Printer Friendly

Hittite zinni-, tuwarni-, zig and related matters.

IN HIS VALUABLE AND STIMULATING Studies in Hittite Historical Phonology H. Craig Melchert (1984, 114-17) makes the important observation that the stem-final vowel of Hit. zinni-zzi because of the consistent spelling -NI-IZ- (never -NE-IZ-) must be /i/, phonemically distinct from the vowels of the csvv. lukke- wasse- (/e/ or secondarily /e/) and of the characterized stems in -ske- -ie- (/e/). His explanation that zinni-, tuwarni- 'break' and hulli- 'fight, destroy'(1) have /i/ from the duratives in -anni- (with inherited /i/) is offered with judicious hesitation, and he himself notes that the points of partial coincidence (only 2sg. pret. *yannis |approximately~ *zinnes, imv. *yanni |approximately~ *zinne, 2pl. *yanniten(i) |approximately~ *zinne-/zinna-ten(i) as against, e.g., 3sg. *yanni |is not equal to~ *zinnezzi, etc.) constitute "a rather slim basis," especially since the hi-inflection of the duratives is opposed to the mi-pattern of zinni-. It is especially disturbing that the supposed analogical process should have operated in prehistoric times with ironclad rigor in a single direction on only the timbre of the stem vowel--not on the endings or inflectional class. Hence the sporadic bi-directional class transfers like zinahhi and iyannizzi, cited by Melchert in support of "considerable mutual influence of these classes," are of later date ("within the historical period"), of different sort, and so neither parallel nor probative.

The source of these difficulties is the etymology. Melchert (1984, 29 n. 61, 95, 114-15) accepts Oettinger's (1979, 151-52) reconstruction of a set root |*seih.sub.1~ with nasal-infixed present stem *si-ne-|h.sub.1~-. This leads to proximate *zin(n)e- or *zin(n)e- and the need to appeal to analogy. Moreover, the etymology lacks adequate support, since the sole extra-Hittite comparandum offered by either scholar is Lat. sino 'permit', and the difficulties raised by this equation are prohibitive. If one speculates that the original meaning of putative |*seih.sub.1~- was 'let go of' (vel sim.), semantic comparison is probably possible, but even so, serious problems remain. Integral to the meaning of Hit. zinni- is a noematic feature of COMPLETION,(2) which is not an essential component of sino or its compounds.(3) Briefly to illustrate we can compare KBo 19.142 ii 20-22 mahhanmakan UDU arkuwanzi zinnanzi nu UZUNIG.GIG UZUSA happinit zanuwanzi UDU-makan hamandan markanzi 'when they finish gutting (i.e., removing the organs of) the sheep, they cook the intestines and heart with fire, but the entire sheep (torso) they butcher' (CHD 3.188) with Ter. Andr. 22-23 dehinc ut quiescant porro moneo et desinant male dicere 'I warn them (my critics) to be silent from now on and to stop speaking ill', further Lucr. 4.400-402 atria versari . . . fit uti pueris videantur ubi ipsi desierunt verti 'it seems to children that the hall keeps turning round when they themselves have discontinued (= are no longer) turning'. Hit. arkuwanzi zinni- designates completion of the process of removing the animal's organs, a necessary preliminary to further preparations.(4) In contrast Lat. desinere denotes merely discontinuance, the abandonment of or cessation from some activity: completion is not entailed. The distinction is important. For example KUB 30.15 obv. 7 (Otten 1958, 66) nu mahhan hastai lessuwanzi zinnanzi means 'when they have finished gathering up the bones', but not 'when they stop gathering . . .', i.e., leaving some bones uncollected. The semantic counterpart to zinni- in Latin is not (de)sinere but finire, finem facere.(5) By far the most serious difficulty with the connection of Hit. zinni- and Lat. sinere is formal. Not only are the conditions under which word-initial *s- appears as Hit. z- quite unclear,(6) but especially ruinous to the etymology is the patent fact that Lat. sinere cannot continue a set, i.e., H-final, base. The a verbo of sinere in Latin is an isolate without parallel, a datum indicative of archaic morphology.(7) In this light the extra-presential patterns pf. si-u-i : pf. sbj. si-r-i- : pple. si-tu-s, which can immediately and naturally be seen as the direct continuation of (anit) aor. *sei- : aor. opt. *sei-s-i|h.sub.1~- : pple. *si-to-, precludes any analogical explanation.(8) There are no grounds whatever for believing in a set root |*seih.sub.1~- or in a punati-type present stem *si-ne-|h.sub.1~-.

Although etymological connection of Hit. zinni- with Lat. sinere cannot be maintained, Oettinger has made a valuable advance in his recognition that zinni- originates as the transitive counterpart to the middle (tantum) verb 3sg. ze(y)-a 'is cooked, is done (food)'.(9) He compares the mechanism to the familiar hark- 'perish' |right arrow~ har-nin-k- 'destroy': from the perspective of Hittite, a derivational not an inflectional process.

Both the semantic and the formal facts pertaining to Hit. ze(y)-/zinni- are well accommodated by etymological connection with PIE *dhu-e|h.sub.1~-(i-), a base extended from the simple root *dheu- by suffix |*-eh.sub.1~-, with subsequent further extension to *dhue|h.sub.1~i- upon reanalysis of present stem **dhue|h.sub.1~-io- |right arrow~ *dhue|h.sub.1~i-, ablaut *dhui|h.sub.1~-.(10) Lat. fi-ni-s |is less than~ *dhui|h.sub.1~-(s)ni-(11) will then be cognate as also ON dvi-na 'hinschwinden', OE dwi-na-n 'dwindle, waste away', OIr. di-th (|is less than~ *dhui|h.sub.1~-tV-) 'destruction, death, loss', etc.(12) Hit. mid. ze(y)-a(ri)(13) continues reanalyzed *dhue|h.sub.1~i-o(-) (type *kei-o), earlier **dhue|h.sub.1~, without further ado and can be ascribed to the protolanguage. The derived transitive is evidenced only in Hittite and may be a formation posterior to the separation of Anatolian from the IE (proper) languages. However, the pattern *io-intransitive: N-transitive is archaic, recurring in Indo-Iranian (with minor adjustments) as the familiar type of Ved. ksiya-te 'perishes, disappears': ksina-ti 'destroys', etc.(14) In any case the infixed derivative 3sg. *dhue-n-|h.sub.1~i-ti |is greater than~ zinni-zzi, 3pl. *dhui-n-|h.sub.1~-enti |is greater than~ zinn-anzi is morphologically natural and accounts for the attested forms with a minimum of difficulty.(15) The reconstructions require merely the assumption that the infixing of *-n- took place after the secondary root /|d.sup.h~|weh.sub.1~y- |is similar to~ |d.sup.h~|wyh.sub.1~-/ with its accent-regulated metathesis of laryngeal and semivowel had established itself in the system.

