Modeling Impression in Probabilistic Transliteration into Chinese

Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing (EMNLP 2006), pages 242–249, Sydney, July 2006. c⃝2006 Association for Computational Linguistics Modeling Impression in Probabilistic Transliteration into Chinese LiLi Xu∗ Atsushi Fujii Graduate School of Library, Information and Media Studies University of Tsukuba 1-2 Kasuga, Tsukuba, 305-8550, Japan fujii@slis.tsukuba.ac.jp Tetsuya Ishikawa The Historiographical Institute The University of Tokyo 3-1 Hongo 7-chome, Bunkyo-ku Tokyo, 133-0033, Japan ishikawa@hi.u-tokyo.ac.jp Abstract For transliterating foreign words into Chi- nese, the pronunciation of a source word is spelled out with Kanji characters. Be- cause Kanji comprises ideograms, an indi- vidual pronunciation may be represented by more than one character. However, because different Kanji characters convey different meanings and impressions, char- acters must be selected carefully. In this paper, we propose a transliteration method that models both pronunciation and im- pression, whereas existing methods do not model impression. Given a source word and impression keywords related to the source word, our method derives possible transliteration candidates and sorts them according to their probability. We evalu- ate our method experimentally. 1 Introduction Reflecting the rapid growth of science, technology, and economies, new technical terms and product names have progressively been created. These new words have also been imported into different languages. There are three fundamental methods for importing foreign words into a language. In the first method—translation—the meaning of the source word in question is represented by an existing or new word in the target language. In the second method—transliteration—the pronunciation of the source word is represented by using the phonetic alphabet of the target language, such as Katakana in Japanese and Hangul in Ko- rean. ∗ This work was done when the first author was a grad- uate student at University of Tsukuba, who currently works for Hitachi Construction Machinery Co., Ltd. In the third method, the source word is spelled out as it is. However, the misuse of this method decreases the understandability and readability of the target language. While translation is time-consuming, requiring selection of an existing word or generation of a new word that correctly represents the meaning of the source word, transliteration can be performed rapidly. However, the situation is complicated for Chinese, where a phonetic alphabet is not used and Kanji is used to spell out both conventional Chi- nese words and foreign words. Because Kanji comprises ideograms, an in- dividual pronunciation can potentially be repre- sented by more than one character. However, if several Kanji strings are related to the same pro- nunciation of the source word, their meanings will be different and will therefore convey different im- pressions. For example, “Coca-Cola” can be represented by different Kanji strings in Chinese. The offi- cial transliteration is “����”, which comprises “�� (tasty)” and “�� (pleasant)”, and is there- fore associated with a positive connotation. However, there are a number of Kanji strings that represent similar pronunciations to that of “Coca-Cola”, but which are associated with in- appropriate impressions for a beverage, such as “����”. This word includes “��”, which …

Content truncated to 500 words