Page 1 of 2
Where can't a verb be?
Posted: April 5th, 2017, 7:28 am
by Alan Bunning
What I am learning from my Matthew 1:20 post, is that a verb can come just about anywhere (even between a noun and its adjective inside a prepositional phrase). In English, the verb often (but not always) serves as a nice boundary between the subject and the object, but in Greek I can’t tell if it serves as a boundary for anything (other than being in the right sentence). So now I am wondering if there is anywhere a verb can’t be?
1. Maybe not between an article and its noun?
2. ?
Re: Where can't a verb be?
Posted: April 5th, 2017, 9:19 am
by Jonathan Robie
Alan Bunning wrote: ↑April 5th, 2017, 7:28 am
What I am learning from my Matthew 1:20 post, is that a verb can come just about anywhere (even between a noun and its adjective inside a prepositional phrase). In English, the verb often (but not always) serves as a nice boundary between the subject and the object, but in Greek I can’t tell if it serves as a boundary for anything (other than being in the right sentence). So now I am wondering if there is anywhere a verb can’t be?
1. Maybe not between an article and its noun?
2. ?
Yes. For instance, I think this would be ungrammatical: * τὸ γὰρ ἐν αὐτῇ γεννηθὲν ἐκ ἐστιν πνεύματός ἁγίου. But perhaps the most important thing to notice is this: English sentences have verb phrases, and the position of the verb is important for recognizing them. Greek does not have verb phrases, it has clauses that are governed by a verb, and the position of the verb within a clause is nowhere near as fixed as it is in English.
But there are an awful lot of places where verbs can be, and the rules are underspecified in most grammars. I know you are a computer scientist - so am I - so let me use a little insider language. We'd both love it if someone would just show us the formal grammar of the Greek language. We don't have that yet. But we do have the equivalent of the parse tree - that's what treebanks do. The treebanks are not yet perfect, but they are pretty good, and they are being improved. And you can download them, query them, and look for examples of various constructions. If you have a set of parse trees, you can infer a grammar. Currently, the grammars that can be inferred are well short of 100%, though, even for English.
Even in English, you can stick a verb between an article and its noun if you do it the right way: "the churning sea", "the boiling water". And I bet most native English speakers couldn't easily tell you all the cases that can come up in English. One of the things we do on B-Greek is ask questions about these rules and see if we can find the answers. There are certainly constructions in which a Greek verb occurs between an article and a noun ... we can easily find some examples, but I don't know that anyone has an exhaustive list of the rules yet. If you have the energy, you could use the data in our treebanks to generate rules that govern the instances we have. It would take some time. It might make more sense a year from now, since we are improving some important aspects of one set of treebanks between now and then.
Re: Where can't a verb be?
Posted: April 5th, 2017, 10:52 am
by MAubrey
I think it might be good to switch our point of view. In terms of the "rules" it's better to say that the verb itself is more or less fixed in its position and that the other constituents are moved around it.
When the verb is between the noun and the adjective, the verb hasn't been moved to inside that phrase. Instead that phrase has been split around the verb.
This is an important distinction if we take seriously the idea that the structure of a sentence has a definable relationship with the meaning of the sentence. And when we get an ordering like:
Adj - V - Noun
Noun - V - Adj
We can see that it's the meaning of the Noun Phrase that's being affected, not the verb (though of course, the significance is likely going to be debated for the time being. The verb continues to predicate normally without any difference.
English instances of verbs appearing between nouns and their modifiers are of a different kind though:
"the churning sea"
Art - V - Noun
"the sea churned"
Art - Noun - V
Here the change is in the verb, rather than the noun and article. In the first, the verb is no longer predicating and as such there is proposition made and nothing is asserted, while in the second with the normal order, we do have a proposition.
These are actually the sorts of tests that we use in linguistics for proving constituency. And here, the English variation is evidence that the article and the noun form a unit of their own.
Jonathan Robie wrote: ↑April 5th, 2017, 9:19 am
Currently, the grammars that can be inferred are well short of 100%, though, even for English.
If it wasn't for world Englishes, we'd probably be in the low 90's.
Re: Where can't a verb be?
Posted: April 5th, 2017, 11:18 am
by Alan Bunning
Jonathan Robie wrote: ↑April 5th, 2017, 9:19 am
There are certainly constructions in which a Greek verb occurs between an article and a noun ... we can easily find some examples, but I don't know that anyone has an exhaustive list of the rules yet.
I would like to see an example of that, and if possible an example of it where it is not a participle.
Re: Where can't a verb be?
Posted: April 5th, 2017, 12:33 pm
by Jonathan Robie
Alan Bunning wrote: ↑April 5th, 2017, 11:18 am
Jonathan Robie wrote: ↑April 5th, 2017, 9:19 am
There are certainly constructions in which a Greek verb occurs between an article and a noun ... we can easily find some examples, but I don't know that anyone has an exhaustive list of the rules yet.
I would like to see an example of that, and if possible an example of it where it is not a participle.
Participles are, of course, common in this kind of construction. This query looks for 3-word noun phrases with a determiner where one of the words is a participle:
Code: Select all
for $np in //wg[@class='np' and w[@class='det']]
where count($np//w) = 3
and $np//w[@class='verb' and @mood = 'participle']
return
<match>
{ $np/ancestor::sentence/p }
{ $np }
</match>
I found 366 of these, e.g. Ἰησοῦς
ὁ λεγόμενος Χριστός. I have not read through them and verified them, there may be false positives.
Let's try it with infinitives:
Code: Select all
for $np in //wg[@class='np' and w[@class='det']]
where count($np//w) = 3
and $np//w[@class='verb' and @mood = 'infinitive']
return
<match>
{ $np/ancestor::sentence/p }
{ $np }
</match>
I found 60 of these, e.g. μέλλει γὰρ Ἠρῴδης ζητεῖν τὸ παιδίον
τοῦ ἀπολέσαι αὐτό.
For indicative, I found only 1: ὁ δὲ Ἰησοῦς εἶπεν αὐτῷ
Τὸ Εἰ δύνῃ, πάντα δυνατὰ τῷ πιστεύοντι. A false positive because it is a substantized verb, but still interesting word order.
For subjunctive, I get another false positive because it is a substantized verb, but an interesting one for word order: Luke.19.48 καὶ οὐχ εὕρισκον
τὸ τί ποιήσωσιν· ὁ λαὸς γὰρ ἅπας ἐξεκρέμετο αὐτοῦ ἀκούων.
No imperatives or optatives.
Re: Where can't a verb be?
Posted: April 5th, 2017, 7:32 pm
by MAubrey
Jonathan Robie wrote: ↑April 5th, 2017, 12:33 pmLet's try it with infinitives:
...
I found 60 of these, e.g. μέλλει γὰρ Ἠρῴδης ζητεῖν τὸ παιδίον
τοῦ ἀπολέσαι αὐτό.
This looks like a regular articular infinitive.
Jonathan Robie wrote: ↑April 5th, 2017, 12:33 pm
For indicative, I found only 1: ὁ δὲ Ἰησοῦς εἶπεν αὐτῷ
Τὸ Εἰ δύνῃ, πάντα δυνατὰ τῷ πιστεύοντι. A false positive because it is a substantized verb, but still interesting word order.
For subjunctive, I get another false positive because it is a substantized verb, but an interesting one for word order: Luke.19.48 καὶ οὐχ εὕρισκον
τὸ τί ποιήσωσιν· ὁ λαὸς γὰρ ἅπας ἐξεκρέμετο αὐτοῦ ἀκούων.
No imperatives or optatives.
Arguably, you could group these with the articular infinitives that are complements of a verb and treat the article as a complementizer.
I think that'd be a reasonable rule for regularizing the data.
Re: Where can't a verb be?
Posted: April 6th, 2017, 1:17 am
by Stephen Hughes
It is assumed that a verb cannot be between a preposition and its noun. If a preposition immediately precedes a verb, it is assumed to be prefixed to it, not directly related to another nominal element.
Re: Where can't a verb be?
Posted: April 6th, 2017, 7:30 am
by Jonathan Robie
MAubrey wrote: ↑April 5th, 2017, 7:32 pm
Jonathan Robie wrote: ↑April 5th, 2017, 12:33 pmLet's try it with infinitives:
...
I found 60 of these, e.g. μέλλει γὰρ Ἠρῴδης ζητεῖν τὸ παιδίον
τοῦ ἀπολέσαι αὐτό.
This looks like a regular articular infinitive.
Yeah, and a lot of the results I got for infinitives follow this pattern. So I should have used a query that looks at the structure, instead of lazily looking for noun phrases with three words. Here's the structure of that:

