Toegepaste Taalwetenschap in Artikelen - Volume 27, Issue 1, 1987

Corpus linguistics is a rich source of information for lexical research. The use of computers can make this source easily accessible. Computer readable corpora may be enriched in various degrees: from untagged upto fully syntactically labeled. In this paper it is shown that a syntactically labeled corpus provides a better lab for lexical research than one that has been tagged for word classes. For this purpose we use some example investigations that were actually carried out on an existing analyzed corpus of English.

Many fields of language and speech research need a tool to get full information about phenomena that occur at the lexical level of language. To this end, a countrywide accessible database is being built with lexical information of the Dutch and English languages. CELEX is the co-operative effort of the University of Nijmegen, the Max-Planck-Institute for Psycholinguistics, the Institute for Dutch Lexicology, the Institute for Perception Research, and the Dr. Neher Laboratories of the Dutch Telecommunication Company.The design of a coherent lexical database with generally usable information about, amongst other things, morphology and syntax, leads to a challenging conflict. On the one hand, the objective of CELEX demands that the representation of all occurring phenomena should be independent - as far as possible -of specific language theories. On the other hand, the value of a lexical database is to a great extent dependent on the relational structure, which is based precisely on linguistic theories. In other words, a lexical database, which makes it possible to test theories, is itself a necessary tool for the construction of a lexical database.The pragmatic approach of the CELEX-project means that countless applications will be possible, but choices will have to be made. One application will be to research the optimal structure of theoretically-founded lexica and lexical databases.

The accessibility of lexical information stored on computers is not only important for the human computer user, but also for programs that process natural language. The requirements with respect to the content and structure of a computer dictionary are different than for a printed dictionary and depend on the specific function of the language processing system. For example, the lexical knowledge needed for hyphenation is different from that for the correction of spelling errors. A number of such language manipulating modules which are designed in the context of the Nijmegen Language Technology project are discussed with special attention to the dictionaries which they use.If one tries to integrate the different functions of an author system, integration of the different kinds of lexical knowledge in one database is an obvious task to consider. When designing a computer lexicon with many kinds of information, the redundant information which one finds in a conventional dictionary can be extracted and represented in the form of abstract categories and rules. Taking that approach not only makes the lexicon more compact but also makes lexical concepts explicit and usable in different ways.

Dictionaries should be designed with a special set of users in mind and for their specific needs". This claim has been made for at least 25 years, but the information gathered in this period about dictionary users is very scant. More is said in the lexi-cographical literature about what dictionary users should do than what they do in actual practice. Yet is it not very difficult to collect this sort of information, especially when computer based dictionairies are available. Meanwhile (concise) dictionaries are being produced. This article discusses in what dictionary makers can take into account the requirements of a specific set of users.

An ALP (=Advanced Learner's Dictionary)- model is presented in which both the learning object and the learning subject are central.The implications of such a model are made explicit and (some aspects of) English ALD' are discussed from that point of view.The aspects dealt with cover among others:- macrostructure- semantic component- syntactic component- metalanguage- overall organisationFinally the model is illustrated by means of sample entries.

Van Dale Basiswoordenboek Nederlands (Elementary Dictionary of Dutch), to appear in May 1987, differs from existing school dictionaries in a number of ways.The number of entries is considerably larger. It contains ± 25.000 entries, whereas the currently most frequently used school dictionary (Dijkstra' School Dictionary) contains ± 14.000 entries.The definitions are characterized as follows. Usually a logical definition is given. The definition of the word in a rather arbitrary context as in Dijkstra, is not very practical because for a child it is difficult to ascertain what part of the explanation is paraphrased by this context and what part is a definition of the entry. Synonymous definitions are only used to define difficult words by means of an easy one; in the position of the easy word a logical definition is provided. In this way circular definitions are avoided. In order to demonstrate the relation of a cluster of derivations to a root word we employ a morphosemantic definition for words belonging to such a cluster.The Basis woordenboek pays attention to usage in the Dutch-speaking part of Belgium. It contains ± 1200 entries, meanings and expressions that are typical of the Dutch language as spoken in Belgium.Proverbs and expressions are to be found at a fixed location, i.e. in the entry of the first noun that occurs in the proverb or expression. If there is no noun, the (first) adjectives, and if there is no adjective either, the first verb is the entry where the proverb or expression is to be found. Approximately six months after the publication of the Basis-woordenboek a lesson series on how to use a dictionary will be marketed by the same publisher, specifically aimed at how to use this dictionary.

One of the marks of a good translation is the use of the precise terminology, i.e. that technical terms are translated correctly into the target language or - if there is no direct translation - are paraphrased. With the increasing specialisation of tech-nical fields and the rapid growth in the number of new terms, the search for the correct words tends to be very time-consum-ing. As a result the need is often felt to record equivalent terms for future translation work. For this a terminology data bank can be used, either decentrally on stand-alone equipment or centrally on a mainframe computer.Even if the translator continues for the time being to keep his own collection of terminology on file cards, he can still make use of a generally accessible terminology data bank such as EURODICAUTOM in his search for terms. Unlike a "normal" dictionary, the terminology bank provides additional information such as definitions, relationships between terms and especially the sources of this information. This makes it easier to assess the reliability of the translation given.This article considers the possibilities offered to the translator by a terminology bank and also further developments such as the connection of a data bank to a word processor or a machine translation system. We are now observing the development of a new branch of technology: CAT - Computer Aided Translation.

The official languages of the Communities have to be treated on a basis of equality. Rules and regulations adopted by the Community are valid in all the Member-States. Therefore,every Institution has recourse to a translation service. Translators cannot be familiar with the terminology of all the different areas the Commission deals with. Much time is spent on terminological search work. The sources, however, are often incomplete, outdated, in-accessible or difficult to manage. In order to increase the efficiency of the translation process, a terminological database, called EURODICAUTOM, has been developed by the Commission.Via the computer the user has direct access to all stored terminological information. EURODICAUTOM is a simple, flexible and dynamic system, where the search procedure has been attuned as much as possible to the method used by the translator. The underlying idea is that, if a complete answer to a problem cannot be given, incomplete answers, ordered according to decreasing relevance, can be of some help to the user.The author concludes by saying that terminology is an area, which is always in movement and therefore requires a dynamic, practice-oriented and multi-disciplinary approach.