Were enclitics considered part of a word for stress purposes?

Question

One rule of Latin stress is that it can never go farther back than the antepenult: the third syllable from the end. For example, we have cár-men "song", cár-mi-ne "with a song", and car-mí-ni-bus "with songs", but never *cár-mi-ni-bus.

When an enclitic is added to a word, does it count against this three-syllable limit? For example, "with a song" is cár-mi-ne; if I want to say "…or with a song", would that be cár-mi-ne=ve, or car-mí-ne=ve?

EDIT: Or perhaps car-mi-né=ve, a third option proposed by Joonas Ilmavirta in the comments?

(Ideally I'd like an answer about Classical Latin, but unfortunately stress wasn't marked in Classical times. It should be possible to answer about later Latin, though, because later texts sometimes used an acute accent to mark stress.)

Answer 1

As you mentioned, it is very difficult to say anything clear about stress in Classical Latin, because there is little evidence, either direct or indirect, of the position of the stress.

As Joonas Ilmavirta and Alex B. pointed out in the comments, there was supposedly a special rule in Latin about stressing the immediately preceding syllable, regardless of whether it is heavy or light, in words suffixed with a monosyllabic enclitic. The way that this rule violates the main heavy/light rule is often illustrated with the example Musaque [muːˈsa.kwe]. The "qu" here is not expected to make the preceding syllable heavy: "qu" is typically analyzed either as a singleton consonant with a secondary articulation of labialization (/kʷ/) or as a complex onset composed of a singleton /k/ followed by a glide /w~u̯/.

Although I've only seen examples with -que in discussions of this alleged rule, it is also described as applying to other monosyllabic enclitics like -ne and -ve (Mester 1994, p. 48; C. M. Weimer's answer to the question Are there exceptions to the Latin stress rules?).

Your question about whether we see an acute used on the syllable before -ne or -ve in postclassical texts is a good one. There is literature that says that the acute accent was used this way; "Accent Notation in the Classical Languages and its Influence on Lithuanian Accent Notation" by Mindaugas Strockis, cites Steenbakkers (1994) (see section 3.2.1 of Strockis). I don't know which specific documents show this feature, or how common it is (Strockis says that -ne and -ve were less frequent than -que).