The weak stem *dhui-n-|h.sub.1~i- finds a canonical parallel in *|q.sup.u~ri-n-|h.sub.2~-, reflected (with modifications) by Ved. krini-te, OIr. crenaid, Toch. B. karna-(sk-) 'buy(s)', type *CRi-n-H-.(16) The strong stem violates Strunk's (1973, 67) rule that *-n- occurs before the root-final segment when this is not immediately preceded by /e/, or before /e/ when it is.(17) Strict adherence to this rule would lead us to expect either *dhu-n-|eh.sub.1~i-ti, a patently impossible trisyllabic stem, or *|dhueh.sub.1~-n-i-ti. I think the latter is also unlikely, but in any case the leveling of *zeni-zzi (vel sim.) to zinni-zzi after pl. zin-nanzi would be natural.(18) There are no grounds for reconstructing a stem *dhui-n-|eh.sub.1~- parallel to *|q.sup.u~ri-n-|eh.sub.2~- (cf. RV krina-ti), for the fundamental full-grade root structures are entirely disparate: *|q.sup.u~ri(i)|eh.sub.2~- or *|q.sup.u~|reih.sub.2~- |is not equal to~ *|dhueh.sub.1~i-. The accent-ablaut structure of the reconstructions proposed here resembles that of the non-samprasarana set-root stems of the type *ster-n-|h.sub.3~- (Lat. 3sg. sterni-t, OIr. 3sg. sernaid): *str-n-|h.sub.3~-.(Ved. 3pl. strn-anti).(19) The canonical acceptability of a stem with major phase-structured/|C.sub.0~e-n-Hy-/ is solidly supported by the annai-duratives if Melchert's (1984, 116) reconstructions pres. 3sg. *ie-|nh.sub.2~i-ei |is greater than~ *yanni |approaches~ yannai, pret. 3sg. *ie-|nh.sub.2~i-s |is greater than~ yanni-s are correct.(20) The familiar samprasarana type CReC- |approaches~ CR-n-eC- is represented in Hittite by the type of huek-zi |approaches~ hunik-zi (Strunk 1979b) and of tarnahhi if from *tr-n-|eh.sub.2~- as widely believed. A possible further example of the *ster-n-|h.sub.3~-ti type in Hittite is 3sg. su-u-ni-iz-zi 'immerses, plunges; sows' KBo 6.26 i 34 (Laws 166 OH/NS, cf. pret. 3sg. -ni-(e-)it 39, 44), StBoT 17 B obv. 29' (OH/NS), if Melchert (1984, 29 with n. 61) is right to suggest that the synchronically disparate verbs suni(ya)- 'immerse, sow' and sunna- 'fill' originated in the ablaut variants of the n-present to a root *seuH-. The plene-spelled vowel of the singular su-u-ni- may well have been accented so that a reconstruction *seunH-ti (= ||sewn.sub.e~Hti~) |is greater than~ suni-zzi: *sunH-enti |is greater than~ sun-nanzi 'they fill' serves better than a sg. *su-n-eH-ti. The word for 'autumn', ze-e-na-, has been connected etymologically with zinni-, and I believe the connection should be maintained. The semantics are unforced, since the autumn as the final time for sowing,(21) i.e., the end of at least the agricultural year, can readily stem from a base meaning 'come to an end'. Formally a reconstruction *|dhueh.sub.1~-no-, i.e., an old derivative from the suffixed root *dhu-|eh.sub.1~- prior to its contamination by *i, originally at home in a verbal stem formant, seems equally natural.

The formal canon *|C.sub.0~eH-no- is firmly established in the function of deriving (substantivized) adjectives directly from verbal roots. Cf. *|h.sub.1~ei- 'go' |approaches~ suffixed root *|h.sub.1~i-|eh.sub.2~- |approaches~ *|h.sub.1~|ieh.sub.2~-no- |is greater than~ Ved. nt. yana- 'car, vehicle; course, way, journey' (earliest RV 4.43.6, the ajiram yanam of the Asvins),(22) *|steh.sub.2~-no- |is greater than~ Ved. nt. sthana-place, abode; standing' (RV +), OCS m. stan '(army-) camp', *|deh.sub.3~-no- |is greater than~ Ved. nt. dana- (and dana-) 'gift, giving' (RV +), m. dana- 'gift(-horse)' (RV 3x), Lat. donum 'gift', et al.(23) Within Hittite *CeC-no- derivatives seem evidenced by pesna- 'man' |is less than~ *pes-no-,(24) isna- 'dough' |is less than~ *ies-no-,(25) perhaps parsana- 'panther (|is less than~ *speckled)' |is less than~ *pers-no-,(26) et al. An alternative possibility is that ze-e-na- is a secondary derivative built to a lost root noun, i.e., (as if) *|dhueh.sub.1~ (-) |approaches~ *|dhueh.sub.1~-no-.(27)

A potential difficulty with the above analysis is the lack of any indisputable example of Hit. z- |is less than~ *dhu- (before front vowel). Neither, however, are there cogent countercases. Ablaut and Sievers' variants, *dhu-C- |is greater than~ duC-, *dhuo- (probably) |is greater than~ da-,(28) *dhuuV- |is greater than~ dumV- (duwV- ?), present noncomparable environments and so are nonprobative. This lack of parallels is not surprising in view of the small number of lexically distinct morphs with the initial shape *dhuE- securely to be reconstructed for PIE. Moreover, the number of Hittite lexemes with extra-Anatolian etymologies of a quality adequate to vouchsafe reliable preforms is also small, at least compared with the considerably larger number available between, say, Vedic and Greek, or indeed within IE proper, generally. As a case in point the verb tuwarni/a- 'break (trns.)' has often been seen (since Goetze 1954, 403) as stemming from a PIE (or PIH) root *dhuer-, Skt. dhvara-ti.(29) However, for both dhvara- and Hit. tuwarni- alternative explanations are available so that the cognation of the forms and hence the putative PIE root *dhuer- become doubtful. In the Rig Veda the root "dhvr-" 'injure, damage, harm'(30) exhibits only the present stem dhurva-: dhvara- is a patent innovation of MS verse--replacing the dhurva- of corresponding mantras in other YV sam-hitas--and TS prose in word-play with adhvara-.(31) The RV furnishes pres. dhurva- 3x (books 6, 8, 10), to which s-aor. adhursata 5.12.5, derived nn. dhurvan- (nt.) 9.61.30, dhurti- (f.) 9x (books 1, 7, 8). The material gives ample evidence that in the oldest stratum of Indic available to us the fundamental root shape was dhur- and the fundamental form was the present dhurva-, for which the reconstruction *|dhrh.sub.2~-ue- immediately suggests itself: cf. turna-ti |is less than~ *|trh.sub.2~-ue-, jurva-ti |is less than~ *|grh.sub.2~-ue-. The evident extra-Indic cognate to dhurva- is Gk. 3sg. |Theta~|Rho~|Alpha~|Upsilon~|Epsilon~|Iota~ 'breaks', which may be no more than a thematicized replacement of |Theta~|Rho~|Alpha~|Upsilon~-|Tau~|Iota~ |is less than~ *|dhreh.sub.2~u-ti. For the type compare Hit. /|tarh.sup.w~-tsi/ |is less than~ *|terh.sub.2~i-ti (Oettinger 1979, 222-23). Thematicization of earlier root athematic presents active was a highly productive process in pre-Greek, cf. Gk. |Theta~|Epsilon~iv|Omega~: Ved. han-mi, Hit. kuen-zi; Gk. pple. |Epsilon~|Kappa~|Omega~v: Ved. vas-mi, Hit. wek-mi; Gk. |Epsilon~|Rho~|Alpha~|Omega~ |center dot~ |Epsilon~|Rho~|Alpha~-|Mu~|Alpha~|Iota~, etc. (details Specht 1936, 211-26). Oettinger's (1979, 222) explanation of Ved. sg. turva-ti (as if |is less than~ *|trh.sub.2~ue-ti) as a secondary development of Indo-Iranian based on pl. turvanti (|is less than~ *|trh.sub.2~ue-nti) is plausible. Hence the *|dhrh.sub.2~ue-ti underlying Ved. dhurvati may well be a secondary replacement of *|dhrh.sub.2~ue-ti. On the other hand, if pre-Greek shared in the innovation with Indo-Iranian, then for *|dhrh.sub.2~ue/o- we expect Gk. *|Theta~|Rho~|Alpha~|Epsilon~/o- |is greater than~ 3sg. *|Theta~|Rho~|Alpha~|Epsilon~|Iota~, cf. pres. 3sg. *|Tau~|Rho~|Omega~|Epsilon~|Iota~ (|Phi~ 293) |is less than~ *|trh.sub.3~ue- (Peters 1980, 340). Attested |Theta~|Rho~|Alpha~|Upsilon~|Epsilon~|Iota~ would then be leveled after full-grade forms like |Tau~|Rho~|Alpha~|Upsilon~-|Mu~|Alpha~ (regular |is less than~ *|dhreh.sub.2~u-mn) together with innovatory "super" zero-grade *|Theta~|Rho~|Alpha~- built secondarily to *|Theta~|Rho~|Alpha~- after built inherited |Alpha~: |Alpha~ in, e.g., |Sigma~|Tau~|Alpha~: |Sigma~|Tau~|Alpha~- |is less than~ *|sth.sub.2~-: *|steh.sub.2~-, etc. (cf. Peters 1980, with lit.). This seems a more likely explanation of the root shape of |Tau~|Epsilon~|Theta~|Rho~|Alpha~|Upsilon~(|Sigma~)|Mu~|Alpha~|Iota~, |Theta~|Rho~|Alpha~|Upsilon~|Sigma~|Iota~|Sigma~, |Theta~|Rho~|Alpha~|Upsilon~|Sigma~o|Sigma~ etc., than that they are direct continuations of *|Theta~|Rho~|Alpha~|Upsilon~- |is less than~ PIE *|dhrh.sub.2~u-(C) without metathesis as against the metathesized |Theta~|Rho~|Upsilon~- of Horn. aor. psv. 3sg. |Theta~|Rho~|Upsilon~|Lambda~i|Chi~|Theta~|Eta~ (|Psi~ 396) 'was injured, smashed' |is less than~ *|dhruh.sub.2~-(s)l. . . , with derivational links to |Theta~|Rho~|Alpha~|Upsilon~|Omega~ already severed in prehistoric times. The final result was nonablauting |Theta~|Rho~|Alpha~|Upsilon~-.

