What counts as a preferred or dispreferred answer to a question and how can it be recognized? These questions have been widely discussed in conversation analysis in recent years, and some well-established diagnostic features of dispreferred answers have been identified. In this paper, we turn to a less well-established feature: gaze aversion. Kendrick and Holler (2017) have previously suggested that answers to polar questions that invert the question’s polarity are accompanied by gaze aversion to a statistically significant extent. They argue that this gaze aversion indicates the answer’s dispreferred status. We were able to replicate their findings using a large collection of English and German conversational data for polar question-answer sequences. However, we found an even stronger tendency for the answerer to avert their gaze during or after wh -questions, for which, per definition, the concept of preference in terms of polarity matching does not apply. Therefore, we extended the focus of our study to both polar and wh -questions and to alternative concepts of preference: type-conformity and pragmatic preference. Only the latter was found to be associated with gaze aversion in a statistically significant way. Thus, we considered an alternative explanation for answerers’ gaze aversion, also suggested by Kendrick and Holler (2017): answer complexity. Logistic regression analysis revealed that answer complexity, measured by the number of TCUs and absolute length, as well as hesitations, are significant predictors of gaze aversion, while none of the preference types are. These results demonstrate that gaze aversion does not index dispreference so much as it indexes that a complex answer is about to follow.
Auer et al. (Mon,) studied this question.