Sumerian Language Use at Garíana on Orthography, Grammar, and Akkado-Sumerian Bilingualism* WALTHER SALLABERGER LMU MÜNCHEN

Home , Ma (cuneiform), Ta (cuneiform), Um (cuneiform)

Sumerian Language Use at GarÍana On Orthography, Grammar, and Akkado-Sumerian Bilingualism* WALTHER SALLABERGER LMU MÜNCHEN

Among the administrative texts from the Ur III period, the GarÍana documents stand out for their rich vocabulary and a relatively high quota of phonographic writings. This evidence becomes most fascinating since almost all scribes, who wrote the documents in the prestigious Sumerian language, bore Akkadian names. This study covers various aspects of language that contribute to our topic: writing and phonology, morphology, and syntax; thereby frequencies and proportions are given much attention. The fact that the GarÍana scribes were mostly native Akkadians transpires relatively rarely in various impositions in morphology and syntax, mostly in the message domain of single speakers. The mastering especially of the diversified verbal morphology or the invariably Sumerian basic vocabulary is a clear sign of a high degree of language acquisition, and the phonographic spellings even hint at imitation of Sumerian phonemes. The integration of special expressions from Akkadian in a basically Sumerian vernacular agrees well with the fact of a far-reaching acquisition of the non-native Sumerian language. The active Akkado-Sumerian bilingualism of the GarÍana scribes would have been unthinkable without constant communicative contacts with the native speakers of Sumerian in the region.

The archive from GarÍana, a settlement in the the GarÍana documents stand out by their rath- province of Umma at the banks of the lower er frequent use of a variety of verbs and the for- Tigris, provides a unique occasion to investi- mulation in phrases. Furthermore, this homo- gate the usage of Sumerian at the end of the geneous corpus boasts a relatively high quota of third millennium. Compared to the bulk of the phonographic writings, which allows for a bet- more than sixty thousand published adminis- ter comprehension of the actual language than trative texts from this period, the time of the does the usual standard orthography. This evi- Third Dynasty of Ur (Molina 2008), which dence becomes important and most fascinating present themselves most often as lists with an because of the specific sociolinguistic situation: addition of some few administrative key terms, the population of the military settlement of

* David I. Owen deserves my deepest gratitude for from discussions with Jan Keetman and, especially, providing me with transliterations of the GarÍana Wolfgang Schulze, who provided indispensable texts shortly after their acquisition and he encour- linguistic background information and literature aged me to undertake this study. The study of this and who encouraged me to describe the treated fascinating text material led to a rewarding ex- features more precisely, which eventually entailed change with David, especially in the first years of hours of counting references. the work on the archive. This study has profited

335

336 Walther Sallaberger

GarÍana was mainly of Akkadian and other ori- 1. On the GarÍana Archive and Its Scribes gins, but it was situated within the Sumerian- The sheer existence of a “GarÍana archive” is speaking area of southern Babylonia; most owed to the ceaseless efforts of David I. Owen scribes, who wrote the documents in Sumeri- and Rudolf H. Mayr, who collected and pub- an, bore Akkadian names. lished more than 1400 cuneiform tablets that The discussion is ultimately devoted to this were looted at the place called GarÍana in specific situation of language contact. There- antiquity (Owen and Mayr 2007). Although fore, it concentrates on those aspects of language the archaeological context is lost, the coher- in which differences from standard Sumerian ence of the archive is remarkable: apart from a usage can be noted. A first section addresses the handful of earlier texts all dated documents relationship between written and spoken lan- belong to the ten years between fiu-Suen 4 and guage within this corpus, and a phonological Ibbi-Suen 4. They are concerned with the detail, the spelling of the infinite -e-d-e-forms, internal organization of the household and the becomes a firm witness for the representation military camp headed by the general and phy- of spoken Sumerian in these documents. On sician fiu-Kabta and his wife, the princess the other hand the imposition of Akkadian on Simat-IÍtaran (cf. Heimpel 2009, 2–4). The the Sumerian vernacular used by the GarÍana probable etymology of GarÍana as GaraÍ-Ana scribes is investigated in regard to both syntax “the camp of An” (with garaÍ as loanword of and lexicon. Akkadian kar⁄Íu “military camp”), the dedica- The exceptional situation of GarÍana allows tion of the main temple to the warrior god Nergal, the role of fiu-Kabta as general, the sta- for an exemplary view on Sumero-Akkadian tioning of “soldiers” (aga -ús, TÉL 171, ITT 3 bilingualism in southern Babylonia at the end 3 6174) and the high number of “troops” (eren , of the third millennium, just before the eco- 2 TCTI 2 3543) at this place, and the short period nomic, political, and social catastrophe at the of documentation between fiulgi and IÍbi-Erra end of the Third Dynasty of Ur put an end to indicate that GarÍana was mainly a military the old city Sumerian states and eventually to camp in the South, founded most probably by Sumerian as their vernacular. Ur-Namma or fiulgi of Ur. Methodologically this study has tried to The proper names reveal that persons of avoid single anecdotal observations, which mostly Akkadian origin were settled here. tend to be misleading, but to cover compre- Thus GarÍana differs sharply from the sur- hensively the various aspects of language that rounding ancient cities of Umma and ªirsu, contribute to our topic, and thereby frequen- which were dominantly inhabited by Sumeri- cies and proportions have been given more ans. Although, of course, a single name does attention than it is unfortunately too often the not reveal the actual language use of the person case in linguistic studies of Sumerian. But only bearing that name, the total evidence is very this broad coverage of the evidence allowed a clear in this regard, as the following tabulation, sound application of linguistic models of lan- based on randomly selected name lists of four guage contact and eventually led to an interpre- contemporary sites (Sallaberger 2004), illus- tation that covered all aspects of the analyses. trates:

Total of names Of other or unclear City counted Sumerian Akkadian linguistic affiliation1 Nippur 1433 52 % 21 % 28 % Umma 546 68 % 15 % 17 % GarÍana 172 9 % 68 % 23 % ªirsu 1111 63 % 14 % 23 %

Sumerian Language Use at GarÍana 337

GarÍana was thus inhabited by a dominant- flummid-il‹ (IS 4/01, maÍkim) ly Akkadian-speaking population, whereas the fiarakam (fiS 9/11, witness) administration of the provincial governor at fi2li⁄num (fiS 5/04, maÍkim) Umma controlled a dominant Sumerian popu- fiu-Adad (fiS 8/12, ™iri3) lation. fiu-Erra (fiS 8/06, maÍkim) The distribution of the personal names fiu-K›bum (fiS 9/09, witness) among the scribes active at GarÍana agrees with T›ram-il‹ (IS 2/02–4/01, ™iri ) the general picture. In the list of scribes attested 3 Ur-Eanna (fiS 8–IS 1/12, ™iri3, at GarÍana that is given below, the following maÍkim) information is added: the period attested in the U‰i-nawir (IS 2/05-06, ™iri ) documents for the person indicated as “scribe” 3 (dub-sar), but no attestations of the same per- Of the 23 names of scribes, son without that title; the administrative 2 are Sumerian (fiarakam, Ur-Eanna); function (™iri “go-between, responsible” or 3 18 are Akkadian (Adad-tillat‹, AÓ›ma, maÍkim “commissioner, deputy” as designa- AÓu-waqar, Aw‹lumma, Ea-d⁄n, tions of involvement in transaction) or the role Ea-Íar, Erra-b⁄ni, Ibni-Adad, Il‹- as (eye) witness; the attestation as scribe of the ‰ul›l‹, I‰‰ur-Suen, L⁄-q‹p(um), household in wage lists (text nos. 394, 399, Puzur-Ninkarak, flummid-il‹, fiu- 406). Those persons whose title “scribe” is Adad, fiu-K›bum, fiu-Erra, known only from the seal inscriptions are not T›ram-il‹, U‰i-nawir); considered (Owen and Mayr 2007, 429–439): 3 are of various or uncertain linguistic Adad-tillat‹ (fiS 6/05–fiS 8/03, ™iri3) affiliation (Babani/Ba-ba-ni, Enlil- d AÓ›ma (IS 2/01, ™iri3) bazaDU/ En-líl-ba-za-DU, AÓu-waqar (IS 1/01–3/01, ™iri3) fi2li⁄num/fie-NI-a-LUM). Aw‹lumma (IS 1/12–2/01, ™iri ) 3 In the wage lists two or three scribes are Babani (fiS 8/11–IS 3/01, ™iri ) 3 listed (Ea-d⁄n, Ea-Íar, L⁄-q‹p(um), Puzur- Ea-d⁄n (household scribe no. 399, Ninkarak); according to the anonymous fiS 6/06; no. 406, fiS 6/11) inspection lists between two and five scribes Ea-Íar (fiS 6/06–9/07, ™iri3, were employed at the building work at GarÍana maÍkim; household scribe (4 scribes: no. 379, fiS 6/05/22; 2 scribes: no. no. 394, fiS 6/05) 402, fiS 6/08; 5 scribes: no. 556, date lost). It DU Enlil-baza (IS 2/06, ™iri3) should be emphasized that the responsible Erra-b⁄ni (fiS 8/08, witness) scribes appear without or with different titles Ibni-Adad (fiS [x]–IS 3/08, ™iri3) over a longer period of time (see on the names Il‹-‰ul›l‹(AN.DÙL) (fiS 9/02, eye witness?) Heimpel 2009, 38–43). The Akkadian scribes I‰‰ur-Suen (IS 2/07, ™iri3, maÍkim) of the Akkadian settlement of GarÍana wrote L⁄-q‹p(um) (no. 394, fiS 6/05; no. 399, their administrative texts exclusively in Sume- fiS 6/06) rian, not in Akkadian. One may adduce two

Puzur-Ninkarak (fiS 6/05–7/12, ™iri3; arguments derived from the tradition of writ- household scribe no. 394, ing at this period to explain this behavior: the fiS 6/05; no. 399, fiS 6/06; scribal education and the tradition of adminis- no. 406, fiS 6/11) trative texts. Teaching of cuneiform writing

1 This category includes not only Hurrian, Elamite, and unanalyzable names (e.g., of the so-called ‘Ba-

nana’ type), but also names such as Urdu2-DN “servant of DN” that can be read both in Sumerian and Akkadian.

338 Walther Sallaberger was intrinsically linked to the Sumerian lan- corpora. In this regard one may apply the rule guage; the syllabary and the ideograms are on the relationship between lexicon and lan- based on Sumerian and the traditional school guage use observed for administrative docu- texts, most of all the lexical lists, were Sumerian ments: An obligatory strict text pattern and a even long after the disappearance of Sumerian restriction of the lexicon to administrative key as an everyday vernacular. Secondly, Sumerian terms allows a reduction of grammatical mark- was the language of administration both in ers; the relations between various elements, southern Babylonia and, more importantly in persons and objects, are indicated by the for- this regard, in the state organizations, testified mulaic pattern and are not marked grammati- by the archives from PuzriÍ-Dagan. cally as phrases. On the other hand, lexical It is probably anachronistic to assume an diversity implies the use of grammatical rela- active royal language policy that had influ- tions and their markers. Everyday language enced the writing of documents,2 since Akka- thus appears in administrative documents in dian was used as well for administrative docu- those phrases that deviate from the standard ments in northern Babylonia (Ishan Mizyad, formulas used in an archive.3 see Hilgert 2002, 21–23) and also for royal Although both scholarly education and inscriptions. Furthermore the differences be- administrative traditions may partly explain the tween the provincial archives of ªirsu and of use of Sumerian in the GarÍana documents, the Umma further rebut the assumption of a state- lexical variation strongly suggests that this was directed technical language for administrative a vernacular actively used by the scribes. Fur- scribes. Nonetheless, there is no doubt that in ther evidence for this assumption will be pre- the state of Ur Sumerian was the more presti- sented at the end of the article. gious language—prominently used in royal texts (year dates, hymns, many inscriptions)— 2. Orthography and Phonology: Infinite Verbal and it was the vernacular of the economically Forms in -e-d-e dominating provinces of Umma and especially 2.1 Unorthographic Writings ªirsu. In the GarÍana documents one encounters a GarÍana Sumerian can definitely not be number of so-called unorthographic writings. labeled a kind of “simplified Sumerian” that Words written regularly with simple or com- would apply only administrative key terms. pound logograms in the large state or provin- Such a documentary pseudo-Sumerian that cial archives are represented by basic phono- builds on some technical expressions and intro- grams usually of the CV and VC pattern. A few duces artificial Sumerograms is attested for examples of words that are usually written dif- example at Old Babylonian Uruk in the nine- ferently in other Ur III archives may suffice teenth century (Sallaberger 2000, 274f.). GarÍa- here (for other examples see Kleinerman and na documents, however, are characterized by a Owen 2009, 175 s.v. Syllabically written Sum- lexical variation that is rare in contemporary erian):

2 Of course, the high prestige of Sumerian as reflect- matische Markierungen (mit Hilfe von Kasusmor- ed in the royal texts must not be underestimated phemen) zu verzichten. Umgekehrt bedeutet lexi- and one notes that ﬁulgi enhanced the education of kalische Vielfalt, ein Abweichen von den Schlüssel- scribes by the founding of schools. begriffen, auch Grammatizität. [...] Nach dieser 3 Based on a study of patterns of administrative doc- Regel tritt die Syntax der Sprache ihrer Zeit in uments from the Old Sumerian to the Old Baby- Wendungen und Sätzen auf, die vom Standardfor- lonian period, Sallaberger (2000, 276) concluded: mular des jeweiligen Archivs abweichen. [....] “Die verbindliche Abfolge der einzelnen Elemente Wichtig ist noch die (weitgehende) Abhängigkeit einer Urkunde und die Reduktion der lexikali- des Formulars vom Lexikon: andere Verben als die schen Mittel auf wenige Schlüsselbegriffe erlauben administrativen Schlüsselbegriffe ziehen syntak- es, zur Darstellung wechselseitiger Bezüge auf gram- tische Formulierung nach sich.”

Sumerian Language Use at GarÍana 339

Ía-ra-ab-du (GarÍana) = Íár-ra-ab-du (stan- scribes would have been employed (cf. dard Sumerian) “field surveyor” Veldhuis 2008). It is in the same line of argument that the ™e26-en (in verbal form in-da-™e26-en) = ™en “to go” number of unorthographic spellings is especial- ú-lá-bi = ul -la-bi “quickly” ly high in the private or small administrative 4 archives at Nippur (Wilcke 2000, 34–49 and The use of unorthographic writings and the 66–80). In his treatment of these writings creative use of syllabograms implies that the Wilcke (2000, 47f.)5 observed that phono- Sumerian written at GarÍana was not consis- graphic writings tend to be more frequent tently acquired graphically, that it was not sim- when a lexeme appears not isolated but in ply a prefabricated professional language of combinations, reduplicated, or with prefixes or scribes. Instead the writings “by ear” can be suffixes. taken as unequivocal evidence that these Sum- erian phrases were based on spoken Sumerian 2.2 -e-d-e Writings as a phonetic reality. For the sake of clarity it 2.2.1 The Evidence has to be repeated that at GarÍana unortho- An especially rewarding field for the study of graphic writings appear in a context of freely phonology at GarÍana is constituted by the infi- phrased Sumerian texts (see above), but, of nite verbal forms ending in _V/-e-d-e (stem course, unorthographic writings have to be eval- ending in a vowel/marû marker e, imperfective uated differently in other environments such as -d-, directive -e) “in order to.” Most referenc- the fixed phrases of post-Sumerian legal texts es can be found in the worker-inspection and (e.g. in Old Babylonian Kisurra) or religious the work-assignment accounts from formula- texts. tions of the type: The relatively high number of unortho- ™uruÍ/geme VERB-(e)-dè gub-ba “ graphic writings, as well as the greater variation n 2 n in the formulations, is a characteristic feature of men/women on duty in order to do VERB” smaller organizations that is shared, for exam- (nos. 1–306, Owen and Mayr 2007, groups 4 I and II). ple, by the texts from Irisa™rig/ml-Íarr⁄k‹. Here less “paperwork” has to be performed, The following list includes all pertinent whereas the large state or provincial scribal GarÍana forms and it notes relevant parallels offices produced more concise and more stan- from other Ur III archives.6 In cases of variants, dardized texts. Apparently the larger amount of the frequencies are given. Furthermore, the scribal work implied a better scribal formation class of marû-stem formation is indicated (I reg- on the job and in the central state organizations ular, II duplicated, III augmented, IV irregular such as at PuzriÍ-Dagan the best-educated royal verbs).

4 David I. Owen is preparing an edition of texts from schreibt er es mit Zeichen, die er beherrscht und this archive (Nisaba 15, forthcoming), and I am die der Lautung des Wortes, wie er es im Ohr hat, grateful to him for sharing with me numerous ex- nahekommen.” amples. 6 In cases of ambiguity reconstructed forms have 5 Wilcke (2000, 47f.) writes: “[D]er Grund für das been excluded from the counts. Most references Abweichen von der Norm [ist] nicht zur Schau can be verified and checked by the verb list in gestellte Gelehrsamkeit, sondern die Unsicherheit Kleinerman and Owen (2009, 231–377) and des jeweiligen Schreibers. Er kennt das Wort, ist through Manuel Molina’s indispensable “Database sich aber über die korrekte Zeichenwahl im of Neo-Sumerian Texts” (BDTNS) (bdts.filol. Unklaren. In den Zeichenlisten, nach denen er die csic.es); in some cases (e.g., ra-ra not listed in verb Schrift gelernt hat, standen die Wörter isoliert, list) references have been added. The comparative ohne Prä- und Suffixe, nicht redupliziert, nicht in data from other Ur III archives derive from person- Wortkoppelungen oder anderen engen Verbin- al collections and Manuel Molina’s BDTNS. dungen, die die Lautstruktur verändern. Darum

340 Walther Sallaberger

GROUP A: STANDARD ORTHOGRAPHY. ke4-dè (ca. 116™) = *kè(AK)-dè (“I”) “to Some verbs comply with the standard orthog- make”; in other Ur III texts written kè- raphy, where the logographic writing corre- dè, cf. Attinger (2005, 62) sponds to the expected phonemic realization. ku-bu-dè (gu5-bu-dè) (no. 1309:11) = The class I stems listed here end in a vowel like *gub-bu-dè (I/IV) “to prop up” (note all class II verbs: that gub-bu-dè is not attested at GarÍana dù-dè (I) “to build” despite the ubiquituous gub-ba); cf. regular gub-bu-dè, e.g., UET 3 697, NATN du8-dè (I) “to caulk” 882 ge4-ge4-dè (only restored contexts) (II) “to ku-úr-gu-ru-dè (no. 972:40, no. 975: 42, return” no. 1090:1-3) = ? (II)8 ™á-™á-dè (II) “to place” ne-ne-de4(TE) (no. 1016:21) = *ni10-ni10- Ói-dè (I) “to mix, blend” dè (II) “to circuit”; cf. literary text Íu

lu-dè (I) “to stir, mix” ni10-ni10-dam NATN 8 ii 14

ra-ra-dè (no. 262:21) (II) “to hit, beat” la-Óa-dè (2™), la-Óe-dè (1™) = *laÓ(4)/5-e-dè (IV) “to bring”; cf. la-Óe-dam, e.g., sa10-sa10-dè (II) “to buy, pay” TCL 5 6047:15, 6163:28; LA° -dam su-su-dam (II) “to repay, replace” 5 SNAT 83, SAT 2 352(?), LA°5-e-dam tà-tà-dè (II) “to apply” TCTI 2 3484 zi-zi-dè (II) “to remove” 9 si-™e6-dè/Íé-™e6-dè = *Íe™6-e-dè (I) “to boil, cook”; no corresponding forms GROUP B: PHONOGRAPHIC WRITING: attested Some verbs are consistently written phono- te-te-dè (de4-de4-dè?) (no. 222: 4) = *de5- graphically, not logographically. Note that all de -dè (II) “to collect”?10 Óamˇu stems of class I verbs end in a consonant: 5 ti-te (dì-de ) = *di-dè (IV), in: káb ti-te dì-de see below ti-te 4 4 (dì-de4) (no. 271: 6) “to check, calcu- gu5-bu-dè see below ku-bu-dè late”; regularly written di-dè/-dam, cf. ì-li-dì (ca.34™), ì-li-dè (2™ in text no. 53)7 = Attinger (1993, 396. 404. 574)

*íl-e-dè (I) “to lift”; in Ur III texts usu- ù-ru-dè = *uru3-dè (I) “to level”; in other ally reduplicated: íl-íl-dè; reading íl not Ur III texts written ùr-dè, cf. Yoshikawa assured in NATN 491: 4' (ge íl-dè) (1988, 63)

7 10 The reading of the variant i-l[i]?-dè, no. 46: 25', re- No. 222 reads: 1 ™uruÍ LÚ.TU9 ... PN lú-kí™-ge4-a mains uncertain, since the same text writes ì-li-dì lugal-ta (for -da) lú te-te-dè in-da-™é26-en “1 man, in line 20'; documents tend to keep the same or- a fuller, went with PN, the royal messenger, in orthography for a word (cf. ì-li-dè 2™ in no. 53). der to ... men.” Kleinerman and Owen (2009, 360) 8 According to Kleinerman and Owen (2009, 302) understand the verb as te “to approach” (marû stem this stands for kúr gur-ru-dè “to return in the fu- te-™e26); the reduplication would point to a plural ture.” However, this does not fit the contexts, object: “in order to approach men.” However, in which are about flour treated for meat. In letters phonographic writings te represents /de/. Thus I the marû stem of gur appears in nam-mi-gur-re. No tentatively propose to understand it as a writing for verbal core like /kuru/ of the reduplication class (II) de5-de5-dè (II) “to collect.” The combination of is listed by Attinger (2009) Annexe 2. Perhaps we are de5 with persons is known, e.g., from NG 209: 13 dealing with a form of the marû stem KARA .KARA lú-enim-ma-bi ba-de5-de5-ge-eÍ-[àm] “its witness- 2 2 es have been collected (i.e., withdrawn, brought of KARA2 “to illuminate, light.” Do the texts thus refer to the coating of meat with flour? somewhere else?)”; cf. also NG 190: 2–3. 9 Add reference no. 65:21', listed by Kleinerman and Owen (2009) as sim “to sieve, filter.” Sumerian Language Use at GarÍana 341

GROUP C: LOGOGRAPHIC WRITING WITH- NATN 17 etc.; a-á™-dè-e NATN 397; á- OUT INDICATION OF MARÛ VOWEL *-E: ™á-dam NATN 134 (Yoshikawa 1988, 73) Other verbs are written only with logograms ba-al-dè (I) “to excavate” (no. 104:37 // (transliterated below in small capitals), which no. 105:35, no. 155: 26 restored)13; cf. allow no conclusion on the pronunciation; ba-al-da(-Íè) MVN 11 50, ba-al-e-dè here some references to other Ur III texts are TCS 1 216,14 nu-ba-al-e-da AOAT 25 added: 444 no. 9, ba-al-a-dè TPTS 2 5:4 DUB-DUB-dè (II?) “to heap” (no. 385); cf., lu-LU°-dè (I) “to clean, wash” (no. 783:3); e.g., CT 32 49 cf. luÓ-dè UET 3 49415, luÓ-da RTC È-dè for /edede/ (è.de-d-e) (III) “to go/ 307; luÓ-dam JCS 32 230 6 N-T 254 bring out”; cf. È-dè, e.g., UET 3 836. tùm-dè (IV) “to bring”; regular form, no 854, more often reduplicated form (plu- -e expected (on tùm see Meyer-Laurin ral object) È.È-dè, see Yoshikawa (1988, 2010). 56); È.È-dè may also replace the marû stem, e.g., in NRVN 1 264, see Krecher GROUP D: LOGOGRAPHIC WRITING OF VER- 1995, 165–73 BAL BASE WITH CONSISTENT PHONOGRAPHIC GAZ-dè for /gazade/ (gaze-d-e) (I) “to crush” RENDERING OF MARÛ VOWEL *-E: (no. 1013); cf. Yoshikawa (1988, 55) This group includes standard writings of class I G≤D-dè for /gidide/ (gíde-d-e)11 (I) “to verbs that end in a consonant in the Óamˇu stem stretch, draw”; cf. NATN 816 (cf. taÓ-Óe-dam, zi-re-dam). u GUL-dè for /gul de/ (gule-d-e) (I) “to KALA.KALA-ke4-dè (I?, II?) “to reinforce”; decompose”; cf. UET 3 894 cf. KALA.KALA-ge-dè UET 3 486 etc., KEfiÉ-dè for /keÍedrede/ (keÍedre-d-e) (I) KALA-ge-dè UET 3 1504 viii 14 “to bind, fasten”; cf. in texts from Ur the KU5-ru-dè (I) “to cut (etc.)”; cf. Ur III reg- variants KEfiÉ-dè UET 3 1212, KEfiÉ-dè- ularly KU5-dè, but TMHNF 1/2 271 dè UET 3 873, KEfiÉ-re-dè UET 3 1531 KU5-ru-dè (2™) (Yoshikawa 1988, 58f.) u SUR-dè for /sur de/ (sure-d-e) (I) “to LÁ-e-dè (I) “to hang, weigh (etc.)”; cf. LÁ-dè, twist, twine”; cf. SUR-dè, e.g., NATN lá-dam (Yoshikawa 1988, 58: 7, 59: 11) 596, Yoshikawa 1988, 60: 13; SUR-e-dè si-ke4-dè (I) “to insert, pack into” (2™, nos. UET 3 1465 16 274-275) ; cf. Ur III regularly si-ge4- 17 GROUP C': OTHER VERBS WITHOUT INDICA- dè , but si-ke-dè NATN 114

TION OF marû VOWEL *-E: SÌ-ke4-dè (I) “to apply, place, tamp”; cf., á-á™-dè (I) “to measure out” (no. 299:5, e.g., sì-ge4-dam RTC 288, sì-ge4-dè restored in no. 298:5)12; cf. á™-da UET 3 556, 613, sì-ge-dè UET 3 557, sì- (Yoshikawa 1988, 72f.), á™-e-dè, e.g., dè UET 3 290

11 15 Probably reduplicated in a different context: [G]≤D. A reading luÓ-dè in UET 3 787:2 could not be ver- G≤D-da[m] no. 525:7. ified by collation. 12 This reference to Íe á-á™-dè “to measure out grain” is 16 Kleinerman and Owen (2009, 343) assume a vari- not listed under á™ in Kleinerman and Owen 2009. ant writing of sè/sì in the phrase má saÓar ég a-ab- 13 These two references are missing in Kleinerman ba si-ke4-de “to pack clay (for) the sea dike on the and Owen 2009, 253, but in no. “106:36” listed boat” (or: “to sink boats with clay for a sea dike”?); there ba-al does not appear in the transliteration of si-g is used with both má(-a) “to fill on the boat” the text. and saÓar “to pack clay” in GarÍana (see ibid. p.344f.). 14 In OrSP 47-49 211 (CDLI P 125101) -da is the co- 17 BDTNS has also si-ge-dè in TMH NF 1/2 69 mitative case: i7 ba-al-la-da Íu-zu ™á-ra-(ta) “(hirelings) who helped at the excavation of the river.” (CDLI P 134381) and RA 19 41 no. 50 (CDLI P 342 Walther Sallaberger

SILA11-™á-dè (I) “to knead dough”; no NAª4-dè (3™, including 2™ in no. 92), comparable forms NAª4-™á-dè (14™), al-NAª4-™á-dè (2™) (I) fiÚM-mu-dè (I) “to give”; thus regularly Ur “to crush”; no comparable forms fiÚM III, exceptionally -e-dam CST 537, SÀG-dè (5™), SÀG-ke4-dè (2™ in text no. fiÚM-e-dè RA 19 41 n.50 (CDLI P12 7692) 250) “to beat”; cf. for the final consonant TA°-Óe-dam (I) “to add” (no. 1381); TCS 1 173:5 Óa-sàg-ge BDTNS lists taÓ/Óe-dam/da/dè (9™), TAB-bé-dè (I) (10™), TAB-ba-dè (1™) (I) “to taÓ-e-dam (3™), taÓ-dam (5™), taÓ-Óu- pile, stack”; cf. TAB-bé-dè AOS 32 P 1, dam (sic, 1™) (accessed 12/02/2009). TAB-bé-da TCL 2 5540, TAB-dam, e.g., te-™e6-dè (III) in: Íu te-™e6-dè “to receive”; NRVN 116; see Yoshikawa (1988, 72) cf. Íu ti-dè NATN 59:9 (Yoshikawa For similar variants cf., e.g., 1988, 62), Íu TI-dam; Íu TI-™e26-dè BPOA 1 382 D≤M-dè (MVN 14 323, Umma), D≤M-me- dè (UET 3 812//1498 vi, Ur), D≤M-e-dè zi-li-dè (1™), zi-li-dì(TI) (2™), zi-lí-dè (3™) (SNAT 535 e.a., Umma) = *zil-e-dè/zi-le-dè (I) “to strip”; only zi-la “stripped off” attested in Ur III SAR-dè (UET 3 620), SAR-re-dè (UET 3 texts. 666), both at Ur zi-re-dam (I) “to cancel”; regularly Ur III 2.2.2 Discussion GROUP D': LOGOGRAPHIC WRITING OF VER- 2.2.2.1 Verbal Stems with u and Vowel BAL BASE WITHOUT OR WITH PHONOGRAPH- Assimilation at GarÍana IC RENDERING OF MARÛ VOWEL *-E: The GarÍana references hint at a vowel assimi- Despite the homogeneity of the GarÍana lation that pertains to the marû -e before the archive some of these infinite marû forms occur -d morpheme, as it was until now attested only in various forms (cf. already in group B: ì-li-dì/ for certain verb stems with the vowel -u-. The ì-li-dè, la-Óa/Óe-dè): latter vowel assimilation is a regular feature: in the Ur III data base BDTNS, the sequence ÀR-dè (2™), ÀR-ra-dè (4™) (I) “to grind”; (-)Íúm-mu(-) “to give” resulted in 545 hits cf. ara3(ÀR)-dè SAT 2 1123, SAT 3 1643; ÀR-e-dè SAT 3 2128, UET 3 151(?) (not checked); for (-)Íúm-e(-) only three unequivocal examples were found (search of BALA-dè (6™, including 4™ in no. 609), 2009/11/25). Since a writing *Íúm-me is BALA-a-dè (2™, no. 39:38, no. 45:18), apparently not attested, the writing Íúm-e pri- BALA-e-dè (2™, no. 390, 1523), BALA-a- 18 marily indicates the morpheme -e and e-dè (4™, nos. 200–203) (I) “to transfer, marû BALA does not aim at a phonographic rendering of to rotate”; cf., e.g., -dè RTC 333, 20 BALA-e-dè RTC 336. 341 (Yoshikawa the respective vowel. 19 1988, 58), UTI 5 3191 ku-bu-dè (gu5-bu-dè) = *gub-bu-dè

GA6-dè (4™), GA6-™á-dè (ca. 55™) (I) “to car- (group B), ku-ur2-gu-ru-dè (B), ù-ru-dè ry”; cf. in Ur III texts reduplicated ga6- = *uru3-dè (B), ku5-ru-dè (D), Íúm- ga6-dè (thus also, pace BDTNS, CST mu-dè (B); cf. Ur III gub-bu-dè, su-bu- 627) dè

20 127692); but in both cases si-ge4-dè (ge4, not ge) is The writing of the underlying morpheme and not possible or even probable according to the photo- its actual realization was described as “graphie mor- graphs. phophonologique” by Attinger (1993, 99ff.); the 18 On the various forms cf. Heimpel (2009, 242). term “morphographic” is a transposition of the ex- pression to be used together with “logographic” 19 BPOA 1 964: 4 ™uruÍ .... 11 tu Íe bala-a-da ummaki 9 and “phonographic.” (transliteration only) is unclear to me. Sumerian Language Use at GarÍana 343

2.2.2.2 Verbal Stems with a unknown from other Ur III documents, as attest- Besides the already known change of -e-d > ed by the following examples: -u-d, the GarÍana texts show also -e-d > -a-d á™: á-™á-dam Nippur, NATN 134 after a stem containing a as the only or as the (Yoshikawa 1988, 73) last vowel (sila11.™); the attested verbs end in ba-al: ba-al-e-dè Umma, TCS 1 216; nu- H/’, Ó, b, ™, r. As it is the case with verb stems ba-al-e-da ªirsu, AOAT 25 444 no. 9; with u (e.g. du10.g-e-d-), certain “/a/ verbs” ba-al-a-dè Umma, TPTS 2 5:4 do not follow this vowel assimilation, namely sàg-e-d- and kala.g-e-d-. dab5: dab5-ba-dè in Irisa™rig, examples cited by Heimpel (2009, 62f.); other Ur III

(a) Vowel assimilation -e-d > -a-d without texts have dab5-dè, or reduplicated dab5- variation: dab5-dè (Yoshikawa 1988, 57) SILA11-™á-dè (D) In conclusion, there is sufficient evidence ÀR-dè (2™), ÀR-ra-dè (4™) (D') for a variant -a-d-e < -e-d-e in the Ur III period, which can be understood by analogy with GA6-™á-dè (ca. 55™), GA6-dè (4™) (D') two related phenomena, the vowel assimilation NAª -dè (3™, including 2™ in no. 92),21 NAª - 4 4 -u-d-e < -e-d-e and the sporadically attested ™á-dè (14™), al-NAª -™á-dè (2™) (D'). 4 assimilation -e > -a in words ending in _a in Note that ™á could also be read ™e26, but for the ergative case (ama-a; see Attinger 1993, an /e/ vowel GarÍana texts would probably use 211). However, contrary to -e > -u, which ™e6 instead, as shown by the example si-™e6-dè characterizes the marû stem of class I verbs as (group B). well (Íúm-mu, gub-bu), -e- > -a- is restricted (b) Vowel assimilation with a/e variation to the combination _a_-a-d-e (< _a_-e-d-e); thus any confusion with nominalized or infi- BALA-dè (6™, including 4™ in no. 609), nite perfective forms ending in -a is avoided. BALA-a-dè (2™, no. 39:38, no. 45:18), BALA-e-dè (2™, no. 390, 1523), BALA-a- 2.2.2.3 Verbal Stems with i and e e-dè (4 , nos. 200–203) (D') ™ It is more difficult to discover a differentiation la-Óa-dè (2™), la-Óe-dè (1™) = *laÓ(4)/5-e-dè between /e/ and /i/, since in the cuneiform (B) writing system the same sign often covers both (c) Without vowel assimilation the e and the i vowels (e.g. ni = ni, né; ri = ri, re). In the Sargonic period “sign pairs” of SÀG-dè (5™), SÀG-ke4-dè (2™ in text no. 250) (D') closely related values were used to distinguish between CV syllables with /e/ or /i/ (e.g. gi4 KALA KALA . -ke4-dè (D) vs. ge, le vs. lí; see Sommerfeld 1999, 18–21; LÁ-e-dè (D) Hasselbach 2005, 32–35. 39–73). Such rules TA°-Óe-dam (D) seem not to apply in the Ur III period, espe- TAB-bé-dè (I) (10™), TAB-ba-dè (1™) (D'). cially in the syllabary of GarÍana (note, e.g., Here the /a/ coloring remained an that GarÍana writes li-iq-tum for standard Ur III exception. lí-iq-tum, Óu-ri-um for Óu-rí-um). Most of the /-a-d-e/ spellings appear as variations (a) i probable of logographic writings or of a morphographic ì-li-dì (ca.34™), ì-li-dè (3™ in text no. 53) = rendering as -e-dè, but la-Óe-dè/la-Óa-dè points *íl-e-dè (B) to a variation as linguistic reality. Although only zi-li-dè (1™), zi-li-dì(TI) (2™), zi-lí-dè (3™) (D) the GarÍana texts offer a broader basis for this phe- (b) e probable: nomenon of vowel harmony, it is not completely si/se-™e6-dè = *Íe™6-e-dè (B)

