List of Articular Infinitives

Post Reply
Jonathan Robie
Posts: 3743
Joined: May 5th, 2011, 5:34 pm
Location: Durham, NC
Contact:

Re: List of Articular Infinitives

Post by Jonathan Robie » March 7th, 2015, 5:47 pm

I think these three queries get at least the vast majority of instances, and they can be combined using an | operator:

Code: Select all

//w[@mood='infinitive' and @head='true']/preceding-sibling::w[@class='det'] 
|
//w[@mood='infinitive' and @head='true']/ancestor::wg[@class='cl'][1]/preceding-sibling::w[@class='det']
|
//w[@mood='infinitive' and @head='true']/ancestor::wg[@class='cl'][1]/parent::wg[@class="cl" and count(*) = count(wg[@class="cl"] | w[@class='conj'])]/preceding-sibling::w[@class='det']
They do include some false positives, so you have to read through the results. I suspect that with a little attention to the roles of the clauses, or perhaps to verbs like μέλλω and ἄρχω, these results could be improved to remove the majority of the false positives.
0 x


ἐξίσταντο δὲ πάντες καὶ διηποροῦντο, ἄλλος πρὸς ἄλλον λέγοντες, τί θέλει τοῦτο εἶναι;
http://jonathanrobie.biblicalhumanities.org/

Stephen Carlson
Posts: 3021
Joined: May 11th, 2011, 10:51 am
Location: Melbourne
Contact:

Re: List of Articular Infinitives

Post by Stephen Carlson » March 7th, 2015, 7:17 pm

Thanks for keeping at it, Jonathan.

A couple of points/questions:

1. How can I make my own queries? Is it online or must something be downloaded? Or is it not yet ready for public release?

2. The determiner of an articular infinitive has to be neuter, so things like ὁ μέλλων ἔρχεσθαι should not have hit. Is there a way to query gender?

3. It looks like the structure of articular infinitives is not consistently coded in the database. Does this means that some of the effort involves basically reverse engineering the variety of strategies that were used to parse the text? I don't know who, how many, or according to what standard the syntax trees were done.
0 x
Stephen C. Carlson, Ph.D.
Melbourne, Australia

Jonathan Robie
Posts: 3743
Joined: May 5th, 2011, 5:34 pm
Location: Durham, NC
Contact:

Re: List of Articular Infinitives

Post by Jonathan Robie » March 7th, 2015, 8:23 pm

Stephen Carlson wrote:Thanks for keeping at it, Jonathan.

A couple of points/questions:

1. How can I make my own queries? Is it online or must something be downloaded? Or is it not yet ready for public release?
It is already released on github, at the links I provided above. But "it" is the data, you also need an XQuery processor. So you might consider getting BaseX or eXist or Zorba or Saxon to do your queries. I think BaseX is a good environment for this.

I will eventually have an online version of this, but I can't give any concrete timelines on that.
Stephen Carlson wrote:2. The determiner of an articular infinitive has to be neuter, so things like ὁ μέλλων ἔρχεσθαι should not have hit. Is there a way to query gender?
Absolutely. Take a look at this README, particularly the markup scheme. Anything present in the markup can be queried.
Stephen Carlson wrote:3. It looks like the structure of articular infinitives is not consistently coded in the database. Does this means that some of the effort involves basically reverse engineering the variety of strategies that were used to parse the text? I don't know who, how many, or according to what standard the syntax trees were done.
So far, I don't think that they are inconsistently coded, I think that they occur in a variety of syntactic structures, but I'd love to discuss any possible inconsistencies or ideas for improvement. The article sometimes governs more than one infinitive, those infinitives may occur in more than one clause, and the underlying structure needs to be represented. I think the Greek results in many forms, we represent those forms, and you have to take that into account when you query.

