1 Theoretical Framework
1.1 What Is a Loanword?
A lexical loanword or borrowing can be defined as any lexical item that is produced inside a target language based on a form from a model language. The distinction between a loanword and borrowing could be treated as a point of view: the modeled word is a loanword from the perspective of the model language and a borrowing from the perspective of the target language. Therefore, we can safely employ the two terms as synonyms.
To distinguish a loanword from other phenomena of lexical exchanges between languages—namely, calques—it is important to look at the material that is involved in the transfer process. In the case of a loanword, an entire linguistic sign of the target language is reproduced in the model language, that is, both its expression (signifiant) and content (signifié). The two elements of the sign may undergo more or less drastic changes in the transfer: the signifiant may be phonetically, phonologically, and morphologically adapted to a greater or lesser degree, and the signifié may also be altered—typically, a word is not borrowed with all of the meanings that it has in the model language.
Loanwords represent the most superficial expression of linguistic and cultural contact. They may occur in situations of loose or indirect contact, even between distant languages. Indeed, the model language of a given loanword may not coincide with its source language. Instead, intermediaries may be involved in the transmission between the ultimate source and a given target language; in evaluating the phenomena of adaptation, the direct model is often more relevant than the ultimate source of a given loanword.
1.1.1 Loanwords vs. Heterography
As is well known, Hittite texts do not only include Hittite words. Just as Babylonians and Assyrians used to include Sumerian words in their texts in Akkadian, so the Hittites filled their texts with Sumerian and Akkadian words—the so-called Sumerograms and Akkadograms, which are generally referred to as ‘logograms’ (word signs). We prefer to label them ‘heterograms,’ following Kudrinski and Yakubovich (2016), because: 1) Sumerograms are often root signs rather than word signs, with the word (and even the stem, in the case of derivatives) frequently resulting from the addition of phonetic complements; 2) several Sumerograms consist of more than one sign, sometimes with a non-compositional meaning; and 3) Akkadograms are mostly written with syllabic signs, so they cannot be regarded as logographic.
There is a crucial difference between such heterograms and loanwords, which should be stressed to avoid confusing the two phenomena. Loanwords are foreign words that have entered the lexicon of the target language, thus becoming part of its langue. Heterograms belong only to the written language (and possibly to a very restricted oral dimension that remains functional in writing). They are foreign graphic words meant to represent—and thus be read as—their corresponding words in a different language, that is, the language of the text. For example, the Sumerogram EN and the Akkadogram BĒLU(M) represent the Hittite noun išha- ‘lord’ when they occur in Hittite texts. Although evidence for direct dictation of heterograms seems to exist,1 it was probably simply a scribal practice and cannot prove the existence of these words as loanwords in the langue. In the case of Akkadograms, the distinction between a heterogram and a true loanword is sometimes not straightforward, but the two phenomena are easily discernible in most cases.
1.1.2 Loanwords and Related Phenomena
Before discussing in detail the criteria for analyzing and classifying loanwords, it is worth briefly elucidating some concepts referring to phenomena that are related to or can be confused with loanwords—namely, Wanderwörter, Kulturwörter, glosses, code-switching, and code-mixing—to have a clear metalinguistic framework. The terms Wanderwort and Kulturwort are often used interchangeably. However, a distinction is sometimes made, although it is quite blurred and probably redundant because it has more to do with our limits than an actual state of affairs. ‘Wanderwörter’ can be defined as words, generally denoting objects, techniques, or commercial products, that are used in a significant number of languages that are not necessarily close to each other in space or time. Kulturwörter are also words used in many languages but lack a clear etymology, so they cannot be unequivocally traced back to a given language.2 Whether one decides to distinguish the two concepts or consider the two terms synonymous, we are still dealing with loanwords—very successful loanwords—even if we are not able to determine their ultimate origin and fully evaluate their paths of diffusion. We can analyze them with the methodology that we apply to loanwords because not being able to trace the ultimate source of a word and perfectly reconstruct all the steps of its diffusion does not necessarily prevent us from determining the direct model responsible for its transmission to a target language—that is, the last step, which is often the most relevant for investigating the linguistic strategies used in adapting words and the cultural implications of the borrowing process.
Glosses, code-switching, and code-mixing are different phenomena. While loanwords, including Wanderwörter and Kulturwörter, belong to the langue—that is, they are part of the linguistic system of a given language and thus are expected to be familiar to almost all speakers of that language—code-switching and code-mixing phenomena pertain only to the parole; they are strictly individual and usually confined within a single act of communication. The term ‘code-switching’ generally refers to the use of different languages by the same speaker in a single communicative act. This can take place in different ways. When the alternation between the different languages occurs within a single sentence, we speak of code-mixing (or intrasentential code-switching).
As for glosses, they are simply isolated foreign words that are not part of the lexicon of the language in which the text is written. They are usually mentioned so that they can be explained—for example, nu hattīli tahaya halzai tahayan=ma=za hattili LÚŠU.I halziššanzi “He calls out tahaya in Hattian—in Hattian, the barber is called tahaya” (lit. “they call the barber tahaya”) (IBoT 1.36 i 65–66). This concept of a gloss should not be confused with the ‘glosses’ in Hittite studies, the so-called Glossenkeilwörter, which are foreign words—mostly but not exclusively Luwian—that are embedded in Hittite texts and marked by the Glossenkeil. As will be discussed, such words represent code-switching phenomena or even true loanwords.
It is not always straightforward to distinguish between these different types of phenomena, particularly between loanwords and code-switching, in ancient languages that are known only from written texts. Frequency of occurrence is a criterion: true loanwords, being part of the lexicon of a language, are expected to occur multiple times in a corpus at different periods and possibly even in texts of different genres unless they are technical terms specific to certain textual typologies. Conversely, code-switching and code-mixing phenomena are likely to be nonce words employed to solve an immediate communicative contingency. However, a full assessment of individual cases may be hampered by the fragmentary and incomplete nature of the documentation and the possibility that the two phenomena coexist—for example, that a nonce word used on a specific occasion eventually becomes a loanword, integrated into the lexicon of the language, or that a true loanword is used in a code-switching or code-mixing context. For instance, a non-native speaker may not be aware that a word in his language, which he is using because he does not know its equivalent in the language in which he is communicating, was borrowed from his native language. Similarly, nothing prevents a gloss included in a text from corresponding to a true loanword or showing some superficial and mechanical phenomena of phonological and/or morphological adaptation like a loanword (compare, for example, the Hattian gloss tahayan in the Hittite accusative case in the passage quoted above).
1.1.3 Typology of Loanwords
Loanwords can be classified using different criteria. The most relevant from a linguistic point of view is adaptation. According to this criterion, loanwords can be divided into two categories, adapted and non-adapted loanwords. A loanword is regarded as adapted when the signifiant of the model word accommodates the phonological rules of the target language and is also assigned its inflectional morphology. Consequently, an adapted loanword is not immediately recognizable as a word of foreign origin to an ordinary speaker of the target language. Phonological adaptation should not be confused with phonetic adaptation, which is mechanical and almost always occurs in the target language. Morphological adaptation can only concern overt morphology because the acquisition of some mandatory morphological features required by the target language is unavoidable.3
As will be shown, derivational morphology may also have a role in the process of adaptation, and sometimes even more complex phenomena occur. For example, folk etymology may result in the alteration of the original signifiant to match a word in the target language based on some phonetic similarity and a real or supposed semantic correspondence.4 In a non-adapted loanword (also referred to as ‘foreign word’), the original signifiant remains unchanged, thus preserving a structure consistent with the phonological rules of the model language; the only changes that can occur in the target language concern mechanical processes of phonetic adaptation. Overt morphology of the target language is also expected to be absent in a non-adapted loanword. For a native speaker of the target language, a non-adapted loanword is usually clearly identifiable as a foreign word.
Loanwords can be classified not only by adaptation but also by integration. Integration is the degree to which a given loanword is acclimatized to the lexicon of the target language. This is measured by the possibility of forming derivatives through productive word-formation rules of the target language.5 Such base productivity is independent of adaptation because non-adapted loanwords may produce derivatives, whereas nothing prevents a fully adapted loanword from remaining isolated in the lexicon of the target language and not being used to form other words. Base productivity strictly concerns word-formation rules of the target language and should not be confused with the possible occurrence, in the target language, of derivatives of a given loanword that are independently borrowed from the same model language through a direct or indirect path. Furthermore, integration as defined here has nothing to do with the possible use of derivational morphology in the process of adaptation of loanwords. A derivational morpheme of the target language may be selected to adapt a loanword. For example, a professional noun borrowed from a model language may be adapted by adding a derivative suffix that usually forms professional nouns in the target language. This is not a matter of base productivity; it merely represents an adaptation strategy.6
We employ in our analysis the following categorization of loanwords, which combines the criteria of adaptation and integration:
|
Adaptation |
Integration |
||
|---|---|---|---|
|
Type I |
– |
– |
= non-productive foreign word |
|
Type II |
– |
+ |
= productive foreign word |
|
Type III |
+ |
– |
= non-productive adapted loanword |
|
Type IV |
+ |
+ |
= productive adapted loanword |
Such a typology should not be understood as hierarchical, and the four types do not represent mandatory steps on a path toward adaptation and integration. A loanword entering the target language may be adapted immediately and become productive. Furthermore, a loanword in a given target language does not necessarily fall into only one of the four types outlined above. Sometimes, a loanword may belong to more than one category in the target language, depending on factors such as the chronology of attestation or sociolinguistic variation.
1.1.4 Borrowability Scales
From a strictly linguistic perspective, no compelling reasons make a borrowing necessary or, on the contrary, prevent it. However, some general tendencies can be identified. Some linguistic signs are more likely to be borrowed than others. Differences in ‘borrowability’ relate to the types of morphemes involved and the parts of speech to which words belong. Based on these tendencies, borrowability scales can be established, that is, hierarchies of borrowings meant to measure and predict the greater or lesser ease of borrowing linguistic items, which is also related to the degree of contact between two languages.7
First of all, it is universally accepted that lexical items are more easily borrowed than grammatical morphemes.8 This can be represented as follows:
lexical > non-lexical
The first borrowability scale was formulated by Whitney (1881:19–20). Linguists have devised several others over the decades.9 The scales differ from each other because they are based on case studies or differing amounts of comparative data, but all identify nouns as the parts of speech most likely to be borrowed.
A borrowability scale functions at different levels and can be read in different ways. Let us take as an example the scale established by Haugen (1950:224) on data relating to American Norwegian and American Swedish: nouns > verbs > adjectives > adverbs-prepositions, interjections. Such a scale, as described by Haspelmath (2008), can be assigned the following values, although they are sometimes difficult to distinguish clearly:
-
temporal: elements on the left side of the scale are usually borrowed before those on the right;
-
implicational: a language that contains borrowed elements on the right is also expected to have those on the left;
-
quantitative: borrowed elements on the left are expected to be more numerous than those on the right;
-
probabilistic: elements on the left are more likely to be borrowed than those on the right.
Because different borrowability scales exist, they cannot be regarded as universal. The circumstances of borrowing may vary, resulting in considerable divergences among language contact situations.10 Therefore, borrowability scales identify general tendencies rather than absolute rules.
1.2 Calques and Their Typology
While a loanword is the replication of a linguistic sign from a model to a target language, a lexical calque—also called loan translation—involves the transfer of only one of the components of the sign, the signifié. It becomes associated with a signifiant belonging to the target language that already exists or is created for the purpose through the productive word-formation rules of the target language. Although complex typologies of calques have been established,11 sufficient for our purposes are the two macro-categories into which all other subtypes fall: structural and semantic calques.
Structural calques involve the creation of a new signifiant in the target language—a word, phrase, or more complex structure—to receive the signifié transmitted by the model language. The process usually involves translating the constitutive elements of the model word into corresponding elements in the target language to obtain an expression that optimally matches the deep structure of the word in the model language, regardless of a more or less perfect match of the surface structure.
Semantic calques, also called loan shifts, involve the transfer of a signifié from the model language. The signifié becomes associated with the signifiant of an existing sign in the target language that is formally or semantically similar to the sign in the model. A semantic calque extends the meaning of an existing word through language contact; no word-formation process is involved.
2 The Languages Involved
Excluding the phenomena that occurred during the Old Assyrian age (on which see the brief discussion in Chapter 4), the languages that must be considered when discussing lexical interference with Hittite and, more generally, interference involving Anatolian during the Late Bronze Age, can be categorized into two main groups: 1) the languages and cultures that were in synchronic direct or almost direct contact with Hittite, Luwian, or Palaic and 2) the languages and cultures of the wider ancient Near East and Mediterranean region. The languages of the former group are those from which loanwords were borrowed synchronically, in an almost direct fashion. The languages of the latter are those that shared some non-inherited lexical material with Hittite—material that circulated in a wider areal context. This categorization entails a conventional distinction between loanwords, generally borrowed from language X into language Y, and Wanderwörter. The origins of the latter were not always discernible; they were borrowed by more than one language in an area and sometimes showed up in languages spoken by cultures that virtually never interacted with one another.
