Eyebrow raising in dialogue : discourse structure, utterance function, and pitch accents

• Main Author: Flecha-García, María L.
• Published: University of Edinburgh 2006
• Subjects: 410
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.650956

This thesis investigated eyebrow raising in a small corpus of task oriented English dialogues. Applying a standard dialogue coding scheme (Conversational Game Analysis, Carletta et al. 1997), eyebrow raises were studied in connection with discourse structure and utterance function. Supporting the prediction, more frequent and longer eyebrow raising occurred in the initial utterance of high level discourse segments (<i>transactions)</i> (utterance = ‘<i>move’</i> in Carletta et al.) than anywhere else in the dialogue. Additionally, eyebrow raises were more frequent in instructions than in requests for or acknowledgements of information. Instructions had also longer eyebrow raising than any other type of utterance. The start of a lower discourse level (<i>conversational game)</i>, and queries, did not have more frequent or longer eyebrow raising than any other position in the dialogue and any other type of utterance, respectively. In order to explore their location inside utterances eyebrow raises were also studied in relation to intonational events, namely pitch accents. Results showed evidence of alignment between the two events (the brow raise start and the pitch accent start). On average, eyebrow raises occurred 0.06sec (±0.46) before the nearest pitch accent on the speech signal, though, of course, not all pitch accents occurred next to a brow raise. To investigate what could explain the alignment between the two events, pitch accents aligned with brow raises were compared to all other pitch accents in terms of: phonological characteristics (<i>primary </i>vs<i>. secondary </i>pitch accents, and <i>downstep-initial </i>vs.<i>non-initial </i>pitch accents) and information structure (given vs. <i>new information </i>in referring expressions, and the <i>last quarter of an utterance word length </i>vs.<i> earlier parts </i>of the utterance). Brow raises proved to be aligned more frequently with <i>downstep-initial </i>pitch accents, but not with <i>primary</i> pitch accents, and they were not related to information structure of referring expressions or to the last part of an utterance.