Why difference in specific word counts between Acc and BW?

[mortenjensen]

Hi all,

In relation to our recent investigation into difference in tagging in Accordance itself, I have a question that I really would like to know the answer for - though I am aware, it might be difficult to get: How can there be a huge difference between Accordance and BibleWork's word-counting in the NA-text? I came across this, when correcting assignments from my students, who use one or the other:

- number of "kai" in Mark: Accordance: 1100 - BibleWorks: 1091. - if enlarged to the entire NT, the difference is: 9161 (acc), 9018 (bw).

 

This should really not be possible, since both programs get the text directly from the bible societies. 

 

I can better "accept" the tagging differences - like: number of infinitives in NT: 1629 (acc) vs. 1756 (bw) - or number of finit verbs: 1873 (acc) vs. 1874 (bw), since the tagging involves grammatical choices of course.

 

But any ideas about the difference in word-counting?

 

Thanks,

 

Morten

[Λύχνις Δαν]

My first guess would be that they are not the same text exactly. You'll want to somehow get a comparison of some examples of the text for a word where the counts are not the same. And are they using the same edition of the NA text in BW and Acc. ? You can fairly easily get a table of counts of και by chapter in Mark using Analysis tab Book/Chapter Stats, Table report. If you can get something similar in BW you can then look at examples side by side and see.

 

thx

D

[Helen Brown]

Have you checked how the different programs treat the crasis such as kagw?

[Dieudonne]

I think one of the reasons is because Accordance counts variations like κἀκεῖνοι. 

[Jordan S]

From the help files: 

 

• If a word is not enclosed in quotation marks, it is treated as a lexical form.
• To find an inflected form in a tagged text, enclose it in plain quotation marks.

Try searching "και" (in quotes) and you will get something closer to what you are looking for.

 

When searching without the quotes, you can use the analysis tab and set the analysis display (cmd-T) to show the inflected forms. So you get something like what you see below. Looks like the NA-28 has an extra και. 

 

καί and, also, even, and yet, but = 9162

Κἀγώ = 1

Κἀγὼ = 3

κἀγώ = 9

κἀγὼ = 63

Καὶ = 513

Καί = 3

καὶ = 8469

καί = 34

Κἀκεῖ = 1

κἀκεῖ = 9

Κἀκεῖθεν = 2

κἀκεῖθεν = 8

κἀκεῖνα = 4

κἀκεῖνοι = 7

κἀκεῖνον = 3

κἀκεῖνος = 6

κἀκεῖνός = 1

κἀκείνους = 1

κἀμὲ = 3

κἀμοὶ = 4

κἀμοί = 1

κἂν = 17

[Λύχνις Δαν]

There is a way to find crasis directly in Acc (*@"χρασις"). There are actually too many of them in the Greek NT to account for this difference, unless BW only handles some of the ones Acc handles, which it well might do.

 

And yes I found the extra και (UBS4 vs NA28) - its in James chapter 2 I think. I would guess it's almost certainly due to UBS 4 being NA 27 versus the NA 28.

 

Thx

D

[mortenjensen]

Thanks to all of you, this explains! Great to learn new things about Accordance all the time - I like the xrasis-search and the ability to show inflected forms in the Analysis-window!

Morten

[mortenjensen]

Btw: I am amazed that some information of the use of "kai" (and other words) is tagged into Accordance. If you cmd-T the Analysis-window, you can add "tag" and Accordance lists its sorting of the use of "kai" - for example with 20 "conjunction coordination adversative". Would you not consider this regular syntax-info?

So it seems that Accordance is tagged quite a bit beyond mere words: crasis, some syntax-info - and maybe even more? Are there any other hidden gems like the  *@"χρασις"-search?

Morten

[mortenjensen]

There is a way to find crasis directly in Acc (*@"χρασις"). There are actually too many of them in the Greek NT to account for this difference, unless BW only handles some of the ones Acc handles, which it well might do.

 

And yes I found the extra και (UBS4 vs NA28) - its in James chapter 2 I think. I would guess it's almost certainly due to UBS 4 being NA 27 versus the NA 28.

