Fireside Friday, August 15, 2025 (On Latin Pronunciation)

Hey folks, Fireside this week! The new semester is starting up next week, so things may be a little more fireside-y than usual over the next few weeks, but I do promise we will get to the end of “Life, Work, Death and the Peasant” eventually. That said, since I am teaching Latin rather than history this semester, I thought it might be fun this week to talk briefly about Latin pronunciation (at least in the English-speaking world). I’m going to try to keep this simple-ish and I am sure some of those simplifications will draw the ire of more skilled philologists, but mostly here I am interested in the fact of multiple pronunciation systems, rather than their precise differences. There are already any number of guides online, ranging from the simple to the very, very complex, on the precise differences between these systems.

There are a lot of pronunciation systems for Latin. This is a product of the language surviving as the language of education and diplomacy across much of Europe long after its living component had fragmented into the Romance languages. In a sense, we could even argue that the topic of ‘Latin pronunciation’ ought to include phonetic shifts between Spanish, Portuguese, French, Italian and so on, since those languages are, in a very real sense, ‘modern Latin’ as it is lived and breathed today. But as Latin lived on as an undead language, its local pronunciation in each part of Europe acquired the quirks of local pronunciation, such that by the modern period Latin was pronounced slightly differently in each.

As a result, the modern resident of an English-speaking country is likely to encounter three different pronunciation systems for Latin. ‘Three’ will also be the number for most speakers of other European languages, except for Italians, for whom there will largely be only two. But I am most familiar with the Anglophone Latin triplet, so that is where we will focus.

In the Anglophone countries, we term these systems ecclesiastical, traditional and classical. I’m not going to go through all of the differences between the systems here, just some of them, but we can exemplify them by picking a single Latin phrase and walking through how its pronunciation changes based on systems. A good one, because it has a lot of shifting sounds, is the legal principle stare decisis (which we ought to write stāre dēcīsīs to capture vowel length, that’s long-short, long-long-long), with the added benefit that we may educate a great many lawyers, many of whom seem to have adopted a pronunciation for the phrase which is – somehow – not correct in any of these systems.

We can start with what is generally called ecclesiastical Latin. Ecclesiastical Latin formed a standard pronunciation during the Carolingian Renaissance, but in practice its modern form is basically the globalization of the traditional Italian pronunciation of Latin that developed in the centuries since then. Other regions had diverged from that Caroline foundation in other directions, but for the last century or so the Catholic Church has held the traditional Italian version to be the ‘official’ pronunciation of the Church. As a result, ecclesiastical Latin’s pronunciation is essentially a simplified form of the original (‘classical’) Latin, with an admixture of Italian pronunciation elements.

The main simplification is vowels: whereas classical Latin vowels can be either short or long and change their pronunciation accordingly, in ecclesiastical Latin, each vowel gets one sound. In effect, I’d say that (compared to Classical), a, i, o, and u are always long, while e is always short. Thus filii, which in Classical Latin is long-short-long (fee-li-ee)¹ in ecclesiastical Latin each vowel has the same length, fee-lee-ee. Likewise, classical Latin unus is long-short (oon-uhs)² but in ecclesiastical Latin pronounced oon-oos.³

Meanwhile, the consonants get more complicated; where classical Latin has one sound per consonant (no hard or soft letters), ecclesiastical Latin has some letters carrying double-duty. The most immediately audible example is c: before e, i, y, ae, and oe it is pronounced ‘ch’ but in all other cases maintains the hard Latin c (like an English k in all cases); likewise ‘g’ is soft (like a ‘j’ sound or the ‘g’ in ‘gel’) before that same set of vowels but other wise hard (like ‘gap’). Thus ecclesiastical Latin says for genus jee-noos (cf. classical geh-nuhs) but for galea ‘gah-lay-ah.’ ‘V’ has a ‘v’ sound. There are other differences (see this handy guide), but you can see how Italian has influenced the pronunciation in ecclesiastical Latin.

