Terminological Ontology and Cognitive Processes in Translation

Terminological Ontology and Cognitive Processes in Translation* Fumiko Kano Glückstad Dept. of Intl. Language Studies and Computational Linguistics, Copenhagen Business School Dalgas Have 15, DK-2000 Frederiksberg, Denmark fkg.isv@cbs.dk Abstract. This work is based on a terminological ontology method proposed by Madsen et al. (2004; 2005). The main purpose is to bridge a semantic relation between domain specific terms in two languages by mapping two language-dependent terminological ontologies. To explain why this method is preferable from the view of the cognitive process in translation, the terminological ontology method is contrasted with a study made by De Groot (1997) in the area of bilingual memory research. Keywords: terminology, ontology, bilingual memory, ontology matching, translation * Acknowledgement: I would like to thank my supervisor, Hanne Erdman Thomsen, for helpful advices on my project. Copyright 2010 by Fumiko Kano Glückstad 1 Introduction Imagine a situation where non-English speaking European and Asian are debating in English about the issue of the academic degree systems in their respective countries. While a German might be explaining about the Doctor of Science (Habilitation) as their highest academic degree, a Japanese might be having the highest academic degree in Japan - Ph.D. level - in his mind. This imagined conversation shows a typical situation revealing a deep inherent misconception between the two parties since each of them has their own conceptual - and correct - understanding of the highest obtainable academic degree in their respective countries. Statistical Machine Translation (SMT) is one of the most widespread machine translation approaches. However, when considering rare language-combinations, the lack of “direct” parallel corpora is a critical concern for successfully applying the SMT approach. In case of the rare language-combination: Danish and Japanese, it is possible to collect a reasonably sufficient amount of corpora between Danish and English as well as English and Japanese. However, the availability of bilingual corpora that directly link Danish and Japanese are very limited. Hence, a transitive translation technique using English as a pivot language is often employed in translation of rare language-combinations. A question is how well the transitive machine translation approaches can convey original conceptual meanings of Source Language (SL) words into Target Language (TL) translations in the transitive translation. When considering human translations between rare combinations of languages, human translators typically look up SL-English dictionaries to identify meanings of SL words and then use English-TL dictionaries to determine appropriate TL translations. In such situations, the problem of word sense ambiguation becomes critical. Since a lexical representation often carries several meanings, the transitive translation approach using English as pivot simply amplifies the probability of selecting an inappropriate sense in a TL. Thus the problem of polysemy becomes especially serious in the process of transitive translations. PACLIC 24 Proceedings 629 This present work is primarily based on a terminological ontology method (Madsen et al. 2004; 2005), a domain-specific ontology which is used in standardized terminology works such as ISO 704 (2000). Since the traditional study of terminology is widely described as the standardization of terminology, a concept …

Content truncated to 500 words