Wolfgang de Melo
When reconstructing the pronunciation of Latin, Classicists have predominately focused on the speech of the educated elite in the first century BC. Traditionally, scholars have asked themselves what Cicero and Caesar, Virgil and Ovid sounded like, not the man who sold scriblitae, a type of pastry, on the streets of Pompeii, or the midwife living in an insula, a block of apartments, in the poorer parts of Rome. Thankfully, these days we take a broader view and examine variation in speech. Even though our evidence for non-elite varieties is often more limited, there is a reasonable amount of material that can teach us how local accents differed from the dialect of Rome and what second-language learners sounded like. But this piece is not about the reconstruction of pronunciation per se, although we shall occasionally touch on that topic; rather, this piece is about ancient attitudes to different kinds of accents. What did speakers of Latin think of those who spoke a local dialect, or of those who had a foreign accent? How similar were their attitudes to modern ones? In what follows, I use ʻdialect’ for variation in words and in pronunciation, but ʻaccent’ for variation that is restricted to the latter.
As it is instructive to compare ancient views with current ones, I want to start my piece with my own observations of the highly diverse opinions on accent variation held by native speakers of British English. I could make similar observations for speakers of German or Flemish, but my awareness of such issues was sharpened when I moved to Britain as a graduate student to devote myself fully to the study of the noble art of linguistics; nothing makes you notice people’s attitudes as clearly as the combination of being a new arrival in a country and being given the powerful tools of linguistic analysis. Next, I will turn to the question why people have local or foreign accents in the first place; how do they come about, and why do they typically persist? Once we have this background information, we will look at how ancient authors felt about local and foreign accents and compare their attitudes with modern ones. I want to end this piece with some comments on what today’s students of Latin should focus on when learning to pronounce Latin.
Attitudes to local and non-native accents: an impressionistic overview
One statement that I have come across surprisingly often is that the British Isles are losing their dialects rapidly. The advent of the radio and the television are said to be the main culprits. Standard British English used to be called Public School Pronunciation; despite being based on the usage of the well-off classes of the southeast, it became a supra-regional variety, but as its name indicates, it was socially restricted to those who acquired it in ʻpublic’ schools (a confusing name for those outside Britain, since these are private schools). Public School Pronunciation was renamed Received Pronunciation (RP) and is now often called BBC English, even though these days not all BBC presenters actually speak BBC English. However, the name is still apt because the BBC did give access to more prestigious types of pronunciation to ever larger groups of people.
So just how rapidly are dialects disappearing? In the First World War, Wilhelm Doegen, a German sound pioneer, made extensive recordings of British prisoners of war. These are now kept in the British Library in London, and listening to them is an amazing experience: the sheer variety of accents and dialects is astounding. Much of this variety is indeed gone. However, it may come as a surprise that it has not been replaced by BBC English. Rather, while a few generations ago a dedicated dialectologist could have pinned down precisely which village near Birmingham or Manchester or Liverpool a speaker came from, such a dialectologist would not be able to state the precise village any longer, but could still classify according to the great cities nearby. Variation persists, but the local kind has given way to the supra-regional.
To the linguist, each dialect and each accent is equally valuable. There is no better or worse. But most people do make value judgments. To a speaker of Received Pronunciation, not all regional British accents are equally easy to understand. Such speakers may in fact find many American varieties, such as those spoken on the East Coast, easier to follow than, say, the dialects of Glasgow or Liverpool. But the value judgments attached to British dialects are not exclusively based on mutual intelligibility; rather, they have at least as much to do with class associations. This is why those in public life sometimes modify their pronunciations. Margaret Thatcher, Prime Minister from 1979 to 1990, came from a humble background and famously took elocution lessons, not because her variety of English, Lincolnshire, would have been hard to understand, but because it did not sound prestigious enough. By contrast, Tony Blair, Prime Minister from 1997 to 2007, came from a highly privileged background and was privately educated; during my student days, he was widely mocked on the radio for affecting Estuary English in a somewhat desperate attempt to come across as a man of the people.
People get judged and mocked for their accents on a daily basis. However, what is interesting here is that there has been a remarkable shift. Accentism and mockery directed against foreigners with a different first language has become rather uncommon; it is simply no longer politically correct to mock foreign accents. But prejudices against and mockery of native speakers of English with regional accents that are not considered posh are still extremely widespread. The general attitude seems to be that foreigners can’t help their accents, while native English speakers should know better. Interestingly, I have also observed highly educated foreigners, with mild foreign accents, ridiculing the local accents of native English speakers. It is harder to disentangle what exactly is going on in such cases; it may be an attempt to show off one’s own linguistic competence, coupled with deep-seated insecurities.
