Cohesive devices in conversations

Abstract

The literature on conversational coherence is filled with devices that are believed to be the bases of coherence, but no test has been performed to determine if those devices are found more frequently in coherent than incoherent messages. This research was designed to find out how coherence is cued between turns in conversations. In Phase 1, the assumption that conversations contain explicit cues to coherence was tested by asking subjects to reconstruct conversations from randomized constituent remarks. Only if connections between turns were explicitly cued would subjects have any basis for placing turns together beyond levels expected by chance. The findings indicated that turns were consistently paired well beyond chance, suggesting that explicit cues to connections between turns in conversations do exist. The next step was to determine what those cues were. In Phase 2 a coding system was developed and applied to the pairs of turns identified in the reconstruction task as either coherent or incoherent. Syntactic, pragmatic and lexical devices were found in all conversations, but only lexical devices were found significantly more often in coherent than incoherent pairs. The findings address current controversies about how different types of devices cue conversational coherence.