Coreference processing in L1 and L2

Cho, Hee Youn

Permalink

https://hdl.handle.net/2142/18487 Copy

Description

Title

Coreference processing in L1 and L2

Author(s)

Cho, Hee Youn

Issue Date

2011-01-21T22:42:32Z

Director of Research (if dissertation) or Advisor (if thesis)

Christianson, Kiel

Doctoral Committee Chair(s)

Christianson, Kiel

Committee Member(s)

• Yoon, Hye Suk James
• Shih, Chilin
• Ionin, Tania

Department of Study

Linguistics

Discipline

Linguistics

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Coreference
• Anaphor
• Accessibility
• Subject
• Referential ambiguity
• English
• Korean
• Second language

Abstract

In order to build a coherent representation of a discourse, the language processor must be able to resolve coreference relations between an anaphor and its antecedent. Theoretical and empirical evidence has suggested that discourse entities differ in their relative accessibility, and that this guides the choice of anaphors and identification of their referent (e.g., Ariel, 1990; Walker, Joshi, & Prince, 1998). According to the accessibility account, there is an inverse relation between the accessibility of a discourse entity and the degree of reduction of an anaphor used to refer to the entity. In seven self-paced reading experiments, this dissertation investigated the determinants of accessibility, the way accessibility interacts with referential ambiguity of anaphors, and the extent to which coreference processing in L2 is guided by the same factors that operate in L1 processing. Experiments 1 and 2 (English) tested the prediction that an entity in the subject position is more accessible than an entity in the object position as measured by the relative ease of comprehending temporarily ambiguous pronouns disambiguated to the subject and the relative ease of establishing coreference using pronouns compared to repeated names for a subject antecedent. Another goal of Experiments 1 and 2 was to investigate whether the function of pronouns as a marker for the most accessible antecedent is influenced by their referential ambiguity. The results revealed a clear preference for interpreting ambiguous pronouns as referring to the subject. However, name anaphors did not result in additional reading time costs compared to ambiguous pronouns even in the subject antecedent condition. Given delayed disambiguation (Experiment 2), there was an overall preference for repeated names. These results suggest that the processing of anaphors is influenced by the relative accessibility of discourse entities, but the ability of pronouns to coherently refer to accessible antecedents may be greatly modulated by the need to uniquely identify the referent of an anaphor. Experiments 3 through 5 investigated the interpretation of null pronouns and repeated names as a window into how relative accessibility of discourse entities is determined in Korean. Accessibility was manipulated by varying the grammatical role (subject vs. object) and the order of mention (SOV vs. OSV) of an entity. The results consistently revealed an advantage for subject antecedents. Although not as strong as subjecthood, the results also indicated increased accessibility of scrambled objects: the on-line subject advantage was less pronounced in the OSV order when it was presented in felicitous discourse contexts, and off-line comprehension questions were answered more accurately when the anaphor had an object antecedent compared to a subject antecedent in OSV. Experiments 6 and 7 examined the extent to which coreference processing in L2 English by Korean speakers is guided by accessibility and referential ambiguity considerations. Unlike native speakers, L2 learners showed an overall preference for repeated names in comparison to grammatically ambiguous pronouns, independent of the extent of referential ambiguity and the grammatical role of the antecedent. These results suggest that processing resources of L2 learners are limited. As a consequence of this limitation, L2 language processors face a particularly difficult challenge when having to take into account the multiple sources information needed to interpret pronouns. Taken together, these results indicate that coreference processing is influenced not only by the accessibility of the antecedents but also by the referential ambiguity of anaphors and the available processing resources of the processor. Implications are discussed in relation to theories of reference focusing specifically on what underlies increased accessibility of subjects and scrambled objects, why referential ambiguity matters, and why L2 coreference processing is different from L1 processing.

Graduation Semester

2010-12

Permalink

http://hdl.handle.net/2142/18487

Copyright and License Information

Copyright 2010 Hee Youn Cho

Owning Collections