UniversalDependencies / UD_Irish-IDT

Irish data
Other
6 stars 7 forks source link

Features added for 1600 words #68

Closed kscanne closed 3 years ago

kscanne commented 3 years ago

Added features only in cases where it occurs for all instances of the given (surface form, lemma, POS) triple in my lexicon. So Gen=Masc is ok for (bhfear, fear, NOUN) but not Number=Sing or Case=NomAcc since it could be genitive plural.

The vast majority of these changes should be uncontroversial, although my comments in issue #63 arise a few times in the context of nouns that are "common in form, genitive in function" as in the old Christian Brothers grammar on p.30. See the Case=NomAcc assigned to "Chathair" in sentence 3434 (carn ainmfhocal iontach, dála an scéil: "láithreán gréasáin Chomhairle Chathair Bhaile Átha Cliath") or to "Bhéal" in "Gaeilge Bhéal Feirste", etc.

Other examples mentioned in #63 are the ones that might be considered errors, e.g. the "ar son a ndíograis" example in sentence 3130. This patch assigns NomAcc because of the surface form.