Notes for classicists

When I first began developing my style of reciting Greek, in the mid-seventies, the main authority on the sound of spoken Greek was Allen’s Vox Graeca. More lately we have Devine and Stephens, whose impressive Prosody of Greek Speech has made me reconsider many things, though not as yet to significantly alter my practice. I will list here the main ways in which that practice differs from what others have done or recommended, and briefly discuss my approach to some of the points in question.

1. General philosophy: There are people who can acquire perfect pronunciation in languages not their own, but not in the absence of living models. Short of the discovery of ancient voice recordings, our pronunciation of ancient Greek is going to remain “foreign” no matter what we do. What, then, should be our criteria for a “good” pronunciation? Think of foreign accents in English: there are some that do not prevent a foreigner from speaking clearly, eloquently, and with good rhythm, while there are others that reduce the simplest greeting to pure babble. Throughout my practice I have assumed that clarity, good rhythm, and an intonation that is expressive without emphasizing the wrong syllables, are far more important than the exact shade of each vowel and consonant. To be sure, these things interact – everything in pronunciation, more or less, interacts with everything else. Individual phonemes must obviously be such as to yield the correct quantities. They should also, so far as possible, be clearly and consistently distinguishable. But this still leaves a lot of leeway. In any widely spoken language, after all, there is tremendous variety even in native pronunciations.

But how does one really judge rhythm? A line of poetry, as regrettable numbers of students can attest, does not necessarily feel rhythmic just because it is correctly scanned. Are there objective standards for what is expressive? Can we measure eloquence?

Reciters will inevitably be guided by their own feelings of rightness, subjective to a degree of course, but by no means entirely so. The voice, after all, is a physical thing, and quickly gives its own measure of what is workable and what is not. If certain articulations, certain vocal habits, seem to increase fluency, stamina, ability to memorize, maintain rhythm, and vary expression, while others just as clearly do the opposite, the reciter can hardly be faulted for trying to cultivate the former and shed the latter, no matter what the historical evidence may seem to say.

None of which guarantees that every listener will find the result either eloquent, expressive, or rhythmic. Those who don’t are free to develop other styles. One thing we need not try to do, given both the scale of our ignorance and the nature of the art, is confine the “good” to a single, approved type.

2. Pitch excursions: I make no attempt to confine accentual rises in pitch to any particular interval. The statement of Dionysius of Halicarnassus that appears to specify the interval of a fifth (De Compositione Verborum, xi) has caused a lot of trouble, in my opinion, and is partly to blame for the sing-song quality of many “reconstructed” pronunciations. Allen’s interpretation of the passage as referring to a sort of mean or average (VG, p. 120) seems plausible to me, and Devine and Stephens I think largely concur (PGS, pp. 171-172).

There are deeper reasons, though, behind the tendency to “sing” the accents. English speakers are not very aware of how we are using pitch in our everyday speech. What we call a stress accent is often thought of in terms of loudness, though a little experimentation will show that it is almost entirely a matter of pitch and duration, reinforced, in English at least, by the tendency to reduce or neutralize all unaccented vowels to some degree. Loudness as an independent factor in addition to these doesn’t really come into it. Were you to somehow artificially make all unaccented syllables louder than accented ones (say by audio processing of recorded speech), while keeping pitch, duration, and vowel quality the same, it would certainly sound odd, but I doubt that anyone would perceive the accents as having shifted. Whereas if you performed the same sort of reversal with either pitch or duration, one certainly would.

What we perceive in a stress accent is emphasis; what is actually happening though, is that (at least) two different kinds of emphasis – pitch and duration – are being used more or less interchangeably. Which one we use for any particular syllable is partly a matter of our language’s prosodic rules, and partly a matter of choice. It’s certainly something that can be quite freely varied for expressive purposes. But this is done unconsciously, and all we really perceive is greater or lesser emphasis on certain syllables, which it is easy – though not at all accurate – to equate with loudness.

The upshot with regard to pitch, though, is that in one of its main uses in our language – its use in accentuation – we don’t really hear it as pitch at all. We hear it just as we hear duration – as emphasis or stress. By and large we only hear pitch as pitch in English when we are focusing on the larger intonational curve of speech, its “melody.”

But in Greek pitch and duration had different functions and were clearly separated. There would have been no tendency to use higher pitch to signal a “long” syllable or lower pitch to signal a “short” one, or to prolong a syllable because it was accented, or shorten it because it was not. All these are things we do all the time in English, and these are the habits we have to unlearn to properly recite Greek. Once we can separate pitch and duration confidently, automatically, most of the battle is won.

Until we have done so, though, there is a temptation to “sing” the accents merely in order to hear them. In practice that usually means trying to place them at an interval that we can recognize as “musical,” such as a fifth, rather than at some less distinctive interval, which our native speech habits will tend to make us classify as “stress” rather than pitch.

In my own practice, I use these less distinctive, “non-musical” intervals as freely as any others to render the accents. Roughly speaking, I require only that an accented syllable represent some kind of perceptible local peak in intonation, which may be very slight, very pronounced, or anywhere inbetween. The actual interval chosen for each accentual rise and/or postaccentual fall will depend on the needs of expression and emphasis and the general intonational curve of the phrase in which it occurs. I find that by this method I can read any Greek verse passage in strict meter, with just about any expression or emphasis I choose.

Naturally, the speech melodies that emerge from such an approach, though different from those of one’s native speech, are still going to be conditioned by them, or by other languages one has actually heard, and are not likely to match the ancient speech melody. This, I would say, is of minor importance. That melody will not be heard again. What we can, and should try to do, is use speech melodies that seem expressive to us, and at the same time violate nothing we know with any degree of certainty about the rules of accent and quantity.

On the matter of intonation over phrases and sentences Devine and Stephens have a good deal to say. For one thing, they give cogent arguments for a general fall in pitch over the course of a sentence (PGS, pp. 435-441). This accords pretty well with my own practice (and with intuition), and is also a feature mostly ignored by reconstructions of the more sing-song variety.

2. Acute vs circumflex: I interpret the circumflex as an acute on the first half of a long syllable followed by a fall in pitch on the second half. An acute-accented long syllable I interpret as the same, minus the fall on the second half. On this interpretation it is admittedly difficult, in certain positions, to consistently distinguish the two, and I don’t claim that I always manage to do so. This doesn’t particularly trouble me. The difference must have been small, though perceptible, as the story of the actor Hegelochus shows. There are of course other views of how these accents differed in contour; all I can say is that doing it this way feels comfortable to me; there are probably other ways of doing it convincingly, but I don’t know that the overall effect, because of that alone, would be dramatically different.

3. Grave accents: When I first began reciting I consistently treated grave-accented syllables as unaccented. Devine and Stephens have argued, mostly from the evidence of musical settings (PGS, pp. 180-183), that the grave represents a lowered, not a totally suppressed, acute. It is fair to say that my treatment of the grave is at present somewhat equivocal. Since it is put where it is by orthographic rule, it seems at least possible that it doesn’t represent anything very consistent as regards intonation. There are times when I feel a grave-accented syllable can take a full, acute-sized rise in pitch; this would be mostly when I wanted to emphasize the word in performance, or to put a slight pause after it. At other times I pronounce the syllable with a slight rise, at still others, with none at all. Considering that my intonation of unaccented syllables is also pretty free, it would be hard for me to give definite rules for how I treat the grave.

4. Vowels and diphthongs: In general I follow the pronunciations recommended by Allen in VG, except for the pronunciation of epsilon + iota, which, after trying for awhile to pronounce as a mid-close long vowel as he recommends (VG, pp. 69-75), I am now gravitating back to pronouncing as a true diphthong. Not because I think that is more correct: I simply prefer it. The important thing, it seems to me, is that this sound, whether diphthongized or not, be clearly differentiated from the more open sound of eta.

Diphthongs traditionally written with iota subscript I consistently pronounce as such.

5. Consonants: Again I follow Allen, but in the pronunciation of theta, phi, and chi, I have gone back and forth. I used to use what I would think of as a transitional pronunciation between the original aspirated consonants and the fricatives they evolved into: theta I would pronounce roughly halfway between English t and th, and phi roughly like f or pf, but placing the upper and lower lips together (i.e. a bilabial fricative) rather than placing the lower lip against the upper teeth. Chi I pronounced as a fricative, more or less like German ch. Though I think all this is defensible, and certainly doesn’t sound bad, I am now moving back towards “true” aspiration of these consonants.

6. Fluency and speed: I have tried to approximate a plausibly “normal” rate of delivery, and to make the language flow like normal speech. Probably the biggest factor here, after becoming comfortable with the pitch accent, is to have internalized a pronunciation of long and short vowels, single and double consonants, and consonant combinations that will automatically yield the correct quantities. One cannot recite properly when one is worried about scansion. Any real sense of rhythm – especially in elaborate meters like those of choral lyric – tends to break down or slip away in a pokey, hesitant delivery. Scansion of course is still indispensable, if only for recognizing metrical shortening and lengthening and clarifying doubtful quantities.

Inadequate internalization also makes it impossible to confidently slow or speed the rate of delivery for expressive purposes, or simply for variety, or to pause appropriately (see the note after the Homer readings), without fear of losing the rhythm.

7. Ictus: As indicated above, one feature of my pronunciation that is liable to seem “wrong” to some is my use of pitch intervals that seem to “stress” the accents, a thing many say must not be done in Greek. The reason usually given is that this somehow disrupts or supplants the true ictus of the verse. To me this seems to result from a confusion. Ictus is not stress, though in English and other stress-accented languages stress is what produces ictus when ictus is present. In Greek, as we know, or should know, this is not the case: ictus arises from a repeated pattern of quantities, or not at all. No mere accent, however prominent, can shift it from its true position, provided the correct quantities are being observed.

In the kind of stress accent Greek eventually acquired, stresses have replaced the original rises in pitch on the same syllables. Or to put it differently, the accents have largely kept their pitch, but also acquired duration. Whatever ictus modern Greek verse has doubtless comes, as in English, from the accents. This would, indeed, seem to argue for some original tension, not yet amounting to conflict, between accent and quantity, which later helped erode the unaccented long syllables and tip things towards a stress accent. To my ears it is this tension that gives real life to the rhythm. The syncopated sound of short accented syllable followed by unaccented long, the very distinct sounds of spondees depending on which syllable is accented, the constant divergence and falling back in line of accent and ictus, these things, far from obscuring the rhythm, give it far greater definition and memorability, making it easier to internalize, to the point where one can be confidently free with it.

Carefully confining pitch rises to “musical” intervals, and trying to avoid any sense of “stress” on the accents, to my mind deflates this tension and results in a much deader rhythm, which decreases fluency and makes the verse much harder to memorize. Without tension, rhythm becomes mere empty pattern, like a dripping faucet.