eLatin eGreek eLearn

More wired than a Roman Internet café


W.S. Allen, in his “Vox Latina”, dismisses the idea that Latin had a pitch accent, despite the description of this accent in great detail by a number of Roman grammarians writing prior to the fourth century AD. Allen states that the accent is “a minor detail of the Greek”. This would be like saying that the musical accent of Italian was “ a minor detail of Italian”. In fact, the survival of the pitch accent, albeit in modified form, in Italian, and the survival of tonality in the five main Romance languages descended from Latin. provides evidence that educated Romans adopted it into their Latin. Cicero himself speaks of the musicality of Latin, likening Spoken Latin to a form of singing. Further evidence exists in the adoption of the tonal accent into Hebrew recitation. Indeed, the Jews adopted the Greek system, including the method for manually marking the tones. (Manuum variis motibus altitudinem, depressionem, flexus vocis significabant) Talmudic texts were published with accents for this tonal singing, until well into the mediaeval period. This accent has similarities to the Greek accent , and probably developed in imitation of the Greek recitation of the Laws to a chanted tune.

Edgar H. Sturtevant, "The Pronunciation of Greek and Latin" University of Chicago Press, 1920, gives a much more developed analysis of the accent than Allen does, and he reaches the opposite conclusion. In paragraph 214 of The Pronunciation of Greek and Latin, Sturtevant sets forth the summary of his argument:

"214. The evidence compels us to conclude that in the period of the classical and post-classical literature the Latin accent involved both stress and high pitch upon the same syllables. For stress we have abundant evidence also for both the pre-classical and the latest periods; but we learn directly of the Latin pitch only for the period from about 100 B.C. to about 300 A.D. It is probable, however, that it existed both earlier and later. In fact, it is not unlikely that the considerable element of pitch in the modern Italian accent is a direct inheritance from Latin."

Bennett, along with David (see below), both of whom I regard as authoritative on this matter, come down in favour of the "Greek" accent. Herman and Wright in “Vulgar Latin” also hold the view that the accent in Classical times was a tone accent (pg 36).

One major plank of the argument regarding Classical Latin and tone versus stress, (Vulgar Latin, J Herman) is defeated by Hungarian, which “has a very strong stress accent involving intensity, while at the same time a whole operating system of vowels based on distinctions in length”.

In other words, a clear strong stress accent and a vowel system based on phonological length distinctions are not ipso facto incompatible. Yet one hears this recited again and again by Classicists, educated linguists and laymen alike, so often has this notion been repeated, that is has taken on authority simply by dint of repetition. I am not sure with which linguists this canard arose – for canard it surely is. There is no empirical scientific evidence for this opinion, only evidence that weighs against. Indeed, as Bennett notes, no human language has either an exclusive tonal accent or an exclusive loud-soft or stress accent. Some languages lean more towards the stress accent than the tonal accent, and others vice versa but the only human speech that would be devoid of tonal variations would be a totally monotone language, which, as far as one may suppose, does not exist, except in the minds of some misguided Latinists.

Classical Latin had both a stress accent, with tonal differentiation, and vowel length distinctions. Earlier Roman Grammarians assert quite explicitly that Latin used a tonal accent, similar to the Greek, and only from the fourth century onward to Roman grammarians talk about relative loudness, as opposed to pitch. (pg 36 Vulgar Latin, J. Herman & R. Wright, 2000, Penn State Press.)

The question of the nature of the Classical Latin accent was initially argued for cogently in English by Abbott, in his paper “The Accent in Vulgar and Formal Latin” (Classical Philology, II ppp 444 ff). Abbott held the view that the accent of the common people continued to be one of stress, but educated Romans developed an accent in which pitch predominated. This view is reasonable enough, when we consider to what extent Roman literature is based on the Greek. Also, educated Romans spoke Greek, with its pitch accent. This view is also supported by R.G. Kent ( Transactions of the American Philological Association, LI, pp19 ff), and Turner (Classical Review, 1912, pp147 ff).

Kent writes “In the middle of the second century BC the Greek teachers of the Roman youth set a fashion of speaking Latin with a pitch accent, for as Greeks they kept this peculiarity of their mother tongue when they learned Latin. From that time on, Latin was spoken with a pitch accent by the highly educated class, while the general populace retained the stress accent” (quoted on pg 55 of “Accentual Change and Language Contact” J. Salmons, 1992, Routledge.

