Academic journal article Journal of International Technology and Information Management

Machine Translation in a Multilingual Electronic Meeting

Academic journal article Journal of International Technology and Information Management

Machine Translation in a Multilingual Electronic Meeting

Article excerpt

INTRODUCTION

As global trade and travel continually increase, there is a growing need to support communication among people from different countries. However, human interpreters are difficult to schedule and expensive, and as a result, only a small percentage of spoken encounters are interpreted (Fugen, Waibel, & Kolss, 2007).

In a multilingual setting, group members can try to use a common language (e.g., English) or use interpreters to speak the equivalent comment in other participants' languages. Alternatively, group members can participate in an electronic meeting (e.g., an electronic chat room) that provides participants with a means of sharing comments anonymously and simultaneously (Hung, Tang, & Shu, 2008). These electronic meetings have been shown to be superior to traditional, oral discussions in many cases (Chen, Ramano, & Nunamaker, 2006; Dennis & Valacich, 1993), and there is a demand for multilingual support in these settings (Lim & Yang, 2008). Using such a system, group members can type comments in their own native language and use an interpreter or machine translation (MT) to understand others' comments, type comments and have staff members provide the translations, or type comments and have completely automated translation, as this paper proposes. After a case study showing the problems encountered with staff supported translations, this paper describes a new, locally developed multilingual electronic meeting system that provides automatic translation among 41 languages. Early results show that for many languages, the system will provide highly comprehensible translations.

INTERPRETATION AND TRANSLATION IN MULTILINGUAL MEETINGS

The vast majority of multilingual meetings are interpreted, that is, a human hears spoken words in one language and speaks them again in one or more other languages (Nolan, 2005). Interpretations can be whispered to a few participants if the majority of a group shares a language and a small number do not speak it. Alternatively, sound-proof booths can be set up that allow linguists to listen to a speaker and interpret into a microphone for the target audience.

Electronic meetings are often more productive than traditional, oral meetings when eight or more people need to share ideas anonymously (Gallupe, Dennis, Cooper, Valacich, Bastianutti, & Nunamaker, 1992; Fjermestad, 2004; Kepuska, Gurbuz, Rodriguez, Fiore, Carstens, Converse, & Metcalf, 2008), and in a multilingual setting, machines can assist with online, interactive translation (O'Hagan & Ashworth, 2002). Some researchers believe that MT can already deliver satisfactory accuracy in an electronic meeting, and even if not perfect, human real-time performance has probably reached its peak, leaving room for MT to catch up with or even surpass that of linguists (Fugen et al., 2007; Strong, Ghosh, & Conlon, 2008). Even if not correct, perhaps most of the key words in a foreign comment can be translated accurately enough for a basic understanding of its meaning (Sprung, 2000), and a group member can use the gist of the text to determine if an idea should be ignored or pursued further (Somers, 2003). Finally, translation in an electronic meeting might be more accurate than interpretation in an oral discussion because in the latter, group members might speak too softly, slur words, or otherwise be difficult to understand.

One way of providing interactive translation in an electronic meeting is with a linguist reading online comments while dictating to a typist. However, once there are more than five or six participants with two or three languages, the human translator will likely be overwhelmed, even with the assistance of MT (O'Hagan & Ashworth, 2002). This maximum group size that one translator can support is likely to be even smaller when languages such as Japanese and Chinese are used because of the delay experienced entering the characters through a standard QWERTY keyboard. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.