Comparative Studies in Greek and Indic Meter

Appendix B. Dovetailing: Speculations on Mechanics and Origins

Similarly, I posit a conflation of two types of suture to explain the origins of the iambic trimeter of Attic Drama:

1̄̆ 2̄ 3̆ 4̄ + 1̄̆ 2 3 4 5 6 7 8 =
1̄̆ 2̄ 3̆ 4̄ 5̄||6 7 8 9 10 11 12
(with trace of forward dovetailing)
1 2 3 4 5 6 7 8̄̆ + 5̆ 6̄ 7̆ 8̄̆ =
1 2 3 4 5 6 7||8̄ 9̆ 10̄ 11̆ 12̄̆
(with trace of backward dovetailing)


The products of non-dovetailing, …4|| and …8||, are not extant in the dodecasyllables of Attic Drama; they are clearly preserved, however, in the dodecasyllables of the Rig-Veda.

By contrast, Indic has preserved catalexis in trimeter, since there are frequent interrelations of dodecasyllabic Jagatī and hendecasyllabic Triṣṭubh in the Rig-Veda. These interrelations, however, are interpreted differently by Oldenberg, who thinks that the dodecasyllable is derived from the hendecasyllable rather than vice versa. [13] Granted, the relationship between acatalectic and catalectic trimeter verses may have become reinterpreted in Indic versification as an additive process (11 + 1 vs. 11) instead of a subtractive process (12 vs. 12 – 1). Consider the Rig-Vedic meter known as ātijagatī (8.97.13, etc.), which consists of 12 + 1 syllables and which presupposes that the Jagatī dodecasyllable equals a Triṣṭubh hendecasyllable + 1. [14] Nevertheless, it does not follow that the original relationship between the Indic dodecasyllable {285|286} and hendecasyllable was likewise additive. The comparative evidence of the cognate Greek trimeters suggests the opposite, and the Gāthic evidence adduced by Oldenberg [15] is inconclusive. Admittedly, there seem to be no Gāthic dodecasyllables with caesura pattern …4||, only hendecasyllables. [16] On the other hand, there are Gāthic dodecasyllables with caesura pattern …7||, [17] for which there is a direct analogue in the dodecasyllables of Greek iambic trimeter. [18] Since the patterns …4|| in 11 and …7|| in 12 are mutually irreconcilable and since neither of the two can be derived directly from the other, the absence of …4|| in 12 (or of …7|| in 11) would be of no probative value.

I conclude this phase of my argumentation with a synopsis on forward and backward dovetailing:






Indic Greek
backward dovetailing inoperative, but traces in the metrical segmentation after syllable 7 in trimeter backward dovetailing inoperative, but traces in the metrical and phraseological segmentation after syllable 7 in trimeter
also, traces in the catalexis of syllable 8 in dimeter and of syllable 12 in trimeter forward dovetailing inoperative, but traces in the caesura pattern …5|| of trimeter also, traces in the synchronic catalexis of syllable 8 in dimeter and of syllable 12 in trimeter forward {286|287}dovetailing operative; also, traces in the caesura pattern …5|| of trimeter

It bears repeating that besides being synchronically operative in attested Greek Lyric, forward dovetailing is also attested in synchronically inoperative phases. In a given pattern M⏓||^N (where ⏓ designates a final syllabic addition to the meter M and ^ designates an initial syllabic subtraction from meter N), the underlying plain meters M and N can be perceived only so long as M⏓ and ^N remain together with intervening ||; once separated, however, M⏓ and ^N can become reinterpreted as new meters E and F, no longer synchronically connected with plain M and N. True, E = M⏓ via hypersyllabism and F = ^N via acephaly, but these are diachronic rather than synchronic equations. Forward dovetailing is the synchronic mechanism, while hypersyllabism in E and acephaly in F are products of forward dovetailing only after it has ceased to be synchronically operative in a given instance.

I propose that we see an example of meter E in the Hipponacteum, ⏓ ⏓ – – ⏓, [19] derived from the Glyconic ⏓ ⏓ – ⏓ (meter M) via hypersyllabism; an example of meter F is the Telesilleion, ⏓ – ⏓, [20] likewise derived from the Glyconic, {287|288} via acephaly. But now an additional complication is noticeable: where M = N, three rather than two segmentations are possible. Since both the Greek and the Indic evidence indicate that identical dimeters can occur in a series,

M||M||M…||M||,


let us consider what would happen when forward dovetailing is applied:

M⏓||^M⏓||^M⏓…||^M||


If M = Glyconic, then I would imagine the following original contrast between a plain series of M vs. a dovetailing series:

⏓ ⏓ – ⏓||⏓ ⏓ – ⏓||⏓ ⏓ – ⏓…||⏓ ⏓ – ⏓||
⏓ ⏓ – ⏓ ⏓||⏓ – ⏓ ⏓||⏓ – ⏓ ⏓…||⏓ – ⏓||


The final segment ⏓ – ⏓, according to this hypothetical scheme, becomes the Telesilleion, while the Hipponacteum results from the first segment, via generalization of ⏓ ⏓ – – ⏓ over a potential ⏓ ⏓ – ⏓. But the latter type of pattern is still apparently attested as a synchronic combinatory variant of the former in the stanza structure of Alkman’s Partheneion (1 P). Consider this contrast between lines 35 vs. 49 and 77: [
21] {288|289}

ϝέργα πάθον κακὰ μηcαμένοι

τῶν ὑποπετριδίων ὀνείρων
ἀλλ’ Ἁγηcιχόρα με τείρει

– –
– – – – –


As for the medial segment ⏓ – ⏓ ⏓, I propose that it may become generalized as either ⏓ – – ⏓ or ⏓ – ⏓. The first pattern is likewise found as a component in the stanza structure of Alkman’s Partheneion. Consider line 37:

ὁ δ’ ὄλβιοc ὅcτιc εὔφρων
– –


This octosyllabic type is well known for the Lesbian poem (Sappho 94D = Adespota 976P) which begins:

δέδυκε μὲν ἀ cέλαννα
– –


The alternative generalization which could have stemmed from the medial segment of my hypothetical schema, ⏓ – ⏓, has the shape of a meter known as the Prosodiakon, a basic component in the Dactylo-Epitrite versification of Pindar and Bacchylides. Significantly, the segment ⏓ – – ⏓ is still attested as a variant of ⏓ – ⏓. For example, the former pattern is {289|290} joined with a preceding Lekythion (– – ⏓ – ⏓) in the stanza structure of Alkman’s Partheneion (lines 36-37):

ἔcτι τιc cιῶν τίcιc
ὁ δ’ ὄλβιοc ὅcτιc εὔφρων

– –


The latter pattern, ⏓ – – ⏓, is the equivalent of ⏓ – ⏓ in such familiar segments as Pindar’s Pythian 1.1:

χρυcέα φόρμιγξ Ἀπόλ-
λωνοc καὶ ἰοπλοκάμων
– – –
– –


As is his wont in constructing Dactylo-Epitrites, Pindar here has made a word-bridge where tradition calls for a word-break. In fact, the essence of Dactylo-Epitrite versification is that distinct metrical members have eventually become fused members by way of word-bridging. [
22]

The closing of the octosyllabic segment ⏓ – – ⏓, as Watkins has noticed, [23] is exactly opposite to that of the Glyconic ⏓ ⏓ – ⏓: that is to say, trochaic … – – ⏓ vs. iambic … ⏓. It remains now to apply the hypothetical schema {290|291} of sequential forward dovetailing to the plain iambic dimeter in Greek. The obvious result is the trochaic dimeter:

⏓ – – ⏓ – ⏓||(⏓ – – ⏓ – ⏓||…)
– ⏓ – – ⏓||(– – ⏓ – – ⏓||…)


The corresponding result in Indic would have implications for trimeter as well as dimeter:

csigim-appBfig2
csigim-appBfig3


An iambic dimeter sequence 8 + 8 + 8 (= 24), if it is resegmented as 12 + 12 (= 24), would lead to an iambic trimeter of the shape symbolized by bracket A. Bracket B symbolizes a stage transitional to a shift from iambic trimeter to trochaic trimeter, which in turn is symbolized by bracket C. The caesura pattern …5|| is, according to this hypothetical construct, a reflex of such a transitional segment B, with 5 + 7; the latter results from the model of segmentation A, with 4 + 8. This 4 + 8 of segmentation A would be in turn a sequential result of segmentation A’, with 8 + 4. [
24] A conflation of the patterns labeled B and C would result in a Triṣṭubh hendecasyllable with caesura {291|292} pattern …4|| or …5|| and with the last five syllables shaped …⏓ – – ⏓. The Triṣṭubh sequence …⏓ – – ⏓ is identical to the opening of segmentation B, [25] or in other words, to the sequence preceding caesura at …5||.

