Lucas Champollion
Supervisors: Cleo Condoravdi and Aravind K. Joshi
Phone: +1 (212) 998-8692
Address: Lucas Champollion
NYU Department of Linguistics
10 Washington Place, room 412 (fourth floor)
New York, NY 10003
United States
Phone: +1 (212) 998-8692
Address: Lucas Champollion
NYU Department of Linguistics
10 Washington Place, room 412 (fourth floor)
New York, NY 10003
United States
less
InterestsView All (31)
Uploads
The constraints on these constructions involve concepts that are generally studied separately: aspect, plural and mass reference, measurement, and distributivity. In this talk, I will give an overview of my dissertation, which provides a unified perspective on these domains, connects them formally within the framework of mereological semantics, and uses this connection to transfer insights across unrelated bodies of literature. A generalized notion of distributivity is proposed and formalized as a parametrized higher-order property called stratified reference: a predicate that holds of a certain entity or event is required to also hold of its parts along a certain dimension and down to a certain granularity. The dimension parameter is a thematic role in the case of 'each' and 'all', a measure function in the case of pseudopartitives, and time or space in the case of 'for'-adverbials. The granularity parameter involves pure atoms in the case of 'each', pure and impure atoms in the case of 'all', and very small amounts of space, time, or matter in the cases of pseudopartitives and 'for'-adverbials. Stratified reference is used to formulate a single constraint that explains each of the judgments above. The constraint is exploited to improve on existing characterizations of distributivity, atelicity, and monotonicity of measurement.
The framework results in a new take on the minimal-parts problem that occurs in the study of atelic predicates and mass terms. It scales up successfully from temporal to spatial aspect, and it explains why pseudopartitives and other distributive constructions are sensitive to the difference between intensive and extensive measure functions. It provides a fresh view on atomic and cover-based theories of quantificational distributivity. The framework is also used to account for the scopal behavior of 'all' and of 'for'-adverbials with respect to cumulative quantification and dependent plurals. Together with a novel theory of collective predication, the framework also provides an account of the differences between such predicates as 'be numerous' and 'gather' as they interact with 'all'.
With the help of two parsed and coreference-annotated corpora, this paper estimates the impact of the split-architecture proposal. The findings of this work are as follows: (1) Subject pronouns in authentic texts behave the same way in main and subordinate clauses. (2) The number of sentences in which a split architecture would behave differently than a system that treats both cases the same way is close to zero. Therefore, a separate treatment of resolution within and across units is unlikely to improve the performance of any system. This result casts a doubt on the split-architecture proposal, and more generally on approaches that directly incorporate psycholinguistic results into performance-oriented algorithms for anaphora resolution without assessing the relative importance of the phenomena that underlie them.
The constraints on these constructions involve concepts that are generally studied separately: aspect, plural and mass reference, measurement, and distributivity. In this talk, I will give an overview of my dissertation, which provides a unified perspective on these domains, connects them formally within the framework of mereological semantics, and uses this connection to transfer insights across unrelated bodies of literature. A generalized notion of distributivity is proposed and formalized as a parametrized higher-order property called stratified reference: a predicate that holds of a certain entity or event is required to also hold of its parts along a certain dimension and down to a certain granularity. The dimension parameter is a thematic role in the case of 'each' and 'all', a measure function in the case of pseudopartitives, and time or space in the case of 'for'-adverbials. The granularity parameter involves pure atoms in the case of 'each', pure and impure atoms in the case of 'all', and very small amounts of space, time, or matter in the cases of pseudopartitives and 'for'-adverbials. Stratified reference is used to formulate a single constraint that explains each of the judgments above. The constraint is exploited to improve on existing characterizations of distributivity, atelicity, and monotonicity of measurement.
The framework results in a new take on the minimal-parts problem that occurs in the study of atelic predicates and mass terms. It scales up successfully from temporal to spatial aspect, and it explains why pseudopartitives and other distributive constructions are sensitive to the difference between intensive and extensive measure functions. It provides a fresh view on atomic and cover-based theories of quantificational distributivity. The framework is also used to account for the scopal behavior of 'all' and of 'for'-adverbials with respect to cumulative quantification and dependent plurals. Together with a novel theory of collective predication, the framework also provides an account of the differences between such predicates as 'be numerous' and 'gather' as they interact with 'all'.
With the help of two parsed and coreference-annotated corpora, this paper estimates the impact of the split-architecture proposal. The findings of this work are as follows: (1) Subject pronouns in authentic texts behave the same way in main and subordinate clauses. (2) The number of sentences in which a split architecture would behave differently than a system that treats both cases the same way is close to zero. Therefore, a separate treatment of resolution within and across units is unlikely to improve the performance of any system. This result casts a doubt on the split-architecture proposal, and more generally on approaches that directly incorporate psycholinguistic results into performance-oriented algorithms for anaphora resolution without assessing the relative importance of the phenomena that underlie them.
(1) John dialed a wrong phone number for five minutes.
But this generalization is not exceptionless. For example, in a
context in which we discuss the daily intake of patients, we can say (2) without implying that he took the same two pills repeatedly. (Champollion 2010, Deo and Pinango 2011, Del Prete 2011)
(2) The patient took two pills for a month and then went back to one pill.
This reading of (2) is distributive: the predicate "take two pills" is distributed over salient subintervals of the month. I will extend the pragmatics-based theory of distributivity from Schwarzschild (1996) to explain why a distributive reading is available in sentences like (2) but not (1).