CC BY
35
A. V. Shatskov
ANOTHER VERBAL SUFFIX IN HITTITE?*
В данной статье рассматривается возможность выделения суффикса *-k- в ряде хеттских глаголов (в том числе hassik- 'насыщаться', malikk- 'ослабевать', nink- 'напиваться', hark- 'исчезать, гибнуть') на основании их сопоставления с глаголами других анатолийских и индоевропейских языков. Для большинства из этих глаголов этот формант обычно считается частью корня.
Ключевые слова: этимология, морфология, глагол, суффикс, сравнительно-историческое индоевропейское языкознание, хеттский язык.
The Hittite verbal system differs significantly from the system reconstructed for Proto-Indo-European (further PIE). The reconstruction of the latter is based mainly on Greek and Indo-Iranian data. One of the major differences is that Hittite (and Anatolian) verb does not exhibit any variation of the stem, so characteristic of a PIE verb. That is, a verb in late PIE had finite forms derived from one of the three stems: Present, Aorist or Perfect, rather than immediately from the root. As for Hittite, there is a specific imperfective stem formed with the suffixes -ske/a-, -anna/i- or -ssa-, but it is rather an Anatolian innovation. Thus, there is only one stem for all the tense forms.
The lack of aspectual stem differentiation in Anatolian is even more puzzling as all the tools used in the formation of PIE present and perfect stems are present in Hittite, be that verbal suffixes -ske/a-, -ye/a-, -nu-, -ahh- etc., or infixation and reduplication. This article presents evidence for one more Hittite verbal suffix -kk-, which previously has been undetected and may be related to Gr. -к- of the perfect stem.
Let us start with two verbs, hassik- 'to satiate oneself' and malikk- 'to become weak, weaken', where an element -ikk- can be easily distinguished.
The verb malik(k)- 'to weaken, become weak', which is attested once in KUB 30.10 Rs. 4 ma-li-ik-rku'-un1, is related to malisku-
* Исследование проведено при поддержке гранта РНФ № 14-18-03585 «Праиндоевропейский корень и основообразующие форманты»).
1 It can be equally read as ma-le-ek-rku1-un.
'soft'. Puhvel (HED M : 30) compares these words to Gr. ^a^aKO^ 'soft', p^a^ 'weak, soft', stupid', Goth. -malsks in untilamalsks 'rash, reckless' etc. and reconstructs PIE *mleh2k-2. This is problematic since the /k/ of the root should not have been dropped before the -ske/a- suffix, cf. wekiske/a-, imperfective stem of wekk- 'to ask for, wish' or salikiske/a-, imperfective stem of salik- 'to touch, to invade'. Rieken (2013: 275f.) connects malisku- directly to Goth. -malsks and the stem *molh2-sko- with the subsequent loss of the laryngeal due to Saussure's effect. Her explanation is attractive, but it leaves the final -e/ik- of malik(kk)- unexplained. One is tempted to compare it with equally obscure -ik- of hassik- 'to satiate'.
The verb hassik- is attested more than a dozen times, e.g. in ha-as-si-ik-ke-er KUB 17.10 I 20 (Middle Hittite copy of an Old Hittite text) or ha-as-si-ig-gal-lu KUB 36.93 Rs. 6, though there are also some spellings with single -s-, e.g. ha-as-si-kan-du KBo 15.10+ III 38.
In Palaic there is a related verb has- 'to be satiated of drinking'. While many Palaic words and contexts are still not very clear, there is a general agreement on the meaning of the passages with has-:
KUB 32.18 Vs. 7 a-ta-a-an-ti ni-ip-pa-si mu-sa-a-an-ti a-hu-wa-an-ti ni-ip-pa-as ha-sa-a-an-ti (similarly Vs. 9, 18). 'They eat, (but) do not have enough, they drink but do not quench their thirst'.
(Carruba 1970: 8, most recently Yakubovich 2009: 177).
The Palaic cognate has- clearly shows that -ikk- in hassik- is a suffix rather than a part of the root. Already Oettinger (1979: 194) analyzed hassik- as *h2es-K-. However, further etymology is not very clear. HED H: 231 compares Pal. has- and Hitt. hassikk- to Greek aro 'to satiate, take one's fill', but Beekes 2010 : 146 attributes the Greek verb to *seh2- (LIV: 520f.), which hardly had any initial laryngeal. The root *h2es- still could be envisaged if has- and hassik- are related to Lat. inanis 'empty, hollow' < *n-h2es-ni- as suggested by Klingenschmitt (1994: 242f., s. also LIV: 273). However, a crucial point for all the etymological attempts is the geminated spelling -ss- in Hitt. hassik-. It can hardly go back to just PIE /s/, for initial /hs-/ would have been spelled ha-si-, he-si- or hi-si-, s. Hoffner, Melchert 2008: 12f.
