Syntax query (construct relations excluding certain nouns)

This search query ought to be really simple, but it doesn't work. Am I doing something wrong or is there a hitch in the system? Here it is, along with some of the results on the left:

I want to look for all cases in which a noun with the semantics "deity" is used in the construct state (e.g. יהוה in Ps 24:10). I don't want to see the term אלהים, however. So, as you can see in the search, I typed in the lexeme אלהים and tagged it "Match anything but this." As you can see in the result, I get hundreds of hits with אלהים in them.

I then tried this version:

Again, to no avail.

I would have thought that this is a simple search: cases for a deity in the construct state, except for the term אלהים.

Find more posts tagged with

Comments

Kevin Becker

After wrestling with this for about 1/2 hour I'm ready to say it's probably a bug in the search engine or syntax resource. What convinced me was when I compared (using the wonderful Passage lists in the Beta) the results of a search for just a Phrase (Construct, Deity) and then a Phrase (Construct, Deity) with elohim and elh5 excluded. The excluded search produced hits from 13 less verses. Looking at these 13 verses I could not tell any difference between then and the elohim hits that made it past the excluded search. Perhaps I missed something, but it sure doesn't look right to me.

Any ideas from other syntax searchers? Should this be upgraded to a bug report?

davidphillips

https://community.logos.com/discussion/comment/124133#Comment_124133

I'm getting different results. A search for this criteria just excluding אלה results in 328 results.

Adding אלהים as an exclusion keeps the 328 results.

However, running the same search and looking for places where אלהים MUST BE present results in 0 hits.

Further, running a phrase search for construct, deity gives me 771 results, so the exclusion appears to be working.

Can anyone confirm this? I'm running on my non-Beta PC right now.

Dave Hooton

As you can see in the result, I get hundreds of hits with אלהים in them.

I agree with David that the exclusion term IS working, Philip. I don't understand your claim when Gen 9:26 (as shown in your graphic) satisfies the condition, with a lexeme different to אלהים

I then tried this version:

This version probably produced much less results because you insisted the segment be a child of the phrase. If both queries had that constraint then the difference is 2 results, because the second query demands that the lexeme NOT appear anywhere in the phrase.

davidphillips

https://community.logos.com/discussion/comment/124216#Comment_124216

I think the problem is whether or not אלה (AF 5) is different than אלהים. To be honest, I would have grouped them together. Both link to the same entry in HALOT. However, Andersen-Forbes treats them as separate lemmas, and as a result, While Genesis 9:26 does satisy the condition according to AF, it does not according to what Philip was looking for. That's why I showed the fourth search, where looking for any instance of אלהים in the construct produces 0 results. אלהים is never in the construct in AF, rather אלה (AF 5) is used instead. At least that appears to be what's going on.

Dave Hooton

https://community.logos.com/discussion/comment/124229#Comment_124229

At least that appears to be what's going on.

Hopefully Philip will clarify that!

Philip E. Sumpter

https://community.logos.com/discussion/comment/124306#Comment_124306

Guys, that was really helpful.

I see my mistake now. My problem was, as David P. says, that I thought the gloss "god/God" was covered by just one "lexeme": אלהים. I now see that it is covered by a host of other "lexemes" (whatever that means for A-F, it's not defined in the glossary), including אלה 5.

So, what have I learnt? I should check if my understanding of the "lexeme" corresponds to that of A-F.

Now for a little rant on the complexity and inconsistency of A-F on this issue:

I think that A-F need to clarify what they mean by "lexeme," as their criteria for what actually constitutes a lexeme are not clear and inconsistent. As far as I can see, A-F create "lexemes" according to different criteria: etymological (roots), semantic, and morphological. The result is a huge amount of lexemes, and it is not clear to me how one can possibly be sure that one has got all one's bases covered when doing a search (especially in light of the last criterion, see below)!

For example, the etymological criterion (for want of a better category): I can see the justification in assigning a number of different "lexemes" (whatever those are) to the gloss "god/God." According to ThWAT (German dictionary), there are three words which are used synonymously for "g/God": אל, אלוה, and אלהים. So I guess it makes sense to have three separate lexemes.

The problem is that no one can decide whether these "words" are in fact different words or whether they are derivatives of just one word (root). Don't they all stem ultimately from אל? ThWat can't decide. If so, you wouldn't be etymologically justified in having three "lexeme" entries.

Either way, A-F complicate the matter by adding a fourth "lexeme" (root?) which I can't find anywhere: אלה 5. Theoretically I'd have thought this would represent אלוה, but they have a separate "lexeme" for both (and it isn't the Aramaic term). How am I supposed to know to check this "lexeme"?

They complicate by further dividing the "lexeme" אל into two different semantic categories: its meaning as "god" and its meaning as "God." What is a "lexeme," then, if it refers to both etymology and/or semantics? They are also inconsistent here (or make a mistake). According to their categorization, אלהים means "God" with a capital-G, but in Ps 86:8, the term clearly means "gods."

I thus had to tag 5 different lexemes in order to cover what is generally considered to be one word (or three, if you are so inclined).

Another example:

If we look at the word for "river, stream" (נָהָר), we have the aforementioned criteria (etymological: נהר for river, נהר for "to shine"; and semantic: נהר for "river" and נהר for "stream") plus another criterion: morphological. The plural of river is usually masculine (נהרים), sometimes feminine (נהרות). This ambiguity happens in Hebrew. Yet, if one wants to get hits for the plural form of the word, one has to be aware that A-F have added an extra "lexeme," נהרה, for this feminine plural form. It isn't assigned to נָהָר. The problem is that this "lexeme" does not actually exist (in the sense of "stream, river"). It is a theoretical construct designed to correlate to the feminine plural form, one you won't find in any dictionary (I can't). But this phenomenon of varying genders simply happens in Hebrew. If A-F want to be consistent here and mark every switch in gender, they would have to give an extra "lexeme" for רוח (spirit), which is usually feminine but sometimes masculine. They don't do this, however, probably because the masculine version doesn't look different (i.e. it has the same morphology; it's only the adjectives qualifying it that are conjugated differently).

The upshot is that it is very difficult (impossible, given the mixed criteria?) to know that I've tagged all the possible "lexemes" that A-F imagine to belong to a singular dictionary word. For example, if I didn't realize that they invented an extra lexeme for נהר with a fem. plural ending, I would not have received hits for Ps 24:2.

Rant over. I could turn this into a blog post ...

I'd appreciate corrections if I've misrepresented them.

davidphillips

https://community.logos.com/discussion/comment/124476#Comment_124476

The problem is that no one can decide whether these "words" are in fact different words or whether they are derivatives of just one word (root). Don't they all stem ultimately from אל? ThWat can't decide. If so, you wouldn't be etymologically justified in having three "lexeme" entries.

Philip,

I think that is getting to the root (hah!) of the problem. I really appreciate the work Logos and others have put into addressing issues of homographs (see http://www.logos.com/support/lbs/hebrewhomographs and http://www.logos.com/support/lbs/homographindicators), but I think that in some instances AF adds confusion with excessive homographs. I would have made the same mistake you did in the search and only caught it because of Kevin Becker's search. I think there is a two part "solution" to this. 1) Be aware of the problem and take note of how many homographs are listed when doing a search and check them out to see if any of them need to be included. 2) Look at Dave Hooton's "A Strategy for Syntax Search" on the logos wiki (http://wiki.logos.com/A_Strategy_for_Syntax_Search), particularly "Construct a text query that will be your control." A simple text query can help identify places where you might miss valid hits while performing a syntax search and allow you to make some tweaks.

I suppose the third solution would be to solve all the homograph problems conclusively and then have AF retag everything [8-|]

Dave Hooton

https://community.logos.com/discussion/comment/124476#Comment_124476

Guys, that was really helpful.

Philip, thanks for clarifying your objectives. I enjoyed the "blog" part as my knowledge of Hebrew is minimal. It would be good if Logos would now comment ...

MJ. Smith

https://community.logos.com/discussion/comment/124476#Comment_124476

So, what have I learnt? I should check if my understanding of the "lexeme" corresponds to that of A-F.

I've also discovered that my linguistics/Proto-IndoEuropean terminology sometimes does not match Hebrew specialist for root/stem/lexeme - in fact, sometimes even the Greek seems to have a life of its own. Unfortunately, there doesn't seem to be a magic pill to learn the idiosyncrasies of the various resources.

Philip E. Sumpter

https://community.logos.com/discussion/comment/124609#Comment_124609

Thanks for the responses again guys (this forum is awesome). I post my concerns about A-F's definition of a "lexeme" in a separate post. To reiterate, my issue isn't that my (or anyone's) knowledge of Hebrew isn't good enough to grasp A-F's tagging scheme (I think they're justified in have three lexemes for God: אל, אלוה,אלהים), my issue is the mixed and inconsistent criteria they have which makes it difficult to know for sure whether one has tagged all the relevant "lexemes" (e.g. their seemingly inconsistent introduction of semantic and morphological distinctions).