7.0175 Robustness Conference (1/286)
Elaine Brennan (EDITORS@BROWNVM.BITNET)
Fri, 10 Sep 1993 16:00:24 EDT
Humanist Discussion Group, Vol. 7, No. 0175. Friday, 10 Sep 1993.
Date: Fri, 10 Sep 1993 15:20:13 +0200
From: Onderzoeks Instituut Taal en Spraak <ots@let.ruu.nl>
Subject: Robustness Conference
PRELIMINARY PROGRAMME, September 1993
***************************************************************
THE ROBUSTNESS OF THE LANGUAGE FACULTY:
COPING WITH INCOMPLETE INFORMATION
Organized by the Research Institute for Language and Speech (OTS)
on the occasion of its 5th anniversary
28-30 October 1993
Theme Description
The human language faculty shows a remarkable robustness with respect
to incomplete information. Many possible features are not realized in the
signal of a normal linguistic utterance; and on the meaning side too, the
interpretation is highly underdetermined by the expression itself. Yet, in
the normal case, understanding is not in any way hampered by this. It
may well be that the availability of knowledge from other cognitive
domains contributes to the overall success, or perhaps this robustness is
caused by the modular structure of the linguistic system itself. Information
that disappears from one module would be compensated for via another
module. The central question of this conference will be how this
robustness of the language system can be explained, focusing in particular
on the role of non-linguistic information and higher order redundancy.
Preliminary programme
Thursday, October 28 1993
10.00 Welcome, S. Nooteboom (director OTS, Utrecht)
10.15-11.30 Theme: Language Acquisition
How is it that the child is able to extract the
necessary information from an incomplete
analysis of language data to proceed successfully
in the acquisition process?
Speaker: K. Wexler (MIT)
Comments: J. Weissenborn (MPI, Nijmegen)
11.30-12.00 Coffee
12.00-13.15 Theme: Aphasia
Aphasia can be characterized by the existence of
blockades in the transmittance of information
between various cognitive domains. What kind of
strategies are used to avoid these blockades?
What role does the modular structure of the
language faculty play here?
Speaker: G. Dogil (Stuttgart)
Comments: L. Blomert (MPI, Nijmegen)
13.15-14.30 Lunch
14.30-15.45 Theme: Language Processing
Various cognitive domains, partly linguistic,
partly also extra-linguistic, are involved in the
interpretation of language utterances. It is often
assumed that these various processes operate in
parallel fashion. With such a parallel form of
processing interpretation can be the result of
different independently operating subprocesses;
that is to say, information within one module is
not accessible to the other module. This
presupposes that processes act as filters on each
other's output. A feasible alternative is that
modules are not informationally encapsulated and
that processes do affect the operation of one
another. The discussion will address the question
in how far models of human language processing
shed light on the optimalisation of processing
modals of machine languages.
Speaker: M. Tanenhaus (Rochester)
Comments: L. des Tombe & S. Krauwer (Utrecht)
15.45-16.15 Tea
16.15-17.30 Theme: Sign Language
The development and acquisition of sign
language is a typical example of the robustness
of human linguistic competence. The canonical
medium is not available and its role is taken over
by another medium with principally other
constraints. This also means that the language
faculty is not bothered by such constraints in its
operation. The obvious question is what
compensatory mechanisms are at work, and how
the activities of the various modules are
influenced by the alternative medium.
Speaker: D. Perlmutter (San Diego, UCSD)
Comments: A. Mills (Amsterdam)
18.00- Reception
Friday October 29 1993
Session: Computational Linguistics and Logic Session
Theme: Partial information
In computational studies of linguistic competence the notion of
"information" is an important one, in particular partialness of
linguistic information, and the dynamics of reasoning with respect
to this partial information. In feature grammars unification is the
central operation for combining partially specified linguistic
descriptions. In categorial frameworks unification is enriched by a
dimension of type inference. The general research goal in these
computational models of linguistic competence is the
characterization of what one might call "linguistic inference": a
theory of reasoning by means of linguistic objects. It has proven
fruitful not to separate linguistic inference from other modes of
inference which play a role in broader cognitive frameworks.
Linguistic inference is sensitive to the structure of linguistic
constructs in the form and meaning dimensions. Current
developments in logic offer an excellent starting-point for the
systematic study of such a structure-sensitive inference.
9.00-12.30 Invited speakers are: I. Sag (Stanford), R. Kempson (SOAS,
London), H. Verkuyl (Utrecht), J. van Benthem (Amsterdam)
12.30-14.00 Lunch
Session: Phonology
Theme: Overdetermination and underspecification in phonology
In phonology the problem of robustness takes the form of
constraints on the distance between lexical and phonetic
representations. Such constraints can be sought in lexical
representations (the theory of underspecification), or in rule
application (theory of lexical phonology). In addition, the
robustness of lexical representation can perhaps also be explained
by means of overdetermination in prosodic structure (syllable
structure, metrical structure). Central research questions associated
with robustness are the following. Are lexical phonological
representations maximally underspecified, or only partially? What
is the role of prosodic structure in the characterization of lexical
representation? Which constraints are the null positions in
phonological representation (empty syllables, onsets) subject to?
14.00-17.30 Invited speakers are: P. Kiparsky (Stanford), R. Kager
(Utrecht), K.P. Mohanan (Singapore), S. Anderson (Johns
Hopkins)
Evening lecture
20.00-21.00 D. Lightfoot (Maryland)
21.00- Drinks
Saturday, October 30
Session: Phonetics
Theme: Phonetic Underspecification
In natural speech many features of speech sounds are not always
realized. In producing artificial speech it can be observed that
realization of all features leads to an unnatural result. It seems
that the human perceptual mechanism is so specifically designed
for under-information that is caused by natural sloppiness that
complete information is perceived as over-information. In light of
this fact the question arises what the rules are for realizing the
features of speech sounds. What are the thresholds of tolerance,
and what are these determined by? What is acceptable sloppiness?
To what extent are these determined by the nature of the
production and perception processes, and to what extent by formal
and substantive properties of the utterance itself? Are there
parallels to be drawn with demands for underspecification in
syntax?
9.00-12.30 Invited speakers are: B. Lindblom (Austin), C. Darwin
(Sussex), L. Pols (Amsterdam), S. Anderson (Johns Hopkins)
12.30-14.00 Lunch
Session: Syntax and Semantics
Theme: Contextual vs Grammatical Conditions on Interpretation
In addition to elements which can independently have a referential
function (such as common nouns and pronominals) the language
system also has elements which do not have such a function.
Anaphors form a large group of the latter class, their
interpretation being dependent on other elements. In principle,
anaphors are underspecified for one or more grammatical
properties. A fundamental question in this regard is what is the
connection between the syntactic/lexical property of the absence of
features, and the semantic property that independent interpretation
is impossible. Under strict conditions on the non- linguistic
context, however, certain types of anaphors can get an
interpretation without having a linguistic antecedent. These then
occur as logophors. This leads to two fundamental questions: 1.
What intrinsic properties of anaphors determine their ability to
show up as logophors? 2. How can contextual information show
interaction with strictly grammatical conditions on interpretation?
14.00-17.30 Invited speakers are: T. Reinhart (Tel Aviv/Utrecht), M.
Diesing (Cornell), J. Huang (Irvine), D. Delfitto (Utrecht)
Organization
Peter Coopmans Research Institute for Language & Speech
Martin Everaert Utrecht University
Eric Reuland Trans 10, 3512 JK Utrecht
Wim Zonneveld tel:+31-30-536006, fax:+31-30-536000, e-mail:ots@let.ruu.nl
***************************************************************
Conference sites:
- Academy Building
Domplein, Utrecht
- CSB Building
Kromme Nieuwe Gracht 39, Utrecht
***************************************************************
Accommodation:
The organisation will not take care of hotel accommodation. Please
contact the VVV Tourist Information Office:
Utrecht VVV Tourist Information Office
address: VVV Utrecht
Vredenburg 90
Postbus 19107
3501 DC Utrecht
Holland
tel.: +31-6-34034085
fax: +31-30-331417
***************************************************************
Registration
Fee for registration before October 1 / after October 1
Employed: Dfl 60 Dfl 90
Unemployed/student: Dfl 40 Dfl 60
Payment: All payments must be made in Dutch guilders.
***************************************************************
- You can transfer the appropriate amount to our bank account:
Coopmans en/of Buenen, Inz.Congres
Account no 40.84.68.939
ABN-AMRO Bank
Postbus 362
3500 AJ Utrecht
Reference : OTS-Robustness registration fee
A copy of the bank transfer should be sent to us together with
your registration form. Make sure you add transfer charges.
- You can use MasterCard/Eurocard and VISA credit cards.
***********************cut here********************************
Registration Form
Mr/Ms ............................................................
Family Name ............................................................
First Name ............................................................
Affiliation ............................................................
Address ............................................................
............................................................
............................................................
Tel ............................................................
Fax ............................................................
Email ............................................................
Conference rate:..............
Enclose a copy of the bank transfer, or fill in and sign below if you
pay by credit card.
Please charge [ ] Mastercard/Eurocard [ ] VISA
Card number: ..........................................................
Expiration date: ..........................................................
Amount: ..........................................................
Name: ..........................................................
Address: ..........................................................
Signature: ..........................................................
Send this form, with full payment, before 1 October 1993 to:
Robustness Organizing Committee
OTS
Trans 10
NL-3512 JK Utrecht
The Netherlands.
Tel: +31-30-536006
Fax: +31-30-536000
Email: OTS@let.ruu.nl