This study sought to determine if subjects could discriminate turn-final and turn-medial utterances taken from different types of natural conversation and presented out of context. The main aim of the study was to see if subjects could make this distinction when presented with either an audio recording of the utterances, thus testing the role of prosody, paralanguage plus syntax and semantics in the turn-taking process, or a typescript of the utterances which served to control for the effects of syntax and semantics. The study also investigated whether perceived drawl was associated with turn endings and if judgements about the apparent presence of drawl on the final or stressed syllable of the terminal clause correlated with the ability to distinguish turn-final and turn-medial utterances. The study revealed that subjects could discriminate tum-final and turn-medial utterances but only when presented with an audio recording of the samples and in addition when these samples were taken from conversations involving disagreement (but not from those conversations involving agreement). Subjects were found not to distinguish turn-final and turn-medial utterances in terms of the perception of drawl, when the extracts were presented in random order, but they could do this when the extracts from each individual speaker were presented consecutively. The perceived presence of drawl did not, however, seem to correlate with the judgement of the completion to turns. The results demonstrate the importance of prosodic and paralinguistic elements in the regulation of speaker turns in conversation.