Our lowfat trees are transformed from Randall Tan / Andi Wu's Global Bible Initiative syntax trees, and they rely on their interpretation. The transformation makes them easier to query and display.
0 x
ἐξίσταντο δὲ πάντες καὶ διηποροῦντο, ἄλλος πρὸς ἄλλον λέγοντες, τί θέλει τοῦτο εἶναι;
http://jonathanrobie.biblicalhumanities.org/

Jonathan Robie
Posts: 3743
Joined: May 5th, 2011, 5:34 pm
Location: Durham, NC
Contact:

Re: List of Articular Infinitives

Post by Jonathan Robie » March 8th, 2015, 7:16 pm

Jonathan Robie wrote:It is already released on github, at the links I provided above. But "it" is the data, you also need an XQuery processor. So you might consider getting BaseX or eXist or Zorba or Saxon to do your queries. I think BaseX is a good environment for this.
Incidentally, all of the XQuery processors I mentioned are available for free.
0 x
ἐξίσταντο δὲ πάντες καὶ διηποροῦντο, ἄλλος πρὸς ἄλλον λέγοντες, τί θέλει τοῦτο εἶναι;
http://jonathanrobie.biblicalhumanities.org/

Jonathan Robie
Posts: 3743
Joined: May 5th, 2011, 5:34 pm
Location: Durham, NC
Contact:

Re: List of Articular Infinitives

Post by Jonathan Robie » March 8th, 2015, 8:02 pm

Three more comments:

1. I just took a look at the PROIEL treebank, which has a high quality dependency treebank of the Greek New Testament. Advantage: the relationship between an articular infinitive and the corresponding article is very direct, and you don't have to worry about constituents. Disadvantage: for the few examples I looked at, it doesn't handle cases like Matthew 11:1, where a single article governs more than one infinitive.

2. Writing the query is essentially the same thing as defining what you mean by an articular infinitive in terms of syntactic structure.

3. Queries like this really are a great way to figure out how to better design our treebank, so I really am very open to thinking about ways to improve the existing structure. But I don't think we can get around what I said in #2, sometimes the hard part of writing a query is thinking clearly about what exactly you are looking for.
0 x
ἐξίσταντο δὲ πάντες καὶ διηποροῦντο, ἄλλος πρὸς ἄλλον λέγοντες, τί θέλει τοῦτο εἶναι;
http://jonathanrobie.biblicalhumanities.org/

Stephen Carlson
Posts: 3021
Joined: May 11th, 2011, 10:51 am
Location: Melbourne
Contact:

Re: List of Articular Infinitives

Post by Stephen Carlson » March 9th, 2015, 1:13 am

Jonathan Robie wrote:
Stephen Carlson wrote:3. It looks like the structure of articular infinitives is not consistently coded in the database. Does this means that some of the effort involves basically reverse engineering the variety of strategies that were used to parse the text? I don't know who, how many, or according to what standard the syntax trees were done.
So far, I don't think that they are inconsistently coded, I think that they occur in a variety of syntactic structures, but I'd love to discuss any possible inconsistencies or ideas for improvement. The article sometimes governs more than one infinitive, those infinitives may occur in more than one clause, and the underlying structure needs to be represented. I think the Greek results in many forms, we represent those forms, and you have to take that into account when you query.
For example, I wouldn't have expected articular infinitives with arguments to be coded differently from those without (i.e., with or without an intermediate [cl ... ] constituent between the article and the infinitive, respectively). Conjunctions are often hairy in syntax trees; no surprise there.
0 x
Stephen C. Carlson, Ph.D.
Melbourne, Australia

Stephen Carlson
Posts: 3021
Joined: May 11th, 2011, 10:51 am
Location: Melbourne
Contact:

Re: List of Articular Infinitives

Post by Stephen Carlson » March 9th, 2015, 1:22 am

Jonathan Robie wrote:This does, however, have a few false positives such as this one (due to the function of ἄρχω, which the query does not account for):
inf-04.png
The text is Mark 14:71 ὁ δὲ ἤρξατο ἀναθεματίζειν και ὀμνύναι .... Actually, I think the issue is that ὁ is miscoded as a determiner instead of a pronoun.
0 x
Stephen C. Carlson, Ph.D.
Melbourne, Australia

Jonathan Robie
Posts: 3743
Joined: May 5th, 2011, 5:34 pm
Location: Durham, NC
Contact:

Re: List of Articular Infinitives

Post by Jonathan Robie » March 9th, 2015, 7:51 am

Stephen Carlson wrote:
Jonathan Robie wrote:This does, however, have a few false positives such as this one (due to the function of ἄρχω, which the query does not account for):
inf-04.png
The text is Mark 14:71 ὁ δὲ ἤρξατο ἀναθεματίζειν και ὀμνύναι .... Actually, I think the issue is that ὁ is miscoded as a determiner instead of a pronoun.
That's a bug. I just entered a bug report for this, we'll fix this:

https://github.com/biblicalhumanities/g ... t/issues/6

By the way, anyone can enter bug reports like this.
0 x
ἐξίσταντο δὲ πάντες καὶ διηποροῦντο, ἄλλος πρὸς ἄλλον λέγοντες, τί θέλει τοῦτο εἶναι;
http://jonathanrobie.biblicalhumanities.org/

