II. Hephaestion and Aristides

The ancient handbooks on metre and music do not comment directly on melody’s role in strophic responsion, but neither do they rule it out. In his work On Poems, Hephaestion defines responsion in very general terms (3.2 = 64.22–23 Consbruch): ‘Compositions in responsion (κατὰ σχέσιν) are those which the poet writes in submission to reprisal (ἀνταπόδοσις, lit. “giving out in exchange”) and recurrence (ἀνακύκλησις, lit. “circling back”).’Footnote ⁶ According to Johannes M. van Ophuijsen, the terms ‘reprisal’ and ‘recurrence’ could refer to the historical distinction between the binary stanza pairs of antistrophic song (which entail a sense of reprisal or exchange) and the multiple recurrences of a single stanza within monostrophic song.Footnote ⁷ Alternatively, given that Hephaestion presents these qualities as a pair (καί) rather than as alternatives, we might see ‘reprisal and recurrence’ as referencing the combination of audible and visual markers associated with antistrophic performance: the ‘reprisal’ of metre/song and the ‘circling back’ (ἀνακύκλησις) of the direction of dance.Footnote ⁸ In either case, however, Hephaestion’s description is concerned with the larger formal structure of responsion and does not clarify what role (if any) melody played in communicating that structure in performance.

A similarly general description of strophic responsion concludes book 1 of Aristides’ treatise On Music (1.29):

Aristides’ references to responsion between sequential sections of the strophe and antistrophe are again general enough to include both metre and melody. This final passage comes immediately after a section discussing metre, but it is worth noting that book 1 began with sections on harmonics and rhythmics. While the three sections are never explicitly integrated, the book’s closing comments on responsion could potentially look back to all three of the preceding sections, if responsion applied to rhythm and melody, as well as metre.Footnote ⁹

Neither Aristides nor Hephaestion explicitly mentions melody’s role in strophic responsion, but some clue might be found in Hephaestion’s description of the comic parabasis, where he distinguishes between sung and spoken forms of responsion (On Poems 8.2 = 73.5–10 Consbruch):

Hephaestion defines the antepirrhēma as a metrical response to the epirrhēma, but identifies the antistrophe’s responsion specifically in terms of song (τὸ τῷ μέλϵι ἀντίστροφον).Footnote ¹⁰ What that entails in terms of performance is not specified, but the context could suggest that it is something more than mere metrical responsion.

III. Aristotelian sources: Problems 19.15 and Rhetoric 3.9

The earliest unambiguous evidence for the repetition of melody in antistrophic song appears in book 19 of the Aristotelian Problems, which discusses questions related to harmonics.Footnote ¹¹ In Problems 19.15, the author addresses the question of why nomoi were through-composed, while choral songs were composed in antistrophic pairs (διὰ τί οἱ μὲν νόμοι οὐκ ἐν ἀντιστρόφοις ἐποιοῦντο, αἱ δὲ ἄλλαι ᾠδαὶ αἱ χορικαί, 918b13–14).Footnote ¹² The answer is exceptionally detailed and attributes the difference to practical considerations regarding performance: the degree of mimesis required and the skill of the performers. In explaining how mimesis and musical skill influence a song’s structure, the author indirectly makes clear that antistrophic responsion applied to melody as well as rhythm.

First, the author observes that nomoi were performed by a single professional competitor, who was capable of longer and more ‘multi-formed’ (πολυϵιδής) song, and who could therefore achieve a higher degree of mimesis in performance ([Pr.] 918b14–16). This heightened mimesis was facilitated specifically by the lack of melodic repetition (918b16–18):

According to Problems, the distinctive quality that allows solo songs to be more mimetic than antistrophic pairs is the use of a melody that is constantly changing (τὰ μέλη … ἀϵὶ ἕτϵρα γινόμϵνα). Such a distinction is only relevant if antistrophic song did not likewise feature a constantly changing melody and instead required that τὰ μέλη be repeated.

