Research Showcases Literature

2.1. Gudea Cylinder A. Observations on the text’s micro-structure. by Lucrezia Menicatti, MA

2.2. Notes on the construction of meaning in an Old Babylonian bilingual proverb about exotic animals. by Dr. Frank Simons

2.3. Parallelism in Ugaritic Poetry. by Clemens Steinberger

2.4. When Repetition Fails – The Curious Case of the Hittite Šar Tamḫāri. by Henry J. A. Lewis

2.1. Gudea Cylinder A

Observations on the text’s micro-structure

Written by Lucrezia Menicatti (MA)

How to cite: Menicatti, L., 2020, “Gudea Cylinder A: Observations on the text´s micro-structure,” Project REPAC (ERC Grant no. 803060), 2019-2024, at (accessed day/month/year).

© Musée du Louvre/ Philippe Fuzeau

Gudea Cylinder A is a complex and refined literary work. As Suter 2000, 132-133, remarks, the motif of the divine fate frames the entire text in a huge ring-composition structure. In the following, we will offer an example of REPAC’s approach to Ancient Mesopotamian literary compositions by focusing on some of the devices shaping the text, with its contiguous and near-contiguous textual units, on the micro-level. These devices include structural formulas, textual and inter-textual parallelisms, and narrative repetitions.

By focusing on such phenomena, REPAC’s research seeks a deeper understanding of how meaning was created in Ancient Mesopotamia. Our aim in the present context is more modest; we simply seek to highlight the role played by some forms of repetition as structuring devices in a sophisticated written product such as Gudea Cylinder A.

Cylinder de Gudea
© 2006 RMN / Franck Raux

The Ring Composition Frame

The theme of divine fate encloses the entire Gudea text in a ring composition frame. The motif is introduced in the opening formula u4 /an-ki- a nam tar-[re]-/da\, “When fate was being decided in heaven and earth” (lit. “in the day in which fate was being cut in heaven and earth”).

This formula is borrowed from the epic genre and occurs in Sumerian narrative from as early as the Early Dynastic period (Black 1992, 93-95). This opening therefore constitutes an intertextual reference that places the hymn within a long tradition. Furthermore, the formula is meant to set the narrative action in a mythological past, thereby giving a mythological character to Gudea’s deeds.

The theme of divine fate and will occurs again at the closure of Cylinder B (CB 24, 9-17), when the Eninnu receives divine blessing, having been ‘joined together’ with heaven and earth (Suter 2000, 132-133).

Formulaic Repetitions and Repetitive Narrative Patterns

Formulaic Repetitions

Successive series of narrative sequences constitute the internal structure of Gudea’s Cylinder A, and fixed sets of expressions which we define structural formulas mark the transition from one of these narrative sequences to the subsequent one. These expressions consistently occur in the same narrative context – that is, at the end or at the beginning of a given narrative section. The initial part of Gudea’s text provides evidence for three such formulas.

One of these, sipa-zi gu3-de2-a gal mu-zu gal i3-ga-tum3-mu (“the right shepherd, Gudea, learnt much and put much into action”) occurs at four different places in the text. It marks the transition to a new action sequence, or a pause between two events (Averbeck 1987, 264-265). It is used after Nanshe’s revelation (CA 7:10), after Gudea has celebrated the rituals for Ningirsu (CA 12:20) and after the building of the Eninnu is finally completed (CA 25:22-23). The presence of the prefix /nga/ in the verbal form i3-ga-tum3-mu is a significant marker, since it is rare and was definitely not productive anymore by the Gudea’s time, but it only occurs in fixed expressions (Jagersma 2010, 513).

Two more formulaic repetitions occur in the initial part of Cylinder A in a similar context. One is the clause ga-na ga-na-ab-dug4 (“I really must tell it to her”), which introduces Gudea’s speech to the goddess Nanshe. The second consists of a longer clause, ama-ĝu10 ma-mu- ĝu10 ga-na-de6 (“I shall bring my dream to my mother”) which opens the section in which Gudea tells Nanshe about his dream.

These two formulas also involve sound repetition. In the first case (ga-na ga-na-ab-dug4) the directive expression /gana/ reproduces the exact same sounds of the two prefixes at the beginning of the following verbal form, the modal prefix ga- and the indirect object prefix –(n)na-. The second formula includes an alliteration of /m/ and /ĝ/ as well as the assonance of /a/ and /u/ (ama-ĝu10 ma-mu- ĝu10).

Repetitive Narrative Patterns

The ‘naming of the stelas’ (CA 23: 8 – 24: 7) is a long narrative sequence that shows repetitive patterns, and in particular structural parallelisms. Gudea names each of the seven stelas which he has erected surrounding the Eninnu. The naming of each stela takes up five lines.

The first line consists of an opening expression, which identifies the stela by the place where it was erected – an example from CA 23: 13 consists of na kan4-sur-ra bi-du3-a-na ‘to his stone, which he set up at the Kan-sura gate’. This follows the fixed structure na ‘stone’ — locative noun phrase – verbal form bi2-du3-a-na (‘that he erected’). The locative noun phrase changes according to the context, but the noun na ‘stone’, that functions as the absolutive of the relative clause, and the nominalized verbal form bi2-du3-a-na ‘which he erected’ are repeated in all the seven episodes.

The second line consists of the noun phrase na-du3-a ‘stela’ (literally ‘erected stone’) in the absolutive. The name of the given stela follows, starting with a periphrasis referring to Ningirsu. In the first occurrence, the sequence reads: na-du3-a lugal kisal si “the stela, ‘the king who fills the courtyard…”

The name of the stela continues in the third line, which is always repeated identically and places Gudea’s name alongside Ningirsu’s (gu3-de2-a en dnin-gir2-su2-ke4, ‘Gudea, the lord Ningirsu…’). Ningirsu is always the agent of the sentence, whereas Gudea does not always maintain the same logical function. In the first occurrence, this name is the direct object, in the second it is in the terminative, and so on. But no case marker follows his name. The notation of the case marker after a vowel was not common in Gudea’s time, but its systematic omission here may indicate an intention to preserve the symmetry and repetitive pattern of the passages.

The main verb of the clause (mu-zu) appears in the fourth line, preceded by a noun phrase with an adverbial meaning – the first occurrence includes gir2-nun-ta mu-zu ‘he knows him from a princely way’.

The fifth line consists of the closing formula na-ba mu-še3 im-ma-sa4 ‘to this stone he gave it as a name’, which is identical in every instance.

Similar examples of structural parallelisms can be detected in many more places in Gudea’s text. Examples include the sequence of Gudea’s prayers which is repeated three times, first, to Ningirsu, secondly to Gatumdug, and thirdly to Nanshe. Many more of these patterns may be identified. Because of their repetitive narrative structure, these sequences resemble very closely the ‘typical scenes’ and the ‘themes’ that characterise oral and aural poetry (Lord 2005, 133-134). The wider ramifications of this phenomenon cannot be discussed here exhaustively, but they are currently being studied within REPAC.

Download here.

Selected Bibliography

Averbeck, R.E. 1987, A Preliminary Study of Ritual and Structure in the Cylinders of Gudea (2 volumes). Unpublished PhD Dissertation, Dropsie College.
Black, J. 1992, ‘Some Features of Sumerian Narrative Poetry’ in M. E. Vogelzang and H.L.J. Vanstiphout (eds.) Mesopotamian Epic Literature: Oral or Aural? Lewiston, NY, 70-101.
Edzard, D. 1997, Gudea and his Dynasty. The Royal Inscriptions of Mesopotamia – Early Periods 3/1, Toronto.
Falkenstein, A. 1966, Die Inschriften Gudeas von Lagaš: Einleitung. Analecta Orientalia 30, Rome.
Jagersma, B2010 A Descriptive Sumerian Grammar. Unpublished PhD Dissertation, Universiteit Leiden.
Lord, A.B. 2005, Il cantore di storie, G. Schilardi (trans.). Lecce.
Römer, W. H. Ph. 2010, Die Zylinderinschriften von Gudea. Münster.
Suter, C.E. 2000, Gudea’s Temple Building: The Representation of an Early Mesopotamian Ruler in Text and Image. Groningen.
Thureau-Dangin, F. 1905, Les Inscriptions de Sumer et d’Akkad. Transcription et Traduction. Paris.

2.2. Notes on the construction of meaning in an Old Babylonian bilingual proverb about exotic animals

Written by Dr Frank Simons BA (Hons), MPhil.