2 Cf. a possibly related root *meld- 'to become mild, soft', for which see e.g. LIV: 431.
Since the root cannot be just *h2es-, -ss- must reflect some kind of assimilation. The assimilated sound could be a laryngeal, e.g. hassa- 'fireplace' < PIE *h2eh^-eh2-, Skt. asa- 'ashes', Lat. ara 'altar' (Kloekhorst 2008: 322). In this case we would rather expect a plene spelling **hassik-; moreover, the stem *h2e/oHs-ik- looks highly unusual. Alternatively -ss- could go back to a nasal and sibilant. The sequence *-VnsV- was assimilated to Hitt. -ss-, see Melchert 1994: 163. The fate of *-ms- is less clear, see e.g. the detailed discussion in Kimball 1999: 331f.; Melchert (1994: 164f.) suggests assimilation of *-ms- to *-ns-, while Kloekhorst 2008 : 319f. argues that *VmsV yields Hitt. -ss-, as in hassu- 'king' < *h2emsu-, while *CmsV yields Hitt. -anz- as in hanzassa-'offspring'. Note however that there is a very limited data to confirm any of these scenarios. However, a stem of the type *h2eNs-k- or *h2Ns-k- with the plain *-k- would regularly yield hassikk- with an anaptyctic -i- (see e.g. Hoffner, Melchert 2008: 43), as in e.g. hassikke- < *hann-ske/a-, imperfective stem of hanna- 'to judge'.
If there is a nasal in the root of hassik-, the root *h2em(H)-s-3 'to pour' (Klingenschmitt 1982: 118f., LIV: 265, Arm. aman 'pot, vessel', amana- 'to fill') comes to mind. The suffix *-s- is relatively rare in Hittite, but it is assumed e.g. for tamass/tamess- 'to oppress' from PIE root *demh2- and half a dozen other verbs (see esp. Kloekhorst 2008: 823). For Hittite, this could mean a stem like *h2m-s-k- that developed into *hamsk-, then to *hansk- and further to hassik-4. As for Palaic verb, one has to assume that has- reflects the zero grade *h2m s-, with the syllabic nasal yielding just -a-. Such a development is also attested in Hittite, e.g. in katta 'down' < *kmto, but its conditioning is clear neither for Hittite nor for Palaic. Still, even if the proposed root etymology is incorrect, I believe that hassik- goes back to a root of the type *h2Ns-k-'; *h2eNs-k-' with an
3 The laryngeal in the Auslaut is reconstructed in LIV because Arm. amanais believed to go back to an infixed stem, but it is in fact not required as the ana-type of conjugation was productive, s. Kocharov 2011: 272f.
4 Alternatively it could be *h2m-s-k- > *hamsk- > *hamsik- > *hassik- if we accept the rule VmsV > VssV; given the scarcity of reliable examples it is hard to say for certain. The reflex accented *e of the full grade would however be at least sometimes spelled with plene, so hassik- can hardly go back to the accented full stem *h2em-s-k-
The root-final laryngeal for this root ddoes not fit in for this scenario. The stem *h2mh2-s-k- would yield **ha(m)me/isk-, as *dmh2s-ent- > Hitt. dax-mi-es-sa-an-t-, cf. Kloekhorst 2008: 823f.
unaccented full grade is also possible, but it is hard to imagine such a stem for a PIE verb, unless the full grade was levelled.
Note, however that malikk- (can be read also as malekk-) cannot reflect a full grade of the root *mleh2k- or full grade of the suffix *mlh2ek-, since it would yield **mlakk- or **malakk- respectively, so malikk- may only reflect the zero grade *mlh2k-. While we have no secure evidence of a regular phonetic development for such a cluster, one is prompted to assume here a vocalization of the syllabic resonant and emergence of an anaptyctic vowel.
Besides hassikk- and malikk- where the suffix -kk- may be clearly distinguished, there are several other Hittite verbs with unclear etymology that can contain this morpheme.
A well-attested verb nink- 'to drink one's fill, to get drunk' still has no convincing etymology. Various attempts to connect it to Hitt. nakki- 'heavy, important' or ninink- 'to set in motion' are improbable, cf. Kloekhorst 2008: 607. The latter connection is based on the translation of ninink- as 'to soak' in KBo 8.47 Vs. 10 karittiyas nininkanta "floods will soak in" that matches Akk. milu ina nagbi illakuni "floods will soak into groundwater" (so HED N: 116), but such a translation is incorrect as the corresponding Akk. alaku means 'to go', matching well the standard meaning of ninink-'to set in motion, move'.
