Taking the Duncan & Fiske stuff, but going at it through the Prague school, i.e., focusing on the indexical, as opposed to iconic or symbolic, nature of signs, looking at whether turn exchange is a contextual feature of conversation indexed by linguistic and/or nonlinguistic forms, and whether other contextual parameters are also indexed at turn exchanges. Indexes (straight from Peirce, via Silverstein) may be referential, as in shifters, or not-if they're not, they signal a value of one or more contextual variable, independent of the referential potential of speech. By virtue of its indexical properties, a device may be seen as automatized (basically neutral) or foregrounded. Occurrence of a token signals the value of a context parameter, the function that describes the relation between token and type is the "rule of use" or "rule of indexicality." Turn taking is indexical: actions of interlocutors are indexically related to the context parameters of turn exchange. Turn taking is optional; actions in the exchange are indexically related to the opportunities for smooth exchange. Sacks looks at turn constructional units as syntactic, the types grammatically characterizable. Within the CA system, nonverbal behavior is unimportant. D&F's system gives weight to both verbal and nonverbal elements, cues, differentiate smooth from simultaneous exchange. Not known whether the nonverbal indexes are redundant to the syntactically motivated units observed in CA [but could the two be really separated, except experimentally?]. Denny hypothesizing that turn taking is an indexical system, behaviors about turn exchange as indexes, with different forms of marking indexing different things. Strategy [in D&F's terminology, i.e., like creativity, agency] then becomes study of the use of indexes.
Once index is delineated in analysis, can rank it on a continuum of pragmatic markedness-from unmarked (expected in the context, not really noticed), to marked. Marked indices invoke/create some contextual parameter which would not otherwise have been supposed. Present article is looking for indexes, in particular nonlinguistic, that effect smooth turn exchanges, see if these are necessary in turn exchange. Would expect one-to-one correspondence of verbal with nonverbal elements comprising an indexical form establishing locations of turn exchange.
Finds that gesture has a clear effect on the rate of transition. Different elements differently distributed in their probabilistic salience; Certain set(s) of elements can predict turn exchange with a probability near to 1.0, regardless of the speech event [calls CA somewhat into question-it's not just about the grammatical symbols]-most complete predictor is grammatical completion, gaze, auditor gaze, and pause length combined: if all these aren't simultaneously present, the index is pragmatically marked, and a more stylized speech form will result (e.g., questions and answers, length of next turn). Marked exchange restrict subsequent speech, organize the linear unfolding of talk. The location of turn exchange as a locus for the instantiation of indexical signals-brings up question of why turn exchange not always indexed: in some cases, illocutionary force of speech is enough. That illocutionary force may render the index of turn exchange irrelevant, making the location of turn exchange a site for marking other aspects of conversation.