How to cite: Simons, F., 2021, “Notes on the construction of meaning in an Old Babylonian bilingual proverb about exotic animals,” Project REPAC (ERC Grant no. 803060), 2019-2024, at 10.25365/phaidra.261 (accessed day/month/year).

Among the less well-attested collections of proverbs from the Old Babylonian period, one bilingual example stands out as particularly interesting. The collection is preserved on just two tablets, one of which remains unpublished (courtesy J. Matuszak). The published tablet, N 3395, is a two column school tablet from Nippur, first edited by Lambert (1960: 272-3), and later re-edited more comprehensively by Alster (1997: 288-9). 1 The entire collection contains somewhere in the region of 20 proverbs, though the manuscripts are not exact duplicates so it is impossible to be sure of its original extent. The collection as a whole has several interesting features, but here we will consider just a single proverb:

In each of the first four lines a very rare word – Sumerian in 3 instances, Akkadian in the other – is paired with a relatively common one in the other language: d i – b i – d a is otherwise attested only in a lexical list (Civil 1971: 179),2 t i l – l u – u g only in the royal praise poem Šulgi B (Castellino 1972: 36-37, l. 59; ETCSL l. 58.), and g u l – l u m and margû are hapax legomena (contra CAD M/1: 278 s.v. margû A. See Simons forthcoming A: §2c). The better attested d ì m – š á ḫ is known from a handful of texts, mostly lexical, as a word for bear (Simons forthcoming A: §§2a-2b), and with the exception of margû the Akkadian equivalents are all perfectly commonplace – imēru, šurānu and pīru are the usual words for donkey, cat and elephant respectively.

At first glance, the superabundance of rare words in this proverb is unusual and difficult to understand. On closer inspection, however, it is clear that the choice of this succession of rare words seems to have been motivated by the assonance and consonance of their constituent parts. The words g u l – l u m and t i l – l u – u g are phonetically related, revolving around /g/ and /l/, while d i – b i – d a and d ì m-š á ḫ share the almost homophonous initial sounds /dib/ and /dim/, as well as a similar pattern of vowels. The final syllable of g u l – l u m nearly forms a palindrome with the succeeding first syllable of m e – l u ḫ – ḫ a.3 Similarly, the juxtaposition of d ì m – š á ḫ, m a r – ḫ a – š i, margû and paraḫše emphasises the repetition of the consonants /m/, /š/, /ḫ/ and /r/. The same consonants are also notable in the other Akkadian animal names, imēru, šurānu, and pīru. It seems likely that this influenced the use of the word margû as the equivalent to Sumerian d ì m – š á ḫ, which is otherwise only equated with dabû ‘bear.’ In addition, the sign DÌM is composed of the signs GAL and LUGAL which, were they to be pronounced aloud, would resound with g u l – l u m and t i l – l u – u g.

Given the fact that the rare words explicitly refer to foreign animals, it seems wholly plausible that they are not in fact Sumerian or Akkadian per se, but rather foreign names of foreign animals. This is almost certainly the case for the otherwise unknown margû. The CAD understands margû as a foreign word (CAD M/1: 278 s.v. margû A), presumably on the basis that an Akkadian etymology gives either a deverbal noun from ruggû ‘to wrong, to make illegitimate claims’ (CAD R: 404 s.v. ruggû) + ma-, or a quadriliteral root *mrgˀ. The language of Marḫaši (the Jiroft civilisation) is almost completely unknown, but as the animal in question is said to be ‘of Marḫaši’, and the word margû is evidently a loanword from an uncertain language, it is perhaps within the bounds of reason to suggest that margû is a remnant of this language. The same may also be suggested of g u l – l u m and the language of Meluḫḫa (the Indus Valley civilisation), and perhaps of d i – b i – d a and Elamite, though I can find no plausible candidate in the Elamisches Wörterbuch (Hinz & Koch 1987). It is also plausible that this is ultimately another loanword from the Indus Valley civilisation. No country is given for the t i l – l u – u g but it is equally likely to be a foreign word.4

Clearly this proverb is a work of some poetic skill. The euphony present throughout the first four lines demonstrates that the words were carefully chosen, and, as Steinkeller has pointed out, the whole proverb is also geographically organised, with the lands listed in order from west to east (Steinkeller 1980: 9). This led Civil to suggest that the animals may stand figuratively, or through alliteration or pun, for the lands from which they are said to come (Civil 1998: 11-12, n. 6). The animals dealt with in the proverb, however, are at least plausibly identifiable with actual animals, and the practice of presenting exotic animals as diplomatic gifts make it likely that they actually came, or were thought to have come, from the lands in question. This will be discussed at greater length in a forthcoming paper (Simons forthcoming B) which will deal with the identities of the animals involved.

Crisostomo has recently demonstrated that Sumerian and bilingual proverb collections were assembled using the same sorts of analogical techniques as were lexical lists, and that individual proverbs could be generated, among other methods, through interlingual phonological analogies (Crisostomo 2019: 154-155). That is to say, phonetic correspondences between Sumerian and Akkadian words and phrases could play a large role in the development of proverbs. This offers a rather better way of interpreting the proverb discussed here. As we have discussed, there are clear interlingual analogies in the proverb between Sumerian, Akkadian, and whichever foreign languages the animal names came from. Following Crisostomo’s argument, these should be understood as the basis from which the text developed – the euphonic juxtaposition of foreign names for comparably powerful animals and foreign place names is the root of the proverb. The succession of very rare words we have examined here is, therefore, not merely an aesthetic choice, but is in fact fundamental to the development of meaning in this text.

1 Alster notes that Lambert’s edition was made before the tablet was baked. It has also been collated by Castellino 1972: 117 and by Civil 1998: 11 n. 5. The edition given here follows that of Alster.

2 Izi? “C“ iv 35. d i – b i – d a = e-me-ru ‘d i b i d a = donkey’. This is a Middle Assyrian tablet (VAT 9714; CDLI P282498) provisionally assigned to the acrographic lexical series Izi = išātu by Civil, but with the proviso that it is likely a development from the exclusively Old Babylonian series N í g – g a = makkūru, and its exact identification is therefore uncertain.

3 It is possible that both d i – b i – d a a n – š a 4 -a nki – n a and g u l – l u m m e – l u ḫ – ḫ aki are sandhi spellings, which is to say that the animal name and the place name have rolled into one – d i b i d a n š a n and g u l l u m m e l u ḫ ḫ a respectively. It is not possible to be certain, however, as both d i b i d a and g u l l u m are so rare that we do not know their normal forms.

4 It has been noted that d ì m – š á ḫ seems to have been borrowed from a Semitic language (Civil 1998: 12). See further Simons forthcoming A: §2a, n. 21.

Download here.


Alster, B. 1997, Proverbs of Ancient Sumer: The World’s Earliest Proverb Collections (2 Volumes). Bethesda, MD.
Castellino, G.R. 1972, Two Šulgi Hymns (BC). Studi Semitici 42, Rome.
Civil, M. 1971, Izi = išātu, Ká-gal = abullu, and Níg-ga = makkūru. Materials for the Sumerian Lexicon 13, Rome.
Civil, M. 1998, ‘“Adamdun,” the Hippopotamus, and the Crocodile’ in Journal of Cuneiform Studies 50, 11-14.
Crisostomo, J. 2019, ‘Creating Proverbs: The Listing Scholarship of the Sumerian Proverbs Collections’ in KASKAL 16, 141-157.
Hinz, W./H. Koch, 1987, Elamisches Wörterbuch (2 Volumes). Berlin.
Lambert, W.G. 1960, Babylonian Wisdom Literature. Oxford.
Simons, F. forthcoming A, ‘Lions and Bison and Bears, oh my! Thoughts on some rare words in the cuneiform lexical tradition I – Animals and Instruments.’
Simons, F. forthcoming B, ‘The Donkey of Anšan – a Rhinoceros in Mesopotamia?’
Steinkeller, P. 1980, ‘The Old Akkadian Term for <<Easterner>>’ in Revue d’Assyriologie 74, 1-9.

2.3. Parallelism in Ugaritic Poetry

Written by Clemens Steinberger

How to cite: Steinberger, C., 2022, “Parallelism in Ugaritic Poetry 01,” Project REPAC (ERC Grant no. 803060), 2019-2024, at (accessed day/month/year).

