syncretism · 6 . historical patterns of analogy present tense perfect tense all dialects...

Post on 23-Aug-2020

13 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Syncretism (an example of a form-meaning mismatch) Andrea D. Sims The Ohio State University

Introduction to Morphology 2017 Linguistic Institute

1

¡ Syncretism = When a single inflected form corresponds to more than one set of morphosyntactic values.

¡ A situation in which morphological form is insensitive to a morphosyntactic distinction.

¡ More than accidental homophony. A systematic generalization about the inner workings of the morphological system.

Intro 2

Accidental homophony vs. syncretism 3

Class IV MESTO ‘place’

Class IV VINO ‘wine’

Nom Sing [mjɛ́stəә] [vjinó] Acc Sing [mjɛ́stəә] [vjinó] Gen Sing [mjɛ́stəә] [vjinɑ́]

Homophony in forms of Russian noun MESTO 'place' (nom/acc sg vs. gen sg) (cf. VINO ‘wine’)

¡ Some patterns of homophony don’t have anything to do with morphology

Accidental homophony vs. syncretism

¡ Others reflect identity within the morphological system

4

Singular Plural

1st person spiele spielen 2nd person spielst spielt 3rd person spielt spielen

Syncretism in forms of German verb SPIELEN 'play' (1pl – 3pl)

Accidental homophony vs. syncretism 5

Singular Plural

1st person spiele spielen 2nd person spielst spielt 3rd person spielt spielen

Syncretism?? in forms of German verb SPIELEN 'play' (3sg – 2pl)

(Systematic) syncretism – criteria?

¡ How can we distinguish between accidental homophony and (morphologically-relevant) syncretism? What descriptive criteria can we employ?

6

Historical patterns of analogy

present tense

perfect tense

all dialects

reconstructed Proto-Romani

Central Finnish,

Balkan, & Vlax

Romani

Sinti Romani

Welsh Romani

Northeast Romani

1st plural -as *-am -am -am -am -am

2nd plural -en *-an -an, -en -an -e, -an -e

3rd plural -en *-e -e -an -e -e

7

Historical development of syncretism in various Romani dialects (2pl – 3pl)

Russian ‘case matching’ effects

Ja ne mogla ponravit´sja tomui, kogoi on I not could please that.DATSG who.ACCSG he nenavidit. hates. ‘I could not please the one whom he hates.’

cf. *Ja ne mogla ponravit´sja kogo on nenavidit.

8

ACC SG kogo ‘who’ ≠ DAT SG komu

Russian ‘case matching’ effects

Ja kupila toi čtoi bylo v magazine. I bought that.ACCSG what.NOMSG was in store.

‘I bought that which was in the store.’

cf. Ja kupila čto bylo v magazine.

cf. *Ja ne mogla ponravit´sja kogo on nenavidit.

9

ACC SG čto ‘what’ = NOM SG čto

More Russian ‘case matching’ effects

*On ne ostavil, tak kak emu nadoela, tarelk-u/a he not keptACC, as him sick.ofNOM, plate-ACC/NOM

s čërnoj kaëmkoj. with black border

‘He did not keep, since he was sick of, the plate with the black border.’

cf. On ne ostavil, tak kak emu nadoela, bljudc-e s krasnoj kaëmkoj. ‘He did not keep, since he was sick of, the saucer with the red border.’

10

ACC SG tarelku ‘plate’ ≠ NOM SG tarelka ACC SG bljudce ‘saucer’ = NOM SG bljudce

‘Metasyncretism’ 11

I ‘window’

II ‘chair’

III ‘bone’

sing plur sing plur sing plur

nom prozor prozori stolica stolice kost kosti acc prozor prozore stolicu stolice kost kosti gen prozora prozōrā stolicē stolīcā kosti kostī /

kostijū dat/loc

prozoru prozor-ima stolici stolic-ama kosti kost-ima

inst prozorom prozor-ima stolicom stolic-ama kosti / košču

kost-ima

voc prozore prozori stolico stolic kosti kosti

Croatian nouns

  Singular Plural Singular Plural Nomina.ve amīcus amīcī cursus cursūs

Geni.ve amīcī amīcōrum cursūs cursuum

Da.ve amīcō amīcīs cursuī cursibus (cursū)

