AI to Help Transcribe Tsugaru Dialect into Standard Japanese
Local dialects are plentiful in Japan - the heaviest in Kagoshima, Kyushu, and Tsugaru up north in Aomori. In the feudal period, heavy dialects often served convenient tools to tell strangers apart for security as in Kagoshima where the Lord of Shimazu did utilize the barrier of dialects for self-defense.
Up in Aomori, the Tsugaru dialect is really heavy and difficult for the outsiders to follow to such an extent that hospital staffs away from Aomori have a hard time understanding the complaints of the local patients.
Hisoraki University and the Tohoku Electric Power Co., Inc. are to jointly develop an AI-based sound recognizer to help find a way out of the fix. The joint research is to start August 1 and to come up with findings on January 13, 2018.
Tohoku Electric will provide data collected at the call center for analysis via the university's AI/language know-how. Findings are expected to help provide means to expedite much-needed communication fluency with the patients.
"Dame" or "not good" in the standard Japanese sounds "mane" in Tsugaru dialect; "aruku" or "walk sounds "asagu", etc. - much too far apart to be recognizable.
The joint study will involve 12 Tsugaru dialect speakers of ages 20-70 from an area called Ajigasawa where the dialect is the heaviest of all Aomori localities. The samples will be processed by the AI sound recognizer with extra language processing know-how into a standard Japanese text and further automatically into summaries.
Hirosaki University had first asked Tohoku Electric to join the program. The university hospital has long been under stress in the communication with the patients as quite a few of the hospital staffs are from outside Aomori and it was essential that the doctors and nurses prepare summaries of their communication with the patients.
Tohoku Electric's call center has had its share of the similar communication barrier with 1.5 million inquiries in Tsugaru dialect a year. (Nathan Shiga)
Source: Nikkei