"I, uh, meant to do that!" - Intentional Disfluency as an Element of Audience Design

Abstract

The purpose of the present study was to determine whether disfluency is produced intentionally as a signal to the listener, as proposed by Clark and Fox Tree (2002), or whether it is produced unconsciously as a reaction to problems during speech planning (Blackmer & Mitton, 1991; Postma & Kolk, 1993). Codability of items was introduced as a factor to increase rates of overall disfluency. It was also of interest to establish whether disfluency rates would pattern with intentional audience design, as opposed to apparent audience design in the form of priming. A positive correlation between disfluency and audience design would lend weight to the idea of intentional disfluency. To exhibit measurable audience design, a participant and confederate carried out a card matching task similar to that used by Haywood, Pickering and Branigan (2003). Utterances were coded as ‘helpful’ when the participant used a noun phrase order which matched the organisation of the box used by the confederate. Attention was divided by means of a dual task paradigm between card description and auditory beep monitoring. Disfluency rates increased with divided attention, whereas levels of helpfulness remained unaffected, supporting the theory that disfluency is produced automatically. The lower an item’s codability, the more likely participants were to be disfluent, but they were less likely to produce helpful audience design. This suggests they are governed by separate systems and further supports disfluency as an automatic reaction.