4.0666 CSLI Calendar (1/383)

Elaine Brennan & Allen Renear (EDITORS@BROWNVM.BITNET)
Wed, 31 Oct 90 22:55:26 EST

Humanist Discussion Group, Vol. 4, No. 0666. Wednesday, 31 Oct 1990.

Date: Wed, 31 Oct 90 15:25:01 CST
From: Mark Olsen <mark@gide.uchicago.edu>
Subject: CSLI Calendar


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
_____________________________________________________________________________

1 November 1990 Stanford Vol. 6, No. 7
_____________________________________________________________________________

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

CSLI ACTIVITIES FOR THURSDAY, 1 NOVEMBER 1990

12:00 noon TINLUNCH
Cordura 100 Connectionist Approaches to Linguistic
Information Processing
David Rumelhart
(der@psych.stanford.edu)
Abstract in last week's Calendar

2:15 p.m. CSLI SEMINAR
Cordura 100 Object Theory, Intensional Logic, and
Situation Theory
Edward N. Zalta
(zalta@csli.stanford.edu)
Abstract in last week's Calendar

4:15 p.m. STASS SEMINAR
Cordura 100 An Analysis of Inference Involving Venn Diagrams
Sun-Joo Shin
(shin@csli.stanford.edu)
Abstract below

CSLI ACTIVITIES FOR THURSDAY, 8 NOVEMBER 1990

12:00 noon TINLUNCH
Cordura 100 Data Management in Environmental
Information Systems
Oliver Guenther
(guenther%dulfaw1a.bitnet@forsythe.stanford.edu)
Abstract below

2:15 p.m. CSLI SEMINAR
Cordura 100 Object Theory, Intensional Logic, and
Situation Theory
Edward N. Zalta
(zalta@csli.stanford.edu)
Abstract below

4:15 p.m. STASS SEMINAR
Cordura 100 Meaning in Speech Acts
Tomoyuki Yamada
(yamada@csli.stanford.edu)
Abstract below
____________

STASS SEMINAR ON 1 NOVEMBER
An Analysis of Inference Involving Venn Diagrams
Sun-Joo Shin
Department of Philosophy
Stanford University

Venn diagrams are widely used to solve problems in set theory and to
test the validity of syllogisms in logic. However, it is a fact that
Venn diagrams are not considered valid proofs, but heuristic tools for
finding valid formal proofs. The purpose of this talk is to present
Venn diagrams as a formal system of representation equipped with its
own syntax and semantics. Moreover, I prove that this system is sound
and complete.

This is a repeat of my talk at the "Situation Theory and its
Applications" conference in Scotland last September.
____________

TINLUNCH ON 8 NOVEMBER
Data Management in Environmental Information Systems
Oliver Guenther
FAW Ulm, Germany
(Research Institute for Applied Knowledge Processing
at the University of Ulm)

The management of geographic data, such as digitized maps, is a major
focus of FAW's research activities in the area of environmental
information systems. The research project RESEDA, for example, is
working on a knowledge-based system for the extraction of
environmental information from satellite images of the earth. In
order to obtain satisfactory results, it is essential to utilize
geographic information that is available on the area investigated.
This information is usually stored in a spatial data base to provide
efficient access to all objects in a given spatial neighborhood.

In this talk, we will discuss two techniques to facilitate the
management of large amounts of data in environmental information
systems. For the modeling of geographic objects we propose a concept
called spatial data-base views. Here only atomic objects (such as
lakes or buildings) are stored physically on disk at the largest
resolution available. Molecular objects (such as cities) are
represented by means of data-base views. This technique results in a
structural object-orientation that helps to avoid redundancy.

For efficient spatial access to these objects, spatial index
structures have to be used. Whereas numerous data structures are
available for the indexing of point data, the generalization to
extended data objects (such as polygons) has proven to be difficult.
The great variance in object sizes that is typical for geographic
applications seems to aggravate those problems. In our talk, we will
present a concept called "oversize shelves" that overcomes some of
these difficulties. Oversize shelves are special disk pages that are
attached to an index structure in order to store very large objects.
____________

CSLI SEMINAR ON 8 NOVEMBER
Object Theory, Intensional Logic, and Situation Theory
Edward N. Zalta
Department of Philosophy
Stanford University

We turn from the analysis of fiction within situation theory to the
general study of intensional logic. We'll begin with a discussion of
Frege's views, the views of the direct reference theorists, and
consider whether and how these two general views are incompatible. We
shall catalog the roles Frege's senses are supposed to play in the
philosophy of language, canvassing work of Tyler Burge, Nathan Salmon,
John Perry, and others. Then we begin to piece together an
understanding of Fregean senses from the framework of object theory.
____________

STASS SEMINAR ON 8 NOVEMBER
Meaning in Speech Acts
Tomoyuki Yamada
Visiting Scholar
Hokkaido University, Japan

My long-term ambition is to develop a philosophically sound and
mathematically rigorous theory of speech acts that provides an
empirically adequate treatment of speech act phenomena in Japanese.
In my current research, I treat meaning as relation between a type C
of circumstance of utterance, a type U of utterance, a type I of
illocutionary act performed, and a type B of background condition,
namely:

C & U => I | B
(C & U involves I given that B)

I will discuss the following questions:

(1) Can those illocutionary acts that do not have truth conditions be
about some portions of the world? And if they can, what kind of
portions can they be about?

