Naming the gods: traditional verse-making in Homer and Old Babylonian Akkadian poetry

: This is an investigation of character-naming expressions in early Greek (ca. eighth–sixth c. BC) and Old Babylonian Akkadian narrative poetry (ca. nineteenth– seventeenth c. BC). It compares the mentions of Zeus and Enlil (the Babylonian chief god)in Iliad Book8andOB Atra-hasis , andproposesathree-layeredclassificationsystem based on degrees of traditionality. The system involves metre and repetition parameters, and accounts for the techniques through which poets in both traditions made the mention sound venerable and ancient. Control cases include other characters in the Iliad (Diomedes, Hector) and OB Akkadian poetry (Isthar, Ea). The resulting figures are commensurate for the two traditions, supporting the hypothesis of a similar degree of orality-literacy interaction. The article seeks to offer a model for fine-grained cross-cultural literary criticism and verse study.


Oral transmission of Near Eastern poetry to the Aegean
Comparison between surviving ancient Near Eastern poetry (Sumero-Akkadian viz.Mesopotamian, Hurro-Hittite, Ugaritic and Hebrew) and the early Greek epic tradition has been developing substantially over the past few decades. 1 The hypothesis of direct or indirect reception of preserved Near Eastern poems by Homer is regarded with increasing caution, 2 and several scholars consider a broader, analogical perspective to be more productive. 3Because, however, the quality and quantity of similarities is such that a degree of interaction between poetic traditions (as opposed to individual works) is beyond doubt, the question of transmission remains open.
will turn to my case-studies and preliminary conclusions ( §4-5).
Oral transmission cannot have happened between Old Babylonian (OB: ca.2000-1595 BC) and Homeric composers (eight/seventh c.BC), separated by almost a thousand years, and a direct line need not have existed between the Akkadian and the Greek traditions.Yet several features of the earliest OB evidence endured into the first millennium.Akkadian epic of the second and first millennia BC features pre-eminently in the continuum of literary forms connected to performance that spanned the major literary cultures of the Eastern Mediterranean and Near East.It is the most longstanding and best preserved.Its products, including Gilgamesh, radiated westward from the second millennium BC, and it is with Akkadian poetry that most Near Eastern literary parallels to early Greek epic have been drawn (West 1997a;Bachvarova 2016;Ballesteros 2021a).As outlined below ( §2), compelling contextual reasons commend the OB tradition as an excellent place to start this investigation.
While comparative work may thus help us substantiate the oral transmission hypothesis, it will also enrich analogical understanding of the literary cultures under scrutiny.Given the limited scope of the examined Akkadian corpus and its temporal and spatial distance from the Greek, however, this article will set the question of transmission aside, concentrating instead on establishing comparative parameters for future research.But it is well to keep in mind the broader context.
These are strong grounds to investigate traditional markers of orality in OB Akkadian texts, whose compositional structures demand to be read bearing their performative dimension in mind.Comparison with Homer proves instructive.

A comparative approach to traditional naming
How should one conduct that comparison?Formulae are an obligatory starting point.The 'Parryan' noun-epithet formula (e.g., 'and in reply to him spoke swift-footed Achilles') is an expression conveying 'a given essential idea' (Achilles in the nominative); crucially, Homer would use one and only one expression when willing to communicate that idea under a given 'metrical condition' (Parry 1971(Parry (1928)): 13).Exceptions exist, but the 'essential ideas' covered by 'formulaic expressions' are so numerous, and the system shows such pervasiveness ('extension') and such a strict connection with metre ('economy'/'thrift'), that it cannot conceivably have been invented by Homer himself.Noun-epithet formulae, therefore, must be traditional expressions.Indeed, Parry (1971Parry ( (1928)): 24-36) showed that Apollonius and Vergil do not apply the system, thus not qualifying as 'traditional' poets.
That 'traditionality' here implies the illiterate oral composition of the Homeric epics, a conclusion at which Parry and Lord arrived through the South-Slavic analogy, has not been demonstrated (Lord 1960). 16Three aspects are central for present comparative purposes.First, the systematic recurrence of a given repeated expression in discernibly identical metrical contexts is a sign of traditionality (though not necessarily of illiterate composition).Second, comparison with living oral traditions supports the idea that the traditional character of certain poetic features should be associated with oral composition, whether preceding or contemporary with our written sources.Foley (1990Foley ( , 1991) ) influentially coined the adjective 'oral-derived' to describe this phenomenon, though one must remember that oral phases of composition do not necessarily precede a given written version, and for this reason among others, a different label-'performance-directed'-seems preferable (see also below §5).Third, the attempts at extending Parry's approach and conclusions to cases beyond noun-epithet expressions illustrate the fundamental role of variation, compositional freedom, and idiosyncratic usage of traditionality.They can offer insights into the poets' artful deployment of their inherited formulaic stock.
