My research deals with the notion of « effort » in speech production and aims at understanding its relationship with different speech and voice disorders (dysphonia, stuttering, parkinsonian dysarthria, …). These troubles can be the consequence of :
an indadapted regulation of speech effort to the communicative situation, because of a distorted auto-evaluation of this effort (over-estimated, under-estimated, from auditory and proprioceptive sensory feedbacks), or because of a distorted perception of the situation and of its communicative constraints, or else because of a dysfunctionning of the regulatory mechanisms themselves (audio-phonation loop, sensori-motor recalibration). In that case, the effort disorder is quantitative (excessive vs. insufficient effort).
a non optimization of this effort, by implementing inefficient coordinations of gestures or by using inefficient communicative strategies, i.e. that require a lot of physiological effort for a poor (acoustic or communicative) result. In that case, the effort disorder is rather qualitative.
My goal is to provide speech pathologists with new knowledge that enable them to elaborate or to improve their methods of porevention, diagnosis and rehabilitation of these different disorders.
__________________________________________________________________________________________________
To that goal, I conduct research projects that aim at characterizing, quantifying and modeling this speech production effort. This requires some methodological work beforehand, in order to develop new measurement techniques (ex. new sensors) and to find relevant descriptors of speech efforts, not only from acoustical and physiological signals, but also from the perceived voice quality.
This also leads me to investigate more fundamental questions about the cognitive mechanisms that underlie the adaptation and regulation of speech efforts (Lombard effect, phonetic convergence, compensation of a perturbed auditory feedback): their principles, their constraints and motivations, the factors that influence them (situation, anatomy, culture, expertise, …), their neural correlates.
I also conduct several studies to understand how some gesture coordinations optimize the production efforts (balance between antagonist forces, source-filter interaction, …) and how some speech cues optimize the communicative efforts (audiovisual redundancy or complementarity).
The methodologies and theoretical knowledge brought by these different projects enable me to characterize the speech production and adaptation for different kinds of individuals, beeing experts of a vocal technique, or on the contrary presenting a speech or communication disorder.This allows me to better understand:
-what explains the efficiency of expert vocal techniques, and what part of these technique can be transfered to prevention or rehabilitation techniques
-what is wrong in the production of people with an effort-related speech disorder, and what can (or cannot) be rehabilitated with a speech therapist, or avoided with a prevention program.
Transversally to these different projects, I have faced several poblems with the statistical analysis of my data, which most of the time are repeated, paired and multiparametric data. Although these issues are very common and shared in the speech community, the methods of statistical analysis usually used are rarely adapted or approved by experts in statistics, for whom our data and questions represent a real research topic. Therefore a non negligible part of my time is spent collaborating with researchers in statistics, on different studies and associated datasets, in order to develop new methods of statistical analysis that are adequate to these studies and that can then be used in future studies with similar experimental designs.
Grenoble Images Parole Signal Automatique laboratoire
UMR 5216 CNRS - Grenoble INP - Université Joseph Fourier - Université Stendhal