Latiné loqui disce sine molestiá!
Learn to speak Latin with ease! ¡Aprende a hablar latín sin esfuerzo!
Apprenez à parler latin sans peine! Impara a parlare latino senza sforzo! Lernen Sie latein zu sprechen ohne Mühe!
Whereas any student can understand that we accurately know the writing usages of the Romans, as well as those of the users of Latin of any other later periods, it is a cause of greater confusion and even scepticism to affirm that we also know with great accuracy the pronunciation of Latin in the different stages of its long history, and in particular in classical times.
It would be too long to explain here how we have come to know the pronunciation of Latin in times for which we have no sound record; but this is a question on which many specialists have worked, and the beginner has to exercise a certain degree of trust in the teacher. We will mention here nevertheless the main instruments that allow us to know how a language like Latin was pronounced in the past.
1. In the first place we’ve got historical and comparative philology. If a Latin word like máter, for instance, has evolved into Italian and Spanish madre, Catalan mare, French mère, Portuguese mãe and in all of those languages without exception the letter m is pronounced the same way, maybe we can hesitate about the pronunciation of the other letters in the word; but we have in principle no reason to think that the m was pronounced any differently in Latin than it is pronounced in all those languages that derive from it. If we then compare Latin not with the daughter languages that evolved from it in later times, but with the sister languages that are related to it because they share an even older common origin, the so called Indo-European languages, and we observe that the letter m has kept exactly the same pronunciation, independently from Latin, into modern Greek μητερα [mi.'tɛ.ra], German Mutter ['mʊ.tʰɐ] and Hindi माता ['mɑ:.tɑ:], which all correspond to Latin máter, then once more we find support to be pretty sure of the way Latin m was pronounced. Similar analogies can be established for all the other letters. Historical and comparative philology can also explain the cases where we observe differences in the related languages, as the presence of an è in French or a u in German where we have á in Latin, and we can thus have an idea of the way in which Latin was and was not pronounced.
2. We have also the written testimony of Latin grammarians, who described how the sounds were articulated, or Latin writers in general who commented on many points of pronunciation.
3. Spelling mistakes we find on inscriptions also illuminate us regarding contemporary pronunciation. If during a certain century everybody writes æquus when they mean ‘just’, and equus when they mean ‘horse’, and we don’t find, even in the writing of the less educated, anyone that confuses the two words, but in the following century many people, even the better educated, start to hesitate and sometimes write equus to mean ‘just’ and æquus for ‘horse’, we can gather two things: first, that in the latter century the pronunciation of æ and e was similar enough to make even educated people make mistakes when they wrote those otherwise unrelated words; and second, that in the former century, on the other hand, the pronunciation of æ and e was still so different that not even uneducated people confused the two words when writing.
4. Finally, poetry gives us a lot of clues about pronunciation, as a poem has to abide by certain rules of sonority that tell us how the words in it should be pronounced to create the adequate rhythm.
With all of these tools and the number of documents that have arrived to us from antiquity, we can be sure that we know quite accurately —there are always some minor obscure areas, of course— how Latin was pronounced during the classical period. The interested student should refer to W. Sidney Allen, Vox Latina, The Pronunciation of Classical Latin, Cambridge University Press 1992 , where a much more detailed treatment of all these topics will be found.
