Research

Sharon Peperkamp (MdC, Paris VIII)	Charlotte Jacquemot (CR, CNRS)

Andy Martin (Post-doctorant)	Marcela San Giacomo (Doctorante, Paris VIII)	Katrin Skoruppa (Doctorante, Paris VI)	Laura Robotham (Shapiro fellow)

We know that in adults, speech sounds are encoded in an abstract phonological representation that incorporates language-specific properties. We investigate this representation in three lines of research. First, we study the way in which it develops during first language acquisition. Second, we examine the processing of speech sounds in adults. Third, we explore deficits in the processing of speech sounds.

The acquisition of phonological representations

Are phonological representations acquired before or after the lexicon ? If they are acquired prelexically, what are the learning mechanisms that allow for this acquisition? We explore these question by means of both modelling studies and behavioural experiments with infants.

The processing of speech sounds in adults

What are the functional models that account for phonological processing observed in adults, and what are the limits of the processing of non-native sounds? What are the consequences of the limits for linguistic theory, in particular for phenomena of languages in contact? And what are the neural bases of phonological processing? The questions are examined using several techniques: adult psycholinguistics, theoretical linguistics and sociolinguistics, as well as brain imagery.

Deficits in the processing of speech sounds

Our studies on the deficits of the speech sound processing system are based on neuropsychological explorations of aphasic patients. Our aim is to account for the phonological deficits observed in these patients within a theoretical framework of speech processing.

Anne Christophe

 

	Perrine Brusini
Thomas Hannagan (doctorant, Paris VI)	Elodie Cauvet

Children who acquire their mother tongue have to lean its words (or lexicon, that is the associations between sound patterns and meaning), and its syntactic characteristics. The acquisition of each of these linguistic domains could be facilitated if infants had access to some knowledge of the other one. Thus, knowing about word meanings should help syntactic acquisition, but conversely, analysing the set of syntactic structures within which a new word occurs can help learners to guess its meaning. This leads to an apparent paradox, or ‘bootstrapping problem’ : the lexicon is necessary to acquire syntax, and syntax is necessary to acquire the lexicon.

In this theme, we pursue several research project that aim at elucidating how early lexical and syntactic acquisition gets bootstrapped. These projects focus in particular on the problems of segmenting the continuous speech stream into words (role of phrasal prosody, and computation of transition probabilities between neighbouring segments), of identifying and using function words to constrain the acquisition of content words (noun/verb categorization, singular/plural), and on the first stages of syntactic analysis (exploiting phonological phrase boundaries on-line to constrain syntactic analysis, exploiting function words to improve the efficiency of lexical access).

Franck Ramus

Baudouin Forgeot d'Arc (Doctorant, Paris VI)	Caroline Bogliotti (Post-doctorante)

Language acquisition relies on children's early cognitive capacities. These cognitive capacities have neural bases, which are in part genetically determined. Our group studies developmental disorders, as they highlight complex causal chains between genetic factors, brain properties and cognitive functions. We study primarily developmental language disorders (dyslexia and specific language impairment), at the cognitive level using psycholinguistic methods, at the neural level using brain imaging, and at the genetic level in collaboration with molecular biology laboratories. We are also interested in other developmental disorders that affect specific cognitive functions, like autism, and more generally in the mechanisms by which the genome constrains the architecture of human cognition.

Emmanuel Dupoux  (Prof, EHESS)	Sid Kouider   (CR, CNRS)	Jérôme Sackur (MdC, ENS)

Antoine Barbot (M2)	Yasuyo Minagawa- Kawai (Post-doctorante)	Sofie Gelskov (Doctorante)	Vincent Roger de Gardelle (Doctorant)
	Marine Buon (Doctorante)	Belonia Gabalda (M2)

 
Though language may be a specialized and relatively autonomous system, (a module, cf, Fodor, 1983), one can question its interconnection with other higher functions. We approach this question from two angles.

The first approach questions the interface of language and conceptual organization throughout development. During the first year of life, in parallel to acquisitions made in the domains of phonology and morphosyntax, an infant also undergoes great development in both the domain of physical object reasoning, and also in understanding of agency and intentionality (see the work of Susan Carey, Elisabeth Spelke, Renée Baillargeon, Gyorgy Gergely, etc). These developments exist independently, and use different properties of stimuli available in the environment. By the end of the first year of life, however, infants are able to make use of these elements to connect the lexical world with the conceptual world. The beginnings of lexical acquisition necessitates the conjunction of word forms that are extractable from speech, with conceptual elements abstracted from representations of the situation at hand. This key moment in development leads us to ask a series of questions on the nature of interactions between language and conceptual organization: does language modify conceptual organization, by allowing the creation of new concepts, or in contrast, might language represent exclusively pre-existing concepts?

The second approach focuses more generally on the role of executive function in the circulation of information across different components of linguistic and nonlinguistic cognition. A majority of models of cognitive architecture propose a static organization: the cognitive system is presented as a complex organization of specialized modules that are interconnected, however the manner in which organization is circulated is said to be fixed. However, this circulation of information is far from being fixed ; depending on the task, linguistic information, for example the identity of a word, will be used to find grammatical information, semantic information, or even spelling. It is thus important to define the parts of the cognitive system that are active automatically, and the parts that depend on a particular strategy and therefore access to information at the level of the central executive. We explore this this question using techniques of masking that permit the disoccation of automatic processing from strategic processing.

---

• 

Intranet