I'm doing some work on my verb inflection script at
This uses a list of verbs extracted from JMdict, and I was trying to overhaul the parsing software from sixteen years ago when I noticed the following.
In JMdict.xml, we find <pos>&v5n</pos> marking all verbs which end with -nu such as "shinu" and a few other bits and bobs. There is also a part of speech (<pos>&vn;</pos>) for "irregular nu verbs". This tag is used to mark shinu but not 焼け死ぬ for example:
<keb>死ぬ</keb>
...
<pos>&v5n;</pos>
<pos>&vn;</pos>
<keb>焼け死ぬ</keb>
...
<pos>&v5n;</pos>
<pos>&vi;</pos>
I don't know what is irregular about shinu compared to other verbs ending in nu, and it seems odd that even other versions of shinu don't have the &vn;:
<keb>凍え死ぬ</keb>
...
<pos>&v5n;</pos>
<pos>&vi;</pos>
<keb>4ぬ</keb>
...
<pos>&v5n;</pos>
<pos>&vi;</pos>
I have looked for some documentation by searching this mailing list for "irregular nu verb" and by looking at
but I could not see very much description.