Pantheon Poets’ approach to Latin pronunciation

What’s the best way to find out how a language which is not your mother tongue should be pronounced?

Ask a native speaker – that is why Pantheon Poets uses them for French and German posts. But, for Latin (and ancient Greek), that isn’t an option. No-one can be absolutely sure exactly what native Latin speakers in the time of Augustus sounded like, and that applies to us at Pantheon Poets as much as to anyone else. But we do have evidence, which has been painstakingly put together by scholars who have combed through the ancient authors for clues. There is a pretty general consensus in the British academic world these days about how letters, syllables and words should be pronounced, and I will tell you later where you can find summaries.


One issue for spoken languages is where the stress comes within a word – taking English as an example, we say “your button is undone”, and it would sound wrong to say “your button is undone”. Lots of languages have conventions about this. Usually, it’s not too complicated in Latin: most of the time, if the syllable before last in a word is long, that is where the stress falls, and if it is short, the stress falls on the syllable before it (and little one-syllable words probably wouldn’t carry much stress at all in most cases). So, taking a couple of well-known lines from Virgil and Catullus, the italics show where the spoken word stress would normally fall:

sunt lacrimae rerum et mentem mortalia tangunt (events bring tears, and human experience touches the mind)

atque in perpetuum, frater, aveatque vale (and for ever, brother, hail and farewell)

But these examples are poetry, not ordinary speech. Poetry in both Latin and Greek depends on metre – patterns of long and short syllables in set combinations – and it is important when it is spoken to make the distinction between the two. In addition, the rules of metre mean that some syllables aren’t sounded at all, or perhaps only in a very vestigial way. So this is how the metrical emphasis falls, on the long syllables shown in capitals:


ATQ( ) in PER-pe-tu-UM, FRA-ter, av( ) AT-que va-LE

If you compare that with the word stress in the first pair of examples, it’s obvious that word stress in normal speech often doesn’t coincide with where the metrical stress falls (namely, on the long syllables). Metre emphasises some long syllables which would be unstressed in normal conversation (sunt, -mae, et, -tem, mor-, -gunt; at-, in, per-, -um, -le), and doesn’t emphasise some short syllables which would be stressed in normal speech (la-, -pet-, av-, va-). When I was at school in the last century, there was a bit of a tendency to duck this issue when reading Latin poetry aloud by just stressing the long syllables, or, in other words, giving metrical emphasis priority over normal word stress. That can feel quite satisfying, but the trouble is that it would probably often have sounded to a Roman like “your but-ton is un-done.”

There are differences of opinion about how to deal with this. Some people say it’s OK simply to pronounce the Latin as if it were normal speech, and let the metre take care of itself. Others stick to old-school metrical emphasis, and just stress all the long syllables. At Pantheon Poets, we try to reflect both normal word stress and the emphasis that metre places on long syllables. This means stressing some short syllables (following normal word stress) without making them long, and leaving some metrically long syllables unstressed without shortening them. Some syllables carry one kind of stress but not the other, some carry both and some neither. So, using italics again for normal stress and capitals for metrically long syllables, this gives the following:


ATQU( ) in PER-pe-tu-UM, FRA-ter, av( ) AT-que va-LE

This isn’t easy, but it is worth the attempt because the interplay of word stress and metre allows more variation in pace and rhythm than just going for one at the expense of the other. It also highlights more sharply passages where word stress and metrically long syllables coincide, as they often do particularly in the second half of a hexameter line, or when there is an important point or a dramatic effect to bring out. These variations and emphases do not happen by chance – they are part of the poets’ artistry, so it would be a shame not to try to give them full scope.

You can judge for yourself by listening to the poetry we post whether you think we achieve what we are attempting or not: I have recorded the three pairs of examples to try to illustrate what I mean and you can listen to them by pressing play at the foot of the page.

Incidentally, you may have noticed that the first and last “a”s in “ave atque vale”, which most English speakers pronounce “Arvay atque Varlay,” are in fact short.


There is an additional difficulty in knowing what ancient Greek poetry sounded like: the scholarly consensus is that until around the fourth century CE, Greek, like modern Mandarin, was a tonal language: in other words, the pitch at which a syllable was spoken mattered, as well as word stress and metrical value. The system of the accents which appear in our modern texts of classical Greek works were invented by ancient scholars many centuries after the classical period itself in an attempt to capture this tonal aspect of the language, which had by then long disappeared from current speech, but they are far from a self-standing guide to how tonality worked and sounded in say, Pericles’s Athens. The advice of the best academic authority on pronunciation I know (Professor Allen, see below) is that, while it is possible to establish reasonably confidently what tonality meant for individual words, it is so difficult to be sure how it worked at the level of whole sentences that it is best not to attempt it when reading aloud, and to concentrate on trying to be as accurate as possible in other respects. That makes practical sense, and it is broadly adopted, for example, in the CDs accompanying the widely-used Joint Association of Classical Teachers’ Greek Course, “Reading Greek,” where accents are generally pronounced as though marking stress, not pitch. But although we can never know just what a performance by an ancient Greek would have sounded like, that need not spoil our enjoyment, provided that we pay reasonable attention to what modern scholarship can tell us with reasonable confidence about pronunciation, that we are consistent in our approach and that we are honest about the limits of our knowledge.

Marcus Agrippa


See the illustrated blog post here.


“Vox Graeca” and “Vox Latina” by Professor W Sidney Allen are the best authorities I know on the pronunciation of Latin and ancient (Attic) Greek. Each is a demanding read, but they have brief and user-friendly “how-to” sections near the end.

“Reading Latin Poetry Aloud” by Clive Brooks is an extremely useful and entertainingly-written practical guide.

There is another handy summary near the beginning of Professor W Fitzgerald’s “How to Read a Latin Poem if You Don’t Know Latin Yet”.

To scroll both versions of the poem at the same time - tap inside one box to select it and then scroll.