LINGUIST List 6.1595

Sat Nov 11 1995

Calls: Southeastern conf on ling, Word sense disambiguation

Editor for this issue: Ann Dizdar <>


  1. Martin Matthew Jacobsen, Secol Conference
  2. Nancy Ide, Call: Word Sense Disambiguation

Message 1: Secol Conference

Date: Sat, 11 Nov 1995 00:45:26 Secol Conference
From: Martin Matthew Jacobsen <mmj3293TAM2000.TAMU.EDU>
Subject: Secol Conference


March 15-16, 1996
College Station, TX

Plenary speakers: Paul Hopper, Elizabeth Traugott

Abstracts are invited for three special sessions, as well as for
general sessions in all areas of linguistics and stylistics. Send 6
copies of a 300-word abstract for a paper or a poster to the
appropriate address below (hard copies only, please). People whose
papers are accepted must be or become SECOL members.


On-line resources in linguistics. Abstracts to Anthony Aristar,
Department of English, Texas A&M University, College Station, TX
77843-4227 by Nov. 1, 1995. E-mail

Ethnography of communication. Abstracts to Kathleen Ferrara,
Department of English, Texas A&M University, College Station, TX
77843-4227 by Nov. 1, 1995. E-mail

Etymology. Abstracts to J. Lawrence Mitchell, Department of
English, Texas A&M University, College Station, TX 77843-4227
by Nov. 1, 1995. E-mail


Abstracts to Southeastern Conference on Linguistics, Department of
English, University of Memphis, Memphis, TN 38152 by Dec. 5, 1995.
For guidelines, contact Joan Weatherly [Phone (901)678-4582; E-mail] or Marvin Ching [Phone (901)678-4520;
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: Call: Word Sense Disambiguation

Date: Sat, 11 Nov 1995 12:09:50 Call: Word Sense Disambiguation
From: Nancy Ide <>
Subject: Call: Word Sense Disambiguation


 Special issue

 Guest Editors
 Nancy Ide <>
 Jean Veronis <>

The discrimination of word senses, word sense disambiguation (WSD), is
of prime importance for all areas involving computerized language
analysis, including corpus-based research, lexical studies,
information retrieval, machine translation, natural language
processing, studies of style and theme, authorship attribution, and
applications such as hypertext browsing.

As early as the late 1950's, WSD was recognized as a critical but
extremely difficult task for automated language analysis. The
intractability of the problem contributed to the abandoning of machine
translation research programs in the early 1960's, when one of the
pioneers in the field, Bar-Hillel, proclaimed that he could see no way
computer programs could ever determine, for example, the proper sense
of the word "pen" in the sentence "The box is in the
pen". Nevertheless, in the past 30 years researchers have continued to
work on the WSD problem, and although the problem is far from solved,
considerable progress has been made.

This special issue is intended to summarize the state of the art in
WSD, identify the main areas of difficulty, and suggest solutions for
improvement. Papers are invited on all areas of WSD, and especially on

 o large-scale knowledge sources for WSD. Given the difficulty of
 building large-scale knowledge sources by hand, researchers have
 turned to "ready-made" resources such as machine-readable
 dictionaries and corpora. However, each of these kinds of resources
 poses problems, and none covers definitively the areas required to
 accomplish WSD.

 o the problem of sense division. The disambiguation of word senses
 involves, a priori, determining what the appropriate sense
 distinctions for a given word are. Many studies have shown that the
 kinds of sense distinctions made by most everyday dictionaries (and
 even some computer dictionary resources such as WordNet) are too
 finely-grained, and in some cases not even appropriate, to serve the
 purposes of language analysis.

 o combination of methods (statistical, rule-based) and knowledge
 sources (associative, collocational, phrasal, morphosyntactic,
 statistical, domain-related, etc.). It is now widely held that no
 single approach is complete enough for WSD, and that a combination
 of sources and methods is required. However, it remains to be
 determined how to most effectively combine methods and knowledge for

 o assessment of the knowledge "needs" for WSD. There is no precise
 quantitative study on what the knowledge "needs" are in order to
 disambiguate a given word in a given context. For example, in the
 sentences, "I write on the page", "I spoke to the page", "The front
 page says..." it is obvious that different kinds of knowledge
 contribute more or less to the proper (and/or easy) disambiguation
 of "page" in the various contexts--i.e., the strong association
 between "write" and "page", selectional restrictions on the verb
 "speak", and collocational information for "front" and "page",

All papers will be peer reviewed. Priority will be given to papers
that have a strong empirical background and report precise,
quantitative results.

SUBMISSIONS Hard copy submissions should be on letter-size paper (8.5
x 11 inches, or A4), double-spaced throughout, including footnotes and
references. The paper should begin with an informative abstract of
150-250 words. Manuscripts must be written in English. Six copies
should be sent to

 Julia Hirschberg, CL Editor
 AT&T Bell Laboratories
 600 Mountain Avenue, 2D-450
 Murray Hill, NJ 07974, USA
 (+1 908) 582-7496;

Submissions in electronic form (LaTeX) must conform to the
Computational Linguistics specifications, which can retrieved from

Electronic submissions should be sent to <>.
All submissions must be received before April 1, 1996.

Computational Linguistics is published quarterly by the MIT Press for
the Association for Computational Linguistics.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue