[Prev][Next][Index]

CSLI Calendar , 27 JUNE 1996, vol.11:NO 31


   
         C S L I   C A L E N D A R   O F   P U B L I C   E V E N T S
______________________________________________________________________________

27 June 1996                      Stanford                     Vol. 11, No. 31
______________________________________________________________________________

      A weekly publication of the Center for the Study of Language and
Information (CSLI), Stanford University, Ventura Hall, Stanford, CA 94305-4115
                               ____________

                CSLI ACTIVITIES DURING 27 JUNE -- 31 JULY 1996

FRIDAY, 28 JUNE
         1:15 - Semantic Workshop
                Cordura Hall, Room 100
                Underspecified Representations and a Theory of
                Utterance Processing 
                Massimo Poesio
                Abstract below

MONDAY, 1 JULY
        11:00 - Machine Translation Seminar
                Cordura Hall, Room 100
                Composition of Transfer Rules in a Multi-Lingual MT System
                Manny Rayner, SRI International
                Abstract below

         2:00 - Machine Translation Seminar 
                Cordura Hall, Room 100
                Statistical Machine Translation
                Dekai Wu 

WEDNESDAY, 17 JULY
         1:30 - Special Syntax Seminar
                Ventura Hall, Room 17
                On the Nature of Compound Verbal Noun Constructions in
                Japanese
                Yo Matsumoto

                NOTE: This is a special issue of the CSLI
                Calendar for the 1995-96 academic year.
                The first issue for 1996-97 will appear 
                towards the end of September 1996.

                               ____________

The CSLI Calendar appears weekly on Wednesdays throughout the academic year.
Announcements, abstracts, and other information to appear in the Calendar can
be submitted to [mailto:incalendar@csli.stanford.edu].

Information about CSLI's research program and past issues of the CSLI Calendar
are available at [http://www-csli.stanford.edu/csli/].  The CSLI Calendar is
also posted each week to [news://nntp-csli.stanford.edu/csli.bboard].

                               ____________


 ***********************************************************************
                             Psychology 118: 
                        PSYCHOLOGY AND TECHNOLOGY
 Instructors: Mija Van Der Wege (mija@psych) and Jeff Zacks (zacks@psych)
                   MW 9:00-10:50, Jordan Hall, room 245
 ***********************************************************************

Psychology has played an important role in the design of airplanes, computers,
telephones, cars*all the technologies that have changed the shape of our
world.  This course provides a grounding in the role cognitive, social, and
perceptual factors play in the design and evaluation of technology.  At the
same time, it explores the question: How do the tools we use change the way we
think?       

Course Overview: Topics

PART 1: PSYCHOLOGY AFFECTS TECHNOLOGY
The trouble with computers
Psychological bases 
Roots: ergonomics, human factors, HCI and CHI
The cognitive modeling approach
The design principles approach
The iterative design and testing approach

PART 2: TECHNOLOGY AFFECTS PSYCHOLOGY
Origins of Cognitive Psychology: The Computer Metaphor
Identity Questions: Minds and Programs
Identity and Community
Social Psychology of Telecommunications
Psycholinguistics of Telecommunications
Evolution

                             _______________

                            SEMANTIC WORKSHOP
                            on Friday, 28 June
                    1:15 p.m., Cordura Hall, Room 100
   Underspecified Representations and a Theory of Utterance Processing
                              Massimo Poesio
                         University of Edinburgh

Underspecified representations have been developed as a way to improve
the performance of NLP systems, and have been mostly studied from a
logical and computational point of view. In the talk I am going to
propose that underspecified representations can be used to obtain a
theory of the semantics / pragmatics interface that is both
linguistically motivated and cognitively plausible. I will first of
all discuss what kind of meaning can be assigned to ambiguous
expressions, and review some evidence concerning ambiguity
processing. I will then propose a theory of underspecification and
disambiguation with underspecified representations that integrates the
central ideas of most current theories of underspecification, and in
addition introduces three novel ideas: (i) underspecified
representations are best interpreted as partial characterization of
what has been said, rather than of the content of utterances; (ii)
theories of nonmonotonic reasoning play a fundamental role not only in
explaining some of the characteristics of ambiguity processing, but
also to account for the notion of precisification that is central to
most people's idea of ambiguity; and (iii) in a theory of
underspecification consistent with (i) and (ii), there is no need to
introduce a special semantics to characterize the meaning of ambiguous
expressions.
                             _______________


There will be two Machine Translation seminars on July 1st.  
I think both of them will be of interest to people who aren't into MT at the
moment, but want to know the state-of-the-art in research.  In
particular, Manny Rayner's talk might be of interest because it's about
a speech-to-speech translation system.  (The title may look a bit
specialized, but he said he'd start with a general overview.)

                        MACHINE TRANSLATION SEMINAR
                            on Monday, 1 July
                    11:00 a.m., Cordura Hall, Room 100
        Composition of Transfer Rules in a Multi-Lingual MT System
                     Manny Rayner, SRI International

There is a well-known argument in the machine-translation literature
that goes something like this. Suppose that we wish to build a
transfer-based multilingual machine translation system, which will
translate from any one of N languages into any other. If the system is
transfer-based, we will need N(N-1) sets of transfer rules, and as
soon as N becomes large the effort needed to write these rules will be
intolerable. Hence the indicated solution is introduce an interlingua,
and only write rules to transfer each language to and from the
interlingua.

In this seminar, I will describe some recent work carried out within
the Spoken Language Translator (SLT) project, which explores an
alternative approach. We retain the normal transfer paradigm, but
exploit the highly declarative nature of the SLT transfer formalism to
construct new sets of transfer rules by automatically composing
existing ones. Thus given three languages A, B and C, we compose
transfer rule sets for A -> B and B -> C to get a set for A -> C.
Although we have found that the composed set of rules for A -> C is in
practice not quite good enough to use as it stands, it is far easier to
improve it into a useful set than it would be to write a new set of
rules from scratch.

The seminar will cover the following topics:

- An overview of the SLT system and project.

- The transfer rule composition operation, which is based on standard
program-transformation ideas taken from logic programming.

- Results of concrete experiments in transfer rule composition,
carried out using the language-triples Swedish -> English -> French
and English -> French -> Spanish. In both cases, the domain was Air
Travel Inquiries (ATIS), with a vocabulary of about 1500 words in each
language.

(Joint work with Ivan Bretan and Mats Wirén, Telia Research,
Sweden).
                             _______________


                       MACHINE TRANSLATION SEMINAR
                            on Monday, 1 July
                    2:00 p.m., Cordura Hall, Room 100
                     Statistical Machine Translation
                                 Dekai Wu
              Hong Kong University of Science and Technology

                             _______________


                          SPECIAL SYNTAX SEMINAR
                          on Wednesday, 17 July
                     1:30 p.m., Ventura Hall, Room 17
           On the Nature of Compound Verbal Noun Constructions 
                               in Japanese
                               Yo Matsumoto
                          Meijigakuin University
                     [yomatsum@mh.meijigakuin.ac.jp]

                             _______________