First, notes to the spreadsheet:
Sense 11: (by chance, by contrast, by candlelight, etc.) These are syntactically and semantically disparate and I would not lump them ODE has done. Since a more or less finite group of phrases fits this category, it might be better to treat them idiomatically (see below). Alternatively, they could easily be split in two groups: those describing the backdrop or a feature of it (by candlelight/moonlight/day/starlight, etc.), and those describing, usually parenthetically, some existential condition assigned to the statement by the speaker (by luck/a twist of fate/coincidence/contrast/chance). The first behave as ordinary predicate adjuncts; the latter are like sentence adverbs.
Sense 14: I would be inclined to split this sense, as there are, syntactically, two patterns here. (1) the “by the day” pattern, which might be thrown in with 15; and (2) the “minute by minute” pattern, which always consists of “by” sandwiched between two instances of the same word (though the second may be modified), and also includes such constructions as “inch by inch,” “letter by letter,” “bit by bit,” etc., which ODE doesn’t seem to account for.
Sense 19: This sense overlaps a bit with various phrasal verbs with “by” (stop by, drive by, come by, pop by, pass by), discussed below.
Sense 22: The presence of “swear” is a marker for many instances of this; those that might be used without swear are a finite group and might better be treated idiomatically (see below).
Other instances of by that may deserve idiomatic treatment (by which I mean testing for context to see if conditions for this meaning are met. These are in addition to those noted under “phrases” in ODE)
These can be divided in various ways, all of which are relevant to KMS and parsing decisions. Throughout this discussion, I wonder: if a preposition is spotted without an accompanying phrase, is it interpreted as an adverb?
(1) separability: The phrasals (all transitive) in which “by” can be separated from the verb seem to be in a very small group. Thus, I can think of only put (sth) by, set (sth) by, run (sth) by, lay (sth) by. Of these, run by is unique in having to be separated for this meaning: run that idea by Sharon and see what she thinks. With the others here, I would think the parser needs to be able to distinguish among sentences such as: We ate most of the tomatoes but put the rest by./We put by nearly 20 pounds of tomatoes./Who put the tomatoes by the back door? The first two are examples of the phrasal verb put by, the third is merely put followed by a sense 18 “by” prep phrase.
(2) transitivity: Intransitive phrasal verbs with by are, I would think, easy to treat idiomatically because “by” will always be found in isolation at the end of a phrase, clause, or sentence. Thus get by, stand by. The phrasals that can be transitive or intransitive are perhaps the most troublesome cases, since the transitive form could be interpreted in two ways. Take, for example, stop by. Stop by on your way home from work is arguably the phrasal verb (or is it merely stop + adverb by?). But what is Stop by the 7-11 on your way home from work. ? Phrasal verb + direct object, or the verb stop plus a sense 19 prep phrase? Other verbs in this category are: drive by, pop by, pass by, drop by, go by, and come by. The only phrasals I can think of that must be transitive are live by and go by in its other sense: Don’t go by what he says, read the instructions. But this could also occur with terminal by: She left the meeting early, if that’s anything to go by.
(3) distransitivity: drop by and drive by could arguably be treated as either (a) ditransitive phrasals: drop that by the Post Office on your way back/drive him by the house so he can see it. Alternatively, these are just (b) verbs with sense 19 prep phrases, but in contexts where the object would be understood, the pattern would presumably look different to the parser though the meaning would be the same: drop that by on your way back/drive him by so he can see it.
B. Idioms in which “by” appears with a fixed form of words and may lend itself to treatment by identifying collocates and context.
(1) hard done by invariable; though not particularly common in AmE.
(2) set store by sth some variation: set (intensive) store by sth; sth by which sb set (intensive) store
(3) swear by and swore by (as noted above) some variation possible: by which sb swore. Sworn by, on the other hand, seems to turn up at least as often in cases where affidavits or statements are involved, making it a sense 2.
(4) instances of sense 11 above. The “candlelight” group is nearly always in the predicate; the “chance” group is parenthetical or introductory, and the sentence can always function without it.
C. Other patterns possibly worth taking note of:
(2)The pattern By + present participle at the beginning of a sentence, or By +adverb+present participle is invariably sense 5: it presents the means by which the task in the main clause is or can be accomplished.