http://www.kingmixers.com/Franklin%20PDF%20files%20copy/Franklin%20Dissertation.%20PDF/3%20%20IndoEuropean%20Music.pdf.

Homer’s Lyre: The Indo-European Music-stream
3.1
Because heptatonic scales are now so familiar, Terpander’s
causes little discomfort. More problematic is the putting aside of “four-voiced song”
(
). In this regard, the fragment has been the subject of a long-
standing controversy. Deubner (1929) argued that the ancient interpretation of these
verses—as seen for example in Strabo (
cf. 3.10
)—should be upheld, that they bear
witness to an historical change of the Greek lyre from four to seven strings. At that
time, the notion of a four-stringed instrument had been rejected as ludicrous by
Wilamowitz and others, who saw it as a back-formation of the Hellenistic period, for
whose theorists, as post-Aristoxeneans, the tetrachord was an important unit of
analysis.
1
Deubner supported his argument with a thorough survey of the then-
available representations from the Mycenaean period through the early Archaic.
Recognizing that in some cases an artist might be limited by space, material, or interest
in realism, he showed that the art of the Geometric period is, on the whole, consistent
in showing instruments of three or four strings.
2
It seemed that the seven-stringed
lyre, which was predominant in the Mycenaean period,
3
began to resurface in the late-
eighth century, being firmly re-established by the middle of the seventh—just when
the Lesbian singer was said to have lived. Deubner (1930) went on to argue that
Terpander’s inspiration came from his knowledge of Near Eastern instruments,
known to have been many-stringed (i.e. seven or more), in support of which he
pointed to Pindar’s portrait of the poet at Lydian banquet tables, where he was
introduced to the
(
cf. 2.15
)
.
3.2
The quickness of nineteenth-century scholars to reject a four-voiced music seems
surprising today, with many traditions now documented which use only a few pitches.
It is not merely that a four-stringed lyre seemed beneath the dignity and imagination
of the Greeks: such traditions were simply not known. Bartók himself, who made
such important contributions to the ethnomusicology of the Balkans, was at first
1
Likewise, Winnington-Ingram (1936), 10ff. cautioned that theories about early Greek music
should not to be based on the tetrachord.
2
Subsequent surveys of four-stringed and other lyres of fewer than seven strings include
Gombosi (1939), 48ff.; Wegner (1949), 222f.; (1968), 2-16; for new seven-stringed
examples from the Archaic period, see Gostoli (1990), XXXIX-XLI. On the issue of
instruments with other than four or seven strings, see below.
3
Anderson (1994), 1-16, gives a good overview of the Mycenaean and Minoan evidence and
of the evolution of Greek instrument types in general; Younger (1998) now provides a
comprehensive collection of the Bronze Age evidence.
Page 2
62
unaware of the existence of the South Slavic epic song tradition in his own backyard.
There is now no a priori reason to doubt the existence of a “four-voiced song”, and
on the whole recent scholarly opinion has accepted either a standard Homeric lyre of
four strings; or, more flexibly, lyres which could intentionally have fewer than seven-
strings.
4
3.3
But the attack against the tradition was relaunched by Maas/Snyder (1989),
vehemently condemning Deubner’s article as “influential and unfortunately
misleading”.
5
(For the record, Deubner was merely trying to confirm what
musicologists had already long entertained on the strength of the ancient tradition; this
is the direction of influence.
6
) On the basis of the Mycenaean and Minoan evidence,
Maas and Snyder maintain that the lyre had always been seven-stringed.
7
Their case
is weakened by the attempt to explain away the ancient traditions, accepted by most
scholars, of ever increasing
in professional instruments of the fifth and
fourth centuries. As they see it, the lyre remained seven-stringed even through the
modulatory New Music of the late fifth century, the traditions being due to comic
hyperbole and post-classical misunderstanding.
8
It is true that the seven-stringed lyre
persisted into later centuries, especially at the popular and educational levels, and this
is a crucial fact (
cf. 7.15
). But the gradual addition of strings by professionals like
Phrynis, Philoxenus, and Timotheus can hardly be doubted.
9
3.4
The mainstay of the argument against a four-stringed lyre has always been the
supposed unreliability of the plastic and ceramic evidence in the Geometric period.
4
Gombosi (1939), 19, 48ff.; (1944), 172; Wegner (1949), 29; Picken (1975) 597f.; West
(1981), 115; cf. (1992), 52, 330 with n.7; Barker (1982-9), 1.43 n. 18; Gostoli (1990)
XXXIX-XLI; Anderson (1994), 61ff.
5
Mass/Snyder (1989), 26, 36, 203.
6
E.g. Hawkins (1776), 1.3ff.; Helmholtz (1895).
7
Mass/Snyder (1989), 203: “Variations of a minor sort probably occurred, but in essence the
seven-stringed lyre remained seven-stringed from before the days of the Trojan War to the
time of Alexander the Great and probably beyond.” In this they have been followed by
Younger (1998), 20 and n. 51, and themselves followed e.g. Shipton (1985), 117 n.21;
Duchesne-Guillemin (1967) and Allen/Halliday/Sikes (1936), 274f. Deubner (1929)
himself allowed for a continuous history of heptatony: “Dass Terpander die siebensaitige
Leier nicht im eigentlichen Sinne erfunden haben kann, ist durch den Sarkophag von Hagia
Triada bewiesen” (195). See further below.
8
Mass/Snyder (1989), 62f.
9
See e.g. Anderson (1994), 140.
Page 3
63
Maas and Snyder warn that the Geometric depictions are not to be trusted, their
crudeness revealing a lack of concern as to the finer points of chordophone
construction; space and material bring further limitations.
10
It is true that the evidence
of ancient paintings has been abused by those who accept every variation in string
number as reflecting some definite tonal reality.
11
But for the Greek evidence of the
Geometric and Archaic periods, Deubner’s judicious examination has been
sufficiently defended by West (1992):
Certainly in some cases we may say that a painter or the maker of a small model had room
for only three or four strings in the space available, given the thickness of his brushstrokes
or the metal strands he could make. But in other cases more strings could easily have been
accommodated; and in view of the quantity of the evidence, besides the existence of a literary
tradition . . . Perhaps some [sc. lyres] had only three, but as between three and four the
artistic evidence does not have the same probative value as it has between four and seven,
and we should expect there to be a standard number corresponding to the requirements of a
particular type of singing.
12
3.5
With two exceptions, Geometric art does in fact show instruments of three or four
strings, and occasionally two or five.
13
The two exceptions come in the late eighth
10
Mass/Snyder (1989), 203.
11
See, for example, the elaborate evolutionary scheme devised by Gombosi (1944).
12
West (1992), 52 and n. 15.
13
The description in Ps.-Plut. de Mus. 1137a-b of early music as “simple and three-
pitched/stringed” (
) probably does not relate to such instruments.
Barker (1982-9) 1, 223 n. 124 persuasively argues that three pitches per tetrachord are
intended, since this passage clearly recalls the earlier description (derived from Aristoxenus)
of the enharmonic of Olympus (cf. 1135a-b), which omitted ‘diatonic
’ (cf. 1.4,
1.12, 1.25, 7.21, 7.39). If Barker’s interpretation is correct, then how does this relate to
Terpander, who has just been mentioned in company with Olympus, and whose hallmark is
always the seven-stringed lyre? If the Libation Music of Olympus were not so neatly
accounted for, one might try to associate
with the three- and four-stringed
Geometric instruments and
—that is, to the older style which
‘Terpander’ seems to have continued performing alongside the new heptatony (cf. 2.29).
Sources which describe an archetypal three-stringed lyre (e.g. D. S. 1.16.1, ps.-Censor. de
Mus. 6.610.1f.) derive from the schematization of the archaic heptachord through its
boundaries (see further 9.38-39); conceivably this could be the reference. But the simplest
solution is to suppose that
applies only to the Libation Style discussed earlier.
Olympus is mentioned by himself in this final statement (
Page 4
64
century,
14
just when Orientalizing elements are beginning to saturate Greek culture.
One cannot apply a double standard. The evidence is only unrealistic only if one
decides in advance that seven strings had been standard from the Mycenaean period
onwards. For the sake of argument, one could assume that four-stringed lyres had
always been the norm, with seven-stringed depictions the result of artistic fantasy and
abundance of space. Provided one observes proper caution—as Deubner did—it is
better to trust the overall reliability of the artists at each period, and confront the
difficulties this raises. If a believable explanation can be found for the various
changes, the artistic evidence will fall into place.
3.6
The Minoan and Mycenaean evidence, then, is a difficulty which has yet to be
overcome. After all, if the lyre had once been seven-stringed, why should there have
been a ‘regression’ to less than seven? Of those who have accepted the Terpandrean
tradition, Deubner was alone in proposing a solution: though the seven-stringed lyre
was developed in Crete during the palatial period, it was slow to supplant an older
four-stringed lyre, and Terpander was merely given credit for the final victory.
15
But
the hypothesis of a gradual diffusion from Crete is now undermined by finds of
seven-stringed lyres at mainland sites in the Mycenaean period.
16
This distribution
suggests a division not between Crete and the mainland, but between palace and
village: lyre-players are depicted in palatial art, and are now attested among the palace
personnel in Thebes (
cf. 4.5, 5.7
). Other scholars who accept that lyres went through a
period of fewer strings in the Dark Age have offered no explanation, though West
(1992) recognizes that the phenomenon has important tonal implications.
17
suppl. Bernardakis
), and though
Terpander has just been mentioned twice in company with him, there the contrast was with
and
, which in other sources are measured as deviations from the
standard seven strings: see further 8.49-65.
14
See West (1992), 51f.
15
Deubner (1929), 198f.: “Terpander es war, der die siebensaitige Leier zwar nicht erfunden,
aber doch für die Griechen zuerst an Stelle der bis dahin üblichen viersaitigen eingeführt und
kanonisiert hat”. Cf. West (1992), 330.
16
See Younger (1998), 61ff.
17
West (1992), 328: “it suggests a more restrained style of singing that used a smaller
compass, perhaps not more than a fifth”.
Page 5
65
3.7
Maas and Snyder object further to Deubner’s textual interpretation of the fragment:
Even if the lines are genuine, they need not refer to the replacement of a four-stringed
instrument with a seven-stringed one; the first line refers only to “four-voiced song”, which
might be taken in opposition to the “new hymns” of the next line, rather than to the
“seven-toned phorminx”. The poet may only be saying that he is casting aside an old form
of song in favor of a new one that is accompanied by the phorminx. The lines do not say
that the phorminx ever had fewer than seven strings.
18
The antithesis which Maas and Snyder wish to create—between “four-voiced song”
and “new hymns”, with the “seven-toned phorminx” irrelevant to the
disjunction—is impossible. By any interpretation,
must mean a
song or style using four pitches. The alternative, a song for four voices, is impossible.
The numerical elements
mark out the true antithesis, while
requires that
–
“voice”) be a single melodic element, as commonly later with
. Now, the Homeric
did not, as a rule, sing without his lyre.
19
If we may
take South Slavic heroic song as the closest extant cognate tradition (
see below
), it is
very likely that the
provided the voice with an accompaniment which, if not
always—or ever—in strict unison, was at least of similar pitch range.
20
An oral poet
uses his instrument primarily to mark rhythm as an aid to composition within the
necessary metrical restrictions.
21
Whatever the value of Saint-Saëns’ comparison
with modern African lyre-technique,
22
and however much melodic composition may
have changed in the melic revolution, it is probable that epic lyre accompaniment
involved a certain amount of heterophony in the form of rhythmic strumming, where
the function of string-pitches would be to provide the singer with his palette of tones.
While this function does not absolutely exclude a close correspondence between vocal
18
Mass/Snyder (1989), 203.
19
Apart from the fact that Demodocus, Phemius, and Achilles are so portrayed by Homer,
there is the explicit literary testimony of e.g. S. E. M. 6.16-17 (166.17f.):
(“The epics of Homer were of old sung to the lyre”);
conversely, Hesiod was considered anomalous for not playing the lyre: cf. Paus. 10.7.3:
(“And it is said that Hesiod too was ruled out of the [sc. Pythian]
competition, not having learned to play the cithara along with his singing”).
20
Nothing about the technical phrase
(Hom. Il. 18.570; Od. 21.411; h.
Merc. 54, 502) implies unison accompaniment. On this expression, see further 5.15.
21
See Lord (1980), 46.
22
See Gombosi (1944), 178f.
Page 6
66
and instrumental ‘melody’—recalling that the word is anachronistic in the description
of Homeric music (
cf. 2.31-32
)—clearly it would do little to foster such a relationship.
3.8
In the South Slavic tradition, the singer accompanies himself with the gusle, a member
of the lute family (i.e. having a fingerboard) whose one string can produce a number
of pitches through stopping. The accompaniment here is obviously monophonic, and
is almost never in strict unison with the voice. Its primary function, while not without
a melodic component in that it makes use of differences of pitch, is rhythmic. At the
same time, it uses approximately the same pitch range as the voice, and often ‘sings’ a
simpler counter-melody—or better ‘rhythm-melody’—as may be seen from Bartók’s
transcriptions.
23
In Greece, where the epic singer’s instrument was a lyre, each string
of which gave out only one pitch,
24
the quantification of ‘voice’—compare the later
use of
in terms like
and
—must have a parallel implication
for string number, and vice versa. This reading of Terpander’s language finds clear
parallels in Pindar’s
and Bacchylides’
of a seven-stringed lyre’s accompaniment.
25
Sophocles too uses
of the
instrument’s voice in the Ichneutai, which dramatized the myth of Hermes’
invention.
26
Compare also Euripides’
and
.
27
Thus,
can only mean a melodic division into four
pitches, accompanied by a
which, if not invariably four-stringed, would
certainly fall short of being heptatonic. Of course, it is possible that an earlier four-
23
Parry/Lord/Bartók (1954); cf. Lord (1960), 37-41; Foley (1999), 71, who notes that both
vocal and instrumental lines have a marked tendency to be phrased according to the
decasyllable; that is, the stichic repetition of the metrical unit is strongly felt by the singer.
24
I leave aside the question of whether lyre-strings were ever stopped to yield additional notes:
see e.g. Gombosi (1944), 179f. Roberts (1980) found in her reconstructions that stopping
produced acceptable results, comparable to pizzicati. But I have heard this technique at the
expert hands of Stelios Psaroudakes, with practically no difference in tone color. Given
Aristophanes’ testimony that students were introducing modulations to their music lessons,
where seven-stringed lyres were standard (cf. 7.17), I think we must accept stopping as a
familiar, if occasional, technique. Yet Homeric accompaniment is an entirely different
question, and I doubt very much that the technique would have been practicable during
composition-in-performance. Cf. West (1981), 116.
25
Pi. O. 3.8; cf. N. 5.22:
’ . . .
B. fr. 20B.1f. (Snell):
.
26
S. Ichn. 297 (Maltese).
27
E. Rh. 548; Med. 196.
Page 7
67
voiced music could later be sung on a seven-stringed instrument by using only part of
its range, and that this is the context of the poem.
3.9
The deeper problem with Maas and Snyder’s criticism is that the reading is not
merely Deubner’s, but that of Strabo and many other ancients. Terpander’s authority
as an innovator rests on information considerably older than the Roman geographer,
antedating the period of theoretical schematization in the late fifth and fourth centuries.
From Aristoxenus comes the fragment of Pindar which attests Terpander’s ties with
Lydia and the invention of the
.
28
Timotheus, too, attests the legend of
Terpandrean novelty (
cf. 8.61-65
). The tradition marks the continuity of Greek musical
memory in the Archaic period—a perfectly believable achievement, given that
historically the preservation of ancient lore had long been entrusted to musicians.
3.10
Finally, Maas and Snyder question the authenticity of the fragment, an issue not
addressed by Deubner. Strabo himself is dubious, they say, citing
(“it is
said”) as evidence of the geographer’s reluctance to commit himself:
29
30
Terpander . . . the first to use a seven-stringed lyre instead of a four-stringed one, exactly as
it is said in the verses ascribed to him.
But
cannot imply a questionable popular opinion: it merely reports the content
of the verses.
(“ascribed”), on the other hand, might have been used
to support the argument, since it raises the issue of the verses’ attribution.
31
But even
if Strabo had his doubts as to Terpander’s authorship, there is no sign that he did not
believe in the tradition itself; quite the opposite, as can be seen from
.
And in fact, as we have seen, belief in the tradition is well-attested throughout
antiquity.
3.11
The issue of authenticity is, at any rate, beside the point. As Deubner argued from the
start, it need not be literally true that Terpander was the first to use a seven-stringed
28
Pi. fr. 125 (S-M).
29
Mass/Snyder (1989), 26 and n.17.
30
Str. 13.2.4.
31
Cf. Wilamowitz (1903), 64 n.1.
Page 8
68
lyre: he may only have been instrumental in popularizing it.
32
West (1992) has taken
this view a step further, suggesting that Terpander need only have been a prominent
name with whom later memory could associate a large-scale change in musical
tastes.
