CC BY
37
I. Yakubovich
anaptyxis in hitt. *SPAND- 'to libate':
One more Case of luvian influence on new hittite
It is well-known that the syllabic cuneiform writing is not fit for the unambiguous representation of word-initial consonant clusters.1 The phonological sequence /C1C2-/ can only be represented as either C1V-C2V- or VC1-C2V- in syllabic orthography, and in neither of the two cases can one be a priori sure whether the vocalic component of the first syllabic sign is phonetically real. In order to approach this problem one has to rely on additional clues, such as the analysis of spelling alternations. By contrast, there are certain spellings, such as CjV-V-C2V- or C}V-VC2-C2V- which rule out word-initial consonant clusters in the respective lexemes. In this paper I intend to discuss one of such spellings as evidence for a contact-induced phonological change that affected New Hittite.
Eight years ago, Alexei Kassian and I published a research paper devoted to the phonetic interpretation of word-initial consonant clusters in Hittite (Kassian and Yakubovich 2002). We argued there, among other things, for the purely graphic character of vocalic prothesis and anaptyxis in etymological *sC- clusters.2 This conclusion, however, is implicitly rejected in the latest descriptive grammar of the Hittite language with the following argumentation (Hoffner and Melchert 2008, § 1.54):
1 During the preparation of this paper, I was able to consult the card-files and preliminary materials of the Chicago Hittite Dictionary, which was supported in part by the National Endowment for the Humanities. I gratefully acknowledge the logistic help of Theo van dem Hout and the staff of the Chicago Hittite Dictionary. This paper also benefited from substance comments by H. Craig Melchert (UCLA) and the editorial assistance of Demetria Nanos (Chicago). I take, however, the sole responsibility for its possible shortcomings.
Here and below, C is used as a cover symbol for obstruents. It is assumed that resonants have a different cover symbol R, but there was no need to use it in this paper.
The consistent spellings is-tV and is-kV without alternate spellings *sV-tV and *sV-kV, for words with etymological initial sequences *st- and *sk- argues that these contain a real "prothetic" vowel, as in foreign words beginning with sequences loaned into Turkish istasyon 'station'... The situation for *sp- is more complicated. It is likely that consistent is-pa-an-t° 'night' also shows a real prothetic i-, while sa/se/si-pi-ik-ku-us-ta 'pin' has preserved /sp-/... For the special case of si(-ip)-pa-an-t° 'to libate' (also with consistent i-vocalism) beside is-pa-an-t°, see Forssman (1994).
In my opinion, the consistency of orthographic prothesis in etymological *sC- clusters does not necessarily speak for its phonetically real character in each and every case. Given that all the Hittite texts available to us represent products of professional scribes, one need not be startled at their ability to faithfully memorize a simple orthographic rule. By contrast, one expects to find variations between is-CV and sV-CV spellings in the instance of those Hittite personal names beginning with *sC- that are preserved in foreign transmission. Indeed, the Old Assyrian sources of the Colony period preserve differences between orthographic renderings of Hittite names, such as Su-pu-da-ah-su (o/k 53: 3) vs. Is-pu-da-ah-su (Laroche 1966, # 473.1), Su-pu-na-ah-su (Laroche 1966, # 1190) vs. Is-pu-na-ah-su (Matous and Matousova 1984: 34 [8 obv. 14]), and Su-pu-nu-ma-an (Laroche 1966, # 1191) vs. Is-pu-nu-ma-an (Laroche 1966, # 475). The first of these names, /spudaxsu/, survived in the Old Kingdom period and is consistently written "Ispudahsu" (i.e. beginning with Is-pu-) in Hittite sources (Laroche 1966, # 473.2-3).3 Hoffner and Melchert are right, however, in the sense that words with etymological *sp- clusters represent the only case among the lexemes beginning with *sC- combinations where the absence of phonetic prothesis in early Hittite can be actually proven. The Old Assyrian tablets of the Colony period do not yield reliable examples of alternations revealing other clusters, and neither do Hittite texts, even though the latter provide combinatory evidence for the
3 Another lexeme that speaks against the phonetic character of prothesis in Hittite is the town name /smiriga/, which was consistently written with initial is-m° in cuneiform transmission (del Monte and Tischler 1978: 149). The hypothesis that that it was beginning with /sm-/ follows from the derived Iron Age Luvian adjective sa-max-ra/i-ka-wa/i-ni (URBS) "belonging to Smiriga" (Hawkins 2004: 364, following C. Melchert). Since Luvian likely dropped word-initial *s- before another consonant in its inherited lexicon (see below), this toponym must have secondarily passed into Luvian from Hittite. On initial *sm- in Hittite, cf. Hoffner and Melchert 2008: 27, fn. 41.
