TOP
Search the Dagstuhl Website
Looking for information on the websites of the individual seminars? - Then please:
Not found what you are looking for? - Some of our services have separate websites, each with its own search option. Please check the following list:
Schloss Dagstuhl - LZI - Logo
Schloss Dagstuhl Services
Seminars
Within this website:
External resources:
  • DOOR (for registering your stay at Dagstuhl)
  • DOSA (for proposing future Dagstuhl Seminars or Dagstuhl Perspectives Workshops)
Publishing
Within this website:
External resources:
dblp
Within this website:
External resources:
  • the dblp Computer Science Bibliography


Dagstuhl Seminar 9313

Universals in the Lexicon: At the Intersection of Lexical Semantic Theories

( Mar 29 – Apr 02, 1993 )

Permalink
Please use the following short url to reference this page: https://www.dagstuhl.de/9313

Organizers
  • H. Kamp
  • J. Pustejovsky



Motivation

The lexicon is generally looked upon as the depository of all that is idiosyncratic about the basic constituents of a language, - as a place where one can find out the meanings of its words and affixes, as well as whatever is special about their syntax, morphology and phonology. A lexicon or dictionary that lives up to these expectations must contain a huge amount of apparently unrelated facts. Traditional dictionaries seem to prove the point. The wealth of information a good dictionary provides is spectacular. But, as lexicographers would be the first to admit, it is not easy to perceive any real systematicity in the welter of data it has to offer.

Lexicographers would also be ready to admit that even the best dictionaries supply only part of the information which they should contain if they were to capture all that is known about the meaning and use of the words to a native speaker. But what is missing from existing dictionaries is not just more of the kind of information they already contain. As linguistic theory deepens we are coming to realize that there are all manner of things about the meaning and function of individual words that are an essential part of linguistic knowledge but none of which can be found in any conventional dictionary.

These more global deficiencies of existing lexica are brought to our attention with particular force by work in computational linguistics, most particularly through the reections such work entails about the lexical information that should be available to sophisticated language processors (such as cooperative question-answering systems or machine translation systems with substantial coverage). Lexica that answer to the needs of such systems must be very different from any dictionary that is on the market today.

In fact, such lexica will not only have to contain a lot of new information, they also will have to be able to respond to particular kinds of query from other components of the system, by supplying just the right information in the right form. These tangible demands that computational architectures place on the retrievability and form as well as the content of lexical information mirror current convictions about the content and organization of the mental lexicon; this is one of the many areas where Artificial Intelligence and Cognition seem to go hand in hand.

A central, practically and linguistically important aspect to the matter of form and content of lexical information is what general principles support the mental lexicon and might serve as guidelines in the design of computational dictionaries. It has long been felt and often been claimed that there is still much to be discovered here, much that has to do with conceptual organization and with global choices that individual languages make with regard to lexical realization of conceptual structure. That such general principles exist is witnessed by, among other things, the numerous productive and semi-productive lexical processes that create new words out of existing ones or that assign old words new meanings. Such processes have long been a topic of linguistic interest. Some of them were subjected to close and illuminating scrutiny during the Dagstuhl meeting. But, as the meeting made clearly visible, they are ultimately just symptoms of a more fundamental conceptual organization.

These days such fundamental questions loom large in many areas of lexicon-related research, and they are reflected in, and often they have inspired, a spectrum of specific problems with which various research groups are currently concerned. It was out of an awareness of this that the idea of a Dagstuhl workshop on Universals in the Lexicon was born; and it was out of that same awareness also that the organizers strove to bring together lexicographers, linguists, computational linguists and philosophers whose professional careers have not so far given much occasion for scientific interaction. Indeed, those who found themselves thrown together between March 29 and April 2, 1993, might well have been described as a motley collection - a description that will no doubt have crossed the mind of more than one of those present. If it hadn't been for the fact - which we came to realize more clearly as time went on - that all of us were in some way or other bothered by the general questions hinted at above, the impression of being just a motley collection might well have prevailed until the end. But as it turned out, a sense of understanding and common purpose did develop. This alone, we think, should have been enough to call the meeting a success.

As a matter of fact the meeting accomplished a good deal more. The lexicographers succeeded in making plain to those linguists, computer scientists and philosophers who still needed persuading that trying to come up with an adequate description of just about any lexical item is likely to bring to light unexpected riches_and complexities; that these complexities arise already within a setting of "classical" lexicography; and that it is anything but a trivial matter to make proper use of the extensive resources that are no available in the form of large electronically stored corpora. Many of the linguists lifted, with their detailed studies of particular words or small groups of words, as many tips of the lexical veil that still hides most of the conceptual structure that lies below it from our view. While it would be foolish not to realize that most of this structure remains invisible, it can nonetheless be said that there is quite a bit that today we can see much more clearly than we could ten or even five years ago.

Finally, the philosophers and some of theoretical and computational linguists pointed towards a new direction in lexical research by stressing the connections between lexical meaning, discourse structure and inference (including, in particular, defeasible inference). This may well have been the first meeting devoted entirely to matters of the lexicon in which this aspect of lexical semantics has come into focus with such force and clarity. It is a direction whose importance may be expected to grow with time. For the connections on which it concentrates are crucial both to the theory of cognition and to the theory of computation. But it is a dimension of lexical knowledge of which we understand precious little at the present time.

Thus the result of the meeting should be characterized as a web of problems rather than a set of solutions. But even in the world of problems the whole is often more than the sum of its parts.

Copyright

Participants
  • H. Kamp
  • J. Pustejovsky