Motor speech disorders (MSD) refer to a broad set of altered speech dimensions (articulation, speech rate, voice, prosody) and of impaired speech processes, which alter dramatically patients’ communication skills in the course of several neurological diseases, including stroke and neurodegenerative diseases. The impaired processes in MSD are localized at the ‘latest’ stages of speech production processes, corresponding to the transformation of an encoded linguistic message into articulated speech. These processes include the planning of invariant speech goals and their adaptation to the required contextual variations, the programming of motor programs and control of detailed neuromuscular commands to execute speech. The distinction and delimitation between phonetic speech planning and speech motor programming is still debated both from the theoretical and the clinical semiological perspectives. In particular, whereas most models posit distinct processes and different brain circuits for the planning of speech goals and subsequent processes of motor programming and execution, corresponding to different speech motor disorders (apraxia of speech versus dysarthria), the surface manifestations of impaired speech planning versus impaired speech programming/execution often overlap and the clinical differential diagnosis is far from being clear-cut.
The present project brings together a multidisciplinary consortium with complementary expertise in speech and language pathology, psycholinguistics, neurology, phonetics, and speech engineering in order to (a) improve the characterization of phonetic speech planning and motor speech programming, (b) identify acoustic, speech behavior and electrophysiological brain markers of these processes in a typically (unimpaired) functioning system and in MSD, with the overarching goal to (c) better isolate and classify speech alterations in MSD and differentiate those who derive from impaired speech planning from those deriving from impaired speech programming or execution.
Neurophysiological, articulatory and acoustic data will be collected experimentally for both healthy speakers and speakers with MSD along with speech elicited in more natural conditions on over 100 French-speaking patients diagnosed with different types of MSD (apraxia of speech, hypokinetic dysarthria, flaccid/spastic dysarthria linked and ataxic dysarthria), as well as a mixed group with unclear diagnosis.
The originality of this project relies on the combination of psycholinguistic, phonetic and speech engineering and computational approaches to the study of speech in typical and impaired speakers, allowing the mapping of results based on experimental approaches and on refined acoustic analysis and automatic processing of a large amount of speech samples.
The research proposed has fundamental, technological and clinical relevance. It will inform theoretical models of speech production through the integration and possible convergence of speech behavior, acoustic markers and neurophysiological data from healthy and impaired speakers. From the technological side, robust automatic speech perception and production models will be developed based on new speech feature representations linked to speech planning and programming. From the clinical side, results will contribute to improve the definition and classification of MSD and will help develop semi-automatic assessment tools to be used in clinical practice.