existence of /tsk-/ and /tsh-/ (Hoffner and Melchert 2008, § 1.127). Therefore, it is necessary to concede that Hittite word-initial spellings is-TV, is-KV, and is-HV remain amenable to different interpretations.
On the other hand, I find no justification for continuing to invoke the descriptively inadequate analysis of Forssman 1994 for the Hittite stem si(-ip)-pa-an-t° / is-pa-an-t° 'to offer, libate'. Forssman (op. cit.) attempted to refine the commonly accepted connection between the Hittite verb under discussion and Lat. spondeo 'to promise, pledge'. He offered the direct comparison between the variant si(-ip)-pa-an-t° and the archaic Latin perfect spepondi (i.e. either IE *spe-spond- > *se-spond- > Hitt. sipant-, or IE *spe-spond- > *spe-pond- > Hitt. sipant-). The variant is-pa-an-t°, according to Forssman, goes back to the zero grade of the present stem * spend- / spnd-. As argued in Kassian and Yakubovich 2002, Forssman's account can be directly refuted on philological grounds. The occurrences of the stem under discussion in the originals of the Old Hittite texts are collected in Kassian 2002: 101-102 and yield the following picture after the restorations:
Table 1
prs. 1 sg. is-pa-an-tah-hi/é 6x (Not available)
prs. 3 sg. is-pa-(a)-an-ti 8x si-pa-(a)-an-ti 27x
prs. iter. 3 sg. is-pa-an-za-as-ki-iz-zi 1x (Not available)
prs. 3 pl. is-pa-an-ta-an-zi 1x si-pa-an-ta-an-zi 7x
One can easily see that there is not a single instance of the spelling si-ip-pa-an-t° in the Old Hittite originals. The spelling si-pa-an-t°, if taken at face value, would have reflected the pronunciation /siband-/ (Hoffner and Melchert 2008, § 1.84-5). There is no way of deriving /siband-/ from earlier */sippand-/ via the standard rules of lenition, while the sheer number of attestations precludes the treatment si-pa-an-t° as scriptio defectiva. Therefore, one can accept Forssman's phonetic interpretation only at the cost of giving up the traditional etymology of is-pa-an-t° / si-pa-an-t°, which would undermine the very rationale of Forssman's reasoning. To this one must add that the historical dissimilation rules cited above are fully and explicitly arbitrary, while the postulated contrast between the historical present stem is-pa-an-t° and perfect stem si-pa-an-t° does not correspond to any synchronic semantic distinctions in our corpus (cf. Kloekhorst 2008: 405-6). In view of these facts, one can only attribute the persistent references to Forssman 1994 to the overall
high reputation of Bernhard Forssman in the field of Indo-European studies, and not to any intrinsic merits of this particular paper.4
At the same time, one must agree that the case of si(-ip)-pa-an-t° / is-pa-an-t° is truly unique, since no other Hittite lexeme shows an alternation between graphic i-prothesis and graphic i-anaptyxis in the rendering of *sC- cluster.5 Given the independent evidence for the preservation of *sp- clusters in early Hittite personal names and the etymological difficulties arising from the reading /siband-/, the prima facie hypothesis in this case is orthographic variation in the rendering of /spand-/. One must, however, explain what prompted the clustering of graphic anaptyxis in two Old Hittite finite verbal forms: 3 sg. /spandi/ and 3 pl. /spandantsi/. In addition to the rest of the verbal paradigm cited in Table 1, they can be contrasted with the nominal derivatives of /spand-/, (i)spantuzzi 'a libation vessel' and (i)spantuzziyassar 'a libation vessel'. These nominal stems are respectively attested nine times and thirteen times in the spelling is-pa- in Old Hittite original texts and never show graphic anaptyxis (Kassian 2002: 102).
