Song 44 of Sappho revisited: what is ‘oral’ about the text of this song?

2016.08.31 rewritten 2024.02.13 | By Gregory Nagy

§0. The phraseology that we see in Song 44 of Sappho reveals the same kind of formulaic structure that we also see at work in the Homeric Iliad and Odyssey. Such a structure, in the case of Homeric poetry, indicates that this poetry originated from traditions of oral performance. So also in the case of the songmaking exemplified by Song 44 of Sappho, there is evidence for a parallel origin from oral traditions.[1]

DeChirico_Hector and Andromache_1926
“Hector and Andromache” (1974), Giorgio de Chirico (1888–1978). Rome, Fondazione Giorgio e Isa de Chirico. Image via flickr.

§1. If we read in isolation the ancient Greek text known as Song 44 of Sappho, our first impression of the form and content of this composition is that we see here an example of epic poetry, not of lyric songmaking. A superficial inference, then, would be that this lyric song is heavily influenced by epic poetry. I argue, however, that Song 44 of Sappho is a lyric composition that is independent of, though interactive with, the traditions of epic composition that we know primarily from reading the texts of the Homeric Iliad and Odyssey. The evidence for this argument comes primarily from the formulaic structure of Song 44.

§2. To make this argument, I first need to offer my own working definition of the terms formula and formulaic. How is a structure formulaic? For an answer, I explain these terms formula and formulaic as referring to traditional forms of phraseology that follow the linguistic rules of oral poetic diction, and I explain oral poetic diction as a special language that interacts with the meters and/or the melodic contours of poems or songs.[2] Examples of such special language, evolving out of everyday language, have been studied most incisively by Milman Parry (especially 1928ab, 1930, 1932) and Albert Lord (1960, 1991, 1995, 2000), especially with reference to Greek and South Slavic oral traditions.[3]As Albert Lord (1960:47) observes, for example, on the basis of his systematic analysis of a sample poem stemming from South Slavic oral traditions that survived into the twentieth century, ‘There is nothing in the poem that is not formulaic’. Lord’s teacher Milman Parry (1928a:10–11=1971:8–9) made a comparable observation about ancient Greek oral traditions, with reference to the all-pervasive formulaic system of Homeric poetry.

§3. With this working definition in place, I proceed to synthesize arguments that I have presented in earlier work about the relationship between Song 44 of Sappho and the epics that we know as the Homeric Iliad and Odyssey.

§4. The metrical structure of Song 44 is stichic and not strophic. The term ‘stichic’ refers to a structure that goes from verse to verse, as distinct from the term ‘strophic’, which refers to a structure that goes from stanza to stanza. The stichic rather than strophic meters of ‘lyric’ are actually attested as usable for extended narratives of a type parallel to ‘epic’, composed in the dactylic hexameter, which is the stichic meter of ‘non-lyric’ par excellence. A prime example of such a stichic structure in ‘lyric’ is Song 44 of Sappho, a composition that focuses on a heroic setting: the song recounts the happy moment when Hector and Andromache got married—before the siege of Troy ever began. This song is composed in a stichic meter, labeled glyc2da by experts in Greek metrics:[4]

#xx–uu–uu–uu–ux# = glyc2da

(The symbols I use here are: ‘u’ = short syllable, ‘–’ = long syllable, ‘x’ = short or long syllable, ‘#’ = word-break at the end/beginning of two consecutive lines.)

§5. All of Scroll II in the ancient canonical collection of songs attributed to Sappho was composed in this stichic meter, glyc2da (Hephaestion 7.7, p. 23.14–17 ed. Consbruch). The meter glyc2da, even though it is stichic in the case of Song 44, is cognate with various strophic meters of Lesbian ‘lyric’. An example of such a strophic meter is glyc1da in Song 94 of Sappho:

#xx–uu–uu–ux# = glyc1da.

§6. I argue that the meter of Song 44 is also cognate with the stichic verse of Homeric poetry, known as the dactylic hexameter. Here is the metrical shape of the dactylic hexameter:

#–uuuu–|uuuuuu–x# = hexameter with penthemimeral caesura

#–uuuu–u|u–uuuu–x# = hexameter with ‘trochaic’ caesura

#–uuuuuu–|uuuu–x# = hexameter with hephthemimeral caesura

#–uuuuuuuu|–uu–x# = hexameter with bucolic diaeresis

(The underline under uu means that the two short syllables uu can optionally be replaced by a long syllable, –; the four vertical strokes (|) that come between syllables uu or –u or u– indicate four different places where formulas can be expected to begin or end within the hexameter. In other words, the caesuras and the diaeresis are boundaries for formulas.)

