(The first half of this discussion is from The Latin Tamer. The second half—finer points— explains things that are especially useful for understanding the audio recordings on the website.)
Nobody knows exactly how Latin sounded in classical times, but scholars have worked out confident ideas about much of it by studying various sources: Latin poetry, which followed known rules; or variations in spelling that point to a particular pronunciation; or patterns in languages that descended from Latin; or statements by Latin grammarians themselves. This chapter summarizes the most common understandings. The recordings at the website for the book will let you hear what these principles might mean in practice.
Pronouncing the Letters.
a. Generally. In Latin there are no silent letters. Every letter is sounded in some way. And every Latin word has as many separate syllables as it has vowels or diphthongs (two vowels said together, such as ae pronounced as ai /aye). So the Latin word cane isn’t pronounced like the English version; it’s pronounced with two syllables: cah-neh. (It’s a command: sing!)
The Latin alphabet was the same as the English one, but without w. I and j were the same letter in Latin, usually written i. This letter i worked both as a vowel (which we’ll talk about later) and as a consonant, which sounded like the y in yacht. (The j sound as we know it in English didn’t exist in Latin.) Nowadays some people who write in Latin use i for the vowel and j for the consonant, but many use i for both.
The letters u and v were originally the same, too—sometimes written more like u and sometimes more like v, depending on the type of writing. This u /v letter doubled as a vowel (this, too, will be described below) and a consonant. As a consonant, it sounded like the English w. (The English v sound didn’t exist in classical Latin.) People writing in or about Latin today often make the u /v distinction (writing the vowel as u and the consonant as v), though not all do.
Thus vīvō is roughly pronounced wee-woe. Vēnī vīdī vīcī is roughly pronounced way-nee wee-dee wee-kee. (Those efforts to capture the pronunciation using English equivalents are rough because the oe and ay sounds in English are diphthongs, as we’ll see again below. An ancient Roman who heard you say those expressions in the way just shown would probably understand you, but would perceive that you weren’t a local.)
b. Consonants. Most Latin consonants are pronounced as you’d expect in English, but there are a few differences to note.
The letters c and g are always hard. So Caesar is pronounced Kai-sar, and Cicero is pronounced Kik-er-oh. The word giganteus is pronounced with both g sounds the same—like the second g in our word gigantic.
The letter r is rolled, or trilled (as in Spanish today). It was compared at the time to the sound of a growling dog.
The letter s always sounds like it does in the English words send or say, not with a z sound as in the English words ease or lose or those.
The combination ph is pronounced just as p but with a little more air—not like the English f. So the word philosophus isn’t pronounced fil-oss-oh-fus. It’s more like pih-loss-oph-us. If you pronounce the p in pater with a candle in front of your lips, the flame shouldn’t waver. If you pronounce the ph in philosophus, it should.
Likewise, the combination th is pronounced as a t with some air in it. The word theatrum is pronounced teh-at-rum.
The combination ch is always pronounced hard, as in the English words chorus (with an added puff of air), not as in the word cheer or beach.
The letter m is mostly similar to the English m. It’s believed that in classical times, though, an m at the end of a word (as in piratam, templum, or patrem) wasn’t pronounced as such; instead it added a nasal quality to the preceding vowel and lengthened it at the same time. But there’s a post-classical history of simply pronouncing the m (and leaving the preceding vowel short and non-nasal), and many Latinists do it to this day. (Nasal vowels are also thought to have occurred in a few other places, at least in some periods; for example an n before an s or an f—as in consul or inferus—may have had the same effect as a final m.)
Consonants that are written double are also pronounced double. Thus the two ns in annālēs (meaning annals) sound like the two ns in unnatural, not like the two ns in annals (which are really pronounced as just one n). The double t in mittere (meaning to send) is held out longer than a single t. The double s in missus (meaning sent) is held out longer than a single s.
c. Vowels. Latin vowels can be either long or short. Those two words refer to quantity—that is, to how long it takes to sound out the vowel. The long a is stretched out longer than the short one. English doesn’t stretch out vowels as much as some other languages do, such as Italian and Latin. But compare the a sound in bat to the sound of it in bad (or in father). In those last words, the vowel lasts a little longer. Or compare the sound of the letter o in got and god; the first is short, the second is long. Latin did more with this kind of stretching. Some commentators suggest that a long Latin vowel should take twice as long to pronounce as a short one. Sometimes two words spelled the same way have different meanings just because the vowel is pronounced long in one of them and short in the other.
Some long vowels are also thought to have produced, starting at an uncertain point, somewhat different sounds than the short versions. Here is the rundown:
Short a: like the a at the start of aha!
Long a: like the a in father—the same sound as short, just sustained longer.
Short e: like the e in bet.
Long e: the vowel sound in the word they or clay. (Roughly. The English sound in those words is actually a diphthong. That means it’s made up of two vowel sounds, with the first one gliding into the other; if you listen carefully, you’ll hear that those two words end with a y sort of sound. The Latin long e, by contrast, is a pure vowel, not a diphthong: it doesn’t have the y sound at the end. This pattern can make it hard to precisely describe long Latin vowels by giving English equivalents.)
Short i: pin, tin, win.
Long i: an ee sound, like the i in machine, or the ee and ea in feel and easy.
Short o: as in the words off or hot spoken with a British accent.
Long o: oh—as in the words drone or clove. (Again roughly, because this o is a diphthong in English—gliding into a w-like sound—while it’s a pure vowel in Latin.)
Short u: sounds like the vowels in foot or put (not like putt ). So sum sounds a bit different in Latin than in English. It’s not pronounced like the word some; it’s closer to zoom, but quicker and starting with an s sound.
Long u: sounds like the vowels in rude or food.
How will you know whether a vowel is long or short? That’s one of the points you pay attention to when learning a word: not just its meaning, but how its vowels sound. In this book, long vowels are marked in the notes with a macron (a line) over them. That practice wasn’t followed in ancient times and isn’t followed in most complete texts of Roman writings now, but it’s very helpful when you’re learning the language.
One last note about vowels. In English, many vowels lose their distinct qualities when they’re in unstressed syllables. Take hippopotamus and rhīnocerōs as examples, since those words have the same spelling and meaning in Latin and in English. (They originally came from Greek, but never mind that now.) In English the vowels in the last syllables of those two words sound pretty much the same, even though they’re spelled differently. In each case, the last vowel is pronounced as a “schwa”—a kind of middling or neutral vowel sound. English vowels often get reduced to a sound like that when they aren’t stressed (and occasionally even when they are stressed). But this doesn’t happen in Latin. The last syllables in hippopotamus and rhīnocerōs contain vowels that produce distinct sounds.
d. Diphthongs, as noted a moment ago, are pairs of vowels that produce one sound. The sound usually has some slight movement within it (the tongue or the shape of the mouth moves from the start of the sound to the end). Latin has several pairs that can work this way:
ae: pronounced aye; haec sounds like the English hike. Caesar = Kai-sar.
au: pronounced ou, as in foul or house. The Latin laudō is pronounced loud-oh.
ei: pronounced as in the English word reign. The interjection ei (alas! woe!)is pronounced that way. But these letters are usually pronounced separately. Deī is pronounced day-ee, not day.
eu: pronounced e + oo as one sound, something like the sound phew or mew. Rare.
oe: pronounced oi, as in boil. The Latin coepit is pronounced coy-pit.
ui: pronounced oo-(w)i in one syllable. (The oo is short, as in book. The i is short, too, like in pit.) These letters form a diphthong only in the following words: cui, huic, and hui (a rare interjection).
The first of those diphthongs (ae) is the most common. As noted, sometimes these combinations aren’t pronounced together (and some people don’t count all of them as diphthongs at all). The most important cases where ei and ui are pronounced as diphthongs are the examples shown above.
Accented Syllables.
In a Latin word with more than one syllable, as in an English word of that kind, one of the syllables gets emphasized. There are basically three rules for knowing which.
Some Latin words have more than three syllables, in which case it might take a moment of attention to get the accent onto the right one: mīrāminī (mī-rā-mi-nī), intelligitis (in-tel-li-gi-tis), incommodus (in-com-mo-dus), ambulābimus (am-bu-lā-bi-mus).
A syllable can be long/heavy for either of two reasons. First, it can be long because it contains a long vowel or a diphthong. Examples: amāre (a-mā-re), vidēre (vi-dē-re), audīre (au-dī-re), ipsīus (ip-sī-us), illīus (il-lī-us), rēgnātis (rēg-nā-tis), negātus (ne-gā-tus), īnfīnītum (īn-fī-nī-tum), inimīcum (i-ni-mī-cum).
Second, a syllable is long (or heavy) if it ends in a consonant. (This is known technically as a closed syllable.) Here are some examples: īnfernō (īn-fer-nō), dēpugnō (dē-pug-nō), honestae (ho-nes-tae), lībertās (lī-ber-tās), libenter (li-ben-ter), magister (ma-gis-ter), secundum (se-cun-dum), respondit (res-pon-dit), negantur (ne-gan-tur), mūtantis (mū-tan-tis), susceptus (sus-cep-tus), neglegentis (neg-le-gen-tis), intolerandus (in-to-le-ran-dus).
This last principle can seem tricky if you aren’t sure where one syllable ends and another one starts. It’s helpful to understand two more rules. First, when a consonant is followed by a vowel, the consonant gets attached to the vowel that comes after it, not to the one before it. So domus is divided into syllables as do-mus, not dom-us. Second, when vowels are separated by two or more consonants, the first consonant usually stays attached to the first vowel. This causes the earlier syllable to end with a consonant, and makes that syllable long. All this is easier to understand with examples; and the examples at the end of the previous paragraph illustrate the idea. The practical point to notice: two consonants in a row are usually a signal that the syllable leading up to them is long—and (if it’s the second-to-last syllable) that it receives stress.
Calling (c) the exception and treating (b) as the rule is arbitrary. You could call (c) the rule and (b) the exception. It doesn’t matter.
There are a few other refinements. Certain consonants (such as x) are treated as if they were two (because they are, indeed, two in sound; for instance, x = k + s); and some combinations of consonants (such as ch, ph, and th) are treated as one consonant (because they represent single sounds). When one of the consonants b, c, d, g, p, or t is followed by an l or r, both consonants often go together with the next vowel for syllabification purposes. So the word cerebrum can be ce-reb-rum according to the usual rule; it can also be ce-re-brum instead. But the general rules above are all that you need most of the time.
Finer points.
When a word ends in m, the combination of that m and the previous vowel (–am, –um, –em, –im) was probably pronounced in many cases as a nasal vowel (rather than as the vowel + a clear m sound). A nasal vowel is a vowel that is pronounced with some airflow from the nose, as in the French word non. But the way that the m is pronounced can vary depending on what comes after it:
—When followed without pause by a word starting with c, q or g (e.g. cum crēderet, pecūniam quaerit, rem gerit), the m takes on a quality like that of English –ng at the end of a word (as in sing or doing). This also applies when a word with a final –m has the enclitic –que attached to it (e.g., virumque) or in compounds like numquam or quamquam.
—When followed without pause by a word starting with n, d or t, m is pronounced as n. E.g., multum nōbīs prōfuit would sound like multunnōbīs . . . This also happens when a word ending in –m has the enclitic –ne attached to it; e.g., multumne would be pronounced multunne.
—When followed without pause by a word starting with m, b or p, m is pronounced . . . m!
—When followed by f, l, r, s, v or a consonantal i (pronounced like the y in yacht), vowel + m has been pronounced as a nasal vowel in our recordings (even though there is doubt about those cases). It has also been pronounced as a nasal vowel when it’s at the end of a sentence or followed by a significant pause.
When the word following the final vowel + m starts in a vowel or diphthong, something yet different tends to happen: the nasal vowel gets elided (like any other vowel; we’ll come back to this in the section on elision).
This phenomenon is thought to have also occurred across what we think of as word boundaries. Thus, it would have affected the preposition in when followed by a word starting with s or f. In the phrases īn sēmitā or īn forō, īn would represent a long nasal i (although the i of in is short in other contexts). Those rules have been followed in our recordings.
In later stages of Latin, we know that y came to be pronounced the same way as i, and some people may already have done it that way in the classical period. Our recordings use the classical-Greek-style pronunciation.
There’s debate about the extent to which elision was applied in prose. It’s pretty much certain that it occurred to some extent, but was it as systematic as in poetry? In our recordings, elisions have usually been made except in a few cases—such as when they would have caused ambiguity, when it seemed natural to mark a pause between two words, and sometimes in monosyllabic words like cum (more on that last point below).
There is also some debate about whether the first vowel was dropped altogether or rather merged in some way with the next one—or if one of those things happened in some situations and the other thing in others. In our recordings, the vowel (or diphthong) has usually been dropped, but occasionally merged when it’s a long i or a long u. Magnī operis may have sounded something like magn’yoperis and exercitū ingentī as exercit’wingentī.
Monosyllablic words like cum or tam can undergo elision in the usual way. Funny as this may sound, cum īret can become c’ īret—unless it was something along the lines of cwīret (probably with some nasalization) if the proponents of the “merging vowels” theory are right. In any case we know that elision or something like it (whatever the exact form it took) could occur there because we see it happen in poetry, where the meter makes it clear. Yet poets didn’t do it very often, which may possibly indicate that there was some uneasiness about it. They didn’t often use such combinations without elision, either; in fact that happened even less often (because it created a hiatus—the presence of two vowels in a row without elision, something that poets usually avoided; see the section on meter). They just seemed to have preferably (though not strictly) avoided those combinations altogether as if they were unsure what to do with them or didn’t like the result much either way.
