Possession type affects resolution of possessive pronouns in English VP ellipsis

Jesse Storbeck,Elsi Kaiser

doi:10.3765/plsa.v3i1.4346

Abstract

The sentence “Bill washed his car, and John did, too” has two possible interpretations if the overt his refers to Bill: (i) a coreferential interpretation, in which John washed Bill’s car, or (ii) a bound variable interpretation, in which John washed his own car. What guides comprehenders’ selection of one over the other? Previous research has identified factors such as processing economy (e.g. Reuland, 2001) and lexical semantic properties of the verb and possessed noun (e.g. Foley et al., 2003; Ong & Brasoveanu, 2014). We extend research on the contribution of possession type to resolution of this type of ambiguous VP ellipsis. We hypothesize that the range of possession types found in natural language varies in the extent to which the possessum is processed as an independent discourse referent or as dependent on the discourse representation of its possessor. Moreover, we expect that such differences modulate the possessum’s availability for coreference and, therefore, affect ambiguity resolution. We conducted an experiment testing how different possession relations modulate adult L1 English speakers’ interpretational preference. Inanimate nouns favored bound variable interpretations more than animates did, supporting our hypothesis that the overt possession’s animacy and its resultant discourse status are important factors in the resolution of the elided possessive pronoun. Follow-up experiments confirmed these results and ruled out nouns’ real-world plausibility of possession as a determinant of interpretational preference. Our results suggest that animate possessions are more likely than inanimates to receive independent status in the discourse and consequently to be available for coreference when the ellipsis is interpreted.

Highlights

Pronoun interpretation can occur via two mechanisms: discourse-level coreference or semantic-level binding (e.g. Heim, 1982; Reuland, 2001)
We hypothesize that possessed nouns differ in the extent to which they are represented as independent discourse referents – as opposed to being dependent on the discourse representations of their possessors – and that animacy is one of a number of factors modulating this distinction
The model assumes that animate possessions are more likely to receive fully independent discourse representations, while inanimate possessions are less privileged and more likely to be conceptualized as dependent on the representations of their possessors

Summary

Introduction

Pronoun interpretation can occur via two mechanisms: discourse-level coreference or semantic-level binding (e.g. Heim, 1982; Reuland, 2001). Heim, 1982; Reuland, 2001) These mechanisms yield different interpretations of the VP ellipsis in a sentence like Example (1a). B. Billi washed hisi car, and Johnj did [VP washed hisj car], too. Bill [ λx ( x washed x’s car ) ], and John [ λx ( x washed x’s car ) ], too. C. Billi washed hisi car, and Johnj did [VP washed hisi car], too. Bill [ λx ( x washed hisBill car ) ], and John [ λx ( x washed hisBill car ) ], too. D. Billi washed hisk car, and Johnj did [VP washed hisk car], too. Bill [ λx ( x washed hisTed car ) ], and John [ λx ( x washed hisTed car ) ], too

Objectives

Methods

Results

Conclusion