Automation of Bilingual Lexicon Compilation

[MT Summit III, Washington, DC, USA, July 1991] Automation of Bilingual Lexicon Compilation I. Arturo Trujillo* David A. Plowman† Computer Laboratory Computer Laboratory University of Cambridge University of Cambridge Cambridge CB2 3QG, England Cambridge CB2 3QG, England Abstract This paper shows that, there are a number of common concepts which are used to define a class of nouns in standard, monolingual English and Spanish dictionaries. An experiment is de- scribed to show how a small sot of such con- cepts was derived semi-automatically by auto- matically analysing the definitions in each lan- guage and then matching equivalent definitions manually. Also, some of the benefits of con- structing such sets are described, together with the problems encountered while carrying out the experiment. 1 Introduction The bilingual dictionary is a crucial component in all Machine Translation (MT) systems which do not adopt the interlingua strategy. Examples include TAUM- METEO, SYSTRAN, METAL, GETA-AR1ANE, the systems at PAHO, EUROTRA and many others (see [Hutchins, 1986], [Nirenberg, 1987], [Slocum, 1988]). Its construction, however, is time consuming and inefficient, especially when translation between more than two lan- guages is involved. One way to automate bilingual lexi- con compilation is to describe each monolingual entry in terms of a set of common meaning primitives and then use these primitives as a search key to match entries from different languages. Pure meaning primitives are an elusive idea, and there is no agreed methodology on how to search for them. For the purposes of bilingual dictionary construction, it might suffice to have a set of common concepts for which there are equivalent words in most languages, and in terms of which other word definitions might be made. Since these concepts would be common to all languages they could be thought of as an interlingua vocabulary in which monolingual lexicons could be written. For this reason they will be referred to as the Lexicon's Interlin- gua (henceforth LI). The Longman Dictionary of Contemporary English (LDOCE) [Procter, 1978] shows that approximately * This work was funded by an SERC, UK, Advanced Course grant. † Now at Harlequin Ltd., Harrington Hall, Barrington, Cambridge CB2 5RG, England. 2200 basic words might suffice to define all other words in English. Now, comparing LDOCE definitions with those found in the Spanish monolingual dictionary 'VOX Diccionario Manual Ilustrado de la Lengua Española' (VOX) [Gili Gaya, 1988], we find that they share a set, of common concepts. It would therefore be interesting to base our search for these concepts on existing mono- lingual dictionaries. 2 Evidence Before comparing word sense definitions from each dic- tionary two points should be borne in mind. Firstly, LDOCE has been written with the foreign language stu- dent in mind and hence definitions have been made as simple as possible. In contrast, VOX contains word ex- planations aimed at the native user of the language. Sec- ondly, VOX does not claim to use a restricted defining vocabulary, thus making its applicability to natural lan- guage processing less straightforward. The following two definitions …

Content truncated to 500 words