Parallelism (often referred to as parallelismus membrorum) is a fundamental characteristic of Ugaritic poetry and its main principle of versification. Poetic parallelism relies on the juxtaposition of two linguistic sequences that share semantic, grammatic, phonetic, or graphic properties. As certain features of the first verse unit are repeated in the next, the two units are tied together. The subsequent verse unit emphasizes, complements, contextualizes, specifies, increases, advances, or contrasts the first unit’s issue. Poetic parallelism is employed in epics and mythological texts (KTU 1.1–1.24), incantations and historiolae (KTU 1.100, 1.114, 1.169), as well as in some prayers and evocations accompanying rituals (KTU 1.23, 1.108, 1.119, 1.161). This brief study aims to provide an overview of parallelism’s implications for the semantics, grammar, phonetics, and visuals of Ugaritic poetic texts (parallelism has been treated several times in Ugaritic studies; a brief bibliography is found on the Ugarit-Portal Göttingen; here, you will also find German translations of selected Ugaritic poetic texts).

Semantic Parallelism: Parallel verse units usually contain semantically related words and phrases (the semantic – more precisely: paradigmatic – relations of juxtaposed lexemes are manifold and cannot all be dealt with here). Parallel expressions may bear the same or a very similar meaning, employing more or less synonymous lexemes like ˁr “city” and pdr “town, city” in ex. a:

a) KTU 1.16 VI 6–7
6 ˁrm . tdu . mt[[x]]
7 pdrm . tdu .šrr

From the city she scares off Môtu (i.e., death),
from the town she scares off the enemy.

However, for many word pairs that are considered synonymous, it is reasonable to assume that the juxtaposed lexemes exhibit minor differences in meaning or bear different connotations (e.g., in the case of Ugaritic / Northwest Semitic words paired with words of foreign origin). Supposedly synonymous terms may harbor different value judgements, originate from different sociolinguistic contexts, and thus evoke different associations for the audience. Therefore, in most cases, we should refer to partial synonymy rather than synonymy. By juxtaposing two partially synonymous terms, the common denotative core of meaning is emphasized, while the peripheral connotations of the two lexemes voice different facets of the superordinate issue.

Apart from partially synonymous terms, hyponyms (sub-terms) and hypernyms (superordinate terms) can be joined in poetic parallelism (cf. Tsumura 1988). In this case, the second term classifies (as a hypernym) or specifies (as a hyponym) the first. A superordinate term following a more specific one allows for the correct classification of the first statement. In ex. b, the phrase ṣbrt aryh “the flock of her kin” indicates that bnh “her sons” in the preceding verse unit refers to the entire clan of the goddess ˀAṯiratu (she is considered the creatress of the gods in Ugarit). Moreover, ˀAṯiratu’s name is replaced in the second unit by the more general term ilt “goddess,” which can equally be taken as a hyponym-hypernym sequence (the juxtaposition of a figure’s name with an epithet is frequently found in Ugaritic poetry):

b) KTU 1.3 V 36b–37
yṣḥ . aṯrt 37 w bnh .
ilt .w ṣbrt . ary!h

He called ˀAṯiratu and her sons,
the goddess and the flock of her kin.

On the other hand, a hyponym following a hypernym specifies the issue, just as ymn “right hand” specifies which yd “hand” (left or right) is meant in ex. c (Tsumura 1988, 259–260):

c) KTU 1.19 IV 53b–54b
qḥn . w tšqyn . yn .
[q]54 ks . bdy .
qbˁt . b ymny

Take (it) and give (me) wine to drink,
t[ak]e the cup from my hand,
the goblet from my right hand!

Likewise, holonyms (whole) and meronyms (part) can be linked in poetic parallelism. In ex. d, yd “hand” is parallel to uṣbˁt “fingers,” which are parts of the yd “hand.” The meronym probably provides a synecdoche for the holonym “hand.” At the same time, the image drawn of the goddess meticulously washing every single finger becomes more detailed:

d) KTU 1.3 II 32b–33
trḥṣ . ydh . bt33lt . ˁnt .
uṣbˁth . ybmt . limm .

Virgin ˁAnatu washed her hands,
the sister-in-law of the peoples (/ of Liˀmu) her fingers.

In addition, multiple hyponyms (of the same hypernym) can occur side by side. The juxtaposed terms derive from the same field of meaning or have a quality in common that is decisive for the story. Set in parallel, co-hyponyms serve to exemplify an entire field of meaning and thus visualize the scene. In ex. e, ḫrṣ “gold” and ksp “silver” are parallel, in this passage co-hyponymous for valuable metals. rqm “sheets” and lbnt “bricks” in turn exemplify different building materials:

e) KTU 1.4 VI 34–35a
34 sb . ksp . l r⸢q⸣m .
ḫrṣ 35 nsb . l lbnt

The silver turned into sheets,
the gold turned into bricks.

Likewise, epithets and names of figures that are assigned the same or a similar role in the narrative occur in parallel. In ex. f, various enemies of Baˁlu are mentioned in parallel, which ˁAnatu is said to have defeated:

f) KTU 1.3 III 43–46a
43 mḫšt . mdd ilm . arš
44 ṣmt . ˁgl . il . ˁtk
45 mḫšt . k{.}lbt . ilm . išt
46 klt . bt . il . ḏbb

I struck down the beloved of ˀIlu, ˀARŠ,
I destroyed ˀIlu’s calf, ˁTK,
I struck down ˀIlu’s bitch, ˀIŠT,
I annihilated ˀIlu’s daughter, ḎBB.

Rarely, parallelism serves to elaborate comparisons. The couple is then made up of a reference word that is used in its literal meaning and a metaphorical expression or a comparative phrase figuratively describing the first one (see bˁl mrym ṣpn “Baˁlu from the heights of Zaphon” // k ˁṣr udnh “like a bird from its nest” in ex. g):

g) KTU 1.3 III 47b–IV 2a
rd . bˁl IV 1 mrym . ṣpn .
ṣ . k . ˁṣr 2 u{.}dnh .

(I fought for the silver, acquired the gold of him)
who expelled Baˁlu from the heights of Zaphon,
who made (him) fly away like a bird from its nest.

In Ugaritic poetry, co-referent parallel expressions (referring to the same issue) are occasionally attached with dissimilar numerals (this phenomenon is to be distinguished from enumerations): two subsequent verse units each contain a numeral (having the same syntactic function), with the number in the second verse unit being higher than the first. The numbers’ arithmetical meaning is secondary; rather, the numbers illustrate the (enormous) extent of a given matter, resulting in an increase (from the lower number to the higher; Segert 1983, 304). In ex. h, the two numbers 77 and 88 are parallel. By juxtaposing the two two-digit palindromic numerals, it is indicated that Baˁlu and his lover (a heifer) slept with each other many, many times. In ex. i, the two numbers 1.000 and 10.000 are connected with the units of measure šd and kmn. Given this passage (the pair of measurements is frequently found in Ugaritic poetry), it is probably not to be concluded that a šd is exactly ten times as large as a kmn; rather, it is intended to showcase the enormous size of Baˁlu’s palace:

h) KTU 1.5 V 19b–21
škb 20 ˁmnh . šbˁ . l šbˁm
21 [ˁ]ly . ṯmn . l ṯmnym

He slept with her 77 times,
she let him [mo]unt 88 times.

i) KTU 1.4 V 56–57
56 alp . šd . aḫd bt
57 rbt . kmn . hkl

The house shall occupy 1.000 šiddu,
the palace 10.000 kumānu!

In antithetic (or contrastive) parallelism, comparatively rare in Ugaritic poetry, terms or phrases with contrasting meanings are juxtaposed (cf. Watson 1986b; see the complementary contrast between d ydˁnn and d l ydˁnn in ex. k, or the directional opposition between low and high in KTU 1.23 32a: hlh [t]špl hlh trm “Look, one gets down low, // look, the other gets up high”). Antithetical parallelism usually involves two opposing agents: at times, they are associated with two contrasting yet coequal issues (see ex. k). However, antithetical parallelism can also serve to view a superordinate issue from two opposing perspectives, the second statement presupposing the first. In ex. j, the antonymous verbal forms ḫt “be smashed” and li “be victorious” are opposed. In this case, Šaˁtiqatu’s victory presupposes Môtu’s expulsion (cf. Watson 1986b, 415):

j) KTU 1.16 VI 1–2a
1 [m]t . dm . ḫt .
šˁtqt . dm! 2 li

[Mô]tu, be smashed!
Šaˁtiqatu, be victorious!

