1.1 The Luwian Corpus in Cuneiform Transmission

Unlike most other languages of ancient Anatolia, the Luwian language is attested in two different writing systems: the Anatolian adaptation of Mesopotamian cuneiform (cf. Chapter 6) and Anatolian hieroglyphs.¹ It was previously believed that these two corpora corresponded to two closely related languages, named faute de mieux Cuneiform Luwian and Hieroglyphic Luwian. By now it has become clear that the differences between some Luwian dialects attested in cuneiform transmission exceed the postulated distance between Cuneiform and Hieroglyphic Luwian, and one of the dialects attested in cuneiform transmission, namely Hattuša Luwian, displays a particular resemblance to the early form of Hieroglyphic Luwian. Therefore, it makes more sense to speak of one Luwian language, and reserve the terms ‘cuneiform’ and ‘hieroglyphic’ for the writing systems associated with specific Luwian corpora (Yakubovich 2010:68–73). However, all the Luwian cuneiform texts predate the abandonment of Hattuša, whereas the bulk of Luwian hieroglyphic texts postdate this watershed event in ancient Anatolian history, so the Luwian cuneiform corpus can be used as the basis for the study of areal phenomena involving the Luwian language before the 13th century BCE. This justifies the separate treatment of Cuneiform Luwian and Hieroglyphic Luwian in the present volume. This chapter will address the evidence of Luwian cuneiform texts, while the Hieroglyphic Luwian inscriptions will find a detailed treatment in Volume 2.

As far as we know, the term ‘Luwian’ does not reflect the name given to the language by its speakers but rather harks back to its Hittite designation. Hittite cuneiform texts feature the adverb luwili ‘in Luwian, as a Luwian’, formed according to the same model as nišili ‘in Nesite/Hittite’.² Although the adverb luwili frequently precedes the direct speech utterances recorded in Hittite, sometimes what follows it is in a different language, which is patently related to Hittite (Mouton and Yakubovich 2021:26). Accordingly, European scholars call this language Luwian, while the variant Luvian appears in the publications of American and some US-trained scholars. The term luwili is probably derived from the geographic name Luwiya, which is likewise attested in Hittite transmission and will be discussed in § 1.3. Regrettably, exactly what the Luwian speakers called their language, either in the second millennium BCE or later, remains unknown.

Luwian texts are still imperfectly understood, and this holds in particular for those of the second millennium BCE. The bulk of the Luwian passages in cuneiform are embedded in Hittite religious compositions. In some cases, the Hattuša scribes translated the embedded incantations into Hittite. In other cases, they chose not to do so, presumably to preserve the ritual efficacy. The incantations tend to seem esoteric to modern readers, and many of the lexemes used in them are specialized terms deployed in witchcraft or other religious performances. Therefore, our ability to understand the Luwian embedded passages largely depends on the availability of Hittite adaptations or comparison with the Hittite narrative frame, which frequently makes it possible to predict their content. The situation is exacerbated by the fragmentary condition of nearly all of the relevant texts: the incantation may be preserved but the narrative frame lost or both may be preserved only partially. The contrast between lexical and grammatical interpretations must also be highlighted: it may be harder to interpret the content than the grammatical structure of some sentences because our command of Luwian grammar surpasses our grasp of the Luwian lexicon.³

With very few exceptions, the Luwian embedded passages were not composed by Hattuša scribes, but rather reflected direct speech utterances that they more or less faithfully recorded. Therefore, the cuneiform script of the available Luwian passages is essentially the same adaptation of the Mesopotamian cuneiform that was used in Hattuša for Hittite texts, including those that framed the Luwian insertions (Hawkins 2003:129). There are, however, some orthographic peculiarities. The sparing use of Sumerograms and the extreme rarity of Akkadograms in the preserved Luwian passages correlates with the passive role of scribes in their transmission: presumably the ‘stenographers’ did not have the time or inclination to transpose the recorded utterances into the complex heterographic code that is typical of the Hittite formulaic passages (cf. Kudrinski 2017:278–279). Numerous instances of multiple plene spellings (vowel replications) in Luwian forms—for example, da-a-u-i-iš ‘eye’, ta-a-wi₅-⸢ya⸣-a-an ‘toward’, [ti]-⸢i⸣-ti-i-ta-a-ti ‘with the eye pupil(s)’ may testify to the care that scribes took in transliterating this less familiar language.⁴

In addition to the embedded passages, the Luwian cuneiform corpus also features isolated forms, likewise mostly embedded in Hittite texts. From the late 14th century BCE onwards, they were commonly marked with special signs that were otherwise used to accompany glosses in certain cuneiform traditions and are therefore known as ‘Glossenkeil words’ after the German term for gloss wedge.⁵ Furthermore, there are Luwian glosses added to some Akkadian texts (Giusfredi 2012; Pisaniello and Giusfredi, 2021). Some of the texts with Glossenkeil words are attributed to the same kings that commissioned the first monumental hieroglyphic inscriptions in the Luwian language. Although the Glossenkeil words yield a smaller corpus than the embedded Luwian sentences, they are important for three reasons. First, they frequently complement the lexicon known from other sources, being embedded in texts of various genres, many of them secular. Second, it is often possible to guess their meanings from the surrounding Hittite context. Third, they tend to preserve a different dialect than the bulk of the Luwian incantations and conjurations (cf. § 1.2).

The patchwork character of the Luwian cuneiform corpus explains why its systematic study trails the investigation of Hittite texts by several decades.⁶ The publication of KUB 35 by Heinrich Otten in 1953 introduced the largest fragments that contained Luwian passages, including those of CTH 758–763, to the community of cuneiform scholars, while Laroche 1959 represented the first Cuneiform Luwian lexicon. Another accomplishment of Emmanuel Laroche was the preliminary classification of the published Hittite-Luwian texts by genre (CTH 757–773 in Laroche 1971). New opportunities for the study of our corpus emerged with Otten’s autographic edition of another large group of Hittite-Luwian fragments, which was published as KBo 29 in 1983. This project was coordinated with Frank Starke’s attempt to transliterate all of the fragments with Luwian insertions, which culminated in the publication of Starke 1985.