If these attitudes towards different accents show prejudice and bias, then surely the recent diversity of accents in advertising must be proof that attitudes are shifting. But must it really? Advertisements are meant to sell stuff, and if virtue signalling about diversity is doing the trick, then advertisers are going to do just that. But in this case we are not even dealing with virtue signalling. If you match accents and goods, you will see very clear trends: a refined Scottish accent (typically Edinburgh rather than Glasgow) is used when people are trying to persuade you to trust them with your money or personal safety, as with banking or flights; agricultural products sell better when the salesperson has a West Country accent; and Essex accents are the preferred option when people are trying to convince you to part with money for gambling, alcohol and ʻmatey’ stuff. Foreign accents are used in similar ways: artisan food adverts feature French and Italian accents, and for perfumes it is most commonly French.
Onomatopoeia, playing with sound effects and imitating the sounds of nature, is more common than one might think; when imitating children’s speech, many people will use words with i-sounds, or will shift their vowels a little in that direction, and this makes sense: children’s voices are higher-pitched than adult voices, and high front vowels like /i/ have an intrinsically higher pitch, so that this is subconscious imitation. But the choice of different accents in advertising has fairly little to do with onomatopoeia and the actual sounds of these varieties. No, what is really going on is that advertisers play with our prejudices: the reliable Scotsman, the West Country farmer, the fun-loving youngster from Essex; and in the case of foreigners, the Italian proud of native culinary traditions, or the refined Frenchman. From a purely acoustic perspective, a French accent should not be sexier than an English or German one; if it is perceived as such, it is simply because French culture is considered seductive and alluring, and this stereotype is then transferred onto the accent. Rather than tearing down prejudices, our clever advertisers capitalize on them and reinforce them. On this note, we can turn to the science behind accents.
The science behind accents: why do people have them in the first place?
But why do people have accents to begin with? Many foreign accents tend to be perceived as lazy, as if second-language learners did not want to make the effort to ʻspeak properly’. And, since class and accent go hand in hand in Britain, some local native accents are perceived as uneducated. Thus there is a big incentive for people who want to succeed in their careers to adjust their pronunciations to the accent that is BBC English. Why are they so often unsuccessful?
First-language acquisition follows a number of universal patterns. Babies and toddlers begin with consonant-vowel sequences, and these consonants tend to be stops, p, t, k, and vowels like a. All languages have at least some of these sounds as well as the most basic syllable template, consonant + vowel. Intonational contours are also acquired early. Other sounds are always acquired relatively late: Czech ř, a raised alveolar non-sonorant trill in phonetic terms (roughly “rzh”), is one such sound, which Czech children usually master last, when they are between four and eight years old. However, even though the articulation of such sounds may come late, babies already begin to understand which sounds matter in their parents’ language and which ones do not; and as soon as they have realized that, they stop paying equal attention to all sounds and focus on those which are contrastive in the language or languages they hear.
There is a window for first-language acquisition which ends when children reach a certain stage of development; for some, this can be as early as eight years old, for others, as late as fifteen or sixteen. After this age, language acquisition normally requires more structured input to be successful. In terms of pronunciation, habits have set in: we can still hear all the sounds of any language, but our brains tend to filter out what is not distinctive in our native languages; and our brains drive the muscles responsible for articulation in ways that take our native patterns as the default.
Second- (and third- and …) language learners, then, have to learn to actively listen to new contrasts and to overcome ingrained patterns of articulation. This requires much more effort than one might think. We can actually see this in images of the brain produced in electro-encephalograms (EEG, the brain equivalent of the ECG). Here, the difference between a first language and a second one shows up very clearly: for the first language, relatively little of the brain lights up, and processing and production is highly efficient, but for a second language, there is a veritable explosion of lights. Second-language learners can still acquire a very high degree of proficiency in pronunciation, and some particularly talented ones may even become indistinguishable from a native, but the time and effort required is considerable, and many learners will not have the leisure or indeed the need to reach these highest levels.
You can perhaps compare the situation of learning a foreign language with learning a musical instrument. If you learn it at a young age and you are talented, you may become world class. If you learn it in adulthood, your progress will be slower not only because you have less time to dedicate to it, but also because your brain is more averse to acquiring and maintaining the neural pathways for new motor skills. This may seem unfair, and also strange from an evolutionary perspective: would it not be advantageous to retain the skill to acquire new languages and motor skills with ease and speed? Well, yes and no. The advantages are obvious, but there are also problems. Filtering out variations in sound that are not contrastive in our native languages and having routine articulations set in our neural pathways saves a great deal of brain energy. That is why things are the way they are, unsatisfactory though that may be for us language lovers.
