Hebrew letter sequence search

[SMacG]

How do I search for a Hebrew letter sequence in the MT?

For example, how do I find all occurrences of the letters “לו אל” in the sequence they are written (ל followed by ו followed by space followed by א followed by ל), regardless of how they are pointed, and regardless of whether they are words in themselves or part of larger words?

I am running Accordance 13.3.3.

Thank you in advance for help.

[Fabian]

I would say choose "Letters" 

I got 9 hits in 9 verses. 

 

 

[SMacG]

Thank you.    In my version of Accordance (13.3.3), searching for "לו אל" under 'Letters" revealed only five verses (Ex 18:27; 2 Sam 9:2, 13:12; Psa 37:7, 2 Chr 16:4), all which were the word לו followed by the word אל.   A construct search with both terms entered as "Letters" revealed the same examples, but there were no examples where לו is the last two letters of a longer word and אל is the first two letters of a following longer word.    How do I make Accordance reveal everything, particularly letter sequences which aren't whole words, and the nine examples you have suggested?   

[Fabian]

Strange you get only 5 hits. With HMT-W4

 

I got this one

Gen. 33,20; Ex. 18,27; 28,7; 2 Sam. 9,2; 13,12.16; Psa. 37,7; Prov. 15,12; 2 Chr. 16,4

 

With the use of wildcard

*לו אל*

 

I get 66 hits in 66 verses.

Gen. 33,20; 35,7.10-11; 45,9; Ex. 4,18; 15,15; 18,27; 23,21; 28,7; Num. 14,40; 20,27; Deut. 4,7; 31,28; Josh. 10,4; 19,51; Judg. 4,5; 16,5.18; 20,23; 1 Sam. 2,25; 10,9; 14,12.33; 17,30; 2 Sam. 9,2; 13,12.16; 1 Kings 8,2.30.35.44.48; 10,26; 2 Kings 2,4.6; 13,15; 18,33; Is. 36,18; Jer. 38,19; 39,1; 49,28.31; 50,14; 51,3; 52,15; Ezek. 32,21; 40,49; Zech. 11,2; Psa. 37,7; 150,1; Prov. 15,12; 24,17; 26,25; Dan. 12,7; 1 Chr. 12,9; 2 Chr. 1,14; 5,3; 6,21.26.32.34; 16,4; 20,30; 32,13.29

Edited May 28, 2022 by Fabian

[SMacG]

By switching to HMT-W4, I now get the same result!   Thank you very much.   Previously I had been searching BHS.   But why the (very substantial) difference?   And is there another text that would yield even more results?   If so, which is optimal, for searching the MT?

[c. stirling bartholomew]

S

6 hours ago, SMacG said:

By switching to HMT-W4, I now get the same result!   Thank you very much.   Previously I had been searching BHS.   But why the (very substantial) difference?   And is there another text that would yield even more results?   If so, which is optimal, for searching the MT?

 

What is optimal is what yields the results you are seeking. Historically  literal string searches in Hebrew texts were difficult. It would appear that problems got resolved at some point with the recent HMT-xx editions. When I tried the same search in HMT-T 3.5 2001, I got 37 hits. I did some work on it. Apparently one reason was the leading asterisk in the string *לו אל* would not match לוֹ אַל־אָחִי because it was on a leading word boundary. I didn't do an exhaustive analysis of the problem since it appears to be working now. People publishing academic papers using research done on digital platforms need to be well aquatinted with their digital tools.  I learned something today.

 

Postscript: I never attempted to use literal string search in Hebrew texts. Decades ago I assumed they were unreliable. It's been decades since I worked with Hebrew.   

Edited May 30, 2022 by c. stirling bartholomew

[c. stirling bartholomew]

correction: The word should be acquainted . 

 

I became fascinated with this problem and after further analysis discovered that I could solve the problem with construct windows which is my normal first choice in doing searches of tagged texts. 

 

I'm not particularly skilled at posting screen shots so I will just explain in words what I did. I set up HMT-T linked to construct number one which was simply two inflected words *לו אל* -within 1 constraint. This gave me the same 37 hits as before. Then I set up a second HMT-T linked to construct number two which was LEX ל= followed by mas 3p sg suffix followed by inflected *אל within 2 constraint, not contents of HMT-T 1 constraint. This produced 22 hits. 

 

Don't think the problem has anything to do with the tagged text. The search engine wasn't really intended for string searches. The architecture* of the software was built around tagged texts. That was what it was from the beginning when Roy Brown and Martin Abegg published "A Time for Tagging."   

 

*architecture is probably the wrong word. 

 

POSTScript: For string searches TLG is the obvious choice. I assume there is a Hebrew equivalent. 

Edited May 31, 2022 by c. stirling bartholomew

[SMacG]

Thank you, very interesting.   The question however remains: is there a reliable way to perform a letter string search of the MT?   I imagine it can't be difficult for the software to perform a literal search, because it is easy to export letters without accompanying marks, so scanning the text for a letter string, without marks but including spaces, ought to be straightforward, yet the results from searching BHS differ significantly from HMT-W4, which is worrying.  

[Donald Cobb]

Perhaps this will get you what you are looking for:

 

(אל*) <FOLLOWED BY> <WITHIN 1 Words> (*לו <OR> לו)

 

I just tried this in BHS, doing a "letters" search, and it gives me 112 hits. A couple of examples:

 

וַיֹּאמֶר לוֹ אֱלֹהִים

 

‏וּשְׁמַע בְּקֹלוֹ אַל־תַּמֵּר

 

I also came across a bug, though. I tried doing this with the construct search but when I tried to put an * in a "letters" box, Accordance immediately crashed. Can anyone reproduce this?