21 The reading na™4 is based on na™a4(KUM). 344 Walther Sallaberger

te-™e6-dè (D) 3. Additional Notes on the Syllabary at (c) undecided cases: GarÍana zi/ze-ri/re-da A study of the phonograms that was used by

si-ke4-dè (D) the GarÍana scribes provides another clue that sì-ke4-dè (D) the spoken language had a strong impact on the With the i sequences one notes that the orthography and resulted in the substitution of vowel assimilation in rare cases even affects the traditional standard writings. The deviations final directive -e in ì-li-dì and zi-li-dì (Owen from the standard orthography constitute further evidence for the sound changes that both and Mayr 2007 transliterate de9 for dì). This fact and the writing zi-lí-dè are the main arguments Sumerian and Akkadian underwent before or to suggest a similar vowel assimilation for these during the early Ur III period (see, e.g., Jagers- two verbs. ma 2000 [2005], Keetman 2004a, 2004b). The case of sì-ke -dè is hard to decide. ke 4 4 3.1 Specific Sumerian Phonemes represents generally the combination of geni- At GarÍana, the Sumerian phoneme ™ is always tive and ergative, _k-e, and at GarÍana ke4 is also an unorthographic writing for kè(AK). The correctly employed also in phonographic and Old Babylonian sign-list Proto-Ea 243 gives in unorthographic writings (signs ™á/™e26, ™e6, both gi-i (1™) and ge-e (2™) for the word “reed [™u10]). The so-called dr phoneme appears in mat.” The rendering -ke -dè represents a the traditional orthography rá(DU) with the 4 marker -a of infinite perfective forms, e.g., unique feature of the GarÍana orthography, 22 keÍe2-rá “bound,” ku5-rá “cut,” but before -e- whereas elsewhere the sequence -gi4-dè is used d-e it appears as [r] in ku5-ru-dè (see above). for both si-g and sì-g. Since gi4 represented /gi/, not /ge/ in the Sargonic period, this could be The use of phonograms for stops in unor- thographically written Sumerian is consistent taken as an argument for a reading sì-gi4-dè instead of sì-ge -dè; for GarÍana, however, this with the Akkadian system at use at GarÍana, as 4 far as it can be reconstructed. ti was used for the would force us to assume a unique “ki4” at use verb de “to bring” and for -dè ([di]), and sim- besides ke4. 6 The unorthographic phonographic writ- ilarly in Akkadian one finds dì(TI)-um, PN i- ings of GarÍana thus express a vowel assimila- dì(TI)-ìr-ra. Rarely does the phonogram di(DI) d tion that was elsewhere mostly hidden behind appear in Akkadian as in the PN ÍamaÍ-di-in, the standard logographic rendering. As we but it had not yet replaced dì(TI). Similarly the have seen above, the unorthographic replace- variations between ta- and tá- are a reflex of the ment of logograms by phonograms appears undergoing replacement of tá(DA) by ta (see the mostly if lexemes are embedded in grammatical index of names Kleinerman and Owen 2009, forms with prefixes and suffixes. The regularity 615). These writing conventions conform to of the vowel assimilation with certain verbs and the tendencies evaluated by Keetman (2004a). the unorthographic rendering both prove that A unique feature of GarÍana orthography is the vowel assimilation was based on the actual the use of -ke4 in -ke4-dè in infinite verbal spoken language. The Sumerian that was writ- forms (_g-e-d-e, see above). In Akkadian con- ten by the GarÍana scribes is thus based on the texts both ki and gi occur at GarÍana (note, e.g., living, actual use of Sumerian, as the probably below p.357f. ar-ki-LUM vs. Ía-pá-ar-gi-lum). bilingual inhabitants of GarÍana heard it in their Does this mean that -ke4 marks a specific Sum- environment of the Umma province. Further- erian vowel quality of /ke/ vs. /ki/ (see above more, since progressive vowel assimilation on -ke4 in _k-e-d-e, and note unorthographic does not occur in Akkadian, an imposition on ke4 for kè)? Or should we assume that the velar the side of the scribes can safely be excluded. stop was considered voiceless both in ke4 for kè

22 ™eÍ No. 855 reads gu-za sìr-da for “keÍe2-da.” Sumerian Language Use at GarÍana 345

“to make” and in sì-ke4-dè for standard sì-gi4- PN i-ta-è-a “who came forth from her,” dè by the Akkadian scribes and they hesitated more often at ªirsu and Umma, in-ta-è- to employ the sign KI? The second option a especially at PuzriÍ-Dagan becomes more likely if one considers hints at nin-i-ti Umma passim, // lugal-i-TI ªirsu Akkadian perceptions of Sumerian phonemes (e.g. TCTI 1 732), Umma (e.g. Torino 2 as identified in the following paragraphs. 703), which should probably be inter- 3.2. Prefix /i-/ Written ì- and i- preted as nin/lugal-i-tìl “the Lady/Lord The Sumerian verbal morpheme /i-/ appears became/was life” by comparison with in two orthographies even in similar forms: Presargonic nin-ì-tìl (BIN 8 39 iii 25) or the writing lugal-ì-tìl (e.g. ªirsu RTC i-: u4 dur11-ra i-me (no. 466), i-[me-a] (no. 399 iv 6; more often at Ur) 444) The writing of the verbal prefix with i- u4 é NN-Íè i-re-Ía-a (no. 480), u4 kaskal-Íè appears more often in Old Babylonian legal i-re-sa-a (no. 529) documents (but not yet in the Isin craft archive ™eÍ u4 guÍur i-im-de6-a of the time of IÍbi-Erra and ﬁu-iliÍu), in royal ì-: ì-™ál, ì-Ói, ì-lá, ì-™en-na-a, ì-im-™en-na- inscriptions beginning with the second Isin

a king ﬁu-iliÍu (RIME 4.1.2.1:22 i-ni-ib-ku4- The same morpheme /i-/ is written with ku4-a), and in Old Babylonian literary texts. both I and NI(ì) as most clearly shown by the The sound change in Akkadian (yi > i) led variation of i-im vs. ì-im-. Therefore, the to a phonological equivalence of I (yi > ’i) and former distinction between I for /yi/ and NI(ì) NI (’i), so that now both signs could express the for /’i/ was no longer valid (note also writings sound /(’)i/ in Sumerian. As a result, the value like i-la-ak-nu-id, ilak-nu’’id, whereby ’il “god” of a phonogram was defined by its use in writ- is written now with i-, formerly /yi/). The ing Akkadian. This implies that writing, to a interchange of I and NI in the Ur III period is large extent, represented language directly in its known from Akkadian contexts (Hilgert 2002, phonological (or phonetic?) form. 120). However, outside GarÍana the Sumerian verbal prefix was always rendered by the tradi- 3.3 SV and ﬁV Series: Plural Suffix -eÍ + tional writing ì- in Ur III documents, whereas Nominalizer -a Written -sa and -Ía i- is only rarely attested in some Sumerian per- The sequence -eÍ-a, suffix of the 3rd person sonal names: plural agent in the preterite -eÍ plus nominalizer a, is written -Ía in Sumerian, but in GarÍana this traditional writing can be substituted by -sa. Some examples:

(™uruÍ) má PN G.-ta U.-Íè in-G≤D-Ía-a (™uruÍ) má ... G.-ta P.-Íè in-G≤D-sa-a “men, who hauled the boat of PN from G. to “(men) who hauled the boat from G. to P.” (no. U.” (no. 238) 253 etc.)

u4 é NN-Íè i-re-Ía-a “when they went to the u4 kaskal-Íè i-re-sa-a "when they went on expe- house of NN” (no. 480) dition" (no. 529)

in-TI(di3)-Ía-a (no. 1062) “who carried” mu-un-TI(di3)-sa-a “who brought” (no. 243, 304)

u4 bàd in-dù-sa-a “when they built the wall” (no. 396 etc.)

u4 igi Íu-kab-tá inda3 in-gu7-Ía-a “when they ate bread in front of ﬁu-Kabta” (no. 479) 346 Walther Sallaberger

The plural morpheme alone is written -eÍ, -éÍ, probably an existing difference between /s/ or -uÍ (in-tùm-uÍ, e.g., no. 27). and /Í/ was only imperfectly perceived by the This orthographic variance cannot be ex- Akkadian scribes at GarÍana so that they could plained on the basis of the Sumerian evidence use both signs interchangeably in phonograph- alone, since there SA and the SV series of pho- ic spellings.27 Interestingly, they employed SA, nograms and fiA and the fiV series were always which was not attributed a specific value in distinguished; note at GarÍana, e.g., Ía-ra-ab- Akkadian, as if a “foreign” phoneme should be du for Íár-ra-ab-du, but sa “bundle.” In Akka- represented. dian, however, the former distinction between This case is slightly different from the one the Sargonic fiV- and the SV-series, represent- described before for the coincidence of i- and ing etymological t (θ) and etymological Í, s, ì- as writings for the verbal prefix: The latter respectively, had disappeared by the Ur III feature continued to be used in the Old Baby- period; now both series represented /Í/.23 lonian period, whereas the permutability be- Since Akkadian etymological s was represented tween SA and fiA and the use of SI for /si/ and by signs of the ZV series (ZA, ZI, ZU), the SV /Íe/ in Sumerian remained a local phenome- series was deprived of any function in Akkadi- non of the GarÍana documents. For an evalua- an. So in the writing of Akkadian words and tion of the evidence see below p. 364f.. names at GarÍana, SA is never attested, only fiA.24 The equivalence of the fiV series and the 4. Observations on the Sumerian Grammar at SV series appears in the exchangeability of SI GarÍana and fiI (in: ták-si-ru-um, ták-Íi-ru-um) or even SI The characteristic style of administrative docu- and fiE (in the personal name si-um = Íe-um, ments with their short notes and asyntactic lists name of the father of Adalal). Both SI and fiI hinders an investigation of the Sumerian gram- appear in Akkadian names and words.25 mar. The most obvious field of study, the From the viewpoint of an Akkadian sylla- grammatical relations including the case system bary, one can easily understand that SA repre- and its reflections in the verb, is difficult to sented not only the traditional Sumerian value evaluate, since administrative documents often /sa/, but also /Ía/ and that SI was used for /si/ do not write case markers. Case markers are and for /Íi, Íe/ (note, e.g., SI for the verbs si-g especially rare with personal names, and in and Íé-™, respectively, the latter in the unor- many frozen phrases the genitive case was sys- 26 thographic writing SI(Íé)-™e6-dè, see above). tematically disregarded. In Sumerian everything points to a continua- The observations collected here are based tion of the difference between /s/ and /Í/, on a repeated close reading of the GarÍana written by the SV and the fiV series, respective- texts. In most cases I have tried to find parallel ly, although admittedly an unequivocal differ- examples from other Ur III texts in order to entiation remains a thorny issue for many investigate if a specific phenomenon occurs lexemes and in diachronic perspective. Thus, only here, although unfortunately this was not

23 Hilgert (2002, 128-133); for the continuation of ﬁI: ták-Íi-ru-um; personal names: a-Óa-am-ar-Íi, é- this pair in southern Old Babylonian, see Kogan/ a-nu-ùÓ-Íi, Íi-ba-aq-r[a-at], Íi-ma-tum, Íi-nu-ri, Íi-pu- Markina (2006, 569). uz-ri. 24 ™eÍÍa-bar-gi-lum; personal names: Ía-lim-be-lí, Ía-lim- 26 In a somewhat similar way the Sumerian simda,

dú-ri, lú-Ía-lim etc., pí-Ía-aÓ-DINGIR, pi5-Ía-Óa-lum, loanword from Akkadian Íimtum “brand, branding Ía-at-èr-ra and other Ía-at-names; ip-qú-Ía, ù-qì(?)- iron,” appears both as sim-da and Íim-da. pá-Ía, note also Ía-Ói-ti, Ía-at-ri-tum, al-la-Ía-ru-um, 27 Cf. also the unorthographic writing of ur-dasar-lu- Ía-bar-zi-na-at, taÓ-Ía-tal, si-ri-Ía. Ói as ur-Ía-lu-Ói in some texts (nos. 43–45). 25 SI: si-ì-tum, ták-si-ru-um; personal names: i-ˇib-si-na- at, ì-lí-si-na-at, si-mu, dub-si-ga, si-ri-Ía, la-te-ni-si, ku-un-si-ma-at. Sumerian Language Use at GarÍana 347

possible for all features treated below. A more ba-an-Íi-: sa10 systematic study of the Sumerian grammar at ba-na-: Óa-la (for Óal), zi (zi-g) GarÍana may perhaps be helpful, but it would ba-ra-: du , Íe™ gain in importance only by a careful compari- 8 6 son with the bulk of the Ur III documentation. bí-ib-: sù-ub bí-in-: *a5 (bí-na for bí-in-a5, no. 4.1. Finite Verbal Forms at GarÍana 1048:2), du11, gi4 In administrative texts verbs occur mostly in ga-ra-ab- su (no. 1499, GarÍana?) infinite forms, of which a special type has been Óa-ba-: -tur-re treated above in section 2. Finite forms are rar- Óa-ba-ab-: Íu -ti-™e er, but they occur also in the few letters and 26 legal documents. Generally, the verbal system Óa-ba-Íi-ib- Íu -ti (no. 1499, GarÍana?) seems to obey the Sumerian grammatical rules Óé-eb-: -sá-e apart from the few exceptions that will be not- Óé-en-: -™á-™á ed in the paragraphs that follow. As an exam- Óé-na-ab-: -Íúm-mu ple, I refer to a lexical investigation of the verbs ì-: dab , ™ál, ™en, Ói, lá, tab; in no. used for “to bring” (Sallaberger 2005), where 5 1511: -dan6-dan6, -e-re, U+NU the GarÍana evidence helped substantially to (GarÍana?) solve the puzzle not only because of the unor- i-: me, -re-Ía/sa-a (™en, pl.) thographic writings, but also because the subtle differentiations of the plural forms depending ì-im-: ™en on the object for the verbal stem and the agent i-im-: de6 for the suffixes were strictly followed. ì-in-: DU? Since it cannot be the purpose of this study ib-: -kur-re (for Íu gur?), -tab-bé to unravel the verbal morphology, the differ- (both no. 1511, GarÍana?) entiation and flexible variation of verbal forms íb-: ga6, gíd, gu7, na™ (all collective, met in the GarÍana corpus are illustrated with trans. pret.), su-su (su-g, trans. the help of two lists, one of the prefixes and the present-future), ra, sá (intrans.) other of the lexemes combined with more than íb-ta-: Ói, zi one prefix chain. The few examples of pre-pat- terned phrases that are the topic of the next in-: (sing.) da (no. 1463), du8, ge- paragraphs have to be read against the picture en, gu7, ku5, lá, la-aÓ (for laÓ4), obtained by these lists. pà, sù-ub (for Íub?); (plural) Prefix chains of finite verbs (the list in- -dù-sa-a, -ge-né-eÍ, -gíd-sa-a, TI cludes references from the text nos. 1462–1527, -gu7-Ía-a, - -Ía-a (for de6), “Tablets of Unknown Provenance” and “Sup- -tùm-uÍ (for túm) plemental Texts” with the exception of 1466, in-da-: ™ál, ™e26-en (™en) 1478, 1479, 1490, 1491, 1493 from Irisa™rig/ in-na-: ge17-ga (intrans.) ml-Íarr⁄k‹ and elsewhere): in-na-an-: Íúm (trans.)

al-: -na™4-™á, -ús-sa in-Íi-: sa10 (no. 1477)

ba-: a5/aka, dù, gi-in/ge-en/gen7 (= mu-un-: (intrans.) ™ál (no. 1057), -til-la- ge-n), gul, ™ar (no. 1259, 1369), ta (for tìl), (trans.) (sing.) TI (for Ói, Óu™, keÍé, Íu ti, úÍ, zal, zi de6), (plural) -gíd-Ía-a, -li-sa-a (for íl?), -TI-sa-a (for de6) ba-a-: ge-en/gen7 (= ge-n), ™ar, keÍé mu-na-: ba? ba-ab-: gu7, sù-ub, ÍeÍ4, taÓ na-ba-ab-: -Íúm-mu-ne ba-an-: dé, ku4 (both intrans.), -du11-ga (no. 1053?), Íu ti, tùm (for na-mi- -gur-re (no. 1499, GarÍana?) túm), zuÓ (all trans.) nam-mi-: -gur-re 348 Walther Sallaberger

nu-: ™ál, tuku, zi-g su ga-ra-ab-su (no. 1499, GarÍa- na?), íb-su-su nu-ub-: -gi4-gi4-(da) (no. 1477), ra, tuku (also no. 374?) sù-ub: ba-ab-sù-ub, bí-ib-sù-ub

nu-ub-ta-: -gi4-gi4-(da-Íè, no. 1048:8), zi Íúm: Óé-na-ab-Íúm-mu, in-na-an- nu-um-: ™ar (intrans.) Íúm, na-ba-ab-Íúm-mu-ne nu-un-: éÍ (no. 1057:7!) ti: Íu ba-an-ti, Íu ba-an-ti-éÍ, Íu

nu-un-ni-: ku4 Óa-ba-ab-ti-™e26, Íu Óa-ba-Íi- ib-ti (no. 1499, GarÍana?) ù-na-a-: du11 ! (u4-)um -: de6 (no. 1173, 1361!, “u4 dub tuku: nu-tuku, nu-ub-tuku túm”) túm: ba-an-tùm, in-tùm-uÍ, in-la-aÓ Verbs with more than one prefix chain: zi-g: ba-zi, ba-na-zi, íb-ta-zi, nu-zi, a5/ak: ba-a5/aka, bí-na nu-ub-ta-zi

de6: i-im-de6-a, ì-in-DU?, in-TI-Ía- a, mu-un-TI(-sa-a), (u -)um!- 4.2 Prefabricated Verbal Forms in the Year 4 Dates and Elsewhere de6 (no. 1173, 1361!, “u4 dub túm”) Most finite verbal forms are attested in the date dù: ba-dù-a, in-dù-sa-a formulae and as administrative key terms as Íu ba-ti or ì-dab . The year dates can be phrased in du : ba-ra-du , in-du 5 8 8 8 two ways, with the king named (in the ergative du : ba-an-du -ga (no. 1053?), bí- 11 11 case) and the verb with the ventive prefix in-du , ù-na-a-du (read ba- 11 11 mu-, or without the royal name and thus with an-zuÓ for “ba-an-du ”); nu- 11 the “medium” prefix ba-. However, at GarÍana un-éÍ (no. 1057:7!) scribes more often write non-standard mu- gíd: íb-gíd, in-gíd-sa-a, mu-un- Óulu for ba-Óulu in the intransitive construc- gíd-Ía-a tion. Although mu-Óulu (mu-n-Óulu-ø, “he ge-n: ba-gi-in/ge-en/gen7, ba-a-ge- destroyed it”) would be somehow grammati- en/gen7, in-ge-en, in-ge-né-eÍ cally correct, if we assumed an omission of the gi4: bí-in-gi4, nu-ub-gi4-gi4-(da) noun in the ergative in the sentence, such an (no. 1477), nu-ub-ta-gi4-gi4- abbreviated phrase is not used in Ur III date (da-Íè) formulae. gu : ba-ab-gu , íb-gu , in-gu -a, in- ki 7 7 7 7 mu LAND (= ma-da za-ab-Ía-li , si-mu-ru- gu -Ía-a ki 7 um ) mu-Óulu (incorrect): more than ™ál: ì-™ál, in-da-™ál, mu-un-™ál (no. 160™ 1057), nu-™ál mu LAND (= ma-da za-ab-Ía-liki, si-mu-ru- ™ar: ba-™ar (no. 1259, 1369), ba-a- ki um ) ba-Óulu (correct): ca. 49™ ™ar, Óé-en-™á-™á, nu-um-™ar “year: LAND (= Land of ZabÍali, Simur- ™en: ì-™en-na-a, i-re-Ía/sa-a, ì-im- rum) was destroyed” ™en-na-a, in-da-™e26-en Ói: ba-Ói, ì-Ói, íb-ta-Ói With the name of the king, mu-Óulu is keÍe : ba-keÍe -rá, ba-a-keÍe standard (188 hits in BDTNS), but ba-Óulu is 2 2 2 written once although the king is named (no. ku : ba-an-ku , nu-un-ni-ku 4 4 4 294). lá: ì-lá, in-lá With the verb dù “to erect,” the distribu- ra: íb-ra, nu-ub-ra tion at GarÍana is even more inconsistent. A sá: Óé-eb-sá-e, íb-sá quick count of the results obtained by BDTNS

sa10: ba-an-Íi-sa10, in-Íi-sa10 (no. (accessed 2009/12/06), disregarding recon- 1477) structed prefix morphemes, gives the following Sumerian Language Use at GarÍana 349

distribution (note that mu-ne-dù for fiS 6 does text (no. 1329); see the verbal forms listed not show any variation): above. mu KING-e OBJECT mu-dù (correct): 70™ Given the fact that the year date variation does not show any discernible pattern, the sec- mu KING-e OBJECT ba-dù (incorrect): 7 ™ ond option seems far more likely: a year date “year: KING erected OBJECT” could be perceived as a formula that was auto- mu OBJECT mu-dù (incorrect): 46™ matically written down without reflecting its mu OBJECT ba-dù (correct): 31™ correct form and thus its literal meaning. The grammatical peculiarities of stock phrases thus “year: OBJECT was erected” should not be taken as reflecting the command Similar errors can appear sporadically in the of Sumerian grammar, although it remains Ur III corpus. In the tables prepared by remarkable that the royal year dates were treat- Schneider (1936) only eleven aberrant forms ed that way. are listed from the more than 11,000 Ur III tab- The interchange of the verbs with mu- and lets published by that time: ba-prefixes in the date formulae seems to occur mu- instead of ba- (intransitive formula): independently of an actual experience of lan- Schneider (1936, 21): fiulgi year 44 B.tt) guage. A similar phenomenon can be discerned (1™), ibid. 22: fiulgi year 45 B.j) and m) only in two verbal phrases that appear more (2™), ibid. 36 fiu-Suen year 9 B.f) (1™), cf. frequently in the GarÍana corpus, namely ba- ibid. 31 fiu-Suen year 2 B.p) in- instead na-ha-la and al-na™4-™á-dè. The indirect object of ba-, cf., e.g., Syracuse 379 of the verbal form ba-na-ha-la is invariably singular although one expects mostly a plural ba- instead of mu- (presence of KING in *-ne-(a)-: ergative case): Schneider (1936, 24): Amar-Suena year 2 B.c) (2™), ibid. 25 Íidim ù lú-hu™-™á-e-ne ba-na-ha-la (e.g. Amar-Suena year 3 B.e) (1™), ibid. 32 fiu- no. 424; cf. Kleinerman and Owen 2009, Suen year 4 B.e)-f) (3™), ibid. 33 fiu-Suen 316–18. Note that the dative case writ- year 6 A.f) (1™), cf., e.g., NATN 202, ten -er would hardly be expected in an BPOA 6 9 administrative text) Seen in this light, the extremely high por- “it was distributed (to him) to the builders tion of incorrect forms in the GarÍana texts is and hirelings” noteworthy and cannot simply be explained as Even more perplexing is the rare form al-

resulting from the scribe’s inattentiveness. Was na™4-™á-dè “to crush,” where the verbal prefix the differentiation between intransitive ba- and al- appears in a strange combination with the transitive mun-forms losing its binding force? infinite -e-d-e form. This form is attested two Or did the formulaic character of the year dates or three times at GarÍana (nos. 67, 70, and, lead to the distortion, since they were not treat- probably, 217), whereas a correct na™4-™á-dè ed as “real” language but as pre-formulated appears 14 times. All references stem from stock phrases? A similar free variation between identical contexts; the object is invariably munu4 intransitive and transitive, ventive and non- “malt.” Nos. 67 to 69, for example, date to ventive forms seems not to be attested for verbs three consecutive days (ﬁS 6/08/28–30), the in phrases embedded in the main body of the same persons are named, but incorrect al-na™4- texts: the verb gíd “to haul (boats)” appears ™á-dè appears in no. 67, correct na™4-™á-dè in with ventive in the case of a movement toward nos. 68–69. GarÍana (mu-un-gíd-Ía-a), but without ventive The incorrect variant cannot be explained if the boat is moved from GarÍana to another within standard Sumerian grammar. At GarÍa- (in-gíd-sa-a); and sù-ub appears both in a tran- na the prefix al- is used only in idiomatic ex- sitive (bí-íb-sù-ub, collective) and in an intran- pressions concerning the preparation of food:

sitive-passive (ba-ab-sù-ub) form within one al-nag4-gá “crushed,” al-ús-sa “processed” (see 350 Walther Sallaberger

the phrases collected by Kleinerman and Owen this variation, which may be ascribed to the 2009, 364f.). Apparently al- was (partly) de- presence of various active scribes (see Heimpel prived of its function as a “stative” prefix, and 2009, 27), proves how flexible they were when

so the nominalized form al-na™4-™á could be treating the Sumerian language. So apparently reinterpreted as an infinite form *alna™+-a each scribe could write according to his pref- “crushed”; and the suffixes -e-d-e could then erences and so we actually dispose of docu- be attached to this seemingly new verb /alna™/. ments of an individual language use. This error is restricted to rare cases and can thus surely be attributed to a single scribe. 4.4 Plural Formation as Calque of Akkadian? Although a single reference is of little value for 4.3 Inconsistency of Plural Formation grammatical observations, the following exam- In the relatively small organization of GarÍana ple has to be considered in the context of a pos- even details of the transactions were described, sible Akkadian imposition of the scribes on such as the exact work performed at various Sumerian. The usual phrase ka-ga-na ba-ge-en building steps. In the larger archives, one more “it was confirmed in his (own) statement” has often finds simply lists with numbers of persons once been used in the plural, where the text and at the most some short notes on the work reads to be done (cf., e.g., the lists of the men at work ka-ga-na-ne-ne ba-ge-en (no. 1062:9) at ªirsu or at the Ur or Isin craft archives). This ka.g-an(i)-(a?)-anene ba-ge.n-ø tendency to provide more-detailed descrip- tions implies a greater variation in phrasing. An mouth-POSS.3SG-(LOC?)-POSS.3PL obvious point in this regard is the inconsistency MED-confirm-S.3SG in the formation of the plural, both as personal instead of the correct: plural and as neutral (non-personal) collective. *ka-ga-ne-ne-a Both forms are grammatically correct and are ka.g-anene-a well attested in Ur III texts. mouth-POSS.3PL-LOC Personal plural: One may simply suppose a temporal inatten- d ™uruÍ má ... é Íara2-ta G.-Íè mu-un-gíd-Ía- tiveness of the scribe, who did not erase his a “men, who hauled the boat ... from the incorrect -na- (= -ani-a “in his ...”), which he ﬁara temple to GarÍana” (no. 529) had written out of habit. However, one could mu lú Óu™-™á-e-ne-Íè u4 igi PN inda3 in- also assume an unconscious calque on the gu7-Ía-a “because of the hirelings, when Akkadian plural formation: the plural of corre- they ate bread in front of PN” (no. 479) sponding ina p‹-Íu is ina p‹-Íunu with suffixation Collective: of the plural marker. A similar example of an 3 ™uruÍ ... N.-ta B.-Íè má PN íb-gíd “3 Akkadianism in an Old Babylonian literary text men ... hauled the boat of PN from Nip- has been pointed out by Wilcke (1998, 462): pur to B‹t-ﬁu-Suen” (no. 220) the form Íà eren2-na-ka-ni “in the midst of his troops” in GilgameÍ and Akka lines 81 and 99 inda ugula lú Óu™-™á-e-ne íb-gu “bread: 3 7 (instead of *Íà eren -na-na-ka/ke ) follows the the foremen of the hirelings ate it” (no. 2 4 sequence of suffixes of Akkadian . 385) ina libbi ‰⁄b‹Íu Whereas the variation in the year dates led 4.5 Some Observations on the Case System to incorrect forms, here the variation, which Generally, the case system of Sumerian is well does not seem to imply any difference in mean- reflected in the GarÍana documents, especially ing, remains within the traditional limits of the if one takes into account that case suffixes, not language. PuzriÍ-Dagan documents or legal only the genitive suffix, can be missing in ad- documents prefer the 3rd person plural, where- ministrative documents, primarily with proper as the 3rd neuter collective was especially pop- names. In the following, I list deviations from ular in administrative texts from Umma. Thus the standard use of case suffixes. These single Sumerian Language Use at GarÍana 351

observations do not pretend to cover every texts (“was placed upon...”), although it cannot aspect of the case system at use at GarÍana. be excluded that here the terminative denotes the purpose of the placement and the phrase 4.5.1 -ta for Comitative -da thus means “OBJECT was deposited for BUILD- Once at GarÍana the case suffix -ta is used for ING” (thus Heimpel 2009, 42). But in this case the comitative -da, which seems not to be the “préfixe local” (Attinger 1993, 240ff.) -y/e- attested elsewhere in the corpus: of ba-a-™ar (/bay™ar/) would be devoid of any function. An abbreviation for ki(-e) ba-a-™ar 1 ™uruÍ ... PN lú-kí™-ge4 lugal-ta ... in-da- ™e -en “it was placed on the ground/was founded” is 26 extremely unlikely. “1 man ... he went together with the royal messenger PN.” (bitumen) (ﬁID-tum of buildings) ták-si-ru- -Íè ba-a-™ar (no. 1260, cf. 1299) Instead of assuming a semantically incorrect um ablative -ta, which furthermore could not be “(bitumen) was placed upon/deposited for combined directly with a noun of personal/ the ﬁID-tum of buildings, on/for the human gender, it is more probable that the place of repair.” comitative -da was written -ta. Such an ortho- 210 sa ge-NE gu-kilib3 12 sa-ta dì-um é-ki- graphy has to be understood from an Akkadian mu-ra-Íè ba-a-™ar (no. 1325:4-6; cf. point of view, since Sumerian stops were all 1261, 1262, 1286, 1308 etc.; without ter- voiceless with a further differentiation, proba- minative no. 1294) bly aspiration. In the Ur III period, one began “210 bundles of ‘fire’ reed, bales of 12 bun- to write the sign da for the voiced alveolar stop dles, were placed on (or: deposited for) /d/ and ta for voiceless /t/ in Akkadian. the waddle-and-daub of the washing Interestingly, the same use of -ta for -da house.” appears in a text from the contemporary dug archive of Irisa™rig/ml-Íarr⁄k‹, similarly from 63 ù-lu-lu 2 kùÍ-ta a-za-bu-um Íà é-a-Íè an organization closely related to the state: ba-a-™ar (no. 1286: 7f.) “63 clay pipes of 2 cubits each were placed u4 alan lugal-ta im-da-e-re-Ía-a (no. 1478 ii 12) on (or: deposited for) the cistern within the house.” “when they went with (= escorted) the royal statue” Constructions with ba-a-™ar appear more often in Ur III texts, albeit mostly in the con- 4.5.2 Terminative for Locative struction a-gù PN-a-ka ba-a-™ar “it was In the GarÍana texts the verb ba-a-™ar, “it is placed on the account of PN.” But a search in placed upon,” is usually construed with a ter- BDTNS (2009/25/11) revealed no other minative at the noun instead of a locative or examples of the combination with the termina- directive. This use is especially clear in the con- tive -Íè except the probably adnominal context of the use of leather that was “placed on” struction UTI 5 3370 má ... GN-Íè ba-a-™ar another object: “placed on the boat to GN” and the unclear Ontario 2 222. 5 kuÍ udu e-rí-na ma-an-Óara 3-Íè ba-a- 4 The construction NOUN-Íè ba-a-™ar for an ™ar (no. 923:1-2; cf. also nos. 864, 939, expected NOUN-a (locative) ba-a-™ar, as it 968, 1517) appears in the frequent phrase má-a ba-a-™ar “it “5 tanned sheep skins were placed on 3 was placed on the boat,” has perhaps to be seen receptacles” in a wider perspective. The locative case, The same construction NOUN-Íè ba-a-™ar marked by a simple -a, was going to lose appears with the use of bitumen or with reed ground in the late Ur III and the Isin period for repair and building. These passages should against the more prominent case markers ter- be translated in the same way as the leather minative (-Íè) and ablative (-ta), which were 352 Walther Sallaberger

expressed by a whole syllable; they remained in Of course the change in some construc- use or even took over the domains of the loc- tions does not imply that the locative case had ative. A good example is the change of con- completely disappeared at GarÍana. It is still struction of ku4 “to enter”: the locative (-a) was generally employed with gub “to stand” (e.g. still in use in Ur III texts, e.g., é-gal-la ba-an- é-gal-la gub-ba “stationed in the palace,” no. ku4 “it was entered in the palace” (passim in the 506:23) or with ™ál “to be present, to exist” ﬁulgi-simtum archive). Not unexpectedly, (e.g. beÍe™ dub-ba-ka ™ál-la “present in the GarÍana already starts using the terminative: é- tablet basket” no. 1233:13). kiÍeb3-ba-Íè ba-an-ku4 “it was entered in the storehouse” (no. 1031; but ku4 with locative in 4.5.3 The Genitive Case: Presence, Loss and no. 1049 //1050!). In the Isin craft archive, Hypercorrection which dates only a few years later, after the fall As already mentioned in the introduction, the of the Ur III empire, and stems from northern genitive case often is not marked in Ur III Babylonia, one meets similarly both termina- administrative documents, but, as with all case tive é-gal-Íè ba-an-ku4 (BIN 10 117; even Íà é- suffixes, its usage is more consistent in legal gal-Íè ba-an-ku4 BIN 9 399) and locative é-gal- documents and letters at GarÍana and else- la ba-an-ku4 (BIN 10 189). where. Also within administrative texts the genitive is not completely abandoned and it appears especially in the combination with the locative case:

u4 siki-ba-(a)-ka Íu bar-ra (no. 234 etc.) u4 siki-ba-ak-a Íu-ø bar-a day wool-distribute-GEN-LOC hand-ABS open-PFV “released at the day of wool-distribution”

MAfi.EN.GAG lú-kaÍ4 ká é-gal-ka gub-ba (no. 529) MAfi.EN.GAG lú-kaÍ4 ká é.gal-ak-a gub-a dependant runner gate palace-GEN-LOC stand-PFV “dependants and runners, stationed at the gate of the palace” With the frequent verb gub “to stand, to be stationed (intr.)” the genitive+locative is even noted with Íà “interior” (on Íà see p.355): ták-Íi-ru-um Íà é-gal-ka gub-ba (no. 49: 45 etc.) takÍ‹rum Íà é.gal-ak-a gub-a repair interior palace-GEN-LOC stand-PFV “stationed at the repair, in the palace” A noun phrase in the absolutive, ending in a double genitive and thus marked equally -ka (<-ak- ak-ø) appears regularly in the date formula for year fiu-Suen 9:

d d ki d ki mu Íu- EN.ZU lugal urim5 -ma-ke4 é Íara2 umma -ka mu-dù (no. 257 etc.) mu fiu-Suen lugal urim-ak-e é fiara Umma-ak-ak-ø mu-n-dù-ø year PN king GN-GEN-ERG house DNGN-GEN-GEN-ABS VNT-A.3P.SG- erect.Bh-o.3SG “year: fiu-Suen, king of Ur, built the temple of fiara of Umma” Sumerian Language Use at GarÍana 353