Acquiring pronunciation skills in a second language is invariably harder than doing so for a different dialect. The former requires a great deal of learning, unlearning, relearning; the latter often requires no more than some minor adjustments of specific features. Still, changing one’s native accent can also be a challenge; and often it may not be desirable, because one’s local accent is often tied to one’s identity. Why should speakers of the West Country dialect, which is very easy to understand for speakers of BBC English, change pronunciation? Why shouldn’t the speakers of BBC English give up their own prejudices instead?
Since we are on the subject of making small adjustments to one’s native pronunciation, one common preconception needs to be addressed. Often, one hears complete monolinguals say that they would be excellent language learners and would acquire any language without a foreign accent; this is based on the fact that they are pretty good at imitating foreign accents or dialects of their native language. But such an idea is not based on reality. Such imitations require minor adjustments. Learning the pronunciation of a foreign language means so much more than acquiring a few sounds and peppering one’s speech with them; these sounds need to be used in the correct positions and correct contrasts. Time to return to antiquity.[1]
Attitudes to accents in antiquity
Regional variety gives way to supra-regional dialects
All the attitudes towards accents that I mentioned for Britain can be paralleled by ancient ones. Roman writers do not complain, it is true, that local dialects are disappearing, but their comments are testimony to the disappearance of small, regional varieties and to the subsequent appearance of supra-regional ones. Plautus (died 184 BC) makes fun of the Latin of Praeneste. In Truc. 687–91, the uncouth slave Truculentus goes to see a prostitute and offers her a rabo. When asked why he doesn’t say arrabo, “part payment”, he responds:
A facio lucri,
ut Praenestinis conea est ciconia.
I’m saving the A, just as a ciconia (“stork”) is a conea to the inhabitants of Praeneste.
Did the inhabitants of Praeneste really say conea? Unsurprisingly, the word is not attested in inscriptions, so that we cannot know if they had lost the unstressed first syllable. But if our manuscripts transmit reliable spellings here (a big if, admittedly), then Plautus would also be commenting on the ending -ea instead of -ia. And indeed, such spellings are common in Praeneste. Praeneste is the modern Palestrina; what is remarkable to me is that a variety of Latin was mocked that was spoken only 22 miles east of Rome!
When Republican writers comment on dialectal variation, whether it is in pronunciation or in specific words, they focus on small towns in close proximity to Rome. But this changes during the Empire. Now people notice differences between the Latin of Italy and individual provinces. Some more localized variation is likely to have persisted for at least some time, but it goes under the radar because supra-regional variation overshadows it.
For example, St Augustine (AD 354–430), who hailed from North Africa and had grown up speaking the local variety of Latin, compares his own speech with that of the Itali ʻItalians’ rather than that of the Romans (De ordine 2.17.45). And Pliny the Younger (Epp. 9.23.2) relates an interesting anecdote by Tacitus (AD 56–120). Modern scholars do not know for certain where Tacitus was born (though Gallia Narbonensis is the most likely place), and neither did a Roman knight, who could not place Tacitus’ accent and asked him whether he was Italicus, “Italian”, or provincialis, “from the provinces”. Tacitus replied that he knew him ex studiis (“from their studies”), and the knight promptly proceeded to ask him which one he was, Tacitus or Pliny! Dialects and accents persisted, but they had become supra-regional.
Sociolects: aspiration and au/o
A dialect is a regional variety. But varieties can also be determined by social class rather than locality, and in such cases we speak of sociolects. Sociolects are often marked by features of pronunciation. In the first century BC, the Roman upper classes were trained in Greek, and when they used Greek loanwords, they would pronounce and inflect them according to Greek conventions. Greek made a threefold contrast between stops: β was fully voiced (b), π was voiceless but unaspirated (p), and φ was voiceless and aspirated (ph). Similar contrasts can be found in many languages across the globe, for instance in Thai, but in modern European languages they are unusual, so a brief explanation may be helpful.