Another recent study in support of the Pitch accent, is “The Non-European and Semitic Languages”, Saul Levin, SUNY Press, pg 236 ff

“ The ancient grammarians say clearly that the accent of Latin is either acute or circumflex, and they describe it just like Greek. In many details the distributions of acute and circumflex [between the Latin and Greek] agrees remarkably.”.

Levin continues to say “ Some in modern times have wrongly doubted, or rejected altogether, the testimony of the Roman Grammarians about accent. But since Latin literature conforms to the syllabification and vowel quantity of Greek, the literary language of Rome can hardly have failed to employ a pitch accent compatible with such versification and prose rhythm.” He then says even more emphatically, “ It will not do to dismiss the Latin pitch accent as an artificial imitation of Greek. The most classical Latin, the kind most thoroughly described in our sources, is the most thoroughly Hellenized. If Latin was ever free from Greek influence in some prehistoric time, that Latin is unknown to us, and to reconstruct it, be peeling off what we may label the literary, Hellenizing features, is a fantasy……..Admitting that there was a raised pitch does not conflict with the stress which undoubtedly was present in early Latin.”

See also the seminal work of J. Vendryes, “recherches sur l’histoire et les effets de l’intensite initiale en latin” (Paris 1902), which is quoted by Bennett.

“New Comparative Grammar of Greek and Latin” Andrew Sihler 1995, OUP , pg 241 also argues in favour of the pitched accent.

“ Roman Grammarians, down to the 4th C AD, describe the Latin accent in terms only appropriate for a pitch accent. Scholars have been wary as taking this as cogent, however, as not only is the terminology of Roman Grammarians taken over entire from Greek, their statements are often cribbed from Greek sources. Some scholars protest, however, that ancient authorities could hardly have thus identified Greek and Latin accent had there not been at least an appreciable element of pitch in the latter….The familiarity of educated Romans with Greek accent in both practice and theory probably would not have caused them to adopt an element of accent wholly irrelevant for their natural speech, but could have made them more aware of an existing element of pitch, and even to a studied enhancement of it – Latin with a Greek accent, if you will, in oratory or recitations of poetry”

Pulgram 1975, pg 116, quoted in “From Latin to Spanish”, Paul M Lloyd, Diane Publishing, 1987, argues that speakers of Classical Latin adopted the Greek pitch accent, and certainly made an effort to adopt it on formal occasions, if not in general speech.

“A New Theory of the Greek Accent” A.P. David, Oxford University Press, 2006 pg 76-7 is the most recent, and authoritative of the new school of scholars who promote the view that the original statements of Quintillian, etc, are accurate descriptions of Latin as it was spoken. Here is Davis' argument:

“It is also possible that Greek forms with an acute on the antepenult are a product of the reflex described in Vendryes’ Law, if the Latin penult in these words was heard to be pronounced with a circumflex.

Might there have been such a contonation in Latin? A simple synchronic picture, which accords with the traditional account, emerges if we assume a contonation. We are informed by a recent commentator that “ Roman Grammarians, down to the 4th century, describe L[atin] accent in terms appropriate only for a pitch accent” (quoting A.Sihler, see above, pg 241).

Modern scholars, however, tend to see a sort of ‘Greek envy’ in this native description and to be dismissive. But if we frame the new rule for Latin in terms of a recessive contonation, where the voice was required where possible to rise two morae before the ultima – without, in the case of this language, any stipulation as to the quantity of the ultima – the traditional stress rules for polysyllabic words in Latin automatically follow, if the combination of pitch and quantity worked in the way that I have described for Greek. A long penult, with two morae, containing the rise combined with the Latin version of the svarita, would produce a circumflex on the penult (amIcus); [circumflex on the capitalised I] while a short penult (of one mora) would cause the rise to revert back to one mora to the antepenult, producing the Latin acute with a deemphasised svarita (facilis).

In making an authoritative correction, Quintillian actually points to this recessive rule. Discussing errors in accentuation, he cites CethEgus [circumflex on the capitalised E] as properly having a flex on the penult (Institutio Oratorio 1.5):the common error was to pronounce the penult grave in this word instead of circumflex, which apparently rendered the penult short. (A circumflex requires two morae). He implies that this change in the quantity of the penult necessitates an acute first syllable (Cethegus) [with acute on the first e] – an erroneous pronunciation, but one which conforms to the proposed rule. The Latin accent was a recessive contonation, a rise and fall, where the rise occurred, wherever possible, on the second mora before the ultima.”

This is sound reasoning for dismissing W.S. Allen’s view.

As a final point, I would like to note, that one reason why one seldom hears Latin declaimed with this accent, is that one seldom hears Classical Greek spoken with it, even though there is not even a sliver of doubt that Classical Greek was spoken with a pitch accent. Current practice, however, is not necessarily a guide to good practice, and I would advocate the use of the tonal accent, for purely pedagogical reasons – it makes Latin more intelligible, and also makes clearer distinctions between stressed and unstressed, unaccented and accented syllables, and long and short vowels.

In monosyllables, vowels that are long take the circumflex, and vowels that are short take the acute.


Polysyllables take the circumflex accent when the penult is long by nature (This simply means that it has a long vowel, see above), and the final vowel is short. A circumflex can only appear over a syllable with a long vowel.

jûrĭs lûcĕ mûsă spînă

The circumflex accent is thought to have has a slight up-down tone, the acute a straightforward upwards tone. Final unaccented syllables had a slight falling tone. (This is called the grave accent, but this is not written, it is simply understood to be there.)

These accents were applied by the Romans in imitation of the Greeks, and may have been used when reciting poetry and during orations.

Views: 883

Comment by Laura Gibbs on June 29, 2008 at 5:05pm
Evan, this is REALLY interesting, and very clear. I think you are quite right in pointing out that the failure of this point of view to make real headway in common Latin practice today is in large part due to the failure to deal with pitch in the teaching of Greek - and I am guilty of that myself, never having heard any of my Greek teachers over the years attempt anything in the way of pitch-based reading, and having taught my students likewise, stressing the accented syllable, with no meaningful distinction between circumflex, acute, and grave (and here is some more true confession: when I taught first-semester Greek, I did not even try to teach my students the rules for writing the different accent marks, since it would not help their reading; I simply had them put an acute accent whenever an accent would go, allowing them to sort out which kind of accent that should be if they continued their studies of Greek into subsequent semesters) - mea culpa.

Admittedly, I personally don't agonize over this - as someone who studies modern languages and ancient languages, I have decided not to invest my own time in cultivating oral fluency in the dead languages, when I am so aware of how it is only due to living among a community of native speakers that I gained any fluency in Polish and in Italian. In the absence of such a living community of native speakers for Latin (or for ancient Greek), I personally don't strive for reconstructed pronunciation, authentic ancient pronuncitation, etc.

Nevertheless, I learned a LOT from reading the materials you have posted here, and I wish I had known all about this when I was teaching Greek - many of my Greek students knew Latin, and I am sure they would have found this very intriguing indeed, as I do! :-)
Comment by Laura Gibbs on June 30, 2008 at 6:12pm
P.S. I forgot to say that the AMAZING Joseph Herman was a visiting professor at Berkeley when I was in graduate school and I took two classes with him then. He is someone who convinced me that there is still real scholarship in the world: his knowledge of Latin was incomparable... and his knowledge of Hungarian and Italian, too, of course, since he was a Hungarian who taught in Italy. It was one of the high points of my graduate studies. :-)
Comment by Molendinarius on July 1, 2008 at 7:07am
Message forwarded to me for posting here by John Doublier:

De Lingva Latina Recte Pronuntianda

I found Mr. Whalen's criticism of Evan Millner's pronunciation of Latin on the latter's website "Latinum" curious to say the least, all the more so because I am by profession a simultaneous interpreter (as well as a translator) practicing this vocation in a number of modern languages, including a few of Latin's granddaughters. Like other interpreters, I am obliged to improve continuously my language skills I have lived eight years in different foreign countries, working on one or another foreign language to gain the required vocabulary, fluency, and pronunciation. In this office I am compelled to pay due attention to the niceties of stress and intonation, especially in my non-native Spanish, French, Italian, German, Romanian, all of which I have interpreted in trials and court proceedings, lest my listeners be left scratching their heads in partial or total incomprehension.

In all this time I have not come upon one human language that does not exhibit both stress and tonality. Some languages, like English and its Germanic cousins, feature more stress than tonality, but the tones are still there and are easily heard in almost any utterance one can imagine, except perhaps from those emanating from an actor or comedian speaking in a monotone for comic or dramatic or anti-dramatic effect. Even our modern English speech, in its most blighted form, i.e., that heard from the relatively immobile mouths of U.S. Midwesterners and farmers and ranchers living on the Great Plains shows some small variation of pitch.

I cannot imagine the Latin of any period having been spoken in a monotone, especially in the Classical period. I speak to a greater or a lesser degree the five major Romance languages, and none of these offspring of Latin wants for tonal variations. Indeed, Italian, Romanian, and French have a very marked tonality in educated speech.

There may have been some backwaters in the Roman Republic or Empire where where farmers, dung-spreaders, and other locals spoke Latin with lock-jaw monotony. But I cannot believe that the philhellene optimates and litterati of Rome and other urban centers eschewed musical pitch in speaking a Latin which, of course, also featured stress. But both the abundance evidence cited by the great American linguist E. H. Sturtevant in his "The Pronunciation of Latin and Greek " and the marked musicality of Latin's children, esp. French, Italian, and Romanian convince me that Virgil did not compose his hexameters with a monotone buzzing in his ear.

If, however it is the monotonous backwater speech that Mr. Whalen wishes to impart to his pedagogical charges, he is welcome to do so, clamping his jaw shut with surgical wire and avoiding arpeggios up and down the musical scale in favor of atonality.

I am not a Latin teacher, hence do not know with what ancient Roman professions today's students of Latin might identify. But I would wager that those of "cantor" and "musicus" at the court of Augustus would win hands-down over those of "agricola" or "stercorator" on the eastern edge of Dacia (Qvid diceret Ovidius?).

As for Latin's modern avatars, it is impossible to hear a native speaker of Italian uttering that at once ancient and modern Italic word "cantare" without noting the delightfully falling interval of a musical fifth, a characteristic of the Italian language that Verdi and Donizetti exploited to such great advantage and which did not simply appear out of nothing. It is also eerily similar to the Greek circumflex accent as described by Dionysius Halicarnassus and deftly reproduced by Mr. Millner on Latinum).

And "cantare" is what I would suggest that Mr. Whalen do a little more of, especially in his classes of Latin, less whatever acoustic appreciation of Latin that his pedagogic charges have retained be completely obliterated by a stress-heavy pronunciation of Latin which taken to the extreme steals the warmth out of the human heart and the beauty out of human life and language.
John Doublier
Comment by Laura Gibbs on July 1, 2008 at 7:44am
Hmmm, I think I will not let my Oklahoma students know that they are speaking a "blighted" form of English that steals the warmth from their hearts and the beauty from their life and language. Calling one form of English speech or Latin speech "blighted" when compared to another is just the same prejudice shown by Mr. Whalen, inverted.
Comment by Molendinarius on July 1, 2008 at 4:21pm
True, John uses somewhat colourful language. 'largely atonal' maybe, rather than blighted is a more apt description. Though from the heights of the Palatine Hill, methinks that is probably how the patres conscripti thought of hoi polloi labouring down below in the suburba.
Comment by Laura Gibbs on July 1, 2008 at 4:24pm
Yes, we are hoi polloi in Oklahoma, and proud of it... ha ha! :-)
Comment by Molendinarius on July 6, 2008 at 3:43pm
Mr Doublier recently wrote to me, with the following observation:

While listening just now to Mr. MacDonald's recitation of Latin sentences, I heard an example, in several of his sentences, of what for me is the conclusive proof that Latin had a pitch accent.

Mr. MacDonald accents by length, quite properly, the last syllable of "volo". He accents the second O in time by making it last two or more morae. He accents the first "o" by making it last, quite properly, one mora. But he fails to accent the first syllable, as the Latin rules demand.

There are only three ways to accent the first syllable:

(1) lengthen the "o". But this is impossible; we know it was a short "o".

(2) stress, by volume, the "o". But this would either lengthen the first o, or, if held to one mora while pronouncing it in a loud volume, produce the oddest jarring effect when contrasted with the unstressed by temporarily long second o.

(3) Raise the pitch of his voice on the first "o".

The third option is the only viable one. It is the only one which renders feasible maintaining the second "o". on the ultima, long. I invite anyone to try these three options himself, by way of an experiment. A high-pitched short "o" in VO does not conflict with a low-pitched long "o" in "LO". This dual system of accentuation is the only way long vowels could have remained long in unaccented syllable.
Comment by Legoquego on July 19, 2008 at 4:34am
I agree with the article and I do Latin in the way as posited by that article. Now, time for boasting. I can do Latin with a stress and a pitch accent and a mixture of both. I think my pronunciation is just as good as Robert Sonkowsky or Lucus Eques (Luke Amadeus Ranieri, from textkit forums). The last comment by Evan is very straight to the point: one should keep the stress on the correct syllable and continue distinctions between vowel (more like syllable) lenghts. By the way, Estonian is much more atonal than Latin, Italian, Ancient Greek, but sometimes it is funny to speak Estonian with an Italian accent, especially the Latin-derived foreign words, makes for a good joke, but such a joke is less funny as time goes by. Sometimes, theatrical performances are done with a theatrial intonation in Estonian that probably derives from Italian opere. To us Estonians it sounds either artistic(appealing), or annoying, or funny. Although, also equally as much, foreign accents are appealing. I know how to pronounce several languages, but I only ever learned 4 seriously: Estonian (native, similar to Hungarian especially in terms of vowel length distinctions and a presence of stress, in our case a stress on the first syllable and on other syllables in native words, tonality in Estonian is comparable to that of Germanic languages, but we have 14 cases in Estonian.) Though, I found a book containing an antology of ancient Greek poetry translated into Estonian. Apparently, it's very difficult to render Estonian in Greek meters, as we have 3 syllable length distinctions, in a way, too complicated to describe in this post. Latin pronunciation is very easy for me, because Estonian has all the letters in almost exaclty the same phonetical detail, and way more(a total of 9 vowels, and therefore lot of diphthongs), but Estonian lacks the w sound, but we have the j (i glide, like in Latin use). It's very easy to use the w now because I picked it up from English.
Comment by Molendinarius on July 23, 2008 at 4:10am
Some more notes:
Regarding Greek, WS Allen uses the term 'contonation', consisting of a high pitch on the accented mora, and a falling glide on the following mora, if part of the same syllable [i.e. the circumflex accent]. The then states, "not more than one mora ( short vowel) may follow the contonation" (p 124 Vox Graeca)

Roman Jakobson pg 270 "On Ancient Greek Prosody", in "Selected Writings" says:
In Latin the principle of the mora predominates over the syllabic principle: the count of syllables gives only the starting point - the second syllable from the end of the word, from here on there begins the count of morae - the second mora from the end of this syllable determines the position of the accent. Thus, the count of the morae is the immediate determinant of the place of the accent in the Latin word......The quantitiative principle of versification in no way excludes the accent from among the rhythmic factors, as has been shown by the studies on the distribution of the accent in Latin verse and by the recent work of J. Rypka ( in Travaux du Cercle Linguistique de Prague, VI (1936) pg 192 f.

See also J. Safarewicz, "Etudes de phonetique et de metrique latines" (Wilno, 1936) p 73f.
Comment by Molendinarius on July 23, 2008 at 4:25am
more notes:
"New Comparative Grammar of Latin and Greek" Andrew Siher pg 236
In Greek the PIE system of accent was altered in two different ways. First, the free accent was replaced in nouns and adjectives by one in which the accent must fall within the last three syllables; and if the ultima was long, within the last two. Or, expressed in terms of morae ( a short vowel = one mora) the accent could stand on the fourth mora from the end in a case like anthorpos, with a short ultima, otherwise not further back than the third mora from the end. Hence, the circumflex ( in effect acute plus grave) was excluded from the antepenult, and might occur on the penult only when the ultima was short.......The historical Latin accent resembled the Greek in that it could not stand farther back than the third syllable from the end of the word, but beyond this general restriction the resemblances cease. In Latin the penult ( a syllable which plays almost no role in the Greek system) determined the position within these limits, in Greek the governing syllable was the ultima ( a syllable which plays no role in the Latin system). The Latin accent was uniformly recessive...the ruling variable in Latin was the weight of the penult, while the length of the vowel of the ultima is operative in Greek. The Latin accent fell on the antepenult unless the penult was heavy ( to borrow a term from Sanskrit Grammarians), namely containing a dipthong or a long vowel or was followed by more than one consonant [i.e. long by position]. In such a case, the accent fell on the heavy penult."


You need to be a member of eLatin eGreek eLearn to add comments!

Join eLatin eGreek eLearn



© 2017   Created by Andrew Reinhard.   Powered by

Badges  |  Report an Issue  |  Terms of Service