8 Sumerian in the Middle Assyrian Period

8.1 Introduction

Sumerologists are in the fortunate position that their research is based on a huge and long-lasting corpus of royal or votive inscriptions, lexical, literary, liturgical, legal, and administrative texts, to name just a few of the main textual categories. If we take the archaic tablets from Uruk into account, for which a Sumerian background is not more than an unsubstantiated hypothesis, we are facing roughly three millennia of Sumerian scribal lore.¹ Given this enormous timespan, which includes periods of actively compiling texts in Sumerian as well as ones in which older scribal lore was copied and transmitted, it is imperative to consider that there are not just “as many Sumerian languages as there are Sumerologists,”² but indeed there are as many Sumerian grammars, or grammatical nuances, as there are periods or places where Sumerian was transmitted. Maybe this view is too exaggerated considering that the basics of Sumerian language remained more or less the same. However, over such a vast amount of time no language stays untouched or is resistant to modifications and changes in its structure, syntax, or lexicon, let alone to influences from other languages.³ And, of course, there is the texts’ orthography that frequently may conceal grammatical details and, hence, obstructs our perception of Sumerian grammar.⁴ Often, peculiar spellings were coined “errors” or “mistakes,” but this notion should be widely abandoned. Language contact is one, but not the exclusive, motivator for such changes.⁵ In this respect, let us consider K. David Harrison’s view, when he states:

Languages are highly complex, self-organizing systems in constant flux. [...] We all participate in constant change, but no individual speaker controls the speed, trajectory, or character of change. A process of emerging complexity—not yet well understood—gives a language its constantly changing and characteristic shape.⁶

This does certainly not mean that Mesopotamian scribes, and young ones, in particular, made no mistakes; scribal errors do exist, quite frequently at times, but every peculiar spelling must be treated with utmost care.⁷ Stefan Maul argued that the scribes working on bilingual texts during the first millennium BCE were not necessarily unable to comprehend the Sumerian, but that the interpreter or commentator instead wanted to introduce a new text layer, which subsequently led to discrepancies between the Sumerian and Akkadian versions.⁸

From a grammatical point of view, the Sumerian language and its written lore are frequently treated in relative homogeneity, almost concealing the fact that grammar and lexicon may show important differences between sites and periods. Grammarians often deal with linguistic phenomena in texts that cover either long periods of time or whose manuscripts originate from various places or even different scribal milieus.⁹ Such a treatment is certainly reasonable when dealing with a language overview as, for instance, in Marie-Louise Thomsen’s Sumerian Language¹⁰ or Brahm Jagersma’s A descriptive grammar of Sumerian,¹¹ or within a greater linguistic framework.¹² There are, nonetheless, important studies as, for example, Jeremy Black’s Sumerian Grammar in Babylonian Theory,¹³ which provides a more close-up view of the Sumerian language with a focus on a specific period on the one hand and on a specific kind of dataset, namely the rather artificial framework provided by the so-called Old Babylonian Grammatical Texts (short OBGT), on the other.¹⁴

The Old Babylonian period, which is often perceived as a pristine example for the study of scribal education and transmission of knowledge, does not present a coherent treatment of the Sumerian language either. Grammatical lists, such as the Old Babylonian Grammatical Texts published in the fourth volume of the Materialien zum Sumerischen Lexikon are mere glimpses and simply treatments of single verbs or morphological elements. All in all, they do not provide a full-fledged paradigm that can be exploited in order to deal, for instance, with narratives or even the syntax of a sentence.¹⁵ Recently Paul Delnero discussed the variation in a rather coherent group of Sumerian literary texts, which was copied in the Old Babylonian period at various places.¹⁶

This paper deals with a period in which copying Sumerian scribal lore was still at its peak. Scholarly texts deriving from various sites in Babylonia reached its northern periphery. Unfortunately, exactly how texts are transmitted is often unclear. Colophons of the time offer a few clues, but it is clear that sources reached the north under very different circumstances. The title of this paper might imply a comprehensive treatment of Sumerian grammar in the Middle Assyrian period in an area north of the Mesopotamian core and therefore at its periphery. But this is certainly not attempted here because of exactly the reason stated above. Furthermore, the Sumerian found in these texts is usually the Sumerian of the sources. An exception are texts that were compiled in Assyria proper, such as a few texts praising the Assyrian king.¹⁷ The main objective of this paper, however, is to pinpoint some observations on Sumero-Akkadian bilingual texts and subsequently the relationship between the late tradition of a Sumerian source text and its Akkadian translation. The Middle Assyrian period contributes significantly to our understanding of the ancient scribal lore, which is often insufficiently preserved in the areas a majority of the compositions were imported from. The reasons for this temporal—and also geographical—limitation are mostly based on our meagre knowledge about the transmission of lexical and literary texts in the late second millennium BCE on the one hand, and the extraordinary good state of preservation of the Middle Assyrian scholarly texts on the other.

8.2 Translating Sumerian

In order to deal appropriately with translations from the late second millennium, let us first provide some general remarks on the physical appearance of bilingual texts. By the Middle Assyrian period, bilingualism fully infiltrated scholarly texts. Among the many Sumero-Akkadian texts dating to this period there are large numbers of lexical lists, which are already more or less parallel to the tradition of the respective lists in the first millennium BCE. On the other hand, we are dealing with a slightly smaller corpus of bilingual literary compositions.¹⁸ Except for lexical texts, which distribute the Sumerian and Akkadian versions in columns, bilingual texts in the Middle Assyrian period conventionally use an interlinear distribution, which means that each Sumerian line is followed by its Akkadian equivalent.¹⁹

As has been attested previously,²⁰ and in particular during the first millennium BCE, indented lines for the Akkadian equivalents are a widely absent feature in the Middle Assyrian period. In those instances in which the Sumerian version retains a rather short form, the scribes frequently saved space by putting both versions on one single line and separating Sumerian and Akkadian by a so-called Glossenkeil (e.g., , , ).²¹ The famous tablet of the “Astrolabe” B in its Middle Assyrian version KAV 218 represents a subtype of interlinear translations. Due to the tablet’s layout and the division into three columns, the scribe had to break each version several times. In order to keep the Sumerian and Akkadian versions apart, all lines except for the first are indented.²² This subtype, however, still belongs to the category of interlinear translations. Similar to lexical texts, some bilingual compositions distribute the Sumerian and Akkadian versions into separate columns with the Sumerian text on the left and its Akkadian equivalent to the right.²³ It is difficult to decide whether this kind of layout derived from the source used by the copyist, or whether it was restricted to certain genres of scholarly texts. In any case, this type of layout is rather scarce in the Middle Assyrian period outside the genre of lexical texts.

In quite a few instances, both versions demonstrate a tendency towards segmentation into smaller (syntactical) units. The separation of a line into two halves is well attested in literary sources of the first millennium BCE, but is relatively uncommon in the Middle Assyrian period. In the subsequent example attested on VAT 9710 (Lugale IX–XII, line 421), the following segmentation can be observed:

More common are segmentations in even smaller units such as in line 9 of Nin-Isina’s Journey to Nippur :²⁴

A similar kind of segmentation can, of course, occur in the aforementioned column-based bilingual texts as well. The following example comes from tablet VI of the lexical series Ana ittišu:²⁵

Sometimes a text became too long to fit one line. In such a case the scribe needed to abandon any kind of segmentation. Another noteworthy feature concerns the so-called “firing holes.” This frequently attested feature of late library texts, which is produced by punching deep holes into the clay body using a round stylus or pin, requires an in-depth study. Whereas it is perfectly feasible to interpret such holes on large tablets to reduce the strain on the tablet during the firing process, their purpose to do so on smaller or medium-sized clay tablets appears to be negligible and further explanations are possible. Very occasionally the placement of these holes appears to take the syntax of the text into account.²⁶

The relationship between the Sumerian and the Akkadian versions of a composition quite often pose a certain amount of difficulties, which was pointed out by Markham J. Geller:

Like any good translation, Akkadian translations of Sumerian literature had to be cast in idiomatic Akkadian, which often makes it difficult to match the Akkadian and Sumerian texts grammatically.²⁷

Elsewhere Geller challenges the necessity of our separate treatment of the Sumerian and Akkadian versions of a literary composition, in order not to judge the ability of the ancient translator to understand the Sumerian source correctly:

On the other hand, it is questionable whether one must translate each version of a bilingual text independently, which assumes a priori that the ancient translator has failed to grasp the sense or even spirit of the original text.²⁸

The Middle Assyrian scribes had all the necessary tools, such as lexical texts, specialized vocabularies,²⁹ and even paleographical lists³⁰ at their disposal. These reference works play a pivotal role in the transmission of scholarly texts, but whether they were used as tools for an interpreter is difficult to judge based on the textual record.³¹

In terms of their transmission in the last third of the second millennium BCE both versions should be considered as one unit.³² The Middle Assyrian scribes were mostly not concerned with translating the Sumerian of older compositions into Akkadian; they had already copied a bilingual text. However, the advantages of stand-alone translations of either the Sumerian or the Akkadian version are not always self-evident.³³ Henri Limet summarises these issues as follows:

La traduction est l’art de presenter [...] dans une langue un texte qui a été écrit ou prononcé dans une autre. On passé donc d’une langue A, dite “de depart,” le sumérien, à une langue B, dite “d’arrivée,” appelée aussi “langue cible,” l’accadien. La difficulté vient, non seulement de ce que les deux langues A et B different dans leur vocabulaire et leur grammaire, mais aussi de ce que le texte à traduire a été conçu dans une culture qui n’est pas celle de la langue cible.³⁴

Some bilingual compositions of this late stage appear to have been compiled from an Akkadian perspective. The Sumerian of such texts frequently demonstrates a great variety of unusual spellings, which frequently seem impenetrable and almost of arcane and cryptographic nature. Fluent translations of such texts seem to be impossible without the Akkadian equivalent. However, the Sumerian language received the status of a pseudo-original by placing it first.³⁵

The textual record of the Middle Assyrian period remains rather silent about the means of transmitting scholarly texts. Any information about the origin of a source can be gleaned from the colophons.³⁶ The so-called “Astrolabe” B with its sophisticated astronomical menology of the month names is available in its full form thanks to the Middle Assyrian tablet VAT 9416. Most other text witnesses and parallels date to a significantly later date and are much more fragmentary.³⁷ A textual analysis of the Middle Assyrian tablet clearly favors an earlier date. Some of its verbal forms show features of Middle Babylonian texts. As we will see further down, the Sumerian of this composition shows an array of peculiar or at least arcane spellings.³⁸

The colophons on Middle Assyrian tablets, as far as they are preserved, may provide information on the origin of a source text or the family background of the copy’s scribe,³⁹ but colophons never include information on the responsible translator of a Sumerian composition. This is mainly due to the fact that the translated source in its bilingual setting was considered as one inseparable entity.⁴⁰ A rather different case presents itself through two copies of the lexical series Ea. The colophons on Ass. 523 as well as VAT 10172 both refer to the source as being an “old A.A series” (Ass. 523: a.a^meš-tu libir.ra^meš-tu and VAT 10172: geš.gar3 a.a^meš libir.ra^meš). Indeed, both tablets contain archaizing sign forms. One can contrast this kind of lexical tradition with another tablet dated from the Middle Assyrian period which also contains a copy of the first tablet of Ea lacking any older sign forms.⁴¹ Note that both the paleographical sign list AfO 4, plates III–IV written by a certain Marduk-kābit-ahhēšu as well as the Middle Assyrian copy of the creation myth KAR 4 (see section 8.3.6) go back to “old sources.”

Some information can be gained through textual analysis, which might provide clues as to the origin of the source text used by the copyists.⁴² However, even such analyses provide mere glimpses, but fail to give the whole picture and thus many issues persist. An unfortunate fact is the lack of information regarding both the translators of Sumerian texts as well as the exact circumstances of the process of translating these texts.⁴³ The colophons are generally ignorant about these highly intriguing aspects and limit themselves to the scribes, who already had a bilingual copy at their disposal.⁴⁴ Nonetheless, it can be taken for granted that the Middle Assyrian scribes were not the translators of Sumerian compositions, neither lexical nor literary.⁴⁵

In Emar, whose texts date slightly earlier than the Middle Assyrian texts from Assur, the colophons on copies of scholarly tablets are separated by a double ruling from the actual “base text” as well. Over the double ruling the sequence be man be is written in smaller script. Yoram Cohen notes that besides Mesopotamia and Emar this notation is also known from Ugarit and Hattusa. At least for Mesopotamia or Assyria, in particular, it seems rather plausible to see in be a notation that stands for Sumerian til. This is somewhat verified by the parallel al.til also written over the double ruling on VAT 8876. man, on the other hand, still poses some problems. An interpretation of be for bēlu, “lord,” and man for šarru, “king,” appears too far-fetched. Cohen emphasizes that in the Western periphery, this notation may have lost any semantic affiliation and kept only a symbolic value.⁴⁶

A few redactional remarks such as hepi, “it is broken,” not only show that the Middle Assyrian scribes attempted to produce a faithful copy of their sources, but moreover that they did not have to bother with translating or interpreting Sumerian compositions.⁴⁷ If such remarks also occur in the Akkadian version, it is quite clear that the Middle Assyrian scribes already copied from a bilingual source. Such a source text quite certainly can be traced in a center of learning such as Nippur of the slightly earlier Middle Babylonian period. Amid the scarcity of Middle Babylonian literary sources, N 6286 is a comparatively well-preserved bilingual source of “Ninurta’s Return to Nippur,” whose layout puts the Sumerian and Akkadian versions into columns.⁴⁸ Another issue is the fact that we know almost nothing about scribal education in the Middle Assyrian period. There should have existed some means of transmitting the know-how of writing and dealing with “old” scribal lore, either affiliated to an institution⁴⁹ or within the private sphere of skilled scribes or officials in the city’s administration. Compared to the richness of sources for the Old Babylonian period⁵⁰ and the later evidence from first millennium BCE Babylonia,⁵¹ we search in vain for any meaningful information regarding this issue. There are neither archaeological traces nor any valuable hints in the texts that may help in identifying the Middle Assyrian bīt ṭuppāte as locus operandi for the education of young scribes.⁵² For the time being, any reconstruction of teaching methods must remain speculative.⁵³ Additionally, the copies of the Middle Assyrian scribes appear not to resemble school texts.⁵⁴

Yet another issue defying any easy approach is the choice of texts that have been copied in Assur. The composition “Nin-Isina’s Journey to Nippur,” for instance, is otherwise known only from a small fragment originating from Old Babylonian Nippur.⁵⁵ Was it mere coincidence that this composition found its way to Assur? In his recent overview of the lexical tradition in Mesopotamia, Niek Veldhuis noted that the Middle Assyrian corpus of lexical texts contained, besides regular school texts, also “rare archaic compilations, such as the phrasebook Ki-ulutinbiše (also known as ana ittišu), which originated in Old Babylonian Nippur.”⁵⁶

Given all those circumstances and the complex history of text transmission, which is shrouded from view by a lack of information, it is an arduous task to treat the “quality” of a language in such a late stage. When exactly the Sumerian language ceased to be used as vernacular is a cause of much debate and might have happened in various stages after the downfall of the Ur III Empire at the end of the third millennium BCE. All written scribal lore that has been compiled in the two millennia that followed may show deficiencies or peculiarities of any kind.⁵⁷ The decline of Sumerian is due further to the ever-increasing infiltration of Akkadian into the economic and daily life. However, a great share of the textual material used for grammatical observations and setting up an artificial paradigm belong to some extent to the Old Babylonian school milieu.⁵⁸ The variation between copies belonging to the same composition often allows for the identification of scribal errors, Hörfehler, and other deficiencies in copying source texts.⁵⁹

Frequently, secondary literature texts dating after the Old Babylonian period are deemed to contain mistakes or errors by the respective scribe. Just a few attempts were undertaken to interpret unusual spellings as evidence for linguistic change or variations in virtue of regional customs.⁶⁰ But one has to bear in mind that the Old Babylonian texts are, strictly speaking, not a product of speakers of the Sumerian language. Bilingual sources from the Old Babylonian scholarly sphere are relatively scarce. Bilingualism infiltrated royal inscriptions and lexical texts faster than narrative compositions. Complete interlinear translations are practically absent from the Old Babylonian text corpus. Partial Akkadian translations are usually added to Sumerian texts as annotations or glosses.⁶¹

Jerrold S. Cooper states about bilingualism and bilingual texts in the first half of the second millennium BCE:⁶²

But unlike the period after 1600 BCE, when Sumerian texts were as a rule accompanied by an Akkadian translation, in this earlier period, translations were quite rare, often from outlying areas, and by their appearance and quality betray themselves as the work of inferior scribes, either students who needed a ‘pony’ to learn Sumerian, or scribes who never learned Sumerian well enough in the first place. The rarity of these early bilinguals, compared to the thousands of unilingual Sumerian tablets of the same period, is eloquent testimony to the strength of Sumerian tradition in the Old Babylonian (2000–1600 BCE) academy.⁶³

The variation and use of local orthographical as well as grammatical features allows local traditions or even only the preference of a single scribe to be highlighted. Even while comparing the wide array of text witnesses to a given Sumerian literary text in the Old Babylonian period, the variants between the respective manuscripts may be astonishing.⁶⁴

Copies of lexical and literary texts, which came down to us from the Middle Assyrian period, offer important insights, such as possible evidence for dictation. A proper investigation of many of these features, however, is still a desideratum. There is ample evidence that Sumerian at the end of the second millennium BCE was not just widely used in the scholarly tradition, but was given a pivotal role in the scribal sphere as well. The physical appearance of the (bilingual) texts themselves provide enough hints, let alone the fact that in bilingual texts the Sumerian version generally appears first—both in interlinear as well as column-based bilingual texts—, and is followed by the Akkadian translation.⁶⁵ This feature for bilingual narrative texts might have completely derived from the lexical tradition.⁶⁶

Amid the fact that we have only small glimpses of the original textual record at our disposal, the extant texts suffice in order to get a good perception of the scribal lore that was transferred to Assur and copied there. The prominence of compositions such as “Ninurta’s Exploits” or “Ninurta’s Return to Nippur” is noteworthy. Together with the two manuscripts of “Nin-Isina’s Journey to Nippur,” it is astonishing to note that most of the Sumerian literary texts associated with the Middle Assyrian period in Assur deal (at least to some extent) with the topic of the divine journey. Whether this fact has any relevance for a kind of “program” in acquiring cuneiform sources, must remain speculative.

8.3 Text Basis

In the Middle Assyrian period, the elite⁶⁷ of Assur came into contact with a huge amount of literary and lexical texts, which originated in Babylonia. It is almost impossible to reconstruct the transmission paths of these scholarly texts. Following one possible, and not unlikely, hypothesis Babylonian scribes brought their text collections with them when moving to the Assyrian realm.⁶⁸

With the Middle Assyrian period, we enter an age of diplomacy and international relations.⁶⁹ The Assyrian “state” was increasing its power and political hold in the Fertile Crescent. According to a fragmentary passage in the “Epic of Tukultī-Ninurta I,” the Assyrian king plundered Babylonian libraries and brought their contents to Assyria.⁷⁰ There are a couple of Middle Babylonian scholarly texts among the tablets found at Assur, and those might very well have come to Assur on such an occasion.⁷¹

By looking at all the extraordinarily well-preserved Middle Assyrian manuscripts, one might wonder what happened to the sources, the Vorlagen. It is unlikely that all bilingual texts were transmitted orally. This is confirmed, on the one hand, by the great stability in compositions such as “Ninurta’s Exploits,” but also by internal remarks that imply copying from a physical source. Such remarks might occur within the copy itself.⁷² Some of the tablets add til or al.til before the colophon, thus indicating that the copy is “complete.” Also, the Sumerogram gaba.ri quite certainly refers to a physical tablet, which was used as a source text for the copyist. It is, however, not always clear whether every scribe copied from such a tablet, or whether sometimes text witnesses also went back to other forms of transmission, such as dictation. The texts themselves are usually not very explicit, but give nevertheless some small clues. The colophon of the aforementioned text witnesses for “Nin-Isina’s Journey to Nippur” refer to the source as being written ana pî ṭuppi šaṭāri, “according to the wording of the written tablet.” Amid the well-preserved corpus of Middle Assyrian copies at our disposal, the sources are gone. Were they sent back? Were the sources first copied onto perishable material or on tablets, which were then recycled? All these questions unfortunately cannot be answered. One can, however, entertain such a possibility and compare the situation in Assyria in the last third of the second millennium BCE with medieval scriptoria, where manuscripts were copied before being returned to their home institutions.⁷³

The majority of Middle Assyrian scholarly texts discussed in this study was assigned to a possible institutional library with the preliminary designation “Reconstructed Library M 2” by Olof Pedersén. The reconstructed state of this library or manuscript collection is owed to the find conditions at Assur, for the texts assigned to it were found spread over a relatively large area at the site of Assur, more precisely between the Aššur temple precinct and the temple of the gods Anu and Adad.⁷⁴ This Middle Assyrian collection is often referred to as a royal library supposedly established in the reign of Tiglath-pileser I.⁷⁵ The studies by Claudio Saporetti and Helmut Freydank about Middle Assyrian eponyms showed that an affiliation of these texts to the reign of Tiglath-pileser I appears to be unlikely.⁷⁶ Whether the tablets belonging to this reconstructed group were part of an institutional library or part of smaller manuscript collections stored in the houses of various Middle Assyrian officials can no longer be verified.⁷⁷ Be that as it may, by the Neo-Assyrian period specific tablets were selected and moved to the royal libraries in Nineveh, which were assembled by either Esarhaddon or Ashurbanipal.⁷⁸ It is, however, rather unclear why certain tablets were chosen and others not. It is therefore rather surprising that either Esarhaddon’s or Ashurbanipal’s officials chose BM 122625+ containing a copy of sections XIII–XVI of “Ninurta’s Exploits”,⁷⁹ because this copy attests to several deficiencies compared to the tradition of this literary text. The tablet’s scribe Marduk-balāssu-ēreš presumably copied the complete composition onto four large tablets, of which three exemplars survived.⁸⁰ Another example is a god list published as CT 24, 20–46 written by a certain Kidin-Sîn son of Suti’u. This god list is an exceptional case within the Middle Assyrian evidence because Kidin-Sîn copied the text “according to the wording of an old ‘big tablet’” (rev. vi, 8’: a-na pi-i dub.gal-le libir.ra).⁸¹ The same scribe also copied the creation myth KAR 4, which has not been transferred to Nineveh. However, there are copies known of this composition at Kuyunjik. Thus one can entertain the possibility that this composition was copied on clay or wax before being transferred to the capital.

The Middle Babylonian and Middle Assyrian “library” texts from Assur are often considered to be part of a royal collection established by Tiglath-pileser I.⁸² But, as was emphasized by Niek Veldhuis recently, the archaeological and textual evidence speaks against assigning this group of texts to the reign of Tiglath-pileser I.⁸³ According to Pedersén, three libraries date to the Middle Assyrian period. Besides the small library in the Old Palace (M 1) and an even smaller collection of a couple of school tablets found near the Ištar temple (M 3), the largest group of Middle Assyrian library texts has been reconstructed by Pedersén due to the “findspots and external appearance of the tablets” (M 2).⁸⁴ The archaeological context poses many difficulties. It is, for instance, not possible to affiliate the group M 2 with the Aššur temple. As indicated above, it is not entirely certain whether the texts of this group were part of an official collection or belong to several private manuscript collections.⁸⁵ The many findspots of Middle Assyrian material found together with Neo-Assyrian scholarly texts in the southwest of the Aššur temple seem to indicate that the earlier tablets were incorporated into a later (temple) library (N 1).⁸⁶

A brief survey of the manuscript collection M 2 with its affiliations to the later Neo-Assyrian group N 1 shows that it contained a large variety of scholarly texts.⁸⁷ Apart from the literary texts, the most important part comprises copies of lexical lists. An intriguing group are texts dealing with hippology. For this study, the bilingual sources are of particular interest.