The unit ⏓ – – ⏓ is also attested in Greek as the trimeter opening, that is, the sequence preceding caesura at …5|| in iambic trimeter. I propose that the alternate pattern …4||, well attested in Indic, was once present but has been ousted in the standard Greek iambic trimeter of Attic Drama; the caesura pattern …8||, also well attested in Indic, would likewise have been ousted, while …7|| would have been kept as an alternative to …5||. [26] But there is internal Greek evidence for an earlier state of affairs. In the most archaic iambic trimeters attested, namely the corpus of the so-called Iambographoi (notably Archilochos, Semonides, Hipponax), there are survivals of …4|| and …8|| besides …5|| and …7||. Moreover, there is an important distributional constraint:

when …4||, then no …8|| but …7||:
⏓ – –||⏓ – ||– ̮̑⏓ –
when …5||, then no …7|| but …8||:
⏓ – – ̮̑⏓||– –||⏓ –


I cite the following respective examples: {292|293}

τὴν δ’ ἐκ κυνὸc λιτοργόν, αὐτομήτορα (Semonides 7.12W)

vs.

τῇ πάντ’ ἀν’ οἶκον βορβόρῳ πεφυρμένα (Semonides 7.3W)


The symbol  ̮̑ indicates a word-bridge, that is, a constraint against word-break; in the first and second examples above, the  ̮̑ moves the || backward and forward respectively, by one syllable: in other words, backward and forward dovetailing occurs. Taking the posited Indic 4 + 8 and 8 + 4 dodecasyllabic sutures [
27] as our comparative frames of reference, we are now in a position to reconstruct the following stages for Greek iambic dodecasyllables:

inherited sutures         4 + 8 and 8 + 4
Iambographoi             4||3||5 and 5||3||4
Attic Dramatists          7||5 and 5||7


The pattern 4||3||5 is a conflation of 4 + 8 (⏓ – – + ⏓ – – ⏓ – ⏓) and of 8 + 4 via backward dovetailing (⏓ – – ⏓ – ⏓ + ̮̑⏓ – ⏓), while the reverse pattern 5||3||4 is a conflation of 8 + 4 (⏓ – – ⏓ – ⏓ + ⏓ – ⏓) and of 4 + 8 via forward dovetailing (⏓ – – + ̮̑⏓ – – ⏓ – ⏓).

In sum, the iambic dimeter is to the trochaic dimeter

⏓ – – ⏓ –
– ⏓ – – ⏓

as the Glyconic is to the Prosodiakon

⏓ ⏓ –
⏓ –

as the Prosodiakon is to the Hemiepes’

⏓ –
– ⏓

Out of the 24 Pindaric lines, there is a word-break

at x 12 times
at y 4 times
at z 16 times

Out of the 12 Bacchylidean lines, there is a word-break

at x 9 times
at y 0 times
at z 7 times

If we consider y as the point of suture, then this pattern

xyz – –


can be described as D + D. In poetic genres other than Dactylo-Epitrite, we would expect such a suture D + D to be reflected as D||D, that is, by a word-break between the two segments

xy


and

z – –

And yet, the statistics of Irigoin show that Dactylo-Epitrite versification tends to observe this point y not by word-breaking (||) but by word-bridging (  ̮̑ ). The actual word-break is pushed one syllable to the right (z) or left (x). The point x marks a C + B division, and this C||B pattern of word-breaking should be contrasted with the simultaneous D ̮̑D pattern of word-bridging.

Other examples: the pattern {298|299}

xyz


is attested in 55 Pindaric lines and 11 Bacchylidean lines.

Out of the Pindaric lines, there is a word-break

at x 20 times
at y 2 times
at z 35 times

Out of the 11 Bacchylidean lines, there is a word-break

at x 5 times
at y 0 times
at z 7 times

Thus points y and x mark the simultaneous patterns d ̮̑D and c||B respectively.

The pattern

⏓ – xyz – –


is attested in 49 Bacchylidean lines. There is a word-break

at x 26 times
at y 1 time
at z 27 times

Thus points y and x mark the simultaneous patterns B ̮̑d and A||b respectively.

Earlier versification built from ABCD and abcd does not operate on the principles of both word-bridge and word-break: rather, the latter factor suffices. Among the most notable meters of this earlier genre are

Enkomiologikon          – –||– – – ⏓        (= C||b)
Iambelegos                ⏓ – – –||– ⏓        (= b||C)


One sure reason for their preeminence was their archaic dodecasyllabism. Because of the regular word-break, such metrical types are called asynartetic. Typical of a transitional phase between asynartetic and synartetic phases is the approach of Anakreon to versification with the Enkomiologikon: he alternates C||b with an optional variant D||d:

– –||– – ⏓ {300|301}


At times patterns latent in Dactylo-Epitrite versification reveal affinity to familiar patterns of other genres. As an example for comparison, let us consider the following two segments from Pindar (Nemean 1.9 and Pythian 9.22):

κείνου cὺν ἀνδρὸc δαιμονίαιc ἀρεταῖc
θῆραc ἦ πολλάν τε καὶ ἡcύχιον


Diachronically, these are a + A and c + A respectively:

⏓ – – – –
– – –


(When a + B = b||C, then the latter corresponds to the asynartetic Iambelegos.) The A-component, or Prosodiakon (⏓ – ⏓), has a diachronic octosyllabic variant likewise generated from the parent Glyconic; this variant has already been discussed in the context of the Lesbian poem which begins


If we designate this octosyllabic type (⏓ – – ⏓) as A’ vs. the A of the Prosodiakon, then the formula a + A’ yields the famous Alcaic verse (384LP):

ἰόπλοκ’ ἄγνα μελλιχόμειδε cάπφοι
– ⏓ – – – {301|302}


The caesura pattern b||C’ here corresponds to the plain b||C of the Iambelegos. Even more significant, the formula c + A’ yields the familiar Sapphic hendecasyllable (– – ⏓ – – ⏓), as in line 1 of Sappho 1 LP:

ποικιλόθρον’ ἀθανάτ’ ʼΑφρόδιτα


The caesura pattern c||A’ frequently results, as in line 2 of Sappho 1:

παῖ Δίοc δολόπλοκε λίccομαί cε


Word-breaks may also occur after syllable 4 (hence d||C’), and word-bridges are not observed with any regularity. Given these data, we are now in a position to offer an explanation for why the Sapphic hendecasyllable does not operate on the inherited trimeter caesura system of …4|| and …5||. The reason, I suggest, is that the Sapphic hendecasyllable is not a catalectic but an acephalic dodecasyllable.

Footnotes

[ back ] 1. For the basics on this term and on the phenomenon itself, see Maas and Lloyd-Jones 1962:44; cf. also Irigoin (1956 and 1957).

[ back ] 2. See pp. 37-48.

[ back ] 3. Maas 1962:44.

[ back ] 4. See pp. 171f and 174f.

[ back ] 5. The pattern …7|| of Greek iambic trimeter has an alternative, …5||, which will be discussed presently.

[ back ] 6. See pp. 176f.

[ back ] 7. See especially fragment 717P.

[ back ] 8. Fragment 935P.

[ back ] 9. Jakobson 1952:27.

[ back ] 10. See pp. 280f.

[ back ] 11. Cf. Wilamowitz 1921:248f.

[ back ] 12. Arnold 1905:161.

[ back ] 13. Oldenberg 1888:43f.

[ back ] 14. Oldenberg 1888:44n2.

[ back ] 15. Oldenberg 1888:43f.

[ back ] 16. Ibid.

[ back ] 17. Ibid.

[ back ] 18. See pp. 280f.

[ back ] 19. See pp. 283f.

[ back ] 20. Ibid.

[ back ] 21. We must make allowances here for Alkman’s regularization of the Aeolic base, ⏓ ⏓, into – – and – . For a discussion of this innovating convention, where we see the Aeolic base being replaced by the dactyl, see p. 55.

[ back ] 22. Wilamowitz puts it this way (1921:421): “Die Daktyloepitriten sind von Hause aus wirklich Asynarteten, aber allmählich haben die Dichter sich erlaubt, die Diäresen zu vernachlässigen, und so sind sie ἀcυνάρτητα cυνηρτημένα geworden.”

[ back ] 23. Watkins 1963:207.

[ back ] 24. See again pp. 167-190, where I originally derived the trimeter from a combination of 8 + 4 and 4 + 8 patterns.

[ back ] 25. See pp. 176f.

[ back ] 26. See p. 282.

[ back ] 27. As reflected in the caesura patterns 4||8 and 8||4.

[ back ] 28. See pp. 279-281, 287f. The dovetailing of a Glyconic results synchronically in … – u – ⏓||, not … – ⏓||. I am claiming only that the latter had also been possible once, from the diachronic standpoint. The pattern … – ⏓|| could have evolved once the Glyconic was perceived as consisting of an Aeolic base plus two dactyls (see p. 48).

[ back ] 29. Maas 1962:34.

[ back ] 30. For statistics, see Irigoin 1953:64.

[ back ] 31. As listed on p. 294.

[ back ] 32. Irigoin 1953.

[ back ] 33. Irigoin 1953: 18-26.

[ back ] 34. See p. 289. For arguments that this verse was actually composed by Sappho, see Clay 1970.