Very poor kana/kanji conversion in Japanese

[cajuncoder]

Anthy seems to be very bad at distinguishing where one word ends and another begins when converting Japanese text, even with very basic phrases. For example, if I type:
よろしくおねがいいたします
When I hit space to convert it to kanji, I get:
よろしく尾根買いいたします
Which is terribly incorrect. The same phrase is correctly converted by mozc, which yields
よろしくお願いいたします

This occurs on every machine and version of anthy I have tried, across many (debian/ubuntu based) distros, including Debian Stretch & Testing/Buster, and Ubuntu 16.04 and 17.10. It is an extremely common occurrence even with simple phrases, and it makes anthy rather frustrating to use.