Nicole Brisch on Paul Delnero’s The Textual Criticism of Sumerian Literature

Ancient Mesopotamia, the area that today corresponds to the modern states of Iraq and North-East Syria, has given us one of the world’s oldest corpora of literature, if not the oldest. The earliest literary texts probably date to 2600/2500 BCE, though a recent suggestion would date a first attempt at literature even earlier. Scholars still only poorly understand these earliest works of literature due in part to the highly abbreviated nature of the ancient writing system, cuneiform, invented around 3300/3200 BCE. Additional difficulties arise from the language in which (most of) these texts were written: Sumerian is a linguistic isolate that probably died out as a spoken language towards the end of the third millennium BCE. The bulk of Sumerian literary texts date to the Old Babylonian period (c. 2000-1595 BCE), well after the language’s demise as a vernacular. By this time Sumerian had taken on a second life as a written language that predominated in literary and religious texts. Among the most famous works preserved from this period are the earliest literary tales about Gilgamesh, the legendary king of Uruk. But hundreds of other compositions survived, including mythological tales, hymns in praise of kings or gods, and lamentations. Many literary compositions served as school exercises and are attested in multiple manuscripts copied during a relatively short period of time in the mid-eighteenth century BCE, and the overwhelming majority was excavated in the city of Nippur, the religious center of that period.

Anyone who has attempted to edit a work of Sumerian poetry will appreciate the difficulty that study of this literature poses. Currently there is no comprehensive dictionary of the Sumerian language, though Steve Tinney’s electronic version of the Pennsylvania Sumerian Dictionary has helped remedy this gap. The difficulties in reconstructing Sumerian grammar pose a much greater problem. We lack any cognates, and some Sumerologists are reluctant to use linguistic analyses of ergativity and other features alien to most European languages in their analysis of Sumerian grammar. As a consequence, reconstruction is particularly problematic and opinions differ greatly.

Paul Delnero’s The Textual Criticism of Sumerian Literature enters this fraught territory with its focus on the “Decad,” a group of ten literary texts that were copied very frequently in the Old Babylonian schools at Nippur early on in the curriculum. These compositions include some mythological or legendary texts such as Inana and Ebih and the tale of Gilgamesh and Huwawa; the latter is part of the Sumerian Gilgamesh cycle. The collection also contains various royal and divine hymns, a song in praise of the hoe, a hymn in praise of the goddess Inana, etc. Most of these texts were also found in other cities or are also attested in copies of unknown provenience.

Delnero rightly points to the absence of a structured methodology to assess text linguistic variations in the corpus of Sumerian literary texts and promises that the study will “present a detailed treatment of the different types of textual variants that occur in the duplicates of Sumerian literary compositions and to propose a methodology for identifying and critically evaluating these variants.” He deserves our praise and gratitude for attempting this complicated task, as he has clearly exerted an enormous amount of energy to produce this study, valuable because it distinguishes not only mechanical errors but also interpretive, local, and diachronic variants.

A close and detailed study of these variants has also yielded Delnero’s most important result, namely that no “Vorlage” or stemma of Sumerian literary texts ever existed. This is important, because it shows that Sumerian literature was much more fluid and much less fixed than often presumed (see already Michalowski), and certainly than concepts of authorship in a world with copyrights and title pages allow. This fluidity can be seen from different manuscript traditions of any given composition; in some cases lines or entire passages are added or omitted, in other cases entire compositions are known from one city but not from another. A well-known example here is the Sumerian tale of Gilgamesh and Huwawa, which is preserved in two different versions (termed Gilgamesh and Huwawa A and Gilgamesh and Huwawa B by modern scholars). These variations don’t just appear in the oldest sources; they can also be observed in much later Akkadian literary texts, even among the famous Epic of Gilgamesh in its Standard Babylonian version, which dates roughly a thousand years later than the texts studied by Delnero. Such variants were already briefly discussed by Andrew George in his monumental edition of the epic. Thus, one can say that a certain amount of fluidity is characteristic of ancient Near Eastern literatures and reflects that access to texts and the knowledge preserved therein was quite limited in a pre-modern civilization with relatively low rates of literacy.

Evaluating variants can be complicated. Delnero uses the frequency of attestation of a given variant to argue that there was a standard or model form, and uses that model to assess the correctness of a given variant. But, as J. Cale Johnson has already asked, why we should privilege one form over another and presume it is the more correct form just because it is attested more frequently? The very notion of correctness poses difficulties, especially in a language that died out as a spoken language more than four thousand years ago and has no known cognate. Although Delnero recognizes the problems surrounding a notion of correctness, he still writes: “The pedagogical objective of the curriculum implies that certain notions of grammatical and orthographic correctness were already in place at the time the compositions in this corpus were taught and that certain standards for assessing the accuracy of the copies of these texts existed.” But a later analogy illustrates the problem: prior to the eighteenth century. orthographic variation was permissible in English to a greater extent than now, and became became standardized through orthographic and grammatical prescriptiveness (see Milroy and Milroy). Orthography and grammatical correctness depend on the social and political context which may generate such prescriptions, and it has yet to be demonstrated that such prescriptions were ever made. Cuneiform writing — at all times but especially in the early periods — allowed for a certain amount of variability in orthography, and future studies should distinguish between what one may loosely call permissible varation from true variation.

The following example illustrates both the problems of using frequency of attestation and modern notions of grammatical and orthographic correctness in identifying variants. The Sumerian language frequently makes use of suffixes to express grammatical relations. An example is the enclitic copula /am/, which means “it/he/she is” and is often attached to a word for emphasis or as a simple predicate (e.g. lugal-am means “he is (the) king”). The copula /am/ is often spelled using the cuneiform sign representing the syllable /am/, but it is also frequently spelled using the cuneiform sign with the value /a/. In Delnero’s study, the spelling of the copula /am/ using the cuneiform sign /a/ is classified as a variant. This particular variant then appears in three different categories: on page 72 it is a local variant, on pages 138-39 an “idiosyncratic variant,” and on page 172 it is listed under “interpretative variants,” where Delnero states that “[v]ariation involving the defective writing of the copula [sc. /am/] as “-a” is very common in Old Babylonian literary sources.”

The method employed here leaves several questions open: Why is the abbreviated version of the copula /am/ considered to be a variant and not the norm? Which of these forms is orthographically “correct,” the longer or the shorter form? And how can the same variant be sometimes local, sometimes idiosyncratic, and sometimes interpretative? It is my sense that variations such as these should be considered “permissible” within the cuneiform writing system, simply because they occur frequently and may depend on the scribe’s mood, the availability of space on a cuneiform tablet (the cuneiform sign for /a/ is shorter and easier than that for /am/), or aesthetic considerations, among others. Classifying such a spelling as a “local variant,” for example, may simply overstate the issue and lead to skewed results.

Another key factor in the identification of grammatical variants is a thorough understanding of Sumerian grammar (and its pitfalls), and Delnero’s analysis of grammatical omissions and additions (“Interpretive Variants”) could have benefited from an immersion into the recent debates and advances in the study of Sumerian grammar. How are grammatical variants assessed if the analysis of grammar is uncertain? This uncertainty leads to an ambiguity of what constitutes a grammatical omission or addition. For example, one of the omissions Delnero lists is that of the pronominal prefix of the verb /n/, a prefix that agrees with the subject of a transitive verbal form when that form is in the perfective aspect. The use of agreement prefixes and suffixes is still not entirely understood and may also depend on the semantics of the verb, the choice of the verbal prefix, and the aspect of the verb (see, for example, Gragg and Woods). Because Delnero mixes all of the alleged omissions of the pronominal prefix /n/ together, regardless of which verbal prefix and verbal stem they were combined with, it becomes impossible to distinguish where the /n/ may have been omitted in error and where the omission is actually grammatically correct. Delnero could have considered alternative explanations on the omission and addition of the agreement prefix /n/, already in texts from the third millennium (see again Woods).

I do not wish to diminish the value of Delnero’s study, in particular his goal of applying a systematic methodology of textual criticism to one of the world’s oldest literary corpora. But the methodology used here is ultimately not convincing, because too many key factors in identifying variants are not taken into consideration, therefore putting some of the classifications on shaky ground. However, Delnero’s important contribution lies in refuting, hopefully once and for all, the ghost of the Vorlage, or stemma, in Sumerian literature, one of the world’s oldest corpus of literary and religious texts, bringing us closer to an understanding of how these texts were produced and reproduced in ancient Mesopotamia.