A solution to this problem was found by Alexei Kassian and first presented (in a slightly different form) in Kassian and Yakubovich 2002: 34. The spelling 3 sg. pres. si-pa-(a)-an-ti, which rendered the most frequent form of the paradigm, can be interpreted as instance of graphic disambiguation prompted by the existence of the nominal
4 Forssman's account is also rejected in Jasanoff 2003: 78, fn. 39. As pointed out by Jasanoff (ibid: 3,11), the most likely Hittite match to the Indo-European contrast between perfect and present stems is wewakk-i 'he demands' vs. wek-zi 'he requests'. In this case, however, we are dealing with the regular perfect reduplication and the expected contrast between -mi and -hi conjugations. One more point that Jasanoff does not stress is the semantic difference between the two stems, which suggests that perfect reduplication might have been intensive in origin. None of these special traits find any parallel in the case of is-pa-an-t° vs. si-pa-an-t°.
5 The variation between the regular suppe/istuwara/i- and the exceptional ispestuwara- (KBo 42.64 rev. 2) 'animal representation or icon (usually of metal)' offers a remote formal parallel to the graphic phenomenon under discussion. The latter form, however, emerges as secondary on both philological and etymological grounds (the compound in question contains suppi- 'pure' as its first component, which underwent syncope in one New Script text). Therefore this lexeme is not relevant for the discussion of word-initial consonant clusters in Old Hittite.
form is-pa-an-ti 'in the night'.6 From there on, forms with anaptyxis began to spread across the paradigm of /spand-/ 'to libate'. The Old Hittite originals feature only 3 pl. si-pa-an-ta-an-zi 'they libate' but in Middle Script texts we encounter 1 sg. pres. si-pa-an-tah-hi (KUB 30.48 17) and 3 sg. pres. iter. si-pa-an-za-ki-iz-zi (KUB 28.9 i 3), which provide counterparts to Old Hittite forms with graphic prothesis listed in Table 1. Additional paradigmatic forms of the same verb attested in Middle Script are 1 sg. pret. si-pa-an-tah-hu-un (KBo 15.10 ii 10), 3 sg. pret. si-pa-an-ta-as (KBo 15.10 iii 59), 3 pl. pret. si-pa-an-te-er (KBo 15.10 ii 31), 3 pl. pret. med. si-pa-an-da-an-da-at (KBo 12.62 13), infin. si-pa-an-du-wa-an-zi (KBo 13.165 ii 11), part. nom. sg. si-pa-an-ta-an-za (KUB 14.1 rev. 78), 3 pl. pres. iter. si-pa-an-za-kan-zi (KBo 23.27 ii 40), 3 pl. imp. iter. si-pa-an-za-kan-du (KUB 31.88 iii 6). By contrast, forms with graphic prothesis (is-pa-°) never spread beyond those slots in the paradigm where they are attested in Old Hittite, and even there became overwhelmingly rare in later texts. It is fair to say that the generalization of graphic anaptyxis in the paradigm of /spand-/ is completed by the Middle Hittite stage. This orthographic change, however, did not involve the nominal derivatives (i)spantuzzi and (i)spantuzziyassar, which were not perceived as synchronic members of the verbal paradigm and therefore preserved their graphic prothesis throughout the history of Hittite.