33
Indeed, taking Terpander’s floruit (let us say 675) to mark the mainstream
acceptance of seven-stringed tunings, and counting backwards by two generations to
allow for a larger movement of which Terpander was the culmination (perhaps 725),
the Lesbian gleeman may be seen as representing a musical movement which
corresponds very closely to the Neo-Assyrian acme.
34
Though certainly historical,
Terpander—like the archetypal guslar
35
and Homer himself—grew to symbolize the
melic revolution. This is the more attractive for being able to accommodate the
numerous other innovations attributed to him.
36
Thus, as we have seen, the
anonymous corpus of citharodic preludes was ascribed to the Lesbian singer (
2.29
).
Such a situation could also underlie reports that Terpander received credit for some of
Philammon’s musical contributions—though Philammon is himself a mythical
figure.
37
Also of interest is the tradition, parallel to that of Terpander, which makes
Amphion, having added three strings to an earlier four, learn the “the tuning of the
Lydians” (
).
38
These and other such testimonia should
not be regarded as a hopelessly inconsistent jumble of tales, but as a document of
heterogeneous musical change throughout the whole of Greece, with all its regional
subtraditions, over several generations. What is important, then, is not whether the
32
Deubner (1929), 195.
33
West (1992), 330: “Much of this [list of innovations] was no doubt constructed by
projecting Classical citharodes’ practice and repertory back upon the first famous citharode
to be remembered.”
34
It is not surprising, then, that depictions of seven-stringed instruments are quite rare in the
first part of this transitional period, even allowing for the decreased sample size (See West
[1992], 52), while at the same time instruments of fewer than seven strings are also
attested, e.g. a five-stringed lyre from Attica (c. 700): Anderson (1994), fig. 11.
35
On the legendary guslar, see Foley (1999), 49-56.
36
Cf. West (1992), 329f.; Barker (1982-9), 1.43 n. 18.
37
Ps.-Plut. de Mus. 1133b:
del.
ego]
(“They say that the ancient Philammon [of Delphi] composed some of
the citharodic nomes which were used by Terpander in his poetry”)—the odd repetition of
the article suggests an interpolation, glossing Philammon; cf. Suda s.v.
:
38
Paus. 9.5.7-8.
Page 9
69
Greeks believed in the authenticity of Terpander’s poem, but whether they believed
their music underwent significant change in the early Archaic period. In fact, as we
have seen, there were many musical figures from this time who were remembered as
innovators, often with Asian associations (
cf. 2.2
).
3.12
Deubner’s hypothesis, while not answering all the questions raised by this fragment,
nevertheless rests upon sound methodology. The conflicting indicators of the ceramic
evidence are not to be dismissed, but should be welcomed as an opportunity for
discovering greater historical complexity. There are three essential issues, to be
treated one each in this and the following two chapters: the meaning of
in the context of pre-Orientalizing Greek music; the place of the Mycenaean
seven-stringed lyre within this music stream (
4.0
); and the broad changes to earlier
tradition wrought by the heptatonic
of Terpander (
5.0
).
3.13
Accepting that a four-stringed
had been the instrument of the epic singer,
West (1981) suggested that this instrument implied a limited melodic range which
might be typical of the ancestral Indo-European song tradition, and made an ingenious
(and admittedly speculative) reconstruction of the four-pitched tuning with which the
accompanied his melodies. In support of such limited melodic ambitus, West
pointed to Serbo-Croatian heroic song (
see below
) and to the chanting of the Rgveda, in
which the words’ ancient pitch accents are stylized into a three-pitched melody. Given
that the tonal accent was an original part of Proto-Indo-European, he suggested the
possibility that “the practice of ‘singing’ texts by disposing the syllables over a
limited set of fixed notes according to their accents was also Indo-European”.
39
The
Indo-European basis of Vedic song has been challenged by other scholars, who see
the accent-singing as a secondary development.
40
That the Indian vocal art does not
match the Greek system of accents poses no problem in itself, however, for musical
practice would naturally diverge alongside the respective languages. Be that as it may,
‘speech-song’ occurs in many forms throughout the world, and such traditions were
probably already ancient at the time of Indo-European unity.
41
The fact remains that
the hymns of the Rgveda are proven descendants of the Indo-European song tradition,
whatever its melodic art may have entailed originally.
3.14
In theory, at least, the general melodic character of the ancestral art might be
understood in the same way as Proto-Indo-European itself, by deduction from the
39
West (1981), p. 114.
40
Burrow (1973), 115; cf. Anderson (1994), 46.
41
Wiora (1959).
Page 10
70
comparative evidence of the daughter traditions. The progressive linguistic divergence
of these from their reconstructed parent is now quite well understood. Just as verbal
sound laws may be drawn between languages as dissimilar as Greek and Sanskrit, it is
possible to do the same for cultural institutions—which may change much more
slowly than language—and an astonishingly rich picture of Indo-European culture has
emerged in this way.
42
3.15
Recent work on Indo-European poetics has grown from the fundamental
breakthrough in the area of metrics, from which it is necessary to infer an ancestral art,
based on the distinction between long and short syllables and using certain
fundamental patterns, which split into a number of subtraditions.
43
Even more than
linguistic kinship, this brings to life the reality of a unified Indo-European
culture—“so small a language community that dialect differentiation on a spatial basis
played no part”
44
. It also demonstrates the astonishing powers of conservation
possessed by the singers.
45
The reconstruction of poetic diction depends on the
assumption that the metrical element developed side-by-side with a continually
evolving repertoire of word-formulae, built up over generations, which Parry (1930,
1932) proved to be the primary compositional tool of both the Serbo-Croatian and
Greek heroic singer. These formulae were used as building blocks in the improvised
song-telling of traditional stories, no two versions of which were ever the same,
although the storyline itself—the Aristotelian
—might be considered a unique
entity. This process explains the seemingly paradoxical statement of the
Phemius, who asserts “I am self-taught, and a god implanted all sorts of tales in my
mind.”
46
42
See especially Benveniste (1973); Polomé (1982); Watkins (1995).
43
The foundation was laid by Meillet’s (1923) comparison of Greek and Indic verse; Jakobson
(1952), Watkins (1963), and Cole (1969) established the Indo-European nature of Slavic,
Celtic, and Italic metre respectively; for Nagy’s (1974) comparison of Greek and Indic
metre, see below.
44
W. Meid ap. Szemerényi (1996), 30.
45
Further cognate phrases and semantic doublets are catalogued by Schmitt (1967); (1968).
46
Hom. Od. 22.347f.:
’
. As Dodds (1951), 10 explained, “The two parts of his statement are not felt as
contradictory . . . he has not memorized the lays of other minstrels, but is a creative poet
who relies on the hexameter phrases welling up spontaneously as he needs them out of
some unknown and uncontrollable depth; he sings ‘out of the gods,’ as the best minstrels
always do”.
Page 11
71
3.16
Taking this as his departure point, Nagy (1974) made a case study of a pair of phrases
first noted by Kuhn (1853)—Greek
and Sanskrit srava(s)
aksitam—which are phonologically, quantitatively and accentually equivalent (< PIE
*klewos ndh
w
hitom, “imperishable fame”). These, he argued, constitute cognate
poetic formulae which, by a fortunate accident, had been handed down intact over the
millennia by singers who, originally, must have known the same songs and shared an
archetypal repertoire of poetic formulae. According to this approach, elements of
Proto-Indo-European poetic diction may be tracked through the overgrown jungle of
metrical data, like a single tagged animal will reveal the peregrinations of its species.
The formulaic character of
has since been questioned,
47
and many
scholars now believe that “the hierarchical dependence of metrical form on
phonological and phonetic form makes actual reconstruction of metrics an unrealistic
goal”.
48
Nevertheless, the large amount of poetic phraseology which can now be
reconstructed
49
requires us to suppose some coevolution of diction and metre.
3.17
Although many of the Greek and Indic metres as we have them are fixed grids into
which words of matching syllabic quantities were fitted, they cannot have been so
originally; for otherwise we should expect to find cognate poetic material in identical
metres. Because they occur in different but related metres, the fixed classical forms
must have come about incidentally in the course of musical evolution. That is, as
certain word combinations were used again and again in the telling of tales, they
became fixed formulae, which in turn influenced the singers’ compositional process:
“the changes follow the patterns of the stable formulae, because the singer thinks in
those patterns.”
50
As the various formulae were used in tandem, larger patterns began
to result. These in turn could change as phonological developments took place within
the language itself and words acquired new metrical properties.
51
By a continual
feedback process, certain patterns became ossified until they did come to function,
effectively, as grids. It is important to realize, however, that the Greek metrical art as a
whole continued to unfold down into the Classical period, alongside certain fixed
forms. The dactylic hexameter was one of these, its evolution complete some time
47
Finkelberg (1986) argues that
, which occurs only once in Homer, was an
ad hoc creation modelled on other well-established formulae. Yet its frequency in later
poetry might well mean that it was not simply a Homeric borrowing, and that its lack of
repetition in Homer is insignificant.
48
Watkins (1982), 164f., with further literature; see also Ruijgh (1995); Gasparov (1996).
49
See especially Watkins (1995).
50
Lord (1980), 41.
51
Nagy (1974), 50.
Page 12
72
before Homer. Yet, even here, metrical anomalies reveal traces of obsolete phonology,
in some cases antedating Linear B.
52
3.18
These discoveries confound rigid notions of genre distinctions.
*klewos
ndh
w
hitom, for example—if it truly was an Indo-European formula—surfaces in the
epic song of Homer, the non-epic, more ‘lyric’ hymns of the Rgveda, and in a poem
of Sappho which is not hexametric, but which is partially dactylic.
53
As Aristotle
noted, metre is an insufficient criterion for defining genre.
54
Note that ‘lyric’, like
‘melody’, is at any rate an anachronistic term, if one accepts the communis opinio that
the Indo-Europeans knew no form of the lyre before their contact with the Near
Eastern cultural sphere. And of course Greek epic was also accompanied by the lyre.
The discovery of Indo-European metrical kinship makes it necessary to suppose that,
though there will have been different kinds of song, these must have been at one time
indistinct as to their basic musical elements, being drawn from a single, homogeneous
musical language. Because of this, the lack of attested ‘lyric’ poetry contemporary
with Homer need not pose an insurmountable obstacle to establishing the general
characteristics of Greek music prior to the Orientalizing period. Indeed, it is certain
that the conservative Aeolic tradition preserved inherited metrical features in a very
archaic, ‘pre-Homeric’ state. The ‘lyric’ structures used by Sappho and Alcaeus best
reveal the survival of Indo-European features, while the dactylic hexameter represents
but one special—and relatively late—development of these.
55
For the formulae of
Homer, when their internal rhythms are considered outside of their hexametric context,
frequently coincide with the metres used by the Aeolian poets.
56
3.19
Since the metrical and dictional data are drawn from the most ancient material available
in each of the subordinate traditions, and in every case this is song,
57
one must posit
for the Indo-Europeans a unified musical stream of which the metrical and dictional
were two components, and for which there must also have been a melodic aspect. The
52
West (1988), 156f.
53
Sapph. 44.4 (Voigt).
54
Arist. Po. 1447a28ff., esp. 1447b17-20:
(“but Homer and Empedocles share nothing besides
metre, wherefore it is right to call the one a poet, but the other a natural scientist rather than
poet”).
55
See Nagy (1974); West (1982), 29f.
56
Nagy (1974), 15, 27ff.
57
Cf. Lord (1962), 180ff.
Page 13
73
Slavic, Indic, and Greek traditions have provided essential comparanda for a
reconstructed picture of this unified musical stream in its metrical and dictional
aspects (
see below
). Because of this, the analogy of Slavic and Greek epic, though
often abused in the study of Homer, remains fundamentally valid.
58
Obviously, the
two have evolved along very different lines.
But a number of shared
phenomena—formulaic phraseology, metrical anomalies deriving from formula
fossilization, heterogeneous dialectal elements with synchronic and diachronic
dimensions
59
—can only be explained as deriving from a single method of building
poetry. Since no epic tradition has survived into modern Greece,
60
the cognate South
Slavic tradition, different though it may be, is the only evidence which has any real
claim to illuminating the melodic practice of Greek epic.
3.20
From his study of Serbo-Croatian folk songs in the Parry collection, Bartók was
surprised to discover how many melodies were of restricted scope: a full third of those
transcribed spanned only a ‘fifth’, but more often a ‘fourth’ or less.
61
(These interval
measures, deriving from diatony, are for this music approximate and anachronistic.)
For the most part these were ‘ceremonial’ songs, those which could not be dissociated
from certain ritual contexts, taken in the broadest sense—work songs (hay gathering,
harvest), cradle-songs, wedding-songs, laments, children’s songs, calendar songs,
rain-begging songs, and so forth. A number of these ‘genres’ are also alluded to in
Greek sources, beginning with Homer.
62
The equally ancient South Slavic heroic
song is itself essentially ritualistic, serving as the historical record of a people,
performed in prescribed social settings, and constituting “a necessary part of the so-
cial life of the family or of a community”.
63
Tacitus described how the ancient
Germanic peoples “celebrate in ancient songs, which is the only kind of record and
archives that they have” (celebrant carminibus antiquis, quod unum apud illos
memoriae et annalium genus est), mythological stories and the interrelations and
migrations of the kindred tribes, using song also for divinatory purposes.
64
Caesar
left a crucial description of the Celtic druids as guardians of lore through song, the
58
See generally Foley (1999).
59
See Foley (1999), 66-88.
60
Lord (1991), 93.
61
Bartók/Lord (1951), 52-6; 60.
62
See West (1992), 28f.
63
Lord (1962), 181.
64
Tac. Germ. 2-3.
Page 14
74
sacred injunction against the use of writing—an Indo-European characteristic
65
—and
the regional schools where training could take as much as twenty years; Diodorus,
who uses the term ‘bard’, describes their songs of praise and blame, and the lyre with
which they accompanied themselves.
66
Oral composition characterized all of these
traditions, and while they flourished, the individual song was continually new,
describing recent as well as ancient events. Thus, in the living South Slavic tradition,
one encounters jarringly modern details, such as the hero armed with a rifle. At the
same time, however, we must suppose the continual fossilization of certain songs,
where faithful recitation was crucial to the song’s efficacy. In the Greek and Indic
traditions, oral composition as a whole eventually became static. The recitation of the
Iliad and Odyssey, for example, became a ritual in itself, that of preserving and
learning from the poetic monuments of the ancient style. The same may be said of the
Vedic canon, fixed hymns which descend from an originally oral style of composition;
here, however, their ritual preservation was religiously driven.
67
3.21
Having observed such material throughout the Balkans, Bartók posited an ancient pan-
Slavic musical tradition of which limited melodic compass was a defining feature.
68
He had no cause to look beyond the Balkans; but if his hypothesis is right, one might
well suppose that the tradition from which it descended will have used melodies no
65
Caes. B Gall. 6.13-14: neque fas esse existimant ea litteris mandare . . . quod neque in
vulgum disciplinam efferi velint neque eos, qui discunt, litteris confisos minus memoriae
studere (“And they think it unlawful to entrust these verses to writing . . . because they do
not wish the learning to be made common knowledge, and they do not wish those who learn
it to develop the memory less through their reliance on writing”); cf. Polomé (1982a),
166f.: “The reserved attitude of the Indo-Europeans was translated in their piety by a set of
interdictions . . . their tradition was transmitted orally, and after some of them acquired the
skill of writing, a taboo was maintained against putting down in writing their religious
lore.”
66
Caes. B Gall. 6.13-14; D. S. 5.31.2-5:
’
’
(“And among them are also poets of music,
whom they call Bards. And these, singing to instruments like lyres, make songs of praise
and blame”). On the strength of the celtic bardic tradition as late as the seventeenth century,
see further Watkins (1995), 76ff.; Ahl (1991) 136 and n. 14; for the sources used by Caesar
and Diodorus, as well as other ancient testimony, see Rankin (1987), 272-276 et passim.
67
Nagy (1974), 15ff.
68
Bartók/Lord (1951), 4, 52-6; 60; cf. Wiora (1959), 203.
Page 15
75
less restricted in scope.
69
In his seminal work on the existence of pre-pentatonic
‘tonal systems’, Wiora (1959), expanding on Bartók’s hypothesis, showed that songs
of limited melodic compass were to be found, not merely in the folk traditions of
Europe, but throughout the world, whether living or fossilized. In the majority of
cases, a close relationship existed between melody and speech contour,
70
and this led
him to conclude that speech-song had once been a universal or near-universal
phenomenon. Both qualities he attributed to the earliest phase of human musical
activity.
71
3.22
If Indo-European melodic practice was originally connected with tonal accent, some
eventual dissociation of the two would have been unavoidable in the daughter cultures,
since this accent has itself proven to be an evanescent feature. Pitch-accent is,
however, still preserved in Serbo-Croatian, playing a part in the singers’ melodization,
along with a body of stereotyped melodic formulae.
72
What role the tonal accent
played in the creation of these formulae is unknown, yet it may have left its mark in
peculiarities of melodic intonation which, while being subsequently conditioned by the
diatonic scales of Western art music, owe nothing to them in origin. In such singing,
the ‘same’ note will often vary widely.
73
Because of this, it is impossible for us to
speak of a ‘tone-system’, since, strictly, the musical tone is by definition a single,
stable pitch—deriving in this sense from
as
(
cf. 2.20, 4.33
). At the same
69
But this is not to yield to an evolutionary view of melody which regards “one-, two-, three-,
and more-note systems as corresponding to successive historical periods”; for a refutation of
this outmoded view, see Wiora (1959), 185 and as a whole.
70
Wiora (1959), 189 and quoting Hornbostel: “The duality of pitch, rising and falling by one
step, is in primitive music obviously related to accent. ‘The singer gives way to the natural
tendency to sharpen or flatten the note simultaneously as it grows weaker or stronger.’ It is
more natural to change the level in such a manner than to maintain it, that is to repeat the
note continuously from beginning to end of a song.”
71
Wiora (1959), 203f.: “[Such narrow melodies] are, obviously, more ancient than pentatony.
They belong to the most ancient kind of tonal systems known to us, and, what is more, to
the most ancient that ever existed; this can be inferred by systematical
considerations . . . They are evidences of the origin of music following the pre-musical
sound . . . Ancient forms, that were to become blocks and backbones, shine here in the
archaic purity of their origin.”
72
Lord (1960), 37f.
73
Wiora (1959), 200: “If the pitches intoned are measured in cents, the number of cent values
obtained may often be very high, while the number of ‘degrees’ conceived still remains very
small.”
Page 16
76
time, it is clear that this variable intonation is intentional and does form a coherent
‘system’ of some sort.
74
This unique intonation of Serbo-Croatian heroic song, still
uninfluenced by European art music at the time of Parry’s fieldwork,
75
might preserve
elements of the ancient pitch accent to which the pan-Slavic melodic language as a
whole originally answered. The phenomenon of broad musical unity over great
geographical stretches is well illustrated by South Slavic poetic diction, which is a
composite of Bosnian, Croatian, and Serbian dialectal forms. As with Homeric
diction, the singer mainly employs his own dialect, but there are many formulae which
carry with them archaic or alien forms.
76
3.23
As to ancient Greek heroic song, its melodic dependence on accent may be inferred
from the fact that peculiarities of Homeric accent and pronunciation were preserved in
the rhapsodic tradition long enough to receive the attention of Hellenistic grammar-
ians; furthermore, the available fragments of notated music suggest that, unless
prevented by strophic responsion, even Greek melody of the Classical period and later
has a marked tendency to follow the contour suggested by word accent.
77
Indeed, a
kind of accent-melody was addressed by Hellenistic literary theorists, who seem to
have treated ‘euphony’ as a formal art; Philodemus’ refutation shows that the issue
was still alive in the first century
B
.
C
., as do the accents inserted in parts of the text by
the unknown owner of the papyrus.
78
Thus, the evidence does in fact indicate an art
of accent-melody for Greek epic song. Given the identifiable continuity of Indo-
European poetic tradition throughout the daughter cultures, it remains to my mind
quite conceivable that core elements of an ancestral vocal art should have persisted—in
the broadest terms of course—and that Homer represents one heir to this. On the
strength of this alone it should be necessary to explain the presence of heptatonic
melodies in Greek music of the Archaic period, even without the traditions about
Terpander. It is the seven-stringed lyre, and not the four-voiced song, that requires
explanation.
74
Bartók/Lord (1951), 4: “These deviations, since they show a certain system and are
subconsciously intentional, must not be considered faulty, off-pitch singing. This is the
essential difference between the accidental off-pitch singing of the urban amateurs and the
self-assured, self-conscious, decided performance of peasant singers.”
75
Bartók/Lord (1951), 4.
76
Foley (1999), 76f.
77
West (1981), 114; (1992), 198-200.
78
Philodem. Poem. 1.93-4. See Janko (2000), 84, 298-301.
Page 17
77
3.24
The Slavic material proves that Greek song could easily have retained its ancient
character into the Dark Age. The Vedic tradition, no more ancient than Slavic or
Greek oral composition in origin, was effectively arrested at a very early stage in its
development—a millennium and a half earlier than the date hypothesized for a
common Slavic spoken language
79
—and handed down subject to the most rigorous
and centralized conservation that the priests could achieve. This provides an important
diachronic anchor for Indo-European poetics, with or without accent singing as
original. Serbo-Croatian heroic song, on the other hand, like Greek epic down
through the Archaic period, has continued its slow, generational development, without
restriction, into modern times. On an absolute timetable, then, the Homeric art falls
between the dates for which ancient Indo-European attributes are attested in the
parallel subtraditions.
3.25
Yet what supports the notion of an ‘absolute timetable’ when it comes to the
development of a melodic art? It would be necessary to suppose that the two basic
musical elements of the Indo-European tradition, metrical and melodic, evolved in
close company, at roughly the same rate, and both subject to the same conservative
transmission which preserved the vestiges of metrical kinship: for it is the metrical
data in the first place which guarantee a recognizable Indo-European character to the
Greek subtradition at the period in question. But what would prevent the melodic and
metrical elements from evolving at different rates? After all, we find melodies of
octave scope in Greece over two thousand years ago, while still-extant Slavic melodies
preserve traces of the ancestral tradition.
3.26
Moreover, even if Greek
was the sort of narrow-range speech-song proposed,
was Greek music like this as a whole? One might suppose that the non-heroic songs
of the same period—all unattested, and only alluded to by Homer—used melodies of
greater scope, while epic singing survived as a sacred and ancient tradition, as it
certainly was by the Classical period. Vedic song might provide a parallel, existing
alongside the Saman chant, which used the text of the Rgveda but was more
recognizably melodic—i.e. more than a simple stylization of the words’ pitch accents,
and exceeding at times a sixth in range. The beginning of Saman chant must have
been before the Vedic hymns received their finished form.
80
Conversely, the
79
Old Church Slavonic, with records going back to the ninth century
A
.
D
., can be taken as a
close representative of a common Slavic language: cf. Szemerényi (1996), 11.
80
Fox-Strangways (1914), 249 n. 2: “There is nothing to show that the [Saman] chants are
later than the words [of the Rgveda]; in fact, since Samans are often mentioned in the
Rgveda there is a probability, beyond the intrinsic likelihood, that they are older.”
Page 18
78
development of wider melodic range—if this was in fact a secondary innovation—did
not prohibit the continued existence of more ‘archaic’ melodic styles, for
subpentatonic melodies of the type Wiora observed as being global are well
documented in India today. These so-called tribal melodies fall outside the elaborate
taxonomy of classical raga, which span for the most part a range of at least an
octave.
81
The distinction between ‘urban’ and ‘rural’ style seems to have been made
already in antiquity by Bharata;
82
if this is right, the contrast will not have been with
raga per se, which attained to theoretical primacy at a date considerably later than the
Natyasastra.
83
It is not surprising, even at this early date, to find an awareness of the
divergence of art music from a more universal and ancient ‘folk’ style. Indeed, a
departure from such a conservative practice would seem to demand some historical
acknowledgement, like the Terpandrean tradition. The dichotomy provides an
interesting parallel to the emanation of diatony from the cities outwards in the Balkans,
documented by Bartók (
see below
).
3.27
Once this possibility is allowed, what limit is there to the antiquity of such stylistic co-
existence: might there not have been such complexity at the time of Indo-European
unity? Under what conditions would melodic compass have changed, especially if one
rejects a progressive evolution towards heptatony? Could this have happened from
within the tradition, or only in response to some external influence? Clearly, this
raises questions about the nature and transmission of melody—a word which I have
used lightly until now. But as we understand it—a fixed, repeatable tune—the word is
misleading and at least partially, perhaps wholly, irrelevant. As we have seen (
2.31-32
),
musical
is not found in any of Homer’s numerous performance scenes; thus,
even if the word could have had a musical sense in his time, it was clearly unessential
for the style of song he wished to describe. Furthermore, in the Indo-European and
81
Bhattacharya (1968), 46ff.
82
Bhattacharya (1968), 46ff. The dating of the Natyasastra, a treatise on all aspects of
dramaturgy, including music, is uncertain. Ghosh (1934) believed it to be the work of one
hand, written in the second or third century
A
.
D
.; though isolated verbal features did suggest
the preservation of earlier sources; he was reluctant to date this material before the first
century
B
.
C
. Srinivasan (1980) has since argued persuasively that the work is in fact a
hopelessly jumbled and inconsistent compilation; Rocher (1981) has examined the
complexity of its textual history. In any event, at least some of the musical terminology
must be considerably more ancient than the Natyasastra as we have it, since musical
references in older, non-specialized works imply a formal theoretical tradition: cf. Tarlekar
(1975), 161; Fox-Strangways (1914), 114.
