguage, New York Harcoun, Brace and World Borg, E (2004) Mt~tt~1a/ S~ma?tttrs, (hdbrd Oxkxd Uruversity Pres~ Brandom, R (2008) Between Saylng and Dozng Towrirds dtz Artalyttc Pragmattsm,

Oxford Oxford Uruven~ty Press k$~oniberger, S (1989) type^ and Tokens In Lmguistics' In A George (ed.),

Rglertrons on CJtomtky, Oxford BlackweIl, 58-89 Purge, 1 (1974) 'lJemorntrat~ve Construc tlons, Referen~e, and 'Truth', Journal

Ph11ctsophy 71 2o>-x.j

Ca~~pbe i l , It (sg8r) 'Language Acqmsi~on, Psycho1og;;l~al Dualism and the Defi- ntt~on of Praiplatics' In If. Parret. M Sblsa, and J Venchueren (eds), Po~srbilttw and Limitatrt~m r?f l~ragnzattcls. Amterdarrr job1 Beryamuis, 93-10;

Capone, A (200)) Kevlew ot Ltteral ~bfeanrn~q, Journ~l of llngttrstrcc 341 45-1,

('appelen, 1% (2007) "Semantics and Prapancc Sorlie Central Issues' In G Preyer and (; Prter (eds), Cuntezt-Smtttvtty and Serrtuntrc &ltnimalbm, Oxford Clare& -.I

Pre\s, 3-22 Cappelerr, F1 and Lepore, E (1997) 'Vaneties of Quotation', M n d 106 429-50 ('appelen, I I and Leporc, E. (ivy;)) 'Rephe, to the Commentaries' In K Muraugt

arid K Stantort (cds), I'hilusophy and Llngut~tics, Boulder Westmew Press,

27")--8> Cap pelen, 1-3 and Lepore, E. (aooja) Inrevtstttve Semantru, Oxford Blackwell ('appden, fJ and Lepore, E (zoo5b) 'Vxietres of Quoahon Revisited', Belgtan

Journal oJ lrnguutt~s 17 51-76 Cappelen, £4 and Lepore, E (2007) Language Turned Upon Itself, Oxford Oxford

Uruverslty Pres? Carnap, f\ (193 7) ?fie Logrcal Syntax ofhnguage, London- Routledge and Kegan

1h.d Carruthers, P (1996) Languiige, 'I?wught, and Consnousness, Canibndge Carnbndge

Uruvers~ty Press

Page 311: Recanati, François. Truth-Conditional Pragmatics

Carston, R . (1988) 'implicature, Explicature, and Tmth-Theoretic Sexrl;lntics'. In

R. Kempson (ed.), iW'rntuE Rejrue~lt~ziiom: the Itztedkc~, between I ~ t g t a u g ~ 'zrzd Reality, Cambridge: Cambridge Uiliversity Press, 155-8 I .

Carston, K. (1997) Ztxicbrneut and Loosening: Con~plexrrentary Prrrcesses in

Deriving the Proposition Expressed?', lingidistishe Bevichte 8: 103"-.27. Carston, R. ( zooz), ~fiotghts and Ilttemtzces: ?'he I'rugmatia o f Explicit Cotn~r~unica-

tion, Oxford: Blackwell. Carston, R. (2004) "I'ruth-(:onditioi~d C O I I L ~ ~ C and Conversational Inlplrcaturt:'.

In C. Biariclli (ed.), The SertlarztiisiI'rugrntztir.~ Dstirlctittn, Scanftjrd: CSLI Pub- lications, 18-48.

Chierchia, G. (1999) 'Linguistics and Language'. In R. Wilso~i ant1 F. Kcil jecis),

?he iV1IT Encycltlpedia of the Cc;qnitirtc Sricnctls, Cambridge, MA: MIT Press, xci-cix.

Chiercfiia, G. (2004) 'Scalar Ixnpljcat~ires, Polarity I'henornetla, ar~ri the Syntax/

liragmatics Interface'. In A. Beletti (ed.), Strt~ttnrcs and Bt,yctrrd, Oxffrd: C)xf%rd University Press, 39-1 03.

Chierchia, G. (2006) 'Broaden Your Views: Iqlicatures of Llomain Widrriri~g and

the ''Logicality" of Language', Linguictic Inil*lity 37: 535-90 .

Cherchia. G., Fox. D., and Spector, B. (ii>rthcommg) '"'The Grniiimarical View of

Scalar Implicatures and the Rdationship bet~veeri Seixiant~cs and X'rrag~n:~trcs'. In P. Portner, C. Maienborrr, arid K. vor~ Ir-teixsinger (eds), I-idtzJhc?ok qj'Sen?antitli:, Berlin: Mouton-De (;ri~vter.

Chierchia, G. and McC:onneU-Gitrot, S. jic)(~o) lt4ttanirg and C;vtztlmnldr, C'a~?~t>ricige,

M A : MIT VressiBradford Books. 7 ' 10 9- --' Chomsky, N. (1976) 'Con&tiom on Rules of Grarrurur" L,inf~cistic .4~talysis. ,. 3 3

Cbristensen, N. (1967) 'The Alleged Uiscirrctiorr between LJse arltl Merrtio~~',

Philosophical Review 76: 358-67.

Clark, E. (1993) The Lexicon in Ayuisitiotz, Cambridge: Czunbridge Ifnlverslq Press.

Clark, H. (1996) Using I-ungu'zge, Cambridge: Cambridge Urliversiv Press. Claik, H. and Gerrig, K. (rg8q) 'On the Pretense 'Theory oC Irony, Jouraal t$

Expeuimental Psycholqqy: Gerlerdl, r r 3: 12 I -6.

Clark, H. and Gerrig, R. fryyo) 'Quotaaons as Demonstratioris', L~ttqulxgz 66: 764-80s.

Clark, H. and Wilkes-Gibbs, I). (rc~Xfi) "Referring as a Colla!>orativc Process', Cognition 22: 1-39.

Cohen, L. J. (1971) 'Some Remarks on <;rice's Views About the Logijrlcal Partir:les of Natural Language'. In Y. Bar-I-liUel (ed.), I~frtrtqmntics of Norurul I ~ z q g z t i l ~ ~ : s ,

Dordrecht: Reidel, 50--63.

Page 312: Recanati, François. Truth-Conditional Pragmatics

?(>ti I<FFkI< kN( ;I-!;

i :tairc*il. l . f . (1977) (, ~ . t r i tllc (I:o~ivcrratr~~u,tlrst 1 lypoti~esis He ii kfended?', Wrila- .;tiphai-~ii S/uidier 3 r : 8 a--go.

("oircii, i . /. (rg8o) " f low is <:or,cepti~al Iru~ovrrtrorr Xbsslble?', Erkenrstrtic 35: 221-38. i :opc.it,iki., A. ,iiid ihrisc-oc, '1'. (199.2) "1 c x i c : ~ i Ojieratrc>ns in a Urufication Based

I:r,iiii~\wtriic'. j i i j . L'~~~t.ejovsky rind S. IJrrglcr jeds), r".exiciil Srm'zntirs nrzd Know!-

etlqc Kr~mri,ritaticvi, liertirr: Sprirrgcl., 101 16).

C:or:i~xa, 1;. ( . : i io~) itc:jlcc.fir~q the Mind: Incicb"~iric.iir~ izviti Qitk~i-lndmicuiily. Cl)xford: C :Eai-ciiiior~ 1'r1"~.

( oriirr her, I$, t i c - j r 078) 'I .'irrt.rsc. la ciassz cii:?; vrrbes p;rreiltil ttiqtles, et le sig11e

r i a r r ~ r ~ c j ~ i i . ' , I,'idhrcbrc dc. Iiuz,q~.ri.~ri(~r~ct S : j; -i>j.

( ' trrr~riii~r, 1%. di. (1t~8-j 'SLIT 1~: S C ~ S ~it"s ililestloiis tcpt:dcs ct dtern,~tives', Lancqu,aes

0,7: <#5 roc). ('oniciEr~.i. I<. tic (n"i.t), 'I'oiir li':rtnalvre nuiirrtr~iiste iic c-ertrunes cspressiorls de

qu,riitlti.'', 3oi~niid ~) fX) r~ i~pv t ,z f i ( s W: 001 o i . ( 'nrr~i~-rxi\s, n/i ( iyc)~) 'l;xlk itbi)zrc Bclicy; Callbr~dge, MA: MIT Press/Uradforci

Ii!><>l\~. i ' P , i o ~ k o ~ i , I ). (i(>:i/) '(JLIOY:~CIOXI'. I? C ~ T I X I E C " ~ iii 111s Iksuys or? 7"riillz urld Interprrtution,

Oulirrd. ( : l , ~ i c ~ i i t i o r r hcss , I()#.$, .i()-i)-?.

i l%i.:tb3ntc~.. I). (id.) ( ~ ~ ( i j ) llyirvid C>riotiiticins. Ai~lsterdarrl: John Werlja~nins

(" Xk,iqi,in . b ~ i r i l m l ~ ! f l . iqp~aist i iq 17, zoo.{). v,cn i icciriccr, K. ,ir,tl Ikec.rrs, S. (eds) (~096) "l'cmt~rzfi.~*lmb~prsity dnd I-Jndcnpec$c:utiolz,

!it:illiiM 8 1 " t;sl,l l+l!?l ic:;itl<>tls.

1 >(. Sw'ut, X I . (1908) Iizzroil'uitic~t, EO ~ \ ' ~ ~ L S I ~ ; E ~ IAS;II?~EII!I(C .$ern(iritic~, Stanford: CSll ii\iahiii-'~triiris.

I )tii.n.ot, I ). ( i ~ c i c ) ) ' l"ri.siippo.;is tbr: soirc -eiitrrrciws ', L A z ~ i q z ~ t ' Fr~iizpi-qe 4: 30-43.

I ) r l i i o i , 0 ( r y 7 z) Iirn' r l tic- pas iir'rc,, Xx,rrxs: 1 Irrr~r:tnn.

1 ) ~ I L I or, t ). (197.5) 1 . 4 2 prixiitli, ci /t, tdiri,? PJCIS: Mcil~~ts.

A liri-rot, ( ) . j,i,Xo) 'A i~ i lys ib tic: t.txttr ct / i r i g~~s i i quc" Je i'cnonc-iirtion'. In 0. L>tlcrot ct L Z ~ . ~ l.tv .L!ol.s i l l4 ! )tsct~urs. l%r~s: MIIILM, 7 56"

1 )tic.rt>o, 0. (1984) ')Jsciu~ss~' ii'r~ne tlii;.orres polyi_lkloirrque Je I'i~~o~lciatioir'. 111 >. 1 h~c ro t , 1.t l > i r ~ P! I!(, l>i/. P:ir~s: h4ii111?(, 171- 233.

i:gg, ha. ( ? c i r 1,) I.?c.rihErb .Strnrrfrtici ,fbr litrivsl~~.xy~uruiii~)t~ I%errt~r?zencr, Stanford: C:SI ,I ~ ' ~ ~ ~ ? l l ~ ~ < ~ t ~ ~ bns.

ki.lirourric, X'. (to08) ' L'lii. Arg~~nlcut frc)rn tlincbng", l'\lrlc)s(c~~liirul ,"ersj~e~tives aa: Sp--I 10.

I:;iart.trxiriic.x, 6;. jxo;rh) "K.cr~~:irqut's sur [:I rlicorrc ~ i e s pl-iiiiom2nes scalairrs', b i ' ~ l ~ ~ ? l f i k c ) s i (3): 13 30.

I'~iirx,circ.. (: ( i c ) i r c l ) ' 'l 'ypesuf l , ex~cd Irihriiution'. III f:. ICit:fer (ed.), Sttidies in

Syntdi.\ crnil St:rtriarriit.s, l)c>utlrcci,t: K.eicit.1, r 09 37.

Page 313: Recanati, François. Truth-Conditional Pragmatics

l:illrnore, C . (ry8r) 'Yraginatics 2nd tkle L)escnpti()ri ofl)iscoi~rse'. Irl 1% ((:ctle (ed.),

Iiudiiul f+cgrrictfics, New York: Acaderruc Ress, 143--66. Fillmore, C:. j I ($36) ' f'ragr~latically C :ontrolled zero Anapir o~-;l', I)roc.c.clilh?f.s :sf t i le "

Lzitznlrtll 1'vIertir;y ($.tilt NerkeE(3y 1dt;~uistics SL)C~C~Y 1.2: 05- 107.

FiUrnorz, C. (1907) 1 ~ c t l ~ r r : c an 1)cixi.r. St;~nfOr~t: CSI.1 Pt~hl ic t l t i~ i~~. Fodor, J. (200~) 'T,anguage, 'Tliougl~t :*lid C:ompositiondity', it/livtJ and I~rr~~qna~gr 16:

I--IX.

Fodor, J. (2003) Nunw Ccrri~ition~, (.)xford: C)xfbrti U~rivcrsity I'rress. Fox, 12. (zoos) 'I~r~plicatt~re Calctllation, I+uagniatlcs <,r Syrlt.ae?', (ll;tss handout,

Ecole No~xnxle Sripkrie~tre. Fox, I>. (2007) 'Fret. Choice and tlle 'Theory oEScdar X~npiit:atl~r.es'. In U . Sauer--

land and P. Stateva (eds). Z-'rest~~)position and Irrtpliicrturc. irl (~~oti~posifittnd Sen~untici,

London: Pdgrave Macrrrillarr, 7r - 120. t:ox, 1). 2nd 1 Xackl, M. (2006) "J'lle Ul l i~~rsa I I)es~si~)* ~CMeastlt-t~nlrrlt', Li~tqwistil:~

ur?d Pkiiosophy 20: (37 -86.

I:ra~la, I. (~001) The Coilstituents of'Meanirtg. ,lls. Garcia-Carpintero, M. (aoo5) ?J)ouble-Dtity Quoration: 'l'lrc I)ef-i-hi-red Osrensit,n

Account', fkkiun Jonunir1 qf~lincqtristiis 17: 89 r OX.

Gardent, C. (2005) 'lnterp~-t:t:iti~j~as nli~~inlales ct inccrl,r(.atiorxs xrtnxi~nJes'. 111

F. Corblin and C:. C:ardcrit (eds), Intcq>rt;teren (:ontcxtr, tbris: I;avoisier, 271)- 303. Gazdar, G. (1c)7y), fr~z~rriufiu., London: Acacienrit: I'ress. Geurts, 13. (10g8) 'Sc:~lars'. Irt 1'. 1-urirwig and I3. C;rurts (eds). Ix.x.icirlis~lre ,Ct,mantik

c n ~ s ICi?qnifivrv Sichf, 'l'iibingen: C>urlter N a r ~ , 05 r 17. Geurts, B. (LOO<)) 'Scdar 1rnpiic:tt~ires 2nd Local PL-ag~rxirics', Rii~lil dl711 Lii~gudg(5 24:

51 79. (ie~rrts, U. (LOIO) Qui~ntily It~l~~liiaturc.i, (:arribri~Igt:: C:nr~ibridgt~ liriivrrsity I'r-ess. Gctrrts, 13. and Maici-, L.. (200j) '(,)ilot~t.ior~ in C:OIIICST*. lj~~Iqitlr~:~o~r~tal tlj i.itiqr~iitic.~ 17: J C J ~ ) - - ~ X .

Gortiez-l'cxre:~tc, M. (2005) 'lierrl:~i-kr 051 1mpul.e (2uot;ttioli" iBe{qiur~ Jourrzcrl qj' Lir~.pisfics 17: 150-52.

(;reerr M. (1998) '1)irect liefererlcc. arrd In~plicnturc', i)iliiclsoyhicui Sbtdics o r : hr- 90. (;rice, I-". (1957) 'Meailirrg'. I~!li~osoj~~~ic.~d R(~vic.~u 60: 377 88.

Cricr. f'. (1989) St~ddies in the 1 V d y r$' Words, Cairtbridge, MA: Fiarv;~rct University Press.

i i d , A. (zoo8) 'Frer Enricl~inent ur Hidden Incicxicals?', l\4inti c r r d l*rnVqrtqyt 23: 426-jh.

Ileim, 1. (zoos) Ycatures of pranctulis: the sen-tantics of' rrtirriber, gender, anti person'. Lectures at Ecole norn~zrlc st~pci-riclir-i: (Jantia~y LOO^), ordint at http:// www.tiif&ision.ens.tl-/ [last ac.cessed 16 Noveirrbc5r 200gl.

Page 314: Recanati, François. Truth-Conditional Pragmatics

Heini, I. and Kntzer, A. (1998) Semantics in Generative Grammar, OOxford: Blackwell. Il;sSinbothan, J. (i988) 'Conte'uts, Models, and Meanings'. In R. Kempson (ed.),

Mental Reyresetztatiom: 17te Interfclce between Language and Reality, Cambridge: Ckrnbridge University Press, 2 ~ 4 8 .

tlig@nbottnam,J. (2000) 'On Events in Linguistic Semantics'. In J. Hi@nbotham, F. Pianesi, and A. Varzi (eds), Spmking of Events, New York: Odord University Press, 49-79.

Horn, 1,. (1989) A Niiturizl f-lirtory of' Negation, Chicago: University of Chicago Press.

Hornstein, N. (199s) f~gical Fom: From G B to Minirnalisun, Oxford: Blackwell. Iacona, A. (aaoj) Enliglltened Indexicahsm. hrli.

JackendoN; It. (1983) Semantics and Cqnition, Cambridge, M A : MIT Press. Jackendog R.. (1990) Setrtuntic Structures, C:arnbridge, MA: MIT Press. Jackendog R . (1993) 'The Coxnbinatotial Structure of Thought: The F d v of

Causative Corlcepts'. In E. Reularld and W. Abraham (eds), k'urowledge av~d I~nguage IX: I~xical arid Conceptual Structure, llordrecht: Kluwer, 31-49.

Jackendoff, R. (1997) The Arclzitecttire ofthe hnguage Faulty, Cambridge, MA: MIT Press.

J;rckerrdo8: lt. (2002) l~o~lniicltioru 4 Luguage, Odbrd: OxfOrd University Press. Jacobson, P. (1990) 'Raising as Function C:omposition', Linguistics irad Philos<7pizy

13: 423--75 lact.tbsori, Y. (zoos) V~;iriable-Free Semantics: The Case of Quantifier Domain

fitsuictions'. Hantiout, Institut Jean-Nicod. &nip, H. (1975) "Kwo Theories About the Adjectives'. In E. Keenan (ed.), Formal

Semuntii~ qf N~atural L*an91qu<qe, Cambridge: Czarnbridge University Press, I 23-5 5. Katnp, H. and Partee, B. (1995) 'Prototype Theory and Compositionality', Cogni-

tion 57: rLg-91 Kanv, I-!, and kteyle, U. (I 993) From I>iscotrrse to I~tgic, Dordrecht: Kluwer. Kaplan, C). (1978) '1T)TIiATT', Syntm and Semantics 9: 221-43. Kaplan, D. (1989a) 'C)eioristratives'. In J. Almog, H. Wettstein, and J. Perry (eds),

?7zemajom Kayian, New York: Oxford University Press, 48 1-9~3. Kagan, 13. (ry8gk) 'Aftertl~oughts'. In J. Almog, H. Wettstein, and J. Perry (eds),

7Tzerna-bom Kaylan, New York: OAdord University Press, 565-614. Karttunen, L. (1073) 'Presupposition of Compound Sentences', Linguistic Inquiry 4:

169-93. Karttunc;n, L. (1974) 'Presupposition and Linguistic Context', -fieoretical Linguistics

I : 181-14.

Katz, J. J. (1972) Semantic Theory, New York: Harper and Row. Katz, J. J. (r977) Pri7positionul Structure and Elocutionary Force, New York: Crowell.

Page 315: Recanati, François. Truth-Conditional Pragmatics

Kempson, R. (1993) 'Input System, Ariaphor;~. Ellipsis and Operator Binding'. In E. Reuland and W. Abraham @cis), Xflo~dedge and Lngttu<qe 11: h .x i cu l and Corzceptual Structure, Dordrecht: Klumr, 5 r - - f l .

Kennedy, C. (2007) 'Vagueness urd Gramnsar: The Seiilanucs of Rehdve and Absolute Gradable Adjectives', Lir~~uistin and P~zilosctylzy 30: 1--45.

Kennedy, C. and McNdy, L. (2005) "cde Stnlcmre and tile Semantic Tyl?c)Logy of Gradable Predicates', I~znguqqr 8 I: 345- 8.1.

King, J. (2007) ?'he Nature and Stnlcdurcc gm Corltent, Oxford: Oxictrd Uruversi? Press.

King, J. and Stanley, J. (2005) 'Se11lmt.i~~~ P r a p t i c s , and thc liole of Senrantic Content'. In Z. Szabit (ed.), Smantics vt:rstts F~r~zg~n~tii~s, Oxikrd: Glarendon Press, 11144.

Klein. E. (1991) 'Coniparatives'. Irr A. von Stechow and D. W~lniierlich (eds), Semuntik, Berlin: de Gruyter, 673-91.

Klein, W. (1994) Time in I~rgtirge, I,ondon: Routledge. Kiilbel, M. (2008) About Concerns (Sonic C:ol~ments about Recanati's I3c.rsprdiaial

Thought). kfs. Kripke, S. (1977) 'Speaker's Reference and Semantic fief.krencc', Midtt~e.~t Stlrtlies Stz

Philosophy 2: 2 53-76 Lahav, R. (1989) 'Against {2or1lpos1lior1nlity rlte Case of Adjectives', PIiiltls~~p!iic~xI

Studies 57: 261-79. Landman, F. (2000) Events ~rzd I-"lr*nlliry, Dordrecht: ICl~iwer. Lasesohn, P. (1999) 'Prag~natic Hdos', la,z,g~cq~ 75: 532-51. Levinson, S. (198;) Prugmatiu-, Cambridge: (Zan~bridge Universiry Press. Levinson, S. (zooo) Presrrrnptivc. r\;lcarrin,us: ntc, Tlicory 01' Cenpmlizpii C~nvr.n*tion~ab

InapIicirtjrre, Cambridge. MA: MI'T Press.

Lewis, D. (1980) 'Index, Context, and (,ontent'. In S. Kanger anti S. c)bcn;~rr (eds], Philosc7phy unii Gramrnirr, Dordrecht: Redel, pp. 79- roo. lleprilirttl irr his Pulpen

in Philosophical LA&, Cambridge: Callbridge University Press, 1998, 31~4~;.

Lewis, D. (1983) 'Scorekeeping in a Language f;lvn~e'. In his P!zilitsoplri~"al E"iq?c*r.s, vol. I , New York: Olrford Llniversity Press, a 33--49.

Ludlow, P. (1999) Smantics,'l'msc, (2nd 'Tirrie, Garilbridge, MA: MIT Press. Lyons. J. (1975) 'L3eixjs as the Source of Itefererrce'. In E. Keenan (ed.), E20vmal

Semantics qflViitura1 I~n~qciagr, C:ambridge: Cambridge University I'ress, 61- 83. McConnefi-Ginet, S. (1982) 'Adverbs and Logical Forni', L*trzfut(qe (8: r44 -134.

Mackie, J. (1973) 7Ludz, Robability iztzd Riradox, C>xtbrd: Clrrrentforr Press. Maitra, 1. (2007) 'HOW and Why to Be a Moderate Cotltexta:disc'. In G. Preyei. and

G. Peter (eds), Context-Sensitivity and Semczntic rldninzalism, Clxfbrd: Clarerldori Press, 112-32.

Page 316: Recanati, François. Truth-Conditional Pragmatics

"flar.ci, 1 . (~cio0) 'Lfii:trr;ic.rrbtd C:oristirucx~t\ Xlevisitctl', 1,inqrlisfics rmil I-'hilc~soplzy 29: 1.; 5 60.

