Technical reports

The semantics of noun phrase anaphora

David Alan Howard Elworthy

February 1993, 191 pages

This technical report is based on a dissertation submitted February 1993 by the author for the degree of Doctor of Philosophy to the University of Cambridge, Darwin College.

DOI: 10.48456/tr-289

Abstract

Anaphora is a linguistic phenomenon in which one expression, called an anaphor, gains some or all of its meaning from another, its antecedent. In this thesis, I study the semantics of one particular sort of anaphor, where both antecedent and anaphor are noun phrases. Most research in the past has dealt with singular anaphora; I also address plurals.

The two major theories of anaphora are Kamp’s Discourse Representation Theory (DRT) and dynamic logics. While they have yielded many valuable insights into the phenomenon, I think it is time to subject them to some critical scrutiny. There are two main criticisms. Firstly, the interpretation assigned to the lingustic data is not always consistent with language users’ intuitions about it. Secondly the current theories employ semantic formalisms which rely on either specific representational devices or on unconventional logics. I develop a new theory, TAI (Theory of Anaphoric Information), which attempts to rectify both problems.

This thesis starts with a critical re-examination of the linguistic data, and in particular of the so-called “donkey sentences”, which exhibit complex interactions between quantification and anaphora. The following chapter examines DRT and dynamic logics in some detail, considering their successes and failings from both empirical and methodological perspectives.

TAI itself is presented in chapter 4. The theory starts from a conceptual model, which specifies the information needed to interpret anaphors correctly. A logic, L(GQA), is then developed, which derives both truth conditions and constraints on the anaphoric information from formulae derived from natural language sentences. The logic is static and does not rely on structured representations of the sort found in DRT. The translation procedure from linguistic input to L(GQA) formulae captures a significant part of the emprirical weight of the theory, and provides sufficient flexibility to make the required range of readings available.

The last chapter evaluates TAI from a variety of standpoints. The conceptual model is used as a baseline for comparing DRT, dynamic logics and TAI. The relation between semantic logics of TAI and pragmatic aspects of interpreting anaphors is considered. Computational aspects of TAI are also examined: how it relates to Webber’s theory of anaphora, and how the logic could be implemented efficiently. Finally, some directions in which research based on TAI could proceed are identified.

Full text

PDF (13.9 MB)

BibTeX record

@TechReport{UCAM-CL-TR-289,
  author =	 {Elworthy, David Alan Howard},
  title = 	 {{The semantics of noun phrase anaphora}},
  year = 	 1993,
  month = 	 feb,
  url = 	 {https://www.cl.cam.ac.uk/techreports/UCAM-CL-TR-289.pdf},
  institution =  {University of Cambridge, Computer Laboratory},
  doi = 	 {10.48456/tr-289},
  number = 	 {UCAM-CL-TR-289}
}