The over-arching goal of this program of research is to understand how listeners recognize phonological variants of spoken words (e.g., [went, went]) as the words intended by the talker (e.g., [went]). Experimentation can be most productive and performed most knowledgeably with an accurate description (acoustic and phonological) of pronunciation variation in English.
The aim of the present proposal is to make it possible to develop such descriptions by completing phonetic transcription of a 300,000-word corpus of conversational speech. When transcription is finished, not only will we use it for this purpose, but the corpus will be made available to researchers in the speech sciences. User-friendly software to analyze and search the corpus, along with a host of supporting material (instruction manuals, tutorials, user forum) will also be provided to facilitate and encourage its use.
