Curious about Design Goals for Reverse Interlinears

Page 1 of 2 (29 items) 1 2 Next >
This post has 28 Replies | 0 Followers

Posts 709
Russ Quinn | Forum Activity | Posted: Tue, Dec 22 2009 10:17 AM

I am curious as to the design goals of Reverse Interlinears.

Is a goal to provide consistent results for morphological searches across the various versions?

The ESV separates the article from peirasmos in 3:10 while the NASB includes the article in the same field as peirasmos:

This yields different search results from the same search in each version:

Consider the differences between how the ESV and NASB deal with the phrase τῆς μελλούσης ἔρχεσθαι in Rev 3:10.

Of course this also yields no results for a similar search in the ESV and NASB Reverse Interlinears.

 

One would have to know that a search in a morphologically tagged Greek text is required to yield the appropriate results

 

Hence my question  . . . do these results reflect the design goals of searching the Reverse Interlinears?

Are these discrepancies intentional? Is the user expected to learn the nuances of various resources regarding their ability to return expected results? I suspect that the differences between the ESV and NASB concerning the article in the first example is something that should be fixed but I'm not as sure about the second example. Is there an advantage to how the second example differs between the ESV and NASB interlinears? Admittedly I have a hard time seeing it but am definitely open to an explanation.

Or do these differences reflect the need for ongoing tweaking to ensure consistent search results across all versions?

 

Posts 13423
Mark Barnes | Forum Activity | Replied: Tue, Dec 22 2009 10:23 AM

Have you seen this from Rick?

Posts 709
Russ Quinn | Forum Activity | Replied: Tue, Dec 22 2009 10:30 AM

Yes. Saw it right after I posted this Big Smile.

His response there is helpful to explain the alignment of the article to the ESV and might indicate that it is in fact a design goal for the Reverse Interlinears to yield varying results from morphological results.

Also, perhaps I am right that the difference between the NASB and ESV is an error to be corrected.

I would still find it helpful to better understand the design goals.

Posts 1436
LogosEmployee
Rick Brannan (Faithlife) | Forum Activity | Replied: Tue, Dec 22 2009 10:44 AM

Hi Russ

Russ Quinn:
His response there is helpful to explain the alignment of the article to the ESV and might indicate that it is in fact a design goal for the Reverse Interlinears to yield varying results from morphological results.

Word order in reverse interlinears is dictated by the translation line, not by the Greek line. You can mix & match in a reverse interlinear, but the primary purpose of a reverse interlinear is to get a user from a modern, familiar English translation into the Hebrew/Aramaic/Greek that underlies. Doing proximity/word order morph searches within a reverse interlinear is more like a byproduct; but note you'll always be searching with respect to the order implied by the English translation.

If you want to search an original language text sensitive to proximity/word order, then you should search the original language text. If you want to mix in English, you should use an interlinear and not a reverse interlinear *if* word order of the original language is a necessary part of your query.

Russ Quinn:
Also, perhaps I am right that the difference between the NASB and ESV is an error to be corrected.

Yes, I'd say the reverse interlinear alignment of the ESV NT is incorrect in the Rev 3.10 example that's been brought up. The problem I have in editing it is that the alignment was reviewed and corrected by the ESV editors, and I am not sure if we're at liberty to change it if we [Logos] think there is an error. Doubly complicating is that this alignment also exists in print; but I don't know that this needs to be a constraint for editing/updated the electronic edtiion. I'll need to poke around a bit to see how easy/difficult it would be to get changes for such things.

Rick Brannan
Data Wrangler, Faithlife
My books in print

Posts 709
Russ Quinn | Forum Activity | Replied: Tue, Dec 22 2009 11:09 AM

Rick Brannan:

If you want to search an original language text sensitive to proximity/word order, then you should search the original language text. If you want to mix in English, you should use an interlinear and not a reverse interlinear *if* word order of the original language is a necessary part of your query.

Thank you, Rick, for your response.

You have helped me better understand some of the constraints under which you are working with the Reverse Interlinears.

I appreciate the potential of having access to so many various resources. Hence my previous struggle with whether or not there might be advantages that I am not seeing with Reverse Interlinears.

A few friendly, humble suggestions/questions . . .

Would it be better to limit morphological searches to morphologically tagged texts?

Would it not be more user friendly to exclude the Reverse Interlinears as an option for morphological searching or at least return some sort of disclaimer when one attempts a search that we know will produce inaccurate results based on the limitations (or design) of the resource?

Is the burden on the user to learn enough about the particularities of various resources to know whether or not they should use it for a particular search?

Or should it be expected behavior that the search tool will help a user find the best tool for the most accurate results?

For example, when proximity operators are used in a morphological search of Reverse Interlinears, could the search tool recommend searching the best resource instead of just returning zero results with no explanation?

I am very supportive of Logos's emphasis on bringing the ability to interact with the original languages to those who have not had formal training but this is one area where I am concerned that there are a lot of opportunities for misunderstandings.

Please take my suggestions in the spirit of constructive criticism from someone who appreciates all you are doing.

Posts 27449
Forum MVP
Dave Hooton | Forum Activity | Replied: Tue, Dec 22 2009 4:10 PM

Russ Quinn:

Would it be better to limit morphological searches to morphologically tagged texts?

Would it not be more user friendly to exclude the Reverse Interlinears as an option for morphological searching or at least return some sort of disclaimer when one attempts a search that we know will produce inaccurate results based on the limitations (or design) of the resource?

Is the burden on the user to learn enough about the particularities of various resources to know whether or not they should use it for a particular search?

From a user perspective I would not like any type of search to be restricted to certain resources  eg. (coming ANDEQUALS <lbs-morph+el ~ VP??????>) produces Rev 3.10 in ESV. The goal of this search would be unique to a translation such as the RI's and has no bearing on accuracy compared to a Greek text (where one would use a Greek lemma rather than an English word).

Russ Quinn:
I am very supportive of Logos's emphasis on bringing the ability to interact with the original languages to those who have not had formal training but this is one area where I am concerned that there are a lot of opportunities for misunderstandings.

Should the software also restrict or warn users of the limitations of a Strong's number search?

Dave
===

Windows 11 & Android 8

Posts 709
Russ Quinn | Forum Activity | Replied: Tue, Dec 22 2009 4:36 PM

Dave Hooton:

From a user perspective I would not like any type of search to be restricted to certain resources  eg. (coming ANDEQUALS <lbs-morph+el ~ VP??????>) produces Rev 3.10 in ESV. The goal of this search would be unique to a translation such as the RI's and has no bearing on accuracy compared to a Greek text (where one would use a Greek lemma rather than an English word).

And yet a morphological search for articular occurrences of peirasmos in the ESV does not produce an accurate result.
That such an arcane search argument is required to get a desired result really only strengthens my concern that this is not a user friendly approach.

One should expect the software to produce the results that one would expect from the parameters of the search.

Of course this does not address the problem of proximity searches. The results of many proximity searches using reverse interlinears can not be trusted (per Rick's explanation above). My question is simply . . . then why allow them? They only have the potential to mislead.

Dave Hooton:

Should the software also restrict or warn users of the limitations of a Strong's number search?

No, because there is nothing in the software that has the potential to mislead one into thinking that they should expect more from search based on Strong numbers than it is able to deliver.
The problem with the Morph Search Tool allowing you to search a Reverse Interlinear is that it does not yield the results you would expect because of their inherent limitations as a tool for morphological searching. 

Allowing morphological searches on resources that will surely produce less than accurate results is less than helpful. It will be confusing to new users. It increases the learning curve for the software. It leaves the software wide open for criticism by competitors concerning its reliability.

I think the better approach would be to limit morphological searching to morphologically tagged texts where the results are known to be accurate. Reverse Interlinears could then be accessed from the Aligned view of the search results if one chooses.

 

 

 

Posts 709
Russ Quinn | Forum Activity | Replied: Tue, Dec 22 2009 4:49 PM

Dave Hooton:

Should the software also restrict or warn users of the limitations of a Strong's number search?

Simpler answer . . . 

Any search that leads one to believe that it is something that it is not should have a disclaimer.

When the following straightforward query does not yield accurate results, it should not be left to the user to have to discover the ins and outs of how a reverse interlinear is constructed to understand why. 

This could be more user friendly.  

Posts 5337
Kevin Becker | Forum Activity | Replied: Tue, Dec 22 2009 5:11 PM

Russ Quinn:

Simpler answer . . . 

Any search that leads one to believe that it is something that it is not should have a disclaimer.

When the following straightforward query does not yield accurate results, it should not be left to the user to have to discover the ins and outs of how a reverse interlinear is constructed to understand why. 

Russ,

Yes! I know that Logos is using the power of these tools to advertise that one can benefit from Greek without the hard work to learn it. This type of statement is overly optimistic. Using these tools without an introductory grammar and syntax (whether a course or some good self study) will lead to people making unsupportable conclusion about what "the Greek says." Some form of this problem has been around since Strong's numbers were introduced but the software makes it easier to access this information without really understanding how to use it. Just like the country preacher said, "Now the Greek word translated "therefor" here is yap, and whenever you see it you have to ask what the yap is there for."

I am encouraged at least by the high learning curve to perform morph searches. It seems like someone without the benefit of a Greek course or having read a grammar cover to cover will be able to construct much more than a lemma search.

Posts 27449
Forum MVP
Dave Hooton | Forum Activity | Replied: Tue, Dec 22 2009 5:42 PM

Russ Quinn:
And yet a morphological search for articular occurrences of peirasmos in the ESV does not produce an accurate result.

We have just demonstrated that that should be expected  from a translation which will necessarily not reflect the word order of the ancient text. An RI is useful for searches related to the translation but I can't agree that the software should flag its limitations wrt the ancient text.

Russ Quinn:
That such an arcane search argument is required to get a desired result really only strengthens my concern that this is not a user friendly approach.

The query (coming ANDEQUALS <lbs-morph+el ~ VP??????>) was a Bible Search as a Morph Search does not recognise ANDEQUALS. I would much prefer a simpler Search format but this issue is not unique to Interlinear bibles.

 

 

Dave
===

Windows 11 & Android 8

Posts 709
Russ Quinn | Forum Activity | Replied: Tue, Dec 22 2009 6:15 PM

Kevin,

I am supportive of Logos's goals of using their tools helping people learn Greek and Hebrew.

That is why I am concerned that the potential confusion related to a morphological search of a Reverse Interlinear presents an unnecessary hurdle.

Dave,

We agree on the proper use of IR's. You and I are knowledgeable enough to use the tools as they are. I've got an undergraduate degree in Greek, a MDiv, and am completing a PhD in New Testament Studies. I am also somewhat savvy with fairly high level computing. I recognize your abilities because I have learned from you on this forum several times.

My concern with this issue is for what I perceive to potentially confusing to those who are not as strong in one of those skill sets. I'm just trying to say that the software could do more to alleviate the acknowledged problems related to morph searching RIs.

Can we agree on that?

 

Posts 27449
Forum MVP
Dave Hooton | Forum Activity | Replied: Tue, Dec 22 2009 6:59 PM

Russ Quinn:
When the following straightforward query does not yield accurate results, it should not be left to the user to have to discover the ins and outs of how a reverse interlinear is constructed to understand why. 

The main reason can readily be demonstrated by reference to the numbers against ancient words indicating that they are not in the same order as the original text. This is just a matter of education (and the same search can be used in a Greek/Hebrew bible!).

Dave
===

Windows 11 & Android 8

Posts 709
Russ Quinn | Forum Activity | Replied: Tue, Dec 22 2009 7:24 PM

Dave Hooton:

Russ Quinn:
When the following straightforward query does not yield accurate results, it should not be left to the user to have to discover the ins and outs of how a reverse interlinear is constructed to understand why. 

The main reason can readily be demonstrated by reference to the numbers against ancient words indicating that they are not in the same order as the original text. This is just a matter of education (and the same search can be used in a Greek/Hebrew bible!).

 

You are right again, my friend. These things can be demonstrated. My question is should a user be required to know this before they can execute a successful morphological search? What would prompt an unenlightened user to seek to discover this knowledge? Isn't more probable that they would just unknowingly accept the wrong results?

Posts 27449
Forum MVP
Dave Hooton | Forum Activity | Replied: Tue, Dec 22 2009 7:53 PM

Russ Quinn:

My concern with this issue is for what I perceive to potentially confusing to those who are not as strong in one of those skill sets. I'm just trying to say that the software could do more to alleviate the acknowledged problems related to morph searching RIs.

Can we agree on that?

We can both point to one reason (what I called the main reason in my previous post) which alleviates the need for any presumed language/technical skills on the user's part. Any software intervention would be difficult to justify and likely have unintended side effects. Any software warning would create as much confusion as you are trying to avoid. In the 4+ years since the NT Reverse Interlinears were made available in v3 I cannot recall any user having a significant issue about their use vs. the Greek bible. But there have been many, many posts wanting to know how to formulate a query with/without the article in a Greek bible!

 

Dave
===

Windows 11 & Android 8

Posts 709
Russ Quinn | Forum Activity | Replied: Tue, Dec 22 2009 8:45 PM

Dave,

I have to confess I am having some difficulty knowing quite how to respond to your comments. 

In the past I have found your posts to be very helpful but I am not following your train of thought. 

Dave Hooton:

Any software intervention would be difficult to justify and likely have unintended side effects. 

I'm not sure how to respond to such a broad statement concerning unspecified software solutions. You are obviously more knowledgeable about all possible solutions than I am able to discuss.

Dave Hooton:

Any software warning would create as much confusion as you are trying to avoid.

Again a pretty broad statement about unspecified solutions. I find it almost impossible to interact with these types of assertions in an intelligible way. However, I find it difficult to believe that a simple warning accompanying proximity searches of the ESV Reverse Interlinear that encourages a more appropriate search would create more confusion than wrong results with no explanation.

Dave Hooton:

In the 4+ years since the NT Reverse Interlinears were made available in v3 I cannot recall any user having a significant issue about their use vs. the Greek bible. 

Here again, Dave, I am unable to debate your absolute knowledge of all issues users have encountered. Since all users have not reported their experiences to me, this type of knowledge is truly beyond my capability to critique. One explanation I might offer for the lacuna of feedback regarding this issue is that Reverse Interlinears, while present in L3, did not enjoy the centrality they do in L4. It really never occurred to me to attempt this type of search in L3 since the ESV Reverse Interlinear was way down the list of other obviously unrelated texts like the First Targum to Esther. The interface of L4 seems to encourage choosing the Reverse Interlinears in its morphological searches by placing it at the top of the list of available resources to search. Of course, I can only make suggestions based on what my impressions are.
Dave Hooton:
But there have been many, many posts wanting to know how to formulate a query with/without the article in a Greek bible!
Rightly so. That is what people need to be learning. How to interact with the text in a way that will produce reliable results. Please forgive me if I seem a bit frustrated in this post. I have a good deal of respect for you. That is why I am struggling to try to understand your point.

Posts 13423
Mark Barnes | Forum Activity | Replied: Wed, Dec 23 2009 2:27 AM

Russ,

I find myself more in agreement with Dave here. I'm certainly not in favour of warnings, etc.

There are two issues at stake here:

  • Possible errors in Reverse Interlinears
  • Limitations of Reverse Interlinears

We shouldn't let the first cloud our discussion of the second.

Logos 4 is now stuffed full of databases, morphological and syntax - more than ever before. They offer a choice of databases, because the academy/church has not yet agreed which tagging system is the one that is 'right'. For the same reason we have different English translations (though here the variety is obviously greater) and even Greek Texts.

Scholars/students are quite used to working with a variety of databases created with slightly different methodologies in order to verify results to to highlight the practical differences of those methodologies. We'll continue to do that in Logos 4, I'm sure. Pastors too, at least those seminary trained, will have the same experience, though their greater time constraints will sometimes mean they don't have time to check everything as thoroughly as they would like. The vast majority with language training will do their important morphological searches in NA27 I'm sure, and use reverse interlinears for other purposes.

Those who have not been formally trained would potentially be in danger of mis-interpreting data coming from these searches, but frankly I don't think this is a major issue. Why? (a) Because the problem doesn't actually occur that often. (b) Because this group will tend to be working more from the Passage Guide / Exegetical Guide, where the problems don't occur in the same way.

So whilst I agree there is some cause for concern, my concern is very minimal. I'm certainly willing to pay the price of an incorrect statement in every 100 sermons/Sunday school lessons for the more open access to original languages that reverse interlinears give.

Posts 27449
Forum MVP
Dave Hooton | Forum Activity | Replied: Wed, Dec 23 2009 5:22 AM

I'm sorry, Russ but our posts have crossed a couple of times. Let me attempt a more concerted response:-

1. Is there a compelling reason to be pro-active in this matter?

My experience strongly suggests that most users have trouble formulating the appropriate query and knowing which is the right search method. So they ask! In that process I have occasionally questioned the bible they think is appropriate. An RI has not been used to assert a conclusion about ancient language texts. So I have to allow that most users verify the results they produce and ask "why is it not so?"  or realise why and move on to a more productive method.

 

2. Software Intervention vs Software Warning?

Software intervention is rarely warranted. You see, a Search tool will enable very complicated queries in the right hands and should enable relatively simple queries for the occasional user. The main concern of the software is that the query syntax is valid and will enable a search to be conducted. It is not the job of software to interpret the query with respect to its intended use or the possible accuracy of results. It accurately reports what it finds!

A warning per se can be a two-edged sword. "A proximity query with Greek/Hebrew words in a Reverse Interlinear can be inaccurate" might mean that a reverse interlinear is not to be trusted or that the search method cannot handle RI's. "A proximity query with Greek/Hebrew words will be more accurate if you use a Greek/Hebrew bible" carries the same degree of ambivalence. And what is a "proximity query"?

3. What can be done?

The issue could be documented in the wiki as a Search FAQ.

Dave
===

Windows 11 & Android 8

Posts 709
Russ Quinn | Forum Activity | Replied: Wed, Dec 23 2009 11:36 AM

Mark Barnes:

Russ,

I find myself more in agreement with Dave here. I'm certainly not in favour of warnings, etc.

There are two issues at stake here:

  • Possible errors in Reverse Interlinears
  • Limitations of Reverse Interlinears

We shouldn't let the first cloud our discussion of the second.

I totally agree with you that these are separate issues. That is why I have offered two different examples in my previous posts.

 

  • The issue related to searching for articular uses of peirasmos is an example of a problem related to errors. There really is no way to compensate for this through software because they cannot be anticipated. The only antidote for this is an intimate familiarity with individual resources.
  • My second example of searching for "lemma:ὁ WITHIN 1 WORDS lemma:μέλλω WITHIN 1 WORDS lemma:ἔρχομαι" is an entirely different matter. It can be anticipated that this search will not yield accurate results in either the ESV or NASB because of the way the Interlinear is formatted as a database. It is reasonable to assume that a user would expect accurate results from this search because of the way the Morph search tool presents its options. As everyone has admitted in this discussion, users are required to obtain further information about the intricacies of individual Interlinears to understand the results of a straightforward search. My suggestion is related to the user friendliness of this situation. I would argue that there are more user friendly ways to help users understand which combination of tools are best used to obtain accurate results. These scenarios can be anticipated because we all know that the false results are a function of the limits of these resources. If it is known that some searches of some resources will produce inaccurate results, why wouldn't it be wise to alert the user.

 

Mark Barnes:

Those who have not been formally trained would potentially be in danger of mis-interpreting data coming from these searches, but frankly I don't think this is a major issue. Why? (a) Because the problem doesn't actually occur that often. 

I'm not sure what the relationship is between the frequency of a potential problem and its importance. Surely accuracy is the goal for all aspects of Bible software.

Mark Barnes:

(b) Because this group will tend to be working more from the Passage Guide / Exegetical Guide, where the problems don't occur in the same way.

Correct functionality in one function does not compensate for problems in another.
Two points here, if I may:

 

  1. I appreciate the ways in which L4 leverages the strengths of Reverse Interlinears. I think L4 is ingenious in some of this functionality. I am only arguing that there could be more helpful ways to deal with their use for morphological searches. I would recommend leveraging their strengths where it is appropriate but also avoiding their limitations where it is appropriate. I still wonder if the best solution might be to limit morphological searches to the resources that are known to produce accurate results and then use the Aligned view as an entry point to open the appropriate Reverse Interlinear.
  2. One unintended consequence of discovering that the morphological searches of RIs are not reliable is that perceived unreliability in one part of the software damages the perceived reliability of other parts.

 

Mark Barnes:

So whilst I agree there is some cause for concern, my concern is very minimal. I'm certainly willing to pay the price of an incorrect statement in every 100 sermons/Sunday school lessons for the more open access to original languages that reverse interlinears give.

I am not arguing for limiting access to Reverse Interlinears. As I said above, leverage their strengths where appropriate. They can certainly be a powerful tool for certain applications. Just don't encourage their use where their use is not appropriate.

Posts 1669
SteveF | Forum Activity | Replied: Wed, Dec 23 2009 12:40 PM

Mark, Russ & David etc:

I've appreciated your discussion.

There is one area though where you have no fears.

Some of us with slower computers might out of curiosity attempt this tool -- but I wouldn't recommend it..

Until optimized, the Morph tool is very slow; has no "spinning dial" to indicate anything is even happening, which then leads to other buttons being chosen which may or may not cancel earlier commands;  and in addition it is very "touchy" as to what and where something is actually clicked on.

It therefore (at least on my older 1 core machine) remains at best an elite "curiosity" rather than a useful part of the larger logos "package."

Regards (and respect)

Steve F

 

Regards, SteveF

Posts 5337
Kevin Becker | Forum Activity | Replied: Wed, Dec 23 2009 1:06 PM

This is a fascinating discussion. I think morph searches in the reverse interlinears are generally Ok when looking for a single word ie. a specific lemma (with or without mophology restrictions). It's when you start looking for combinations of features, often by using proximity that these searches break down. What if for more complex searches on an interlinear (constructed with a Boolean operator or WITHIN etc and don't reference the surface text) generate a link above the results that would say Run this search in XXXXX where XXXXX would be the an original language text with the morphology scheme equivalent to the reverse interlinear. If we really wanted to closely attend to Russ' concerns the program could automatically run the second search in the background and compare the number of hits and if they differ greatly give a stronger worded link Get Better results by running this search in XXXXX. Then the user of this tool will be given the chance to learn about the limitations of morph searches in the reverse interlinears and what searches they can do well and what ones are best left to the primary texts.

Page 1 of 2 (29 items) 1 2 Next > | RSS