Toward an MT System without Pre-Editing: Effects of a New Method in ALT-J/E

[MT Summit III, Washington, DC, USA, July 1991] Toward an MT System without Pre-Editing -Effects of New Methods in ALT-J/E- Satoru IKEHARA, Satoshi SHIRAI, Akio YOKOO and Hiromi NAKAIWA NTT Communications and Information Processing Laboratories 1-2356 Take, Yokosuka-shi, Kanagawa-ken, 238-03 Japan E-mail: ikehara%nttnly.ntt.jp@relay.cs.net Abstract Recently, several types of Japanese to English MT (machine translation) systems have been developed, but prior to using such systems, they have required a pre-editing process of re-writing the original text into Japanese that could be easily translated. For communication of translated information requiring speed in dissemination, application of these systems would necessarily pose problems. To overcome such problems, a Multi-Level Translation Method based on Constructive Process Theory had been proposed. In this paper, the benefits of this method in ALT-J/E will be described. In comparison with the conventional elementary composition method, the Multi-Level Translation Method, emphasizing the importance of the mean- ing contained in expression structures, has been as- certained to be capable of conducting translation ac- cording to meaning and context processing with com- parative ease. We are now hopeful of realizing ma- chine translation omitting the process of pre-editing. 1 Introduction Recently, R&D efforts at machine translation of differing language families, such as Japanese and En- glish, have become popular [Tomabechi 1987, Tomita 1987 and MT Summit-I 1987]. But between such dif- fering language families, differences in perspectives and grasping of objects affect the structuring of ex- pressions. These differences in expression structures make it difficult to convert each other in their ex- isting stale mechanically. For example, in Japanese to English machine translation, the more typical the Japanese expression, the more difficult to translate it into English due to differences in the thought pro- cess. As a means of solving this problem, efforts have been made in the area of limited languages [Na- gao 1985] or knowledge-based translation [Nirenburg 1989]. But under existing circumstances, when using a Japanese to English machine translation system, a Japanese expression requires translation into eas- ily translatable Japanese by human effort. In other words, there is a need to re-write the text into more English type of concept before machine translation can be performed. This action of re-writing is normally known as pre-editing [Nagao 1989], Measures involved in pre- editing include use of a single word so as to mean only one meaning, limiting the method of using “joshi” (Japanese post-positional word), auxiliary verbs and other words likely to be interpreted several ways, to replace, in advance, any words which may have been omitted, re-writing of idiomatic expressions to more general expressions. These all represent efforts to re-write into Japanese expressions which are literally translatable into English. In viewing the problem of pre-editing in Japanese to English translation, the problem would appear theoretically to be closely related to the princi- ple of elementary composition method. The ele- mentary composition method hypothesizes that "the meanings of the entire expression is the sum of the meanings of the various portions of the expression" [Nomoto, 1986]. With existing machine translation …

Content truncated to 500 words