Our long-term goal is to understand how humans organize their brains and vocal tracts so that they can speak; only through understanding normal function can we see what happens with disorders. Although it is uncontroversial that most of the speech we hear is produced by a human vocal tract, it is less accepted that speech production and speech perception are intricately linked. Many theorists hold that the vocal tract's acoustic output is dealt with in a purely acoustic manner and that the link would be seen in modifications of the vocal tract shape to achieve particular acoustics. An alternative approach holds that speech consists of gestures (the coordinated activity of articulators), such as the jaw and the lips, achieving a phonetic goal, such as lip closure. The gestural model has allowed an insightful interpretation of many speech production phenomena, and the models have begun to have testable predictions for perceptual theories as well. The proposed experiments expand on this research, showing how perception of gestures is possible in automatic speech recognition, how the consequences of articulation?acoustic, visual and even haptic?are used by perceivers, and how accommodations are made for differences between speakers. This theoretical outlook has been fruitfully applied to problems in language acquisition, language change, and certain language disabilities. The advances from the proposed research should allow even broader applications. The goal is to show how acoustic parameters that cohere because of their origin in articulation are used by listeners. This will be accomplished by acoustical modeling of natural productions, perception of natural speech under modified circumstances (e.g., impaired by noise or enhanced by feeling the articulators saying what is being heard), and measurement of speech with ultrasound and optical markers. These measurements provide a basis for input to our configurable articulatory synthesizer, which can match the size and acoustic output of individual speakers. Stimuli generated from this synthesizer can test hypotheses about what is important in the production patterns we see. The results of these experiments will show more clearly than ever the tight link between production and perception of speech. Relevance: Speech is the primary means most humans use to communicate and maintain social relationships, but it is vulnerable to a range of disorders. We have to understand how it is that speech works normally so we can know what to do when things go wrong. Research along the lines in the present project has already contributed to other grants dealing with such disorders as Parkinson's disease and autism.
Showing the most recent 10 out of 72 publications