Choral songs, by contrast, are said to have required a more conservative approach to melody ([Pr.] 918b21–29):

The simplicity of choral song is described in terms of two accommodations. First, the songs were composed using modal melodies (ἐναρμόνια μέλη).Footnote ¹³ Second, they were structured in antistrophic pairs. We should note here that antistrophic pairs are presented specifically as a means to simplify the melodies (ἁπλούστϵρα … τὰ μέλη), rather than the rhythms. The section closes by listing the two qualities that make an antistrophic pair simple: there is ἀριθμός (‘number’, ‘proportion’, ‘set quantity’) and ‘it is measured by a single unit’ (ἑνὶ μϵτρϵῖται).

The translation above is based on the manuscript text, which is emphatic in its focus on melody, but an emended version of the final section of this passage has recently been taken as evidence that antistrophic responsion was purely metrical and did not extend to melody.Footnote ¹⁴ As transmitted, the final line of the passage reads: ἀριθμὸς γάρ ἐστι καὶ ἑνὶ μϵτρϵῖται. However, in Musici Scriptores Graeci, Carl Jan prints an emendation that has since become widely adopted: ϵἷς ῥυθμὸς γάρ ἐστι καὶ ἑνὶ μϵτρϵῖται.Footnote ¹⁵ The hypothetical corruption of ϵἷς ῥυθμὸς (‘one rhythm’) to the MSS ἀριθμὸς is textually plausible, if we infer that ϵἷς ῥυθμὸς was written with numerical alpha (ᾱ ῥυθμὸς) and that the homophony of υ and ι in post-Classical pronunciation facilitated further corruption.Footnote ¹⁶ Presumably working from this emended text, Armand D’Angour offers a translation that further emphasizes rhythm (‘antistrophes are simple: they have one dance-rhythm and one metre’), from which he argues that ‘It seems significant that the author here should mention the identity in strophic responsion of dance-rhythm and metre, but not of melody.’Footnote ¹⁷ The textual history makes this argument untenable, since the text has been altered precisely to produce the expected emphasis on metrical responsion. By contrast, the language of the unemended text is broad enough to encompass melodic as well as rhythmic repetition, as can be seen in comparable usages of its key terms, ἀριθμὸς and ἑνὶ μϵτρϵῖται.

The word ἀριθμός carries a wide breadth of meanings, but in the context of poetry, it is associated specifically with strophic responsion. According to Hephaestion’s On Poems (3.5 = 65.12–15 Consbruch), poetry composed in responsion (κατὰ σχέσιν) can be differentiated from non-responding poetry by the presence of an ἀριθμός that is ‘delimited’ (ὡρισμένος) or ‘prescribed’ (τϵταγμένος).Footnote ¹⁸ Hephaestion does not use this terminology for stichic poetry, suggesting that it indicates more than a fixed metre or rhythm, and while he sometimes uses it for discussing the ‘number’ of cola within a stanza, he does so only when describing the balanced symmetry of responding stanzas.Footnote ¹⁹ Aristotle likewise uses ἀριθμός as a term for balanced symmetry in Rhetoric 3.9 (discussed in detail below), and while his use of the term is somewhat more expansive, these passages provide an extremely apt parallel for that of Problems 19.15, without any need for emendation: antistrophic composition is simpler because there is a balanced symmetry between responding stanzas. While this sense of symmetry could refer only to metrical responsion, the language is broad enough to encompass melodic repetition as well, especially in the context of a passage concerned primarily with melodic variation.

The same is true for the subsequent statement that an antistrophic pair ‘is measured by a single unit’ (ἑνὶ μϵτρϵῖται). Rather than a reference specifically to metre (‘there is one metre’), the phrase turns out to be a frequent Aristotelian idiom, possibly originating in discussions of mathematical proportionality.Footnote ²⁰ For example, in Eudemian Ethics (1243b), Aristotle repeatedly uses this construction in discussing interpersonal relationships that are not directly reciprocal. The inability to apply a single standard (μϵτρῆσαι ἑνί) becomes difficult whenever two people seek to arrange a quid pro quo relationship (λογίζονται τί ἀντὶ τίνος). This includes students and teachers, who will inevitably get into disagreements because ‘knowledge and money do not share a unit of measure’ (ἐπιστήμη γὰρ καὶ χρήματα οὐχ ἑνὶ μϵτρϵῖται). As an idiom, the phrase describes things that are directly comparable and interchangeable, without carrying any association with rhythm or metre.Footnote ²¹

Accordingly, in Problems 19.15 ἑνὶ μϵτρϵῖται can be understood as describing the consistent parallelism between two stanzas, without any specific reference to metre. Taken as a whole, then, the passage’s closing sentence asserts that antistrophic pairs are characterized by balanced proportion (ἀριθμός ἐστι) and a sense of direct exchange (ἑνὶ μϵτρϵῖται). Used in a context that emphasizes the role of melody in defining antistrophic responsion, we can infer that this sense of balanced exchange was created, at least in part, by the careful repetition of melody.