Noun-epithet phrases that strike Homeric readers as stylistically familiar are frequent in Akkadian narrative poetry, as they are in several other traditions (Hecker 1974: 162-8).In Greek epic, the shaping of traditional expressions for stereotypical character-naming is inextricably intertwined with the nature of the hexameter, a very flexible but appreciably regular verse.Akkadian metre shows no such regularity.Diachronic variation is considerable, but Akkadian verse appears to be organized according to accentual peaks (of varying number), often displaying the parallelismus membrorum based on alliterative patterns that is typical of other Semitic poetry (such as Ugaritic, Hebrew, Arabic). 17Poetic lines do not present a fixed number of syllables, nor any regular quantitative alternation.Thus, however tied to metrical factors (below §4.2), Akkadian noun-epithet formulae do not play the same compositional role as they do in Greek epic.For instance, there is no comparable Parryan 'extension' of the formulaic system, no comparable range of 'ideas' expressed through metrically-determined noun-epithet formulae.Nevertheless, early Greek and Akkadian noun-epithet expressions functioned as performance-directed elements in similar manner.Both were regarded as traditional by composers and audiences, and played a comparable aesthetic function in evoking a venerably ancient poetic universe. 18  Given the differences in metre and in the related traditional texture of the two traditions, this comparison is not centred on noun-epithet formulae, but more broadly on the names of traditional characters, with a view to classifying the types of proper noun expressions ('mentions') based on degrees of traditionality.I propose a spectrum of name-mention traditionality comprising three categories: T = traditional: most likely formulaic/traditional (viz.inherited from the tradition, stylistically fixed, comparatively 'frozen') TB = tradition-bound: based on recognisable traditional patterns ('flexible'), but less likely to have been inherited, as far as we can tell from available data NF = non-formulaic: no discernible widespread (viz.presumably traditional) structure behind the mention (save the metre) This is a spectrum enabling a comparative discourse which respects the traditional rules of each corpus. 19As Homerists routinely remark, our surviving texts account for a fraction of the traditional poetry that was once current in writing and orally.Certainty is precluded as Wisnom (2015) compares Old English alliterative verse.On Akkadian metre, see also Helle (2014), building on von Soden (1981, 1984); Hecker (1974: 101-54) and West (1997b); also Buccellati (1990) and Wasserman (2003: 158-62).E.g., Parry (1971Parry ( (1928)): 127): archaicizing formulae demonstrate 'nobility and grandeur' On formulae and other structures as partes pro toto evoking the tradition, see Foley (1991: 1-59).On traditionality and group solidarity, see Ready (2018: 114-21).Ready (2018) classifies Homeric similes within a traditionality spectrum; Beck (2018) considers Odysseus' nameepithet formulae in a semantic spectrum of contextual appropriateness.
to whether an expression repeated twice is traditional or not (Hainsworth 1993: 16-17;Di Benedetto 1998;Currie 2016).Equally important, poets display a vested aesthetic interest in making an innovation sound traditional, which evidently contributes to further blurring the boundaries between our categories.20Homeric poets or audiences may or may not have conceived of a certain formulation as traditional, but there is an extent to which one can expect to be able to tell if it was indeed so.Yet it is imperative to search carefully for the indigenous parameters governing the traditional expression of poetry.

Naming the gods
This case-study, then, concerns the mentions of Enlil in Atra-hasis and of Zeus in Iliad Book 8.About 60% survives of the OB poem, which totalled approximately 1250 lines (Foster 2005: 228).Since the Atra-hasis lines are roughly between one-half and two-thirds the length of a hexameter, this is an acceptable match for the 565 hexameters constituting Homer's Book.In the two texts, the two chief gods can justifiably be said to be protagonists: both are the most frequently named, and both drive the narrative through their initiative.Zeus is mentioned 34 times, and I count 38 mentions of Enlil (I exclude cases where Zeus is only mentioned patronymically (Kroniōn, Kronidēs, 4x T, 2x NF), and Atr.OB 1.131, 133 ≈ 143, 145 and 2.vi.22 because these lines are too damaged, but include the repeated lines Atr. ).The percentage result of the comparison is as follows: Tab. 1: Mentions of Zeus in Iliad 8 and Enlil in Atra-hasis.Before drawing conclusions on these percentages, which seemingly indicate analogous tendencies (prevalence of T), control cases will require examination ( §4.3).First, however, I will discuss the components of each category.

Zeus
Let us start with Zeus. 21The vast majority of the 23 instances classified as 'traditional' (T) are noun-epithet expressions (17): Tab. 2: Noun-epithet expressions (Zeus in Il. 8).Homeric Hymns.Data are taken from TLG searches.Hexameters consist of six metra or feet, which can take a dactylic or spondaic form, except the last, which can be spondaic or trochaic (long + short) but not dactylic.Dactyls are feet where two short syllables follow a long one (-◡◡); spondees are sequences of two long syllables (--).Introduction to the Homeric hexameter: Nünlist (2015), who follows, as I do, the system of Fränkel (1960); trithemimeral caesura = Fränkel's A4, masculine = B1, feminine = B2, hephthemimeral = C1, bucolic diaeresis = C2.Caesurae occur when a semantic pause caused by word-end does not coincide with the end of the foot; diaereseis when it does.Adherence to traditional metrical positions is thus considerably strict.The high number of realizations of the essential idea 'Zeus' shows the importance of possessing an ample search corpus: none of the expressions is repeated more than twice in Book 8, and it would be impossible to distinguish traditionality without the rest of epos.In fact, 8.22 and 8.250 highlight that indeterminacy.The repetition of Ζῆν' ὕπατον μήστωρ' 'the highest counsellor Zeus' at Hom. Il. 17.339 is unique in epos; we cannot exclude that the poet repeated an expression of his own devising, consciously or not.What seems certain is that, even if the phrase was consciously repeated (which should not be granted by default), the poet wanted the expression to sound traditional, since the two contexts cannot be meaningfully connected as evoking an intratextual allusion.An equally venerable and traditional connotation was doubtless attributed to πανομφαίῳ Ζηνί 'to Zeus of all the oracles', evoking Zeus' role as the ultimate prophetic source.Considering the epic corpus alone, its status as a traditional expression in Parry's sense would be questionable: its only other occurrence comes in the genitive case ([Hes.]fr.150.12 M.-W.).But an identical occurrence in an epigram attributed to Simonides (fifth century BC) may strengthen the possibility (AP 6.52 = 61 FGE = Ep.55.1-2 Sider: Οὕτω τοι, μελία ταναά, ποτὶ κίονα μακρὸν / ἧσο, Πανομφαίῳ Ζηνὶ μένουσ' ἱερά• 'Now thus, slender spear, against the great column / rest, and be sacred to Zeus of all the oracles').Again, an intertext seems unlikely. 22Finally, the 'father Zeus' type (here 8.236, 397, 438, 460) may appear to sit uncomfortably among noun-epithet expressions.The monosyllabic nature of the noun (in the nominative and vocative) and the Latin and Sanskrit cognates Juppiter and Dyauṣpitṛ encourage one to take it, if not as a single word (cf. the convenient division at Ζεὺς δὲ πατήρ 'but/and father Zeus'), at least as behaving like a 'metrical word', in which case it should be treated as NF 'isolated' names (see below). 23owever, the fact that the dactylic vocative Ζεῦ πάτερ, out of several possibilities, appears only at the beginning of the verse, is a clear indication of traditionality (contrast, e.g., the dative Διὶ πατρί: 30x epos, 16x verse-end, 8x before feminine caesura, 6x after hephthemimeral).Taken as a whole, these examples evidence the vastness of the inherited formulaic repertoire, as well as the flexibility allowed by the hexameter-certainly not a surprising result for Homerists, but one which is important for the comparative purpose of defining the traditional ways of naming Zeus.

