First, notes from the spreadsheet:
DIMAP sense 8, ODE sense 2a: This sense of through typically occurs with a limited number of verbs; when the through is uncomplemented, the verb is parsed as phrasal. Examples are break through, come through, get through, muddle through, pull through, win through. Other verbs typical of this sense involve aggression, such as fight, battle, and any other verb that might fill the first blank in _______ your way through (_________). (This reminds me of one of my favorite quotes: “The Americans are determined to hack their way through the language, as their ancestors through the forests, regardless of the valuable growths that may be sacrificed in blazing the trail.” (JYT Greig, 1929)
New sense added (spreadsheet record 7): a majority of instances of ODE Dimap sense 5 are the same as “by way of” sense 1, and via sense 1 and 2. ODE’s def is pretty clunky for this, so I have separated here the simple cases where through is a substitute or equivalent for by way of, in the sense dealing with physical space only. by way of sense 3 (=by means of) and via sense 3 are the same as through sense 12 or 13.
New sense added (spreadsheet record 12): Quirk 9.29 notes the “pervasive” meaning of through which has some correspondence to DIMAP 10 ODE 3, but I think ODE overlooks a distinction that is useful here: the use of through especially when preceded by all (as Quirk notes), or when followed by the whole, or the entire, often means throughout: All through the night she wept/ I was bored through the whole lecture.
New sense added (spreadsheet record 16): This is an easily identifiable sense that follows verbs of attachment. See, e.g., hook.v in Framenet for Frame Attaching. Used to denote two things that can interconnect (fingers, hooks, arms, loops, etc.). I put it here because it bears the same relation to “through” instrumental that my added sense of “by” (i.e., “by the corners) bears to “by” instrumental. The complement and the POA are joined by virtue of the through PP.
Other instances of through that may deserve idiomatic treatment: go through with sth (in all inflections of go) might be worth isolating since it might parse incorrectly and always means the same thing. Go through is idiomatically polysemous (see ODE entry on it) but nearly all senses of through that can be complemented fall into one of the defined senses of through. All other phrasal verbs using through as a particle fall within a defined sense of through prep.
Other Notes:
Much more so than with by, I would stress here that the ODE main sense numbers are probably more informative than the minor sense numbers; in other words, the digit(s) within pars in column D are going to tell you more than the DIMAP sense number about most of these instances. There is a lot of ambiguity in the sense that many instances could as easily be assigned to one subsense as another; sometimes context wasn’t sufficient to tell precisely which sense, as defined in ODE, a particular sentence fit. For example: If you throw a rock through a window, ODE would distinguish the through according to whether the window was open at the time and thus regarded as an “opening”(DIMAP sense 1), or whether closed, and thus an obstacle, so that by throwing you broke the window (DIMAP sense 2), but a one-sentence context does not always tell you this. By the same token, senses that to a reasonably informed speaker might be taken as semantically identical (e.g., The rat ran out through a hole in the wall, the pipe exits through a hole in the wall) are distinguished in ODE by the fact that one instance describes motion (and so is sense 1) while the other describes location (and so is sense 1d, DIMAP 5). I would think that for our purposes, distinguishing these two is pointless.
I would also stress that, simply because a particular lex unit illustrates a particular ODE sense in FrameNet (take, for example, creak v., associated with ODE sense 1e), this is not a sufficient basis for assuming that the verb has some sort of special relationship with a particular sense. It is very easy to imagine sentences with creak illustrating any number of other senses.
FrameNet distinguishes no examples of DIMAP sense 11 (ODE 4: staying here through next Thursday, and only one of sense 7 (ODE 2: midway through the second half, etc.). These are both very common senses. The absence of 11 is probably down to it being N American (so labeled in ODE) and possibly not in the BNC data. There is no good explanation for the lack of labeling of sense 7.
Quirk has some discussion of prepositions that can take here and there as complements at paragraph 7.70 and includes through among these. Such uses would correspond to any of ODE sense 1 and subsenses 1a through 1e, DIMAP 1 to 6. Conceivably ODE sense 3 could also take here or there as a complement, but this would be pretty rare. (I would note that Quirk should have included by among the prepositions listed at 7.70 but did not, and I only discovered this paragraph when checking the index on through.)
This method of looking at instances did not lend itself to identifying mislabelings in Framenet, and so I have not listed any on a second workbook. However, I would note that all verbs under Self_motion identified with sense 8(2a), 9(2b), or 10(3) are not really motion, but rather figurative progress of some kind; I don’t think they should really be classified here, since the Frame seems to be concerned only with motion in space and makes no mention of figurative use (though most of the verbs it uses easily lend themselves to figurative use). This suggests that there is probably scope for a new frame (or some other currently existing frame?) that encompasses these sentences, or alternatively, Self_motion should be definitionally expanded to encompass progress in various pursuits that is likened to progress in space through the use of verbs of motion.