§7. To recap the basics of my argument so far: the epic meter known as the dactylic hexameter is cognate with the lyric meter that we find in Song 44 of Sappho, which can be described superficially as a ‘dactylic pentameter’:

#xx–uu–uu–uu–ux# = glyc2da

§8. But my argument goes further. Not only the meters but also the traditional phrases that we find embedded in these meters are cognate. In other words, the phraseology of Song 44 is cognate with the formulas of Homeric poetry.

§9. In earlier work, where I compared systematically the traditional phrases embedded within the metrical frames of the ‘pentameter’ in Song 44 and of the hexameter in Homeric (and Hesiodic) poetry, I found a general pattern of matching phraseology shaped

…uu–uu#xx… in Song 44


…uu–uu|–x# in Homeric poetry.

A prime example is

…κλέοc ἄφθιτον# (xx)… in Song 44.4


…κλέοc ἄφθιτον | ἔcται# in Iliad 9.413–414.[5]

§10. Further, matching the phraseology

…κλέοc ἄφθιτον | ἔcται# in the epic meter of Iliad 9.413

is the phraseology

…κλέος ἄφθιτον | ἑξεῖς# in the lyric meter of Ibycus SLG 151.47.

In this case, the lyric meter is shaped

#xx–uu–uu–x# = pher1da.

§11. What is most remarkable in such matching patterns is that the matches in wording are independent of the matches in meter. To show this independence, I cite the case of

…ἐπ’ ἄλμυρον # πόντον… in Sappho Song 44.7–8,

where the matching phrase

…ἁλμυρὸν | ὕδωρ# in Odyssey 9.470 etc.

has an underlying variant …ἁλμυρὸν…πόντον as we can see from expressions like

…ἁλμυρὸς ἔτρεφε Πόντος# in Hesiod Theogony 107


…ἁλμυρὸς ἔνδοθι πόντος in Hesiod Theogony 964.

In other words, what happens here is that ὕδωρ displaces πόντος in certain environments, not the other way around. So, in this case, the diction of lyric preserves the basic configuration, while the diction of epic is evidently derivative.

§12. In the case of the correspondence between

‘lyric’ …κλέοc ἄφθιτον # (xx)… in Sappho Song 44.4


‘epic’ …κλέοc ἄφθιτον | ἔcται# in Iliad 9.413–414,

the further correspondence of these phrases with Sanskrit śráva(s)ákşitam in Rig-Veda 1.9.7 points to an Indo-European heritage for the formulas and meters of ancient Greek verbal art.[6] Viewed from the broader perspective of Indo-European poetics, the terms ‘formula’ and ‘meter’ can be readjusted here to accommodate such broader terms as ‘phrase’ and ‘rhythm’. I have devised a formulation with such a readjustment in mind:

At first, the reasoning goes, traditional phraseology simply contains built-in rhythms. Later, the factor of tradition leads to the preference of phrases with some rhythms over phrases with other rhythms. Still later, the preferred rhythms have their own dynamics and become regulators of any incoming non-traditional phraseology. By becoming a viable structure in its own right, meter may evolve independently of traditional phraseology. Recent metrical developments may even obliterate aspects of the selfsame traditional phraseology that had engendered them, if these aspects no longer match the meter.[7]

§13. Others too, besides myself, have argued that the meters found in the songs or ‘lyric poetry’ of Sappho—as also of Alcaeus, and in other Greek lyric poetry—are cognate with the meter of Homeric poetry. Examples include the works of West 1973a and b, Berg 1978, Tichy 1981, Haug and Welo 2001. What all these works have in common, despite many incompatibilities in methodology, is a general agreement that the multiple meters of Greek ‘lyric’ and the single meter of ‘epic’ are Indo-European in origin. But none of these works addresses this question: does ‘lyric’, like ‘epic’, have a formulaic structure?

§14. The work of Gentili and Giannini 1977 likewise argues that the meters of ‘lyric’ and ‘epic’ are cognate—but without positing an Indo-European origin. This work also challenges the positing of any ‘monogenetic’ origin of ‘epic’ meter from any single ‘lyric’ meter. Such a challenge applies to all the works mentioned in the previous paragraph, §13.