Goto's (1987, 191) view that dhurva- is secondary for *dhur-va- ("etwa nach turva-") in turn for **dhvr-va- is based on (1) the assumption of a PIE root *dhuer- (following Schwartz 1966, 121-22) and (2) anit forms within Vedic attested from the Rig Veda on. However, extra-Indic support for *dhuer- is limited to Hit. tuwarni-, for which an alternative etymology is also at hand. Moreover, the comparison dhurva-: |Theta~|Rho~|Alpha~|Upsilon~- is immediately preferable on morphological grounds: the u-stem type **|dhreh.sub.2~-u- |approaches~ is established and paralleled and eliminates both the awkwardness of accounting for the Hittite n-stem as well as the need to appeal to analogy to understand the quantity of Ved. dhur-.(32) The anit forms are based on dhur-, which continues the pre-vocalic zero grade *|dhrh.sub.2~-V with u-vocalism here as elsewhere in accord with the timbre of the fundamental pres. dhur-va-; cf. tura-: tur-va-, next to tira-. Though attested late, the root noun evidenced by dhura |is less than~ PIE inst. *|dhrh.sub.2~-|eh.sub.2~ must be judged an archaic inheritance on the grounds of morphological type.(33) It is this dhur- that founds a new pre-consonantal dhvr- (RV only satya-dhvr-t 10.27.1), whence a gunated dhvar- is easy to understand.(34) Etymological connection of Ved. dhurva-: Gk. |Theta~|Rho~|Alpha~|Upsilon~|Sigma~ precludes comparison with Hit. tuwarni- on formal grounds: note especially *|h.sub.2~ (|is greater than~ Gk. |Alpha~), for which we expect -h(h)- in Hittite. Moreover, within Hittite itself tuwarni- is fairly surely to be connected with dudduwar-'cripple, break',(35) and the pre-vocalic attestations of this base, du-ud-du-wa-RI-es-ta KUB 14.15 ii 6, -ri-is-ta ibid. 14, -ra-an-za(-kan) KUB 36.75 iii 14, discourage reconstructing a preform *dhuer-, for *e would not appear as a in this environment.(36) It is accordingly incorrect to backproject Hit. (-)tuwar- into the protolanguage as *Duer-/*Duor- or the like. We should rather follow Neumann (1958, 223-25) and recognize in -ar an Anatolian suffix that recurs in terms denoting a "korperlich- seelischen Zustand": *nahs-ar 'fear' (|approaches~ nahsar-att-, nahsar-iya-), tusgar-att-, -atar 'joy', suppar-iya- 'sleep, dream', tangar-ant- 'hungry, not partaking'. The point to be stressed is that while the suffix itself is archaic and traceable to the protolanguage, the combination of this suffix with a given root need not be. In this light (-)tuwar- is readily seen as Hit. *tu- + -ar |approaches~ *tuwar with *tu- |is less than~ *D(e/o)u-.(37) Since neither Ved. dhvara- nor Hit. tuwarni- furnishes indisputable evidence for a PIE *dhuer- 'damage', no reliance can be placed on diachronic sound laws stemming from so doubtful an etymology. To return to the question of whether *dhu- |is greater than~ z- (in at least certain environments), the justification for this development rests upon (1) the (at least) orthographic falling together of word-initial obstruents in Hittite(38) and (2) the correctness of the view that *tu- |is greater than~ z-. This view, which goes back to Carnoy (1952, 426; 1954, 233-35) and Pisani (1969, 266-68), was renewed with refinements by Georgiev (1978, 93-97; 1980, 51-52) and Hamp (1980, 64). It has been sharply and insightfully challenged by Melchert (1983). The issue hinges on a fluctuating assortment of etymologies, many of which are without serious merit. Moreover, the possibility of deriving the 2sg. pers. prn. nom. zig. |is less than~ *tu- has received particular attention, although strictly speaking the correctness of some alternative preform would merely make this word nonprobative.

Melchert (1983, 163-64 n. 2) objects that, "none of the examples for *tw |is greater than~ z is persuasive," and cites tu-ekka- 'body' (: Ved. tvac- 'skin') and 2sg. prn. obl. tuel tuedaz as counterexamples. As regards the first point it seems to me that Georgiev's equation of Hit. zah- 'beat, strike': Gk. |Sigma~|Alpha~|Omega~ 'sift' is quite cogent, especially if we take full account of the tertium comparationis, Lith. tvo-ti '(mit Prugel, Stock, Peitsche, Axt, Hand u. dgl.) (drein)schlagen . . .',(39) the semantics of which match those of zah- impeccably.(40) Hence for Gk. |Sigma~|Alpha~|Omega~ a semantic development '*beat |is greater than~ *beat through a filtering device |is greater than~ sift, etc.' is entirely plausible.(41) That |Sigma~|Alpha~|Omega~ continues *tu- is evidenced by pf. mid. pple. |Epsilon~|Tau~|Tau~|Eta~|Mu~|Epsilon~vo|Sigma~ (Pherec. 211 Kock), cmpd. |Delta~|Iota~|Alpha~-|Tau~|Tau~|Alpha~|Omega~.(42) The telic Aktionsart of 'strike, beat (through)' suggests a fundamental root aorist for the protolanguage: 3sg. *|tueh.sub.2~-t |is greater than~ *|Tau~|Alpha~, to which the attested press. |Sigma~|Alpha~|Omega~ (-|Tau~|Tau~|Alpha~|Omega~ |is less than~ *|Tau~|Alpha~|Iota~|Omega~ as if |is less than~ * |tuh.sub.2~io and |Sigma~|Eta~-|Theta~|Omega~, are immediately understandable formations. The attested (late) neosigmatic aor. |Epsilon~|Sigma~|Eta~|Sigma~|Alpha~* (Hp.) is a well paralleled innovation, cf. already Hom. act. |Pi~|Lambda~|Eta~|Sigma~|Alpha~ mid. |Pi~|Lambda~|Eta~|Mu~|Eta~v, etc.(43) As regards Hit. zah(h)both hi-forms and mi-forms occur,(44) and either pattern is compatible with a PIE root aorist.(45) For. Lith. tvo-ti tvo-ja tvo-jo |is similar to~ PIE aor. *|tueh.sub.2~-t cf.sto-jo |is similar to~ *|steh.sub.2~-t. Georgiev's etymology is supported by (1) its foundation on three noncontiguous subgroups and (2) the secure correspondence of semantic and morphologic details. The phonologic equation of *|Ceh.sub.2~- |is greater than~ Hit. Cah(h)-, Gk C|Eta~-C|Alpha~, Lith. Co- is also solid. The sound change *Du- |is greater than~ z- requires corroboration.

Connection of Hit. zank-(il-) in zankilai- 'impose a penance upon, exact a penalty from, fine', zankilatar 'penance, amends, fine' with YAv. Obazjaiti 'in Bedrangnis gerat' and the Germanic group of OSax. thwingan, OHG dwingan, NHG zwingen, etc., is also reasonable. The Germanic material, copiously attested, points to the presence of two semantic features in the earliest monuments: (1) restriction or reduction (as the result of force from without) and (2) distress. Cf., e.g., OHG piduungan sint 'coartati sunt' Gl. I 274.59 (sim. I 405.32),(46) duanc 'artavit' Gl. I 272.59, kaduungan uuesan 'constringi' Gl. I 61.38 (sim. Gl. I 6.12, 7.11, 60.38), gaduunganorom '|s~trictioribus' Gl. I 47.30, piduingo 'infrenabo' Gl. I 613.66-68, duang 'frenum' Gl. II 231.62, duingu 'ango' Gl. IV 327.14 (sim. IV 33.45, II 662.51), dvingent 'premens' Gl. II 643.17, wuo engi ist thiu phorta inti bithungan weg 'quam angusta porta et arta via' Tatian 40:10, ther ni thuingit sinaz muat 'he who does not restrain his passion' Otf. 2.12.91; cadungan ubilu 'adfligor malis' Gl. I 40.24-25 (sim. 27), thuuingit 'adfligit' Gl. I 121.41, duinget 'urget' Gl. II 630.35, piduungan uuirdo 'pessimabor' Gl. I 569.30, piduungun 'affecerunt' Gl. I 646.5, anguzliho, duingit 'anxie, afficit' Gl. II 168.40-41, thar si mihilaz githuing . . . theiz ist abulges dag, arabeito, quisti, joh managoro angusti '(he tells that) then shall be much distress . . . that it is a day of wrath, of labor, of torment and of great anguish' Otf. 5.19.22-24, and explicitly in the context of punishment OSax. endi witi gecos, hard hellie gethwing 'and he (Judas) chose his punishment, the hard pains of Hell' Heiland 5168-69. The semantic features of Germanic are neatly matched by Hit. zankila-, which designates (1) a restriction or reduction of wealth as (2) a punishment for neglect of duty. Cf., e.g., KUB 13.4 iv 7-10 man-|ma~-at istanta nuskatteni nu-smas-at wastul nusmas arianzi . . . ISTU GUD-ia-smas 10 UDU-ia zankilanzi 'but if you keep delaying it (offering to the gods), then that is a sin on your part, and they will consult an oracle; (they will do as the gods direct, and) they will fine you a cow and 10 sheep as well'.(47) Less helpful are YAv. Obazjaiti (Yt 19.58 = 61), ase-Obo.zgatema- (Y 13.2), since the contexts in which they occur do not permit an especially close determination of their semantics. Bartholomae's (1904, 798) 'in Bedrangnis gerat', 'der allerdrangendste, alle allereifrigste' are compatible with the analysis offered above, and as regards ase-Obo.zgatema the glosses Phlv. ranj (Bartholomae, 263) and Neryosangh's (bahu-)klesa-(harsa-tama-) corroborate at least the presence of a feature denoting distress.

