find a Hebrew lex without a suffix

[Kristin]

I was searching for a Hebrew word with 1061 hits, and I searched for all instances with a suffix, and there 253 hits. I then wanted to locate the instances without a suffix, so I added NOT, and that produced 106 hits. That is obviously wrong, since the hits should have been 808 hits.

 

Could someone clarify how to locate the lex of a word which does NOT have a suffix?

 

Thanks,

Kristin

[Donald Cobb]

7 hours ago, Kristin said:

I was searching for a Hebrew word with 1061 hits, and I searched for all instances with a suffix, and there 253 hits. I then wanted to locate the instances without a suffix, so I added NOT, and that produced 106 hits. That is obviously wrong, since the hits should have been 808 hits.

 

Could someone clarify how to locate the lex of a word which does NOT have a suffix?

 

Thanks,

Kristin

 

Hi Kristen,

 

I tried this, just as a quick example: 

‏=מֶלֶךְ־1 ‎<NOT>‏ ‎[SUFFIX]‏

 

Which gives me these results: 

 

What word are you working on? That might help to see what's going on.

 

[Kristin]

Hi @Donald Cobb,

Thank you for your response. The word in question is הִנֵּה. I think I had been doing it as you described, but the results are completely off by hundreds of hits. I can attach a screenshot.

 

Kristin

 

[Douglas Fyfe]

I think you told Accordance "find me a verse with hinneh but not with a suffix in it."

So it's finding you every verse that does have hinneh but doesn't have a suffix.

 

You need to specify that you don't want the suffix attached to hinneh, like this:

 

 

You can also do it in a construct search to ensure it's searching in the correct direction

 

 

Hope this helps clarify.

[Donald Cobb]

Hi Kristin and Douglas,

 

This is an interesting problem. I tried a few things:

 

1) First, a search in BHS on הִנֵּה. Opening the analysis with declined forms gave me this:

 

Notice that הִנֵּה is listed three times without a suffix (477 + 329 + 1), totaling 806 hits. Triple-clicking on either of the two forms (but not the one occurrence with the segol) will give you the 806 occurrences (not sure why the two are separated in one instance – 477 + 329 – and conflated in the other...). 

 

2) Reproducing Douglas's construct search with BHS gives me 807 occurrences (interestingly, there's a difference of one hit between my construct and Douglas's, which picked up 808 hits). Comparing my two searches (806 vs. 807) shows that, for some reason, my first search didn't pick up Gen. 19.2. So there may be a bug in BHS at this point. 

 

3) I then tried Douglas's search using HMT-4, rather than BHS. The results follow Douglas's (808 x). So the problem is between BHS and HMT

 

3) Redoing the first search (הִנֵּה@"=הִנֵּה") with BHS (ETCBC) gave me the same results (806 x). However if I change to HMT-W4 it gives me 807 hits, picking up Gn 19.2. 

 

4) Redoing the first search (הִנֵּה@"=הִנֵּה") with Andersen-Forbes this time gives me 809 results. The difference is in Jer. 18.3 which has הִנֵּה plus a suffix (which is why it doesn't show up in HMT). However, in Andersen-Forbes, הִנֵּה is in brackets, thus separating it from its suffix. So 808 hits does seem to be the correct number of hits.

 

Moral of the story? HMT and Andersen-Forbes seem better equipped than BHS for detailed searches. There seem to be some tagging inconsistencies in BHS that need working out. At least all this gives a little better idea of how to get what you're trying to find, Kristin!

 

[Donald Cobb]

4 hours ago, Donald Cobb said:

 

Moral of the story? HMT and Andersen-Forbes seem better equipped than BHS for detailed searches. There seem to be some tagging inconsistencies in BHS that need working out. At least all this gives a little better idea of how to get what you're trying to find, Kristin!

 

 

Second moral: at least for Hebrew, the construct pane is the superior and easiest way of getting to the results you're looking for.