A stop consonant is one in which there is a complete closure for a brief moment and no air escapes through the mouth or nose during that time. If you say reader, during the d, your tongue touches the alveolar ridge just behind your upper front teeth, and this constitutes the stop closure, which is released for the following vowel. We speak of voicing when the vocal folds in the voice box are vibrating. If you say mmmm or zzzz and you put thumb and index finger on your voice box, you should feel those vibrations. In a voiced stop, the vocal folds are vibrating throughout the closure. In a voiceless, unaspirated stop, they don’t vibrate during the closure, but resume vibrating immediately afterwards. And in a voiceless, aspirated stop, the vocal folds don’t vibrate during the closure and don’t resume vibrating immediately after the closure. The beginning of the following vowel is thus voiceless, and this voiceless vowel is h or aspiration.
Now if your first language is Italian or Spanish or French, or Polish or Czech or Russian, you make a contrast between fully voiced stops and voiceless unaspirated ones there. In English, however, the voiceless stops are aspirated, and the voiced ones may not be fully voiced. But there is one context in which English stops are voiceless and unaspirated: after s. You can try this out: can you hear the difference in the p in pair (aspirated) and spare (unaspirated)? If you find it hard to hear, I can show you how you can see it. Take a sheet of paper and hold it up in front of your mouth; just hold it by your fingertips so that it is hanging down loosely. Now say pair: the aspiration should make the paper shake a little. And now try spare: the paper should remain still.
Latin stops were like those of the Romance languages; the contrast was between fully voiced stops and voiceless unaspirated ones. Greek had those stops, too, but there were also the aspirated ones, and these constituted a difficulty for Latin-speaking learners of Greek. Those who wanted to sound educated would have to learn to aspirate, and to aspirate in the right places. Catullus, in Poem 84, mocks a certain Arrius, who knew how to aspirate, but not where:
Chommoda dicebat, si quando commoda uellet
dicere, et insidias Arrius hinsidias.
Arrius used to say chommoda when he wanted to say commoda (“advantages”), and hinsidiae when he intended insidiae (“ambush”).
Finally, Arrius goes to Syria, and people’s ears get a rest, until news arrives:
…affertur nuntius horribilis,
Ionios fluctus, postquam illuc Arrius isset,
iam non Ionios esse, sed Hionios.
…the horrifying message is brought that the Ionian waves, after Arrius had gone there, were now no longer Ionian, but Hionian.
Arrius was a man who could pronounce aspiration, but clearly did not know where to do so, and so he hypercorrected and used this feature where it was not happropriate.
Hypercorrection of this sort had some medium-term effects in Latin: certain Latin words began to be pronounced with aspiration, probably because they were considered to come from Greek in some form, but the Romance languages preserve no traces of this marginal phenomenon. Cicero mentions it (Orat. 160): he tells us that he used to say pulcer, “beautiful”, Cetegus (personal name), triumpus, “triumph”, and Cartago, “Carthage”, and rightly so, but that later he gave in to common practice and said pulcher, Cethegus, triumphus and Carthago. The first two are Latin in origin; triump(h)us is ultimately from Greek θρίαμβος (thriambos), but made it into Latin through Etruscan; and Carthago is derived from Punic, but with major changes. This is quite a mix of words, and rather than assuming that the Romans aspirated because Punic qart, “city”, had an aspirated final consonant, we should think that the modern Latin forms are the result of hypercorrection.
Another sociolectal distinction in Cicero’s time concerned au and long o. Old au had become a monophthong in some lower-class varieties in Rome, but not yet in the speech of the upper-classes. Those who wanted to sound posh would try to keep au and long o distinct, but there were cases of hypercorrection. “To clap” is plaudere or plodere, and one’s first intuition might be that plodere is a vulgar form. However, that is not true, because the compound is explodere. In words with an old diphthong au, this au changes to long u word-internally because of a sound change called vowel weakening, hence causa, “reason”, and excusare, “to make an excuse”. Since we do not find **expludere, we know for certain that plodere is the original form and that plaudere is hypercorrect.
Trying a little too hard to be cool
Cicero and Arrius shared one trait: both tried very hard to speak like the upper classes. Arrius was not particularly successful, but Cicero was. As a homo novus, a “new man”, without a family background in high politics, he had to do his best to fit in. People like that often become linguistic pedants, sticklers for correctness, and take great pride in speaking the way one ought to. It is not unreasonable to compare Cicero with Margaret Thatcher in this regard. But can we also find ancient Tony Blairs? Posh people who try a little too hard to be cool in their language?