§15. The same challenge also applies to my earlier work.[8] But the arguments presented there were then supplemented by arguments that can be found in my later work, where the dactylic hexameter of ‘epic’ is explained as a convergence of two different formulaic systems contained in two different metrical systems found in ‘lyric’.[9] Those two metrical systems can be described in terms of the nomenclature applied to studies of the meters found in the lyric ‘poetry’ of Pindar: ‘Aeolic’ and ‘dactylo-epitrite’. The ‘Aeolic’ meters of Pindar are cognate with the meters found in the songs of Sappho and Alcaeus, while the ‘dactylo-epitrite’ meters of Pindar are cognate with the meters found especially in the songs of Stesichorus.[10]

§16. The dual ‘etymology’, as outlined in §15, of dactylic hexameter as the singular meter of ‘epic’ accounts for every major word-break pattern in this meter.[11] By contrast, the other etymologies as attempted in the works I listed in §13 cannot account for all these breaks.[12]

§17. There are similar problems with the ‘colon-theory’ of Fränkel 1960, where the operation of ‘cola’ is not correlated with the operation of ‘formulae’ as described in Parry 1928b.[13] Fränkel’s theory cannot account for the full range of formulaic variation in the making of Homeric verse.[14]

§18. I return here to the basics of my ongoing argument: not only the meters of Greek epic and lyric but also the traditional wording that we find embedded in these meters are cognate. In other words, the phraseology of a lyric composition like Song 44 of Sappho is cognate with the formulas of Homeric poetry. It follows, then, that the phraseology of Sappho Song 44 is also formulaic, like the phraseology of Homeric poetry, and that Greek lyric originates from oral traditions, just like Greek epic.[15]

§19. In the light of the formal similarities we see in the ‘lyric’ narrative of Sappho Song 44 with the ‘epic’ narrative of Homeric poetry, the question remains whether even the distinction between ‘lyric’ and ‘epic’ is useful here—beyond the simple fact that the medium of Song 44 must have been accompanied by a ‘lyre’ or its equivalent while the medium of Homeric poetry had become purely recitative. Stichic meters of narrative ‘lyric’ such as the glyc2da of Sappho Song 44, conventionally sung to the accompaniment of a stringed instrument, are doubtless more closely related than is the spoken meter of hexameter to the format of the South Slavic guslar who sings to the accompaniment of the gusle. In this connection, I quote this useful formulation: ‘If there was epic or heroic balladry in (say) 1600 [BCE], its characteristic verse was most likely the glyconic [= glyc], whose cognates are used in Sanskrit and Slavic epic.’[16]

§20. In another project, I presented comparative evidence in arguing that Song 44 of Sappho shows the existence of ancient Greek traditions of women’s songmaking about the epic past.[17] In terms of that argument, such traditions of songmaking were distinct from but related to traditions of epic—which was a form of poetry performed by men, not by women.

A postscript to the featured picture for this essay

The featured picture for this essay is taken from one of the many versions of “Hector and Andromache” as painted by Giorgio de Chirico. This 1926 version is my personal favorite. There is something singularly arresting about the bare back of Andromache, revealed at the tenderly vulnerable moment when she sadly embraces Hector for the very last time. I add here as a coda, with the generous permission of the artist, Fyodor S. Wheeler, a picture that takes a second look at the picture made by De Chirico.

"Porcia and the Painting." Original work by Fyodor S. Wheeler. Image via.
“Porcia and the Painting.” Original work by Fyodor S. Wheeler. Image via.

We see again the vulnerable bareness of the sad woman’s back, but now this woman in the foreground is not Andromache but another sad woman who is looking at a picture of Andromache. This sad woman is Porcia, wife of Brutus. The story is told in Plutarch’s Life of Brutus. For quotation and translation, see Nagy 2015.08.12, as listed in the Bibliography. Brutus will soon die just as Hector will soon die in the Iliad, and Porcia is having herself a really good cry, just thinking about the sadness of it all while she views an ancient picture of that final embrace of Andromache and Hector. Accompanying Wheeler’s picture is his own story about the inspiration of the picture as visualized by De Chirico—and as visualized again and again in the story about Porcia and her viewing of that ancient picture of Andromache:

There’s a story from Plutarch that tells of Porcia, Brutus’s wife, crying every time she saw a certain painting of the parting of Hector and Andromache, as told in the Iliad Book VI. Hector didn’t come back. Brutus didn’t either.