2.1 Languages in Direct or Almost Direct Borrowing Scenarios
The first group of languages includes those of the cultures that had historical connections with the Hittite world. Some, especially those that were already exchanging lexical material with Hittite during the earlier phases of the pre-Hittite and Hittite ages, were geographically Anatolian or, if foreign to the Anatolian peninsula, contextually present in it. As outlined in Chapter 4, evidence exists of lexical contacts involving Old Assyrian, Hattian, and Luwian in the mature and late Middle Bronze Age in the context of the Old Assyrian trading network. It is highly unlikely that Hurrian was involved so early and with direct interference: the words with Hurrian etymology that emerge from the Old Assyrian archives are generally present in northern Mesopotamia as well. One must assume, therefore, that they entered Anatolia via the mediation of Assyrian because they were already integrated into the Assyrian lexicon.12 As for Palaic, no evidence exists that it exchanged lexical material with other languages of the area, although later Palaic texts show traces of interference with Hattian that may have occurred during the Middle Bronze Age and possibly earlier.
During the historical phase of the Hittite kingdom proper, loanwords emerging from the texts composed and stored in Hattuša include, again, Luwian and (very little) Hattian material. However, Hurrian appears more frequently, often mediated by Luwian, possibly as a result of the growing influence of the culture of Kizzuwatna from the 15th century onwards. A more precise description of the status and role of the languages during this phase follows:
The network of languages involved in direct, proven lexical exchange with the languages of Late Bronze Age Anatolia can be represented as a directed graph. The solid lines indicate proven lexical exchange, whereas the dashed lines indicate the presence of common lexical items with unclear paths of diffusion and/or Wanderwörter.
This scheme in Fig. 14.1 is offered not as an exhaustive representation of lexical contacts but rather as an overview of a situation that was certainly far more complex. Historically and geographically speaking, however, this network can be easily contextualized; it strongly predicts the cultural, political, and economic relationships between the groups and polities associated with the languages represented. The Hattian-Palaic-Hittite subgraph describes the situation in the north: the Hattians and Hittites had certainly been in close cultural contact since the Middle and probably Early Bronze Age. In contrast, Pala maintained a marginal profile. It interacted with the Hattian world in ways that are hardly traceable, but its cultural production emerged as a minority phenomenon in historical Hattuša. The complementary subgraph describes the most intensive areas of cultural exchange and political interactions between Hatti and the rest of the ancient Near East during the Late Bronze Age, during which northern Syria was the main catalyst for interactions between the local components, the Anatolian, the northern Mesopotamian, and southern Mesopotamian worlds.
Figure 14.1
A tentative model of the network of languages involved in direct lexical borrowings
2.2 Languages in Indirect Borrowing Scenarios
Widening the horizon, we find cultures whose languages share with the Bronze Age Anatolian idioms only a limited number of words. These words were generally circulating in a wider area, so interactions cannot be denied, but no evidence exists to show that these were direct. Mediated, or indirect contacts usually occur over a longer trajectory (that can be defined in terms of lack or difficulty of connectivity between region, not necessarily in terms of mere metric distance), which, in some cases, can even be diachronic rather than geographical. While others may argue for further, sometimes speculative extensions, we consider the following languages the main indirect contacts of the attested Bronze Age Anatolian idioms:
The list could be longer if we considered all of the languages that contain single or very sparse circulating cultural words that also emerge in Bronze Age Anatolia, with no geographic or diachronic restrictions. We would need to include a number of Indo-European, Semitic, and possibly isolated idioms—for instance, all those containing the words for ‘crocus’ or ‘saffron’ that formally match the Hittite kunkuman (see Rizza 2012). If we did so, our area of interest would expand at least to Central Asia, the focus and methodologies used in this work would soon be lost, and the very concept of contact would become so vague as to be useless. Therefore, we will restrict ourselves to the languages that were in a direct relationship of lexical transfer with those of Bronze Age Anatolia or belonged to the world of Hatti in a well-defined Near Eastern areal context.
3 The Early Northwestern Interface
A distinct subarea of lexical circulation seems to have existed that involved northern Anatolia and featured loanwords as well as occasional Wanderwörter that emerged in Hittite, Hattian, and Palaic (and, to qualify as Wanderwörter, in other languages of the wider area). This subarea was active at the northwestern interface of what, during the Late Bronze Age, became the core area of Hatti. It is suggestive of the existence of a strong cultural and linguistic superposition of Anatolians and Non-Indo-Europeans. The evidence for this ‘northwestern’ interface is undeniable but hardly rich in data. No clear examples of direct loanwords from Palaic to Hittite or Hittite to Palaic exist, which makes it impossible to speculate about immediate connections between the early Hittites and the people of Pala (for a historical and contextual discussion, see Chapters 4, 5, and 12). The phenomena that are documented in the corpus are: loanwords from Hattian to Hittite; loanwords from Hattian to Palaic (as recorded in the Hittite archives); a few areal designations pertaining to the royal sphere that emerged in additional languages of central Anatolia; and the single, very problematic case of a word that seems to have occurred in Palaic and in Hurrian (which will be discussed separately from the other three cases).
3.1 Loanwords and Areal Designations
A discussion on the loanwords from Hattian to Hittite involves the general problem of the interactions between the languages arguably already in very early phases, as the co-existence of the two components in Anatolia must have been a phenomeon of the longue durée in proto-historical phases. No data are available for the Middle Bronze Age documents of the kārum world (see Chapter 4), although the adaptation of Hattian toponyms to Hittite (or, more generally, to Anatolian) is represented by vocalic thematization. For example, Hattuš > Hattuša is a form of adaptation, albeit a very obvious and trivial one.19 As the context of these occurrences is Assyrian, it is impossible to draw any conclusions about the steps of the phonological or morphological adaptation of Hattian toponomastics into Hittite.
Apart from toponyms, the Hattian loanwords that can be recognized in Hittite texts during the Late Bronze Age are exclusively nominals (only a couple of dozen or so); nominals are more easily borrowed than verbs and grammatical words. They are normally, but not exclusively, integrated into the target language as i-themes. The words that designate naturally animated entities (e.g., titles, dignities, or job designations) undergo morphological adaptation as common-gender nouns. In some instances—for example, when a loanword is only attested in an indirect case such as the dative-locative—recognition of the gender is impossible. Neutral gender seems to have been preferentially assigned to inanimate nouns, as in the case of the musical instruments hunzinar and ippizinar, treated, it appears, as r-stems.20 The availability of a suitable inflectional class such as the neuter r-stem in Hittite must have facilitated the selection of the paradigm and gender. In at least one case, GIŠšahi-,21 an inanimate becomes an i-themed, common-gender noun when borrowed into Hittite, which may indicate that the prevalence of i-themed, common-gender nominals in Late Hittite played a role in promoting the preference for this type of stem. Overall, it seems that a preference existed for the conservation of a natural opposition of animate and inanimate nouns in the target language, but the existence of productive and morphologically ‘transparent’ classes also played a role in the selection of the paradigmatic series.
A borderline case of the possible influence of Hattian on Hittite was cautiously suggested by Giusfredi and Pisaniello (2020:216). It concerns the possibility that the quasi-agglutinative natural feminines in the Hittite -šara- (matched by those of the Luwian in -šri-) may calque the Hattian formations with the -ah suffix of Hattian. As the Anatolian morphemes are present in Middle Bronze Age personal names, the interference must have occurred very early if this analysis is correct.22
As was observed previously, the number of loanwords that entered Hittite from Hattian was quite limited. It may be even lower than estimated by Tischler (1979:257) because the Hattian origin of some famous lexical elements belonging to the sphere of royalty has been questioned. Given the scarcity of Hattian loanwords in Hittite, it would be inaccurate to describe the relationship between the two languages as a substrate-superstrate system. If Hattian was the superstrate from the standpoint of prestige, more loans would be expected in the technical, political, or religious semantic spheres. If Hattian was the substrate, one would expect grammatical interference to emerge in texts composed in Hittite, yet such interference is virtually absent. As Goedegebuure (2008) observed, it is Anatolian that might have had some degree of grammatical influence on Hattian as it was written by the scribes in the Hittite kingdom,23 which further complicates the reconstruction of a stable and convincing sociolinguistic model for the Hattian-Hittite relationships in terms of a binarily polarized system.
The issue partly depends, of course, on the nature of the corpus that is available to us. To what degree spoken Hattian was influenced by spoken Hittite or spoken Hittite by spoken Hattian and how many loans existed in the lingua dell’uso is impossible to say. Based on the textual materials, Hittite appears to have been on the receiving end of a narrow channel of lexical transmission, even though, as outlined in Chapters 4 and 9, the Hattian world had a significant cultural influence on the Hittite world and Anatolia, by the Middle Bronze Age, had long experienced the coexistence and cohabitation of different peoples and traditions.
Although the relationship between Hattian and Hittite appears complex and is probably underrepresented in the available corpora as outlined in Chapters 4 and 12, some level of lexical interference seems to have occurred during the early history of Anatolia between Hattian and Palaic, which was the other Indo-European Anatolian language of northern Anatolia. Leaving aside the daring and poorly investigated hypothesis of the borrowing of the contrastive morpheme -bi as -pi (that belongs to a hardly borrowable closed class and can be better explained in other ways24), the possible lexical loans from Hattian in Palaic share the following features:
-
They have been identified based on two criteria: the presence of the notation of syllables starting with a fricative using the sign PI with mater lectionis (waa, wii, wuú) and, in a minority of cases, the recognizability of Hattian roots or Hattian morphology;
-
They are all nouns;
-
They all appear to pertain to the religious or ritual sphere or, in some uncertain cases that, if real, are probably better described as Anatolian areal designations, to the political sphere;
-
They are not matched by a corresponding group of words moving in the opposite direction: Hattian > Palaic seems to represent a unidirectional path of lexical transfer.
The first point in the list, as mentioned in Chapter 12, requires a brief discussion. The presence of a fricative sound /f/ (or similar, fortis or lenis, labial or bilabial sounds)25 has not been reconstructed for the better known Anatolian languages (Hittite and Luwian) and therefore is unlikely to have been an inherited sound in Proto-Anatolian. For this reason, and because these types of sounds were notated with the same graphemic strategy by the Hittites when writing Hattian and Hurrian, it is traditionally hypothesized that all words containing a PI sign with a vocalic mater lectionis in Palaic must have been loans, arguably borrowed from Hattian.
This line of reasoning remains valid in principle, as does the belief that the source for non-inherited material in Palaic was mostly Hattian, because no solid case can be made for an important Hurrian presence in central Anatolia during the Middle Bronze Age. However, caution must be used in dealing with words that are interpreted as Hattian loans based only on the presence of the mater lectionis. Most of the words are sparsely attested, and it is almost impossible to exclude the possibility that a sound law of Palaic could have produced innovations in the phonemic inventory of Proto-Anatolian in given contexts. Furthermore, cases exist in which the mater lectionis was employed improperly by the Hittites for languages that were different from Palaic but used in texts connected to the Palaic world—for example, to write the divine name Hilanzipa (dHi-i-la-an-zi-w[aa-]) in a text (KBo 27.7:7, CTH 751) in which other Hattian-Palaic gods are mentioned and spelled with the waa and wuú signs. More problematic is the case of the rendering of the word warra, which may be connected to Luwian wahra26 and therefore certainly inherited from Proto-Anatolian. It is written waa-ar-ra in KUB 35.164 ii? 9, and this might point to the existence of a sound law that somehow produced a true fricative in Palaic, which would cast serious doubts on the Hattian analysis of other forms that exhibit the same graphemic device and are not analyzable morphologically as Hattian. A clear solution cannot be reached based on the few data available. However, given the Hittite scribes’ apparent lack of familiarity with the Palaic language (see Chapter 12), it remains very plausible that these forms were not recognized as Palaic and were interpreted as Hattian and hypercorrected.
Turning to morphological adaptation, it seems that Hattian words that entered Palaic were assigned a vocalic theme, in much the way that Hattian toponyms were adapted by the Hittites during the Kārum period. We can say little about gender due to our poor understanding of the semantics and the obscurity of some contexts of occurrence; the only apparent rule is the obvious animacy of divine names. A very good example of a rather typical process is the fulašina bread, from Hattian fulašne, which is ascribed to the vocalic class of a-themes and, despite not being animate, assigned to the common gender.27
Fulašina bread is also a good example of morphological integration as it seems to behave as a true loanword, acting as a base for a morphologically Palaic derivate relational adjective, fulašinika-.28 The existence of morphological adaptation within Palaic demonstrates that fulašina was a true loanword. This is not equivalent to stating that adaptation is always necessary to prove that a form was borrowed, but it is a sufficient condition. It acquires special significance in the case of contacts that were not only limited to written corpora but, as in this case, are also attested only indirectly: the Palaic we read was written by the Hittites, in Hittite scribal offices. As discussed in Chapter 12, Palaic in Hattuša, even in the pre-imperial phases, seems to have been a crystallized language that was opaque to the scribes who wrote it; they may have occasionally hypercorrected forms such as Hilanzifa by interpreting them as Hattian, indicating that confusion was possible between the two languages. In such a situation, any Hattian form included in a Palaic context could theoretically have been a terminus technicus belonging to the religious or ritual sphere and may or may not have entered the lexicon of the target language as a true loan. The existence of a morphological derivative such as fulašinika- is, therefore, a precious piece of evidence that supports the existence of true lexical interference between Hattian and Palaic in a stage preceding that in which the Hittites wrote down Palaic and Hattian material.
