morphosyntactic data phylogenetic analysis of
TRANSCRIPT
Phylogenetic Analysis of Morphosyntactic Dataa case study of Negation in Tupí-Guaraní
N. Chousou-Polydouri, L. Michael, Z. O’Hagan, N. Gasparini, F. Rose
SLE, Leiden 2-5 September 20151
Introduction
○ linguistic phylogenetics mostly based on lexicon□ Gray & Atkinson 2003, Bowern & Atkinson 2012, Bouckaert et al.
2012...
○ with some exceptions based on morphosyntax□ abstract typological features (e.g. Dunn et al. 2005, Danielsen et al.
2011)□ cognate sets of morphemes (e.g. Nakhleh et al. 2005)□ parameters of UG (e.g. Longobardi & Guardiano 2009)
○ no previous use of cognate constructions
2
spoken in Amazonia and surrounding regions
Kokama
Omagua
Mawe
Tupinamba
Guaranian languages
Tupí-Guaraní
3
Tupí-Guaraní
○ well established subgroup of the Tupí family○ ~45 languages, many extinct or threatened○ previous classifications
□ 8 subgroups (Rodrigues and Cabral 2002)
□ phylogenetic classification based on lexical data (Michael et al. 2015)
□ agreement in most low-level subgroups, different higher structure
4
phylogenetic internal classification of Tupí-Guaraní (Michael et al. 2015)
5
Tupí-Guaraní
○ much comparative work in morphosyntax□ esp. Jensen 1998, Cabral 2000, 2001, 2007...
○ some reconstructions of morphemes and constructions□ Jensen 1987, 1990, 1997…, Schleicher 1998,
Cabral 2001, Cabral & Rodrigues 2005
○ little use of morphosyntax in classification□ Dietrich 1990, 2009, Schleicher 1998, Rodrigues
& Cabral 20026
TG Morphosyntactic Comparative Project
○ collaboration between Lyon and Berkeley○ morphosyntactic database (constructions)
□ phylogenetic classification □ comparison with classification based on lexicon□ subsequent work using the comparative method
7
TG Morphosyntactic Database
○ 27 TG languages + 2 nonTG Tupí languages
○ Domains:□ Negation, Person Marking, Valency Modifying
Operations, Directives…
○ Constructions organized by functions □ i.e. imperative 2pl, reciprocal object of
postpositions, indexation of 1inclA → 2plP…
○ Method illustration with negation8
Previous work on TG negation
○ Reconstructions of morphemes in PTG:□ *n-...-i standard negation (Jensen 1998, Schleicher 1998)
□ *-eʔɨm privative (N and dep. verbs) (Jensen 1998), negation of nouns and imperative (Schleicher 1998)
□ *ruã/*ruĩ adverbial negation (and more) (Jensen 1998)
□ *eme negative imperative (Jensen 1998)
□ *ani free negative response (Jensen 1998), free prohibitive (Dietrich 2003)
□ *-c(o)we after -i w/ some TAM (Schleicher 1998)
○ No explicit reconstruction of constructions9
Negation Dataset
○ 11 functions: standard negation, negative imperative, free prohibitive, denominal privative...
○ not just a list of morphemes, but whole constructions
10
Negation Dataset
11PR: person marker
Negation Dataset
12PR: person marker
Building cognate sets
○ Cognate = common ancestor□ similarity of form & meaning/function
○ Morpheme cognate sets□ 45 sets□ 20 sets with at least two languages
○ Construction cognate sets□ 23 sets with at least two languages and distinct
from morpheme cognate sets
13
*eʔɨm morpheme cognate set
14
*eʔɨm morpheme cognate set
15
*N-eʔɨm construction cognate set
16
*N-eʔɨm construction cognate set
17
Character Coding
1. Binary presence-absence coding for both morpheme and construction cognate sets
2. Partially ordered and partially polarized multistate coding for morpheme function.
18
state network for function of *eam
19
Free Negation “no”
Deverbal Privative
Standard Negation Constituent NegationDenominal Privative
Standard Negation
2
3
2
○ 43 binary characters total○ 5 multistate function characters
informative for subgrouping
too few characters for building a tree (more data coming...)