 

Thx

D

Daniel: how did you do the comparison to find the extra "kai" in NA28? I tried 

-  [HITS NA28 Greek NT (Sigla)] <NOT> [HITS Greek NT Tagged]

-  [CONTENTS NA28 Greek NT (Sigla)] <NOT> [CONTENTS Greek NT Tagged]

- "και" @- [HITS Greek NT Tagged] 

Without success. Scope is set to "verse" but I tried other settings as well. It should work, not?

Morten

[mortenjensen]

Hi again,

I found another way to find the missing "kai" using Ken Simpsons guide here: http://www.accordancebible.com/forums/topic/15882-why-this-difference-between-na28-and-na28-sigla/

I produces an entire list of differences between NA27/Greek NT Tagged and N28.

Still, I would like to know, if the kai can be found through the search line. 

Thanks,

Morten

[Λύχνις Δαν]

Hey Morten,

 

  The obvious thing to try fails, that being to search for και in two tabs, one for each version. Then open a third tab and do a [CONTENTS TAB1] <NOT> [CONTENTS TAB2]. This because the verse you are looking for contains a second και attested in both versions.

 

  But there is a way to narrow down where to look for the needle. If you search for και in each text in separate tabs, you can then open Book/Chapter stats (using the Analysis icon) as a Table. Do this for each text. Then scroll down looking for the book where the count is off by one. Turns out it's James. Then tweak your query to add [RANGE James] format the stats tab to include chapter detail. Click on the little gear wheel icon on the Table pane. This will show you which chapter. It will show you James 2 is the one. Then open a parallel pane on one of the texts with the range more restricted, to James 2, and click the compare checkbox. Then scroll through looking.

 

Not the most elegant and it would be a real pain if you have to find a lot of differences like this.

 

But in answer to your real question, I know of no way to do this with a simple search string even in multiple tabs. I'll have a poke about but I don't know of anything off the top of my head. I would be interested to hear other suggestions.

 

I just had a look at Ken's. I could have used List Text Differences, though didn't think of it, but you would still have to wander through the list which for the whole text looking for an extra και would be a pain. Though I could have used it in a late stage of my process to simplify my life but there is a lot of noise created by other differences. You really want to be able to do a query on the difference output. (That's an ability I would like in Accordance. I usually export such data and post process it in a script.) Actually Book/Chapter stats having a Verse level would be great here too.

 

Thx

D

Edited May 4, 2015 by Daniel Semler

[mortenjensen]

Thanks, Daniel,

Yes, the problem is, we have a verse with two "kai"s - and so none of the search strings work - also not [CONTENTS TAB1] <NOT> [CONTENTS TAB2], which I tried.

It puzzles me, since I thought the difference between a "content"-search and a "hit"-search was that content would search within the scope (and thus find no difference), while "hit" would compare actual hits and thus filter out the extra kai even in a verse with a second kai. Not?

Morten

[Λύχνις Δαν]

Actually CONTENTS takes the verses hit in the referenced tab and uses that as the search criteria in the new tab. The problem with a search for και on Greek NT and NA 28 is that the hit verses are the same and thus they cancel each other out leading to no result. A HIT search takes the words from the referenced tab and searches for those. In this case there is only one word, και, so again you have an empty set and that fails also. There is a good explanation of this in the doc. file:///Applications/Accordance.app/Contents/Resources/Accordance%20Help/Default.htm#topics/06_braa/hits.htm?Highlight=HITS%20command. And for the CONTENTS command : file:///Applications/Accordance.app/Contents/Resources/Accordance%20Help/Default.htm#topics/06_braa/stand-alone_commands.htm?Highlight=CONTENTS%20command.

 

What we wanted was HIT_LOCATION or some such thing but that is harder than it sounds, or something like [DIFF_HITS tab1 tab2] which of course doesn't exist, but in my imagination might do something like take the actual hit texts and diff them. But really what you want is DIFF and show me where they are different. Currently we have diff and show me all the comparisons which means you have to wade through.

 

Of course the issue here is that the word is the second most common one in the NT.

 

 

Thx

D

[Ken Simpson]

Morten, Daniel is correct. I think about it like this:

 

CONTENTS gives me a list of verses to search from the linked pane

 

HITS gives me a list of words to search from the linked pane

 

Just my shorthand.