Universitt zu Kln

Institut fr Linguistik

Abteilung Historisch-Vergleichende Sprachwissenschaft

WS 2012/13

Materialien zum Blockseminar

(Selected) Topics in Anatolian Historical Grammar

von

Prof. Dr. H. Craig Melchert (UCLA)

Eine bersicht ber die einzelnen

Aufstze finden Sie ber die

Lesezeichenfunktion von Adobe Acrobat.

H. Craig Melchert

30th ECIEC

Harvard University

June 8, 2011

melchert@humnet.ucla.edu

[revised November 2012]

Hittite Auslautsgesetze Revisited

I. Absolute Final Vowels

A. Short Vowels

1. Unaccented *-i: generally preserved after both obstruents and sonorants.

a. In present active verb endings -mi, -si, -weni, -teni and adverbial kuwapi where;

when < *kw-bhi; also in D-LSg kti this < *k-dhi (Szemernyi 1956: 63 &

Georgiev 1971: 65) and contra Melchert (1994: 183) also in indefinite -kki < *-ki

(NB distribution kuiki, kuinki, kuitki, kuedanikkki with particle -kki only when last

vowel of base is -i-!). Compare Latin ci-tr on this side, Lith s this, etc.

b. Regularly lost in assibilated *-ts < *-ti (e.g. anz(a) in front < *h2nti = Latin ante etc. as per Jasanoff 1976: 126), but preserved before -ya also, and through late NH

(Abl -az but -azzi=ya) and also analogically restored in P3Sg and P3Pl active verbal

endings -zzi and -nzi (but NB rare regular outcome in z(a) is or iiyanz(a) they bind; see GrHL 11.8). Also preserved in infinitives in -wanzi < Abl-Inst *-wenti,

probably due to close nexus with following finite verb (so also Kloekhorst 2008:

957, but analogical addition of -i after finite forms unlikely). Specially conditioned

apocope in Imv2Sg t go! < *-dhi and -nut < *-n-dhi (NB shifted accent, as per

Kimball 1999: 192), as in Latin dc, fac etc. (see Weiss 2009: 147).

c. Indeterminate whether regularly lost in final *-ri: -i in medio-passive -ri secondary

after active present endings, so possible that Hittite r above, over = CLuvian

arri < *sri (but Hittite may equally well reflect endingless locative *sr, while

Luvian may be trivial innovation with productive ending). Unlikely with anaptyxis <

*sr (Oettinger 1993: 155) in view of (a)r < *sr-.

2. Accented *-: certainly preserved in monosyllabic k < *k; probably but not assuredly

in accented D-LSg -C < *-C (difficult to exclude entirely -C < DSg *-ei; see

below I.C.5).

3. Unaccented *-u: preserved in active third person imperative endings -tu and -ntu (and

spread from there to first person etc.). Also in conjunctions nu < *nu and u <

asseverative *h1su (thus with Zimmer 1994: 165 & Dunkel 1998: 57, given clear

association of u with preterite predicates, as per Weitenberg 1992: 327-8; notion of

u remade from *s and comparison with OIrish se, as per Watkins 1963: 14,

extremely unlikely). Conjunctions unaccented, as per Kloekhorst (pers. comm.). For

addition of clitics (postpositives) to unaccented elements see Agbayani and

Golston (2010) 6-7.

2

4. Accented *-u: no certain examples.

5. Unaccented *-e: contra Melchert (1994: 183-4) clear regular result is Hittite -a.

a. Enclitic 2Sg D-A -tta < PA *-te with regular shortening of unaccented PIE *t;

likewise with Josephson (1972: 419) Hittite -(a)ta by metanalysis < *-te = CLuvian

-tta and HLuvian /-ta/ (NB never rhotacized and never spelled with t; see Rieken

2008a: 640-1); also in synchronic allative apda < *-dhe parallel to kti < *-dhi

above. Also almost certainly in indefinite -kka (kulka, kuzka, kuika, kuiuka,

kuedaka) < *-ke, in view of OLatin -ce (conceivable but far less likely < *-ko, for

which there is no other evidence in use as a particle). See also I.A.10 below.

b. But lost either by apocope or assimilation in -kku < *-kwe (see for rare preserved sense

and Watkins 1985). NB that preserved -a in examples under 5.a could represent

[-], thus instead of apocope quite possible *-kwe > *-k

w > *-k

wu > -kku.

6. Accented *-: lengthened and raised to -, as in ImpfImv2Sg akkuk drink! and

azzikk eat! < *-sk. On adverb and prefix p see I.C.2 below!

7. Unaccented final *-o appears as -a: certain in medio-passive endings with regular loss

of final *-r (see below III.A.3) 3Sg -tta < *-to and 3Pl -nta < *-nto; also in anda

in(to) < *ndo and geminating -a also, and < *-h2o (*o required by Lycian -ke

also).

