From: Tom Burke <burke@Arch.Stanford.EDU>
Date: Wed, 23 Feb 94 11:34:07 PST
Subject: CSLI Calendar, 24 February 1994, vol.9:18
Message-Id: <9402231934.AA04799@Arch.Stanford.EDU>



        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
______________________________________________________________________________

24 February 1994                 Stanford                       Vol. 9, No. 18
______________________________________________________________________________

      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 24 FEBRUARY -- 4 MARCH 1994

  THURSDAY, 24 FEBRUARY
         2:15 - CSLI Seminar 
                Cordura Hall, Room 100
                If It's All in the Head, then the Head Must Be Bigger
                Than We Thought
                Tom Burke, CSLI
                Abstract below

         4:15 - SSP Forum
                Building 60, Room 61-F
                The Cognitive Processes in the Evolution of Grammar: The
                Chinese BA Construction 
                Chao-fen Sun, Stanford Asian Languages
                Abstract below

         7:30 - Phonology Workshop
                Cordura Hall, Upstairs Lounge
                Consonant Gradation is Epenthesis
                Arto Anttila, Stanford Linguistics
                Abstract below

  FRIDAY, 25 FEBRUARY
        12:30 - PCD Seminar
                Skilling Auditorium
                Randal Walzer

         3:15 - Philosophy Colloquium
                Building 90, Room 91-A
                Must Intentional Systems be Rational?
                Kenneth Taylor, Rutgers Philosophy
                Abstract below

         3:30 - Linguistics Colloquium
                Cordura Hall, Room 100
                Aspect Shift
                Alessandro Zucchi, University of Illinois at Urbana-Champaign
                Abstract below

  THURSDAY, 3 MARCH
        12:00 - CSLI TINLunch
                Cordura Hall, Room 100
                Order Independent Typed Default Unification
                Alex Lascarides, Stanford Linguistics
                Absract below

         4:15 - SSP Forum
                Building 60, Room 61-F
                Using Situation Theory as a Descriptive/Analytic Framework
                in Studies of Interaction
                Keith Devlin, Saint Mary's College
                Abstract below

  FRIDAY, 4 MARCH
         3:15 - Philosophy Colloquium
                Building 90, Room 91-A
                On Translatability
                Alexander Duttmann, Stanford Philosophy

         3:30 - Linguistics Colloquium
                Cordura Hall, Room 100
                Classifier Noun Incorporation:  Lexicon or Syntax?
                Mark Baker, McGill University / CASBS
                Abstract below

                               ____________

The CSLI Calendar appears on Thursday of each week (more or less) throughout
the academic year.  Announcements, abstracts, and other information to appear
in the Calendar on a given Thursday should be submitted by e-mail to
incalendar@csli.stanford.edu by 5:00 p.m. on the previous Tuesday.

Past issues of the CSLI Calendar, a quarterly schedule of upcoming CSLI
events, and other information about CSLI are available on-line by way of a
CSLI Gopher at kanpai.stanford.edu.  The Calendar, with all available
abstracts, is also posted each week to the csli.bboard newsgroup.

                               ____________

                               CSLI SEMINAR
                          on Thursday, 24 February
                    2:15 p.m., Cordura Hall, Room 100
                  If It's All in the Head, then the Head
                      Must Be Bigger Than We Thought
                                Tom Burke
                                   CSLI
                         burke@csli.stanford.edu

The "it" in the title of this talk refers to "thinking".  Heads are just the
size they are, so if the title makes a true claim, then thinking is not just
all in the head.  G. H. Mead (back in the first decades of this century) put
forth the peculiar view that thinking takes place in a domain of head/world
interaction, not just in the head.  For example, human/computer interaction
constitutes one kind of head/world interaction.  Following Mead, we would
conclude that thinking is not just in the human part of human/computer
interaction but in the interaction itself.  For another example, in the form
of a question, do people with perceptual or motor disabilities conceptualize
the world differently than people without such disabilities?  According to
this view, they do, insofar as they interact with the world differently and
their conceptualization is embodied in this interaction.  How might this bear
on issues of "equal access" to public information networks?  Is information
only in the computer part of human/computer interaction or in the interaction
itself?

In support of this strange claim about the nature of mind, Mead concocted an
evolutionary story, drawing heavily on social and cultural aspects of human
nature, outlining how it is plausible to think of mentality in these
interactive terms.  In this talk I will trace out this story so far as I
understand it, to try to show how a smart guy like Mead could think such an
odd thing.