This understanding of Problems 19.15 is corroborated indirectly by a combination of two later sections. In Problems 19.30, the author claims that certain melodic modes (the Hypodorian and Hypophrygian) do not allow for antistrophes and are therefore appropriate only for actor song. Section 19.48 addresses the exceptional nature of these two modes, explaining that they ‘have melody the least’ (μέλος ἥκιστα ἔχουσιν).Footnote ²² By attesting that less melodic modes would impede the use of antistrophic song structure, these passages indirectly affirm the role that melody played in communicating responsion to an audience.

The description of antistrophic form found in Problems 19.15 is also paralleled in Aristotle’s more secure works. In Rhetoric 3.9, Aristotle explains the contrast between the ‘continuous’ and ‘periodic’ styles through a musical analogy (3.9.1):

Because Aristotle assumes that his audience would already be familiar with dithyrambic preludes and antistrophic songs, he does not describe their musical structures directly. Instead, he proceeds to describe the structural features of the two rhetorical styles, which presumably resemble the musical forms with which he has identified them. While his comments do not address the role of melody in particular, they reinforce the presentation given in Problems 19.15 and provide broader insight into how antistrophic performance might frame the interpretation of tragic lyrics for an Athenian audience.

In Rhetoric 3.9.2, Aristotle defines the continuous style (λέξις ϵἰρομένη)Footnote ²³ as speech where the end cannot be anticipated from the structure, an arrangement that is ‘unpleasant on account of its boundlessness’ (ἀηδὲς διὰ τὸ ἄπϵιρον). This emphasis on length and variability recalls the contrast between strophic and astrophic song in Problems 19.15, as does Aristotle’s subsequent definition of the periodic style (λέξις κατϵστραμμένη)Footnote ²⁴ at Rh. 3.9.3:

Just as in Problems 19.15, Aristotle uses the term ἀριθμός to describe the proportionate balance that characterizes antistrophic pairs and, in this case, rhetorical periods. Here, too, he characterizes a balanced structure as easier to learn, comparing this outcome to the effect of writing in verse (τὰ μέτρα). While Aristotle mentions metre, it is important to note that ἀριθμός here is not associated with metrical rhythm: it is a quality found in both antistrophic pairs and periodic speech (in prose), but absent from dithyrambic preludes (in metre) and the continuous style. The passage makes clear, as Robert Fowler argues in detail, that ἀριθμός is a term for ‘the general principle of proportion and balance’.Footnote ²⁵

Aristotle expands his choral metaphor in Rhetoric 3.9.6, where he indirectly sheds light on the choreographic design of antistrophic stanzas. In describing the proper proportioning of rhetorical speech he compares the listener to someone moving as part of a group:

Aristotle asks us to imagine the listener as someone following the speaker as they walk along a circuit with a determined boundary (ὅρος). If the speaker stops before the turning post, the listener bumps into him; if the speaker continues on beyond the expected turning point (τέρμα), he leaves the listener behind. The imagery echoes a passage of the Protagoras (314e3–315b8), in which Plato describes the sophist walking back and forth across a courtyard as he lectures to a group that paces with him, gracefully parting and reforming with each change of direction. Plato compares the group’s coordinated alternations to a chorus (Prt. 315b2), indirectly affirming the choreographic model attested more directly in later authors: that the strophe and antistrophe were danced in alternating directions, giving antistrophic dance a balanced pattern of reversal.Footnote ²⁶ This connection between alternating motion and antistrophic dance is even more accessible in the context of Rhetoric 3.9, where antistrophic song is explicitly introduced as a frame of reference. In this sense, Aristotle extends the Platonic metaphor to make a new point: just as a speaker’s physical motion might resemble a chorus’ alternating dance, his language, too, should have the proportioned balance (ἀριθμός) of choral song.