The Middle Assyrian period offers one of the most pristine sources for the transmission of bilingual texts in Mesopotamia, much of which is owed to the good state of preservation of most of the tablets. Amid the creation of various Assyrian scholarly texts, the prime focus of the Assyrian kings was the south, and it is this period that presents most of the major compositions in a form that is rather reminiscent of the “standardized” editions in the first millennium BCE.⁸⁸ The elite living and working in Assur or Assyria in this period drew an enormous amount of knowledge from the south, from Babylonia.⁸⁹ Despite the propagandistic and possibly ahistorical view presented in the Tukulti-Ninurta Epic, most sources of scholarly texts might not have been brought to Assyria as war booty, but might have accompanied scribal families of Babylonian descent who settled in Assyria and brought their manuscript collections with them.

The most homogenous group of colophons can be found among the aforementioned array of tablets originating from the collection M 2 at Assur with stray finds that had been identified as belonging to the group N 1. Currently (at least) 23 tablets can be assigned to the three sons of the royal scribe Ninurta-uballissu.⁹⁰ Amid the relative abundance of sources, there is no information available on this family, which goes beyond the names, occupations, and family relations of these scribes. Ninurta-uballissu’s title “royal scribe” suggests that he held a high position within the Middle Assyrian state administration. However, there are no legal or administrative documents known so far that shed any light onto this individual. Such a lack of information regarding this family in everyday documentation is quite intriguing, but might simply be due to the incomplete dataset that is at our disposal. All the information on the individual careers of the royal scribe’s sons is also only known from the colophons of scholarly tablets. Unless all these gaps in our documentation are merely coincidence, these comparatively productive scribes certainly did not share the same destiny as the (Babylonian) scribe Marduk-nādin-ahhē, whose tempus operandi falls into the reign of the fourteenth century king Aššur-uballiṭ. This scribe, whose Babylonian origin is evident,⁹¹ moved to Assur probably shortly after the new Marduk temple was inaugurated and built a house ina ṣilli bīt Marduk.⁹² In contrast to him, the societal backgrounds of our scribes Marduk-balāssu-ēreš, Bēl-aha-iddina, and Sîn-šuma-iddina lie in the shades of time. Judging from the layout of the extant colophons, it seems likely that the last-mentioned scribe learned his skills from a different tutor.⁹³

The texts selected for this study are by no means numerous. This is due to the fact that observations aim at focusing on bilingual text sources going beyond single word-to-word equivalents. Hence, lexical lists such as Ea or Ura will not feature here.⁹⁴ Consequently, this survey deals with bilingual literary compositions and those lexical texts that contain phrases and expressions. Among the texts discussed in the subsequent sections are one lexical and four literary texts as well as a text that might be called astronomical or “technical literature.”⁹⁵ The most important lexical text for this brief survey is certainly the series Ana ittišu which was well known in the Middle Assyrian period.

Amid the rather problematic connotations of this term, the Middle Assyrian texts discussed below can be considered, in general, “canonical” compositions in the sense that their contents can sufficiently be compared to later first-millennium BCE successors in terms of lexicon, grammar, and sequence of entries or lines.⁹⁶ In this study, this term is used in quite a superficial sense; it ought not to be taken literally. In Mesopotamia, this term was frequently used for compositions, which were standardized to a high degree within the stream of tradition. This process of standardizing a composition does not necessarily imply any rigid copying of texts sign by sign. This terminology is usually applied to witnesses of compositions, which preserve the same wording. Variants are frequently attested and a relative flux in the textual integrity may be quite evident.⁹⁷ The grammatical analysis of the Sumerian language in this late stage of its transmission cannot limit itself to the dataset provided by lexical texts alone. They usually—Ana ittišu is an exception—provide not enough context for a given lexeme. Hence, bilingual narrative compositions offer crucial insights into both the use and the understanding of Sumerian grammar. One of the most important sources for such an analysis is the long composition “Ninurta’s Exploits,” also known by its incipit as Lugal-e. This literary text offers the opportunity to trace the modifications in the text from the first half of the second until the second half of the first millennia BCE. Several text witnesses dating to the Middle Assyrian period are known from finds at Assur and Nineveh. Also the much shorter composition “Ninurta’s Return to Nippur” is preserved through manuscripts from these two places. It should be noted that the text witnesses from Nineveh were moved there from Assur in the Neo-Assyrian period. An intriguing case represents the account on the divine journey of the goddess Nin-Isina to Nippur. We have here a hardly known composition, which by chance entered the scribal repertoire of two Middle Assyrian scribes. All these three compositions show some relation to the topic of Sumerian divine journeys. Whether this is a coincidence or the texts were chosen on purpose, can no longer be verified due to the gaps in the documentation. Due to the shattered archaeological context the texts were found in, the question of either private or institutional libraries arises. According to Dominique Charpin, the term “library” is frequently used inadequately, since the contents of libraries are categorised and scholarly works usually derive from purpose-driven collecting. In constrast to libraries are archives, which contain the written sources pertaining to either an individual, a group of people or an institution.⁹⁸ Charpin states that the intentional firing of a tablet in ancient times can be a criterion for a library, although this feature is not ultimate proof of its existence.⁹⁹ For the Middle Assyrian texts, which will be studied in the subsequent sections, the term manuscript collections is preferred.¹⁰⁰

There are many further texts that might awaken our curiosity and be worth studying in much greater detail. In order to keep this study within reasonable limits, only a small selection of examples has been chosen.

8.3.1 The Lexical Series Ana ittišu

The lexical series usually referred to by its Akkadian incipit Ana ittišu (henceforth Ai) or, less frequently, by its Sumerian equivalent ki-ulutin-bi-še₃¹⁰¹ is unusual compared to other members of this genre. The composition has more in common with grammatical texts, since many passages resemble paradigmatic features.¹⁰² The composition was edited by Benno Landsberger in the first volume of the Materialien zum Sumerischen Lexikon (abbrev. MSL). Landsberger only included the later sources from Assyria and was not aware of any earlier versions, such as the forerunners from Nippur dating to the Old Babylonian period (so-called Proto-Ai).¹⁰³ The Middle Assyrian scribes possibly imported the complete series from Babylonia. The colophons on VAT 9552 (= Ai III) and VAT 8875 (= Ai VI) refer to sources from Nippur. So far no traces of this list can be found among the Kassite or Middle Babylonian lexical tradition.¹⁰⁴

The Middle Assyrian text witnesses are extraordinary because of their excellent state of preservation compared to many of the later sources and can therefore be considered one of the major sources for the reconstruction of this lexical series. The subsequent discussion is based on the following two manuscripts in particular:¹⁰⁵

(1)VAT 9552 is the upper half of a copy of Ai III (ur₄ : hamāmu) written by Bēl-aha-iddina of the Ninurta-uballissu family.¹⁰⁶

(2)VAT 8875 is an almost completely preserved text witness of Ai VI (sib₂-ta : elātu). The tablet was written by Marduk-balāssu-ēreš and checked by his brother, the aforementioned Bēl-aha-iddina.¹⁰⁷

It is rather probable that the Middle Assyrian scribes of Assur imported the whole series of Ana ittišu.¹⁰⁸ Unfortunately, the colophon does not mention any scribe’s name. All other text witnesses containing copies of Ai originate from the context of the sons of the royal scribe Ninurta-uballissu.

A full discussion of grammatical features between the Sumerian and the Akkadian versions is not attempted here. The subsequent paragraphs only contain a few glimpses.

Example 1: Ai III (VAT 9552) obv. i, 10

•

buru₁₄¹⁰⁹ nu-ub-da-me-a

This example contains one of the rare attestations for a Sumerian equivalent to the Akkadian preposition lāma, “before.” Later grammatical texts such as NBGT I, 423 offer the entry nu-da : la-ma.¹¹⁰ A morphologically rather close parallel can be found in NBGT IV, 19: nu-ub-dam : la-ma.¹¹¹ The sample taken from the grammatical text, however, conceals the presence of the verbal base me, “to be,” as is clearly shown by the line in Ai. Example 2 deals with a very similar verbal chain.

Example 2: Ai III (VAT 9552) obv. i, 11

•

buru₁₄ nu-ub-da-g̃en^kin-a

In this instance, the preposition lāma is not attested in its usual prepositional use (e.g., lām ebūri; Ai III obv. i, 10), but as the conjunction “before.” Thanks to the pronunciation gloss kin, the Sumerian version clearly uses a hamṭu base. Examples 1 and 2 show that the prefix chain nu-ub-da-° alone renders the Akkadian conjunction lāma.

Example 3: Ai III (VAT 9552) obv. i, 22

•

[ur₃-r]a-˹ta˺ ka ba-ab-še₃

This entry on the Middle Assyrian tablet might contain an orthographic error. Against the usual equivalent ab for the Akkadian word aptu, “window,” this copy of Ai reads ba-ab. Since the genitive is not marked, the Sumerian version should be understood as ka-ab:ba-še₃ instead. It is noteworthy that Bēl-aha-iddina uses the same spelling in the subsequent entry: [ur₃-ra-t]a igi ba-ab-še₃ : ki.min a-na pa-ni ˹ap-ti˺. One might even entertain the possibility that the scribe confused Akkadian aptu with the close semantic term bābu, “door,” and transposed the latter onto the Sumerian version.

Example 4a: Ai III (VAT 9552) obv. ii, 38–39 // Ai VI (VAT 8875) obv. i, 26–29

•

a₂-tuku a-na i₃-g̃al₂-la¹¹³ / teš₂-a se₃-ga-bi in-ba-eš

•

a₂-˹tuku a˺-na i₃-g̃a₂-g̃a₂-a / igi-^dutu-še₃ / teš₂-a se₃-ga-bi / in-ba-e-ne

Example 4b: Ai III (VAT 9552), obv. ii, 40–41 // Ai VI (VAT 8875) obv. i, 30–33

•

˹a₂˺-tuku a-na i₃-g̃al₂-la / [te]š₂-a se₃-ga-bi in-ba-e-ne

•

a₂-tuku a-na i₃-g̃al₂-la / igi-^dutu-ka / teš₂-a se₃-ga-bi / in-ba-eš

The respective Akkadian versions do not differentiate the Sumerian verbal chains in terms of aspect. Both hamṭu in-ba-eš and marû in-ba-e-ne are rendered i-zu-zu. Whereas the former should be analyzed izūzū, the last-mentioned should be a present form izuzzū. As quoted above Ai VI contains a parallel to these entries with a couple of variants. Whereas Ai III renders Sumerian i₃-g̃al₂-la in both entries as ibbašû, Ai VI differentiates between ibbaššû for i₃-g̃a₂-g̃a₂-a and i-ba-šu-u₂ for i₃-g̃al₂-la. Noteworthy is also the inconsistency between igi-^dutu-še₃ and igi-^dutu-ka, which are both rendered ina mahar Šamaš.

More interesting is the Sumerian equivalent to the Akkadian adverb mithāriš, “each one.” All instances of Ai attest to the form teš₂-a se₃-ga-bi.¹¹⁵ Line 490 of “Ninurta’s Exploits” has the form teš₂-a-ra-ke₄.¹¹⁶ This form appears to be rather close to teš₂-ba ri-a-g̃a₂ in the Old Babylonian text. Another bilingual source dating to the Neo-Assyrian period (ms. j₁) renders teš₂-a si-ga for mithāriš instead. The Sumerian form is already attested in the Middle Assyrian period within the tradition of Ai.¹¹⁷ The form teš₂-a si-ga clearly goes back to Old Babylonian teš₂-a se₃-ga as attested in royal inscriptions.¹¹⁸

Example 5: Ai III (VAT 9552) obv. ii, 5–7

The third tablet of Ai provides two Akkadian equivalents for the Sumerian expression ku₃-im-ba: (1) butuqqû, and the loanword (2) ibissû (< i-bi₂-za), “deficiency, loss.”¹¹⁹

It is noteworthy to look on the greater context of these lines in Ai III and compare the overall sequence of entries with other text genres. The Sumerian literary composition Inana C incorporates some of these terms in line 123:

Tab. 8.1: ⁱSee Sjöberg (1975, 190–191).

It is extraordinary that the literary text, which was rather popular in the Old Babylonian period and survived through many copies, presents these terms widely in the same sequence. Precursors of the list Ai were already known in Nippur in the first half of the second millennium.

The tablet’s scribe Bēl-aha-iddina used some peculiar sign forms on his copy of Ai III, such as KU₃ (; VAT 9552 obv. ii, 20). It appears that the same sign form occurs in this scribe’s copy of “Nin-Isina’s Journey to Nippur” (VAT 9308 [text B] obv. 25).¹²⁰

Example 6: Ai VI (VAT 8875) obv. ii, 18–25

Similar to grammatical lists, this lexical series includes a couple of paradigms as well. In this example, each Sumerian entry is duplicated and translated with a form of either the verb baqāru or ragāmu. In the first pair of entries, the Sumerian form inim g̃a₂-g̃a₂, which appears to contain a reduplicated form of the verb g̃ar, is rendered by the Akkadian infinitives. In the remaining entries, the Sumerian word inim is followed by a locative postposition. The second group renders the finite verbal form either as a I/2 stem preterite or a I stem perfect. The sequence is progressing afterwards. The third group adds a nominalising morpheme °-a and the last group negates the finite verbal chain. Instead of *ša (lā) ibtaqru or *ša (lā) irtagmu, these lines render the Sumerian expressions as infinitive constructions.

Example 7: Ai VI (VAT 8875) obv. ii, 33–34

In this example the negative morpheme *nu- is added paradigmatically without influencing the subsequent syllable. This phenomenon is not completely unknown. The second tablet of Ura contains many Sumerian verbal forms. Entry 70 contains the form in-na-an-sum followed by nu-in-na-an-sum in the subsequent entry.¹²¹ Whether or not this particular orthography is influenced by the separate negative particle in the Akkadian equivalents remains uncertain. Examples such as Ai VI (VAT 8875) obv. i, 47—^gugu₃ li-bi₂-in-sum : ul iš-ru-ur—show that the phonetic adaptation of the negative modal prefixe in the Sumerian verbal chain was known.¹²²

Example 8: Ai VI (VAT 8875) obv. iv, 23–27

In this example, the Sumerian version contains the plural base of the verb du₁₁, which renders, as expected, the Akkadian verbal form iqbû, “they said.” Nonetheless, the different orthographies bi₂-in-e-eš and in-na-an-eš are noteworthy. The latter can be compared to in-na-an-ne-eš attested in lines 275 and 278 of “Inana’s Descent”¹²³ as well as in lines 241 and 243 of the “Nippur Lament.”¹²⁴

8.3.2 “Nin-Isina’s Journey to Nippur” (abbrev. NJN)

According to the colophon on the Middle Assyrian copy VAT 9304, this composition had 49 lines of text.¹²⁵ This short šir₃-nam-šub was already known through a small fragment dating to the Old Babylonian period. An almost perfectly preserved manuscript of the Sumerian text also dating to the Old Babylonian period in a London private collection has now been published.¹²⁶ The only information so far about the subsequent history of this composition comes from the aforementioned colophons on the two parallel Middle Assyrian text witnesses, which were written by the two brothers Marduk-balāssu-ēreš and Bēl-aha-iddina. Both brothers also checked each other’s copy (igi.kar₂). These colophons are comparatively precise as to the source’s provenience, which is stated to originate in a tablet in the possession of a certain Iqīša-Ninkarrak.¹²⁷ The few observations that follow are based on a composite text derived from both Middle Assyrian manuscripts.

Example 9: NJN, line 3

•

e₂-ta hul₂-la-ni nam-ta-e₃ u₄-g̃a₂-nun-na-g[in₇]

The syntax in both the Sumerian and Akkadian versions differs quite significantly. Whereas in the Sumerian text the “joy of the goddess” comes forth of the temple, it is the goddess herself who leaves the “house of her joy” in the Akkadian interlinear translation. The Akkadian form la-li-ša cannot be used as subject here. Hence, the translation renders Sumerian *e₂-hul₂-la-ni-ta.¹²⁸ Another irregularity in this line is the verbal form nam-ta-e₃ with its rendering ittaṣâ in the Akkadian version. Here, it is appropriate to have a comparative look at a couple of further examples of na-preformatives in the respective text corpus:

This line allows a comparison to Gudea Cyl. A viii, 1: gu₃-de₂-a eš₃ e₂-ninnu-ta zalag-ga nam-ta-e₃, “indeed Gudea came out again from the shrine Eninnu with a radiant face.”¹²⁹ The Akkadian rendering ittaṣâ is to be understood as separative I/2 stem, although one rather expects *luttaṣâ, “she indeed came out.” The Akkadian directional phrase ina kummiša, “in/from her cella,” renders Sumerian u₄-g̃a₂-nun-na-gin₇, “like day(light) of the/her cella.”

Example 10: NJN, line 5

•

sila-dag̃al-uru-na-ke₄ mi-ni-in-dib-be₂ uru-ne₂ mu-un-da-sa₂

The translation of this line (possibly dating to the Middle Babylonian period) renders the Sumerian finite verbal chains quite differently. While the first one corresponds to an infinitive construction (ana bâ’i), the second one is given as durative (išannan). In contrast to lexical (word-to-word) attestations, this different treatment is due to the fact that the verbal forms appear in a context.¹³⁰ Finite verbal chains of the Sumerian verb dib, “to pass, to walk along,” often contain a dimensional locative or directive infix. In Akkadian, however, the verb bâ’u is transitive.¹³¹ The verb dib occurs also on VAT 8884¹³² rev. 8–9: ˹e₂˺-šu-me-ša₄ pa-e₃ dib-dib-be₂-ke₄ : ana e₂.šu.me.ša₄ šu-pi-iš i+na ba-’-ka.¹³³

Example 11: NJN, line 6

•

g̃idlam-a-ni ur-sag̃ ^dpa-bil₂-sag̃ hi-li-a mu-un-du

This line is part of a lengthy description of a divine procession of the goddess Nin-Isina to the quay in Isin. While g̃idlam-a-ni, “his/her spouse,” clearly refers to the goddess’ spouse Pabilsag̃, the Akkadian translation misinterprets this detail by providing the genitive construction hīrat qarrādi, the “warrior’s wife.” Thus, according to the Akkadian, the goddess herself is still subject. In this context, however, we expect her spouse to be part of the procession.

8.3.3 “Ninurta’s Exploits” (lugal-e ud-me-lam₂-bi nir-g̃al₂; Lugal-e)

The composition nowadays referred to as Lugal-e is one of the most frequently copied texts in the Old Babylonian period. Far more than a hundred manuscripts are known for this period alone.¹³⁴ Manuscripts were found among the school tablets in “House F” at Nippur. In the Old Babylonian period, the texts or parts of it belonged to the curriculum of apprentice scribes. “Ninurta’s Exploits” is among fourteen literary texts that were copied by advanced scribes.¹³⁵ The attestation of “Ninurta’s Exploits” for the early second millennium is comparable to the Standard Babylonian “Gilgamesh Epic” in the first millennium BCE.

In his edition, Jan Dijk discussed the various tablet types through which the composition is preserved. The Old Babylonian period attests to a few Type I tablets, which contained the whole text of approximately 730 lines in twelve columns. Most text witness, however, encompass much shorter sections, either half or a quarter, or even a sixteenth.¹³⁶

Whereas all Old Babylonian sources of this composition were only transmitted in Sumerian, it can be assumed that the Akkadian translation goes back to the Middle Babylonian period, although secure bilingual text witnesses dating to this period are missing so far.¹³⁷

The Middle Assyrian period offers the best evidence for the bilingual text before the first millennium BCE. By this time, the composition was divided into sixteen sections, which go back to the Old Babylonian format of the im.gid₂da-tablets. The extant colophons locate the sources in Nippur. It is quite certain that the Middle Assyrian apprentice scribe Marduk-balāssu-ēreš of the Ninurta-uballissu family produced a copy of the whole composition, which was inscribed onto four large tablets. He wrote the well-preserved four-column tablet VAT 9710,¹³⁸ with sections IX–XII. He was also responsible for BM 122625+¹³⁹ containing sections XIII–XVI. This copy was selected and transferred to Nineveh in the Neo-Assyrian period. Last but not least VAT 9306¹⁴⁰ is a fragment of a four-column tablet comparable to the aforementioned two texts. Due to its paleography and the placement of sub-colophons between the sections, it is beyond any doubt that this fragment with sections I–IV was written by the same scribe as well.

The list above represents extraordinary sources for the composition “Ninurta’s Exploits.” It is not certain whether its scribe Marduk-balāssu-ēreš decided himself to combine four sections on each tablet or whether this arrangement was already present on his source. Be that as it may, the Middle Assyrian texts from Assur also attest to several tablets, which only contain one section of the text.

(1)VAT 10565¹⁴¹ is the damaged upper half of a one-column tablet, which contains the third section of “Ninurta’s Exploits.”¹⁴² Its colophon is comparable to the one on VAT 9441 + VAT 10648 + VAT 11216, an extract tablet of “Ninurta’s Return to Nippur.” Unfortunately, this type of colophon does not contain any information on the responsible scribe.¹⁴³

(2)VAT 10628¹⁴⁴ is the badly damaged lower part of a possibly one-column tablet.¹⁴⁵ The preserved lines can be assigned to section XII.

(3)VAT 10643¹⁴⁶ is just a small fragment. It probably contains the same recension as BM 122625+¹⁴⁷ cited above, because it also inserts lines 524–530 between 568 and 569. This could either mean that Marduk-balāssu-ēreš did not intentionally forget the respective lines on KAR 14, but that this discrepancy was already present already in the source he used, or BM 122625+ used VAT 10643 as its source or vice versa. Based on the distribution of text on obverse and reverse, the fragment should have contained just sections XII and XIII. It could be argued whether Bēl-aha-iddina wrote VAT 10643. If so, this case is comparable to “Nin-Isina’s Journey to Nippur.”¹⁴⁸

In his study, Stefano Seminara thoroughly discussed the bilingual version of “Ninurta’s Exploits.” Therefore, I will only highlight specific peculiarities that occur in the Middle Assyrian text witnesses.

Example 12: Lugal-e, line 97 (tablet III) (VAT 10565 obv. 15–16)

The Akkadian verb rendering the Sumerian compound bu-ug-tu—za, “to destroy, to kill,” is badly damaged.¹⁴⁹ Dijk¹⁵⁰ reads u₂-[a]b-[bit(?)]. Based on the sign remains the identification of the sign ab is rather questionable. CAD L, 106 s.v. lāsimu favors a reading u₂-ṣa[b-bit]. The sign form erin₂ appears to be much closer to what is still visible on the tablet. However, there is also a third possibility: nab. The lexical text Nabnītu E (= VII = VAT 8755)¹⁵¹ reads on rev. i, 44 bu-du-ug : 6(diš) ša mim₃-ma in a section starting with patāqu. In contrast to the manuscript of “Ninurta’s Exploits,” the lexical text preserves the older form bu-du-ug instead of bu-ug-tu. Interpreting the sign remains as nab may lead to a potential verbal form unappil, which derives from the verb napālu, “to tear down, to demolish,” and in stem II, “to turn upside down.”

Example 13: Lugal-e, line 378 (tablet IX) (VAT 9710 obv. i, 7–9)

Tab. 8.2: ⁱThe construction hur + negation was discussed in Cavigneaux and Al-Rawi (2000, 37), where the authors refer to the first lines of the Sumerian composition “Bilgames’ Death” based on manuscript M1 from Me-Turan (hur nu-mu-un-da-an-zi-zi: “A cause du calque (?) akk., nous transcrivons dans ce cas hur; pour l’hésitation entre ur₅ et mur [...]. L’akk. hur(ru), s’ils’agit d’un dérivé de √’hr ‘être en arrière’ (dans d’autres langues sémitiques des dérivés de cette racine portent aussi le sans ‘autre’) donne à penser que l’étymologie du mot est sémitique” (Cavigneaux and Al-Rawi 2000, 37, note 73). The reading of this lemma must be deduced from the Neo-Assyrian version: hu-ur. See also Falkenstein (1938, 19–20 s.v. line 7).

The greatest discrepancy between the various manuscripts is the verbal form ba-an-ta.¹⁵² The Akkadian version has urabbûšu, for which the Neo-Assyrian manuscript e₁ provides the expected verbal base tu-ud. The extant Old Babylonian text witnesses have either til (ms. O₁) or us₂ (mss. L₄ and X₄), which should be interpreted as phonetic variants (based on the readings uš₂ and uš). The Middle Assyrian base, however, defies any suitable explanation.¹⁵³

Example 14: Lugal-e, line 380 (tablet IX) (VAT 9710 obv. i, 12–3)

The Middle Assyrian text is the only text witness that provides the spelling nu-nus for munus, “woman.”¹⁵⁴ The Old Babylonian as well as first millennium sources read munus instead.¹⁵⁵ Another noteworthy variation is the verbal base tu-ud in post-Old Babylonian manuscripts. All extant text witnesses from the first half of the second millennium read du₁₁ instead. The co-occurrence of the verbal bases tu-ud and du₁₁ is attested in the “Tale about the šumunda-Grass” as well. Lines 10–13 contain the following parallelismus membrorum: an in-du₁₁ {x} ki in-tu-ud / u₂.du₆&du₆.še.sar in-ga-an-tu-u[d] / ki in-tu-ud an in-du₁₁ / u₂.du₆&du₆.še.sar in-ga-an-tu-u[d].¹⁵⁶

Example 15: Lugal-e, line 383 (tablet IX) (VAT 9710 obv. i, 18–9)

Tab. 8.3: ⁱThe composite transliteration derived from manuscripts e₁ and a₂; see Dijk (1983, II, 110).

The first millennium version appears to be closer again to the Old Babylonian text than to the Middle Assyrian recension. While both the text of the early second millennium as well as the late recension use the Emesal form of the enclitic copula °-g̃en as in ga-ša-an-g̃en, “I am the lady,”¹⁵⁷ the Middle Assyrian text contains the independent pronoun g̃a₂-e. In light of the Akkadian stative bēlēku, a copula would be more suitable.¹⁵⁸ It is not unlikely that the interpreter had issues with the spelling °-g̃en for the Emesal enclitic copula of the first and second persons.¹⁵⁹ While the use of the independent personal pronoun in place of the commonly used enclitic copula is surprising, but not inexplicable, the subsequent form dili-ša₄.ab is difficult to interpret in light of the other versions. Both the Old Babylonian and Neo-Assyrian texts read dili-mu-ne instead, which fits quite well with the Akkadian equivalent ēdiššīya, “I alone,” offered by the Middle Assyrian recension onwards. Although a satisfactory solution of this form might escape us, one can pinpoint lines 3 and 5 in “Bilgames’ Death” according to manuscript M1: ša₃-aš-ša₄.¹⁶⁰ Also the bilingual letter from Mari published by Dominique Charpin reads in obv., 22: [lugal (ša₃) aš-š]a₄ [...], whose Akkadian offers a-na lugal gi-it-ma-lim [...].¹⁶¹ The Akkadian adjective gitmālum is known as gloss and thus equivalent to aš-ša₄ in Proto-Izi I, 174. There it is preceded by dili-ni accompanied by the gloss we-di-iš-ši-šu.¹⁶² The lexical series Izi was copied by the Middle Assyrian scribes and even by a member of the Ninurta-uballissu family.¹⁶³ The close proximity of these two lexemes in a lexical text already known from the Old Babylonian period onwards might be no coincidence for the problematic form attested in the Middle Assyrian text. This, however, does not solve the sign ab. BM 122625+¹⁶⁴ obv. ii, 40–41 equates dili-a with e-diš-ši-šu.¹⁶⁵ Ana ittišu should also not be unmentioned in this respect. Its sixth tablet (VAT 8875) equates dili-ni-ni with i-di-iš-ši-šu.¹⁶⁶

Finally, da-gi₁₆-sa deserves a brief discussion. This form is already attested in the Old Babylonian manuscript L₄ and represents most likely a hybrid spelling.¹⁶⁷ It is a mixture of da-ri₂ derived from Akkadian dārû, and gi₁₆-sa, the Sumerian term for “eternal.”¹⁶⁸

Example 16: Lugal-e, line 419 (tablet X) (VAT 9710 obv. ii, 3–4)

Frequently, Sumerian verbal chains attested in later periods contain a hiatus as, for instance, in the given example between ga and en.¹⁶⁹ A similar phenomenon occurs in line 4 of the creation myth KAR 4: [...] mu-un-gi-na-eš-a-ba or in Ai VI rev. ii, 33–34, which reads na₄-kišib mu-sar-ra-ne-ne ib₂-ra-ra-eš. Stefano Seminara correctly points out that “[q]ueste grafie denunciano la natura artificiale della lingua sumerica della recensione bilingue” and “presenta un’insolita grafia franta, forse esito di un eccesso di scrittura analitica”.¹⁷⁰

Example 17: Lugal-e, line 420 (tablet X) (VAT 9710 obv. ii, 5–6)

The Sumerian expression šu-siki-mu-še₃ is only clear by checking the Old Babylonian text witnesses, which have šu-se₃-ke-mu-še₃ instead.¹⁷¹ Thus šu-siki, “hairy hand,” is certainly a phonetic variant. Whether such a variant was caused by either a memory error or by dictation is difficult to answer.¹⁷² The Middle Assyrian recension of “Ninurta’s Exploits” does not offer an abundance of such phonetic variants, but the rather technical text of the “Astrolabe” B does offer quite a few.¹⁷³

Example 18: Lugal-e, line 424 (tablet X) (VAT 9710 obv. ii, 13–14)

As was pointed out by Seminara, there is a lexical variation between ug₂-gal (O₁) or pirig̃-banda (S₁) in the Old Babylonian period and u₄-gal in the Middle Assyrian text.¹⁷⁴ The Akkadian interpreter understood u₄-gal literally and rendered it ūmu rabû, “the great storm.” The same phenomenon occurs in the subsequent expression a₂-kal-ga-bi-še₃, which re-interprets the Old Babylonian form usu(a₂.kal)-bi-ta. This does not mean that the elements of the sign group a₂.kal are always treated individually in the Akkadian translation. Already in the subsequent line usu is rendered with Akkadian emūqu. Splitting Diri compounds into their elements and interpreting them is not uncommon in cuneiform sources. It is a particular feature of late commentaries and can be compared to the hermeneutic method of etymography.¹⁷⁵

Example 19: Lugal-e, line 506 (tablet XI) (VAT 9710 rev. i, 33’)

The Old Babylonian text differentiates in all available manuscripts between the Emesal form mar-za, “rites,” and me-zu, “your me (cult ordinances).” Both concepts are merged together in all later recensions of this line. Instead of the Emesal form, both occurrences use g̃arza. It is quite likely that in the second instance g̃arza can be considered a phonetic misinterpretation of me-zu, since the signs me and bar are paleographically quite similar.¹⁷⁶ This discrepancy is indicated by the Akkadian equivalent lū parṣūka, “they may be your rites,” which is not substantiated by the Sumerian version.

Example 20: Lugal-e, line 541 (tablet XII) (VAT 10628 obv. 2–3)

Tab. 8.4: ⁱ Ms. H₂ has šagina-me-en instead and reads afterwards gu₃-g̃ar-ra-gin₇; for the score, see Dijk (1983, II, 147).ⁱⁱAccording to van Dijk (1983, II, 147), there are remains of the sign bu at the beginning of the line. However, there is not enough space for p[u-uh₂-ru-u]m-ma. Even for a shorter spelling (compare the Old Babylonian pu-uh-ru₂ in ms. H₂) space is limited. The lexical list Proto-Izi II, 142 has me-lam₂ with the gloss pu-uh₂-ru; see Civil et al. (1971, 45). The available space in VAT 10628 would be enough for [me-la]m₂-ma, but this is not a common equivalent of Akkadian puhru. Compare for this sign sequence, though in another context, also the hymn Šulgi D, line 388: ni₂ me-lam₂-ma gu₂ hu-mu-ni-us₂, “May you lift (your) head with a terrifying splendour” Klein (1981, 88–89).ⁱⁱⁱ There doesn’t appear to be enough space to fit a second person °-g̃ar-re-en-na-gin₇ as expected by the Akkadian translation.^iv See for this reconstruction the Neo-Assyrian manuscript z₁ and compare lines 419–422 of “Ninurta’s Exploits”; see the score in van Dijk (1983, II, 119–120). Line 422 equates im-hu-luh-ha-en-na-gin₇ with ki-i tu-gal-li-ta-ni.

The Middle Assyrian version (collated from the original) offers a couple of variants. The title šagina(gir₃.nita) appears to be missing. Instead, it is plausible to assume that the Sumerian line attests to the compound verb gu₂—g̃ar, “to submit.”¹⁷⁷ Thus a reconstruction of the Akkadian verb kanāšu, “to submit,” a known equivalent of the aforementioned compound verb,¹⁷⁸ appears to be suitable.

Example 21: Lugal-e, line 545 (tablet XII) (VAT 10628 rev. 1–2)

Tab. 8.5: ⁱ This part of the line differs in every single manuscript. The quoted version is attested in manuscript H₂. J₂ has ki-šu₂-[u]b-e, which is more revealing in light of the Middle Assyrian text; A₄ has ki-šu₂-šu₂-da.

Unfortunately, the only sufficiently preserved later version does not provide any clues toward an understanding of the modifications that took place in the late second millennium BCE.¹⁷⁹ Line 161 of “Ninurta’s Return to Nippur” is similar in content but does not show significant variation between its Old Babylonian and Middle Assyrian recensions.¹⁸⁰ In this line, the verbal compound giri₁₇ ki-su-ub ha-ma-ab-ak-ke₄-e-ne is translated as liš-ki-nu-u₂-ni, “they may prostrate themselves,” in its Akkadian version. The quoted line from “Ninurta’s Exploits” uses the Akkadian verb šukênu as well, but deviates greatly in the Sumerian line. The enigmatic ki-ag̃₂ (badly damaged but visible) appears to have somehow slipped into this line. It is noteworthy as well that the scribe wrote su-up-pa-ni instead of su-ub-ba-ni.

Example 22: Lugal-e, line 675 (tablet XV) (BM 122625+ rev. i, 6’-7’)

Based on the verbal base e₃ in the Middle Assyrian recension, the Akkadian should have some form of the verb (w)aṣû, “to come out (etc.).” The interpreter, however, translates the Sumerian verbal form with itammû, “they utter.”¹⁸¹ The use of this particular verbal base might have been caused by the preceding line, which renders he₂-en-na-˹e₃˺ with Akkadian liš-ta-p[i-šu(?)]. There, the verbal base fits the context.

Example 23: Lugal-e, line 724 (tablet XVI) (BM 122625+ rev. ii, 13’-4’)

In this final example, the unusual spelling si—si for the compound verb si—sa₂ should be highlighted. Although this phenomenon has been discussed elsewhere,¹⁸² it should be emphasized here again that the same spelling occurs in another text copied by the same scribe. This text, “Nin-Isina’s Journey to Nippur,” renders si mi-ni-ib₂-si with the Akkadian verbal form uš-te-šir₃.

8.3.4 “Ninurta’s Return to Nippur” (an-gin₇ dim₂-ma; Angim)

The composition known as “Ninurta’s Return to Nippur” was the second major Sumerian literary text about the deeds of the warrior god. Unfortunately its state of preservation in the Middle Assyrian period is poor compared to “Ninurta’s Exploits.” So far, it is available through three text witnesses:¹⁸³

It is not unreasonable to assume that the last-mentioned two tablets were used by Marduk-balāssu-ēreš to produce his copy of the complete text.¹⁸⁴ Unfortunately, the poor state of preservation of his copy does not allow for clear answers. The last lines, however, run parallel on both text witnesses including the omission of line 202 compared to the Old Babylonian and Neo-Assyrian evidence.

Jerrold S. Cooper collected the “errors” or variants in the transmission of this text. His classification includes (1) perceptual errors, (2) unmotivated alteration, and (3) motivated alteration.¹⁸⁵ The Akkadian interlinear translation was affected by variants as well.¹⁸⁶

Example 24: Angim, line 162 (VAT 8884 obv. 18–19)

This is one of the few lines of Angim that allows for a diachronic overview of the Old Babylonian and Neo-Assyrian text layers. The example clearly shows that the Middle Assyrian recension of this line is a kind of mixture between the Old Babylonian sources and the later tradition that followed in the first millennium BCE. The Sumerian expression ug.ug-ta is rendered with ina emūqīšu, “in his strength.” Whereas ug is used for lābu, “lion,” in this line as well, both the Old Babylonian and Neo-Assyrian versions have pirig̃ instead. Sumerian ug for Akkadian emūqu is found again in line 164.¹⁸⁷ According to Jerrold S. Cooper, the orthography ug.ug is erroneous and was caused by the similar paleography of the signs pirig and ug in Babylonia.¹⁸⁸ It is noteworthy to find the same phenomenon in line 4 of the composition, which is preserved in Assur through manuscript aA copied by Marduk-balāssu-ēreš. This material might be taken as a hint that either all Middle Assyrian manuscripts of “Ninurta’s Return to Nippur” derive from the same source text(s) or one version was copied from an already existent copy in Assur.

Example 25: Angim, line 83 (VAT 9441+ obv. 27’)

The sign remains at the beginning of this line are inconclusive, but the given space does not necessarily support either [^g̃ešgigir-z]u or [[...]-z]a. It seems that the possessive suffix has been moved toward the end of the line in the Middle Assyrian version.

The joined tablet VAT 9441+ allows for a diachronic examination of this line. The Middle Assyrian version appears to be closer to the Old Babylonian. The Neo-Assyrian text re-interprets gu₃-du₁₀, “pleasant voice,” and replaces it by gu₃-de₂, “call” (Akk. rigmu). Unfortunately, the Akkadian version on the Middle Assyrian text is almost completely broken off. In line 428 of “Ninurta’s Exploits,” the Middle Assyrian version renders gu₃-de₂-zu as ši-si-it-ka (= VAT 9710 obv. ii, 21–22). Nevertheless, there are a couple of other possibilities for the Middle Assyrian text. VAT 8884¹⁸⁹ rev. 17’) reads gu₃-du₁₀ or inim-du₁₀ in line 200. Since the score in Cooper¹⁹⁰ is misleading, it is given here again without incorporating the different variants in the Old Babylonian and Neo-Assyrian manuscripts:

In contrast to line 83 cited above, the Neo-Assyrian recension stands much closer to the Old Babylonian text and it is the Middle Assyrian version that deviates quite substantially. Here, the Middle Babylonian editor probably took the imperative du₁₁-[mu-un-na-ab] already present in line 188 of the Old Babylonian text. The first millennium BCE version, however, has a finite verbal form in the Sumerian line (mu-un-na-ab-be₂), which is translated as preterite (iq₂-bi-ši) both in the Middle Assyrian and first millennium sources. The beginning of this line, again, offers intriguing variation, which pertains to the Akkadian interpretation. Whereas the Middle Assyrian line renders the logogram group ka.hi as ṭēmu,¹⁹¹ the Neo-Assyrian version translates its constituents separately: amātu ṭābtu.¹⁹²

8.3.5 “Astrolabe” B

The Middle Assyrian scribes of Assur did not indulge in the mere copying of the Sumerian literature that was passed down to them from Babylonia. Among the texts that came to Assur from Babylonian sites was also quite technical literature, such as the so-called “Astrolabe B.” As was pointed out elsewhere, this designation is of course a misnomer, since the relevant texts do not represent measuring tools for the rising of stars.¹⁹³ Apart from the two known examples of circular Astrolabe-texts, this kind of scholarly literature is usually treated in lists or tables.¹⁹⁴ The text of Middle Assyrian “Astrolabe B” has forerunners and several successors in the first millennium BCE. Although the origins of this text might reach far back, it appears that the various parts of this text were put to writing not earlier than the Middle Babylonian period. The Kassite dynasty in Babylonia was the driving force for many aspects of scholasticism, and scientific thought such as it appears in texts like “Astrolabe B” demonstrate this quite well. The Middle Assyrian copy was written by Marduk-balāssu-ēreš and checked by his brother Bēl-aha-iddina, both members of the Ninurta-uballissu family.¹⁹⁵ As was pointed out by Wayne Horowitz, who recently collected all related material and presented it in a thorough study, the Middle Babylonian tablet containing a short version of the Sumerian text represents one of the precursors of this tradition.¹⁹⁶

Since this composition is now available in an up-to-date study by Horowitz, this short treatment will limit itself to a few passages.¹⁹⁷ KAV 218 contains a couple of forms that are clearly Middle Babylonian. It is quite likely that Marduk-balāssu-ēreš had already a bilingual source at his disposal, which also included the other parts as the star catalogue.

Among its four sections only the first part is of interest here. It contains a bilingual menology for the twelve month names of the Babylonian calendar. The relationship between the Sumerian and the Akkadian shows many peculiarities that should be highlighted here in greater detail. As was discussed in section 8.2 above, due to its tabular format, the “Astrolabe” can be considered a sub-type of interlinear translations. Regarding its Sumerian version, the main phenomenon we will encounter in this text is the case that finite verbal forms in the Akkadian “translation” are frequently represented by bare verbal bases or infinite verbal constructions in the Sumerian version. Another phenomenon, orthographical in nature, quite frequently uses rarely attested readings in the Sumerian text, which could either be interpreted as auditory or memory errors, or erudite or arcane ways of writing Sumerian. The following observations concentrate on the better preserved menologies in the first two columns of the tablet. In order to properly discuss the text, each menology is fully transliterated with indication of the respective line number on KAV 218. Variants in other manuscripts are given in the discussion. The versions are provided, against the original, in columns.

Nisannu (I)

Tab. 8.6: ⁱ Note that in line 43 the scribe wrote the verbal form syllabically (iš-ša-ka-an) instead of using the mixed orthography (gar-an (line i, 8). The parallel in Sm 755 reads iš-šak-kan.

In the treatment of the first month, the Akkadian compiler interpreted Sumerian bara₂ in two ways: In the first instance it is translated with Akkadian šubtu, “dwelling.” The other occurrence offers the equivalent šarru, “king.” Although none of these equivalents take the primary semantic meaning of bara₂, “dais,” into account, the expression bara₂ il₂-la bara₂ g̃ar-ra could easily be understood literally.¹⁹⁸ There are a few instances in Sumerian literature that support the meaning “ruler,” therefore taking the dais as symbol for the king.¹⁹⁹

After collation the first sign in the subsequent line is certainly sur instead of gar. Thus, Akkadian šurrû appears to be a loanword of the Sumerian form sur-ra-an.²⁰⁰ The whole expression this term appears in is omitted both in the earlier version VS 24, 120 as well as in later related texts. It is, however, included in the Neo-Assyrian copy Sm 755.²⁰¹ But it is safe to say that this entry was already available in the source Marduk-balāssu-ēreš used for his copy. Was the inclusion of this entry triggered by the phonetic similarity between vsarru and šurrû?

Ayyaru (II)

The Sumerian phrase gu₄ si-sa₂-e-ne is rendered by the Akkadian alpū ulteššerū, which is one of the few characteristic Middle Babylonian forms in this text. Compare this to line 11 of “Nin-Isina’s Journey to Nippur,” which equates si mi-ni-ib₂-si with Akkadian ušteššer.²⁰² The subsequent phrase in the Sumerian version is ki-dur₅ gal tak₄-tak₄, which is rendered ruṭubtu uptattâ in the Akkadian version. Lexical attestations of ruṭubtu, “wet land,” are practically non-existent. This lexeme may be connected to ruṭibtu, “flooded ground,” whose equation after the lexical series Igiduh I, 295 is ki-dur₅.²⁰³ Sumerian gal clearly is a phonetic variant of g̃al₂ and thus belongs to the Sumerian compound verb g̃al₂—tak₄, “to open.” On the Middle Babylonian fragment VS 24, 120 we read in obv. 3 [...] ki-dur₅ ˹g̃al₂ tak₄˺-tak₄.²⁰⁴ The final difficult phrase is the Sumerian ^g̃ešapin dur-dur-ru-ke₄, which is equated in the Akkadian text with epinnū irrahhaṣū, “the ploughs are devastated.” It should be noted that “Astrolabe” B is the only lexical occurrence for dur = rahāṣu, so far. Nevertheless, the lexical series Antag̃al attests to the equation dur₂-dur₂-ru : min(rahāṣu) ša ašābi.²⁰⁵ The third tablet of the lexical series šarru²⁰⁶ has on CT 18, plate 29–30 (K.2054)²⁰⁷ from the Kuyunjik collection the equation usan₂+kak () : rahāṣ ūme (rev. ii, 20). A glance onto the ligature in this list from Ashurbanipal’s library reveals that the latter part equals the sign dur (). Horowitz assumes a semantic link between the two verbs rahāṣu A, “to trample, to destroy,” and rahāṣu B, “to wash, to bathe,” since the cause of destructions of the former is the weather god Adad.²⁰⁸

Simānu (III)

In the menology of the third month, the phrase lugal u₂-šub sig₄-ke₄ is difficult to interpret. The Akkadian version has šarru nalbāna ilabbin. Usually, sig₄ alone should not mean “to make bricks.” We would expect here additionally the base du₈, which is well attested in lexical texts.²⁰⁹ Syntactically this phrase runs parallel to the preceding one: iti u₂-šub-lugal-ke₄. For a lexical attestation see, for instance, Proto-Izi I, 263: sig₄-du₈^la-ba-a-nu.²¹⁰

Du’ūzu (IV)

In this menology, one can highlight a couple of intriguing orthographical spellings. The conjunction °-bi-da is written °-bi-id-da in line i, 40.²¹¹ In line 43, the scribe uses kid₂-kid₂ to render Akkadian šisītu, “cry.” The sign kid₂ or tak₄ cannot be traced per se in lexical lists neither as equivalent for šisītu nor the infinitive šasû. Nonetheless, the lexical series Diri attests in I, 231 to a logogram group gada.tak₄.si with the equivalent šisītu.²¹² It might not be too far-fetched to propose that kid₂ is actually a phonetic reflection on Sumerian gu₃-de₂, the common equivalent to šasû and its derivatives.²¹³ In this line, the Akkadian interpreter did not take iti into account. Was is forgotten in virtue of the phonetic similarity to it in ši-si-it?

Abu (V)

Tab. 8.7: ⁱ Wayne Horowitz reads [ki-ne sar-sa]r-re-ne; see Horwitz (2014, 68).ⁱⁱ Both the reconstruction of the verbal form as well as the verb base itself are uncertain. For a bilingual attestation of gi-izi-la₂—gur₃, “to carry a torch,” see R IV p. 26, no. 3, 41–42: [gi]-izi-la₂ gur₃-ru gi₆-gi₆-ga zalag₂-ga-ab : [nāš] di-pa-r[i] mu-nam-mir ek-le-ti; cited after CAD D, 156 s.v. dipāru. Alternatively, we could also expect the verbal base il₂ and possibly a form il₂-la-de₃. Wayne Horowitz reads only the nominal part; see Horowitz (2014, 68).ⁱⁱⁱ Although there is plenty of space available in this line, the scribe wrote (or copied) just three wedges over each other in order to indicate the numeral “9.” This is a common administrative practice. Its Akkadian representative tušu’û is one of the rare syllabic spellings of this numeral.

The month name Abu contains a passage that is quite clear in its Akkadian version, but less so in the Sumerian “source.” The Akkadian has Girra ištu samê urradam-ma itti Šamaš išannan, “Girra descends from heaven and rivals Šamaš.” In the Sumerian version, this passage goes as follows: ^dkaxne am-ta-e₁₁-de₃ / ki-^dutu-ra nim-nim-mu-de₃. For the Akkadian verb šanānu, we would expect a form containing the base sa₂ in the Sumerian text.²¹⁴ In line i:27 is rendered as stative šanin in its translation.²¹⁵

The base nim is problematic. This logogram is also attested in the fourth month (line i, 42), where it is equated with Akkadian harpu, “early” (line i, 48). An alternative reading in the menology of the month Abu could be tum₄. This reading may solve the vowel harmony: tum₄-tum₄-mu-de₃. Also compare it to line 143 of “Ninurta’s Return to Nippur,” which has in its Old Babylonian text the following: e₂ ki-bal tum₄-tum₄ gešbu(geš.ru) ^kušguru₂₁(e.ib₂.ur₃)-g̃u₁₀ mu-da-an-g̃al₂-[la-am₃], “I bear those which carry off the temples of rebellious lands, my throwing stick and shield.”²¹⁶

Tešrītu (VII)

Tab. 8.8: ⁱ See, for instance, Diri Nippur ix, 18: [x-d]u-ul : mu.bu : ni-ru-um; see Civil, Farber and Kennedy (2004, 32).ⁱⁱ The sign ba should indeed be read here with unvoiced consonant. The manuscript Sm 755+ (Çağirgan (1985, text B); see the photo in the CDLI database, no. P426447) reads in obv. ii, 15: šu-pa-tu. Unfortunately, its Sumerian pendant is not preserved.

The Akkadian construction šu-pa₂-a-tu is problematic. In the previous edition of the text by G. Çağirgan, the signs after the divine name in the Sumerian line are read gup₂(li)-pa šu-nir-ne-ne, which led to the translation “shrines are purified.”²¹⁷ The sign li certainly needs to be connected with the divine name Enlil.²¹⁸ The Neo-Assyrian manuscript Sm 755+²¹⁹ has in obv. ii, 10 [...mul mu.b]u-keš₂-˹da˺ ^d+en-lil₂-la₂. The remaining part could be read pa/g̃idri²²⁰ šu-nir-ne-ne, “sceptre (and) emblems.” In light of the Neo-Assyrian version Sm 755+ (šu-pa-tu), it is not unlikely to interpret it as stem III of (w)apû, “to make visible.”²²¹ This interpretation would not literally translate the Sumerian term, but provide a descriptive equivalent: “the visible ones.”²²² For favoring šubtu, “dwelling, shrine,” one should look at the ṣâtu commentary CT 41, 42 (CCP 3.4.9.M) line 3, which reads: šub-tu₄ šu.nir : šu.nir : kak-ku : min : mun-dah-ṣ[u].²²³ Horowitz refers the lexical list Antag̃al, where the equation šu.nir = šurinnu occurs within a group of designations for shrines.²²⁴

The Akkadian verbs ūtallalā and ūtabbabū are parallel. The first Sumerian equivalent ku₃-ku₃-mu seems to have its conjugation prefix in suffix position, which normally indicates an imperative or a defective writing for a “pronominal conjugation.”²²⁵

Last but not least the Sumerian phrase ka₂ abzu(-)ta(-)e₃ is worth a remark.²²⁶ This solution is closer to the Middle Assyrian text, which has kispu in the Akkadian translation. The Akkadian interpreter approached this expression and translated bāb apsî ippattê,²²⁷ “the gate of the Apsû is opened.” The Sumerian base e₃ is not a common equivalent of Akkadian petû.²²⁸ A similar case is found in the menology of the fourth month. There, the Sumerian phrase numun nim ta-e₃-de₃ is translated numun-ni har-pi šu-ṣi-i. The base e₃ appears to be merged with the dimensional marker.

Arahsamnu (VIII)

Tab. 8.9: ⁱ The use of the relatively complex sign dar₃ instead of simple tar is noteworthy. Sm 755+ has indeed pa-ṭar in obv. ii, 22.

In line 41 the scribe deliberately wrote ^gešmah () instead of the expected ^gešapin (). Both signs share a couple of graphical similarities but, one wonders whether the sign mah was mistakenly copied for *apin.na (see line 37) or the scribe considered the reading /al₆/ of the sign mah and therefore anticipated the subsequent ^gešal-la.

Quite intriguing is the Sumerian phrase a-da-min₃ di₅(ri)-di₅-de₃, which was interpreted in the Akkadian version as ultēṣû, a Middle Babylonian form of šutēṣû, “to quarrel.” According to the lexical attestation in Nabnītu M (= XXVII), line 269 the Sumerian should be a-da-min₃ di instead.²²⁹ This orthography is not uncommon. See, for instance, the initial line of the Old Babylonian “Tale about the šumunda-Grass”: ab-ba na mu-un-de. The verbal form clearly needs to be connected with na de₅, “to instruct.”²³⁰

Finally, Sumerian ur (line 39) represents in light of its Akkadian equivalent erēšu, “to cultivate” in line 43 certainly a peculiar writing. Similar to di₅ for di it appears to be another phonetic variant, in this case for uru₄(apin).²³¹

8.3.6 The Creation Myth KAR 4

This text was last discussed and edited together with the Old Babylonian unilingual fragment IB 591²³² by Wilfred Lambert.²³³ Most of this composition is known thanks to the Middle Assyrian tablet, which itself already constituted a copy of an imported source. The text received much attention in the past, not to mention its side-by-side presentation with the Silbenalphabet A.²³⁴ The textual record shows that this list of syllables goes beyond being a conventional learning tool and lexical text. The Kuyunjik text refers to the combined copy of the Silbenalphabet and the creation myth as “second tablet” (dub 2(diš).kam₂.ma) of a series, whose incipit is given as me me [kur₂]-˹kur₂˺ i-li. In the colophon, this information is preceded by a catch-line referring to the Atra-hasis epic.²³⁵ The serialization of this creation myth in a larger context might have been a first millennium invention. Lambert sees the Silbenalphabet as the first composition in the series and the creation myth as the second, since the former is also attested in its own accord. But the Kuyunjik tablet clearly puts the Silbenalphabet and the creation myth side by side, as does the Middle Assyrian text. Be that as it may, KAR 4 represents an intriguing text within the corpus of the Middle Assyrian scribal lore. It was copied by a young scribe called Kidin-Sîn, son of Suti’u. There is only one other text known from this scribe, a copy of the god list AN : Anu on a large tablet with twelve columns text, which found its way into the royal libraries of Nineveh.²³⁶

It remains uncertain whether at this occasion also the text of the creation myth was copied from the Middle Assyrian source and brought to Nineveh. Instead of the broken areas designated as such on the Middle Assyrian tablet, the Kuyunjik has, besides the entries of the Silbenalphabet, unintelligible traces of the Sumerian and Akkadian versions.²³⁷ Unfortunately, Kidin-Sîn’s colophons do not insert a date, as is known from a couple of other scribes in library M 2 such as the aforementioned Marduk-balāssu-ēreš and Bēl-aha-iddina from the Ninurta-uballissu-family. Therefore, his copies cannot be placed within a chronological framework. Whereas on KAR 4, Kidin-Sîn serves in the rank of ^lu2dub.sar tur (Akkadian ṭupšarru ṣehru), “young scribe,” he is a.ba on the copy of the god list. This writing is a comparatively rare Sumerogram for ṭupšarru in the Middle Assyrian period.²³⁸ Based on this difference in the given occupations, the god list should date later when Kidin-Sîn was farther advanced in his career.²³⁹

The colophon categorises this composition as “secret lore” (ad.hal, pirištu) followed by the expression mūdû mūdâ lukallim, “may the knowledgeable show (it) to the knowledgeable.” In the Middle Assyrian textual record this expression is rather unique. Possibly its scribe Kidin-Sîn took this expression over from the tablet he copied from, which probably originated from Middle Babylonian Nippur. The well-preserved tablet CBS 6060 containing an interesting collection of correspondences between objects such as trees, plants, or animals and deities²⁴⁰ bears a colophon, which includes the same formula as well: zu-u₂ {A} zu-a li-˹kal˺-lim.²⁴¹ This expression is followed by the wish that “the ignorant must not see (it)” (nu zu-u₂ nu igi-mar).²⁴²

However, it is less the implications of secret lore or esoteric knowledge that shall be highlighted here, but rather the bilingual tradition of this text. Karl Hecker notes that “der Text war schon in der Antike stark verderbt überliefert und ist daher nicht überall sicher verständlich.”²⁴³ Among the instances of bilingual texts in the Middle Assyrian period KAR 4 is one of the rare cases in which the Sumerian and Akkadian versions are written in separate columns.²⁴⁴ Whether this arrangement was caused by the presence of the Silbenalphabet A is uncertain.²⁴⁵ A new hand copy of KAR 4 is presented on p. , below.

Example 26: KAR 4 obv. 10–11

The Akkadian translation can partly be reconstructed from the Kuyunjik tablet. Although this text offers the reading du₅-ru for the sign ku in the Middle Assyrian copy, KAR 4 seems to use the noun ki-tuš(ku) here as the verbal base. This becomes clear from the equivalent ūšibū, “they sat down,” in the subsequent line.²⁴⁶ On rev. 18 of KAR 4 we read as follows:

Similar to ki-tuš above, the reflexive pronoun ni₂-zu is inserted within the verbal chain. The translation however contains the suitable—though without a corresponding personal suffix—expression ina ramānīšunu.

Example 27: KAR 4 obv. 16–17 and 19–20

In both instances, the Sumerian and Akkadian versions show a clear discrepancy between the verbal forms. Whereas the Sumerian text uses the suffix for the second person plural, the Akkadian translation indicates a first person plural: i nīpuš and i nītepuš or i nibni. The Kuyunjik text differs quite substantially from the Middle Assyrian recension. In lines 16–17 it uses the verbal prefix ga-ab-° for the cohortative while keeping the °-en-ze₂-en as suffix. The verbal base bal with a corresponding Akkadian verb epēšu in the Middle Assyrian version is noteworthy. The Neo-Assyrian text, however, uses the verbal base du₃ in line 16 instead, but keeps bal in line 19. The interlinear translation of this later recension, nonetheless, has the expected i nuš<bal>kit. Also line 20 differs quite substantially from the Middle Assyrian text. Here, the Sumerian verbal form reads mu-un-me-e-e-ze₂-en. Its scribe clearly interpreted the verbal base as du₁₁ with its marû stem e, “to speak.” This explains why instead of i nibni the Akkadian translation in the Neo-Assyrian text has i nibbi.²⁴⁷ Similar discrepancies of the distribution of suffixes occur in this composition elsewhere. In line 25 of KAR 4, for instance, the Sumerian verbal chain im-ma-an-tag-en-ze₂-en is equated with i ni-iṭ-bu-ha in the corresponding Akkadian text.²⁴⁸

Example 28: KAR 4 obv. 21

Here, the Middle Assyrian text clearly uses the base sur for the Akkadian verb izuzzu. Jan Lisman interprets this spelling as an unorthographic writing.²⁴⁹ The later Neo-Assyrian text uses (correctly) the base su₈ in the infinite form su₈-ge-eš. Omitting the initial wedge of the sign sur would lead to g̃ar, which is semantically much closer to the Akkadian verb i/uzuzzu.²⁵⁰

Example 29: KAR 4 obv. 26

The infinite verbal form mu₂-mu₂-e-de₃ corresponds to the cohortative i nibnâ.²⁵¹ Both previous and subsequent Sumerian lines contain the second person plural, which is rendered with the first person plural in the respective Akkadian translation. It cannot be ruled out completely that the suffix –e-de₃ is a defective spelling for –en-de₃-en.²⁵² For a possible Old Babylonian example, see Sîn-iddinam A (text B = Wagensonner 2007, 545–546), line 21’: se₃-ga-de₃-en; see the commentary (Wagensonner 2007, 554). The suffix °-e-de₃ is frequently translated by an infinitive construction as well.²⁵³ One example is, for instance, obv. 29: gi-de₃ : a-na ku-un-ni.²⁵⁴

Example 30: KAR 4 obv. 27

Similar to Example 19 above the Akkadian translation uses the same equivalent for two (seemingly) different terms in the Sumerian text: (1) a₂-g̃eš-g̃ar-ra and (2) eš₂-gar₃.²⁵⁵ In all likelihood, it seems that the former is an erudite spelling for the latter, both imitating the Akkadian lexeme.

Example 31: KAR 4 obv. 32–33

In this example, the adjective gal referring to the gods in the Sumerian text, was re-interpreted and assigned to denote a quality of the house or abode. A similar case can be found in line 41 of “Nin-Isina’s Journey to Nippur”:

•

bara₂-gal-mah-ba si^sic mi-ni-in-g̃ar-re-eš [...]

While the Sumerian line qualifies bara₂ with both adjectives gal and mah, the interpreter understood gal as adverb and used rabîš.

Example 32: KAR 4 rev. 13–14 (= line 54)

The Sumerian verb zil is usually not equated with dešû, “to be abundant.”²⁵⁶. Already line 7 on the reverse corresponds to line 14 cited above. Instead of duššû, the interpreter uses the verb rubbû, “to enlarge.”²⁵⁷ The Middle Assyrian scribes in Assur knew the base zil, but apart from KAR 4 it appears to have never been used to render a form of Akkadian dešû. VAT 9541, which contains a Middle Assyrian excerpt from Ea V, equates this verbal base with ša₂-la-tu, “to split off” (obv. i’, 13’).²⁵⁸ In line 48 of “Nin-Isina’s Journey to Nippur”²⁵⁹ the Akkadian form u₂-da-aš₂-ša-ši renders Sumerian mu-un-na-ab-šar₂-re. The base šar₂ is a common equivalent for Akkadian dešû.

A last remark shall be made on the ending °-NE-ta-a in line 13 of the example cited above. The Sumerian conjunction °-bi-da appears in various readings. The “Astrolabe” B renders it °-bi-id-da. It is therefore likely to see in our line another way of spelling this conjunction: mušen-bi₂-ta-a.

8.3.7 Varia

Jerrold S. Cooper published an overview of the bilingual texts found at Assur (and Nineveh), which date to the period in question.²⁶⁰ No attempt is made in this brief section to provide any exhaustive treatment of additional texts.

In 1976, Wilfred G. Lambert edited a fragment, which he believed to date to the reign of the Assyrian king Tukultī-Ninurta I based on linguistic parallels to KAR 128 and 129.²⁶¹ Based on a paleographical analysis between the text published by Lambert and texts written by Marduk-balāssu-ēreš, Markham J. Geller concluded that all these texts might have been copied by the same scribe and that “a Tukulti-Ninurta prayer, albeit containing statements by the king himself in the first person, could have been composed in the reign of Tiglath-Pileser I, during a period when the Assur scribal schools were thriving and productive.”²⁶² Lambert deemed the Sumerian of BM 98496 as “obscure in the extreme.”²⁶³ The layout of the Sumerian and Akkadian versions is column-based. Unfortunately, most of the Akkadian text is gone, leaving the Sumerian text with many peculiar spellings intact: Examples are the obvious adverb zi-ne₂-eš in obv. i, 10 as a variant to more common zi-de₃(ne)-eš or the two consecutive verbal chains mu-un-dir-dir-re nam-bal-la₂-e (obv. i, 8), which probably need to be understood as non-orthographic renderings of *mu-un-dir-diri-ge nam-ba-la₂-e.²⁶⁴

Another example dating to this period is a “bilingual hymn to Ninurta” edited by Wilfred G. Lambert in his Babylonian Wisdom Literature.²⁶⁵ This bilingual text follows the usual interlinear layout. Unfortunately, the tablet does not preserve a colophon. It differs, however, quite substantially from the previous text. Its Sumerian version is to a great extent well understood, as demonstrated by the following example:

Example 33: VAT 10610 rev. 16–17

•

sila-dag̃al abul u₂-zug sil₆-la₂ g̃al₂-la dib-be₂-da-zu-[ne]

This line allows for a comparative analysis to the composition “Nin-Isina’s Journey to Nippur” discussed above. Line 5 contains both sila-dag̃al (: rebītu) and the verbal base dib (: bâ’u).²⁶⁶ Noteworthy is also Sumerian g̃al₂ for Akkadian malû. Line 43 of NJN contains the intriguing syllabic spelling g̃a₂-la-ni in order to render malât.

The Middle Assyrian scribes from Assur copied collections of incantations as well. A case in point is the fragment VAT 9833 (= KAR 24), which belongs to the series Utukkū lemnūtu.²⁶⁷ As was pointed out by Andrew R. George, this fragment is part of the “same tablet as BM 130660” edited by Markham J. Geller.²⁶⁸ A reconstruction favors a six-column tablet (see p. !!! Page Reference !!! below). As was noted by Geller, two of the Neo-Assyrian text witnesses from the Kuyunjik collection contained six-column tablets as well.²⁶⁹ Since VAT 9833 comes from an archaeological context, it is not entirely clear how BM 130660 happened to enter the collection of the British Museum. It is known that twenty boxes with finds from Assur were taken to London in the early 1920s, before Walter Andrae could tend to their shipment to Berlin, and that some objects were extracted from them.²⁷⁰ Geller notices about this text that “[a]lthough the script is indicative of a library hand, the errors in the text attest to the scribe’s carelessness or ignorance.”²⁷¹ Its scribe Marduk-kābit-ahhēšu is known from at least one other scholarly text. He copied the paleographical list AfO 4, plates III–IV.²⁷²

Geller discussed the variants in the Middle Assyrian copy compared to versions dating to the first millennium BCE. Besides orthographical variants, lexical or semantic differences are particularly revealing. As Geller pointed out, the Middle Assyrian text uses quite frequently rare equivalents when rendering the Sumerian.²⁷³ We get a similar perspective by looking on other texts used in this study.

Example 34: BM 130660 obv. ii, 24–25 (= Tablet 13–15, 106)

•

e₂-a mu-lal₂ g̃iri₃-ni ha-ba-an-gi

Lexical texts provide two Sumerian equivalents for the Akkadian verb e’ēlu, “to hang up”: šu-ur-g̃ar and ri. Stem I/3 seems to be triggered by the reduplication of the grapheme la₂. Noteworthy is also the verbal base gi, which is rendered here by stem IV of the Akkadian verb parāsu. The Neo-Assyrian text K.111+²⁷⁴ reads ha-ba-an-tar at the end of the line (rev. ii, 8) and renders it li-ip-ru-us in the subsequent Akkadian translation. Although the verbal base tar is usually equated with parāsu elsewhere in the Middle Assyrian evidence,²⁷⁵ there are a few hints for the usage of gi : parāsu in the Middle Assyrian version of Utukkū lemnūtu. The Chicago Assyrian Dictionary cites a Late Babylonian bilingual litany, which was published by Wilfred G. Lambert.²⁷⁶ This text reads in obv. 11 as follows: an-ra a mu-ni-ib₂-gi₄-a-ni : e-liš mi-la ip-ru-su.²⁷⁷ The Neo-Assyrian fragment K.5255 (= CDLI P395959) offers in rev. 8–9 the Akkadian equivalent še-pi ip-ru-su for g̃iri₃ mu-un-se₃-ki-ta. Hence, the verbal form in Utukkū lemnūtu cited above should rather be read ha-ba-an-sig₁₇.²⁷⁸

8.4 Conclusions

Although the main focus of this study is to pinpoint a selection of orthographical and morphological peculiarities, it is clear from the previous pages that many obstacles remain while dealing with Sumerian texts of a late period. Even if all these texts had been copied at the same place and date within a rather narrow time frame, many issues persist. On the one hand, the data is obscured by the sometimes rather complex ways and methods of textual transmission. In the case of Assur in the Middle Assyrian period, in general, and the group of the M 2 texts, in particular, there are many uncertainties regarding text acquisition and distribution of source texts. This is mainly due to the fact that the archaeological context was already disturbed in antiquity and the boundaries between the Middle Assyrian texts in this group and texts that had been assigned to the later temple library N 1 are not always clear. A place for copying tablets in this period, as for instance a scriptorium, has never been found. In the Old Babylonian period, the transmission of Sumerian literary texts was mainly triggered by the scribal education in the “schools,” the e₂-dub-ba-a. But there is no evidence for such an institution in the last third of the second millennium BCE. While we are in the lucky position that there is even archaeological evidence for such institutions in the first centuries of the second millennium and also to some extent for the first millennium BCE, we know astonishingly little about textual transmission in the latter part of the second millennium.

On the other hand, variations between different texts often hamper our understanding of specific grammatical problems. Furthermore, just a minor part of the extant texts give us data about the provenances of the sources. This information is mostly general in a way that allows no further investigation. Textual or linguistic analysis of the Akkadian translations may sometimes give clues, especially in light of particular Middle Babylonian forms,²⁷⁹ but here too we have no knowledge of the degree of redaction undertaken by the Middle Assyrian copyists. As long as we do not have more extensive sources for the Middle Babylonian tradition of Sumerian compositions, we have no way of knowing for sure whether these young scribes blindly copied sources or did redaction work themselves.

The Sumerian may contain peculiarities as well. Phonetic variants appear to be quite common.²⁸⁰ Even if they pose problems for the modern reader, the Akkadian translations frequently solve a great deal of issues. In this paper, it was deliberately decided not to speak about Sumerian in the Middle Assyrian period in terms of quality. We are not deemed to judge the Sumerian of this period. There are a great deal of peculiar writings, either phonetic variants or widely abbreviated verbal forms. One might even substantiate the claim that the former originates either from dictation or from minor hiccups in the scribe’s memory.²⁸¹

Recently, Eckart Frahm argued that the Kassite rulers “initiated some of the earliest editorial projects that led to the emergence of the new corpus of ‘canonical’ texts that remained in use until the end of cuneiform civilization.” A clue to such an editorial endeavor is provided by a scribal note on the later hemerological tablet KAR 177, according to which scholars copied and selected from seven tablets originating from places such as Sippar, Nippur, Babylon, and so forth, and gave (the new edited compendium) to the Kassite king Nazi-maruttaš.²⁸²

As I have tried to show throughout this short study,²⁸³ the Akkadian versions of the Sumerian compositions discussed above should not be considered a secondary layer of text.²⁸⁴ Originally derived from glosses and annotations, they eventually became part of the stream of tradition. Both Sumerian and Akkadian versions of a given line were treated as one unit.

The Sumerian language and its scribal lore were able to preserve its status and importance long after its demise as spoken language at the end of the third millennium or slightly later. Whereas the Akkadian language infiltrated and soon dominated the socio-economic life, many areas of the religious and cultural sphere still thrived from the presence of Sumerian texts. Lexical texts, both those dealing with the shapes and readings of signs as well as thematic word lists, were the essential tools for dealing with Sumerian semantics. But word lists present the Sumerian out of context. Bilingual texts that put whole Sumerian phrases and their Akkadian equivalents side by side can be seen in this light as well. They extend the lexical corpus by providing context. In doing so, they kept the Sumerian language alive and removed it from the artificial environment of lexical lists.

Highly learned literature, such as the treatment of the various month names of the year in the “Astrolabe” text presented above, are often considered erroneous due to the fact that their Sumerian appears to reveal deficiencies on a morphological basis or orthographical details that seem peculiar compared to more classical Sumerian literature of the early second millennium BCE. But these texts and their compilation need to be located in the arcane realm of the scribal art. When the scribe of the “Astrolabe” uses the sign dur for the Akkadian verb rahāṣu instead of dur₂, it must not necessarily be an error or misinterpretation. In a recent article about the various text layers that remain hidden within the orthography of a word, Stefan Maul argues that the Akkadian versions on late bilingual texts should be considered more as comments than simple transpositions into Akkadian.²⁸⁵ This view has many merits, in particular, in light of annotations on text witnesses of Sumerian literature dating to the early second millennium BCE.

Acknowledgements

This contribution benefitted from the involvement in the project part “Episteme als Konfigurationsprozess: Philologie und Linguistik im ‘Listenwissen’ des Alten Orient” (under direction of E. Cancik-Kirschbaum and J. Klinger) within the framework of SFB 980 “Episteme in Bewegung.” I would like to thank the organizing committee of the conference “Crossing Boundaries: Multilingualism, Lingua Franca and Lingua Sacra” (Berlin, 8–10 November 2010) for inviting me to publish this work. This study originates in an unpublished master thesis on Sumerian divine journeys.²⁸⁶ In this thesis, the various Sumerian compositions were discussed and re-edited. Among the known texts were two bilingual tablets of a composition now known as “Nin-Isina’s Journey to Nippur” dating to the Middle Assyrian period.²⁸⁷ The new edition led to an investigation of the responsible scribes and their wider familial and occupational background during several research stays at the Vorderasiatische Museum in Berlin and the British Museum in London. Previous studies focused primarily on the information gleaned from the respective colophons.²⁸⁸ For abbreviations, see the list provided by cdli:wiki.²⁸⁹ Lexical and literary compositions are referred to by their incipits or by the subsequent abbreviations (in alphabetical order): (1) Lexical: Ai = ki-ulutin-bi-še₃ : Ana ittišu; Aa = a : a₂ : nâqu; Antag̃al = an.ta.g̃al₂ : šaqû; Diri = diri : si.a : (w)atru; Ea = ea : a : nâqu; Erimhuš = erim.huš : anantu; Ura = har-ra : hubullu; Igiduh = igi.duh : tāmartu; Izi = izi : išātu; Kagal = ka₂.gal : abullu; Lu = lu₂ : ša; Nabnītu = sig₇.alan : nabnītu; Sa/Sb = Silbenalphabet A/B; (2) Literary: Lugal-e = “Ninurta’s Expoits”; Angim = “Ninurta’s Return to Nippur”; NJN = “Nin-Isina’s Return to Nippur.” Corresponding entries in the CDLI database²⁹⁰ are given by the respective P-no.

8.5 Appendix

References

Footnotes