Possible Bug: Syntax Search Result

Page 1 of 2 (24 items) 1 2 Next >
This post has 23 Replies | 3 Followers

Posts 53
Mike Hogue | Forum Activity | Posted: Thu, May 12 2016 8:00 AM

Hello Syntax-Search gurus!

I constructed the attached search with the following result:

The result seems to be pulling the order with the conjunction first rather than the clause first, perhaps due to the "crossed" graph structure having an out-of-order text. Interestingly it does not pull Gal 2:10, which does have the structure I was trying to find. Is this a bug, or am I doing something wrong?

My line of inquiry came from this thread on Syntax Searching:

https://faithlife.com/comments/583563?utm_source=notifications&utm_medium=email&utm_content=faithlife-notification&utm_campaign=faithlife

As such, I saw Mike Meiser's construction using a "negative" construction...but it seems to me the structure should be able to be found with a "positive" construction as well.

Thanks in advance,

Mike

Posts 151
Ryan Robinson | Forum Activity | Replied: Thu, May 12 2016 8:49 AM

No bug.

Your issue in the "match skip levels" on the first verbal clause. 

Embedded within the ADV is a verbal clause (CL -> V -> ἐρχόμενον).

I don't have tons of time. So, hopefully that makes sense. Let me know if you need clarification.

Posts 53
Mike Hogue | Forum Activity | Replied: Thu, May 12 2016 9:03 AM

Thanks Ryan.

The embedded clause still appears after the ἱνα in the tree structure, so I think I disagree with you, but am still open to correction. Why would it pull that one when the "conj" is clearly before the clause that contains the embedded clause in the "ADV" in the tree structure??!

When I undo "matching skips levels" I get zero results, which still doesn't make sense.

Posts 151
Ryan Robinson | Forum Activity | Replied: Thu, May 12 2016 9:25 AM

It does make sense because you are skipping that level of CL.

It is looking for any verbal clause that would come before the ἱνα. The hierarchical structure (CL that comes after the ἱνα) is being discarded because of "matching skips levels".

Let me know if that doesn't make sense. :) 

Posts 53
Mike Hogue | Forum Activity | Replied: Thu, May 12 2016 9:49 AM

Wait...skipping levels suddenly means that the lines don't apply anymore? i.e. skipping levels produces a situation where the lines drawn to that point are ignored?  That seems counter-intuitive, does break the order of the structure, and I would submit should be considered a design flaw or a bug. The order, even with "matching skips levels", should be the structural level for a syntax search...not word order. Your explanation does explain the result, though.

This leads me to another question:

If I reverse the order and undo "matching skips levels" it doesn't catch Acts 19:4...why? There is an immediate clause with the conjunction and a verbal clause underneath it. Removing the "anything" doesn't change the results.

Posts 151
Ryan Robinson | Forum Activity | Replied: Thu, May 12 2016 11:49 AM

Matching skips levels means I'm looking for X anywhere within a clause. Thus, why that was happening in the last search

To speak to why it wasn't getting it with the new search. (This is a terrible picture. All I have to work with at work.)

Posts 53
Mike Hogue | Forum Activity | Replied: Thu, May 12 2016 12:43 PM

Not to continue being dense, but when I mouse over the "CL*" for that node it shows that it is a verbal clause. Why are you saying it isn't? 

Posts 151
Ryan Robinson | Forum Activity | Replied: Thu, May 12 2016 1:19 PM

You aren't being dense. I had a brain fart. You're right. It is a verbal clause. And this in fact is a bug! I've noticed this in the past and forgot about this. 

It has to do with the ADV clause that is fronted and it being embedded within a clause that is actually after the ἵνα. The crossing over arrow kinda messes things up.

I'd encourage reporting this as a bug. I doubt it will go anywhere since syntax search isn't very prioritized for Faithlife from what I'm seen (that isn't a jab just an observation). 

Posts 151
Ryan Robinson | Forum Activity | Replied: Thu, May 12 2016 3:24 PM

I love having these conversations about Syntax Searching. It is an amazingly powerful tool that too many don't know how to utilize. 

Posts 53
Mike Hogue | Forum Activity | Replied: Thu, May 12 2016 3:44 PM

Thanks Ryan - I appreciate you engaging on this. 

I agree that it is powerful...it just doesn't always seem to give the results it should. 

Do you know how to go about submitting this as a bug?

Posts 151
Ryan Robinson | Forum Activity | Replied: Thu, May 12 2016 4:04 PM

Mike Hogue:

it just doesn't always seem to give the results it should. 

What other issues have you had? This is the only one that I've run into that really has caused issues.

Mike Hogue:

Do you know how to go about submitting this as a bug?

I think you'd just have to put it under "logos 6" and put BUG or something. 

Posts 53
Mike Hogue | Forum Activity | Replied: Thu, May 12 2016 4:22 PM

https://faithlife.com/comments/345154

Ryan,

Here's another thread with another problem, and it seemed that unless everything was defined at the same level, the search didn't work. Several (including you) said "I don't know why..."