Lines
One way to deploy and evoke tradition is, thus, to use noun-epithet expressions.The second, which accounts for about a quarter of the T instances (6/23), consists in a phrase repeatedly found in the same metrical position and coupling the name Zeus with verbal or nominal expressions generally themselves recurrent in the corpus (Tab 3):

Enlil
Given the nature of Homeric poetry, epithets proved reliable witnesses to identify traditional (T) mention-types for Zeus.Noun-epithet expressions are also prominent with Enlil, constituting almost half of the identified T instances.The formula qurādu Enlil 'the warrior Enlil' (or, in the genitive, qurādi Enlil) recurs 10x in OB Atra-hasis, and it always concludes the line (1.8 [≈ 125 = 137], 69, 92, 112, 2.v.27, vi.32, 3.vi.5, vi.12).The difference with the Homeric practice is striking: none of the Zeus noun-epithet expressions recurs as many times in the equivalent space of Book 8; and Atra-hasis does not present a variety of epithets comparable to that in Homer.A Parryan explanation for this would also be the simplest: it comes down to the complexity of the hexameter, which obliged generations of poets to devise many 25 Il.8.420-4 athetized by Aristarchus, 8.421-4 by West.But see Kelly (2007: 398-99).ways (i.e.traditional expressions) to accommodate the rhythm and facilitate oral versification.Akkadian poets simply did not need to do so.Why is it, then, that qurādu/-i Enlil always comes at the end of the line?The answer probably is that this epithet was traditionally bound to larger-scale traditional forms, in which the noun-epithet came at the end.This is certainly the case in eight out of the ten OB Atr.occurrences.Three times it comes in the formulaic verse mālikšunu/-kunu qurādu Enlil 'their/your (pl.) counsellor, the warrior Enlil' (cf.3.viii 11 atta mālik ilī rabûti 'you, counsellor of the great gods' and OB Anzû 1.2 = SB 1.85); and five times it occurs when Enlil is the addressee in the speech-introduction couplet of the type :

Anu pâšu īpušamma issaqar ana qurādi Enlil
Anu made his mouth ready, he spoke to the warrior Enlil Study of these couplets in Atra-hasis discloses one traditional rule dictated by the metre, namely that the speaker does not receive an epithet, whereas the addressee always does.As one anonymous reviewer perceptively suggests, 'presumably this rule follows from the balanced structure of the OB verse: the speaker's name is followed by several verbs and nouns that convey the act of speaking, leaving no room for an epithet, whereas the addressee is introduced only by a brief preposition, which allows the poet to supplement the name with an epithet and thereby to balance the couplet.' The rule permits identifying a second type of T naming of Enlil (5x OB Atr.), when he is the speaker in speech-introduction couplets and thus comes without an epithet (OB Atr. 1.85,105,2.v.22,2.vi/vii (?) [= Schø. ii.16'],3.vi.41). 26Here, the name is repeatedly connected to a traditional turn of phrase and qualifies as T despite standing without an epithet (compare Tab. 3 above).One similar case is Enlil illakā dimāšu 'Enlil, his tears were flowing down' at 1.167, where the phrase illakā dimāšu/-ša is stereotypical in Akkadian poetry (Hecker 1974: 78-9;George 2003: 836;Jiménez 2017: 94-7;Ballesteros 2021a: 19-20;add Schø. Atr. iv.10').Tellingly, it came to interact with an equivalent expression in the Hurrian tradition that we now find in the Hittite version of Gilgamesh and in Hurro-Hittite poetry: 'tears flowing forth like canals', a formulaic simile not found in the Akkadian Vorlage (Klinger 2005: 119-20).This is important evidence exemplifying the kind of transfer or adaptation that is possible when two poetic idioms come into contact that share an underlying traditional poetic syntax (cf.Ballesteros 2021a; forthcoming; above §1).
