LINGUIST List 3.864

Fri 06 Nov 1992

Qs: Survey of Computational Linguistics Courses

Editor for this issue: <>


Directory

  1. Bonnie J. Dorr, Survey of Computational Linguistics Courses

Message 1: Survey of Computational Linguistics Courses

Date: Thu, 5 Nov 92 06:31:18 -0500
From: Bonnie J. Dorr <bonnieumiacs.UMD.EDU>
Subject: Survey of Computational Linguistics Courses
We recently sent out a surface mailing of the Survey of Computational
Linguistics courses for the Association for Computational Linguistics.
Much to my dismay, I just discovered that some of the questionnaires
were collated and stapled incorrectly. (In the incorrect versions,
the order of pages in the stapled copy was 214365 instead of 123456.)
I apologize for this error. Here is the on-line version of the
survey:
[----------------------------------------------------------------------
 To: Instructors of Classes in Computational Linguistics
As a follow-on to the Directory of Graduate Programs recently compiled
by Martha Evens, the Association for Computational Linguistics will
publish a new edition of the Survey of Computational Linguistics
Courses. This is a revised version of the 1986 survey, published in
Computational Linguistics (vol. 12) by Robin Cohen, which is intended
to be a systematic compilation of syllabi from individual classes that
teach computational linguistics (i.e., it is not an enumeration of
classes taught in CL programs as in the Directory of Graduate Programs).
The new version of the survey will be published in the Computational
Linguistics journal in 1993.
We are eager to include two types of classes: those that teach
computational linguistics as the sole topic and those that teach
computational linguistics as one of several topics. The survey will
allow us to share with colleagues ideas on how to teach computational
linguistics. It will also provide an idea of how the field of
computational linguistics is being portrayed to potential new
researchers.
Our listing will include the name and address of the University and
Department(s) offering the class, the name and number of the class,
the type of class, and information about the syllabus (e.g., topics,
texts used, software used, format, workload, enrollment, duration,
frequency, and assistance). In addition it will include some
statistics on the responses (i.e., total number of classes having
particular characteristics) and a bibliography of the most frequently
cited references.
We would appreciate your response to the survey as soon as possible.
The intention is to complete the report by early 1993. You may send
information electronically to: cl-surveyumiacs.umd.edu
or by mail: Ms. Sandy Tsue
	 CL-SURVEY
 Institute for Advanced Computer Studies
	 A.V. Williams Building
	 University of Maryland
	 College Park, MD 20742
 USA
(Electronic transmission is preferred.)
Thank you very much for your time.
Sincerely,
Bonnie J. Dorr
[----------------------------------------------------------------------
SURVEY OF COMPUTATIONAL LINGUISTICS COURSES
INSTRUCTIONS
This survey, originally designed by Robin Cohen, and now revised by
Bonnie Dorr, attempts to gather information as painlessly as possible
for both respondents and surveyor. The primary change that has been
made since this survey was last conducted is that it now incorporates
questions about different types of software that people use in their
CL courses. Brief answers are solicited to the questions that follow.
For multiple choice questions, simply type in the appropriate
answer(s) from the list. For open-ended questions, use sentence
fragments separated by semicolons. For illustration, a sample
response is provided after the questionnaire. If possible, return
your response electronically as one file of question numbers and
answers, using the message header "ACL Survey Response." Hardcopy
is acceptable as well.
Feel free to include a copy of course description handouts, if
available, for filing with the ACL. Regrettably, we will not be able
to transcribe syllabi for the journal survey, but these could serve as
the basis for a more extended treatment later.
Note: If you have taught your course a number of times, respond
according to the latest version of the course. In Q12, you may allude
to topics covered and techniques used in previous versions.
[ ----------------------------------------------------------------------
CONTROL INFORMATION
Name:
Department:
Institution:
Address:
Net address:
Name and number of course:
LEVEL
A. Is the course
 - undergraduate?
 - graduate?
 - cross-listed, undergraduate and graduate?
CL STATUS
B. In any given year that you teach this course, is it a course in
computational linguistics (CL) or is CL just one of several topics covered?
 - only CL
 - topics other than CL
Note: Respond to the remainder of the questionnaire with respect to the
computational linguistics part of the course, only.
ACL SURVEY QUESTIONS
TOPICS
Q1. What topics are covered in the course?
 Suggestion: list the topics, one per line, in the order they are
 addressed in the course. Use brief summaries of the topic name,
 followed by names of researchers used as references for each topic in
 parentheses, where appropriate e.g., lexical acquisition and
 use of corpora (Boguraev & Briscoe, 1987; Zernik, 1990).
REFERENCES
Q2. What kinds of reference materials are used in the course?
 - course text(s); specify author & name (publisher & year, if known)
 - recommended reading texts; specify as above
 - papers on various topics; specify if different from those in Q1
Note: If your texts are not written in English, please provide a brief
English translation of the title.
FORMAT
Q3. What format of teaching is used?
 - formal lectures
 - paper presentations by professor
 - paper presentations by students
 - guest lectures
 - open discussions
 - other; please specify
WORKLOAD
Q4. What workload/method of assessment is used in the course?
 - midterm exam(s)
 - final exam
 - assignments with programming
 - assignments without programming?
 - course project
 - class presentations
 - other; please specify
Note: You may provide further details--e.g., midterm exam: take-home, no
programming; class presentations: including one on project topic.
SOFTWARE WRITTEN BY STUDENTS
Q5. Which of the following do students write?
 - parsers
 - generators
 - morphological processors
 - tagging programs
 - discourse processors
 - language translators
 - ATNs
 - semantic network processors
 - lexical-semantic processors
 - knowledge-based processors
 - parallel processors
 - connectionist processors
 - statistical processors
 - other; please specify
 - none
SOFTWARE PROVIDED TO STUDENTS
Q6. Which of the following are provided for students?
 (Label them C if commercial source, N if non-commercial outside
 source, H if built in house; also specify type, e.g., unification
 parser, chart parser, etc.)
 - parsers
 - generators
 - morphological processors
 - tagging programs
 - discourse processors
 - language translators
 - ATNs
 - semantic network processors
 - lexical-semantic processors
 - knowledge-based processors
 - parallel processors
 - connectionist processors
 - statistical processors
 - grammars
 - lexicons
 - corpora
 - semantic networks
 - data bases
 - knowledge bases
 - other; please specify
 - none
SOFTWARE AVAILABLE FOR DISTRIBUTION
Q7. If you've developed software and distribute it, please specify
 type (e.g., grammars, lexicons, parsers, generators, etc.) and
 mode (e.g., by diskette, tape, cdrom, anonymous ftp (give
 addresses), etc.).
ENROLLMENT
Q8. About how many people take the course each time it is offered?
DURATION
Q9. Does the course last one term or a full year?
 - one term
 - full year
FREQUENCY
Q10. How often is the course offered?
 - more than once per year
 - once per year
 - every other year
 - occasionally
ASSISTANCE
Q11. Are there teaching assistants/markers assigned to the course?
 - yes; conduct tutorial sessions
 - yes; only mark and hold office hours
 - no
 - other; please specify
COMMENTS
Q12. Include here any other comments about the course--what you would have
liked to do ideally, what you plan for the future, what lessons you've
learned.
[ ------------------------- Sample response ------------------------
Name: Bonnie J. Dorr
Department: Computer Science
Institution: University of Maryland
Address: A.V. Williams Building, College Park, MD 20742
Net Address: bonniecs.umd.edu
Course: Computational Linguistics (CMSC 828)
Level(A): graduate, cross-listed in linguistics
Status(B): CL only
Q1. TOPICS
The course is divided into 3 sections: (1) Syntax, (2) Lexicon,
and (3) Machine Translation and Generation. The course outline is
as follows:
- Introduction to CL: techniques and analytical tools for comparing
 linguistic theories and comparing computational practice; levels of
 representation and their importance in a computational theory.
- Syntax: introduction to basic concepts and theories; bottom-up
 vs. top-down parsing; Earley algorithm; Tomita's algorithm; ATN's.
 (Sells, 1985; Kay in "Readings," 1986; Earley, 1970; Tomita, 1987;
 Woods, 1970; Kaplan, 1972.)
- Contemporary syntactic models: government-binding theory;
 principle-based parsing; deterministic parsing; unification-based
 approaches; GPSG; LFG. (Berwick & Fong, 1990; Marcus in "Readings,"
 1986; Berwick & Weinberg, 1984; Sells, 1985; Shieber, 1986; Kaplan
 & Bresnan, 1982.)
- Lexicon: lexical representations; semantic roles; primitives;
 case frames; thematic relations; predicate-argument structure;
 conceptual dependencies; spatial fields; nonspatial fields;
 compositionality; linking routines. (Gruber, 1967; Fillmore, 1968;
 Jackendoff, 1972, 1983; Schank, 1973; Dorr, 1991.)
- Lexical Acquisition: non-representational vs. representational
 approaches; statistical methods; use of corpora; acquisition of
 syntactic information; bootstrapping semantics from syntax. (Boguraev
 & Briscoe, 1987; Klavans & Tzoukermann, 1990; Church & Hanks, 1990;
 Zernik, 1990; Pustejovsky, 1987; Hindle, 1990; Brent, 1991.)
- Machine translation and generation: basic theory and technology;
 major characteristics and tradeoffs; mapping techniques; problems
 in MT and generation; divergences; mismatches; lexical selection;
 syntactic realization; parameterization. (Slocum, 1988; Thurmair, 1990;
 Kaplan et al., 1989; Abeille et al., 1990; Arnold & Sadler, 1990;
 Lindop & Tsujii, 1991; Barnett et al., 1991; Dorr, 1991.)
Q2. REFERENCES
No required texts. Papers on various topics (see Q1) are assigned on
a weekly basis. Some reading texts are recommended:
- Allen, J., Natural Language Understanding, Benjamin-Cummings,
 New York, NY, 1987.
- Berwick, R., and Weinberg, A., The Grammatical Basis of
 Linguistic Performance, MIT Press, Cambridge, MA, 1984.
- Grosz, B., K. Sparck-Jones, and B. Webber, Readings in Natural
 Language Processing, Morgan-Kaufman, Los Altos, CA, 1986.
- Hutchins, W., Machine Translation: Past, Present, Future,
 Ellis Horwood, Chichester, England, 1986.
- Sells, P., Lectures on Contemporary Syntactic Theories,
 University of Chicago Press, CLSI, Chicago, IL, 1985.
- Zernik, U. (ed.), Lexical Acquisition: Using on-line Resources
 to Build a Lexicon, Lawrence Erlbaum, Hillsdale, NJ, 1987.
Q3. FORMAT
- formal lectures
- open discussions
- paper presentations by students
- two guest lectures
Q4. WORKLOAD
- 2-4 assigned readings per week
- One written assignment
- One programming lab
- Class presentation
- Final term paper
Q5. SOFTWARE WRITTEN BY STUDENTS
- simple generator and lexical-semantic processor for a language
 translation program
- small grammar and lexicon
Q6. SOFTWARE PROVIDED TO STUDENTS
- parser (N; Marcus Parser)
- grammar (N; Marcus style rules)
Q7. SOFTWARE AVAILABLE FOR DISTRIBUTION
- none
Q8. ENROLLMENT
- 20
Q9. DURATION
- one semester (14 weeks, 2-1/2 hours/week) with one semester
 follow-on taught in the linguistics department (LING 819).
Q10. FREQUENCY
- every other year
Q11: ASSISTANCE
- none.
Q12: COMMENTS
- The goal of the course is to introduce topics, issues, and theories
 in computational linguistics, to relate the field to linguistics
 and AI, and to provide the background necessary for analysis and
 evaluation of computational models of natural language understanding,
 generation, and translation.
- The distribution of students is approximately 60% computer science,
 30% linguistics, and 10% engineering.
- Students work in groups of 4 for the laboratory. The goal is to
 allow linguists and computer scientists to be able to benefit from
 each other's knowledge while also giving students a chance to have
 hands-on experience with some of the concepts that are presented
 in class at a more abstract level. A minimal amount of programming
 in Lisp is expected.
- Sample topics chosen by the students for final papers include
 syntactic models of parsing, logic programming approaches to
 generation, connectionist processing models, speech processing
 models, knowledge representation, lexical semantics, and machine
 translation. Students are expected to compare and critique
 different approaches to natural language processing and to
 give a 20 minute in-class presentation of their paper.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue