Boiled down and oversimplified there are two schools of thought, and there are good reasons to support either of them. One believes that the recitation of poetry occurred within regular intervals of time; the other that poetry was read “prosaically” that is, as if one were speaking them unrestrained by an arbitrary meter, pausing at the end of complete ideas rather than at the end of verses but whose metrical pattern could nevertheless be easily tracked by the moraically tuned ear of the ancients. The latter of these two schemes likely has the majority vote in most countries[1], I think especially since it allows for more freedom in interpreting the text that would otherwise be bogged down by artificial impediments.

My purpose is to come to a better understanding of how the ancients typically exercised their poetry, and I am inclined to think that, at least in public recitations, there was likely an expected formula, which, if left unobserved, resulted in the ceremonious expulsion from the stage, expulsions which Cicero has given us reason to believe were not a totally rare occurrence[2].

The attestation of the public’s reception of recitals is a key ingredient in our understanding of ancient poetry. The fact that the uneducated public was capable of, at least sometimes, so clearly discerning the moraic meter that they would explode anyone so uncultivated in their craft as to commit an error gives the impression that the meter was sufficiently expressive. I am almost willing to grant that it was possible for the ancient public to be this pedantic in the midst of the “prosaic” recitation of dactylic meter that I mentioned above, but the consideration of meter that was much less regular and presented on a public stage immediately obfuscates the issue. Is it possible the that fickle iambic senarii of Terence and Plautus, which were delivered upon the public stage, were still subject to the same scrutiny as the regular and predictable dactyls and spondees of heroic verse? Perhaps. I am not ready to make any pronouncements in this regard. Though it is interesting that the iambic senarii of Roman comedy were the only metrical form in comedy to which there was no instrumental accompaniment[3].

Some have argued that the Roman ear was more tuned to catch the subtle differences between heavy and light syllables than ours, whose are more apt for hearing stress. I have to imagine that this is true to an extent, but the hangup for me is on the issue of time or tempo.

It seems to me impossible, regardless of how tuned one’s ears may be for discerning the weight of syllables, that the public could effortlessly detect minor metrical errors if verse were pronounced no differently than if the same sentences had happened to fall into the mouth of one speaking with a companion in the forum, pausing at the end of complete ideas and clauses and without the confines of an arbitrary meter. Since syllable weight[4] is ultimately an issue of time, how could the Roman in the midst of a prosaic recitation possibly catch the differences between something like:

[5]causa pedestris (dactyl spondee)

audit amaras (dactyl spondee)

audit causas (spondee spondee)[6]

It is possible and maybe even likely that the second example, spoken prosaically at a certain rate of speed could sound like a “dactyl spondee”: since the first ‘a’ in amaras would be absorbed, so to speak, into the single cluster (au-di-ta-ma-ras), but it is also very possible that a pause fall between those two words, a pause which would obscure a prosodic characteristic that is nearly without exception in poetry, even if it occurs at the end of complete thoughts or between changes in speaker, where there would naturally, that is in a prosaic discourse, be breaks in time. With the pause, the listener would hear a spondee (in so far as the final syllable becomes heavy by its position before a pause) and an iamb followed by a long syllable.

Similarly, in an iambic verse, how could anyone, even a trained musician pay close enough attention to catch when a playwright has, instead of an iamb, used the temporally equivalent but metrically inappropriate trochee? For example, Micio’s iambic and sufficiently elided verse:

Aliquid. Vah quemquamne hominem in animo instituere aut

could have been written in this way,

Aliquid. Vah quemquamne mortalem in animo instituere aut

The same amount of time is taken to pronounce the two.

If one were to take up the task of reciting something from the dactyls of Horace, dactyls that have been noted for the discrepancy they exhibit between the end of a verse and the end of a sentence,[7] but were to pronounce it without any strict regard for metrical shape, he would undoubtedly pause (a mora), where he might detect the end of a complete thought (a sentence). Add to this the simple pauses that we make between words, clauses etc. in the act of more vehement or dramatic expression, and you would have, I think, a very confused-sounding jumble in terms of audible meter, but one that if visually scrutinized according to the customarily agreed-upon vowel length for each individual word, would still, in fact, be considered “dactylic hexameter”.

Perhaps, in Horace’s case, a conceptually real but rhythmically unmanifested pattern is all that is required. It seems possible that poetry could be written that was not intended to be recited along in a metrical tempo but to be read quietly to oneself[8]. In the act of reading a text with the eyes, the customary knowledge of syllable weight could be sufficient in recognizing the intrinsic dactylic hexameter.

To make what I am talking about a little more clear, take this example from Horace’s Epistulae:

I bone, quo virtus tua te vocat, i pede fausto,

grandia laturus meritorum praemia. quid stas?’

post haec ille catus, quantumvis rusticus, ‘ibit,

ibit eo quo vis qui zonam perdidit’ inquit.

The editor has added punctuation to mark where one idea stops and the next begins, places where we would likely pause in speaking. A pause at the marked punctuation would disrupt a fixed tempo, signaled by a “sublatio” and “positio” as Quintilain calls them, but neglecting the pause results in a very jumbled meaning. We have a dilemma.

This kind of discrepancy between verse ends and clause ends is frequent in Horace, but in Vergil, the end of a line is also typically the end of a sentence or a clause, a construction for which it is much easier to conceive a continuous tempo-controlled recitation. Take these lines as an example:

Talia perstabat memorans fixusque manebat.

nos contra effusi lacrimis coniunxque Creusa

Ascaniusque omnisque domus, ne vertere secum

cuncta pater fatoque urgenti incumbere vellet.

Abnegat inceptoque et sedibus haeret in isdem.

Each line of poetry contains a complete idea or a self-sustained chunk of an idea without bleeding into the next verse. Vergil intentionally structures his sentences in this way, and only breaks the pattern to draw attention to his doing so, such as in this example where he uses enjambment to highlight Sinon’s fatal lie:

‘Cuncta equidem tibi, rex, fuerit quodcumque fatebor

vera’, inquit ‘neque me Argolica de gente negabo’

This may indicate that these two poems are two different genres of poetry, intended to be consumed in two different ways.

In Catullus’ Carmen 85, there are three distinct expressions that we would classify as complete sentences, comprehending a subject and a verb: 1) odi et amo. 2) Quare id faciam fortasse requiris. 3) Nescio sed fieri sentio et excrucior. If one were to put forth these three sentences in an everyday conversation, he would naturally pause to distinguish each line of thought with one of the pauses likely sustained longer than the other. If one were being more dramatic, there would likely be included intentional pauses between some of the conjunctions. This is how we naturally convey ideas. However, were one to do the same in the recitation of poetry, I believe that, though not totally obliterated, the underlying metrical pattern is by no means obvious, especially considering the fact that ‘nescio’ could be trisyllabic in some instances and dissyllabic in others (or, if you prefer, contains a short o in some situations and a long o in others).[9] Perhaps it is more clear with dactylic poetry, but if we are reading the lines as if in an everyday conversation, the sentences, that is the trains of individual ideas rather than the verses provide the barriers within which we could potentially sense, or expect to feel, recurring patterns. I believe that if one were listening to Horace’s dactyls read to you in a natural way of speaking, pausing at internal punctuation and not at verse ends, they would have to actively count feet to know exactly when to expect the metrically necessary dactyl and spondee ends, something that I don’t think your average Roman spectator would either know to do or care to do. Speculation, perhaps, but not without its grounding in a common-sense understanding of human nature and the nature of the general public with regard to matters of music.

It seems to me that the recitation of at least certain types of poetry in specific settings must have been sufficiently distinct from quotidian speech that metrical errors were obvious. One way to make the meter more obvious is to divide each verse into consistent intervals of time, a practice that makes sense considering a few facts which, for the sake of keeping this post within 4,000 words, will only mention in passing here: First, instrumental arrangements, which demand some regulation of time, frequently accompanied the poets of the ancient world. Second, Quintilian makes several statements in the Institutio from which we can infer that poetry was confined to specific intervals of time and not to the arbitrary taste of the one reciting[10]. Finally, The regular application of elision in all enviornments of poetry[11] also leads me to believe that an external time constraint forced at least some of the elision where it may not have necessarily been practiced in every-day speech.

Some have proposed, considering the musical ignorance of the audience, that the meter must have been made even more manifest by not only restricting the verses to consistent intervals of time but also by the intensification and rearrangement of each individual word’s accent such that, at least in dactylic poetry, it can be heard at the beginning of each metrical foot. Such a recitation certainly clarifies the meter, especially if it is recited with cadence and regularity, that is, in rhythm:

Árma virúmque canó Troiáe qui prímus ab óris

And this is the way, at least in the United States, that your average teacher is going to recite these lines for their students.

Quintilian seems to resolve the whole question for us by citing an example in the Aeneid wherein the accent “volucres” must be shifted from the antepenultimate syllable, where it naturally lies, to the penultimate, lest the final two syllables make an iamb instead of the metrically required spondee.[12] However, this shift in accent must have also included a lengthening of the syllable (a lengthening that he does not mention) in order for the final two syllables to form a true Latin spondee. It is well documented that in dactylic poetry, specifically in Vergil, metrical ictus, that is, where we would expect an accentuated syllable at the beginning of each foot, and natural word accent very seldom coincide.[13] If we are under the assumption that we must move a word’s natural accent to coincide with the metrical ictus, it seems peculiar that with multiple shifts of the natural word accent occurring in almost every line of Vergil’s poetry, Quintilian would cite an example of shifting the accent which also includes a lengthening of the syllable, a not-so-common phenomenon. It may not be a direct statement, but I think that it is safe to assume that at least Quintilian does not visualize the continuous manipulation of natural word accent in poetic recitations.

Moreover, he makes the claim that in the second half of the Aeneid’s first line only one accent is actually heard, “Troiae qui primus ab óris”.[14] This is likely due to the fact that when speaking, we often string words together in such a way that they could be interpreted as single sounds, and therefore pronounced not as individual words but as one long word. This makes sense considering the way that we divide poetic syllables in Latin, estimating a strand of words like a single sound to be divided.  

The fact, too, that Quintilian uses words like ‘digitorum crepitus’ and ‘sublatio’ and ‘positio’[15] indicating that a regular time was kept and perhaps even audible in certain kinds of recitation leave, in my mind, little room for the idea that the ancients publicly read poetry “dramatically” with pauses and intensification left to the whim of the reenactor. How could any absolute standard have possibly been applied?

This is where I start to speculate a little more based on the admittedly shaky concept I have of the history of poetry, so please take what I have to say below with a grain of salt. Today, we tend to think of poetry as a personal and more individual experience, consumed quietly and without ornament. It may be recited out loud in an academic setting or among small pockets of like-minded friends, but for the ancient Greeks poetry was, at least originally, a much more public display that was almost always, if not always tied to instrumental performance. From what I understand, rhapsodes would take three whole days during certain festivals to recite a Homeric work. Theater took place on a stage in front of an audience. The rest of the Greek poetic corpus seems to have been composed to accompany music (or vice versa). How exactly these performances took place, I am not sure. Roman poetry, however, as an imitator of Greek poetry, was not always made for this public display, nor was it always accompanied by music.[16] Certain public musical performances remained like those that were acted in the tragic and comic theater, but it seems to me that the Romans were one of the forerunners in creating a more personalized experience of poetry, poetry that was designed to be consumed by the individual at home, not by the public crowd.

The language of the Romans, like the Greek’s, admitted poetry that was based on the weight of each syllable. However, unlike Greek, Latin did not, as far as we understand, natively utilize a pitch accent, a kind of accent which I assume adds a degree of musicality and grace to a poetic performance. As imitators, the Romans were capable of rhythmically achieving the same types of poetry that Greek could, but as possessors of a stress accent, it was not linguistically equipped to exercise poetry in the exact same way. The inclusion of stress accent as opposed to pitch accent changes the shape or expectation of Roman poetry, since we naturally expect stress to coincide with an underlying tempo, an expectation that presumably you do not have to the same degree with changes in pitch.

So, from what I gather from all of the aforementioned evidence, we are left with the following general requirements. We need a recitation of poetry that simultaneously maintains a word’s or a string of words’ natural accent, and is also moraically clear enough that the general population would be able to easily detect errors that they can hear, not see. It seems sufficiently likely, too, that meter was accompanied by some kind of time-keeping artifice. Whether that artifice was only visible, or if it was also made audible by, say, the regular beat of a drum or a snapping of the fingers, is unclear to me. I can only conclude that public recitation of poetry must have more or less maintained a word’s natural accent while also under the duress of rhythmic time. This combination addresses both of the problems above but creates a new albeit superficial one, namely one of unornamented dryness.

There is little room for emotional display, as the cohibition of the meter restricts the actor to maintaining a tempo within a beat and does not permit him to linger in affectation. The recitation that I posted online was, admittedly, drier than it needed to be, but the limitation remains, and I think that with a potential musical accompaniment, or by reciting by singing with changes in pitch to imitate music, the recitation could become more colorful without damaging the meter.

I don’t take any issue with those who would choose to recite classical poetry in a way that is more palatable to our modern ears. In fact, I too recite poetry in this way. Latinitium just released a reading of Ovid’s account of Pyramus and Thisbe that I think does a great job with this kind of reading. The presence of a clandestine metrical artifice that accompanies a more natural delivery of speech or a plain reading of script is clever and intriguing. It is likely that this kind of recitation existed in the ancient world as well, and I suspect that Horace expected at least some of his dactyls to be recited or read in this manner.

Erasmus, who was one of the first to invigorate interest in restoring the native sounds and syllable measures to Latin, and whom I presume to have been much more knowledgable than myself in these matters, wrote a colloquy called the “Impostura” in which one character promises to trick the other without his notice. The “trick” is that, unbeknownst to his friend, he had been speaking in meter since the beginning of the dialogue. Erasmus’ concept of meter, at least in this instance, is one that may not have relied upon fixed intervals of time, otherwise the “trick” would have been obvious. Erasmus mentions another similar example in his treatise “de conscribendis epistolis” in which he played a similar “trick” on a friend by writing a letter to him in verse and promising that he would not be able to detect this “trick” before the end of the letter. Quintilian, too, acknowledges the metrical patterns of speech delivery that go unnoticed, but that provide rhythm and flow to an oration.

No matter where you fall on the issue, at the bare minimum we know that ancient poetry relied upon the relationship between long and short syllables. So long as those are maintained in some way, we are closer to the way in which the ancients recited their poetry.

Our longest post yet. Hope you enjoyed .

             – Rogerius

[1] Depending on whom you are asking. In my experience in the U.S., the former is more common in primary and secondary schools. I am not sure what the consensus is among scholars.

[2] C.f Cic. De Orat. iii.196; id. Or. 173; id. Paradoxa. 3.26. I don’t want to overestimate the ability of the masses for intuiting such metrical distinctions. We have reason to believe that it happend somewhat regularly based on Cicero’s mention of it, but I am not sure we can jump to the conclusion that the crowd was booing actors off the stage because of minor errors in all genres of poetry, only where it may have been obvious. A single but noticable hicup may have been enough for expulsion. However, as I will touch on a little later, if the actor or recitor were constrained to a beating time, errors in pronunciation would be much more glaring. I have heard that some Roman author mentioned how much more difficult it was to hear iambic poetry than dactylic, but the quote eludes me.

[3] See R.H. Martin’s introduction to his commentary on Terence’s Adelphoe.

[4] Allen prefers to talk of vowels as long or short and syllables as heavy or light, since only vowels can actually be made longer. Consonants are distinguished as long by the length of a stop or pause.

[5] The following examples are arbitrary. Someone could replace them with words you would be more likely to hear in a discourse and between which there was more likely to occur a full stop.

[6] C.f Allen; “There is nothing to suggest that the untrained Roman ear would have appreciated the quantitative (as opposed to phonemic and semantic differences between, say, tange and tangent, or the quantitative equivalence of the latter to tangēs.” Rhythm and Accent pg. 339.

[7] C.f. This discrepancy was first brought to my attention by Sellar’s commentary, a book that I do not have on hand at the moment for an exact page reference: Sellar, W.Y. The Roman Poets of the Augustan Age: Horace and the Elegiac Poets. London: Oxford University Press Amen House, Oxford at the Caredon Press.

[8] I was first introduced to the idea of poetry being read rather than recited by A. J. Woodman’s introduction to his commentary on Horace’s Odes, book 1.

[9] Most internet sources reckon the final -o in “nescio” short by brevis brevians. I don’t have a physical commentary on hand to see how scholars scan this word, but I think it more likely that the three syllables were compressed by synizesis, rendering i a semi-vowel. But this is off topic.

[10] Take, for example, his statement that even children understand that a long syllable is two beats of time and a short syllable one. Inst. Or. 9.4.47

[11] The most interesting being the changes in speaker that occur in the middle of a verse in Roman comedy.

[12] Quint. Institutio Oratoria, i, 5. 25-29

[13] Allen cites many sources for this claim in Rhythm and Accent and asks the following question on pg. 340: “Given, then, the extent of disagreement between accent ad strong position in the first part of the Latin hexameter, how could our unsophisticated hearer be made aware of any pattern in it?”

[14] Same as above: Quint. Institutio Oratoria, i, 5. 25-29

[15] Quint. Inst. Or. IX,4.48. See also Horace Ode iv, 6. 35-36: “Lesbium servate pedem meique/pollicis ictum”.

[16] See footnote 8.