                               ____________

                          SYMBOLIC SYSTEMS FORUM
                         on Thursday, 24 February
                    4:15 p.m., Building 60, Room 61-F
               The Cognitive Processes in the Evolution of
                  Grammar: The Chinese BA Construction 
                               Chao-fen Sun
                         Stanford Asian Languages

No other Chinese construction has attracted the attention of Chinese linguists
more than the BA construction.  In my talk I will discuss its history and show
how BA changed in the last 2,000 years from an Old Chinese verb meaning 'hold,
take, etc' into a Mandarin clausal operator which requires (i) an object verb
word order, (ii) an obligatory referential object, (iii) a temporally bounded
proposition, etc.

As a metonymic process in Middle Chinese, BA was first grammaticalized from a
serial verb construction into a preverbal prepositional accusative case
marker.  I will also discuss how this change can be viewed together with the
emergence of the passive marker BEI at the time to offset a communication
ambiguity caused by a separate linguistic change in Middle Chinese which had
neutralized an Old Chinese syntactic restriction.  After this first major
meaningful shift which turned the verbal BA into a grammatical marker in
certain contexts, BA went through a process of grammatical enrichment in
analogy with an Old Chinese preverbal object marker YI.  This metaphorical
extension enabled BA to take on all of YI's preverbal functions in marking
undergoer and instrument cases, triggered YI's decline in use, ultimately
replaced it and survived into Modern Mandarin with YI's legacy.

In spite of its Middle Chinese innovation, in Early Mandarin BA was still
commonly used as a verb.  In addition, the distributional constraints (ii) and
(iii) mentioned above were practically non-existent in early Early Mandarin.
At this time the nominals after BA can be referential or non-referential.
However, BA's function in marking instrumental case became impossible in late
Early Mandarin.  This may be related to its second meaningful shift which
transformed BA into a prima facie high transitivity marker.  In Middle Chinese
the majority of the non-referential NPs after BA go with the instrumental
case.  After its use to mark instrument became rare, interestingly it could
not co-occur with any non-referential NPs any more.  Cross-linguistically,
high transitivity is found to be correlated to a referential objective
nominal. The constraints (ii) and (iii) mentioned above are exactly the
features found correlated to the notion of transitivity in Hopper and
Thompson's study.

I claim that this process of transitivization is directly responsible for the
loss of BA's verbal function.  As a result, the BA in the Middle Chinese
biclausal serial verb construction, BA NP V (NP), went out of fashion and was
grammaticalized into an operator of a monoclausal construction with the
following V as the main predicate of the proposition in late Early Mandarin.
A syntactic test will be discussed to show how the grammatical status of BA at
different times can be determined.  Typically a serial verb construction can
go with a purposive construction in Chinese at different times.  While BA
could easily go with a purposive construction at earlier times, it could no
longer do so after the transitivization process in late Early Mandarin.

                               ____________

                            PHONOLOGY WORKSHOP
                          on Thursday, 24 February
                    7:30 p.m., Cordura Hall, Upstairs Lounge
                    Consonant Gradation is Epenthesis
                               Arto Anttila
                           Stanford Linguistics
                        anttila@csli.stanford.edu

In this talk, I argue that the phonological processes called Consonant
Gradation (CG) and t-Deletion (TD) in Finnish are not weakening phenomena to
be formalized as deletion/delinking (the standard analysis), but rather
instances of constraint-driven epenthesis.  My analysis provides evidence in
support of Optimality Theory (Prince and Smolensky 1993, McCarthy and Prince
1993).

Under the deletion hypothesis, an underlying double stop is degeminated before
a branching rhyme (CG: CVX.XVC --> CV.XVC, X = /p,t,k/), and an onset /t/ is
deleted after a light unstressed syllable (TD: CV.tV --> CV.V).  Examples:

   so.net.ti/so.ne.ti+n         sonnet/sonnet+GEN               (CG)
   man.sik.ka/man.si.ka+n       strawberry/strawberry+GEN       (CG)
   va.ras/va.ras.+ta            thief/thief+PTV                 (no TD)
   ta.lo/ta.lo.+a               house/house+PTV                 (TD)

This hypothesis is problematic:

(1) CG presupposes an underlying geminate, but length turns out to be
noncontrastive: roots like */soneti/ (as opposed to /sonetti/) are
systematically absent.

(2) C-final loanwords are nativized by adding /i/ to achieve CV-finality
(/idealismi/ `idealism'), but if C = /p,t,k/, we have gemination (/galluppi/
`gallup') and the nativized form feeds CG (/gallupi+n/).  The standard
analysis provides no motivation for the gemination.

