In this paper, wepresent a prototype MTsystem named SYUWANwhich can translate Japanese into Japanesesign language. Oneof the most important problemsin this translation is that there are very few entries in a sign languagedictionary, comparedwith Japaneseone. To solve this problem, whenthe original input worddoes not exist in a sign language dictionary, SYUWANapplies sometechniquesto find a similar wordfrom a Japanese dictionary and substitutes this word for the original word. As the result, SYUWAN can translate up to 95%of wordswhichare morphologically analyzed. Computer