Latin pronunciation for the continuing student

On bulletin boards and in magazines dealing with classical homeschooling, one question that arises over and over again is, “What sort of pronunciation should we use in teaching Latin?” The options usually boil down to two: the reconstructed classical pronunciation, and the Italianate ecclesiastical pronunciation. Both have their champions, and the discussions that follow in their defense usually generate more heat than light. A lot of the discussion is usually centered on which one is right.

Asking “Which pronunciation is the right one?” is an exercise in historical reductionism doomed to fail. One cannot define an entire spectrum from a single point, and the history of Latin as a living language extends for somewhat over two thousand years. Either is right. Neither of them is satisfactory for all occasions.

Typically, the most attention is given this question by parents just starting out in Latin instruction. At this point, the question is more or less moot, and any real anxiety is out of proportion with its pedagogical significance. While learning forms — declensions and conjugations — it doesn’t matter much how you pronounce them, as long as you learn those forms, what they mean, and what they’re for. For practical purposes, therefore, my own suggestion is to pick one — whether purposefully or arbitrarily — and use it consistently for the first year or two. You’re probably better off choosing a pronunciation matching the kinds of texts used in the introductory text. With something like Wheelock’s Latin Grammar, which draws most of its examples from classical authors, you probably want to go with a classical pronunciation. If you’re using a course like Henle’s, which is based on ecclesiastical texts and ecclesiastical authors, then it only makes sense to go with that as your pronunciation standard. If your chief reason for learning Latin at first is to be able to sing church music, that’s a good reason to start with an ecclesiastical pronunciation as well.

Later on, though, pronunciation will become significant, especially when one begins to deal with literary products. Poetry in particular is at least largely about the sounds of a language. I’ll discuss that a little bit later. First, however, it’s probably worth dispelling some of the widespread misinformation that gets circulated.

The one I’ve heard most frequently is, “There are no recordings of classical Latin speakers. It’s clearly impossible to know how the language was pronounced.” This is generally used as a way of dismissing the classical pronunciation, though a parallel argument could be used as easily to dismiss any other system. Unfortunately, those who make this argument are merely asserting that they don’t know how to figure something of this sort out. But there are those who do.

At the subtlest level, yes — there are things we don’t know. We’d give a lot to be able to plant even one microphone in the Forum to pick up just one of Cicero’s orations. But we actually do know, with fair accuracy, how the major inventory of language sounds were produced. Historical linguistics is a slow and painstaking process, but over its long history people really have taken those pains, and so there is now a substantial body of data available for analysis.

Detailing all those sources of information is beyond the scope of this discussion, but a few examples may suffice. We do have a few grammatical and literary discussions about mispronunciations, of course. These are at least somewhat interesting. But they usually document the egregiously odd — such as Catullus’ harangue against a certain Arrius, who added initial “h” sounds to a lot of words that should have begun with a vowel. Those are colorful, but provide less information than we might wish, and almost no information about what was normal. There are, moreover, relatively few of them.

Just as one might read novels and the publications of the popular press today without learning a great deal about how we pronounce English, one could stare at a page of Cicero for the next ten years and learn little or nothing about how Cicero pronounced it. It would help you very little in distinguishing classical from ecclesiastical pronunciations.

But those are literary texts, and literary texts are not the only tools of the discipline. The real treasures for the historical linguist are errors. Some of the papers I get from my students, for example, could provide more information about how we speak than a ten-year run of National Geographic or New Yorker: those who write “I might of known” instead of “I might have known” are providing virtually irrefutable evidence that, in its auxiliary usage, “have” is normally pronounced much the same way as “of”. That will tell us something about the loss of the initial h; it will also tell us that in “of” the final f is like a v. The fact that one sees, with increasing frequency, comparative phrases formed with “then” rather than “than” illustrates the fact that in an unemphatic position (as these connective words almost always are), the vowel itself tends to settle down to about the same middle schwa sound (ǝ).

Our surviving evidence from the ancient world is (unsurprisingly) short on student papers, but they are not short of inscriptions scratched into stone of one sort or another. Some of these are quite elegant; others are primitive — the desperate efforts, for example, of a grieving parent who wants to memorialize his dead son or daughter as best he can. Often that best is riddled with misspellings. The inscriptions themselves are often rather moving, reaching across centuries with an uncommon universality, but in addition, almost every one of them tells us something about the language.

Anyone interested in the detailed conclusions about classical Latin, and the fastidious work that has gone into reconstructing it, would be well advised to take a look at W. Sidney Allen, Vox Latina. It’s fairly dry going, unless you have the philological bent, but it’s worth reading if you do. It argues every point with very solid evidence.

Of course the “we can’t know” argument is not the only one out there. Others are more belligerent and random. One of the more bizarre ones I’ve encountered over the last few years includes the reflection that “if it’s good enough for Dante, it’s good enough for me.” This sounds full of conviction, but substitutes triumphal ignorance for reason. Anyone even glancingly familiar with rhetorical fallacies will identify it as an appeal to inappropriate authority. Dante, writing a little more than 1300 years after Vergil (whom he regarded as his master), had no better direct access to recordings of Classical Latin than we do, but certainly lacked all the comparative evidence that has been marshalled over the last two centuries. To read Vergil as Dante did is probably a useful exercise, if you are interested in learning what Dante was hearing. It tells us virtually nothing about what Vergil was writing, however.

