Syntax question

Kevin Becker
Kevin Becker Member Posts: 5,604 ✭✭✭
edited November 2024 in English Forum

I'm studying pistis cristou and using the opportunity to figure out syntax search. I designed a syntax search to catch pistis cristou and pistis iesous cristou. It's not as good as a Morph search (some false positives, but I'm learning).

My question is, why does the search results show the text from the verse multiple times? It makes the results difficult to use/interpret. Is this a bug/issue or is my search poorly constructed?

Also, any ideas on how to improve this kind of search? Would there be a way to catch pistis iesous only too?

EDIT: This screenshot is from Beta 2, but SR-7 has a very similar behavior.

Comments

  • Robert Pavich
    Robert Pavich Member Posts: 5,685 ✭✭✭


    I'm studying pistis cristou and using the opportunity to figure out syntax search. I designed a syntax search to catch pistis cristou and pistis iesous cristou. It's not as good as a Morph search (some false positives, but I'm learning).

    My question is, why does the search results show the text from the verse multiple times? It makes the results difficult to use/interpret. Is this a bug/issue or is my search poorly constructed?

    Also, any ideas on how to improve this kind of search? Would there be a way to catch pistis iesous only too?

    EDIT: This screenshot is from Beta 2, but SR-7 has a very similar behavior.


     

    Kevin,

    Right off the bat...remember, I'm not the Logos scholar that many are here but my first Q would be: Why use a syntax search on a phrase?

    I'm not that great at syntax searching but in my mind; syntax searches are more for "word functions" and not "phrases" right?

     

    What a basic search like: pistis cristou OR pistis iesous cristou do for you?

    I'm just asking and maybe I'm ignorant of what you're trying to accomplish....

    Robert Pavich

    For help go to the Wiki: http://wiki.logos.com/Table_of_Contents__

  • Robert Pavich
    Robert Pavich Member Posts: 5,685 ✭✭✭

    Kevin,

    Again, being ignorant, I did this search and I think that's your intent right....to see either "faith in" Christ or "faith in" Christ Jesus right?

    image

    You could do it the cheating way [:D] search BWS on pisteow and look a the contexts and that will tell you this answer also.

    If you want to know "what things are "faith in" directed to? Then I think THAT'S  a syntax search maybe....someone can help me....

     

    Robert Pavich

    For help go to the Wiki: http://wiki.logos.com/Table_of_Contents__

  • Russ Quinn
    Russ Quinn Member Posts: 711 ✭✭

    Which instances of pistis christou or pistis Ihsou christou should be translated "faith in Christ" is a hot debate among scholars.

    English translations disagree on particular instances so a English search would not be a reliable way to find all the occurrences.

    I would probably want to use a Morph search but that is because I haven't gotten my head around the Syntax search engine.

    Kevin, it would be helpful for me if you continued journaling your efforts on this.

    Thanks,

    Russ

  • davidphillips
    davidphillips Member Posts: 640 ✭✭

    Kevin,

    The reason your search returns multiple hits per verse is because of the way the Cascadia Syntax graphs (and opentext.org graphs) are constructed. If you take Romans 3:21-22 as an example and fill in the search completely (see below) so that every phrase is included, you will only get one hit per occurrence.

    image

    But what appears to be happening is that since you are allowing matching to skip levels between the word and phrase, it counts a hit at each phrase the word could be coming from. I can't quite figure out how it's counting, but I'm fairly certain the issue lies in the fact that the Cascadia Syntax charts make abundant use of "phrases" (the same thing will happen with opentext.org in a similarly constructed search where you skip levels between the word group and words, but with less "extra" results). In other words, the more you allow matching to skip levels and the "higher" (father left) in the search you do so, the more extra results you'll return.

     

  • Robert Pavich
    Robert Pavich Member Posts: 5,685 ✭✭✭

    I did this in the OpenText...if that helps....

    image

    Robert Pavich

    For help go to the Wiki: http://wiki.logos.com/Table_of_Contents__

  • davidphillips
    davidphillips Member Posts: 640 ✭✭

    Also, I would agree with Robert that this is a good use for a morphological query, since you are searching for a tight phrase. The one he suggested works well. If, however, you were interested in any place πιστεως governs the phrase ιησου χριστου, a syntax search might help, but you'd want to construct it a little differently to allow for the separation of the words. This search gets an extra hit in James 2:1, where πιστις is separated by the phrase του κυριου ημων:

    image

    Also, you'll get some double hits since ιησους is optional. It will be counted once with it present and once without it present.

    A good rule of thumb (that I've found) is this: If you're interested in places where words appear close together, use a morph search. If you're looking for places where certain words are related in a certain way (though they may be physically separated), use a syntax search.