20
Character Coding
Ancestral State Reconstruction
○ quantitative technique○ traces evolution of a feature○ reconstructs state of feature at interior
nodes○ uses explicit model of evolution○ typically tree is fixed
□ TG tree based on lexical data (Michael et al. 2015)
21
Ancestral State Reconstruction
○ Cognate characters□ maximum likelihood reconstruction□ model: gain vs loss ~1:15 (Bayesian estimate)
○ Morpheme function characters□ parsimony reconstruction□ model: state network
22
23
Results: Morphemes
presence of *ani (morpheme)
probability of presence
Proto-Tupí-Guaraní
nonTG
presence of *ani (morpheme)
24
presence of *eam (morpheme)
25
presence of *eam (morpheme)
26
Results:PTG morphemes
*-eʔɨm privative *ruã/*ruĩ constituent negation *eme negative imperative*ani free negation *-c(o)we negation with certain TAM
reconstructed at a lower level (subgroup III plus Southern subgroup)
27
28
Results: Constructions
presence of *PRimp-V emePR: person marker
presence of *PRimp-V eme
29
presence of *t-PRind-V eme
30
presence of *t-PRind-V eme
31
*PR-V eme negative imperative*ta-PR-V eme negative directive*n-PR-N,V-i standard negation
reconstructed as a construction with two affixes, rather than a circumfix
*N-eʔɨm denominal privative
PR: person marker
Results:PTG constructions
32
function of *eam33
Results: Functions
Free Negation “no”
Standard Negation
Deverbal Privative
General negator
Conclusions
○ 6 PTG negators (ani, -eʔɨm, -i, nda-, -ruã, eme) out of 45 cognate sets
○ 5 PTG negative constructions○ many reconstructions: shallow group○ some examples of grammaticalization (e.g.
*eam)○ many cases of functional extension (e.g. -ã
in Siriono)
34
Conclusions
○ Phylogenetic methods & morphosyntax:□ morphology already treated just like lexicon□ syntactic data can also be used□ both add information refining the classification
○ ancestral state reconstruction techniques help□ determining where in the tree an element can be
reconstructed□ determining the proto-function of that element
35
Acknowledgements
○ Data harvesters □ K. Bartolomei, N. Chousou-Polydouri, E. Clem, W.
Daviet, N. Gasparini, P. Granado Columba, L. Michael, Z. O’Hagan, F. Rose
○ Additional data contributors□ E-M. Rössler, S. Meira
○ RefLex development and support□ S. Flavier
○ Funding: □ France-Berkeley Fund, Labex ASLAN
36
Sources
Aché field notes
Paraguayan Guaraní Guasch1996Kaiowá Cardoso 2008Mbyá Dooley 2006Tapiete Gonzalez 2005Chiriguano Dietrich 1986Guarayu Höller 1932Siriono field notesYuki Villafañe 2004Omagua field notesKokama field notes, Vallejos 2010Tupinambá Lemos Barbosa 1956Tapirapé Praça 2007Tocantins Asuriní Harrison 1975Parakana Silva1999
Avá Canoeiro Borges 2006Tembé Duarte 2007Anambé Juliao 2005Araweté Solano 2009Xingu Asuriní Pereira 2009Kayabí Dobson 88, 97Parintintin Pease 2007[1968]Kamaiurá Seki 2000Wayampí Copin 2012Emerillon Rose 2011Ka'ápor Lopes 2009Guajá Magalhaes 2007Awetí Reiter 2011Mawé Franceschini 1999, field notes
37
Higher structure comparison between Rodrigues & Cabral 2002 (on the left) and our TG classification (on the right)
(colors according to the 3 main branches of Rodrigues & Cabral 2002)38
function of *eme
Neg. Imp. & Neg. Dir. Neg. Imp. Neg. Dir. Neg. Int.MCN & FCN
Tupinamba (Lemos Barbosa 1956:91-92)e-î-pysyk umé2sg.imp-3.P-takeneg.imp‘Don’t take it.’t’ o-î-pysyk umédir 3A-3P-takeneg.dir‘He should not take it !’
39
ani PR-V (Negative Directive or Negative Imperative)
40
presence of *-c(o)we morpheme
Guarayu (Höller 1932:29)nd-a-mae-i-chi-raneg-1sg-look-neg-neg.fut-nfut‘I will not look’
41
*ani (free negation)
42