Porson’s bridge is an intriguing thing.
The rationale for it is not at all clear. It may be (my own suspicion) associated with the fact that the sequence short-long-long (or light-heavy-heavy, in the preferred terminology for syllables as distinct from vowels) has clausular function in many early Greek verse-forms, and the “bridge” in iambics propels the line beyond giving any such impression. It will have been instinctive on the part of the composers, who were probably not aware of its being a “law.” (There’s no inhibition against it earlier in the line, at caesura, but then that’s so much earlier in the line.)
That it’s not operative in the comic trimeter, but only in the tragic and other early iambic verse, is easier to explain. It’s of a piece with the much laxer form that stichic meters (iambic trimeter most notably) are free to assume in comedy. Comedy avails itself of all sorts of metrical licences that are unacceptable in tragedy, and non-observance of Porson’s Law is just one of them.
Precise definition of the phenomenon is a little tricky. The Goodwin quote you got from Wikipedia, while not exactly wrong, is not very satisfactory, and certainly has no explanatory value. The only way to make sense of the apparent exception for long(heavy) monosyllables is in terms of apposition, specifically pre-position, which is what Devine and Stephens do. If a monosyllabic “word” is not so much an independent word as part of a word-unit, i.e. attached to what follows it, it is best not regarded as a monosyllable at all, and hence does not break the bridge.
As to what “words” meet this criterion, in addition to D&S there’s a very brief but useful treatment in M.L. West’s Greek Metre, if I remember rightly. What “monosyllabic words” qualify as not breaking the bridge is determined by observation of actual practice, which leads to theory (not the other way around). Not all prepositives qualify, it seems, but in general it’s safe to think of a monosyllabic prepositive as being not a word in its own right but rather the beginning of a single unit. It only looks like a separate word on the page.
To clear up a misapprehension in your initial post, neither the length of the vowel nor the quality of the syllable is affected. Longs remain long, there's no reason to think that they were in any way “reduced.” I don’t recall that D&S say otherwise (haven't checked), but if they do I think they’re wrong. I’d be surprised if they do, for in other publications they have quite rightly insisted on a binary opposition between heavy and light syllables, just as between long and short vowels (there’s no such thing as an anceps syllable). The prepositive function is enough.
And to settle hoti’s business, no, the iota of oti is never lengthened, any more than the omicron. A short vowel is always short. Homer could make oti scan long-short instead of short-short by doubling the tau, but he couldn’t lengthen the iota to make it scan short-long or long-long. (Not that there’s not plenty of prosodic fudging with vowels too.) The final element of a verse (iambic trimeter included) always “counts” as long even when the vowel is short, by virtue of its verse-final position; the vowel itself is not lengthened. But all this gets us well away from Porson’s Law.
- I doubt that any one of us would discern the difference between galhn’ orw and galhn orw. But at least, thanks to the likes of Devine and Stephens, we can understand what the difference was.
Apologies for the length of this post, which may not even be helpful.