Hesitations and cognitive status of noun phrase referents in spontaneous discourse

Speech disfluency is typical of spontaneous speech. Dialogues as well as spontaneous monologues are full of hesitations, filled pauses (um's and ah's), false starts, fragmented words, repairs, etc. The main goal of this thesis is to test the hypothesis that there is a relationship between hesitations in spontaneous speech production and the perceived cognitive (memory and attention) statuses of noun phrase referents for participants in the conversation. This hypothesis was motivated by the idea that the production of hesitations may be closely linked to the cognitive processes involved in the production of referring expressions.; The primary hypothesis asks the question whether the frequency of hesitations immediately prior to production of a referring expression (i.e. noun phrase) in conversational speech depends on how restrictive the form of the referring expression is in the range of cognitive statuses (Gundel et al. 1993) it is normally associated with. The secondary hypothesis abstracts away from the formal characteristics of noun phrases and asks the question whether there is a relationship between the perceived cognitive status of the NP and the frequency of hesitations preceding it. The hypotheses are motivated by the idea that the processing load on the speaker is dependent on how restrictive a given expression is. The restrictiveness is expressed in terms of the relationship between the form and the possible cognitive statuses of its referent in the case of the primary hypothesis and in terms of the unidirectional entailment of cognitive status categories in the case of the secondary hypothesis.; The empirical data collected in this thesis are consistent with the predictions of the primary and the secondary hypothesis; however, the data also suggest that, for various reasons, not all hesitations behave uniformly with respect to the hypotheses. When all the different kinds of hesitations are combined and treated as a generic hesitation, the results confirm the hypothesis. However, when considered separately, the different hesitation types display individual differences. This may indicate that different types of hesitations have different functions and that some of them may be more sensitive to the processes discussed in this thesis than others.