A rather unique case is found in ex. k. The verb ydˁnn, which occurs in the first line as part of the relative clause d ydˁnn “the one who knows him,” is repeated in the second verse unit (again as part of a short relative clause). Here, however, the verb is negated: d l ydˁnn “the one who does not know him.” The contrast between the god who knows Yarḫu and the god who does not is further illustrated in the main clauses: one hands the moon god food, the other beats him with a stick. The two phrases exemplify two contrasting attitudes, one benevolent, the other harsh (cf. Segert 1983, 300; Watson 1986b, 419):

k) KTU 1.114 6b–8a
il . d ydˁnn 7 yˁdb . lḥm . lh .
w d l ydˁnn 8 y[[x]]lmn ḫṭm

The god who knows him (i.e., Yarḫu) passes him food,
yet the one who does not know him beats him with a stick.

Lastly, it is to be noted that antithetical parallelism can be achieved by juxtaposing two protagonists’ names who are hostile. In ex. j, the names of the opponents Môtu and Šaˁtiqatu are juxtaposed and combined with antonymous verbs (the antithesis between the two is reinforced by the opposition of male and female gender; Watson 1986b, 415). In KTU 1.6 VI 17a, the names of Môtu and Baˁlu, the two gods facing each other in battle, are opposed. In this case, however, the names are combined with identical verbal forms: mt ˁz bˁl ˁz “Môtu was strong, Baˁlu was strong” (Steinberger 2022, 75–76).

Antithesis is formally identical to merism: here, however, the focus is not on the difference, but on the whole area that lies between the two contrasting terms (usually two spatial references). In ex. l, the terms šmm “skies” and nḫlm “wadi” illustrate the two opposite regions (above and below) from which usually water springs, but where now oil and honey flow. The parallelism expresses the extent of the paradisaical state that prevails from the very top to the very bottom:

l) KTU 1.6 III 12–13
12 šmm . šmn . tmṭrn
13 nḫlm . tlk . nbtm

The skies rained oil,
the wadis ran with honey.

Syntactic and Morphologic Parallelism: The verse elements that semantically match usually bear the same syntactic function. Thus, parallel verse units often contain equivalent constituents (see, e.g., ex. n: each colon comprises subject, accusative object, and verbal predicate). Syntactically parallel elements may correspond morphologically (regarding word class and specifications in conjugation or declension); however, parallel elements do not necessarily have to be morphosyntactically identical. In so-called asymmetrical constructions, juxtaposed verse parts exhibit minor morphosyntactic differences, allowing the poet to pepper parallel verse units with small variations and thus make speech flow more vivid (Gzella 2007). Hence, parallel terms are sometimes given different suffixes, though the meaning seems to remain unchanged. Not least, parallel words that are otherwise largely identical often show minor morphologic transformations (e.g., verbs with energic suffix are often parallel to verbs without energic suffix; cf. UG2 500; see also ex. n, where the verbal form nbln differs from the parallel verb nbl only by the suffix -n; see also below on repetitive parallelism and the polyptoton).

The repetition and transformation of the first verse unit’s word order in the second and third unit gives rise to different structural varieties of parallelism. Exemplarily, we shall look at the verse composed of two or three cola (cf. Steinberger 2022, 61–63, for an overview of verse units in Ugaritic poetry). In parallelism’s most basic form, the structure of the first colon is maintained in the following colon (the sentence elements are repeated in the same order; see ex. a, f, j). In chiastic constructions, the elements of the first colon are repeated in reverse order in the second (see ex. m, featuring anadiplosis; note that modifiers and particles here merge with a main constituent, forming one unit that is transposed as a whole; cf. Watson 1983, 259–260):

m) KTU 1.17 V 10b–11
{ hlk . kṯr } 11 { k yˁn . }
{ w yˁn . } { tdrq . ḫss }

He indeed saw the coming of Kôṯaru,
yes, he saw the approaching of Ḫasīsu.

In verses with every colon containing more than two (mostly three) constituents, usually two of them join to form a compound colon clause (consisting either of the verbal predicate and a nominal constituent or of two nominal constituents). The compound clause and the remaining single constituent are taken up independently in the subsequent colon. The elements of the compound clause can be rearranged (they are chiastic to the elements of the corresponding compound clause), while the compound clause, seen as a whole, is in the same position in each colon. Likewise, the constituent independent of the compound clause is in the same place in each colon (see ex. n, e, l; cf. Watson 1983, 261–263):

n) KTU 1.3 V 33b–34b
klnyy . { qšh 34 nbln . }
klnyy . { nbl . ksh }

We all want to bring his jug,
we all want to bring his cup!

Occasionally, the constituent that is independent of the compound clause is placed at the beginning of the one colon and at the end of the other colon (it is arranged in chiastic order). The elements of the compound clause, however, are rendered in the same order. In ex. o, the verb yqḥ follows the interrogative pronoun mh in both cola; the direct object (mt uḫryt // mt aṯryt) is once at the beginning of the colon and once at the end (cf. Watson 1983, 260):

o) KTU 1.17 VI 35b–36a
mt! . uḫryt . { mh . yqḥ }
36 { ⸢mh . yqḥ . } mt . aṯryt

Death at the end – what can take it away?
What can take away death in the final stage?

These are but the most basic connections between cola, each containing the same constituents. It is to be noted, however, that often elements of the first colon are omitted in the second or third colon, while in other cases elements are added in the second or third colon, yielding, e.g., terrace verses (cf. Watson 1986a, 208–210), staircase verses (cf. Watson 1986a, 150–156) or elliptical verses (cf. Miller 1999).

Visual Parallelism: In addition to grammar and semantics, parallelism occasionally affects the layout and the phonetics of poetic texts, reinforcing and extending the connection between juxtaposed elements. In visual (or graphic) parallelism, the same or similar signs are arranged one above the other in two or more successive lines, yielding a recurring pattern of text that is visually perceptible (cf. Yogev / Yona 2018). Unlike other forms of poetic parallelism, visual parallelism only appeals to the writer and reader of a text, but not to the listener.

On the Ugaritic tablets, the beginnings of successive lines are at times shaped visually parallel. In KTU 1.15 III 7–12, e.g., the sequence {tld . pġt} (“she shall bear the girl”) is repeated at the beginning of six successive lines (although the phrase is not fully preserved in each line; see the WSRP photos UC15303965 and UC15304134): the corresponding signs are arranged one above the other. Further examples of line-initial visual parallelism are found in KTU 1.4 (cf. Yogev / Yona 2018): {klnyn} in IV 45–46 (repeated twice); {mṯb} in IV 52–57 (repeated five times); {ḥš} in V 51–54 (repeated four times); {špq . il} in VI 47–54 (repeated eight times); {ˁm . ġr} in VIII 2–3 (repeated twice, with ˁm again repeated at the beginning of l. 4).

Phonetic Parallelism and Rhyme: At times, the phonetic shape of poetic texts is influenced by parallelism (Pardee 1988, 51–57 / 182–185). Phonemes from the first verse unit are taken up in the next, yielding different forms of rhyme (note that the study of Ugaritic rhyme is complicated by the fact that the Ugaritic writing system is primarily consonantal). Either whole syllables correspond, or only the vowel or the consonant sequence (assonance vs. consonance).

Initial rhymes build on the phonetic similarity between the first parts of two or more subsequent verse units (cf. Watson 1999, 184). In ex. p, the words tant and thmt (which are neither semantically nor grammatically parallel) are linked by an initial rhyme. Except for the vowel of the penultimate syllable and the case ending, tant and thmt share similar syllables: /ta/ is followed by a laryngeal (/ˀ/ and /h/), the vowel /a/ and a nasal (/n/ and /m/); in both words, /t/ is the last consonant (note, however, that the vocalization of tant is debatable; cf. Bordreuil / Pardee 2009, 168; UG2 270):

p) KTU 1.3 III 22b–25
rgm 23 ˁṣ . w . lḫšt . abn .
24 tant . šmm . ˁm . arṣ
25 thmt . ˁmn . kbkbm

rigmu ˁiṣṣi wa LḪŠ-(a)tu ˀabni
taˀanîtu šamîma ˁimma ˀarṣi
tahāmāti ˁimma
(n)na kabkabīma

The word of tree and whisper of stone,
the whispering of the skies with the earth,
of the floods with the stars!

