The interaction between speech recognition dictation technology and language use: a sociolinguistic study

To explore the interaction between speech recognition dictation technology and language use, four different types of literacy users from diverse ethno-linguistic backgrounds were asked to orally compose a series of college-level writing tasks using the technology. The interactions between the participants and the software were videotaped and transcribed; the transcriptions include the participants' verbal and non-verbal behaviors, and the actions of the participants and the computer. The analysis of the transcripts draws on Halliday's conceptual framework of register, emphasizing the relationship between the language forms and the features of the context. The findings indicate the manner in which the technology influenced the development of text and patterns of interaction. Three of the participants adopted a strategy of perfecting the composition as much as possible during the speech sessions. This strategy included making changes in the surface features of their talk in order to accommodate the software. Speech adjustments, associated with what I call “speech recognition talk,” included lexical choice, utterance length, and pronunciation. The other strategy adopted consisted of quickly dictating a draft with little oral editing or linguistic accommodation to the software. These two strategies shifted to various degrees over time. For the tenor of discourse, two of the participants gained more control over oral phenomena that did not translate well into writing such as filled pauses, false starts, meta-comments, and uttering the imprecise wording of oral punctuation commands. As part of the two-way interaction, the software improved in its recognition rate as these same two participants updated their voice templates during the editing sessions. With respect to the mode of discourse, participants used a novel combination of spoken and written features of language, indicating a distinct skill-set associated with speech recognition-mediated literacy. The conclusions of this study offer a better understanding as to the influences of the situational features associated with speech recognition dictation technology on spoken and written language. The recommendations include the expansion of possibilities associated with the transfer of modes afforded by this technology, as well as insights into the constraints that the technology imposes on communication.