From that fact, one may speculate that such words perhaps tended to not be elided in ordinary speech, though this is very uncertain. In our recordings, they’re often not elided (but sometimes are). (Note one monosyllabic word that is never elided: the interjection ō! If it were elided, what would remain of it?)
Now let’s come to the two exceptions alluded to in the first paragraph of this section: cases where the first vowel of the second word, rather than the last of the first word, gets elided. The e of est and es is elided when it follows a vowel (including a nasal one). So bonum est becomes bonum’st and pulchra es becomes pulchra’s (the elision is sometimes reflected in the spelling—e.g., you’ll occasionally see things like bonumst—; but the elision generally occurs in pronunciation even when est or es is spelled fully—which is most of the time). This type of elision—the elision of the initial vowel of a word, as opposed to the final one—is called prodelision.
Other exceptions are as follows:
—One might sometimes wonder what happens to the accent when a word is contracted. Take the genitive singular form cōnsiliī. It’s accented cōn-si-li-ī. But cōnsiliī can be contracted into cōnsilī. What happens to the accent then? Does it remain where it was before (cōn-si-lī), or are the usual rules applied anew to the new form (cōn-si-lī)? In our recordings, we’ve chosen to keep the accent where it was in the non-contracted form. This seems consistent with what happened to words like adhūc, etc. (see above). The process, though not exactly the same, is similar.
—When a word has an enclitic attached to it (such as –que, –ne or –ve), the whole of the word + its enclitic becomes what, for accentuation purposes, is essentially a new word to which the usual accentuation rules must be applied. So the accent will not necessarily remain where it was in the main word. For instance, virum is stressed vi-rum, but virumque becomes vi-rum-que.
Some ancient grammarians said (or implied) that those enclitics always shifted the stress to the syllable preceding them, period. Thus, e.g., the nominative magnaque would become mag-na-que although that’s against the usual rules, the second-to-last syllable being short. There is doubt whether the claims of those grammarians are accurate. The second-to-last syllable in such compounds is long more often than not; e.g., magnusque, magnumque, magnīque, etc. Maybe those grammarians took that as the rule and overgeneralized. It’s also possible, though, that at least some people did put the stress on the second-to-last syllable even when it was short, by analogy with the many other forms where it was long. Our recordings follow the usual accentuation rules when dealing with enclitics: the syllable before the enclitic doesn’t receive the stress if it’s short.
—It is believed that when two words formed a very close unit together, they could be treated as a single prosodic word (i.e., be treated as a single word for accentuation purposes). This would have happened particularly often with prepositions, especially monosyllabic ones like ab, ex, in, etc. The preposition and the word following it would form a single word for accentuation purposes. Thus in viā would be stressed in-vi-ā. Our recordings follow that rule most of the time where monosyllabic prepositions are concerned.
Disyllabic prepositions such as inter, super, propter, intrā, etc., may have worked a little differently; but they probably formed or tended to form “a single word” at least with personal pronouns (mē, tē, sē, nōs, vōs). Thus, e.g., propter mē would be stressed prop-ter-mē. Our recordings usually treat disyllabic prepositions as forming a single prosodic word with the word following them when that word is a personal pronoun or a demonstrative, but not otherwise.
In some words, the phenomenon was so common that the version ending in a short vowel became the standard. The dative pronouns mihi, tibi and sibi, for instance, originally ended in a long i: mihī, tibī, sibī. But by the classical period, mihi, tibi and sibi with a short final i had become the more common forms (though the forms with a long i still appeared occasionally).
This shortening also sometimes happens in words where it’s less common. For example, the ablative singular form of domus is usually domō with a long o; but with the iambic shortening it can become domo with a short o. The short-o variant of this word occurs in one of our recordings from Plautus.
The phenomenon is called iambic shortening because the combination of a short syllable followed by a long is called an iamb.
Ecclesiastical Pronunciation.