Accusa.ve amīcum amīcōs cursum cursūs

Abla.ve amīcō amīcīs cursū cursibus

Voca.ve amīce amīcī -- --

‘Metasyncretism’

Latin nouns AMĪCUS 'friend' (class II) and CURSUS 'course' (class IV)

Syncretism + suppletion 13

singular dual plural nominative človek   človeka   ljudje  accusative človeka   človeka   ljudi  

genitive človeka   ljudi   ljudi  dative človeku   človekoma   ljudem  

locative človeku   ljudeh   ljudeh  instrumental človekom   človekoma   ljudmi  

Slovenian noun ČLOVEK ‘person’

(Systematic) syncretism – some criteria

¡ More likely within a conditioning morphosyntactic value (e.g. within plural, as in German) than across values ¡  Includes but not limited to natural classes

¡ Historical patterns of analogy ¡ Systematic at least at the time of the change

¡ Resolution of syntactic conflicts (‘case matching’)

¡ Repetition across the morphological system – multiple inflection classes/lexical categories/etc.

¡ Shared morphological idiosyncrasy (e.g. suppletion)

14

Russian Adjectives 15

MASC INAN NEUT INAN FEM INAN

NOM SG nov-yj nov-oe nov-aja

ACC SG nov-yj nov-oe nov-uju

GEN SG nov-ogo nov-ogo nov-oj

DAT SG nov-omu nov-omu nov-oj

LOC SG nov-om nov-om nov-oj

INST SG nov-ym nov-ym nov-oj

NOM PL nov-ye

ACC PL nov-ye

GEN PL nov-yx

DAT PL nov-ym

LOC PL nov-yx

INST PL nov-ymi

¡ Some syncretisms involve several paradigm cells.

¡ How should we formally account for the one in bold?

Russian adjectives (illustrated with NOVYJ ‘new’)

Underspecification

¡ Based on the given data, we can employ a classic kind of underspecification, within a realizational theory

<NOVYJ, {CASE: GEN, NUM: SG, GEN: F}>

<nov-, {CASE: GEN, NUM: SG, GEN: F}>

Rules:

{CASE: NOM, NUM: SG, GEN: F} à -aja

{CASE: ACC, NUM: SG, GEN: F} à -uju

{NUM: SG, GEN: F} à -oj

16

¡  In inanimate nouns, all classes have acc = nom syncretism in the plural, and 3 of 4 in the singular but the inflectional form itself differs

I zakáz

'order'

IV bljúdo 'dish'

II kómnata 'room'

III tetrád'

'exercise book'

sing

ular

nom zakáz bljúdo kómnata tetrád' acc zakáz bljúdo kómnatu tetrád' gen zakáza bljúda kómnaty tetrádi dat zakázu bljúdu kómnate tetrádi loc zakáze bljúde kómnate tetrádi inst zakázom bljúdom kómnatoj tetrád'ju

plur

al

nom zakázy bljúda kómnaty tetrádi acc zakázy bljúda kómnaty tetrádi gen zakázov bljúd kómnat tetrádej dat zakázam bljúdam kómnatam tetrádjam loc zakázax bljúdax kómnatax tetrádjax inst zakázami bljúdami kómnatami tetrádjami

Syncretism in Russian nouns 17

Syncretism in Russian nouns

I otec

‘father’

IV čudovišče‘monster’

II ženščina ‘woman’

III mat’

‘mother’

sing

ular

nom otec čudovišče ženščina mat' acc otca čudovišče ženščinu mat' gen otca čudovišča ženščiny materi dat otcu čudovišču ženščine materi loc otce čudovišče ženščine materi inst otcom čudoviščem ženščinoj mater'ju

plur

al

nom otcy čudovišča ženščiny materi acc otcov čudovišč ženščin materej gen otcov čudovišč ženščin materej dat otcam čudoviščam ženščinam materjam loc otcax čudoviščax ženščinax materjax inst otcami čudoviščami ženščinami materjami

¡  In animate nouns, all classes have acc = gen syncretism in the plural, but only 1A has this pattern in the singular.