Furthermore, end rhymes are attested in Ugaritic poetry, frequently involving homeoptota. (Often, it is not clear whether rhymes occur only by chance or whether they were deliberately used to connect subsequent verse units. The question arises not least in the case of rhymes between recurring grammatical elements, including homeoptota, e.g., when parallel words share the same prefixes or suffixes, or they are connected with identical particles. These forms of rhyme could also be seen as a by-product of grammatical parallelism.) In ex. p, the lexemes abn and arṣ both start with /ˀa/ and end with the genitive ending /-i/. In ex. q, the verbal forms at the end of the two cola show the same vowel sequence (they are both analysed as L-stems; UG2 577 / 650) and have the same pronominal suffix (-k). Furthermore, the second syllable of both words starts with a guttural (// and /ˁ/):

q) KTU 1.4 IV 38b–39
hm . yd . il mlk39 yḫssk .
ahbt . ṯr . tˁrrk

himma yadu ˀili malki 39 yuḫâsisuki
ˀahbatu ṯôri tuˁâriruki

Or does the love of ˀIlu excite you,
does the passion of the bull arouse you?

Repetitive Parallelism: In Ugaritic studies, repetition is commonly considered a form of parallelism (cf. Pardee 1988, 169–170). Here, the juxtaposed terms derive from the very same lexeme. The word forms may be identical, corresponding morphosyntactically; at times, however, the lexeme is modified morphosyntactically in the second unit (which is the case with polyptota). Needless to say, the repetitive elements overlap phonetically and, if the verse units are arranged one above the other on the tablet, yield visual parallelism. The verbatim repetition of entire cola within a verse is rare (at times, individual elements of parallel verse units correspond verbatim, while the others diverge; see ex. a; note, however, that whole narrative sections may be repeated verbatim, which is to be considered repetitive parallelism at its largest scale). In ex. r, the parallel cola are identical except for bn standing in parallel to bnm, the same lexeme with an additional particle –m:

r) KTU 1.15 III 20–21
20 w tqrb . wldbnlh
21 w tqrb . wld . bnm lh

So, her time came to bear him a son,
so, her time came to bear him a son.

Outlook: Poetic parallelism influences the vocabulary, grammar and structure of Ugaritic poetic texts and occasionally affects their phonetic form and their graphic layout. Parallelism links verse units of different lengths. The various scales at which parallelism comes into play have not been addressed in this study. The references discussed above are verses composed of two or three cola. However, parallel links are found on a smaller scale as well, i.e., between phrases within single cola (cf. Watson 1984). On the other hand, poetic parallelism is at work in strophes connecting several verses (cf. Steinberger 2022). Furthermore, parallelism affects the overall structure of poetic texts given that whole narrative sections can be built in parallel (often repetitive-parallel). Thus, a close study of parallelism reveals both the micro- and the macro-structure of Ugaritic poetic texts.

Download here.

Selected Bibliography

Bordreuil, P. / D. Pardee, 2009, A Manual of Ugaritic. Linguistic Studies in Ancient West Semitic 3, Winona Lake, IN.
Gzella, H. 2007, ‘Parallelismus und Asymmetrie in ugaritischen Texten’ in A. Wagner (ed.) Parallelismus membrorum. Orbis Biblicus et Orientalis 224, Fribourg / Göttingen, 133–146.
Miller, C.L. 1999, ‘Patterns of Verbal Ellipsis in Ugaritic Poetry’ in Ugarit-Forschungen 31, 333–372.
Pardee, D. 1988, Ugaritic and Hebrew Poetic Parallelism – A Trial Cut (ˁnt I and Proverbs 2). Supplements to Vetus Testamentum 39, Leiden / New York / Copenhagen / Köln.
Segert, S. 1983, ‘Parallelism in Ugaritic Poetry’ in Journal of the American Oriental Society 103, 295–306.
Steinberger, C. 2022, ‘Das Versmuster A-B │ A()-B() in der akkadischen und ugaritischen Poesie’ in Journal of Northwest Semitic Languages 48/1, 61–82.
Tsumura, D.T. 1988, ‘A “Hyponymous” Word Pair: ˀrṣ and thm(t) in Hebrew
Watson, W.G.E. 1983, ‘Strophic Chiasmus in Ugaritic Poetry’ in Ugarit-Forschungen 15, 259–270.
– 1984, ‘Internal Parallelism in Ugaritic Verse’ in Studi Epigrafici e Linguistici sul Vicino Oriente Antico 1, 53–67.
– 1986a, Classical Hebrew Poetry – A Guide to its Techniques (2nd ed.). Journal for the Study of the Old Testament – Supplement Series 26, Sheffield.
– 1986b, ‘Antithesis in Ugaritic Verse’ in Ugarit-Forschungen 18, 413–419.
– 1999, ‘Ugaritic Poetry’ in W.G.E. Watson / N. Wyatt (eds.) Handbook of Ugaritic Studies. Handbook of Oriental Studies I.39, Leiden / Boston, MA / Köln, 165–192.
Yogev, J. / Y. Shamir, 2018, ‘Visual Poetry in the Ugaritic Tablet KTU 1.4’ in Journal of the Ancient Near Eastern Society 33, 203–210.

2.4. When Repetition Fails
The Curious Case of the Hittite Šar Tamḫāri

Written by Henry J. A. Lewis

How to cite: Lewis, H.J.A., 2024, “When Repetition Fails—The Curious Case of the Hittite Šar Tamḫāri, Version 01,” Project REPAC (ERC Grant no. 803060), 2019-2024, at (accessed day/month/year)

  1. Introduction

    As a subset of the small field of Cuneiform studies, Hittitology has been slow to approach questions of poetics, let alone repetition and parallelism specifically. This is not to say that interest in the study of Hittite poetics has been absent (cf. already Hrozný 1929). But most analyses of Hittite poetics focussed largely on the question of a possible existence of metrical patterns among those texts designated ŠÌR, ‘song’ (esp. the Song of Release), or on works imported from Mesopotamia, whose native poetic features were better understood (for a history of Hittite poetics see Francia 2012a).
    Only with the philological maturity of the field in the past two or three decades have scholars begun to investigate the poetic features of a broader range of Hittite texts. Notable contributions have included articles by Rita Francia (2010; 2012a; 2012b; 2018) and Calvert Watkins (1995; 2010). However, there are now two, recent monograph-length studies that have made poetics their principal focus with a concomitant, deeper engagement with theory: Daues and Rieken’s Das persönliche Gebet bei den Hethitern (2018; esp. Daues’ chapter 5); and Marinaeu’s The Literary Effects of Discourse Patterns in Hittite Texts (2020 PhD Thesis). The theories used in these studies are essentially the same as those found in Assyriological scholarship (Jakobson 1987; Berlin 2008; Jefferies/McIntyre 2010; cf. De Zorzi 2022). Thus, as in Assyriology, the concept of parallelism in Hittitology can be understood as “the activation of linguistic equivalences and/or contrasts within or among words, phrases, lines, or entire texts” (Berlin 2008, 151-152), as well as its function in emphasizing the poetic message (Berlin 2008, 141; De Zorzi 2022, 368).
    The present contribution to REPAC focusses on a single text: the Hittite Šar Tamḫāri. The narrative is one of the most vexatious in Hittitology, largely due to the sheer number of, what seem to be, incomprehensible errors committed by the scribe, whether from a faulty understanding of Old/Middle Hittite or failed attempts at deliberate archaization (Rieken 2001). However, in the preparation of a new edition, some points of repetition and parallelism have emerged that bear witness to the text’s well-structured nature. At the same time, the scribe who copied this tablet endangered such a structure when he struggled to write repeated lines, raising the question as to why things went so wrong for him.
    This showcase will demonstrate the cases of poetic parallelism and repetition in the surviving text, before explaining how it was almost undone by the scribe. The poetics of the Akkadian recension of the text (EA 359) are also brought into dialogue with the Hittite version, to see if their relationship can be in any way clarified.
  1. Manuscripts