Melchert (1994: 165) suggested that nink- has the same element -K- as hassik-, so that the root is *nem- 'to take'; he pointed to Spanish tomar 'to take' > 'to drink' as the semantic equivalent. One can also adduce Russ. принять that can also mean 'to have a drink' aside the basic meaning 'to accept'. Kloekhorst (ibid.) has two objections to this etymology. First, he believes that proposing an extension -K- is ad hoc, but as I try to show in this article, this element is a legitimate suffix. Secondly, he argues that the root *nem- means rather 'to allot', as seen in Gr. ve^ro 'to allot, distribute', cf. also LIV: 453, where the meaning 'to take' of Germanic and Baltic verbs is explained as taken from the Middle forms. Still I believe that this etymology stands, as we sometimes see similar semantic developments in other verbs, e.g. Hitt. da- 'to take' from PIE *deh3- 'to give'.
The verb hark- 'to perish' and its derivatives harnink- and harganu- 'to destroy' are generally compared to OIr. org(a)id 'to smite' and Arm. harkanem 'to smite, smash' (LIV: 301, Kloekhorst 2008: 306f.). Puhvel (HED H: 167f.) however rejects it on the
grounds that the meaning of hark- does not match those of Old Irish and Armenian verbs. It is usually implied that Armenian and Old Irish verbs had preserved the 'causative' meaning which must initially was characteristic only to the present infixed stem. The problem here is that there are no original infixed forms from this root preserved in these languages; according to LIV: 301, Arm. harkanem is derived from the aorist stem *hark- <*h3rk-. Note also that OIr. org(a)id 'to smite' and Arm. harkanem 'to smite, smash' can in fact go back to PIE *perg-, an extended variant of the root *per- 'strike' (LIV: 473), where har(e)- < *pr-, the suppletive aorist to Arm. harkanem, also belongs to, cf. discussion in Klingenschmitt 1982 : 215f. Puhvel himself (l.c.) connects hark- to Gr. ep%o^ai 'to go' which might be acceptable semantically, but is phonetically impossible as e-colouring laryngeal had been lost in Hittite (see e.g. Melchert 1994: 65).
An alternative is to connect hark- to Toch. AB ar- 'to cease, come to an end', again assuming the suffix -k- here. The Tocharian verbs were derived from the zero grade of the root *h3er- 'to arise' by Hackstein (1998: 228f.) or from the root *h2erH- (LIV: 271f., for the final laryngeal cf. also Malzahn 2010: 527f. and Hitt harra- 'to grind, crush' if it also belongs here). Hitt. hark- is unlikely to go back to *h2erH-, as *h2rHk- would likely have yielded something like **harikk-, cf. the discussion of malikk- in the present article. Yet we may deal with the root *h2/3er- here with different suffixes/root extensions.
The verb kink- and its derivative kinganu- are scarcely attested and their meanings are not firmly established. Kink- is used several times in a text on glassmaking (KBo 8.65 Vs. 7, 11, 16, Rs. 5 and KBo 18.201 III 10) and seems to be a technical term (Riemschneider 1974: 268ff., 274). The best preserved context is that of kinganu- in KBo 19.129 Vs. 31-32: [... -]ma 1 GA.KIN.AG TUR DDJGisnuri kinganuzi '... but one small cheese at the dough-bowl he k.y (HED H: 182f.) Oettinger (1979 : 178f.) suggested a meaning 'to smother, smear'. Puhvel (HED H: 183) rather believes it to belong to metallurgical or glassmaking terminology, something similar to 'to spray' or 'to flush'. He further thinks of the meaning 'to attach, fasten' and tentatively adds Lat. cingo 'to gird, equip', Skt. kancate 'to bind' as possible cognates. However, the root *gem- 'to press, grasp' will pass better here. If so, Hitt. kink- goes back to *gem-k-, though this of course remains a mere speculation.
