That is, how do people tell ...VC V... from ...V CV... in languages that have such a distinction?
I haven't been able to find anything regarding this.
1 Answer
It's impossible to tell from pronunciation alone on a universal basis. I could be possible in a specific language if you know the phonetic rules governing vowels and consonant allophones. In some dialects of English, coda consonants have special pronunciations so that /t/ is glottalized and unreleased in [hɪˀt̚] but aspirated syllable-initially in [əˈtʰon]. However, this does not have the requisite "same following" context. You can however get close to a minimal pair with "attack" and "that axe" where "tt" is aspirated in "attack" and flapped in "that axe". This correlates with the presence vs. absence of a word boundary. Then it depends on your analysis of these allophonic processes, whether you posit a difference in syllable position that the rules care about, or the presence of a word boundary.
There are few good candidates for such a distinction across languages, there an intervocalic consonant can syllabify to the left or to the right depending on the word. There is a general view that syllable-structure is not part of lexical representations, it is built by rules, therefore such a difference could only arise if "something else" is going on. That something else might be grammatical boundaries ("juncture"), or it could be phantom consonants or epenthetic vowels – otherwise, we expect VCV to syllabify as [V.CV] and there shouldn't be a contrast, following normal syllabification rules. Of course, a language might require a stressed syllable to be heavy so that /ˈVCV/ becomes [ˈVC.V] and /VˈCV/ becomes [Vˈ.CV], but then the syllabification difference isn't minimal.
There are very many competing analysis of the English facts, so many that you cannot just assert without argumentation that English has such a contrast. Instead, you have to first decide what kind of theory you are operating with – what mechanisms are available? In dialects that have a difference in production of "t(t)" dependent on whether the obstruent is intervocalic at a word-internal versus phrasal level, one can avoid claiming syllabification differences by assigning aspiration relative to word-internal makeup, where "at" is not intervocalic – it just becomes so between words. If you insist on positing syllabification on all phonological outputs, you would then readjust syllabification in "at axe" without reassigning aspiration based on derived syllabification.
In other words, you have to know the rules of the language, and really you also have to be able to justify your claims about that those rules are.
