Ongoing Projects   

Neurobiology of speech functions

Our research on speech functions, funded by NSERC, focuses on understanding the nature of the neural mechanisms that allow us to produce, perceive and comprehend speech sounds (phonemes), syllables and words, as well as to produce the voice (phonation). Using functional MRI and TMS, we examine the mechanisms involved in recognizing and discriminating speech sounds. Another avenue of research focuses on understanding the neurobiological factors such as brain structure and functioning that underlie the wide individual differences that exist in speech and language skills among adults. We are also interested in understanding the interplay between perceptual and motor mechanisms for speech.

An enduring challenge in speech research in trying to understand how the brain processes speech comes from the lack of a simple relationship between the acoustical properties of the speech sounds and their underlying phonological and motor representations. That is, very different acoustic features lead to the perception of the same speech sounds. One hypothesis is that this process is accomplished by extracting articulatory (motor) information from the incoming speech signal and using that information to categorize speech sounds (cf. motor theories of speech perception). Briefly, the idea is that the perceiver uses his or her repertoire of speech motor programs to identify the speech sounds in an incoming auditory signal. Our work focuses on understanding the nature of the contribution of motor and premotor areas in speech perception and production using functional and anatomical brain imaging and TMS.

Since much of our work focuses on understanding sublexical speech mechanisms, mainly syllable-level mechanisms, we need as much information as possible on the syllables that are used in the language that we study, Québec spoken French. As part of a research project funded by the Social Sciences and Humanities Research Council (SSHRC), we have developed a set of computer tools to study the natural statistical dependencies that exist between the syllables of spoken French in Québec (SyllabO). We created a large corpus of over 300K spoken syllables (e.g. transcribed conversations, news bulletins). Such tool is essential to study issues such as whether the syllable ‘sa’ is highly predictive of the syllable ‘lut’, which will allow us to study how the human mind process natural language statistics. Knowledge of the statistical properties of syllables is also useful to create stimuli controlled for their frequency of use. This will allow us to examine, for example, how accurately people produce frequent vs. less frequent syllables. Knowledge of spoken syllable frequencies is also important for speech pathologists in determining rehabilitation targets and evaluating the impact of articulatory of specific deficits.

While the mechanistic aspects of speech production (e.g. muscle tone, movement velocity, force and range) as well as language-specific (linguistic) mechanisms (e.g. syntax and semantic mechanisms), have been investigated in some detail, other aspects of verbal communication remain largely unexplored. One aspect of the neural organization of speech that is particularly unclear is the degree to which the speech system is integrated into a common action control system, an idea that is at odd with the dominant idea that the neural machinery for speech production is dedicated to the specific task of producing language. General action control mechanisms, sometimes collectively referred to as “motor cognition” include the intention to act, the formulation of a motor goal, the selection of the appropriate motor programs, their temporal and sequential organization and initiation. Our objective is to understand the manner and extent to which the neural speech system in integrated into a common action control system. This knowledge has tremendous importance for the development of more comprehensive neurobiological models of speech and language, which typically do not integrate common action control mechanisms, focusing instead on language-specific mechanisms. A better understanding of the role of general action control mechanisms in speech production may lead to the development of new speech rehabilitation strategies. Indeed, if the neural systems implementing one function (e.g., grasping, swallowing) are shared to some degree with another function (e.g., language and speech), treatments targeted to take advantage of this shared neural substrate may have a more ‘‘global’’ effect that crosses the boundaries of the specific targeted function, a phenomenon that is known as spread of effect. Hence, understanding the typical organization for sensorimotor and language functions is a necessary prerequisite for developing such targeted treatments, which may both be more effective and more economical than current treatments for a number of language disorders (e.g., aphasia, anomia, stuttering).