My class finished reading the Iliad and ugh so many parts where I saw Hector and Andromache as Brutus and Porcia. Why must I be a classics fan.


Allen, W. S. 1973. Accent and Rhythm. Prosodic Features of Latin and Greek: A Study in Theory and Reconstruction. Cambridge.

Allen, W. S. 1987. Vox Graeca: The Pronunciation of Classical Greek. 3rd ed. Cambridge.

Barnes, T. G. 2011. “Homeric ΑΝΔΡΟΤΗΤΑ ΚΑΙ ΗΒΗΝ.” Journal of Hellenic Studies 131:1–13.

Barnes, T. G. 2013. “Drakeis, dedorke, and the Visualization of kleos in Pindar.” Harvard Studies in Classical Philology 107:73–98.

Beissinger, M., J. Tylus, J., and S, Wofford, S.,  1999. Epic Traditions in the Contemporary World: The Poetics of Community. Berkeley and Los Angeles.

Berg, N. 1978. “Parergon metricum: der Ursprung des griechischen Hexameters.” Münchener Studien zur Sprachwissenschaft 37:1136.

Clark, M. 1994. “Enjambment and Binding in Homeric Hexameter.” Phoenix 48:95–114.

Clark, M. 1997. Out of Line: Homeric Composition beyond the Hexameter. Lanham, MD.

Devine, A. M., and Stephens, L. D. 1984. Language and Metre: Resolution, Porson’s Bridge, and their Prosodic Basis.American Philological Association, American Classical Studies no. 12. Oxford.

Devine, A. M., and L. DStephens, L. D. 1994. The Prosody of Greek Speech. Oxford.

Duban, J. M. 2016. The Lesbian Lyre: Reclaiming Sappho for the 21st Century. West Hoathly (W. Sussex). On pp. 616–618: survey of views about whether the ἄφθιτον at I.09.413 is predicative or attributive. On p. 618: periphrasis of theoeikelos at I.24.258–259.

Fränkel, H. 1960. “Der kallimachische und der homerische Hexameter.” Wege und Formen frühgriechischen Denkens(ed. 2) 100–156. München.

Gentili, B., and P. Giannini. 1977. “Preistoria e formazione dell’esametro.” Quaderni Urbinati 26:7–37.

Haug, D., and E. Welo. 2001. “The proto-hexameter hypothesis: perspectives for further research.” Symbolae Osloenses76:130–136.

Lord, A. B. 1960. The Singer of Tales. Harvard Studies in Comparative Literature 24. Cambridge, MA.

Lord, A. B. 1991. Epic Singers and Oral Tradition. Ithaca.

Lord, A. B. 1995. The Singer Resumes the Tale (ed. M. L. Lord). Ithaca.

Lord, A. B. 2000. 2nd ed. of Lord 1960. Edited, with new Introduction (vii–xxix) by S. Mitchell and G. Nagy. Cambridge, MA. 3rd. ed., by D. F. Elmer, 2019. Hellenic Studies Series 77, Publications of the Milman Parry Collection of Oral Literature 4.

Meillet, A. 1923. Les origines indo-européennes des mètres grecs. Paris.

Nagy, G. 1974. Comparative Studies in Greek and Indic Meter. Harvard Studies in Comparative Literature 33. Cambridge, MA.

Nagy, G. 1990a. Pindar’s Homer: The Lyric Possession of an Epic Past. Baltimore. The Appendix at pp. 439–464 supplements the argumentation in Nagy 1974.

Nagy, G. 1990b. Greek Mythology and Poetics. Ithaca, NY. Revised paperback edition 1992.

Nagy, G. 1996. Homeric Questions. Austin.

Nagy, G. 2010|2009. Homer the Preclassic. Printed | Online version. Berkeley and Los Angeles.

Nagy, G. 2010. “Language and Meter.” Chapter 25, A Companion to the Ancient Greek Language (ed. E. J. Bakker; Blackwell Companions to the Ancient World 2010) 370–387.

Nagy, G. 2011. “The Aeolic Component of Homeric Diction.” Proceedings of the 22nd Annual UCLA Indo-European Conference (ed. S. W. Jamison, H. C. Melchert, and B. Vine) 133–179. Bremen. In Nagy 2012 v1.

Nagy, G. 2013. The Ancient Greek Hero in 24 Hours. Cambridge MA.

Nagy, G. 2015.02.27. “Song 44 of Sappho and the Role of Women in the Making of Epic.” Classical Inquiries.

