Where did one speak luwili? Geographic and linguistic diversity of Luwian cuneiform texts

Размещено на http://www.allbest.ru/

Размещено на http://www.allbest.ru/

Where did one speak luwili? Geographic and linguistic diversity of Luwian cuneiform texts

Alice Mouton, Ilya Yakubovich

Centre National de la Recherche Scientifique, Paris;

Institute of Oriental Studies, Russian Academy of Sciences;

The purpose of this paper is to assess complications in Luwian dialectal geography in the second millennium BCE, which became apparent in the course of the ongoing work on the edition of Luwian cuneiform texts. On the one hand, a number of Luwian incantations embedded into the ritual traditions of Puriyanni and Kuwattalla (CTH 758-763) and traditionally assigned to the dialect of Kizzuwadna in the southwest of Asia Minor can now be linked to the Lower Land in the central and central-western part of Asia Minor. The increasing Kizzuwadna features of the Kuwattalla tradition, including the Hurrian loanwords in the respective texts, likely reflect its secondary evolution at the court of Hattusa. On the other hand, a large group of Luwian conjurations that is booked under CTH 764-766 can now be linked to the town of Taurisa situated to the northeast of Hattusa. Their language shows dialectal peculiarities, while their formulaic repertoire finds non-trivial parallels in Hattic and Palaic texts. The concluding part of the paper addresses the relevance of these new empirical findings for the dialectal classification of the Luwian language.

Keywords: Luwian language; Hurrian language; Hattic language; Palaic language; Hattusa; Kizzuwadna; Lower Land; Taurisa.

ИльяЯкубович, АлисМутон.Гдеговорилиluwili?Географическое и лингвистическое многообразие лувийских клинописных текстов

Задачей данной статьи является разбор уточнений диалектной географии лувийского ареала во втором тысячелетии до н.э., ставших необходимыми в процессе работы по изданию лувийских клинописных текстов. С одной стороны, лувийские заклинания в ритуальных традициях Пуриянни и Куватталлы, традиционно относимые к диалекту Киццувадны, характерному для юго-востока Малой Азии, также обнаруживают связи с ареалом Нижней страны, расположенной в центральной и западно-центральной частях малоазиатского региона. Усиление влияния Киццувадны на традицию Куватталлы, включая увеличение числа хурритских заимствований в соответствующих текстах, отражает их вторичное бытование при дворе Хаттусы. С другой стороны, происхождение значительной группы лувийских заговоров, соотносимой с каталожными номерами CTH764-766, можно связать с городом Тауриса, локализуемым к северо-востоку от Хаттусы. Язык этих заговоров имеет свои диалектные особенности, а их формульный репертуар обнаруживает нетривиальные параллели в хаттских и палайских текстах. В заключительной части статьи обсуждается значение новых эмпирических выводов для общей классификации лувийских диалектов.

Ключевые слова: лувийский язык; хурритский язык; хаттский язык; палайский язык; Хаттуса; Киццувадна; Нижняя страна; Тауриса.

Introduction

The Luwian language was deployed in writing in Asia Minor in the second and early first millennium BCE and is attested in two distinct writing systems: the adaptation of Syro- Mesopotamian cuneiform and the indigenous Anatolian hieroglyphs. This paper is written under the auspices of the international project “Luwili: Luwian Religious Discourse between Anatolia and Syria”, co-directed by both authors of this paper and funded by the ANR (France, ANR-17- FRAL-0007-01) and DFG (Germany, YA 472/2-1). Susanne Gцrke (Marburg), H. Craig Melchert (Carrboro, NC), and David Sasseville (Marburg) read the draft of this article and made valuable remarks, while Vladimir Shelestin (Moscow) advised us on specific issues related to the topic of our investigation. Manfred Hutter (Bonn), Elisabeth Rieken (Marburg), and Zsolt Simon (Mьnchen) made available to us their forthcoming papers. All the individuals and organisations mentioned above deserve our heartfelt gratitude. All the linguistic generalisations about the Luwian corpora, unless attributed otherwise, are made on the basis of the ACLT database (web- corpora.net/LuwianCorpus/search/) and can be independently replicated by other users. The following specialised bibliographic abbreviations are used in the text of this paper: CHD - The Hittite Dictionary of the Oriental Institute of the University of Chicago, H.G. Gьterbock, H. Hoffner, and Th. van den Hout (ed.); CTH - Catalog der Texte der Hethiter(www.hethport.uni-wuerzburg.de/CTH); KBo - Keilschrifttexte aus Boghazkцi, Leipzig, Berlin; KUB - Keilschrifturkunden aus Boghazkцi, Berlin. The Luwian hieroglyphic inscriptions, most of which are edited in Hawkins 1995 and Hawkins 2000, usually represent independent compositions: some of them have parallel versions in other languages, but these are recorded in different writing systems. In contrast, the Luwian cuneiform texts, the bulk of which is published in transliteration in Starke 1985, are almost invariably embedded into the Hittite narrative frame. The Luwian insertions are commonly introduced by Hit- tite sentences such as KUB 9.31 ii 20-21 nu lu-u-i-li ki-is-sa-an hu-uk-ki-is-ke-ez-zi `(s)he conjures thus in Luwian'.

