Lexical Morphology in Machine Translation: A Feasibility Study

Proceedings of the 12th Conference of the European Chapter of the ACL, pages 130–138, Athens, Greece, 30 March – 3 April 2009. c⃝2009 Association for Computational Linguistics Lexical Morphology in Machine Translation: a Feasibility Study Bruno Cartoni University of Geneva cartonib@gmail.com Abstract This paper presents a feasibility study for im- plementing lexical morphology principles in a machine translation system in order to solve unknown words. Multilingual symbolic treat- ment of word-formation is seducing but re- quires an in-depth analysis of every step that has to be performed. The construction of a prototype is firstly presented, highlighting the methodological issues of such approach. Sec- ondly, an evaluation is performed on a large set of data, showing the benefits and the limits of such approach. 1 Introduction Formalising morphological information to deal with morphologically constructed unknown words in machine translation seems attractive, but raises many questions about the resources and the prerequisites (both theoretical and practi- cal) that would make such symbolic treatment efficient and feasible. In this paper, we describe the prototype we built to evaluate the feasibility of such approach. We focus on the knowledge required to build such system and on its evalua- tion. First, we delimit the issue of neologisms amongst the other unknown words (section 2), and we present the few related work done in NLP research (section 3). We then explain why implementing morphology in the context of ma- chine translation (MT) is a real challenge and what kind of aspects need to be taken into ac- count (section 4), and we show that translating constructed neologisms is not only a mechanical decomposition but requires more fine-grained analysis. We then describe the methodology de- veloped to build up a prototyped translator of constructed neologisms (section 5) with all the extensions that have to be made, especially in terms of resources. Finally, we concentrate on the evaluation of each step of the process and on the global evaluation of the entire approach (sec- tion 6). This last evaluation highlights a set of methodological criteria that are needed to exploit lexical morphology in machine translation. 2 Issues Unknown words are a problematic issue in any NLP tool. Depending on the studies (Ren and Perrault 1992; Maurel 2004), it is estimated that between 5 and 10 % of the words of a text writ- ten in “standard” language are unknown to lexi- cal resources. In a MT context (analysis-transfer- generation), unknown words remain not only unanalysed but they cannot be translated, and sometimes they also stop the translation of the whole sentence. Usually, three main groups of unknown words are distinguished: proper names, errors, and ne- ologisms, and the possible solution highly de- pends on the type of unknown word to be solved. In this paper, we concentrate on neologisms which are constructed following a morphological process. The processing of unknown “constructed ne- ologisms” in NLP can be done by simple guess- ing (based on the sequence of final letters). This option can be efficient enough when the task …

Content truncated to 500 words