APPSy

The project "Asymétries Phonétique et Phonologique de la syllabe" (APPSy) seeks to investigate a universal in the structural organization of the syllable, namely the attraction of consonant segments before the vocal nucleus (attack) to the detriment of the position following the vocal nucleus (coda). The study will be conducted within the framework of the Frame /Content theory by MacNeilage (1998) according to which the organization of speech into sequences of consonant and vowel is the outcome of the jaw oscillation, linking directly the universal CV structure to the cycle of the jaw movement: the consonant is produced when the jaw is raised whereas the vowel is realized when the jaw is in its lowest position. However, this theory does not explain the overrepresentation of the CV syllable structure compared to the inverse structure VC which represents less than 5% of the world languages’ syllables and which fits as well into the jaw cycle. More generally, the theory does not explain one of the standards regularly used in descriptions of phonological processes, namely the empirical principle of maximization of attacks or MOP (Onset Maximization Principle), which occurs mainly in the syllabification of intervocalic consonants, assigning those consonants in priority into the attack position to the detriment of the coda one. In seeking to assess the impact of the jaw’s biomechanical properties on segments’ production, Redford (1999) from a study on American-English shows that there is an asymmetry between the two phases of the jaw cycle: the closing phase is stiffer, faster and with less amplitude than the opening phase. The APPSy project’s aim is to explore this articulatory asymmetry which could explain the MOP by studying, with both additional procedures EMA® and EMG, the links between the jaw movement and various syllable structures, from a selection of 7 languages which differ both in the complexity of their consonantal articulation and in the complexity of their syllable structures. To complement this first experimental part, the project aims at doing a large original study to quantify syllable structures’ universal tendencies from three sources: data from phonologyzed and syllabified dictionaries; dialectal data from linguistic atlas which will help to identify and analyze processes of complexity emergence in syllable attack; multilingual corpora of adults and children’s continuous speech, partly composed of available corpora but which need to be extended within the framework of this project. The different data will allow observing the frequency of syllable structures in lexicons and the reorganization of these structures in spontaneous speech in adults and children compared to phonetic changes. They will also help to verify the predictions made from experimental results in the first part of the project. The third part will study the applicability of mechanical properties of the jaw movement and their acoustic consequences in the detection of syllable boundaries. The purpose of this last component will be on one hand, to propose an innovative methodology based on the phonetic facts (part 1) and playing a juncture role in cutting syllables in order to integrate them into an automatic syllabification tool (which will be tested on the project’s corpora) and on the other hand, to consider the syllable as a modeling unit in continuous speech recognition. In this domain, unlike the phoneme, the syllable has received surprisingly little consideration as a basic modeling unit for the automatic recognition of speech. Thus, the project will be about assessing the importance of such modeling for the automatic processing of natural languages.