The rhetorical value of this structural arrangement is made clear in the following section (Rh. 3.9.7), where Aristotle introduces the concept of antithesis.Footnote ²⁷ He illustrates this technique with a long list of examples featuring balanced two-part constructions (μέν/δέ, καί/καί, ἤ/ἤ, etc.), and concludes by commenting on their pleasurable quality (3.9.8):

This description complements Aristotle’s earlier comments on the pleasure of language presented in balanced periods. On the one hand, formal parallelism allows the listener to form an expectation based on the first clause for how the second clause will end, and thus to feel a sense of satisfaction and accomplishment when that expectation is met. At the same time, balanced oppositions spur the listener to compare or contrast ideas presented in corresponding positions, and to feel an intellectual pleasure in recognizing those correspondences.Footnote ²⁸ In comparing this periodic style to antistrophic song, Aristotle indirectly suggests that his audience would be used to experiencing these same pleasures during a tragic ode, in which each antistrophic pair would similarly combine parallelism and reversal in its formal structure.

Overall, then, the indirect description of antistrophic performance in Rhetoric 3.9 affirms the more explicit discussion in Problems 19, reinforcing the emphasis on balanced parallelism. The phrase Problems 19.15 uses to describe this quality, ἀριθμὸς γάρ ἐστι καὶ ἑνὶ μϵτρϵῖται, is general enough that it could refer to the larger sense of balance and parallelism found in both choral song and rhetorical periods. The aesthetic effect might be the same in both, allowing the audience to anticipate the conclusion and feel pleasure in recognizing the correspondence of elements. As Problems 19.15 makes clear, this formal effect was created specifically by the repeated melody (μέλος), in contrast to astrophic actor songs, where the lack of melodic repetition (τὰ μέλη … ἀϵὶ ἕτϵρα γινόμϵνα, 918b16–18) allowed for greater musical mimesis.

IV. Dionysius of Halicarnassus, On Literary Composition 19

The performative conventions hinted at in the earlier Aristotelian sources are stated explicitly by Dionysius of Halicarnassus. In discussing the use of variation (μϵταβολή) as a part of literary style, Dionysius describes the restrictions that antistrophic structure places on melodic variation (19.15–23):Footnote ²⁹

Dionysius is very clear that in responding strophes and antistrophes, the composer must not change the ‘song’ (οὐχ οἷόν τϵ ἀλλάξαι μέλος). Our understanding of antistrophic responsion therefore depends on the precise meaning of μέλος here. The prevalent interpretation, reflected in Stephen Usher’s Loeb translation, is that ‘writers of lyric cannot vary the melodies of strophe and antistrophe,’ that strophic responsion entails melodic repetition. An alternative interpretation has been proposed by Ruijgh, who suggests that μέλος here must refer only to the modal scale, rather than an actual melody.Footnote ³⁰ Dionysius himself mentions tuning systems (enharmonic, chromatic and diatonic) in the explanation that follows, but he also specifies that responding stanzas must preserve the same ‘directions’ (ἀγωγαί). To firmly establish what Dionysius’ testimony means for melodic responsion, we must first determine how he uses these two key terms: μέλος and ἀγωγαί.

Across On Literary Composition, Dionysius occasionally uses the term μέλος to refer to ‘song’ in general, and this is especially true in passages where he distinguishes between fully sung melodies and the carefully polished sound-patterning he recommends for rhetorical speech.Footnote ³¹ A more specific meaning is established in chapter 11, when he introduces the four core elements of pleasing speech: melody (μέλος), rhythm, variation and appropriateness. In discussing μέλος, Dionysius contrasts the natural intonation of spoken language with the stylized pitch patterns of sung music (Comp. 11.88–110). He illustrates this point with a detailed discussion of the opening melody of Euripides’ Orestes: moving syllable by syllable through several lines, he describes each melodic contour (where the pitch rises, falls or stays the same) without any interest in the scale system or the size of intervals used. Here, as throughout the rest of his discussion of μέλος, it is clear that he understands the term as indicating a specific sequence of pitch contours.