(3) TD repairs the ill-formed string *CV.tV by deleting /t/. It turns out that
*CV.tV is banned inside roots as well, which is unexpected since TD is known
to be a cyclic rule.

(4) Suffixal consonants are exclusively coronal (/puu+ta/ `tree+PTV',
/juos+ten/ `run+INSTR').  Suffixes like /-pa/ or /-ka/ are systematically
absent.  Under the deletion analysis, this has to be stipulated in the
lexicon, which is unsatisfactory.

I propose an analysis where the "weak" forms (/soneti/, /mansika/) are
underlying, but suboptimal due to a high-ranking constraint which bans a
single stop between two open syllables: *CV.XV, X = /p,t,k/. Such structures
must be syllabified CVX.V and the missing onset is supplied by epenthesis. I
also argue that suffixes are simply vocalic and that coronal consonants are
inserted by epenthesis. This analysis is superior to the deletion analysis in
a number of ways:

(1) CG and TD disappear as rules; geminates and suffixal coronals are
FILL(ONSET) violations induced by higher constraints.

(2) We can explain why suffixal consonants are all coronal (/va.ras.+ta/).
Epenthetic consonants get their place specification in one of two ways: either
by spreading from the stem-final consonant, which are independently known to
be exclusively coronals, or by default, coronal being the universal default.

(3) The systematic absence of roots like */soneti/ is no more problematic;
both cyclic t-Deletion and the gap in the root inventory follow from the same
constraint.

(4) The nativization process can be stated simply as V-epenthesis (/idealism/
--> /idealismi/), since gemination (/gallup/ --> /galluppi/) now follows by
the independently needed onset epenthesis.

                               ____________
                                     
                  SEMINAR ON HUMAN-COMPUTER INTERACTION
                          on Friday, 25 February
                  12:30--2:00 p.m., Skilling Auditorium
                              Randal Walzer 
                          randoid@well.sf.ca.us

                               ____________
                                     
                          PHILOSOPHY COLLOQUIUM
                          on Friday, 25 February
                    3:15 p.m., Building 90, Room 91-A
                  Must Intentional Systems be Rational?
                              Kenneth Taylor
                            Rutgers Philosophy
                         taylor@csli.stanford.edu

A number of thinkers have offered apriori arguments for what I will call the
no intentionality without rationality thesis.  According to that thesis it is
some kind of necessary condition on X's being an intentional system, that X be
largely rational.  In this talk, I argue that if broadly information-theoretic
approaches to the naturalization problem for intentionality of the sort made
prominent by, among others, Fred Dretske, Jerry Fodor, and Ruth Millikan are
on even roughly the right track, then it is highly doubtful that even minimal
rationality is any kind of necessary condition on being an intentional system.
I proceed by beginning at a relatively low level of nature, trying to see, as
we move up nature's hierarchy toward ever more complex creatures whose
information bearing states play ever more complex roles in their cognitive
lives, just where we find the first stirrings of both rationality and
intentionality.  We find, I argue, analogs, precursors, and rudiments of
intentionality at relatively low levels of nature.  Rationality, on the other
hand, emerges only at much higher levels of nature with relatively complex
cognitive systems.  And long before we get to cognitive systems that are
clearly rational, we find systems which are clearly intentional.  That the
intentional and the rational first emerge at different levels of nature is
good reason for denying the no intentionality without rationality thesis.

                               ____________

                          LINGUISTICS COLLOQUIUM
                          on Friday, 25 February
                    3:30 p.m., Cordura Hall, Room 100
                               Aspect Shift
                            Alessandro Zucchi
                University of Illinois at Urbana-Champaign
                          zucchi@cogsci.uiuc.edu

The classification of predicates in the basic categories state, event, and
process has become a familiar working tool for linguists investigating the
syntax and the semantics of aspect.  It is well-known (since Dowty 1979) that
the different tests proposed to identify these predicate classes do not yield
entirely consistent results.  In particular, according to this battery of
tests, some predicates behave as if they belonged to more than one aspectual
class (aspect shift).  I argue that the traditional strategy of appealing to
lexical ambiguity of predicates to cope with these inconsistencies is
inadequate.  I present a treatment of the progressive and of durational
adverbs, based on Parsons's (1990) event semantics, which avoids some of the
problems that arise for the lexical ambiguity approach.  I argue that the
investigation of aspect shift leads us to two main theoretical conclusions:
(1) The progressive of copular predicates and the progressive of non-copular
ones are sensitive to different aspects of the predicates to which they apply.
(2) For-phrases are sensitive to whether VPs describe culmination-gradable
events (events for which it is a vague matter what constitutes a completion).