(2) How should the contents of orders, requests, promises, etc., be
characterized? In what respects are they different from, and in
what respects similar to, contents of statements, reports, etc.?

(3) Is it possible for us to complete a list of conditions of felicity
for an illocutionary act without using phrases like "in normal
circumstances"? If it is not, how is it possible to write a
theory of speech acts?
____________

SYMBOLIC SYSTEMS FORUM
Color Appearance:
An Approach Based on Illuminants and Surfaces
Brian A. Wandell
Department of Psychology
Stanford University
Thursday, 1 November, 4:15 p.m.
Building 60, Room 61G

I describe a new approach to understanding color appearance based on
representing the illuminants and surfaces that underly natural-image
formation. I will review (briefly) the foundations of color science
and classic theories of color appearance. Then, I will describe how
color appearance theories can be improved by explicitly including
illuminants and surfaces, as well as photoreceptors, in the color data
representations.

Following week: Believing Computers, Yoav Shoham, Department of
Computer Science, Stanford University.
____________

PHONOLOGY WORKSHOP
The Role of Fundamental Frequency in Signaling
Affect and Contrastive Stress: Evidence for a Dissociation
Gerald McRoberts
Department of Psychology
Stanford University
Thursday, 1 November, 7:30 p.m.
Ventura 17

The naive intuition that English questions have a rising intonation
pattern while statements have a falling pattern has not been fully
supported by empirical study. Attempts to relate the irregular use of
final rise to categories of questions have not proved successful.
Indeed, even among Yes-No questions, which seem to be the most
regular, only some 40-50% show a final rise (e.g., Cohen 1972; Fries
1964; Lee 1980). Others have suggested that paralinguistic factors
may be associated with differences in the amount of final rise (e.g.,
Crystal 1969; Lee 1956, 1980; Jassem 1972).

A somewhat different approach is suggested by a series of studies of
the perception of intonation (e.g., Hadding and Studdert-Kennedy 1964,
1973). These studies demonstrated that as an f0 prominence preceding
the final rise was scaled up, the amount of final rise needed for
listeners to reliably judge a contour to be a question decreased.
This "trading relation" was such that for particularly high f0
prominences a falling final contour was often judged to be a question.
Furthermore, the influence of the f0 prominence on the final rise
appeared only when listeners made (linguistic) question-statement
judgments, not when they judged the direction of the final rise,
suggesting a phonetic (as opposed to auditory) basis for the trading
relation.

A series of experiments was carried out to investigate whether the
findings of Hadding and Studdert-Kennedy also pertained for the
production of question-statement intonation. The results suggest that
the "trading relation" between f0 prominence and final rise occurs
when f0 prominence is used to convey a linguistically relevant
contrast (i.e., contrastive stress), but not when f0 prominence is
used in conveying the affective state of the speaker. These results
are discussed in terms of different articulatory mechanisms involved
in the production of f0 prominences for affect and contrastive stress.
____________

PHILOSOPHY DEPARTMENT COLLOQUIUM
Criterian and Human Nature
Bernard Gert
Department of Philosophy
Dartmouth College
Friday, 2 November, 3:15 p.m.
Building 90, Room 91A

No abstract available.
____________

LINGUISTICS DEPARTMENT COLLOQUIUM
Grammaticalization and Argument Structure
(dissertation proposal)
Henry Smith
Department of Linguistics
Stanford University
Friday, 2 November, 3:30 p.m.
Cordura 100

What role, if any, grammatical relations (GRs) should play in linking
theory has been the subject of a running debate. Discussion to date
has focused on the trade-off between complicated representations and
complicated linking (e.g., case) rules. When it comes to giving an
account of grammaticalization in case rule systems, however, the no-GR
analysis provides a simpler explanation. For the italicized NPs (1-4)
from Icelandic, the key lies in formalizing the hierarchy of rules
behind sentences (1-4):

(1) Mig vantar hni'f. (me-A lacks knife-A)

(2) Barninu batnaDi veikin. (child-the-D recovered-from disease-the-N)

(3) E'g hef e'tiD (I-N have-1sg eaten)

