How is stress realized in Latin phonetically?

Question

I have heard that Latin does not lengthen stressed syllables. If so, are they pronounced louder or with altered articulation, maybe a higher pitch?

Answer 1

We don't exactly know.

In Greek, we have a pretty good idea of how the accent worked: it involved a change in pitch, which was supposedly close to the interval of a fifth (as the grammarian Dionysius laid out the accent system quite precisely in musical terms).

In Latin, however, it's much less clear. Some of the grammarians talk about it, but never in clear terms: Priscian claims that every vowel has "height/depth, width/breadth, and length" (altitūdō, lātitūdō, longitūdō), for instance, and uses these to lay out ten varieties of the letter A. It seems that his "height" is the accent, divided into acute, grave, and circumflex; "length" is vowel length, as we measure it today; and "width" is whether or not there's an H before it (!).

Priscian's acute, grave, and circumflex accents are obviously inspired by Greek. His examples (when unaspirated and long) are árae àrārum âra, and (when unaspirated and short) ábeō àbīmus. So it seems like "grave" means unaccented, and there's some sort of distinction between "acute" and "circumflex" that only happens for long vowels. This could imply a pitch accent like in Ancient Greek, since that's what the diacritics originally meant, and only a pitch accent can be expected to vary between the first and second mora of a long vowel.

On the other hand, the oldest Latin grammarian (Donatus) gives a simpler explanation of the accent:

accentus in ea syllaba est quae plus sonat
The accent is on that syllable which is sounded more

This implies that the difference could involve amplitude rather than frequency: stressed syllables are louder, rather than pitched differently. Donatus was also writing much "closer to the source" than Priscian was.

Or, it's possible that the Latin accent involved both amplitude and frequency, like in English. The only thing we can be fairly sure of is that it didn't involve lengthening the vowel at all.

"Can we learn anything from PIE?" you might ask, and this is a good thought. But the Latin accent was completely separate from the PIE one: PIE's pitch accent died out in pre-Latin, and was replaced by a different system (stress always on the first syllable like in Etruscan), and then this system died out too and was replaced by the rules we know today. So PIE can't tell us anything useful about how the later accent systems worked.

My main source on all of this is Key's "Language: its Origin and Development". Various other sources also cite Priscian's ten varieties of A and three distinctions, but Key is the first one I found to provide quotes (since I can't find Priscian's actual text anywhere online).