Underuse of "acl" in Irish data

[tlynn747]

While acl:relcl is prevalent in the Irish data, "acl" is rarely used.

In fact, they're usually labelled as xcomp:pred (e.g. predicative adjectives) as though there was a substantive verb present

sent_id = 2071 text = Chaith sé bunús deich momaite ag breathnú muid, níor chonaic sé na haghaidheanna smeartha le snas roimhe.

Could be hard to find (and likely to be rare) but would be good to identify a pattern to identify them.

[tlynn747]

Would "arna" constructions be candidates for acl?

According to Teanglann: arna, compound of AR2 and poss. a. A5. (Used with vn.) 1. On his, her, its, their, having been . . . . ~ chríochnú dom, when I had completed it. ~ fhoilsiú ag, published by. 2. Geom:~ leanúint, produced. S.a. MÁRACH.

Previously, we have been using the nmod label to attach the verbal noun to a previous noun in the sentence. In this example below, eisiúint attaches to deimnithe as an nmod.

E.g. sent = 610 deimhnithe imréitigh cánach arna eisiúint ag na Coimisinéirí Ioncaim.

sent 240 aon rialachán arna dhéanamh faoin alt seo 'any rule made under this article' ?

[tlynn747]

Similarly

Some instances of xcomp:pred could perhaps fit the UD description of the regular ACL label

https://universaldependencies.org/u/dep/all.html#acl-clausal-modifier-of-noun-adjectival-clause

e.g.

sent_id = 2071 text = Chaith sé bunús deich momaite ag breathnú muid, níor chonaic sé na haghaidheanna smeartha le snas roimhe.

[kscanne]

There's another unusual case that should probably fall within the scope of this issue; these are examples that look like the "content clauses" described in the UD guidelines for acl. These are rare in traditional Irish, but seem to be growing more frequent, presumably under the influence of English. Here are some candidates in the treebank (though I didn't search thoroughly):

Sentence 740: "...iarratais go gcuirfí buíon neamhspleách i mbun an fhiosraithe" (requests that an independent team be put in charge of the inquiry) Sentence 764: "...an teoraic nach raibh gá níos mó le haon chaiticeasma" (the theory that there was no longer a need for any catechism) Sentence 3233: "...fianaise go bhfuil ár straitéis ... ag dul chun cinn go maith" (evidence that our strategy... is proceeding well)

In these cases the VERB is currently ccomp of the head noun. Should be acl if I'm understanding the UD recommendation correctly.

[tlynn747]

Updated most instances of "arna X" to acl. Ones remained unchanged include those that are arguments of a verb and not a noun

[tlynn747]

There's another unusual case that should probably fall within the scope of this issue; these are examples that look like the "content clauses" described in the UD guidelines for acl. These are rare in traditional Irish, but seem to be growing more frequent, presumably under the influence of English. Here are some candidates in the treebank (though I didn't search thoroughly):

Sentence 740: "...iarratais go gcuirfí buíon neamhspleách i mbun an fhiosraithe" (requests that an independent team be put in charge of the inquiry) Sentence 764: "...an teoraic nach raibh gá níos mó le haon chaiticeasma" (the theory that there was no longer a need for any catechism) Sentence 3233: "...fianaise go bhfuil ár straitéis ... ag dul chun cinn go maith" (evidence that our strategy... is proceeding well)

In these cases the VERB is currently ccomp of the head noun. Should be acl if I'm understanding the UD recommendation correctly.

Have addressed these and similar. Used GREW-MATCH to find ccomps attached to NOUNs. Many were valid (Verbal nouns) but many similar to above examples also found.

Need to further discuss the xcomp:pred for before next release as I'm still not clear as to whether they are adjectival (which they would be in English) or like relative clauses without relative particle+Verb atá.