8. Accented final *- appears as - certainly in p(a)r < *pr (for Hittite p(a)r as

locatival before as well as allative forth see p(a)r and(ant)tar providence as per Puhvel 1991: 101 and duwn p(a)r long ago and renewed pran par ahead

(of time) in Melchert 2008).

9. Unaccented final *-a results in -a in Pret1Sg -a < *-h2e [Ha] and Pret2Sg -tta < *-th2e [tHa].

10. Accented final *-a: no good examples.

B. Long Vowels

1. Long * > (probably shortened when unaccented). Given syntax (occurs with

indicative), prohibitive negative l represents dissimilated *n attested in other

Anatolian languages (perhaps starting from *n=man would that not, as per

Oettinger 1994: 310). Derivation from Imv2Sg *leh1 (Kloekhorst 2008: 523 et al.)

leaves syntax unexplained. Pret3Pl -er < *-r.

2. Long * > a when unaccented, as in OH neuter n-stem N-APl arma bread rations

< *-m (Gertz 1982: 298-9) and N-APl plurale tantum karza swift < virtual *krts-

(Nussbaum, p.c., revising Rieken 1999: 391-2).

3. Long * in 2Sg subject *t required by /u:/ in 1Sg subject -uk (NB spelling with

precludes preform in *-uh2, since latter would lead to *u-uk /o:g/; see Kloekhorst

2008: 59 on a-a-u-u /a:sso:/ goods < *-uh2). But preform *tuh1 also possible (see

IV.C.2 below).

3

C. Diphthongs

1. Unaccented *-oi > -e securely in enclitic OH 3PDat -e < *-soi.

2. Accented *-oi > -e securely in pronominal AnimNPl -e < *-oi in k these, ap

those; they and also NeutN-APl k and ap, as per Jasanoff 2008: 145-8. Also

likely in p hin- < *poi, as per Eichner (1973: 78) contra Melchert (1994: 133). NB HLuvian pa-si-ya- throw away does not exist! Attested pa-sa-y-tu-' (KAYSER

18) highly unlikely to spell /passya-/. Preform *poi accounts for shape of pute/a- carry, bring and compatible with all other verbs with p-, including pai- go (but

all attested forms easily derivable from *pi-h1i- with accented preverb and zero-

grade root, contra Kloekhorst 2008: 616-8 with unbelievable spelling interpretations

and needless complex sound changes). New suggestion by Puhvel (2011: 191-2) that

Luvo-Hittite paku(wa)- forsake, abandon continues *pa-sk(e)u- (cf. Goth. Af-

skiuban idem) not nece. Incompatible: could show variant *po-. NB still no certain

Hittite-Luvian equations. In any case, no evidence for *pe-.

3. Unaccented *-ei probably results regularly in OH P3Sg -e in war(a)e plucks,

harvests < virtual *wrsei (see also OH/OS mazze resists), but evidence is less

than robust.

4. Accented *-ei: no sure examples.

5. Unaccented *-ai > Hittite -e certainly in OH Pres1Sg -e < *-h2ei [-Hai].

6. Accented *-ai: no sure examples.

7. Accented *-i in amphikinetic neuter nouns in -i with shifted accent, as in ati bone; bones < *h2esth2i etc.

8. Accented *-i > -(i) in kuli fallow land < virtual *kwlh1-i < *k

wlh1-o-

(see

Oettinger 1995: 211-14) and in utn land < *ud-ni < *wdeno- (see Oettinger

2000: 185ff.). Derivation of NeutN-APl kue < *kwi (Melchert 2004: 140) possible,

but *kwi as in k and ap (see above I.C.2) more likely.

II. Interior Vowels in Final Syllables

A. Short Vowels

1. Short *i appears as i whether accented or unaccented: kui, kuin, kuit and likewise

alli, allin, alli etc.

2. Short *u appears as u whether accented or unaccented: g(a)nut by the knee < virtual

*gn-t and dau, daun, dau.

3. Accented short *e > in Inst kt < *kd (Melchert-Oettinger 2009: 60) and in Pret1Pl

and Pret2Pl -wen (NB exx. such as , -ten < *-wn, *-tn (Yoshida 1997: 191-2).

4. Unaccented short *-en# > -an in supine -kewan (Yoshida 1997: 192) and local

particle -an < cliticized *en. Unaccented *-es# > -i in npi sky, heaven <

*nbhes (perh. better example is audit- new-born (animal) < *sm-wetes-t- *of

the same year.

5. Accented short *o > in AnimNSg k, ap and NeutN-ASg apt. Accented *-m#

> /-o:n/ (spelled -u-un) in AnimASg kn, apn (Kloekhorst 2008: 99, revising

4

Melchert 1994: 186-7). NB makes very problematic derivation of mn how; as; if,

when < *mm and equation with Lycian m as; thus! Dubious derivation < Inst *mh2m [mm] by Dunkel (1997: 72-4).

6. Unaccented short *o > a in NeutN-ASg -an < *-om, as in pdan place < *pdom

(NB iugan yoke < *yugm must then be analogical in either vocalism or accent)

and in Pret1Sg *-an < *-om, remade as -an-un.

