The Prosody of Sluicing: Production Studies on Prosodic Disambiguation

Abstract

With this thesis, I investigate the prosodic realizations of different sluicing structures, as produced by either trained or untrained native speakers of English. Sluicing is a subtype of ellipsis where the major part of a wh-question has been elided, leaving only a wh-remnant behind (Ross, 1969). From this follows that sluicing can be ambiguous if the wh-remnant has more than one possible antecedent in the preceding un-elided clause. If one of these possible antecedents is located within an island to extraction, the respective sluicing structure is called complex sluicing (Konietzko, Radó, & Winkler, submitted; Ross, 1969; Merchant, 2001). The perception, especially of simple sluicing, has been examined to some extent (Frazier & Clifton, 1998; Carlson, Dickey, Frazier, & Clifton, 2009), finding that listeners prefer a prosodically or syntactically focused NP to be the antecedent of an ambiguous wh-remnant. However, the prosodic production side has not been empirically investigated to date. With this thesis, I thus explore the relationship between prosody and the disambiguation of different sluicing structures in spoken language. With three production studies, I investigate how various sluicing structures with different antecedent types are produced by speakers who are either trained or untrained with respect to the ambiguity of the target items and prosody as a disambiguation technique. I present the results of a pilot production study that examined globally ambiguous simple sluicing structures with contextual disambiguation and two production studies that examined temporarily ambiguous simple and complex sluicing structures with morphological disambiguation. Four preceding acceptability judgment studies made sure that there were no additional factors interfering with the prosodic realizations of the different sluicing structures. The three production studies found that both trained as well as untrained speakers use prosodic prominence as a disambiguating factor to emphasize which NP serves as the antecedent of a contextually or morphologically disambiguated simple or complex sluicing structure. However, an early, sentence-initial NP is more frequently disambiguated than a late, sentence-final NP, both by trained and untrained speakers. In complex sluicing, only a sentence-initial NP is prosodically disambiguated, only by trained speakers. Moreover, trained speakers generally make more frequent use of prosody as a disambiguation technique and they produce stronger prosodic cues than untrained speakers. With this thesis, I thus show that prosody, in the form of prosodic prominence, is used by native speakers of English to indicate the meaning of an information-structurally triggered ambiguity. With this finding, I add further support to Romero (1998), Frazier and Clifton (1998) and Carlson et al. (2009), who argue that a constituent with a prosodic focus is preferably taken as the antecedent of the wh-remnant. Moreover, I add support to Remmele, Schopper, Winkler, and Hörnig (forthcoming 2019), who found that even untrained speakers use prosodic phrasing to resolve a structurally ambiguous word sequence. Furthermore, I contradict Wasow (2015) and Piantadosi, Tily, and Gibson (2012), who argue that one form of disambiguation suffices, thus rendering additional prosodic cues redundant. The results of this thesis thus contribute significantly to the research about the prosody of sluicing and the research about prosodic disambiguation in general.

Gefördert durch die Deutsche Forschungsgemeinschaft (DFG) - Projektnummer 198647426