The Event-Dependent Individuals and the Ambiguity of Donkey Sentences

Language, Information and Computation(PACLIC 11),1996, 197-207 The Event-Dependent Individuals and the Ambiguity of Donkey Sentences Eun-Joo Kwak Language Research Institute, Seoul National University ekwak@plaza.snu.ac.kr Abstract This paper addresses multiple proportional readings in donkey sentences and deals with a question of why the disambiguation of proportional readings is affected by the topic structure of a sentence. The correlation between proportional readings and topic structure has been observed by Heim (1990) and Chierchia (1992) but has not been given a satisfactory explanation yet. Based on my previous study that indefinite NPs are ambiguous between quantificational and cardinal readings, and proper readings are determined by the topic structure, I will show how appropriate readings of donkey sentences are derived from the semantics of indefinite NPs and the syntactic representation of topic structure. 1. Introduction In this paper I address the ambiguity of donkey sentences, namely multiple proportional readings. Kadmon (1987) notices that a conditional donkey sentence may have three different proportional readings, i.e., an asymmetric reading oriented to either a subject or an object, and a symmetric reading, while a relative-clause donkey sentence is unambiguously asymmetric to a head noun. Recently, it has been observed by Heim (1990) and Chierchia (1992) that the proportional readings of a conditional donkey sentence are disambiguated by the topic structure. As always resonated in the literature of donkey sentences, a solution to the proportion problem hinges on the semantics of indefinite NPs and quantificational adverbs. Two main analyses have been suggested for this problem, i.e., Heim (1990) and Chierchia (1992). Both of the analyses assume that indefinite NPs are existential quantifiers, and quantificational adverbs are quantifiers over situations or events. Based on this assumption, they assign multiple proportional readings to donkey sentences; however, neither of them give a convincing explanation for the role of topic structure, namely disambiguating proportional readings. I start with my previous study, Kwak (1995), about the ambiguity of indefinite NPs and the semantics of quantificational adverbs. For independent reasons, I have proposed that indefinite NPs are ambiguous between quantificational and cardinal readings, which are sensitive to topic structure, and cardinal readings of indefinite NPs are not generalized quantifiers but 'event-dependent' individuals. Quantificational adverbs are defined as quantifiers over events. Based on this study, I consider how proper readings of donkey sentences are derived from the semantics of indefinite NPs and the syntactic representations of topic Section 2 concerns multiple proportional readings of donkey sentences and sensitivity to topic structure. In section 3, previous analyses will be critically reviewed. Finally, in section 4, I will summarize my previous study, and show how this work leads to an appropriate account for the current issue. 2. Multiple Proportional Readings of Donkey Sentences The proportional reading of a relative-clause donkey sentence as in (1) has been first pointed out by Partee (1984). 197 (1) Most farmers who own a donkey beat it. Partee argues that the correct truth condition of (1) concerns the proportion of donkey-owning farmers as to donkey-beating farmers. Inspired by Partee, Kadmon (1987) observes …

Content truncated to 500 words