journ=?iso-8859-1?Q?=E9e_?=Labex EFL 7 lundi octobre 2013

[Anne Abeillé]

Demi-Journée d’étude du LabEx EFL

Grammaire expérimentale

Autour de Joan Bresnan, professeur invitée du LabEx EFL

 

Lundi 7 octobre 2013

 

Université Paris Diderot, Batiment Sophie Germain

Salle 2011, Rue Albert Einstein, 75013 Paris

 

14h-14h30

Mathieu Avanzi, Elisabeth Delais-Roussarie & Philippe Martin (CNRS, Paris Diderot)

Contraintes et variation dans les découpages prosodiques

 

14h30-15h

Jonathan Ginzburg, Sara Moradlou (Paris Diderot) Evolving One Word Grammars

 

15h-15h30

Thierry Nazzi (CNRS Paris Descartes), Nayeli Gonzalez-Gomez, Lisa Hsin, Jennifer Culbertson, Isabelle Barriere & Geraldine Legendre (J Hopkins)

Studying early comprehension of subject-verb agreement crosslinguistically: new evidence from Spanish

 

15h30-16h Pause café

 

16h-18h Séminaire de Joan Bresnan (Stanford) : Probabilistic knowledge of grammar : empirical studies

 

Abstracts pour la journée du 7 octobre

 

M. Avanzi, E. Delais-Roussarie, P. Martin.

De nombreux travaux consacrés à la structuration prosodique ont montré que les découpages prosodiques sont construits à partir deux classes de contraintes :

(i)  des contraintes qui rendent compte des relations entre la structure prosodique et les structure syntaxique et informationnelle ;

(ii) des contraintes veillant à la bonne formation rythmique des énoncés (cf., entre autres, Martin (1986, 1987 & 2009) ; Delais-Roussarie (1996 1 2000) ; Selkirk (1995)). A partir d’exemples extraits de divers corpus et portant essentiellement sur les découpages prosodiques observés dans les constructions disloquées et pour les SN sujets,  nous nous fixons dans cette communication un double objectif :

- montrer que les contraintes souvent décrites pour rendre compte des découpages n’en donnent qu’une image partielle, dans la mesure où il existe une importante variabilité ;

- montrer que les mismatchs entre contraintes rythmiques et contraintes linguistiques (c’est-à-dire liées à l’appariement entre la structure prosodique et les structures syntaxique et informationnelle) ont un coût important dans le traitement cognitif.

 

J. Ginzburg, S. Moradlou

We show how to classify productions produced by children from the Providence Corpus of CHILDES in the earliest stages of child/adult interaction using an extension of a semantically-based taxonomy used to analyze adult non-sentential utterances (Fernandez and Ginzburg, 2002). We formalize the type system for classifying these productions using a combination of KoS (Ginzburg, 2012) and Type Theory with Records (Cooper 2012). We consider the implications of this work for existing work in computational learning theory (Clark and Lappin 2011).

 

T. Nazzi et al.

French-speaking children show comprehension of subject-verb (SV) agreement as early as 2;6 in preferential looking/pointing tasks using video stimuli (Legendre et al., 2010; Barrière et al., 2011). By contrast, using identical stimuli and methods, Spanish- and English-speaking children do not show robust evidence of comprehension even at 3;0 (Legendre et al., to appear). Studies using picture selection have revealed comprehension only at 5-6yrs for English (Johnson et al. 2005), 4;8 for Spanish (Pérez-Leroux, 2005). While late comprehension in English and Spanish is surprising (e.g. relative to production; Johnson et al. 2005), early success in French suggests it is not universal properties of SV agreement, but rather its language-particular instantiation that may be troublesome for learners. For example, the cue reliability of a particular morpheme, the syntactic complexity of the constructions it appears in, and its perceptual salience likely affect comprehension of SV agreement generally (Sundara et al., 2001; Legendre et al., to appear). We evaluate whether early comprehension in (Mexican) Spanish can be revealed when task demands are lowered. In particular, we manipulate use of nonce forms (in non-critical portions of stimulus items), and positional saliency of the morpheme, showing that these factors impact success.

Three experiments testing comprehension of SV agreement were run in Mexico City using a touch-screen pointing task. Experiment 1 replicated results from Legendre et al. (to appear), in which auditory stimuli consisted of a transitive verb+nonce object (e.g. agarra el micho ‘he throws the micho’ vs. agarran el duco ‘they throw the duco’). Nonce objects in such stimuli were suggested by Kouider et al. (2006) to encourage children to treat simultaneously-visible stimulus scenes as distinct. Forty children participated (M=50.0mos, range 38.4-64.7), and performed at chance level, not distinguishing singular from plural stimuli (M=54.4% accuracy, t(39)=1.26, p=0.21) with no effect of age (median-split analysis: M=55.3%, p=0.20 and 53.6%, p=0.28 respectively). In Experiment 2 transitive verbs were again tested, but the word objeto ‘object’ replaced all nonce nouns. Forty age-matched children (M=51.6mos, range=41.03-61.5) succeeded with these minimally altered stimuli (M=61.6%, t(39)=3.92, p<0.001), and this was found in both younger (M=62.5%, p=0.01) and older (M=60.6%, p=0.02) children. In Experiment 3 we tested intransitive verbs, thus there was no object at all, and the agreement marker appeared in utterance-final position. In addition to being syntactically simpler, utterance-final position was found by Sundara et al. (2001) to significantly impact the perceptual salience of the agreement morpheme –s for English-speaking children. Forty participants (M=50.8mos, range=37.9-61.1) again succeeded (M=58.8%, t(39)=3.34, p<0.005), and here the older children performed above chance level (M=62.1%, p<0.01), but not the younger children (M=54.4%, p=0.17).

Spanish-speaking children as early as 3;4 show SV agreement comprehension when task demands are lowered. In particular, when children were not required to process nonce lexical items (even when these are not critical to task), and when syntactic complexity was decreased resulting in increased perceptual saliency of the marker. These findings underscore the importance of task/stimulus-specific features when testing early morphosyntactic development, and suggest that previous results may have under-estimated Spanish-speaking children’s competence.

 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/parislinguists/attachments/20130930/7f879562/attachment.htm>