7. Accented *-ms# > /-o:s/ (in ku-u-u and a-pu-u-u), as per Kloekhorst (2008: 99),

revising Melchert (1994: 186-7). Fate of unaccented *-oms (and *- s) remains less

than certain. Kloekhorst (2008: 56-7) assumes --u is regular spelling for AnimAPl

in OH, thus /-us/, replaced by /-os/ in NH, based on later spellings with -u-u. But

NB that neither *-oms nor *- s should ever have been spelled plene! Of 10 exx. of

--u cited by Kloekhorst, 8 (including all OS and MS) are of type ta-lu-ga--u,

ar-a--u, kap-p--u, entirely parallel to ta-lu-ga-e-e, ap-p-ez-zi-e-e (cf. a-an-te-ez-zi--u missed by Kloekhorst), argued convincingly by Kloekhorst

(forthcoming) to reflect use of -e- to mark [y] after -Ci- sign. Thus use of surely

parallel, to mark hiatus-filler ([] or [w]), not vowel quality. Change to use of u

likely to align with real quality of following vowel, not OH to NH sound change, as

per Kloekhorst (see below III.B.1 for similar change with /-on/). Clinching evidence

wrongly dismissed by Kloekhorst (2008: 168): APl a-ku-u-u-=a also sea-shells

(OH/OS) to u-stem aku-! Shows even *-ums > *-s > *-s > /-o:s/ (shortened when

unaccented). Thus in all likelihood also same result for *-oms and *- s.

B. Long Vowels

1. Accented long * > in hysterokinetic collectives in *-r (still likely adur brushwood, twig; splintersee Rieken 1999: 346-7and certain in *appr transactions, business required by appiriya- *market-place > cityRieken 1999: 317). Also in AnimNSg atrz(a) star < *h2str+s.

2. Accented long * > in amphikinetic plurals such as widr waters < *wedr (with

shifted accent that cannot be due to the kwetwores-Regel!). Shortened to a when

unaccented: OH/MH NSg antuwaa human < virtual *n-dhweh2-s.

3. Accented long * > in duwn par long before < *[dwa:m] (by Stangs Law from

underlying ASg *dwh2m distance, long time). See Melchert 2008.

4. Unaccented long * > a in NSg kear(=i) < *ghsr hand.

5. Unaccented long * > a probable in pran in front < *prm ~ Grk. across,

opposite with Dunkel 1997: 72 contra Melchert 1994: 135.

C. Diphthongs

Long *-is with secondary accent > amphikinetic NSg -i (but ASg -in unlikely to be

regular continuant of *-y- ). PIE status of rare matching -u < *-us uncertain.

III. Sonorants

A. Final Non-Syllabic

1. Final *-m merges with final *-n as -n (see various examples cited under II. above).

Date of merger (Common Anatolian or pre-Hittite) difficult to determine.

5

2. Final *-Vr# > -Vr: N-APl *wedr with accent shift > widr and Pres3Sg of type *tuk-

r > *tukkr tuqqri is visible; matters (Yoshida 1990: 112-4).

3. Final *-Vr# > -V: N-APl *-Vr > -V in partawa wings, attata pieces of wisdom etc. (Melchert 1988), Pres3Sg of type *ki-tor > kitta lies (Yoshida 1990: 112-4).

4. Final *-l: no sure examples. uw/l thread could represent hysterokinetic

collective *suh1-l (after Eichner 1973: 54), but more likely substantivized

neuter*suh1-lom with syncope, as is certain for similar cases (Rieken 2008b: 249).

See also Rieken 2008b: 250 for pronominal GSg in -l < syncopated *-los/m, contra

Oettinger 1999: 264.

B. Final Syllabic

1. Final *- > -un, as in athematic Pret1Sg. Probably /-om/, based on sporadic NS

spellings pa-(a)-u-un and e-ep-pu-u-un (thus with Kloekhorst 2008: 609), but plene

spelling can hardly represent vowel length, and MH/MS shows rather pa-a--un

(3x). Thus use of --/-u- not true plene, but to mark transition sound in hiatus

/pa:_on/ < *pi-h1y- . Cf. II.A.7 above on AnimAPl ending.

2. Other final *- > -aR (via *-oRsee immediately below). NB development must be

ordered after loss of final *-r after unaccented vowel in III.A.3 above! Examples for

*- # in N-ASg wtar water, uttar word etc. < *- and -tar < *-t (types in -mar

and -ear are secondary Hittite creations). Also in Pret3Pl -ar < *- such as

apaiyar spied (pace Kloekhorst 2008: 244). Result of post-vocalic *-w is

likewise -war, as in generalized N-ASg of verbal noun. Examples for *- in N-ASg

-man < *-m (lman name < *h1nh3-m etc.) and -an < secondary *- (lgan

inclination etc.): for latter see Melchert 2007-2008[2010]. Probable *- in N-ASg of

deverbatives in -al (ard-al saw, iiyal- binding etc.). Syncopated *-lom less likely in view of unsyncopated -la- < *--lo- in adjectives (then animate agent-

nouns) such as genzuwala- merciful (see Rieken 2008b: 251).

