Revisiting Preference Organization in Context: A Qualitative and Quantitative Examination of Responses to Information Seeking

ABSTRACT

Quantitative studies applying conversation analysis to the study of the timing of answers to sequence-initiating actions expose anomalies in terms of what is known about preference organization. After briefly describing preference organization, anomalies in answer-timing research, and one explanation for such anomalies, this article presents one qualitative and one quantitative study of responses to one thickly contextualized action: positively formatted polar interrogatives implementing information seeking with a relatively ‘unknowing’ stance. Data include 249 questions gathered from videotapes of unstructured conversations. Qualitative results suggest that, rather than two basic answer types (i.e., affirmation/disaffirmation), there may be three: unconditional affirmation, unconditional disaffirmation, and conditional. Quantitative analyses of time to answer, eyeball shifting, and pre-beginning behavior suggest that unconditional disaffirmation may not be dispreferred relative to unconditional affirmation. Instead, conditional answers may be dispreferred. Results begin to reconcile anomalies and expand our current understanding of preference organization. Data are in American English.

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.

Notes

¹ I thank Kobin Kendrick for providing these unpublished times, which were derived from Kendrick and Torreira’s (2015) data. These are median times reported because compared to means, they are less affected by outliers and skewed data and because they are most comparable to the medians reported in the present article. These delay times represent Kendrick and Torreira’s ‘offset 3’ time, which was measured just before “the first word of the base TCU” (p. 266) but after silence, “clicks, in-breaths, and the prefatory particles ‘u(h)m’ and ‘well’” (p. 266). ‘Offset 3’ is an excellent measure of answer delay for the purposes of analyzing preference organization, insofar as silence, breathing, tongue-clicks, Uhm, and Well are all canonical practices for breaking contiguity (Sacks, 1987) and projecting dispreferred answers.

² ’These times reported by Kendrick and Holler (2017) are not precisely comparable to those reported by Kendrick and Torreira (2015). These times represent Kendrick and Holler’s ‘offset 2’ time, which did include silence, tongue-clicks, and breathing but did not include behaviors such as Uhm and Well. Kendrick and Holler did not measure Kendrick and Torreira’s (2015) ‘offset 3’ time, which did include behaviors such as Uhm and Well. Compared to ‘offset 2’ times, ‘offset 3’ times are arguably a more true measure of preference-related delays, insofar as Uhm and Well are canonical practices for delaying and projecting dispreferred answers (Heritage, 2015; Schegloff, 2007). Prior to Kendrick and Holler (2017), Stivers et al. (2009) dealt with the timing of responses to polar questions. However, Stivers et al. measured the equivalent of Kendrick and Torreira’s (2015) and Kendrick and Holler’s (2017) ‘offset 1’ time, which is the “time elapsed between the end of the question turn and the beginning of the response turn” (Stivers et al., p. 10591, emphasis added). Stivers et al. considered the response turn to begin upon the initiation of any vocal response, including audible breathing, tongue clicks, Uhm, Well, etc. (p. 10591). Thus, while Stivers et al.’s times (as well as Kendrick et al.’s ‘offset 1’ times) are extremely useful in terms of determining how quickly selected speakers can respond, generally, these times are not the best indicator of answer-delay, per se, and thus less useful for analyzing preference organization. Note also that Stivers et al. report mean offset times, which (relative to medians) are more biased by outliers and skewed data.

³ Further distinguishing conditional answers as either ‘conditional affirmation’ or ‘conditional disaffirmation’ is difficult. By their very nature, conditional answers simultaneously affirm some of questions’ terms or agendas, yet disaffirm others. While some conditional answers include Yes/No-type tokens (e.g., Yes, but…, No, but…), some do not, and it is not clear that initial (or subsequent) (dis)affirming units should be privileged in terms of categorizing answers as (dis)affirmation. (Remember Sacks’s, 1987, famous answer: Have you friends? – > I have friends. So called friends. I had friends. Let me put it that way). That said, I did roughly categorize conditional answers as ‘conditional affirmation’ (median delay = 588 ms) and ‘conditional disaffirmation’ (median delay = 466 ms). Note the potentially counterintuitive result of conditional-disaffirmation answers having shorter delays, which further reinforced my decision to not subcategorize.