bansp
commented
1 year ago I find this to be an interesting didactic case, for when one wants to show students how lexicography differs from linguistics. Intuitively, for a linguist, conflating (let's call it) 'quantitative' valency with 'qualitative' valency is a bad move. (The former having to do with argument slots for a predicate, the latter with the semantics of (some of) those arguments).
BUT I can't honestly say it's an impractical lexicographic approach. If the Wikipedia list of suggested values is used:
impersonal, intransitive, transitive, ditransitive, tritransitive, I can see that working in a dictionary (although my first reaction to seeing that list was that it represents some kind of didactic approach dressed up as general truth).
I'm not sure how reflexive would work with this, but it becomes a matter of convention and of documenting that convention. (My questions would be, e.g., whether the same tag would be used for He perjured himself and He killed himself -- in a dictionary, maybe yes; or whether he gave himself a smack is an issue -- probably not, if you can stack those tags, as suggested above, and if documentation is provided).
There is a separate property "transitivity" in lex0. We also need to encode properties like "impersonal", which are traditionally grouped under "valency". So we would be inclined to encode intransitive "rain" as
Reflexivity could also be added here.
Cf: https://en.wikipedia.org/wiki/Valency_(linguistics)