3. Final *-(C)w > *-CwoR > -CuR at least for liquids. Result surely /-CoR/, with

regular lowering of *o > a as in regular *- > -aR blocked by preceding *w, which

is then lost by dissimilation. Not unmotivated resyllabification of *-w > -uR

(Melchert 1994: 132): e.g. enkur offering. For interpretation as /-CoR/ NB spellings of a-ni-u-ur with analogical postconsonantal treatment of *-Cw and

secondary accent, for which see examples like pt-tu-u-la- snare, hypostasized from

genitive *that of spreading out to virtual *pth2-w , *peth2-wn-s *spreading out

(thus w/ Rieken 1999: 472 after Puhvel 1979: 211, pace Kloekhorst 2008: 681). NB

cannot belong to type of syncopated *--los/m (Rieken 2008b: 252-3), since latter

show consistent expected /u:/, as in i-i--ul treaty. Examples for *-w # lacking.

C. Interior Syllabic

For final *- s# in AnimAPl ending see II.A.7 above.

IV. Obstruents

A. Final Stops

1. Final *-t (after voiceless obstruent) preserved by addition of prop-vowel in mi-

conjugation (where NB postvocalic ending was preserved as per preceding): Pret3Sg

6

ta was < *h1s-t. Generalized after all obstruents: NB Pret3Sg e-ku-ut-ta drank

/e:wta/. See Yoshida 1993: 23 with refs.

2. Final *-d (after V and R) preserved as -d: pronominal NeutN-ASg apt, -at etc. < *-od,

endingless locative iwat day, Inst < Abl *-d in kiart(a) hand, wedand(a)

water, kt this (NB for original ablatival sense kitpantalaz from this time on and

suppletive enclitic possessives in -it for ablative). See Melchert-Oettinger 2009: 54.

Loss of final stop in NeutN-ASg of stems in *-nt probably already PIE. Final *-g in

-ug *g after loss of final *-h2 also kept (see below IV.C.4). Contra Melchert

1994: 85 generalization of voiced stop here probably PIE: Goddard 2007: 122-3 w/ refs.

3. Whether i-conjugation Pret3Sg - continues *-s-t with regular loss of *-t or simply *-s cannot be determined on phonological grounds.

4. No sure examples for original final labial stop. Preservation of final stops of all points

of articulation in Imv2Sg may be due to trivial analogical restoration/maintenance:

p take!, ark hold!, etc. NB eku drink! with labiovelar /:gw/ vs. l pour! below.

B. Final *-s

1. Regularly preserved after vowel: see various nominal cases endings and OH/MS

Pret2Sg ending -V.

2. Also maintained in AnimNSg of stems in dental stop and *-nt-: karz to kard-

innards, waz day *dyw-ot-s, -anz(a) < *-e/ont-s, etc.

3. Maintained after other stops and -- by anaptyxis, especially i-conjugation Pret3Sg in -i (arguably after loss of final *-t in *-s-tsee IV.A.3 above): akki

died, uwappi hurled, newaa renewed, etc. and with analogical vocalism also maniyai handed over etc. See Melchert 1994: 174 with refs. Also surely analogical in MH atte pricked (Oettinger 1993: 155). Also after *h1 ([] or [h]) in N-ASg i mouth < *(h1)h1(-)s (Melchert 2010).

4. NB lost in secondary final sequences *-Car/ls < *-Cr/los with syncope and new

anaptyxis: AnimNSg uppar bowl < virtual *h2pros, AnimNSg arnikzil restitution < *-tlos (see Melchert 1993 and significant revisions by Rieken 2008b:

246-50). Suggests Imv2Sg examples such as kar cut! may show analogical

restoration/maintenance.

C. Final Laryngeals

1. At least *h2 regularly lost after vowel, via intermediate step of lenition: NB with

Eichner (1973: 59) miyauwant- mature (single --!) reflecting *mya < *m(h1)y-h2 growth. Geminate -- of mnaanda facing, opposite < allative *mneh2 plus anda into (Nikolaev 2010: 68) due to initial *h1 in adverb?? I.e., virtual

*mneh2 h1endo? Next stage of loss of lenited laryngeal with compensatory

lengthening indirectly attested in -tar < *-t (NB with regular lenition after

accented long vowel, contra Melchert 1994: 69; see Melchert 2007: 2-3 after

Jasanoff) and -war < *-w . Also directly in lexicalized goods (NB spelled a-

a-u-u, thus /a:sso:/ with lowering of /u:/ before *h2: see Rieken 2005: 537 with

refs. and Kloekhorst 2008: 59) and OH substantival mekk a lot < virtual *mgh2ih2.

Otherwise final *-Vh2 represented by short vowel (-a, OH and MH -i and -u, as per

7

Watkins 1982: 282-3, pace Prins 1997: 209-11 and passim, et al.). Absence of any

direct lexicalized - < *-eh2 more likely accidental than due to difference in

chronology of shortening of - vs. - and - (see Watkins 1982: 281).