A possible counter-example to this usage might be found in On the Style of Demosthenes, in another passage discussing intonation melodies. According to Dionysius, Demosthenes studied ‘the melodies of words’ (τὰ μέλη τῶν ὀνομάτων, 48.33–34) and was puzzled to find that the same words did not always have the same effect, due to ‘differences in the melodies’ (τῶν μϵλῶν διαφοράς, 48.43–44). In the discussion that follows, Dionysius compares the adjustment of how words are pronounced to the tuning of musical scale systems, writing that intonation can subtly change the words’ effect, ‘just as chromaticism (χρῶμα) relates to mode (ἁρμονία) in musical compositions’ (Dem. 48.45–46).Footnote ³² This comparison makes clear that scalar tunings have a role in melody but does not suggest an equivalence between ἁρμονία and μέλος. Instead, Dionysius appears to understand melody (μέλος) as depending on two factors: the melodic contours (which in speech are determined by the accentual intonation of the words), and the scale system within which these contours are rendered (which the orator can shape in performance). In this sense, a scale system is one part of what defines a melody, but the melody itself is more narrowly defined by its specific sequence of pitches.

This two-part definition of μέλος fits neatly into Dionysius’ description of antistrophic responsion.Footnote ³³ After asserting that one ‘cannot change the melody’ (οὐχ οἷόν τϵ ἀλλάξαι μέλος, Comp. 19.16), he specifies that this restriction applies to both the scale system and the ‘directions’ (ἀγωγαί). In particular, he writes that in responding stanzas poets must ‘preserve the same directions’ (τὰς αὐτὰς ἀγωγὰς φυλάττϵιν).

The noun ἀγωγή has a range of meanings nearly as broad as the verb from which it is derived (ἄγω): it can refer to physical guidance, the upbringing of children or styles of oration and musical performance.Footnote ³⁴ Uniting these varied meanings is a basic sense of ‘direction’ or ‘guidance’ in setting a course of action.Footnote ³⁵ As a technical term in music theory, ἀγωγή can have two meanings, depending on its context. In discussions of rhythm, it denotes how the performer chooses to render the quantities of the notes: that is to say, the tempo of a musical performance. This is explicitly defined in several fragments of Aristoxenus, as well as in Aristides Quintilianus’ treatise On Music, and it may also be implied in several earlier sources.Footnote ³⁶ It is important to note, however, that the use of ἀγωγή in discussions of rhythmics is presented as a secondary meaning, which Aristides feels the need to distinguish as ‘rhythmic direction’ (ἀγωγὴ ῥυθμική, 1.19.1).

The primary meaning of ἀγωγή in the context of music theory is related to harmonics. Both Aristoxenus and Aristides explicitly define ἀγωγή as melodic movement through consecutive notes, in contrast to intervallic movement.Footnote ³⁷ Aristides introduces the term immediately following his description of the enharmonic, chromatic and diatonic genera (16.19–22):

In this context, ἀγωγή is used for a style of melodic movement within the notes of a given scale, particularly the use of adjacent notes in sequence.Footnote ³⁸ Aristides makes clear that these melodic styles are not part of the scale system (which should be established beforehand) but part of the process of composing a specific melody (ὅτϵ … ποιούμϵθα τὴν μϵλῳδίαν, 16.20).Footnote ³⁹

Later, in chapter 29, Aristides discusses the process of melodic composition in greater detail, breaking it into three steps: choosing the range (i.e. musical register), fitting together the scale and composing the specific melody (τῆς μϵλωδίας ἀπϵργασία, 29.18), which he labels with the term ‘usage’ (χρῆσις). As a part of ‘usage’, he identifies three forms of melodic motion, of which ἀγωγή (‘direction’) is the first. Within this category he identifies three subcategories of ἀγωγαί: ascending, descending and a combination of the two. While the plural form ἀγωγαί never actually appears in this passage, one might infer that it would describe a sequence of such ‘directions’ – that is, the rises, falls and turning points of a fully constructed melody.

