Développement du langage et cognition


Research in our team focuses on syntactic representations and processes. We are interested in identifying general principles that guide syntactic production and comprehension, in both adults and children. One key aspect of the research conducted is the cross-linguistic approach by which the performance of babies or adults from languages differing on critical syntactic properties is compared. Such comparisons allow determining universal characteristics of syntactic representations and processes as well as specificities lying in cross-linguistic variation. The approach is that of cognitive science, combining constructs from linguistic theory hypothesized to describe the syntax of natural languages, and experimental methods developed in psycholinguistics to explore syntactic performance by way of measures of timing (in paradigms of speech initiation, self-paced reading, on-line grammaticality judgements, preferential looking) and errors (in spontaneous production or in paradigms of sentence elicitation).

The specific research topics covered in our team are:

Grammatical agreement

Agreement can be considered as one of the best indexes of the syntax-internal dynamics, hence providing a privileged window on the structural specificities of the human language. The psycholinguistic studies conducted in our lab aim at identifying the structural conditions under which agreement errs. The major phenomenon of interest is interference, also called attraction, by which agreement features from an element which is not the agreement controller, are incorrectly transmitted to the agreement target (e.g., *The program on the computers ARE bugged). Experimental studies are conducted in language production (by way of elicitation tasks) and language comprehension (by way of self-paced reading techniques or various on-line grammaticality judgment tasks) in order to determine how structural variations in the sentence modulate the occurrence of interference errors. Experimental data are then accounted for in terms of constructs of syntactic theory (like the c-command/precedence distinction, movement and traces, AGREE). The finding that these constructs, elaborated on independent grounds from linguistic research, adequately capture performance effects brings support to the assumption of a tight connection between the language processor and the grammar.


Hierarchical structure in artificial grammar learning

Language acquisition is at first glance one of the most difficult learning tasks which infants have to face; yet children learn to speak without any apparent effort. How do they reach this competence? Language is a complex system and it is difficult to disentangle the role of multiple factors involved in its acquisition. For this reason, researchers have looked more and more to artificial grammar learning paradigms (see Photos, 2007 and Folia et al., 2010 for a review). In this context participants are exposed to structured material following certain regularities. These regularities mimic aspects of natural language structures but are simplified relative to the realistic language learning conditions. Our project exploits this principle and aims to investigate which kind of information facilitates regularities’ extraction. In particular, we study the capacity to extract regularities based on one of the most salient principles of grammar (Chomsky, 1957), namely the hierarchical organization. Multiple factors which may contribute to language acquisition are examined; in particular, the role of prosody and the role of semantics are taken into account. This project aims to contribute to the debate about adults’ capacity to extract complex rules from the input (i.e., bi-conditional rule based on constituents’ structure), and the mechanisms involved in this task.


The acquisition of word order and subordination

Children’s first utterances already respect the basic word order of their language. However, the initial nature of the underlying syntactic representations of word order, i.e. head-complement order, remains unclear. Whereas data from preferential looking paradigms (e.g. Hirsh-Pasek & Golinkoff, 1996 ; Gervain et al, 2006) suggest abstract word order knowledge during the second year of life (grammatical hypothesis), Weird Word Order studies (e.g. Abbot-Smith et al, 2001; Matthews et al, 2005; Matthews et al, 2007) postulate evidence for verb-specific, lexically encoded word order until age 4 (constructivist hypothesis). In order to decide between those contradictory hypotheses, we conduct language comprehension experiments using pseudo-verbs which combine both experimental paradigms. Our aim is to examine the word order format in 19-month-old French and Japanese children, two languages which show an opposite canonical word order (NVN and NNV respectively). This approach is interesting from a cross-linguistic point of view, comparing a language rich in case marking morphemes (Japanese) with another one without case marking (French), and will give us insight about how young infants interpret grammatical and ungrammatical sentences in their language. First findings indicate that as early as 19 months, infants have an abstract representation of word order of the target language, supporting the grammatical hypothesis.


The role of executive control in the acquisition of syntax

Children acquire quite sophisticated grammatical knowledge by the age of 5, but children at this age still have great difficulties in comprehending temporarily ambiguous sentences (garden-path sentences) like "Put the frog on the napkin in the box." Here, children seem to be "stuck" on the initial, incremental analysis (i.e., napkin = destination for putting the frog), and fail to reach the correct, adult-like interpretation (i.e., box = destination) (Trueswell et al 1999). This surprising difficulty in sentence reanalysis could be taken to indicate that children's immature sentence processing mechanisms undergo a qualitative change during the course of development, or that understanding ambiguous sentences requires domain-general executive control abilities, whose development continues rather slowly until early adolescence. We're testing the latter hypothesis in two ways: First, we examine to what extent children's sentence revision performance correlates with their performance in tasks that measure their executive control abilities. Evidence for a correlation between the language task and executive function task would support our hypothesis. Second, our project compares sentence revision performance in monolingual and bilingual children. Bilingual children have been shown to demonstrate more mature executive function abilities (Bialystok, 1999), presumably due to their daily experience in inhibiting their linguistic systems, and for this reason we expect bilinguals to outperform the monolinguals in comprehending garden-path sentences. This project not only sheds light on how children's sentence comprehension behaviors develop, but also has implications for psycholinguistic models of sentence reanalysis.


The development of syntax in autistic children

Research has accumulated evidence for a deficit in language use in autistic patients. However, very little is known about their syntactic competence. Moreover, the few existing studies were all conducted on English speaking populations. The first aim of our project is to provide a fine screening of French speaking autistic children’s syntactic abilities. Preliminary work by our collaborator, Stéphanie Durrleman from the department of Linguistics suggests that both children and adults have significant difficulties with various syntactic structures, in particular those involving movement (relative clauses). The second aim of our research is to investigate the relation, in autistic children as well as typically developing control children, between syntactic processing and two cognitive functions: Theory of Mind (ToM) and executive control. Research reported in the literature suggests that autistic children have difficulties in processing complement clauses (e.g., John told Mary that the car was stolen). Interestingly, these difficulties were found to correlate with various ToM tasks, verbal and non verbal. Some authors have argued that the syntax of complement clauses may provide a representational basis for ToM in allowing to represent two levels of reality (the truth conditions of the main clause are independent of the truth conditions of the embedded clause). However, those studies showing a tight link between complement clause syntax and ToM rely on multidimensional tasks involving not only syntactic processing but also memory, executive control and even ToM itself. We are currently working on a better way of investigating the issue by testing more finely children’s syntactic processing of complement clauses. Some recent accounts of autism have proposed that a core deficit in executive control characterizes this population. If this hypothesis correct, then autism provides a privileged way to explore the relationship between syntax and executive control, described in the previous project.