But before we proceed with the classical pronunciation of Latin, we need to try and clarify the tremendous confusion that reigns regarding the so called 'ecclesiastical' pronunciation. As it was only natural, after the fall of the Roman empire, Latin, which survived as a learned language, came to be pronounced in every country according to the conventions of the respective vernacular; thus the Latin word lætitia [laɛ̯.'tɪ.tɪ.ʲa] received in Italy the pronunciation [le.'ti.tsja] like the Italian name Letizia, in Spain [le.'ti.θja] like the name Leticia, in England [lə.'tʰɪʃ.ə], etc. (cf. Op. cit., "Appendix B: The pronunciation of Latin in England", pp. 102-110). These traditional pronunciations, varying from one nation to another, were in use in each country in all areas of Latin, both in scholarly contexts and in ecclesiastical ones, of course indistinctively as there was never a conscience that there should be any division between the two. It is wrong to think therefore, as not a few people in modern times have mistakenly been led to believe, that there existed in history a specifically ecclesiastical pronunciation of Latin or that the Italian proununciation would ever have been used by non Italians. Against these varying traditional pronunciations, it’s true, there had been throughout history different erudite initiatives to promote a unified pronunciation of Latin; yet these were not based on the Italian vernacular, but rather on the testimony of the ancient sources, usually understood to support the principle of "one letter one sound" independently of the spelling context, as seemed to be suggested by phonetic descriptions of the letters of the alphabet like the one provided for instance by Martianus Capella in his Dé nuptiís. These erudite unifying tendencies, of course, affected the Latin used both inside and outside the church indistinctly, as for centuries scholars and ecclesiastics were the same people. As a matter of fact, this type of pronunciation was still in use in the court of pope Leo XIII, as anyone who listens to the 1902 recordings of the Italian castrato Professor Alessandro Moreschi can easily perceive (he says [pon.'ti.fi.kem], ['e.ti.am], etc.). Yet these erudite attempts failed for centuries to displace the various national pronunciations that traditionally prevailed in the different countries, both inside and outside the church.
As erudition became increasingly sophisticated, the simple principle of "one letter one sound" was refined by further inspection of the ancient sources, notably by Erasmus and his successors, and the tendency to bring the pronunciation of Latin ever more closely to the one it had in classical times became the main aim of the better educated, who were also ever more disconnected from ecclesiastical milieus. After the fascinating advancement that the science of philology experienced in the last hundred years —scientific comparative philology didn’t really start before the 19th century, with the development of interest in Sanskrit that came along with the European colonization of India—, it was finally possible to recover the classical pronunciation of Latin to remarkable detail. The brilliance of this intellectual achievement of philological science made that in a very short time the unified erudite pronunciation of Latin finally displaced the diverse vernacular traditional pronunciations from all areas of scholarship all over the world (except, partially, in Italy); but as erudition and church now went along completely diverging ways, the different traditional pronunciations prevailed in the ecclesiastical context. This century saw therefore for the first time in history a division between the pronunciation of Latin inside and outside the church, and the birth of the label of 'ecclesiastical' for what was nothing other than the traditional vernacular pronunciations of Latin. As had traditionally been the case, these vernacular pronunciations were still different in the different countries, and in response to the successful internationalisation of the erudite pronunciation, the ecclesiastical authorities started a battle to unify ecclesiastical Latin too; but instead of doing so through the finally thriving unified pronunciation of the erudites, which was in origin no less ecclesiastical than the vernacular ones, they decided to try and spread the traditional pronunciation of one country, Italy, to the rest of the world. Pope Pius X expressed this wish in a letter to the archbishop of Bourges in 1912 (cf. Op. cit., p. 108) and the anhistorical identification of ecclesiastical Latin with the vernacular pronunciation of modern Italian has been spreading since then. Yet even in the 1980s the last Spanish priests I heard using Latin were still firmly sticking to the Spanish pronunciation, and I know well that the vernacular German pronunciation of Latin is still the norm in that country. In any case, the demise of the use of Latin in the church seems to be making both the traditional local pronunciations in general, and the Italian one in particular, equally irrelevant.
Leaving therefore the ecclesiastical Babel behind, we will now proceed to describe the classical pronunciation of Latin —which is sometimes given the name of prónúntiátió restitúta—, as the only authentic standard across borders, and indeed the only thriving one. As said before, the philologist has books available in which they can find lengthier explanations; but the majority of students should have enough with the following. For comparative purposes, British English so-called Received Pronunciation (RP) has been chosen as the standard.
Latin has six vowel letters, the last one of which only appears in words of Greek origin:
a e i o u y
Latin has twelve main vowel sounds, since each one of the vowel letters can represent either of two sounds, one we call short and another we call long. The long sound can be indicated in writing by means of the apex, although, unfortunately, in the majority of texts this extremely beneficial mark is not used (in others it is substituted by the metrical macron, with the subsequent confusion between vocalic quantity and syllabic quantity):
Generally speaking we can say that the long sound is achieved by slightly lengthening in pronunciation the short sound; although this lengthening comes hand in hand with a greater effort of the articulatory organs, which slightly modifies not only the duration (quantity), but also the timbre (quality) of the sound of the vowel.