Morphologically *zankil recollects hurkil |is less than~ *|h.sub.2~urgh-el(-) and suil 'thread' |is less than~ *|suh.sub.1~-el.(48) Puhvel's (1971; 1991, 401-2) insightful determination of the meaning of hurkil as "a sexually based capital crime" originally 'hanging matter, strangulation' and its cognation with OHG wurgen, etc., 'strangle', etc., encourages the speculation that *|h.sub.2~urgh-el denoted 'punishment by hanging, strangulation' and *tungh-el 'punishment by restriction, constraint'. This in turn permits not only an immediate understanding of the morphological parallelism but also accounts in particular for the final segment of *tuengh- (with countercanonical |-VCE~ . . . |+ASP~) as against the *tuenk- of Lith. tvenk-ti 'stauen, (einen Seufzer) unterdrucken'. Note further that *-el-derivatives recur--as *el-o-adjectives--in Germanic for each root: OHG dwengil (*pwang-il-a-) 'exactor' Gl. II 431.32; OSax. wurgil 'rope', MHG wurgel 'strangler'. Melchert (1984, 119-20) argues on the grounds of "the only attested OH spellings" that Hittite nouns in -il, -zzil were /-il/ (not /-el/), /-tsil/.(49) This seems acceptable, but his argument that the agreement Hit. -zzil = Luv. -til requires PAnat. *-til from *-ti- + *-il (not *-el or *-el), which in turn represents PIE *-i- + *-l-, is not convincing. The suffix *-el-, like *-ti-, is archaic, an element of the protolanguage; but the combination of the two is an innovation of Proto-Anatolian that occurred significantly later in time under conditions largely unclear to us. For neuter derivatives we expect originally proterodynamic inflection: NA *tuenk-l, G *tunk-el-(o)s. Hit. zankil- and hu-(u-)ur-ki-il hu-ur-ki-la-as evidence Anatolian leveling of both root and suffix vocalism after the oblique stem, but the not infrequently plene-spelt NA hu-u-ur-ki-il (citt. Puhvel 1991, 401) suggests that the accent either remained mobile or was leveled onto the initial syllable. We arrive at Anat. *-el |is greater than~ *-il, and it is this *-il that combined with *-ti- to form *-til. The word for 'thread(s)' exhibits multiple spellings: su-u-e-el HT 1 iii 9 (and duplicates(50)) and may well reflect a collective (with Schindler 1975) in *-el as against an ordinary singular in *-el (for *-l) continued by su-u-il KBo 15.10 (+) i 7; Mastigga i 29, ii 5, 10.(51) The noncontiguity of Baltic and Greek vouchsafes archaic status for the *tuer(H)- underlying Lith. tver-ti '(um-, ein-)zaunen, verbinden (eine Wunde), fassen, (er)schaffen' and Gk. |Sigma~o|Rho~o|Zeta~ (f.) 'funerary urn'. Cognation of Hit. zeri- 'cup' is entirely possible from the viewpoints of phonology and semantics, although it must be noted that for items of (cultic) realia non-IE etyma are likely. Inflection as a neuter i-stem lays no ground for objection: meni- 'face, chin', OIr. men mena 'mouth', Lat. men-tum 'chin'; lesi- 'liver', Arm. le-ard 'liver'; huelpi- 'newborn animal',(52) et al.

The second point raised by Melchert against the development *tu- |is greater than~ z-, i.e., that *tu- is preserved in twekka- 'body' (: Ved. tvac- 'skin') and in the pronominal forms 2sg. gen. twel, abl. twedaz, loses force in view of the fact that each example is closely linked to samprasarana ablaut variants in tu-. Cf. twel twedaz : dat.-acc. tug; twekka- : gen. tu-ug-ga-as-ta-as KUB 30.10 obv. 9',(53) abl. tu-ug-ga-az- KUB 17.10 iii 10 (Telipinu Myth), KBo 26.132, 3' ("Anatolian" Myth), tu-ug-ga-za- KBo 13.99 rev. 13' (Ritual), et al. This ablaut is surely old.(54) However moribund the pattern may have become by historical times, there can be no real doubts as to its prehistorical vitality or its power to hold--or model the restoration of--(prevocalic) *tu- in paradigmatic alternation with its morphophonemic counterpart *tu-. The real question, then, is why nom. sg. zig was not so held or leveled. In his careful study of the second singular pronoun in Anatolian, Melchert (1983), after a detailed critique of patently cumbersome earlier attempts to explain the exclusively nom. Hit. zi-g, Pal. ti as continuations of *te, a form established only as accusative (: nom. *tu), adds further objections of which the most cogent is the assertion that accented *e does not yield Hit. i.(55) Hence the correct explanation of Hit. zig must start from the nom. *tu, not from *te. Melchert's own explanation, however, requires the assumptions that (1) original nom. *tu : obl. *twe- was replaced by nom. *tu : dat.-acc. *twu : obl. *twe- and that (2) in only the specific environment after a dental consonant a conditioned sound change of (long) *u |is greater than~ *iu (or |+PALATAL~ u) |is greater than~ i operated. Hence in the nom. *tu |is greater than~ *|t.sup.y~u |is greater than~ *|t.sup.y~i |is greater than~ Hit. zi (+ -g), whereas in the dat.-acc. *twe |approaches~ *twu |is greater than~ tu (+ -g) with no palatalization and nonspecial reflex of *u after *w. Although Melchert's analysis is probably not impossible, I do not think it is the right solution. The assumed replacement of inherited nom. *tu : acc. *tue by *|t.sup.y~u (or *tiu) : *tuu, which lacks any independent motivation, is difficult both morphologically, in that the formal marking of a case distinction has shifted to a bizarre and unparalleled internal opposition |+PLTL~ : -u- (or -i-: -u-), and phonotactically, in that the viability of strings shaped/Cwu |center dot~/has not been demonstrated for Proto-Anatolian.(56) The chief difficulty, however, is simply that the postulated sound change *u |is greater than~ *|.sup.y~u |is greater than~ i after dental lacks the corroboration of additional Hittite examples,(57) and counterevidence may be at hand in OH tu-u-wa- 'far' if |is less than~ *du- |is less than~ *|duh.sub.2~-io- (Eichner 1978, 160 n. 69) : Ved. du-ra- 'far'.(58)