A seemingly double genitive is employed, er with the incorrect “double genitive” -ka as however incorrectly, in the year date of fiu- in the former date. Suen 6. The standard formula shows the dative Somewhat similar is the notation of a gen- case, and this is the most frequent construction itive before the locative in the following (143™ for fiS 6, but formula often restored): phrase, where egir-ra-ka (egir-“GEN”-LOC) mu ... na-dù-a maÓ den-líl dnin-líl-ra mu- appears instead of expected egir-ra (egir-LOC): ne-dù gú-ba kí™ bí-na, egir-ra-ka in-sù-ub (no. “year (king fiu-Suen ...) erected the sub- 1048) lime stele for Enlil and Ninlil” “He did work on its bank.28 Afterward/at A variant construction is without -ra but with the backside he neglected it” (assuming a simple genitive, “year: king fiu-Suen ... erect- that sù-ub stands for Íub).29 ed the sublime statue of Enlil and Ninlil for them.” This simple genitive before the absolu- 4.5.4 Ergative for Absolutive tive should be written -lá, but this appears only Perhaps the most remarkable occurrence of an once in this year date, whereas an incorrect incorrect use of a case suffix is the appearance double genitive -ka is written 35™ for the date of the ergative marker -e in an intransitive fiu-Suen 6. phrase: mu...na-dù-a maÓ den-líl dnin-líl-ka u4 lugal-e ba-úÍ-[(a)] (no. 257: 6) mu na.dù.a maÓ Enlil Ninlil-ak-ak-ø u4 lugal-e ba-úÍ-ø(-a) ERG MED SG NMLZ year stele sublime DN DN-GEN-GEN-ABS day king- -die.Bh-3 (- ) “when the king died” mu-ne-dù Since the verbal form ba-a-™ar is clearly intran- mu-ne-n-dù-ø sitive, the same explanation probably holds true VNT-IO.3PL-A.3P.SG-erect.Bh-o.3SG for the almost unintelligible phrase: The only way to explain this variant seems to be a hypercorrection of the genitive (as a dou- u4 di™ir-re gaba na-wi-ir-ilum ba-a-™ar-ra ble genitive), which would have been slowly (no. 251) disappearing in certain contexts. Such a disap- “when a god was placed opposite of pearance is the less surprising since noun- Nawir-ilum”(?)30 noun-combinations without genitive marking The first example is unequivocal: the ergative exist frequently in Sumerian (e.g. qualifications lugal-e appears as subject in an intransitive sen- indicating the material). Furthermore, as with tence. Can one assume that the misuse of the the interchange of the verbal prefixes ba-/mu- ergative case instead of the unmarked absolu- (p. 348f.), the year dates show especially bewil- tive was influenced by the case system of Akka- dering variants at GarÍana. The year formula dian, where the subject is always in the nomi- for fiu-Suen 8, which runs absolutely parallel to native case? Unfortunately it is hardly possible d d fiu-Suen 6, mu (...) má-gur8 maÓ en-líl nin- to investigate the distribution of ergative (-e) líl-ra mu-ne-dìm “year (king fiu-Suen) con- and absolutive (-ø) in the GarÍana corpus, since structed the sublime boat for Enlil and Ninlil,” we deal mostly with proper names and profes- appears most often with -ra (143™), sometimes sions that are more often not case-marked in with a correct simple genitive -lá (7™), but nev- administrative texts.

28 30 For the writing bí-na for bí-in-a5(AK) “he made” Heimpel (2009, 332) understands the phrase differ- cf. Attinger (2005, 54). ently: “the day the breast of Nawir-ilum was made 29 For Íub in a similar context see the fragment RA to settle by the god.” This interpretation obeys the ergative case, but the transitive construction is dif- 10, 66 no. 68: 8': gú i7 gana2-ba Íub-ba “neglected(?) at its bank of the field’s canal.” ficult to reconcile with the intransitive form ba-a- ™ar. See the list of finite verbs above (section 4.1.) for the generally trustworthy rection of verbs. 354 Walther Sallaberger

4.6 Time Specifications: u4 as “subjunction” u4 bàd in-dù-sa-a “(food for builders) when and “preposition” they built the wall” (no. 396). Specific circumstances or even the motivations Similar constructions are often attested (36 for certain transactions are often indicated by clauses counted in the GarÍana corpus).31 temporal expressions in administrative docu- However, u4 “day, time,” can also be com- ments. For example, food was distributed to bined with infinite verbal forms, namely the boat haulers “when they hauled the boat” (e.g. no. 529), but certainly the return from the (a) with infinite perfective forms in -a: trip also justified the expenditure of food. u4 al-tar bàd gub-ba “(they have eaten) Temporal designations in Sumerian are formu- when: stationed (at the) construction work of the wall” (no. 392; cf. also nos. lated as a locative u4 ...-a “on the day, at the time,” whereby further appositions, most often 395, 442, 471, 495, 1007) a nominalized verbal clause, may qualify the (b) with infinite imperfective, future forms in noun u4, “day, time.” -d-e (directive): Besides the formulation as a temporal u4 bàd é-a dù-dè “(for heaps of flour) at the phrase with a nominalized finite verb (type PN day, when: to build the wall of the u ... ì-im-™en-na “(for) PN, when he ... 4 house” (no. 383, cf. also nos. 386, 387; came”), some documents, such as the messen- no. 90 does not belong here) ger texts from ªirsu, prefer an apposition construed with an infinite verb (type PN ... ™en- (c) with infinite simple form:32 na/du-ni “(for) PN, when he went/was u4 al-tar é-... “when: construction work of going”). The phrase mu ...(-ak)-Íè “because” (various houses)” (nos. 424, 428–430, appears more rarely, and here both infinite and 435, 349–441) finite nominalized verbal forms occur (e.g. mu (d) with the so-called pronominal conjugation aga -ús a-tu -a-ka é-gal-la ku -ra-ne-Íè “for 3 5 4 (infinite form + pronominal suffix): the soldiers, having entered the palace at the (time) of bathing,” Sallaberger 1993, vol. 1, u4 PN U.-Íè du-ni “(for heaps of flour) at the day, when: ﬁu-Kabta going to 67f.; PNN dam-gara3-me-éÍ mu na4 mu-ni-in- Umma” (no. 488, 2™) ku4-re-éÍ-Ía-a-Íè “(for) PNN, the merchants, because they had imported stones,” CT 32 25). Such conflations of two different construc- Rarely the purpose of the expenditure can be tions are probably not, or at the most only rare- indicated by an -e-d-e form (e.g., 7 gud en ly, attested in the corpus of Ur III texts. The Óu™-dè “7 cattle, in order to appoint the high- examples that I had thought to have located priest,” BIN 3 352). through searches in BDTNS or in my own col-

At GarÍana the introductory u4 “day, time” lections (and disregarding untrustworthy pub- appears in various combinations (for an over- lications) always turned out to be misreadings view see Kleinerman and Owen 2009, 144f. or wrong restorations of broken passages.33 s.v. Relative Clauses). The standard construc- The background of the peculiar GarÍana tion of a nominalized clause is e.g. combination of u4 with infinite verbs can be

31 32 The phrase in 1173 // 1361 kiÍeb3 PN u4- For u4 dub túm (no. 1173, 1361) see the preceding ! um (“DUB”)-de6 “when the sealed document of footnote. 33 PN will be brought” can be analyzed as u4 “when” For “u ” i -da zi-zi-dè ™en-na Sigrist, Princeton 2 + um-de so-called “prospective” (conditional); it 4 7 6 428, read eren2 i7-da..., as also noted by Notizia can also be understood as an unorthographic writ- (2009, 126); similarly Fish, CST 34 i and iii ! ! ing of the prospective as u4-um -de6; this interpre- eren (u ) ba-Íim-e-Íè ™en-na; Riedel, RA 10 209 tation implies that no nominalizer -a is written. 2 4 BM 103435 (= Verderame/Politi, Nisaba 8, 58) u4 d d ki ki amar- zuen-ke4 Ía-aÍ-ru ù Íu-ru-ud-hu-um [mu?]-Óulu-a. Sumerian Language Use at GarÍana 355

elucidated by a related time specification, the 26). Generally, in Ur III documents, the phrase indication of a festival day in the subscript of Íà GN-a “in GN” replaced the former locative some expenditures for persons. completely. The GarÍana scribes apparently took this u4 izim gur buru14 “at the festival of the return/Kor measure of the harvest” (no. path a step further and used u4 “at” as a tempo- 996) ral “preposition,” in a certain respect as a com- plement to the local Íà “within.” Although the u4 (izim) e-lu-núm-(ma) “at the el›num loss of case suffixes and the emergence of prep- (festival)” (no. 19, 980, 1032) ositions from former common nouns could u4 taq-ri-ib-tum bàd “at the offering (cere- represent an internal Sumerian development, mony) at the wall” (no. 1035) the impact of Akkadian should be acknowl-

u4 izim a-bu-um-(ma) “at the abum festi- edged especially at GarÍana. The Akkadian val” (no. 278, 509) model on which u4 was calqued can probably be identified: the subjunction in›ma, “when,” u4 izim pa-è “at the festival of appearance” (no. 1030) literally “in the day...” (in+›m-), and thus corresponding exactly to the Sumerian construc- The peculiarity of the GarÍana phrases tion u4 ...-a “in the day, ...,” could also be used emerges clearly as soon as similar specifications as a preposition in Old Babylonian (see the dic- in the Ur III corpus are compared. Though tionaries and von Soden 1995, 210 § 115 s and many documents, indeed, note a festival as time §116 b). This corresponds exactly to the use of specification (see, e.g., the texts collected in Sal- u4 at GarÍana. laberger 1993), a combination “u4 izim” (“time: 34 To summarize: The noun u4 “day, time” festival”) seems not to be attested elsewhere; began not only temporal clauses at GarÍana, but the simple standard construction izim NN “fes- also infinite phrases and even nouns, a peculiar tival NN” fulfils the same function. feature that is without parallel in the vast Ur III Obviously, the GarÍana u has to be under- 4 corpus. u4 emerged as a kind of temporal stood as the temporal companion to the local “preposition” similarly to the widespread local specification Íà “in” (< “heart, interior”), Íà “within,” which gradually replaced the loc- which appears as a subscript identifying the ative case suffix -a in certain contexts. The use place of a transaction. The term Íà NOUN-ak- of u4 agrees with the scope of the Akkadian a “within NOUN” (literally “in the interior of subjunction and preposition in›ma. NOUN”) was never written as such in the sub- The temporal “preposition” u4 is apparent- script, but appeared as Íà NOUN(-a), especially ly attested only at GarÍana and due to the end of ki with place names (e.g. Íà urim5 -ma, “in Ur”). the Ur III state this development did not gain In this way Íà began to function almost as a further influence. But the tendency that Sum- local preposition and partly replaced the loca- erian nouns started to serve as “prepositions” tive case, especially at the specific position in continued for example in the spread of Íà for all the subscript of a tablet.35 local specifications in the Isin craft archive In the Presargonic and Sargonic periods, (Sallaberger 2000, 272) and ultimately the case the simple locative case -a had sufficed at this suffixes were replaced by “prepositions” de- position, but in the Ur III period this is met rived from nouns in Old Babylonian adminis- only rarely and first of all in relatively early texts trative texts (Íà “in,” ugu “over, above,” ki from the time of ﬁulgi (e.g. é-sa™-da-na nibruki- “from,” in OB Uruk after Sallaberger 2000, ka “in Esa™dana-Nibru” OIP 115 1, ﬁulgi year 274). Furthermore it would be interesting to

34 35 The reference Sweet, ARRIM 01 23 H 36c: 34' u4 Heimpel (2009, 23f.) went even so far as to see Íà izim?-maÓ?, a text published only in a preliminary in some expressions as equivalent to the Akkadian transliteration, remains too uncertain (cf. Sallaberg- relative particle Ía. In all the examples he cites, how- er 1993, vol. 2, 113 note b). ever, the locative meaning “in” cannot be excluded. 356 Walther Sallaberger

know whether the specific use of Akkadian u4 ní™-mussa é PN-Íè in-na-aka-a in›ma as preposition and not only as subjunc- “when he made the dowry for PN’s house” tion in southern Old Babylonian did not fol- (Amorites 15, dated ﬁ 48/06/15) low a Sumerian vernacular that was more widespread, although we are aware only of the The expected order is also attested:

GarÍana evidence. u4 é-uÍbar(!ÚR)-ra-na ní™-mussa in-aka-a “when he made the dowry for his father-in- 4.7 Inconsistency of Word Order law’s house” (Toronto 2 205, dated AS In Sumerian phrases the absolutive case tends 01/12/12) to be situated directly before the verb. In administrative texts the central object(s) treated Although these observations are based only in the document with its quantities and quali- on a few characteristic contexts and verbs, the fications is always placed at the beginning of observed frequency of the incorrect word the text (cf. Attinger 1993, 154; Sallaberger order at GarÍana is remarkable; the standard 2000). The topic of a large number of docu- word order of Sumerian had lost its binding ments is workers and consequently their num- character. Parallel developments are not yet ber and qualification were noted; often an known, although they may be hidden in the additional phrase describes their work. Within vast corpus of Sumerian Ur III texts. this phrase the place of the absolutive case is directly before the (mostly infinite) verb. 5. The Akkadian Component of the Lexicon Regularly, the transport of goods is phrased The final aspect to be considered for the use of like this: Sumerian at GarÍana is the lexicon. Akkadian words are amply attested in administrative doc- n ™uruÍ (...) GN -ta GN -Íè OBJECT ga -™á 1 2 6 uments of the Ur III period, mostly specific (or finite form) technical terms for concrete nouns. Akkadian “n men (+ qualification), having carried words appear neither as adjectives nor as verbs OBJECT from GN1 to GN2” or in combination with Sumerian verbs. The This construction appears also at GarÍana: growing number of technical Akkadian terms in Sumerian results from the close contact of ≈ ™uruÍ aÍgab u4-3-Íè kar-ta kuru13-Íè Íe the two languages in the late third millennium. íb-ga6 (no. 229, also in no. 263) “≈ weavers for 3 days carried grain from Since we deal with technical terms em- the quay to the granary” ployed in specific contexts, a simple count of the frequency of Akkadian words would not But whereas the correct word order appears yield satisfactory results. But one can investi- only twice, the direct object is placed more gate if the Akkadian terms used at GarÍana are often, namely 34 times, before the local desig- known also from other contemporary archives. nations: Most Akkadian words are listed by Klein- 12 ™uruÍ in-u kar-ta ™á-nun-Íè íb-ga6 (no. erman and Owen (2009, 13f.) (= CUSAS 4), 28) others can be found in the glossary (ibid. 1– “12 men, they carried straw from the quay 230). The comparison with other Ur III ar- to the storehouse” chives is based on BDTNS, my own collec- A search of the frequent combination Íe tions, and the Akkadian dictionaries (AHw,

“grain” with ga6 “to carry” in the Ur III data CAD, MAD 3). Attestations from other ar- base BDTNS did not lead to any references for chives are given only in the case of rare words this irregular word order in documents from or references that are perhaps more difficult to Umma. Rarely, however, the displacement of retrieve. Some of the translations and meanings the object toward the beginning of the phrase given stem from the project on a Sumerian can be discovered in Ur III documents, as in glossary, partly based on the work of Hagan the following example: Brunke and Fabienne Huber Vulliet. Sumerian Language Use at GarÍana 357

5.1 Akkadian Words Attested in Other Ur III darken.” du-ú-um is always combined Documents with the verb aka “to make.” Since var- a-bu-um, “abum festival” (not listed CUSAS ious scribes were active at GarÍana with 4, 13) different writing habits (note Heimpel 2009, 27), we may deal probably with a-ge4-um, a qualification of textiles, perhaps “dyed, colored”(?) the same word as dì-um; the verbs aka “to make,” ga6 “to carry,” lu “to stir, mix” a-za-bu-um, “cistern”; cf. Pre-Sargonic CT are attested also with im sumur “plaster” 50 47:5 ter-kù ugula a-za-búm-ma-ke 4 (where the verb tà is used instead of ™ar lú Óu™-™á ba-de “Terku, the overseer of 6 with dì-um). the cistern, took the hireling(s) with ™eÍ him”; PDT 2 1366 reeds v.s. for the a- dú-ul-bu-um, “plane tree,” cf Gudea Cyl. za-bu-um “of the ﬁu-Suen garden” A xv 32 and St.B v 57 dú-lu-bu-um tu9 e-lu-núm (not listed CUSAS 4, 13), a festival á-gu4-Óu-um, aguÓÓum, perhaps a belt, sash al-lu-Óa-ru-um, “(a mineral tanning agent),” é gi-na-tum, “barracks,” passim Ur III; read- usually written al-la-Óa-ru in Ur III, but ing, meaning and etymology according al-lu-Óa-ru-um MVN 11 142 (Reichska- to Heimpel (2009, 165) derived from lender), note that the entry al-lu-Óu-ru-um kin⁄tum, kinattum, “servants”; note there CUSAS 4, 15 is a ghost entry (all refer- p.164f. for some writings, including é- ences are restored) Íu-tum, thus interpreting é GI.NA.TUM as a Diri compound (cf. for ar-ki-LUM; see CUSAS 4, 22 s.v. ar-gi (KI)- 5 é.GI.NA.AB.DU = Íutum, “(a kind of food product?),” also in 7 Íutummu CAD núm ﬁ/3 412, and MSL 15, 26 Proto-Diri NATN 825 besides onions/garlic and N :02 [é.g]i.na.ab.du = [ ]); its fish; probably different from Ur III 3 10 Í utummu ar-ga- Akkadian interpretation is as uncertain as , an aromatic(?) núm its meaning. ba-tab-tuÓ-Óu-um (in CUSAS 4, 25 read ba- tu9Óa-bu-um, tu9Óa-um, “a cloth for (covering) tab tuÓ-Óu-um), a valuable accessoire of chairs” female dress , “(a spice plant),” instead of geba-ti-um, “(a reed container),” cf. Ur III Óa-Íu-um ÓaÍûm as in the single GarÍana reference no. ba-ti-um, pa -ti-um 4 548:14, Ur III texts write Óa-Íi/Íu-a-núm bí-ni-tum, “a crossbeam” (thus CUSAS 4, (see dictionaries s.v.) 30, following the unsubstantiated idea of , “mayor,” same writing Ur III CAD s.v. “a type of crossbeam”), but Óa-za-nu-um only in NRVN 57:11, usually written according to Heimpel (2009, 181) bí-ì- “container”; see Heimpel for the Óa-za-núm tum s.v. ™eÍ latest treatment of the Ur III references. Óu-pu-um, “wheel rim,” written mostly Óu-pu-um, also Óu-pù-um BÙ-ZU-LUM, cited s.v. pù-sú-lum “(a food product?)”; if read correctly, the variant Óu-ri-um, “(a spice),” elsewhere written Óu- bù-su-lum in no. 1272:4 would point to rí-um; no. 1092 and no. 1190, there the reading pussulum, attested in Ur III among garden products; see below sub dì-um, “wattle-and-daub”; cf. Heimpel b) s.v. e-ri-um (2009, 177f.) for some other Ur III refer- kà-ma-am-tum, “(a vegetable; a spice plant)”

ences; attested with verbs ga6 “to carry,” li-iq-tum, literally “selection,” a qualifica- ™ar “to place,” and lu “to mix” tion of the dying agent alluÓaru (CUSAS du-ú-um, “wattle-and-daub” (variant)(?); 4, 111 “leftovers?”), Ur III written lí-iq- CUSAS 4, 39 “cella platform,” accord- tum ing to Heimpel (2009, 250f.) unclear, ma-ad-li-um, “bucket,” Ur III written ma- perhaps related to Akk. du’’umum “to ad-lí-um 358 Walther Sallaberger

ma-az-Óa-ru-um, “a vessel?, a sieve?,” only e-ri-um, “frond, leaf of the date palm” in no. 1299; also in Ur (CUSAS 4, 45; translation cannot be kuÍmaÍ-lí-um, “leather bucket” (not listed verified with the help of AHw and CAD; CUSAS 4, 13); see Hilgert (2002, 81) CAD E 325 erû E belongs to erû C “grinding slab” according to MSL 15, , “pickled, salted (meat),” mu-du-lum mud- 185); no. 511:42 and no. 1036: 75, 1 sila ; Ur III written rarely also 3 dulum mu-du- e-ri-um between zú-lum and sa gi-ne, lu, see Hilgert (2002, 81) offering for gods; word unknown; per- kuÍ na-aÓ-ba-tum, “leather case for precious haps variant of Óu-ri-um? objects” ™eÍkà-na-at-Óu-um, “(a tree or a aromatic na-kab-tum (not listed CUSAS 4, 13), product from a tree),” only in no. 1375; Akkadian nakkamtum (Brunke 2008) word not listed in the dictionaries; a sà-Óumzabar, “a drinking vessel” variant of Akk. kanaktum (which would ™eÍ sé-er-dum, serdum, “olive tree,” rare in correspond to Sumerian Íem-ge17)? Neo-Sumerian texts, see PDT 2 918 i ™eÍla-ri-um, “a part of the plow,” literally 21, RTC 216: 1 (olive oil) “branch” (larium), only no. 1255:2; per- (™eÍ) si-ì-tum, Í‹tum, “remainder, rest” haps similar to Sumerian á-apin “plow handle” tá-ki-ru-um, takk‹rum, “a heavy, valuable cloth” ní™ lá-lá-rum, lallarum, “(a dessert, pastry),” or ninda lá-lá-rum, only in no. 972 and ták-si/Íi-ru-um, takÍ‹rum, “repair (of archi- no. 975 tecture)” qí-il-pu-um, “skin, peel; bark for tanning?,” [tuÓ-Óu-um read ba-tab-tuÓ-Óu-um] although Akkadian qilpum is attested u4-tuÓ-Óu-um, “a kind of bread,” known at only MB/LB, the proposed etymology is Ur and in the form ì-tuÓ-Óu-um at Isin convincing zi-ba-tum, “a commodity in the textile ra-ˇum, r⁄ˇum, “drain pipe, runnel,” see industry” (CUSAS 4, 222: “rear part (of Heimpel (2009, 184f.); perhaps corre-

certain implements)”), also written zi-ib- sponding to Sumerian a-pa4 ba-tum (UET 3 1505) ™eÍÍa-ba-ar-gi-lum, Íapargillum/supurgillu, ™eÍzi-ri-gum, zir‹qum, “shaduf,” Ur III zi-ri/ “quince” (not listed CUSAS 4, p.13; see rí-gúm p.175), only in no. 1375:9; attested oth- erwise only in lexical lists and in Assyrian 5.2 Akkadian Words at GarÍana not Attested ﬁID-tum “?”, booked as Íit-tum “remainder” in Other Ur III Corpora (which is usually, also at GarÍana, written e-pe-eÍ (ep¤Í) na-am-dur11-ra, “the making si-ì-tum); but of unclear interpretation of sickness”; see Heimpel (2009, 60) (Heimpel 2009, 186, similarly reads Íit- e-pí-a-tum (epi⁄tum), “baked,” only in frag- tum “rest”). In parallel contexts in no. ment no. 1209:2 ninda e-pí-a-tum; CAD 1299 and no. 1265 (restored), referring ﬁID E 247b cites 3 NT-850:1 5 sila3 ninda e- to kinds of oil designated as -tum pu-um (“OAkk,” which includes Ur III; BUILDING (...)-Íè ba-a-™ar “was placed Akkadian or Sumerian context?); corre- on the Í. of ... (and for ...)” 36 sponds to du8 “to bake” taq-ri-ib-tum, taqribtum, “offering”

36 U. Gabbay’s translation “lamentation” given by actionis of qurrubum “to present (an offering),” fits Kleinerman and Owen (2009, 191) cannot be ver- all contexts in CUSAS 4. ified. An etymologically correct taqribtum, nomen Sumerian Language Use at GarÍana 359

zi-mi-tum, “(a kind of plant)”; note OAkk bu-ti-um?, “(a specification of textiles),” zi-me-tum OSP 2 184: 10 (same reference unclear in no. 807:1: 4 túg-bu-ti-um? CUSAS 4, 160 fn. 101) ki-ri-ip, “a (oil) jar,” according to CUSAS 4, 99, thus referring to OB Mari kirip- 5.3 Akkadian Words at GarÍana, but pum. However, in no. 972: 17 and 88 Corresponding Sumerian Form Used read ki-ri-™á according to the photo Elsewhere CUSAS 3, pl. xxviiif.; no. 975 is a paral- a-ga-mu-um, agammum, “marsh,” in the riv- lel text, the reference in no. 511:65 er designation i7 a-ga-mu-um (CUSAS 4, (CUSAS 4, 99) is not preserved. ki-ri-™á 652), elsewhere written a-ga-am is unclear, but cf. perhaps ki!(fiU)-rí-™á a-ga-ri-nu-um, agarinnum, “pit/basin for BIN 10 33:2. mixing mortar” (CUSAS 4, 3 does not mul-ˇum, “a comb,” according to CUSAS list attestations of writings), also Sumer- 4, 123; in no. 923: 6 partly broken [mu]l- ian forms a-™á-ri(2)-na, a-™á-ri-in as else- tum; the phonology of Ur III Akkadian where in Ur III documents. The (-Ít- > -lt- appears half a millennium lat- Akkadian form agarinnum is attested only er) and the context do not allow an in nos. 7 ([fiS 6/05] 11) and no. 8 ([fiS 6/ interpretation muÍˇum “comb.” 05/12]), thus it appears to be the personal habit of a scribe active at this time (fiS 5.5. Discussion 6/05), perhaps scribe L⁄-q‹p(um), who is We obtain the following numbers for Akkadi- attested only in no. 394 (fiS 6/05) and an words in GarÍana documents: no. 399 ([fiS 6/06] restored), but not in no. 406 (fiS 6/11); L⁄-q‹p(um) appears in 36 Akkadian words are attested also in the first text with Puzur-Ninkarrak, who other Ur III documents; may be responsible for the writing im a- 12 Akkadian words at GarÍana are not ™á-ri-na already in no. 9 (fiS 6/05/13) or attested in other Ur III corpora; im ™á-rí-na in no. 10 (fiS 6/05/17). 3 Akkadian words substitute the corresponding Sumerian forms; kara3-Íum, kar⁄Íum, “leek” (CUSAS 4, 93 s.v. kàr-Íum); Ur III uses the Sumerian 6 uncertain entries. ga sar equivalent garaÍ6 The absolute number and the proportion of 5.4 Uncertain Entries37 Akkadian words strongly depend on the semantic fields treated by the various archives. a-Íi-um, “(a specification of textiles),” read- However, several aspects suggest that the ing and contexts uncertain Akkadian component of the lexicon is of ak-lu-um, in broken context in no. 925:14', exceptional quality at GarÍana. Not only do the reading and interpretation uncertain, texts use Akkadian words that are not attested aklum “overseer” hardly possible in other contemporary documents, but, more ™eÍbe-rum, “(a wooden object),” the context importantly, in some cases one can even guess of the reference (no. 807:2) does not the corresponding Sumerian equivalent (see point to a wooden object; note further- epi⁄tum, larium, r⁄ˇum). The case of replacement more that the two signs be-rum are BAD. by the Akkadian form is obvious in the instanc- Afi = zara6 (a textile), which would fit the es in which in other archives the same word context of textiles. appears in its Sumerian form without the Akka-

37 The following words are not Akkadian: za-rí-in “coarsely cleaned (wool) material”; for dugku-kur- rú (Owen and Kleinerman 2009, 104) read instead dugKU-kur-DÙ. 360 Walther Sallaberger dian ending (agammum, agarinnum, kar⁄Íum). of which 18 out of 23 bore Akkadian names, Exceptional is the Akkadian infinitive ep¤Í “to but only two Sumerian ones. make” in the status constructus, combined Despite their Akkadian background all with a phonographically written Sumerian rec- GarÍana documents were written in Sumerian, tum, na-am-dú-ra “illness.” The unortho- the language that was, most probably, essential graphic writing definitely excludes a logo- in the scribal education. The analysis presented graphic interpretation of Akkadian *ep¤Í above pointed out both the impressive pres- mar‰›tim (v.s.). Instead, we deal with a mixture ence of Akkadian words and the errors in the of words, a switching of codes so typical for grammar of Sumerian. Based on this evidence bilingual environments (cf., e.g., Appel and an Assyriologist would most probably recon- Muysken 1987, 117ff.). The orthography dis- struct the socio-linguistic scenario as follows: cussed earlier reveals that the GarÍana scribes the Akkadian scribes remained linguistically wrote their documents close to the spoken lan- attached to their native language Akkadian, but guage. The unfiltered mixture of languages they had learned Sumerian imperfectly at within one nominal construction is further school and used Sumerian simply as a scribal proof of this habit. and administrative language. Since they still spoke Akkadian at home, they naturally used 6. Akkado-Sumerian Bilingualism at GarÍana many Akkadian words in their texts. 6.1 The Socio-linguistic Setting: Bilingualism Such an impression is, according to current and Diglossia understanding, correct in some basic facts, The southern provinces along the Tigris, namely that the names indicate that Akkadian Umma and ªirsu, can be considered as the was probably most scribes’ native language, economically dominating region of the Ur III that Sumerian was the administrative language, state, as is evidenced most clearly by their and that scribal education was probably linked to Sumerian. heavy duties in the term of office (bala) to maintain state organizations. This region can However, a more-detailed and coherent interpretation of the data presented above leads aptly be labeled as the heartland of Sumer, as to a somewhat different and more refined pic- attested by the high portion of Sumerian names ture. Progress is possible by taking into account and the use of Sumerian as vernacular in ad- some basic results of the broad field of linguistic ministrative documents. Besides the provincial research on bilingualism. So our evidence could organization with a largely Sumerian popula- be checked against a wide spectrum of socio- tion there also existed settlements of the royal linguistic scenarios and a great number of lin- sector in the south. GarÍana is not only one of guistic features. It goes without saying that the largest of these garrisons, but also the first many aspects simply cannot be treated because one from which we dispose of cuneiform doc- of the enormous gap that separates us from an uments. ancient language we know only imperfectly in In the organization that was headed by its written form. But even such a restricted cor- general ﬁu-Kabta and his wife, the princess pus offers sufficient linguistic data to allow a Simat-IÍtaran, at GarÍana a high proportion of coherent interpretation that respects linguistic the active population was of Akkadian back- methodology and results. The conclusion builds ground, at least according to their proper directly on the analyses presented above and names.38 This is especially true for the scribes, thus does not repeat the evidence in detail.

38 Arguments over why and how personal names can be used in order to detect the distribution of languages in the ancient Near East have been put for- ward by Streck 2004; for the case of Sumerian cf. Sallaberger 2004. Sumerian Language Use at GarÍana 361

Most scribes active at GarÍana, a newly or logographic values can be derived, and so founded garrison in the region of Umma, bore Sumerian served as a language in scribal teach- Akkadian names, so it is safe to assume that ing even after it had disappeared as a freely for- their native language was Akkadian. Their first mulated language in documents, i.e., from the language Akkadian thus has to be considered late twentieth century onward. Given the their linguistically dominant language. They ac- markedly different socio-linguistic situation in quired Sumerian as their second language, the the Ur III state, one simply cannot adopt the language in which they wrote the cuneiform well-known model of scribal education in the documents. Sumerian was the language of the Old Babylonian period that followed a curric- state organizations, as the royal archives from ulum that was largely designed in the Isin peri- PuzriÍ-Dagan amply testify, which were writ- od at the time when Sumerian was vanishing as ten in Sumerian without exception. In this dis- vernacular.40 Whereas the teacher in the Old cussion we confine ourselves to the language Babylonian period was most often a L2 speaker use within the administration during the Ur III of Sumerian, such a scenario can hardly be pos- period (twenty-first century B.C.E.) and do not tulated for the scribal education in the state sec- consider the language at the royal court, in tor of the Ur III state that we are dealing with poetry and religion. But it is obvious that king here. Furthermore, the acquisition of Sumerian ﬁulgi’s reference to his Sumerian background39 writing at school was combined with natural or the use of Sumerian as the dominant lan- second-language acquisition in the communi- guage in royal texts contributed to the prestige cation with Sumerian native speakers in the of that language in the state of Ur, even if per- surrounding traditional settlements of Umma haps more inhabitants were of Akkadian or province. The orthographic variations men- some other tongue. So Sumerian is character- tioned above (p.339) clearly demarcate the lim- ized by need (in the context of the administra- its of scribal education and hint at the impact of tion) and prestige (as the royal language), the spoken language acquisition. two features that characterize a socially dominant In our discussion we are concentrating on language (Van Coetsem 1988, 13f.). the Akkadian scribes, but the personal names The process of language acquisition is not and the linguistic setting of the Ur III state sug- known directly from the evidence, but some gest that speakers of other tongues were present basic facts are quite certain. The complicated in state organizations and households as well, cuneiform script had to be learned in a long most prominently Amorites, Elamites, and Hur- process of scribal formation, probably within rians. In this multilingual context, Sumerian, the institution in which the scribes would serve which has been identified as the socially dom- later, which implies frequent instruction by the inant language at GarÍana, may have served as father or in the case of newly founded organi- the common language in the institutional do- zations by more experienced scribes. Cunei- main we are discussing here.41 Under such conform writing is intrinsically linked to Sumerian ditions one may think of a diglossia situation as a language, from which most phonographic with Sumerian as the more prestigious lan-

39 Rubio (2006) has argued that fiulgi hymns B and C IÍme-Dagan or Lipit-EÍtar; the reference to Isin indicate that Akkadian was the native language of kings in model contracts; the prominence of the fiulgi (accepted by Michalowski 2006, 175f.); a city of Nippur as the center of Sumerian, which more coherent interpretation of the relevant pas- transpires in all text genres including the lexical sages, which would contradict fiulgi’s claim to be lists, a fact that is historically correct for the Isin pe- “of Sumerian seed,” is provided by Keetman riod only. (2010). 41 Bilingual communication characteristically takes 40 More exactly, the formative period was the second place in communication outside the family; the half of the twentieth century B.C.E.. In this regard background of the family is acknowledged by the one may cite the dominance of Isin texts around definition of a native language. 362 Walther Sallaberger guage in public life, thus the “high” variety vations can be fitted in. I think it is most appro- according to the traditional terminology, and priate to present Van Coetsem’s linguistic Akkadian as “low” form. Nowadays, diglossia model of contact-induced change as sketched is seen as one form of bilingualism, character- by a specialist in Contact Linguistics (Winford ized by the social dominance of one language 2007) who used this model to test different that entails linguistic standardization, as recent methodological approaches: expressed in a definition like the one by Appel Van Coetsem’s major contribution was to fur- and Muysken (1987, 26): “Studies [...] have led, ther refine the traditional distinction between in fact, to a gradual redefinition of the term borrowing and ‘interference’ [stemming from diglossia: it is now used to refer to bilingual the classic of U. Weinreich, Languages in con- communities in which a large portion of the tact: Findings and problems, 1953] by defining speakers commands both languages, and in these types of cross-linguistic influence more which the two languages are functionally dis- precisely, and above all, by distinguishing the tinguished in terms of H [i.e., “high” variety, kinds of agentivity they involve. Van Coetsem’s W.S.] and L [“low,” W.S.].” framework distinguishes between two types of cross-linguistic influence, or what he calls ‘trans- 6.2 Contact-induced Linguistic Change fer types’, namely, borrowing and imposi- according to Van Coetsem (1988, 2000) tion.[...] The latter is largely equivalent to terms The GarÍana Akkadian scribes show a remark- like ‘interference via shift’, ‘transfer’ ‘indirect able linguistic proficiency in Sumerian; they diffusion’, and ‘substratum influence’ that appear in the literature. Borrowing and imposi- did not simply use prefabricated phrases and tion, in this framework, are not seen as ‘mech- forms. As described above in detail in sections anisms’ or ‘processes’, but rather as vehicles of 2 and 3, even phonetic variation was represent- contact-induced change. In both cases, there is a ed in a way hardly known from the larger source language (SL) and a recipient language archives so that the GarÍana documents count (RL). These terms serve as alternatives to various among the most important sources to recon- other terms that have been used in the literature, struct the phonology of Sumerian. This indi- such as ‘donor language’, ‘substrate’, ‘replica lan- cates a strong presence of spoken Sumerian. guage’ and the like. On the other hand, various deviations from the The direction of transfer of linguistic features standard grammar were noted and the high is always from the source language to the RL, portion of Akkadian lexemes stands out both in and the agent of transfer can be either the recip- quantity and quality. The data seemingly pro- ient language or the source language speaker. In duce a contradiction in the interpretation, since the former case, we have borrowing (RL agentivity), in the latter, imposition (SL agentivity). we note a basically Sumerian phonology and (Winford 2007, 25 f.). morphology, but Akkadian influence in the deviations of morphology and especially in the At GarÍana, therefore, we deal mainly with SL lexicon. (= source language) agentivity, since the scribes’ The influential model of bilingualism elab- native language is Akkadian, and the Akkadian orated by Frans Van Coetsem42 (1988, 2000), “influence” in Sumerian (the RL = recipient lan- which largely integrates the terminology and guage in contact-induced changes) introduced classification of previous studies, allows us to by the scribes is, therefore, called “imposition.” disentangle the various contact phenomena at Also highly relevant to the distinction between work, and eventually the contradictions of the borrowing and imposition is the notion of lan- GarÍana evidence will disappear and all obser- guage dominance. As Van Coetsem (2000: 84)

42 Frans Van Coetsem (1919–2002) specialized in Germanic historical linguistics; he taught at Cornell University, the very university where the GarÍana texts are now housed. Sumerian Language Use at GarÍana 363

explains, difference in linguistic dominance is thus more amenable to change. This is partly the main criterion for distinguishing between why borrowing tends to be mostly lexical, and recipient language and source language agentiv- to have little if any effect on the recipient lan- ity. In the former case, the recipient language is guage grammar. On the other hand, in imposi- the dominant language of the speaker, while in tion, where the source language grammar is the latter case, the source language is the dom- more stable and resistant to change, grammatical inant language. When we speak of dominance features can be transferred more readily, leading here, we are referring to linguistic dominance, to significant structural change in the speaker’s that is, the fact that the speaker is more proficient version of the RL. There may well be differenc- in one of the languages in contact. This must be es in degree of stability within different aspects of distinguished from social dominance, which the grammar, which may lead to different refers to the political or social status of one of the potential for transfer. Thus certain function languages. The socially dominant language may morphemes tend to be transferred more readily or may not be the linguistically dominant lan- than others, and word order, for instance, seems guage of the speaker. Of course, dominance to be transferred more readily than, say, embed- relationships may change over time, both in the ding strategies. (Winford 2007, 26). individual speaker. And such shifts in dominance may result in different outcomes, or lead 6.3 The Degree of the Acquisition of Sumerian to attrition of the previously dominant language. The GarÍana scribes must have reached a high These considerations require us to distinguish degree of linguistic proficiency in Sumerian the agents of change from the kinds of agentivity according to the general use of Sumerian gram- they employ in introducing changes to an RL. The fact is that the same agent can employ either mar: the differentiation of the verbal prefixes or type of agentivity, and hence both transfer types, correct stem formation in Óamˇu and marû stems in the same contact situation. This is particularly also with irregular verbs clearly dominates over true of highly proficient bilinguals, though not the few deviations from standard Sumerian list- restricted to them alone. ed above (e.g., the incorrect reconstruction of It is impossible to describe from the outside a verb /alna™/). the grade of linguistic proficiency the GarÍana Phonology is a much more difficult matter scribes have reached in Sumerian. However, since much is hidden behind orthography. The the documents we dispose of are written in variants and the many unorthographic spellings Sumerian, so here the scribes are to be seen as at least provide some indications, but it is “recipient language (RL) agents” in the contact impossible nowadays (at least for me) to recon- situation with Akkadian. The model thus helps struct the phonemic systems of Akkadian and to differentiate between various types of influ- Sumerian at that time and, moreover, a com- ence. This becomes the more important, since bined phonemic system employed in a bilin- Van Coetsem (and, of course, others) identify gual environment. The phonological changes various domains of the language that are more that have been observed both for Akkadian and open to imposition (source language agentivi- Sumerian at the end of the end of the third mil- ty, i.e., the Akkadian scribe’s influence on lennium (e.g. Gelb 1961, 3140, Jagersma 2000, Sumerian) or to borrowing (recipient language Keetman 2004a, 2004b) are clear examples of a agentivity, i.e., the Akkadian influence on the phonological redistribution that affected both Sumerian of a native speaker). languages (cf. Van Coetsem 1988, 115ff.). Gen- erally one considers “vocabulary the least stable Differences between recipient language and language domain, and phonology and grammar source language agentivity are also related to what Van Coetsem (1988: 25) calls the ‘stability (morphology and syntax) the more stable ones.” gradient’ of language. This refers to the fact that (Van Coetsem 1988, 26, his emphasis; cf. id. certain components of a language, such as pho- 2000). The direction of contact-induced nology, morphology and syntax, tend to be change depends on the stability of the phe- more stable and hence resistant to change, while nomena investigated: “In short, the transfer of others, such as vocabulary, are less stable and material from the source language to the recip- 364 Walther Sallaberger

ient language primarily concerns less stable erian /b, d, g/ now corresponded to Akkadian domains, particularly vocabulary, in borrow- /b, d, g/, whereas the /p, t, k/ correspondence ing, and more stable domains, particularly pho- remained unchanged (Gelb 1961, 31–40; cf. nological entities, in imposition” (ibid. 3). The Keetman 2004b). The sound change in Sume- phonemic changes in both languages are thus rian is reflected not only in early and late Akka- sufficient evidence for the intense bilingual dian loanwords, but also in the Akkadian contact, in which speakers from both languages syllabary, which began to adopt the “new” had brought their “accent” to the other lan- Sumerian system already during the Ur III guage as an imposition. period. A short explanation may be helpful. In Concerning the Akkadian scribes at GarÍa- the Sargonic period, Akkadian /da/ and /ta/ na, we confine ourselves to a simpler, but basic were written with the sign DA, which in Sum- question: Did they adapt the Sumerian sounds erian was pronounced as [ta]. Sumerian TA to their own Akkadian system or did they imi- stood originally (probably) for aspirated voice- tate the different system?43 Although, of course, less [tha]; but after the sound change [th] > [t] we cannot know how much of an Akkadian (parallel to [t] > [d]) in Sumerian, the sign “accent” may have existed in spoken language, could be used for similar Akkadian plain voice- the phonographic writings are indicative in this less [ta]. J. Keetman (2004a) has pointed to the regard. First, the typical Sumerian phoneme ™ innovations in the Akkadian syllabary of the Ur is kept in various environments, most notably III period, which he explains as reflection of a in unorthographic writings. Secondly, some sound change in Sumerian;44 within this pro- irregularities in the syllabary seem to indicate cess also the Sumerian “dr phoneme” changes that the phonemic system was enriched by to /d/ or /r/ (Jagersma 2000). “foreign” phonemes, faint reflection of the The new evidence of the GarÍana corpus imitation of Sumerian by the GarÍana scribes. concerns first the sibilants. Both the SV and the In this regard a few words may suffice here. fiV series of phonograms represented Akkadian The study of loanwords has shown that the sys- /Í/, although at GarÍana only fiA, but never SA, tem of Sumerian stops did not correspond to was used for /Ía/. Sumerian still retained the the Akkadian one in the third millennium. difference between the phonemes represented Originally, Sumerian disposed only of voiceless by the SV and the fiV series. However, this dif- (aspirated and plain voiceless) stops, whereas ference was only imperfectly adapted by the Akkadian had both voiced and voiceless (and, GarÍana scribes, since they used SA and fiA to of course, “emphatic,” ejective) stops. In early represent Sumerian /Ía/ (in verbal plural loanwords both Sumerian /b, d, g/ and /p, t, forms) or SI to represent Sumerian /si/ and /Íe/ k/ were reflected in Akkadian as /p, t, k/. A (see p. 345 above). If we concentrate on the sound change affected the Sumerian system at unequivocal case of SA and fiA, the global situ- the end of the third millennium so that it dis- ation in the GarÍana documents can be repre- posed of voiced and voiceless stops; thus Sum- sented as follows:

43 See Van Coetsem (1988, 10f.) on the basic differ- 43 Keetman (2004a, 198–200) still adhered to the ence between imitation and adaptation. Seen in the opinion that Sumerian was a dying language during (typical) RL agency, imitation implies that a for- the Ur III period. The arguments for Sumerian as eign phoneme of the SL is kept, but that this alters a living language in southern Babylonia during the the phoneme system of the RL, whereas in the case Ur III empire have been elaborated by Sallaberger of adaptation the SL phoneme system remains sta- (2004), who votes for a sudden death of the lan- ble, but the foreign word loses its foreing character guage after the perturbations at the end of Ur III; (one example: imitation is the case when a German see also Woods (2006). speaker pronounces “drive-in” with an English/ American /r/, but adaptation if he uses a German /r/). Sumerian Language Use at GarÍana 365

Akkadian Sumerian Phonogram ﬁA SA ﬁA Phonemes (as perceived by /Ía/ /sa/ /sa ~ Ía/ scribes)45

Apparently Sumerian /Í/ was identified 6.4 Imposition and Borrowing in Morphology either as similar to Akkadian /Í/ or to the Sum- and Lexicon erian /s/. More importantly, however, the use As outlined above, imposition and borrowing of the SA sign, which does not occur in Akka- differ in the perspective of the language agent. 46 dian words, implies the imitation of a pho- Imposition is a change (in the recipient lan- neme foreign to Akkadian. The variation in the guage) based on source language activity, thus writing of Sumerian /Ía/ with both the signs SA Akkadian influences that affect the Sumerian of and ﬁA thus reflects the imitation of Sumerian the GarÍana corpus. These changes should not phonemes by the Akkadian scribes with the occur in a similar way in the archives of Umma help of their phonographic repertoire. or ªirsu, which stem from dominantly mono- Another example is the rendering of Sum- lingual Sumerian environments. KE erian /g/ as 4 in the forms si-ke4-dè or sì-ke4- Imposition, which Weinreich and others dè, thereby avoiding both KI and GI, which have called “interference,” first of all concerns were both employed in Akkadian as ki and gi the more stable domains of language, phonol- (p. 344). Also the differing writings of Sumeri- ogy and morphology (Van Coetsem 1988, 3 et an comitative /da/ with DA in the prefix chain passim). Such an imposition can be observed in but with TA as case suffix (p. 351) or the use of the field of morphology at various instances: TI and TE for -dè (pp. 340 and 342–44) point one notes deviations in the Sumerian word in the same direction. order (p.356), an incorrect re-interpretation of The variant writings at GarÍana thus reflect the verb na™4 as /alna™/ (p.349f.), once an the imitation of Sumerian phonemes even in incorrect combination of plural pronoun with writing, a reflection that has become visible case (p.350), deviations in the case system such as only because of the unorthographic spellings by a hypercorrection of the genitive (p.352f.), erga- phonograms. This variation implies that there tive instead of absolutive (p.353), and the grad- still existed a difference between the phonemic ual supplementation of the locative by the systems of the two languages at this period and terminative (p.351f.). Within the vocabulary, that the loanword patterns give a simplified only function words can be taken as signs of picture of the processes of phonological redis- imposition, and here the only example is the tribution. The bilingual situation at GarÍana use of u4 “when” as a kind of temporal prepo- allows a glimpse at the phonological systems at sition/subjunction. The phrase ep¤Í namdura, work that is not possible with the more strict literally “to make sickness,” with the Akkadian orthography employed in monolingual con- verb ep¤Íum “to make” as a primary vocabulary texts like at Umma or ªirsu. word combined with the Sumerian abstract noun, is a unique example of code-switching. It is important to note that imposition involves both externally and internally induced change: SL (source language) agentivity implies

45 /Ía/ and /sa/ are used as symbols to represent pho- 46 Note that text no. 1473 with the personal name ì- nemes within their system whatever their phonetic lí-sa-tu does not stem from GarÍana. character may have been. 366 Walther Sallaberger not only a direct implementation of features of and grammar” (Van Coetsem 1988, 59). In pid- the SL (here Akkadian) in the RL (here Sum- gin “apparently the minimal acquisition of erian), but “language contact favors another English primarily concerns vocabulary” (ibid. kind of change in the RL, namely one that rep- 74), which is homogenous and derives from the resents the activation of latent developments in RL. With a comparable imperfect acquisition the RL itself (developmental trends). This does of Sumerian one would expect a simple vocab- not involve the direct participation of a SL” ulary and a sensible reduction of morphology (Van Coetsem 1988, 41). So only the incorrect (cf. ibid. 63), which is definitely not the case. pronoun-cases combination ka-ga-na-ne-ne Such examples give more weight to the GarÍa- (p.350) or the ergative lugal-e in an intransitive na data as evidence for a high progress in lan- sentence as “subject case” (p.353) have to be guage acquisition that leads to linguistic pro- called externally induced changes. The other ficiency. And better acquisition generally im- instances, including the use of u4 “when” sim- plies less imposition. ilar to Íà “within,” are internally induced The most prominent Akkadian component changes, and in these cases references could be of the GarÍana texts is constituted by the lexi- given for similar or parallel developments in con. The Akkadian lexicon employed com- other archives. prises almost exclusively nouns, most often a Another way of incomplete acquisition of special vocabulary, whereas words from a basic the RL by SL agents is reduction. It “primarily or primary Akkadian vocabulary do not appear affects inflectional morphology, but it also at all. This, however, is a typical situation for affects the domain of vocabulary, i.e., its total- borrowing, a case of RL agentivity (see above); ity as well as its individual items” (Van Coetsem the vocabulary, and here especially nouns, is 1988, 55). Perhaps the best case for morphology the least stable component of language. There- is the frozen form ba-na-Óa-la (p.349) for ba-ne- fore, this is, on the one hand, the first aspect of Óa-la (if it is not sign of a different vowel assim- a foreign language that is acquired and, on the ilation). However, it is hard to find further other, foreign words are most easily integrated examples of reduction. into a language (RL agentivity). The high por- On the contrary, the variation both in the tion of Akkadian nouns within Sumerian texts Sumerian lexicon and in the verbal morpholo- would thus be explained most aptly as borrow- gy confirm that the acquisition of Sumerian by ings by Sumerian speakers who come in con- the native Akkadians must have reached a high tact with Akkadian. This situation of borrow- degree, an impression that is confirmed by the ing describes well the appearance of Akkadian comparison with Sumerian texts from other ar- words in Sumerian texts in the Ur III period in chives. Van Coetsem (1988, 11f. and 18) iden- general, but for the moment it fails to explain tifies two steps in language acquisition in SL the GarÍana evidence. On the other hand, the agentivity, namely adaptation as primary mech- imposition features discussed above in the fields anism, a compensation for incomplete or defi- of morphology and syntax agree with the fact cient acquisition, and imitation as secondary. that native speakers of Akkadian wrote Sume- According to the evidence reviewed thus far, rian. in fact imitation prevails over adaptation in the To overcome this difficulty, the frequency stable language domains. Function words and of the various features has to be considered. As the basic vocabulary in the GarÍana documents we have noted above, the deviations in the are Sumerian, never Akkadian; no reduction field of grammar are most often exceptions can be observed in the actively employed Sum- (note the clear case of alna™a or the single gram- erian lexicon and its idiomatic expressions. matical deviations). Such variations in language The results of the analysis are the more strik- use indicate the personal preferences of the var- ing if compared with other scenarios. “Naïve ious scribes employed at GarÍana, which are language-learning” concentrates on acquiring also discernible in the differences of the layout lexemes, but “scant regard for pronunciation of the tablets (cf. Heimpel 2009, 27). The devi- Sumerian Language Use at GarÍana 367

ations, features of an imposition of the SL distinction between the two transfer types Akkadians on the RL Sumerian, thus belong to [namely imposition in SL agentivity and bor- the language usage of individuals in their mes- rowing in RL agentivity, W.S.] that strongly sage, which is termed “inclusion” by Van Coet- favors a form of convergence of the contacting sem (1988, 77–80). But these features were not languages” (Van Coetsem 1988, 87). yet integrated into the code, the language in its A total neutralization of the two transfer societal use.47 types takes place when “a bilingual’s proficien- In the case of far-reaching language acqui- cy in his two languages became comparable” sition, the two standard transfer types of lan- (Van Coetsem 2000, 84), and now stability guage contact (imposition as SL agentivity and forces of equal value oppose each other. borrowing as RL agentivity) may even be Whereas, as already indicated, secondary voca- reversed. bulary (i.e., contentives) is of minimal stability, primary vocabulary, functors (pronouns, prep- In the standard aspect of RL agentivity the RL is a native or first language, that is, there is in rela- ositions etc.) are of median stability (see in tion to the RL a clear distinction between non- detail Van Coetsem 2000, 105ff.). The Akkadi- nativeness (nonprimary language) and an words in the GarÍana Sumerian texts can, nativeness (primary language) ... in other cases, without any doubt, be identified as secondary however, there is no such distinction between vocabulary. Thus, stability factors play a deci- nonnativeness and nativeness. For example, the sive role in the transfer from Akkadian to Sum- RL as spoken by the SL speaker in SL agentivity erian. According to the observed linguistic is a nonnative or nonprimary language ....; the features, GarÍana Sumerian cannot be com- distinction between nonnativeness and native- pared to the so-called mixed languages, where ness is then not applicable. When the RL is a selection and manipulation occur as the result nonnative or nonprimary language, RL agentiv- of free transfer in both directions (Van Coet- ity may also be found. (Van Coetsem 1988, 79). sem 2000, 239ff.). Seen as a language only, Sumerian at GarÍana was a non-native lan- Sumerian borrowed from Akkadian in the guage for the Akkadian scribes, but it served as domain of secondary vocabulary, the least sta- recipient language (RL) in the case of borrowing, ble segment of language, in order to meet com- namely the inclusion or integration of a high municative needs in the garrison of GarÍana, portion of a specific Akkadian vocabulary. This whereas the more stable parts of the language situation is the less surprising, since “in strata (phonology, morphology, primary vocabulary) distinctions (such as in diglossia) the difference are Sumerian with deviations in the message in linguistic dominance between the RL and domain, which could be explained as imposi- SL and thus between the two transfer types may tion by the Akkadian scribes. be minimal” (Van Coetsem 1988, 78). Thus the So in the end, after applying the model of high portion of Akkadian nouns in the GarÍana Contact Linguistics developed most promi- vocabulary does not contradict the high degree nently by Van Coetsem (1988, 2000), who, of of acquisition of Sumerian by the native Akka- course, built on previous researches since dian scribes of GarÍana, but, on the contrary, Weinreich, the various features of GarÍana borrowing is exactly a feature that occurs when Sumerian described above can be integrated the former non-native speakers master their into a surprisingly coherent picture. second language (as RL) in such a way that they The GarÍana scribes were mostly native integrate “foreign” (SL) vocabulary. “It is pre- Akkadians. This background, however, tran- cisely this interaction with the lack of a clear spires relatively rarely in various impositions in

47 Even the variation in the vocabulary between the Sumerian form a™arin and the Akkadian agarinnum thus belongs to the message domain. 368 Walther Sallaberger morphology and syntax, and more often we incited by Edzard (2000), who, like Micha- deal with inclusion in the message domain of lowski (2006) or Rubio (2006),48 was especially single speakers in certain situations. The mas- informed by the evidence of literary texts in the tery, especially of the diversified verbal mor- broadest sense. Not surprisingly, the concen- phology or the invariably Sumerian basic tration on the language of literature led to such vocabulary, is a clear sign of a high degree of diverse interpretations as the survival of Sume- language acquisition, and the phonographic rian into the middle of the Old Babylonian spellings even hint at imitation of the Sumerian period (Edzard 2000) or an early demise of phonemes by the scribes. The integration of Sumerian most notably by Michalowski (2006), special expressions from Akkadian into a basi- who explicitly bases his considerations on the cally Sumerian vernacular agrees well with the case of Sanskrit (ibid. 171ff.). fact of a far-reaching acquisition of the non- In a different methodological approach, native language Sumerian, whereas the Akka- Sallaberger (2004) evaluated the language used dian context at the GarÍana garrison was in everyday communicative situations, in pas- responsible for the exceptionally high portion sages of administrative and legal documents and of Akkadian words. The acquisition of Sumer- in name-giving. Woods (2006) has partly inde- ian by the scribes thus extends far beyond a pendently developed similar arguments. In mere learning of a scribal language or an admin- very broad lines the language situation at the istrative idiom. This agrees perfectly with the end of the third millennium presents itself as first evaluation of Sumerian as a living language follows: a dominant Sumerian population in- at the very beginning of the philological anal- habited the South of Babylonia, especially the ysis, which was based on the lexical and mor- Tigris provinces of Umma and ªirsu, a fairly phological variation and the coherent phonol- mixed population lived in Middle Babylonia ogy that transpires in the phonographic spell- (Nippur–Isin region), but a dominantly Akka- ings (pp. 338–40). So Sumerian was present at dian population in the North. Gar-Íana as part of an active bilingualism of the GarÍana teaches us that we should not scribes, simply because it is not a “canonical” regard the region of a province, a former city Sumerian and because some of the variations state, as being linguistically homogenous: the are based on an active influence of Akkadian state organization of GarÍana employed domi- within the bilingual environment. The active nantly Akkadians in a Sumerian environment. bilingualism of the GarÍana scribes would have The active Akkado-Sumerian bilingualism of been unthinkable without constant communi- the GarÍana scribes has been the topic of this cative contacts with the native speakers of Sume- study. rian in the region. The fact that GarÍana Sumerian recogniz- ably differs from the Sumerian of other south- 6.5 On Akkado-Sumerian Bilingualism in the ern Babylonian archives, those from Ur, ªirsu, State of Ur or even Umma, the province capital situated This microstudy on the active Akkado-Sume- close to GarÍana, is the best evidence for the rian bilingualism at GarÍana in the Ur III period actual presence of Sumerian as dominant ver- contributes also to the larger debate on the rela- nacular at these places, and thus generally in the tionship between Sumerian and Akkadian, South. The language of documents stemming especially at the end of the third millennium, from the provincial administration that deals about one century before Sumerian disap- with persons of Sumerian background appears peared as a vernacular used in everyday con- more homogenous and consistent if compared texts. A renewed interest in this topic was with the products of the bilingual GarÍana

48 On Rubio (2006) see the careful argumentation by Keetman (2010). Sumerian Language Use at GarÍana 369

scribes. But the Sumerian linguistic environ- the imposition may have been integrated into ment clearly dominated the language use at the the language code. Such a development cannot multilingual settlement of GarÍana. be traced, however, due to the collapse of the Although at first sight one might be in- Ur III state immediately after the end of the clined to regard the Akkadian vocabulary in the GarÍana documentation. Only a few compari- GarÍana texts as a first sign of the attrition of sons can be made with the language of the Sumerian, the linguistic analysis presented Sumerian documents from Isin, which were above has proven the contrary. The texts dem- written only a few years after those of GarÍana onstrate a vivid bilingual situation with close and stem from a different region, namely mid- contact of both languages. And although the dle Babylonia with its higher portion of Akka- portion of personal names at GarÍana might dian population in the late third millennium. suggest that Sumerian would have been of The dramatic events at the end of the Ur III minor importance, people were motivated to period implied a far-reaching collapse of settle- acquire this socially dominant language to a ment patterns in Sumer and the social catastro- high degree of proficiency. phe eventually led to a rapid disappearance of However, as noted in the analytical part of Sumerian as an everyday language (Sallaberger this study, the GarÍana texts do deviate from 2004). Furthermore, no documents from the the standard corpus of Sumerian, especially in South stem from the period of political and the state archives or those from Umma and social disorder after the fall of Ur.49 Therefore, ªirsu. As the discussion has shown, the impo- it is impossible to know whether some of the sition of the scribes remained generally within modernizations of Sumerian that appear at Gar- the speech domain of the individual and often Íana were more widespread or remained just a had not yet become part of the language code local phenomenon. at GarÍana. Sumerian started a second life as the lan- We may not underestimate the effect of the guage of education, scholarship, poetry and changes caused by imposition from the side of religion during the Isin period. Even a superfi- the source language (SL) agents, the Akkadian cial glimpse at some of the features treated scribes, as soon as they would have been inte- above for GarÍana Sumerian reveals the funda- grated into the language code. “The SL contri- mental difference between the third-millenni- bution is essentially an SL penetration into the um Sumerian from an active bilingual context RL. Consequently, imposition (SL agentivity) and the literary language taught and spoken is not an addition to the RL as in borrowing mainly by non-native speakers. The Sumerian (RL agentivity), but rather a catastrophic from the late nineteenth century onward fea- change that affects the core of the RL” (Van tures significant changes in morphology (cf., Coetsem 1988, 40). And furthermore: “Viewed e.g., Wilcke 1998; Zólyomi 2000; Huber 2001, as a function of time, imposition will involve a 171–179); the phonological system is hidden greater concentration of structural change than behind morphographic writings (like dative -ra borrowing. ... This means that the rate of struc- even after vowel), Akkadian imposition is visi- tural change will be more gradual in RL agen- ble in the use of function words as the enclitic tivity (borrowing), and more abrupt in SL particle -ma, and in standard Sumerian texts agentivity (imposition)” (Van Coetsem 1988, the pronunciation is hidden behind a standard- 41). The GarÍana texts still show sporadic ized orthography that prefers logograms. But, imposition in the message domain, but in a sit- contrary to the GarÍana evidence, these Sume- uation of permanent bilingualism of this kind rian texts from the Old Babylonian period

49 The catastrophe at the end of the empire of Ur was (2008) for new documents concerning the agricul- apparently survived by the main temples as the core tural administration by the temples of Nin™irsu and institutions of LagaÍ and ªirsu; see Richardson of Bawu under Rim-Sin. 370 Walther Sallaberger

hardly use any Akkadian vocabulary. These REFERENCES systematic differences are taken as a most wel- Appel, René and Pieter Muysken come confirmation of the reconstruction of an 1987 Language Contact and Bilingualism. active bilingual environment at GarÍana. London: Edward Arnold. Attinger, Pascal ABBREVIATIONS 1993 Eléments de linguistique sumérien-

GarÍana texts are cited according to the scheme ne. La construction de du11/e/di “no. 414” etc., which refers to Owen and «dire». OBO Sonderband. Fribourg: Mayr (2007). References to cuneiform texts Éditions universitaires & Göttin- follow BDTNS: Database of Neo-Sumerian gen: Vandenhoeck & Ruprecht. Texts (http://bdts.filol.csic.es). 2005 “A propos de ak «faire»,” ZA 95, In the glosses the following abbreviations 46–76. 208–75. have been used: 2009 “Tableau grammatical du sumérien 1, 2, 3 person (problèmes choisis),” http:// A agent (in the verb) www.arch.unibe.ch/content/ ABS absolutive (case) mitarbeitende/mitarbeiterinnen/ Bh Óamˇu base prof_dr_p_attinger/e8254/e8548/ ERG ergative (case) GEN genitive e8549/e8639/ LOC locative Tableaugrammaticalrevu_ger.pdf MED “medium marker” ba- (Jagersma) (accessed 2009/12/27). NMLZ nominalizer -a Brunke, Hagan O object (in the verb) 2008 “The nakabtum. An Administrative P personal/human gender Superstructure for the Storage and PFV perfective -a (in infinite verbal Distribution of Agricultural Prod- forms) ucts,” Kaskal 5, 111–26. PL plural Edzard, Dietz Otto POSS possessive 2000 “Wann ist Sumerisch als gespro- S subject of intransitive clause (in the chene Sprache ausgestorben?” verb) Acta Sumerologica 22, 53–70. SG singular VNT ventive Gelb, Ignace J. In the final section note furthermore: 1961 Old Akkadian Writing and Grammar. MAD 2. 2nd ed. Chicago: The RL recipient language (of contact- University of Chicago Press. induced change) SL source language (of contact- Hasselbach, Rebecca induced change) 2005 Sargonic Akkadian. A Historical and Comparative Study of the Syllabic Texts. Wiesbaden: Harrassowitz. Heimpel, Wolfgang 2009 Workers and Construction Work at GarÍana. CUSAS 5. Bethesda, Md.: CDL Press. Huber, Fabienne 2001 “La Correspondence Royale d’Ur, un corpus apocryphe,” ZA 91, 169–206. Sumerian Language Use at GarÍana 371

Jagersma, Bram Molina, Manuel 2000 [published 2005]: “Sound Change 2008 “The Corpus of Neo-Sumerian in Sumerian: the so-called /dr/- Tablets: An Overview,” in S. J. phoneme,” Acta Sumerologica 22, Garfinkle and J. C. Johnson (eds.), 81–88. The Growth of an Early State in Meso- Keetman, Jan potamia: Studies in Ur III Administra- 2004a “Erneuerung des Syllabars im Ur tion. Biblioteca del Próximo Oriente III-Akkadischen: Schriftreform Antiguo 5. Madrid: Consejo Supe- rior de Investigaciones Científicas, oder Sprachwandel?” ZA 94, 186– 201. 19–53. 2004b “Verschlusslaute, Affrikaten und Notizia, Palmiro Frikative im Sumerischen,” Archiv 2009 I testi dei messaggeri da Girsu-LagaÍ Orientalní 72, 367-83. della Terza Dinastia di Ur. Nisaba 22. Messina: Di.Sc.A.M. 2010 “Enmerkar und Sulge als su- merische Muttersprachler nach lite- Owen, David I. and Rudolf H. Mayr rarischen Quellen,” ZA 100, 15– 2007 The GarÍana Archives. CUSAS 3. 31. Bethesda, Md.: CDL Press. Kleinerman, Alexandra and David I. Owen Richardson, Seth 2009 Analytical Concordance to the GarÍana 2008 “Ningirsu Returns to His Plow: LagaÍ and Girsu Take Leave of Ur,” Archive. CUSAS 4. Bethesda, Md.: CDL Press. in P. Michalowski (ed.), On the Third Dynasty of Ur. Studies in Honor Kogan, Leonid and Katja Markina of Marcel Sigrist. JCS Suppl. 1. Bos- 2006 [Review of Hasselbach 2005], Babel ton: American Schools of Oriental and Bibel 3, 555–88. Research, 153–58. Krecher, Joachim Rubio, Gonzalo 1995 “Die marû-Formen des sume- 2006 “ﬁulgi and the Death of Sumerian,” rischen Verbums,” in: M. Dietrich in P. Michalowski and N. Veldhuis and O. Loretz (eds.), Vom Alten (eds.), Approaches to Sumerian Litera- Orient zum Alten Testament. ture. Studies in Honour of Stip (H.L.J. Festschrift für Wolfram Freiherrn von Vanstiphout). Cuneiform Mono- Soden zum 85. Geburtstag am 19. Juni graphs 35. Leiden: Brill, 167–79. 1993. AOAT 240. Kevelaer: Butzon Sallaberger, Walther & Bercker [et al.] 141–200. 1993 Der kultische Kalender der Ur III-Zeit. Meyer-Laurin, Vera Untersuchungen zur Assyriologie 2010 “Die marû-Basen der sumerischen und Vorderasiatischen Archäologie 7. Verben túm „hin-, wegführen” Berlin & New York: W. de Gruyter und xe6 / de6 „bringen, liefern“,” 2000 [published 2005]: “Textformular ZA 100, 1–14. und Syntax in sumerischen Verwal- Michalowski, Piotr tungstexten,” Acta Sumerologica 22, 2006 “The Lives of the Sumerian Lan- 249–72. guage,” in: Seth L. Sanders (ed.), 2004 “Das Ende des Sumerischen. Tod Margins of Writing, Origins of Cul- und Nachleben einer altmesopota- ture. Oriental Institute Seminars 2. mischen Sprache,” in P. Schrijver Chicago, Ill.: The Oriental Institute and P.-A. Mumm (eds.), Sprachtod of the University of Chicago, 159– und Sprachgeburt. Münchner Fors- 86. chungen zur historischen Sprach- 372 Walther Sallaberger

wissenschaft 2. Bremen: Hempen, American Schools of Oriental 108–40. Research, 223–30. 2005 “‘bringen’ im Sumerischen. Lesung Wilcke, Claus

und Bedeutung von de6(DU) und 1998 “Zu ‘GilgameÍ und Akka’. Überle- túm(DU),” in R. Rollinger (ed.), gungen zur Zeit der Entstehung Von Sumer bis Homer. Festschrift für und Niederschrift [...],” in M. Manfred Schretter zum 60. Geburtstag Dietrich and O. Loretz (eds.), dub- am 25. Februar 2004. AOAT 325. sar anta-men. Studien zur Altoriental- Münster: Ugarit-Verlag, 557–76. istik. Festschrift für Willem H.Ph. Schneider, Nikolaus Römer. AOAT 253. Münster: 1936 Die Zeitbestimmungen der Wirt-schafts- Ugarit-Verlag, 457-86. urkunden von Ur III. Analecta Ori- 2000 Wer las und schrieb in Babylonien und entalia 13. Roma: Pontificio Istituto Assyrien. Überlegungen zur Literalität Biblico im Alten Zweistromland. Sitzungs- Soden, Wolfram von berichte der Bayerischen Akademie 1995 Grundriß der akkadischen Grammatik. der Wissenschaften, Philosophisch- AnOr. 33. 3. erg. Aufl. Roma: Pon- Historische Klasse Jg. 2000, H. 6. tificio Istituto Biblico. München: C.H. Beck in Kommis- Sommerfeld, Walter sion. 1999 Die Texe der Akkade-Zeit 1: Das Winford, Donald Dijala-Gebiet: Tutub. Imgula 3/1. 2007 “Some Issues in the Study of Lan- Münster: Rhema. guage Contact,” Journal of Language Streck, Michael P. Contact – Thema 1 (2007) 22–40 2004 “Die Amurriter der altbabyloni- (www.jlc-journal.org, accessed 2009 schen Zeit im Spiegel des Onomas- /12/27). tikons. Eine ethno-linguistische Woods, Christopher Evaluierung,” in J.-W. Meyer and 2006 “Bilingualism, Scribal Learning, W. Sommerfeld (eds.), 2000 v. Chr. and the Death of Sumerian,” in S. – Politische, wirtschaftliche und gesell- L. Sanders (ed.), Margins of Writing, schaftliche Entwicklung im Zeichen einer Origins of Culture. Oriental Institute Jahrtausendwende. 3. Internationales Seminars 2. Chicago, Ill.: The Ori- Colloquium der Deutschen Orient- ental Institute of the University of gesellschaft [...]. Saarbrücken: SDV, Chicago, 91–120. 313–55. Zólyomi, Gábor Van Coetsem, Frans. 2000 [published 2005]: “Structural Inter- 1988 Loan Phonology and the Two Transfer ference from Akkadian in Old Types in Language Contact. Dord- Babylonian Sumerian,” Acta Sumer- recht: Foris. ologica 22, 335–60. 2000 A General and Unified Theory of the Transmission Process in Language Con- tact. Heidelberg: Winter. Veldhuis, Niek 2008 “Orthography and Politics: adda,

‘carcass,’ and kur9 ‘to enter’,” in P. Michalowski (ed.), On the Third Dynasty of Ur. Studies in Honor of Marcel Sigrist. JCS Suppl. 1. Boston: