Shari R. Speer,1 Amy J. Schafer2 & Paul Warren3
1 The Ohio State University, 2 University of Hawaii, 3 Victoria University of Wellington
Previous research comparing spoken questions with and without wh-gaps demonstrated prosodic differences at the potential gap location, suggesting that syntactic gap locations may be directly indicated in prosodic structure [1]. However, recent findings suggest that when potential gap location and syntactic complement/adjunct structure are manipulated separately, prosodic phrasing reflects 'sense-unit' constituents sensitive to these relations, rather than directly reflecting the location of a syntactic gap [2]. We examined the production of questions with wh-gaps like those in (1) and (2) below, both of which have complement to-clauses.
(1) Which trianglei do you want to change the position of ____i this time?
(2) Which trianglei do you want ____i to change the position of the square?
Our production paradigm elicited utterances using a boardgame-based conversation task. Seven pairs of US English speakers and 6 pairs of NZ English speakers, naive to the study's intent, played a series of games. They manipulated gamepieces from start to goal positions, using a limited set of object names and syntactic constructions that were learned easily within an initial practice round and repeated throughout the game series. Players could not see each other's gameboards, and had to communicate cooperatively to complete moves. Permissible utterances included syntactic ambiguities of several types as well as unambiguous ones. Utterances produced were 'quasi-spontaneous' speech, reflecting the pragmatic goals of speakers engaged in a complex conversational task.
The wh-gap studies mentioned above employed reading-based production tasks, measuring prosodic marking reflected in word and silence durations, and F0 excursion across the gap site. We examined game-task productions from two dialects of English, measuring the incidence of wanna-contraction as well as the temporal and intonational indications of prosodic structure across the gap sites. Wanna-contraction was more frequent in the US data, and more likely in both data-sets when there was no gap between want and to, as in (1) above. Phonetic analyses indicated that both groups of speakers consistently lengthened the word and post-word silence preceding the gap location. ToBI transcriptions of matched utterance sets showed a range of intonational phrasing patterns in the vicinity of the potential gap, with prosodic constituent breaks most often patterning with the location of major syntactic breaks. Our results do not necessarily imply that syntactic gaps are directly pronounced in the phonetic structure of an utterance. Nevertheless, we argue that phonological regularities, in combination with the prosody-syntax correspondence rules, may often be available to comprehenders.
References
[1] Nagel, H. N., Shapiro, L. P., & Nawy, R. (1994). Prosody and the processing of filler-gap dependencies. Journal of Psycholinguistic Research, 23, 473-485.
[2] Straub, K., Wilson, C., McCullum, C., & Badecker, W. (2001). Prosodic structure and wh-questions. Journal of Psycholinguistic Research, 30, 379-394.