The problem is not to motivate the -u- of the accusatives Hit. tu-uk, Pal. tu-u, which is attested, but to explain the disparity of the nominatives Hit. zi-ik, Pal. ti-i, and the *tu evidenced by IE proper. The path to a solution begins with the recognition that Anatolian supports Brugmann's (1911, 384) view that in the independent personal pronoun an opaque suppletive nominative in formal contrast with the stem shared by the accusative and oblique cases was characteristic of the first person (in all numbers) but not of the second.(59) In other words the unity of stem evidenced by the Hit. 2pl. nom. sum-es = dat.-acc. sum-as and the familiar unity of the IE (proper) 2sg. nom. *tu = acc. *tu-e(60) combine to indicate that the pre-Anatolian form for both cases was an undifferentiated *tu.(61) While the pronoun of the second person plural was readily differentiated by endings that recur in the nominal system, in the second singular the undifferentiated *tu continued as accusative (and dative), while a new nominative *tu-i was created by the addition of the (stressed) deictic suffix -i, which recurs in Hit. nom. as-i, acc. un-i 'this, that' and is to be compared with Gk. -i in ovtooi, etc.(62) Hence *tu-i |is greater than~ *tui is 'you there!',(63) and the source of the formation is not the nominative but the vocative.(64) Since the pattern N = V |is not equal to~ A was well established, *tui served easily as a nominative, now formally distinguished from (inherited) acc. *tu. We arrive at Palaic ti : tu, Hit. zi-g: tu-g by phonological replacement and in Hittite the secondary addition of -g.(65) It is now easy to understand why Hit. zig was not restored to *tuig in the manner of the oblique forms. As a vocative it must have had a very high currency at the most fundamental levels of usage. Moreover, the more-or-less marginal position of the vocative as regards the nominal paradigm will have played a role: cf., e.g., Russ. voc. boze 'O God!' with unleveled -z-. As a nominative it simply fell in with the pattern of the first person: ug |is not equal to~ amm-ug -el -edaz, so now zig |is not equal to~ tu-g -el -edaz. So far as I see now, cogent evidence for the assibilation of word-initial *t before *i in Hittite is lacking, and Pisani's (1969, 266-68) observation that the process in Hittite, as in non-Doric Greek (type rooic, etc.), was restricted to noninitial syllables may well be right. This casts still more suspicion on Melchert's *tu.(66) The hypothesis that *Du- |is greater than~ Hit. z- is reasonably well supported by *tu-i |is greater than~ zi-g, *|tueh.sub.2~- |is greater than~ zah-, *tunk- (or *tuenk-) |is greater than~ zank-, *tuerH- |is greater than~ zer-i and *|dueh.sub.1~io |is greater than~ zea. As for the phonological processes underlying this development we know far too little about either the phonetic detail or the systematic phonemics of historical Hittite to do more than speculate. A possibility is that the change originated before front vowel with assimilatory fronting followed by glide-incorporation and derounding: *DuE- |is greater than~ *DuE- |is greater than~ *|D.sup.u~E- |is greater than~ *|D.sup.y~E- |is greater than~ zV-. This requires the assumptions that the continuations of *-|eh.sub.2~- and *-nC- (in ablaut to *-enC-) still exhibited the feature |+FRONT~ during the early stages of the process. If this scenario is correct, the apparently similar phenomenon of non-Doric Greek, e.g., ooi |is greater than~ *tuoi, which is restricted to voiceless *t and independent of the timbre of the vocalism, would be only a distant and imperfect parallel. It is admittedly somewhat disturbing that old *d(h)iV- |is greater than~ Hit. sV-.(67) However, the renovation of an inherited phonemic opposition is no surprise, and an explanation in terms of relative chronology would not be unreasonable.

1 Pace Melchert (1984, 115) the ambiguity of hu-ul-LI-IZ-zi hu-ul-LI-IT (with either /e/ or /i/) cannot be settled by collateral pret. 3sg. hu-ul-li-is, which occurs in a thirteenth-century copy of the Zalpa text (KBo 3.38 rev. 24' = StBoT 17, Expl. B) where Exemplar A in old script exhibits hu-ul-LI-IT (KBo 22.2 rev. 8'). Elsewhere the NH Expl. B also exhibits correctly hu-ul-LI-IT (obv. 31' and probably 15'). The evident explanation of hu-ul-li-is is simply that it is a cross between the "hu-ul-li-it" of the OH exemplar and the hu-ul-la-as, now attested in the NH Bronzetafel Bo 86/299 (Otten 1988) i 98, of the scribe's own speech. Oettinger's (1979, 263) "Fehlerhaft" is quite justified. What is evidenced by hu-ul-li-is is that a thirteenth-century copyist read the already obsolete hu-ul-LI-IT with /i/, but this need not have been the phoneme of living OH speech.

2 Cf. Neumann, 1973, 298, "zinna- heisst nicht 'aufhoren' . . . sondern immer 'fertig sein mit etwas'."

3 Within Latin the semantic history of sinere was in all probability (1) 'put, place'--maintained in the pple. situs 'placed, situated' and in the derived n. situs 'site, situation, region' and renewed by the cmpd. ponere 'place' |is less than~ *po-sinere--(2) 'leave (alone)', (3) 'allow, permit'. Cf. Ernout and Meillet 1932, 628-29. In none of these senses is a semantic feature of COMPLETION with regard to either nominal object or verbal action at hand.

4 See CHD 3.188 for the sequence of the processes involved.

5 The common Hittite construction zinni- + INFINITIVE 'finish . . . -ing' is matched in Latin by finem facere + GERUND (also common). In the domain of verbal activity the correspondence is especially striking. Cf., e.g., Sallust Cat. 52 postquam Caesar dicundi finem fecit 'when C. finished speaking' (sim. Cic. Sest. 136) with KUB 9.31 i 42 man memiyawanzi zinnizzi 'when he finishes speaking' (sim. KUB 33.106 ii 10f., KUB 17.14 obv. 22' = StBoT 3.60. al.) (each after a quoted text) and the ellipses after quoted texts Ov. Met. 1.557-66 . . . deus . . . dixit . . . finierat Paean 'the god . . . said, "TEXT" . . . Apollo had finished (speaking)' (sim. 13,123. 14.441, al.) with Tunn. ii 18-21 nu memai . . . mahhanma zinnai nas arai 'and she says, "TEXT." When she finishes (speaking), she stands up' (sim. AOAT 31.206:22', KUB 53.4 iv 28'f., al.). Contrast the very different COMPLETION-free sense of desinere in Caes. B.C. 3.19.7 mediam orationem interrumpunt subito undique tela . . . tum Labienus: 'desinite ergo de compositione loqui, nam nobis nisi Caesaris capire relato pax esse nulla potest' 'suddenly missiles from all sides broke off the discussion midway . . . then L. (said), "stop talking then about an agreement, for we can have no peace until Caesar's head has been brought in".' 6 Melchert 1984, 114 n. 70. Oettinger's (1979, 152 n. 40) idea that *s- |is greater than~ z-, "in Nachbarschaft von n bzw. m," fails before such reasonable equations as Hit. sanh- 'seek' : Ved. san- 'gain' (Sturtevant 1933, secc. 74, 87; Puhvel 1979, 299; Oetringer 1979, 182; et al.) as well as before the evident derivational relation Hit. ze(y)- 'be cooked, done' |approaches~ zinni-, rightly pointed out by Oetringer himself (details below), with nasal only in the forme fondee.

7 Ernout and Meillet 1932, 629.

8 Or (better) *kpei-, etc., see Leumann 1977, 177 (after Froehde). On siri- see now G. Schmidt 1985, 88-89 with n. 143.

9 Followed by Melchert 1984, 114 n. 70. See further Lehrman 1985, 155.

10 On the genesis of secondary roots of the shape CeHU- and their position in the morphological system of PIE, see especially Schindler 1969, 145-56; Mayrhofer 1986, 171-75.

11 Better than *dhi|g.sup.u~-sni- (: figo), Leumann 1977, 343; cf. Brugmann 1906, 288-89. For the formation cf. pe-ni-s with nasal element recurring in Hit. pes-na- 'man' (Neu and Otten 1972; Puhvel 1982, 179), am-ni-s (|is less than~ *ab-ni-) 'river' : Hit. hapa- 'river' (Puhvel 1991, 114-15 with lit.). For further examples, see Brugmann 1906, 286-89.

12 Cf. Pokorny 1959, 260-61. The simple root *dheu- (Gmc. *diwan 'die') was apparently punctual (or momentative) 'end'; the extended root *dhu-|eh.sub.1~, terminative 'come to an end, reach completion'. (For the terms see Hoffmann 1970, 29.)

13 ze-e-ia StBoT 25.54 ii 20' (OS), ze-e-a-ri Pap. (KBo 5.1) i 29, 36; further details Neu 1968, 206-7; Oettinger 1979, 515.

14 Cf. Insler 1972, 100-103.

15 For the development of intervocalic *-RH- |is greater than~ -RR-, see Watkins 1975, 376-78; Melchert 1984, 66 n. 33. For stressed *e |is greater than~ i before n+C (C |is not equal to~ dental stop), see Melchert (pp. 103-4) with the apt example of Hit. innara- : CLuv. annari-.

16 Cf. Strunk 1984, 156-57; Mayrhofer 1989, 410-11--both with lit.

17 The morphological properties of secondary roots of the canon |C.sub.0~eH-U- have not been thoroughly examined, and Strunk quite reasonably left them out of consideration in formulating his rule. Moreover, the formation of a derived n-causative to *|dhueh.sub.1~-i- may well have occurred only in Anatolian or pre-Hittite. There is no evidence that requires assigning it to the protolanguage. Neither the Latin nor the Germanic formations are comparable. 18 The synchronic transitive to ze(y)a(ri) is of course not zinni-, which had broken lexical ties with its founder before historical Hittite, but zanu- (from old script), a later formation after the productive pattern of nu-causatives. For the variation ze- : za- see Oettinger 1979, 515 n. 2.

19 For details, see Strunk 1973, 68; 1979a, 87-102, with lit.

20 Melchert follows (with significant modifications) Jasanoff in connecting the Hit. annai-duratives with Sanskrit secondary presents of the type is-anya-ti 'presses forward'.

21 Hoffner 1974, 41-42; cf. Jakob-Rost 1963, 183-84.

22 Perhaps also Lat. Ianus. See Kircher (1979, 160-61) for a well balanced appraisal of this etymology.

23 For further examples see Brugmann 1906, 259.

24 Neu and Otten 1972, 181-90; Puhvel 1982, 179; Oettinger 1986, 22.

25 Puhvel 1984, 384.

26 Cf. Oettinger (1986, 22), who reconstructs a zero-grade *prs-no-, although parsina- 'cheek' (p. 23) is backprojected as *pers-|neh.sub.2~-. Since *e |is greater than~ Hit. a before (*)RC, (Kronasser 1966, 25, with lit.) at least under certain conditions, it is difficult to be certain about the ablaut grade reflected by CeRC-bases. In any case *pers-no- |is similar to~ *p(o)rs-|neh.sub.2~- has at least an equal claim. Incidentally I see scanty grounds for speculating that the unattested Hittite word for 'Kaufpreis' was *us-na- with a zero-grade *us- sharply deviant from the *ues-no- reflected by Ved. vasna-. Arm. gin, Lat. venum, and the *uos-|neh.sub.2~- of Gk. wun, Lesb. ovva 'contract'. See Ringe 1984. Gk. wvoc 'price' is in my view not the continuation of a PIE variant *uos-no- but merely evidences pre-Greek leveling after *ues-|neh.sub.2~-. There is no reason why a fundamental *ues-no- should have been subsequently leveled after its own forme fondee *us-no-ie (|is greater than~ Hit. *usnai- |is greater than~ usne-sk-, |approaches~ usniia-). See Neu 1980, esp. p. 88. It is worth noting that the above reconstructions conform to the phonotactic canons of Beekes' (1985, 156-58) stage A II, i.e., the archaic phase of the protolanguage subsequent to the syncope of unstressed *e and *o when *o, but not yet *e, had been (re-)introduced in unaccented positions.

27 For possible examples, see Kronasser 1966, 181-82.

28 See Melchert (1984, 51-52) for the rule *u |is greater than~ 0, "between dental stop and *o in pre-Hittite." Secure examples only for *d: dan |is less than~ duo- (Puhvel 1978). idalu : CLuv. adduwali-. The notion that *du- and *dhu- developed differently (e.g., Weitenberg 1985, 125) cannot be sustained without due attention to the vocalism. Moreover, tuwarni- (see text below) must now be excluded from the discussion.

29 Benveniste 1962, 85; Eichner 1972, 75-76 (*dhuorneie-); Mayrhofer 1976, 741; Josephson 1979, 99; Oettinger 1979, 151 (*dhurne-); Melchert 1984, 36, 114 with n. 71 (*dhwerne-ye-); Weitenberg 1985, 125. Further discussion in text below. 30 Distinct from dhru- 'deceive'. Narten 1964, 157 n. 434; Schwartz 1966, 121-22; Hoffmann 1980, 95; Goto 1987, 190 n. 352.

31 Narten 1964, 157-58 with nn. 434, 437. Goto (1987, 190-91 with n. 354) gives additional detail.

32 The type of turva- would probably be an adequate model, but hvara-te / hurcha- (adduced by Goto, pp. 191, 352-53) cannot serve to explain the long a of dhurva-, which was so firmly established in (pre-)Rigvedic as to found, e.g., aor. adhursata, whereas hurcha- appears first in Yajurvedic prose (details Goto, p. 352). Old is RV hvara- = YAv. zbare- so that if this verb exerted any influence at all on dhur-, it can only have been to help model neological dhvar-V after old hvar-V upon the creation of new hur(cha-) identifiable with inherited dhur-.

33 Cf. the discussion of dhura, dhuradhuram, and nom. sg. dhur 'Verletzung -er' (homophonous with 'yoke-pole') by Hoffmann 1975, 116-17.

34 Hoffmann (1980, 95) aligns satya*dhvrt- 'die Wahrheit verletzend' with satyam dhurvant- (RV 10.87.12). Nevertheless, a certain association of dhur- with dhru-/druh- 'deceive' can also be indicated. RV 8.27.15 pra vah samsamy adruhah . . . na tam dhurtir varuna mitra martyam yo vo dhama-bhyo 'vidhat 'I praise you, O (gods) beyond deceit, . . . no harm/deception (comes upon) the mortal, Varuna (and) Mitra, who has satisfied your ordinances'. RV 1.36.15-16 paht dhurter aravnah . . . yo asmadhruk 'save us from harm/deception from the nongiver . . . (smite the nongivers and him) who deceives us'. Quite uncertain is dhvar-as- (f.) 'malevolent spirit', since both Rigvedic attestations are patently aligned with 'deception': 2.23.5 na tam amho . . . titirur na dvayavinah / visva id asmad dhvaraso vi badhase 'neither trouble . . . nor duplicity has overcome him: you drive away from him all dhvarases (O Brahmanaspati)', 4.23.7 druham jighamsan dhvarasam anindram tetikte tigma tujase anika 'wishing to slay the Lie, the Indraless dhvaras, he (Indra) fine-sharpens the (arrows') sharp faces to let fly'. Hence it seems unlikely that this word is an early gunation of dhvr- 'harm' with equally early semantic assimilation to dhru- 'deceive'. We should leave open the traditional view that 'deceive' in PIE was *dhuer- with zero grades *dhur-V- : *dhru-C- (for the metathesis see Mayrhofer 1986, 161-63, with lit.). The last variant gained importance by dint of the extended *dhru-gh- and the common dhru-ti-, so that the secondary full grades of Ved. drogha- and YAv. dravaya- are not as "complicated" as Schwartz (1966, 122 n. 3) reported.

35 Forrer 1926, 53 n. 1; Gotze 1933, 214-15; van Brock 1964, 142.

36 Equally unacceptable is *-dhuor-. which should yield *-dar- (v. n. 28 supra), and if samprasarana ablaut is invoked, we expect *duddur- like kukkur- to kuer-.

37 Neumann posited a base noun *dudduwar 'Lahmung, Steifheit' built to the "intensivreduplizierte Wurzel" *duddu- and left tuwarni- out of the discussion. Unreduplicated *tuwar, however, matches *nahsar, suppar, etc., better formally. In this case tuwarni- exhibits factitive -n-; du-dduwar-, intensive reduplication, regardless of whether it proves to be an intransitive oppositum to tuwarni- as Gotze (1933, 214-15) thought or a virtual synonym with van Brock (1964, 142). If Neumann is right to see PIE *dheu- (Pokorny 1959, 260) as the etymon and 'lameness' as an archaic semantic component (recurring in Germanic), the semantic prehistory would be 'cease to function, come to an end' (*dheu-) |is greater than~ 'become/make lame' (dudduwar-) |is similar to~ 'break bone' (tuwarni-, OH-Laws) |is greater than~ 'break (hard materials generally)', vel sim. Josephson (1979, 99) connected CLuv. lawarr-, which he glossed 'to break' (Laroche 1959, 63 'briser'). Although Josephson's reconstructions, *dhwer-no- and *|dh.sup.y~ewer-, are unacceptable, it remains formally possible to reconstruct a PAnat. *D/le/ou- + -ar (for *D/1- cf. t/labarna-, etc.) and maintain cognation. However, the meaning 'to despoil, strip' given by CHD 3.49 invites caution.