The phenomenon is certainly not restricted to our times. Cicero’s contemporary Publius Clodius Pulcher was a member of the gens Claudia, one of the most ancient and noble families in Rome.[2] Yet in his efforts to present himself as a man of the people, he even changed his name from Claudius to Clodius. Various emperors also affected common speech. Suetonius (Vesp. 22) reports the anecdote that Vespasian used to say plostra ʻwaggon’ instead of plaustra; when corrected by the grammarian Mestrius Florus, he jokingly called him Flaurus, with a hypercorrection alluding to φλαῦρος, “shabby”. It is unlikely that such high-ranking figures did not care about the way they spoke; rather, these are deliberate attempts to show how down to earth the speakers are. In a way, this kind of deliberately casual speech is also a display of power: upstarts like Arrius and Cicero have to prove that they fit in; but members of the traditional elite want to show that they can afford not to care.
City speech, country speech
Cicero praises the sounds of the city, of Roman speech (Brut. 171):
id tu, Brute, iam intelliges, cum in Galliam veneris. audies tum quidem etiam verba quaedam non trita Romae, sed haec mutari dediscique possunt; illud est maius, quod in vocibus nostrorum oratorum retinnit quiddam et resonat urbanius. nec hoc in oratoribus modo apparet, sed etiam in ceteris.
You will understand that immediately, Brutus, when you have arrived in Gaul. Then you will hear certain words not current in Rome, but these can be changed and unlearned; what is more important is that there is a certain ring and more refined sound in the voices of our orators. And this does not only appear in the orators, but also in the others.
In other passages, too, Cicero emphasizes the purity of the Roman accent. Of course Cicero himself came from Arpinum, a little outside, and he was mocked as not being a proper Roman by his opponents (Sall. Cat. 31.7). But nowhere is he mocked for a non-urban accent. Clearly, Cicero had obliterated any such traces. As a social newcomer, he strove for the Latin equivalent of RP.
But just as some British accents other than RP can have positive connotations, like reliability and trustworthiness in the case of Edinburgh, so non-Roman accents could also be perceived as positive. In Brut. 259, Cicero describes Cotta as “broadening his letters” (dilatandis litteris). From other pieces of evidence (De orat. 3.46, Varro Rust. 1.2.14), we can deduce that this refers to a variant pronunciation of the long vowel i going back to the earlier diphthong ei. In the speech of Rome, ei yielded long i, but in the countryside, the corresponding outcome was long e. This affected words like spica (Rome) / speca (countryside), meaning “ear of corn”. Cicero (De orat. 3.42) tells us very clearly that for Cotta, this pronunciation was a deliberate affectation, an imitation of country speech, because it was considered more archaic; and for many Romans, more archaic meant better, purer, in accordance with the customs of the ancestors.
Ancient accentism
When Cicero talks about Cotta’s affectation of a country accent, he is quite restrained because Cotta is a friend. But it is obvious that he thought that this artificial accent was a bit much. Elsewhere, Cicero felt entitled to mock a foreign accent. I would assume that this was not a regular habit of his, but if he thought it would help his cause, he had no hesitation to do so: Quintilian (Inst. 1.1.14) tells us that when he was defending Fundanius in court, a Greek witness on the opposing side struggled with Latin pronunciation and could not pronounce f- in the name Fundanius, for which he substituted h-. At the time, Greek φ was still pronounced as an aspirated ph, not yet as f, and the Greek sound closest to Latin f- was h-.[3] Such sound substitutions are common: think of some German or French speakers who struggle with English th- in thing or this and who substitute s- or z-. Quintilian tells us that Cicero ʻmocked’ or ʻlaughed at’ this witness (irridet). In today’s courts, such behaviour would not be considered acceptable, but in Cicero’s time it helped his cause. And even today it is not unusual for British undergraduates to put on a fake German or French accent to poke fun at others.
But such is human nature. Earlier, I mentioned a rather less common phenomenon that I have witnessed as well, namely highly educated foreigners with very mild accents mocking native English speakers for their regional accents or lower-class sociolects. This, too, happened in antiquity. Gellius (19.9) has an illuminating passage. At a dinner party, presumably in Rome, Antonius Julianus made an appearance. He was a teacher of rhetoric and Hispano ore, “of Hispanic speech”; at this point in time any Spaniard would have been a native speaker of Latin, with the kind of supra-regional accent that I discussed earlier. A group of Greek scholars, highly educated in Latin literature, but presumably not native speakers of the language, started to mock him as barbarous and agrestis, “rustic”, and as a clamator, “shouter”. I would suspect that this group of Greeks wanted to show off their Latin competence, but also felt somewhat insecure about their own accents.
Describing accents in antiquity
If you’re not trained in linguistics, it is really difficult to describe an accent. A French accent, for example, could be marked by lack of aspiration, by not making clear vowel length distinctions, by pronouncing r as a uvular fricative, and by stressing final syllables. Yet those not trained in linguistics may well hear all of these features very clearly, and they may perhaps even be able to imitate them, but they will still struggle to describe what exactly is going on. Ancient authors had similar problems. They could easily say that someone sounded different, but how would they describe such differences?