The one exception to the 'epithet-less speaker' rule is the fifth occurrence of qurādu Enlil in 26 On speech-introductions in Akkadian poetry, see Sonnek (1940); Hecker (1974): 174-80; Vogelzang (1990).speech-introductions.Here the narrative pace is unusually compressed and does not allow for the full standard couplet of speech-introduction (2.v.24-7):27 'Let them br[ing to me] the two [comrades] Let them [come] before me.' 25 The two comrades were brought [to him], And the warrior Enlil spoke to them A closer look shows that the breaking of the rule is drawn by another traditional tendency.To understand better the poetic technique behind these lines, we need to look at the third T way of naming Enlil, namely quatrains.A traditional quatrain can be defined as one composed of two successive couplets (distichs) that are identical except for any grammatical adjustments and one word in the first distich which the second distich varies in the same verse position (Hecker 1974: 146-51, naming this technique 'd[ie] variiert[e] Wiederholung eines Verspaares').The reader of Sumero-Akkadian poetry, even in translation, will immediately recognise this elegant feature, which has long been connected to performance (West 1997a: 593-4).One example from the divine strike scene may suffice (OB Atr.1.70-3, transl.Lambert and Millard):

mišil maṣṣarti mūšum ibašši 70 bītu lawi ilu ul īdi mišil maṣṣarti mūšum ibašši bītu lawi Enlil ul īdi
It was night, half-way through the watch, 70 the temple was surrounded, but the god did not know It was night, half-way through the watch, the temple was surrounded, but Enlil did not know!These structures account for six T mentions of Enlil in OB Atr.(1.44 = 59, 1.73, 1.83, 1.90, 1.96).Now, OB Atr.2.v.24-7 (above) is not precisely a traditional quatrain, given the considerable variation in the second line of the distichs.But a parallel at 1.87-90 suggests that the same technique is in play.That quatrain likewise spans a direct order and its implementation, and displays the correspondence between maḫrīya ('my presence'), which Enlil says of himself, and the narrator's naming of the god.It reports Enlil's first reaction to the news of the revolt (OB Atr.1.87-90): Nuska, edil bābka, kakkīka liqi, izziz maḫrīya!' Nuska idil bābšu kakkīšu īlqi, ittaziz maḫar Enlil 90 Nuska, bar your gate, take your weapons and stand before me!' Nuska barred his gate, took his weapons and stood before Enlil.90 OB Atr.2.v.24-7, then, appears to conflate the techniques relating to the traditional quatrain with those of the speech-introduction, with an unusually compressed version of the latter in place of the expected quatrain closing.As seen above, standard speechintroductions place the verb issaqar '(s)he spoke' in the second line of the couplet, followed by ana ('to') plus noun-epithet expression (e.g., issaqar ana qurādi Enlil).In OB Atr.2.v.24-7, the traditional speech-introduction legitimized the poet to preserve the noun-epithet expression after issaqar.Meanwhile, the compression of the speech-introduction (drawn by the 'quatrain' urge to be explicit about Enlil's name in the second distich) made the composer opt for an unusual nominative, resulting in the breaking of the epithetless addresser rule.
To sum up: the 22 traditional (T) mentions of Enlil in OB Atr.(out of 38) include 10x qurādu Enlil formulae, 5x Enlil as speaker in speech-introduction structures, 1x Enlil coupled with a conventional expression (illakā dimāšu), and 6x in traditional quatrain structures.
As in our Greek sample, tradition-bound (TB) expressions are the least numerous of the three categories (I count seven of them).Their identification invites comparison with the uncertainties surrounding the Greek evidence, highlighting the spectral nature of both systems.As observed above, brief (prepositional) phrases of the kind 'from Zeus' normally offer no hint at traditionality in the sense of a recurring structure.But cases like 'Father Zeus' (Ζεῦ πάτερ), 'but father Zeus' (Ζεὺς δὲ πατήρ), or 'but Zeus me (acc.)'(αὐτὰρ ἐμὲ Ζεύς), whose traditionality is beyond question given the recurring metrical position (including outside the Iliad), alert us to the possibility that a phrase like 'against Zeus' (Διὸς ἄντα) may well have been traditional too, even if it recurs only twice and in the same passage.It was certainly meant to sound traditional.The comparable case in Akkadian is the phrase itti 'with' + proper noun.In the case of 'against Zeus' (Διὸς ἄντα) the 2x short-range repetition is insufficient for us to classify it as traditional.Two examples from OB Gilgamesh are also noteworthy.They come at a short distance, from Gilgamesh's famous heroic exhortation to Enkidu, where he contrasts a glorious fight with the certainty of death and the impossibility of living like the gods (OB Gilg.As with 'against Zeus' (Διὸς ἄντα), there can be no demonstration that itti Enlil in Atrahasis can be taken as an inherited feature.Yet the metrical argument is stronger in the Akkadian case, as is the (related) force and memorability of the phrase, suggested by the quoted occurrences.Such elements seem sufficient to classify itti Enlil as TB, though this case again highlights that traditionality works as a spectrum rather than by fixed boundaries.No hard-and-fast rules exist to exclude that itti + proper noun should just be treated as non-formulaic (NF).Conversely, accepting it as a poetically-marked traditional expression might bring itti Enlil on a par with T cases where the noun is connected to self-standing structures such as speech introductions or quatrains.
A different type of TB expression involves variations upon traditional quatrain patterns.
The poet utilizes this technique to express Enlil's displeasure at mankind's din which induces him to devise destruction.Lines are repeated three times (one for every scourge, except the final Flood), arguably to mark out sections in performance, and contributing to the crescendo effect , restored based on the parallels OB 1.352-60 = 2.i.1-9; cf.NA S rev.iv.1-12, x rev.1-11, and SB Si 5.45-54;George and Al-Rawi 1996): The [land] was bellowing like a [bull] by [their noise] the god was disturbed 355 [Enlil heard] their outburst, [he spoke] to the great gods: The variation is elaborate, and again involves an overlap with speech-introduction structures (cf.above on 2.v.26-7).Since the subject is just called 'the god' (ilu) in the couplet 1.354-5, audiences would have expected a traditional quatrain, with the ensuing distich substituting Enlil for ilu.Such a quatrain may well have been performed.What we do have is an unorthodox-and, one may feel, highly effective-speech-introduction couplet.Technically, the position of the noun at the beginning of line 1.356 is attracted there by its function as addresser (see above).Yet this also works as the second half of a traditional quatrain in so far as (a) it resumes the content of the preceding distich and (b) satisfies the audience's expectation that the name be spelled out.Since it is repeated three times, one cannot exclude this is in fact a T formulation.But it seems best interpreted as a poet's creative elaboration specifically designed to introduce Enlil's reaction to noise in the Flood poem.