A possible reason why the hypothesis of Kassian and Yakubovich 2002 has thus far received little scholarly notice is that we offered no explanation for the graphic variant si-ip-pa-an-t°. It would be entirely gratuitous to claim that the variant si-pa- / si-ip-pa-is limited to the orthography, since Hittite is generally consistent in preserving the distinctions between single and geminate stops, while
6 A case where is-pa-an-ti 'in the night' can be potentially misinterpreted as a verbal form is available e.g. in in KBo 31.8+ iv 14-16 man LUNAR INA E dInar is-pa-an-ti NINDAharsaus parsiya (Dardano 2006: 28). Note, however, that orthographic disambiguation of homonymous word-forms does not always serve utilitarian purposes. For example, it is difficult to imagine a French context where one could confuse a 'has' and a 'at', or the partitive article du and the participle du. In my opinion, the preference for the alternative spelling si-pa-an-t° 'to libate' may ultimately reflect the growing linguistic awareness of Hittite scribes, just as the introduction of diacritics in eighteenth-century France was a conscientious reform undertaken in the spirit of Cartesian rationalism.
the spelling si-pa- is clearly a more economic way of rendering the word-initial consonant cluster. Accordingly, one must accept that the variant si-ip-pa-an-t° reflects real phonetic anaptyxis. Chronological considerations suggest, however, that this variant had nothing to do with the regular phonetic treatment of *sp- in Old and Middle Hittite. I propose that it was rather connected with the imperfect learning of Hittite by Luvian native speakers, which was particularly characteristic of the Empire period.
The first instances of the spelling si-ip-pa-an-t° go back to the early 14th century BC. It is easy to see from Table 2 below that all the "Middle Script" attestations of this orthography are limited to the forms of third singular present and cluster in two groups of texts. On the one hand, five instances of the form si-ip-pa-an-ti are grouped in one column of a tablet containing the script of a little-studied festival for the Moon-god and the Storm-god (CTH 630). The same tablet has another innovative feature, namely the development titha- > tetha- 'to thunder', which is otherwise typical of New Hittite (Yakubovich 2008: 391). On the other hand, we have six occurrences of the same form (apparently) belonging to four separate tablets describing a festival, or a group of festivals, celebrated by the prince (CTH 647). This collection of texts represents an object of the ongoing study by Piotr Taracha. One of the intermediate results of this project published by the Polish scholar is that the bulk of the tablets united under CTH 647 introduce a provincial spring fest that was eventually incorporated into the AN.TAH.SUM festival (Taracha 2005: 714). This opens a possibility that the innovative spelling si-ip-pa-an-t° was connected in this case with regional scribal practices. Although the last conjecture is a mere guess, it remains the fact that the uniform spelling si-pa-an-t° characterized the overwhelming majority of "Middle Script" texts.
Tab
e 2
1. KBo 21.85 iv 12, 14, 19, 26, 27 si-ip-pa-an-ti CTH 630.A Also si-pa-an-t° (e.g. i 26)
2. KUB 20.88 vi? 6 [si]-ip-pa-an-ti CTH 647.II.1a
3. KUB 20.88 vi? 17 si-ip-pa-an-ti CTH 647.II.1a
4. KUB 34.128 obv. 12 si-ip-pa-an-ti CTH 647.II.2d
5. KBo 46.147 6 si-i[p-pa-an-ti] CTH 647
6. KBo 34.199 2 si-ip-pa-an-ti CTH 647 Also si-pa-an-t° (l. 8)
7. KBo 40.325 5 si-ip-p[a-an-ti] CTH 647 Same tablet as 6.
The spelling si-ip-pa- became more widespread in the Empire period (~1350-1200 BC), when it is attested in many forms of the paradigm, even though the variant si-pa-an-t° still remained quite common. The measurements of the volume of the relevant card-files in the CHD catalog suggest that the first orthography is about ten times less frequent than the second one in the overall corpus of Hittite texts. To a large extent, however, this is the artifact of how the verb 'to libate' is distributed across time and textual genres. For example, although oracular inquiries were actively documented in the Hittite Empire and contain frequent references to libations, the scribes in charge of keeping divination records would normally use the Sumerogram BAL as shorthand writing for the verb 'to libate'. As for the scripts of festivals and rituals, where this verb naturally occurs most frequently, they have been shown to develop during the Empire period primarily through the compilation and adaptation of the existing texts. Therefore there is always a strong possibility that si-pa-an-t° occurring in a New Script religious composition reflects historical orthography.