So does it matter what kind of pronunciation you use, and if so, why? To start with, no. It will obviously not affect your conversation with native Romans. It will probably not vastly affect your understanding of Latin texts. Some of the best classicists I have known have had very peculiar pronunciation. They seemed to get along. The English have had a long tradition of some of the finest classical scholarship in the world, coupled with with some of the worst pronunciation imaginable.

But if you want to deal with authors on their own terms, you probably need ultimately to learn and use two (or perhaps more) different ways to pronounce Latin. Sure, you should start with one method while you’re learning the ropes. But if you really want to appreciate the Latin that was written over a space of a thousand years, you have to be ready to adapt. It’s not really that hard, and the fruits of the exercise are considerable.

What’s wrong with reading classical Latin as if it were Mediaeval Latin? It’s not merely that it’s wrong. It’s not, I would argue, morally wrong, and if you can read and appreciate Cicero’s orations while reading them with a thick Italian or English accent, fine. But you will lose the music of the language, and especially with poetry, that’s important. Just as a brief illustrative case, let’s look at two consonants and a diphthong that are treated differently in Classical and Mediaeval pronunciations.

  • In classical Latin pronunciation, the letter C is invariably hard — like our K. It does not vary with position. In ecclesiastical Latin pronunciation, it will change to something like our CH sound (as in “church”) when followed by an I or an E.
  • Similarly, in classical Latin pronunciation, the letter T is invariably hard. In ecclesiastical Latin pronunciation, it will change to something like our S or TS sound when followed by an I or an E.
  • The diphthong AE in classical Latin is a true diphthong — beginning with A (as in “amen”) and gliding into an E or I sound — much like our word “eye”. In the ecclesiastical pronunciation, it is flattened to the equivalent of E — much like what we call a “long” A in modern English.

So in the classical pronunciation, the word “caelum” (heaven) comes out to something like “kylum”. In ecclesiastial pronunciation, it is going to be more like “chaylum”.

Consider the implications in the following fragment from the beginning of Bk. II of Vergil’s Aeneid. It’s written in the ancient meter reserved for epic and didactic poetry, dactylic hexameter. The meter is quantitative, and the lines are unrhymed.

A few lines into the book, one encounters the remarkable lines:

…Et iam nox umida caelo
praecipitat, suadentque cadentia sidera somnos.

…And now dewy night from heaven
descends, and the sinking stars bid us to sleep.”

Vergil achieves something remarkable here (and he knows it’s good: he quotes himself later in Bk. IV):

In a classical Latin pronunciation, the vowels are dark and muted; and the two words in the middle of the line contain an internal rhyme (suadentque cadentia), are followed by two words alliterating in S. The effect is lulling and hypnotic.

In an Italianate ecclesiastical pronunciation, all that is ruined. Praecipitat becomes something like praychippytot; cadentia becomes more like cadensia, which piles up one S-sound too many at the end of the line, so that the whole thing begins to hiss like a basket full of vipers.

Lest I seem to be exhibiting a bias in favor of the classical pronunciation, let me hasten to point out that one can achieve a similar train-wreck by reading mediaeval verse in the wrong way, too. Take the following example from the beginning of the monumental De contemptu mundi by Bernard of Cluny. It’s written in something also called dactylic hexameter, but it’s of a completely different sort. It’s qualitative (stress accent, rather than duration); its lines are rhymed internally (but always at word-end) at the end of the second and the fourth dactyls, and couplets are end-rhymed.

Hora novissima, tempora pessima sunt — vigilemus.
Ecce minaciter imminet arbiter ille supremus.
Imminet imminet ut mala terminet, aequa coronet,
Recta remuneret, anxia liberet, aethera donet.
Auferat aspera duraque pondera mentis onustae,
Sobria muniat, improba puniat, utraque iuste.
Ille piissimus, ille gravissimus ecce venit rex.
Surgat homo reus; instat homo deus, a patre iudex.
Surgite, currite simplice tramite, quique potestis;
Rex venit ocius ipseque conscius, ipseque testis.

To read this in a classical voice is to crush its rhymes: ocius and conscius in the last line there are meant to rhyme, but won’t, unless one follows the ecclesiastical norms for how to handle C; if one keeps a classical diphthong pronunciation of AE, the end-rhymes between onustae and iuste are obliterated. The driving, almost manic energy of Bernard’s apocalyptic lines drains away.

My point here isn’t to champion one form of pronunciation over another. It’s to recommend that a maturing Latinist — and I would include anyone who has done three or four years of Latin with Scholars Online — should learn to adapt his or her reading to the text at hand. If nothing else, it’s an act of humility before the material at hand, and that is probably a good thing in and of itself.

Leave a Reply