Mike

Posts 151
Ryan Robinson | Forum Activity | Replied: Thu, May 12 2016 5:02 PM

Alright. I think I got it. It has to be like this:

Reason being is the asterisk shows that there only ONE head word (e.g. in Matthew 7:26 it is ἀκούων). So, in order to have two different participles you have to go down another level as shown in the search query. 

Let me know if this still doesn't make sense. 

Posts 26013
Forum MVP
Dave Hooton | Forum Activity | Replied: Thu, May 12 2016 5:38 PM

Mike Hogue:
I agree that it is powerful...it just doesn't always seem to give the results it should. 

You would understand better if you highlighted the results for the subordinate clause (red border) rather than the containing clause, as below

Then you will see that the conjunction (immediately) follows it as requested.

For Gal 2:10 the conjunction is embedded in the Verbal clause; which requires a somewhat convoluted query involving Gap.  But it follows the Object function.

You don't really need Anything for this query.

Dave
===

Windows 10 & Android 8

Posts 53
LogosEmployee
Michael Meiser (Faithlife) | Forum Activity | Replied: Thu, May 12 2016 11:44 PM

Mike Hogue:
As such, I saw Mike Meiser's construction using a "negative" construction...but it seems to me the structure should be able to be found with a "positive" construction as well.

I noticed my name being evoked, so I thought I would weigh in. When we are dealing with postpositive conjunctions like δε, and in this instance ινα, most syntax queries are not adequate to find these instances. What we have here, as Dave hinted at in his post, is a case of the dreaded "gap." 

In Acts 19:4, it seems logical that the terminal node immediately precedes the verbal clause from the clause visualization. However, this is not how the computer understands the syntax tree in this situation. In reality, the terminal node is neither preceding nor following the verbal clause, even though visually it appears that way. There is a trump card being played here known as a "gap." You can identify gaps visually by intersecting clausal lines. See highlighted image below.

In essence, the terminal node is in fact buried within the verbal clause even though it is not technically a part of the clause. For instance, here is the search syntax for finding this particular instance of ινα.

If we look at the search results, notice that the word ινα is not highlighted with the rest of the clause.

I think the short answer to your question is that this is not a bug. I am in no way saying that this is the clearest way to represent what is going on with the syntax of this verse, rather, these are the rules that the database has been constructed to follow. (BTW, I didn't build this sandbox, I just like to play in it.) ;-)

Luckily, most of these are outliers to the majority of searches we run. I'm hoping that by this explanation I can help you understand a bit more how to play by the rules of the game as opposed to what "seems" to be the right way. I am open to being proven wrong on any of this as most of these details were arrived at by my own deductions and trial and error over time.

Free Logos Tutorials on YouTube: Digital Bible Study Made Easy

Posts 53
Mike Hogue | Forum Activity | Replied: Fri, May 13 2016 5:12 AM

Thanks Mike - great explanation. I notice also that this "gap" search also catches Gal 2:10...

This is definitely counter-intuitive...namely that the actual structure as understood by the computer has the ἱνα inside a gap in the clause when the picture looks different. The in-program documentation for gaps is somewhat cryptic -- "Gap allows searching within the “gaps” that occur in non-continuous constructions;" 

Per the thread above...I guess the implication is that "matching skips levels" really doesn't throw out the lines on the graph...it's just that with "gap" constructions (wherever there are crossed lines) the lines on the graph don't really represent what is going on.

How would the computer view Matt 11:7? (not the postpositive δε, but the other gap). Why is τί considered part of the gap?

Posts 151
Ryan Robinson | Forum Activity | Replied: Fri, May 13 2016 5:38 AM
Mike M. Awesome explanation. That was super helpful in understanding that. This always makes me think about how there is no documentation out for these things and I scratch my head not understanding why. Mike H When you say "in-program documentation" what are you referencing? Thanks!
Posts 151
Ryan Robinson | Forum Activity | Replied: Fri, May 13 2016 5:43 AM

I'm also noticing this could of been a lot easier search for the question from the Faithlife group. There are only to false hits (2 Cor 2:5 and 1 John 5:13).

Mike Meiser:

I'm guessing the only reason you didn't go for this, Mike, is it isn't as safe to be exactly what he was looking for? 

Posts 151
Ryan Robinson | Forum Activity | Replied: Fri, May 13 2016 5:56 AM

You know what is interesting, I remembered awhile back I had an issue with a gap and Dave pointing me towards using Opentext. I was opposed in my heart ;). But I did it and it was way more simple.

So, I recreated a search similar to the one on on the Faithlife group with Opentext and got wayyyy more results that the Cascadia Search.

This seems to work better in my opinion.

Posts 53
Mike Hogue | Forum Activity | Replied: Fri, May 13 2016 5:56 AM

Ryan - in Logos, in a Syntax Search window, hit "F1", then follow the link to "Syntax Search Document."  The resulting help file includes "available search terms."

It's not incredibly helpful for something like syntax searches, though.

Page 1 of 2 (24 items) 1 2 Next > | RSS