<?xml version="1.0" standalone="yes"?>
<Paper uid="W99-0110">
  <Title>Comprehension of Coreferential Expressions</Title>
  <Section position="1" start_page="0" end_page="82" type="abstr">
    <SectionTitle>
Abstract
</SectionTitle>
    <Paragraph position="0"> The ways in which the form of referring expressions interacts with the structure of language are reviewed. Evidence from a number of different methods - quantitatively analyzed judgements of acceptable coreference, reading time, and corpus frequency of different types of coreferenfial expressions - converges on a fairly simple description of patterns of coreference.- A model is presented which integrates important aspects of Discourse Representation Theory and of Centering Theory in order to provide an account of how referential expressions are interpreted as part of constructing a discourse universe from a series of utterances.</Paragraph>
    <Paragraph position="1"> Introduction The study of coreference in generative linguistics has led to a very strong emphasis on how the hierarchical structure of sentences interacts with the form of referring expressions to constrain coreference (Chomsky, 1986; Reinhart, 1976). The resulting principles, embodied in the Binding Theory, provide rules that are of some use to researchers in natural-language processing because they provide information about disjoint reference - what an expression cannot refer to in certain circumstances. However, beyond that use theoretical work on the Binding Theory does not directly bear on questions central to language processing. Questions of how to resolve potentially ambiguous expression such as pronouns, or how meaning more generally is built up incrementally from linguistic expressions in context, are beyond its scope.  We (Gordon &amp; Hendrick, 1997; 1999) have used the basic methods of experimental psychology to take a close look at the phenomena of coreference and disjoint reference involving full expressions (names and descriptions) that have been cited in support of Principle C of the Binding Theory. I The results show that the interaction of form of referring expression and language structure is far simpler than it has been taken to be in the Binding Theory. Further, results on the judged acceptability of different configurations of referential expressions are consistent with the results of experiments that use reading time as ~ano online measure of language comprehension (Gordon, Grosz &amp; Gilliom, 1993; Gordon, Hendricl~ Ledoux &amp; Yang, 1999). Further, those results are consistent with the frequency of different types of coreferentiai configurations in corpora of naturally-occurring language (Ariel, 1994; Carden, 1982; van Hock, 1997). The pattern of&amp;quot; coreference that is observed with these three types of measures (intuitive judgments, reading time, and frequency in a corpus) is accounted for by a model (Gordon &amp; Hendricl~ 1998) that incorporates aspects of Centering Theory  Reyle, 1993).</Paragraph>
    <Paragraph position="2"> * '. Principle C of the Binding Theory states that an r-expression cannot have a c-commanding antecedent. A constituent ct is said to &amp;quot;c-command another constituent 13if the first branching node that dominates a dominates 13 as well.</Paragraph>
    <Paragraph position="3">  Research in psycholinguistics, both ours and that of others, supports a fairly simple generalization concerning the ease of establishing coreference in sequences of different forms of referring expressions. Coreference is most easily established in *name-pronoun sequences, less -easily established in name-name sequences, and least easily established in pronoun-name sequences. An example of these types of sequences taken from Gordon and Hendrick (1997) is shown below along with the proportion of naive subjects (college students at the University of North Carolina) who judged that it was grammatically acceptable for the expressions in bold-face to refer to the same person.</Paragraph>
    <Paragraph position="4"> John's roommates met him at the shop. .94 John's roommates met John at the shop. .37 His roommates met John at the shop. .23 It should be noted that according to the Binding Theory coreference ought to be acceptable in all thesentences shown above; for both the name-name and pronoun-name sequence the second referring expression does not have a c-commanding antecedent and therefore should be free to corefer with the first referring expression? This is one example of consistent differences that we found between the accepted empirical predictions of the Binding Theory and the quantitatively analyzed judgments of competent native speakers who were naive to linguistic theory.</Paragraph>
    <Paragraph position="5"> In Gordon and Hcudrick (1997; 1999) we find that this pattern of relative acceptabifity between the different types of sequences is shown for categorical judgments, for ratings of grammaticality, for isolated sentences, for sentences in discourse context, and for different types of unreduced expressions including names,</Paragraph>
  </Section>
class="xml-element"></Paper>