In non-technical contexts, descriptions are incredibly vague and unhelpful. What did Antonius Julianus’ Hispanum os actually sound like? Presumably Gellius’ readers would know what a Spanish accent was and could imagine his way of speaking. Cicero tells us (Arch. 26) that young Archias lent his ears etiam Cordubae natis poetis pingue quiddam sonantibus atque peregrinum ʻeven to poets born in Corduba, sounding somewhat fat and foreign’. An interesting description, but in practical terms entirely useless.
Of course a court speech is not the ideal environment to talk about phonetic details, but even the ancient language professionals are not overly helpful. Cicero, in his treatises on oratory, does mention issues of aspiration and the like. Varro (Ling. 5.97) tells us that where Romans say haedus, “goat”, people from outside say hedus. And later grammarians treat divergent accents under the vitia orationis, “faults of speech”; in the 5th century, Pompeius, himself from North Africa, says that it is a vice of the Africans to use only the ʻfat’ l and not the clear one (15.6–8).[4] Judgmental attitudes are the norm, but if we combine such statements with inscriptional data, we can get a little closer to what local accents sounded like.
Confidence and anxiety
When people respect or mock specific accents, and when they describe some as correct and others as having vitia ʻfaults’, this can lead to confidence or anxiety in people whose speech deviates from what is deemed the norm. We can observe such anxieties in the emperor Lucius Septimius Severus, born in 145 in Leptis Magna, a city located in present-day Libya. His father was Phoenician, speaking Punic, a Semitic language closely related to Hebrew, while his mother was Italian, and when he became ruler in 193, he made history by being the first provincial to occupy this position: provincial not just by place of birth, but also by his paternal ancestry. Septimius Severus had native fluency in Punic, but he was also educated in Latin and Greek. His sister, on the other hand, had not received the same degree of education, and when she came to visit her brother, the emperor, in Rome, her lack of language skills caused him severe embarrassment, as the Historia Augusta reports (Sept. Sev. 15.7): cum soror sua Leptitana ad eum venisset vix Latine loquens ac de illa multum imperator erubesceret … redire mulierem in patriam praecepit (“when his sister from Leptis had come to him, barely speaking any Latin, and the emperor was deeply embarrassed about her… he ordered the woman to return to her country”). We don’t know much about what Septimius Severus sounded like himself, but for an emperor it must have been awkward to have a sister around who could not communicate in the language of the ruling classes.
More directly related to accent rather than general linguistic competence are some telling remarks by St Augustine. Unlike Septimius Severus, Augustine merely had some kind of passive knowledge of Punic; Latin was his first and dominant language, but of course he had the local accent. In a remarkable passage alluded to earlier (De ordine 2.17.45), he says to his mother Monica:
si enim dicam te facile ad eum sermonem perventuram, qui locutionis et linguae vitio careat, profecto mentiar. me enim ipsum, cui magna necessitas fuit ista perdiscere, adhuc in multis verborum sonis Itali exagitant et a me vicissim, quod ad ipsum sonum attinet, reprehenduntur. aliud est enim esse arte, aliud gente securum.
For if I said that you would easily attain a state of language which is free from fault in expression and speech, I would indeed be lying. For I myself had a great need to learn these things thoroughly, and yet the Italians still rebuke me on the issue of many sounds in words, and they are in turn criticized by me as far as sound itself is concerned. For it is one thing to be confident in one’s training, and another to be so in one’s birth.
Here we can see Augustine admitting that he is insecure about his pronunciation; not because he has a non-native accent, but because he has a native, but provincial, accent. What I find charming is that at the same time he realizes just how arbitrary such value judgments about native accents are, he has enough self-confidence to criticize the Italians in turn!
So far, every ancient phenomenon could be paralleled with a modern one that I have encountered myself. But there is one for which I have no corresponding experience. The poet Martial (born between AD 38 and 41, died between 102 and 104) was a native Latin speaker from Spain, as was Marcella, probably his patroness. In one poem (12.21), he praises her exquisite way of speaking:
Municipem rigidi quis te, Marcella, Salonis
et genitam nostris quis putet esse locis?
tam rarum, tam dulce sapis. Palatia dicent,
audierint si te vel semel, esse suam.
nulla nec in media certabit nata Subura
nec Capitolini collis alumna tibi;
nec cito prodibit peregrini gloria partus,
Romanam deceat quam magis esse nurum.
tu desiderium dominae mihi mitius urbis
esse iubes: Romam tu mihi sola facis.
Who would think that you, Marcella, are a native of stiff Salo and born in the same place as me? So exceptional and sweet is your taste. The Palatine will say, if it hears you just once, that you belong there. No woman born in the middle of the Subura, no nursling of the Capitoline Hill will compete with you. Nor will there appear a glorious foreign offspring quickly who would be more fit to be a Roman bride. You bid my desire for my mistress the city be milder: you alone bring Rome before me.
This sentiment is entirely alien to us. We have all heard non-native speakers being praised for their excellent pronunciation; it is a great compliment for second-language learners to be told that they sound native. However, it is unimaginable today for an American or an Australian to be complimented on their beautiful British pronunciation. It would not just be a little off, it would be downright offensive, as if their native varieties were second-rate types of English. In my two decades in Oxford, I have indeed encountered a handful of Americans who tried, quite successfully, to acquire an RP accent; there were slight traces of their native pronunciation, but by and large they could pass. Such deliberate accent shifts are normally the result of some deep-seated insecurities; and those of us who knew that they had obliterated previous pronunciation habits would never mention this fact in their presence. A compliment in this respect would have felt entirely inappropriate. And yet Martial does not seem to find it problematic to praise a fellow Spaniard for sounding like a Roman from the most refined part of the city. Let us move back to the present.
Back to our day: some thoughts on learning Latin pronunciation
Now that we have seen the variety of attitudes towards accents in antiquity, it is inevitable that two particular questions should come up: what would the Romans make of our attempts to pronounce their tongue? And how accurately can our pronunciations reflect theirs?
To begin with the latter issue, there is a great deal that we can achieve with sufficient talent and practice. We can make all the phonemic distinctions and differentiate between minimal pairs like malum, “bad”, and mālum, “apple” (similar to hut and heart in RP, where r is not pronounced before consonant). There is also a surprising amount of phonetic detail that we can recover: we know that most voiceless stops were not aspirated, unlike in English; that l had allophones similar to the ones in RP; and that r was a tap or trill as in Italian or Spanish (tap in Spanish pero, “but”, trill in perro, “dog”).
With everything we know, we would probably be quite easy to understand; but we would probably not sound natural to a native speaker of Latin. There are many details we cannot recover. Here is an English minimal pair, identical in every phoneme, but still rhythmically distinct:
(a) Take Gray to London.
(b) Take Greater London.
The diphthong is longer in Gray than it is in Greater. This is because vowels and diphthongs are longer in monosyllables than in disyllabic words. Similar contrasts below the phonemic level exist in most languages and may well have existed in Latin, too; but they are irretrievably lost.
Intonation is another area that we cannot recover. The last time I had to see the doctor, she asked me, as I entered her office, “How are you?”. The pitch contour was a fall-rise, as is normal in English ʻphatic’ questions, that is, questions which are not really asking for information, but are simply there to establish communication. After the equally stereotypical “Fine, how are you?,” she asked me again, “So, how are you?” This time the pitch contour was a fall, typical for genuine questions of this kind. Similar distinctions exist in all languages, but they are very language-specific, and for Latin, they are sadly lost.
Incidentally, even if we knew the system of intonational contrasts in Latin, its exact realization would elude us. Both English and German have a rise-fall (although it is rarer in English than in German). However, we cannot simply equate the two. A rise-fall takes some time to produce, but if the utterance is on the short side, what does one do? English speeds up the tune, a phenomenon known as compression, but German does not and simply cuts it off before it is finished, a phenomenon known as truncation. Our native phonological knowledge can play tricks on what we believe we hear: when Germans expect to hear a rise-fall, but the utterance is too short and the computer shows that only the rising part has happened, they will still believe that they have heard a rise-fall. I simply mention this as a reality check – we need to know the limits not only of reconstruction, but also of our personal abilities.
That said, what advice would I give a student wishing to learn Latin pronunciation? My most important piece of advice is to put most of one’s energy into phonemic contrasts like vowel length distinctions. When I was at school, we were expected to scan Horace’s odes and various other metres, but we had learned Latin entirely without quantities, a bizarre, but remarkably common situation. It is not only that quantities help us with pronunciation when reading Latin aloud and when speaking it freely, two skills which the current education system does not emphasize nearly enough; they also help when writing verse or prose with clausulae, two skills which the enemies of our subject have not yet managed to abolish. Unfortunately, if you have learned Latin without quantities, you will have to relearn them for every single word. When I started university and realized my shortcomings in this respect, I sat down in the Easter break and used a good dictionary to mark every long vowel in Books 1 and 7 of Caesar’s De bello Gallico; I then recorded them on cassette (yes, it was that long ago) and listened to these recordings obsessively until I knew them by heart. Not everyone has the time to do this or parents tolerant enough to put up with such weird behaviour. But however you do it, you will have to learn quantities, or you end up like the foreign student in my building who said that for his holidays he was going home “to the bitch” (= beach).