Finally, the verb mark- means 'to divide, allot'. It has been connected to Skt. marc- 'to damage, destroy', to Lat. mere- 'trade, goods' or to Lat. margo 'rim, border' and Goth. marka 'boundary'. The latter etymology is accepted by Kloekhorst (2008: 559) who however notes that there are no verbal forms for the root *merg- outside Anatolian. In this respect it is tempting to return to the Benveniste's connection of Hitt. mark- to Gr. це(роцш, 'receive as one's portion', Lat. mereo 'to receive as one's wage, earn, deserve', cf. HED M: 74. Semantically it is a very good match, and if so, mark- goes back to the root *(s)mer- 'to receive a share' (LIV: 570). Problematic here is the fact that already in Old Hittite texts this verbs shows both hi- and mi-conjugation forms and the form ma-a-ar-ka-ah-hi in KBo 17.3 IV 30 must reflect a Proto-Hittite *mork-hi. The hi-conjugation is not characteristic for other verbs that presumably contain the suffix -k-.
There is one more verb, salik- 'to touch, to approach, to invade' that may belong here. We find plene spellings in both the first and the second syllables, e.g. sa-li-i-ga KBo 17.18 II 17 and sa-a-li-ga KBo 17.42 7, but the latter spelling is found only in late copies. Therefore, even if forms sa-a-li-ga do reflect some later phonetic developments, for instance breaking of the initial cluster #sl-, in Old Hittite the stem was slig-. The long /i/ cannot be explained as an anaptyctic vowel or a reflex of a short accented /e/, so it may go back either to an accented /i/ or to a diphthong /ei/ (cf. Kloekhorst 2008: 712). The stem-final velar is always spelled single, so formally it does not correspond to the -kk- of hassik- and malikk-, but *-kk-would have been lenited to -k- after a long accented vowel.
The etymology of salik- remains disputed. It was compared by Melchert (1994: 330) to OIr. sligid 'to strike (down)', which according to LIV: 566 belongs to the root *sleig- 'to smear, smooth'. Semantically this etymology does not look attractive. Kronasser (1957: 125) compared it to the root *sel- 'to creep' (LIV: 528, Lith. selu). The connection to a homonymous root *sel- 'to leap, run' (LIV: 527f., Gr. а^оцш,, Lat. salio) is more plausible. If so, the Hittite verb can be analyzed as sl-ik-, with the second part being a composite suffix *-i-kk-. Such a combination is not attested elsewhere in Hittite; perhaps this is a trace of some kind of older combination of suffixes, cf. Казанский 2010.
Note that there also could be a complex suffix -ikk- in hassikk-and malikk-, but in those verbs it is unaccented and thus different from that in salik-. However, assuming the suffix in hassikk- and
malikk- to be just /k/, which is phonetically well possible, makes them structurally similar to nink- and hark-, the only significant difference being the zero grade of the root; therefore I believe this is a much more plausible analysis. Likewise, the final consonant in hark-, nink- etc. may reflect also voiced velars as well, but it is more economical to assume there was one velar suffix in (proto-)Hittite, which is preserved in half-a-dozen verbs, rather than several suffixes preserved in even less number of verbs.
Kronasser (1957) in an article on guttural and dental roots extensions adduced some more examples with an alleged velar extension, like malk- 'to spin', which he connected to TochAB malk-'to put together'. Some of these etymologies gained recognition, but unlike the verbs discussed in this article they have the root final element *-k- also in other branches, so it is generally believed to be a part of the root, cf. the root *melK- in Kloekhorst 2008: 550. As for the Hittite verbs discussed here, *-k- is likely to be added in proto-Anatolian or even proto-Hittite period, after the split of the Indo-European family and perhaps after the split of proto-Anatolian as well, as it seems to be in the case of Pal. has- and Hitt. hassik-.
This suffix was no longer productive at the time of the earliest Hittite texts, so its function is difficult to determine. Note however that at least four of the discussed verbs denote some kind of achieving a state, that is hassik- 'to satiate oneself', nink- 'to drink one's fill, to get drunk', malikk- 'to weaken, become weak' and hark-'to perish'. Mark- 'to divide, allot' on the other hand clearly denotes an action; it however has other morphological peculiarities, like hi-conjugation forms, so it may actually not belong here.
Given the scarce information we have on the suffix *-k- in Hittite, one can only tentatively suggest its counterparts in other branches. One comparison however is quite tempting - it is the Greek suffix -к- used in formation of the perfect stem in Greek, e.g. in теб^ка; its origin remains unclear, cf. e.g. Sihler 1995: 576. This is however a matter for a further research.
References
Казанский 2010 - Казанский Н. Н. Грамматика порядков и проблемы праиндоевропейской реконструкции // В пространстве языка и культуры: Звук, знак, смысл. Сб. ст. в честь 70-летия В. А. Виноградова / Ред.: В. З. Демьянков, В. Я. Порхомовский. М.: Языки славянских культур, 2010. С. 589-597. Beekes 2010 - Beekes R. S. P. Etymological Dictionary of Greek. Leiden: Brill, 2010.
Carruba 1970b - Carruba O. Das Palaische. Texte, Grammatik, Lexikon.
Wiesbaden: Harrasowitz, 1970 (= StBoT 10). Hackstein 1998 - Hackstein O. Tocharisch und Westindogermanisch: Strukturell uneinheitliche Laryngalreflexe im Tocharischen (Uridg. *-Uh1C- vs. *-Uh2/3(C)- und *#h1RC- vs. *#h2/3RC-) // Sprache un Kultur der Indogermanen. Akten der X. Fachtagung der Indogermanischen Gesellschaft. Innsbruck, 22.-28. September 1996 / Ed.: W. Meid. Innsbruck, 1998. P. 217-236. HED - Puhvel J. Hittite etymological dictionary. Berlin; New-York;
Amsterdam: Mouton, 1984-. Hoffner H., Melchert 2008 - Hoffner H., Melchert C. A Grammar of the
Hittite Language. Winona Lake: Eisenbrauns, 2008. Kimball 1999 - Kimball S. Hittite Historical Phonology. Innsbruck, 1999. Klingenschmitt 1982 - Klingenschmitt G. Das altarmenische Verbum.
Wiesbaden: Reichert, 1982. Klingenschmitt 1994 - Klingenschmitt G. Die Verwandtschaftsverhältnisse der indogermanischen Sprachen // In honorem Holger Pedersen. Kolloquium der Indogermanischen Gesellschaft vom 26. bis 28. März 1993 in Kopenhagen / Eds.: J. Rasmussen, B. Nielsen. Wiesbaden: Reichert, 1994. P. 235-251. Kloekhorst 2008 - Kloekhorst A. Etymological Dictionary of the Hittite
Inherited Lexicon. Leiden: Brill, 2008. Kocharov 2011 - Kocharov P. On ana-presents of Armenian. Indogermanistik und Linguistik im Dialog // Akten der XIII. Fachtagung der Indogermanischen Gesellschaft vom 21. bis 27. September in Salzburg / Eds.: Th. Krisch, Th. Lindner. Wiesbaden: Reichert, 2011. P. 271-278. Kronasser 1957 - Kronasser H. Gutturale und dentale Erweiterungen beim hethitischen Verbum // Studies Presented to Joshua Whatmough on his Sixtieth Birthday / Ed.: E. Pulgram. 's-Gravenhage: Mouton, 1957. P. 121-129.
LIV - Rix H. et al. Lexikon der indogermanischen Verben. 2nd Edition.
Wiesbaden: Reichert. Melchert 1994 - Melchert H. C. Anatolian Historical Phonology.
Amsterdam; Atlanta: Rodopi, 1994. Oettinger 1979 - Oettinger N. Die Stammbildung des hethitischen Verbums. Nürnberg: Hans Carl, 1979. Rieken 2013 - Rieken E. Sekundäre denominale u-Stämme im Hethitischen // Multi Nominis grammaticus. Studies in Classical and indo-European Linguistics in honor of Alan J. Nussbaum on the occasion of sixty-fifth birthday / Eds.: A. Cooper, J. Rau, M. Weiss. Ann Arbor; New-York: Beech Stave Press, 2013. P. 274-284. Riemschneider 1974 - Riemschneider K. Die Glassherstellung in Anatolien nach hethitischen Quellen. Anatolian Studies presented to Hans Gustav Güterbock on the occasion of his 65th Birthday. Eds.: K. Bittel, Ph. H. J. Houwink ten Cate, E. Reiner. Istanbul, 1974. P. 263-278. Sihler 1995 - Sihler A. New Comparative Grammar of Greek and Latin.
Oxford University Press, 1995. Yakubovich 2009 - Yakubovich I. Sociolinguistics of the Luvian Language. Leiden: Brill, 2009.
A. V. Shatskov. Another verbal suffix in Hittite?
Hittite verbs hassikk- 'to satiate oneself' and malikk- 'to become weak' have an obscure stem-final -kk-, which did not belong to the root, as is clear from the comparison with Pal. has- and Hitt. malisku- respectively. In several more verbs, like hark- 'to perish', the final -k- is also likely to be a suffix. The function of this suffix which already was not productive in Old Hittite is unclear but it can tentatively be compared to -k- of the perfect stem in Greek.
Keywords: morphology, etymology, Hittite, comparative linguistics, suffix, verb.