Such a state of affairs has a sociolinguistic explanation: the available cuneiform tablets with Luwian passages all emanate from the chanceries of Hattusa, where the main written language was Hittite, a close relative of Luwian. In contrast, the Luwian language was not regarded as suitable for the composition of official cuneiform texts in Hattusa chanceries. There are cases where direct speech utterances are introduced by the adverb luwili “in Luwian”, but then translated into Hittite. The embedded Luwian passages that avoided translation usually represent incantations, and one can assume that they were recorded in the original language because of their illocutionary force. A different sort of code-switching is the use of isolated Luwian words in Hittite texts. These occur in a wide variety of textual genres, predominantly in the texts written in the New Hittite ductus, and are frequently marked by special signs known as “gloss wedges” (Glossenkeile).

It was traditionally assumed that the two graphic systems deployed for writing Luwian were used for recording two different dialects. Consequently, the taxonomic terms Cuneiform Luwian and Hieroglyphic Luwian became standard among the Anatolianists (see e.g. Mel- chert 2003: 170-175). This basic dichotomy was challenged in Yakubovich 2010, primarily with reference to the status of the “Glossenkeil words”. luvian area hattusa kuvattalla

Developing some observations that are already found in Melchert 2003 and van den Hout 2006: 236, Yakubovich (2010) argued that these foreign words in Hittite cuneiform texts essentially reflect the same dialect as that of the hieroglyphic inscriptions. The lack of association between the “Glossenkeilwords” with specific texts or even genres led him to the hypothesis that they reflect the Luwian dialect of the elites of Hattusa, the authors of the bulk of the compositions found in the cuneiform archives of this city. Since the longer Luwian hieroglyphic inscriptions of the Late Bronze Age all belong to the kings of Hattusa, it is easy to see how they can reflect the same dialect. The name “Empire Luwian”, proposed for this dialect in Yakubovich 2010, reflects the assumption that the prestigious variety of Luwian spoken in the capital was imitated by the provincial elites of the Empire of Hattusa (also known as the Hittite Empire). This explains why Iron Age Luwian / Late Luwian, the dialect of hieroglyphic inscriptions emanating from various Syro-Anatolian States (“Neo-Hittite States”) and continuing the cultural tradition of the Empire of Hattusa in 12^th-8^th centuries BCE, represents the descendant of Empire Luwian.

As for the Luwian incantations embedded in Hittite cuneiform texts, they must reflect dialects other than Empire Luwian on linguistic grounds. Some of them, e.g. the Luwian passages in a festival text KUB 35.133(+) (CTH 665), may have been dictated in the Luwian dialect of Hattusa before the formation of the imperial koine (cf. Yakubovich 2010: 21). Another special case are the Songs of Istanuwa (CTH 771-772), which were thought to reflect a particular Luwian dialect at least since Laroche 1959. These incantations, however, are relatively short and do not form the core of the Luwian cuneiform corpus. The label that Yakubovich (2010) assigned to the Luwian dialect determined to constitute the main counterpart of Empire Luwian was Kizzuwadna Luwian.

The new primary dichotomy in the classification of the Luwian dialects received a favourable response in the last decade. Among the papers who expressed support for the new solution are Melchert 2013: 159-160, Rieken 2017: 301-302, Giusfredi 2018: 80; it is also presented without objections in Hawkins 2013: 28.possessive construction marking the plurality of the possessors. We shall return to this problem in the concluding section of this paper. For the time being, however, it is only important to observe that Simon 2016 accepts

Kizzuwadna Luwian as a taxonomic unit. It seems, however, that this near-consensus is pri-

Simon 2016 represents an attempt to refute Hurrian influence on Kizzuwadna Luwian, which was presented in Yakubovich 2010 as the motivation for the key innovation of this dialect, namely the rise of a special marily driven by the evaluation of linguistic isoglosses separating the two basic groups of texts, such as the distribution of the innovative acc.pl ending /-ntsi/ or the possessive constructions with plural possessors. Some scholars, notably Hawkins (2013), also address in passing the possible role of Hattusa scribes in the formation of Empire Luwian, but the geographic reality behind the term “Kizzuwadna Luwian” so far played little, if any role in the validation of the new hypothesis.

This is perhaps not a coincidence, because the positive arguments advanced thus far for the justification of the label Kizzuwadna Luwian are limited. There are only two cuneiform texts with Luwian insertions that contain internal references to Kizzuwadna. One of them is the well-preserved ritual of Zarpiya (CTH 757), which contains both Luwian and Hittite incantations, probably implying metaphorical code-switching (Yakubovich 2010: 282-283). The other is a tiny fragment KUB 35.8 (Starke 1985: 43).^{* The analysis of the formulaic content of the fragments undertaken with the framework of the Luwili project is conducive to grouping together KUB 35.8 with KBo 29.36, KBo 29.47, KUB 32.124, KUB 35.65, and KUB 35.68, all featuring Luwian incantations. The peculiarities of KUB 35.8 that speak in favour of such a cluster is the mention of a sheep in combination with a bovine, presumably as sacrificial animals, and the reconstructed reference to the tandem of an evil woman and an evil man. We assign the number CTH 763.1 to this group and believe that it eventually became integrated into the Kuwattalla tradition (for the general problem of Kizzuwadna influence on the evolution of the Kuwattalla tradition, cf. Section 2.2 below). At the same time, the form} si-wa-an-na attested in KUB 35.8 i 6 can hardly be separated from Hitt. siwanna/i-, a noun of unknown meaning occurring in the texts of the Tunnawiya tradition (CHD S: 488a). Note in particular its occurrence in the fragment KUB 35.146, which features code-switching between Hittite and Luwian, but also exhibits significant parallels with the Tunnawiya tradition, as convincingly argued in Pisaniello 2015. None of the other Luwian cuneiform texts available to date can be unambiguously assigned to Kizzuwadna, while some of them contain internal geographic references pointing in a different location. The main rationale behind assigning a variety of texts with Luwian insertions to Kizzuwadna was the presence of Hurrian loanwords in the respective compositions.

