EVOLUTIONARY CHANGE IN INDO-EUROPEAN MOTION EVENT ENCODING

Annemarie Verkerk
SCCR 22/25-02-2011

TALMYS MOTION TYPOLOGY

Spanish (verb-framed)!! ! ! ! La! ! botella ! entr !! a ! ! the! bottle ! moved.in ! to! ! ! ! ! ! ! ! PATH !! ! ! ! ! ! la ! ! the ! ! ! ! ! ! ! cueva ! otando cave ! ! oating ! ! ! MANNER Talmy (1985: 69)

Dutch (satellite-framed) De ! es ! ! dreef ! ! de ! grot !in the ! bottle ! oated ! the ! cave ! into ! ! ! ! ! MANNER ! ! ! ! PATH The bottle oated into the cave

The Many Ways to Search for a Frog

Linguistic Typology and the Expression of Motion Events

Dan 1. Siobin

1. INTRODUCTION
J. Linguistics 46 in (2010), f Cambridge University Press 2009 The chapters this 331377. volume, along with the extensive list of frog-story doi:10.1017/S0022226709990272 First published online 30 November Appendix II, provide a rich database for the exploration of2009 particular

Kyoko Hirose Ohara, Seiko Fujii and Charles J. Fillmore

m to solve as you might have thought. Presented at the Fifth struction Grammar, Austin, Texas. ). Grammatical constructions and linguistic generalizaruction. Language 75,1-33. Not as hard a problem to solve as you might have thought.

s ofcognitive grammar: Volume I: Theoretical prerequisites. ss. analysis of Pawnee kinship usage. Language 32,158-194. nge expressions in Japanese and their cognitive and lin. Sweetser (Eds.), Spaces, worlds, and grammar (124-156). ress. grammar, and multilinguality in the Japanese FrameNet. al Conference on Language Resources and Evaluation. hio Ohori, Ryoko Suzuki, Hiroaki Saito & Shun Ishizaki project: An introduction. Proceedings of the 4th Internasources and Evaluation. rminology in Hebrew: A study in Lexical Semantics. Dis-

studies in questions of language use and acquisition. The studies reported in Part I reflect a range of 1 languages different types, it possible to focus on therevisited role of linguistic The of typology ofmaking motion expressions l typology in narrative construction. A recurrent concern in those studies is the Oone HN Bthe EA V E R S themes of Frog, where are is of dominant expression of motion, which J you? In one way or another, all of the studies confront Talmy's by now familiar Department of Linguistics, The University of Texas at Austin typology of verb-framed and satellite-framed languages (Talmy 1985, 1991, 2000b). Briefly, the typology isBconcerned E T H L Ewith V I Nthe means of expression of the path of movement. In verb-framed languages ("V-languages") path is expressed Department Linguistics, (,enter', 'exit', Stanford 'ascend', University etc.), whereas in satelliteby the main verb in a: clause of framed languages ("S-languages") path is expressed by an element associated SH IAO WE I THAM with the verb ('go in/out/up', etc.). This dichotomy has engendered a good deal of research and debate in the literature on motion-event descriptions over the past Department of East Asian Languages and Literatures, Wellesley College decade or SO.2 In this concluding chapter on typological perspectives I suggest factors "conspire" to15 produce a range of frog-story that several different sorts (Received 20 of March 2008 ; revised January 2009) varieties. These varieties result from combined influences of linguistic structure, on-line processing, cultural practices. Talmy's typology designed to This paper provides and a new perspective on the options availablewas to languages for characterize lexicalization patterns, and it(2000) has provided important insights into the encoding directed motion events. Talmy introduces an inuential two-way overall set of structures define adopt individual languages. However, the typology typology, proposing that that languages either verb- or satellite-framed encoding of motion events. This augmented by Slobin (2004b) and Zlatev & alone cannot account fortypology discourseis structures, because language use is determined Yangklang (2004) with a third class of equipollently-framed languages. propose It is striking howclassification much has been We learned by by more Revising than lexicalization patterns. Talmy's typological that the observed options can instead be attributed to : (i) the motion-independent application of the V-Ianguage/S-Ianguage contrast, and it still plays a part in the morphological, lexical, andhere. syntactic resources languages make available for encodof complex event constructions mix of factors considered But a fuller account of narrative organization will ing manner and path of motion, (ii) the role of the verb as the single clause-obligatory require attention to a range of morphosyntactic, psycho linguistic, and pragmatic

lexical category that can encode either manner or path, and (iii) extra-grammatical factors that yield preferences for certain options. Our approach accommodates William Croft, Johanna Barodal, Willem Hollmann, the growing recognition that most languages straddle more than one of the previously proposed Violeta typological categories : a language may show both verb- and satelliteSotirova, and Chiaki Taoka framed patterns, or if itNew allows equipollent-framing, even all three patterns. We further University of Mexico, USA, University of Bergen, Norway, show that even purported verb-framed languages may not only allow but actually University of Lancaster, UK, University of Nottingham, UK, prefer satellite-framed patterns when appropriate contextual support is available, a and Kobe College, Japan situation unexpected if a two- or three-way typology is assumed. Finally, we explain the appeal of previously proposed two- and three-way typologies : they capture the encoding options predicted to be preferred once certain external factors are recognized, including complexity of expression and biases in lexical inventories. 1. Introduction

PARALLEL CORPUS
Alices Adventures in Wonderland (Lewis Carroll) Through the Looking-Glass and what Alice found there O Alquimista (Paulo Coelho) 308 motion sentences

SWEDISH LATVIAN LITHUANIAN ENGLISH IRISH DUTCH POLISH GERMAN FRENCH PORTUGUESE ROMANIAN SERBO-CROATIAN ITALIAN ALBANIAN GREEK ARMENIAN PERSIAN NEPALI HINDI RUSSIAN

MOTION ENCODING STRATEGIES

! ! ! ! ! satellite-framed: ! ! ! ! ! ! ! verb-framed: !! path-only: ! deictic:! ! ! ! manner-only: !! coordination: ! ! ! ! ! ! Alice laughed so much at this, that she had to run back into the wood ! ! for fear of their hearing her; she had to enter the wood she had to run in the wood she had to go into the wood she had to run and go back into the wood she had to run to go back into the wood she was in the wood ! she had to enter the wood running / at a run / quickly

subordination: ! ! other: ! ! ! !

manner+path verb: ! she had to run+enter the wood

MOTION ENCODING IN IE
1.0 proportion per sentence 0.0 0.2 0.4 0.6 0.8

originals

russian

polish

lithuanian

swedish

german

dutch

latvian

english

irish

serbocroatian

greek

italian

portuguese romanian

french

hindi

persian

armenian

albanian

nepali

MOTION ENCODING IN IE
1.0

path-only verb-framed satellite-framed

originals russian polish lithuanian swedish german dutch latvian english irish serbocroatian greek italian portuguese romanian french hindi persian armenian albanian nepali

proportion per sentence

0.0

0.2

0.4

0.6

0.8

PHYLOGENETIC COMPARATIVE METHODS

PHYLOGENETIC COMPARATIVE METHODS

PRINCIPAL COMPONENT ANALYSIS

0.10 Polish Lithuanian 0.05 Russian Latvian German Serbo-Croatian Portuguese Italian Romanian Greek Albanian French Second principal component

65% of the variance is explained by the rst principal component Take the score of each language on the rst principal component

0.00

-0.05

English Dutch Swedish Nepali Armenian Persian

-0.15

-0.10

-0.20

Irish -0.3 -0.2 -0.1 0.0

Hindi 0.1 0.2 0.3

First principal component

TESTING HISTORICAL SIGNAL

RESEARCH LETTER
Indo-European
Hittite

Phylogenetic trees: - ! from Dunn et al. (2011) - ! build on lexical data (Swadesh lists) - ! estimated using a Bayesian Markov ! Chain Monte Carlo approach

