>Currently released model supports most of the languages except East Asian (Chinese Simplified, Traditional, Japanese, Korean, Thai). You should expect good res

>Currently released model supports most of the languages except East Asian (Chinese Simplified, Traditional, Japanese, Korean, Thai). You should expect good results if a language uses space as a main token delimitter. Any plans to change this? What does MS fall back on for these languages?