Today the availability of complete annotated corpora of Luwian cuneiform texts and the ongoing work on their translation undertaken within the framework of the Luwili project facilitate their analysis at a deeper level. In what follows we intend to offer a more nuanced version of the geographic and linguistic classification of Luwian incantations in cuneiform transmission. In Section Two, we intend to argue that a large group of cuneiform rituals with Lu- wian insertions yield conflicting cues as to their origin, showing significant parallels with the ritual tradition of the Lower Land of Hittite sources (parts of Classical Lycaonia and Cappadocia) as well as convergence with the Kizzuwadna rituals. See Yakubovich 2014 (submitted 2009) for a new etymology of Cappadocia, derived from what appears to have been the Hittite designation of the Lower Land. An independent account advocating the same connection but exhibiting formal differences is Casabonne 2012. In Section Three, we shall endeavour to demonstrate that another group of Luwian cuneiform texts, which received only cursory treatment in Yakubovich (2010), reflects the tradition of the town of Taurisa, probably situated to the northeast of Hattusa. Part of the texts belonging to both traditions is recorded in Middle Script, which pleads for their written fixation no later than in the Early New Kingdom period (early 14^th century BCE). In the concluding Section Four, we shall explore the repercussions of the new data for the prehistory of the Luwian language and the development of its individual dialectal isoglosses.

1. Ritual Tradition of the Lower Land

1.1 Tunnawiya and Kuwattalla

The goal of this subsection is to address similarities between the ritual traditions of Tunnawiya and Kuwattalla, which in turn represent an argument for the Lower Land origin of many fragments with Luwian insertions. Before proceeding to the discussion of individual texts, it is appropriate to address briefly our understanding of the word “tradition”. We accept the arguments presented in Miller 2004 and Christiansen 2006 in favour of the ongoing modification of ritual texts in Hattusa chanceries and believe that the list of secondary products of court scholarship is likely to be extended in the future. The elite group of scholar-scribes, discussed in van den Hout 2015, would supply the most likely milieu for the modification of the pre-existent ritual texts. At the same time, we accept the historicity of ritualists mentioned in the Hittite sources, and believe that at least some rituals were recorded in close cooperation with their practitioners. An illustration of such a phenomenon has been provided in Marcuson and van den Hout 2015. The last assumption is particularly necessary in the instance of rituals with Luwian incantations, which reflect dialects other than that of Hattusa and therefore could not represent a product of Hattusa scribes. For the discussion of dialectal interference in the course of copying Luwian incantations, see Yakubovich 2010: 28-29. This means in practice that the tradition associated with a particular ritualist (based on the combination of internal references and the commonality of structural features) may contain both original texts recorded from the respective performer and the result of their adaptation and amalgamation in scribal circles (or by other ritual practitioners). In contrast with Miller 2004: 522, cf. the more nuanced discussion in Pisaniello 2015: 31-32.

The Lower Land is the name assigned in Hittite sources to the southern part of the Central Anatolian Plateau from the 14^th century BCE onward. Manfred Hutter (2003: 243-247) discussed at some length the festival of Huwassanna, worshiped as the divine queen of Hupisna, as a specimen of Lower Land religious literature. This is a straightforward assumption, given that the Bronze Age toponym Hupisna can be identified Hellenistic KuЯiaxpa, the name of a town in southern Cappadocia (Laroche 1979: 67, fn. 25). The rituals of Ambazzi constitute another likely specimen: besides the Luwian names of supernatural entities mentioned in CTH 391 (Alauwaimi and Tarpatassi), the sacrificial practice described in CTH 391 is similar to that of the Hupisna festivals (Mouton 2012: 133-134). But the most straightforward candidates for exponents of the Lower Land tradition within the corpus of Hittite rituals is a group of compositions attributed to the Old Woman Tunnawiya (CTH 409).

The name Tunnawiya can be most straightforwardly explained as “sent by (the god of) Tunna” (Mouton 2015: 86, modifying the hypothesis of Yakubovich 2013a: 102-103). From the structural viewpoint, this name is parallel to some other female theophoric onomastic compound containing surface toponyms, such as Halpawiya “sent by (the Storm-god) of Aleppo” or Ziplantawiya “sent by (the Storm-god) of Zippalanda”. Note that if one accepts the earlier interpretations of these names as `woman of Tunna', `woman of Aleppo', `woman of Zippalanda' etc., their connection with the respective toponyms becomes only more straightforward. For reasons to reject the interpretation of the Luwian element /wija-/ as `woman', see Yakubovich 2013a. It is, however, to be noted that unlike the cults of Aleppo and Zippalanda, the cult of Tunna did not play a prominent role in the Kingdom of Hattusa at the state level. Therefore, an individual carrying the name Tun- nawiya is likely to have a family connection with the town of Tunna, situated in the Lower Land (probably the archaeological site of Zeyve Hцyьk - Porsuk) and well-attested in Hittite sources (del Monte and Tischler 1978: 439). There is no contradiction between this hypothesis and the fact that Tunnawiya is called ^MUNUSSU.GI i^URUHATTI `Old Woman of Hattusa' in KBo 21.1 i 1, because she may have been practiced in Hattusa in adulthood, or, somewhat less likely, the scribe may have associated her with the whole Kingdom of Hattusa rather than its capital.

Another consideration fleshes out the connection between Tunnawiya and South-central Anatolia. A distinct feature of the Tunnawiya tradition is the presence of the ^DIM ariyattalli-, which can be literally interpreted as “Storm-god of the Crag”. The rationale for interpreting the adjective /arittali(ja/i)-/ as referring to an elevated landscape feature is its perceived connection with the Luwian verb /ari-(ti)/ `to raise, rise'. The traditional interpretation of Luw.*ariyatt(i)- was `mountain' but Gйrard (2006) plausibly argued that the principal Luwian word for `mountain' is /watt(i)-/. Given the geographic distribution of the divine epithet /arittali(ja/i)-/, it is likely that it refers to a specific landscape feature that was prominent in the area under discussion, such as the protruding rock formations, typical of Cappadocia and the areas immediately to the south and constituting the local tourist attraction. Now the same deity occurs with a different possessive suffix in a curse formula of an 8<sup>th</sup>-century hieroglyphic text KU- LULU 1 (cf. Hutter 1988: 67-68), a new translation of which is provided under (1) below. The interpretation of ha-pa-za-nu-wa/i- as `to make attached' is based on the comparison of this verb with Luw. /xab(a)i-(di)/ `to bind', commonly used in ritual incantations, such as (6) below, and /xabantsu-/ `loyal, attached', for which see Melchert 1988: 236-240. The broader context of KULULU 1 makes it clear that after being restrained with the help of a mountain the evil-doer will be devoured by a supernatural dog. For our purposes, it is significant that both Kululu and Tunna ultimately belonged to the geographic area known as Tabal in the early first millennium BCE. It is, therefore, likely that the cult of the “Storm-god of the Crag” is identified with the area of Tabal, which in turn exhibits significant overlap with the late-second-millennium Lower Land (cf. Hutter 2003: 248). Note that KULULU lead strip 1 makes references to the towns of Upper and Lower Tun(n)a (Hawkins 2000, II: 506-507). The preferred tentative hypothesis of Hawkins 2000, II: 431-432 is to locate the twin towns in the immediate vicinity of Kululu, separating them from Tunna of the Hittite sources, although he admits various possibilities (cf. Hawkins 2000, II: 432, fn. 75). For fairness' sake one must observe that the cite of Kululu is situated outside the borders of the historical Lower Land, not far from the Bronze Age town of Nesa and modern Kayseri. The Lower Land was, however, the most probable source of Luwian migrations or language shift into this area, which must previously have been populated by the speakers of Hittite/Nesite.

The anthropological side of the Tunnawiya rituals is addressed in Mouton 2015. Comparing them with three established clusters of ritual texts, Arzawa rituals, Kizzuwadna rituals, and texts with Hattian cultural background, Mouton comes to the conclusion that they display certain parallels with all of the three groups: the ritual use of combs is shared with Arzawa, that of wine for purification and ritual spitting into the mouth of an animal-substitute are common with Kizzuwadna, the conjuration of the nine/twelve body parts of the patient is inherited from the Hattian substrate. At the same time, the Tunnawiya rituals possess specifictraits which they do not share with any other ritual tradition: the ritual use of a model of the wawarkima- door element and the mention of the “Storm-god of the Crag” are unique within the ritual corpus of Hittite Anatolia. This situation is perfectly consistent with the geographic location of the Lower Land at the crossroads of Anatolia, in the centroid of a triangle formed by Arzawa, Kizzuwadna, and the region of Hattusa.

The research history, summarised in Mouton 2015: 85-86, bears out the identification between the Tunnawiya tradition and the town Tunna in the Lower Land as the mainstream solution, advocated already in the first edition of the best-preserved ritual text belonging to this tradition (Goetze/Sturtevant 1938: 28, cf. Hutter 1988: 56, Hutter 2003: 248, Miller 2004: 453, Yakubovich 2010: 20). Miller (2004: 452-458) adduces a number of specific textual arguments against the Kizzuwadna origin of the compositions mentioning Tunnawiya. It is, therefore, our aim to pursue the implications of this discussion for the geographic origin of a closely related tradition, which is associated with the attendant woman Kuwattalla and the Old Woman Silalluhi (CTH 759-763). The compositions of this group, some of which are recorded in Middle Script, are treated among the Kizzuwadna rituals in Hutter (2003: 253-254) and Hutter 2019, while their incantations are booked as specimens of the Luwian dialect of Kizzuwadna in Yakubovich 2010: 18-20 and Mouton 2014: 579.

We now submit that the attempts to assign the Tunnawiya and Kuwattalla traditions to separate quarters would be extremely unlikely in view of the close and non-trivial similarities between the ritual formulae used in the two groups of texts. A particularly well-studied case is that of the Hittite incantations embedded in the “Ritual of the Ox” (KUB 9.4+) and the quasiparallel Luwian incantations in KUB 35.43(+). The “Ritual of the Ox”, although lacking internal attribution, was safely assigned to the Tunnawiya tradition based on the parallelism of many of its parts with those of the taknaz da- ritual of Tunnawiya (Beckman 1990, cf. Mouton 2015: 81). Starke (1985: 136) linked KUB 35.43(+) to the Kuwattalla tradition in view of the transparent similarities of its Luwian incantations with those assigned to Kuwattalla's “Great Ritual” on the basis of their colophons. Both attributions stood the test of time, and consequently the two texts are now booked under CTH 409 and CTH 761 respectively. But long before they were made, Laroche (1959: 147-151) effectively used the formulaic parallelism between the same texts in order to approach the interpretation of the Luwian incantations. This explains why these two texts had originally been assigned the number CTH 760 (Laroche 1971: 136). Laroche had originally reserved this number for the texts of the Kuwattalla tradition with demonstrable parallels in the Tunnawiya tradition. Note that the attribution of KUB 35.43(+) to Tunnawiya is still argued based on the structural features of this text in Marcuson 2016: 290. It is fair to say that KUB 9.4+ and KUB 35.43(+) functioned as quasi-bilinguals for decipherment purposes. This point can be illustrated with the following parallel passages, which refer to the manipulations involving a scapegoat and the ritual patron:

