Better alignment of various Hebrew TEXT modules in cross highlighting and cross textual searches

[Accordance Enthusiast]

Very often verses that contain the same LEMMA more than once are incorrectly linked to corresponding texts. This also often occurs where Ketiv Qere shows up. The Ketiv in HMT-W4 is erroneously linked to the Qere in ETCBC.

 

This also gives false searches where SUPPOSEDLY the tagging is different in HMT than ETCBC, but in reality the tagging is the same and the cross-linking is wrong.

 

Please correct this in the near future!

 

[Jordan S]

The search results are exactly what you are asking for: The same verb occurs in the same verse in both the HMT and ETCBC. In the case of the former it occurs as an imperfect and in the case of the latter it occurs as not an imperfect.

[Accordance Enthusiast]

Hi Jordan, I'm not dumb!

 

If you look at the tagging you will see that HMT-W4 and the ETCBC have the SAME parsing, whereas the search should bring up examples where they are different.

 

BOTH HMT and ETCBC parse the second occurrence of Yiqra in Gen 49:1 as an IMPERFECT!

 

Use attached Accordance file and you can see the false results.

ETCBC vs. HMT-W4 verbs.accord

[Accordance Enthusiast]

Look at the second example,

 

The problem is that the wrong two words are cross-highlighted while they are not the same word.

 

[Jordan S]

Cross-highlighting (which I agree is imperfect and could use some improvement) uses an algorithm to identify matching words across the two panes. This feature is unrelated to your search and the search results. The two features are independent.

 

The imperfect that is highlighted in the left pane and the InfAbs in the right pane (both from עזב) are the reason this verse is included in your results. The one on the left is the VERB imperfect and עָזֹב on the right is the "not-imperfect" from your construct search. If you scroll through your results, you'll find in the right pane (MT-ETCBC) a "not-imperfect" form for each of the highlighted verbs. I understand that you are desiring a word-to-word comparison of tags, but Accordance is not searching for that (at least that's how I understand it). That's all I was trying to point out. 

[Accordance Enthusiast]

Hello Jordan,

 

Thank you for your response.

 

Please re-evaluate my request. I really think this is a bug that needs to be corrected.

 

This search is indeed a word-to-word comparison of the tags. I added all the search criteria in ONE column in the construct window. Thus it should NOT find two different words, but the SAME word with different tagging.

 

NB: if this search were to find every occurrence of two cognate verbs in the same verse it would come up with many other hundreds of hits.

 

You see, in every search hit I found with this search, the cross highlighting is wrong. This is more than mere co-incidence. I am doing a cross-textual search, and how does Accordance know which words in various texts are equivalent? It seems to be using the same algorithm used for cross-highlighting. Thus every time there is a mistake in the cross highlighting there will be a mistake in cross textual searches.

[Accordance Enthusiast]

Here is a similar search which works well and has ALMOAST NO FALSE results:

 

Edited April 1, 2022 by Anonymous

[Accordance Enthusiast]

This is INDEED a word-to-word comparison search,

and in the above results in the second example the cross-highlighting and cross textual searching is really correct!

 

Regards

[Jordan S]

7 minutes ago, Anonymous said:

Here is a similar search which works well and has NO FALSE results:

 

 

See Deut 6:11. 

 

[Accordance Enthusiast]

OK I see. 

 

But again this is linked to the cross highlighting issue.

 

Most of the other results worked because the cross-highlighting works.

 

Whenever the result is wrong the cross highlighting is wrong too.

 

Do you agree?

[Jordan S]

I agree that the cross highlighting is imperfect. I would have to verify (tomorrow) that the issues are connected. I don't think that they are. I think it's coincidental due to the fact that these discrepancies require multiple forms of the same word in the same verse. This is known to cause complications with cross highlighting. 

 

You have raised interesting search questions and also morphological questions regarding Westminster and ETCBC. I would like to dig into these some more.

[Accordance Enthusiast]

Hell Jordan,

 

Thank you for your response and for looking into this further tomorrow or next week.

 

Please note that there are approx. 1800 verses where the same LEXICAL form appears twice. 

 

Very few of these come up in the searches described above, showing that the above searches are indeed word-to-word searches.

 

I will be very thankful if these mismatched hits could be excluded in future.

 

It is very interesting to see the various places where the HMT and ETCBC differ in tagging.

 

regards 

 

 

[Accordance Enthusiast]

From all I've seen the two issues are related.

 

If they are not, somehow the same mistakes occur with both the search results and the cross highlighting.

 

Anyway, 

 

Thank you for your help and for trying to solve the problem in future

[Accordance Enthusiast]

Hi Jordan,

 

I'm eagerly waiting for you to respond and report this bug so it can be fixed.

 

If you are not going to report it then tell me so I can report it via e-mail to Accordance to be corrected...

[Jordan S]

Sorry I missed this post. I reported the issue on April 1. 

[Accordance Enthusiast]

OK, thank you very much!