The Pronunciation of Classical Latin

The circumflex accent ( ˆ ) is used to denote the length of the vowels.

Orthography is conventionalized, and the contemporary Roman grammarians’ comments lack clarity, so that to a considerable extent it is necessary to extrapolate from later developments in Romance in order to describe it. On the whole, linguists think that Latin probably sounded something like Italian, though some features make it close to Castillian Spanish.

By the time of the Late Roman Republic (i.e. in the 1st c. BC) the Latin alphabet consisted of 23 letters, named and pronounced as shown in the following table:
 

Letter	Name	Phonetic  Value	Notes
A	a	[a] [a:]
B	be	[b]
C	ce	[k’] [k]	Initially c was written to mark a softened [k’] sound before the anterior vowels e and i and the diphthongs ae and oe, while k was written before a, o, stressed u and the consonants and was pronounced in a hard manner. Because the difference between the two consonants was not significant to the speakers and there were not grammatical functions associated with it, it became a common practice to mark both sounds by the letter c.
D	de	[d]
E	e	[] [:]
F	ef	[f]	The letter f probably represented by classical times a labiodental sound pronounced with the lower lip touching the upper front teeth like its English equivalent) but earlier it may have been a bilabial (pronounced with the two lips touching or approaching one another).
G	ge	[g]
H	ha	[h]	H was pronounced only by educated speakers even in the classical period, amd references to its loss in vulgar speech are frequent.
I	i	[i] [i:] [j]	The [j] sound (technically called consonantal i) appears in the beginning of the words before a vowel or in the middle of the words between two vowels, as in ius [jus] and cuius ['kujus]; the compound words preserve the [j] sound of the element, that begins with it, cf. coniunx [konjunks] and adiectivum [adjektivum]. For convenience we will use the letter j (named jota) to mark the [j] sound, as it is common from the Middle Ages onward; so we will write the above words as jus, cujus, conjunx and adjectivum.
K	ka	[k]	The usage of the letter k was preserved in several words, as Kalendae, Kaeso, Karthago (also Carthago) etc.
L	el	[l]
M	em	[m]
N	en	[n]
O	o	[] [:]
P	pe	[p]
Q	qu	[q]	The letter q was written before unstressed semivowel u [=w] to mark a voiceless labiovelar sound [q], as in quinque ['qwinqw] (presented also as ['kwinkw]) with the accent on i. When u was stressed and thus not of semivowel type, there was written c instead of q, cf. cui ['kui] to whom (with accent on u) vs. qui ['qwi] who? (with accent on i).
R	er	[r]	R was probably a tongue trill at the classical period, like in modern Castilian, but there is earlier evidence that in some positions it may have been a fricative or a flap.
S	es	[s]	It is suggested that Latin s had a pronunciation like that of modern Castilian (with the tip, rather than the blade, raised behind the teeth, giving a lisping impression). In early Latin it was often weakened in final position.
T	te	[t]
V	u	[u] [u:] [w]	The [w] sound was pronounced before vowel, as in solvo ['slw] or quartus ['kwartus], while before consonant in the beginning or the middle of the word and after consonant at the end of the word was heard [u], as in unda ['unda], natura [na'tu:ra] and natu ['natu:]. In the modern printed Latin texts u stands usually for [u] and v for [w].
X	ex	[ks] [gz]
Y	ypsillon	[y] [y:] => [i]	The letter y was introduced from the contemporary Greek alphabet to mark the [y] sound, inexistent in Latin, but frequently found in the numerous Greek loan words.
Z	zêta	[dz] => [z]	Like y, z was also introduced from the contemporary Greek alphabet to mark the [dz] sound, especially at the beginning of the words.

In addition to the consonants shown, educated Roman speakers evidently used a series of voiceless aspirated stops, written ph [], th [], ch [x], originally borrowed from Greek words but also occurring in native words (pulcher beautiful, lachrima tears, triumphus triumph, etc.) from the end of the 2nd century BC. The rh, surely not distinguished phonetically from r, was written in the beginning of the Greek loan words only (like rhetor, rhombus etc.).

The sound [] (as in English ‘sing’), written ng or gn, may not have had phonemic status (in spite of the pair annus/agnus year/lamb, in which [] may be regarded as a positional variant of [g]).

Consonants written double in the classical period were probably so pronounced (a distinction was made, for instance, between anus old woman and annus year). When consonantal i appeared intervocalically, it was always doubled in speech.

In the technical vocabulary of the Roman grammarians the consonants p, t, c, k, q, b, d, g were called mutae mute and the consonants l, m, n, r liquidae liquid.
 
 
Vowels’ Quantity and Quality

The Latin vowels differed in their quantity (i.e. time of pronunciation) into short, pronounced for a time of one mora, and long, pronounced for a time of two morae. Because the system of vowel length was lost after the classical period, it is not known with any certainty how vowels were pronounced at that period; but, because of later developments in Romance, the assumption is that the vowel-length distinctions were also associated with qualitative differences, in that short vowels were more open, or lax, than long vowels.

The distinction between short and long vowels was grammatically significant, cf. hora hour vs. horâ at (this) hour, now, and since early times the Romans tried to mark the difference by writing two consequent letters (eg. AA for [a:]), by putting an “apex” (`) as at the end of the Roman Republic or an acute accent (') as by the time of the Empire. This practice was, however, never universally and uniformly accepted. In the late Middle Ages it became conventional, especially in the manuals, to mark the long vowels by putting a stroke over the letter (eg.  for [a:]) or by circumflex accent over it (eg. â for [a:]), while the short vowels were marked by a small bow over the letter (, , , , ).

In Classical Latin the length system was an essential feature of verse, even popular verse, and mistakes in vowel length were regarded as barbarous. In later times, however, many poets were obviously unable to conform to the demands of classical prosody and were criticized for allowing accent to override length distinctions.

Classical pronunciation also used some diphthongs pronounced by educated Romans much as they are spelled, especially ae (earlier ai), pronounced perhaps as an open long e [:] in rustic speech, au (rustic open long o [:]), and oe (earlier oi, late Latin e). Moreover a neutral vowel was probably used in some unaccented syllables and was written u or i (optumus, optimus best), but the latter rendering became standard.

The Roman grammarians tried to establish formal rules about determining the vowels’ length. According to them vowel before another vowel or h was always short, e.g. puer boy, veho carry. Vowels were long by nature, as in corôna crown, or by position, when a vowel was followed by two or more consonants (x being considered two consonants ), as in locûsta lobster; the combination of a mute and a liquid consonant did not, however, lenghten the preceding vowel, thus the e in tenebrae darkness, the i in arbitror think, believe and the u in volucris bird were thought short because br, tr and cr all were combinations of a mute and a liquid consonant. The diphthongs were considered always long.
 
 
Syllable (Syllaba)
 
Syllable consists of a vowel or a diphthong with or without one or more consonants. The Roman grammarians considered the syllable short if its vowel was short, and long if its vowel was long..They elaborated several major rules about the syllables:

1. A consonant between two vowels is united with the following vowel: me|di|ci|na.
2. The double consonants are separated between the syllables: stel|la.
3. A consonantal group is united with the following vowel if there is a Latin word beginning with this consonantal group, e.g. di|sci|pli|na, because in Latin there are words like scire to know and plenus full that begin wit these consonantal groups; in the word pro|vin|cia the group -nc- is divided because there is no any word in Latin to begin with nc-.
4. The group st is always divided anyway, though there are words in Latin that begin with it: has|ta.
5. The compound words are divided according to their elements: dis|tribuo, trans|eo etc.

 

The way vowels developed in prehistoric Latin suggests that there was a heavy stress accent on the first syllable of each word, but in later times the accent fell on the penultimate syllable or, when this had light quantity, on the antepenultimate (much as in modern Italian). The nature of this accent is hotly disputed: contemporary Roman grammarians seem to suggest it was a musical, tonal accent and not a stress accent; thus Varro contrasts the pitch (altitudo) of a sound with its length (longitudo) (De Lingua Latina, 210, 10-16, GS). This view is traditionally uphold by the French linguists. If this were so, the acoustic effect of Latin would be quite different from later Romance. It is possible, however, that Latin grammarians were merely imitating their Greek counterparts, as all the Latin grammar was built up on the pattern of the Greek one (See The Origin of the Latin Grammar Terms). It is implied that the fact that in Latin accent is linked with syllable vowel length makes it unlikely that such an accent was tonal. Probably it was a light stress accent that was normally accompanied by a rise in pitch; in later Latin evidence suggests that the stress became heavier.

The system of syllable quantity, connected with that of vowel length, must have given Classical Latin distinctive acoustic character. Broadly speaking, a light (or open) syllable ended in a short vowel and a heavy (or closed) syllable in a long vowel (or diphthong) or a consonant. The distinction must have been reflected to some extent in late Latin or early Romance, for, even after the system of vowel length was lost, light syllables often developed in a different way from heavy syllables.
 
 
Later changes in Pronunciation
 
In the course of time the classical pronunciation have gradually changed. By the epoch of the Late empire (3^rd and 4^th c.) the diphthongs ae and oe were equalized with the short e and au with the short o. The pronunciation of y coincided with that of i and by then the letter name ypsillon was replaced in the popular language by i graeca Greek i.

The most important phonetic change of that period was, however, the palatalization of c and g before the anterior vowels e and i and the equalized with them ae, oe and y. In this position c was changed in Gaul and Spain to [ts] and in Italy and on the Balkans to [], cf. F. ciel [sjel], Sp. cielo ['jel], Port. ceu [seu] vs. It. celo ['el], Rum. cer [er], all from L. coelum. G was palatalized in all Latin speaking world to []. The consonantal i [j] was also transformed to [].

T before short i was palatalized everywhere to [ts], if not preceded by s, t, x; thus we have natio ['natsj] nation, lectio ['lktsj] lecture, but ostium ['stjum] mouth (of river), Bruttium ['bruttjum] name of province in Southern Italy, mixtio ['mikstj] mixture, totius [t'ti:us] (with long i).

Ch and th lost their aspiration in the pronunciation and were equalized with c [k] and t.

The pronunciation of the imperial period was used later throughout the Middle Ages with some variations influenced by the particularities of the living local idioms. The International Congress of living Latin, held at Avignon in 1957, recommended the usage of the classical pronunciation. The following congresses at Lyon (1961), Strasbourg (1963), Rome (1966), Bucharest (1970), Malta (1973), Dakar (1977) and Trier (1981) reaffirmed this recommendation.