2. Final *-Vh1 probably reflected in neuter i-stems reflecting old duals: e.g. GI

lzi

(balance) scales < *-t < *-tih1 (see Puhvel 1984: 270 and Rieken 1994: 52, who

cites further plausible examples). Far less certain is -a (synchronically interpreted as

collective plural) < thematic *-oh1. Other examples cited by Puhvel (1969: 61) are

extremely dubious, but kuwa eyes may indirectly continue *-oh1 taken from

thematic examples (so cautiously Watkins 1986: 6133

). Compare IV.C.4 below. 2Sg

subject *t in zk could represent *tu-h1 (Schrijver 1999: 157), but real *t with

monosyllabic lengthening also possible. See I.B.3 above for impossibility of preform

*tuh2.

3. No clear examples for final *-Vh3. Imv2Sg d probably reflects regularly *dh3 (or

alternatively *dh3), but generalization of dominant allomorph d- from rest of

paradigm cannot be excluded.

4. At least *h2 regularly lost after stop: OH NeutN-ASg mk much < *mgh2, OH/OS

CollPl aniat=et his regalia < *-ot-h2 (assumption of spelling error entirely

gratuitous), 1P subject g I *g < *gh2.

5. Final *-h2 also lost after sonorant in ke/itkar at the head < *kd here + endingless

locative *krh2 (see Nussbaum 1986: 96ff.). Preservation of *h1 or *h3 in same

environment implausible. Imv2Sg wal strike! < *welh3- shows trivial restoration/ preservation just like , an (a-an-a/a-na-a) seek, search!< *senh2-.

8

References

Agbayani, Brian, and Chris Golston. 2010. Second-position is first-position:

Wackernagels Law and the role of clausal conjunction. IF 115.1-21.

Dunkel, George. 1997. B. Delbrck and the Instrumental-Ablative in *-m. In E. Crespo &

J. L. Garca Ramn (eds.), Berthold Delbrck y la sintaxis indoeuropea hoy, 63-83.

Wiesbaden: Reichert.

. 2007. Chips from an Aptotologists Workshop I. In A. Nussbaum (ed.), Verba

Docenti. Studies in historical and Indo-European linguistics presented to Jay H.

Jasanoff by students, colleagues, and friends, 53-61. Ann Arbor/New York: Beech

Stave.

Eichner, Heiner. 1973. Die Etymologie von heth. mehur. MSS 31.53-107.

Georgiev, Vladimir. 1971. Die Herkunft der hethitisch-luwischen Dativ-Lokativendungen

des Singulars. IF 76.59-65.

Gertz, Janet. 1982. The Nominative-Accusative Neuter Plural in Anatolian. New Haven:

Yale University Ph.D. dissertation.

Goddard, Ives. 2007. Phonetically Unmotivated Sound Change. In A. Nussbaum (ed.),

Verba docenti. Studies in Historical and Indo-European Linguistics Presented to Jay

H. Jasanoff , 115-30. Ann Arbor/New York: Beech Stave.

GrHL = Hoffner, Harry A. Jr., and H. Craig Melchert. 2008. A Grammar of the Hittite

Language. Part I: Reference Grammar. Winona Lake: Eisenbrauns.

Jasanoff, Jay. 1976. Grk. , Lat.amb et le mot indo-europen pour lun et lautre.

BSL 81/1.123-31.

. 2008. *-bhi, *-bhis, *-is: following the trail of the PIE instrumental plural. In: J.

Rasmussen & T. Olander (eds.), Internal Reconstruction in Indo-European:

Methods, results, and problems. Section Papers from the XVI International

Conference on Historical Linguistics. University of Copenhagen, 11th-15th August,

2003, 137-149. Copenhagen: Museum Tusculanum.

Josephson, Folke. 1972. The Function of the Sentence Particles in Old and Middle

Hittite. Uppsala: Skriv Service AB.

Kimball, Sara. 1999. Hittite Historical Phonology. Innsbruck: Institut fr

Sprachwissenschaft der Universitt Innsbruck.

Kloekhorst, Alwin .2008. Etymological Dictionary of the Hittite Inherited Lexicon.

Leiden/Boston: Brill.

. forthcoming. The Phonological Interpretation of Plene and Non-Plene Spelled e in

Hittite. In B. Nielsen et al. (eds.), The Sound of Indo-European. Selected papers from

the conference held in Copenhagen 16-19 April, 2009.

Melchert, H. Craig. 1988. Final -r in Hittite. In Y. Arbeitman (ed.), A Linguistic

Happening in Memory of Ben Schwarz, 215-234. Louvain-la-Neuve: Peeters.

. 1993. A New Anatolian Law of Finals. Journal of Ancient Civilizations 8.105-13.

. 1994. Anatolian Historical Phonology. Leiden/Atlanta: Rodopi.

