<OT> Descriptive and Explanatory Adequacy in Linguistics (DEAL)
Rutgers Optimality Archive
roa at ruccs.rutgers.edu
Wed Jul 20 08:32:03 PDT 2005
Conference Announcement
Full Title: Descriptive and Explanatory Adequacy in Linguistics
Short Title: DEAL
Location: Berlin, Germany
Contact Person: Hans Broekhuis <Hans.Broekhuis at uvt.nl>
Meeting Email: DEAL at uvt.nl
Web Site: http://let.uvt.nl/deal05
Call Deadline: 15-Sep-2007
Meeting Description:
This workshop addresses problems concerning descriptive and explanatory
adequacy in linguistics, especially the Minimalist Program and
Optimality
Theory.
The current state of affairs in the minimalist program (MP) and
optimality
theory (OT) clearly illustrates the general tension between descriptive
and
explanatory adequacy. The description of the differences between the two
frameworks in the next three paragraphs is somewhat coarse and
oversimplified
in order to highlight the general tendencies.
MP is a very austerely formulated theory, but this seems to go at the
expense
of descriptive adequacy: the empirical scope seems to be restricted to
syntactic and semantic phenomena that involve or can be directly
related to
movement and features, and during the last decade much effort has been
devoted
to incorporating established insights of earlier phases of the
theory. Although this endeavour has been successful to a certain
extent, much
still seems to be out of reach of the theory in its present formulation.
OT, on the other hand, can readily be applied to virtually any
imaginable
subfield of linguistics. It is widely accepted as the standard
phonological
theory, and proposals exist that extend its empirical domain to the
fields of
syntax, semantics and pragmatics. The explanatory adequacy of OT is
limited,
however, which is clearly related to the fact that OT only offers a
general
scheme for the formulation of grammars; it has a system of ranked
violable
constraints at its core, but OT grammars can be specified in various,
sometimes even incompatible ways. A great advantage of this freedom is
that OT
has brought together researchers from various theoretical frameworks
(such as
GB, MP, LFG, HPSG, etc), but a drawback of this is that the body of
work that
could be referred to as OT-syntax lacks internal consistency, so that
OT-syntax is still far from acquiring a reasonable degree of explanatory
adequacy.
The problem concerning explanatory adequacy is actually enlarged by the
fact
that there is no clear consensus about what constitutes the core of
OT-syntax,
the universal set of constraints CON. Contrary to what seems to be the
case in
OT-phonology, there is no well-established and widely accepted set of
substantive syntactic constraints. There is even disagreement on the
general
format of these constraints: whereas in phonology most constraints are
assumed
to be either faithfulness or markedness constraints, this is not
clearly true
for OT-syntax. This, in its turn, is related to the question what forms
the
input of the OT-syntax. Many possibilities come to mind: a numeration, a
pre-established phrase marker, a semantic representation, etc.
Many researchers construe MP and OT not only as competing but also as
incompatible theories. Since OT is not committed to a particular
representational formalism, this is by no means obviously the case: it
could
well be imaginable that current MP could be reformulated in OT-terms.
This
might actually solve one major problem for current OT-syntax since it
would
restrict the syntactic constraints such that they can only be phrased
in terms
of (the outputs of) the operations of the generator, which under this
scenario
would be the computational system CHL, and the legibility conditions
imposed
by LF and PF. Further, it is clear that MP and OT-syntax in their
present
formulation focus on different aspects of the grammar: whereas the
former is
mainly concerned with the derivation of structure, the latter evaluates
the
syntactic representations created by the generator. Of course, MP
acknowledges
that the output structures must be evaluated by interface/bare output
conditions, but an explicit formulation of these conditions is still not
provided. Within OT-syntax it is sometimes acknowledged that conditions
are
needed on the generator, but so far explicit proposals are lacking. It
is
conceivable, therefore, that MP and OT are actually complementary
theories,
which can cancel each other's weaknesses. Of course, incorporating
insights
from MP into OT, or vice versa, can be obtained in a variety of ways.
The
questions whether this is desirable, and, if so, in which way this must
be
obtained are largely empirical in nature.
In this workshop we want to address questions that involve the
descriptive and
explanatory adequacy of MP and OT, and the relation between the two
frameworks. Below we provide some of the questions that could be
addressed in
this conference. Of course, this set of questions is not exhaustive.
The way
these questions are formulated is certainly biased, and expresses our
belief
that the insights of MP and OT can profitably be combined into a more
comprehensive theory. Of course, contributions that contest this belief
and
argue for the primacy of one of the two frameworks are also welcome at
this
workshop. It goes without saying that the questions below are just a
random
subset of the set of questions that can be addressed in this workshop.
The minimalist program
- Empirical scope. So far the empirical scope of MP seems to be
restricted to
syntactic phenomena involving or directly relatable to movement. Is
this an
inherent restriction of MP? If not, how can MP be extended in order
to also
cover cases that cannot be related to movement. If so, how should the
facts
that MP cannot account for be addressed?
