Pitch and prominence

Two ways to make a syllable stand out. Every language marks one syllable of a word as prominent. In a stress language the marked syllable is pronounced with more force. In a pitch-accent language it is pronounced on a higher note.

A school grammar of 1902 imagined two such languages. In one, every syllable was said on one note, but one of them was louder. In the other, every syllable was equally loud, but one of them was higher. English is predominantly the first kind, and Greek the second.^[1]

The traditional terms follow that division: ancient Greek has an accent but no stress.^[2] Phonetically, in a pitch-accent language the tune of the word is carried by pitch alone, without systematic changes of length or loudness. In a stress language, by contrast, a change of pitch comes bundled with greater length and intensity.^[3]

Stress is made, pitch is heard

Allen’s Accent and Rhythm draws the line in a different place. For Allen, stress is a motor act: a pulse of the chest. The listener judges it by the force the speaker has used, so its criteria lie in the speaker. Pitch is an acoustic effect, judged by the ear, so its criteria lie in the listener.^[4]

What is heard as stress has no simple correlation with loudness. Duration and pitch are often the stronger cues.^[5] The pitch cue is a change of pitch, not necessarily a rise. In Hindi the stress goes with a step down. The English speaker tends to hear the high unstressed syllable as the prominent one, and, when speaking Hindi, to stress it.

Differences between languages in the relation of stress to pitch produce wrong judgements of where the accent falls.^[6] The weak stress patterns of Indian languages, with their unfamiliar pitch cues, are hard for English speakers to recognise at all.^[7]

What English ears do with pitch

English is a stress language in which pitch is one of the main marks of stress. A stressed English vowel is longer and louder than an unstressed one, and usually higher. Raising or lowering the pitch of a syllable is enough to make a listener hear it as stressed. Of the three cues, pitch and length are the strong ones, and loudness by itself is the weakest.^[8]

The consequence for a reader of Greek is direct. The ancient accent was a rise in pitch on one mora, with a fall on the next.^[9] An English speaker who hears that rise, or produces it, hears a stress, because in English a rise is the sign of a stress. And when an English speaker tries to raise the pitch of a syllable, the length and loudness that go with it in English come along uninvited. The accent is not misheard so much as translated.

Length is translated too

In English, and still more in Italian, a long vowel in an open syllable is a stressed vowel. So a word like δείκνῡμι, with its long unaccented υ, is heard by such listeners as stressed on the second syllable. In Ranieri’s judgement, learning to keep quantity and accent apart is the hardest and most important step in acquiring the ancient phonology.^[10]

Conversely, a stress lengthens the vowel it falls on. So it tends to wipe out a contrast of long and short vowels, which is why few languages keep independent distinctions of both. English escapes only because its stress is largely non-phonemic, and so can be laid on a short vowel without lengthening it.^[11]

English rhythm is timed by its accents. Some English syllables plainly take longer than others. Yet English speakers do not register the difference as metrically meaningful. Vowel length tells bit [bɪt] from beat [biːt], but it counts for nothing in English verse.^[12]

Why the distinction matters for metre

The ancient descriptions of the accent speak only of pitch and melody. They never speak of loudness or added length.^[13] That is why the classical metres can be built on syllable length alone and take no notice of the accent. The accent did not lengthen or strengthen anything.

A reader who turns every acute into an English stress is reading a different language. In that language long and short have been replaced by loud and quiet, and the quantitative pattern of the line goes under. The same translation was performed by Greek itself in antiquity, and again by the scholars of the Renaissance.^[14] Whether the pitch can be produced on its own, and how, is the next question.^[15]

Notes

1. ^ Thompson 1902, pp. 7, 427.
2. ^ Probert 2006, p. 8.
3. ^ Devine and Stephens 1994, p. 204; Probert 2006, p. 56.
4. ^ Allen 1973, pp. 76–78, 83.
5. ^ Allen 1973, p. 74.
6. ^ Allen 1973, p. 75.
7. ^ Allen 1973, p. 275.
8. ^ Devine and Stephens 1994, pp. 204–205.
9. ^ Horrocks 2010, p. xx.
10. ^ Ranieri 2021, p. 15.
11. ^ Allen 1973, pp. 81–82.
12. ^ Vandiver 2023, pp. 149, 154 with nn. 45–46.
13. ^ Devine and Stephens 1994, p. 208.
14. ^ See the From pitch to stress tab.
15. ^ See the Pitch without stress tab.

Is it possible at all?

Yes: languages do it. A word’s tune can be carried by pitch alone, with no systematic change of length or loudness on the accented syllable. Where that is so, the accent is not a stress but a pitch accent. Japanese is the usual example.^[1]

A language with a pitch accent may also have a stress pattern of its own, laid out independently of the accent. Bantu languages, Mandarin and others do this. Nietzsche challenged scholars to name a language that keeps force of voice apart from the accent, and the challenge has many answers.^[2] Allen rejects the compromise view on which pitch, stress and duration are inextricably mixed in an accent. He rejects, too, the claim that the truth lies somewhere between pitch and stress.^[3]

Whether classical Greek was of the pure kind is a separate question, and a disputed one. The main views are these:

• The ancient descriptions name pitch and nothing else, and there is no metrical evidence of a stress. Devine and Stephens conclude that the arguments for a stress component have not been accepted. They treat the Greek accent as pitch alone.^[4]
• For Allen the strongest argument is the verse. Classical spoken verse makes no attempt, in any part of the line, to bring accent and metre into agreement. It would have done so if the high pitch had carried a stress. In classical Greek he finds no evidence of any inherent link between stress and pitch. But he thinks Greek probably had a weak stress of its own, placed by syllable weight and unconnected with the accent.^[5]
• Horrocks thinks some increase in loudness almost certainly went with the rise. It was secondary, and not contrastive.^[6]
• Probert holds that the evidence cannot decide between two types. A Japanese-type accent is pitch alone. A Swedish-type accent is pitch with some length and loudness. She leaves the question open.^[7]

What is not disputed is this. A contrastive stress is one that can by itself tell two words apart, as in English content [ˈkɒntent] “what is inside” and content [kənˈtent] “satisfied”.^[8] Greek had no contrastive stress, and its rhythm was timed by syllable weight. That is why its verse counts long and short syllables and ignores the accent.^[9]

Why it is hard

Two things stand in the way of an English speaker. The first is physiological. The vocal folds vibrate faster when the air pressure beneath them rises. So a louder syllable tends to be a higher one, and a higher one a louder one. Singers have to train to send the pitch up while the volume goes down.^[10]

The second is linguistic. In English a rise in pitch is a cue to stress, and stress brings length and loudness with it. So the habits of a lifetime push the three together.^[11] Ranieri notes the same from the classroom. Few learners speak a language that uses pitch independently of stress. It is hard for them to turn pitch into a free device, rather than a mark of the word’s prominent syllable. That is one reason why he chooses, in his own speech, to let a stress reinforce the pitch.^[12] It is a choice, not a finding, and the recordings that follow it carry both.

A reader trained in the Erasmian tradition has a third obstacle. This is a stress on the written accent, or on the Latin penult, learned with the first paradigms.^[13] Allen himself found no convincing tonal recitation. He advised the student not to strive for one, but to stress the written accent while keeping the vowel lengths true. That at least puts the prominence in the right place.^[14] The exercises on the next tab are for the reader who wants to go further.

What to aim for, and how to practise it, is on the Practising the tune tab.

Notes

1. ^ Devine and Stephens 1994, p. 204; Probert 2006, pp. 56–57.
2. ^ Allen 1973, p. 94.
3. ^ Allen 1973, p. 91.
4. ^ Devine and Stephens 1994, p. 208; Probert 2006, p. 57 n. 11.
5. ^ Allen 1973, pp. 261, 274–275; see the Stress and ictus tab.
6. ^ Horrocks 2010, pp. 118, 169–170.
7. ^ Probert 2006, pp. 57, 123–124.
8. ^ Roach 2004, pp. 240, 243.
9. ^ Horrocks 2010, pp. 41 n. 4, 45; Devine and Stephens 1994, p. 208.
10. ^ Devine and Stephens 1994, p. 208.
11. ^ Devine and Stephens 1994, pp. 204–205.
12. ^ Ranieri 2021, p. 26.
13. ^ See the From pitch to stress tab.
14. ^ Allen 1968, pp. 118, 142–144.

Pitch or stress? Our advice

This section is metrehelp.org’s own advice, not any scholar’s. Allen, whose reconstruction this site follows, advised readers not to strive for a tonal rendering, and to put a stress on the accented syllable instead, keeping the lengths of the vowels.^[1] That is one view, and a sound one for many readers. We would put it more simply: the right choice depends on what you want Greek for.

• If your goal is to read Greek for its meaning, a dynamic accent, a stress on the marked syllable, is perfectly serviceable. It fixes the accent of each word in your memory, and it is how many readers learn. Keep the long vowels long if you can: it costs little, and it is the rhythm of the verse.
• If you want to recite poetry, or to come as close as you can to the sound of ancient Greek, aim to master the pitch accent. Let the voice rise on the accented mora and fall after it, keep the loudness even, and give every long vowel its full two beats. The rest of this tab sets out how.

The steps below use these six words. The Classical Attic examples include three human readings: μοῦσα and δῶρον by David Chamberlain, and ἄνθρωπος by Yokocentaur. The others are synthetic (∿), with pitch accent and even vowel loudness.

What to aim for

The target is a tune laid over an even rhythm. Each short syllable takes one beat, and each long syllable two, at the same loudness. The accented mora rises, by something like a fifth at most. A fifth is Dionysius of Halicarnassus’ measure of the melody of speech. It can be compared with the sixth of Norwegian and the fifth of Vedic recitation.^[2]

The fall comes on the following mora, either within the word or on an enclitic that follows it. An enclitic is a small word, such as τε or τις, that leans on the word before it. A circumflex rises and falls within its own long vowel. The grave is an acute at the end of a word that stands before another word. It is some reduced form of the rise, and its exact value is unknown.^[3]

Ranieri’s two-level model is a practical simplification. The accented mora is high and the next mora low, and there is a gentle downward drift through the sentence.^[4]

Practical steps

1. Separate the three dimensions on a single syllable. Say ah at one pitch and one loudness for two beats. Then say it again with the pitch sliding up and the loudness held. Then say it with the loudness rising and the pitch held. You can watch the three curves, pitch, intensity and duration, as you do it.^[5]
2. Count morae, not syllables. Tap the rhythm of λόγος (short-short), ἄνθρωπος (long-long-short) and δείκνῡμι (long-long-short). Speak the syllables on one note and at one loudness, until the lengths are right without any push.
3. Add the rise. On the accented mora let the note climb, and keep the volume flat. Let it fall on the next mora. Check that the intensity trace stays level while the pitch trace moves.^[5]
4. Do the circumflex within one vowel. In μοῦσα and δῶρον, go up and down inside the long vowel, and keep the following syllable low.
5. Carry it into a line of verse. Keep the long-short pattern as the floor, and the accents as the tune above it. Compare the traces with the synthetic examples, which were built exactly this way.^[6]

If you learned Greek with a stress

Many readers learn Greek with a stress on the accented syllable, as Modern Greek has it. If you have, and you now work on the pitch accent, you will need to unlearn that habit. A stress makes the accented syllable louder, and usually longer too. In Greek verse that does double harm: the tune is lost under the loudness, and a stressed short vowel starts to sound long, so the metre goes with it. These steps are our suggestions.

1. Find your habit. Say λέγω and ἄνθρωπος as you usually do, and notice where you push. Record yourself and look at the loudness (intensity) trace: a peak on λέ- or on ἄν- is the stress.^[5]
2. Flatten the word. Say it again on one note and at one loudness, every syllable as loud as the next. Keep only the lengths: in λέγω a short λέ- and a long -γω, in ἄνθρωπος long, long, short.
3. Watch the short accented vowels. In λέγω the accent stands on a short vowel and the long vowel follows. A stress turns this round, a long λέ- and a clipped -γω. Keep λέ- to one beat and give -γω its two.
4. Put back only the pitch. Let the note rise on the accented mora with no extra breath. If the loudness climbs with the pitch, hum the word first with the lips closed, then open them and keep the same tune.
5. Use the same test on a line. Read a line of Homer on one loudness and let the accents move only the note. The long and short syllables then carry the rhythm on their own, as the metre needs.

The scholarly question of what the accent was, the evidence of the musical documents and the dispute about a stress in classical Greek are not repeated here.^[7]

Notes

1. ^ Allen 1968, pp. 118–119, 143–144.
2. ^ Allen 1973, pp. 232–233.
3. ^ Allen 1968, pp. 115–117; Horrocks 2010, p. xx.
4. ^ Ranieri 2021, pp. 26–27.
5. ^ ^{a b c} See the Sound lab page.
6. ^ See the Comparison tab.
7. ^ See the Accent page of the Metre section.