(3) KUB 35.43(+) ii 16-18 (CTH 761), cf. Starke 1985: 144

i-pa-la-a-ti-du-wa-anhu-i-[n]u-wa-ah-ha a-du-ut-tai-pa-la-a-ti-en

ibaladi=du(w)=anx^winuwaxxaa=du=ttaibaladin

left.INSTR=he.DAT=he.ACC.C drive.1SG.PRTPTCL=he.DAT=PTCL sinisterness.ACC.SG

la-at-tai-sar-u-i-la-t[i-p]a-du-wa-anhu-u-i-nu-wa-ah-ha

lattaisarwiladi=ba=du(w)=anx^winuwaxxa

take.3SG.PRTright.INSTR=COORD=thee.DAT=he.ACC.C drive. 1SG.PRT

a-du-ut-taat-tu-wa-li-inh[a-a]t-ta-as-ta-ri-inla-at-ta

a=du=ttaattuwalinxattastarinlatta

PTCL=he.DAT=PTCLevil.ACC.SG.Cterror.ACC.SGtake.3SG.PRT

`I drove him to his left, so that he took his sinisterness. I drove him to his right, so that he took his evil terror.'

In fact, the comparison of (2) and (3) is conducive to a stronger claim, namely the reconstruction of Luwian incantations in the redactional history of the “Ritual of the Ox”. As observed in Beckman 1990: 51, the vacillation between the second and the third person pronouns with reference to the ritual patron in (2) implies that the text “is obviously not in order here”. The comparison with (3) helps to qualify the origin of this error: the Luwian indirect object pronominal clitic /=du/ can mean both “to you (sg.)” and “to him”, and the distinction between the second- and third-person interpretations can be established only by context. The corpus analysis of the Kuwattalla rituals suggests that the ritual patron is always addressed there in the third person, hence the translation proposed for (3) above. But the scholar-scribe responsible for translating some of the relevant incantations from Luwian into Hittite and embedding them into the Tunnawiya tradition did not attempt to generalise over the Kuwattalla corpus. Therefore, he was understandably confused, since the pragmatics of the incantation under discussion is compatible as such with both second-person and third-person interpretations.

If the parallel discussed above were isolated, one could argue that it represents an instance of secondary convergence between originally unrelated traditions. This is, however, not the case. The structural similarities between the texts attributed to Tunnawiya and Kuwattalla played a prominent role in the recent dissertation Marcuson 2016, written within the anthropological paradigm. Focusing on the role of the “Old Woman” in these ritual texts, Hannah Marcuson did not either tackle the question of the traditional CTH numbering of the latter or pursue the implications of her analysis for localising the Kuwattalla tradition, presumably because this topic lay outside the immediate scope of her dissertation research. But the degree of similarity between these texts, as discussed by Marcuson, militates against the assumption that they reflect practices of two different geographic areas. Therefore, if one accepts the evidence for connecting Tunnawiya with the Lower Land, we gain a serious argument for assigning the origin of the Kuwattalla tradition to the same region. A comprehensive list of non-trivial similarities between the incantations of the Tunnawiya and Kuwattalla traditions was provided by Alice Mouton in her presentation at the conference “Contacts in Pre-Hellenistic Anatolia and Ancient Near East - From Languages to Texts” (Verona, 25-27 February 2021). The content of this talk will be incorporated into the philological edition of the Puriyanni and Kuwattalla traditions, currently in preparation by both authors of this paper.

Once this step has been made, this opens the possibility that the ritual for the purification of the house attributed to Puriyanni (CTH 758) likewise can be connected with the Lower Land rather than Kizzuwadna. The Luwian incantations embedded within this text do not show any resemblance to those of the Kizzuwadna ritual attributed to Zarpiya (CTH 757) but find close parallels within the Kuwattalla tradition. Thus, in both cases we find incantations prescribing the activities of the scapegoat. One feature they have in common is the literary figure consisting in the use of merisms for the classification of the negative phenomena to be carried away: “past or present/future, internal or external, of the living or the dead...” (Mouton and Yakubovich 2019). Another feature, which also finds parallels in Tunnawiya's incantations, is the enumeration of the scapegoat's body parts (cf. e.g. KUB 35.54 iii 9-11, Starke 1985: 68 vs. KUB 35.43(+) iii 24'-27', Starke 1985: 143). An additional set of incantations common to the Puriyanni and Kuwattalla rituals introduces Luw. /talupp(i)-/ `a lump of dough', which apparently also has the ability to carry away the miasma (cf. e.g. KUB 35.55:5'-7', Starke 1985: 70-71 vs. KUB 32.9(+) obv. 2-6, Starke 1985: 87). The mentions of a “pure taluppi-” in the so-called Ritual of Kizzuwadna (CTH 479.1: see Ьnal 2017, § 4'-5') might reflect the permeability of certain ritual traits among neighbouring regions (in this case Lower Land and Kizzuwadna). For more instances of the same phenomenon, cf. the discussion in the following section. Since the origin of Puriyanni is not mentioned anywhere in the text, the hypothesis of its connection with the Lower Land must also be given a fair hearing.

A linguistic argument in favour of localising both Kuwattalla and Puriyanni traditions in the Lower Land comes from the syntax of Luwian incantations in the respective corpus. As maintained in Mouton and Yakubovich 2020, their distinctive feature is the proleptic construction, which combines verbal fronting and clitic doubling. The exhaustive English-language discussion of the proleptic construction will be presented in an appendix to our forthcoming edition of the Puriyanni and Kuwattalla traditions. For example, the literal way of saying `The ritual patron is breaking the evil tongue' found in these incantations is “(He) is breaking it, the ritual patron, the evil tongue” (KBo 29.3+ iii 17'). As argued in the same paper with reference to Adiego 2015, this Luwian construction is situated halfway between the verb-final syntax typical of most Luwian dialects and the Lycian construction with nasalised preterit, which ultimately reflects the grammaticisation of redundant clitic pronouns appended to clause-initial verbal forms (Mouton and Yakubovich 2020: 213-214). Since the Luwian prolep- tic construction is demonstrably innovative, linking its origin to the dialectal area that was adjacent to the territory of (pre-)Lycian language community represents the most economical solution. The Lower Land, situated as it was in the central-western part of Asia Minor, clearly qualifies better as such an area that Kizzuwadna, even though one cannot tell precisely how far this innovation eventually spread to the east.

A religious argument in favour of localising Puriyanni ritual tradition in the Lower Land is the mention of the divine epithet parattassi- `of impurity' attributed to the Storm-god of the Open Country (KUB 7.14(+) i 2-3). This epithet can only be found in one other religious text, namely KBo 29.33+ iii 6' (CTH 694.1) which is a fragment describing a festival for Huwas- sanna, the most important distinct goddess of the Lower Land. In this fragment, the epithet also qualifies a Storm-god. We are grateful to Laura Puйrtolas Rubio for bringing this point to our attention.

1.2 Kuwattalla, Silalluhi, and Mastigga

Against such a background, one has to re-examine the arguments that were traditionally adduced for the Kizzuwadna connections of the Kuwattalla tradition. They were recently summarised in Kaynar 2017: 190-191 and Kaynar 2019: 108 with reference to the earlier work of other scholars (Hutter 2003, Yakubovich 2010, Beckman 2011, Melchert 2013). This list includes the appearance of several Hurrian theonyms, such as Hebat, Sawoska, and Ninatta, the use of the West Semitic loanword /xalal(i)-/ `pure', a reference to purification by blood, and the Hittite technical terms keldi- and nakkussi-, both of Hurrian origin. We intend to argue that

Kizzuwadna features of the relevant rituals mostly arose in the course of their adaptation in Hattusa.

The hypothesis of secondary interference is compatible with what we know on extralin- guistic grounds. The incipits or colophons of several texts belonging to the group under discussion attribute them not to the attendant woman Kuwattalla but to the Old Woman Silal- luhi, or the tandem of both ritual practitioners. As already suggested by Starke (1985: 74), the texts of the “Great Ritual” with colophons mentioning Kuwattalla alone, namely KUB 35.24+ and KUB 32.9(+), belong to the oldest layer of the tradition. The work of the Luwili project was conducive to confirming that these two pieces belong to the same manuscript, to which one can also assign the smaller fragments KUB 32.10+, KUB 35.23, and KBo 29.15. We classify this earliest version of the Great Ritual as CTH 761.1. No Hurrian loanwords have been identified within this group thus far, which does not mean that they could not exist in the lost portions of the relevant manuscript, but suggests that they were infrequent. In contrast, Silalluhi's name certainly has a Hurrian origin, cf. the Hurrian professional title silalluhi (Richter 2012: 375). Accordingly, it is tempting to hypothesise that certain Hurrian elements were introduced into the Kuwat- talla tradition when Silalluhi established collaboration with Kuwattalla or undertook a revision of her rituals. It seems, however, unlikely that the “Old Woman” Silalluhi was acting alone, presumably the modification of the tradition reflected the expectations of the ritual patrons.

If the Hurrian elements represent a secondary phenomenon within the tradition of Ku- wattalla, it is worth asking where and when they may have been added. On can approach the answer to this question from the prosopographic viewpoint. We know that king Arnuwanda I and Queen Asmunikkal granted land to the attendant woman Kuwattalla, presumably in reward for her services (Hutter 2003: 253). If we exclude a hypothesis of two different women sharing the same name and title, this is as close as we can get to actually proving the connection between Hattusa and the first written record of the Kuwattalla tradition. But the implied chronology also accommodates well the Hurrian influence upon its subsequent development. We know that new Hurrian rituals were composed in Hattusa and/or Sapinuwa during the reign of Tudhaliya II/III, son of Arnuwanda I, who is also known under the Hurrian name Tasmi-Sarri. Furthermore, several Hurrian compositions, of which the Song of Release and the Kumarbi cycle are the best-known examples, probably reached Hattusa at about the same time. The question of Hurrian impact on the state cult of Hattusa in the Early New Kingdom is likely to acquire a new dimension after the comprehensive publication of texts from Ortakцy, but in the meanwhile see Corti 2017b as a recent stance on this complicated issue. The prestige of the Hurrian religion in Hattusa in mid-fourteenth century BCE may have also inspired the efforts of Silalluhi leading to the adaptation of the Hittite-Luwian rituals from the Lower Land.

A likely trace of such an adaptation is the appearance of Hurrian concepts in the incipits KUB 35.18(+) and KBo 29.3+ The direct join KBo 29.3 + KUB 35.45, made first by Annelies Kammenhuber but largely ignored in subsequent scholarship, was recently reaffirmed in Sasseville 2020: 113-114. For the parallel description of scapegoat activities in this text and the Ambazzi ritual, see Marcuson 2016: 295-296. introducing the combined performance of the ritual katta walhuwas (literally “of striking down”) and the “Great Ritual”. Cf. e.g. KUB 35.18(+) i 2-7: ma-a-an an-tu-uh-s[i] kat-ta wa-al-h[u-u-wa-as SISKU]R si-pa-an-du-wa-ni na-as-ta ma-ah-ha'-an 4-NA' U4.3.KAM kat-ta wa-al-hu-u-wa-as S[ISKU]R as-nu-me-ni [n]a-an I-NA U4.3.KAM pa-ra-a GAL-77- pвt a-ni-u-u[r a]p-pu-ы-e-ni nu ki-i tum-me-ni `When we perform the ritual katta walhuwas for a person, and when we complete the ritual katta walhuwas on the third day, on the (same) third day we take up the “Great Ritual”, and we take the following (implements)'. The katta walhuwas ritual is the Hittite rendering of the name of the dupaduparsa- ritual (Hutter 2019: 381 and Sasseville 2020: 111 with ref.). This version of the Kuwattalla tradition, which we book under CTH 760, demonstrably postdates CTH 761.1. The first one is attributed to both Kuwattalla and Silalluhi, the performer of the second one is a practitioner from the town Ziluna, whose name has not been lost in a lacuna. We know, however, that Ziluna lies in a likely Hurrian milieu, on a road from Hattusa to northern Syria; therefore, according to the hypothesis of Sasseville 2020: 113, the performer from Ziluna is most probably to be equated with Silaluhhi. In both instances the references to the keldi sacrificial rite and smearing feet with blood are found in a close juxtaposition in fragmentary contexts. The second fragment also mentions the nakkussi (scapegoat) rite. One must stress that the discussion here concerns the use of Luwian forms that are ultimately derived from Hurrian *nakkosse `release', and not the scapegoat rite as such. The latter represents an integral part of the Puriyanni and Kuwattalla traditions, but is also well known in Arzawa, where direct Hurrian influence can safely be ruled out. So far as we can judge, both incipits describe essentially the same implements and the difference between the two is mainly stylistic. The restored translation presented below is based on the assumption that the number of sheep used for individual rites must total eight in each case.

Furthermore, it is appropriate to point out that the variation in animal offerings is directly attested through the fragmentary incipit (6), whose attribution to the Kuwattalla tradition was recently stressed in Hutter 2019: 384. It is easy to see that the total only four sheep are required for this version of the ritual, and only one sheep is necessary for the ikkunatt-rite, whereas twice as many sheep are mentioned in (4-5) in each of the two cases. Furthermore, although smearing feet with blood and the nakkussi-rite can be restored in (6), the keldi-ritewas demonstrably absent in this version of the ritual, at least, it does not occur in its expected position. The passage under discussion confirms the hypothesis that individual rites could be added or removed as the tradition evolved, which in turn implies that the lost incipit of CTH 761.1 could easily lack references to any of the Hurrian concepts.

(6) Bo 4388: 3'-5' (CTH 763), cf. Fuscagni 2007: 70-71

3'. nu4 UDU^HВ-pвt Щ MВS.GAL [...]

4'. 1 UDU ik-ku-na-at-ta-as 1 UDU [...]

5'. is-har-nu-ma-an-zi 1 MВS.GAL-ma A-N[A ...]

`Four sheep and one billy goat [...], one sheep of the ikkunatt-rite, one sheep [...], one sheep [.] they smear [their feet], and one billy goat for [...].

In the light of this general observation, one can now consider the empirical evidence for distribution of Hurrian features within the Kuwattalla tradition. The most solid cluster is formed by the derivatives of Hurr. *nakkosse `release'. The Middle Script fragments KBo 9.141 and KUB 35.15 (CTH 761.2), both characterised by the archaic spelling BE-EL SISKUR for `ritual patron', contain the description of a nakkussi-rite and Luwian incantations featuring the forms /nakkussaunta/ `we released a scapegoat' and /nakkussaxidi/ `with the release of the scapegoat'. Yet the style of both fragments, which feature long Hittite narrative passages, is different from that of CTH 761.1, where the extended Luwian incantations are punctuated by very laconic Hittite instructions. It is, therefore, perfectly possible that despite their archaic outlook, the manuscripts collected under CTH 761.2 reflect a version of the Great Ritual that had evolved with Silalluhi's collaboration. Another interesting case is KBo 10.42 iv 4', where the Hittite instrumental form nakkussit `with the scapegoat', occurs at the very end of the tablet, almost immediately before the colophon. Although the Hittite instrumental forms are archaic by definition, the restoration of the colophon suggests that the manuscript is attributed to both Kuwattalla and Silalluhi. The other manuscripts featuring Hittite nakkussi- or its Lu- wian cognates appear to be more innovative.

