[supermatt]

It seems googling for "what did middle english sound like" often comes up with a link to this video as the definitive example, but how do we know it was pronounced phonetically, or what the phonemes were?

It sounds lovely, and I would like to imagine it sounding like that - but I don't see how we can reconstruct its pronunciation reliably over 500 years later.

[laxmis]

You might be surprised. Phonetics and historical linguistics are proper subjects of study, not quackery. I've heard teachers say that in England people used to pronounce Latin much like English until the specialists rescued much of the original pronunciation of Latin for the public. Today every Latin primer, especially the British and the American ones, insist that you read Latin in the reconstructed way.

WS Allen in his book, Vox Latina, actually explains how Latin might have been pronounced in the Roman period. Interestingly his theories have universal acceptance today.

[dwohnitmok]

You can't ever be 100% sure, but there's a variety of ways to get confidence that you're getting close.

Rhyming lines in poetry are a great way to constrain what the various phonemes were. The meter is also a great way to back out what sounds were silent and which were not. (Speaking of which I disagree with the video's choice to pronounce a lot of the final "-e"s in words; that's a trait of early Middle English that was already on its way out by the time of Chaucer and its pretty clear from certain rhyme choices that it's silent in a lot of Skelton's words: see how e.g. Skelton rhymes "Jerusalem" with "creme" and how everything is nice iambs, unless you start pronouncing "-e"s everywhere).

500 years is also within a time horizon where you can get pretty reliable internal reconstruction clues (examining inconsistencies in modern English and thinking about what common features might have led to those inconsistencies) to try to get a better grasp of the phonemic inventory of a language.

Spelling also is a great clue into what the phonemes of the language were (although this is complicated by digraphs and trigraphs).

Then you can use a variety of methods to try to get an idea of what the phonetic realizations of those phonemes were. In this poem in particular we have English words rhymed with foreign words which is a great clue. You can also compare different dialects of English to try to figure out what the original sound is of a word that has now fractured into many different pronunciations.

And of course you can examine what people of the time say about the pronunciation of their own language (this is how we e.g. are pretty confident in r being a trill or a tap, based on descriptions from both domestic and foreign speakers, see e.g. https://english.stackexchange.com/questions/298566/why-and-w... for a great summary of some of the evidence).

All in all it's kind of like playing a game of Sudoku. Rhyming and other clues can set up a set of relationships among different sounds and what would seem like an otherwise sparse set of clues (a comment by some writer here, a common misspelling there, some translation here) can cause cascades of phonetic realizations of different phonemes that magnify the impact of those clues.

And the sanity check to make sure that all this makes sense is being able to provide reconstructed sounds that are consistent with new descriptions, rhyming patterns, and foreign transcriptions as new evidence rolls in over time.

[supermatt]

Fascinating stuff!