Discourse Issues in the Translation of Japanese Email

The translation of email presents unique challenges to translation systems. At the same time, features peculiar to email constitute new sources of information for these systems. A corpus of Japanese email is examined and its particular characteristics discussed: visual formatting, use and nature of openings and closings, visual devices for capturing aspects of speech (e.g., nonstandard punctuation, spelling, and use of English), and patterns of discourse marker, formal/informal verb form and verb tense use. It is suggested that author-specific features (use of closings, number of sentences, size of paragraphs, and use of visual devices) allow customization of translation. Implications of these features for modifications to translation systems are discussed.