  • davidphillips
    davidphillips Member Posts: 640 ✭✭

    I did this in the OpenText...if that helps....

    You beat me [:)]

  • davidphillips
    davidphillips Member Posts: 640 ✭✭

    Kevin,

    A final thought. The fact that all those extra hits appears isn't that big of a deal. You still get the verses listed on the left. Rather than scanning through each hit, you can simply go to each verse and look at them yourself. If there happen to be two hits in a single verse, you should see it, since you know where to look.

  • Kevin Becker
    Kevin Becker Member Posts: 5,604 ✭✭✭

    Kevin,

    Again, being ignorant, I did this search and I think that's your intent right....to see either "faith in" Christ or "faith in" Christ Jesus right?

    image

    You could do it the cheating way Big Smile search BWS on pisteow and look a the contexts and that will tell you this answer also.

    If you want to know "what things are "faith in" directed to? Then I think THAT'S  a syntax search maybe....someone can help me....

     

    Robert,

    Thanks, I did something very similar to this yesterday too. I was thinking that Morph searches do proximity fairly well, however the more proximity you introduce to catch variations in how a phrase is constructed the more false hits you generate. In this case I was searching for faith of Christ, Faith of Christ Jesus, or Faith of Jesus. Since the syntax graphs group together words and help you to see what modifies what, my hope was that a syntax search might be flexible enough to catch phrasal variations. It appears that morph search is probably the best tool for this evaluation. Tracking down every occurrence of pistis with a genitive name or pronoun after it is probably in my near future.

     

  • Russ Quinn
    Russ Quinn Member Posts: 711 ✭✭

    The hit in James 2:1 is critical to the debate because it represents a nonPauline author's use of a similar construction.

    Please bear with my newbieness to the Syntactical search engine but why are all of the operators before "Unordered" necessary?

    If a response like "Read the manual" is appropriate, I understand.

    But if a simple explanation is possible, I would appreciate it.

  • davidphillips
    davidphillips Member Posts: 640 ✭✭


    Tracking down every occurrence of pistis with a genitive name or pronoun after it is probably in my near future.

     

    That would be an excellent place for a syntax search!

  • Kevin Becker
    Kevin Becker Member Posts: 5,604 ✭✭✭

    Kevin,

    A final thought. The fact that all those extra hits appears isn't that big of a deal. You still get the verses listed on the left. Rather than scanning through each hit, you can simply go to each verse and look at them yourself. If there happen to be two hits in a single verse, you should see it, since you know where to look.

    Thanks David for you help. You are right the Morph search is a better tool for this job. I'm glad I wrestled with the syntax search yesterday, though, I understand it a little better.

    Ultimately when verse lists return this sort of search/study will be easier.

  • davidphillips
    davidphillips Member Posts: 640 ✭✭


    Please bear with my newbieness to the Syntactical search engine but why are all of the operators before "Unordered" necessary?

    If a response like "Read the manual" is appropriate, I understand.

     

    Russ,

    That's a great question! One of the great things about syntax searches is that you can be very specific about what you're looking for. The challenge is that if you're too specific you might miss valid hits, but if you're not specific enough you could get a lot of junk. In this case, I included the "Word Group" and "Head term" operators because I was confident I wanted instances where πιστις and ιησου χριστου were in the same syntactical unit. Had I not specified them (or more appropriately, when I first tried the search without them specified!) I received hits like Romans 5:1, where the 3 words appear in the same clause, but are not related. Sometimes you have to construct multiple searches to narrow down what you're looking for. Dave Hooton put together a great primer on having a strategy for syntax searches here: http://wiki.logos.com/A_Strategy_for_Syntax_Search

    Does that make sense?

  • Rick Brannan (Logos)
    Rick Brannan (Logos) Member, Logos Employee Posts: 1,862

    This is actually a good example to start with a syntax template.

    Here I started with Cascadia's "Adjectival Modification" template, which is pretty strict — it will basically find πιστις Χριστος and nothing else once lemmas are inserted in the query.

    I added an additional lemma of Ιησους to the sibling phrase object. Rather than worry about order/hierarchy of Χριστος Ιησους or Ιησους Χριστος, I figured it would be easier if I just supported them both at the same spot.

    I also selected "Matching Skips Levels" on the sibling phrase object. This is what ends up causing most of the "multiples" to occur. One effect is everywhere that both Ιησους and Χριστος occur as lemmas will be a double hit. Another effect is that where the same words occur embedded in structures, multiple matches may occur because while the words are the same, the phrase structures contained within the hit(s) are different.

    I end up reporting 51 results, but there are 15 verses in the result list (including hits in Acts James and Revelation). Using Beta 2.

    image

    Rick Brannan
    Data Wrangler, Faithlife
    My books in print

  • Rick Brannan (Logos)
    Rick Brannan (Logos) Member, Logos Employee Posts: 1,862


    Tracking down every occurrence of pistis with a genitive name or pronoun after it is probably in my near future.

     

    That would be an excellent place for a syntax search!

    Here I used the search I did prevously as a starting point.

    [side note: did you know that you can use a previously saved search as a template for a new search? Open a new syntax search via File->Syntax Search. Click on the one you want to start with in the "Previous searches". Now you've just populated your new search with your old one as template.]

    In this instance, I'm not skipping levels because of the intent to tighly tie πιστις to what follows. Louw-Nida 93A is the proper noun, personal name subdomain. I also disallowed matching Ιησους as a lemma (note that Χριστος is not categorized as a name by LN). 39 results, including Ro 4.16-17 which speak of the faith of Abraham. Query below:

    image

     

    Rick Brannan
    Data Wrangler, Faithlife
    My books in print

  • Russ Quinn
    Russ Quinn Member Posts: 711 ✭✭

    Does that make sense?

    David,

    It is beginning to make more sense for me. This concrete example of an issue I have worked with before has provided good stimulation for me to dig a little deeper into how these resources work.

    Thanks for the link to Dave's primer. I hadn't seen that and will study it this afternoon.

    One thing I have realized in just the last little bit is that some of the information I assumed would be in the Syntactic resources is not actually there.

    I was surprised to see that there seems to be no information in Cascadia, Lexham or Opentext on the syntax of these genitives (i.e. objective, subjective, source, etc). I assumed that these designations would be part of the tagging system in some way.

    Obviously I have a lot to learn about these resources.

    Russ

  • Rick Brannan (Logos)
    Rick Brannan (Logos) Member, Logos Employee Posts: 1,862

    One thing I have realized in just the last little bit is that some of the information I assumed would be in the Syntactic resources is not actually there.

    I was surprised to see that there seems to be no information in Cascadia, Lexham or Opentext on the syntax of these genitives (i.e. objective, subjective, source, etc). I assumed that these designations would be part of the tagging system in some way.

    Obviously I have a lot to learn about these resources.

    Hi Russ.

    This sort of information is found in the Lexham Syntactic Greek New Testament. Here's a listing in the "Expansions and Annotations" resource for Ro 3.22. Note how it classes Ιησους and Χριστος.

    image

    There is presently an issue on my tasklist regarding searching for the "Syntactic Force" information using the syntax search features; it is non-functional at present. I also have some ideas on how to make this information available in the sentence graphs for the Lexham SGNT, but that will take a little more work.

    Hope it helps.

     

    Rick Brannan
    Data Wrangler, Faithlife
    My books in print

  • spitzerpl
    spitzerpl Member Posts: 4,998 ✭✭✭

    There is presently an issue on my tasklist regarding searching for the "Syntactic Force" information using the syntax search features; it is non-functional at present. I also have some ideas on how to make this information available in the sentence graphs for the Lexham SGNT, but that will take a little more work.

    I can't wait till you guess are able to get through the "gotta fix" checklist so you can get to the "get to play" checklist :-)

  • George Somsel
    George Somsel Member Posts: 10,153 ✭✭✭


    I'm studying pistis cristou and using the opportunity to figure out syntax search. I designed a syntax search to catch pistis cristou and pistis iesous cristou. It's not as good as a Morph search (some false positives, but I'm learning).

    My question is, why does the search results show the text from the verse multiple times? It makes the results difficult to use/interpret. Is this a bug/issue or is my search poorly constructed?

    Also, any ideas on how to improve this kind of search? Would there be a way to catch pistis iesous only too?

    EDIT: This screenshot is from Beta 2, but SR-7 has a very similar behavior.


    I limited my search to the NA27 and did a morph search.  For some reason L3 is better for this than L4 since in L4 "lemma:πίστις BEFORE 2 WORDS lemma:χριστός@ngsm" produces no result.  If I use "lemma:πίστις BEFORE 2 words lemma:Χριστός" on one computer I get 5 hits

    Rom 3.22
    Gal 2.16; 3.22, 26
    Phil 3.9

    while on the other I get none -- the morph search in L4 is flakey.  I cannot specify "lemma:χριστός" and the form "ngms" for it and get a result on either computer.  Much better is the morph search in L3 "[πίστις=N???] before 2 words [χριστός=NGSM]" which eliminates the spurious Gal 3.26 since that is governed by the preposition ἐν and is therefore in the dative.  Note that this catches both instances with or without the intervening Ἰησοῦ.

    george
    gfsomsel

    יְמֵי־שְׁנוֹתֵינוּ בָהֶם שִׁבְעִים שָׁנָה וְאִם בִּגְבוּרֹת שְׁמוֹנִים שָׁנָה וְרָהְבָּם עָמָל וָאָוֶן

  • davidphillips
    davidphillips Member Posts: 640 ✭✭

    Rick,

    Thanks for chiming in. I always forget about the templates, and it was good to see how you constructed the searches.

    George,

    Could you post a screenshot of your failed morph search? I ran it and it works fine:

    lemma:πίστις BEFORE 2 words lemma:Χριστός@NGSM gives 15 results in 4 verses:

    Rom 3:22, Gal 2:16, Gal 3:22, Phil 3:9

  • Kevin Becker
    Kevin Becker Member Posts: 5,604 ✭✭✭

    George,

    Could you post a screenshot of your failed morph search? I ran it and it works fine:

    lemma:πίστις BEFORE 2 words lemma:Χριστός@NGSM gives 15 results in 4 verses:

    Rom 3:22, Gal 2:16, Gal 3:22, Phil 3:9

    I'm not George, but what he is talking about is the difference between χ and Χ, for some reason the lack of capitalization throws Logos off.

  • Russ Quinn
    Russ Quinn Member Posts: 711 ✭✭

    Hope it helps.

    Rick,

    That is very helpful. Thank you for the direction.

    I have primarily used other tools for morphological searches in the past.

    I am having some difficulty getting my head around how the syntactical resources work.

    I found the tutorials on syntactical searching in L3 here: http://www.logos.com/logos3/videos.

    Would you still recommend them as a starting place for newbies or would it be better to wait for newer materials that are specific to L4?

    Thanks,

    Russ

  • Kevin Becker
    Kevin Becker Member Posts: 5,604 ✭✭✭

    This is actually a good example to start with a syntax template.

    Here I started with Cascadia's "Adjectival Modification" template, which is pretty strict — it will basically find πιστις Χριστος and nothing else once lemmas are inserted in the query.

    I added an additional lemma of Ιησους to the sibling phrase object. Rather than worry about order/hierarchy of Χριστος Ιησους or Ιησους Χριστος, I figured it would be easier if I just supported them both at the same spot.

    I also selected "Matching Skips Levels" on the sibling phrase object. This is what ends up causing most of the "multiples" to occur. One effect is everywhere that both Ιησους and Χριστος occur as lemmas will be a double hit. Another effect is that where the same words occur embedded in structures, multiple matches may occur because while the words are the same, the phrase structures contained within the hit(s) are different.

    I end up reporting 51 results, but there are 15 verses in the result list (including hits in Acts James and Revelation). Using Beta 2.

    Rick, thanks. This is really helpful.

    However, it has highlighted how poorly I understand this area. I recreated this search and it prompted some questions.

    image

    I apologize for embedding them in a picture, but it seemed like the best way to express my questions.

  • Robert Pavich
    Robert Pavich Member Posts: 5,685 ✭✭✭


    Which instances of pistis christou or pistis Ihsou christou should be translated "faith in Christ" is a hot debate among scholars.

    Russ, thanks for giving me that background on it...I didn't know that

    Robert Pavich

    For help go to the Wiki: http://wiki.logos.com/Table_of_Contents__

  • George Somsel
    George Somsel Member Posts: 10,153 ✭✭✭


    Rick,

    Thanks for chiming in. I always forget about the templates, and it was good to see how you constructed the searches.

    George,

    Could you post a screenshot of your failed morph search? I ran it and it works fine:

    lemma:πίστις BEFORE 2 words lemma:Χριστός@NGSM gives 15 results in 4 verses:

    Rom 3:22, Gal 2:16, Gal 3:22, Phil 3:9


    Note that the search string was copied from your own post.

    image

     This was done on my old computer since my laptop is on its way to the repair center to replace the DVD so I resorted to installing L4 on my old computer.  I had the same result on my laptop.  L4's morph search is flakey.  They really should bring back the old check-box method -- you check the boxes and add the lemma and IT WORKS.

    george
    gfsomsel

    יְמֵי־שְׁנוֹתֵינוּ בָהֶם שִׁבְעִים שָׁנָה וְאִם בִּגְבוּרֹת שְׁמוֹנִים שָׁנָה וְרָהְבָּם עָמָל וָאָוֶן