38 Melchert 1984, 69, with n. 128.

39 Senn and Salys 1963, 756.

40 That the essential and inherited meaning of zah- was 'beat, strike' is evidenced both directly in contexts of punishment, e.g., KUB 5.6 ii 49 halluwairr-a kuies nas PANI DINGIR-LIM zahanzi 'and those who quarreled--shall they beat them in the presence of the divinity?' (see Sommer 1932, 278 n. 4, for discussion and further citations), and indirectly by secondary acceptations 'bekampfen, raufen, storen, etc.' (see Kronasser 1966, 378, with citations) readily derived from 'beat'. The way to the meanings 'battle, fight' of the derivv. zahhai- zahhiya- is made plain by the reciprocal middle of Laws sec. 174 takku LU.MES zahhanda ta 1?-as aki 'if men exchange blows (|approaches~ fight) and one dies'.

41 Cf. Georgiev's 'schlage durch'. It is clear from, e.g., Hdt. 1.200, where fish are first dried in the sun, then ground fine in a mortar and then owoi dia oivdovoc 'they bolt it through linen', that the procedure must have required a certain force.

42 Lejeune 1972, 110, 275, 305. Cf. Chantraine 1983, s.vv. diattaw, onow.

43 Further details Chantraine 1961, 180-81; V. Schmidt 1968, 16-17 n. 49, with lit.

44 Details Oettinger 1979, 446-47.

45 See Barton 1985, with lit.

46 The OHG glosses (Gl.) are cited from Steinmeyer and Sievers 1879-1922.

47 Cf. Unal 1978, 115-16. with lit.

48 For the formation see Eichner 1972, 73-74, 85 n. 6, who rightly notes its unproductive and so archaic character.

49 Apart from the derivatives in -zzil Melchert's corpus omits *zankil and adds alii- 'flower', esharwil- 'blood-X' and passuwil- 'mountain stream (?)'. The first exhibits radical -(l)il (Puhvel 1984, 32-33), the second is a secondary derivative hardly comparable with inherited primaries, while the third is too obscure to be probative. We return to Eichner's hurkil *zankil suil. 50 KUB 9.31 iii 20, 9.32 i 7.

51 Note also su-u-i-il KUB 12.51 i 8; 7.3, 7, 13; su-u-el JCS 24.84 (Malli A) iii 5', su-i-el KUB 41.1 (Malli H) iii 13'.

52 See Puhvel 1991, 332, and, for a general discussion of the problem of neuter i-stems, Brosman 1978; Beekes 1987, 46-47 et passim.

53 Kantuzzili's prayer. Note the full-grade acc. tu-ek-kam-ma-an of the next paragraph (ibid. 14').

54 So Melchert 1984, 52-53.

55 Melchert 1984, 92. Note, however, pp. 111-12, 135-36, 150 for the development *e |is greater than~ i after *|h.sub.2~.

56 Melchert (1983, 159) recognizes that *twu is an "awkward sequence" and indicates the possibility of its being "immediately dissimilated to *tu." But what would have motivated the analogical creation of the awkwardness in the first place? The assumption of a direct replacement *tiu : *tue |approaches~ *tiu : *tu would be no improvement.

57 Duly acknowledged by Melchert (1983, 163). Moreover, the exclusion of short *u arouses suspicion, and the extra-Anatolian parallels are not exact: Osc. tiurri 'turrim' exhibits short u, the Middle Chinese rule applies to *a and *o as well as *u, etc.

58 Differently Melchert 1984, 30.

59 Cf., e.g., Hit. ug: amm-ug -el -edaz with, say, Lat. ego : me mei mihi; Hit. wes : anz- Ved. vay-am : asm-, Goth. weis: uns(-), etc.

60 Cf., e.g., Szemerenyi 1978, 275; 1985, 253 with lit.

61 If the protolanguage ancestral to Anatolian exhibited nom. *tu : acc. *tu-e, then Anatolian has neologically leveled. (Cf. Normier 1977, 210 with n. 93, and see Brugmann 1911, 414-15 for further parallels within IE.) It is in this context that Benveniste's (1962, 73) connection with the vocalism of other pronominal accusatives, e.g., apun 'ilium', may be pertinent. If the protolanguage exhibited undifferentiated *tu as against 1sg. nom. *eg- : acc. *m-e(-) (cf. Villar 1989), then the creation of acc. *t(u)-e(-) (after *m-e-) in IE proper represents a shared innovation significant for the evaluation of the Indo-Hittite hypothesis. An argument in favor of the latter assumption is that known Anatolian levelings show the spread of accusative forms to the nominative, e.g., HLuv. amu, NHit. ammug, anzas (as nomm.), perhaps already OH sumes if *ius was ever really ancestral to Anatolian. In either case the creation within Anatolian of a differentiated nominative *tui is a later development.

62 For this particle see Chantraine 1983, 452; Puhvel 1984, 5-6 with lit.

63 For quasi second-person usage in Greek cf., e.g., tic ovtooi; 'who are you?' Ar. Achar. 1048.

64 The pseudo-vocatives in -i to Hit. u-stems may reflect this same *-i. (For alternative guesses, see Eichner 1982-83 and Neumann 1982-83.) The innovation probably began where it is best attested, with the words *istanu-i 'O Sundeity!' and hassu-i 'O king!', where the suffix would denote distance and so respect for the majesty of a superior directly addressed. Note further such collocations as KUB 7.1 i 15 . . . dUTU-/zi-ik az-zi-ik-ki ak-ku-us-ki 'thou, O Sundeity, eat (and) drink!' for *istanu-i *tu-i '(O) Istanu yonder! (O) thou yonder!'. In this way we immediately understand the peculiar restrictions on the distribution of this formation. (For philological detail, see Eichner 1982-83, 233-36 and the lit. there cited.)

65 I provisionally follow Melchert (1983, 152-53) in the view that the final velar of Hit. zi-ik tu-uk is secondary vis-avis Pal. ti-i tu-u on the grounds that, "Palaic does not share with Luvian the loss of word-final stops." But Melchert cites examples only of preserved final dental (kuit, kat, kuwat) not of final velar, so that the case of, say, NE I -ly, it : OSax. ik -lik, it encourages caution.

66 And on Cowgill's (1965, 169 n. 56) view that the original nominative was *ti(H).

67 See Puhvel 1979, 301-2 and the lit. cited n. 26. Of course graphic (z-) can be either /dz-/ or /ts-/ just as graphic (s-) can be /z-/ or /s-/.

REFERENCES

Bartholomae, C. 1904. Altiranisches Worterbuch. Strassburg.

Barton, C. 1985. Hittite me-ri-ir, epp- and a note on the ablaut of root verbs. KZ 98:13-19.

Beekes, R. S. P. 1985. The origins of the Indo-European nominal inflection. Innsbruck.

-----. 1987. Indo-European neuters in -i. In Festschrift for Henry Hoenigswald, ed. G. Cardona and N. H. Zide. Pp. 45-56. Tubingen.

Benveniste, E. 1962. Hittite et indo-europeen. Paris.

Brosman, P.W. 1978. The gender of Hittite i-stems. In Linguistic and literary studies in honor of Archibald A. Hill, vol. 3, ed. M. A. Jazayery, E. C. Polome, and W. Winter. Pp. 19-27. The Hague.

Brugmann, K. 1906. Grundriss der vgl. Grammatik der idg. Sprachen. 2d ed. Vol. 2, pt. 1. Strassburg.

-----. 1911. Grundriss der vgl. Grammatik der idg. Sprachen. 2d ed. Vol. 2, pt. 3. Strassburg.

Carnoy, A. 1952. Dialectologie proto-indoeuropeenne. Orbis 1:423-27.

-----. 1954. Etymologies hittites. La nouvelle Clio 6:231-35.

Chantraine, P. 1961. Morphologie historique du grec. Paris.

-----. 1983. Dictionnaire etymologique de la langue grecque. Paris.

Cowgill, W. 1965. Evidence in Greek. In Evidence for laryngeals, ed. W. Winter. Pp. 142-80. The Hague.

Eichner, H. 1972. Die Etymologie yon heth. mehur. MSS 31:53-107.

-----. 1978. Die uridg. Wurzel *|H.sub.2~reu 'hell machen'. Die Sprache 24:144-62.

-----. 1982-83. Zur Genese der heth. Vokative auf -i und -e. KZ 96:233-40.

Ernout, A., and A. Meillet. 1932. Dictionnaire etymologique de la langue latine. Paris.

Forrer, E. 1926. Forschungen. Vol. 1, no. 1. Berlin.

Georgiev, V. I. 1978. Die Herkunft des heth. Suffixes -zi-el, -zi-il und der Lautwandel ide. tw |is greater than~ heth. z. KZ 92:93-97.