    The Hittite Šar Tamḫāri is listed on the Konkordanz of the Hethitologie Portal Mainz under CTH 310. Manuscripts date to the New Hittite period (c. 1400-1200 BC) (with one Late New Hittite duplicate), but exhibit features that date to at least Middle Hittite (15th c. BC) and are possibly as old as Old Hittite (c. 1650-1500 BC) (Rieken 2001). Thus, the narrative is almost certainly more archaic than the tablets on which it is preserved.
    Several fragments currently classified as CTH 310 on the Konkordanz ought to be treated with caution, since they cannot conclusively be identified as belonging to the Šar Tamḫāri. In toto, only CTH 310.1 (KBo 3.9), 310.3/5 (KBo 12.1 // KBo 22.6 + KBo 22.97), KUB 48.98, and 310.4 (KBo 13.46) belong with certainty to the text, though each (except for duplicate KBo 12.1) narrates the story in a slightly different way. The latter two, KUB 48.98 and KBo 13.46, appear to narrate a scene also found in KBo 22.6, and therefore add little to the story. Therefore, most of the following analysis draws upon KBo 22.6 + KBo 22.97, the largest and best-preserved manuscript. Reference to KBo 3.9 is made in the reconstruction of the story. For the only published edition see Güterbock 1969; for transliterations see Rieken 2001, Groddek 2008, Torri and Barsacchi 2018a, and 2018b.
  1. The Story

    Anyone familiar with the Akkadian Šar Tamḫāri (EA 359; see Westenholz 1997 and Haul 2009) will recognise that the Hittite recension shares in a similar plot, but the sequence of story-events is somewhat different, and the Hittite version adds its own scenes. As assessed by Güterbock (1969, 14), the text appears to be “keine wörtliche Übersetzung, sondern eine freie Nacherzählung der akkadisch überlieferten” (Güterbock 1969, 14). In addition, the beginning of the largest fragment (KBo 22.6+) is broken at its beginning, middle, and end. Large swathes of the plot must be inferred, leading to the following reconstruction (see also Gilan 2014, 54):

    1) CTH 310.1 (KBo 3.9)— After a (possible?) list of epithets (Obv. 1ʹ-7ʹ), we find Sargon in the gatehouse (Éḫilamni) of Akkad (Obv. 8ʹ). He is addressing someone about roads, presumably related to him by merchants (Obv. 10ʹ), which lead to Purušḫanda (Obv. 9ʹ-14ʹ).

    2) CTH 310.3/5 (KBo 12.1 // KBo 22.6+ KBo 22.97)— At some point, Sargon goes to sleep and Ištar appears to him in a dream. She assures him that he will be victorious in his conquest of Purušḫanda. Sargon awakes and tells his heroes (some of whom had been reluctant to campaign) that Ištar has guaranteed him victory. Having set out, Sargon bridges and crosses the Tigris (ÍDAranzaḫ), sacrificing to both the river and the bridge. (Obv. I 1ʹ-20ʹ).

    3) Obv. I 20ʹ-29ʹ— Meanwhile, Enlil appears in a dream to Nūr-Daḫḫi, the king of Purušḫanda. He warns him that Sargon is coming but assures him of his divine weapons (compared to destructive acts of nature) and lack of equal.

    4) Obv. II 1ʹ-16ʹ— Though highly fragmentary, it appears as if Nūr-Daḫḫi and his people (warriors?) are in dialogue about the topographical difficulties one must face to reach Purušḫanda (Obv. II 4ʹ-10ʹ). Confounding Nūr-Daḫḫi’s estimation, Sargon then, on a sudden, arrives (Obv. II 11ʹ-16ʹ).

    5) Rev. III 1ʹ-13ʹ— Sargon repeats Nūr-Daḫḫi’s speech to him from the previous section (1ʹ-7ʹ). Nūr-Daḫḫi acknowledges Sargon’s superiority in another speech (8ʹ-13ʹ).

    6) Rev. IV 1ʹ-7ʹ— An act of investiture is performed where Sargon is seated on a golden throne.

    7) Rev. IV 8ʹ-32ʺ— Sargon remains in Purušḫanda for three years and five months. As he sets out for Akkad, his heroes remind him that they have ‘done nothing’ to the land. They urge him to cut down three types of trees in the gatehouse of Purušḫanda to make into different objects, tear down the walls of city, and make an image of himself with Nūr-Daḫḫi to install on the gates. Sargon acquiesces and the tablet breaks off.

Parallelism and Repetition in the Hittite Šar Tamḫāri

Parallelism exists on the macro- and microscopic scale in the preserved text, with symmetry occurring between scenes themselves and the details they contain. When analysing micro-structures, the Hittite text is separated by clause as opposed to tablet line, as is customary in Hittite stylistic analyses (cf. ‘colon’ in Daues and Rieken 2018, 181).

4.1 Macro-Structures

4.1.1 Ring-Composition

If the reconstruction of the story (above) is accurate, one can initially note that ring-composition is taking place: the action begins in the éḫilamni of Akkad and ends in the Éḫilamni of Purušḫanda.

The ring-composition can, however, only be adduced for the story (or ‘plot’; see also Gilan 2000, 88). The manuscripts only partially attest to the phenomenon. KBo 22.6 may have preserved a scene in the Éḫilamni of Akkad at its incipit, but that portion of the text is lost. Likewise, KBo 3.9 may have had a scene in the Éḫilamni of Purušḫanda at its close.

A similar structuring may also have been used in a Mesopotamian version of the story. The Nineveh recension (K13228) seems to reflect the same opening as in KBo 3.9. However, the Amarna recension does not bear witness to this feature, beginning in media res (Meriggi 1973, 200; Westenholz 1997, 108).

4.1.2 Parallel Dream Sequences

The text constructs the two dream sequences of Sargon and Nūr-Daḫḫi ((2) and (3) above) in parallel. Not only is the act of deity (Ištar/Enlil) visiting king (Sargon/Nūr-Daḫḫi) mirrored in both sequences, but the content of the speeches too. For example, though fragmentary, I understand KBo 22.6 Obv. I 5ʹ to express Sargon’s wish that his army is not hindered by adverse winds:

A semantic parallel thus emerges: the weapons, compared by Enlil to a šalli ḫuwanti, ‘a great wind’, resume Sargon’s wish for the army to be unhindered by adverse weather, though the repetition is stylistically varied through use of Sumerogram in the first instance and Hittite in the latter. Dramatic irony is thereby created: the ‘wind’ that the army will face is really the metaphorical wind of Nūr-Daḫḫi’s weapons.

There are subtle differences between parallels. Ištar, if we are to take Sargon’s address to the soldiers at face value, has guaranteed the king victory. In terms of narrative suspense, this is somewhat counterintuitive: the stakes are low for Sargon. But it casts a rather sombre shadow over Nūr-Daḫḫi in his dream. Whether this is a “lying dream like that of Agamemnon’s in Book 2 of the Iliad” as Bachvarova (2016, 172-173; see also Haul 2009, 270; Mouton 2007, 15; and Gentili 2000, 367-269) would have it is debatable. The obverse breaks off before the end of the speech, so that all we can say with certainty is that Enlil attempts to embolden Nūr-Daḫḫi, but not necessarily lie (unless annauliš=wa=[tta?] ŪL kuiški ešzi, ‘Equal to you is there no one’ (Obv. I. 25ʹ-26ʹ) is a genuine (false) promise?). Thus, from a modern perspective, Nūr-Daḫḫi becomes a tragic figure, emboldened to fight, but doomed to fail (although whether a Hittite audience would perceive him as such is unprovable). One imagines that the parallels between these scenes were much fuller, but the broken tablet precludes further analysis.

4.2 Micro-Structures

4.2.1 The Speeches of Nūr-Daḫḫi/Sargon

A better preserved example of parallelism is the repeated speech of Sargon/Nūr-Daḫḫi, which can only be analysed between Rev. III 2ʹ-6ʹ, where it is least broken (story sections (4) and (5) above). In these lines we find a combination of a syntactic parallelism and semantic chiasmus. The speech was likely first related by Nūr-Daḫḫi to reassure himself and/or his troops/people, insofar as the king believes that the difficult topographical features on the road to Purušḫanda will stop Sargon. The features are organised as follows:

The fronted negative rhetorical question, ŪL=war=an aranzi, makes the parallelism clear, with the final instance applying to two geographical features rather than one. This asymmetrical end reminds one of a notable Indo-European stylistic figure, termed by Watkins (2010, 329-335) the a a b triad: “the main verb occurs triadically, twice as one lexical item and the third time climatically as a semantically similar but more highly marked variant.” Here however the triad is of substantives, not verbs. The variant in the speech above of two topographical features for one emphasizes the difficulty of the journey, the accumulation of features mirroring the accumulation of difficulties.