Good old Cicero had to learn to aspirate in certain words. If your first language is English or German, you will have to unlearn aspiration for almost all words. This challenge is quite different from learning quantities. English does have a phonemic distinction between long and short vowels (beach and bitch, or feet and fit for those of a more sensitive disposition); so here the issue is simply relearning which vowels are long and short in Latin, which is not a challenge of articulation, but of memory. On the other hand, learning not to aspirate voiceless stops and to produce voiced stops as fully voiced does require learning new articulation skills. But once we have got the hang of that, we merely have to make it a habit, and no memorization is required.
If you are a school teacher, there may be further complications, this time with your pupils. In English, the contrast between p and b is partly one of aspiration. If you pronounce Latin the way the Romans did, your pupils may well hear your p as a b. If I may end with an anecdote, I knew a wonderful and hugely enthusiastic teacher of Ancient Greek who had worked through Sidney Allen’s Vox Graeca (3rd ed.,1987) and was now convinced that he ought to pronounce φ, θ and χ as ph, th and kh, the way Plato and Aristotle still did, and not like in fig, thing and Loch Ness, the way speakers of Modern Greek do. That was all well and good, but he had not learned to pronounce π, τ and κ (p, t and k) without aspiration and rendered them the English way, as ph, th, kh. Huge confusion ensued, because words like φῶς, “light”, and πῶς, “how”, had now become homophones. The moral of the story is this: make sure you prize phonemic contrasts highly, even at the expense of phonetic accuracy; and once you have internalized them, work on the phonetic details.
Above all, though, don’t sweat it. The Romans and Greeks had many varied attitudes to accents, ranging from derision to respect. But if you could speak to time travellers from the future trying their very best to speak to you in your own tongue, you would probably feel touched and inspired. I find it hard to imagine that the ancients were any different.
Wolfgang de Melo is Professor of Classical Philology at Oxford. He has published on early Latin, especially Plautus and Roman comedy, and on Varro. He teaches linguistics and comparative philology and has a special interest in linguistic typology.
Further Reading
A huge amount of information on Latin accents can be found in two books by the late J.N. Adams: Bilingualism and the Latin Language (Cambridge UP, 2003) and The Regional Diversification of Latin 200 BC – AD 600 (Cambridge UP, 2007). Although both books are primarily concerned with linguistic features rather than attitudes towards them, this latter theme is also very prominent.
Those interested in English pronunciation, both RP and other varieties, should look at A.C. Gimson and A. Cruttenden, Gimson’s Pronunciation of English (8th ed., Routledge, London, 2014). An earlier edition of this work is the first linguistics book I ever read – it certainly got me hooked! P. Trudgill, The Dialects of England (2nd ed., Blackwell, Oxford, 1999) is also a fascinating read.
Wilhelm Doegen’s recordings can be accessed online here. And finally, here is a short clip on accentism, an interview with Jane Stuart-Smith, professor of phonetics at Glasgow.
Notes
| ⇧1 | We may also mention here what is sometimes called ʻforeign accent syndrome’, when a person has a stroke and suddenly speaks with a ʻFrench’ or ʻRussian’ accent without ever having learned the language. Such stories make it into the newspapers, but they never stand up to scrutiny. To the untutored ear, such a person may indeed sound French or Russian, but that is based on a few superficial features. The stroke has affected the parts of the brain in charge of motor control, and some features of pronunciation have changed; this may result in a superficial similarity to a particular language that has one or two comparable features of pronunciation. But other features will remain distinct, and the English speaker with a new ʻFrench’ accent will not be able to pronounce French any better than anyone else. |
|---|---|
| ⇧2 | Note also the form of the third name, Pulcher with aspiration rather than Pulcer without. |
| ⇧3 | The two sounds are acoustically similar. Spanish (but not Portuguese) changes Latin f– to h– (still written, but by now lost in pronunciation); cf. facere, “to do” > hacer or faba, “bean” > haba. |
| ⇧4 | For more on this, take a look at my previous Antigone article on Latin spelling. |