Show simple item record

dc.contributor.authorWisenburn, Bruce
dc.date.accessioned2016-03-28T19:06:30Z
dc.date.available2016-03-28T19:06:30Z
dc.date.issued2005
dc.identifier.isbn9780496895632
dc.identifier.isbn049689563X
dc.identifier.other305378080
dc.identifier.urihttp://hdl.handle.net/10477/44831
dc.description.abstractConverser is a computer program designed to serve as an augmentative/alternative communication (AAC) aid. AAC users have difficulty communicating due to a slow rate of expressive language and due to a passive communication style during interactions. The goal of Converser is to recognize and parse interlocutor speech input in order to generate conversationally relevant responses for the AAC user. This may improve rate and decrease passiveness in communication. This dissertation focuses on Converser's use of natural language processing to recognize speech input and identify noun phrases. Converser then produces two responses related to the interlocutor input: "Tell me more about <noun phrase>" and "I want to say something about <noun phrase>." Converser also allows the user to select the noun phrase by itself to combine with spelled text. In the main study, seventeen dyads (consisting of a user and an interlocutor participant) were used to test Converser's efficacy for communication. The dyads participated in two communication tasks: a conversation and an interview task. Two conditions were applied: a simple alphabet board aid without Converser (Alpha-only condition), and an identical aid with Converser (Alpha-Converser condition). Objective measurements were made concerning the user's rate and Converser usage. Subjective data was gathered through rating questionnaires and written comments. The results showed that the Alpha-Converser condition generated a faster communication rate than the Alpha-only condition. User participants also rated the speed of communication faster in the Alpha-Converser condition. User ratings of quality and interlocutor ratings of speed and quality showed no difference between the two conditions, although the participant comments about Converser were positive. These results suggest that Converser may be efficacious for both rate and quality for AAC.
dc.languageEnglish
dc.sourceDissertations & Theses @ SUNY Buffalo,ProQuest Dissertations & Theses Global
dc.subjectHealth and environmental sciences
dc.subjectApplied sciences
dc.subjectSpeech recognition
dc.subjectNatural language processing
dc.subjectUtterance
dc.subjectAugmentative and alternative communication
dc.titleThe use of natural language processing for augmentative-alternative communication utterance generation
dc.typeDissertation/Thesis


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record