                               ____________

                               CSLI TINLUNCH
                          on Thursday, 3 March
                    12:00 p.m., Cordura Hall, Room 100
               Order Independent Typed Default Unification
                             Alex Lascarides
                           Stanford Linguistics
                          alex@csli.stanford.edu

In this talk, I will define an order independent, and hence fully declarative,
version of default unification on typed feature structures (TFSs).  The
operation extends that described by Young and Rounds in ACL93 in two important
respects.  First, default information in an FS that is typed with a more
specific type will override conflicting default information in an FS typed
with a more general type, where the specificity is defined by the subtyping
relation in the type hierarchy.  Second, the operation is able to handle TFSs
where reentrancies are default.  This enables the operation to have at least
the same potential for linguistic application as Bouma's and Carpenter's
definitions of typed default unification, but with the added advantage of
being declarative.

I will demonstrate the utility of this version of default unification in
several linguistic applications.  First, I will show how it can be used to
define multiple orthogonal default inheritance in the lexicon in a fully
declarative fashion.  Second, I will describe how it can underpin feature
propagation principles, such as the Head Feature Convention in GPSG, without
restrictive procedural assumptions.  And finally, I will discuss the pragmatic
phenomena of blocking and semi-productivity, and argue that they militate
against a purely abbreviatory use of default specifications in lexical
entries.  If there's time, I will show how the representation of default
information in TFSs enables defaults to survive beyond the lexicon, and so
interact with discourse processing heuristics in the way that blocking and
semi-productivity demand.

                               ____________

                          SYMBOLIC SYSTEMS FORUM
                           on Thursday, 3 March
                    4:15 p.m., Building 60, Room 61-F
        Using Situation Theory as a Descriptive/Analytic Framework
                        in Studies of Interaction
                               Keith Devlin
                           Saint Mary's College
                         devlin@csli.stanford.edu

Situation theory was originally conceived as a mathematical theory that would
serve as a foundation for situation semantics.  Much of the early development
of the theory was carried out at Stanford's CSLI.

Situation semantics attempts to provide mathematical structures to serve as
the formal "meanings" of various parts of speech.  Mathematical structure is
intended to correspond to the structure of language.  Situation semantics is
able to capture many features of language use normally classified as
"pragmatics", for example indexicality and various other issues to do with
context.

In joint work published last year, Duska Rosengerg and I used situation theory
for another purpose, namely as a descriptive/analytic tool in the study of
socially conditioned interaction.

We refer to the method we evolved as "zooming".  Using this technique, an
initial analysis looks little different from a traditional analysis in
sociolinguistics.  Where our approach offers something new is that we make use
of the mathematical tools provided by situation theory to give increased
precision at various key points of the analysis.  The role played by situation
theory is in large part a methodological one.  At every stage of the analysis,
we have to ask various precise questions.  This forces us to adopt both a
uniform framework and a consistently high degree of precision, even at stages
that might not seem problematic.  When a problem is encountered, we "zoom in"
on that part of the analysis, increasing the mathematical precision until a
level of detail is reached that is sufficient to provide a resolution to the
problem.

Given two hours, I could probably give a reasonably good account of this work.
I plan to speak for much less than an hour and give the overall flavor.

                               ____________
        
                          PHILOSOPHY COLLOQUIUM
                            on Friday, 4 March
                    3:15 p.m., Building 90, Room 91-A
                            On Translatability
                            Alexander Duttmann
                           Stanford Philosophy

                               ____________

                          LINGUISTICS COLLOQUIUM
                            on Friday, 4 March
                    3:30 p.m., Cordura Hall, Room 100
            Classifier Noun Incorporation:  Lexicon or Syntax?
                                Mark Baker
                        McGill University / CASBS
                         baker@casbs.stanford.edu

Over the past ten years there has been a sometimes vigorous debate over the
nature of noun incorporation which cuts across a variety of grammatical
frameworks.  I and others have defended the view that it is syntactic, with
the incorporated noun counting as (part of) the grammatical object of the
verb.  However, an important lexical alternative has been proposed by Di
Sciullo, Williams, and S. Rosen which looks superior in certain respects.  In
this talk I will seek to clarify the conceptual and empirical issues at stake.
Then I will show how an understanding of the nonconfigurational syntax that is
typical of languages having noun incorporation affects the logic of the
arguments and their empirical predictions.  Finally, I will test some of these
new predictions with data from agreement, disjoint reference, and question
formation in Mohawk and certain other languages. The results support the
syntactic analysis for a wide range of cases.

                               ____________