Call for Papers
Local Modelling of Non-Local Dependencies in Syntax
Workshop, 30th meeting of the German Linguistic Society (DGfS)
February 27-29, 2008
Universität Bamberg
Organizers:
Artemis
Alexiadou
Institut für Linguistik/Anglistik
Universität Stuttgart
|
Tibor
Kiss
Sprachwissenschaftliches Institut
Ruhr-Universität Bochum
|
Gereon Müller
Institut für Linguistik
Universität Leipzig
|
Workshop Description
Non-Local Dependencies
Syntactic dependencies may be non-local in the sense that they involve
two positions in a phrase structure whose correspondence cannot be
captured by invoking notions like ``clause-mate relation'' or
(non-extended) ``predicate/argument structure''. A classic example
that instantiates such a non-local relation is the existence of
long-distance movement dependencies in natural languages (e.g.,
wh-movement, topicalization, etc.), where the displaced item and its
base position can in principle be separated by arbitrarily many
intervening clause boundaries. However, there are many other
syntactic dependencies that can also be non-local in this sense. For
instance, reflexivization is often confined to minimal
predicate/argument structures, but it may also apply non-locally in
certain contexts, in certain languages (without necessarily being
amenable to an account in terms of logophoricity). Control of the
subject of an infinitive by an argument belonging to a matrix clause
also emerges as a non-local operation, at least in some
analyses. Furthermore, many languages (among them, e.g., Tsez,
Itelmen, and Hindi, but also, strictly speaking, Icelandic) exhibit
instances of non-local agreement. Case assignment, too, may in
principle be non-local (i.e., it is not necessarily confined to
minimal predicate/argument structures); and tense relations between
clauses are non-local almost by definition. Finally, a particularly
clear example of a non-local dependency is the binding of pronouns
that are interpreted as variables.
Local Modelling
By postulating successive cyclicity in the case of displacement
phenomena (i.e., Comp-to-Comp movement), a non-local dependency was
(to some extent) modelled as a local phenomenon in classic
transformational grammar. Subsequently, an even more local treatment
of movement dependencies was developed by Gerald Gazdar in the
framework of GPSG, by adopting Slash features that are passed on in
minimal subtrees; essentially, this kind of approach is still
maintained in HPSG analyses. Interestingly, recent analyses within the
Minimalist Program (including some of Chomsky's own work) converge
with Slash feature percolation approaches in that they assume that
displacement phenomena involve minimal local movement steps -- not
only to the edge of each phase (i.e., clause or predicate phrase), but
actually to the edge of each XP (also see Jan Koster's recent work on
gap phrases). In the same vein, it has recently been proposed that
reflexivization should be modelled in a strictly local way (by
invoking feature percolation or extremely local movement steps) --
both within HPSG analyses and Minimalist analyses. Analogous
considerations apply in the case of the other non-local dependencies
mentioned above.
Goals of the WorkshopAgainst this background of growing convergence among syntactic
theories, the goals of the workshop are these: 1) to bring together
researchers working on the local modelling of non-local dependencies
from different theoretical points of view; 2) to discuss advantages
and disadvantages of local treatments of non-local dependencies; and
3) to compare different theoretical approaches. As far as this last
point is concerned, we believe that it may turn out that local
analyses of non-local phenomena developed in different kinds of
syntactic theories (and spanning the generative/declarative dichotomy)
can be shown to not only share identical research questions, but also,
to a large extent, identical research strategies. Needless to say,
these considerations are not confined to HPSG and the Minimalist
Program; they also apply to syntactic theories in which local
approaches to non-local dependencies are either an important building
block per se (e.g., LFG, categorial grammar, in some sense also TAG),
or in which local analyses have recently come to the fore as viable
alternatives to standard, non-local approaches (e.g., optimality
theory). Recurring questions arising in this general area of research
include the following: How can asymmetries between different kinds of
(basically non-local) dependencies be accounted for (e.g.,
displacement may often be non-local to a higher degree than
reflexivization)? And how can asymmetries between different languages
with respect to the same kinds of (basically non-local) dependencies
be accounted for?