- Computational system. In the earliest studies within the MP, it was
claimed
that CHL is universal and that differences in the output should be
attributed exclusively to the morpho-syntactic features in the
lexicon:
language variation is due to the set of strength/epp-features in the
lexicon. In general practice, the effect of these features is that
each
numeration leads to a single converging syntactic representation.
Although
the universality of the CHL is still maintained, the claim that the
locus of
language variation is the lexicon is not. In recent work, it is
assumed that
the output of CHL is evaluated by the interpretative component INT,
which
seems to consist of a set of language-specific filters. This implies
that
CHL overgenerates, and that the selection of the acceptable structure
is
actually due to the evaluation of the output of CHL. This raises the
question whether we still need to postulate lexical
strength/epp-features or
whether they can actually be assumed to be part of INT.
- Interface/bare output conditions. What are the interface conditions,
and how
should they be formulated? In which way are they applied to the
structures
created by CHL? What is the relation of the interface/bare output
conditions
to the filters of the interpretative component INT? More
specifically, is
INT part of the semantic bare output conditions? If so, is there also
a
phonological interpretative component?
Optimality Theory
- Input. It is not a priori clear what the input of the OT-generator
is. Is it
a numeration of the sort proposed in MP, or is it some other object
like a
semantic or some prefabricated syntactic representation. If the
latter, what
is the origin of these representations?
- Generator. Although it is generally acknowledged that the generator
has
certain properties that affect the candidate set, studies in OT
normally do
not explicitly address this. Furthermore, these properties normally
do not
play a role in the explanations for the empirical data. Nevertheless,
a
fully elaborated theory should make clear what operations the
generator
consists of, and what effects these operations have on the output of
the
system. In principle, enrichment of the generator may reduce the
candidate
sets, which in its turn may void the need of certain constraints and
would
shift part of the explanatory power to the generator. Is there
anything in
OT that would block such a move? Is such a move desirable? May such a
move
contribute to making the theory explanatorily adequate?
- Evaluator. Focus in OT-syntax seems to be on descriptive adequacy,
and since
the constraints are crucial in obtaining that, it is not surprising
that the
set of proposed constraints is growing abundantly. In a sense
OT-syntax
resembles early TG-theory, where the focus was on formulating and
discovering the properties of the newly introduced means of
transformations. However, if we can learn something from the history
of
generative grammar, it is that we can only achieve explanatory
adequacy if
we place severe bounds of the admissible means of describing the
data. It is
therefore necessary to develop a constrained and principled theory of
constraints: What is the general format that the constraints should
adhere
to, and what are the types of constraints we may expect to occur?
The Minimalist Program and Optimality Theory
- Should MP and OT-syntax be considered two competing theories? If so,
what
properties of the two systems make them intrinsically incompatible?
What are
the prospects of solving the problems concerning the
descriptive/explanatory
adequacies of the two systems?
- Should MP and OT-syntax be considered as frameworks that simply focus
on
different questions, so that they are actually complementing each
other? If
so, could they be profitably combined into an overarching theory?
- Should we incorporate aspects of OT into MP (e.g. could we
reformulate the
filters in INT and/or the interface conditions into a OT-format) or
should
we incorporate MP into OT (e.g. could some version of CHL be the
OT-generator). How would combining aspects of MP and OT contribute to
solving the descriptive/explanatory adequacies of the two systems?
Practical information
The workshop is organized by the Linguistics departments of the
University of
Tilburg (the Netherlands), the University of Potsdam (Germany) and the
Zentrum
f¸r Allgemeine Sprachwissenschaft, Typologie und Universalienforschung
(ZAS)
in Berlin (Germany).
The workshop will be held at the ZAS from Saturday December 17, 2005,
until
December 19, 2005. The first two days and the morning session of the
third day
will consist of 15 talks of fifty minutes each (including a 10 minutes
discussion). Five of these talks will be given by the following keynote
speakers:
1) Edwin Williams, Princeton University, New Jersey
2) David Pesetsky, MIT, Cambridge, Massachusetts
3) GÈraldine Legendre, Johns Hopkins University, Baltimore, Maryland
4) Jane Grimshaw, Rutgers University, New Brunswick, New Jersey
5) Chris Collins, Cornell University, Ithaca, New York
The afternoon session of the third day will be devoted to a round table
discussion of the five keynote speakers chaired by Henk van
Riemsdijk. Abstracts are solicited for the remaining 10 slots. The
abstracts
will be reviewed anonymously. Please keep to the following instructions
concerning your abstract:
- Submission is only possible in electronic form, preferably in
pdf-format but
we also accept .rtf, .doc, or plain text files.
- Send one copy that includes your name and affiliation, and one
anonymous copy.
- Abstracts may not exceed two pages of text with an at least one-inch
margin
on all four sides.
- Abstracts must employ a font not smaller than 12 point.
- Each page may include a maximum of 50 lines of text.
- Abstracts may include an extra page for references (not examples).
- Abstract should be sent to DEAL at uvt.nl
- Abstract should be received by September 15, 2005.
- Notices of acceptance will be sent out before October 15, 2005.
More information about the Optimal
mailing list