I've now looked into this. It is an interesting test case.First, the data:
From our valency lexicon-in-progress, we have this brief entry:
פחד (qal) ‘be afraid, fear s.t./s.o.’ 1/2 [S V NP/PP] 22x†
As in English, the complement of ‘fear’ verbs are frequently PP, and therefore it is difficult to clearly distinguish at times between adjunct PP and complement PP. For instance, the על PP, which governs a clause in Jer 33:9 is best treated as a causal adjunct clause and not the “thing” that is feared. The cognate accusative constructions of Deut 28:67 (relative head), Ps 14:5 par. 53:6, and Job 3:25 seem to confirm that the verb is both bivalent and monovalent.
(1a) Monovalent without modification (10x): Deut 28:66; Is 12:2; 33:14; 44:8, 11; 60:5; Jer 33:9; 36:24; Psa 78:53; Prov 3:24.
(1a) With מן-PP complement of what is feared (6x): Is 19:16–17; Mic 7:17; Ps 27:1; 119:161; Job 23:15.
(1b) With adjunct אל-PP of what is feared (2x): Jer 36:16; Hos 3:5.
(2) Bivalent with NP (4x): Deut 28:67; Ps 14:5; 53:6; Job 3:25.
As for Ps 27:1, it is an inconsistency that I have now fixed (and will be reflected in the summer release along with the new texts; to be released in late August). Thank you for discovering this. To all syntax users: please report apparent errors like this so we can check them (and either fix them or explain the difference in tagging).
On the relationship between the ל and מי in Gen 32:18 -- I am not sure why you wrote that one is tagged as an adjunct. This does not reflect the tagging: both are tagged the same way, as complements.Second, the search:
To get a verb and *all* its complement, one need simply search for a verb = lexeme (e.g., פחד) + complement. (See screenshot #1, below
To distinguish between accusative (e.g., NPs with or without את) and oblique (prepositional) complements, one can specify the complement as a preposition (see screenshot #2, below)
To get just the accusative complements, *eventually* it will require one to simply specify the complements as a noun. Currently, though, there is still some squish in the searching and it returns hits that are not accurate (it overlooks a preposition and takes the prepositional complement as the verbal complement). This type of thing is still being addressed.
As a fix in the meantime, one could do both the searches illustrated below and then either manually subtract the oblique hits from the whole list (for smaller searches) or use the HITS command (I think! -- a true power user would have to jump in here to correct me or illustrate it, since I've forgotten how it works -- where you take two searches and coordinate the hits from each).
Finally, a note on the concept of an "indirect object". This is not a clear syntactic category operative in Hebrew grammar (unlike some Indo-European languages, which use a specific morphological case for the indirect object), it is not included in the syntax database. Hebrew grammar only distinguishes between complements (a category that includes "accusative" and non-"accusative (= oblique)" complements) and adjuncts. Since the verb + complement relationship can be manifested in a variety of ways (depending on the verb), it is desirable to search for both accusative and oblique (= prepositional) complements. To arrive at those hits that are analogous to English indirect objects (which, to be clear, are quite like Hebrew in that they do not represent an defined morpho-syntactic category but depend on specific verbal lexical semantics), one must then identify the sub-class of verbs (e.g., נתן, שׂים) that arguably take so-called indirect objects and after that sort the oblique complement results. Screenshot #1
16 downloadsScreenshot #2