Notes
-------------------------------------------------------

An interesting point is that ellipsis of VPs seems to be used quite heavily with coordinated NPs:

Source: a:300
Vetted: s
Judgment: g
Phenomena: crd
Weesterlön wel det ja hol en di Aussendörfer
Weesterlön wel-t det ja hol en di Aussen.dörf-en
Weesterlön want-3SG that yes gladly and the outer.village-PL
Weesterlön and the outer villages want it

>> This doesn't look like NP coordination to me.  It looks like
S coordination + VP ellipsis.

There are many complementizers corresponding to the 'wh' words in English that can be used in interrogative clauses, but "weder" (whether) or "of" (also glossed as whether) seem to be the ones that are used for yes-no interrogatives.

>> Words such as 'what', 'when', 'how' etc generally aren't
considered complementizers.  'whether' is though.

In Frisian, the semantic agent is focused by default. There are several methods of shifting focus, including passivization, lexically specified focus on arguments of certain verbs, and defocusing pronouns.

>> Semantic agent or active subject (for most verbs, e.g., not "fo")?

We had some difficulty transferring these observations into the questionnaire, especially since our source was very unclear in the sections that we could find relating to information structure. The semantic agent is usually in preverbal position and usually the focus of the sentence, and given the free word order of everything except the verb, it seems reasonable that the preverbal element is focused. The verb "fo" seems to act something like an overt topic marker where it appears, but it isn't an affix, adposition, or modifier, and contributes to the main semantics of the sentence, so it didn't fit into the topic section of the questionnaire. There was not enough information given in our source for us to say that anything special is done with contrastive focus in Frisian, so we have selected the "My language uses the same position to express contrastive focus as non-contrastive focus" option and left the rest of the focus section blank.

>> Let's consult with Sanghoun and see what he has to say.

A lot of incorrect ambiguity has been added due to what seems to be a bad interaction between V2 word order, subject-verb inversion, and possibly argument optionality. Examples like:

Source: author
Vetted: f
Judgment: g
Phenomena: pronouns
ik iit höm
ik iit höm
I  eat it.OBJ
I eat it

are now getting odd parses with the V OBJ portion being analyzed as a VP with question word order and accepting the true subject of the sentence as a complement to the left of the verb.

>> At least one factor is the ambiguity of höm, which can be
masc or neut --- why not underspecify that?  (You can select
both in the questionnaire, and it will create the disjunctive type
for you, or you could do that by hand.)

Another source of ambiguity is that the 1sg-nonpast-verbal-inflection
LR seems to be optional.  That probably should be an obligatory position
class, shouldn't it?

That leaves just two parses, one of which involves the inversion
LR.  That rule puts the subj on the comps list, so now there are two
complements.   You're right that there shouldn't be that ambiguity,
but I suspect the semantics is correct in both cases.

It seems like what we want here is not the inversion lexical rule for
questions (since you don't need it to get the verb to the left of the
subject) but rather a constraint that associates V-initial position
with questions (and non-V-initial with non-questions).  Have you looked
at Germanic CLIMB yet?  Post to GoPost and we'll see if we can get
Antske to weigh in on this one.

This is causing some overgeneration in examples that are ungrammatical due to agreement as well... the grammar is parsing these examples as having the verb agree with what it thinks is the inverted subject, though it is actually the object, and the disagreement between the true subject and the verb is ignored as the grammar thinks that the subject is just another argument and doesn't require agreement. An example of an ungrammatical sentence that is parsing for this reason:

Source: author
Vetted: f
Judgment: u
Phenomena: agr
dü fo jil
dü fo jil
you.SG get.PRES money
you're getting money

>> Aha --- the inversion lexical rule seems to be providing an "escape hatch"
for uninflected forms.  That is, it's not properly integrated into the 
morphotactics.  (Which is not surprising, the questions library is ancient.)


Unfortunately, the matrix has a bug that causes our indefinite determiner "en" and our conjunction "en" to get smashed together into one (inconsistent) lexical entry:

>> Noted. Thanks.


