INFER command exact phrase usage

[Joe Weaks]

If I search for:

 [iNFER=i 15,0,0 GNT-T] <AND> [RANGE Luke] 

where the "GNT-T" window is displaying Mark, then my result is giving me every place where Luke has an exact phrase of 15 or more words in one verse that is an exact match of 15 or more words in a single verse in Mark (precise inflected form with no intervening or added words).

 

I just wanted to check my logic.

 

A real limitation here is that a phrase with a verse marker in the middle of it fails.

[RobM]

Your logic is correct. However, if you click on "Lemmas" rather than "Words" you will get more results (which you may not want).

 

I just performed your search (but changed it to "Lemmas") and found three verses in Luke instead of two. Luke 9:24 was added: Luke 9:24 was connected with Mark 8:35. The difference between the two is: Luke has an aor subj of apoluw whereas Mark has the fut ind of opoluw.

[Joe Weaks]

(which you may not want)

I do not want.

 

Thanks.

 

I'm really bummed about this Verse field limitation. It stymies some great use of this feature for two corpuses that share literary dependency.

[RobM]

I do not want.

 

Thanks.

 

I'm really bummed about this Verse field limitation. It stymies some great use of this feature for two corpuses that share literary dependency.

 

I agree that the Verse field does limit the search. At the same time, however, if you want to find places where the string of words begins near the end of one verse and goes into the beginning of the next verse, you could simply shorten the length of words in the phrase so that it could pick up the shorter phrases, but still display the results as if they were all one connection. Does that make sense?

 

For example:

If you want to be able to find a parallel with the following (the bolded part)

 

Mark 5:2 When He got out of the boat, immediately a man from the tombs with an unclean spirit met Him,

Mark 5:3 and he had his dwelling among the tombs. And no one was able to bind him anymore, even with a chain;

 

Set the search to look for phrases 5 words long with no ambiguity built in, and if there are results for the last phrase of 5:2 as well as the first phrase of 5:3 and both of those results appear in one verse or across two verses (as here), then the search will find it.

 

There is one downside, however. Opening up the search to 5-word-phrases, you will get more hits, some of which may be irrelevant. But, I think this is a decent work-around.

 

What do you think? (Does this address what you are lamenting?)

[Joe Weaks]

What do you think? (Does this address what you are lamenting?)

Rob, really, it simply highlights the lament. You haven't thought through all the steps that you are wishing would work.

 

For one, we don't know if there are 1, 2, 3, 4, 4, 6, 7, 8, 9, 10, 11, 12, 13 or 14 words of the phrase that are in verse A and with a corresponding 14, 13, 12 ,11, 10, 9, 8, 7, 6, 5, 4, 3, 2, or 1 words of the phrase that fall in verse B. All of those configurations are possible.

 

And, in your example in searching for "an unclean spirit met Him, |verse break| and he had his dwelling among the tombs", both of these phrases would "hit" only if a verse break happens at the same place in each corpus.

 

This does not present a doable workaround.

Thanks for input, though. It's a great feature and can do alot, still.

[RobM]

You haven't thought through all the steps that you are wishing would work.

 

Actually, I think I have...

 

For one, we don't know if there are 1, 2, 3, 4, 4, 6, 7, 8, 9, 10, 11, 12, 13 or 14 words of the phrase that are in verse A and with a corresponding 14, 13, 12 ,11, 10, 9, 8, 7, 6, 5, 4, 3, 2, or 1 words of the phrase that fall in verse B. All of those configurations are possible.

 

I grant you that we don't know how long the string of words would be in Verse A along with Verse B. Many legitimate connections would be missed, I agree. However, by searching for phrases with fewer words in them the possibility of finding such connections becomes much more possible. Granted, the search can't be narrowed to only find 15 word phrases that wrap from verse A to verse B (especially if the wrapping occurs in Mark but not Luke, for example). However, by identifying phrases of 5 or 4 or even 3 words it will highlight them all and you simply need to sift through and do the "search back" search to find the corresponding passage in the other book. Using this method (I searched for 4-word phrases using Lemmas), I identified in a couple minutes the connections between Luke 5:22-24 and Mark 2:8-10; Luke 9:16-17 and Mark 6:41-42. It's really a matter of whether or not you're willing to sift through the data. By setting up searches for longer phrases, sure there will be fewer and more 'juicy' connections, but much is left out. By setting up searches for shorter phrases more legitimate connections will be surfaced as well as many combinations of random words.

 

And, in your example in searching for "an unclean spirit met Him, |verse break| and he had his dwelling among the tombs", both of these phrases would "hit" only if a verse break happens at the same place in each corpus.

 

 

Regarding your objection about my example. You're half right and half wrong. When the search compiles the list of phrases to search for in your base text, it will compile the phrases "an unclean spirit met Him" and "and he had his dwelling among the tombs." Now, simply set up the search to find 5-word phrases. Let's say in the search text, as you object, the verse break is not in the same place, but occurs halfway through the first phrase. What you need to realize is that the words cut off in the first phrase are added to the next verse in the second phrase ending up with: "an unclean spirit" and "met Him and his dwelling among the tombs". I grant you that the search will not surface "an unclean spirit", but it will surface "met Him and his dwelling among the tombs". You simply have to go to that hit and look at the preceding verse and lo and behold the rest of your base text is there! Presto, the work-around! And it works regardless of the direction of the unevenness of the verse break within a phrase. I grant you it makes the sifting through of the data more tedious, but it is still surfacing data that you probably wouldn't have found otherwise (or at least it would have taken much longer to surface it any other way).

 

Is that more clear?

[Joe Weaks]

Thanks for the thoughts on this, Rob.

 

It's really a matter of whether or not you're willing to sift through the data.

Yes, of course, which is also the point... the lament. To have anything meaningful to say about exact matching phrases between two literary-dependent corpuses, the least you would have to do is search for is 3 word phrases (that's 570 verses) and trust that your sifting finds the connections, but you're still left with incomplete and imperfect data. The ?,0,0 query has the goal of finding exhaustive results. In this approach, your only sound option is to search for 1 word phrases, and then your sifting has been reduced to doing the whole thing manually (2 words is the smallest phrase you can search for with INFER).

 

Faith in searching for smaller chunks fails consistently for another reason. The INFER command ignores phrases whose makeup consists significantly of very common words (think articles). So, a search for [iNFER 2,0,0 GNT-T 2] FAILS in just the second verse, when it does not find τὴν ὁδόν σου

[RobM]

Yes, good discussion. Though it seems we may have been saying the exact same thing regarding the search's abilities; I think the difference was in perspective: I was pointing out how far it can get you in doing intertextual studies whereas you were lamenting about how it doesn't get you quite as far as you'd like. Believe me, I feel your pain. I don't know if you gathered this or not, but I was the one who wrote the ETS paper that David blogged about recently. Last fall when I was doing the research that that paper is based on I was doing exactly what I've been describing to you in Hebrew and that before they refined the search to leave out the noise of strings of articles and prepositions. When I would perform the search there would literally be around 400-500 verses found connecting Judges and Exod 20-40 and around the same amount for Judges and Deut. Sift is exactly what I did, and it took a long time. But at the end of the day, I could not have done what I did with the INFER search in an entire year much less a single semester without it.

 

I guess what I'm saying is, I sympathize with your lament, but I also recognize how difficult the same study would be without the search. And like I said in my paper, it depends on whether or not you are doing rigorous research or just preparing for a Bible study.

[RobM]

Now, I also agree, it definitely would be cool if the search could be adjusted to ignore verse boundaries. I believe I've (and others) asked about that under a different topic, and the response was something about it being too complicated to change based on things that I'm not knowledgeable enough to comment on.

[Joe Weaks]

I don't know if you gathered this or not, but I was the one who wrote the ETS paper that David blogged about recently.

Rob, I knew a paper was given on the INFER command, but I did not make the connection. I am lucky then to have posted on a day when it garnered your attention. It was very helpful to think things through with someone well-versed in what can be a powerful, but enigmatic command. (I suspect we've had a few onlookers thinking "huh?")

[RobM]

Rob, I knew a paper was given on the INFER command, but I did not make the connection. I am lucky then to have posted on a day when it garnered your attention. It was very helpful to think things through with someone well-versed in what can be a powerful, but enigmatic command. (I suspect we've had a few onlookers thinking "huh?")

 

It was also helpful for me as well because I've been using the search primarily in OT studies and have not used it much in the NT (except for tinkering). There are not huge differences in how the search should be used in either testament, but there are issues (some of which were touched on in the thread) that are unique to both that could effect one's methodology.

 

I suspect we've had a few onlookers thinking "huh?")

Probably

[Helen Brown]

Just a quick point. The phrases searched are overlapping, and if they are exact even a shorter phrase search will immediately highlight the longer passages that are matches. Then you should quickly be able to scroll through your results to find those that have large matching sections. I agree that overlapping phrases could be highlighting from different parts of the source, but I think this is unlikely to be a significant problem.

[RobM]

Just a quick point. The phrases searched are overlapping, and if they are exact even a shorter phrase search will immediately highlight the longer passages that are matches. Then you should quickly be able to scroll through your results to find those that have large matching sections. I agree that overlapping phrases could be highlighting from different parts of the source, but I think this is unlikely to be a significant problem.

 

Helen,

This is something I was assuming at least in what I wrote and how I read Joe's posts. But that is a good point to make, especially for those who perhaps had trouble following my and Joe's conversation: Building ambiguity into the search does does not leave out the longer phrases in the results; rather, the search perceives those longer phrases as several shorter phrases right next to each other and amongst each other using many of the same words.

 

I agree that overlapping phrases could be highlighting from different parts of the source, but I think this is unlikely to be a significant problem.

 

Actually, I would not consider this a problem at all but a fascinating situation. If you have two phrases from two different locations with different contexts combined together in a new context, the author is doing something quite interesting that could shed new light on the meaning of a passage as well as reveal another rhetorical strategy of a given author. In my paper I demonstrate that the author of Judges is doing just that with two passages in Exodus: the author combines two phrases, one indicating blessing and another indicating curse. The author took a phrase that used to promise blessing and turned it upside down to use it as a curse prediction against the people (see my discussion of Judg 2:3 in my ETS paper).