Nagy, G. 2015.10.22. “Diachronic Sappho: Some Prolegomena.” Classical Inquiries.

Nagy, G. 2016.08.31. “Song 44 of Sappho revisited: what is ‘oral’ about the text of this song?” Classical Inquiries.

Page, D. 1955. Sappho and Alcaeus: An Introduction to the Study of Ancient Lesbian Poetry. Oxford.

Parry, A. 1971. ed. The Making of Homeric Verse: The Collected Papers of Milman Parry. Oxford.
GN – The first part contains Parry’s work on the Homeric texts, before he undertook his fieldwork research in the former Yugoslavia. The second part combines his experience in fieldwork with his expertise in the organization of Homeric poetry.

Parry, M. 1928a. L’épithète traditionnelle dans Homère: Essai sur un problème de style homérique. Paris. tr. in Parry 1971:1–190.

Parry, M. 1928b. Les formules et la métrique d’Homère. Paris. tr. in Parry 1971:191–234.

Parry, M. 1930. “Studies in the Epic Technique of Oral Versemaking. I. Homer and Homeric Style.” Harvard Studies in Classical Philology 41:73–147. repr. in A. Parry 1971:266–324.

Parry, M. 1932. “Studies in the Epic Technique of Oral Versemaking. II. The Homeric Language as the Language of Oral Poetry.” Harvard Studies in Classical Philology 43:1–50. repr. in A. Parry 1971:325–64.

Tichy, E. 1981. “Hom. ἀνδροτῆτα und die Vorgeschichte des daktylischen Hexameters.” Glotta 59:28–67.

West, M. L. 1973a. “Greek Poetry 2000–700 B.C.” Classical Quarterly 179–192.

West, M. L. 1973b. “Indo-European Metre.” Glotta 51:161–187.

[1] An earlier version of this essay appeared in Nagy 2016.08.31.

[2] The formulation here derives in part from Nagy 1990b:29–35.

[3] On oral lyric songmaking, see especially Lord 1995 ch. 2.

[4] Details in Nagy 1990a:49.

[5] Nagy 1974 ch. 5 and Nagy 1990a Appendix. The Appendix of Nagy 1990a at pp. 439-464 supplements the overall argumentation in Nagy 1974.

[6] Nagy 1974:145.  On the syntax of κλέοc ἄφθιτον | ἔcται # in Iliad 9.413-414 and of śráva(s)ákşitam in Rig-Veda 1.9.7, I offer both comparative and internal analysis in Nagy 1990a: 244-245 n126 and Nagy 1990b: 122-127.

[7] Nagy 1974:145; see also Allen 1973:13–14, 258; further analysis in Nagy 1990a:39–42. I take this opportunity here to record my firm conviction that the work of Allen 1973 is of lasting value, and it cannot be replaced by the later work of Devine and Stephens 1984 and 1994.

[8] That is, to Nagy 1974 ch. 5.

[9] Nagy 1990a: 439-464, the Appendix.

[10] For more on Pindar’s meters, see Barnes 2013; on lyric and epic meters in general, see Barnes 2011.

[11] The argument is presented in Nagy 1990a:439-464, the Appendix.

[12] See also Nagy 1990a: 461n116 for a rejoinder to Berg’s critique of Nagy 1974.

[13] Further observations in Nagy 1974 ch. 3.

[14] Nagy 1990b:29–35; see also Clark 1994, 1997.

[15] Nagy 1974 ch. 5. The formulaic behavior of the phraseology in Song 44 of Sappho is evident also in other ways. One example, which calls for further research, is the distribution of dative plural -οιc with or without word-final -ι, so, -οιcι vs. -οιc. The relevant forms are surveyed by Page 1955: 67. Two cases merit special attention: line-final φίλοιc in Sappho F 44 line 12 and line-final θέοιc at line 21. In line-medial position, such forms generally show -οιcι. By contrast, the definite article in the dative plural is τοιc and ταιc even in line-medial position. See also Page p. 208 on Alcaeus “G2” (now F 130) line 15, cυνόδοιcί μ’ αὔταιc, and on Sappho F 160 lines 1-2, τάδε νῦν ἐταίραιc | ταὶc ἔμαιc τέρπνα κάλως ἀείσω. Page notes a similar pattern of distribution in the phraseology of Archilochus and Anacreon: line-medial -οισι vs. line-final -οιc.

[16] West 1973a:188.

[17] Nagy 2015.02.27.