‘Um, er’: how meaning varies between speech and its typed transcript

We report a small empirical study on the way the transcription used to represent speech affects its meaning. We show that ‘disfluencies’ in speech indicate far more uncertainty in the speaker when transmitted in text than when transmitted in recorded sound. This has important implications for how transcribed interviews should be edited when they are being used to convey meaning rather than the organization of phonemes. We propose the implications of different ways of representing speech in text could be a new subject for investigation. Presented here is one possible empirical approach to such studies.