Wonky search or wonky me?

[Peter Bekins]

I am trying to put together a search to analyze the use of prepositions with the addressee of a verb of speaking. Either I am missing something simple in the logic of the construct window or a recent update to the search algorithm is going awry.

 

When I search for all instances of דבר-2 as a finite verb (perfect, imperfect, wawConsecutive, imperative) from the main window there are 938 hits.

 

 

When I try to replicate this search in the construct window I get the same 938 hits.

 

 

When I add the criteria that the finite verb should be a predicate the hits drop to 810 (since all the books aren't yet tagged for syntax).

 

 

But when I add the criteria that the predicate should occur within a predicate phrase (which should be redundant) the hits drop all the way to 115.

 

 

Note that this step is important since I want to next filter the adjuncts of דבר based on prepositional marking. 

 

From a cursory scan of the results, I cannot deduce any rhyme or reason to why these particular clauses hit but the others did not. Any ideas?

 

Pete

 

[Robert Holmstedt]

Pete,

 

Seriously? You should know the answer to your post title question. Of course it's you.

 

Now, with that morning fun out of the way, this is not good. I'm bumping it up to the higher authorities (= programmers) to see what's going on.

[Joel Brown]

Peter, we actually found this bug at the recent SBL conference, and happened to be working on it anyway.  I can now confirm it is fixed, so you should get the same, correct, results when running the search in the next rev.

[Robert Holmstedt]

That, I say, that is welcome news!

 

(Who can name the cartoon character I just channeled?)