One can interpret the couplet where Atra-hasis addresses the city-elders similarly.Here the variation upon the expected repetition works within the compass of the couplet (as opposed to the quatrain) and takes a chiastic form (OB Atr.3.i.42-3): My god is [in a strife] with your (pl.) god, [Enlil] and Ea are furious with one another.
One may note the itti + noun foot at the beginning; there is little ground to suspect that Ea u Enlil is a standardized expression, though it recurs at 2.vi.22 (otherwise illegible), which is unlikely to have contained the same text as 3.i.43since it precedes Enlil's speech.
Homer is keen on variations marking decisive moments too.One TB mention of Zeus is worth quoting (Il.8.1-2): Ἠὼς μὲν κροκόπεπλος ἐκίδνατο πᾶσαν ἐπ᾽ αἶαν, Ζεὺς δὲ θεῶν ἀγορὴν ποιήσατο τερπικέραυνος Dawn the saffron-robed was spreading over the face of all the earth and Zeus who delights in thunder summoned the gods' assembly (Ζεὺς τερπικέραυνος 4x Hom., verse-end, cf.above) The striking-and unparalleled-hyperbaton (separation of two normally adjacent words) emphasizes the epithet.As a new day begins, the line foreshadows that Zeus will at last take steps towards a Greek defeat as he had promised, and will actively help the Trojan leader Hector through an exceptional display of his thundering abilities .29 This can profitably be compared to how Enlil is named in the chief rebel's rallying speech to the assembly of the insurgent gods (Atr.OB I 43-6 ≈ 57-60; cf.SB Si I 49-51; George and Al-Rawi 1996: 158): [ilam] mālik ilī qurādam alkānim i nišši'a ina šubtīšu Enlil mālik ilī qurādam alkānim i nišši'a ina šubtīšu [The god], the counsellor of the gods, the warrior, let us go and remove him from his dwelling!Enlil, the counsellor of the gods, the warrior, let us go and remove him from his dwelling!Here the resounding fullness of Enlil's titles, highlighted by the disjunction, stresses the daring of the rebels menacing the chief god.The effect is technically predicated on the use of the traditional quatrain, which appears to determine the hyperbaton.Like in the Iliad, this might have a foreshadowing resonance-though an ironic one, since the Enlil who shuts himself in the palace, bursts into tears, and forsakes the divine assembly hardly acts as 'the warrior' (for similar irony surrounding this epithet, see OB Anzû 1.2 = SB 1.85).The foreshadowing disjunction of a noun-epithet expression makes this an excellent parallel to Il. 8.2.And although Enlil mālik ilī qurādam, on account of the traditional quatrain, should be classified as T (rather than TB as Ζεύς…τερπικέραυνος or Enlil ištēme rigimšin), all these examples illustrate a cross-cultural analogy in the effects achieved by playfully varying upon inherited patterns.Does the analogy extend to the quantitative level viz. the relative percentages of use of traditional, tradition-bound, or non-formulaic ways of naming characters?NF occurrences of the name Enlil in OB Atra-hasis are nine (1.14, 84, 104, 168, 196, 2.viii.35, 3.i.48, 3.iii.39, v.41).This makes the percentage difference with NF occurrences of Zeus in Iliad Book 8 relatively negligible.
It is now worth returning to those percentages: Tab. 6: Mentions of Zeus in Iliad 8 and Enlil in Atra-hasis (again).The data are considerably more useful for comparing the ways in which each poetry is traditional than the degrees to which each one is so (and much less for comparing degrees of orality).Traditionality in Homer is not the same as in OB Akkadian epic.The noun-epithet formularity as tied to the hexameter system is probably the most evident differential trait here.Disparity in T percentages is not insignificant, but more important is the fact that, in Homer, T mentions are made up of 74% noun-epithet structures (17/23), but just 43% (10/23) in Atra-hasis.In the context of noun-epithet expressions, as noted above, variability is incomparably more pronounced in Homer, since there is only one noun-epithet structure for Enlil.Again, this is predicated on the difference between metrical systems and composition as a function of them.On the other hand, the larger proportion of T cases that are not noun-epithet structures makes the Akkadian technique a more varied one in the ways it expresses traditionality generally.

Zeus in
Another major divergence lies in the degree of verbatim repetition, which is much more pronounced in Akkadian poetry, and is also a major factor affecting percentages.(If one, for instance, should not accept that 1.356 = 2.i.5 = [2.ii.41] counts as TB but consider it as NF, the latter would shift from 23.5% to 31.5%).Homeric messenger speeches are routinely repeated, often with crucial variations (Kelly 2007: 325-9;now Battezzato 2019;Ready 2019: 34-51, 75-97).But the range of repeated scenes in Akkadian poetry is incomparably larger-most famously perhaps the fourfold repetition of Tiāmtu's illegitimate enthroning of Qingu in Enūma eliš, occupying much of tablets II and III (Enūma eliš 2.11-48 (with 2.15-48 = 1.129-62) = 3.15-52 = 3.73-110).This phenomenon increases with time: later versions of OB poems augment or introduce repetition (Cooper 1977;Vogelzang 1986;1988: 192-234 on Anzû; Wisnom 2023).Repetition is helpful for memorization, but this difference in scale seems best attributed (however speculatively) to aesthetic preferences, perhaps dictated by performance modes: recitation for Homer, higher musical variation and song in Atrahasis permitting audiences to enjoy equal or diverging musical arrangements of the same line blocks (on song in Atr.see above §2; below §5 on Agushaya).
Nevertheless, fundamental commonalities-shared traditionality modes-remain and should not be downplayed.The existence of common features such as noun-epithet expressions, formulaic speech introductions, or typical situations such as assemblies has long been acknowledged (albeit little studied).But the successful application of the same taxonomy has shown that poets also had analogous ways of playing with traditionality.Crucially, both Homeric and OB Akkadian poets were keen for poetry to sound traditional.
