Syntax Search Problem

Page 1 of 2 (23 items) 1 2 Next >
This post has 22 Replies | 0 Followers

Posts 582
Gary O'Neal | Forum Activity | Posted: Sat, Mar 6 2010 7:08 PM

I've been following Terry's and other's discussion of trying to get more help from the forums in actual Bible study. One of the areas I really struggle with is in learning how the syntax search works. Some of you (Kevin, Mike, and Dave to name a few) really seem to have your mind wrapped around this, but I'm not confident in any results I get back. I've been through the wiki pages and the videos but still need more practical help.

Could one of you throw out a problem for us to attempt to search for, let us post our responses and see how you would have done it differently?

πάντα εἰς δόξαν θεοῦ ποιεῖτε

Posts 26033
Forum MVP
Dave Hooton | Forum Activity | Replied: Sun, Mar 7 2010 5:39 AM

OK. The Genitive Absolute is my challenge:-

a. Structure
...
1) a noun or pronoun in the genitive case (though this is sometimes absent);
2) a genitive anarthrous participle (always);
3) the entire construction at the front of a sentence (usually).

b. Semantics
....
1) This construction is unconnected with the rest of the sentence (i.e., its subject-the genitive noun or pronoun-is different from the subject of the main clause);
Wallace, D. B. (1999; 2002). Greek Grammar Beyond the Basics

Solutions to start with: Matt 1:18; 2:1, Act 12:13 (in OpenText look for an Embedded EC or Secondary clause SC. In Cascadia look for a clause that it is not the Head clause ie. CL  as opposed to CL*). For construction use an Unordered group. See http://wiki.logos.com/A_Strategy_for_Syntax_Search

Start with whatever database you find easiest (OText or Cascadia). We won't hold you to ensuring the result is an anarthrous participle.

Dave
===

Windows 10 & Android 8

Posts 582
Gary O'Neal | Forum Activity | Replied: Sun, Mar 7 2010 7:21 AM

Dave Hooton:

OK. The Genitive Absolute is my challenge:-

a. Structure
...
1) a noun or pronoun in the genitive case (though this is sometimes absent);
2) a genitive anarthrous participle (always);
3) the entire construction at the front of a sentence (usually).

b. Semantics
....
1) This construction is unconnected with the rest of the sentence (i.e., its subject-the genitive noun or pronoun-is different from the subject of the main clause);
Wallace, D. B. (1999; 2002). Greek Grammar Beyond the Basics

Solutions to start with: Matt 1:18; 2:1, Act 12:13 (in OpenText look for an Embedded EC or Secondary clause SC. In Cascadia look for a clause that it is not the Head clause ie. CL  as opposed to CL*). For construction use an Unordered group. See http://wiki.logos.com/A_Strategy_for_Syntax_Search

Start with whatever database you find easiest (OText or Cascadia). We won't hold you to ensuring the result is an anarthrous participle.

Dave

Thanks for posting this challenge. Here's my first attempt -- I couldn't figure out how to only find anarthrous participles.

 

 

 

πάντα εἰς δόξαν θεοῦ ποιεῖτε

Posts 582
Gary O'Neal | Forum Activity | Replied: Sun, Mar 7 2010 12:59 PM

Ok, second attempt. I'm not sure how to compare these two searches. They both return the control verses. Reckon next would be to start going through and seeing what they returned that wasn't genitive absolutes. This attempts to make sure the participle is anarthrous and provides a choice of either noun or pronoun in the genitive case. Not sure that this is actually needed--it might just complicate the search further.

 

πάντα εἰς δόξαν θεοῦ ποιεῖτε

Posts 5337
Kevin Becker | Forum Activity | Replied: Sun, Mar 7 2010 7:03 PM

Challenge #2

2nd class conditional.

Protasis: εἰ + indicative; past tense (aorist, imperfect); (negative: μή) Apodosis: (ἄν) + indicative; past tense (aorist (past time); imperfect (present time);

Daniel B. Wallace, Greek Grammar Beyond the Basics - Exegetical Syntax of the New Testament ( (Zondervan Publishing House and Galaxie Software, 1999; 2002)), 689.

