WTM construct phrase search incorporating noun, adjective, or participle

[doc_dave]

I'm looking for "long" construct chains. One can readily do something like this:

'*@n???c+Sx* *@n???c+Sx* *@n???c+Sx*

to find three construct nouns in a row.

But adjectives and participles can also be in the construct state. Because the first element after the "@" cannot be a "?", how can one set up a search which looks for construct forms, but is agnostic as to whether that construct is a noun, adjective, or participle? I've been puzzling over this one, and searching the forum, and still none the wiser. (I do realize that some sequence will be unlikely or impossible!)

Any suggestions?

[Donald Cobb]

I'm looking for "long" construct chains. One can readily do something like this:

'*@n???c+Sx* *@n???c+Sx* *@n???c+Sx*

to find three construct nouns in a row.

But adjectives and participles can also be in the construct state. Because the first element after the "@" cannot be a "?", how can one set up a search which looks for construct forms, but is agnostic as to whether that construct is a noun, adjective, or participle? I've been puzzling over this one, and searching the forum, and still none the wiser. (I do realize that some sequence will be unlikely or impossible!)

Any suggestions?

I just gave your construction a try, but placing brackets for the first element, i.e., '*@[na]???c+Sx* *@[na]???c+Sx* *@[na]???c+Sx*

This includes adjectives in your query. I also tried enlarging it to verbs [nav] but that didn't turn up anything, perhaps because verbs have a more complex parsing.

That will at least get you a little further. Someone else might be able to chime in.

Regards,

Don Cobb
Aix-en-Provence, France

[Jim Wert]

This worked for me using WTM:
'*@??*c+S* *@??*c+S* *@??*c+S*
After each @ I had to press Esc (escape) to get the command line to accept the first ?.
This returned 92 verses, 89 of which have 3 construct lemmas in a row. Psa 127:2 has 5 in a row, Isa 23:9 has two sets of 3, separated by two non-construct lemmas, and Dan 12:7 has 4 in a row.
I used ?? to avoid the particle conjunctions I got with only one ?.

In Aramaic I didn't find any consecutive constructs.

--Jim

[Jim Wert]

I just gave your construction a try, but placing brackets for the first element, i.e., '*@[na]???c+Sx* *@[na]???c+Sx* *@[na]???c+Sx*

This includes adjectives in your query. I also tried enlarging it to verbs [nav] but that didn't turn up anything, perhaps because verbs have a more complex parsing.

That will at least get you a little further. Someone else might be able to chime in.

Regards,

Don Cobb
Aix-en-Provence, France

Don: I modified you query to
'*@[nav]*c+S* *@[nav]*c+S* *@[nav]*c+S*
and got 994 verses back.
I don't understand why I got so many more hits than when I used
'*@??*c+S* *@??*c+S* *@??*c+S*

It seems to me that ??*c should be a subset of [nav]*c rather than the other way around.
Looking at some of the *@??*c+S* query results, they seem to all begin with a verb construct and end with a noun construct. My take is that BW is not handling this query correctly.
Running these 2 queries using GSE gave me exactly the same results, so GSE is handling the question marks the same way as the command line searches.
I'll bring this to the attention of BW support.

--Jim

[Donald Cobb]

This worked for me using WTM:
'*@??*c+S* *@??*c+S* *@??*c+S*
After each @ I had to press Esc (escape) to get the command line to accept the first ?.
This returned 92 verses, 89 of which have 3 construct lemmas in a row. Psa 127:2 has 5 in a row, Isa 23:9 has two sets of 3, separated by two non-construct lemmas, and Dan 12:7 has 4 in a row.
I used ?? to avoid the particle conjunctions I got with only one ?.

In Aramaic I didn't find any consecutive constructs.

--Jim

Jim,

This is a little mystifying for me. I tried your query and came out with 1198 hits, with different elements (i.e., nouns or verbs at the beginning, etc.) at the beginning and at the end.

I also had the idea that using the "or" function with the GSE might give the best results. But I don't recall how to make the elements follow without any intervening words. I got this far. Someone else will no doubt be able to show me what needs to be added.



One other thing doc_dave : with the construction I tried yesterday ('*@[na]???c+Sx* etc.) I came up with the same number of hits as with '*@n???c+Sx* I assume, therefore, that adjectifs aren't placed in construct states. I also tried '*@a???c+Sx* etc., and got no hits.

Don

[Donald Cobb]

I think I've found what you need.

A couple things. In the previous post, I mentioned that *@a???c+Sx* in triple succession got no hits. This is because there is one ? too many. It works with *@a??c+Sx*.

BUT, what you need to do is have BW search for instances in which either verbs, adjectives or participles are in a construct state to each other (x 3). To do this, you have to use the GSE and the iel function. This is what I came up with:



I put nouns, adjectives, and participles in construct states (with or without suffixes) in the iel list (I made a list, then saved it, to keep from writing the same construction three times). This means that the GSE searches on each one of those as an OR search. By doing setting up that same box three times, you come up with all the verses with three connected nouns, adjectives, or ptcples in construct state to each other. I come up with 1022 hits.

I hope that helps! Others may want to verify this or may come up with something more precise.

Don

[MBushell]

Don: I modified you query to
'*@[nav]*c+S* *@[nav]*c+S* *@[nav]*c+S*
and got 994 verses back.
I don't understand why I got so many more hits than when I used
'*@??*c+S* *@??*c+S* *@??*c+S*

It seems to me that ??*c should be a subset of [nav]*c rather than the other way around.
Looking at some of the *@??*c+S* query results, they seem to all begin with a verb construct and end with a noun construct. My take is that BW is not handling this query correctly.
Running these 2 queries using GSE gave me exactly the same results, so GSE is handling the question marks the same way as the command line searches.
I'll bring this to the attention of BW support.

--Jim

Jim,

The ?? construction requires at least two characters before the 'c' and is thus more restrictive and will produce fewer hits.

Mike

[doc_dave]

I greatly appreciate all the thought and time that have gone into the responses here! I won't be able to digest and evaluate until later in the week, but just wanted to say a quick "thanks" in case you wondered were the "OP" had gone...

[Jim Wert]

This is a followup to several of my entries, and several responses from others.

I discovered that my strange counts came from a keying error I made -- I missed an *, so the first construct in the string had to be a verb infinitive construct, since that is the only word that will match <*@??c+*>.

I took a simple-minded approach to finding strings of constructs, figuring that I would include everything that the WTM coding scheme calls construct.
My search, <'*@??*c+* *@??*c+* *@??*c+*> does, I think get all constructs.
construct is always coded "c", but the code "c" does not always mean construct.
construct is always the last code before the "+".
<*@?*c+*> gets constructs, as well as "Particle conjunction".
The reason Don got more hits using my search (he didn't make the typo I did) is this more general search also finds Numeral constructs.

If you don't want to include Numeral constructs, then you can use <*@[nav]*c+*].
And remember that "escape" allows you to enter the "?".

--Jim