[Kristin]

@Douglas Fyfe and @Donald Cobb

Thank you both for the screenshots, and that is really interesting about the WITHIN 1 Word. I tried that and it worked. I had never used that before since I had thought the 1 was one more. (The word, within 1 word next to it), so it is good to understand that, and it for sure does seem concerning that the results are different. It appears that if you just do one search you don't really have security. For example, I just did the construct and got 903 hits. So now the construct has given 807, 808, and 903. Maybe I am doing something wrong and will post a screenshot.

 

25 minutes ago, Donald Cobb said:

Second moral: at least for Hebrew, the construct pane is the superior and easiest way of getting to the results you're looking for.

 

@Donald Cobb,

My understanding is that the construct search is not as reliable as looking at the text itself (as far as I remember, I have learned that in Webinars, so not just because of this). That said, do you think the WITHIN command that @Douglas Fyfe used is more trustworthy, or is the WITHIN command as sketchy as the construct search?

Kristin

 

[darrylmy]

Regarding the hits discrepancy between 903 and 808, @Douglas Fyfe is using the search character '=' (which is used to specify an exact form, or in this case it seems the proper lexeme) whereas @Kristin's search does not which accounts for the discrepancy in hits. The analytics from Kristin's search also includes the results "הֵנָּה־1 (הן) here = 47" and "הֵנָּה־2 (הוא) they (f.) = 48" (accounting for the difference). Just in case it might also help, I'm using v. 2.2 of the HMT module.

[Donald Cobb]

46 minutes ago, Kristin said:

 

@Donald Cobb,

My understanding is that the construct search is not as reliable as looking at the text itself (as far as I remember, I have learned that in Webinars, so not just because of this). That said, do you think the WITHIN command that @Douglas Fyfe used is more trustworthy, or is the WITHIN command as sketchy as the construct search?

Kristin

 

 

@Mark Allison might be able to weigh in here, but I was under the impression (my feeling is that Mark said this in some post somewhere) that, as soon as searches begin to get more complicated, the search construct is more reliable. 

 

But, yes, as you said, once you start looking beyond "quick and dirty" results, it's probably wise to search in a couple of different ways. Based on the different searches, the 808 figure found in HMT and confirmed with A-F seems to be the most reliable one.

[c. stirling bartholomew]

Negating some element in the CONSTRUCT pane is logically subtle. Accordance will typically refuse a negated argument it it will cause logic error in the search. You will be told more or less what you did wrong so you can correct it. Sometimes it isn't perfectly clear instruction. I tested this kind of construct pane search of HMT-T and negating suffix <within 1> of the preceding substantive worked as expected and the numbers were correct.

 

The frequent comments reflecting unreliable search results are a reflection on the state of the current releases. These search patterns were tested and tested thousands of times decades ago. The ancient language modules searches were very reliable from version 1.1 to version 9.6.8. We hammered on these relentlessly. I had to work really hard to find errors for the 3.x release. The errors didn't just happen randomly I had to dream up off the wall counter intuitive combinations of  search arguments to cause a problem. I can recall Roy Brown calling and explaining why I was getting the results I reported. There were complexities which are not always immediately understood in the logical relationships between the elements specified in the the construct pane. The construct pane is a much more lucid representation of the logic and it is capable of disambiguating what would be an impossible search without it.

 

Postscript:

 

Used the construct pane search for Kritin's word הִנֵּה. Numbers came out correct: 897 252 1149

 

TEXT Module:

Hebrew Masoretic Text

Groves-Wheeler Westminster Hebrew Morphology (Release 3.5) ©1991, 1994,1999, 2001 Westminster Theological Seminary, Philadelphia, Pennsylvania  USA.
All rights reserved.

Text used by permission of the United Bible Society, based on the Michigan-Claremont-Westminster machine-readable text.  All rights reserved.

Version 3.1
 

Edited July 25, 2023 by c. stirling bartholomew