The next question is how representative our case study is.This is key not only from a quantitative perspective, but also because narrative poetry is not a mechanic process, and how characters are mentioned must be determined to at least some degree by their attributes and actions, and generally by poets' choices about them.Looking at other characters, and, for OB Akkadian poetry, other poems, shows that the proposed taxonomy holds, and that the traditional modalities identified in Atra-hasis were indeed so.They also strengthen the impression of commensurability between the two corpora (for a synopsis see Tab. 8 below §4.3).

Ea, Diomedes, Hector, Ishtar, and Ea again
Nevertheless, the naming of Ea in Atra-hasis (after Enlil, the most frequently mentioned character) may seem to strengthen the comparative divide. 30Out of 25 counted mentions, only 14 can be taken as T (56%, compared to Zeus' 67,5%).Again, one finds noun-epithet expressions, speech-introductions, and traditional quatrain structures..98, 1.100, 1.102, 1.254, 2.v.18, 3.i.45.TB cases (16%, 4/25) include the itti Ea occurrence at 1.201 and the quatrain variation at 3.i.43(both mentioned above §4.2);an association of Ea with the root 'pš 'to make' with which he is traditionally linked (3.vi.14 ša lā Ea īppuš 'who but Ea could have done this?'); 31 and a variation upon the quatrain pattern whereby the god is mentioned in the first distich rather than in the second as is usual (1.364-7). 32 Ea's case-study consists of just 25 cases, and the difference in percentages compared to those for Enlil and Zeus should not be overemphasized.Nevertheless, the higher proportion of NF seems to justify the impression that Akkadian naming patterns may be less de- pendent on tradition than the Greek.That impression is strengthened by looking at the occurrences of Diomedes' name in Iliad Book 5, where he is the protagonist.Book 5 is the longest of the poem, and being largely about battle scenes it is among the most traditional thematically (see Fenik 1968: 9-78;Krischer 1971: 23-7;and Andersen 1978: 47-94 on the poet's manipulation of conventions).In Book 5 Diomedes' name occurs 25 times, the vast majority of which are T instances (84%, 21/25), with both TB and NF scoring a meagre 8% (2/25).Each of the 21 T mentions is a noun-epithet formula, 33 and both TB instances resemble one considerably. 34These data are strongly suggestive of a high degree of traditionality in Homer, one that appears distant from those concerning Ea in Atra-hasis.
Data for all the examined case-studies are below: before feminine caes.11x epos.
Tab. 8: General mention distribution.The examined control cases confirm the fundamental tendencies outlined by the initial Enlil/Zeus comparison.Greek epic tends to use more T expression.The difference is not immense, however, and can be explained by the compositional importance of noun-epithet formulae in the hexameter context.Variations, at any rate, exist in both traditions, as the extremes Diomedes/Hector and Ea (in Atr.)/Ishtar (Agushaya) exemplify.The broader picture also confirms the value of the proposed taxonomy in illuminating the peculiarities of each corpus such as, again, the Greek noun-epithet formula, or the Akkadian traditional quatrain.

Zeus in
This picture also has limitations.To distinguish between types of mentions, one relies on statistics and on the extent of corpora which are limited by definition and cannot be trusted to encompass the entirety of the tradition.Our percentages can therefore be partially flawed due to borderline cases (as repeatedly discussed) and should not be taken rigidly.
A further limitation is that much Akkadian poetry is missing from this research.Preliminary observation of the OB Gilgamesh, for example, shows a lower reliance on noun-epithet structures, though traditional speech-introductions and quatrains are as prominent as one could expect.Research should also be extended to later evidence: especially intriguing are changes between OB and first-millennium versions of the same poem.This will sharpen understanding of the traditionality of epic diction, including its diachronic and synchronic variations-an aspect that could also be important for Hellenists, since comparisons could be profitably made between Homeric epic and other archaic hexameter poetry.
Yet the work done indicates fundamental tendencies and an overarching homology.Per-haps most importantly, the proposed taxonomy shows that in both contexts traditionality worked as a spectrum, with the fundamental aesthetic implication that traditional items were immensely valued by composers and audiences.

Concluding Remarks
The Agushaya Song was composed under the patronage of king Hammurabi of Babylon (reign.1792-1750), as evidenced by the following passage (Agushaya B v 23-9): 40 u šarrum ša annâm zamāra [m] idāt qurdīki tanittāki išmūni 25 Ḫammurapi <ša> annâm zamā [ram] ina palîšu tanit<tā>ki ⌈x x⌉ innepšū lū šutlumšu addār balāṭ [u] And the king who heard from me (viz.Ea/the performer) this song, the sign of your (viz.Ištar's) heroism, and your praise, 25 Ḫammurabi, <who> (heard) this song your praise, it is during his reign [...] it was composed, may he be granted life forever!This composition can be dated precisely.But what matters here is that a song presented to Hammurabi and meant to be used at a civic festival , arguably in Babylon, was surely not an 'oral composition' of the improvised and illiterate type, however unclear the institutional relationship between musicians, composers and scribes may be.41  Moreover, Agushaya yields positive evidence for its being a performance text-beyond, that is, internal references such as that just quoted, and those occurring, in a much more circumscribed manner, in Atra-hasis (Finkel 2014: 298-308 deems the 'Ark tablet', possibly belonging to Atr., a performer's aide-mémoire).The Agushaya manuscripts are accompanied by scribal notations in Sumerian (rubrics) detailing the composition's strophic structure as a function of performance.Each rubric of the type ki-ru-gú 1/2-kam-ma 'this is the 40 Text from SEAL (but insert the <-ta> at 27).For the difficulties of these lines, see Groneberg (1981): 127-33;CAD T 170-1;Foster (1991): 24; Shehata (2010): 202-3.41 Cf. above n. 12 On literacy in OB Mesopotamia, see Wilcke (2000); also Charpin (2010: 7-24); Van De Mieroop (2022: 1-35).
first (/second etc.) song' is matched by one ĝiš-gi4-ĝál-bi, 'its antiphon' (for kirugu 'strophes' and ĝišgiĝal 'antiphons', see Shehata 2009: 344-9).Finding stylistic features entirely comparable to those in Atra-hasis, and the same forms of traditionality, suggests that those forms and features were integral to the texts' performative dimension.This is so despite stylistic differences, for example in the shape of speech-introductory couplets and strophic structure (there is less uniformity here than in the Greek hexameter corpus).It seems unlikely, therefore, that we are confronting conventional 'relics of orality' in Hecker's sense (see above §2).