Examples

Luke 7:39; John 5:46; 1 Cor 2:8

See also: Wallace pg 694-696

[I haven't tried this one yet, but I'll work on it tomorrow Smile]

Posts 26033
Forum MVP
Dave Hooton | Forum Activity | Replied: Mon, Mar 8 2010 5:14 AM

Gary O'Neal:
Ok, second attempt. I'm not sure how to compare these two searches.

I'm not sure that Optional is handled correctly - whether "May or may not appear" or "0-1", especially at the clause level. So I drop it. The anarthrous case is handled well but you will note some simplifications in my solution.

 

Dave
===

Windows 10 & Android 8

Posts 5337
Kevin Becker | Forum Activity | Replied: Mon, Mar 8 2010 6:09 AM

Dave Hooton:

I'm not sure that Optional is handled correctly - whether "May or may not appear" or "0-1", especially at the clause level. So I drop it. The anarthrous case is handled well but you will note some simplifications in my solution.

I agree with the dropping the may or may not be present for the subject-less Genitive Absolutes; it introduces too much noise into the results. Dave I compared my search (functionally equivalent to yours) against Wallace's example lists. There were a couple of examples that Cascadia had in a CL* so I dropped Head = No in Clause 1 (Although most of these were Noun/Pronoun-less Gen Absolutes but Acts 13:2 is an example that would be excluded by specifying Head=No). Also, I found an example where the Participle was classified as Verbal Complement to I added that as an option. I retained 2 words for the Noun/Pronoun simply because I like the way it looks Smile. So, here's the picture:

Posts 582
Gary O'Neal | Forum Activity | Replied: Mon, Mar 8 2010 12:26 PM

Dave and Kevin

Thank you both for your help. I believe I understand each of the changes you made in my approach. The participle isn't set to be the head because verbs are naturally heads of their clauses--right? Also, could you clarify why the subject function appearing 0 or 1 times would not work.

Now, how do we tell how accurate our search is? I did a search online to try to find out how many genitive absolutes there are in the NT and came across this article http://www.macdiv.ca/coned/documents/GApaperforOctober.pdf, which gives a number of 312 for the entire NT. It also said there were 52 in Matthew. Our search returns 40. Do either of you know of an accepted list of ga's that I could check these results against?

Also, Kevin I have briefly started on the 2nd class conditional problem. I'll try to post something later today, but my initial problem was getting the apodosis to show any results. I started by setting up a search that would find the protasis and that worked fine, but whenever I tried to add the apodosis search it always returned no results. I'll play with it a little more and then try to post something later today.

 

πάντα εἰς δόξαν θεοῦ ποιεῖτε

Posts 5337
Kevin Becker | Forum Activity | Replied: Mon, Mar 8 2010 1:48 PM

Gary O'Neal:
Also, could you clarify why the subject function appearing 0 or 1 times would not work

It's not that it won't work. It should perform the same as "may or may not appear." However, the difficulty is that this essentially makes the search catch every anarthous genitive participle, the majority of which are probably not genitive absolutes, hence my comment about introducing too much noise (that along with multiple hits for introduced by matching skips levels).

One of the difficulties with Syntax search is that these forms (Genitive Absolute, 2nd class conditionals) is that writers don't always follow the "rules." For example there's at least one 2nd class conditional that doesn't have the particle αν but you know it's 2nd class because it's negated by μη but if you make αν optional you'll grab a bunch of 1st class conditionals in your search results (Perhaps there's a way to say include αν's absence only if μη is used, but I haven't figured it out.)

In the end, it's probably not about syntax searches being inaccurate but incomplete due to grammar/syntax "errors" and/or ellipsis. If you want scholarly complete lists you're going to have to construct both syntax searches and Morph searches so broad that they get lots of false hits you manually sift out. Once verse lists are re-introduced it should be easier to compare the variation in search results introduced my more permissive terms.

Posts 582
Gary O'Neal | Forum Activity | Replied: Mon, Mar 8 2010 2:24 PM

Kevin Becker:
It should perform the same as "may or may not appear." However, the difficulty is that this essentially makes the search catch every anarthous genitive participle,

I thought I was asking for the subject (noun or pronoun) to appear one or zero times. By asking for the subject to appear 1 or 0 times, I am in fact asking for anything as a subject, right?

 

πάντα εἰς δόξαν θεοῦ ποιεῖτε

Posts 582
Gary O'Neal | Forum Activity | Replied: Mon, Mar 8 2010 5:11 PM

Kevin Becker:

Challenge #2

2nd class conditional.

Protasis: εἰ + indicative; past tense (aorist, imperfect); (negative: μή) Apodosis: (ἄν) + indicative; past tense (aorist (past time); imperfect (present time);

Daniel B. Wallace, Greek Grammar Beyond the Basics - Exegetical Syntax of the New Testament ( (Zondervan Publishing House and Galaxie Software, 1999; 2002)), 689.

Examples

Luke 7:39; John 5:46; 1 Cor 2:8

See also: Wallace pg 694-696

[I haven't tried this one yet, but I'll work on it tomorrow Smile]

Here's my attempt at this one.

 

This search returned all the control verses. It found 41 instances in all. Boyer, in "Adverbial Clauses: Statistical Studies" (GTJ, 11:1 (Spr 90): 71-96), says that there are 48 of these in the NT, but 7 have an implied verb. I haven't checked the results to see if they are the same 7, but I doubt I'm that lucky.


.

 

πάντα εἰς δόξαν θεοῦ ποιεῖτε

Posts 582
Gary O'Neal | Forum Activity | Replied: Mon, Mar 8 2010 5:19 PM

Gary O'Neal:
Here's my attempt at this one.

Two things stump me about this problem. First, when I added ἄν as a lemma in the apodosis, I get no results. I did go back and add it in the text section and came back with 38 results.

Second, Boyer said there were 6 instances where the order of the protasis and apodosis is reversed. To attempt to catch these, I added an Unordered group after the first clause but this returned no results.

 

πάντα εἰς δόξαν θεοῦ ποιεῖτε

Posts 26033
Forum MVP
Dave Hooton | Forum Activity | Replied: Mon, Mar 8 2010 9:32 PM

Gary O'Neal:
I thought I was asking for the subject (noun or pronoun) to appear one or zero times. By asking for the subject to appear 1 or 0 times, I am in fact asking for anything as a subject, right?

The 1 does what you expect. The 0 will return a result in which a Subject appears nowhere within the parent term, but specifying 0-1 within an Unordered term results in too many permutations of the results. Another query shows there are 379 participles without a Subject, so the most meaningful method is to combine two queries as one.

Note that I have adopted part of Kevin's suggestion, but I decided to show only the head Subject e.g. 1 out of the 3 possible results for Matt 1:18. This will return 656 results. Unfortunately there are some false positives because a subordinate clause may also be a head clause. This is the value of comparing Syntax results with those from independent sources, but also illustrates the weakness that any method of analysis will be biased and/or include errors.

Of the results from Wallace it is missing Matt 8:16 and Jn 4:50. Matt 8:16 would be included if I added the Verbal Complement to the second structure as per Kevin (giving an extra 32 results) but Jn 4:50 appears to be a typo for Jn 4:51 (no genitive participle).

Dave
===

Windows 10 & Android 8

Posts 26033
Forum MVP
Dave Hooton | Forum Activity | Replied: Mon, Mar 8 2010 10:27 PM

Kevin Becker:
In the end, it's probably not about syntax searches being inaccurate but incomplete due to grammar/syntax "errors" and/or ellipsis.

My OpenText query matches my first Cascadia query wrt Wallace's results ie. they both include the same verses and also omit the same verses except that OpenText includes Acts 13:2! However the OpenText query has 50 more results than Cascadia, but are very close when I remove the Head = No constraint.

Kevin Becker:
If you want scholarly complete lists you're going to have to construct both syntax searches and Morph searches so broad that they get lots of false hits you manually sift out. Once verse lists are re-introduced it should be easier to compare the variation in search results introduced my more permissive terms.

Yes, I desperately miss the Verse List.

Dave
===

Windows 10 & Android 8

Posts 5337
Kevin Becker | Forum Activity | Replied: Tue, Mar 9 2010 3:57 AM

Dave Hooton:

Note that I have adopted part of Kevin's suggestion, but I decided to show only the head Subject e.g. 1 out of the 3 possible results for Matt 1:18. This will return 656 results. Unfortunately there are some false positives because a subordinate clause may also be a head clause. This is the value of comparing Syntax results with those from independent sources, but also illustrates the weakness that any method of analysis will be biased and/or include errors.

Of the results from Wallace it is missing Matt 8:16 and Jn 4:50. Matt 8:16 would be included if I added the Verbal Complement to the second structure as per Kevin (giving an extra 32 results) but Jn 4:50 appears to be a typo for Jn 4:51 (no genitive participle).

Good observations Dave. I noticed the John 4 typo too and added a note to my copy of Wallace. The benefit of embedding the noun/pronoun option in the subject node is that it reduces the duplicate hits introduced by skipping levels. I'm going to take your suggestion and modify my search. I wish there was an option that the search algorithm could filter out the multiple hits on a structure so the hit count would be more accurate. As is, it's possible to have two people comparing search results and the one with the higher hit count could have actually found fewer examples of the structure.

Posts 5337
Kevin Becker | Forum Activity | Replied: Tue, Mar 9 2010 4:28 AM

Gary O'Neal:
Two things stump me about this problem. First, when I added ἄν as a lemma in the apodosis, I get no results. I did go back and add it in the text section and came back with 38 results.

I was able to use αν in the lemma box only if I copied and pasted it in.

Gary O'Neal:
Second, Boyer said there were 6 instances where the order of the protasis and apodosis is reversed. To attempt to catch these, I added an Unordered group after the first clause but this returned no results.

The syntax search doesn't allow for nesting of unordered groups. Logos disallowed it because of the complexity that it could cause. To accomplish it you will have to add an OR and then switch the order of the protasis and apodosis manually.

Let me trot out some of my searches aimed at this problem.

This was my complicated search, trying to catch each feature of the text but it takes way to long to process and with so many duplicate hits I wouldn't attempt to evaluate its effectiveness until verse lists come back.

A more elegant solution is to not search for everything but just some of the key things

However, some of the things you said prompted this move to simplicity

I also tried an OR coming off of Clause 3 to include a Terminal Node 3 looking for μη (because of the example in Wallace w/o αν but there was a lot of false positives because of it.) However, when you get this simple you run the risk making a search that might be better suited for a Morph query.

Posts 582
Gary O'Neal | Forum Activity | Replied: Tue, Mar 9 2010 5:23 PM

Kevin Becker:
The benefit of embedding the noun/pronoun option in the subject node is that it reduces the duplicate hits introduced by skipping levels.

How do you embed the noun/pronoun? I'm assuming this should be done through the morph dialog, but I can't enter both.

πάντα εἰς δόξαν θεοῦ ποιεῖτε

Posts 5337
Kevin Becker | Forum Activity | Replied: Tue, Mar 9 2010 5:35 PM

Gary O'Neal:
How do you embed the noun/pronoun? I'm assuming this should be done through the morph dialog, but I can't enter both.

Put this in the Morph line

@NG @JG

Posts 582
Gary O'Neal | Forum Activity | Replied: Tue, Mar 9 2010 6:43 PM

Kevin

Thanks for you help. I believe I follow what you are doing--just need to spend more time with these so that I get a feel for what works and when to use certain structures. Two questions -- Is there a reason to prefer Terminal node over Word? I notice you did that in the 2nd class conditional search. Also in the 2nd class conditional search, you included a search for the prefect tense. Aren't these limited to aorist, imperfect and pluperfect?

Finally, to keep from a separate post, thanks for the info on embedding multiple morph forms.

πάντα εἰς δόξαν θεοῦ ποιεῖτε

Posts 5337
Kevin Becker | Forum Activity | Replied: Tue, Mar 9 2010 7:00 PM

Gary O'Neal:
Is there a reason to prefer Terminal node over Word? I notice you did that in the 2nd class conditional search.

The terminal node will have all the details from the word level. So from a functional level it is almost equivalent to a word. Also, since it is one level higher than Words that is one less level for the search to jump over which I believe returns fewer multiple hits (although I haven't tested this assumption).

Gary O'Neal:
2nd class conditional search, you included a search for the prefect tense. Aren't these limited to aorist, imperfect and pluperfect?

Yes, I was wrong to include Perfects in my search. I'll have to strike it from my search.

Page 1 of 2 (23 items) 1 2 Next > | RSS