A chiastic ordering of the speech, by semantic field, is also evident: earth (‘mountains’); water (‘floodwaters’); water (‘meadows’); earth (‘paths’). This chiastic pattern is reinforced by the adjectives qualifying the nouns: both ḪUR.SAGMEŠ and KASKALḪI.A have two adjectives and share one in ‘ḫatugaeš’; conversely, arunaš and Ú.SALḪI.A are modified by one adjective each, the former, however, in a genitive construction arunaš l[elḫurtimaš(?)] that is akin to the adjective-description (‘floodwaters of the sea’ = ‘flooding sea’) and thus stylistically varied again.

An alternative interpretation of the same passage could also be suggested (see Marineau 2020, 45-46, for discussion of Fabb’s (2004) ‘formal multiplicity’ as aesthetic): Rev. III 2ʹ-4ʹ (A+B) represent the perimeters of the landscape, the mountains and the sea, whereas Rev. III 5ʹ-6ʹ (C) the area in between these natural borders, the meadows and the roads. The speech thus emphasizes the expanse of territory by framing the extremes of the landscape, a technique already present in Sargonic inscriptions (e.g., Rimuš E2.1.2.9, 1-17) and popular in OB literature, termed merism: “conceptual totality is expressed, concretum pro abstracto, by the use of two antipodal terms” (Wasserman 2003, 61). Our example differs slightly, in that the ‘middle area’ is, unlike in typical merism, ‘defined’ (‘meadows’ and ‘roads’), when normally the area between extremes is left unsaid. Regardless, the text still engages in a very similar kind of poetics.

Further, if it is correct to suggest that the Akkadian and Hittite versions are related here (Gilan 2000, 64-65; 83; Soysal 2017, 221 fn. 21), then we can validly recognise the creativity of the Hittite scribe in their re-ordering and choice of certain elements to create an effect. The Akkadian recension reads (EA 359, Rev. 17ʹ-18ʹ; cf. Westenholz 1997, 126-127; Haul 2009, 422-423 and 440-442):

[a]dīni Šarru-kēn ( lā illakannaši liklaššu! kibru mīlu šadû (ḪUR.SAG) gapšu līpušu apu qilta lišapīšu ḫubuta qalla kiṣ!ṣari

“[T]ill now Sargon has not come to us: may the riverbank, the flood, (and) the mighty mountain hold him back!

May the reed thicket make a forest, may it make appear to him a wood, a forest of knots(?)”

Despite the challenging Akkadian, the shared elements of flood, mountain, and reed thicket emerge. A tricolon in kibru mīlu šadû is observable, in addition to a parallelly formed set of statements in EA 359 rev. 8ʹ-9ʹ (Haul 2009, 264). According to Westenholz (1997, 107), “The outstanding feature of the poetic structure [of the Amarna recension] is the abundant use of parallelism. Many lines contain several synonymous-parallel clauses, usually incremental in nature”. It is much the same in the Hittite text, but the parallelisms are arranged differently. Therefore, by an analysis of the parallelism and repetition of these lines, the Hittite “freie Nacherzählung” is shown to be much more, and the programmatic remark of Francia (2012a, 82) is handily confirmed: “Gli Ittiti andarono alla ricerca di uno stile proprio, non limitandosi a riportare pedissequamente il testo di partenza, ma andando alla ricerca di espedienti stilistici tali da conferire alla traduzione una struttura poetica e un’originalità propria.

4.2.2 The Destruction of the Gatehouse

What original poetic structure was gained, however, was almost undone by the scribe towards the end of the composition. In story section (7) we read (rev. 14ʹ-22ʹ):

The content of these lines is repeated immediately after Sargon is said to acquiesce to the soldiers’ demands (rev. 27ʹ-30ʹ), only now as a narrative summary (narrative-time less than story-time) and not narrative scene (story-time and narrative-time are contemporaneous), e.g.: [GIŠḫika]r=ašta karšada nu=at!=a[pa GI]ŠBANŠURMEŠ DÙ-at ta=za=kan [LÚMEŠ U]R.SAG- adanna ti[e]r (‘He cut out the [ḫika]r-tree and made it! in[to] tables, so that upon (them) [the her]oes began to eat’). Not only was the relative clause omitted, but the scribe also used short-forms of previous words (the tree name (Güterbock 1969, 25) and Sumerogram DÙ for Hitt. iya-). The result is to have the repeated lines convey the same content as their previous iteration in a faster manner (9 lines vs. 4 lines), demonstrating narrative speed (for these narratological terms see Genette 1980 and 1988). It is also suggestive that more attention is to be paid to the direct speech version, perhaps not unsurprising in a text that prizes dialogue (as can be seen above in 4.2.1).

The parallelism of the reported version is fuller: the actions are demanded as a tricolon, each of which can be distinguished into three clauses: tree in gatehouse; cut; and fabricate. Each clause then corresponds to a parallel clause in the next iteration, thus: (A)—(D)—(H); (B)—(E)—(I); and (C)—(F+G)—(J+K).

Immediately it becomes apparent that the final two members of the tricolon are more elaborate than the first. They add a desired result to the act of fabrication. Thus, a triad of a b b emerges, a kind of reverse of the pattern adduced for the speech of Sargon/Nūr-Daḫḫi (a a b).

A rhyming pattern is also produced from the repeated use of the 3rd imperative ending -du that occurs at the end of each clause following the introductory tree-clause. And for the 1st and 2nd instances of the rhyme, the same verb is repeated: karš- (‘to cut (out)’) and iya- (‘do, make’).

The resulting effect is to depict the soldiers demanding Sargon destroy Purušḫanda in a chant-like manner, i.e., repetitively and in quasi-rhyme. Such a depiction is reminiscent of the Soldier’s Song (CTH 16; KBo 40.368), notably “with ‘young men/warriors’ attempting to ‘cut’ (karsikanzi) a mountain and singing a song” (Weeden 2013, 89).

But these considerations are only borne out of an emended text. The poetic qualities of the above lines were almost completely corrupted by a struggling scribe and their many errors (see also Gilan 2000, 56; Rieken 2001, 579-584). Each clause makes use of different clitics, or falsely attaches clitics to different words or clauses (e.g., (B), nu=war=a(t)=št[a]; and (E) nu=war=a(t)=〈šta, written nu-wa-ra-at-ta). Interestingly, the duplicate of this tablet (KBo 12.1) also preserves the same errors in its extant overlaps, suggesting another scribe struggled with their copying. The repeated relative clause uses arta (ar-, ‘to stand’) once, only to erase it twice thereafter. More than an error, the double-erasure exhibits intentional disregard for poetic structure. And the parallel exhortation to ‘cut’ (karšandu) is written only once correctly, despite occurring three times: the first iteration omits -an, and the third -du! Notwithstanding these errors, the scribe also made a perplexing choice in opting for (the still grammatically acceptable) conjunction ta in (J), when they had twice previously used nu (C + F).

In order to maintain the poetic style of these lines, the scribe had simply to write the repetitions verbatim with minor substitutions: interchange of tree-type and product made. Why he had such difficulty in doing so is a mystery.

One reason might simply be that the maintenance of a poetic structure was not the scribe’s priority. Many of these errors have been accounted for by Rieken (2001) as failures in deliberate archaization—a later scribe attempted to make their copy appear older, but had a faulty understanding of Old/Middle Hittite. Thus, in an attempt to archaise, the scribe neglected the parallelism of these lines (e.g. the conjunction ta in (J) when nu would have made better grammatical sense and preserved the parallelism; cf. CHD (L-N s.v., ‘nu A’, 468b) for ta as “properly only [conjunctive in nature in] OH”). In addition, if the scribe were unaccustomed to earlier language features, their confused use of older clitics might have a knock-on effect on the parallelism. One could also suggest that, if the scribe were still learning their craft, these skilful aspects of writing were still beyond them.

In short, there is no clear answer. But the Hittite Šar Tamḫāri is a telling reminder of the pitfalls of applying stylistic, parallelistic analyses to Hittite texts, especially when it comes to contrasts (‘foregrounding by deviation’): “In Hittite, deviation can be difficult to detect due in large part to the uncertainty of whether a perceived deviation was caused by scribal error or deliberate intention” (Marineau 2020, 44).

5. Conclusion

This short survey on the largest parallelistic and repetitive features of the Hittite Šar Tamḫāri represents the tip of what was almost certainly once an iceberg of poetic structure. It is highly plausible that the text made use of more stylistic features no longer extant. Nevertheless, the stylistics of the composition are clear.

