De la phonologie aux formes lexicales: liaison et cognition en français contemporain – PHONLEX

It is well known that the concept of word, which lies at the heart of most classical morphological and syntactic descriptions, cannot be identified in any direct way from acoustic signals. The problem is particularly acute in French where various phenomena such as liaison, enchaînement (forward syllabic linking) and phrasal stress, have led many phonologists and phoneticians to question the relevance of the word for a proper characterisation of the sound structure of this language. A number of current psycholinguistic models are also critical of the notion of word and treat it as an emerging unit, inferrable from the stocking and processing of larger constructions (cf. example-based models defended by some members of this project). Moreover, various specialists stress that the central role accorded to the word by a long tradition, up to even the most recent formal models, is largely based on spelling systems and the mental representations that language-users have of the link between sound structure and orthographic systems. The main objective of this project, on the basis of a thorough analysis of French liaison, is to reexamine the relationship between phonological and lexical structure, while paying due attention to the orthographic dimension. - - From the standpoint of linguistic description and theorization, too many studies have neglected the multifactorial nature of liaison. First of all, the classical trichotomy (obligatory, optional, forbidden) has not been tested against large date bases making room for diachronic, geographical, stylistic and social variation. Indeed, taking variation into account entails a reconsideration of the role played by orthographical systems. One of the major objectives of our project is (i) to test the correlation between phonology and spelling through (neuro)psycholinguistic experimentation and (ii) to devise new hypotheses concerning the cognitive representation of spoken and written systems. A thorough examination of developmental patterns will also prove crucial as it will allow us to compare the performance of pre-literate children with literate adults. In particular, we will relate the frequency of liaisons observable within the input and the actual usage of children. In this area, the qualitative and quantitative results derivable from the PFC project (Phonologie du Français Contemporain : usages, variétés et structure) will afford us a better understanding of the nature of the (average) input facing a child in the process of acquiring liaison. - - As far as description and theorization are concerned, a truly multifactorial study of liaison will first of all require a fine-grained phonetic analysis of lexical syllables and syllables resulting from liaison. At the morphological level, the relation between inflection, derivation and liaison will be reexamined and related to corpus frequencies and to word-templates. The analysis of transcribed corpora will allow us to establish a quantiative index of the relative autonomy of the lexico-graphical word. From a syntactic point of view, the formalization of phrasal structure will be correlated with statistical information. A computational tool will be devised which will yield a robust morphosyntactic labeling of transcriptions of spontaneous speech and predict liaison sites while integrating variation parameters. Finally, from a semantic, pragmatic and discursive standpoint, a typology of prosodic patterns will be established. More generally, the phenomenon of liaison in French will be compared with similar sandhi phenomena in languages where they are attested. - - The originality of the project lies in the exceptional collaboration between specialists of neighbouring areas such as phonology, morphosyntax, prosody, experimental phonetics, (neuro)psycholinguistics, acquisition, natural language processing and corpus linguistics. Our aim is therefore to offer an integrated treatment of a crucial morphophonological phenomenon of