Returning to our pattern phase, stāre dēcīsīs, none of those vowel length markers matter because the sounds are all the same in ecclesiastical Latin. But that ‘c’ proceeds an ‘i’ and so is going to be a ‘ch’ sound. So we’ll get stah-reh deh-chee-sees. The cheese stands (alone) indeed. Likewise che-sar for Caesar with that simple short e for the Latin ae-dipthong.

Next, there is traditional English Latin pronunciation, which in an Anglophone country is just going to get called ‘traditional Latin.’ No one learns this is school anymore, but it persists in common Latin words, phrases and especially names that entered the English lexicon prior to the reconstruction of Classical Latin in the late 19th and early 20th century. There is an irony that on the one hand these pronunciations are the most familiar to folks, but that no one actually uses this system anymore (we don’t teach it) and it is insanely complex and idiosyncratic, far more so than actual classical Latin, made yet more difficult because we still use these pronunciations, but don’t train anyone in them. It is, of course, complex and idiosyncratic because the basic rule here is to pronounce Latin as if it was English (with some exceptions) and English pronunciation is already a mess of exceptions and strange conventions, even before it is layered over another language.

Vowel lengths vary and do not necessarily follow classical vowel lengths, but they do generate different sounds. Quite a few consonants differ their sounds based on the letters around them and the overall syllable structure. Mostly these pronunciations feel intuitive to English speakers – this is, after all, mostly just “Latin pronounced like it was English” – but the moment you step off of the intuitive pronunciation being right, folks are out to sea because of how many letters shift sounds is odd ways. So you get that soft ‘c’ in Lucius, Caesar or Cicero, but a hard ‘c’ in Marcus and Caecus (both of them – kae-kus).

What makes this lingering traditional pronunciation a pain is that even in classics we generally hold on to it for names, for the obvious reason that if I talk to my students about ki-keh-roh they are going to be very confused, as opposed to Si-ser-oh, a figure they are vaguely aware of. For the most part, no one learns the entire traditional system (I sure didn’t), but instead learns each persisting traditional pronunciation individually. It is striking to me that while there is a movement among Hellenists to move away from traditional spellings and pronunciations of ancient Greek works to closer to their original (thus Seleukos (sel-e-ooh-kos) instead of Seleucus (seh-lou-kuss or even sometimes seh-lou-suss)) there doesn’t seem to be much of the same among Latinists, probably because everything thinks Ki-keh-roh sounds silly and doesn’t want to have to say it.

In any case, for stare decisis, traditional Latin pronunciation says stah-ray dee-sise-is (first ‘i’ is long, making an ‘eye’ sound, second is short, so decisis rhymes with ‘crisis’). Gaius Julius Caesar is pronounced Geye-uhs Jool-e-uhs See-zer.

Finally, there is classical Latin pronunciation. This is our best guess at how Latin would have sounded in the city of Rome during the first centuries (both of them) or so. This did not come down to us through a living tradition – you can see what a mess the living traditions are – but instead had to be reconstructed. We use a mix of methods to parse this out: poetry gives us a window into correct vowel length, while phonetic values can be revealed through things like how Latin was transliterated into other languages (e.g. Latin ‘c’ always becoming Greek kappa, never sigma), the misspellings Romans frequently made (which signal which letters might be confused for each other) and comments by ancient authors (especially grammarians) on proper pronunciation. While some uncertainty must always remain we are pretty confident that classical Latin pronunciation is very close approximation to how Latin was actually spoken in antiquity, in Rome and Latium. Of course there must have been (and in many cases we can know there were) a mountain of regional accents in antiquity.

I have to admit, I really admire classical Latin pronunciation’s directness and simplicity. There are no silent letters (also true in ecclesiastical Latin), the consonants all have one-to-one letter-to-sound relationships (so long as we’re not putting two letters together) and word stress (‘accent’) follows really nice predictable rules (unlike Greek). I find this handout guide, designed for use with Wheelock’s Latin remarkably handy for understanding the rules.