The beginning of the twenty-first century marked a new stage in Luwian studies. The work by J. David Hawkins and Anna Morpurgo-Davies on deciphering Luwian hieroglyphic texts greatly improved our understanding of Luwian grammar and the Luwian lexicon. The first collective monograph devoted to the subject, published in 2003, contains a grammatical survey of the Luwian language based on the entire corpus, regardless of the writing system used (Melchert 2003b). This pioneering paper has not lost its value, although for best results it should be read in conjunction with later surveys (Yakubovich 2015a and Melchert 2020a). The growth of the accessible Luwian cuneiform corpus and improved understanding of Luwian morphology made possible the production of a new Luwian lexicon (Melchert 2022).

1.2 Luwian Dialect Geography

Luwian belongs to the Anatolian group of languages and thus is a close relative of Hittite. The relationship between Luwian and the Palaic language, likewise attested in cuneiform transmission, is even closer. For example, the Luwian Sun god Tiwad and the Palaic Sun god Tiyat are derived from the same stem, whereas the Hittite cognate of this stem is šiwatt- ‘day’. However, Palaic and Luwian were not mutually understandable and thus represented separate languages. The closest relatives of Luwian are the languages of the Luwic group,⁷ which are best attested through alphabetic inscriptions of the first millennium BCE. The most important among those are Lycian and Milyan, known from the rock-cut tombs and other monuments of classical Lycia, and the recently deciphered Carian language, spoken in Caria in the extreme southwestern part of Anatolia. Previously it was believed that at least some of these languages were direct descendants of Luwian. Now most linguists have embraced the view that the Lycian and Carian languages are more closely related to Luwian than to Hittite or Palaic but are not the direct descendants of the known Luwian dialects. One may wonder whether those who spoke their ancestors in the second millennium BCE would have understood some of the attested Luwic dialects (in particular, Ištanuwa Luwic), but in the absence of any preserved Proto-Lycian or Proto-Carian texts, it is impossible to answer this question.

When we turn to the attested dialects of the second millennium BCE, traditionally classified as Luwian, there are sufficient grounds to distinguish between Kizzuwatna/Lower Land Luwian, Tauriša Luwian, Hattuša Luwian (Empire Luwian), and Ištanuwa Luwian/Luwic. This proposed taxonomy is based on the conventional Luwian corpus (contrast § 1.3) and is by no means exhaustive, given the almost certain existence of cuneiform fragments bearing other Luwian dialects. Future additions to the Luwian corpus may make possible the more precise dialectal identification of some passages. For now, these are the only four dialects that can be defined from their linguistic peculiarities.⁸

The best-known dialect (or dialectal continuum) is associated with a vast area stretching from the Lower Land, situated on the central Anatolian Plateau, to Kizzuwatna, roughly corresponding to classical Cilicia. The texts featuring minor Luwian insertions that are securely associated with the Lower Land are the festival of Huwaššanna (CTH 690–694), a goddess worshipped as the divine queen of Hupišna, and the rituals attributed to the Old Woman Tunnawiya (CTH 409), whose name links her to the town of Tunna (Mouton and Yakubovich 2021:28).⁹ Furthermore, the extensive Luwian incantations associated with the earliest layers of the Kuwattalla tradition (CTH 761) show striking parallels to the incantations of the Tunnawiya tradition and therefore must also be associated with the Lower Land (Mouton and Yakubovich 2021:30–31). In contrast, a Hittite-Luwian ritual text that had paramount importance for the initial decipherment of the Luwian language is attributed to Zarpiya, a physician from Kizzuwatna (СTH 757). In addition, there are reasons to think that Old Woman Šilalluhi, who was responsible, at least in part, for the secondary adaptation of the Kuwattalla tradition, hailed from Kizzuwatna.¹⁰ It does not seem possible to discriminate between the Luwian passages emanating from Kizzuwatna and the Lower Land on purely linguistic grounds. A typical linguistic feature of this dialectal continuum is the proleptic construction, which will be addressed in more detail in § 2.2.¹¹

The Hittite-Luwian fragments collected under CTH 764–766 can be subsumed under the label of the Tauriša tradition. They feature an unusual divine triad, namely, the Sun god, the goddess Kamrušepa (probably the adaptation of the Hattian Katahzifuri), and their offspring, the Tutelary God of Tauriša (Mouton and Yakubovich 2021:38–40). While the Luwian Sun god Tiwad is also common in the Hittite/Luwian texts from Kizzuwatna/the Lower Land, the other two deities are not grouped with him there. The texts of the Tauriša tradition belong to the genre of conjurationsdefined by means if Hittite hukmaiš or Akkadographic ŠIPAT as opposed to the Sumerograms SISKUR(.SISKUR) ‘rituals’ or EZEN₄ ‘festivals’. Their Hittite colophons do not mention a specific performer but specify the occasion of their performance—usually childbirth or children’s diseases. The beneficiaries of these performances are known as DUMU.NAM.LÚ.U₁₉.LU, meaning ‘human child’ or ‘human being.’ The conjurations are usually shorter than rituals, but their Luwian portions may contain not only incantations as such but also historiolae. The town of Tauriša was situated in northern Anatolia, probably to the northeast of Hattuša. The clitic particle /=wa/ is used frequently in Tauriša Luwian but not in Kizzuwatna Luwian, whereas the clitic particle /=g^wa/ appears to represent an exclusive feature of Tauriša Luwian (Mouton and Yakubovich 2021:42–43).¹²

The Luwian dialect of Hattuša, which apparently spread through the Hittite Empire during the last century of its existence, was discussed in some detail in Yakubovich 2010:15–73 (but cf. Melchert 2003b:171–175). In cuneiform transmission, it manifests itself in Glossenkeil words and other Luwian foreign words occurring in Hittite texts. Presumably most of these forms reflect code mixing in dictation to Hattuša scribes, but some of them may have entered the scribal jargon. A borderline case consists of isolated Luwian forms in texts of Kizzuwatna inspiration: depending on whether they were written down by Hattuša or Kizzuwatna scribes, such texts may or may not feature Hattuša Luwian forms (cf. Chapter 15, § 5). The lack of cohesive cuneiform passages, or even sentences, written in Hattuša Luwian limits our abilities to investigate this dialect, but an innovation that separates it from both the Kizzuwatna/Lower Land continuum and Tauriša Luwian is the spread of the nom.pl.c. ending /-ntsi/ to acc.pl.c., at the expense of the inherited ending /-nts/. This feature is shared by the Luwian hieroglyphic texts from the Neo-Hittite states of the first millennium BCE, implying that their dialect (Late Luwian) was a descendant of Hattuša Luwian. Therefore, what is empirically observed as the Luwian dialect of Hattuša can alternatively be called Empire Luwian.¹³

The largest group of festival texts with Luwian (or Luwic) insertions addresses the celebrations that are associated with the towns of Ištanuwa and Lallupiya (CTH 771–773). Both groups feature repeated indications that the participants sing (SÌR-RU) the Luwian passages, while the form /paju/ ‘may he give’ represents their common dialectal archaism.¹⁴ The location of the town of Lallupiya is unknown, but Ištanuwa is usually associated with the Sakarya river valley in the northwestern part of Anatolia. The choral singing accompanying the ritual acts in the Ištanuwa and Lallupiya traditions finds parallels in the texts of Hattian festivals and in archaic and classical Greece (Hutter-Braunsar 2022) but not in other Hittite-Luwian texts.¹⁵ Because the Luwian insertions in the texts of CTH 771–773 are overwhelmingly short and only indirectly related to the Hittite matrix text, they are particularly difficult to interpret. Yet, based on what we know, it is highly likely that the dialect of Ištanuwa also represents an outlier among the dialects treated in this section. According to Melchert 2022, its idiosyncrasies concern phonology (the likely absence of the change *dw- > /k^w-/), morphology (3sg.impv. ending -u), and semantics (=mi ‘to me’ used in a nonreflexive function).¹⁶ Those three cases must represent archaisms. The sociolinguistic and historical implications of this analysis are considered in the following section.

The discussion above should suffice to demonstrate that Luwian occupies a unique position among the ancient languages of Anatolia in the degree of its attested dialectal variation. The contrast with Hittite should be particularly clear: the matrix language of the texts of all of the four Luwian dialects hardly varies except for the geographically irrelevant distinction between Old, Middle, and New Hittite. The reason for this contrast is the erstwhile oral character of the embedded Luwian passages. The observed dialectal differences demonstrate beyond reasonable doubt that the Luwian language was spoken in several distinct areas of Asia Minor in the second half of the second millennium BCE.

1.3 The Location of Luwiya and Luwian Dialectal Filiation

As mentioned in § 1.1, the appellation Luwian, as applied to the language, is derived from the geographic name Luwiya. Therefore, determining the location of the land of Luwiya would directly impact the discussion of the Luwian homeland. Unfortunately, the land of Luwiya is mentioned only in the Hittite Laws, which complicates its identification. The laws distinguish between the land of Hatti (Hattuša), which constituted the core of the kingdom, and the lands of Luwiya and Pala, which were subject to certain legal provisions that did not apply elsewhere in the kingdom. For example, the murder of a merchant was generally punishable by a fine of 100 minas of silver, but the assailant was further expected to replace the merchant’s goods in the lands of Luwiya and Pala (Hoffner 1997:19). Abducting a free person from the land of Hatti and taking him or her to the land of Luwiya could lead to the summary confiscation of the abductor’s property, whereas the abduction of a Luwian national and the delivery of that person to the land of Hatti could be reversed by law but was not otherwise punished (Hoffner 1997:29–31). The most straightforward interpretation of this discrepancy is that the inhabitants of Luwiya and Pala enjoyed fewer legal protections in Hatti than the populace of its core area.

One consideration that helps to constrain the geographic location of Luwiya is its replacement with Arzawa in a passage of a late version of the Laws. Unfortunately, this cannot be taken as direct evidence, because the core area of Arzawa was demonstrably situated on the Aegean coast, which did not belong to Hattuša when the Hittite Laws were first recorded. It seems likely, however, that it reflects the assumed location of Luwiya on the western periphery of the Kingdom of Hattuša, which was contested with Arzawa at the time that the revised version of the Laws presumably came into being.¹⁷ Furthermore, it is possible that Purušhanda, an important town to the west of Kaneš/Neša and the westernmost destination of Old Assyrian trade, represented an important center of the Luwian-speaking area in the early second millennium BCE (Forlanini 2017:136 with fn. 77). Finally, the Hittite Laws describe Luwiya as a peripheral location under Hittite jurisdiction: the reward for bringing back a runaway slave increased progressively with distance, depending on whether the escapee was returned from a location nearby, this side of the river, that side of the river, Luwiya, or an enemy land (Hoffner 1997:32).¹⁸ The cumulative weight of these arguments supports the localization of Luwiya in the western part of the central Anatolian Plateau.

While the land of Luwiya arguably constituted the Luwian core area in the late third and early second millennium BCE, Luwian speakers must have spread throughout Anatolia by the time the Hittite Laws were composed. The absence of the toponym ‘Luwiya’ in later Hittite texts arguably reflects its growing anachronism as a result of further Luwian migrations, some of which presumably correlated with the emergence and spread of the attested Luwian dialects. Hattuša Luwian probably developed last. Its first traces can be observed in the 14th century BCE (Yakubovich 2010:33–36). The spread of Luwian across the Antitaurus mountain range from the Lower Land to Kizzuwatna was possibly facilitated by the expeditions of Hattušili I and Muršili I to southeastern Anatolia in the 17th and 16th centuries BCE, although it is risky to link military campaigns to large-scale migrations.

A far more intriguing case is that of Tauriša Luwian, recently identified in Mouton and Yakubovich 2021. Mouton and Yakubovich suggest that Luwian speakers had a presence in the Hattuša region independently from their presence in the town. The Tauriša dialect is distinct from that of Hattuša, and its important isogloss is the high frequency of the particle =wa. This raises the possibility that the Luwian speakers who triggered the language shift in Hattuša in the 14th to 13th centuries BCE were not the newcomers from the Lower Land or Kizzuwatna but rather those who lived next door. The timing of the initial arrival of Luwians in north-central Anatolia is uncertain but possibly could be relegated to prehistory. This is not a new proposition: even before the identification of the Tauriša tradition, Goedegebuure (2008:171–174) argued that Luwian newcomers had triggered the syntactic restructuring of the Hattian language (cf. Chapter 15, § 3.1). An alternative solution would be interpreting the dialect of Tauriša as the outcome of a resettlement (or perhaps transportation) of the Luwian population under the direction of Hattuša in the historical period. The choice between these two solutions must take into consideration the contact-induced features of Tauriša Luwian (cf. § 2.2 below).

The adverb luwili introduces the Luwian insertions in the Kizzuwatna/Lower Land dialect as well as that of Tauriša. The absence of embedded cuneiform passages in Hattuša Luwian need not to imply that the adverb luwili was not used in connection with the utterances in the Hattuša dialect. There are texts in which this adverb introduces the direct speech of Hattuša residents, although it is rendered in Hittite (Yakubovich 2010:264–265). The cases of the Songs of Ištanuwa and Lallupiya are different. The lexeme luwili is not attested in this corpus, but the adverb ištanumnili ‘in Ištanuwian’ is deployed once in connection with singing in CTH 772, albeit not in a Luwian context (Starke 1985:319 with fn. 117a). The implications of this finding do not seem to have been fully appreciated in earlier studies. The contrast between the expressions “he sings in Luwian,” in KUB 35.1 and KUB 35.2(+), and “he [s]ings in Ištanuwian,” in KUB 53.15+ ii 15′, suggests that the Hattuša scribes did not regard the Ištanuwa dialect as a variety of Luwian. From their perspective, ‘the language of Ištanuwa’ had the same taxonomic status as ‘the language of Luwiya’ or ‘the language of Pala.’¹⁹

Naturally, the Hattuša scribes may have been mistaken in their assumptions, but there are independent reasons to give them the benefit of the doubt. On the one hand, as mentioned in the previous section, the dialect of Ištanuwa is the most aberrant and archaic of the attested ‘Luwian’ dialects. On the other hand, while the land of Luwiya is a plausible starting point for the eastward migrations to the Lower Land and then Kizzuwatna, Luwian migrations to the north, to the Sakarya river valley, require special pleading.²⁰ Without assuming such population movements, one may hypothesize that the dialect of Ištanuwa and Lallupiya belonged to the Luwic continuum but was distinct from Proto-Luwian, spoken further south in about 2000 BCE. If the region of Ištanuwa did not belong to Luwiya, its inhabitants had no reason to call themselves Luwians (whatever the term was in the Luwian language). If the people of Ištanuwa remained autochthonous at the time of the Luwian migrations, they had no reason to identify themselves with Luwiya at a later point either. The Hattuša scribes who recorded the festival of Ištanuwa may have been aware of this fact, but if they were familiar with Luwian, they may also have reacted to the perceived differences between the Ištanuwa dialect and the Luwian that they spoke/learned at home.

The formal implication of this hypothesis is that the term Ištanuwa Luwic is probably to be preferred to Ištanuwa Luwian. The distinction may seem to be hairsplitting, but the issue is nonetheless important given the parallel with the sociolinguistic situation in Arzawa. The Arzawa core area can be identified with the Meander (Menderes) valley and the region surrounding it, where the Carian and Lydian languages were spoken in the first millennium BCE. The Arzawa personal names transmitted in cuneiform are surely Luwic but structured differently than those associated with Hattuša Luwian, and there are no textual indications that the population of Arzawa spoke luwili. Caution dictates that one should not prejudge the question of whether the dialect of Arzawa is more closely related to Luwian or Carian, and therefore Arzawa Luwic was proposed as a neutral label (Yakubovich 2013:116). But if one extends the distinction between Luwian and Luwic to the second millennium BCE, the question of where and how to draw a boundary between the two arises. The only operational test appears to be the elicitation of contemporary judgment: only those second-millennium dialects that can be linked to the adverb luwili deserve to be called Luwian. The dialect of Ištanuwa fails such a test and thus emerges as the best-attested member of the Luwic dialectal continuum in the Late Bronze Age.

1.4 The Areal Context

The sheer size of the territory inhabited by Luwian speakers makes it impossible to generalize about their linguistic neighbors. Each Luwian community associated with a specific dialect must be treated on a case-by-case basis. The most obscure case is that of Luwiya, the supposed local homeland of the Luwians: we are not aware of other languages that were spoken in west-central Anatolia in the late third to early second millennium BCE. Projecting the situation backward from almost 1000 years later, one could argue that the western part of the Anatolian plateau had been settled by speakers of various Luwic dialects, such as Ištanuwian (cf. the previous section). We do not know whether non-Indo-European languages lingered in this area or whether the ancestors of the Lydians ventured east of the Aegean basin. Proto-Luwian was probably in contact with pre-Hittite in the east, but the effects of this contact are inferable from Hittite rather than Luwian data (Melchert 2020a:245).²¹

We are on firmer ground when we turn to the sociolinguistic situation in southeast Anatolia. Although the Hurrians were newcomers to Asia Minor,²² they were established in this area by the early second millennium BCE. We know from the archives of Kaneš that part of this territory once belonged to the kingdom of Ma’ama. Its ruler, who bore the Hurrian name Anum-hirbe, corresponded and presumably shared a common border with Waršama, king of Kaneš (Miller 2001). One of the major urban centers ruled by Anum-hirbe was Haššu, probably once situated in the area of modern Gaziantep. There are reasons to believe that Haššu and its vicinity constituted one of the important Hurrian cultural centres (Wilhelm 2008:190–193). Hattušili I retaliated for the Hurrian raids into central Anatolia by conquering and plundering Haššu, and his reign may be regarded as the start of substantial Hittite (and presumably Luwian) presence in this region.

We know little about the ethnolinguistic landscape of the Cilician Plains in the first quarter of the second millennium BCE because no indigenous sources are available to us from this period (Novák 2010:401–402).²³ However, it seems probable that Luwian speakers arrived in this area in the first half of the second millennium BCE, an event that may or may not have been connected with the military campaigns of Hattušili I and Muršili I in southeastern Anatolia and Syria.²⁴ Around 1500 BCE, the Cilician Plains and adjacent parts of southeastern Anatolia gained independence from Hatti and came to be known as the kingdom of Kizzuwatna, which became a dependency of the Hurrian kingdom of Mittani later in the 15th century BCE (Miller 2004:7). The incorporation of Kizzuwatna into Hatti around 1400 BCE sparked interest in Kizzuwatnean religion and culture at the court of Hattuša. The accumulation of Hurrian and Hittite-Hurrian compositions attributed to Kizzuwatna practitioners was a manifestation of this curiosity.

Although we have no evidence for Semitic population groups settled in southeastern Anatolia in large numbers, both Akkadian and West Semitic must have constituted an important part of the cultural context of this region from the early second millennium BCE. The West Semitic (Amorite) population was linguistically dominant in Syria; the kingdom of Yamhad, which had its center in Aleppo, probably maintained close commercial ties with Cilicia.²⁵ The Assyrian trade network controlled commerce to the east of Cilicia, presumably including the area of Haššu, and to its north (on the other side of the Antitaurus Mountains). The demise of the Assyrian trade in the 18th century and the disintegration of Yamhad a century later may have affected commerce, but by that time Akkadian had become the language of cuneiform literacy in Anatolia. Presumably it continued to play this role, alongside Hurrian, in the short-lived kingdom of Kizzuwatna (Yakubovich 2010:274–275). In contrast, Hittite and Luwian had little influence on the West Semitic vernaculars of Syria after Hatti conquered northwestern Syria during the reign of Šuppiluliuma I (mid-fourteenth century BCE) and established residencies in Karkemiš and Aleppo.

The situation in north-central Anatolia must have been quite different. The Hattians are generally believed to have inhabited this area before the Indo-European migrations. An Indo-European population group that settled there in prehistoric times consisted of Proto-Palaic speakers. This is suggested by the cumulative evidence of several interference features between Palaic and Hattian, although few of these features are conclusive individually (cf. Chapter 13, § 2.3). The advance of the Hittites to the area of Hattuša and its surroundings is supported by historical sources. The advance began in the mid-18th century BCE, during the reign of Anitta, who boasted of razing Hattuša to the ground, although the integration of the Hittite and Hattian elites, which culminated in the rebuilding of Hattuša, probably was not achieved until the following century. After Luwian speakers settled in the same milieu, they must also have interacted with their new neighbors. Therefore, it is unsurprising that the Luwian texts of the Tauriša tradition show parallelisms with Hattian, Hittite, and Palaic texts at the level of narrative formulae (Mouton and Yakubovich 2021:44–46) and that the linguistic structure of New Hittite underwent numerous changes under the influence of Hattuša Luwian (Rieken 2006, cf. Chapter 14, § 3.4).

Yet the contact-induced features of Luwian dialects need not conform only to the observed areal configurations: they can also reflect the migration itineraries associated with the speakers of particular dialects or elucidate the status of the respective dialects in multilingual societies. We will address contact-induced features through this lens in the remaining sections of the present chapter.

2.1 Loanwords in Luwian

This discussion of lexical transfers into the Luwian language is based on two recent publications. Simon (2020b) provides a meticulous update on loanwords from Hurrian vis-à-vis the state of affairs described in Starke 1990 and Melchert 1993. Melchert (DCL) accepts most of Simon’s conclusions in his new dictionary and offers a parallel revision of data on borrowings from other languages. The list below is based on Melchert 2022, with the omission of some examples whose status is less assured, slight semantic modification of certain others, the adaptation of Luwian nominal stems to the style of the ACLT project.²⁶

This brief survey confirms the impression that Hurrian constitutes the principal source of loanwords in Luwian. As already illustrated in detail in Starke 1990:210–221, most Hurrian nouns whose stems ended in -e/i, were adapted to the class of neuter it-stems when borrowed into Luwian (cf. Chapter 14, § 5). The same is true of Akkadian loanwords, several of which are suspected of having been borrowed via Hurrian. From the semantic viewpoint, the majority of borrowed nouns designate transferrable artifacts (such as vessels or items of furniture) or refer to technical terms used in rituals and divination. The predominance of inanimate nouns in (a) and (b) must have facilitated their adaptation as neuter stems in Luwian. The adjectives and verbs belonging to these groups are formed using Luwian denominative suffixes, which do not represent obvious adaptations of any source language forms and therefore presumably reflect language-internal derivation.²⁷

A lesser-known fact about Hurrian and Akkadian loanwords is the lack of their correlation with the Kizzuwatna Luwian dialect. The number of Empire Luwian loanwords in (a) and (b) is comparable to that of Kizzuwatna Luwian loanwords.²⁸ If we were to limit our Kizzuwatna Luwian corpus to actual Luwian passages (as opposed to the isolated Luwian foreign words and loanwords embedded in Hittite texts), then the number of Hurrian loanwords in Kizzuwatna Luwian would be even more modest: just one borrowed noun in -it-, namely, hamrit-, could be assigned to Kizzuwatna Luwian under such a procedure (see below in this section). But, as was already mentioned in § 1.2, the Luwian foreign words in Hittite texts primarily reflect code-switching in dictation or adoption into scribal jargon, rather than being vestiges of Luwian texts translated into Hittite.

The conclusion to be drawn is that the nouns in -it-, constituting the bulk of Hurrian and Akkado-Hurrian loanwords in Luwian, were primarily transferred in the milieu of the elites and literati, whether in Kizzuwatna or Hattuša. In the light of this, their status is similar to that of the bulk of Hurrian loanwords in Hittite. This implies, however, that the borrowed nominal it-stems cannot be attributed to a particular Luwian dialect, but rather represent features of a sociolect. The Hurrian technical terms for rituals and divination imply specialized knowledge, while the mobile property items with Hurrian names can be interpreted as luxury goods circulating as wealth finance. Judging by the situation in Hattuša, the literati were primarily recruited from social elites, so there is no contradiction between the two categories. The likely political context for the acquisition of Hurrian loanwords at the court of Hattuša was the annexation of Kizzuwatna between the late 15th and early 14th century BCE.

The residue of the Hurrian, Akkadian, and West Semitic loanwords in Luwian incantations is too meager to draw firm conclusions regarding the circumstances of their transfer. These are hāpiri(ya/i)- ‘of the bedouin, seminomads’, hamrit- ‘cult-house, sanctuary’, nakkuššāhit- ‘scapegoat rite’, nakkuššā(i)- ‘to perform a scapegoat rite’, *šaur(a)- ‘tool/weapon’ (in a compound wašha-šaur(a)-), zammitāt(i)- ‘grits’, and halal(i)- ‘pure’. If anything, they show that the Hurrian and Akkadian superstrate influences on Kizzuwatna Luwian were diminishing rapidly outside of royal circles. It is remarkable that all of these lexemes, except for *šaur(a)- and zammitāt(i)-, have cognates in Hittite and/or Hattuša Luwian.²⁹

The clearest example of an exclusive transfer from Akkadian to Kizzuwatna Luwian is the adaptation of the term for a ritual patron. Its cultural background was the widespread character of privately commissioned rituals that were carried out by reputable practitioners in both Kizzuwatna and the Lower Land. The Akkadian bēl niqê, literally ‘lord of the offerings,’ was translated word for word into Luwian as /malxassassis *nijas/ (usually spelled malhaššaššiš EN-aš). The calque acquired the terminological meaning ‘ritual patron’ (i.e., a private commissioner of a religious performance) and became very frequent in the texts of Luwian incantations. In contrast, a unitary term, ašhušikkunni, was coined for ‘ritual patron’ in Hurrian, while the Sumerogram EN SISKUR had the same role in Hittite scribal jargon but apparently without a stable phonetic reading.³⁰ The seemingly isolated character of this calque probably results from the fact that Mesopotamian and Hurrian influences on Kizzuwatna Luwian phraseology have yet to be addressed in the same detail as lexical transfers.

Another surprise is the relatively small number of Hittite/Hattian loanwords in Luwian and their lack of correlation with the northern Anatolian Tauriša tradition.³¹ The only borrowing attested in Tauriša Luwian is parnink(i)- (a disease), supposedly of Hattian origin. A partial explanation for this state of affairs is the parochial character of Tauriša conjurations, which are not associated with a state cult or reputable performers. Yet, even under such conditions, if the local Luwian community formed an enclave within a multilingual region for an extended period, one would expect to find more adstrate interference in the preserved texts. The next section offers a full account of this seeming mismatch.

2.2 Structural Interference

Since Luwian texts are attested across vast tracts of Asia Minor and beyond and span the period between the Late Bronze Age and Iron Age, the Luwian language is frequently invoked in discussing contact-induced changes in the region. In some instances, such changes are common to several languages and are best discussed in the context of the Anatolian linguistic area. To this category belong, for example, the second-millennium constraint on word-initial r- (Kocharov and Shatskov 2021), the neutralization of the voiceless/voiced (or fortis/lenis) distinction in word-initial obstruents (Melchert 2020a:274–275), and the grammaticalization of possessive adjectives (Luraghi 1998). Such phenomena will not be discussed here in any detail as this will be done in Chapter 15. There is likewise little point in addressing here instances of structural interference in which Luwian is the donor language because such a discussion is most appropriate in the chapters devoted to the respective target languages. We will instead concentrate on two structural changes that cut across the Luwian dialectal area and are thus conducive to refining our understanding of Luwian dialectal prehistory.

The first isogloss is verbal fronting, which is observed on a number of Luwian incantations from Kizzuwatna and the Lower Land and reconstructed in Lycian A. In Luwian, this syntactic configuration is frequently accompanied by the use of cataphoric second-position clitics, which double the subjects of intransitive clauses and the objects of transitive clauses. For example, KBo 29.6(+) i 23′–24′ awidu=pa=aš=ta ma[l]haššaššiš EN-aš ha[ratnāti] waškulimmā[ti] can be literally translated as “May he come, (namely,) the ritual patron, from offense and fault,” and KBo 29.3+ iii 17′ ma[m]malway=an EN SÍSKUR-iš adduwalin EME-in as “May (he) crush it, the ritual patron, (namely,) the evil tongue.” The pragmatic reason for verbal fronting is usually the alignment of the predicate with the topic or narrow focus.³² This innovation is synchronically restricted to the Luwian incantations of the Puriyanni and Kuwattalla/Šilalluhi traditions but can also be reconstructed in Tunnawiya and Maštigga rituals, where it occurs in the texts of Hittite incantations that presumably were translated from Luwian (Rieken 2011:500–502).

At the same time, the construction with verbal fronting must be reconstructed as a core feature of the ancestor language of Lycian A.³³ Such an approach appears unavoidable to account for the phenomenon of Lycian nasalized preterits, the forms that regularly appear in transitive clauses after the particle m(e)=, in contrast to the preterits without nasalization deployed in other contexts (Adiego 2015:8). An important correlate of the same distribution is the tendency to place the verb before the subject in sentences with the nasalized preterit (‘Imbert-Garrett’s Rule’). Thus, TL 103.1 ebẽñnẽ : xupã : m=ene : prñnawatẽ : tebursseli “This tomb, the one to build it (was) Tebursseli” contrasts with TL 40a.1 [p]ajawa : manax[in]e : prñn[a]wate : prñn[aw]ã : ebẽñnẽ “Pajawa (titled) Manaknine built this building.” The construction illustrated by TL 103.1 is very frequent in the Lycian A corpus, and the use of the nasalized preterit is a grammatical rule in such constructions.

Adiego (2015:26) persuasively argued that “at a certain point in the prehistory of Lycian … verb-“fronting” implied a clitic doubling.” At that point, the virtual archetype of TL 103.1 would have the shape *prñnawat(e)=ẽ tebursseli ebẽñnẽ xupã ‘(He) built it, Tebursseli, (namely) this tomb’. What eventually emerged as the marker of the nasalized preterit, must then have been an object clitic in its expected second position. This construction is, of course, parallel to the Luwian one ‘May (he) crush it, the ritual patron, (namely) the evil tongue’ (see above in this section), and nothing contradicts the assumption that it was already present in the dialect of the Lukka tribes, the ancestors of the Lycians, in the second millennium BCE. The use of the particle m(e)= and the left dislocation of the topic in front of this must be regarded as subsequent inner-Lycian innovations. This secondary process was likewise accompanied by clitic doubling involving =ene ‘it’, or its allomorphs

The Luwian dialectal change, particularly its Lycian counterpart, signal a radical departure from the consistent verb-final word order reconstructed for Proto-Anatolian and typical of Hittite, Palaic, Tauriša Luwian, and Late Luwian. The fact that verbal fronting was accompanied by clitic doubling in both Luwian and Lycian suggests that we are dealing with a common innovation, not merely two typologically similar changes.³⁴ Yet its dialectal distribution represents a challenge. Even in the Kizzuwatna/Lower Land dialectal continuum, verb fronting remained an optional device, governed by pragmatic factors, as did clitic doubling in sentences with verb fronting. This speaks against treating the similar change in Proto-Lycian as an outcome of language shift; moreover, the acquisition of Proto-Lycian by Luwian dialectal speakers lacks historical support. The opposite scenario, namely the language shift in Kizzuwatna and the Lower Land imposed by the Lukka tribes, is even less likely because such a migration of the Lukka tribes would have been reflected in the historical sources of the second millennium BCE. It seems most probable that the construction with verb fronting and clitic doubling originated somewhere in southern Anatolia and then spread across a large area as a pragmatic device in a situation of language maintenance. This hypothesis implies that the dialectal continuum of the Luwic languages remained transparent for such innovations; that is, the ancestor of Lycian A and the neighboring Luwian dialects must have remained mutually understandable.

Another areal isogloss of interest for Luwian dialect geography is the possessive construction with a plural possessor, which was first described in Melchert 2000. The Luwian possessive adjective in /-ass(a/i)-/ dialectally inserts a special morpheme /-nts-/ between the possessive suffix and the ending to mark the plurality of the possessor. The suffix /-nts-/, however, has a limited functional distribution: it is compatible only with the endings of nom.-acc.pl.n., dat.sg., dat.pl., and instr. For example, the Luwian term for the ‘lord of the offerings’ is malhašš-ašši-š EN-aš, as discussed in the previous section, but its dative plural form is malhašš-ašša-nz-anza EN-anza: literally, ‘to the lords of the offerings.’ Both the placement of the morpheme /-ants-/ after the possessive marker (the etymological genitive case ending) and its restricted paradigmatic distribution are typologically unusual and call for a historical explanation.³⁵

In Yakubovich 2010:51–53, it was argued that the Luwian instrumental forms in /-assa-nts-adi/, the most frequent ending signaling the plurality of the possessor, represented the starting point for the formation of the new construction. More specifically, /-ants-adi/ was added on top of the pre-existing possessive construction as a calque of the Hurrian =až=āe, the dedicated possessive instrumental form marking the plurality of the possessor. The morpheme =až= normally functions as the plural marker in Hurrian, although in this case it absorbed the genitive marker =ve= through a historical sound change (*=až=ve=āe > =až=āe). Accordingly, it was calqued by the Luwian plural marker /-nts-/, while the instrumental ending /-adi/ represents the expected calque of its Luwian counterpart =āe. The calque outlined above represents an instance of structural imposition, likely to be implemented during a language shift from Hurrian to Luwian. Later the morpheme /-nts-/ was extended by analogy to other Luwian forms marking plural possession, but only in those cases where its combination with the subsequent inflectional endings would not generate inadmissible consonant clusters.

The hypothesis outlined above was premised on the assumption that the morphological expression of the plurality of the possessor was limited to the Kizzuwatna dialect of Luwian.³⁶ This claim, however, was no longer tenable after the identification of the Tauriša tradition in Mouton and Yakubovich 2021. In that tradition we find an incontestable example of the relevant construction, namely, KUB 35.103(+) iii 10 wa^!-ya-am-ma-na ú-li-ip-na-aš-ša-an-za “wayamana of the wolves” (cf. ibid. iv 9 [ú-li]-ip-na-aš-ša-an-[za]), also the probable restoration KBo 29.52 r. col. 4′ a-pa-aš-ša-a[n-za-a]n pár-n[i] “in/to their house” and the possessive adjective KUB 35.90 r. col. 7′ ^dUTU-ša-an-za-a[n] “of the Sun gods” (dat.pl.) in a broken context. Furthermore, the sequence malhašš=ašša=nz=anza EN-anza ‘to the lords of the offerings’ and other relevant case forms of the same phrase, although nominally occurring in Kizzuwatna/Lower Land incantations, can be reinterpreted as scribal insertions because the ritual patron was presumably called by name in the actual performance. There are reasons to believe that at least the ritual texts attributed to Kuwattalla were first recorded in Hattuša (Yakubovich 2010:277–278).

In Mouton and Yakubovich 2021:49, this new evidence was taken as an argument for the pandialectal character of marking the plurality of the possessor in Bronze Age Luwian. While formally possible, such a hypothesis does not illuminate the distribution pattern of the relevant construction, which has no counterpart in Late Luwian or Lycian.³⁷ An alternative way of coping with the new facts is admitting the diffusion of the construction from Kizzuwatna to the Lower Land, Hattuša, and Tauriša. However, the horizontal transfer must have had a strong sociolinguistic motivation: the possessive construction with a plural possessor is too unusual and asymmetrical to have been borrowed solely for convenience reasons.

