Many statistical machine translators like Google Translate are very sensitive to the availability of bilingual corpus. In this case GT seems to have learned tha

Many statistical machine translators like Google Translate are very sensitive to the availability of bilingual corpus. In this case GT seems to have learned that -子 means either a dumpling (餃子) or a female name ending with -ko, but haven't seen enough corpus to determine that the preceding 淳 is pronounced either Atsu- or Jun- in given context so it is guessing. Combined with the user-contributed corpus this can be rather disastrous: several machine translators had translated Japanese "初音ミク" [1] to Korean "이명박" [2] ;-) [1] https://en.wikipedia.org/wiki/Hatsune_Miku https://en.wikipedia.org/wiki/Hatsune_Miku [2] https://en.wikipedia.org/wiki/Lee_Myung-bak https://en.wikipedia.org/wiki/Lee_Myung-bak