- Screen Shot 2017-04-06 at 7.17.24 AM.png (14.47 KiB) Viewed 4237 times
I get a lot with that structure that end in the pronoun. Here's another structure I get:

- Screen Shot 2017-04-06 at 7.20.25 AM.png (10.39 KiB) Viewed 4237 times

- Screen Shot 2017-04-06 at 7.22.19 AM.png (14.02 KiB) Viewed 4237 times
MAubrey wrote: ↑April 5th, 2017, 7:32 pm
Arguably, you could group these with the articular infinitives that are complements of a verb and treat the article as a complementizer.
I think that'd be a reasonable rule for regularizing the data.
I think the underlying data is fairly regular here - here's that last structure. But now that Lowfat is decoupled from GBI, we can make changes as we see fit.

- Screen Shot 2017-04-06 at 7.28.23 AM.png (10.38 KiB) Viewed 4237 times
Re: Where can't a verb be?
Posted: April 6th, 2017, 8:25 pm
by Stephen Carlson
MAubrey wrote: ↑April 5th, 2017, 10:52 am
I think it might be good to switch our point of view. In terms of the "rules" it's better to say that the verb itself is more or less fixed in its position and that the other constituents are moved around it.
This is a good way to put it. I should clarify that this basically applies to non-enclitic finite verbs. Enclitic verbs pattern with enclitics, and participles and infinitives pattern with nouns in some respect.
Re: Where can't a verb be?
Posted: April 6th, 2017, 8:35 pm
by Jonathan Robie
Stephen Carlson wrote: ↑April 6th, 2017, 8:25 pm
MAubrey wrote: ↑April 5th, 2017, 10:52 am
I think it might be good to switch our point of view. In terms of the "rules" it's better to say that the verb itself is more or less fixed in its position and that the other constituents are moved around it.
This is a good way to put it. I should clarify that this basically applies to non-enclitic finite verbs. Enclitic verbs pattern with enclitics, and participles and infinitives pattern with nouns in some respect.
Can you point me to some things I could search for in the treebanks to see the difference between enclitic verbs and non-enclitic verbs? Or an article or somesuch that describes this?