This technical sense of ἀγωγή, found in both Aristoxenus and Aristides, seems exceptionally well-suited to the context of Dionysius’ description of antistrophic responsion. Dionysius uses the word in the context of harmonics rather than his later discussion of rhythmics, and he specifically echoes the theorists in introducing ἀγωγή as a compositional step that follows the choice of scale system, writing that ‘whether [composers] undertake enharmonic, chromatic or diatonic melodies, they must preserve the same “directions” (ἀγωγαί) in all the strophes and antistrophes.’ Admittedly, this usage of the term ἀγωγή is not directly parallel to that of Aristides, who uses it specifically for stepwise melodic movement, which he distinguishes from melodic repetition and intervallic movement. It is worth noting, however, that Aristides does not offer an alternative term for ‘melodic movements’ in general. Given that ἀγωγή has a broad inherent meaning (from ἄγω) and that stepwise motion appears to have been the most common form of melodic movement in practice, it would appear a convenient term to fill that gap.Footnote ⁴⁰

Dionysius himself appears to use ἀγωγή in this broader sense elsewhere, when he observes that Demosthenes ‘composes remarkable rises and falls of pitch, sometimes making the direction (ἀγωγή) more dignified, other times more lovely’ (Dem. 44.4–6). Here, ἀγωγή explicitly describes rises and falls of pitch, without any reference to the contrast between stepwise and intervallic motion.Footnote ⁴¹ Moreover, because Dionysius is discussing an author he knows only through writing, his description of pitch variations must refer to the patterning of pitch accents in the words themselves, rather than the pitch intervals with which Demosthenes originally chose to pronounce them.

Returning once more to Dionysius’ definition of antistrophic repetition, we see that the wording he chooses is consistently compatible with the long-standing interpretation of the passage as evidence for precise melodic repetition between strophes and antistrophes. To his knowledge, at least, responding stanzas repeat the same melody (μέλος), which is realized both in the choice of scale system and in the specific sequence of melodic rises and falls (ἀγωγαί).

V. Athenaeus and the Alphabet Tragedy of Callias

The main evidence cited in opposition to Dionysius’ testimony comes from another late source. The humorously titled Alphabet Tragedy of Callias appears to have been a late fifth-century Old Comedy with paratragic elements, though the date, authorship and even title are debated.Footnote ⁴² The play survives only via Athenaeus’ Learned Banqueters, where it is twice mentioned briefly (7.276a, 448b) and later described at length (10.453–54). Within this third passage, Athenaeus describes a choral song with a distinctive musical structure and closes with the puzzling claim that Euripides imitated this song in his Medea and that ‘on account of this, as it seems, the others all admitted the antistrophes from this source into their tragedies’ (διόπϵρ οἱ λοιποὶ τὰς ἀντιστρόφους ἀπὸ τούτου παρϵδέχοντο πάντϵς, ὡς ἔοικϵν, ϵἰς τὰς τραγῳδίας, 10.453e).

Ruijgh influentially took this claim as evidence that Euripides’ Medea was the first tragedy to use the same music in responding stanzas and that earlier strophic song did not feature melodic repetition.Footnote ⁴³ Because Ruijgh was working under the once widely adopted hypothesis that Athenaeus’ text was shortened by a breviator who left out small words, he ‘restores’ small words throughout his reading, in order to give the text what he views as its proper sense.Footnote ⁴⁴ Thus, he understands Athenaeus’ patently absurd claim that Medea was the first tragedy to introduce ‘antistrophes’ in general (τὰς ἀντιστρόφους, 10.453e) to be a shortening of ‘this kind of antistrophe’ (τὰς τοιαύτας ἀντιστρόφους), referring to the specific musical structure that Euripides is said to have borrowed from the Alphabet Tragedy ode.Footnote ⁴⁵

In Athenaeus’ description of this ode (10.453d–e), he repeatedly comments on the use of metre and melody together, suggesting that this was a distinctive element. He introduces the song by saying that the chorus consisted of women divided into pairs, singing lyrics that appear to build syllables out of letters: βῆτα ἄλφα βα, βῆτα ϵἶ βϵ, βῆτα ἦτα βη, etc., with each responding stanza using a different initial consonant.Footnote ⁴⁶ In describing the first strophe, he says that the song was ‘metrical and at the same time melodificated’ (ἔμμϵτρος ἅμα καὶ μϵμϵλοπϵποιημένος, 10.453d). This pairing is echoed at the transition to the second stanza, which Athenaeus refers to as ‘an antistrophe of the metre and of the song’ (ἐν ἀντιστρόφῳ τοῦ μέλους καὶ τοῦ μέτρου, 10.453d). After describing two stanzas, Athenaeus sums up the rest of the ode thus: ‘and likewise, on each of the remaining syllables, all the women have both the same metre and the same melody in the antistrophes’ (καὶ ἐπὶ τῶν λοιπῶν συλλαβῶν ὁμοίως ἑκάστων τό τϵ μέτρον καὶ τὸ μέλος ἐν ἀντιστρόφοις ἔχουσι πᾶσαι ταὐτόν, 10.453e).Footnote ⁴⁷ Here again, there is a ‘curiously superfluous’ emphasis on the pairing of metre and melody.Footnote ⁴⁸