Parallelism and repetition feature most prominently in the speeches of the Šar Tamḫāri. While this may be because the majority of the text (and Hittite literature in general; Weeden 2013, 80) is dialogue, it may also be no coincidence that the Hittite Song of Release also exhibits a high degree of poetic style in its direct speeches (Francia 2010, 65-71). Perhaps this was where the scribes working with foreign material felt most comfortable in applying such features, though this remains supposition.

Working with imported literature did not, however, limit a scribe’s poetic capabilities. Indeed, they could apply their own native stylistics. This was shown when comparing the Hittite and Akkadian versions of Šar Tamḫāri: the latter, often assumed to be the Vorlage, also exhibited parallelism, but of a different arrangement to the Hittite version. The result cannot determine which version of the text came first. But it does suggest that the Hittite text shared the same preferences for parallelism, while simultaneously freely reworking parallelism in its own way. There is, then, a shared poetic goal, only achieved via different means. This may have bearing on the possible Hittite origins of the Amarna recension (Beckman apud Westenholz 1997, 105), insofar as both attempted the same stylistics. But the differences are also in line with what one finds between the Akkadian and Hittite versions of the Annals of Ḫattušili I (Marineau 2020, 68-69), helping attest to a native poetics applied to works of ‘translation’.

Towards the text’s end, however, our understanding of the poetic structure of the Hittite Šar Tamḫāri is problematised by what seems to be a plethora of errors. While their causes remain speculative, one reason could be due to priorities: the scribe who wrote our most complete manuscript was concerned with archaising as opposed to stylistic arrangement. As such, the Hittite Šar Tamḫāri demonstrates that scribes were not solely concerned with the specific poetics of parallelism and repetition, but rather had a repertoire to draw upon. Part of that repertoire, I suspect, is archaising: the conscious use of an older register in a newer text to imbue the latter with an antique quality. Such a consideration is borne, nonetheless, from the fruitful avenue of research into parallelism and repetition.

Download here.


Bachvarova, M.R. 2016, From Hittite to Homer: The Anatolian Background of Ancient Homeric Epic. Cambridge.
Berlin, A. 2008, The Dynamics of Biblical Parallelism. Grand Rapids.
Daues, A./Rieken, E. 2018, Das persönliche Gebet bei den Hethitern: Eine textlinguistische Untersuchung. Studien zu den Boğazköy-Texten 63. Wiesbaden.
De Zorzi, N. 2019, ‘Literature as Scholarship: Some Reflections on Repetition with Variation and the Construction of Meaning in the Šamaš Hymn 112-117’, Kaskal 16, 159-182.
-2022, ‘Parallelism and Analogical Thought in Babylonian Poetry – Case Studies from Ludlul bēl nēmeqi, the Babylonian Theodicy, and the Šamaš Hymn’, Wiener Zeitschrift für die Kunde des Morgenlandes 112, 367-394.
Fabb, N. 2004, Language and Literary Structure: The Linguistic Analysis of Form in Verse and Narrative. Cambridge.
Francia, R. 2010, ‘The Poetic Style of the Direct Speeches in the Hittite ‘Parables’ of the ‘Epos der Freilassung’ in J. Klinger/E. Rieken/C. Rüster (eds.) Investigationes Anatolicae. Gedenkschrift für Erich Neu. Studien zu den Boğazköy-Texten 52. Wiesbaden, 65-72.
-2012a, ‘Storia degli Studi sulla Poesia Ittita e una Nuova Chiave di Lettura di un Testo ‘Classico’: CTH 414’, Studi Micenei ed Egeo-Anatolici: Supplemento 1, 79-106.
-2012b, ‘La Struttura ‘Poetica‘ degli Exempla Ittiti nel ‘Canto della Liberazione’, Centro Mediterraneo Preclassico 3 (3), 77-107.
-2018, ‘The Grammar of Hittite Poetry’ in E. Rieken (ed.) 100 Jahre Entzifferung des Hethitischen: Morphosyntaktische Kategorien in Sprachgeschichte und Forschung: Akten der Arbeitstagung der Indogermanischen Gesellschaftvom 21. bis 23. September 2015 in Marburg. Wiesbaden, 83-90.
Genette, G. 1980, Narrative Discourse: An Essay in Method. Ithaca.
Genette, G. 1988, Narrative Discourse Revisited. Ithaca.
Gentili, P. 2000, ‘Il Re Sognatore’ in S. Graziani (ed.0 Studi sul Vicino Oriente Antico Dedicati alla Memoria di Luigi Cagni. Istituto Universitario Orientale Dipartimento di Studi Asiatici Series Minor 61. Napoli, 355-373.
Gilan, A. 2000, Sargon in Anatolia: The “King of Battle” in a Hittite Context [MA Thesis]. Freie Universität Berlin.
-2014, ‘Epic and History in Hittite Anatolia: In Search of a Local Hero’ in D. Konstan/K.A. Raaflaub (eds.) Epic and History. New York, 61-65.
Groddek, D. 2008, Hethitische Texte in Transkription KBo 22. Dresdner Beiträge zur Hethitologie 31. Wiesbaden.
Güterbock, H.G. 1969, ‘Ein neues Bruchstück der Sargon-Erzählung ‚König der Schlacht’, Mitteilungen der Deutschen Orient Gesellschaft 101, 14-16.
Haul, M. 2009, Stele und Legende: Untersuchungen zu den keilschriflichen Erzählwerken über die Könige von Akkade, Göttinger Beiträge zum Alten Orient 4. Göttingen.
Hrozný, B. 1929, ‘L’invasion des Indo-Europeens en Asie Mineure vers 2000 av. J.-C. ’, Archiv Orientalni 1, 273-299.
Jakobson, R. 1987, Language in Literature. Cambridge.
Jeffries, L./McIntyre, D. 2010, Stylistics. Cambridge.
Marineau, R.N. 2020, The Literary Effects of Discourse Patterns in Hittite Texts [PhD Thesis]. University of Chicago.
Meriggi, P. 1973, ‘Zu einigen Stellen hethitischer historischer Texte’, in E. Neu/Ch. Rüster (eds.) Festschrift Heinrich Otten. 27. Dezember 1973. Wiesbaden, 199–208.
Rieken, E. 2001, ‘Die hethitische šar-tamḫāri-Text: archaisch oder archaisierend?’, in G. Wilhelm (ed.) Akten des IV. Internationalen Kongresses für Hethitologie. Würzburg, 4.–8. Oktober 1999. Wiesbaden, 576–585.
Mouton, A. 2007, Rêves Hittites. Contribution à une Histoire et une Anthropologie du Rêve en Anatolie Ancienne. Culture and History of the Ancient Near East 28. Leiden/Boston.
Soysal, O. 2017, ‘Sargon, Nūr-Dagal/N ve Purušanda’, in Ö. Süleyman/H. Hüryılmaz/A. Türker (eds.) Samsat’tan Acemhöyük’e eski uygarlıkların izinde: Aliye Öztan’a ar- mağan (From Samosata to Acemhöyük trailing the ancient civilizations: studies presented to honour of Aliye Öztan). İzmir, 217–229.
Torri, G./Barsacchi, F.G. 2018a, Hethitische Texte in Transkription KBo 12. Dresdner Beiträge zur Hethitologie 50. Wiesbaden.
-2018b, Hethitische Texte in Transkription KBo 13. Dresdner Beiträge zur Hethitologie 51. Wiesbaden.
Wasserman, N. 2003, Style and Form in Old-Babylonian Literary Texts. Cuneiform Monographs 27. Leiden/Boston.
Watkins, C. 1995, How to Kill a Dragon. Aspects of Indo-European Poetics. New York/Oxford.
-2010, ‘An Indo-European Stylistic Figure in Hittite: The a a b Triad and the Climatic Formula … V1 … V1 … Vj’, in J. Klinger/E. Rieken/C. Rüster (eds.) Investigationes Anatolicae. Gedenkschrift für Erich Neu. Studien zu den Boğazköy-Texten 52. Wiesbaden, 329-335.
Weeden, M. 2013, ‘Poetry and War among the Hittites’, in H. Kennedy (ed.) Warfare and Poetry in the Middle East. London, 73-98.
Westenholz, J.G. 1997, Legends of the Kings of Akkade. Mesopotamian Civilizations 7. Winona Lake.