Canaanite in cuneiform.

• Author: von Dassow, Eva

It has become a truism that Akkadian, the principal Semitic language of ancient Mesopotamia, was the lingua franca of the Near East during the second millennium B.C.E. This is stated, more or less in so many words, in any number of works on the ancient Near East, which usually offer the Amarna letters, the trove of correspondence between Egypt and other states that was found at the site of Akhetaten (Tell el-Amarna), as the parade example of Akkadian as lingua franca. (1) But is the truism true?

The idea that Akkadian was in common use as a written language throughout the ancient Near East, Egypt included, tacitly assumes the exact identity of writing with language: it assumes, that is, that what people write represents at face value the language in which they mean to communicate. According to this theory, if a scribe in Hatti or Egypt, Canaan or Cyprus, writes in cuneiform using sign sequences that spell Akkadian words, he means to write in the Akkadian language, regardless of whether what he writes exhibits features of his own or another language as well as errors in Akkadian. But this idea conflates the modality of encoding linguistic expression with linguistic expression itself. It need not be the case that the signs with which a text is written directly represent the language in which it is written, and to assume that this is the case is inherently problematic when the writing system in question is one such as cuneiform, which tends to employ a variety of frozen graphic sequences (e.g., logograms) dissociated from language-specific referents. When such a writing system is borrowed from one language community into another, the assumption that the language of a text is directly represented by the writing of the text becomes so problematic that it should be treated as a proposition requiring demonstration rather than an axiom to be taken for granted.

With regard to the ostensible use of the Akkadian language, written in cuneiform, outside of Mesopotamia, in many instances the evidence offers reason to jettison the assumption that writing is a face-value representation of language, and to consider alternative possibilities. The present article develops an alternative hypothesis concerning one such instance, which involves a subset of the Amarna letters that are usually cited to illustrate Akkadian-as-lingua-franca, namely the use of cuneiform by Canaanite scribes during the Late Bronze Age (c. fifteenth-thirteenth centuries B.C.E.).

The language that Canaanite scribes used for correspondence in cuneiform during the Late Bronze Age has been an object of scholarly attention since the discovery of the Amarna tablets over a century ago. This language, that is, what the Canaanite scribes wrote, appears to be a hybrid produced by grafting the scribes' native Canaanite onto their borrowed Akkadian. The resulting Canaano-Akkadian hybrid, which was initially thought to incorporate proto-Hebrew forms into a barbarized Akkadian dialect, is considered in current scholarship to be an autonomous dialect with its own linguistic system and its own rules of morphology and grammar. (2) Recently, many features of this dialect have been catalogued and analyzed by Anson Rainey in his four-volume work Canaanite in the Amarna Tablets. The description of Canaano-Akkadian that can be abstracted from Rainey's work reveals a strange composite: in texts written in this hybrid, sentences composed of Akkadian words are arranged in Canaanite syntax; Akkadian words are made to function according to the rules of Canaanite grammar; Akkadian words are provided with Canaanite affixes; Akkadian words and morphemes are recombined to produce otherwise nonexistent forms; and Canaanite words, besides being deployed as glosses, are used alongside Akkadian ones. Such a peculiar array of features (detailed in specific terms below) prompts asking what kind of language salad was this; who used it with whom, and how?

In this article, I propose that the hybrid of Canaanite and Akkadian in which Canaanite scribes wrote was not a language of any kind, but an artifact of these scribes' use of cuneiform, and furthermore, that the language underlying their communication in cuneiform was not Akkadian but Canaanite. The Canaanite use of cuneiform would then be an instance of alloglottography, to borrow a term from Ilya Gershevitch, who defined it as "the use of one writable language for the purpose of writing another language": the Canaanite scribes used Akkadian words, spelled in cuneiform, to write Canaanite. (3) In order to elucidate the basis for my hypothesis, I shall first survey some of the features of the Canaanite scribes' usage that appear to be symptoms of their "hybridization" of Canaanite with Akkadian. Second, I shall address the question whether this hybrid was a language or a means of writing one; third, proceeding on the theory that the Canaano-Akkadian hybrid is the Akkadographic writing of Canaanite, I shall outline a model to explain how the Canaanite scribes' use of cuneiform might have worked; and last, I shall explore the question of how the Canaano-Akkadian writing system might have developed, based on the evidence of cuneiform texts found in Canaan, in particular those that may reflect writing instruction.

The texts under consideration in this inquiry consist mainly of those Amarna letters that were written by scribes from Canaan, with the addition of the few extant letters addressed to Canaanite rulers, and these Amarna letters are supplemented by a diverse assortment of roughly 50 tablets and fragments written by Canaanite scribes and found at sites in Canaan. (4) The tablets and fragments found in Canaan span about three centuries as well as numerous different sites, and many of them are fragmentary or barely legible. Only at a few sites, notably Taanach and Kumidi, have remnants of what were clearly archival groups turned up, comprising letters along with other kinds of texts. Thus the assemblage of Late Bronze Age cuneiform tablets found to date in Canaan provides rather little material that is directly comparable with the fairly coherent assemblage of Canaanite letters found at Amarna, although it is of great importance for studying the use of cuneiform in Canaan (on which see the final section of this article). The relative paucity of tablets found in Canaan, and of Canaanite tablets from periods preceding and following the Amarna period itself (mid-fourteenth century), somewhat constrains the scope and approach of investigation; the principal focus is necessarily on the Amarna tablets, and the perspective is for the most part synchronic.

Prior to commencing the discussion, it is important to observe that designations such as "the Canaanite scribes" and "the Canaanite-Akkadian hybrid" are convenient simplifications. Study of the tablets written by scribes from Canaan suggests that there were as many Canaanite cuneiform idiolects as there were Canaanite cuneiform scribes, and the work of William Moran and his successors has shown that these scribes did not constitute a unitary group, either linguistically or in terms of their training. (5) Unitary designations are employed here as shorthand, for the sake of efficient expression, but are not meant to collapse the differences among various text groups, scribes, and their uses of written language. Another prefatory note is that when speaking of Late Bronze Age Canaanite, we are speaking of a largely unknown language, the characteristics of which must be inferred on the basis of contemporary Ugaritic, later Hebrew and Phoenician, and the very same array of Canaanite glosses and "Canaanizations" in the cuneiform tablets written by Canaanite scribes which are the object of investigation; when describing these Canaanizations, therefore, it is often necessary to draw on Western Semitic languages generally, rather than Canaanite specifically.

FEATURES OF "HYBRIDIZATION"

Cuneiform texts written by Canaanite scribes exhibit, besides the outright substitution of Canaanite words for Akkadian ones, a variety of features that blend Canaanite and Akkadian morphology and grammar. Many of these features, which tend to be somewhat inconsistent and variable, involve (1) employing Akkadian words and morphemes according to the rules for their Canaanite counterparts, (2) collapsing Akkadian grammatical distinctions that did not exist in Canaanite, and (3) grafting Canaanite morphology onto Akkadian words. The following paragraphs survey examples of such features, with references to Rainey, CAT I-III, where text citations supporting the description of each feature are found. (6)

(1) The Akkadian negatives ul and la were sometimes interchanged and employed in accord with the rules for their West Semitic (functional) counterparts la and 'al, rather than in accord with Akkadian rules (CAT III: 209-26). Similarly, the Akkadian interrogative pronouns mannu and minu, "who" and "what," were sometimes interchanged, so that the word chosen sounded like the appropriate Canaanite interrogative: thus, Akkadian minu was used to mean "who" and mannu was used to mean "what," in accord with the vowelling of their Canaanite counterparts miya and mah(/man(nv)) (CAT I, ch. 6, esp. pp. 105, 111-12). Akkadian prepositions were sometimes employed in functions that they did not perform in Akkadian, but that were appropriate to their Canaanite equivalents. For instance, Akkadian ana, "to, for," was occasionally used where West Semitic languages would use l- while Akkadian would use ina; conversely, Akkadian ina, "in, from," was used in the functions of West Semitic b- as well as in its standard Akkadian functions (CAT III: 12-14, 21, 31-35). The Akkadian temporal conjunction inuma was employed for the functions of the West Semitic subordinating conjunction ki. (7) And, while the Akkadian coordinating conjunction -ma was completely ignored, the conjunction u, which in Akkadian expresses only simple coordination...