AG 1, AG 2, AG 3, AG 4   Info

Expert Audience

-- Not specified --

Saarbrücken

Currently visiting:
Division of Informatics
University of Edinburgh (Edinburgh, Scotland)

Web site: http://www.cogsci.ed.ac.uk/~gj/
.../index.html links to publications
.../dissertation.html drafts of my dissertation
.../dissertation/dgl/wo.html word order fragments

ABSTRACT

In the Prague School, information structure (IS) is considered an
inherent aspect of linguistic meaning (LM), expressing how a
sentence's LM depends on, and may affect, the discourse context in
which the sentence is uttered. As has been frequently pointed out,
languages may use word order, tune, and morphology to realize
information structure. When realizing information structure, no
language avails itself solely of just one of these means - but when
does a language use one means or the other?

To try and find an answer to this question, I look at the realization
of information structure from a cross-linguistic (typological)
perspective, in the following senses:

(1) How can we formulate hypotheses that predict what means a language
employs to realize for example the informational focus, based on
its language type?

(2) How can we formulate grammar architectures that elucidate how
languages differ, or share commonalities, in the ways they express
information structure?

In this talk, I address (1) and (2) using insights from language
typology, dependency grammar and categorial grammar, and multilingual
grammar modeling. The talk begins with a brief overview of Dependency
Grammar Logic (DGL). DGL is grammar framework that formalizes Praguian
dependency grammar using extended categorial grammar for a detailed
description of form ('syntax') and hybrid logic for representing
linguistic meaning ('semantics'). In DGL, a grammar consists of a
lexicon and a set of structural rules ('tree-structure rewriting
rules'). A multilingual grammar architecture is a network of packages
of structural rules in which some packages may be available to more
than one language, and where packages are organized hierarchically by
a principle of "monotonic extension". The first illustration of DGL
and multilingual grammar architectures comes in the form of a detailed
description of word order. After that, I present a (semi-) typological
approach to describing the realization of information structure,
addressing (1), and the resulting formalization in DGL, addressing
(2). I end the talk with a discussion of how these formalizations
relate to a performance model of processing mixed and free word order,
and what the role of information structure could perceivably be in
such a model.

REFERENCES

DGL and the realization of information structure is the main topic of
my dissertation. Performance modeling of mixed/free word order
phenomena is joint work with Shravan Vasishth (Ohio State University,
http://www.ling.ohio-state.edu/~vasishth/).

Valia Kordoni

--email hidden

passcode not visible

logged in users only