83
On the relative lateness of the raga system, see Widdess (1995).
Page 19
79
other epic traditions, words for ‘song’—e.g. Homeric
—are frequently
ambivalent as between “sing” and “tell”,
84
and this suggests a fundamental
subordination of ‘melody’ to words. All recorded epic melody is simple and
repetitive, since the demands of composition-in-performance force the singer to give
most of his attention to choice of word and phrase. In the Serbo-Croatian material
collected by Parry and Lord, there is an inverse relationship between richness of
poetry and complexity of tune. Avdo Mededovic, for example, regarded by Parry and
Lord as the last singer of Homeric stature, was a guslar of middling ability who was
sometimes reduced to running his bow over the string in a drone as he unleashed great
torrents of poetry.
85
Here we see the essentially rhythmic function of the epic
singers’ instrument, despite its participation in the realm of pitch. The primacy of
poetic invention thus operated as a sort of balancing mechanism to check the
elaboration of melodic language. The Indo-European singer of tales was simply not a
melodist as we would understand it.
3.28
If the metrical model presented above is valid, and if Indo-European art song involved
the melodization of pitch accent—or if its melody answered to any aspect at all of the
language—there are several consequences. Where rhythm and melody are drawn
from internal characteristics of word and phrase, we must suppose a pattern of
melodic evolution analogous to the metrical. Any given combination of words would
have a unique pattern of pitch accents to be navigated melodically according to some
standard strategy, while allowing for the ‘tactical’ variations of regional subtraditions
and the individual singer. The conventions which governed these maneuvers might be
called ‘melodic syntax’, depending as it would on the syntax of words. A collocation
which stood the test of time, fixing its place in the poets’ repertoire, would be
accompanied by a melodic fragment which was also more or less formulaic, to be
absorbed by a young singer as he learned the words and rhythm of each formula.
Such a fragment comes a step closer to our notion of melody as a unique and
memorable pattern of pitches. The sum total of these fragments would comprise a
melodic ‘vocabulary’ which would follow the same the cycles of growth and decay as
the diction and metre.
86
Yet during composition-in-performance, the melodic and
metrical will have unfolded independently; for each new phrase which took advantage
of an established metrical cliché would be accompanied by its own accentual pattern.
The counterpoint of tonal and metrical modulation provides an ever-different
84
Lord (1962), 180ff.; West (1981), 113.
85
Lord (1980), 57, 68.
86
For the concept of musical styles being governed by syntactic and morphological ‘rules’ and
analyzable through a modified linguistics approach, see e.g. Sloboda (1985).
Page 20
80
‘accompaniment’ to the tale being sung. It was this hypnotic interplay in Serbo-
Croatian heroic song which fascinated Bartók in the end.
87
3.29
The metrical parallel suggests that a limited number of melodic phrases could emerge,
and that these would eventually drive the process of oral composition, and no longer
be driven by it. Thus, the South Slavic guslar, though his formulaic diction is
continually varied, uses a fixed set of melodic formulae in a single metre. To press the
analogy of metre and melody, we must suppose a state of the world—at least for Indo-
European oral composition—in which melodies were not sung to certain fixed
pathways such as those offered by piano scales or the strings of a lyre. Such melodic
routes would result in time, a ‘tonal’ system derived through the ritualization of
speech as song.
88
3.30
This model of melodic evolution produces a continuum of fixity, ranging from the
determined melodic fragments of traditional formulae to the fluid customs by which
non-formulaic phrases would be navigated in performance. It also follows that each
of the subtraditions would hold in common a different body of melodic fragments and
syntax, varying directly in proportion to the culture’s overall divergence from its
parent and sisters. Accordingly, the melodic evolution of the Indo-European traditions
should show patterns of synchronic and diachronic change closely akin to those of the
associated languages. This is borne out by Bartók’s study of melodic distribution,
which shows patterns precisely analogous to those of dialectal dispersion. In the
Balkans, the ‘same’ melody is found in various regions, most easily recognized by
features such as section structure, metrical character, ambitus, and melodic contour.
The actual ‘scale’ or pitch-set of each proved unhelpful to Bartók, who found it
difficult to decide when two melodic variants were distinct enough to warrant classi-
fication as separate tunes. Sometimes, when tunes which had been accepted as
87
See Lord (1960), 37f. I was privileged to hear some this material in October 1997, by the
kindness of M. L. Lord and M. Kaye. Selections from the Parry Collection have finally
been made available with the reissue of Lord’s The Singer of Tales (Cambridge, Mass.,
2000).
88
Wiora (1959), 203: “When we make music or listen to it, we have the keyboard, the tonal
system, the totality of our scales in our minds; these make up the area on which we move,
and of the whole tonal area which we realize in its entity, we let now one now another ring
out. In ancient singing, on the contrary, the dividing lines between notes and the system
imagined against it existed but in the kernel. The usual systems or the ‘succession of notes
used’ in this singing are actually scales, i.e. keys, but in a broader sense, [from] which we
must exclude everything that is added to it by the music of great civilizations”.
Page 21
81
variants were ranged by degree of variation, the first of the series could no longer to be
heard as a variant of the last, but was a strikingly distinct melody.
89
The same
phenomenon has been observed in India—where the elaborate classification of raga is
concerned with just this—as well as England and elsewhere.
90
Propp observed the
effect in the distribution of Russian fairy-tales.
91
In song dispersion, such scalar
differences within the ‘same’ tune might fairly be seen as ‘regional pronunciations’,
directly analogous to dialectal differences in the spoken language. The closer the
relationship between speech and song, the more these patterns will present a
synchronic view of the Indo-European musical tradition in dispersion.
3.31
Clearly, such patterns must have existed in ancient Greece, and may be at the root of
the musical
. The word, which in a non-musical sense means ‘custom’ or
‘common law’, may also designate a musical entity which was open to interpretation
and variation, but at the same time uniquely recognizable.
92
Compare the memorable
duel in Stevenson’s Kidnapped, where each piper plays variations, known only to him,
of a tune known by both; or the friendly competition set up by Parry between Avdo
Mededovic and his colleague Mumin Vlahovljak.
93
The use of
to describe
birdsong might support this interpretation. When Alcman claims to know all the
of birds, the human musical
is clearly implied; indeed, this fragment is the
first witness for both uses of the word, giving very early authority to the Greeks’
recognition of the analogy.
94
Elsewhere Alcman explicitly compares human and
birdsong,
95
and Hesiod attests a musical sense to
in his description of the
89
Bartók/Lord (1951), 15, 34.
90
Fox-Strangways (1914), 151ff.
91
Propp (1975), 114: “The distinction between theme and variant is totally impossible. Here
there can be only two points of view. Either each alteration gives a new theme, or all tales
provide one theme in diverse variants. As a matter of fact, both formulations express the
same thing: the entire store of fairy tales ought to be examined as a chain of variants.”
92
See e.g. West (1992), 215-7; the testimonia for
are assembled by Grieser (1937).
93
Lord (1980), 68.
94
Alcm. 40 PMGF; cf. 39; Alc. fr. 307c (Voigt)?; Pratin. fr. 1.5 (PMG 708); Ar. Av. 210,
1346; etc. Barker (1982-9), 1.250 rightly argues that the classical writers were largely
speculating in their discussions of the Archaic
, and that Alcman’s usage is not
technical in the same way; but the literary evidence is sufficient to show that the musical
existed in some form throughout the Archaic period; cf. Chadwick (1996), 206f.
95
Alcm. 1.100f. PMGF; 39; Anth. Pal. 7.19.1f. (Leonidas Tarentinus), which calls Alcman
’
(“the singer-swan of wedding songs”); cf. h. Hom.
19.16f.; Anth. Pal. 9.184.9 (Anon.).
Page 22
82
Muses singing the laws (
) of the gods.
96
But the association of birdsong with
the lyre was already very ancient, being attested in the Mycenaean period by paintings
and instrument design,
97
and obliquely in Homer’s archer-citharist simile, where
Odysseus’ bow string is said to sing like a swallow.
98
The derivation of both
and
(“pasture”) from
(“to distribute”)—if this is correct
99
—accords
well birds’ use of song to delimit territory, a function which combines the musical and
‘legal’. Moreover, though each species has a single and distinctive call, there is a
continuum of variation throughout the range.
100
It seem likely therefore that the
Greeks would have been aware of the ‘same’ song being varied by region, just as they
were familiar with the various spoken dialects.
3.32
We get frequent references in later literature to individuals, such as Terpander or even
Timotheus, who composed
.
At first sight, this seems to contradict the
identification of the
as a traditional tune in dispersion. Yet in the melic
revolution, with its emphasis on individual innovation, the word’s older meaning
would naturally have evolved alongside the musical changes. The waning of the
heroic song tradition, as represented by the increasingly classical status of the Iliad
and the Odyssey, clearly attests the effacement of traditional distribution patterns. Of
course,
could be quite properly applied to a piece by Timotheus once it had
been adopted by other musicians, given a musical culture in which broad interpretive
powers were granted to the individual musician. In the Archaic period, especially,
there must have been a certain amount of such freedom. The songs of Sappho, for
example, were widely known and survived in recognizable form into the Classical
period, yet the hypothesis of distribution in fixed form through notated ‘sheet music’
96
Hes. Th. 66f.:
. Ps.-Arist. Pr. 19.28 offers the explanation that prior to literacy laws were
sung, and reports that this was still true among the Agathyrsoi of Thrace; Plato develops
the association of the legal and musical
extensively at Lg. 656c-660c, 799e; cf.
Phdr. 278c; cf. Mart. Cap. 9.926 Graecarum quippe urbium multae ad lyram leges
decretaque publica recitabant; Clem. Al. Strom. 1.16.78:
97
Anderson (1994), 4-7, 12f.; Younger (1998), plates 13 (Pylos fresco, Chora Museum, LH
IIIB2-C), 14.1 (Pyxis, Chania Museum 2308, early LM IIIB).
98
Hom. Od. 21.406-11, cited in 5.14.
99
Chadwick (1996), 206f. has recently questioned this derivation, “and even if the connexion
is proved, more research is needed on the history of the noun”.
100
This was pointed out to me by Dr. Richard H. Backus, Woods Hole Oceanographic
Institution (communication).
Page 23
83
is problematic in the extreme. Here, perhaps, the traditional art of melodizing poetry
was still operating somehow within the new melic music—like the epic-melic fusion
of the Homeric Hymns (
cf. 2.30, 5.0
)—so that, throughout the Greek world, trained
musicians could give a musical performance directly from the text, if the
or
were specified. If this is correct, it is strong evidence for the early existence
of a broadly unified Hellenic melodic art, of which epic singing was one exemplar.
3.33
This picture of ‘melodic dialectology’ might also illuminate two further aspects of
later Greek music. The first is the existence, attested as early as Alcman (
cf. 2.31
), of
(later
) distinguished by region: Phrygian, Lydian, Dorian, Ionian,
Cretan, Aeolian, and so forth. The use of such names to describe the octave species is
a development of the later fifth century, perhaps one contribution of Eratocles,
101
and
here it is right to suppose some systematization of early Greek practice through the
regularizing effects of diatony (
cf. 1.4, 1.12
). Again, this does not exclude the
possibility that the diatonic octave species were also known throughout the Archaic
period (
cf. 1.12
); prior to Eratocles they may have had no names, or other
names—such as Terpander’s sevenfold division of the citharodic
(
cf. 7.25,
10.37
)
.
Such regional styles are likely, at least in part, to have been conditioned by
“the isolationism of the Dark Age”,
102
though ‘Phrygian’ and ‘Lydian’ may
document, however distantly, the exchange of musical ideas with these cultures (
cf.
1.22, 2.5, 2.11, 2.15
). Though there is a continuum of variation even within a spoken
dialect, nevertheless there must be enough distinctive features found throughout the
dialect area to warrant its unique classification. The regional metrical traditions are
likewise marked out by conventions proper to each.
103
The same must have been true
of the otherwise continuously varying melodic dialects, of which the early
were
expressions.
3.34
This is consistent with what is known about the education of singers in the Indo-
European traditions, whether we suppose for the Greeks formal, regional schools of
song like those attended by the Celtic bards, travelling
such as Terpander and
Alcman who might effectively impose certain standards on many communities during
the course of their travels, or a combination of the two—recalling that Terpander was
said to have established a musical ‘school’ in Sparta (
cf. 2.38
), and the Lesbian
may have been similar. The legendary guslar of South Slavic tradition was also
101
West (1992), 227; on Eratocles, see further 7.0.
102
West (1973), 181.
103
See West (1982).
Page 24
84
remembered as travelling widely;
104
his comprehensive repertoire and reputed
influence over singers everywhere symbolizes the broad unity of the tradition. The
professionalism of Greek musicians is well illustrated by a host of performing names,
historical and legendary—Araros the son of Aristophanes, Chairis, Choricius,
Chorocles (father of Phrynichus?), Cycleus (father of Arion), Demodocus, Encomius
(father of Pratinas?), Epicharmus, Eumolpus, Eunomos, Harmonides the aulete,
Molpis, Phemius, Philochorus, Polymnestus, Polyterpus, Spendon of Sparta,
Stesichorus, Terpes, Terpsicles—not to mention both Homer
105
and Terpander
himself. These may have been taken as stage names by each, or given at birth by
musician fathers intending to pass on their trade. Thus Demodocus and Phemius
cannot be dismissed as fictitious merely because of their names; they may once have
been renowned performers whose memory Homer honored for reasons of
professional courtesy. In the Indo-European cultures, singers often formed a
hereditary caste within the aristocratic stratum.
106
In Sparta, at least,
-playing
was passed from father to son.
107
Serbo-Croatian heroic song still seems to have
been quasi-hereditary in the early twentieth century.
108
3.35
This inherited melo-dialectal variation may have been one tributary to the Aristoxenean
genera and their diverse shadings (
cf. 1.24, 2.35
). In cases of musical syncretism, it is
common to find that two styles, after coexisting for some time, may coalesce into a
distinct practice which features the strongest elements of each.
109
In the case of the
African-American syncretism, we find the imposition of inherited ‘blue’ notes on
European diatony. Likewise, the Terpander fragment suggested a juxtaposition and
segregation of two musical styles in the early phase of seven-stringed music (
cf. 2.29
).
By the end of the Archaic period, however, and probably well before (
cf. 7.39
), the
‘clear’ diatonic tunings were being ‘colored’ by musicians like Lysander of Sicyon,
who introduced
(“well-shaded colors”).
110
We might then see
Aristoxenus as the first ethnomusicologist, making careful measurements against a
diatonic grid through the application of Aristotelian methods of classification. The
104
See Foley (1999), 52f.
105
Nagy (1979), 297-300.
106
Watkins (1995), 71
107
Hdt. 6.60:
.
108
Lord (1960), 22.
109
See Nettl (1985), and below.
110
Philoch. FGrH 328F23 = Ath. 14.637f-638a. See further 7.33.
Page 25
85
musicians’ instincts must still have been alive and well for him to be able to make
such nice distinctions; for the Aristoxenean measurements do not seem arbitrary or
even particularly approximate.
111
3.36
It is against this continuum of melodic change that the fossilized ceremonial songs
observed by Bartók and Wiora, and alluded to by Homer, must be understood. If
each accompanied some ritual—a complex of actions resistant to significant
change—the ceremonial song would more or less drop from the ongoing process of
formation and reformation which characterizes a tradition of oral-composition.
Having achieved perfection, it would not be recomposed on each occasion, and would
endure as a snap-shot of the music-stream in flow, becoming what we understand as
melody: a fixed tune. Kinship between languages is most easily revealed by those
words which have preserved their ancient meanings the longest; those which,
designating something basic and unchanging, had themselves no reason to change.
Likewise, it is through the ceremonial songs of the cognate subtraditions that the
character of Indo-European song must be established, as these are the only diachronic
data available. While we must suppose that e.g. the harvest songs of each tradition
will be as unlike-sounding as the respective languages, a continuous channel of
preservation for the ‘form’ itself requires no imagination.
3.37
The dependence of metre and melody on poetic language, if it is right to suppose this
as characteristic of Indo-European song, would go a long way towards explaining the
seemingly analogous phenomena of linguistic and melodic distribution. The singer of
oral-formulaic speech-song requires a fluency which seems as much linguistic as
musical.
112
As with language, the handing-down of such an art is conservative in the
extreme. A new generation does not devise a new language, but learns that of its
parents. Likewise, the accomplishment of the younger singer is in the emulation of
the elder, and this professionalism—in the case of South Slavic heroic song, the
111
See Winnington-Ingram (1932).
112
Lord (1962), 184: “As a boy he hears the old men sing, and he absorbs the stories and
becomes acquainted with the phraseology and language of the poetry and with its rhythms.
They become a part of him and his young mind begins to remember the tales and to form
his thought in the pattern of the song. The process in the early stage is as unconscious as a
child learning to speak, when he first listens to the sounds his elders are making”; cf. Lord
(1960), 22.
