Gaze and Recipient Feedback in Triadic Storytelling Activities

ABSTRACT

This article focuses on the relationship between gaze and recipient feedback in triadic storytelling activities. Our starting point to investigate this relationship is an article by Bavelas et al. (2002), who report a statistically significant interaction between feedback and so-called gaze windows in dyadic storytelling activities. The pattern they found involves three interdependent phases: a narrator shifts gaze to a recipient to elicit feedback (phase 1), narrator and recipient are mutually gazing at each other while the recipient gives feedback (phase 2), and the narrator shifts gaze away from the recipient (phase 3). In our triadic data, where gaze was captured by mobile eye tracking glasses, gaze is equally used to elicit feedback. However, feedback is most often not produced during mutual gaze between narrators and recipients. The gaze window pattern is instantiated in only roughly one third of the feedback tokens in our corpus. We illustrate coexisting gaze patterns and discuss reasons why mutual gaze between narrator and feedback-giving recipient is less prevalent in our data.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. Turn constructional units (TCUs) are units of turns-at-talk. They are not linguistic units (such as, e.g., clauses) but defined with respect to turn-taking: at TCU boundaries a turn is potentially complete (cf. Selting, 2000).

2. Following Lerner (1992) and Bavelas et al. (2002), the collaboration of the recipients is indeed a prerequisite as fundamental to the success of interactional storytelling activities that they should be considered co-narrators. We nonetheless use the term recipient in this article. This is to avoid terminological confusion as we also look at storytelling activities with two co-narrators who recount a shared experience to a third party.

3. Cf. Schegloff (1982) on continuers such as mhm and uh huh; Heritage (1984) and Goodwin (1986) on change-of state tokens such as oh; Schegloff (1982) and Jefferson (1983) on acknowledgment tokens such as yeah; and Goodwin and Goodwin (1987) and McCarthy (2003) on assessments, among many others.

4. Our study differs from the study design of Bavelas et al. (2002) in all three aspects, and we pay close attention to these differences when discussing our results.

5. They report intercoder reliability rates on a test sample (two dyads) of 100% for mutual gaze and 83.9% and 86.75% for onset and offset of mutual gaze.

6. The default gaze pattern for narrators is to regularly change gaze orientation toward and away from the recipients, while recipients most of the time sustain gaze at the narrator (Argyle & Cook, 1976; Duncan & Fiske, 1977; Goodwin, 1981; Kendon, 1967; Rossano, 2012a, 2012b).

7. Bavelas et al. (2002) did indeed investigate both verbal feedback tokens and nods, but these were ultimately lumped together in one category for quantitative analysis.

8. One recording involved both an extended joint storytelling activity and various storytelling sequences with one narrator only. In another recording, participant first talked about a joint visit to the movies and this storytelling activity was followed by a spontaneous, nonelicited co-narration of a joint visit to the opera. Therefore, the total number of participants is lower than 21.

9. We did not, however, investigate whether there is a potential gender influence on speakers’ or recipients’ gaze behavior.

10. The sampling rate for SMI eye trackers is 30 Hz and 50 Hz for Tobii eye trackers.

11. We have obtained written consent from all participants to show the stills from the eye trackers and external camera.

12. This is to exclude that we code saccades (i.e., rapid eye movements from one point of fixation to another during which visual perception is reduced) as fixations.

13. We are aware of the fact that this “verbal feedback” category is a quite heterogeneous, both with regard to their interactional function and its relationship to gaze. In order to not make the following analysis overly complicated, we decided not to zoom in on the details between the different categories of verbal feedback but to address them elsewhere.

14. This methodological decision is a measure of caution to make sure we do not come to different results than previous research because laughing (and hence gaze aversion) is very common in some, but not all, of our conversations and was presumably not too frequent in the interactions between unacquainted participants studied by Bavelas et al. (2002).

15. Intonation units are defined as “a stretch of speech uttered under a single coherent intonation contour” (Du Bois et al., 1993). The final pitch movement of an intonation phrase is interpreted in terms of whether it projects completion or continuation. Recipients orient to these intonation contours to projects possible points of completion (TCU boundaries and TRPs, supra) at which turn-taking may occur.

16. We therefore cannot exclude that our results also differ from what Bavelas et al. (2002) found because they accepted, for example, much longer phases of mutual gaze after feedback has been given or treated much longer narrator gazes at the recipients as gazes that serve to elicit a response. However, other reasons that we discuss in the Discussion and Conclusions section may be weightier.

17. Unlike Bavelas et al. (2002, p. 575) we do not add .5 seconds to each gaze window. Given the fact that we use eye tracking data we can directly measure the length of each mutual gaze phase and therefore do not manually add time to the observed length mutual gaze. P is the average mutual gaze in storytelling activities in each conversation. All other activities that the interactants were engaged in during the recordings were not taken into account to calculate p.

18. Each completion of a TCU establishes “a transition – relevance place (TRP) where a change of speakership becomes a salient possibility that may or may not be realized at any particular TRP” (Clayman, 2012, p. 151).

19. Her co-teller Nora could hence take the turn at this TRP.

20. The fact that in example (4) Zac—the nongazed at recipient—only offers a short continuer, while Den gives a rather strong assessment may be related to the narrator’s gaze at him as he may feel a stronger urge to give relevant feedback. This issue, however, cannot be dealt with in depth in this article.

Additional information

Funding

The research presented in this article was supported by the Ministry of Science and Arts Baden Württemberg and the ESF (Margarete von Wrangell Habilitation Programme).