Tocharian A Tocharian B Armenian Mod Albanian G AncientGreek Greek Mod Latin Rumanian List Sardinian C Catalan Portuguese ST Spanish Italian Ladin Walloon Provencal French Gothic Old Norse Swedish List Danish Riksmal Icelandic ST Faroese Old English English ST Luxembourgish Pennsylvania Dutch German ST Dutch List Afrikaans Flemish Frisian Irish B Scots Gaelic Welsh N Cornish Breton ST Lithuanian ST Latvian Old Church Slavonic Polish Russian Byelorussian Ukrainian Czech Slovak Lusatian L Lusatian U Slovenian Serbocroatian Bulgarian Macedonian Kurdish Persian List Tadzik Baluchi

W Kashmiri Nepali Bih Be O

DATA + TREES
French 0.15 0.6 Portuguese 0.12 0.5 1 Italian 0.10 Romanian 0.12 0.4 Irish -0.10 Dutch -0.17 0.4 1 1 German -0.15 English -0.05 1 Swedish -0.20 Polish -0.17 0.75 0.7 Russian -0.25 1 Serbo-Croatian 0.04 1 Lithuanian -0.16 1 Latvian -0.11 0.37 0.44 Nepali 0.23 1 Hindi 0.06 1 0.43 Persian 0.12 Armenian 0.13 1 Albanian 0.21 Modern Greek 0.06

HISTORICAL SIGNAL

TESTING HISTORICAL SIGNAL

French Portuguese Italian Romanian Irish Dutch German English Swedish Polish Russian Serbo-Croatian Lithuanian Latvian Nepali Hindi Persian Armenian Albanian Modern Greek French Portuguese Italian Romanian Irish Dutch German English Swedish Polish Russian Serbo-Croatian Lithuanian Latvian Nepali Hindi Persian Armenian Albanian Modern Greek

the likelihood of real trees is signicantly different from likelihood of trees with zero lambda (p < 0.01)
(Pagel 1999)

TESTING HISTORICAL SIGNAL

French Portuguese Italian Romanian Irish Dutch German English Swedish Polish Russian Serbo-Croatian Lithuanian Latvian Nepali Hindi Persian Armenian Albanian Modern Greek Russian German Portuguese Latvian Italian Swedish Irish English Armenian Polish Nepali French Romanian Lithuanian Modern Greek Hindi Dutch Persian Albanian Serbo-Croatian

the kappa score provided by this analysis shows that historical signal is present
Blomberg et al. (2003)

TESTING FOR HISTORICAL SIGNAL

DONE

French 0.15 0.6 Portuguese 0.12 0.5 1 Italian 0.10 Romanian 0.12 0.4 Irish -0.10 Dutch -0.17 0.4 1 1 German -0.15 English -0.05 1 Swedish -0.20 Polish -0.17 0.75 0.7 Russian -0.25 1 Serbo-Croatian 0.04 1 Lithuanian -0.16 1 Latvian -0.11 0.37 0.44 Nepali 0.23 1 Hindi 0.06 1 0.43 Persian 0.12 Armenian 0.13 1 Albanian 0.21 Modern Greek 0.06

 go

back in time

ANCESTRAL STATE ESTIMATION

ANCESTRAL STATE ESTIMATION

French Portuguese Italian Romanian Irish Dutch German English Swedish Polish Russian Serbo-Croatian Lithuanian Latvian Nepali Hindi Persian Armenian Albanian Modern Greek

PIE?

ANCESTRAL STATE ESTIMATION

French Portuguese Italian Romanian Irish Dutch German English Swedish Polish Russian Serbo-Croatian Lithuanian Latvian Nepali Hindi Persian Armenian Albanian Modern Greek

PIE = satellite framed? (Talmy 2007, Acedo Matelln and Mateu 2008)

ANCESTRAL STATE ESTIMATION

Maximum Likelihood transition rate: -0.5 0.5 ? ? ? ? ? ? PIE? ? ? ? ? ?
French Portuguese Italian Romanian Irish Dutch German English Swedish Polish Russian Serbo-Croatian Lithuanian Latvian Nepali Hindi Persian Armenian Albanian Modern Greek

? ? ?

Lithuanian Russian Greek Armenian

Irish Armenian Albanian Greek

Irish Armenian Greek Albanian

Lithuanian Russian Greek Armenian

n. 345

n. 346

n. 347

n. 348

German Dutch English Portuguese French Irish Lithuanian Russian Armenian Greek Albanian

Dutch German English Portuguese French Lithuanian Russian Irish Armenian Greek Albanian

Lithuanian Russian Irish Portuguese French German Dutch English Armenian Greek Albanian

German Dutch English Portugues French Irish Lithuan Russian Armenian Greek Albanian

n. 619

n. 620

n. 621

n. 622

Dutch German English French Portuguese Irish Lithuanian Russian Armenian Albanian Greek

German Dutch English Lithuanian Russian Armenian Portuguese French Irish Albanian Greek

-0.5 0.5 ? German ? Dutch ? English ? Lithuanian ? Russian ? Irish ? French ? Portuguese Armenian ? Greek n. 888

Albanian

Dutch German English French Portugue Irish Lithuanian Russian Armenian Alban Gree

n. 886

n. 887

n. 889

Portuguese French

German Dutch

German Dutch

Portuguese French

0.23

Nepali Albanian

ANCESTRAL STATE ESTIMATION

French Armenian Portuguese Romanian Persian Italian Modern Greek Hindi Serbo-Croatian

English

Irish Latvian German Lithuanian Dutch Polish Swedish

-0.25

Russian

0.23

Nepali Albanian

ANCESTRAL STATE ESTIMATION

Italian Modern Greek Hindi Serbo-Croatian

English

Irish Latvian German Lithuanian Dutch Polish Swedish

-0.25

Russian

Root estimate PIE: between -0.02 and 0.09

French Armenian Portuguese Romanian Persian

INCORPORATING INFORMATION FROM ANCIENT LANGUAGES

French Portuguese Italian Romanian Latin Irish Dutch German English Swedish Polish Russian Serbo-Croatian Lithuanian Latvian Sanskrit Nepali Hindi Persian Armenian Albanian Modern Greek

PIE?

THE EVOLUTION OF THE PIE PREVERB SYSTEM

PIE! ! ! ! ! ! ! Latin!! ! ! ! ! ! ! Romance

! ! ! ! ! ! Preverb ... Verb ! ... Preverb Verb! ! ! ! ! ! !

! Prep.+Noun ... Verb ! ! ! ... prex-Verb!! ! ! ! ... Verb

Watkins (1964), Vincent (1999), Iacobini & Masini (2006)

THE EVOLUTION OF THE PIE PREVERB SYSTEM

Latin se-que ! ! ! rursus ! in ! osti-um ! ! ! ! dom-us ! ! ! in-ced-ere 3SG.F.REFL.ACC-and ! back ! in ! entrance-N.ACC.SG ! house-F.GEN.SG ! in-go-PRS.INF and found herself walking in at the front-door again. in tenebr-as se ab-rip-uit quam cel-emme pot-uit into darkness-F.ACC.PL 3SG.REFL.ACC away-tear-PFV.3SG how fast-ADV be.able-PFV.3SG and skurried away into the darkness as hard as he could go.

THE EVOLUTION OF THE PIE PREVERB SYSTEM

PIE/Sanskrit! ! ! ! later Sanskrit! ! ! ! ! modern lang.

! ! ! ! ! ! Preverb ... Verb ! ... Preverb Verb! ! ! ! ! ! !

! Prep.+Noun ... Verb ! ! lost ! Noun+Post. ... Verb ! ! ... prex-Verb!! ! ! ! ... Verb
Watkins (1964), Bloch (1965)

THE EVOLUTION OF THE PIE PREVERB SYSTEM

Sanskrit t! ! ! p!t-ay-ati ! ! ! pak"#a$ away/out ! y-CAUS-3PL! ! bird.PL she makes the birds y away

Delbrck (1893: 648)

INCORPORATING INFORMATION FROM ANCIENT LANGUAGES

French Portuguese ? Italian Romanian Latin ? Dutch Irish German English Swedish Polish Russian Serbo-Croatian Lithuanian Latvian Sanskrit Nepali ? Hindi Persian ? Armenian Albanian Modern Greek

PIE?

? ?

CORRELATED EVOLUTION
skim glide skurry crawl rush swim roll hurry tumble soar oat jog march run saunter creep walk y drift slip

passear caminhar nadar precipitar se trotar correr esgueirar se rastejar pular voar
(Slobin 2004)

CORRELATED EVOLUTION
! language ! ! 1.! Russian! ! ! 2.! Swedish!! ! 3. ! Polish! ! ! 4. ! Lithuanian! ! 5.! Dutch! ! ! 6.! German! ! 7.! Latvian! ! ! 8.! Irish! ! ! 9.! English! ! ! 10.! Greek! ! ! 11.! Hindi! ! ! 12.! Italian! ! ! 13.! Persian !! ! 14.! Portuguese! ! 15.! Armenian! ! 16.! French! ! ! 17.! Albanian! ! encoding ! -0.23! ! -0.18! ! -0.15! ! -0.14! ! -0.14! ! -0.12! ! -0.09! ! -0.08! ! -0.03! ! 0.09! ! 0.08! ! 0.13! ! 0.14! ! 0.15! ! 0.15! ! 0.18! ! 0.24! ! manner verb class 33 23 26 26 21 30 26 14 30 16 16 16 15 20 15 13 11

CORRELATED EVOLUTION
size of manner verb class 10 15 20 25 30 35

Russian German Lithuanian Polish Swedish Dutch

English Latvian

Portuguese Italian Greek Armenian Hindi Persian Irish French Albanian

0.4 0.5 0.6 0.7 0.8

5 0.2

0.3

scale of encoding patterns

Lithuanian Russian Greek Armenian

Irish Armenian Albanian Greek

Irish Armenian Greek Albanian

Lithuanian Russian Greek Armenian

n. 345

n. 346

n. 347

n. 348

German Dutch English Portuguese French Irish Lithuanian Russian Armenian Greek Albanian

Dutch German English Portuguese French Lithuanian Russian Irish Armenian Greek Albanian

Lithuanian Russian Irish Portuguese French German Dutch English Armenian Greek Albanian

German Dutch English Portugues French Irish Lithuan Russian Armenian Greek Albanian

n. 619

n. 620

n. 621

n. 622

Phylogenetic Generalized Least Squares

Dutch German English French Portuguese Irish Lithuanian Russian Armenian Albanian Greek German Dutch English Lithuanian Russian Armenian Portuguese French Irish Albanian Greek German Dutch English Lithuanian Russian Irish French Portuguese Armenian Greek Albanian

Dutch German English French Portugue Irish Lithuanian Russian Armenian Alban Gree

n. 886

n. 887

n. 888

n. 889

Portuguese French

German Dutch

German Dutch

Portuguese French

CORRELATED EVOLUTION
p = 0.05
250 frequency 0 0.0 50 150

0.2

0.4

0.6

0.8

1.0

probability of slope

CONCLUSION
An approach to motion events that takes into account patterns of usage gives us a more ne-grained and productive perspective Patterns of motion encoding diversity are not random but historically patterned, and comparative analysis needs to take this into account In order to take into account this history we need ways to combine traditional historical linguistic methods with phylogenetic comparative methods

Thank you!

GEOGRAPHICAL DISTANCE

Mantel test (Spearman correlation): Mantel coefcient 0.095 Two-tailed p-value: 0.369

GEOGRAPHICAL DISTANCE
0.4 data.dists 0.0 1.5 0.1 0.2 0.3 0.5

2.0

2.5

3.0 log.lang.dists

3.5

4.0

4.5

CORRELATED EVOLUTION
Coefficients: ! Estimate ! Std. Error ! t value !Pr(>|t|) (Intercept) ! 2.73756 0.31165 ! 8.7841 ! 1.041e-05 *** log(encoding) ! -1.01505 ! 0.36695 ! -2.7662! 0.02189 * --Signif. codes: 0 "***# 0.001 "**# 0.01 "*# 0.05 ".# 0.1 " # 1 Residual standard error: 0.01612 on 9 degrees of freedom Multiple R-squared: 0.4595,!Adjusted R-squared: 0.3995 F-statistic: 7.652 on 2 and 9 DF, p-value: 0.01144