-----. 1980. Hethitische Etymologien. BalkE 23(2):49-52.

Gotze (Goetze), A. 1933. Die Annalen des Mursilis. Leipzig.

-----. 1954. Review of Hethitisches Worterbuch, by J. Friedrich. Lg. 30:401-5.

Goto, T. 1987. Die "I Prasensklasse" im Vedischen. Vienna.

Hamp, E. 1980. On IE *tu |is greater than~ Hittite z. KZ 94:64.

Hoffmann, K. 1970. Das Kategoriensystem des idg. Verbums. MSS 28:19-41.

-----. 1975. Der ved. Typus menamenam. In Aufsatze zur Indoiranistik, ed. J. Narten. Pp. 113-19. Wiesbaden.

. 1980. Das Verbaladjektiv von hvr bei Panini. StII 5-6:87-98.

Hoffner, H. A., Jr. 1974. Alimenta hethaeorum. American Oriental Series, vol. 55. New Haven.

Insler, S. 1972. Vedic mamatsi, amamadur and iyate. KZ 86:93-103.

Jakob-Rost, L. 1963. Zu den heth. Bildbeschreibungen, II. MIO 9:175-239.

Josephson, F. 1979. Assibilation in Anatolian. In Hethitisch und Indogermanisch, ed. E. Neu and W. Meid. Pp. 91-103. Innsbruck.

Kircher(-Durand), C. 1979. Heritage indo-europeen, apport etrusque et sabin, vestiges de substrat italique dans quelques noms de divinites romaines. In Universite de Nice, Centre de Recherches Comparatives sur les Langues de la Mediterranee Ancienne, Document, no. 4. Pp. 146-87.

Kronasser, H. 1966. Etymologie der heth. Sprache. Wiesbaden.

Laroche, E. 1959. Dictionnaire de la langue louvite. Paris.

Lehrman, A. 1985. Simple thematic imperfectives in Anatolian and Indo-European. Ph.D. diss., Yale University, New Haven.

Lejeune, M. 1972. Phonetique historique du mycenien et du grec ancien. Paris.

Leumann, M. 1977. Lateinische Laut- und Formenlehre. 2d ed. Munich.

Mayrhofer, M. 1976. Kurzgefasstes etymologisches Worter-buch des Altindischen. Vol. 3, no. 26. Heidelberg.

-----. 1986. Lautlehre. Part 2 of W. Cowgill and M. Mayrhofer, Indogermanische Grammatik, vol. 1. Heidelberg.

-----. 1989. Etymologisches Worterbuch des Altindoarischen. Vol. 1, no. 6. Heidelberg.

Melchert, H. C. 1983. The second singular personal pronoun in Anatolian. MSS 42:151-65.

-----. 1984. Studies in Hittite historical phonology. Gottingen.

Narten, J. 1964. Die sigmatischen Aoriste im Veda. Wiesbaden.

Neu, E. 1968. Interpretation der heth. mediopassiven Verbal-formen. Studien zu den Bogazkoy-Texten, no. 5. Wiesbaden.

-----. 1980. Die heth. Verben des Kaufens und Verkaufens. WdO 11:76-89.

Neu, E., and H. Otten. 1972. Hethitisch 'Mann', 'Mannheit'. IF 77:181-90.

Neumann, G. 1958. Hethitische Etymologien II. KZ 75:221-25.

-----. 1973. Review of Das Ritual der Malli aus Arzawa gegen Behexung, by L. Jakob-Rost. KZ 87:295-99.

-----. 1982-83. Zur Genese der heth. Vokative auf -i und -e. KZ 96:241-44.

Normier, R. 1977. Idg. Konsonantismus, germ. 'Lautverschiebung' und Vernersches Gesetz. KZ 91:171-218.

Oettinger, N. 1979. Die Stammbildung des hethitischen Verbums. Nuremberg.

-----. 1986. 'Indo-Hittite'-Hypothese und Wortbildung. Innsbruck.

Otten, H. 1958. Hethitische Totenrituale. Berlin.

-----. 1988. Die Bronzetafel aus Bogazkoy. Wiesbaden.

Peters, M. 1980. Attisch tpavua: griech. Etymologie und idg. Labiolaryngale. In Lautgeschichte und Etymologie, ed. M. Mayrhofer, O. E. Pfeiffer, and M. Peters. Pp. 328-51. Wiesbaden.

Pisani, V. 1969. Questioni ittite. Athenaeum 47:266-71.

Pokorny, J. 1959. Indogermanisches etymologisches Worterbuch. Bern and Munich.

Puhvel, J. 1971. Hittite hurkis and hurkel. Die Sprache 17:42-45.

-----. 1978. Remarks on 'two' in Hittite. KZ 92:98-107.

-----. 1979. Some Hittite etymologies. In Florilegium anatolicum: Melanges offerts a Emmanuel Laroche. Pp. 297-304. Paris.

-----. 1982. Review of Hethitisches Worterbuch, fasc. 4, by J. Friedrich and A. Kammenhuber. JAOS 102:177-79.

-----. 1984. Hittite etymological dictionary. Vols. 1 and 2. Berlin and New York.

-----. 1991. Hittite etymological dictionary. Vol. 3. Berlin and New York.

Ringe, D. A. 1984. Ionic ovovnueva. Glotta 62:45-56.

Schindler, J. 1969. Die idg. Worter fur 'Vogel' und 'Ei'. Die Sprache 15: 144-67.

-----. 1975. L'apophonie des themes indo-europeens en R/N. BSL 70:1-10.

Schmidt, G. 1985. Lat. amavi amasti und ihre idg. Grundlagen. Glotta 63:52-92.

Schmidt, V. 1968. Sprachliche Untersuchungen zu Herondas. Berlin.

Schwartz, M. 1966. Iranian *draw-'to lead astray'. JRAS 119-22.

Senn, A., and A. Salys. 1963. Worterbuch der litauischen Schriftsprache. Vol. 4. Heidelberg.

Sommer, F. 1932. Die Ahhijava-Urkunden. Munich.

Specht, F. 1936. Griechische Miszellen 6. Zur Vermeidung von Wortern mit drei kurzen Silben. KZ 63:207-26.

Steinmeyer, E. von, and E. Sievers. 1879-1922. Die althochdeutschen Glossen. 5 vols. Berlin.

Strunk, K. 1973. Methodisches und Sachliches zu den idg. Nasalinfixprasentien (ai. 7. Klasse). IF 78:51-74.

-----. 1979a. Anhaltspunkte fur unsprungliche Wurzelabstufung bei den idg. Nasalprasentien. IL 5:85-102.

-----. 1979b. Heth. huekzi, heth. hunikzi und die idg. Nasalinfixprasentien. In Hethitisch und Indogermanisch, ed. E. Neu and W. Meid. Pp. 237-56. Innsbruck. -----. 1984. Reflexions sur l'infixe nasal. In E. Benveniste aujourd'hui, vol. 2, ed. J. Taillardat, G. Lazard, and G. Serbat. Pp. 151-60. Paris.

Sturtevant, E. H. 1933. A comparative grammar of the Hittite language. Philadelphia.

Szemerenyi, O. 1978. On reconstruction in morphology. In Linguistic and literary studies in honor of Archibald A. Hill, vol. 3, ed. M. A. Jazayery, E. C. Polome, and W. Winter. Pp. 267-83. The Hague.

-----. 1985. Introduzione alia linguistica indeuropea. Milan.

Unal, A. 1978. Ein Orakeltext uber die Intrigen am hethitischen Hof. Texte der Hethiter, no. 6. Heidelberg.

van Brock, N. 1964. Les themes verbaux a redoublement du hittite et le verbe indo-europeen. RHA 75:119-65.

Villar, F. 1989. Los pronombres personales hetitas. Aula Orientalis 7:117-22.

Watkins, C. 1975. Die Vertretung der Laryngale in gewissen morphologischen Kategorien in den idg. Sprachen Anatoliens. In Flexion und Wortbildung, ed. H. Rix. Pp. 358-78. Wiesbaden.

Weitenberg, J. J. S. 1985. Review of Investigationes philologicae et comparativae. Gedenkschrift fur Heinz Kronasser, ed. E. Neu. BiOr 42:120-28. @ @
COPYRIGHT 1993 American Oriental Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Barton, Charles R.
Publication:The Journal of the American Oriental Society
Date:Oct 1, 1993
Words:9523
Previous Article:Myth and authenticity: deciphering the Chu Gong Ni Bell inscription.
Next Article:Some proposed emendations to the text of the Koran.
Topics:

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |