word study text comparison - search numbers do not match

[mbcvida]

I was doing a brief word study on Sheol [KEY H08619] and comparing different english translations - I noticed that the NIV11 translates the word as "realm of the dead" - however when using the text compare discover that the count for realm is listed as 100x rather than the 25x the phrase appears. It appears they have counted all four words in the phrase rather than the initial word 'realm' - could this be corrected so that the entire phrase is counted as 1, rather than counting all four words in the phrase

Edited May 28, 2023 by mbcvida
correction

[Dr. Nathan Parker]

I'll test this next week. In the meantime, if anyone else confirms the bug, let me know.

[Dr. Nathan Parker]

I'm looking at this one further. Not sure if it's a "bug report" or if if needs to be filed as a "feature request". I'll ask one of my team members to clarify.

[Wing Kei Wong]

Yes, this problem really exists in NIV with enhanced G/K numbers. 

 

There are three observations. 

 

1) Let's check the Text Compare result of the word ἀφορίζω (GK G0928) below,

 

The word in NIV is rendered as "set apart" 3 times (Acts 13:2; Rom 1:1; Gal 1:15), but the counts of "Set" and "set" are 6 in total. Moreover, the Word Pie Graph seems to recognize the phrase "set apart", but the display of Text Compare is not consistent. 

 

For reference, the count of "Mounce NT with enhanced G/K numbers" column is correct. In this text, the word is also rendered as "set apart" 2 times (Acts 13:2; Rom 1:1), which is reflected in the Text Compare as "Set" and "set".

 

2) By checking the Analysis of the search of set@[KEY G0928], another kind of wrong counting appears. See the following,

 

3) Finally, let's check the text, which may help the developer to trace the problem.

For NIV text, the mouse hovering effect of the corresponding word is shown below. The whole phrase "set apart" is highlighted in the same colour. 

 

For Mounce NT, the same phrase is displayed with a slight difference. The colour of "apart" is pale blue. 

This may reflect the underlying mechanisms behind the same phrase in these two texts are different, which leads to a different result in counting. 

 

Aside from the bug fix of word counting, I propose to show the whole phrase in Text Compare for the text with enhanced G/K numbers, as shown in the Word Pie Graph. 

 

 

 

[mbcvida]

Here is a simple example of my issue - search for keyword [KEY H08619] (sheol) in NIV84 = 66x; search for keyword [KEY H08619] in NIV2011 (enhanced) = 143x -- the hit counts are totally off in the enhanced NIV due to counting all words in the phrase rather than the keyword itself. Using NT [KEY G2663] NIV2011 5x, NIV84 2x -- again the enhanced version is counting each word in the phrase rather than the keyword itself, making the word counts unreliable

[Dr. Nathan Parker]

Thanks for the followups! I'm having Tech Support look into this one.

[Brian Phillips]

This is still not right. The Word Study for the key number (G0928) is showing two hits for 'set' and three hits for 'apart', but these two words work together as one phrase used to translate one Greek word. I.e., there is not a separate ἀφορίζω for 'set' and another for 'apart' in the Greek. But, if you look at the interlinear for the NIV11, you can see that ἀφορίζω shows up twice—once under 'set' and once under 'apart'. I don't think that's how an interlinear should work. It should show the actual Greek text, not a one-to-one correlation to each English word. 

 

Even weirder is when I click on 'set' in the pie graph. It lists two hits, but results in the same three hits as 'apart'. The difference is that there are two hits where "set apart" occurs, but then one hit where "set me apart" (Gal 1.15) occurs—with the intervening 'me'.

 

If you look at Acts 13.3, 'after they had fasted' shows up with blanks in the Greek, as is proper, until the word for 'fasted' (νηστεύω). But then in 13.4, we get errors similar to the above where ἐκπέμπω shows up three times in the Greek for the English phrase 'sent on their way'. 

 

The error with ἀφορίζω in Acts 13.2 appears in the NASB95 interlinear too.

 

 

 

 

 

 

 

 

 

 

Edited August 3, 2023 by Brian Phillips
More examples

[Wing Kei Wong]

@Brian Phillips

 

Your reported case is not a bug but seems to be a technical limitation of the texts you used. 

 

The two texts you reported (NIV with G/K numbers and NASB95) can only do word-to-word mapping. As for your testing case, the words "set" and "apart" are treated independently, and each of them is linked to the original word ἀφορίζω separately. Thus the results "apart=3" and "set=3" (2+1, combining the case S and s) are expected. (I also agree that it is not convenient to use)

 

Two months ago, I also reported similar testing, but the texts I used were "enhanced G/K numbers" (NIV and Mounce NT). Both of these texts are supposedly capable of doing word-to-phrase mapping. That is, "set apart" is treated as one phrase and mapped to the original word ἀφορίζω. However, in my test, the count result was incorrect (as also firstly reported by @mbcvida). These errors are related to the texts with "enhanced G/K numbers", a bit differing from your case. 

 

Edited August 3, 2023 by Wing Kei Wong

[Brian Phillips]

3 hours ago, Wing Kei Wong said:

The two texts you reported (NIV with G/K numbers and NASB95) can only do word-to-word mapping. As for your testing case, the words "set" and "apart" are treated independently, and each of them is linked to the original word ἀφορίζω separately. Thus the results "apart=3" and "set=3" (2+1, combining the case S and s) are expected. (I also agree that it is not convenient to use)

 

Thank you for the explanation. That seems to be to an odd approach to mapping the Greek and English, and it's not consistent either. As you can see in my example above in Acts 13.3, the phrase "after they had fasted" is mapped to one Greek word. Doing it the way you described above (sounds like someone from Accordance explained this to you?) makes it really difficult for an end user to derive counts of the Greek usage.

 

What good is an interlinear that doesn't map the real text of the Greek without inserting repetitions to accommodate the English? It's not really an "interlinear" with that approach.

 

It's hard enough to figure out which texts are tagged or not since the website is sometimes inconsistent with the way products are coded or described. But now I'm supposed to know which texts can only do word-to-word mapping (because these have no use for me). 

Edited August 3, 2023 by Brian Phillips