Translation using Information on Dialogue Participants

Translation using Information on Dialogue Participants Setsuo Yamada, Eiichiro Sumita and Hideki Kashioka ATR Interpreting Telecommunications Research Laboratories* 2-2, Hikaridai, Seika-cho, Soraku-gun, Kyoto, 619-0288, JAPAN { syamada, sumita, kashioka} @itl.atr.co.jp t Abstract This paper proposes a way to improve the trans- lation quality by using information on dialogue participants that is easily obtained from out- side the translation component. We incorpo- rated information on participants' social roles and genders into transfer rules and dictionary entries. An experiment with 23 unseen dia- logues demonstrated a recall of 65% and a preci- sion of 86%. These results showed that our sim- ple and easy-to-implement method is effective, and is a key technology enabling smooth con- versation with a dialogue translation system. 1 Introduction Recently, various dialogue translation systems have been proposed (Bub and others, 1997; Kurematsu and Morimoto, 1996; Rayner and Carter, 1997; Ros~ and Levin, 1998; Sumita and others, 1999; Yang and Park, 1997; Vi- dal, 1997). If we want to make a conversation proceed smoothly using these translation sys- tems, it is important to use not only linguis- tic information, which comes from the source language, but also extra-linguistic information, which does not come from the source language, but, is shared between the participants of the conversation. Several dialogue translation methods that use extra-linguistic information have been pro- posed. Horiguchi outlined how "spoken lan- guage pragmatic information" can be trans- lated (Horiguchi, 1997). However, she did not apply this idea to a dialogue translation system. LuperFoy et al. proposed a software architec- *Current affiliation is ATR Spoken Language Trans- lation Research Laboratories Current mail addresses are { setsuo.yarnada, eiichiro.sumita, hideki.kashioka} @slt. atr. co.jp ture that uses '% pragmatic adaptation" (Lu- perFoy and others, 1998), and Mima et al. pro- posed a method that uses "situational informa- tion" (Mima and others, 1997). LuperFoy et al. simulated their method on man-machine inter- faces and Mima et al. preliminarily evaluated their method. Neither study, however, applied its proposals to an actual dialogue translation system. The above mentioned methods will need time to work in practice, since it is hard to obtain the extra-linguistic information on which they depend. We have been paying special attention to "po- liteness," because a lack of politeness can inter- fere with a smooth conversation between two participants, such as a clerk and a customer. It is easy for a dialogue translation system to know which participant is the clerk and which is the customer from the interface (such as the wires to the microphones). This paper describes a method of "polite- ness" selection according to a participant's so- cial role (a clerk or a customer), which is eas- ily obtained from the extra-linguistic environ- ment. We incorporated each participant's so- cial role into transfer rules and transfer dictio- nary entries. We then conducted an experiment with 23 unseen dialogues (344 utterances). Our method achieved a recall of 65% and a preci- sion of 86%. These rates could be improved to 86% and 96%, respectively (see Section 4). It …

Content truncated to 500 words