A PRELIMINARY EXAMINATION OF CONVERSATION ANALYTIC TECHNIQUES - RATESOF INTER-TRANSCRIBER RELIABILITY

Conversation analysis has enjoyed recent acceptance in mainstream comm unication research. Despite successes, it remains criticized by quanti tative researchers for a variety of reasons. One such reason is that, to date, conversation analysts have not felt obligated to demonstrate ''inter-transcriber'' reliability for the use of transcription notatio n. This paper argues for determining ways to address this criticism. T oward that end, 22 participants were asked to transcribe a randomly se lected segment of naturally occurring conversation. The transcripts we re compared and percentages of agreement were computed for overall agr eement as well as agreement regarding verbal content, pauses, overlaps , and intonation. Additionally, education level was considered as a pr edictor of transcription agreement. Results indicate that multiple tra nscribers are capable of producing similar transcripts within the acce ptable tolerances of a priori coding schemes. With the exception of pa uses, all transcription notations were agreed upon at a rate of 75% or better. The data also suggest that there is very little difference in transcript accuracy rates between undergraduates and graduate student s. Thus, education level on this dimension does not appear to have an affect.