Page 26
86
legacy of minstrelsy in the medieval courts—“may be of great moment in maintaining
a tradition”.
113
3.38
Poetic diction, while it is in constant flux alongside the spoken language, changes in
some ways more slowly, protected as it is by the ritual of performance and by
professional convention. Many archaic words and hapax legomena are preserved in
Homer, the precise meanings of which the
himself may not have known,
feeling only “the atmosphere and the fragrance, and of course the actual magic, that
clung about them.”
114
The very purpose of the tradition was conservation: the
singers kept alive the memory of the past, the deeds of old heroes, and the technique
and musical lore of the ancient teachers. Thus Mnemosyne (Memory)—the mother
of the Muses—was deified by the singers, personifying the unfathomable depths of
musical tradition. To break with the technique of one’s elders would destroy the
medium which preserved these
. This mission of conservation was most
rigorous in Vedic tradition, where its success over the millennia is easy to measure.
Here too, however, unintentional, cumulative change is shown by the disagreement of
ancient theory and current practice.
115
3.39
Thus the melodic element of an Indo-European musical tradition could well have
evolved in strict company with, and no faster than, the metrical element, poetic diction,
and the language itself. In each case this evolution was so conservative as to allow
only unintentional and cumulative change—paradoxically through the contributions
and influence of individual singers—to the tradition’s essential features. This is the
essence of historical language kinship, for despite the huge differences between the
Greek, Slavic, and Indic languages, the proven kinship between them, and the essential
sameness which lurks just beneath the phonological detritus, is more striking still. In
Indo-European metrics, the basic distinction between long syllables and short is one
constant, and certain fundamental patterns survived in recognizable form. On the
basis of the Greek and Slavic parallels, it seems possible that some sort of accent- or
word-melody may have been an equally stable part of Indo-European oral
composition.
3.40
But was limited melodic compass one of the essential qualities of Indo-European
song? This would follow on the hypothesis of accent-melody, since, according to
113
Lord (1962), 181f.; for the young singer’s training, Lord (1960), 20-29; for the singer’s
status in the medieval courts, Lord (1960), 16.
114
Murray (1927), 42f.
115
See Fox-Strangways (1914), 246f.
Page 27
87
Dionysius of Halicarnassus, the Greek pitch-accent spanned an approximate (i.e. non-
resonant) fifth.
116
If so, under what conditions would melodic range change? Why
do we find the Greeks using heptatonic scales in the Archaic period, but vestiges of
narrow and ancient melody in pockets of the Balkans today? Bartók’s research
provides the key. As he discovered, songs with the archaic pan-Slavic qualities were
distributed more densely in the territories where Turkish influence had been minimal;
by contrast, melodies of wider range were found in areas, such as Bulgaria, where
there had been active Turkish settlement for some centuries.
117
A second and more
erosive force has been European art music, flowing outwards from the cosmopolitan
cities. In areas less subject to this diatonic stimulus, songs of restricted melodic scope
persisted, albeit dwindling steadily. This is a good illustration of how a resonant tone-
structure may serve as an international musical standard, a ‘metric system’ to which
archaic intonation may be assimilated (
cf. 2.8
). The same phenomenon has been
observed in modern Greek folk music.
118
3.41
With the expansion of melodic range, the possibilities of tonal contextualization
greatly proliferate. The traditional melodies of the Balkans had often come to be
heard as fragments of diatonic scales, their ancient tonal contexts forcibly reinterpreted
against the bimodality—major and minor—of Western art music.
119
At the same
time, syncretism involves mutual adjustment. These diatonic scales have been adapted
to local needs, whereby typically Slavonic ‘accidentals’ have been superimposed on
the diatonic substrate to create such syncretic pitch structures as the octatonic scales
used by Bartók and Stravinsky. I know of one Rumanian melody which uses the
sequence
D-C#-C-A
—identical to an ancient Greek chromatic tetrachord. These
parallels are important for understanding the Greek music of the post-Orientalizing
116
D. H. Comp. 11 (126.3f. Roberts):
(“Now the melody of speech is
measured by one interval, closest to that which is called a ‘fifth’”). It is important that the
interval is made approximate: it divorces pitch accent from the tonal intonation proper to
resonant intervals. Thus, any attempt to recite Greek language or poetry by modulating the
voice woodenly between musical intervals does no justice to the subtle tonality implied
here.
117
Bartók also noted the difference between rural and urban Turkish styles. Both use melodies
of wide scope, but the peasant style is of Central Asiatic origin, while the urban is heir to
the Near Eastern tradition of octave scales: see Bartók/Lord (1951), 55 n. 45. The latter
surely goes back to the heptatonic tradition of Mesopotamia.
118
Beaton (1980), 9.
119
Cf. Bartók/Lord (1951), 59-60; Wiora (1959), 203.
Page 28
88
period. South Slavic song demonstrates the self-sufficiency of ancient oral
composition, proving that there was no evolutionary imperative towards an expansion
of melodic range, the ‘perfection’ of melodic intonation in conformity with resonant
tone-structures, or even the crystallization of accents into precise melodic pitches.
Limited compass and vagrant, non-diatonic intonation were, broadly speaking, perfect
and unevolving traits within the evolving tradition, invulnerable in the absence of the
external stimuli which might induce mutation.
3.42
If the Slavic parallels are valid, the melodic tradition of those Indo-Europeans who
came to Greece should have pursued its course until deflected by similar
circumstances. Accordingly, the expansion of melodic compass, the standard use of a
seven-stringed lyre, and the primacy in later theory of a resonant tone-system, must be
explained by the Greeks’ contact with a cultural sphere in which these things were
standard. Mesopotamian diatony, as we have seen, was the core of such a tradition,
already constituting a system of great refinement—one might say completion—and
disseminated widely throughout the Near East. If this musical culture was relatively
stable throughout its range, as the texts indicate (
cf. 1.6
), the Greek and Indo-Iranian
encounters with the Mesopotamian musical sphere involve a constant which may help
to explain such Greco-Indian parallels as those noted by Fox-Strangways (1914).
For the Greek Orientalizing period itself, the relationship of Cyprus and the Aegean
periphery to the central and expanding superpower of eighth-century
Assyria—carrying the cumulative weight of a millennium and more of Mesopotamian
culture—might be compared with the rapid modification of hundreds of ancient
musical traditions which is going on today, due to the ‘global village’ effect.
120
The
Balkan material studied by Bartók is itself a good illustration of this. Apart from the
analogous cultural dynamics of the ancient and modern situations, diatony features in
each case as an essential musical catalyst.
3.43
As I see it, then, the Terpandrean tradition documents the Greek musical experience of
the Neo-Assyrian period. The model presented in this chapter is reductive, of course,
120
See especially Nettl (1985), 20, who distinguishes two broad levels of response, which give
rise to a wide array of syncretic phenomena. “Modernization” is “the incidental movement of
a system or its components in the direction of Western music and musical life, without,
however, requiring major changes in those aspects of the non-Western tradition that are
central and essential.” “Westernization” is “the substitution of central features of Western
music for their non-Western analogues, often with the sacrifice of essential facets of the
tradition”. If one substitutes ‘Oriental’ and ‘Orientalization’, many of the cases surveyed by
Nettl offer stimulating ways of thinking about musical change in the Greek Archaic period.
Page 29
89
and pre-Orientalizing Greek music must have been more varied than it allows. The
tradition of oral composition, to which the model is most relevant, was but one element
of a larger musical culture; the Homeric art was but one subtradition of this,
descending from a specialized development of the Mycenaean period. It is essential to
remember, however, that the hexameter derives from the same matrix as the lyric
metres of Lesbos, and this strongly supports the idea of a unified musical tradition
transcending generic distinctions. Likewise, Bartók was able to identify universal pan-
Slavic characteristics—the same limited range and non-resonant intonation which I
have proposed as characteristic of the Hellenic tradition. The fundamental problem in
understanding Homeric song is that we simply do not know what it sounded like, and
never will. All the same, it is safe to say that Greek music as a whole in the Geometric
period conformed, in the most general terms, to a ‘system’ of its own which
contrasted sharply with Near Eastern diatony, and that these disparate elements could
form the basis of a lasting musical syncretism. Clearly, we cannot suppose that every
expression of this musical culture followed the conventions of “four-voiced song”.
Yet oral composition was central to this tradition, being both its high art music and a
key instrument of cultural preservation. The Ionic epic
was, besides, the
preeminent art-music of the eighth century, and the
, as a professional class,
would have been the principal adapters during a large-scale musical movement.
Terpander’s juxtaposition of
and
thus brings
into focus the principle musical forces involved in the melic revolution, and reveals two
key elements of the Greco-Asiatic syncretism.