So far, we have discussed those phenomena of lexical interference of the northwestern interface that may be uncontroversially assigned to the sphere of direct borrowing from one language to another. However, a few words exist that emerge in Hittite and one or both of the other languages of the northwestern area. These are, most notably, the titles tabarna/labarna, attested in all the three languages as well as in Luwian, although with some variations in the functional features, and the female equivalent tawannanna, which is attested in Hittite and Hattian but not as yet in Palaic. Labarna, with lateral onset, appears, in its earliest occurrence, on a tablet from Kaneš level Ib (La-[ba]-ar-na-áš in Kt 88/k 713 obv. 3; Donbaz 1993:145–146),29 which is particularly rich in Anatolian onomastic material (cf. Tuthaliya tù-ut-hi-li-a at rev. 16 et passim, also with the ‘ending’ tù-ut-hi-li-áš at rev. 29). It was probably the personal name of a seal owner and witness. While the text is in Akkadian, the ending áš in this and other names would contain a trace of an a-theme if the integration and interpretation are correct. In Palaic, the word tabarna occurs as a divine title or epithet (cf. Tiyaz tabarni in KUB 35.165 obv. 22). In Hattian (cf. Soysal 2004:152), it seems to be an epithet used to refer to the king or a title, but its function is unclear because it appears mostly in mythological texts that were already integrated into the Hittite tradition. In Hittite, Labarna was the personal name of one or more kings who predated Hattušili I, but this does not indicate that it was originally a personal name; we know from Kaneš that it had become an anthroponym well before the Hittite age. It remained a title used for the king throughout the existence of the kingdom of Hatti and emerged in the Luwian Bronze Age corpus with lateral onset Labarna as both a divine epithet (KUB 35.133 ii 13) and the title of the king (KBo 19.155:6). It survived as a personal name, with no evidence of it being still also a title, in Iron Age hieroglyphic Luwian texts.30 Linguistically, it is still debated, with Soysal (2004:152) proposing a Hattian etymology and other scholars (notably Starke 1983:405–406, Melchert 2003a:19–20, and Yakubovich 2010:229–231) defending an Anatolian origin. Presently, the Anatolian interpretation appears to be dominant. Regardless of the word’s etymology, it is clear that it circulated widely, originated in the cultural koiné of the northwest, and survived long enough to enter the Luwian tradition and be used as a personal name well into the Iron Age.
The feminine equivalent, tawannanna, was used similarly as a title and personal name. It appears in both Hattian and Hittite documents and is considered Hattian by Soysal (2004), but alternative Anatolian analyses have been proposed (Melchert 2003a:19–20). As in the case of tabarna/labarna, the final word has not been said on the etymology of tawannanna, but it seems to be an areal designation that belonged to at least two traditions and originated in the northern cultural area.
The third ‘areal’ title that has been described as a Hattian loanword into Hittite is the title for the crown-prince, tuhukanti. In this last case, it must be stressed that the title is discussed here because it is alleged to have a Hattian etymology. Rieken (2016b) has argued that the word was Luwian; if it was, then its absence from the Hattian and Palaic corpora would remove it from the group of northern areal designations of royal dignities. However, Rieken has not proven definitively that the root is also Luwian (or Anatolian), although she has shown that the word’s ultimate derivational morphology is Anatolian.31
In sum, the fact that a Hattian etymology is no longer the only or best explanation for the political titles labarna/tabarna and tawannanna does not alter the fact that these designations (and possibly also tuhukanti-) seem to have circulated within a cultural area at the northwestern interface of the Late Bronze Age core area of Hatti. As such, they behave as short-range, local Wanderwörter, proving once more the intensity and continuity of the interlinguistic and intercultural exchanges among Hattian, Palaic, and Hittite (and, at least in the case of labarna/tabarna, via Hittite to other languages of Anatolia).
3.2 The Problem of the Elusive ‘Loanword’ hašira- ‘Dagger’
Thus far, it has been possible to defend a model in which the northwestern area of circulation of ideas and words was relatively well delimited. Although this area was not isolated, it had a limited and finite set of cultures and languages involved. This is consistent with the historical and archaeological descriptions presented in Chapters 4 and 12. However, a single piece of evidence appears to contradict this scenario. In the Palaic text KBo 19.152 i 12, the sequence ha-ši-i-ra-am=pí is attested.32 After assimilation of the final /n/ to the bilabial stop, the form is analyzable as an accusative singular of the common-gender, a-themed noun hašira-, which is replaced in the parallel passages in KBo 19.153 iii 7 and 19, in a very similar context, by the more ‘Hittite-sounding’ GÍR-an=pát. This sumerographic writing indicates that the meaning of hašira- must be close to ‘dagger,’ prompting the comparison with a Hurrian word, hašeri, which occurs in Boğazköy and whose meaning would also be ‘dagger’ (Richter 2012:139). As is often the case with these types of unexpected shared forms, Hurritologists and Hittitologists proffered different explanations. Richter (2012:139) ascribed the word to the verbal root haš- ‘to be strong’; Vine, cited in Melchert (2007:257), proposed an Indo-European etymology going back to a root *h2es- ‘to cut (vel sim.)’. Recently, Simon (2021) proposed an ultimate Hurrian origin but a mediation by Hattian.
However, the most important problem is not to understand which of two formally defendable etymologies is correct but rather how a loanword could have entered Palaic from Hurrian, because, as discussed in several parts of the present book, it is very hard to defend the hypothesis of a significant Hurrian presence in central Anatolia before the mature phase of the history of Hatti in the late 16th century BCE.
There is currently no certain reading for the Hittite word or words written with the sign GÍR in the texts composed in Hattuša. If a hypothetical circulating word existed both in Hurrian and Palaic, it may have been present in Hittite as well, thereby qualifying as a Wanderwort. This would make the historical scenario more credible but would not explain where the word originated and why and how it was transferred. As it denotes a realium, its circulation within Anatolia would not be unconceivable, especially if it dealt with a ritual item employed in ritual contexts. But it would still be difficult to imagine where Hurrian would fit in the cultural scenario.
In light of the semantic vagueness of Richter’s Hurrian etymology (‘be strong’ > ‘dagger’), an alternative explanation might be a northern Mesopotamian origin of the form, possibly in the Assyrianized linguistic environment of the kārum society. In Akkadian, the verb hasārum, hesērum means ‘to blunt, chip, trim’ (CAD H:176), and a G-stem participle may have served as the base for a loanword. As the Hittite and Boğazköy-Hurrian ⟨Š⟩ may have rendered a non-palatalized sibilant, there would be no phonetic obstacles to this hypothesis.
While none of these explanations is conclusive nor solves all of the problems, we maintain that the simple presence of one apparent loan from Hurrian to Palaic or from Palaic to Hurrian is not enough to assume direct lexical exchange between the two languages, especially since such an exchange would be very difficult to contextualize historically and geographically. The only solid conclusion is that the formal and semantic match appears to be convincing, but the two forms—the only two clearly attested manifestations of Wanderwörter in this context—probably belonged to a wider scenario of lexical circulation.
4 Akkadian and the Languages of Anatolia
The long-standing prominence of the Akkadian language in the kingdom of Hatti is well known and has been widely investigated.33 The adoption of the cuneiform writing by the Hittites during the reign of Hattušili I implied the acquisition of a whole cultural world. This world included the two main Mesopotamian languages, Sumerian and Akkadian, which were indissolubly bound to the cuneiform script. Not only was Akkadian a constitutive element of the Hittite cuneiform script, surfacing in the form of heterograms (or Akkadograms)—that is, Akkadian words syllabically spelled to represent Hittite words—but it was also widely present in the local textual tradition produced by the Hittites from the establishment of the Hittite kingdom. Old Hittite kings wrote their annals, edicts, and other administrative documents in Akkadian as well as Hittite; Akkadian literary works were copied and stored in the Hittite archives as part of the Hittite scribal curriculum; these literary works were used as models to draft original Hittite compositions; the same occurred with Akkadian technical texts, which were the repositories of ritual, medical, and mantic knowledge. Finally, when the kingdom of the Hittites became an international power, Akkadian was the language of diplomacy, in which treaties and letters were composed.
Given this situation, it would not be surprising to find phenomena of lexical interference in both directions between Akkadian and the different languages spoken in the kingdom of Hatti. Such phenomena are indeed attested and have been extensively studied by Anatolianists. However, the issue of Akkadian loanwords in the languages of Anatolia is not as simple as it might seem. The consideration of three methodological points relating to such loanwords is in order:
-
Is it really Akkadian? The identification of an Akkadian loanword should be assessed based on either a) its etymology, which should unambiguously guarantee that it belongs to the Akkadian language; or b) its occurrence in chronologically appropriate Akkadian texts; the latter also allows the assignment to Akkadian of words not originally Akkadian. Note that Semitic does not mean Akkadian: in some instances, a Semitic origin for a given word can be established, but the possibility exists that Akkadian is not the model language.34
-
Which Akkadian? Akkadian varied considerably in space, over time, and according to textual genres. An analysis of the phenomena of lexical interference between Akkadian and Hittite should ideally take all of these factors into account.
-
Through which path of transmission? The Akkadian language was used across the ancient Near East, which means that it served as a model for virtually all of the languages spoken in that area—languages that also had relationships with one another. Therefore, the path through which an Akkadian word may have reached Hittite or another language of ancient Anatolia was not necessarily direct. It could have included one or more intermediaries and thus different adaptation processes, which were specific for each of the languages involved.
A further issue to be considered concerns the distinction between true Akkadian loanwords and Akkadograms, which is mostly unproblematic but may sometimes complicate the interpretation. The issue is not whether the Akkadograms were dictated in Akkadian because how the text was produced does not relate to the language. To give an example, the unusual spelling BE-LU-uš-ša-an occurring in HKM 52:25 and HKM 80:5, whose phonetic complementation does not match the Hittite noun išha- ‘lord’ in nom.-voc.sg., which is required by the context,35 probably indicates that the Akkadian word was dictated,36 which does not mean that it was read in Akkadian and did not represent the Hittite išha-. The point is that it is not always clear whether an Akkadian word occurring in a Hittite text, with or without a Hittite ending, should be understood as an Akkadogram (with or without a Hittite phonetic complement) or an Akkadian loanword (adapted or not) in Hittite. For example, the Hittite hapax NA₄y]a-aš-pu-un ‘jasper’ (acc.sg.) in KUB 15.5+ i 4 depends on the Akkadian (j)ašpû- ‘id.’ (possibly a loanword in turn), but the Akkadogram NA₄YA-AŠ-PU is found on the same line, meaning that an Akkadographic interpretation as NA₄Y]A-AŠ-PU-un, with a Hittite phonetic complement, is also possible.37 Similarly, the Hittite acc.pl. NINDAtap-pí-in-nu-uš (a kind of bread) only occurs in the building ritual KUB 32.137+ ii 16.38 The stem form NINDAtap-pí-in-nu is consistently used elsewhere in the text. The noun patently relates to the Akkadian tappinnu (< the Sumerian dabin = ZÌ.ŠE),39 and the ritual shows clear Mesopotamian influences;40 however, it is difficult to decide if the noun should be regarded as an Akkadogram, NINDATAP-PÍ-IN-NU (with the Hittite phonetic complement -uš in ii 16), representing a different Hittite word, or if NINDAtap-pí-in-nu-uš was a Hittite nonce word, an occasional Hittitization of an Akkadian noun.41
The examples above reveal the importance of distinguishing between loanwords and nonce words in the study of Akkadian lexical interference in Hittite. Nonce words are not so different from code-switching phenomena because they can be regarded as occasional Hittitizations of Akkadian words, occurring most frequently in Hittite translations or adaptations of Akkadian texts. Thus NINDAtappinnuš, which occurred only in a Mesopotamian-influenced ritual as mentioned above, is probably a nonce word (unless an Akkadographic interpretation should be preferred). Similarly, the Hittite GU₄alu-, the heavenly bull in the Hittite version of the Gilgameš epic, < the Akkadian alû, may represent an occasional Hittitization of the Akkadian word if it is not to be understood as a proper name, although an Akkadographic explanation cannot be excluded (i.e., nom.sg. GU₄A-LU-uš, acc.sg. GU₄A-LU-un, GU₄A-LU-Ú-un). The Hittite kumra- (a priest), < the Akkadian kumru, which only occurs in the Kizzuwatna ritual fragment KUB 59.60 ii 8′, 9′ (acc.sg. LÚSANGA kumran, stem form LÚSANGA kumra), may also be a nonce word, although the preserved text does not show obvious Mesopotamian influence.
In other cases, a full assessment remains problematic. For example, it seems reasonable to regard the Hittite huripta- ‘desert’ (only dat.-loc.pl. hu-ri-ip-ta-aš) < the Akkadian huribdu, only attested in the myth of Elkunirsa und Ašertu, as a nonce word in a translation text, but the occurrence of a possible derivative verb huriptai- in a festival for Ištar (KUB 45.46:9′) would point instead to a true, productive loanword despite the inclusion of Hurrian recitations in the text.42 Even if they cannot be strictly regarded as loanwords, nonce words are still worth considering because they can show the strategies implemented in adapting foreign words into Hittite.
As mentioned, one of the major challenges in studying Akkadian loanwords and nonce words in Hittite is identifying the path of transmission: distinguishing between words that entered Hittite directly from Akkadian and those that arrived through the intermediation of other languages, specifically Hurrian and/or Luwian. Only words transmitted directly can explain how Akkadian loanwords were adapted in Hittite; loanwords mediated by Hurrian and/or Luwian are classed as Hurrian or Luwian loanwords in Hittite.
Identifying paths of transmission is not always easy, although some criteria can be established. These concern both adaptation strategies and the type of text in which the words occur. Thus an originally Akkadian word occurring in Hittite transmission with a Hurrian suffix points to Hurrian intermediation. Some examples include the Hittite pūhugari- ‘substitute’ < the Akkadian pūhu, adapted with the Hurrian suffix -ugar-; the Hittite irimpi(t)-/irippi(t)-/eripi- ‘cedar’ < the Akkadian erēnu ‘cedar’ (< the Sumerian erin), with the Hurrian suffix -bi, also occurring in texts belonging to the Hurrian milieu (see below for the alternation between the common gender i-stem and the neuter gender stem in -it); the Hittite ša(n)kunni- ‘priest’ < the Akkadian šangû, šaggû (< the Sumerian sanga), with the Hurrian suffix -nni-; and the Hittite MUNUSentanni- and MUNUSentašši- (a priestess) < the Akkadian entu (fem. of enu < the Sumerian en), with the Hurrian suffixes -nni- and -šši-, the former also attested in Hurrian context (ēntani in KUB 27.34 iv 17′).43 In the case of the noun mitga(i)mi-/mittaka(i)mi-/mintaka(i)mi- ‘sweet bread’, a Luwian intermediation is clear because it is a Luwian participle from the unattested verb *mitkai-, which is allegedly related to the Akkadian matāqu ‘be sweet’, matqu ‘sweet’ (thus mitga(i)mi- = ‘sweetened’).44
Akkadian loanwords and nonce words occurring in Hittite translations or adaptations of Akkadian texts can be safely regarded as directly transmitted from Akkadian to Hittite, without any intermediation, as in the case for GU₄alu- and huripta- (mentioned above). Note, however, that Hurrian is often believed to have had a role in the transmission of Akkadian literature to the Hittites. Theoretically, then, a Hittite translation of an Akkadian literary work could derive from a Hurrian model.
The paths of transmission of most of the allegedly Akkadian loanwords in Hittite remain ambiguous except for the few Akkadian words in which Hurrian intermediation is patent due to the presence of a Hurrian derivational morpheme and an even smaller number of words for which a direct path seems to be the best solution because of the type of text in which they occur. Most do not show any unambiguous derivational morphemes but rather occur in Hittite as i-stems (sometimes with a secondary stem in -it). Examples include LÚapiši- ‘exorcist’ (< the Akkadian āšipu, with metathesis),45 É apuzzi ‘storehouse’ (< the Akkadian abūsu),46 hazzizzi(t)- ‘ear; wisdom’ (< the Akkadian hasīsu ‘ear’), huruppi-, a kind of dish (< the Akkadian huruppu, a metal dish), kappi- ‘bowl’ (< the Akkadian kappu), DUGkazzi(t)-, a container for liquids (< the Akkadian kāsu ‘cup’), kazmi(t)- ‘sample’ (tentatively < the Akkadian kasmu ‘cut, chopped’),47 lahanni-, a flask (< the Akkadian lahannu), makalti-/magalzi-/makanti- ‘(eating) bowl(ful)’ (< the Akkadian mākaltu, a bowl, from akālu ‘eat’), magari- ‘wheel (of the chariot)’ (< the Akkadian magarru, mugarru), GIŠpaini(t)-, ‘tamarisk?’ (cf. the Akkadian bīnu, but the preservation of the original /ai/ in Hittite excludes a direct borrowing from this Akkadian form),48 tuppi- ‘(clay) tablet’ (< the Akkadian ṭuppu < the Sumerian dub), LÚummiyanni-, an official (< the Akkadian ummiānu < the Sumerian um-mi-a, um-me-a ‘expert’), and LÚzakkinni- ‘prefect’ (< the Akkadian šaknu, šākinu). Since Hurrian generally adapted Akkadian words as i-stems,49 and given that some of the words listed here also occur in Hurrian texts and/or in Hittite texts belonging to the Hurrian milieu,50 a Hurrian intermediation is often invoked in the transmission of these words from Akkadian to Hittite. Nevertheless, a direct transmission from Akkadian cannot be entirely excluded in some instances because the i-stem may derive from the Akkadian oblique stem or reflect a common Hittite strategy of thematization, perhaps also influenced by the Luwian i-mutation in New Hittite.
A likely and productive Akkadian loanword for which a direct path of transmission can be suggested is the Hittite arzana-/aršana-, always occurring with per-/parn- (the Sumerian É) ‘house’, < the Akkadian arsānu (a kind of groats). It is likely that the phrase arzanaš per-/parn- (also É arzanaš) originally meant something like ‘porridge-house’ (that is, an inn or hostel) but examples like acc.sg. É arzanan and abl. É arzanaz (or better Éarzanan and Éarzanaz) seem to point to a metonymic extension of arzana- to denote the building. The noun also became productive in Hittite with this new meaning, as evidenced by the derivatives LÚarzanala- ‘attendant of the arzana-house’ and arzanai- ‘quarter, billet’.51
Given the difficulties involved in distinguishing between direct and indirect Akkadian loanwords, the adaptation strategies of Akkadian words in Hittite are not easy to investigate. Except for the few examples only attested as stem forms (non-adapted loanwords or Akkadograms?), thematization offers multiple possibilities that perfectly match the Akkadian case endings: a-stems (arzana-, huripta-, kumra-), u-stems (GU₄alu-, NINDAtappinnu-, NA₄yašpu-), and i-stems (see the list above) are attested, with the latter being particularly prevalent but suspected of being Hurrian loanwords instead. In the attested forms that allow an assessment of gender, the Akkadian words are generally assigned to the common gender in Hittite, with the important exception of tuppi- ‘(clay) tablet’, which is consistently neuter (neuter stems in -it result from Luwian intermediation, as will be shown below).
Hurrian loanwords also occur in Akkadian, such as the Akkadian ambassu, allegedly ‘park, game preserve,’ which is attested in late sources. Possibly it derives from the Hurrian ambašše, a derivative of the Hurrian verb am- ‘burn’ (although with an odd semantic shift in Akkadian). Similarly, the Akkadian apu ‘hole, opening in the ground’, found only in Neo-Assyrian, may derive from the Hurrian āpi ‘sacrificial pit’ (or possibly on Anatolian āpi(t)-, which is a Hurrian loanword), perhaps deriving in turn from the Sumerian ab ‘window’ (which is borrowed and adapted in Akkadian as a feminine noun, aptu ‘id.’).52 Finally, the Akkadian maninnu ‘necklace’, only attested in peripheral Akkadian, mostly in texts belonging to the Hurrian milieu, may ultimately derive from the Indo-Iranian *mani- through Hurrian intermediation, pointing to language diffusion from Mittani to the west (see Chapter 13).53
Examples of the opposite path of transmission, from Anatolian to Akkadian, can be found in the Siege of Uršu text (CTH 7), in which some Hittite words occur both as foreign words and in Akkadian guise.54 In this case, however, we cannot speak of true loanwords. They should rather be regarded as occasional code-switching phenomena.
An example of true Anatolian loanword in Akkadian is argamannu ‘red-purple wool; tribute’ (the latter meaning only attested at Boğazköy), which occurs quite late in the Babylonian and Assyrian documentation.55 It should be compared with the Hittite argama(n)- and the Luwian arkamman-, which surely mean ‘tribute’ but probably also ‘purple-dyed wool’ as emerges from the Manapa-Tarhunta letter (KUB 19.5+).56 Although a Semitic origin is sometimes assumed, an Indo-European etymology seems to be assured for the Hittite and Luwian nouns (both from PIE *h1érk- ‘cut, divide’).57
Besides loanwords and related phenomena involving the transfer of a whole linguistic sign, the lexical interference between Akkadian and the other languages of Anatolia also emerges in structural and semantic calques involving the content. We will first consider structural calques (or loan translations). Hittite compounds built with a genitive and the noun išha- ‘lord’ (e.g., hannešnaš išha- ‘lord of judgment’, that is, a legal opponent) are generally believed to be based on similar Akkadian expressions with bēl ‘lord’ and a noun in the genitive case (e.g., bēl dīni, also ‘lord of judgment’) that occur quite frequently as heterograms in Hittite texts (also with the Sumerian EN in place of the Akkadian bēl).58 It is also possible that Hittite independently built new words, based on an initial set of examples, without needing direct Akkadian models. This would explain the absence of an Akkadian equivalent for nouns such as mukešnaš išha- ‘lord of the ritual’. In that case, Akkadian would not have caused merely lexical interference but rather introduced a new word-formation rule into the target language.
The Hittite compound šiyannaš per ‘seal house’ (written as É šiyannaš or, fully heterographically, as É NA₄KIŠIB) is possibly a calque on the Akkadian bīt kunukkim, and the Hittite anišiwat ‘today’ (lit. ‘this day’) and appašiwatt- ‘future’ (lit. ‘after-day’) may be calqued on the Akkadian ūmu annû and (w)arkiat ūmi, respectively (also note the Sumerogram EGIR.U4KAM).59 Phraseological calques seem also to be attested, such as the Hittite išhiul išhiya- lit. ‘to bind a binding’ (that is, to conclude a treaty) < the Akkadian riksa rakāsu, often occurring in treaties, or the Hittite aššul hatrae- ‘write favor’ < the Akkadian šulma šapāru, frequently attested in Hittite letters, which largely derived from Akkadian models in the formulary.60
In some cases, it is difficult to assess whether a given Hittite compound is modeled on an Akkadian word or its Sumerian counterpart occurring in Hittite texts as a heterogram. Thus the Hittite šuppiwašharSAR ‘onion’ could have been based on the Sumerian sum-sikilsar (also attested as a heterogram in Hittite) or perhaps the Akkadian šamaškil(l)u, šusikilu, borrowed from Sumerian. The Sumerian model, being more transparent than the Akkadian model, is perhaps a better candidate for the Hittite compound.61
As for semantic calques (or loan shifts), some technical meanings attested for Hittite words are probably derived from an Akkadian model. For example, the Hittite parkuešš- ‘become pure’ also had the legal meaning ‘be proven innocent’, thus semantically matching the Akkadian ebēbu(m) ‘be pure; be innocent’.62 Another possible example is the Hittite haštai- ‘bone’, which also denoted a measure of length, possibly after the Akkadian eṣemtu(m) ‘bone; fraction of a cubit’; however, the latter meaning has only been attested in Neo-Assyrian,63 so the hypothesis of a semantic calque in Hittite remains uncertain. Finally, the specific type of groom defined in Hittite as LÚantiyant-, lit. ‘the one who entered (his wife’s family)’, that is, a son-in-law, may attest to a calque of the Mesopotamian erēbu(m) marriage.
Considering lexical transfer in the opposite direction reveals that Hittite represented a source of linguistic contents for Akkadian. This is shown by some phraseological calques on Hittite structures occurring in the Akkadian versions of treaties and political documents issued by the Hittite kings (see Chapter 8). These calques were probably not systemic, and most should be understood as interference phenomena at the single-document level.
5 Hurrian, Luwian, and Hittite between Hatti and Kizzuwatna
As most recently shown by Mouton and Yakubovich (2021),64 second-millennium BCE Luwian, also referred to as Cuneiform Luwian, was not homogeneous. Different varieties should be distinguished: 1) the Luwian of Hattuša, also called Empire Luwian, which surfaced in the Glossenkeilwörter and later developed into Iron Age Luwian; 2) the Luwian of Kizzuwatna and the Lower Land, attested by incantations embedded in rituals from the two regions; 3) the Luwian of Tauriša, also attested in conjurations in ritual texts; and 4) the Luwian (or Luwic? See Chapter 11) of Ištanuwa, attested only in a couple of festival texts belonging to the cult of that city.65 Attestations for a fifth variety, the Luwian of Arzawa, are very scanty, being limited to personal names and possibly sporadic Luwian words that were included in Arzawa rituals written completely in Hittite.66 Thus, when studying lexical interference between Luwian and Hittite, we should try to distinguish between the different areas and varieties involved rather than considering Luwian as a whole. The two most important topics to be investigated separately are the interface between Kizzuwatna and Hatti, which is dealt with in this section and involved Hurrian, the Luwian employed in the different ritual traditions, and Hittite, and the relationship between Empire Luwian and Hittite at Hattuša, to which the next paragraph is devoted.
With the annexation to Hatti of the formerly independent kingdom of Kizzuwatna and the general expansion of the Hittite influence under Tuthaliya I, several rituals belonging to the Hurro-Luwian milieu reached Hattuša, where they were copied and variously adapted for different purposes. Two different groups of so-called Kizzuwatna rituals were identified (Miller 2004; Yakubovich 2010). The first group consisted of rituals with Luwian incantations that show no traces of Hurrian elements, possibly pointing to the migration of ritual practitioners to Hattuša, where the texts were written by Hittite scribes under dictation. The second group consisted of rituals with both Hurrian and Luwian elements, perhaps originally recorded at Kizzuwatna and later copied at Hattuša.67 While recently a new and more complex model of the Luwian traditions and dialectology has been proposed (cf. Chapter 11, with references), the distinction based on the presence or absence of Hurrian material remains significant for our current purpose.
Through the path of transmission to Hattuša, several cultic technical terms, both Luwian and Hurrian, entered Hittite. Luwianisms in Hittite rituals were collected by Melchert (2013), who distinguished between words of Luwian origin and words whose origin was ultimately Hurrian. Melchert showed that the rituals belonging to the first group include, besides Luwian incantations, a larger number of originally Luwian words than rituals written completely in Hittite with both Luwian and Hurrian elements. In the latter group, true Luwianisms are markedly less common than words of Hurrian origin. Interestingly, but maybe not surprisingly, some of the Luwianisms occurring in the Hittite sections of the rituals with Luwian incantations belong to Empire Luwian (cf. the acc.pl. NINDApartanninzi and NINDAwartanninzi in KUB 17.12 vs. the expected acc.pl. ending -nz(a) of the Luwian of the rituals), as do some Luwianisms in Hittite rituals that have Hurrian and Luwian elements but no Luwian incantations (cf. the acc.pl. nišhinzi šūntinnānzi in the ritual of Ammihatna, KBo 5.2 iii 29). If these words belonged to Empire Luwian rather than merely attesting to a sporadic interference of the Luwian of Hattuša on foreign Luwian loanwords in Hittite, the degree of lexical interference of non-local Luwian on Hittite might need to be reconsidered. One might wonder, e.g., whether Kizzuwatna/Lower Land and Tauriša Luwian were confined almost exclusively to the incantations in these rituals, whereas the Luwianisms in the Hittite ritual framework mostly belonged to the Luwian dialect spoken in the Hittite capital. Possible exceptions could be some specific technical terms related to the Kizzuwatna cult (mostly of Hurrian origin), which, however, were also not immune to Empire Luwian influence (cf. the aforementioned nišhinzi šūntinnānzi) because Kizzuwatnean Hurro-Luwian technical terms probably also entered Empire Luwian (e.g., the acc.pl. šehellinzi in the oracle report IBoT 2.129 obv. 23). Therefore, the adaptation strategies of Luwian nominals in Hittite will be dealt with in the following section, which is dedicated to lexical interference between Empire Luwian and Hittite.
Hurrian loanwords could have entered Hittite through a twofold path based on the different adaptation strategies found in Hittite. Some reached Hittite through Luwian intermediation, as is shown by the typically Luwian strategy of adaptation of foreign words as stems in -it,68 reflected in Hittite as neuter gender dental stems (with the dental stop only appearing in oblique cases).69 A secondary thematization as a common gender a-stem occurs sporadically, as in the case of harzazuta- ‘breadmash’.70 Other Hurrian loanwords may have had direct Hurrian models unless they also came from Luwian and were recategorized as simple vocalic i-stems by dropping the expected dental consonant. These words are generally assigned to the common gender. Sometimes two allotropies are attested for the same word, perhaps reflecting direct and indirect paths of transmission (e.g., the common ahrušhi- vs. the neuter ahrušhi(t)- ‘censer’, the common eripi- vs. the neuter irippi(t)- ‘cedar’, and the common kazmi- vs. the neuter kazmi(t)- ‘sample’). As mentioned above, the Akkadian words that entered Hittite via Hurrian (and sometimes also Luwian) intermediation also had a twofold path of transmission.
Since we are dealing with materials transferred to Hattuša from elsewhere, it is not to be expected that lexical interference occurred in the opposite direction at a systemic level, that is, from the languages spoken in the Hittite capital city to Kizzuwatna Luwian and Hurrian. Only few Hittite words occur in Luwian passages embedded in Hittite rituals: the noun halmaššuitti- ‘throne’, found in a ritual fragment (KUB 35.67 ii 2′); the genitival adjective hurkilašša/i-, built on the Hittite hurkil-, which occurs in a Luwian phrase in the ritual of Zuwi (KUB 35.148+ iv 13′),71 and the noun GIŠkattaluzzi(t)- ‘threshold’, which was adapted as a stem in -it- in the ritual of Puriyanni (KUB 35.54 iii 29).72 Furthermore, Empire Luwian had a limited influence on Luwian passages included in Luwian-Hittite rituals written at Hattuša. This took the form of non-systemic grammatical interferences that only affected single documents, such as the occasional replacement of original plural accusatives in -nz(a) with the innovative Empire Luwian forms in -nzi, modeled on the plural nominative.73 As mentioned, some Empire Luwian words also seem to occur in Hittite rituals with Luwian passages.
6 Luwian and Hittite at Hattuša
As is acknowledged after Yakubovich (2010), most of the Luwian words occurring in Hittite texts, with and without Glossenkeil, reflect the Luwian dialect spoken at Hattuša, Empire Luwian, which shows some innovations compared to the older Luwian of the rituals and later developed into the Luwian variety attested in the hieroglyphic inscriptions. Thus, in the bilingual environment of the Hittite capital city, it was this Luwian dialect that mostly influenced the Hittite language, both at the lexical and grammatical level. As previously mentioned, Empire Luwian influences can also be found in Luwian rituals, both in the incantations (plural accusatives in -nzi replacing expected forms in -nz(a)) and in Hittite ritual instructions (Luwian loanwords showing unique Empire Luwian features).
Being genealogically related, Luwian and Hittite were structurally similar. Because they had the same grammatical categories, complex adaptation strategies of Luwian loanwords would be unlikely in Hittite. In considering lexical interference between Hittite and Luwian, the major problem is rather to identify Luwian loanwords unambiguously, which can only be done on etymological grounds. The Glossenkeil and unique Luwian endings could only prove the existence of a given word in Luwian, not its original Luwian status. Thus the Hittite :aggatiuš ‘hunting net’ (acc.pl. in KUB 8.56 i 12″) surely derives from a Luwian model because it matches the Hittite cognate ēkt- and complies with Čop’s Law. Conversely, the unambiguous Luwian abl. GIŠkalmušati, which occurs in Hittite context (KUB 44.60+ iii 9), only attests to the existence of a Luwian noun kalmuš- ‘lituus’ but cannot indicate whether the noun’s ultimate origin was Luwian.74
True loanwords, which can be expected in different texts because they belong to the langue, must be distinguished from code-switching phenomena, which are confined to individual acts of parole, although the strategies of adaptation to the Hittite language—for words that were adapted—could have been the same in both cases. The presence of the Glossenkeil may be a helpful criterion, although not entirely reliable. As suggested by Yakubovich (2010:396–410), the lexical function of the Glossenkeil to mark single words is sometimes best explained as a mark of incompetence code-switching. From Muršili II’s reign forward,75 scribes with non-native knowledge of Hittite marked, more or less consistently, Luwian words they were not able to translate into Hittite and that were perceived as stylistically inappropriate. This marking probably was intended to highlight those forms to be assessed by senior scribes and possibly replaced in edited copies of the text.76 However, the use of the Glossenkeil was not consistent and depended on how an individual, presumably non-native, Hittite scribe assessed a given word. Therefore, a Luwian word marked by the Glossenkeil in a given text, even if understood as a code-switching phenomenon by a non-native scribe who considered it stylistically inappropriate, could have corresponded to a true Luwian loanword in Hittite, whose acceptability might have been assessed differently by a native Hittite senior scribe.
Contact between Luwian and Hittite at Hattuša was much more intense than contact among Hittite and the other languages that we have discussed; it also affected the structural levels of the language (see Chapter 15). Lexical interference in this situation also went beyond the mere transmission of nominals to involve other parts of speech, such as verbs and adverbs. In the following sections, we focus on Luwian lexical loanwords in Hittite, divided according to their parts of speech. Limited evidence for the opposite path of transmission—that is, from Hittite to Empire Luwian—also exists. For example, /xassusara-/ ‘queen’, occurring in Iron Age Luwian hieroglyphic inscriptions, probably was a Hittite loanword traceable to the Empire period,77 as well as the Luwian noun /mugissar/ occurring in the KARAHÖYÜK inscription (12th c. BCE), clearly reflecting the Hittite mukeššar ‘invocation ritual’ because the form is not compatible with the phonological and morphological characteristics of the Luwian language.78 There are also some possible calques. It has been recently suggested that the phrase annān tiššā(i)- occurring in Puduheba’s letter to Ramses II (KUB 21.38) is based on the Hittite kattan handā(i)- ‘to match with’. The Hittite model with kattan would explain the unexpected occurrence of the Luwian annān ‘below’ instead of *ānna ‘with’ (Hieroglyphic Luwian CUM-na).79 Furthermore, the meaning of ‘defeat’ or ‘conquer’ for the Luwian verb muwa- occurring with the reflexive particle -ti could be calqued on the Hittite -za tarhu- ‘id.’.80
6.1 Lexical Interference between Empire Luwian and Hittite: Nominals
Empire Luwian nominals—regardless of their status as true loanwords or code-switching phenomena (see above)—occur in Hittite texts either as foreign words (i.e., with Luwian endings) or adapted forms. Given the structural similarity between the two languages, Luwian words adapted to Hittite generally maintain their original stems and grammatical genders. Luwian nouns showing the i-mutation can be adapted either as i-stems or a-stems or both. Some examples of common-gender nouns include the Luwian armanna/i- ‘lunula’ > the Hittite armanni-, the Luwian halhalzāna/i- (a body part) > the Hittite (:)halhalzana-, the Luwian harpanalla/i- ‘rebellious; rebel’ (adj. and n.) > the Hittite harpanalli-, the Luwian hutanu(i)- ‘mng. unkn.’ > the Hittite hūtanui-, the Luwian nakkušša/i- ‘scapegoat’ (< Hurrian) > the Hittite nakkušša/i-, and the Luwian papašāla- ‘esophagus’ > the Hittite UZUpappaššala/i-.81 Some neutral-gender nouns include the Luwian GIŠir(h)wit- ‘basket’ vs. the Hittite GIŠērhui-, the Luwian NINDAtannaš- (a type of bread) > the Hittite NINDAdannaš-, and the Luwian upatit- ‘landgrant’ > the Hittite upati-. We find a mismatch between Luwian and Hittite in sporadic cases. For example, the neuter stems in -it- dupšāhit- (a ritual) and waškuit- ‘offense’ are adapted as the Hittite common-gender i-stems dupšāhi- and waškui- (with secondary stem wašku-).82 As mentioned previously, gender mismatches between Luwian and Hittite may result from the different paths of transmission when original Hurrian words are involved, so a Hittite common-gender i-stem may not reflect the corresponding Luwian neuter stem in -it-.
In some cases, it is not clear whether a Luwian noun in Hittite transmission was adapted to Hittite or occurred as a foreign word. There are limited instances of overlap between the Luwian and the Hittite noun declensions (nom.sg. in -š, acc.sg. in -n, and dat.sg. in -i), and nouns in the genitive case can also present problems. Several Luwian words in Hittite transmission occur in the genitive case with the ending -aš. They are generally regarded as forms adapted to the Hittite language, with the Hittite genitive ending. It is generally assumed that Luwian lost the genitive case, and its function was taken over by a relational adjective in -ašša/i-, occurring in agreement with its head noun. However, while the loss of the genitive case can be ascertained for the Luwian of the ritual traditions, for which we have proper texts, it is by no means assured for Empire Luwian, only attested in isolated words in cuneiform texts. Furthermore, if we take into account hieroglyphic texts, which used a Luwian dialect that, as mentioned, was closely related to Empire Luwian, we find considerable evidence for genitives in /-as/ alongside relational adjectives. Lycian—another Luwic language—also shows the genitive case in -ah and the relational adjective. Therefore, the loss of the genitive case could be regarded as an innovation that only affected the Luwian of the rituals. This makes it impossible to establish whether a Luwian word occurring in Hittite in the genitive case should be regarded as an adapted form or not.83
6.2 Lexical Interference between Empire Luwian and Hittite: Verbs
Apparently Luwian was the only language from which Hittite borrowed verbs. Luwian verbs occur in Hittite both as foreign words (i.e., with Luwian endings) and adapted loanwords.84 The latter were consistently assigned to the Hittite mi-conjugation, which was more productive than the hi-conjugation or perhaps the unmarked one or just more similar to the Luwian conjugation.
As in the case of nouns, the major challenge in analyzing Luwian verbs borrowed in Hittite is identifying them. While verbs that occur frequently in Hittite texts and consistently show unique Luwian endings can be regarded as Luwian non-adapted loanwords (e.g., tapar- ‘rule, govern’, which only shows Luwian endings in Hittite texts), the same is not true for hapax legomena and adapted forms for two reasons: 1) the status of a true loanword cannot be assigned to hapax legomena—or perhaps even to words that occur sporadically—because we cannot exclude code-switching phenomena, which are conceptually different from loanwords; and 2) adapted forms—that is, verbs with Hittite endings—can only be recognized as Luwian loanwords when they have an etymologically unambiguous Luwian verbal root (further considerations are in order when derived stems are involved, as discussed below).
As mentioned, the presence of a Glossenkeil is insufficient to identify an original Luwian word,85 and the original Hittite or Luwian status of the base simply cannot always be determined. For example, the verb hašp- ‘overcome’ occurs with both Hittite and Luwian endings from Middle Hittite texts and is generally assumed to be Luwian in origin.86 However, there is no etymological ground to unambiguously assign the root hašp- to Luwian and thus regard the Hittite verb as a borrowing: they could have been simply cognates. The fact that later Luwian forms are marked by the Glossenkeil is not relevant because they could have resulted from code-switching by a scribe who was unaware of the existence of a cognate verb in Hittite.87
6.2.1 Root Verbs and Deverbal Derivatives
If the ultimate base is an original Luwian verb, underived stems are not problematic: an original Luwian root verb occurring in a Hittite context with Hittite endings can be identified as a Luwian verb borrowed in Hittite (unless it is a case of code-switching). The same may apply to derived verbal stems with unique Luwian suffixes (e.g., the imperfective Luwian suffix -zza- vs. the Hittite -ške-) and perhaps also to derived verbal stems built with a unique Hittite deverbal suffix, which can simply be a matter of adaptation.
The distinction between true loanwords and code-switching phenomena can be evaluated by considering multiple elements, such as the number of occurrences, the presence of the Glossenkeil, and other contextual information. Some assessments are clear-cut. For example, the unique form :a-wi5-ya-ah-ha[(-), which occurs in the fragmentary prayer KUB 36.96:12′, must belong to the unambiguous Luwian root verb awi- ‘come’ (vs. the Hittite cognate uwe- ‘id.’) and should probably to be explained as a ‘badly Hittitized’ imperative form, :awiyahha[ru] (Melchert 2022 s.v.). It is a likely example of code-switching due to its unicity, unexpected form, the presence of the Glossenkeil, and the existence of other Glossenkeilwörter in the same context, which point to a general Luwian influence on the text. Nothing argues for the existence of a true Hittite verb awi- borrowed from Luwian.
Other cases are less straightforward. For example, the Hittite šappai- ‘peel, trim’, derived from the Luwian verb šappa-, is only attested in KUB 8.50+ iii 23′ in a Hittite context (:šap-pa-at-ta, together with other Luwian Glossenkeilwörter). The Luwian status of the base is assured, the Hittite corresponding verb being the poorly attested šippai- (Melchert 2022 s.v.).88 The only Hittite adapted form, ša-ap-pa-a-iz-zi, is found in the Hittite medical text KUB 44.63+ ii 11′, without the Glossenkeil and in a fully Hittite context showing no traces of Luwian interference. Therefore, it should possibly be regarded as a true loanword even though it is a hapax. Also note that the only two assured occurrences of the Hittite šippai- are found in the MS ritual KUB 29.7+,89 so we could also imagine a later replacement of the original Hittite verb with a borrowing from Luwian.
The Hittite hap(p)uš- ‘make up for, replace’, which occurs frequently in Hittite texts (cf. HED H:133–134), is sometimes regarded as a Luwian loanword because of the Luwian pres.1sg. ha-pu-uš-wi5 (KBo 31.169 obv. 8′, in a Hittite context) and two occurrences of the Hittite ha-pu-ša-an-zi with the Glossenkeil in the oracle report KUB 16.66:5′, 6′. However, the etymology is unclear,90 and the Hittite and Luwian verbs may be cognates.
The Luwian verb šiwari(ya)- ‘deny, withhold’ (?)91 occurs with unique Luwian endings in some NH letters, sometimes marked by the Glossenkeil. The verb is found twice with Hittite endings—the unusual pres.2sg. ši-wa-ri-eš-ši in KUB 23.97 and the pret.3sg. ši-wa-ri-ya-it in KUB 21.38 (probably twice)—which may represent sporadic Hittitizations rather than a loanword.
The Hittite hapax paštarnuwanzi found in the Hittite-Luwian ritual KUB 35.146 ii 14′, probably belonging to the tradition of Tunnawiya’s rituals,92 surely represents the Hittite adaptation of the Luwian causative verb paštarnuwa- ‘cause to fly up’,93 derived from the Luwian verb paštari-.94 A Luwian origin seems assured if, as suggested by Puhvel (HED PA:193), the base should be explained as pa-štar-, with the Luwian prefix pa- matching the Hittite pe-. But, again, the single occurrence of the Hittite verb is insufficient to establish whether it reflected a loanword or a code-switching phenomenon.
A clear example of a Luwian verbal loanword in Hittite is pušša(i)-, the imperfective stem of puwa(i)- ‘pound, crush’, built with the suffix -šša-. The Hittite verb is attested nine times and is only once marked by the Glossenkeil, which perhaps makes the hypothesis of code-switching less likely.95 Although the base verb puwai- is attested in Hittite and may be a cognate of the Luwian verb puwa(i)-, there can be little doubt that the derivative pušša(i)- should be regarded as a loanword, both because the imperfective suffix -šša- is not productive in Hittite, where it only occurs with four verbal stems (halzai- ‘call (out)’, iya- ‘do, make’, šai- ‘press’, and warrai- ‘help’),96 while it is very productive in Luwian, and especially because the Hittite pušša(i)- belongs to the mi-conjugation, whereas the Hittite verbs in -šša- belong to the hi-conjugation. For the same reason, the Hittite hapallašai- ‘injure’97 and the Hittite tišša(i)- ‘shape, prepare’,98 both belonging to the mi-conjugation, probably derived from the corresponding Luwian verbs if they were built with the suffix -šša-.
The Hittite tiyaneške-, which occurs only twice in Tunnawiya’s ritual (KUB 7.53+ ii 10, 17), probably represents the Hittitization of the Luwian verb tiyanišš(a)- ‘fill, stuff’ (?), which occurs in the same context (pres.1sg. ti-ya-né-eš-šu-i in KUB 7.53+ ii 12). Unlike the examples quoted above, it replaces the productive Luwian imperfective suffix -šša- with its Hittite productive counterpart, the suffix -ške- since an imperfective form is required by the context. However, even if we take it as a loanword—although the occurrence of the corresponding Luwian form in the same context instead argues for code-switching involving morphological adaptation—the ultimate origin of the base cannot be determined, nor can the exact derivational process be established.99
Finally, the issues surrounding the Luwian verb tapar- ‘guide, rule’ and the forms related to it are quite complex. The Luwian (:)tapar- is quite frequently attested in Hittite texts as a non-adapted loanword (i.e., always has Luwian endings).100 Although some forms are marked by the Glossenkeil, the frequency with which the verb occurs and the textual typologies in which it is found point to a true loanword rather than a code-switching phenomenon. Related to this root is the Hittite verb taparriya(i)-, also Luwian in origin, but whose explanation is less clear. Different scenarios can be envisaged. According to Starke (1990:259), followed by Melchert (2003b:207), the Luwian tapar- is a secondary stem back-formed on the Luwian tapariya-, which is only attested in Hieroglyphic Luwian and which would also be the source of the Hittite taparriya(i)- in this hypothesis. Kloekhorst (EDHIL:831) suggested instead that the Luwian tapar- was the base of all the other derived stems, including the Luwian tapariya-, which he too regarded as the source of the Hittite verb. Sasseville (2020:31, 271–273) proposed a more complex scenario, in which 1) tapar- would be a primary root verb; 2) the Hieroglyphic Luwian tapariya- would be a denominative verb based on the action noun tapariya- ‘command’; and 3) the latter would derive from an unattested Luwian verbal stem *tapari-. For the Hittite forms, Sasseville distinguished between the denominative taparriyai- and the adapted verb tapariye/a-, which would reflect the unattested Luwian verb *tapari-. If one accepts Sasseville’s scenario, the situation can be summarized in the following schema (Fig. 14.2), which only takes into account the relevant tapar-related words.
6.2.2 Denominal and Deadjectival Stems
Verbs derived from nouns or adjectives through a unique Luwian suffix may be safely regarded as originally Luwian verbs, provided that the Luwian suffix cannot be confused with a homophonic Hittite suffix. But when verbal stems are derived from nouns or adjectives through ambiguous or uniquely Hittite suffixes, we cannot establish definitively whether the derived verb occurring in Hittite is an adaptation of the Luwian verb or merely a derivative verb built in Hittite on a noun borrowed from Luwian (pointing to the productivity of the noun), even when the original Luwian status of the base noun is assured and the denominative verb is attested in Luwian. As an example, the verb anahidai-, which only occurs with Hittite endings in Hittite texts, is clearly based on the Luwian noun anahit- ‘sample’, which is a loanword from Hurrian adapted in Luwian as a stem in -it- (see above), whereas an alleged Luwian verb *anāhidā(i)- is not attested. How should we explain the Hittite verb anahidai-? Is it the adaptation of an unattested Luwian verb—that is, a Luwian verb borrowed in Hittite (scenario 1)—or a denominative Hittite verb based on the noun anahit-, borrowed from Luwian and become productive in Hittite (scenario 2)?101
Figure 14.2
The process of borrowing of tapar- into Hittite
Scenario 1
the Luwian anahit- (n.) → the Luwian *anāhidā(i)- (v.) → the Hittite anahidai- (v.)
Scenario 2
the Luwian anahit- (n.) → the Hittite anahi(t)- (n.) → the Hittite anahidai- (v.)
Among the almost 50 verbs that can be possibly regarded as adapted Luwian loanwords in Hittite, excluding lexicalized Luwian participles (e.g., NINDAwalipaimiuš in KBo 24.29 iii 6), several denominative verbs are found (anahidai-, annara(i)-, appalai-, arrahhani(ya)-, aštaniya-, hapā(i)-, huntariyai-, kušalā(i)-, lila(i)-, malāi-, patalhai-, putalliya-, taparriyai-, tarkummā(i)-/tarkummiya(i)-, zammurā(i)-, and zappantala-). However, the Luwian status of the base is not proven for all of these, nor for some deadjectival verbs (hantalliyai-, hantiyai-, maruwāi-).
Choosing for the moment to leave aside the problem of the distinction between true loanwords and code-switching phenomena, let us pretend that all these Hittite verbs can be regarded as possible Luwian loanwords. To evaluate the possibility that a given Hittite verb is a borrowing from a Luwian denominative verb and not an inner-Hittite formation based on a Luwian noun borrowed in Hittite, we summarize the relevant data in the following table, which provides the following information for each Hittite verb allegedly Luwian in origin: 1) the attestation—in Luwian texts and/or Hittite transmission—of the Luwian denominative verbs (with Luwian endings); 2) the attestation of the Hittite base nouns (with Hittite endings) that were allegedly borrowed from Luwian; 3) the attestation of the Luwian base nouns (with Luwian endings) that were originally Luwian or borrowed from another language.
|
Hittite verb (= with a Hittite ending) |
Luwian denom. verb (= with a Luwian ending) |
Hittite base noun (= with a Hittite ending) borrowed < Luwian |
Luwian base noun (= with a Luwian ending) |
|---|---|---|---|
|
anahidai- |
no |
yes |
yes102 |
|
annara(i)- |
no |
no103 |
yes |
|
appalai- |
no |
yes |
no104 |
|
arrahhani(ya)- |
yes |
(no) |
(no)105 |
|
aštaniya- |
(yes)106 |
(yes) |
(yes)107 |
|
hapā(i)- |
no108 |
(no)109 |
(yes)110 |
|
huntariyai- |
(yes)111 |
no |
(yes)112 |
|
kušalā(i)- |
no |
no |
(no)113 |
|
lila(i)- |
yes114 |
yes |
yes |
|
malāi- |
yes115 |
no |
yes |
|
patalhai- |
Yes |
yes |
no116 |
|
putalliya- |
Yes |
no |
no117 |
|
šapiyai- |
yes118 |
yes |
yes |
|
taparriyai- |
(yes)119 |
no |
(yes)120 |
|
tarkummā(i)-/tarkummiya(i)- |
yes |
no |
no121 |
|
zammurā(i)- |
no122 |
no |
(no)123 |
|
zappantala- |
yes |
no |
no |
If the Luwian status of a noun base is assured on etymological grounds—which is by no means always obvious—the most favorable situation for assuming that a Hittite verb is directly borrowed from a Luwian denominative verb is when the base noun is not attested with Hittite endings and the denominative verb is attested with Luwian endings. Conversely, when the allegedly Luwian base noun is also attested in Hittite in an adapted form, an inner-Hittite derivation of the denominative verb from the Hittite noun cannot be entirely excluded, even when the Luwian denominative verb exists, and all the more so if it is not attested. If the Luwian status of the base noun cannot be etymologically determined due to the lack of unambiguous Luwian phonological features, the probability of a borrowing from Luwian decreases, even if the Luwian denominative verb is attested. Despite what is sometimes claimed, the occurrence of a denominative verb with unique Luwian endings cannot be taken as proof of the Luwian status of its base noun. It can only show that the noun existed in Luwian and was productive; only the etymology can guarantee its original Luwian status. Without unique Luwian phonological features, we cannot distinguish a genuine Luwian word from a possible Hittite cognate or exclude the possibility of a Hittite loanword in Luwian. This can be summarized in the following table:
|
Assured Luwian base noun |
Hittite noun borrowed < the Luwian |
Luwian denominative verb |
Probability of Hittite verb borrowed < Luwian verb |
|---|---|---|---|
|
+ |
– |
+ |
++/– |
|
+ |
– |
– |
+/– |
|
+ |
+ |
+ |
+/– |
|
+ |
+ |
– |
–/+ |
|
– |
+ |
+ |
–/+ |
|
– |
– |
+ |
–/+ |
|
– |
+ |
– |
––/+ |
|
– |
– |
– |
––/+ |
The addition of the loanword vs. code-switching issue complicates the situation further, although the problem of the dependence of the Hittite verb on a Luwian verb vs. its derivation from a Hittite noun borrowed from Luwian can perhaps be partly bypassed because a Hittite verbal form depending on a Luwian model that results from code-switching is likely to be taken as an extemporary Hittitization of a Luwian verb or possibly an occasional form built on a Luwian noun but not a genuine Hittite form built through an inner-Hittite derivation process based on a Hittite noun.
6.3 Lexical Interference between Empire Luwian and Hittite: Other Parts of Speech
Sporadically, other parts of speech were involved in the borrowing process, such as the Luwian adverbs ziladuwa and ziladiya ‘in the future’.124 The full integration of such adverbs in the Hittite lexicon is proven by the edited version of Muwattalli II’s prayer to the assembly of gods (CTH 381), KUB 6.45+, in which the marked :ziladiya occurring in the draft (KUB 6.46 iv 23) is accepted without the Glossenkeil (iii 54).125
A special case is the occurrence of the Luwian adverb šarra in a gloss in Hittite included in the Akkadian medical omens, KBo 9.49 obv.? 14′ […]auliš šarra tarru artari “the auli- [probably the spleen] stands up firmly.” This occurrence appears to be unique because the Hittite adverbs šarā and šer are consistently used in the Hittite context. Since the glosses in medical omens probably reflected a less controlled language than that used in more official texts, plausibly closer to the spoken language of the scribe, two scenarios could account for šarra: 1) the scribe was a Luwian native speaker, who included a form of his language—an occasional code-switching phenomenon—in a text that had not been finalized by going through a process of standardization that might have produced a full translation into Hittite; 2) the colloquial Hittite language of the Empire period was even more Luwianized than we see in official documents, so genuine Hittite local adverbs were the standard in official Hittite texts, but some original Luwian adverbs might have belonged to the lower registers of Hittite.126
7 Concluding Remarks
The Indo-European Anatolian cultures of the Bronze Age were part of a network that involved other local cultures as well as cultures from the Syro-Mesopotamian interface. Over the centuries, this scenario triggered the direct and indirect borrowing of lexical material in the form of loanwords and Wanderwörter. The Hittite archives, with their wealth of textual documentation, shed light on borrowing patterns that involved several of the surrounding languages. Different ‘paths of circulation’ can be identified. Hattian, Palaic, and Hittite seem to have constituted a preferential ‘circle’ in the very early phases—apparently the only circle that involved Palaic. In the more mature phases of the Hittite history, the rate of lexical exchange with Luwian and with Hurrian intensified. Akkadian seems to have remained mostly peripheral with respect to direct borrowing, although important cultural calques can be identified.
Besides highlighting the direction and date of these patterns of lexical interference, we followed up on our previous study (Giusfredi and Pisaniello 2020) by categorizing the material by the strategies used for morphological adaptation. Due to the different grammatical structures of the languages involved, such process proved very helpful also to offer an improved contextual description.
Cf. Weeden 2011:10–11.
Cf., e.g., Rubio 2005:330–331 fn. 80.
For example, in a target language with a pervasive gender system, all borrowings are necessarily assigned to a grammatical gender, regardless of the addition of a dedicated morpheme.
Cf., e.g., the adaption of the German Steinbock to the Italian stambecco ‘Alpine ibex’ instead of the expected *stambocco because becco is an Italian word meaning ‘buck’ or ‘ram.’ Similar alteration phenomena may also involve grammatical suffixes. For example, the Old Persian *ganzabara- ‘treasurer’ became the Lycian gasabala-, with the sequence -ara- of the Persian model—which is not a morpheme, but part of the lexical element bara-—altered to -ala-, probably in order to match the Lycian ‘professional’ suffix (for a full discussion and references, see Volume 2.
On the distinction between adaptation and integration, see, e.g., Gusmani 1986 (who labels them integrazione and acclimatamento, respectively), followed by Cotticelli-Kurras 2012:75–76.
Cf., e.g., the Akkadian pūhu(m) ‘replacement, substitute’ > the Hurrian pūhugari- ‘id.’. The Akkadian word is adapted in Hurrian through a derivational suffix whose exact meaning is unclear.
See Thomason and Kaufman 1988:74 for a comprehensive discussion of borrowed linguistic items relative to the degree of language contact.
Cf., e.g., Tesnière 1939:85; Deroy 1956:66, and Thomason 2001:69.
See, e.g., Haugen 1950:224, Moravcsik 1978, Muysken 1981, Field 2002:36–40, and Matras 2007:71.
Cf. Campbell 1993.
Cf., e.g., the typology provided by Cotticelli-Kurras 2007:95–96.
Cf. Chapters 4 and 10.
For the identification, among loanwords of various origins, of Anatolian words in Ugaritic, see Watson’s contributions (in particular Watson 1995, 2015, and 2018).
The painstaking identification of non-Semitic loanwords in Ugaritic was mostly accomplished by Watson (1995, 1996, 1998, 1999, 2000, 2006, 2009, 2010, and 2015).
See Giusfredi 2018 for a discussion of this Wanderwort. On the problem of Indo-European words borrowed by Sumerian, see also the illuminating and very critical discussion by Rubio 1999.
See above, Chapter 13, for a discussion. The limited amount of lexical material available does not reduce the historical importance of the cultural connections between the ancient Near East and the Indo-Iranian regions and peoples.
Del Olmo Lete and Sanmartín 2015:26.
Examples include the famous case of the Hebrew kwb‘, Ugaritic kpḫ, and Hittite kupahi-, which is, however, ultimately Hurrian in origin: kufahe. See also Hoffner 1964, Del Olmo Lete and Sanmartín 2015:447, Puhvel HED K:257–258 (with mention of other Western alleged cognates), and Oreshko 2018:105. The same path of diffusion can be assumed for cases that received less scholarly attention, such as ’bws (from the Hurrian abuzi/Hittite apuzzi via the Ugaritic ı͗bsn) and ’gn (from the previously mentioned Hurrian/Hittite aganni via the Ugaritic a͗gn, possibly also with involvement of Egyptian). We do not include a full list, which can be easily found in Noonan (2019). However, we must correct a mistake in Noonan (2019:107): ḥtwl is not a direct derivation from the Hittite huttulli but rather is present in Ugaritic as ḥtl (Del Olmo Lete and Sanmartín 2015:107).
Most notably, in the recently published letter of Wiušti KBo 71.81 (Barjamovic and Schwemer 2018:89), the name of the city Ša-la-ti-wa-ra=ma at rev. 8 is thematized as an a-stem in an Anatolian fashion and, perhaps significantly, the name of the city of Hattuš (e.g., obv. 3 Ha-tù-uš), shows the usual lack of a vowel theme.
Giusfredi and Pisaniello (2020:215); for the attested forms cf. HW2 III:726–728 and IV:72–73.
The form is attested both in Hattian and Hittite contexts (as a common-gender accusative in the latter case); see Soysal 2004:695.
The element is present in the Kārum-period female personal names ending in -hšušar, on which see Kloekhorst 2019:235–239.
Cf., however, Chapter 4 for discussion.
Palaic exhibits two morphemes, -pa and -pi, which might have been allomorphs; even if they weren’t, they would still be explainable as inherited materials. See also Chapter 12.
See Simon 2012 for a discussion of the phonetics of the fricatives in Hattian.
DCL s.v.
For the paradigm of attested forms, see Carruba 1970:79. The availability of both a dative-locative and a nominative guarantees the recognition of common gender.
Attested in the plural fulašinikeš (Carruba 1970:79).
The form Zabarna (e.g., Kt a/k 1263 obv. 7) is also attested and may be an alternative writing of the same name, although the phonetic reasons for this rendering remain elusive.
The name is attested, for instance, in the Hieroglyphic Luwian inscription KULULU 4, § 5, and in the Assyrian annals, e.g., those by Ashurnasirpal II (Lubarna, man of Patina, in RIMA 2 A.0.101.1 iii 81–82).
Rieken 2016b; see Giusfredi and Matessi 2021 for a discussion.
Carruba 1970:21–23, for the context in which it occurs.
Cf., e.g., Schwemer 2005–2006, Dardano 2012, and Dardano 2018.
Cf., e.g., kaparta-/kapirta- ‘rodent’, usually traced back to a PIE compound but recently regarded as a Semitic loanword by Kroonen (2016), < Proto-Semitic *ʕakbar-t-, fem. of *ʕakbar- ‘jerboa’ (cf. the Akkadian akbaru; fem. personal name Akbartu) through Hattian intermediation to explain the loss of expected initial h- (see Giusfredi and Pisaniello 2020:223–224, with references). However, a loanword from the Akkadian akbartu with metathesis in the first syllable (or possibly aphaeresis and anaptyxis) may be a more economical solution.
Cf. Weeden 2011:175–176.
This is not unusual because some scribal mistakes seem to indicate that Sumerograms were sometimes dictated in Sumerian (cf. Cotticelli-Kurras and Pisaniello 2021, with references).
Hittite phonetic complements after Akkadograms are infrequent but not unknown (cf. Weeden 2011:10–13).
With dupl. KBo 40.20+ l.c. 3′ (NINDAtap-pé-e-nu-uš).
Cf. CAD T:182–183.
See G. Torri (ed.), hethiter.net/: CTH 415 (INTR 2012-07-30).
See also Giusfredi and Pisaniello 2020:223.
Cf. HW2 III:752.
Other cases are more problematic, e.g., the Hittite našarta- ‘concubine’, allegedly < the Akkadian esertu, and the Hittite nakappi-/nekappi- (a kind of bowl), allegedly < the Akkadian kappu ‘bowl’, because the nasal prefix cannot be easily traced back to a Hurrian morpheme, although Hurrian intermediation is often invoked for these forms (see the discussion in Giusfredi and Pisaniello 2020:224–225, 228 fn. 73, with references).
For a comprehensive discussion on Akkadian loanwords in Luwian, see Chapter 11.
Invoking Hurrian intermediation in order to explain the metathesis is not really necessary since it could be explained by an association through folk etymology between the āšipu and the verb epēšu ‘do, work, perform incantation, etc.’ (cf. Otten 1974–1977:178). Note, however, that LÚapiši- and regular LÚĀŠIPU occur in the same text, which possibly makes a direct borrowing of LÚapiši- from Akkadian unlikely (cf. Schwemer 2005–2006:226). For a similar metathesis, cf. the Hittite gurzip(p)ant- ‘wearing a hauberk’, from an unattested noun *gurzipi-, related to the Akkadian gurpisu, gursipu (also occurring as an Akkadogram GUR-ZI-IP, KUR-PÍ-ŠI in Hittite), which is generally regarded as a foreign word, possibly from Hurrian (cf. Richter 2012:228–229).
Only occurring as a stem form, which also permits it to be interpreted as an Akkadogram (É A-BU-US-SÍ).
Cf. Pisaniello 2017.
A preform *baynum should be reconstructed, attested as baynu at Ebla (see Kogan and Krebernik 2020:104). Because a single occurrence of paini is found in the Akkadian of Nuzi, probably to be explained as the code-switching of a Hurrian native speaker (see Kogan and Krebernik 2020:137), Hurrian may be the source of the Hittite word (see also Richter 2012:286).
A list of Akkadian loanwords in Hurrian can be found in Neu 1997.
For all of the references and further discussion on individual examples, see Giusfredi and Pisaniello 2020:226–228.
See Yakubovich 2006:44–45, contra Dardano 2018:355, with references.
Other Hurrian words in Akkadian, also occurring in Hittite as loanwords, include huprušhu (a vessel), huburnu (a small container), and namallu/namullu ‘plank-bed’ (cf. the gloss na-ma-al-lum = er-šu SUKI “bed in Subarean” in CT 18 4 ii 27). The Akkadian utuplu (a fabric or weaving) may also be a loanword from the alleged Hurrian source of the Hittite adupli(t)- (a kind of festive garb), borrowed via Luwian intermediation.
On Hurrian loanwords in Akkadian, cf. Giusfredi and Pisaniello 2020:228–230.
Hittite forms are hūškiwanteš ‘lingering’ (KBo 1.11 rev.! 14), lahnit ‘with lahni-’ (rev.! 15), šehuwen ‘we soiled?’ (rev.! 15), kurziwanieš ‘helmet wearing’ (rev.! 15), kulēššar ‘hesitation’ (rev.! 17). Possibly Akkadianized Hittite forms include the verbal forms taštanazzukā (obv.! 14′) and lištazzukū (obv.! 20′), which have been suggested to be built on the Hittite verbal stem šeške- ‘rest’, kula’ūtam (rev.! 18), an abstract probably reflecting the Hittite kulēššar, and eddūtam (rev.! 31), perhaps meaning ‘ration’, tentatively explained as an abstract noun built on the Hittite verb ed- ‘eat’. On these words, see the commentary to the Uršu text in Beckman 1995, with references, and Wilhelmi 2022:350–351.
Cf. CAD A/2:253.
See Singer 2008. See also Giusfredi and Pisaniello 2019:28–34 for a discussion on this noun and further references.
Cf. DCL s.v.
Cf. Dardano 2018:358 for other examples.