English has very accurate equivalents for the Latin short vowels e (as English bed), i (bit), o (British RP pronunciation of pot) and u (put), and for the Latin long vowels á (British RP pronounciation of father, similar to General American pronunciation of odd), í (machine), ú (goose). Not so much for the rest of vowels: Latin short a is somewhat between the vowels in British RP cut and cat, Latin long ó is similar to British RP pronounciation of thought and even closer to certain less common American pronunciations of force, and Latin long é is unlike anything an English speaker is accustomed to articulate. The sound of vowels y and ý, originally Greek, is also foreign to English; they sound like French u, that is by rounding the lips as for the oo in goose, but saying an ee as in geese.
The main difficulty native English speakers are likely to experience with Latin vowels derives from the fact that long vowels in English usually appear only in accented syllables, so it will prove challenging to put them in non accented ones, where they appear in Latin just as frequently as they do in accented ones; in fact, in non accented syllables, all vowels have a tendency in English to become a mere schwa [ə], which is a neuter vowel (the first a in about) that sounds like something in between Latin a, e, i, o, u and y, while at the same time being none of them in particular. This diminished pronunciation of the vowels hinders comprehensibility enormously and must be avoided at all costs in Latin.
A secondary problem is that some of the Latin vowel sounds (e and o) never appear in an open syllable in English, so there will be a tendency to change them into diphthongs in words like 'mare', whose final e will tend to be pronounced as in English café, or 'homó', which will be made to rhyme with English homo. This should also be carefully avoided.
We now indicate the pronunciation of each vowel by means of the symbols of the International Phonetic Alphabet, and we provide some examples that can be listened to by clicking on each one of them:
Besides the twelve above, Latin has four short nasal vowel sounds, which appear only at the end of words and are indicated in writing by an –m which is called m cadúca and is pronounced as a consonant only when it’s followed by a word that doesn’t begin by a vowel or an h1. English, unlike French or Portuguese, has no nasalised vowels, so these may prove challenging for native English speakers. To nasalise a vowel, the air has to be sent up through the nose. In Latin, they are as follows:
Latin has six diphthongs, that is pairings of vowels belonging to the same syllable. Only the first three diphthongs are really frequent, whereas the second three appear only in a very small number of words. These combinations of vowels can also appear as hiatus, i.e. belong to separate syllables. In the case of the first three pairings, the hiatus will happen mainly in words of foreign origin. No ideal and consistent way to distinguish diphthongs from hiatuses in spelling has yet been devised. The one I follow is described below. It is ambiguous only in the case of eu. Most people and modern editors of Latin texts don’t even attempt to distinguish diphthongs from hiatuses in writing at all, and one is meant to know in advance which case is which.
To put it in a different way, diphthongs are vocalic clusters that, within one and the same syllable, start with one vowel sound and end with another. Looking at the Latin spelling, it should be absolutely straightforward how they are going to sound (although this will look much less straightforward for native speakers of English, whose vocalic system is very different from others): æ starts with a and ends with e (similar to English I), œ starts with o and ends with e (similar to boy), au starts with a and ends with u (just as in cow), eu starts with e and ends with u (similar to a very posh British post), ej starts with e and ends with i (just as in hey), uj starts with u and ends with i (without equivalent in English).
Latin has nineteen consonantal letters, the last one of which only appears in words of Greek origin:
b c d f g h j k l m n p q r s t v x z
As has been said above, the letters j and v are only graphic variants of i and u, and the ancients didn’t allocate any precise distinguishing value to the difference; nevertheless the pronunciation of the consonants differs from that of the vowels and we consider the written distinction later established to be advantageous in order to mark this difference, and therefore to be worth observing. Of course, it makes no sense to observe it in one case and not in the other.
The letters c, k and q represent exactly the same sound. The letter k is just an archaism, which only appears in a few words before a (e.gr. kalendæ) and q is only used before a semivocalic u followed by a vowel (e.gr. qvis).
Latin has twenty-four consonantal sounds, which we now indicate by means of the symbols of the International Phonetic Alphabet, and we provide some examples that can be listened to by clicking on each one of them:
Bilabial and velar voiced plosives: b, g
They are pronounced exactly as in English, although one has to remember that the g has always the sound as in girl, never as in gist.
Dental voiced plosive: d
The corresponding sound is pronounced in English with the tongue pressing on the upper alveolar ridge (the fleshy part at the front of the palate where the roots of the teeth are contained), whereas in Latin, as in most other languages (whether Romance ones or not), it is pronounced with the tip of the tongue pressing on the teeth themselves.
Voiceless plosives: p, t, c
The corresponding sounds in English are pronounced with a strong explosion of air after them, which is clearly observable if words like pot, top or car are pronounced in front of a flame or a loosely hanging piece of paper situated close to the mouth. This explosion should be avoided, so that the correct pronunciation of Latin 'per', 'ter', 'cor' will not make the flame or the paper tremble. The puff of air is weaker in English after an initial s- (so called liquid s, as in spot, stop, scar), and it’s this type of unaspirated sound that should be aimed at instead. In fact, the strong puff of air makes the English sounds rather similar to Latin ph, th, ch below. As for d, it has to be noted that the articulation of t in Englih is alveolar, but in Latin it’s dental.
Aspirated plosives: ph, th, ch
These Latin sounds, all derived from Greek, are in fact quite similar to the strongly aspirated initial p-, t-, c- of English pot, top or car. One should be able to extinguish a candle by saying 'pha', 'tha', 'cha' in front of it. As for d and t, the articulation of Latin th is dental, not alveolar.
Labiovelars: gu [gw], qu [kw], su [sw]
The Latin combination 'gu' can sound [gw], as in guacamole, but only after an n; for instance, in the disyllabic 'san.guis'. Otherwise the combination 'gu' is just a 'g' plus a vocalic 'u', as in 'e.xi.gu.i.tás'. The Latin combination 'qu' is always pronounced [kw], as in queen. The Latin combination 'su' can be pronounced [sw], as in suave, but only in a very few Latin words, like 'suá.de.ó' or 'Sué.tó.ni.us', that need to be learnt specifically. Otherwise it’s just an 's' and a vocalic 'u', as in 'su.a'. Some editors used to spell the above cases as gv (sangvis), qv (qvis), sv (svadeo), in the past. This makes a lot of sense and should probably be reinstated.
Nasals: m, n
They are pronounced as in English, except in two cases. An n before s or f made the previous vowel long, possibly nasalizing it, and possibly disappearing altogether from pronunciation as a consonant (cf. 'cónsul', 'ínfimus'). As said in the vowels section above, a final -m had undoubtedly disappeared altogether from pronunciation as a consonant in classical times (except before other consonants) and indicated just a nasalisation of the previous vowel, which nevertheless remained short (cf. 'dictum').
Liquids: r, rh, l
The Latin r is pronounced as in Italian. Basically it’s a single flap of the tongue when it’s single (which sometimes sounds as a 'd' to a native English ear) and a lengthened trill when double. There are useful tongue-training exercises for native speakers of English to pronounce a correct Latin r. One of the most useful ones is to say train without the vowels (i.e. trn) repeatedly, whereby the articulation of the r can be brought forward in the mouth to that of the t and the n that contain it. Through repeated practice and increasingly dwelling on the r, the student eventually learns to roll the r.
In Greek words there appears an aspirated rh which is pronounced as a combination of a single r, as described above, and an h at the very same time.
The Latin l has two distinct sounds (something ignored by many), just as in English: a clearer one, as in English low, and a darker one, as in English goal. The clearer one is only used before an 'i' (as in 'límes') or when double (as in 'ille').
Fricatives: f, s
The 'f' is as in English.
The 's' is also as in English, although one has to remember that it has always the sound as in dose, never as in rose.
The 'h' is as in English.