One tricky quirk for Classical Latin is vowel length: vowels can be long or short and vary their pronunciation accordingly. In practice, learning the language, you just have to memorize these lengths. While textbooks and other student texts will dutifully print long-marks over long vowels, Latin was not written with these marks in antiquity and most academic texts do not include them. This can be tricky because vowel length can be grammatically important: amīca is in the nominative (the subject of a sentence), but amīcā is in the ablative and doing all sorts of fun ablative things. But in practice in most Latin texts, you are going to see amica (as a Roman, reading, would) and have to figure out from context if that final a is long or short and thus what case the word is in.

On top of this, of course, classical Latin has no ‘j’ but instead ‘i’ can be either a vowel or a consonant (thus Iulius, not Julius), where it has a ‘y’ sound like ‘yikes.’ Meanwhile, the Romans often merged ‘v’ and ‘u,’ writing them as the same letter (‘v,’ thus VNVS, ‘one’), where the consonantal form (by tradition written as ‘v’ in modern texts) gets a ‘w’ sound. Once you realize that ‘v’ has the phonetic value of ‘w’ (double-u) the fact that it shares a letter with ‘u’ actually makes a lot of sense, they are similar sounds.

In any case if you take Latin today, you will be taught classical pronunciation.

Returning to our paradigm phrase stāre dēcīsīs – back with our long marks because we need them – the long ‘a’ reads as in father, the short ‘e’ as in pet, the long ‘e’ as in ‘they’ and the two long ‘i’s as the i in machine. The ‘c’ is hard and the ‘r’ should be lightly trilled, giving us stah-reh dey-kee-sees. Gaius Julius Caesar is rendered as Geye-uhs Yoo-lee-uhs Keye-sar.

So there you have it. In Latin pronunciation, the oldest system is also the youngest and we use it in scholarly contexts, except when we don’t in order to avoid sounding silly.

On to Recommendations:

I want to start with this remarkable piece by fellow classicist Clifford Ando on the financial and quite frankly moral crisis at the University of Chicago, which just announced the pause – and likely closure – of basically all of its graduate language programs. As Ando notes, the sins of U-Chicago may be unusual in scale but they are not rare in presence: the basic pattern of universities making massive, largely speculative investments in expensive (mostly STEM) fields that put otherwise economically stable institutions at risk has shown up again and again as higher education crumbles. The current administration’s decision to slash medical and science grants has turned this ulcer into a bleeding wound, but the practice was not viable in the long-term even before last year. Yet, driven by an ideological bias that favors fields with direct and narrow ‘job application’ – even when those fields now have worse job outcomes – universities continue to double down into this error, pillaging humanities departments that, because they teach more cheaply, run at a ‘profit’ for the university, in order to invest more heavily into business and STEM, which they can no longer afford.

Unfortunately I see no signs this self-destructive ideological behavior will stop. Instead if anything, as a recent, remarkably foolish, short-sighted piece in The Dispatch suggests, critics of the university will continue to double-down on a STEM-first, business-and-law-second, humanities-never strategy even as the current administration burns down the last thin supports holding up even the empty suggestion of its plausibility. It is going to be a dark few years in academia, for everyone.

Meanwhile, there is a new Pasts Imperfect (note they have moved off of substack), the main essay of which (by Lauren Donovan Ginsberg) tackles exactly this theme. Ginsberg notes that humanities departments are regularly bring in more than they cost, despite the absence of grant funds, because we’re relatively cheap to run and teach more per year, while at the same time humanities degrees remain in substantially higher demand on the job market than conventional wisdom would suggest, often outpacing students with degrees in ‘professional’ subjects more directly tied to a specific career path. As always, Pasts Imperfect, now that it is back from hiatus, is a great regular read to keep track of what is going on in the world of Classics.

Meanwhile on the historical, enthusiast side, I should note that Tod Todeschini is running a kickstarter fundraiser in an effort to fund another edition of testing medieval bows and arrows against medieval armor to suss out the limits of protection and penetration. Tod’s first two videos (and the supplemental videos he did around them) are honestly some of the most rigorous research work on this question I have seen, rivaling and in some cases decidedly surpassing scholarly efforts over the past several decades. Right now the project looks to be fully funded, but there are stretch goals. Students often ask why more of this sort of research isn’t done and the answer is generally “the funding doesn’t exist” – at least, not for this kind of experimental work in the humanities (which get a tiny fraction of the research funding of other fields). So if you want to see more rigorous testing of weapons and arrows banging off of or through medieval armor, now is your chance.

