Understanding Language Through TalkBank

Abstract

Advances in computer technology have produced a flood of new data sets for understanding human language. However, nearly all of these new data sets are based on written, rather than spoken, language. This means that, despite their importance, open-access data on naturally occurring spoken-language conversations are much more difficult to obtain. The major exception to this is the TalkBank system, which provides online multimedia data for 15 types of spoken-language data: language in aphasia, child language, stuttering, child phonology, autism spectrum disorder, bilingualism, conversation analysis, classroom discourse, dementia, psychosis, right hemisphere damage, Danish conversation, second-language learning, traumatic brain injury, and daylong recordings in the home. This article reviews these resources and describes the ways that they are being used to further our understanding of language learning and usage.