Double consonants: x, z
The 'x' represents the double sound [ks]. One has to remember that this is never softened into [gs] or [gz].
The sound of 'z' is dubious. It was either a [z] sound, as in English zero; or a [dz] sound, as in Italian zero (not as in Italian zio). The options are similar enough to leave the choice open to individual preference; also, it was a Greek letter, foreign to Latin, so people not educated in Greek would probably have had trouble to pronounce it, and would have adapted it in different ways, mostly as a mere 's', as can be appreciated in the Latin term 'massa', from Greek μαζα. Between vowels, the letter z is always pronounced as double, as the Latin adaptation 'massa' testifies, but it is never written geminated as other consonants: a word like 'gaza' is therefore pronounced ['gaz.za] or ['gadz.dza] (as Italian gazza).
Semivowels: j, v
The 'j' (which many people now write as a mere 'i', following spelling conventions and considerations alien to Latin) sounds as English 'y' in yet. When between vowels, it’s always pronounced (never written) as double: a word like 'ejus' is therefore pronounced ['ɛj.jʊs].
The 'v' (which the people above usally write as 'v', following again spelling conventions and considerations alien to Latin, and which only very few of those who advocate the end of 'j' are at least consistent enough to drop and substitute by 'u') sounds as English 'w' in wet (more probably as Hindi व ).
Most of the consonant sounds can appear doubled in writing (e.gr. appónere, accipere, terra, commúnis, etc.) and they must be pronounced also as double ([ap.'po:.nɛ.ɾɛ], [ak.'kɪ.pɛ.ɾɛ], ['tɛr.ra], [kɔm.'mu:.nɪs]). Although Italians will have no problem with this, all others need to pay attention to pronounce the two consonants clearly. One needs also to remember that the consonants j and z are always spelt as single but pronounced as double when between vowels (cf. ejus ['ɛj.jʊs], gaza ['gaz.za] or ['gad.dza]).
In order to know where to put the accent or stress in a Latin word, and also to understand the structure of Latin verse, we first need to know how to divide a Latin word into syllables.
1a. Every syllable has a vowel as its nucleus (no consonant can form a syllable on its own without a vowel).
1b. There will normally be as many syllables in a word as there are vowels, one vowel per syllable. So vowels that appear together regularly belong to different syllables:
me-us, su-us, a-vi-a, ó-ti-um
1c. Exception: the groups seen above (æ, , au, and less frequently eu, ei, ui) are usually not divided, but pronounced within the same syllable, and we call them diphthongs; on rare occasions (usually foreign words, but not always) those groups of vowels do belong to different syllables, a phenomenon we call hiatus. Cases of hiatus have to be learnt specifically.
Diphthong æ: æ-ri-us
1d. Note: It is a very common tendency to hear things like a disyllabic pronunciation of 'tertia' ['tɛr.tja]. This is wrong and should be carefully avoided. The combination 'ia' is no diphthong, so the i and the a belong to different syllables and the word is trisyllabic and should be read 'ter-ti-a'. In fact, in combinations of i+vowel, a glide in the form of a [j] (as in Eng. 'y') happens between the two to produce ['tɛr.tɪ.ʲa].The same goes for u+vowel, where a [w] sound appears; thus 'quattuor' sounds ['kwat.tʊ.ʷɔɾ] (not disyllabic ['kwat.twɔɾ]).
2a. A consonant between two vowels forms a syllable with the following vowel, not with the preceding one as it often does in English. The simplest syllabic structure of Latin words therefore is v-cv-cv-cv , or, of course (cf. point 1 above), cv-cv- -cv-cvc:
2b. Any group of consonants between two vowels is separated: the last consonant of the group will form a syllable with the following vowel and the rest will belong with the preceding syllable: cvc-cv , cvcc-cv , etc.:
gut-ta, gut-tur, cel-la, com-mú-ni-tás
2c. Exception 1: the combinations ch, ph, th, qu (qv [kw]), and in some cases gu (gv [gw]) and su (sv [sw]) before vowels, represent single phonemes and cannot be divided; on the other hand, the letter x represents two phonemes [ks] that can belong to different syllables:
pul-cher, sym-phó-ni-a, syn-the-sis, e-qvus
2d. Exception 2: a consonantal cluster of múta cum liquidá (a stop with a liquid) cannot be divided. Mútæ are b/p/ph, d/t/th, g/c/ch, but we also consider f for this rule; liquidæ are l and r:
com-pró-mis-sum, cóns-truc-ti-ó, íns-críp-ti-ó
All syllables in Latin, whether accented or not, should be allocated a fairly regular amount of energy and should be pronounced equally clearly, much unlike other languages like English or Portuguese where most unaccented syllables are hardly dwelled on.
3. Syllabic quantity
It is very important to distinguish between vocalic quantity (short and long vowels) and syllabic quantity (short and long syllables). A syllable is said to be short if it has in it:
a) a short vowel followed by no consonant within the same syllable: a-ni-ma, ca-ve-a
A syllable is said to be long if it has in it:
a) a long vowel: dú-có, Ró-má-ní
Now that we know how to identify Latin syllables and their quantity, we can easily work out the position of the accent or stress in a Latin word. We just need to pay attention to the penultimate syllable of that word, and follow this rule:
If the penultimate syllable is long, the accent falls on it:
má-tró-na (a housewife), ac-cen-tus
If the penultimate syllable is short, the accent goes back one syllable and falls on the antepenultimate (if the word has more than two syllables of course):
Má-tro-na (the river Marne), fa-mi-li-a
Its worth noting in this respect that a vowel before another vowel will regularly be short (e.gr. the i in per-vi-us); although there are some exceptions, mainly in foreign words (e.gr. the é in Mú-sé-um).
Only a few words have exceptionally an accent on the ultima for historical reasons, like il-líc or Ar-pí-nás. These will have to be learnt independently.
The last thing we need to know about Latin pronunciation is a phenomenon we call synalpha which takes place when words are put together in connected speech. When a word that ends with a vowel (including a nasalised vowel, i.e. one followed by m cadúca), or with a diphthong, comes before another word that starts with a vowel or diphthong (including aspirated ones, i.e. those preceded by h), the last syllable of the preceding word and the first one of the following word coalesce, combining into one joint syllable, where the last vowel of the preceding word becomes just a semivocalic sound or glide:
sí mé obsecret [si:.'me̯ɔp.sɛ.kɾɛt]
cum arcessor [kʊ̯̃ar.'kɛs.sɔɾ]
siquidem hercle possís [sɪ.'kwɪ.'dɛ̯̃ɛ̥r.kɫɛ.'pɔs.si:s]
Some vernaculars like English, and even more so German, have a little glottal stop before words that start with a vowel, which prevents synalpha (the fusion of the involved vowels into a single syllable) producing hiatus instead (a pronunciation of the vowels in distinctly separate syllables). This breaks the flow of the language and makes Latin sound very unnatural. It must be avoided.
Despite widespread confusion, it must be noticed that synalpha is neither elision nor diphthong. Elision would mean the complete disappearance from pronunciation of the final vowel of the preceding word, as if we were to pronounce sí mé obsecret as [si:.'mɔp.sɛ.kɾɛt]. Elision happens in many vernaculars (cf. Italian luniversità, Catalan magrada, French jaccuse), but Roman writers clearly indicate that this is not the case in Latin. Diphthong too is different from synalpha, in that in the diphthong the vowel that acts as a semivowel is the second one, whereas in the synalpha its the first one. This has important implications in poetry, because a dipthong makes its syllable always long (the semivowel closes the syllable and makes it long by position), whereas the result of a synalpha can be short or long depending on the quantity of the first syllable of the following word (the semivowel comes before the vowel, and so doesnt close the syllable). We pronounce ita ut æquum fuerat as ['ɪ.ta̯ʊ.'taɛ̯.kʊ̃.'fʊ.ʷɛ.ɾat], where the syllable with the synalpha [ta̯ʊ] remains as short as if it had been [tʊ] or [tɾʊ], whereas a diphthong [taʊ̯] would have made it as long as say [taɛ̯] or [tar].