 

 

[SMacG]

Thank you.

 

I tried a construct search for לו* followed by (within 1-1 words of) *אל.   This returned 22 verses in HMT-W4 and 19 in BHS.

 

In contrast, Fabian’s suggestion of performing a Letters search with the wildcard

*לו אל*

returned more verses (66 in HMT-W4, 56 in BHS).

  

Why is this, and why does BHS produce fewer results?

 

For example, the first two verses from Fabian’s search suggestion in HMT-W4 are missing from BHS, but should not be:

לוֹ אֵל (Gen 33:20)
נִגְלוּ אֵלָיו (Gen 35:7)
  
I am confused!    Can anyone shed light?
 

[Donald Cobb]

I too am getting inconsistent results with my construction, depending on which resource I use. It would be good for someone on Accordance staff to weigh in here (also for the problem with Accordance crashing with the *). 

[c. stirling bartholomew]

The results of this search are compromised by Ketiv/Qere pairs; Unless that is what you were looking for in the first place?  

 

Quote

 

Ezek. 40:34‏ וְאֵלַמּוֹ [וְ][אֵלַמָּיו] לֶחָצֵר הַחִיצוֹנָה וְתִמֹרִים אֶל־אֵלוֹ [אֵלָיו] מִפּוֹ וּמִפּוֹ וּשְׁמֹנֶה מַעֲלוֹת מַעֲלוֹ [מַעֲלָיו]

 

‎Ezek. 40:36‏ תָּאוֹ [תָּאָיו] אֵלוֹ [אֵלָיו] וְאֵלַמּוֹ [וְ][אֵלַמָּיו] וְחַלּוֹנוֹת לוֹ סָבִיב סָבִיב אֹרֶךְ חֲמִשִּׁים אַמָּה וְרֹחַב חָמֵשׁ וְעֶשְׂרִים אַמָּה

 

 

I recall asking but don't remember the answer: How do we get the search engine to ignore the Ketiv/Qere pairs? It is a royal pain to work around. 

Edited June 2, 2022 by c. stirling bartholomew

[Donald Cobb]

20 hours ago, SMacG said:

Thank you.

 

I tried a construct search for לו* followed by (within 1-1 words of) *אל.   This returned 22 verses in HMT-W4 and 19 in BHS.

 

In contrast, Fabian’s suggestion of performing a Letters search with the wildcard

*לו אל*

returned more verses (66 in HMT-W4, 56 in BHS).

  

Why is this, and why does BHS produce fewer results?

 

For example, the first two verses from Fabian’s search suggestion in HMT-W4 are missing from BHS, but should not be:

לוֹ אֵל (Gen 33:20)
נִגְלוּ אֵלָיו (Gen 35:7)
  
I am confused!    Can anyone shed light?
 

 

Ok, I just tried my search again, with a slight refinement over what I posted yesterday:

 

(אל, אל*) <WITHIN 1 Words> <FOLLOWED BY> (*לו, לו)

 

HMT-4 and Andersen-Forbes both give me 132 hits. The two seem to be perfectly consistent. When I do the same search in BHS, it gives me 112 hits and, in the ETCBC text, 108 hits. I have to look through the latter more closely, BUT comparing HMT-4/Andersen-Forbes with BHS, I see that the discrepancy comes, in each case, from the fact that a letter referring to the critical apparatus is placed between the two groups of letters, e.g.:

 

וַיַּעֲלוּ ^a אֶל־הֹר

 

Because it's a "letters" search, the siglum counts as a letter and skews the results. I haven't yet looked at the ETCBC text, but Stirling's comment on the qere/kethib readings could well account for the other differences.

 

The moral of all this: for a very specific letters search, HMT-4 or the Andersen-Forbes text seem to be the best suited. So, different texts for different purposes.

 

I also said this earlier: "I also came across a bug, though. I tried doing this with the construct search but when I tried to put an * in a "letters" box, Accordance immediately crashed. Can anyone reproduce this?"

 

I'd still like to know if this is a general bug, or if it's just me.

 

Edit: I just picked out the differences between BHS (112 hits) and ETCBC (108 hits). The difference is with these texts, which ETCBC does not pick up: Judg 16,18; 1 Kgs 8,2; Prov 15,12; 1 Chr 12,9. I can't see why one would pick these up and not the other.

 

 

Edited June 2, 2022 by Donald Cobb
Added comment

[Mark Allison]

@Donald Cobb I haven't encountered an issue with the asterisk causing a crash in a letters search. Is it happening in only one particular Hebrew Text, or is it happening in all of them?

Also, we're away of the issue with a miscount in letters searches in Hebrew texts that include sigla. We're working to fix it asap.

[Donald Cobb]

On 6/7/2022 at 1:06 PM, Mark Allison said:

@Donald Cobb I haven't encountered an issue with the asterisk causing a crash in a letters search. Is it happening in only one particular Hebrew Text, or is it happening in all of them?
 

 

Hi Mark,

 Thanks for looking into this. I just tried it again with a couple different editions of the Hebrew text. I couldn't reproduce it. Maybe restarting Accordance did the trick.