Turn order and turn distribution in multi-party storytelling

Abstract

In this paper we examine turntaking patterns in conversational storytelling. It has long been noted that turntaking in every-day narrative differs on a number of counts from turntaking in regular conversation. The differences, however, have, at best, been researched qualitatively based on casual observations and small datasets. Here, we base our analysis on two specialized corpora of conversational narrative, the Saarbrücken Corpus of Spoken English (SCOSE) containing American English 4- and 5-party stories and the Narrative Corpus (NC) containing British English 4- to 7-party narratives, as well as the conversational component of the British National Corpus (BNC). The analysis is decidedly quantitative and statistical in orientation. Specifically, we are concerned with turn order and turn distribution in conversational multi-party narrative. The aims are twofold. We wish to examine the validity of Sacks’ description of storytelling as “an attempt to control a third slot in talk, from a first” (Sacks, 1992:18), a turn order pattern we refer to as the N-notN-N pattern. We further investigate whether individual speakers’ turntaking styles have an impact on turn distribution, a measure intimately related to turn order. Moreover, given the structural differences in the data at hand (the SCOSE being raw-text, the NC being densely annotated) we employ largely different methodologies particularly in addressing turn order. The results on turntaking styles suggest that this factor cannot account for the noticeable increase in the narrator's turn share as soon as the conversational activity moves into storytelling. The results on turn order reveal the N-notN-N pattern's statistical overrepresentation in all multi-party narrative types examined. The implications of this finding are far-reaching. First, Sacks et al.’s dictum that turn order is not fixed in advance does not hold true for conversational narrative. Also, turn order in conversational narrative is not locally controlled, on a turn-by-turn basis, but globally, on the basis of the activity the conversationalists are involved in, viz. storytelling.Second, a fundamental correlate of the N-notN-N pattern is the avoidance of double-responses, that is, of two consecutive response turns following the narrator's turn. This avoidance suggests that the turn order system underlying multi-party narrative is that of 2-party talk. Further, the double-response avoidance suggests the possibility that the source of the turn-order bias in narrative is a tacit agreement between the recipients to promote the single-recipient filling the single-response slot to a ‘spokesperson’ taking the turn on behalf of all other recipients. We also note the possibility of there being a recipient-subsystem for turntaking at the single-response slot interacting with the narrator-recipient turntaking organization but still, to an extent, working on its own terms.