May, I<. (a08j) 1nyic;ti [:bntz, <:an~hntigc, MA: M1' 1' L1rri.ss.

MrtchuEl. J. (1987) ']?.re I:clnnal .Vvt?taiztics (?f l-'oirit (!/' Liic~w. PLil ) I)issertatinn, I)c:part- rrwrrt 116' I .r rigtirstic.~, i,_lt~lvcrsrq oi. Massachnset~~.

Mnrgar1,J. (ru;rS) " l 'wu 'l"ypc-s of (.:c>rivcriRtrn i n lr~liiretrc Spcecl~ Acts'. S y ~ t i i x l~rzii Setnailiio 9: LO* -Xi>.

Mcw~it, A. (LOOH) 'lnrentions, Ciesturcs auld S:&erlcc in C>i-d i~~aiy and I)ef?rred I >eiiitrnstr;invr Kr~ercnc-e', iC/Iind trnd I d ~ r l g 1 4 ~ ~ ~ ' 2 3 : I 45 -64.

Ncde, "d ( l ~ ( i * > ) '(:olur~ng ;lxlci C:ompusncitrn'. In K. Munsugi a11d K. Stdirlton (eds), l>h;"C)iol7ir)t crad X.ir;qtbi\tics, Boulder: Vd%esr\ricw~, 3 j -82.

Nr;ilc, S. (~croo) "111 hemg Exldicrz: C:ornrneti& oil St~nley and Szab6, and on I+J~ .~I ' , 234i~~l i ' 1 t ? d I.dfl.qti'l<gt~ 1s: 284.--94.

Nc,aIc, S. LOO'^) 'C )11 I~C~QO~X'. 111 M. ( )'Ktrrir/\cl dncJ C. Wnshingo~l (ecis), .5itcr'iting ,";erndnt it3: i i s i a y on t11e I%iloco!)hy of ' ,'&,/?a P+rry. Ca~xxhricige, MA: M l I I"res, 2, 'j i -9.4.

k i o v c ~ i k , I. cr dl . (2,002) '1 . ~ z ~ g ~ l i ~ t i ~ ~ - l k l a g ~ r ~ a ~ ~ c 1'a;ii.turs in Interpretit~g lli~junctions',

'Ihiflil:ir\q L i r r L E Rnucrlii#!q S(.l): so.i.--3~(i.

Nriurbecg, (;. ( I V ~ O ) " ' b ' i i i . Nori-tiilrijrrenlccs oC Cerrral~rrc Solrrtic'r~~s: ii-"olyserr~y', Iainfiuktiis Litjil 3 " h i i o ~ ~ ~ ~ ~ ~ 1 ~ ~ 3 : 143 -84.

Niriibc:rg, 4 ;. (iiiyci) '(7re Iiyquarics ~fIlrt~ir~aiio;orr. St:inli>rci: c :SI,I. Yuril,c:rg, (i.. ( 1411>2) ' Two X<ill(ls (EIIx~IIcLX~C;C~~CV'. IXX C. Barker arrcl I>. I3t>wty (ecis),

I)rc~i.cit:dirt,q> o f ,Y~r i i , " I ' IC , ' I ' t~ t : O h i o Stxtr linrvt.rsrc-\j, Workiixg i'aperr iii Lir~guis-

tics no. 40: 2x3-301

Niiriljery, C;. j i i ~ j ) "iiiciexicaliry and L)e~.rs'. d4it<qrdislic.s utid Pf?i/o.so~?hy, ~(i: 1-43.

Nunberg, (;. ( r ~ ) c J $ ) "'I'r~msfes~ of Mt-a~itng', .7o idn7~! qf'Serrzantits 12: rap-32.

Nrrnberg. C;. ~ r i i l Xarrlcrr, 1%. (r[):~s) 'Systtrtnatic. X\olyser.ny in Lexicolohy 2nd I.,exrccsgr:ipiif. Irr i l. I'orn~x~ol:~. I<. V;irar~tol:i, X ' . . 'I'oioneri, and j. Scl~opp (eds), r"%oi.cic?,c[iir~s c!fliunzle.~ 2, I.J111vcrsXty ~ > ~ ' I ' I I I I ~ c ~ c , ['art TI, 387--98.

1>>1gi7?ji, rjl ( L U O S ) '(:o111?~6)(os~tio1iaIitv ;LYIL~ C:C)II~ZXC'. 111 (;. 1)reyel- ;inti (>. Prtcr (a&), C;)nk*wvizlut~l"i>;izr i n I%ik)-sof)izy: Knowlrl{qi., Mc,i~niw, ur7d Thith, Oxford: Clarendon ~ Y s s , 303 -48.

Pagxi, 1" iaritl X.k4lrt1e:i-, j. (2007) 'C:orrtrxt, i:oritent ; a i d (:or~~murlication'. In

C;. I'reyrr :xrrd C;. Ik t r r (eck), C,'untc:~~t-?;c.r~-siti~~iiy utrd St'muntic ~Vfinimulkln, C)xfirril. l:larerri"ic,r 1'1-ess, 2 5 4 ) ~ .

k%arit~vu.ira,J. ( ~ 0 0 - 3 ) 'So111e Lssue:s ofsyntax ancl iierrrarrtics ol'Vcri>al Modifications', M'l"E 2003, Par~i .

fxarsorrs, ' r. (ioc.so) I.'rrintrr in the Sarnantitx ($' I r~~lzsh , I>i%partntcnt of I . i~l~i~ist ics, U( :l A

Page 317: Recanati, François. Truth-Conditional Pragmatics

I-'artee, 13. (1973) ‘Seine Srructurd Rna1t~gi:res licstweeri 'f'enscs ;inti I%ront-iur~s in

Cilgiish', Jc~ttrrlal of l'iiifosop,yhy 70:-60 i -9. Kcpriiztetl ill i",rrtei- (.iool), yo 8.

Partee, B. (1984~1) 'Co111posihon:dity'. t r l t:. X,atld~r~ail .*iltf F. Veltil-tari (c(ls),

Varieties of Forrr~al Senzuntici, t)ur<lrecht.: Fctris, LY i -jr;. Kt-prrilt-ed 111 l"artce

(2004), I 53-81. Partee, 13. (198411) 'Nominal allti Il'ernpord Aliaphora'. l . i ~ ~ ~ u i , f l i c rrnd I"i~ili).soplzy 7 :

243-86. Partee, M. ( I $39) 'Binding i~npliclt Variables irk (Jirantif-lrd (-:urrtc:xts', CXS 25:

342-65. Ikeprinted in Partee (znoq), 259-81. Partee, B. (1998) 'Lexical Sernxltics ;md (:ort~~>t~sitioxrdity'. 111 I). C)sIlerson .xnd I i .

I,astuk (eds), r4tr ltzvitatiotz to Cogniti:le S'C~C;T!FWC(:, VOL. I , 211~1 edn, (.;arrlbridgc. MA: MI'T I'ress/Hradft)rd Books, 3 I I -+)o.

I';u-tee, B. (2004) C~r)tnr)u~i/iona/ity itz Fortnit/ Scnr~zrzti~s, f_)xfurci: Blat-kwcli.

Partee, 33. (fortl~cotning) "I>rrivative Ad,jectives: Stlbscctivc 1.rlus <:oc*rc:iorl'. In

13.. 13%~1erk, U. 1te$c, and 'r. E. Z i ~ ~ ~ ~ ~ ~ e r i r t a r r n (cds), f-'re.~r*pp~si/ioris itrid I)iicoullie. Ainsterdnni: Elsevier.

Penini, E. (forchcomit~g) 'Does C:oritestuatisrri Make C.:orl11rlunic:at1t3i1 a Mlraclc!' To appear in :\/flirt~~irito.

t'cny, J. ( 1986) "l'htr~tgbt witlltrur R epri~ci~r~t;itic m', I)rt~ccetliv~~s qf thr rln'stc~tciii~rr

Society, sup. vol. 60: r37--j1. lieprinted (wit11 a potscript) irr fJt-rry (~oocr), 171 -88.

Perry, J. (LOOO) X R t Problem ofthe I i~sml iu l Itz(i('3il(.<i/ /rr[d (.Ither E55i~ys, cxp;tntleti ettn. Stanfiord: CSLI I~ublications.

Perry, J. (2001 ) KeJ2revlce arzd frzdt.?ric~rlify, St~nthrd: (:SLI. Potts, C:. ZOO^) Rze I e i i of C3orzver1liotiu/ lrr.~plii~~tcires, Oxfitrd: Clxtord IJnrversity

Press.

Ibtts, C. (2007) "l'lze Ilin~ensiorrs ok' Cfuotation'. In C. Barker ailti V. .Jacohsori (eds), Direct <~orr~p~?siiiorzality, Oxiibrd: Oxfi,rd Urriversity Press, 403- 3 1.

Predelli, S. (ryyli) 'Utterance, I~tteqretatio~l, ; ~ n t l the I.ogic ~~.-IIIJcsIc:~s', iLJi~?(l linii

I ~ r g u a g e 13: 400-14.

I'redelli, S. (2003) 'Scare Quotes and Their ILrlation t.0 (3tller Scrnaritic Issues', Litgiiislics and P\zilosophy 26: 1-28.

I'redeui, S. (zoo5cl) Contexts: ~Mratiin~q, i17r:dfh, Czntl the I,'.st of I L ~ ~ ( ; ~ M U , ~ C , Oxfi,rti:

Oxford Uiliversity Press.

Predelh, S. (20056) "'Subli~ninabie" Messages. Scare Quotes, the Use I-lyputl-t- esis', Bekiun Journal qfLinquirtiis 17: I j3-66.

Prior, A. (1971) O h j e ~ (?1'Tllzou'q/lt, ed. by 1'. Geac1l aid A. Keriny, (.)xford: Clarendon Press.

I'utnam, 13. (1975) 'The Mearii~lg of "Meaning"'. 111 his Pizi!oso~thii'~l I'upc.rs, 1101. 3: Mind, I~wqtl trgr and Reality, C:arnbridge: (larrrbridgc Un ive~ i t y Press, 21) -71.

Page 318: Recanati, François. Truth-Conditional Pragmatics

312 REFERE . -- .- -- - - -. ---- -. . -

Quine, W. V. 0 . (19 51) ibfatlzernutual Logic, and edn, Cambridge, M Harvard University Press.

Quine, W. V. 0. (rg6o) Yariables Explained Away'. In Iris Selected Lcrgic Papers, Chnbridge, MA: f-iamard University Press, 1995, 227-35.

Quine, W. V. 0. (1962) Methods o j l q i c , and edn, London: Routledge and Kegan Paul.

lxauh, G. (ed.) (1983) Essuys ort Deixir, Tubingen: Narr. Read, S. (~997) Quotation and Reach's Puzzle', Acta Analytica 19: 9-20. Recanati, F. (1079) I A Tramparence et I'Enonciation, Paris: Editions du Seuil. Recanati, F. (1~81/1987) Meanirg ilnd Force, Cambridge: Carribridge University

Press. (Eng. trans. of Les Enoncis pdomzat$, Paris: Editions de Minuit, 1981).

K.ecanati, F. (1987) 'Contextual Dependence and Definite Descriptions', Proceed- ings ofthe Aristotelian Society 87: 57--73.

Recarrati, F. (1989) 'The Praginatics of What is Said', Mind and Language 4: 295- 329. Kcprirtted in S. Davis, S. (ed.), 13la'qmatics: A Reader, Oxford: Oxford University Press, rt)gr, 97---120.

Recanati, f;. (1993) Direct Reference: From Language to nlought, Oxford: Blackwell. Recanati, F. (ryg5a) 'Le prOsent Cpistolaire: une perspective cognitive', L'lnfomu-

tiun grammatic~zk: 66: 38--44.

Recarrati. I;. (19yjb) 'The Alleged Prioriry of Literal Interpretation', Cognitive Scieni~ 19: 207-32.

Itccanati, I-:. (ry96) 'I>otxrains of l>iscourse', Lingt~istics and Pjiil~1sophy 19: 345-75. Recamti, I;. (1997) 'Can We Believe What We Do Not Understand?', Mind and

Ldr~bguxe 12: 84-- IC)O.

Kecanati, F. (1998) 'P~ragrtiatics'. in E. Craig (ed.), Koutledqe Emyclopedia 4Philoso- phy, London: Routlerige, vol. 7, 620-33.

I-Cccanati, F. (tc)c)c)) 'Sittlatictns and the Structure of Content'. In K. Murasugi and 1-I. Stainton (ecis), l)!tik~~soy)hy and Linguistics, Boulder: Vestview Press, 113-65.

liecamti, F. (~ooocr) 'Opacity and the At~itudes'. In Orenstein, A. and Kotatko, l'. (eds), f(rzowle4y:, Innguczge and Ln'qic: Questions for Quine, Dordrecht: Kluwer, 367-40(i

Recxnati, F:. (zooob) Clrutio Z)bliqhtu, Orutio Rectu: A n Essay on ~C.feturepresen~ution, (.?anrbritfge, MA: M1T t-'rcss/Bradford Books.

Recanati. F. (200111) '(Jperi Quotation', '!Mind 110: 637-87. Kecariati, F. (zoorb) Are "here" and "now" indexicals?', Texte 27/28: ITS-27. Kec:mati, F. (zoorc) 'Irnplicit Arguments and Otlier Unarticulated Constituents'.

Conmitinication to SPR 1 , Donostia-San Sebastian, 23 November 2001.

liecanati, F. (zooxd) 'What is Said', Synthese 128: 75-91.

Page 319: Recanati, François. Truth-Conditional Pragmatics

Recanati, F. (zooie) "Utstabiliser le sens" IRrvut: Ivifem~ztiowle de I'failosoy~hic 216: 197-208.

Kecmati, F. (3002) 'Unamculated Coi~stituerrts', ljvgrni~ricr and I'fzil~>.soplty 25: 399-34 5.

Recanati, F. (2003) 'Embedded Iqiicaturczs', jnlzilosoplzi~al 1-'erspective.5 r7: 299-3 32 .

Recanati, F. (2004) literal Matling, Catlibridge: Cambridge University Press. Recanati, F. (2005) 'Ueiuis and Anaphora'. In S. Szab6 (ed.), S~mnrziics 11s Pr<gnzatics,

New York: Oxford University Press, 286-316. Kecanati, F. (2.007) Perspcctit~ul Tlzoiiglzt: il Plea-fir (A,.:lt~demtclj Rclrrtivism, 0xcibr.d:

Clarendon Press. Reimer, M. (1991) Ben~onstratives, Uernoristmdons and I~ernonstrata', Hrziimoph-

ical Studies 63: 197-202. Reirner, M. (2005) 'Too Counter--Intuitive to Believe? T'ragrnatic Acco~xnts of'

Mixed Quotation', Bckiara Journal c?f Idirl,uuistics 17: 167-86. Roberts, L. (1997) 'How Demc)nstratiorrs Goitnect: with Refererttral li~tentiorls',

Philosophical Studies 75: rye-loo. van Kooij, R . and Scliulz, K. (2004) Zdiaustive Irlterpretatiori of C:orr?_plex

Sentences', JOM-ma1 ojLoyir , I.ilrgtiage avid Irtfirmiliion 13: 491 51%

RuUmann, PI. (2004) 'Fist axid Second Person Ih-cmouns as Bound Variables', Linguistic Enquiry 35: 159-68.

Ruinelhart, D. (1979) 'Sorne Proble~rls with tire Notiorr ofl.itera1 Meaning'. In A. Ortonv (ed.), ,Ifetaphor gtrrd IPFlttu~hr, 2nd eiin, Garribridge: C:'m~bridge liniversity Press, r()93,71-8;..

RusseU, B. (3006) 'Against C;~-anx~wticd Gompuration of Scdar linplicat~ires', Jonrnal Of' Semantics 23 : 361.- 82.

Sadock, J. (1981) 'Alniost'. Irr T". Gole (ed.) Ktrdirlzl I-'vu~maticli, Necv Y o r k : Aca-

deinic Press, 2 j7d7~.

Sag, I. (1981) 'Fonxld Semntics llrld Es~rdirignistic Context'. Irr 1'. Cole (cil.), Radical _F'ra'qmatics. New York: Acadeinir: Press, r?73--94.

Saka, P. (3005) 'Quotat~onal Coxrsb-uctions', f ~e (g i~znJ~~unw1 ofLingtrisricr 17: 187212.

Sauerland, U. (2004) 'Scalar Implicarures in Corrtplex Sentences', Lirt~uO~iis land

Philosophy 27: 367-91. Schlenker, P. (2003) 'A Plea for Miixzsters', I.ira'e~rzlictiic and I>krilosc?phy 36: 29--120.

Schlenker, 1'. (2004) ' C O I ~ ~ ~ X F oflhi)t~ghf anti Context of U~era ice : A Note 0x1 Free Inhec t Discotuse and the Iltistoricd I+rescrit', Mirzd arid hrgttcagc ro : L7cP)04.

Schlenker, P. (forrhcomng) 'Inifexrc:dir): and L)e Sr Reports'. In 1'. Pormer, C. Maienborn, and K. vorl 1-teusi~rger (rds) , EIundOook uJ' ,Ye~niznti~i, X-ledin:

Mouton-De Gruyter. Searle, J. (1969) Speech Acts, Carrtbridge: Chnbridge University Press. Sede , J. (1975) 'Itidirect Speech Acts', Syntax uttd Semantics 3: 59 82.

Page 320: Recanati, François. Truth-Conditional Pragmatics

3 14 il. C.I'F:I".i:PJ( '175

Scarie. J. (1980) "'l'kre Barksounti of Meaning'. 111 J. Scarle, F. IGefer, and M. k%rc.r\vri-ir (riis), ,Spet,clz ALL 'lj(l~(~y anti Prq;niiititi, I)ordscclit: ILeidcl, 321--31.

Secrrle, j. (xg8j) Itirt~niioraality, (::ainbricige: (:an~britige University I%ess.

Scarlc*, j. (1,311,~) 'ri?rc Kcdisccwcri, cf?lre :\/lir~d, ('arali~niige, M A : MI'!' I'rres. Sli;:lBi, 1'. , f I,?jrcc)v,r, I?... :inti kJ:anevov~,j. (t~cts) (1980) 'X73ize ~Cfizeanirt~ c f t k e Sentence if? iis

Sirir~cniriic t~rad I'rquriilric ,4spcbi.h, I loidrrc-lr 1: I<i~rdci.

Siegc.1, S. (2oo.4 '''l'hc. I< ole of- Pcrirpnoil in X)cxrlcsrrstrr,~tive ILc&:rrnce7, Pl~iltnu?71zers' Iniprini 2 r

So;irrlrbs, S (fi,r-clicirxljriig} "'l'hi. (.;;tp ber\\iet-ri Mcaxlixrg arxi Assertion'. T o appear in M . I - lacW :rid I t . '~iior-~~toxr ((:cis) Ass~i.rirkq, ;\.1(i~rrsir.g, dnd Inzplyirg.

Spector, H. (200,3) '\;i.ii?~r I~rrplrc.sr~ircs: locai or Glnlxd?' Fkaper prexmttdto the lirvisirrri o f 1 abc~r tvurksboy. l..~ k3rc*rcsci1r, Ji lr ie z o q .

";i"i.cior, U. (LOO(,) . A i j ~ t p ~ t . s de lit pr~~qnscititjuc~ J C ~ t p ! r t i t r ~ ~ n lo~<it/cirs. Yhl) dissertation, 17. P:111s VIP.

iipi.rl,cr, Ll. ~ r r d W~lsori , 2). ( r 0 8 r ) 21rorly ' i l l t i tirc Use-Meuition I?istinction'. In P. C:ok (eci.), X.iatii~~z/ I'r(gz~itttits, Nt:w York: Ac,ide~nic PTCSS, 295-3 ~ 8 .

Sppcr-bcr, 1 ). ,rrrcl W~lsoxl, 1) . ( I g8hLz) iieiivc~rrtr: i,'orrrrrzurzilatit,rz a~zd C'o~nition, C)xior-ti. Lli:~ckwt.li.

Spcrlri.r, I). dx iJ Wilsoxi, 1). (igX(ih) "I.otsc- i',rlk', I)roic.c.dirti:i; cf i l .~c Avistofeli'rt? Sorit'ty

x o : 153 - r / L .

St;iii~rtrri, 1< (ri joa)) 'lSerrl,rrks o r ] rlie ?;>~iltas ,irid Setn-mami~cs of Mixed Q~rotation'. 1x1 I<. M iirahiigl :iii~i It. S131r11011 ( ed~) , I4tif i)sclpi~y &~trcii i . i r~qi* isf i ts, Iic~ulder: Vcst- vrcw he\ \ , 259-78.

"iaii.i,rker, K. ( 1 ~ 7 4 ) 'I'x.;~grm~t~c X~resri~?~~ositiorrs'. It1 M. Mnriitz and 1). Unger (eds), S ~ ~ l a n r i ~ s 121211 [j/t6llil,>sojr~hy, NCLV Yock: Ncw yor-k U~liversity press, 197-314.

Stxi~x:r ler , I<. (itiyv) C:orrtc~xt btnii (lorttt?nr, Oxiilrci: Oxtbrd Uxiiversity I'ress. St;cirle.y, Q. (~troo) Y (:ontext and Y ,ci"gical I:onrr', i.irtqtri.stics ~ n d Philosophy 2 3 : 3y1-434.

Strarricy, J. (300 2 ) 'M'iking It Artic.til;~~~d', iWfzd ifiril Lurgttq~e 17: 149-68.

St~r~ic:y, J. ( ~ o o ~ t i ) 'St:rxXlntics in (:ontext'. III C;. Prever. and G . Peter (eds), (A'onim~uuFisu~r in I%iliiu.so~~h,hy: K~zotirhdqc., h!i.crnilg, und il-mth, CMord: CLarendon i'rei?, 2.21--53.

Stanley, j. (~c io jb) 'Kcview oi l:rarlt;c,rs Krc:iir;rt~'s I.iteral Ibreaninq', ~Votrc Dume i%/~ilocophiiiii Keaac,ii>\, ox~line at I l t t~?: / /~~C/~>~.~ld.cciu/ [last accessed 16 November 2ooc)]

".;a~i,lcy, .I. (*)iooi~.) I.i~~~j(ld~z,qe jn (;OYII(:XI, Cl?;ii>r~i: C)xti>rd [Jniveaity Press.

vorl Stci.ilow, A. (1984) YCc,rrlp,rrrtig Sexx~dntic 'I'hrorirs of ( : o m ~ ~ a f i s o i ~ ~ ~ o n r n a i of S(wcrnricr j: i -7'7.

Ster-n, J. (~ooo) IC?e~~iphor in (:ontc~:\rf, C::ix~rhriilge, MA: MI'1' Press.

St iw i , 1. LOO^) 'Metaphor, I ,ttcr:d, .tiid I,itebra/i\rrr', il/lind (2nd L L Z P ~ U < I ~ Z S ( ! L 1 : 243-79.

Page 321: Recanati, François. Truth-Conditional Pragmatics

Strawsorl, Y. (~997) Etitily ilnd ldcntity, Oxlbrci: C;lareiidorl X%nbss.

Szabi,, %. (moo) 1'ri)blerr~s <f-C:omn~~:iosirit,nalit.y, New York: C;arl:ind. Sxabci, %. ( ~ a o t j 'Adjectives in (~ontcxt'. In K. M. I '~mish ;irrci I. Kerresei (eils),

I-'ersl~~ctives 012 r%,fnmti~, I:>rizgntatus, and Ijiccours~, Airistcrt-lal-rr: L)t:njarruins, r r g-46. 'Tarski, A. (1933) "The Conrc-pt of'X'rtttEs it1 I%rm;~iized I,axlgrrtlges7. 111 111s 1,ogk;

Semantics, Md~lnr~ttI.lc*mcrtia, trans. J. Woodger, Oxfi,rd: C.:Larent-lon L+ess, 1956, 152-78.

'I'aylor, K. (2001) 'Sex, Breakhst, arid L)escriptus Internrptus', ,Syryntlztre 128: 4s--61. Tesrubre, 1,. (1950) Elinzt~tzts dt, cSynt~~xe Sttz~~~ur(zlc, PLIL~S: Kiincksieck. 'I'ravls, C:. (1975) Saying arld tJftcir!r.rtcindinLq, Ostiord: Lilackwell. 'Travis, C:. (1c)g.l) 'C)II Constr-airrts of Generality', 13roc~edirgs ofthe ilristotcrlim Socie'ty

44: 165--88. Uxlger, Y. (1075) [erzorrrntr. (3xfc)rci: C1;trendon I"-rss.

Vuilhn~~le , M . (~c)go) G'u~~mrtruire terrti?orell~ des rGcirs, l'nris: Mjrruit.

Wdker. R. (1975) 'C:onversa~.iond Il-~iplicatttre'. In S. X-ilackburn (ed.), h4carziyq,

R<fcrtmre and fir~ssity, Gax-nbridge: Cambs~cige U~uveriity I'ress, r33 -8 I .

Weinstein, S. (1974) 'Trtlth and IJemonstuarivc.s', ,Vatis 8: 179--84.

Weiskoyf, D. (2007) '(:oinpo~tnd Nor-v*tnals, C:ont-ext, a t ~ d Cctrtlposiriona1ity', Syrzthesc~ 156: 161-2~14.

Westeustalrl, I]. (tbrthcotnirlig) "Cotixpositictnality i r ~ I(;tplan Style Sen1;lrttic.s'. In

W. Hinzerr, E. Mdchei-y, alid M. W e m ~ n g (cds), Yhr lidndhook 4 (:ourzpsitiori-

ality, Oxi'urci: C)sf?~rd University I+-ess. Wierzbicka, A. (1974) 'Thr Serrwntic-s of 1)irrc.t ;-rtld Incixrect. I)iscoursr', f+ipers in

1,ingzlistics 7: 267-307. Wilsuu, 13. (~975). Prcsuppositiaiz and !\;ort-'l'r~iih-C:o~zditit)~~aI Semat~ti~s, L~IICIOI~:

Acndenlic Ikess. Wilson, 1). (aooo) 'Metarepreseiitation ixi 1.iligiiistic C;ox~lrlsrrnical-ic>~~'. 111

L). Sperber (eci.), r2llutarcprrsentt1tions: z-1 ~Vultidiscif~linovy Pers~fectiire, Nrw Work: Oxford UrGwrsity Press, 4.1 I-'@.

Wilson, 1). ;ulct Sperber, U. (2001) "l'ruthfulness a11cl Relevaxice', ~Witid I I I : 583- 43 3.

Page 322: Recanati, François. Truth-Conditional Pragmatics
Page 323: Recanati, François. Truth-Conditional Pragmatics

Index of Names

Anand, P. 97n, 1031-1, 21213, 218n Anderson, S. 2 t i

Anccombre, J.-C. 147-8, 244

Ausan,J. L. 7011, 184, 201, 207. 229

Each, E. 211

Bach, K. r6n, 112, 135, 138, 141r1, rq6

149, 169, 18311. 191, 244. 275 Bar-I-fillel, Y. 2 57. 281

Barwise, J . 112, 102

Beauzbe, N. aron

Beribaji, Y. 272n. ~ S j n . 297n

Benveniste, E. 7011, LIS

Bezuidenhout, A. 17n, I<%)

Bierwisch, M. 10 j r ~

Bloorn, P. 7 Bolinger, U. 7 Borg, E. 2, yo11 Brandonr, K. 3

Briscoe, 7'. 85n

Bromberger, S. 223n Burge, T. 19

Campbell, R. i49

Capone, A. roGn

Cappelen, 11. 2-3, 5-6, 8-9 , r7,43, 50,

2:: -17, 21C, 212I1, 23211, _"J;-;,

239, 242, 250, 261, 264-8, 276,

292-3, 295-8, 300-1

Carnap, R. 233 Carruthers, I?. 137

Carston, K. 68, 92, r 31-4, 14611, 147.

(:hierchka, C. 4711, 105.--6, r28, rjo, 132, 1:{4, 136-7, 152, rj~$--X, 165, t72?

174-80

Clml?zsky, N. 29, 136 Christenser~, N. 22 i

C:krk, E. 7 CLrk, 1-1. 8, 195, LO!), 215, 232--.5, L 3.7.

231, 2 3 ~ g . 239, 247, 262, 301

Colzen, 1.. J. 30, 31, 46, i 50 , 155.-6,

r58.-'>, 17311, r77 Cr,pest:aka, A. 85x1

C o ~ i z ~ a , E. I tin

<:orxu!ier. B. 'it x53n, 163, IBOXI, 195,

227-8. 237n. a68n

Crxrxrxx~ins. N. 121--2

L)avidson, L). j r . 82-;, 89-90, 2 15. sto - 2 , 2:3-8, 240, 263n, 26549

L3e Brahalltcr, 1'. 26811, q o n , arm,

?OII'l, 30211

1311. Swart, E l . 40, 51

L)ucmr, (3. 147-8, r5o, 158, 161, 164,

I!)?-.!&, 202, 23411. 243-4

i>rrrnl?lett, M. 28611

Egg, M. 21, 7011, 13711 E!?,,,,,, 2'. 11, 95, rcjx!, rsq,

r l j , 13811

r,t~rcornucr, G 147 -8, 154

Fdlrnorr, (:. 7yn, So, ~9ir.7, 208-9,

213-14

Fodor, J 30-2, 46-7, ~3s--C"

Page 324: Recanati, François. Truth-Conditional Pragmatics

ib2,ri.k re, J. r B i , 6 ("1

kf;lrc*r, ii 276.. 278 g, 2c);n

iM<iitr't, I. ,711

M:tt-t.~, 1,. \)O, 1 ~ ~ 8 , 113, ~ j g - - ~ p

M:iy, K , r 3711

Mc (:ontxcll -C;irict, 1;. ro j 6, i r i , i j o

Mrtzlic-ll, J. ,761

R/lorrt;cg~~e, it. j i

Morga~i, J. 149

McNalIy, k. 06

R/lo~rnt, A. iXln

Nealc, S. 88- yo. 9611, 138, 24-1, 27511

Nevrns, A. a r rn Noveck, 1. 178, i & o i ~

Page 325: Recanati, François. Truth-Conditional Pragmatics

Pagin, f ' . g, 4 j--7, j:in, 7bn, 12811

J'ancvova, J. 84

tJarsans, 'F. cj:~n, I 05 --0

Pame, B. 73,75.79,08 101, rogn. r 18,

r6Xn, 185- 6, 211

Xbelletier, J. 9, 4jn

13errini, 1.:. 8

IJerry, J. 22, 9611, I 21, 102, 240

Peters, S. 13711

Potts, C. 271, 27511, 570, 2x9, 29111

I'redclli, S. 129, 197, ~ 4 - - - 7 , 25711.

26811, s72n, 2,75n, 18 r - 2

Prior, A. 23 j

I'uma~n, 1-1. I 8

Kead, S. azi

Kvcanati, F. 3 , 5, 7, 12, 1.1, I O ~ , 17n,

20- 3 , Lj, 40 I , 53, 5811, 78, 81--2, SO, 112, 123, 12811, 120, 140, 165, 168,

170, 173n, 179. 18311, 1991, 195-8,

202, 206, 21011, 215, 24J11, 24j,

25711, AS!>, ,262, 209, 273 0, 2S1,

283-4, 201

Keicl~enbarh, I-I. i o t n

Reimer, M. I 8313, 297n

Keyle, U. 206

Roberts, 1.. l8jrl

Kullmann. 1 1. I 8;

Run~elliart, 1). 40 -I

Russell, B. r8on

S;tk:), P. joiri

Satterl;rricI, l J . r8on

Sclilenker, 1'. ~oorl, i i > r n , l.+ln, r R j 1 1 ,

r Y 7, r97. ~ o o , 202 3, , ~ > o 1.3, 2 I j,

218, 168 Scltu', I<. r8ii11

SCYLrltx, 1. 3<l, 33 4, .if), 40, 3, 47, 140,

169, 177. 211, 2.3311, 236

Sgdl, 1'. 84

Sicgt:I, S. I X j r l

Soanlcs, S. 311

Spcctor, H . i ~ 9 1 , I 8ori

Spcrbrr, 1). O j . 05, hS, .7hrt, 97n,

123 .f, 1 3 1 4 , 1~$0n , 16711, 1 7 ~ , ~SCOCI ,

8 , 223, z,%R, 26811,

27211

Slairlto~t, 1:. 2 !oil, 24111, 266,

26811

Sta11i;xker. I<. 16?., r64 5 , 173". 192x1,

21211, 218ri

Stanlcy. J. 4 7, ro r2, r.gil, 27, 90, rr:),

rzg, 166, 1-72 4, 177, 283

Ster11,I. 130 4 1

Str:~wson, 1'. 00

Sz:jhG, %. 5311, j4 8, ')on, 139

van 1)ceilttcr. K. rj711

van I\ooij, R . 18cin von Stcc-how. A. ti2

Vuillat~~tle, iM. zo5, 2og1r

Walker, I t . rg9 ~ O i

Wcinsteitl, S. I9

Page 326: Recanati, François. Truth-Conditional Pragmatics

Wieiskopt; 17. 37n Wilson, D. 63.68, 123-4, 131--j, 172,

VVesters&hl, D. yn, 46, 47ri 177, zXon, ryg, 22.3, 23+1, 248.

Whrder, S. 116--17 268n,272n Wierzbicka, A. 23411 Wittgenstein, L. 7 , 28611

Wrlkes-Gibbs, D. 811

Willian~son, 1'. 22711, zOSn Zaenen, A. Ysn, 118, 169

Page 327: Recanati, François. Truth-Conditional Pragmatics

Subject Index

Adjectives 30, 38, 40, 49-76

Gradable 50-7 I

Intersective 38, 49-53, 70--6

Privative 50, 53, 71 -2

Relative 49-5 5 , 66-8, 7111

Adjunct I In, 84, 141

Austinian proposition 23-4, r14, I

Availabilrty 12x1, 14., 143-6, 157~

165,170

Caricell2bdity 141, 152, 154-5, 175,

254. 268, 290-302

Character 2--3, 16, 37-9, 41-4. 168.

280, 282 -yo

Circunlstaiice of evaliration 13, 2.3- 49 122, 182, 189, z6o-r

(Zoercion 41-2, 69-71

Colour pre&crites 5 3 9

Commutiicatior~ 1, 6-9, 194. 106-8?

s l r , azg , 291

Compositionality gn, 10, 15, as, 27--'+7, . - >;, 62~1, 73, ~ 2 2 , ;;s4, ~ . 5 2 - ~ 4.9

156-7, 168, 244-5, 23.8, 254.-i9

266, 2 7 ~ ~ 1 , 2 7 5 , a98

Cotiditional truth-conditions 3 , 19

Conditio~lals 155, r60-I, 161, 173.

176-7 Context-dependence sce Context-

serisitivity

(loxitext--sexisitivitv r --3, (n, 711, 12,

iIJ-10, 25, 27, 37-9, /il-.{, 46-7,

51, 66, 127-8, 135, 139, 15711, 165~

r68r1, 173, 177, 1x2, 276

Contest--slUft 23n, 2 5 , 18 i-218,

259-62, 26x11, 27y-+jO

Contextual ~lonraixl scstririccioa lor.

r ro -15, ILZ-3, 141x1, 176

<:c,iitextuahstri 5n, 13, 17--22, 43,

467, 5.5 Convelitiolzd i~nplicatore i 5 2 , 16i a,

244.-5, 247, 249, 262, z7.$--.5, 284..

2.87, 297ri

C:i,rm-crsat~on;d iri~plitrnturc I 5- 16, 6.3. 95. 143-52, 156, 158, 160-2, r7u- I ,

244, 255, 278 E.nlbeddrti 25, 14;--80

Sc:d*x 128, 1~913, 148-49, i Sa -5,

157-8, 160-1, 1&p72, 17.5 4)-

178, r80

Generalized 146-5 r Covert optiorrals id(>, I 39-41. r Xnn ~ ' : ~ r n ~ u i a t i v e / i i ~ ~ ~ ~ c ~ ~ ~ r ~ u l n ~ v c s w Hybrid

lXSC

Page 328: Recanati, François. Truth-Conditional Pragmatics

Elcliorc arsc 189, too. 21( j , 2 2 6 8, 2 j i ,

,240 2 , 2.18 r)? "LO 9, 2hSr1, 2711,

z74. -'7'/, 277, %"ac,ii, 2x2, 2x4 00,

a)&!. "k, 3 0 1

I h - j c h u w r ~ t 14, 10, 2 2 5, *$L , \o,

81 t), 01 5? I IL [ , 111 13, I I ~ , I 22.

r ~5~ 128, 134, i.40, i 6 5 , 174 80, 25.5 6, ?'sf), 263, .? 7 0 ( j , LC)() t,

-4 0 '2

X:.citnvo(-ax~o~x" 40 7, 5 r 3 , .ii, 70

I ~ V C I K S 897 41 1

Nty,;ic~t; t : o ~ 6n

s t ~ e ~ u c ~ l z c s of' I.(<> 60

l i v c i i ~ seni,lntrcs X;i, 3

k:xir,irr~txfica~it>ii ?;n

Page 329: Recanati, François. Truth-Conditional Pragmatics

Meaning Elirnirtativism 2 r Mrarung shift 74, I 15-21, r 24, r 8011

Metalir~glistic character 283 .$,

286- 7, 289

Met;rp't~or 4, 41-2, 68, 118, r j8-9,

r?bGri

Metaphysics, us senlanrics 87-94

Meteorologcal predicatcx see Weatlicr

verbs

Metonymy 4, 19, 41 -2, 44. 7011, 118,

128, r jy-40

set also Predicate transfer

Mirrucry see Echoic use Mii~iir~al proposition gn, 13 14, 16, 2 I ,

24, 131 Minim&sni 5-7, 12. 10, 43

Mixed quotation 25, 214 r 8, 220,

236 -45, 249, 75i, 253. 255, 258, r?jc)n, 261, 265-6, 290 --302

Modified Oc.carn's Razor ((:;rice) 62,

155, 299 --joO Moci~~larity a--t~, 131, 13 j

Mociulation 5, 7, 9-12, 16, rc) 25,

39 -47, 55-76, t r r , 125, 128,

r 39-4.1, 168 70, 173, 179 80

Narrator 193, 305, 281

Non-rut nrai tnearting

see Speaker's rirearlirig

Occasio~l rrre;uri~~g xe' Sta~liijr~g ~lti(~tllling

Optiorlal varklblcs 98 -102, 125

Optionality criterion 10, 57--8, 65,

Sq, 140

Oratio recta/&rect speech rqg, ry i~ -9,

203--4, 21 I , 215, 218, 327, 237, 26'1

C)stension 7, ry5, 215, 221-2, 228, 141,

369-71, 291, 302

C3vergeneration 6. ro

(>\iotatic>li 1n:trks LO(,, 21 9- 22. ~3 1,

233 8, 240, 243 t), 252 j, 2.56 8,

~ 0 2 3, 266, 200 70, 373, 27'4. 5,

280, 283 90, 3(>jrl, 296-7, 300-2

Q,)rtct tatiarl

(-:lased zrcj -20. 2.8 -40, 247 0 ,

253. IOJ 0 , zhr) 71, 273 j, L ~ O I ~ ,

287 8

llfybrid sce t-lvbrid use Mixed see Mixvli y~rotat iot~

Opcn 219

Sub-clausal 241, 2'71- j, 282 (;),

29) .2, Lqrt, 3 0 0

c>ucrtatjorlal point 247- 9 , 2110, 293

Page 330: Recanati, François. Truth-Conditional Pragmatics
Page 331: Recanati, François. Truth-Conditional Pragmatics