Jonathan Robie
Posts: 3743
Joined: May 5th, 2011, 5:34 pm
Location: Durham, NC
Contact:

Re: List of Articular Infinitives

Post by Jonathan Robie » March 9th, 2015, 8:11 am

Stephen Carlson wrote:
Jonathan Robie wrote:So far, I don't think that they are inconsistently coded, I think that they occur in a variety of syntactic structures, but I'd love to discuss any possible inconsistencies or ideas for improvement. The article sometimes governs more than one infinitive, those infinitives may occur in more than one clause, and the underlying structure needs to be represented. I think the Greek results in many forms, we represent those forms, and you have to take that into account when you query.
For example, I wouldn't have expected articular infinitives with arguments to be coded differently from those without (i.e., with or without an intermediate [cl ... ] constituent between the article and the infinitive, respectively). Conjunctions are often hairy in syntax trees; no surprise there.
That comes from two rules we use to build the lowfat trees:
  • A clause is defined as a verb with all of its arguments and adjuncts (subject, objects, adverbs, adverbial prepositional phrases, etc.)
  • In the lowfat trees, a word group always contains at least two sub-nodes. If a verb has no arguments or adjuncts, we do not add a clause node.
The following XQuery is close, but I think there are some cases that it doesn't take into account:

Code: Select all

for $np in //wg[@class='np']
where $np/w[1][@class='det' and @gender='neuter']
for $infinitives in $np/w[@mood='infinitive' and @head='true']
   |  $np/wg[@class='cl' or @class='vp']/w[@mood='infinitive' and @head='true']
return $infinitives
Let me break that down.
  • Iterate over noun phrases

    Code: Select all

    for $np in //wg[@class='np']
  • Keep only the ones where the first word is a determiner with neuter gender

    Code: Select all

    where $np/w[1][@class='det' and @gender='neuter']
  • Find infinitives that are in words in the noun phrase itself

    Code: Select all

    for $infinitives in $np/w[@mood='infinitive' and @head='true']
  • And also those found in clauses or verb phrases contained in the noun phrase

    Code: Select all

     |  $np/wg[@class='cl' or @class='vp']/w[@mood='infinitive' and @head='true']
That finds 249 instances, so I think we're missing some cases, I can look at that later. When I do that, I'll mail you a list of what I found.
0 x
ἐξίσταντο δὲ πάντες καὶ διηποροῦντο, ἄλλος πρὸς ἄλλον λέγοντες, τί θέλει τοῦτο εἶναι;
http://jonathanrobie.biblicalhumanities.org/

Jonathan Robie
Posts: 3743
Joined: May 5th, 2011, 5:34 pm
Location: Durham, NC
Contact:

Re: List of Articular Infinitives

Post by Jonathan Robie » March 9th, 2015, 8:29 am

Jonathan Robie wrote:That comes from two rules we use to build the lowfat trees:
  • A clause is defined as a verb with all of its arguments and adjuncts (subject, objects, adverbs, adverbial prepositional phrases, etc.)
  • In the lowfat trees, a word group always contains at least two sub-nodes. If a verb has no arguments or adjuncts, we do not add a clause node.
Incidentally, the GBI syntax trees, also found at our github site, have a different organization. You can try querying each set of trees to see which you prefer for this. They are also in XML, so you can use the same XQuery processor for each set of trees.

And Dag Haug's PROIEL GNT is also in XML, and also on github.

So it's pretty easy to get your hands on at least three complete treebanks for the Greek New Testament that use different models. Ours are derived from the GBI trees, though, so they aren't based on an independent analysis.
0 x
ἐξίσταντο δὲ πάντες καὶ διηποροῦντο, ἄλλος πρὸς ἄλλον λέγοντες, τί θέλει τοῦτο εἶναι;
http://jonathanrobie.biblicalhumanities.org/

Post Reply

Return to “Syntax and Grammar”