Hebrew/Greek Word study issues 14.0.2

[miketisdell]

I have been using 14.0.2 for a little bit this morning and it appears that the most serious issues have been address (However, I haven't yet checked syncing). 

 

However, when I did a word study on תורה and added the LXX to my text compare, I got many wrong answers. 

 

If we look at the first wrong answer i.e. "ὁ" and see the verse returned, we can quickly see the problem i.e. looking at Gen. 26:5 we can see the Hebrew and Greek translation. While the LXX offers a correct translation of this phrase that includes the article, the article does not translate תורה; in this case, the article isn't even present in the Hebrew text because construct nouns in Hebrew cannot take an article. I have color coded the matching phrases in both text to show how they should be matched. The problem is that the word study matched "ὁ" with "תורה" (presumably because it is legitimately part of the Greek phrase "καὶ τὰ νόμιμά μου" that translates ותורתי), however, the correct match with "תורה" is "νόμιμά"

 

Note for clarity: "τὰ" is just a different case of "ὁ"

 

 

 

 

There are similar issues with the matches for καὶ, σύ, ἐν, etc... i.e. the are correct part of a phrase that translates a Hebrew phrase that includes תורה but they are not being used as a translation of תורה

 

 

[John Fidel]

I concur with your issue. It really has to do with the MT/LXX Parallel tool. If you open that in parallel with the LXX and HTM you can see what is happening. See screenshot below. There are multiple Greek words used to match the Hebrew and each is listed separately in the WS report.

 

Edited December 2, 2022 by John Fidel

[John Fidel]

I would suggest the solution would be for the WS to include all the Greek words together rather than separate just like the MT/LXX

 

[Michel Gilbert]

@miketisdell and @John Fidel

 

I’m not a programmer, but yesterday’s post about a word study of יהוה with its link to Mark’s post about GK and Strong coding suggests there might be a more systemic problem.  Original language texts seem to tag only the beginning of a phrase, e.g., an article that shows up in Greek or English text comparisons.

 

Edit: The K/Q was probably tagged in different ways in original texts and translations.

 

Edited December 2, 2022 by Michel Gilbert

[John Fidel]

Often time the problem is which word in the English text is tagged or in the case of the Mounce NT, how many words are tagged. The example below shows demonstrates this. Down is tagged and written is not, nor is it included under the search for written. Fixing this requires review and retagging the English text, at least i think so.

 

[ScottDF]

@John Fidel, you are quite right. I regularly send in suggestions for corrections to tagging, especially of the English texts. It doesn't take long and they are incorporated into the text.

 

The original manuscripts are infallible, the tagging, not so much. 😐

 

Edited December 2, 2022 by ScottDF

[John Fidel]

As long as users understand the limitations when using the tools. Most advanced users understand the limits of number tagging English texts to the original languages. However, some users may not especially if they are working with the original languages without understanding them. I am guilty of that most of the time, but try and work through it.

[miketisdell]

15 minutes ago, John Fidel said:

I concur with your issue. It really has to do with the MT/LXX Parallel tool. If you open that in parallel with the LXX and HTM you can see what is happening. See screenshot below. There are multiple Greek words used to match the Hebrew and each is listed separately in the WS report.

 

That is exactly what is expected in the parallel tool; however, they cannot take every word in the phrase used in the translation and equate it with the single word in the Hebrew text. Looking at the tagging in the LXX should allow them to rectify this issue. 

 

I did notice that that cross highlighting has similar problems when looking when highlighting the Hebrew to Greek i.e. the conjunction and pronominal suffix are not highlighted but the entire Greek phrase is highlighted.

 

 

But when looking at cross highlighting from Greek to Hebrew the entire Hebrew phrase is highlighted as each word in the Greek phrase is highlighted. This could be better but it is acceptable. 

[miketisdell]

52 minutes ago, John Fidel said:

I would suggest the solution would be for the WS to include all the Greek words together rather than separate just like the MT/LXX

 

That would be a better solution, but I wonder if tagging could be looked at to make a better connection. 

[A. Smith]

I get incorrect results when adding the LXX as a comparison text to a GNT word study. Note the word is present in the LXX, as evident from the word usage search: 

[A. Smith]

The same null result obtains when graphs are added to include LXX usage. 

[mgvh]

@A. SmithHmmm.... something odd is going on...

I see that you were using NA 28 Greek NT (Sigla) text.

Change that to the NA 28 Greek NT (i.e., not-Sigla) text. Doing so does cause the LXX instances to appear for me.

You can also use Greek Bible (LXX + GNT) to get results, but that GNT is Westcott-Hort and not NA28.

Edited December 3, 2022 by mgvh

[A. Smith]

1 hour ago, mgvh said:

@A. SmithHmmm.... something odd is going on...

I see that you were using NA 28 Greek NT (Sigla) text.

Change that to the NA 28 Greek NT (i.e., not-Sigla) text. Doing so does cause the LXX instances to appear for me.

You can also use Greek Bible (LXX + GNT) to get results, but that GNT is Westcott-Hort and not NA28.

This a may be. But that’s silly. We should be able to search the modern critical texts. 

Edited December 3, 2022 by A. Smith

[miketisdell]

On 12/2/2022 at 8:54 AM, Michel Gilbert said:

@miketisdell and @John Fidel

 

I’m not a programmer, but yesterday’s post about a word study of יהוה with its link to Mark’s post about GK and Strong coding suggests there might be a more systemic problem.  Original language texts seem to tag only the beginning of a phrase, e.g., an article that shows up in Greek or English text comparisons.

 

Edit: The K/Q was probably tagged in different ways in original texts and translations.

 

Original language texts do not tag only the beginning of a phrase. 

 

The Hebrew words, suffixes, and prefixes are tagged with numbers to a Hebrew lexicon and the Greek text is tagged with numbers to a Greek lexicon, but there is nothing that ties these tags together. Tov's MT-LXX DB appears to be used for linking the Hebrew and Greek texts together and it ties equivalent phrases together. In the example I provided the Hebrew uses a noun with prefix and suffix which equates to entire Greek phrase. The problem is that the word study is taking each word in the Greek phrase and equating it with the single Hebrew word selected in the word study which is incorrect. The easiest fix would be to tie the entire selected phrases in both DB's together. In order to tie the words together, one would need to have some sort of DB that links the specific words together i.e. tagging the LXX to the Hebrew like English bibles are tagged to the Hebrew. 

ETA - The issue of a single Hebrew "word" being equated with an entire Greek (or English) phrase is quite common because things like prepositions and pronominal suffixes are attached in Hebrew but stand alone in many non Semitic languages. 

 

Edited December 3, 2022 by miketisdell

[mgvh]

1 hour ago, A. Smith said:

This a may be. But that’s silly. We should be able to search the modern critical texts. 

Agreed. I was just pointing out why it is an anomaly. This gives the Accordance folks a clue on how to fix it.

[Michel Gilbert]

@miketisdell

Hi, Mike,

When I posted, I was thinking about the history of Bible software in general and how early on they would have written algorithms for aligning interlinears. I thought that they would have searched for the beginnings of words to do this. My post wasn't one of my best; what you said makes more sense.