This evidence illustrates a point repeatedly made in this volume: evolutionary scenarios leading from oral, illiterate, improvised to written and meditated composition can often be inadequate, and sometimes misleading.They are of limited utility in assessing the type of poetry addressed in this article, either for the mechanisms of composition or for the aesthetic effects poets and audiences were after.Stylistic analysis alone is unlikely to determine degrees of orality or literacy, even in the case of highly traditional poetry like the Homeric.But comparison with the Babylonian practice, which this article has for the first time attempted, can be productive in several ways.
Starting from Homeric orality, the proposed model permits a commensurate accounting of traditionality which respects cultural specificities.It may also highlight the importance of reading Babylonian poetry by doing justice to its traditional diction and performative dimensions.For scholars (such as Homerists) involved in attempts to connect traditionality and orality, the Babylonian comparandum is likely to be sobering.Fully developed literacy in the urban Old Babylonian context is not a deterrent for high degrees of traditionality meant for performance.In fact, this comparison may prompt reconsideration of Foley's category of 'oral-derived' poetry, as well as of Lord's (not unrelated) concept of the 'transitional text' (especially Lord 1986;1995: 212-37).These have proved excellent models for reading Homer and other narrative poetry conceived for performance.However, a connection between the texts' traditional features and their historical status as documents for the passage between orality and literacy does not necessarily reflect reality.In fact, these models convey teleological views (from the oral to the written) which the Babylonian case among many others, as this volume attests, simply disproves.Rather than of 'oral-derived', we should now more accurately-and more productively-speak of 'performance-directed' texts, at least in the Old Babylonian case.This category needs to be explored through further study, but it promises to have historically diagnostic advantages.At the same time, it incorporates the major gain of Foley's approach: it is useful for investigations of comparative aesthetics, and for understanding interactions between composers and audiences.
The proposed classification reveals fundamental analogies in modes of composition and poetic conception between early Greek and Babylonian poetry.Seemingly, the most im-portant is the nexus between a performance context and the force of traditionality.For ancient poets and audiences, a newly crafted phrase had to fit pre-existing parameters that qualified the expression as worthy of the epic medium.This had the effect of concealing the technical operation behind it.But we can assume that expert poets and audiences would look beneath the surface, enjoying traditionality even more as a result.Modern readers can presume to approximate that competence, and comparison has once again proved as good an avenue as any for doing so.
Yasumura 2011;Pucci 2018;al mentions (Zeus in Il. 8).strictly tied to certain metrical positions.Like Parry's noun-epithet formulae, they are unlikely to have been coined by the author of the Iliad.Further down the spectrum of traditionality, one encounters expressions identifiable as tradition-bound (TB).Their precise shape is unparalleled in the corpus, but composed of certain features that the audience would recognise as epic and traditional.24Thesearegenerallytraditionalepithetsorotherfrozenexpressions in specifically reshuffled combinations.I count four of them for Zeus in Iliad Book 8:Tab.4:Tradition-boundmentions(Zeus in Il. 8).The evidence is insufficient to decide whether a given TB expression was in fact traditional or just made to look so.One cannot exclude that the powerful hyperbaton at Il. 8.2 (see below) was well known among groups of singers and considered common property.This is perhaps more likely in the case of Ζεύς τ' αἰγίοχος καὶ Ἀθήνη (8.287), given the Odyssean parallel.The case of Διὸς δόμον αἰγιόχοιο (8.375) likewise displays a degree of indeterminacy, since other Διὸς ◡◡ αἰγιόχοιο expressions are widespread in epos (notably Διὸς νόον αἰγιόχοιο, cf. on 8.143).Nevertheless, the evidence does permit grouping these instances together, at least notionally, because they are unique in a corpus vast enough to allow for other types of expressions to recur up to 35 times.The case of 8.412 brings us closer to our final category, the nonformulaic instances (NF) displaying no hints at a traditional status.What differentiates Διὸς δέ σφ' ἔννεπε μῦθον from NF cases is the relevance of the metrical position: Διὸς δ(έ) recurs only after the feminine caesura, though its metrical shape would allow for several other possibilities.This suggests that its recurrence was intimately tied to dactylic halflines (hemiepes) of the type most famously represented by Il. 1.5 (also Od. 11.297, Cyp.fr.particularly tied to any of them, though there is a prevalence after bucolic diaeresis (seeKelly 2007: 168-9for the referential resonance of this expression).Nevertheless, it bears repeating that this is a 'spectrum' of traditionality, and the available evidence is frequently not sufficient for strongly informed opinions.The matter is particularly acute with the 'metrical words' of the 'father Zeus' type which (unlike Ζεῦ πάτερ) recur sporadically, such as Ζεὺς δὲ πατήρ 'but Zeus the father' (T, see above).One should not exclude that an expression like Διὸς ἄντα 'against Zeus' (NF) was in fact common, especially considering the importance of Olympian conflict in the tradition (e.g.,Yasumura 2011;Pucci 2018; Ballesteros, forthcoming).What seems certain is that the poet, by emphasizing the expression through the short-distance repetition, made it sound traditional (Il.8.423-8): 25 ἀλλὰ σύ γ' αἰνοτάτη, κύον ἀδεὲς, εἰ ἐτεόν γε τολμήσεις Διὸς ἄντα πελώριον ἔγχος ἀεῖραι.Ἣ μὲν ἄρ' ὣς εἰποῦσ' ἀπέβη πόδας ὠκέα Ἶρις, 425 αὐτὰρ Ἀθηναίην Ἥρη πρὸς μῦθον ἔειπεν• ὢ πόποι αἰγιόχοιο Διὸς τέκος, οὐκέτ' ἔγωγε νῶϊ ἐῶ Διὸς ἄντα βροτῶν ἕνεκα πτολεμίζειν• But most dread are you, you shameless dog, if truly you will dare to raise your mighty spear against Zeus.'And having said so swift-footed Iris went away 425 but Hera then said word to Athena: 'O dear, daughter of Zeus who bears the aegis, no more shall I allow that we two should fight for mortals' sake against Zeus; I shall return to borderline cases, but it is first worth looking at Enlil's naming in OB Atrahasis.
