Kazuma Hatano
Abstract
The growing role of technology in society, particularly machine translation, has grown due to its enhanced accuracy. Yet, communication quality depends not only on translation precision but also on user factors. It is vital to understand communication characteristics when using machine translation for ...
Read More
The growing role of technology in society, particularly machine translation, has grown due to its enhanced accuracy. Yet, communication quality depends not only on translation precision but also on user factors. It is vital to understand communication characteristics when using machine translation for its future application in daily life, business, and education. This study investigates these characteristics in face-to-face communication using consecutive speech machine translation technology. Participants were international students at a Japanese university who conversed using speech machine translation. Despite mistranslations, they continued to communicate, either understanding most of the message or giving up due to incomprehension. Notably, the turn-taking order was alternated mechanically, even when there was a long wait for the other person to speak. Unlike previous studies, gestures were not used to overcome communication barriers due to mistranslations; rephrasing was preferred. This study proposes the concept of “speech production load,” the burden of device operation and translation waiting time, as a determi-nant of the relationship between a speech machine translation system and user communication behavior in the context of modern technology.