Gaze Direction Signals Response Preference in Conversation

In this article, we examine gaze direction in responses to polar questions using both quantitative and conversation analytic (CA) methods. The data come from a novel corpus of conversations in which participants wore eyetracking glasses to obtain direct measures of their eye movements. The results show that while most preferred responses are produced with gaze toward the questioner, most dispreferred responses are produced with gaze aversion. We further demonstrate that gaze aversion by respondents can occasion self-repair by questioners in the transition space between turns, indicating that the relationship between gaze direction and preference is more than a mere statistical association. We conclude that gaze direction in responses to polar questions functions as a signal of response preference. Data are in American, British, and Canadian English. The morphology of the human eye has evolved to allow those around us to recognize the direction of our gaze. The stark contrast between the white sclera and the darker pupil of the human eye exposes our gaze direction, whereas the dark sclera of nonhuman primates serves to camouflage theirs (Kobayashi & Kohshima, 1997). Our unique sensitivity to gaze direction is evident in the behavior of infants who can discriminate between direct and averted gaze just 2 days after birth (Farroni, Csibra, Simion, & Johnson, 2002). In the behavioral repertoire of most species, direct gaze is predominantly a signal of aggression or threat (with the possible exception of mother-infant dyads among some of the great apes; Gomez, 1996), while gaze aversion tends to signal submission. In much of the animal kingdom, gaze thus functions as an antithetic signal (Darwin, 1872). The scope of functions that gaze fulfills in human interaction is vast in comparison (see Rossano, 2013). The functional complexity of human gaze is perhaps most obvious in conversation. Goffman (1964) wrote of the “eye-to-eye ecological huddle” that is characteristic of human interaction, referring to the strong human tendency to mutually orient to one another, thus facilitating mutual visual perception. Despite this tendency, however, we do not constantly maintain eye contact. Rather, face-to-face conversation consists of a complex interactional dance, as it were, with frequently alternating periods of gazing at the other and gazing away. In his landmark study of gaze direction, Kendon (1967) identified, among other patterns, a general asymmetry in the behavior of speakers and recipients: Whereas recipients tend to maintain gaze to speakers, speakers alternate their gaze toward and away from recipients as they speak (see also Argyle & Cook, 1976; Bavelas, Coates, & Johnson, 2002; Ho, Foulsham, & Kingstone, 2015). CONTACT Kobin Kendrick [email protected] Department of Language and Linguistic Science, University of York, Heslington, York YO10 5DD, United Kingdom. We are grateful to Linda Drijvers, Ludy Cilissen, and Sean Roberts for assistance. We would also like to thank the anonymous reviewers for their valuable suggestions and comments. Color versions of one or more of the figures in this article can be found online at www.tandfonline.com/HRLS. Supplemental files referred to in this article can be accessed on the publisher’s website. © 2017 Taylor & Francis RESEARCH ON LANGUAGE AND SOCIAL INTERACTION 2017, VOL. 50, NO. 1, 12–32 http://dx.doi.org/10.1080/08351813.2017.1262120 There has been much speculation on what determines the occurrence and duration of mutual gaze in conversation. Kendon (1967) proposed three main functions of gaze direction. First, he argued that gaze direction serves a regulatory function: Speakers avert their gaze to signal that they are about to take a turn, may continue to look away in order to hold the turn, and may redirect gaze to the recipient in order to yield the turn. Kendon’s observations had little impact on the development of the CA model of turn taking by Sacks, Schegloff, and Jefferson (1974), in which gaze direction serves as resource for next speaker selection (Lerner, 2003) but does not otherwise organize transitions between speakers. Outside of conversation analysis, however, the regulatory function of gaze became a central part of the cue-based model of turn taking developed by Duncan et al. (Duncan, 1972; Duncan, Brunner, & Fiske, 1979; Duncan & Fiske, 1977) and continues to garner attention (e.g., Bavelas et al., 2002; Ho et al., 2015). Second, Kendon (1967) proposed that gaze direction serves amonitoring function: By looking at the recipient, speakers gather information about his or her attentional state, facial displays, intention to take a turn, and so on. As a corollary to this, he proposed that “paying attention to one’s interlocutor and planning what to say are incompatible activities” (p. 34), suggesting that speakers avert their gaze in order to avoid interference while planning upcoming units of talk. In the subsequent literature, few have investigated the monitoring function of gaze per se (Goodwin, 1980), focusing instead on its cognitive costs (Beattie, 1978). In one line of experimental research, participants are asked not to avert their gaze as they normally would (e.g., to maintain eye contact with the experimenter as they perform an experimental task) and consequently perform worse than those who are free to avert their gaze naturally (Beattie, 1981; Doherty-Sneddon, Bonner, & Bruce, 2001; Markson & Paterson, 2009). Such results have been understood to reveal evidence of cognitive interference caused by the automatic processing of the interlocutor’s facial signals. The monitoring function of gaze is thus one element of what has become the “cognitive load hypothesis” of gaze aversion, namely, that speakers avert their gaze to reduce the cognitive costs associated with the monitoring function. Third, Kendon (1967) argued that gaze direction serves an expressive function, whereby participants regulate the level of emotionality and arousal in the interaction. In one particular conversation, Kendon found that mutual gaze was inversely related to the rate of smiling, one measure of emotionality. If the level of emotionality becomes too high, Kendon maintained, participants may avert their gaze to express embarrassment and reduce arousal. According to Kendon, the expressive function of gaze also concerns the speaker’s need for affiliation, as mutual gaze appears to increase in affiliative and cooperative interactions (p. 48). Here, we propose one specification of the expressive function of gaze direction, namely, that gaze maintenance and aversion serve as resources for the construction of socially affiliative and disaffiliative actions, among their other uses. In conversation, speakers use the resources of language and the body to form recognizable actions (Levinson, 2013; Schegloff, 1996), which cohere to form sequences of action (Schegloff, 2007). One of the most basic and ubiquitous forms of action sequence is the adjacency pair, in which one action, a first pair-part, makes conditionally relevant another, a second pair-part (Schegloff, 2007). For some adjacency pairs, the first and second pair-parts are of the same type (e.g., greetings are followed by greetings), but for most, next speakers must select from a set of alternative seconds (e.g., invitations may be followed by acceptances or declinations). The construction of such alternative second pair-parts typically reveals systematic asymmetries: While one may be produced quickly, take a simple form, and promote the accomplishment of the activity, a preferred response, the other may be delayed by silence or turn-initial particles, have a relatively complex design, and forestall the accomplishment of the activity, a dispreferred response (Pomerantz & Heritage, 2013; Schegloff, 2007; Stivers et al., 2009). The concept of preference in conversation analysis was developed to explain such systematic asymmetries, which serve to maximize opportunities for affiliative actions and minimize opportunities for disaffiliative ones (Heritage, 1984; Sacks, 1987). We hypothesize that the systematic asymmetry in the construction of preferred and dispreferred responses should be visible in the gaze behavior of participants. Specifically, we predict (a) that gaze aversion should occur more frequently in dispreferred responses to polar questions than preferred RESEARCH ON LANGUAGE AND SOCIAL INTERACTION 13