There is clearly something distinctive about the use of metre and melody in the Alphabet Tragedy ode, and Ruijgh proposes that this is the introduction of melodic repetition. To facilitate this interpretation, he proposes several textual restorations in the final sentence (10.453e): he takes ‘both the metre and the melody’ (τό τϵ μέτρον καὶ τὸ μέλος) as the breviator’s shortening of ‘not only the metre, but also the melody’ (οὐ μόνον τό μέτρον ἀλλὰ καὶ τὸ μέλος), and he emends ‘all the women … in the antistrophes’ (ἐν ἀντιστρόφοις … πᾶσαι) to ‘all the antistrophes’ (αἱ ἀντίστροφοι … πᾶσαι).Footnote ⁴⁹ It is these changes that produce his proposed understanding of the key passage: ‘all the antistrophes have not only the same metre, but also the same melody.’Footnote ⁵⁰

It must be emphasized that the breviator theory which provided a basis for Ruijgh’s speculative adjustments of Athenaeus’ text has been universally rejected in more recent scholarship.Footnote ⁵¹ Without Ruijgh’s interventions, the emphasis is not on the addition of melodic responsion, but on the performance of responding sections by the same women, who are earlier described as being arranged in pairs (ἐκ τῶν σύνδυο, 10.453d). While the musical design of Callias’ song is beyond the scope of the present study, the unemended text suggests that the ode is better understood to have been distinctive for its combination of sung lyrics and metrical recitation, performed in responding positions by the same paired choreuts. Thus Athenaeus describes even the initial strophe as ‘metrical and at the same time melodificated’ and later refers to an ‘antistrophe of both the song and the metre’ (ἀντιστρόφῳ τοῦ μέλους καὶ τοῦ μέτρου, 10.453d), a phrase that parallels ancient descriptions of comic parabases, in which sung and chanted sections alternate within a single stanza.Footnote ⁵² In this case, we might infer that Medea was well known for its use of a similarly mixed performative style.Footnote ⁵³

Overall, Athenaeus’ unemended language gives no indication that either Alphabet Tragedy or Medea was remarkable for the introduction of melodic repetition in metrically responding stanzas, and such an innovation would appear very unlikely in the context of what is known of Euripidean music. Ancient authors often comment on Euripides’ tendency towards greater musical variety: he is said to loosen the formal restrictions followed by the earlier tragedians, and this is borne out in the lyric sections of his surviving plays, which feature an increasing variety of metres and a decreasing use of antistrophic responsion.Footnote ⁵⁴ The introduction of a new restriction, which requires the melody to repeat rather than vary freely, is hardly what one would expect of Euripides, such that there is no need to emend Athenaeus’ testimony to conform with this idea.

In fact, the scholarship arguing that melodic repetition was a Euripidean innovation has generally used Athenaeus as evidence not for melodic repetition itself, but for its supposed ‘corollary’: that while earlier tragedians based their melodies on the rise and fall of the pitch accents, Euripides chose to free his melody from the restriction of following the pitch accents. Such a proposition is more plausible as a potential Euripidean innovation, but it does not provide a logical motive for introducing melodic repetition: if Euripides wanted to disregard accents, he could easily do so without also introducing a new restriction on melodic variation. The connection between these two phenomena, while not strictly logical, has its basis in a history of scholarship linking the question of melodic responsion to that of pitch accents’ role in Classical melodies, as discussed in the next section. This connection is not, however, mentioned by Athenaeus, and it should not frame our interpretation of the testimony he provides.

