There will be a few people doing "but how do we know" and the answer is (as I understand it) with some uncertainty, structural clues in things like rhymes, where the written form clearly indicates sounds which had to sound alike. If you follow the clues through language shifts over the centuries, you can arrive at a sense which ones preserved the rhyme, and which didn't and a huge amount of inference added in gets you back where you expect things started.
Latin poetic meter didn't employ rhymes, it used interplay of long and short vowels. Poetic evidence is how we know the vowel quantities of many words. Much of how latin pronunciation was reconstructed comes from comparative reconstruction of the Romance languages, and loans from and into other langauges.
There will be a few people doing "but how do we know" and the answer is (as I understand it) with some uncertainty, structural clues in things like rhymes, where the written form clearly indicates sounds which had to sound alike. If you follow the clues through language shifts over the centuries, you can arrive at a sense which ones preserved the rhyme, and which didn't and a huge amount of inference added in gets you back where you expect things started.
The same with the great vowel shift.
Bill and Ted and So Crates.
Latin poetic meter didn't employ rhymes, it used interplay of long and short vowels. Poetic evidence is how we know the vowel quantities of many words. Much of how latin pronunciation was reconstructed comes from comparative reconstruction of the Romance languages, and loans from and into other langauges.