As for the other two Hurrian features reflected in the incipits, they are fairly likely to reflect the secondary modification of the Kuwattalla tradition in the course of its written transmission. The only reference to smearing feet with blood outside the incipit section is the Lu- wian foreign word KBo 29.6(+) rev. 18' a-as-har-nu-um-m[i-ti] occurring in Hittite context, just as a-as-har-nu-um-ma-in-ti does in (4). The relevant fragment can be attributed to the late- thirteen-century scribe Pariziti based on its ductus and therefore can be assigned to CTH 762. The next paragraph of the same fragment, KBo 29.6(+) rev. 20'-23', signals the arrival of the next day (the number is unclear), while the following one, KBo 29.6(+) rev. 24'-26', and refers to a sheep offering. It is tempting to see here the reference to a keldi-rite, but this oblique piece of evidence is isolated within the available corpus. Furthermore, the reconstruction of the sequence of ritual acts within the Kuwattalla tradition suggests that the rite involving smearing feet with blood and the keldi-rite to follow occur after all the other identifiable rites. Such a peripheral position is obviously compatible with the hypothesis of a later addition. ²² In fact, An additional candidate for a Luwian technical term of Hurrian origin was the ikkunatt-rite.According to the tentative proposal of Hutter 2019: 383-384, this term represents a derivative of Hurrian egunni `pure'. In the meanwhile, however, a convincing Indo-European etymology for this term was offered in Sasseville 2021: 562-563.

The Luwian noun ikkuwar (KUB 35.72 ii? 8'), from which the name of the ikkunatt-rite is ultimately derived, represents a straightforward formal cognate of Latin iecur, Greek pnap, and Vedic yвkrt`liver'. The understanding of the ikkunatt-rite as “the rite of liver-treat” is borne out by the fact that this is the only preserved rite within the Kuwattalla tradition where gods are actually treated with liver. there are no grounds to believe that any of these two rites had already been present in the freestanding version of the Great Ritual (CTH 761).

The hypothesis of secondary Hurrian influence derives further circumstantial support in the analysis of other Hurrian theonyms in the incantations of the Kuwattalla tradition. For example, the Hurrian goddess Ninatta (KUB 35.71+ iii 3'), Istar of Nineveh (KUB 35.71+ ii 7-8), and unspecified ISTAR/Sawoska (KUB 35.71+ iii 2') all appear in the same New Script fragment belonging to the free-standing version of the dupaduparsa-ritual (CTH 759). The formula in KUB 35.71+ ii 6'-9' exhibits close similarity to the one in KBo 29.6(+) obv. 20'-21', but no reference to the Hurrian gods is found in the latter passage, which supports the hypothesis of their secondary insertion. A less trivial issue is the attestation of ISTAR/Sawoska in KUB 35.82:7' (CTH 761.2). The Middle Script fragment KUB 35.82 shares its ductus with several other specimens of the Kuwattalla tradition, including KUB 35.34. On the plausible assumption that these fragments belong to the same manuscript, it displays a number of archaic features, including the designation BE-EL SISKUR for `ritual patron'. Yet the phraseology of the taluppi-rite in KUB 35.34 is not at all similar to its counterpart in CTH 761.1, so there are no arguments for assigning this manuscript to Kuwattalla alone. Cf. immediately above for the discussion of the nakkussi-rite in CTH 761.2.

Several more items of Hurrian origin are attested in the dupaduparsa-ritual (CTH 759). Contra Melchert (2013: 169), who stresses the lack of Hurrian influence on the dupaduparsa-ritual.Thus, the best-preserved tablet of this composition contains a mention of the Syro-Hurrian goddess Hepat (KUB 9.6+ ii 6'). The fragment KUB 35.83(+) can be attributed to the same ritual based, among other things, on the characteristic purification rite involving the gangati-plant. This fragment contains the possessive adjective [h]a-am-ri-ta-as-si-en-zi (ii 6'), which is derived from Hurr. hamri `(type of sanctuary)', and possibly even the adverb [hur-l]i-li `in Hurrian' (iii 18'). The palaeographic analysis suggests that KUB 35.83(+) belongs to the same manuscript as the small fragments KUB 35.40+ and KUB 35.41, which contain colophons attributing the dupaduparsa-ritual to the tandem of Silalluhi and Kuwattalla. There are no versions of CTH 759 attributed to Kuwattalla alone, while all the manuscripts of this group exhibit the features of either New Script or Late New Script.

Nevertheless, not all the instances of “southeastern” influence upon the Kuwattalla tradition can or need be explained in the same fashion. Once we turn to the formulaic repertoire, we find suggestive parallels even in CTH 761.1, the Middle Script version of the “Great Ritual” attributed to Kuwattalla alone. For example, the notion of a `divine path', securely restored in KUB 32.10+ obv. 10', finds parallels in the Hurrian-inspired Salasu ritual and an oracle question concerning Sawoska of Samuha (Hutter 2019: 393-394), which prompts Hutter to conclude that “here we find an element of Hurrian tradition taken up by Kuwattalla in her Kizzuwadnaean surroundings”. The presentation of a pot with vegetable soup in the Kuwattalla ritual is accompanied by the statement that the seeds contained there “will not become seed” (KUB 32.9(+) obv. 25); the same statement is made in connection with the presentation of a pot with dough and black cumin in the Mastigga ritual against the domestic quarrel (Miller 2004: 80-81, § 27). The sufficient assumption for tackling such cases is the formulaic continuity between the rituals of the Lower Land and Kizzuwadna. While this explanation is notionally distinct from the one advanced for Hurrianisms, it agrees well with what we know about the interaction between ritualistic traditions in the neighbouring regions (cf. the preceding subsection).