! !

18

Thought Exercise

Are the mechanisms that we’ve seen so far sufficient to account for Russian noun metasyncretism? Why or why not?

19

The Problem (as identified by Baerman 2004)

¡ Some realizational rules that we need (ignoring accusative for the moment)…

¡ For nominative plural 1.  {case: nom, num: pl, class: IV} à /-a/ 2.  {case: nom, num: pl} à /-i/ ¡  Note: orthographically <y> or <i>

¡ For genitive plural 1.  {case: gen, num: pl, class: I} à /-ov/ 2.  {case: gen, num: pl, class: III} à /-ej/ ¡ Notice that classes II and IV have “zero endings” – a nice

feature of realizational theories is that no rule at all is needed here! The ultimate default…

¡ Notice that these rules apply to both animate and inanimate nouns, so we don’t want our rules to include information about animacy!

20

The Problem (as identified by Baerman 2004)

¡ What about accusative plural? Two syncretisms, depends on animacy ¡ Animate accusative is like genitive

¡  Inanimate accusative is like nominative

¡ Can we handle this with underspecification + Paninian rule ordering? Not really…

¡ The crux of the problem: To capture syncretism across case, we underspecify for case. But we can’t capture both syncretisms at the same time!

21

The Problem (as identified by Baerman 2004)

¡ Possibility 1 (revising the nom rules): 1.  {case: gen, num: pl, class: I} à /-ov/ 2.  {case: gen, num: pl, class: III} à /-ej/

3.  {num: pl, class: IV} à /-a/ #Revised version ¡  Underspecified for case! Captures acc = nom syncretism

4.  {num: pl} à /-i/ #Revised version ¡  Underspecified for case! Captures acc = nom syncretism

¡ What is the problem here?

¡ Answer: We are missing the acc = gen syncretism pattern (all animates)!. So we also need: 5.  {case: acc, num: pl, anim: animate, class: I} à /-ov/ 6.  {case: acc, num: pl, anim: animate, class: III} à /-ej/

22

The Problem (as identified by Baerman 2004)

¡ What about the other way round?

¡ Possibility 2 (revising the gen rules): 1.  {case: nom, num: pl, class: IV} à /-a/ 2.  {case: nom, num: pl} à /-i/

3.  {num: pl, class: I} à /-ov/ #Revised version ¡  Underspecified for case! Captures acc = gen syncretism

4.  {num: pl, class: III} à /-ej/ #Revised version ¡  Underspecified for case! Captures acc = gen syncretism

¡ But now we are missing the acc = nom syncretism pattern (all inanimates)!. So we also need: 5.  {case: acc, num: pl, anim: inanimate, class: IV} à /-a/ 6.  {case: acc, num: pl, anim: inanimate} à /-i

23

The Problem (as identified by Baerman 2004)

¡ So, either way, we have to treat one of the patterns of accusative syncretism as accidental

¡ Despite the fact that according to all relevant ‘tests’ for systematic syncretism, both patterns are systematic ¡ Within number ¡ Resolution of syntactic conflicts (‘case matching’) ¡ Repetition across the morphological system –

multiple inflection classes/lexical categories/etc.

¡ Moreover, which pattern we treat as accidental is randomly chosen

24

The Problem (as identified by Baerman 2004)

¡ Towards a better solution…

¡ … known as rules of referral

25

Another Thought Exercise

I inanimate

I animate

IV inanimate

IV animate

sing

ular

nom Ø Ø -o -o acc Ø -a -o -o gen -a -a -a -a dat -u -u -u -u loc -e -e -e -e inst -om -om -om -om

plur

al

nom -y -y -a -a acc -y -ov / -ej -a Ø gen -ov / -ej -ov / -ej Ø Ø dat -am -am -am -am loc -ax -ax -ax -ax inst -ami -ami -ami -ami

¡ Descriptively, the Russian pattern is always described as acc looking like nom. Why not say that the nom looks like acc? (And why say that acc looks like gen, not that gen looks like acc?)

26

¡ Nominative and genitive always have the same form (within a declension class), regardless of animacy ¡ E.g. the Class I & IV form -a "belongs" to the genitive

¡ The accusative form thus seems to "borrow" the form of the nominative/genitive ¡ This is a directional syncretism

Answer 27

¡ Directional syncretisms are not rare. Another example, from Bonan (an Altaic language of Mongolia):

¡ For each syncretism, which case does the “borrowing” and which does the “loaning”?

nouns 'foliage'

pronouns 'he'

nominative labčoŋ ndžaŋ genitive labčoŋ-ne ndžaŋ-ne

accusative labčoŋ-ne ndžaŋ-de dative-locative labčoŋ-de ndžaŋ-de

ablative labčoŋ-se ndžaŋ-se instrum.-comitative labčoŋ-Gale ndžaŋ-Gale

Directional Syncretism 28

Directional Syncretism 29

I II III IV V

Ergative -(ya)yalili -li ∅ -ndu -li Absolutive ∅ ∅ ∅ -ni -n ̪a Accusative ∅ -n ̪a -n ̪a -n ̪a -n ̪a Locative -n̪i -ŋu -ŋu -n̪aŋu -ŋu Allative -ya -ŋu -ŋu -n̪aŋu -ŋu Dative -ya -ṇi -ṇi -n̪aŋka -ṇi Ablative -ndu -ŋundu -ŋundu -ŋundu -ŋundu Diyari noun case exponents (Pama-Nyungan language of Australia)

Rules of referral ¡ Referral = A rule stated at the level of

morphosyntactic features, to the effect that two sets of morphosyntactic values have the same form ¡ A different kind of realizational rule (within inferential-

realizational frameworks)

¡ Informally: ¡ {CASE: ACC, NUM: PL, ANIM: ANIM} à {CASE: GEN, NUM: PL} ¡ “The form of the accusative plural animate is the

same as the form of the genitive plural, whatever that is.”

¡ {CASE: ACC, NUM: PL, ANIM: INAN} à {CASE: NOM, NUM: PL}

30

Rules of referral

¡ Formally (nearly full Paradigm Function Morphology formalism):

Where σ = {CASE: ACC, NUM: PL, ANIM: ANIM}, RR(<X, σ>) = <Y, σ>, where NaR(X, σ/{CASE: GEN}) = <Y, σ/{CASE: GEN}>

Where σ = {CASE: ACC, NUM: PL, ANIM: INAN}, RR(<X, σ>) = <Y, σ>, where NaR(X, σ/{CASE: NOM}) = <Y, σ/{CASE: NOM}>

31

Rules of referral

¡ Since the referral is written into realizational rules, referrals are ordered relative to other rules by Panini’s Principle (Elsewhere Ordering)

¡ In other words, they are formally just like other realizational rules, except that they require a ‘lookup’ to a different paradigm cell to get the morphophonological form ¡ A paradigmatic relation

32

Big Picture

¡ Syncretism is not a unitary phenomenon – it can range from a couple isolated forms to full neutralization throughout the inflectional system

¡ Correspondingly, different theoretical mechanisms are posited to account for syncretisms at different levels of generality

¡ There is ongoing debate about which mechanisms are best for handling the most stubborn cases of syncretism ¡ Much of the debate comes down to different

evaluation metrics

33

The End

¡ Everything below this is ‘bonus content’

34

Impoverishment vs. Referrals

¡ How should we formally account for directional syncretisms? What kind of mechanism should we add? ¡ Distributed Morphology: Impoverishment ¡ Inferential-Realizational Morphology: Referrals

35

¡ Can Impoverishment account for the Russian pattern of animacy-based syncretism?

I zakáz

'order'

IV bljúdo 'dish'

II kómnata 'room'

III tetrád'

'exercise book'

sing

ular

nom zakáz bljúdo kómnata tetrád' acc zakáz bljúdo kómnatu tetrád' gen zakáza bljúda kómnaty tetrádi dat zakázu bljúdu kómnate tetrádi loc zakáze bljúde kómnate tetrádi inst zakázom bljúdom kómnatoj tetrád'ju

plur

al

nom zakázy bljúda kómnaty tetrádi acc zakázy bljúda kómnaty tetrádi gen zakázov bljúd kómnat tetrádej dat zakázam bljúdam kómnatam tetrádjam loc zakázax bljúdax kómnatax tetrádjax inst zakázami bljúdami kómnatami tetrádjami

Back to Russian… 36

Russian INANIMATE nouns

Syncretism in Russian nouns

I otec

‘father’

IV čudovišče‘monster’

II ženščina ‘woman’

III mat’

‘mother’

sing

ular

nom otec čudovišče ženščina mat' acc otca čudovišče ženščinu mat' gen otca čudovišča ženščiny materi dat otcu čudovišču ženščine materi loc otce čudovišče ženščine materi inst otcom čudoviščem ženščinoj mater'ju

plur

al

nom otcy čudovišča ženščiny materi acc otcov čudovišč ženščin materej gen otcov čudovišč ženščin materej dat otcam čudoviščam ženščinam materjam loc otcax čudoviščax ženščinax materjax inst otcami čudoviščami ženščinami materjami

¡ Can Impoverishment account for the Russian pattern of animacy-based syncretism?

! !

37

Russian ANIMATE nouns

Featural Decomposition 38

Marginal (Peripheral)

Directional Quantificational

Nominative - - -

Accusative - + -

Genitive - + +

Dative + + -

Locative + + +

Instrumental + - -

Genitive 2 - - +

Locative 2 + - +

Jakobson’s case features for Russian nouns and adjectives (second version, 1958)

¡ To be concrete, we can take Jakobson’s famous analysis of case subfeatures ¡ + means marked; - means unmarked

Conclusions from Russian

¡ The bidirectional nature of the syncretism makes it difficult to adequately account for Russian animacy-based syncretism in terms of Impoverishment

39

Evaluation – a one-sided perspective

"... Noyer's proposal entails that in a directional syncretism, the determinant member's morphosyntactic property set should always be less marked than that of the dependent member (prior to impoverishment)... [which] contrasts starkly with the rule-of-referral approach, which makes no predictions about the relative markedness of a directional syncretism's dependent and determinant members... But the issue here is obviously an empirical one, namely: can one maintain Noyer's conjecture that universally, a directional syncretism's determinant member is less marked than its dependent member? The answer, clearly, is no. First, the very existence of bidirectional referrals is incompatible with Noyer's conjecture. Moreover, there are unidirectional syncretisms in which the dependent member is less marked..." (Stump 2001: 238).

40

¡ What generalization(s) can be made about the forms that are boxed? How can you describe this formally?

IA zakáz

'order'

IB bljúdo 'dish'

II kómnata 'room'

III tetrád'

'exercise book'

sing

ular

nom zakáz bljúdo kómnata tetrád' acc zakáz bljúdo kómnatu tetrád' gen zakáza bljúda kómnaty tetrádi dat zakázu bljúdu kómnate tetrádi loc zakáze bljúde kómnate tetrádi inst zakázom bljúdom kómnatoj tetrád'ju

plur

al

nom zakázy bljúda kómnaty tetrádi acc zakázy bljúda kómnaty tetrádi gen zakázov bljúd kómnat tetrádej dat zakázam bljúdam kómnatam tetrádjam loc zakázax bljúdax kómnatax tetrádjax inst zakázami bljúdami kómnatami tetrádjami

Exercise 41

Russian nouns

Singular ones

1.  {case, nom, num: sg, class: IV} à /-o/ ¡ Note: orthographically <e> or <o>

2.  {case: nom, num: sg, class: II} à /-a/ ¡ The homophony with /-a/ in (1) is probably an

accident, and is treated as such here

3.  {case: acc, num: sg, class: II} à /-u/ ¡ The only uniquely accusative ending in Russian nouns!

4.  {case: gen, num: sg, class: I or IV} à /-a/ ¡ There is a better way to do this – avoiding ‘I or IV’, but it

is more advanced…

5.  {case: gen, num: sg} à /-i/

42

top related