And for this week’s book recommendation, I am going to recommend E. Alyagon, Inked: Tattooed Soldiers and the Song Empire’s Penal Military Complex (2023), which in turn was recommended to me by James Palmer and is well worth the read. The book’s ‘hook’ is the remarkable Song dynasty practice of tattooing their soldiers (typically with unit assignments) as a means of exerting control, but Inked goes beyond just the question of tattoos to present a fairly complete vision of the Song military system from the soldier’s eye view. The book covers the emergence of the Song military system and then follows it from recruitment, through training and unit assignment, looks at the families and social status of soldiers and then continues all the way to discharge or revolt. The reader thus gets a vision of the whole system from an organizational and institutional culture perspective (though not arms or tactics) and that overview is delivered remarkably well. The book is remarkably well written and very readable, even for the lay reader (as indeed, when it comes to Chinese history, I am effectively the lay reader).

Alyagon notes in the introduction on older strain in the historical reception of the Song dynasty in which Song soldiers took the blame for the decline and eventual destruction of the Song dynasty (at the hands of the Mongols), despite the Song dynasty itself having a reputation as a period of technological advancement and renaissance, overseen by Confucian scholar-officials. That judgement on Song dynasty soldiers themselves is almost certainly unfair, but one thing that struck me reading Alyagon is that the Song military system does not beat the rap. Emerging out of a desire by the early Song emperors and their bureaucracy administration to bring the army under firm control, the Song military system in Alyagon’s telling emerges as an aggressively dehumanizing institution. Military service was often a legal penalty (and so the status of prisoner-conscripts very low) but at the same time men were also conscripted from the civilian populace or dragooned from soldiers’ families to serve. They were tattooed in an effort to prevent desertion (which created problems for men who served honorably and were discharged – discharge papers could be lost, tattoos were forever) and sorted and assigned based on bureaucratic metrics with few connections to actual performance.

One of the most striking practices that comes up frequently was ‘rotation’ – like the Romans, the Song recognized leaving a commander in charge of a specific army on the frontier long-term was dangerous. But rather than doing the Roman thing and rotating commanders, the Song rotated soldiers, marching their common soldiery long distances, often in difficult conditions away from their home to facilitate rotating armies. It is such a staggeringly inefficient, brutal system of control that, to me, it served well as a summary of the entire system of intense top-down control where every aspect was engineered to suit the whims of the ‘enlightened’ scholar-officials running the system, quite regardless of military effectiveness or basic humanity. Reading, I found I was no longer surprised that, once the power of the Song began to break, these armies swiftly collapsed or switched sides.

Nevertheless, Alyagon’s book is a valuable read for anyone interested in military history, because it is such a complete portrait of the institutional culture of a military system, useful both in itself but also as a comparison point to think about other military systems. I found in reading that I was struck by how the Song system felt like a mirror image of the early Roman imperial military system, but with the status of the soldiers being much lower and as a result the balance of coercion and incentive was far more tilted towards coercion. It is, of course, striking then that the Roman imperial military system, when it faced fragmentation in the context of intense security challenges in its third century survived, whereas the Song system, when it met the same challenge (at about the same point in its ‘life cycle’) collapsed. I did not find myself, by the end of the book, mystified by the difference in performance.

1. ‘ee’ as in the ‘i’ in machine, but ‘i’ as in the ‘i’ in pin.
2. The first ‘u’ as the u in rude, the second ‘u’ as the u in put
3. Attentive Latin students will note this obliterates an important pronunciation distinction in the fourth declension between the nominative singular and the genitive, so that gradus and gradūs would both sound the same – grah-doos – whereas in classical Latin you’d have the slight difference between grah-duhs and grah-doos.