Allan 2008ry' in the present or imperfect tense, poets would use ὀπάζω'grant' (rather than  δίδωμι 'give' as in 8.216, aorist).κῦδοςὀπάζει(/ν) is itself a frozen expression recurring 11x epos at line end (3x with Zeus as subject).At root, these are all clearly recognizable as frozen expressions 1.7, cf.Il.20.15):Διὸςδ'ἐτελείετοβουλή 'and Zeus' plan was being accomplished' (seeAllan 2008; Currie 2016: 1-4; Edmunds 2016).NF instances, then, display the noun alone or in prepositional phrases, with no discernible traditional viz.widespread structure behind the occurrence, save for the metrical constraint (as opposed to the consistent metrical positions defining T).I count seven of them for Zeus in Iliad Book 8:Tab.5:Non-formulaic mentions (Zeus in Il. 8).At first sight, there is nothing obviously differentiating ἀλλὰ Ζεῦ 'but, O Zeus' at 8.242 from Ζεῦ πάτερ 'father Zeus' (8.236 voc.) which was classified as T. The latter's recurrence 34 times in the same metrical position makes all the difference.Similarly, with ἐκ Διός 'from Zeus': it recurs 19 times in the corpus, but in several different positions, meaning it was not And the same goes for itti Enlil 'with Enlil', which recurs only 3x Atr.(1.152 = 1.165, 2.vii.47).But there is perhaps a case to be made that itti + proper noun is an underlying traditional naming manner against whose background itti Enlil resonates.One reason is the considerable brevity of the Akkadian line, meaning that itti Enlil occupies one of three feet of the quick OB verse.This is in sharp contrast with the prepositional phrases like Διὸς ἄντα, occupying less than a fifth of a hexameter: the Akkadian phrase receives far more emphasis.This impression is reinforced by looking at other examples of this structure in OB Akkadian poetry, such as the sentence-end case in the Literary letter to Ninmuga 5-7 (SEAL no.1649): Rather than a casual use of plain 'with someone' expressions, these examples speak for variations on a traditionally pointed-though not necessarily oral-template, which gained strength by the fact that it occupied one of the three (or two, in the case of the Ninmuga passage) feet of the line (also OB Gilg.2.135-8, with the colon marked through line-division:⸢itti⸣ [Ša]mkatim / ippuš [u]lṣam / iššīma īnīšu / ītamar awīlam '[With Shamkat] / he (Enkidu)was pleasuring himself / he lifted his eyes / he saw the man').Even more striking is the deployment of itti Enlil in OB Atr., for it contributes to a long-distance intratextual connection which encapsulates the rivalry between the chief god and the wisdom god Ea.The occurrence at 2.vii.47, in the heated divine assembly where Ea refuses to cause the Flood, Ea's words look back to his own pre-eminence when it came to creating mankind.The destructive obnoxiousness of Enlil is contrasted with the beneficent creative power of Ea (Atr. Tab. 7: Traditional mentions (Ea in OB Atr).
Homeric naming can present balanced percentages (like those for Ea); and Akkadian practice similarly unbalanced numbers (similar to those for Diomedes).Returning to Iliad Book 8, Hector's case is instructive.He is the human protagonist of the book.As among Diomedes' mentions, TB are hardly visible, with only two cases (7,5%) distinguishable from T on scant statistical grounds.35Moreimportantforcomparativepurposes is that traditional occurrences (T) are 16/26 (61,5%), but only seven of them are noun-epithet formulae (27%), in contrast with Diomedes' 21/25 (84%).36Moreover,Hector is named in NF manner 8/26 (31%): 8.90, 117, 148, 235, 301 = 310 (Ἕκτορος ἀντικρύ 'against Hector' before masculine caesura, only here, compare Διὸς ἄντα 'against Zeus' discussed above §4.1),341,348.This is a much higher percentage than with Zeus and Diomedes, but similar to those for Enlil and Ea.Further contextual study (and work on other Homeric books) is required to account precisely for such differences between the ways Diomedes and Hector are mentioned.They may well be caused by the specific actions in which Hector participates, rather than his being a relative newcomer in the oral epic tradition about Troy, as some scholars have suggested.37Whateverthereason, this example illustrates that the proportion of NF mentions can vary considerably in Greek epos too.Hector's mentions thus appear closer to observed Akkadian cases in the relatively high proportion of NF.Equally important, the evidence exists in Akkadian for a relatively high proportion of T. This can be seen by moving away from Atra-hasis and on to another classic of OB Akkadian poetry, the Song of Agushaya, a narrative hymn for the goddess Ishtar (ed.SEAL; Agushaya A+B refers to the two surviving tablets).Here Ishtar acts as the main character as much as Ea does; for both names I count 12 occurrences.