The so-called ecclesiastical pronunciation of Latin differs from the classical in a number of ways. Latin is the official language of the Catholic Church (though other languages have increasingly been used in the Church in recent decades). During the Church’s long history, Latin has undoubtedly been pronounced in a multitude of slightly different ways depending on times and places. Not all those ways can be described here or even known for certain. Nevertheless, there are general tendencies and some kind of a standard that people often have in mind when they talk of “the ecclesiastical pronunciation.” That “standard” has, unsurprisingly, been largely set by the Vatican. As a result, it sounds rather like Latin pronounced with the rules of Italian. (Note: Italian is descended from Latin, and some of the sound changes that led to Italian pronunciation had already occurred in Late Latin. So the ecclesiastical pronunciation is likely similar in some respects to what you might have heard on the streets of late antique or early medieval Rome.) The main distinguishing features of the ecclesiastical pronunciation are as follows:
—The classical distinction between long and short vowels is largely ignored in ecclesiastical Latin, though a long vowel tends to remain long when it’s in an open stressed syllable. An “open” syllable is one that ends in a vowel rather than in a consonant; e.g., mā in a-mā-tus is an open syllable; crēs in crēs-cit is a closed one. The ā of amātus would tend to be given more length in ecclesiastical Latin than the ē of crēscit, though both are long in classical Latin. When a classically long syllable occurs in an unstressed syllable (e.g., the ā in a-mā-vē-runt), it’s often shortened in ecclesiastical Latin. Overall, the ecclesiastical patterns of vowel length should be a little more intuitive to English speakers than the classical ones.
We’ve said that a classically long vowel in an open stressed syllable tends to remain long in ecclesiastical Latin. It can even happen that a vowel in such a position becomes long when it was classically short; but that hasn’t been done in our recordings.
—There are some subtle differences in the qualities of vowels. Some short vowels tend to be more similar in quality to the corresponding long classical vowels—but without necessarily becoming long. For example, a short e in an open syllable (see the previous paragraph for the definition of “open”) tends to sound a bit like the classical long e (just, usually, a shorter version of it). And a short i could be perceived by some ears as more similar to the ee in English bee than to the i in sit, despite being short.
—There are no nasal vowels in the ecclesiastical pronunciation. Final ms are simply pronounced as m, and the combination of a vowel + n occurring before s or f is simply pronounced as that vowel (usually shortened) + n.
—Y is pronounced the same as i.
—The diphthongs ae and oe are pronounced as a simple vowel resembling the classical long e (but not necessarily long).
Note that ae and oe don’t always represent diphthongs in classical Latin. Sometimes they represent two distinct vowels, as in aer (roughly pronounced ah-air) and poeta (roughly po-ey-ta). In those cases, they keep being pronounced as two separate vowels in ecclesiastical Latin.
—In classical Latin, c and g are always pronounced hard, as in English cot and good respectively. That can change in ecclesiastical prounciation. When followed by e or i (or by y, which is pronounced the same as i; or yet again by the classical diphthongs ae or oe, which basically amount to e in ecclesiastical Latin—see above), c is pronounced like the ch in English cheese; e.g., cēdit would be pronounced roughly chay-dit, caelum would be pronounced roughly chay-loom, and cibus would be pronounced roughly chee-booss. G is pronounced as in English allege when followed, again, by e or i (or y, or yet again ae/oe-turned-e). Thus gerit is roughly jay-rit, gigās is roughly gee-gas, and gȳrus is roughly gee-rooss. When the combination sc occurs before those same vowels, it becomes like the English sh sound (sometimes doubled unless it’s at the beginning of a word). Thus ascendit is roughly a(sh)-shen-dit and scelus is roughly shay-loose.
A double c or a double g before those same vowels has about the same quality described for the single c and g above, but the ch or j sound, or rather the first part of that sound, tends to be doubled in the following way: ecce would be pronounced approximately et-chay, and agger would be roughly ad-jair.
—Ti before a vowel is pronounced tsee most of the time. Thus, laetitia is roughly lay-tee-tsee-a, and āctiō is roughly ac-tsee-o. Exceptions are when ti + vowel follows the letters s, t or x. Then it’s simply pronounced tee. So for example molestia is approximately mo-less-tee-a, not mo-less-tsee-a.
—V is pronounced as in English, not w as in classical Latin.
—Ch is typically pronounced like a simple k, without an added aspiration.
—Th is likewise pronounced like a simple t.
—Ph is pronounced f.
—H is typically silent at the beginning of words. Hortus and ortus sound the same. Chiefly in the words mihi, nihil and nihilum, it can be similar to the sound of ch in German ich; or it can sound rather like a k. The former pronunciation has been adopted in our recordings. (Inside some other words—such as Isrāhēl, a variant spelling of Isrāēl—h can be used merely as a silent indicator that the vowels on either side of it are pronounced separately. This use of h, especially in variant spellings, already existed in classical times.)
—Elision mostly doesn’t happen in ecclesiastical Latin.
Recordings using the ecclesiastical pronunciation have been provided for illustrations taken from the Vulgate Bible (and occasionally other ecclesiastical sources).