The overall goal of this project is to determine how listeners recognize words whose surface forms have been modified as a result of lawful phonological processes including English coronal place assimilation. This project will continue the development, testing and refinement of a comprehensive theoretical framework for understanding this problem that emphasizes the importance of fine phonetic detail and the role of general auditory grouping mechanisms in processing. The work is organized around three aims: 1) to determine what role general auditory grouping mechanisms play in the processing of assimilated speech and what level of representation they operate on, (2) to determine how phonetic fine structure that has been shown to influence lexical processing is reflected in the categorization of assimilated speech, and (3) to determine what role different types of implicit linguistic knowledge play in the processing of assimilated items heard in context. These issues will be examined through a combination of off-line perceptual tasks (phoneme identification, AXB discrimination and categorical goodness rating tasks), on-line behavioral tasks (phoneme monitoring and form priming), and the use of techniques combining magnetoencephalography (MEG), electroencephalography (ERP), and functional and structural MRI. Stimuli in these experiments will consist of a combination of synthesized speech, unmodified and digitally modified natural speech, and synthesized speech analogs observing carefully measured or controlled acoustic parameters. This work will provide insight into the relationship between linguistic competence, auditory scene analysis mechanisms and spoken word recognition. As basic research it provides a framework for understanding these processes that may be useful in characterizing and developing effective remediation for communicative disorders, and in developing automatic speech synthesis and recognition technologies that may serve people with a variety of perceptual and communicative disorders.
Showing the most recent 10 out of 12 publications