  • davidphillips
    davidphillips Member Posts: 640 ✭✭

    George,

    Thanks. While I agree that the morph search isn't perfected yet, I have had none of the issues you have. In this instance, I would suggest that it might be a problem, as the yellow box suggests, that you're still in the process of indexing (since this is a new install for you). Try it again when indexing is finished and see if it still has problems. If it works for me and not you, there's an issue. As I've said, any problems I've had with flakiness have been my errors in the past. I hope they continue to improve the morph search, because it's not as good as it was in L3 for various reasons, so I'm not disagreeing with you on that [:)]. I'm indifferent to new morph selection however (though it still seems a smidge buggy at times)

  • George Somsel
    George Somsel Member Posts: 10,153 ✭✭✭


    George,

    Thanks. While I agree that the morph search isn't perfected yet, I have had none of the issues you have. In this instance, I would suggest that it might be a problem, as the yellow box suggests, that you're still in the process of indexing (since this is a new install for you). Try it again when indexing is finished and see if it still has problems. If it works for me and not you, there's an issue. As I've said, any problems I've had with flakiness have been my errors in the past. I hope they continue to improve the morph search, because it's not as good as it was in L3 for various reasons, so I'm not disagreeing with you on that Smile. I'm indifferent to new morph selection however (though it still seems a smidge buggy at times)


    I would have come to the same conclusion were it not for the fact that I had the same result for the same search on my laptop which is what I used for the beta program so that it has been installed for a while and is fully indexed.

    george
    gfsomsel

    יְמֵי־שְׁנוֹתֵינוּ בָהֶם שִׁבְעִים שָׁנָה וְאִם בִּגְבוּרֹת שְׁמוֹנִים שָׁנָה וְרָהְבָּם עָמָל וָאָוֶן

  • Russ Quinn
    Russ Quinn Member Posts: 711 ✭✭


    Which instances of pistis christou or pistis Ihsou christou should be translated "faith in Christ" is a hot debate among scholars.

    Russ, thanks for giving me that background on it...I didn't know that

     

    Robert,

    George Howard has a helpful summary of the issues in his article "Faith of Christ" in the Anchor Bible Dictionary.

    Russ

  • Robert Pavich
    Robert Pavich Member Posts: 5,685 ✭✭✭

    Robert Pavich

    For help go to the Wiki: http://wiki.logos.com/Table_of_Contents__

  • Rick Brannan (Logos)
    Rick Brannan (Logos) Member, Logos Employee Posts: 1,862

    Hi Kevin.

    I finally have a little time to dig into your questions.

    image

    Question 1: "So this box says find a phrase with πιστις". No, not really. It says find a phrase that has πιστις as head of the phrase node. The word-level information of the head term of the structure is encoded at the node as well. Much of the dancing one has to do in OpenText.org involves finding a head within something. Cascadia is head driven, and thus populates this information up the nodes as well. You can specify it at the head and alleviate the need to find a structure general enough to be worth it but specific enough to keep all the chaff out.

    Question 2: "Terminal Node = a Word? Thus one of the terminal notes[sic]  = πιστις?" Two issues here. First, the template should probably just point to a word object; the effect would be the same and it would be less confusing. Every word has a Terminal Node as its head. Sometimes terminal nodes are easier to use when searching for patterns (alleviates the necessity of specifying part-of-speech morphology). Here note the agreement specified with the "Instance" property and the skip levels property. This also gets into your next question ...

    Question 3: "Since the lemma is already in the phrase 1 box, why is this needed?" In short, it is needed to highlight the proper word in the results. The phrase object has a lemma property, and it finds the hit. If we removed the Terminal Node object (along with "Anything", and the Word object, for that matter) it wouldn't affect the hit count at all. But it would affect highlighting. Here we're using the Terminal Node with Instance agreement (and the subsequent word node with instance agreement) to highlight the right stuff in the search results window. The "Instance" agreement basically allows you to jump across all levels of intervening hierarchy and find the Word or Terminal Node object that represents the same word instance as the head of the parent/ancestor node (here a phrase node).

    Question 4: "Why Word here? Why not another Terminal Node? (etc.)" A great question. I don't have an easy rule here apart from noting that sometimes it is easier to specify a Terminal Node and its type (or types) than a Word if you're basically filtering on part-of-speech. I think I used Terminal Node because it was closest and that's just where my brain was. Either could have been used in the same way (agreement on 'Instance', skipping levels) with the same results.

    Question 5: "Does putting multiple functions in a phrase box function like OR?" Yes, the list of criteria is implicity OR'd. So here the lemma of the head of the phrase could be either Ιησους or Χριστος.

    Sorry for the delay in getting answers to your questions. Hope it helps out.

    Rick Brannan
    Data Wrangler, Faithlife
    My books in print