Machine translation in conversations: does it help or hurt?

Have you ever had a conversation through machine translation? Many problems arise: “computer bug” may become “computer insect,” and complex ideas like  “he asked me to figure this out first when we talked last time” may translate very unpredictably.

In a lab study, we asked a native English speaker and a native Chinese speaker to brainstorm together, some using English as a common language, and others using machine translation.

Machine translation helped the native Chinese speakers produce ideas (Figure 1) but both native and non-native speakers viewed machine-translated messages as less comprehensible than English messages (Figure 2).

Figure 1. Idea productivity by English and Chinese native speakers in MT- and English-mediated groups.

cscw2013_figure3

Figure 2. Perceived message comprehensibility by Americans and Chinese in MT- and English-mediated groups.

This result gives hope for combining machine translation and second language ability to better support inter-lingual communication. It may be effective to use machine translation to help people produce messages in their native languages, while leaving incoming messages untranslated and leveraging people’s second language proficiency for comprehension.

For more, see our full paper to appear at CSCW 2013, Machine Translation vs. Common Language: Effects on Idea Exchange in Cross-Lingual Groups
Hao-Chuan Wang, National Tsing Hua University, Taiwan
Susan Fussell, Cornell University, USA
Dan Cosley, Cornell University, USA