ITRI Seminars - Autumn 2000


		ITRI seminars - Autumn 2000 ITRI seminars generally take place 12 noon on Thursdays in room W107 on the first floor of the Watts Building, University of Brighton (Moulsecoomb site). Occasional deviations from this pattern are indicated below. Information on how to find W107 is available on our contact page.
26 Oct abstract		Johan Bos University of Edinburgh Inference and Discourse Interpretation
9 Nov abstract		Krista Varantola Tampere University, Finland User-sensitive lexical databases: a case of lexical knowledge management
16 Nov abstract		Gregory Grefenstette Xerox Research Centre Europe, Grenoble Very Large Lexicons from the WWW
30 Nov abstract		Ruth Kempson Kings College, London On making syntax dynamic: Towards a typology of left-dislocation structures
7 Dec abstract		Wolfgang Teubert University of Birmingham The meaning of globalisation. Corpus linguistics vs. cognitive linguistics

		Previous ITRI seminars
		See also NLP seminars at COGS, University of Sussex

Abstracts

Johan Bos
Inference and Discourse Interpretation

Automatic discourse interpretation requires certain inference tasks for dealing with pronouns and presuppositions, or other phenomena that trigger ambiguities. Ambiguities lead to multiple interpretations, and several acceptability constraints have been suggested recently in computational semantics to filter out undesired interpretations:

a discourse should not contain contradictions;
each contribution to the discourse should be informative.

These constraints on acceptability also hold on subordinated levels of discourse, as suggested by Van der Sandt in his work on presuppositions.

In this talk I discuss the implementation these acceptability constraints in the DORIS system. This system builds DRSs (the boxes used in Discourse Representation Theory) for a given text. To implement the inference tasks required for discourse interpretation, DORIS appeals to several state-of-the-art theorem provers and model builders in a distributed framework. DORIS exemplifies the feasibility of this approach to implement inference, but also highlights the shortcomings in coping with certain linguistic phenomena.

Krista Varantola
User-sensitive lexical databases: a case of lexical knowledge management

I shall focus on user-friendliness in lexical information sources and give a brief summary of the main points in two papers by Virpi Kalliokuusi and Varantola (EURALEX 1998 and 2000). I will then go on to describe an application of these ideas in a project we hope to launch in the near future.

The key issues I wish to elaborate are:

the context-sensitive needs of dictionary-users
lexical databases with interactive, context- and user-sensitive interfaces
frame-semantic thinking in the context of user-sensitive lexical interfaces.

Gregory Grefenstette
Very Large Lexicons from the WWW

Three unrelated technological advances make it possible to conceive of a new type of linguistic resource: Very Large Lexicons containing abstracted examples of all language use. This talk will describe how these technologies interact and how Very Large Lexicons can be built. Examples will be given of extracting abstracted language use of English and other language from the WWW.

Ruth Kempson
On making syntax dynamic: Towards a typology of left-dislocation structures

This talk shows how modelling interpretation on a left-right basis as a process of tree growth provides solutions to phenomena taken in other frameworks to constitute syntactic puzzles, leading to natural cross-language typologies based solely on concepts of tree growth. The case-study I shall give will be the interaction between anaphora, relative clause construal, and left-dislocation characterising cross-language variation in the resumptive use of pronouns in left dislocation structures, while sustaining an entirely unitary, pragmatic process to pronoun construal.

Wolfgang Teubert
The meaning of globalisation. Corpus linguistics vs. cognitive linguistics

A corpus is a principled collection of texts, and as such, it is a sample of the universe of discourse of a discourse community. This discourse is the manifestation of the language of the discourse community. New texts entered into the discourse always refer to older texts. Words used in texts always refer to previous citations of these words, be it in the same text or in earlier texts. What words mean is how they have been use before in the discourse, how they have been paraphrased before and sometimes even how they have been used to refer to something outside of the discourse. In this view, meaning is in texts, and it is not something that has been assigned to texts from outside the discourse. There is no language independent or language neutral taxonomy onto which the meaning of natural language texts and words can be mapped. It is not something that is in the mind, not some mental representation, not a concept, neither in Jerry Fodor's nor in Hilary Putnam's sense. As I see it, meaning cannot be separated from form; from its embeddedness in the discourse, not even where texts seem to refer to some discourse-external reality. I will use the example of the neologism globalisation to show that all we (as linguists) can ever hope to find out about globalisation is contained in its occurrences in the discourse. Outside of discourse, there is no globalisation.

Maintained by Adam Kilgarriff (Adam.Kilgarriff@itri.brighton.ac.uk).
Last updated Monday November 20 2000

©Information Technology Research Inst itute