Cf. Dardano 2018:357–358. Other possible calques are generally believed to be coincidences, e.g., the Hittite pattarpalhi- (a bird), lit. ‘wide-winged’, formally matching the Akkadian kappurapšu, which however seems to designate a different kind of bird (also note that such a designation is quite trivial and actually attested elsewhere, cf., e.g., the Greek
Other possible examples are listed in Dardano 2018:358.
Contra Rieken (1999:313–314), even if the Hittite word could be etymologically explained as a dvandva compound, ‘das Reine und Heilige’ (wašhar being possibly related to the Luwian wašha-), a calque cannot be excluded because, as Rieken herself notes, wašhar was probably also used as a plant name.
See Dardano 2018:357 for other cases.
Cf. CAD E:343.
See also Chapter 11.
The distinction between the different Luwian varieties may have been a diatopic one, but one should note that the Luwian corpora also differed in their textual genres: Hattuša Luwian probably reflected the Luwian language spoken by Luwian speakers at the Hittite court, whereas Kizzuwatna/Lower Land and Tauriša Luwian, as far as we can tell, were used only for recitations in magical rituals. The linguistically conservative Ištanuwa Luwian was used for songs. Therefore, the possibility exists that a diaphasic rather than a diatopic variation is involved, to be understood as variation of the subcode.
E.g., perhaps, the Luwian neuter noun mūranza, the name of the ritual ascribed to the augur Maddunani from Arzawa (KUB 7.54 i 4). See also Chapter 11, in which it is suggested that the label ‘Luwic’ may be more appropriate than ‘Luwian’ for the Luwian of Arzawa. The second-millennium hieroglyphic inscriptions from western Anatolia—which, as suggested by Oreshko (2013), could belong to a different scribal tradition (but see Müller-Karpe 2019)—might be related to the Luwian of Arzawa, but the linguistic material that they provide is extremely limited.
Cf. also Melchert 2013:168–170.
See Melchert 2003a:198. For a full list of Hurrian loanwords in Cuneiform Luwian, see Simon 2020b.
Cf., e.g., Carruba 1967, Giorgieri 2012, and Pisaniello 2017.
Cf. Giusfredi and Pisaniello 2020:218–219.
However, according to Melchert (2013:161), “[t]here is no basis for regarding the ritual of Zuwi as ‘Luvian’.”
Note that a Hittite word, DUGharši- ‘pithos’, also occurs in a Luwian passage that is included in the festival of Ištanuwa (KBo 4.11 obv. 29), but the textual material we know from the tradition of this Luwian center is too scarce for a full assessment of the phenomena of interference between this Luwian variety and Hittite.
See the discussion in Yakubovich 2010:26–38.
See DCL s.v.
Some examples can perhaps be found in MS texts. For an overview of the uses of the Glossenkeil in Hittite, see Melchert 2005, van den Hout 2007, and Pisaniello 2020b, with further references.
This can be seen in the comparison between the two manuscripts of Muwattalli II’s prayer to the assembly of gods (CTH 381), the draft KUB 6.46 and the edited version KUB 6.45. The marked Luwian noun :hūwayalli- ‘runner’ (or perhaps ‘helper’), occurring in the draft as an epithet of the Sun god (iv 53), is replaced by the Hittite kutrui ‘witness’ in the edited version (iv 56), while the marked :ziladiya ‘in the future’ in the draft (iv 23) is simply repeated in the final version without the Glossenkeil (iii 54), probably pointing to its status as a true loanword in Hittite (see Yakubovich 2010:378–379, with further examples in the following pages).
Many thanks to H. Craig Melchert for pointing out this example.
See eDiAna s.v. /mugissar/, mu-ki-SUPER+ra/i ‘invocation ritual’ (
Original hypothesis of a calque by P. Goedegebuure (quoted in DCL s.v. tiššā(i)-); analysis by H.C. Melchert (personal communication).
See DCL s.v. mūwa-2.
See DCL s.v.
Other cases are more uncertain. For example, the common gender Luwian noun hupalla/i- ‘scalp?’ was oddly adapted in Hittite as a neuter s-stem hupallaš-, although a mistake in the textual tradition cannot be excluded (see the discussion in Giusfredi and Pisaniello 2020:214, with references).
See also DCL, § 2.3.2.
The identification of an adapted form can only be based on the presence of a unique Hittite ending. This is generally unproblematic, since Luwian and Hittite verbal endings rarely overlap. However, some ambiguous cases are found. For example, the imperative form :ma-am-ma-an-na-at-tén, which occurs twice in KUB 24.12+, can also be read as :ma-am-ma-an-na-at-tanx, i.e., a fully Luwian form (cf. Sasseville 2020:514, DCL s.v.).
It seems that the Glossenkeil is sometimes the only criterion to assign a word to the Luwian language. For example, the form :ša-am-la-ya-ya-ši (meaning unknown), only occurring in KUB 14.24 (where the Hittite infinitive [(:)ša-am-la-ya-]ya-u-wa-an-zi should perhaps also be restored), is regarded as Luwian only because of the Glossenkeil (the word is included in DCL), although its meaning and thus etymology are unclear, and the pres.2sg. ending -ši is shared by Hittite and Luwian. Similarly, the explanation of the Hittite hu(wa)rai- ‘sprinkle’ as deriving from the Luwian verb hur- ‘give liquid’ (cf. DCL s.v., based on HED H:397–398), attested in the pret.3sg. hu-ur-ta (KBo 8.17:5′) and the derivative huramman- ‘watered pasture’ (KUB 26.43+ i 12), probably is based only on the glossed form :hu-u-wa-ra-an-zi occurring in KUB 6.24 obv. 6′. Also, there is no ground to suggest a Luwian origin for the Hittite uranai- ‘bring a fire-offering’, despite a couple of occurrences with the Glossenkeil (cf. EDHIL:926; HEG U:93–94, DCL s.v.), or for the Hittite aršulai- ‘please, satisfy’ (?), which occurs only once in KUB 16.32+ ii 26, marked by the Glossenkeil.
Cf. Sasseville 2020:264.
Note that some other Hittite verbs previously regarded as Hittitized forms of originally Luwian verbs are explained as cognates—for example, the Luwian (:)tahušiya-/Hittite tuhušiyai- ‘keep silent’ (Sasseville 2020:31–32, with references, followed by DCL s.v.) or the Luwian (:)kiša-/Hittite kišai-, independently derived from the same unattested nominal base (cf. Sasseville 2020:26–27).
Cf. CHD Š:202–203.
Conversely, whether parā šippanzi in KUB 51.15 rev. 3′ relates to to šippai- ‘peel’ is not entirely assured (cf. the discussion in CHD P:114–115).
See EDHIL:299–300, including the discussion of previously proposed etymologies.
See CHD Š:493–494 for the occurrences and meaning.
Cf. Pisaniello 2015b.
The Luwian form :pa-aš-ta-ar-nu-wa-at-ta occurs in KUB 5.24 ii 47 in a Hittite context.
Cf. Sasseville 2020:468–469.
1sg.pres. pu-uš-š[a]-a-mi (KUB 33.120+ ii 45); 3sg.pres. pu-uš-š[a]-iz-zi (KUB 29.7 + KBo 21.41 rev. 24); :pu-uš-ša-iz-zi (KUB 36.25 iv 10); 3sg.pret. pu-uš-ša-a-it (KUB 29.7 + KBo 21.41 rev. 30, 33); 3sg.imp. pu-uš-ša-id-du (KUB 29.7 + KBo 21.41 rev. 31; KUB 33.93+ iii 21′); uncertain pu-uš-ša-a-a[n- …] (KBo 21.15:4′); and pu-uš-ša-a-[…] (KBo 34.64:3′).
However, the Hittite warrišša-, despite the hi-conjugation, may possibly be regarded as an earlier Luwian loanword because synchronically it does not function as the imperfective stem of the rare verb warrai- (see the discussion in Pisaniello 2020a:247–252).
Only ha-pal-la-ša-iz-zi in KBo 6.4 i 22.
1sg.pres. ti-iš-ša-a-mi (KUB 56.1 iii 18; KUB 31.58(+) iii 7′), ti-iš-ša-mi (KUB 31.63+ iii 20′); 3pl.pres. te9-eš-ša-an-zi (KUB 22.51 obv. 12′, 13′; KUB 50.108:10′); 3sg.pret. ti-iš-ša-a-it (KBo 32.14 ii 43); 2sg.imp. :ti-iš-ša-a-i (KUB 36.12 iii 14′); ptc.nom.sg. te-eš15-ša-za (KBo 30.164 iii 6′); ptc.nom.-acc.sg.n. ti-iš-ša-a-an (KUB 21.38 obv. 59′), ti-iš-ša-an (KUB 33.102+ ii 15), te9-eš-ša-an (KUB 33.98+ ii 13); 2sg.pres.impf. :te9-eš-ša-eš-ke-[š]i (KUB 23.1+ iv 20); 1sg.pret.impf. :te9-eš-ša-iš[-ke-nu-un] (KBo 12.38(+) iii 2′); 2sg.imp.impf. :te9-eš-ša-eš-ke (KUB 23.1+ iv 22). See DCL, s.v., for the meaning.
Cf. HEG T/D:365–366 for an overview of the various hypotheses.
See Sasseville 2020:271 for attested forms.
Denominative verbs in -ai- are widely attested in Hittite (cf. GrHL:176–177).
Despite Sasseville’s (2020:82–83) reservations, a Hurrian origin is most likely, as shown by Giorgieri (2012).
The unambiguous Hittite cognate innara- is found.
the Hittite appala- ‘trap’ is attested as the base noun. It is regarded as Luwian in origin by Starke (1990:319–321) and DCL, s.v., because of initial ā- < *h1ép- (vs. the Hittite ē-), which, however, only occurs in […]⸢a-ap⸣-pa-li in KUB 36.106(+) obv. 8′ and in the derived verbal form a-ap-pa-la-a-u-e-ni in KBo 16.50 obv. 14 (MS, CTH 270). In all the other occurrences, the noun appala-, the derived verb appalā(i)-, and also the possible derived noun appaliyalla- do not display the scriptio plena.
According to Sasseville (2020:150), followed by DCL s.v., the noun arrahhaniya- (also attested with Hittite endings) is an action noun built on the verb.
Only the Hittite hi-conjugation pres3sg. aš-ta-ni-ya-i, for which a Luwian origin is generally assumed (cf. Melchert 2005:454–455).
The assumption that the Luwian noun ašta- ‘spell, charm’ (also occurring with the Hittite genitive ending) is the base was challenged by Sasseville (2020:412), who posited an unattested base noun *aštan-.
But the denominative verb xba(i)- is attested in Lycian.
the Hittite hapa- ‘river’ is a cognate and may be the base of the verb hapā(i)- ‘irrigate’, which is regarded as a Luwian loanword purely on semantic grounds (cf. Starke 1990:514 fn. 1896).
According to Sasseville (2020:102–103, with references), the Luwian root noun hapa/i- ‘river’ cannot be the direct base of the denominative verb. Instead, a collective formation in *-eh2- should be posited as the base.
Only attested by the Luwian participle huntariyammaza (with the Hittite ablative ending), assuming that this participle is actually related to huntariyai-.
The base is huwantar (< huwant- ‘wind’). However, it is only attested in Luwian with the meaning ‘sheer cloth’ (with determinative GADA), which does not fit the meaning of the verb semantically.
The alleged noun base *kušal- ‘curry-comb’ would also be indirectly attested as the base of the noun kūšattar and the verb kūšai- (cf. DCL s.v.).
According to Sasseville (2020:27–28, 89), two different denominative verbs may have been derived from the Luwian noun lila-, lila-(ti) and lila(i)-(di); forms with Hittite endings seem to derive from both.
Note, however, that, according to Sasseville (2020:222), the verb may be derived instead from an unattested root verb *mal- (although a denominative origin is not excluded).
According to Melchert (2014:210), the Luwian origin for the noun patalha- is suggested by the existence of the Luwian verb patalhai- (with an awkward cluster /lx/ where an assimilation to /ll/ would be expected).
A nominal base in -alla/i- should be posited. However, Sasseville’s (2020:140) claim that the Hittite verb putalliya- ‘must’ be a loanword from Luwian because the geminate /ll/ reflects the Luwian suffix is unsupported. A Hittite verb can be based on a Hittite noun borrowed from Luwian, and the Luwian suffix -alla/i- also spread to native Hittite bases (cf. Melchert 2005:455–456 for the details). The existence of the Luwian verb is a stronger argument, but the Luwian status of the ultimate base should be proven on etymological grounds.
Only participles.
Only in Hieroglyphic Luwian.
Only in Hieroglyphic Luwian (see above for an account of the tapar-related stems).
The verb must be a denominative from *tarkumman- (Starke 1990:261) or *tarkumma/i- (Sasseville 2020:230).
However, if coming from the PIE root *(s)k’em- ‘to be disgraced’ (Nikolaev 2019:191–192; cf. Eng. shame), Luwian status is assured.
A denominative origin of the verb is suggested in EDHIL:1030 and HEG W-Z:652, but the attested Luwian noun zammurai- is regarded by Melchert (DCL s.v.) as an action noun built on a verbal stem. Melchert adds that “despite Glossenkeil no assurance that stem is genuine Luvian and not Hittite creation.”
On these adverbs, see Rieken 2019.
Cf. Yakubovich 2010:378–379.
For a thorough discussion, see Pisaniello and Giusfredi (2021) and Pisaniello and Giusfredi (forthcoming).