(4) Hann keyrDi bi'linn thessa leiD. (he drove car-the-A this-A route-A)

(5) a. ACC-NP|arg|goal|of vanta
b. ACC-NP|arg|goal|of vanta

(6) a. DAT-NP|arg|goal
b. DAT-NP|arg|goal

(7) a. NOM-NP|arg
b. NOM-GR1

(8) a. ACC-NP
b. ACC-GR

Without primitive GRs the rules are (5a-8a), with GRs (5b-8b). The
rule that applies is the first or, intuitively, the "most specific"
rule. Using (5a-8a) (no GRs), we can reduce this, as in phonology, to
the proper inclusion relation between environments of
rules(restrictiveness). This is not possible for (5b-8b) since some
pairs, e.g., (6b,7b) refer to entities of a different sort. We can
predict the direction and the likely paths of the grammaticalization
of case: restrictiveness decreases over time (e.g., a shift from
lexical ACC to DAT (later NOM) as has taken place in German and is
taking place ("Dative Sickness") in Icelandic).
Restrictiveness-decreasing changes are widely attested; the reverse
are not. In this way, it is possible to unify and extend several
partial generalizations about the "acquisition of subjecthood" (Cole
et al. 1980) and semantic-to-syntactic case shift (Kurylowicz 1965).
____________

COMPUTER SCIENCE DEPARTMENT COLLOQUIUM
Are There Fundamental Limits to Network Performance?
David D. Clark
Laboratory for Computer Science
Massachusetts Institute of Technology
Tuesday, 6 November, 4:15 p.m.
Jordan 040

A number of factors influence network performance: technology limits,
modes of sharing, algorithm design, and implementation options. These
combine in a way that makes it rather complex to predict how a set of
network components will actually interact.

In particular, network protocols seem to be designed in a way that
obscures, rather than emphasizes, the performance consequences of the
design. While this might seem wrong, it is in fact the objective of
the designers. Protocol design proceeds on the assumption that a
single specification should permit different implementation with
different performance goals.

While this flexibility may be desirable, it leads to confusion. Since
protocol specifications give no guidance as to the range of possible
performance, a number of myths have arisen about the limits of
performance, most of which have eventually proved false.

In this talk, I explore where the fundamental limits of networking
arise, and how they relate to the design of hardware and protocol. In
passing, I will review the history of performance mythology, as well
as the role of key protocol abstractions, such as layered design, in
preventing us from realizing the actual performance of networks.
____________

SYNTAX WORKSHOP
Light Verbs and other XCOMP-taking Predicates in Japanese
Yo Matsumoto
Department of Linguistics
Stanford University
Tuesday, 6 November, 7:30 p.m.
Cordura 100

Japanese light verb constructions allow the arguments of a "verbal
noun" to appear in "verbal" case marking. In (1), for example, the PP
"sono chihoo e" (to the region), which is an argument of the verbal
noun "yusoo" (transportation), appears as if it were an argument of the
light verb "suru."

(1) Seehu wa sono chihoo e busshi no yusoo o suru.
government TOP the region GOAL goods GEN transportation ACC did
"The government transported various goods to the region."

Grimshaw and Mester (1988) claim that the construction exemplified by
(1) involves the process of Argument Transfer, by which the arguments
of a verbal noun transfer to the light verb "suru."

The purposes of this paper are twofold: to point out the problems of
the Argument Transfer account, and to propose an alternative account.
First, I point out that examination of light verbs other than "suru"
reveals that the properties of light verb constructions cannot be
explained by an operation on argument structures (as Grimshaw and
Mester conceive it), such as Argument Transfer. These properties
include: (1) the fact that adjuncts can also be "transferred," and (2)
the fact that the obligatory "transfer" of the subject argument of a
verbal noun is in fact a case of control.

Second, I will argue that the light verb construction in Japanese
involves an XCOMP in f-structure (i.e., the verbal noun is the
predicate of the XCOMP subcategorized by the light verb). In this
view, the apparent argument transfer is explained by the principle of
Functional Uncertainty, in much the same way that certain German
infinitival constructions are explained (Zaenen and Kaplan 1990). I
will argue that many properties of light verb constructions that
Grimshaw and Mester did not explain, such as obligatory adjacency of
the verbal noun and light verb, as well as the nonpassivizability of
the light verb, follow naturally in my account. Furthermore, my
account provides a unified account of other XCOMP-taking predicates in
Japanese, such as those in the benefactive and causative constructions
(Ishikawa 1985).
____________

PSYCHOLOGY DEPARTMENT COLLOQUIUM
Child's Theory of Mind
Henry Wellman
University of Michigan
Wednesday, 7 November, 3:45 p.m.
Building 420, Room 050

No abstract available.
____________