VI. Pitch accents and melodic responsion

In the songs with musical notation that have survived from antiquity, composers often coordinate the accentual intonation of the lyrics with the rises and falls of their melody.Footnote ⁵⁵ Comments in Plato suggest that such a practice was characteristic of old-fashioned songs (Resp. 3.398d, 399e–400a), and in Aristophanes’ Frogs the character Aeschylus mocks Euripides for lyrics that distort pronunciation, suggesting that this was not a traditional approach.Footnote ⁵⁶ Nevertheless, it was long assumed that accent could not have played a role in the strophic songs of the Archaic and Classical periods, since the pitch accents of responding stanzas do not consistently fall in responding metrical positions.Footnote ⁵⁷ This assumption has also generally been accepted by scholars arguing in favour of accent-based melodies, who therefore assert that earlier strophic songs would have been through-composed without melodic repetition.Footnote ⁵⁸

This is, however, a false choice. As I have argued elsewhere, the principles for coordinating accent and melody are flexible enough that we would not expect to find perfectly responding accents even in songs where the same melody was precisely repeated with multiple sets of accent-coordinated lyrics.Footnote ⁵⁹ Rather than consistently imitating spoken intonation, ancient composers merely avoid contradiction between words and melody, such that flat melodic contours (i.e. repeated notes) can be set with different intonational contours in different stanzas.Footnote ⁶⁰ In the ancient melodies that survive, flat melodic contours are frequent: 22 per cent of notes are followed by a repetition of the same note.Footnote ⁶¹ If tragic melodies featured a similar rate of repeated notes, then responding stanzas could disagree in the direction of their intonation 22 per cent of the time. In a digital study of accentual responsion in the lyrics of Aeschylus, I found that just 20 per cent of metrical positions conflict in their intonational contour, making it perfectly possible that Aeschylus’ songs could both respect accent and precisely repeat the same melody in responding stanzas.Footnote ⁶²

There are, moreover, many passages of tragic lyric where the accents do trace the same melody in the strophe and the antistrophe, and these provide further positive evidence for the repetition of melody. In the Choephori kommos, for example, the accentuation in the actors’ final stanza pair (Cho. 456–60) allows for matching melodic movements in 90 per cent of metrical positions, including a responding circumflex accent.Footnote ⁶³ Likewise, in Sophocles’ Trachiniae, the chorus’ climactic lament for Heracles and Deianeira traces the same melody in both strophe and antistrophe for all syllables but one (96%), and Heracles’ subsequent monody is also extremely parallel in its accentual contours (92%).Footnote ⁶⁴ Such stanza pairs offer strong evidence for the use of melodic repetition, but accentual divergences in other responding lyrics do not necessarily provide evidence of melodic variation, since they may simply indicate repeated notes in the original melody. The rate of accentual correspondence increases across early tragedy, and I have therefore proposed that differences in accentual parallelism might be understood as indicating a contrast between an older, more monotonous style of song and a newer, more lively approach to melody.Footnote ⁶⁵ We might therefore view stanza pairs with closely parallel accent patterns (such as those discussed above) as distinguished by a melody in constant motion, in contrast to more monotonous passages, which allow for divergent accent patterns in the lyrics.

The melodic data preserved in pitch accents generally support Dionysius’ claim for precise melodic repetition, but they also provide some evidence for minor melodic variations in tragic antistrophes. Most significantly, Aeschylus’ antistrophes have 27 per cent more circumflex accents, on average, than their corresponding strophes.Footnote ⁶⁶ In the musical documents, circumflex accents are typically embellished with a two-note falling figure, presumably reflecting their natural pronunciation.Footnote ⁶⁷ The disproportionate presence of circumflex accents in antistrophes, which are metrically and stylistically identical to the preceding strophes, strongly suggests that ornamental melodic flourishes were used to add new interest when the melody was repeated a second time.Footnote ⁶⁸ This effect is especially prominent in the first stasimon of Sophocles’ Trachiniae (497–516), in which a series of anapaestic metra consistently repeats one accent pattern in the strophe and then changes to a new pattern in the antistrophe, dense in circumflex accents.Footnote ⁶⁹ Thus, while the overall accentual data provide strong evidence that melodic repetition was the norm in tragic antistrophes, they also leave room for small ornamentations and the occasional variation.