Thoughts on the Accentuation of Classical Greek

Thoughts on the Accentuation of Classical Greek
in the English‑Speaking World
Dr Sean Gabb
(Text of a Seminar Paper read out at London Guildhall University, May 2002)

Though Greek has, ever since the Renaissance, been regarded within our civilisation as a language both important and beautiful, its pronunciation has always been questionable. I am not here doubting the correct sound of vowels and diphthongs and consonants. Though when first re-introduced to us by the Byzantine scholars and refugees of the 15thcentury, Greek had a pronunciation very different from that of the ancients, this was quickly noticed, and efforts were made to recover the ancient pronunciation. Antonio of Lebrixa in 1486, Aldus Manutius the Elder in 1512, and Desiderius Erasmus in 1528, began the process of recovery. It was substantially complete long before the 20thcentury. Today, there is a single system of pronouncing Greek in all countries—excepting Greece, where the ancient language continues to be pronounced as if it were the modern—and any doubts that remain are over details that are of no importance to the average student or even scholar. The problem with Greek pronunciation lies in the placing and use of accents.

This is a serious problem. A language is not just a vocabulary and set of grammatical rules. It is also a system of sounds; and much of its appreciation lies in an appreciation of its sounds. This is particularly the case with Greek, in which nearly everything was composed to be performed or read aloud, and where the sound of literature—whether prose or verse – was analysed by the critics with a passion and minuteness largely absent from criticism in the modern languages. When, therefore, we take up any of the Greek classics, it is natural to ask what effect the writer was trying to produce—or perhaps what effect was produced on his Greek audience. We may appreciate Homer read according to the modern Greek pronunciation, just as we may appreciate a Mozart string quartet played by a rock band. But this will not be an original appreciation, and it will not satisfy most people of taste. Where classical music is concerned, there is a strong modern desire to look beyond the performing styles that emerged in the late 19thcentury, to know what was intended or heard in an age of lighter instruments and different performing styles. For the same reason, so much effort has gone over the centuries into recovering and teaching the ancient pronunciation of Greek.

Now, the sounds of a language do consist of vowels and diphthongs and consonants, and quantities of vowels, and things like the seeking and avoidance of clash and alliteration. But not forgetting these, the sounds of a language are also determined by accent, or by the particular emphasis given to certain syllables. English has a powerful stress accent, and good style in English is at least partly felt in the accenting of a passage. Take this from The Bible:

Though I speak with the tongues of men and of angels, and have not  charity, I am become as sounding brass, or a tinkling cymbal. And though I have the gift of prophecy, and understand all mysteries,  and all knowledge; and though I have all faith, so that I could remove  mountains, and have not charity, I am nothing. And though I bestow all my goods to feed the poor, and though I give  my body to be burned, and have not charity, it profiteth me nothing. (1 Corinthians, 13:1-3)

Great literature is beyond analysis. But the magnificence and beauty of this passage surely owes much to its accenting. Shift any accent to a neighbouring syllable, and the effect is marred. Even retranslate the Greek αγάπη not as “charity” but as “love”, and the consequent repatterning of accents diminishes the effect. And so, an original appreciation of Greek literature requires a correct placing and use of accents.

The problem, however, is that an original appreciation seems to be impossible. We do not know how the Greek accents sounded. There is little dispute over the Latin accent. This was one predominantly or wholly of stress. In words of more than two syllables, it fell generally on the penultimate if long or on the antepenultimate if the penultimate were short. The rule seems to have influenced prosody—either directly in the case of the last two feet of a dactylic hexameter, or indirectly so far as accent was determined by quantity; so that particular arrangements of words according to the rules of quantity would lead to fairly recognisable accompanying patterns of stress. The Greek accent, though, was one of tone, and had no easily discerned influence on the writing of verse, which seems to have been purely quantitative. Starting perhaps in the 2nd century BC, and completed perhaps by the 3rd century AD, this accent of tone was replaced by one of stress; and this is how Greek was pronounced by the Byzantines and is now by the modern Greeks.

