A Unified Management and Processing of Word-Forms, Idioms and Analytical Compounds

A UNIFIED MANAGEMENT AND PROCESSING OF WORD-FORMS, IDIOMS AND ANALYTICAL COMPOUNDS Dan Tufts Octav Popescu Research Institute for Informatics Miciurin 8-10, 71316, Bucharest, 1 Fax:653095 Romania ABSTRACT The paper presents a morpho-lexical environ- ment, designed for the management of root- oriented natural language dictionaries. It also encapsulates the basic morpho-lexical processings: analysis and synthesis of individual word-forms or compounds (idioms and analytic constructions). INTRODUCTION lately, a proliferation of computational lexicon environments (CLE) has been noticed, which sign i ftcantly influence the work on natural language (mainly, machine translation) (Byrd et al. 1987), (Nircnburg and Raskin 1987), (Ritchie et al. 1987) etc. With more and more computing power incor- porated, the modern CLEs are capable to process not only individual inflected words or derivatives but also idioms and collocations. Nonetheless, there are many applications in language industry which consider a CLE an unfordable luxury. We believe that such an objection may be refused if the CLE is so designed that it should function in a data-driven manner. We have purposely developed a morpho-lexical management and processing environment aimed at providing an unified and satisfactory solution to a wide range of applications: intelligent text-process- ing, textual information retrieval, natural language interfacing, natural language understanding, ma- chine translation. Also, and more important, the environment is intended to be used for a large class of natural languages (at least for those of which morphology may be described in terms of our para- digmatic model (Tufts 1989)). In order to reach these objectives, we made a clear distinction between the morphological pro. cessings and the knowledge governing them. This distinction Is beneficial not only with respect to natural language independence from the processing environment but also with respect to the d~ired degree of complexity of the process in case. The lack of information in such an approach will not block the system but will produce a simplified result. An interesting characteristic of our system is its capability to treat, besides idioms, analytical com- pounds as well as grammatical and lexical colloca- tions. The work reported here is developed within the context of the paradigmatic theory of morphology as defined in Tufts (1990). The terminology used in the following is taken from the above-mentioned paper. In the same paper, it is shown that learna- bility is the great advantage of paradigmatic mor- phology. The PARADIGM system, described in Tufts (1989) and Tufis (1990) allows a novice user to informally teach the program how to (de)com- pose inflexionai word-forms, that is to enable the morphological processing by a natural language processor. THE MORPHO-LEXICAL KNOWLEDGE BASE Obviously, the main depository of morpho-lexi- cal knowledge is the dictionary, to be discussed in the following. Other morphological knowledge sources are the endings tree and the paradigms table. These data structures do not depend on a specific lexical stock because they encode general linguistic knowledge for the language in ease (parts of speech, relevant categories for the inflexional behaviour, endings, paradigms, etc.). Since their organization and ac- quisition are described elsewhere …

Content truncated to 500 words