An assumption necessary to maintain the calque from Hurrian is the status of the language of the incantations as the Kizzuwatna Luwian koiné. The historical trigger of its formation would be the interaction among the itinerant Old Women and other ritual practitioners. A handy typological parallel is the formation of the Greek epic language as a result of interactions among the rhapsodes. In sociolinguistic terms, the language of the incantations was akin to a professional jargon (Fachsprache), although it must have been passively familiar to the potential patrons to ensure the success of the rituals.³⁸ Naturally, its spread need not have respected political boundaries: if the ritual specialists of the Lower Land belonged to the same professional community as the Kizzuwatna ritualists, they would have been likely to avail themselves of the same ritual language. This would mean learning not only formulae and stylistic devices but also certain grammatical structures, such as the possessive construction marking the plurality of the possessor.

At the same time, just as one investigates the dialectal background of the Greek epic language, it is appropriate to inquire into the dialectal origin of the Kizzuwatna Luwian ritual language. To maintain the hypothesis of the Hurrian calque, one is forced to assume that the Luwian dialect of former Hurrian speakers played a key role in the formation of the new koiné. Although Simon (2016:326) makes a valid point that the precise area where the Hurrian speakers shifted to Luwian remains elusive, there is a linguistic argument for the presence of Hurrians in Kizzuwatna well before Mittani began to exercise influence in this region. As stressed in Trameri 2020:333–336, the dialect of the Hurrian texts emanating from Kizzuwatna and preserved in the archives of Hattuša is distinct from that of the Mittani letter found in El Amarna. Moreover, some of the Hurrian texts have idiosyncratic dialectal features. Therefore, the language shift from Hurrian to Luwian in Kizzuwatna (perhaps in its eastern part) remains the default hypothesis. It can be claimed with more certainty that the Hurrian and Luwian ritual traditions interacted with each other in the same region in the mid-second millennium BCE.³⁹

The marking of the plural possessor in the expression ‘lord of the offerings’ can be explained through the hypothesis that the Hittite-Luwian rituals were first recorded by Kizzuwatna scribes. The presence of such a scribal group in the chancery of Hattuša is not a new idea; the activity of ‘southern’ scribes was linked with the Luwian foreign words in Hittite texts in Güterbock 1956b:138. This hypothesis did not stand the test of time because the dialect of those Luwian foreign words turned out to be distinct from the Kizzuwatna dialect (cf. § 1.2). There is, however, a different argument that points in the same direction, namely, the presence of Hurrian loanwords in Luwian in Hittite texts that have no apparent connections with the Kizzuwatna or Hurrian milieus (cf. § 2.1). The easiest explanation for the extension of the Kizzuwatna scribal jargon to the Hattuša chancery would be the transfer of some Kizzuwatna scribes to Hattuša in the wake of Kizzuwatna’s annexation. Yet if one faced the task of recording Hittite-Luwian rituals emanating from Kizzuwatna or the adjacent regions, it would be only natural to involve scribes of Kizzuwatna extraction familiar with the local variety of Luwian. As long as the Kizzuwatna scribes were able to imitate the language of the oral performers, they would insert the forms in /-assa-nts-/ where appropriate, including into the expression ‘lord of the offerings.’⁴⁰

Finally, the use of the same forms in the Tauriša tradition could reflect the dialect of a Luwian population group that migrated or was transported from Kizzuwatna. Such a solution may seem ad hoc at first glance but gains in attractivity when weighed against the possible alternatives. If the population of Tauriša had spoken Luwian since the prehistoric period, it would be reasonable to expect a considerable degree of interference between the local dialect and Hittite/Hattian. However, this is not the case: the only likely loanword in this corpus is parnink(i)-, a word for a disease, which was supposedly borrowed from Hattian (cf. § 2.1). The hypothesis that the Luwian presence in Tauriša derived from populations transported from Arzawa or a late language shift has even less to recommend it: some fragments belonging to the Tauriša tradition are written in Middle Script (Mouton and Yakubovich 2021:41). This leaves us with the late 15th to early 14th century BCE as the most likely period for the Luwian resettlement in the region of Tauriša; what we know of Anatolian political history in this period would favor Kizzuwatna as its starting point. Granted, we have no written accounts of mass migrations or transportations from Kizzuwatna to Hattuša in the wake of the former’s annexation, but this could be due to the extreme scarcity of historical sources addressing the circumstances of this political event. We do know, however, of the integration of Kizzuwatna texts into the archives of Hattuša (Yakubovich 2010:275, summarizing Miller 2004a); we have reason to believe that Kizzuwatna scribes were relocated to Hattuša as well (see above in this section); and we know that the Kizzuwatnean Goddess of the Night was split and reinstalled in Šamuha (Miller 2004a:259–439).⁴¹ Against such a background, there is nothing counterintuitive in the hypothesis that the rulers of Hatti would draw upon the population of their new province to compensate for the people they had lost to the Kaška ravages in the north.⁴²

As mentioned in § 1.2, the anonymous Tauriša conjurations contrast the Kizzuwatna Luwian rituals attributed to reputable professionals, so it would not be unexpected if they also differed in the linguistic register used. The deviations of Tauriša Luwian from the ritual language of Kizzuwatna may have a twofold explanation. First, if the population to be resettled in Tauriša was recruited in a particular town or region of Kizzuwatna, then certain features of Tauriša Luwian could reflect the colloquial speech of that town/region rather than the standardized ritual language. Second, some of the innovations that characterize Tauriša Luwian may have come into being in the northern Anatolian milieu. This explanation is particularly likely in the instance of the clitic particle /=wa/, which spread beyond its original quotative function in Tauriša Luwian and was grammaticalized as clause-demarcational particle in Hattuša/Empire Luwian (Mouton and Yakubovich 2021:42–43). All else being equal, these two phenomena should be considered together as evidence for a new dialectal continuum but do not constitute an argument for a common substrate and thus are compatible with the scenario of secondary convergence.