This tonic accent of Greek was described and analysed by the ancient critics and grammarians; and they are unanimous in claiming that its melodic properties were essential to appreciating the aesthetic power of Greek literature. When a natural perception of the tonic accent began to fade, they went to the effort of creating diacritical markings to show in the manuscripts its placing and nature. Sadly, for all their efforts, the sound of the tonic accent is wholly lost to us. The Byzantines may have reintroduced an understanding of Greek to the West, but they no more understood the true meaning of its accents than we do. Speaking of them, Edward Gibbon laments:

Of the power of the Greek accents they were ignorant; and those musical notes, which, from an Attic tongue and to an Attic ear, must have been the secret soul of harmony, were to their eyes, as to our own, no more than mute and unmeaning marks, in prose superfluous and troublesome in verse.[1]

For myself, as a speaker of English, I cannot even imagine how this accent might have sounded – an accent able to make verse sound naturally musical but not able to disrupt perception of a rhythm based on patterns of long and short syllables. I have no doubt that this was the case. Sadly, I am unable to appreciate it. I am certainly unable to reproduce it in my own reading of Greek verse or prose.

Perhaps the French do not see this as a problem. Their language seems not to bother much with stress, and many of their scholars continue to insist that even Latin had no stress. But it is a problem for a speaker of English or any other stressed language. We naturally—and perhaps irremediably—look for patterns of stress to make sense of continuous speech of any kind; and where no pattern is given, we impose one of our own. Like someone creeping through a minefield, we need to know where we can press and where not. What pattern can we, as English-speakers, impose without doing too much violence to the intended and received effect of Attic tongues on Attic ears? Broadly speaking, the problem has been given five solutions.

First, there is the Byzantine and modern Greek system of placing a stress on the accented vowels in prose, and in verse of shifting the stress to the first long syllable in every foot. This system has a number of advantages. It represents with fair accuracy the sound of later Hellenistic Greek and the Greek of the Byzantines. This is a particular advantage for anyone who wants to appreciate the rhythmical prose of the later Church Fathers and the liturgies of the Orthodox Church. It also has a natural authority. To read Greek in this way is seriously to misrepresent the sounds that Homer and Thucydides  and Demosthenes were interested to produce. But it does reproduce the effect that these writers had on the contemporaries of Nonnus and Procopius and Libanius. Additionally, it makes much easier work of remembering where to place the accents when writing Greek, and makes the modern language more accessible.

However, there are disadvantages in using the system. It requires the use of two different systems of accentuation, one for prose and one for verse. This is untidy. It also does much to ruin the sound of verse, which becomes a monotonous verbal scanning exercise hardly ever found in the reading of Latin verse outside a classroom. It also leads to the stressing of syllables in an apparently random deviation from the stresses placed on the accented syllables in the reading of prose.

As an example of such a reading, take the opening lines from the Iliad:

μῆνιν ἄειδε θεὰ Πηληϊάδεω Ἀχιλῆος
οὐλομένην, ἣ μυρί’ Ἀχαιοῖς ἄλγε’ ἔθηκε,
πολλὰς δ’ ἰφθίμους ψυχὰς Ἄϊδι προί̈αψεν
ἡρώων, αὐτοὺς δὲ ἑλώρια τεῦχε κύνεσσιν
οἰωνοῖσί τε πᾶσι, Διὸς δ’ ἐτελείετο βουλή,
ἐξ οὗ δὴ τὰ πρῶτα διαστήτην ἐρίσαντε
Ἀτρεί̈δης τε ἄναξ ἀνδρῶν καὶ δῖος Ἀχιλλεύς.

The verse ictus is generally different from the written accents.

If we think of tidying the system, and maintaining the same accentual rules in prose and verse, as we do in Latin, the results for verse are chaotic. As said, in Latin, where stress is determined by quantity, a non‑scanning reading of verse produces an orderly stress rhythm which can be reasonably taken as a counterpoint to the quantitative rhythm.[2]Take, for example, this from Virgil:

huc omnis turba ad ripas effusa ruebat,
matres atque uiri, defunctaque corpora uita
magnanimum heroum, pueri innuptaeque puellae,
inpositique rogis iuuenes ante ora parentum.
(Aeneid, vi.305-8)

The verse ictus falls thus:

huc omnis turb(a) ad ripas effusa ruebat,
matres atque uiri, defunctaque corpora uita
magnanim(um)heroum, puer(i)innuptaeque puellae,
inpositique rogis iuuenes ant(e) ora parentum

In speaking the verse, the accents fall thus:

huc omnis turb(a) ad ripas effusa ruebat,
matres atque uiri, defunctaque corpora uita
magnanim(um) heroum, puer(i) innuptaeque puellae,
inpositique rogis iuuenes ant(e) ora parentum

The works in Latin. Even when the quantitative patterning is obscured by the stress accents, there is a certain regularity in all but the more complex metres. Where hexameters are concerned, there is the nearly universal conicidence of stress and quantity in the two last feet. It works also for hendecasyllable verses. And reciting poetry as if were prose seems to have been the Roman practice. But in Greek, where accent and quantity are apparently unrelated, a non‑scanning reading of verse produces rhythmical effects that do not just overlay and obscure, but obliterate the quantitative rhythms—much as the street pattern of mediaeval London gives no indication to that of the Roman city.

As for the alleged advantage of making modern Greek more accessible, this is easily dismissed. In the first place, there is not that much worth reading—far less than in languages like Czech, which, in spite of a most interesting and entertaining literature, hardly anyone outside Central Europe bothers to learn. In the second, the “purist” Greek used before the 1980s, which was plainly related to the classical language, has now generally been replaced by the far less closely related demotic Greek. I find my knowledge of French a better key to reading Spanish than my classical Greek is to the modern language. Classical and modern Greek are different languages; and it makes no sense to distort the sounds of one to enable making sense of the other.

Recognising the difficulties of the Byzantine system, two Dutch scholars of the 17th century—Isaac Vossius and Heinrich Christian Henning[3]—proposed an alternative system. They suggested that the written accents should be disregarded. Instead, the language should be stressed according to the Latin penultimate rule. This system was adopted in Holland and in the English‑speaking world—though it was displaced in America during the 19th century by the Byzantine and modern system that continued to prevail elsewhere.

There are two advantages to this system. It allows a single accentual reading to both prose and verse. It also allows a reading of verse that is neither monotonous and unnatural nor chaotic. Most English-speaking scholars of Greek have previously learnt Latin; and a reading of Greek verse according to Latin accentual rules allows easy perception of the underlying quantitative rhythm. Indeed, read according to the Latin accent, most lines in Homer end with the characteristic coincidence of verse ictus and stress that helps establish the “beat” of a Latin hexameter.

Never adopted elsewhere, and abandoned in America, the system is now confined to Britain, Holland and those Commonwealth countries that have any interest in classical studies. Even here, it is under attack, on the grounds that there is no evidence that the Greeks ever used it.[4]

The third system is to try to recover the sound of the ancient tonic accent, just as the ancient vowel sounds have been recovered. If this were possible, it would be a complete solution to the problem of accentuation. However, it is generally regarded as not possible. It may be possible to recover the accent of individual words, but not to recover what W. Sidney Allen calls the “tonal syntax”—this being the way in which tonal patterns interacted with each other in continuous speech. I have heard several recordings of Greek verse allegedly using a tonic accent. The result for me is a stress accent overlaid by squeaks and trills—half comical, half ugly.[5]

Of course, no argument from perceived ugliness can be decisive. Much perception of beauty is determined by cultural expectations; and just as the Victorians had to give up their belief in the chaste whiteness of Greek marble when traces of colour were found on the Parthenon friezes, so too would it be necessary to revise opinions of these recordings if they could be shown to be true renderings of Greek verse. However, there is no reasonable evidence that they are true renderings.

The fourth system is to respect the written accents as indications of tone, but to use a statistical analysis of verse to find if classical Greek had a stress accent autonomous of tone. The most distinguished modern efforts in this direction have been made by W. Sidney Allen. He believes that Greek was stressed on the final long syllable in every word, or on the penultimate if the final was short, or on the antepenultimate is the penultimate was short, with perhaps a secondary accent earlier in the word according to the same scheme. Following this rule, he argues, there is an overwhelming coincidence of verse ictus and stress. In short, he turns the Byzantine system on its head. Rather than shift the stresses in verse to fit the metre, he wants to shift the stresses in prose to fit those he claimed he had discovered in verse.[6]

The advantage of this system is that it avoids the double accentuation of the Byzantine scheme, while making sense of verse—and while reconstructing the alleged practice of the ancients.

The disadvantages are obvious. There is no clear evidence that the ancient grammarians were aware of this stress, and while they overlooked much that is now obvious to us, they were not deaf. They could hardly have overlooked a feature of their language that largely determined the patterning of their verse. If it did exist, it vanished from Greek so completely that stress was soon confined to the accented vowels—with no observed linguistic fossils to show an earlier and altogether different state of affairs. Moreover, it does not seem to have been noticed by the Romans, who having a stress accent of their own, ought surely to have noticed one in Greek. Had there been one, it would have been natural for them, when they began to imitate Greek models, to have written verses like

Spargens humida mella soporiferumque papauer
(Aeneid, iv.486)

where stress and verse ictus coincide. Instead, they ignored what it is reasonable to suppose for them would have been the most obvious key to Greek pronunciation, and wrote verses where stress and verse ictus hardly ever coincide.

The fifth system is simply to ignore the problem, and give students no advice on how to accent classical Greek. This is the general rule in school textbooks and home study guides published in England. For example, Abbot and Mansfield lay down elaborate rules for the placing and nature of the written accents, but are silent on how to pronounce them.[7]In their introductory chapter, B.R. Reed and Margaret E. Jervis dismiss the problem as follows:

[T]here is some difference of opinion as to how exactly accents affected pronunciation…. [P]itch accent is not easy to achieve, and beginners need not worry unduly about accents….[8]

L.A. Wilding says that he will insert the accents, “but their use may be put off to a later stage”.[9]F. Kinchin Smith and T.W. Mellish do not even insert them: “The writing of accents in Greek is a conservative tradition from which we might with advantage break away”.[10]Perhaps the intention here is to leave students to adopt the Latin rules of accent—though there is no explicit advice for them to do so.

These, then, are the various systems. All are seriously defective. We do not know how classical Greek sounded, but we can be reasonably sure that it did not sound like any of these. Nevertheless, excepting the last, which is no system at all, one of them must be chosen—unless we are to read Greek with as much attention to its sound as we look for in the instruction manual to an oriental pocket calculator. In the rest of this article, I will suggest in outline which system might be regarded not as the most correct, but as perhaps the most “authentic”.

A starting point is to ask how the Romans pronounced Greek. For the most part, it is unlikely that they pronounced it just as the Greeks did. Learning a foreign language is quite easy. Speaking it with no trace of one’s own accent is a very rare achievement. This is especially the case when the two languages in question have large differences of sound. The English and French, for example, have lived as close neighbours for over a thousand years. Whether in commerce or diplomacy or war, or in social mixing at all levels, the two nations have been indissolubly connected. Each has had a profound cultural influence on the other. Yet the speech habits of each nation have made a perfect spoken grasp of the other’s language almost impossible. French attempts at speaking English have been a standard joke in England since time immemorial. The French similarly laugh at the English. One of the main causes of difficulty has been the different kind of accent in each language. English has a powerful stress accent. Most kinds of French have not, but instead have a weak tonic accent. This being so, it is reasonable to suppose that the Romans, whose accent differed from the Greek much as the English does from the French, would not, with few exceptions, have been very Hellenic in their pronunciation of Greek.

It is doubtful if even the Roman critics understood the difference between the Latin and the Greek accent. Quintilian, for example, discusses accent in Latin wholly as if it were one of tone, writing confidently about acutes, graves and circumflexes—when the philological evidence indicates that the Latin accent was at least overwhelmingly one of stress, and that analyses copied from the Greeks are inapplicable to it.[11]

It is possible, moreover, not merely that the very learned misunderstood the evidence of their own ears, but that many other Romans did not even hear the Greek accents in a meaningful sense. We can see this in the treatment of words borrowed from Greek into Latin. There was a strong tendency to Latinise them—not just in spelling, but also in accentuation. Thus in  the Greek κάμηλoς, the accent was shifted to the long penultimate camElus –  just as κατάληψις became catalEpsis, and θεάτρον became theAtrum.

Significantly, there is evidence that, towards the end of the first century AD, this custom began to die. In two passages from book i of his Institutes, Quintilian—who can be trusted to know the place of an accent, if not its nature – says:

Sed id saepius in Graecis nominibus accidit, ut “Atreus” [‘Atreύς] quem nobis iuvenibus doctissimi senes acuta prima dicere solebant, ut necessario secunda gravis esset, item “Nerei” “Terei”que….

Inde Olympo et tyranno [Όλυμπος, τύραννος] acutam syllabam mediam dederunt, quia [duabus longis sequentibus] primam brevem acui noster sermo non patitur [12]

The natural implication is that while, in the past, the accent on these Greek words had been shifted to follow the Latin rule, the Greek rule was now being followed. Why should this be? It is unlikely that the Romans at the end of the first century AD were better educated or more fastidious in their learning of Greek than they had been a century or so earlier. More likely is that the tendency to keep the Greek accent was due to the growing change in Greek from tonic to stress accent. Just as the going out of a tide reveals contours on the sea bed previously unseen by an observer from the land, so the transformation of the Greek accent might have revealed a patterning unheard by most earlier Romans. Though he is making a different point, Ernst Pulgram would agree:

It is possible to regard the equation κάμηλoςcamElus as demonstrating that the Latin borrower paid attention only to the long vowel in the penultimate syllable, which of course he was then obliged to stress. The question as to whether the speaker of Latin even heard the accentuation of the first syllable of  κάμηλoς cannot be answered on this evidence alone. Quite possibly he could not; for it is known that the proper perception and identification of a linguistically significant pitch by a linguistically untrained person whose own idiom does not operate with pitch, is a difficult if not impossible achievement.[13]

Now, if the Romans before the end of the first century tended to Latinise the accent of individual Greek words because they could fully hear the Greek accent, it may be reasonable to suppose that they did the same in their pronunciation of continuous prose in Greek. There is, of course, no direct evidence that they did. Nevertheless, if many Romans did pronounce Greek according to the accentuation rules of Latin, it might explain the rather odd tension in their verse between quantity and spoken stress.

Nothing is certain about how the Romans recited their verse, but, to elaborate on a point made above, there is a rebuttable presumption that they recited it with the stress accent of each word unshifted, except as may have been required by the elision of open syllables. The consequent stress rhythms then became the perceived rhythms of Latin verse. For many uneducated people, and for many after the quantitative structure of Latin began to break down, these rhythms could even be detached from the underlying structure, so that unscanning verses could be written that still sounded correct. See, for example, the verses of Commodian. Again, see the mediaeval sapphic quoted by L.P. Wilkinson:

Terra marique victor honorande
Caesar Augustus Hludowice, Christi
dogmate clarus, decus aeui nostri,
spes quoque regni….[14]

Why did they do this? One possible reason is that this is how they recited—and perhaps heard—Greek poetry. And so, when they began to write quantitative poetry of their own, they tried to reproduce not just the metres of Greek but also an accentual rhythm that, while not really there, they had imposed on it in an attempt to make sense of its pronunciation. If so, they would have been in the same situation as generations of English-speakers who, confronted with a passage of French verse, have imposed patterns of stressed iambi and dactyls and so forth that most French-speakers would deny are present in the texts being read. They did not exactly reproduce what they may have heard. Latin has fewer short syllables than Greek, and so their own verses had more spondees and a heavier sound. They also took the above average tendency for the end of Greek hexameters to end with a coincidence of Latin accent and verse ictus and regularised this into an almost universal rule.

Some evidence for this—not strong, I accept—can be seen in the imitation in Latin verse of Greek rhythms that emerge from a Latin accentuation. Take, for example, the opening words of the Iliad and the Aeneid: μῆνιν ἄειδε θεὰ – Arma uirumque cano. Or take the coincidence between several lines of Theocritus and Virgil quoted by L.P. Wilkinson, where it can be argued that Virgil is echoing a rhythm imposed on the Greek by a reading according to the Latin rules of accentuation:

Pήpok__r__sθ__ka Dάψniς_tάketo p_poka Nύm_ai

Namneque parnasi uobis iuga, nam neque Pindi[15]

It may be asked why this mistake – assuming it was ever made – was allowed to continue unchallenged throughout the whole classical period of Latin literature. Even granting that the early Roman poets had a bad ear for Greek, there were by the 2ndcentury BC Romans who were as careful in their Greek as some English-speakers have been in their French. Why did these not intervene to correct what they might have seen as a vulgar misunderstanding of the true nature of Greek? There is an answer here that, while unprovable, may have some persuasive force.

There are many instances where a practice is mistakenly imitated from abroad, but which, once established, becomes a custom in its own right, unshaken even when the mistake has been commonly realised. Take the use of the saucer in the English tea service. The Chinese do not use their saucers to support teacups, but to cover them. But the English saucer has its fixed use, and has never been designed for any other use. Again, take the Indian custom of mixing tea, milk and sugar and then boiling them together. This seems to have arisen from a misunderstanding of the English custom, though is now a custom in itself. The same may be true of Latin verse. Almost nothing from the early period has survived, and even less criticism has survived from before the 2ndcentury AD, and so we cannot follow its development more than conjecturally. But it is conceivable that the earliest experiments in quantitative verse were based on a corrupt pronunciation of Greek. Though famously deferential to all things Greek, the Romans were also proud of their own literary achievement, even before the classical period gave them just reason to be proud. Once established, the accentual patterning of their verse would have become part of their literary tradition. Undoubtedly, it survived the Greek transformation from tonic to stress accent, after which the Greeks began to recite their verse with the accent displaced. Equally, the Greeks do not seem to have imitated Roman practice after the transformation, when it might have made sense to continue writing verse with an accentual structure that did not obliterate quantitative rhythm: they had their own customs and made the necessary adaptations without regard to foreign models.

All this being so, the English system of pronouncing Greek according to Latin accentual rules may not after all be that unauthentic. It is not as the classical Greeks pronounced their language—but, as said, there is no other system that reproduces the classical accentuation. Even so, the English system may reproduce how many Romans heard and accented Greek. It may therefore be claimed as the most “authentic” system on offer.[16]


[1]Edward Gibbon, History of the Decline and Fall of the Roman Empire (1776-87), chapter lxvi.

[2]The whole question of how Latin verse was intended to sound is discussed fully and persuasively in in L.P. Wilkinson, Golden Latin Artistry, Cambridge University Press, Cambridge, 1963.

[3]Vossius, De Poematum Cantu et Uiribus Rhythmi, 1673; Henning, Dissertatio Paradoxa, 1684 – both cited by W. Sidney Allen, Vox Graeca, Cambridge University Press, Cambridge, 2nd edition, 1974, p.151. According to Henning, “ergo ut Latine pronunciamus ita et Graeci erit pronunciandum”.

[4]“The ‘Henninian’ pronunciation (still in use in England), which uses for Greek the same rules of accentuation as for Latin, ought to be rejected”, Anna Morpurgo Davies, article on “Pronunciation of Greek”, The Oxford Classical Dictionary, (3rd edition), edited by Simon Hornblower and Antony Spawforth, Oxford University Press, Oxford, 1996, p.1255

[5]Justifying his own dismissal of these recordings, W. Sidney Allen comments:

My objections to attempting a tonal rendering are in fact based less on considerations of ‘perfectionism’ than of practicality and fluency. Since we know so little about Greek tonal syntax, we should either have to treat each word as an intonational isolate (‘one-word sentence’), with the unconcatenated effect that this engenders, or to make arbitrary guesses which would differ from scholar to scholar and school to school. The latter alternative seems evidently undesirable, and the former produces results that bear little resemblance to any tonally accented language that I have heard. (W. Sidney Allen, Comment attached to W.B. Stanrod, “On the Pronunciation of the Ancient Greek Accents”, Didaskalos, 1968, 152.)

[6]W. Sidney Allen, Accent and Rhythm: Prosodic Features of Latin and Greek: A Study in Theory and Reconstruction, Cambridge University Press, Cambridge, 1973, pp.274-334. This is by no means the first attempt to discover an autonomous stress in Greek from analysing the arrangement for words in verse. C.W.E. Miller in 1922 cites Hanssen from 1882 with similar rules for placing an “ictus” in Greek. See Miller, “The Pronunciation of Greek and Latin Prose, or Ictus, Accent and Quantity in Greek and Latin Prose and Poetry”, Transactions of the American Philological Association, 1922, p.186.

[7]Evelyn Abbott and E.D. Mansfield, A Primer of Greek Grammar, various editions since the 19th century.

[8]B.R. Reed and Margaret E. Jervis, Lampas: A New Approach to Greek, Basil Blackwell, Oxford, 1970, p.3.

[9]L.A. Wilding, Greek for Beginners, Faber and Faber, London, 1957, p.17.

[10]F. Kinchin Smith and T.W. Mellish, Teach Yourself Greek, Hodder and Stoughton, Sevenoaks, Kent, 1947, p.xviii.


Apud nos vero brevissima ratio: namque in omni voce acuta intra numerum trium syllabarum continetur, sive eae sunt in verbo solae sive ultimae, et in iis aut proxima extremae aut ab ea tertia. Trium porro de quibus loquor media longa aut acuta aut flexa erit, eodem loco brevis utique gravem habebit sonum ideoque positam ante se, id est ab ultima tertiam, acuet. Est autem in omni voce utique acuta, sed numquam plus una nec umquam ultima, ideoque in disyllabis prior. Praeterea numquam in eadem flexa et acuta, +qui in eadem flexa et acuta+; itaque neutra cludet vocem Latinam. Ea vero quae sunt syllabae unius erunt acuta aut flexa, ne sit aliqua vox sine acuta. (Quintilian, Institutio Oratoria, i,v.xxix-xxxi – available on the Internet at

 Helmut Lüdke comments on such passages:

The Latin grammarians were not just bad phoneticians, they were no phoneticians at all; they were not even good phonologists, but continuously confused phonological and orthographical evidence; besides, they of course lacked the diachronic perspective, and on top of all that they allowed themselves – gravest error of all – to be enticed by the example of Greek grammarians to make statements which with respect to the Latin language were utterly senseless. (Die Strukturelle Entwicklung des romanischen Vokalismus, Bonn, 1956, translated and quoted by Ernst Pulgram, “The Accentuation of Greek Loans in Spoken and Written Latin”, American Journal of Philology, 1965, p.143).

[12]Quintilian, op. cit., i,v.xxiv-xxv, lxii.

[13]Pulgram, op. cit., p.145.

[14]Wilkinson, op. cit., p.108. For a modern example of an hexameter rhythm detached for any underlying quantitative structure, see the opening lines of T.S. Eliot’s Dry Salvages:

I do not know much about gods, but I think that the river
Is a strong brown god – sullen, untamed and intractable,
Patient to some degree, at first recognised as a frontier;
Useful, untrustworthy, as a conveyor of commerce
Then only a problem confronting the builder of bridges.

This sounds much more like a Latin hexameter than most attempts at correctly reproducing quantitative verse in English. See, for example, the opening lines of an attempt I made in 1987 at a comic elegy:

 O weep for Liberace, who born to the ivory keyboard,
now to the grave is borne, though to reside in heaven.

[15]Wilkinson, op. cit., pp.194-95.

[16]This conclusion is unaffected by the statistical attacks cited by W. Sidney Allen – Accent and Rhythm, p.173 et passim – on the lack of coincidence between stress pattern and verse ictus in Greek verse read according to the Latin accent. Since we do not normally read Latin verse according to the verse ictus, it is only the stressed rhythm that is important here.