The primary aim of this thesis is to analyse how interaction with a conversational
partner affects the performance of speakers, their addressees, and those who
overhear the discourse. Speakers frequently appear to adjust their speech to
accommodate their addressees' needs (although the extent to which this is
deliberate is a
topic of much debate), and this often seems to occur as a direct
response to the feedback they receive from their partners. The bulk of the research
in this area has focussed upon the manner in which speakers use feedback to
facilitate the overall success of the interaction (in terms of completing the task at
hand, for example), but fewer studies have investigated how feedback specifically
affects detailed characteristics of the speakers' speech. This thesis attempts to
investigate such a topic, whilst also examining the benefit to addressees of being
able to give feedback, and the benefit to overhearers of hearing addressees'
The thesis begins by examining the effect of feedback on speakers' and addressees'
performances. It first reports a referential communication task which investigated
some of the notable differences between speakers' speech in monologue and
dialogue contexts (Experiment 1). I focussed in particular on how detailed aspects
of language production, such as the length of object descriptions and the
repetitiveness of the language used, varied between these two feedback conditions.
I also looked at how the ability to give feedback aided the addressees' performances
on the given task. Experiment 2 then analysed how the amount of feedback
received by the speaker related to, firstly, the shortening of their item descriptions
over repetitions, and secondly, the increasing consistency of descriptions (in terms
of lexical overlap) during the experiment.
The second section of the thesis focuses primarily on the benefit for overhearers of
hearing other people's feedback. It first reports two experiments that replicated and
expanded on a study by Fox Tree (1999) which showed that overhearers identified
tangrams more accurately when they overheard a dialogue rather than a
monologue. Experiments 3 and 4 tested two explanations proposed by Fox Tree for
this result; firstly, the potential presence of additional perspectives in dialogue (for
example, one partner viewing a tangram as a 'chicken', and the other calling it an
'ice skater'), and secondly the more numerous discourse markers in dialogue in
comparison with monologue. Additionally, the use of repeated descriptions by the
speakers allowed me to analyse the effect of interlocutors' 'conceptual pacts' on
overhearers' task performance. Finally, Experiments 5 and 6 assessed how the
benefit of, firstly, giving feedback (for addressees) and secondly, hearing feedback
(for overhearers) was affected by the difficulty of the task at hand. Overall, this
thesis provides evidence that whether people are producing, receiving or simply
overhearing feedback, it influences their performance in a positive manner, meaning
that, in general, a greater amount of interaction leads to more successful
communication for everyone involved.