MIT Libraries logoDSpace@MIT

MIT
View Item 
  • DSpace@MIT Home
  • MIT Libraries
  • MIT Theses
  • Doctoral Theses
  • View Item
  • DSpace@MIT Home
  • MIT Libraries
  • MIT Theses
  • Doctoral Theses
  • View Item
JavaScript is disabled for your browser. Some features of this site may not work without it.

Coherence in natural language : data structures and applications

Author(s)
Wolf, Florian, 1975-
Thumbnail
DownloadFull printable version (8.363Mb)
Other Contributors
Massachusetts Institute of Technology. Dept. of Brain and Cognitive Sciences.
Advisor
Edward Gibson.
Terms of use
M.I.T. theses are protected by copyright. They may be viewed from this source for any purpose, but reproduction or distribution in any format is prohibited without written permission. See provided URL for inquiries about permission. http://dspace.mit.edu/handle/1721.1/7582
Metadata
Show full item record
Abstract
(cont.) baseline, and that some coherence-based approaches best predict the human data. However, coherence-based algorithms that operate on trees did not perform as well as coherence-based algorithms that operate on more general graphs. It is suggested that that might in part be due to the fact that more general graphs are more descriptively adequate than trees for representing discourse coherence.
 
The general topic of this thesis is coherence in natural language, where coherence refers to informational relations that hold between segments of a discourse. More specifically, this thesis aims to (1) develop criteria for a descriptively adequate data structure for representing discourse coherence; (2) test the influence of coherence on psycholinguistic processes, in particular, pronoun processing; (3) test the influence of coherence on the relative saliency of discourse segments in a text. In order to address the first aim, a method was developed for hand-annotating a database of naturally occurring texts for coherence structures. The thus obtained database of coherence structures was used to test assumptions about descriptively adequate data structures for representing discourse coherence. In particular, the assumption that discourse coherence can be represented in trees was tested, and results suggest that more powerful data structures than trees are needed (labeled chain graphs, where the labels represent types of coherence relations, and an ordered array of nodes represents the temporal order of discourse segments in a text). The second aim was addressed in an on-line comprehension and an off-line production experiment. Results from both experiments suggest that only a coherence-based account predicted the full range of observed data. In that account, the observed preferences in pronoun processing are not a result of pronoun-specific mechanisms, but a byproduct of more general cognitive mechanisms that operate when establishing coherence. In order to address the third aim, layout-, word-, and coherence-based approaches to discourse segment ranking were compared to human rankings. Results suggest that word-based accounts provide a strong
 
Description
Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, February 2005.
 
Includes bibliographical references (leaves [143]-148).
 
Date issued
2005
URI
http://hdl.handle.net/1721.1/28854
Department
Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences
Publisher
Massachusetts Institute of Technology
Keywords
Brain and Cognitive Sciences.

Collections
  • Doctoral Theses

Browse

All of DSpaceCommunities & CollectionsBy Issue DateAuthorsTitlesSubjectsThis CollectionBy Issue DateAuthorsTitlesSubjects

My Account

Login

Statistics

OA StatisticsStatistics by CountryStatistics by Department
MIT Libraries
PrivacyPermissionsAccessibilityContact us
MIT
Content created by the MIT Libraries, CC BY-NC unless otherwise noted. Notify us about copyright concerns.