At the same time, there are a sufficient number of demonstrable cases for the double spelling si-ip-pa-an-t° introduced into the late copies and adaptations of Old and Middle Hittite texts in spite of the pressure of orthographic tradition. Thus, all the extant Middle Script manuscripts of the Mastigga ritual against the domestic quarrel (CTH 404.1) display the single spelling, and so do all the New Hittite copies of the same composition except one. The manuscript II.C2, which is not particularly remarkable for other orthographic deviations, shows consistent double spelling (Miller 2004: 186-87). In a similar fashion, the Old Hittite fragments mentioning the asusala-men (CTH 665) feature the spelling si-pa-an-t°, which was innovative for this period (Starke 1985: 285-88 [ii 11, iii 1, 9], 29092 [i 14, iii 3], 292-93 [ii 4,7]). The Late New Hittite text KUB.35.133+ belonging to the same group is characterized by the consistent double spelling si-ip-pa-an-t° (Starke 1985: 278-83 [ii 26, 40, iii 19, iv 6]). In this particular case, one has to assume that the Late New Hittite ritual represents a creative adaptation of the existing tradition because the Hittite term asusala- is replaced there with its Luvian equivalent asusatalla-.
One New Hittite original composition where the verb 'to libate' occurs quite frequently is Muwatalli's prayer to the assembly of gods through the Storm-god of Lightning (CTH 381). The prayer is available in two main copies: KUB 6.46 and KUB 6.45+. Houwink ten Cate (1968) and Singer (1996: 135-142) have persuasively
argued that the first tablet represents the draft of the second one. Table 3 demonstrates that most forms of the verb 'to libate' in KUB 6.46 have two orthographic peculiarities: the double spelling si-ip-pa- and the omission of the <an> sign. The second feature is consistently eliminated in KUB 6.45+, but the first one is corrected only once. It stands to reason that the scribe of the edited version perceived the omission of the syllable- final nasal as unacceptable practice but had no firm opinion with regard to the spelling of the initial group of consonants. The loss of n before another consonant must be regarded as a stigmatized colloquial feature of Late Hittite (Yakubovich 2008: 401-405), but the variant si-ip-pa- has become stylistically acceptable by the early thirteenth century BC.
Table 3
KUB 6.46 KUB 6.45+
3 sg. pres. i 44 si-ip-pa-ti iv 7 si-pa-an-ti
3 sg. pres. i 48 si-ip-pa-ti iv 12 si-ip-pa-an-ti
3 sg. pres. i 52 si-ip-pa-ti iv 17 si-ip-pa-an-ti
3 sg. pres. i 56 si-ip-pa-ti iv 22 si-ip-pa-an-ti
3 sg. pres. i 60 si-ip-pa-ti iv 27 si-ip-pa-an-ti
3 sg. pres. i 64 si-ip-pa-ti iv 32 si-ip-pa-an-ti
3 pl. pres. it. iii 47 si-ip-pa-an-za-kan-zi iii 9 si-ip-pa-an-za-kan-zi
3 sg. pres. iv 50 (no correspondence) iv 51 BAL-ti
Another hint at the acceptability of the New Hittite spelling si-ip-pa-an-t° comes from the erroneous writing KBo 22.135 rev. 13 si-<<pa>>-ip-pa-an-t[i] (Rüster 1988: 305). Although the tablet where this form occurs shows the New Hittite ductus, it contains a part of the ritual script for the libation to the throne of Hebat (CTH 701), a text that was recorded in Hittite in the early fourteenth century BC. Consequently, we expect that the scribe of KBo 22.135 was copying from the original that contained the form si-pa-an-ti. He began to faithfully reproduce the archaic formation, but then had second thoughts and replaced it with the more modern double spelling, forgetting, however, to delete the <pa> sign. Such a conscientious replacement is conceivable only if he regarded si-ip-pa-an-ti as the standard spelling for his time.
Summing up, the orthographic development si-pa-an-t° > si-ip-pa-an-t° is barely seen in Middle Script manuscripts, where the form si-ip-pa-an-t° is restricted to two compositions, but can be easily deduced from the analysis of New Script tablets. Accordingly, one has to conclude that the phonetic development /spand-/ > /sippand/
was dialectal in the Early New Kingdom period (~1400-1350 BC), but became common among the scribes of the Hittite Empire (~1350-1200 BC). Both the start and the proliferation of this change belong to the period when Hittite was rapidly evolving under the impact of the related Luvian language. The contact-driven morphosyntactic innovations that transformed the outlook of Hittite in the last two centuries of its written attestation have been catalogued in Rieken 2006. They include syncretisms in the nominal case system (e.g. the merger between nominative and accusative plural), restructuring of the relative order of pronominal clitics, secondary use of the reflexive particle as a linker in the nominal sentence, generalization of the verbal suffix -iya- at the expense of the allomorph -iye-, and a number of other changes. No evidence has, however, been presented for the restructuring of the Hittite phonotactic system under the influence of its Luvian counterpart. One possible reason for this disparity is the phonotactic similarity between the closely related Hittite and Luvian languages, while the other one is the difficulty of studying phonotactic distinctions among languages recorded in syllabic orthography.
There is, however, one clear distinction between Hittite and Luvian historical phonologies that has direct relevance for the phonotactics of the two languages. The correspondence between Hitt. istamana- 'ear' and Luv. tummant- 'ear' speaks for the initial cluster simplification *st > t in Luvian (Melchert 1994: 271). The absence of either direct or indirect evidence for word-initial /sC-/ clusters in Luvian suggests that s- would disappear before any consonant (cf. Melchert 1994: 267). But if Luvian had a highly-ranked constraint on word-initial /sC-/ clusters, one would expect that Luvian native speakers learning Hittite as a second language would fail to faithfully reproduce Hitt. /spand-/, just as they would fail, for example, to implement the distinction between the nominative and accusative plural endings that was absent in their own dialect (Yakubovich 2008: 32ff.). From the purely phonetic point of view, the use of anaptyxis for breaking word-initial /sC-/ clusters finds a good parallel in the history of Persian. One can compare, for example, contemporary Tehrani Persian examples, such as sefid 'white' going back to Middle Persian sped 'id.', or setare 'star' cognate with English star.7
7 The irregular substitution p > f in sefid proves that Arabic native speakers modified the pronunciation of this word as a result of the imperfect learning
What sets aside the contact-induced anaptyxis in si-ip-pa-an-t° from the Persian parallel is the lack of other consistent examples of New Hittite i-anaptyxis in etymological *sC- clusters (cf. Forssman 1994: 99). If Luvian native speakers eliminated the cluster in this particular lexeme, one would presume that they implemented the same strategy in other words with similar phonotactics. Orthographic evidence suggests, however that this was usually done via prothesis rather than anaptyxis. This discrepancy is illustrated by the following table, where is-pa-an-f 'night' stands for the usual treatment of lexemes beginning with *sC-. As the prothesis in the verbal stem must have been purely graphic in Old Hittite, so must also have been the prothesis in the homonymous nominal stem. But the same prothesis in the word for 'night' was surely phonetic for those Luvian native speakers who were learning Hittite as the second language, and perhaps in New Hittite in general.
Table 4
Old Hittite Middle Hittite New Hittite
'libate, offer' is-pa-an-f si-pa-an-f si-pa-an-f si-ip-pa-an-f si-ip-pa-an-f (si-pa-an-f)
'night' is-pa-an-f is-pa-an-f is-pa-an-f
Is it possible that the original graphic distinction of Old Hittite eventually generated the phonetic difference between the verbal stem /sipand-/ and the nominal stem /ispand-/ in New Hittite? The answer is yes if we assume that the study of written texts was an important component of the acquisition of Hittite in 14-13th centuries BC. As a parallel, one can cite the graphemic mediation in the development of English lexicon. For example, the replacement of Middle English faute with New English fault was a work of Renaissance grammarians, who wished to give a more archaic appearance to this French loanword and stress its etymological connection with Lat. fallere 'to deceive'. Yet, the pronunciation of this word did not change till the 19th century, when -l- came to be pronounced under
of Persian. One can therefore hypothesize that anaptyxis in this case likewise developed in the context of language contact between Persian and Arabic, which lacks word-initial clusters altogether. It is, in fact, likely that Arabic influence is ultimately responsible for the complete elimination of the Persian consonant clusters in word-initial position, although the distribution between prothesis and anaptyxis as strategies of cluster simplification is not always clear.
the pressure of written tradition. No, doubt, the growing literacy of English speakers must have been decisive for this development.
There are, of course, no grounds to believe that the bulk of the population of the Hittite Empire or even its capital was literate in the cuneiform. It is possible, however, that as Luvian was consolidating its linguistic dominance in Hattusa in the Empire period, the number of Hittite speakers then and there gradually shrank to the point when the scribes came to constitute a sizeable part of this speech community. At the same time, the good command of Hittite remained a necessary prerequisite for being a cuneiform scribe. Therefore, the scriptorium became an important setting for the transmission of Hittite as both written and spoken language. The Luvian native speakers, who by then constituted the majority of junior scribes, would acquire the necessary Hittite skills through a combination of oral instruction and written exercises. Being unable to articulate word-initial /sp-/ clusters, they would use the existing graphic practices as hints pointing at ways of their phonetic substitution. In the case of *spand- 'night' and most other lexemes, the spelling unequivocally pointed in the direction of prothesis, but in the case of *spand- 'to libate', the cuneiform onset si-pa- was more conducive to phonetic anaptyxis. Once the sound change was implemented, the new spelling si-ip-pa-an-t° was introduced as a better match to the phonological /sipand-/.8
There is possible independent evidence that Hitt. *spand- was borrowed into the Luvic languages with anaptyxis. Hajnal (1995: 133-134 with ref.) tentatively reconstructs the Iron Age Luvian professional title (CAELUM. *286.x)sa-pa-tara/i-i-sa 'libation-priest?' (KARKAMiS A 2+3 17a) as earlier */sVpentero/i-/ on the basis of its comparison with Lyc. hppnterus, also a professional title or name of institution. Although the anaptyctic vowel is not transparent in Anatolian hieroglyphic orthography, the change s->h-
8 Another instance of phonetic anaptyxis in New Hittite that was presumably mediated through the graphic system is KUB 58.100 ii 1 sa-a-pi-ku-us-ta-as 'pin', where the plene spelling suggests that /a/ was pronounced. The other spellings of the same lexeme show the variation of the initial sign sa/se/si-pi-ik-ku-us-ta-, which allows us to reconstruct the initial /sp-/, as per Hoffner and Melchert 2008, § 1.54. The optional elimination of this cluster in New Hittite is probably due to interference with Luvian, while the Hittite spelling conventions predetermined the choice of anaptyxis, as opposed to prothesis.
in the history of Lycian did not occur in pre-consonantal position. It is, however, unlikely that the Luvian or Lycian titles are inherited from Proto-Anatolian, because the base verb is not attested anywhere in the Luvic languages, while word-initial *sC- is expected to be simplified in Luvian via the deletion of the first element. If the Luvian and Lycian forms are indeed related to the Hittite stem under discussion, one would rather expect that the hybrid title *sdpantalli-, lit. "pertaining to libation" was coined in the Hittite-Luvian bilingual environment in the Empire period, underwent rhotacism within Luvian, and was subsequently borrowed into Lycian.9 The parallel with si-ip-pa-an-t" is not precise because of different anaptyctic vowels, but this difference can be accounted for through the presence vs. absence of the guiding orthographic tradition.
Thus the analysis of Hittite *spand- 'to libate' not only provides us with one more example of contact-induced linguistic development in Bronze Age Anatolia but also helps us to specify an important mechanism for the transmission of Hittite in the Empire period. We have seen that spelling practices began to play a role in the phonological evolution of New Hittite, which bespeaks a substantial match between the Hittite speech and epigraphic communities in thirteenth-century Hattusa.
Bibliography
Dardano, Paola. 2006. Die hethitischen Tontafelkataloge aus Hattusa (CTH
276-282). StBoT 47. Wiesbaden: Harrassowitz. Forssman, Bernhard. 1994. "Zu hethitisch sipand- und ispand-". In: In honorem Holger Pedersen. B. Nielsen and J.E. Rasmussen (eds). Wiesbaden: Reichert. Pp. 93-106. Hajnal, Ivo. 1995. Der Lykische Vokalismus: Methode und Erkenntnisse der vergleichenden anatolischen Sprachwissenschaft, angewandt auf das Vokalsystem einer Kleincorpussprache. Graz: Leykam. Hawkins, J. David. 2004. "The Stag-god of the Countryside". Indo-European Perspectives: Studies in Honour of Anna Morpurgo Davies. J. H. W. Penney (ed.). Oxford University Press. Pp. 355-69.
9 In this connection, one should compare the hapax legomenon sa-pa-an-ta-al-la 'pertaining to libation (?)' which occurs in KBo 31.8+ i 7 sa-pa-an-ta-al-la-ma DUB 1.KAM HI.A anda UL handa[n] 'the libation (?) tablets are not present' (Dardano 2006: 22). The orthography does not show anaptyxis in this case (one would rather expect **sa-ap-pa-an-ta-al-la), but we are possibly dealing here with an archaizing or shorthand spelling.
Hoffner, Harry and H. Craig Mel chert. 2008. A Grammar of the Hittite Language. Winona Lake, IN: Eisenbrauns.
Houwink ten Cate, Philo. 1968. "Muwattallis' Prayer to be Spoken in an Emergency: an Essay in Textual Criticism". Journal of Near Eastern Studies 27: 204-8.
Jasanoff, Jay. 2003. Hittite and the Indo-European verb. Oxford University Press.
Kassian, Alexei 2002. "Glossary of Verbal Forms and Derivatives from Published Old Hittite Texts". Anatolian Languages. V. Shevoroshkin and P. Sidwell (eds.). Canberra: Association for the History of Language. Pp. 72-136.
Kassian, Alexei and Ilya Yakubovich. 2002 "The Reflexes of Indo-European Initial Clusters in Hittite". Anatolian Languages. V. Shevoroshkin and P. Sidwell (eds.). Canberra: Association for the History of Language. Pp. 10-48.
Kloekhorst, Alwin. 2008. Etymological Dictionary of the Hittite Inherited Lexicon. Leiden: Brill.
Laroche, Emmanuel. 1966. Les noms des hittites. Paris: C. Klinksieck
Matous, Lubor and Marie Matousova-Rajmova. 1984. Kappadokische Keilschrifttafeln mit Siegeln aus den Sammlungen der Karlsuniversität in Prag. Prague: Univerzita Karlova.
del Monte, Giuseppe F. and Tischler Johann. 1978. Die Orts- und Gewässernamen der hethitischen Texte. RGTC VI. Wiesbaden: Reichert.
Melchert, H. Craig. 1994. Anatolian Historical Phonology. Amsterdam-Atlanta: Rodopi.
Miller, Jared. 2004. Studies in the Origins, Development and Interpretation of the Kizzuwatna Rituals. StBoT 46. Wiesbaden: Harrassowitz, 2004.
Rieken, Elisabeth. 2006. "Zum hethitisch-luwischen Sprachkontact in historischer Zeit". Altorientalische Forschungen 33: 271-85.
Rüster, Christel. 1988. "Materialen zu einer Fehlertypologie der hethitischen Texte". Documentum Asiae Minoris Antiquae. Festschrift für Heinrich Otten zum 75. Geburtstag. Ed. E. Neu and Ch. Rüster. Wiesbaden: Harrassowitz. Pp. 295-306.
Singer, Itamar. 1996. Muwattalli's Prayer to the Assembly of Gods Through the Storm-God of Lightning (CTH 381). Atlanta: Scholars Press.
Starke, Frank. 1985. Die keilschrift-luwischen Texte in Umschrift. StBoT 30. Wiesbaden: Harrassowitz.
Taracha, Piotr. 2005. "Zu den hethitischen Ritualen des Königssohn". V. Uluslararasi Hititoloji Kongresi Bildirleri. Y. Hazirlayan and A. Süel (eds.). Ankara. Pp. 707-714.
Yakubovich, Ilya. 2008. Sociolinguistics of the Luvian Language. University of Chicago Ph.D. Dissertation.