. 2004. Second Thoughts on *y and *h2 in Lydian. In M. Mazoyer & O. Casabonne

(eds.), Studia Anatolica et Varia. Mlanges offerts Professeur Ren Lebrun.

Volume II, 139-50. Paris: LHarmattan.

. 2007. Luvian Evidence for PIE *h3eit- take along; fetch. Indo-European Studies

Bulletin, UCLA 12/1.1-3.

9

. 2007-2008 [2010]. Neuter Stems with Suffix *-(e)n- in Anatolian and Proto-Indo-

European. Die Sprache 47.163-181.

. 2008. Hittite duwn (par). In C. Bowern, B. Evans & L. Miceli (eds.),

Morphology and Language History in honour of Harold Koch, 201-9. Amsterdam/

Philadelphia: Benjamins.

. 2010. The Word for mouth in Hittite and Proto-Indo-European. International

Journal of Diachronic Linguistics 7.55-63.

Melchert, H. Craig, and Norbert Oettinger. 2009. Ablativ und Instrumental im

Hethitischen und Indogermanischen. Ein Beitrag zur relativen Chronologie. IncLing

32.53-73.

Nikolaev, Alexander. 2010. Hittite menaanda. JAOS 130.63-71. Nussbaum, Alan. 1986. Head and Horn in Indo-European. Berlin/New York: de Gruyter.

Oettinger, Norbert. 1993. Mittelhethitisch hattes sie schlachtete. Linguistica 32.153-5

(Bojan op septuagenario in honorem oblate).

. 1994. Etymologisch unerwarteter Nasal im Hethitischen. In J. Rasmussen (ed.), In

honorem Holger Pedersen. Kolloquium der Indogermanischen Gesellschaft vom 26.

bis 28. Mrz 1993 in Kopenhagen, 307-30. Wiesbaden: Reichert.

. 1995. Griech. , heth. kuli und ein neues Kollektivsuffix. In H. Hettrich et

al. (eds.), Verba et structurae. Festschrift fr Klaus Strunk zum 65. Geburtstag, 211-

28. Innsbruck: Institut fr Sprachwissenschaft der Universitt Innsbruck.

1999. Zum nordwest-indogermanischen Lexikon (mit einer Bemerkung zum hethitischen

Genitiv auf -l). In P. Anreiter & E. Jerem (eds.), Studia Celtica et Indogermanica.

Festschrift fr Wolfgang Meid zum 70. Geburtstag, 261-7.

. 2000. Heth. udn, armen. getin Land und lyk. wedre/i-. In A. Hintze et al. (ed.),

Anusantatyai. Festschrift fr Johanna Narten, 181-7. Dettelbach: Rll.

Prins, Anna. 1997. Hittite Neuter Singular Neuter Plural. Some Evidence for a

Connection. Leiden: Research School CNWS.

Puvhel, Jaan. 1979. Hittite words with initial pt/pt sign. In E. Neu & W. Meid (eds.),

Hethitisch und Indogermanisch,209-17. Innsbruck: Institut fr Sprachwissenschaft

der Universitt Innsbruck.

. 1969. Hittite anna iwaz. KZ 83.59-63.

. 1984. Hittite Etymological Dictionary.Vol. 1 Words beginning with A. Vol. 2

Words beginning with E and I. Berlin/New York/Amsterdam: Mouton.

. 1991. Hittite Etymological Dictionary. Volume 3: Words beginning with H.

Berlin/New York: Mouton de Gruyter.

Rieken, Elisabeth. 1994. Der Wechsel -a-/-i- in der Stammbildung des hethitischen

Nomens. HS 107.42-53.

. 1999. Untersuchungen zur nominal Stammbilding des Hethitischen (StBoT 44).

Wiesbaden: Harrassowitz.

. 2005. Zur Wiedergabe von hethitisch /o/. In G. Meiser & O. Hackstein (eds.),

Sprachkontakt und Sprachwandel. Akten der XI. Fachtagung der Indogermanischen

Gesellschaft., 17.-23. September 2000, Halle an der Saale, 537-49. Wiesbaden:

Reichert.

. 2008a. Die Zeichen , , und in den hieroglyphisch-luwischen

Inschriften der Nachgroreichszeit. SMEA 50.637-47.

10

. 2008. The Origin of the -l Genitive and the History of the Stems in -l- and -l- in

Hittite. In: K. Jones-Bley et al. (eds.), Proceedings of the 19th UCLA Indo-European

Conference. Los Angeles, November 3-4, 2007, 239-56. Washington DC: Institute for

the Study of Man.

Schrijver, Peter. 1999. Vedic g bhti, g bhyti and the semantics of *ye-derivatives of nasal presents. MSS 59. 115-62.

Szemernyi, Oswald. 1956. Hittite pronominal inflection and the development of syllabic

liquids and nasals. KZ 73.57-80.

Watkins, Calvert. 1963. Preliminaries to a historical and comparative analysis of the

syntax of the Old Irish verb. Celtica 6.1-49.

. 1982. Notes on the Plural Formations of Hittite Neuters. In E. Neu (ed.),

Investigationes Philologicae et Comparativae. Gedenkschrift H. Kronasser, 250-62.

Wiesbaden: Harrassowitz.

. 1985. Indo-European *-kwe and in Hittite. In H. M. lberg & G. Schmidt (eds.),

Sprachwissenschaftliche Forschungen. Festschrift fr Johann Knobloch, 491-7.

Innsbruck: Institut fr Sprachwissenschaft der Universitt Innsbruck.

. 1986. The language of the Trojans. In M. Mellink (ed.), Troy and the Trojan War.

A Symposium Held at Bryn Mawr College, October 1984, 45-62. Bryn Mawr: Bryn

Mawr College.

Weitenberg, Jos. 1992. The Use of Asyndesis and Particles in Old Hittite Simple

Sentences. In O. Carruba (ed.), Per una grammatical ittita, 305-53. Pavia: Iuculano.

Yoshida, Kazuhiko. 1990. The Hittite Mediopassive Endings in -ri. Berlin/New York: de

Gruyter.

. 1993. Notes on the Prehistory of Preterite Verbal Endings in Anatolian. HS 106.26-35.

. 1997. A Further Remark on the Hittite Verbal Endings 1 pl. -wani and 2 pl. -tani.

In D. Q. Adams (ed.), Festschrift for Eric Hamp. Volume II, 198-93. Washington

DC: Institute for the Study of Man.

Zimmer, Stefan. 1994. Indogermanisch *h1su und *dus- im Kymrischen. ZCP 47.176-

200.

Luvo-Lycian Dorsal Stops Revisited

H. Craig Melchert University of California, Los Angeles

1. Introduction There is a broad consensus that at some prehistoric stage the Indo-European stop system had only a two-way phonological contrast of labialized velars (traditionally labiovelars) and non-labialized velars: thus schematically /Kw/ versus /K/. The latter had two conditioned allophones, palato-velars or front velars (traditionally palatals) in fronting environments and non-front velars (traditionally velars) elsewhere: [Kj] and [K]. Formulations of just what constituted a fronting environment can differ markedly (one may compare among many Meillet 1934: 94 and Lipp 2009: 1.7&53). Nevertheless, the highly skewedthat is, non-random and largely complementarydistribution of traditional palatals and velars strongly supports this fundamental premise.

The major point of dispute for more than a century has been whether or not a partial loss of conditioning environment, analogy, and other factors led to a so-called phonemic split in PIE (that prehistoric stage that we reach by direct comparative reconstruction). The result would have been a three-way contrast /Kj/, /K/ and /Kw/, albeit very likely in a limited number of examples and with much of the previous complementary distribution preserved, as is typical for such splits. I contend that the answer to this question is yes, regardless of the facts of Luvo-Lycian (likewise Kmmel 2007: 312).

The typological objection of Sihler (1995: 152-3) and Lipp (2009: 1.3165)

that the presupposed backing of palatal stops to velars in the centum languages is unattested is a straw man. There is no compelling evidence that the PIE stops conventionally labeled palatals were true palatal stops (IPA [c] and []). Evidence from all attested languages is fully compatible with front velars.1 There is in any case no necessity to assume that the phonological merger of /Kj/ and /K/ in the centum languages even took place by phonetic backing of front velars. Since the place of articulation of /K/ in the older centum languages is indeterminate, we may rather assume the more likely generalization of [Kj] in fronting environments (along with other factors leading to an allophonic distribution).

1 Compare Kmmel 2007: 318. I also regard as fully viable his alternative solution (2007: 318-27) of an original contrast of velar, uvular, and labialized velar stops.

2

No explanation is offered by Sihler, Lipp or others for the contrasting data cited in the appendix below. Constructs such as *kreuh2- beside *kruh2- for gore, bloody flesh (with a supposed front allophone before *r + front vowel) or *kew-on- dog (with wholly unattested root full grade!) as a variant of the root generally cited as *(s)keuh1- perceive, notice (Lipp 2009: 1.55&79) are plausible enough, but they are totally unsupported for Proto-Indo-European as reached by direct comparative reconstruction. They are rather the result of internal reconstruction based on PIE and thus by definition represent pre-PIE.2 2. Luvo-Lycian Evidence I The question at hand is thus not whether PIE had a three-way phonemic contrast in its dorsal stops, but whether Luvian and Lycian truly offer further support for that contrast. In Melchert 1987 I claimed that Luvian reflexes of voiceless dorsal stops show a triple contrast before front vowels. Morpurgo Davies and Hawkins 1988 independently reached the same conclusion (as noted in the epilogue to my article, Warren Cowgill had already made this analysis many years before, on the basis of far less evidence). The issue was left open as to whether this contrast was unconditioned or reflected a conditioned palatalization of *k but not *k. However, in Melchert 1989: 23-31 (see especially page 30), I argued for an unconditioned three-way contrast based on further putative evidence from Lycian. This analysis had the important implication that Anatolian had merged neither the PIE labiovelars and velars nor the velars and palatals. It was thus neither centum nor satem, but was, as often, more archaic. This formulation has met wide acceptance (e.g. Fortson 2004: 168, Kloekhorst 2008: 17-18), but occasional rejection (Sihler 1995: 154, Woodhouse 1998, Lipp 2009: 1.275-302).

The crucial Lycian evidence I cited then for *k > ts before *o (i.e., in a non-fronting environment) has not withstood scrutiny. Contra Melchert 1989: 29-30 the suffixes Lycian -(i)s- and Luvian -(i)zza- are not cognate, reflecting *-i(s)ko-. Lycian -(i)s- forms abstracts and names for topographic features: e.g., wazzis- __ship (an office) or Trmis- Lycia (with Borchhardt-Eichner 1997-99: 82-3).3 The true equation is Luvian -zza- = Lycian -za- < *-tyeh2: e.g., 2 Sihler (1995: 154) dismisses such evidence as an artifact of the method. This is methodologically unacceptable: by definition all of PIE is an artifact of the comparative method. One may always legitimately argue that the method has been incorrectly applied in a given case (and that the results are thus false), but one cannot accept other results of the comparative method on a large scale, as Sihler does, and airily dismiss those results that do not happen to conform to ones preconceptions of how some part of the grammar should look. The data in the appendix is based on quite indisputable etymologies. 3 Lycian -(i)s- in my view continues *-sh2o- (cf. Luvo-Hittite -(a)a- in tariyaa- exertion, exhaustion, luliyaa- marshland, etc.), but this is an entirely separate issue. The key point is that it does not match Luvian -(i)zza-, as I had mistakenly claimed.

3

CLuvian waazza- sanctified, holy = Lycian wasaza- (kind of priest), HLuvian ku-ma-za- = Lycian kumaza- priest, as seen by Hajnal (2003: 193). The HLuvian ethnic adjectives seen in /Karkamisa/itsa-/ of Carchemish and /nimu:witsa-/ child (i.e. *of /nimu:wi-/ un-manhood) likewise show the same PIE suffix *-tyo- seen in Lycian Pttaraze/i- of Patara and neleze/i- of the agora (for the Lycian suffix see Gusmani 1961 and Hajnal 1995: 82 with n. 96). With removal of the supposed Lycian evidence showing *k > ts before back vowel, my claim of an unconditioned three-way contrast in Luvo-Lycian reflexes of voiceless dorsal stops cannot be upheld. The purported unconditioned palatalized treatment of voiceless *k had always been suspect, due to the appearance of voiced *g(h)- as Luvo-Lycian k- before back vowels (see below), as per the apposite remarks of Woodhouse (1998: 40) and Lipp (2009: 1.297). 3. Luvo-Lycian Evidence II (voiceless stops) We must reexamine all available evidence anew. Data for the voiceless stops is summarized in Table 1:4

Table 1: Summary of Evidence for the Voiceless Series (C=CLuvian; H=HLuvan)

*k *k *kw

ziyari lies (C) ~ Lyc. sijni ki(i)- (C) to comb kui- who (C) ~ Lyc. ti < *kyo- < *ke s- OCS esati etc. < *kwi- zr-(za)/zart- heart (C/H) kar- (C) to cut kuar-/kur- to cut (C) *kr/*kr d- *k(e)rs- Toch krst-/krst- *kw(e)r- Skt. kti etc. -za- (iter.) (C/H) ~ Lyc. -s- Lyc. tukedre/i statue < *-ske/o- < *twek- Skt. tvc- skin za/i- this (C/H) kup(iya)- to plot (C) < *ko/i- Lith. s etc. < *kup- Skt. kpyati desires Lyc. ti-se some-/anyone Lyc. xupa- grave-house < *-ke ~ Hitt. -kka < virtual *kupeh2- ~ Gmc. in kulka etc. *xufa- house; hill (Schaffner 2001: 143)

azu(wa)- horse (H) ~ Lyc. esbe- HLuv. kat- fight in katunas < *kwo- (Morpurgo Davies 1986: 132-3) zuwan(i)- dog (H) ~ Hitt. kattu- weapon, talon < *kwon- (thus contra Puhvel 1997: 138) zurnid- horn (H) ~ OIr. cath fight, Gmc. *hau-, < *kr ng-id- ~ Hitt. karkid-ant- OCS kotora fight, Germ. Hader & Skt. rnga- < *kat- (Starke 1990: 406-7; Puhvel 1997: 89) wazi- request (H) < *wek-ye/o-5 4 Examples listed above the solid horizontal line in each column I regard as indisputable etymologies, while those below it are merely probable or possible.

4

zarwaniya- of horn (C) < *kerw+ zalma- shield, protection < *kelmn-o- (Melchert 1988: 241-3; but cf. Lipp 2009: 1.27533 for an alternative Hurrian interpretation) zalla- trot (C), /zallal-/ vehicle (H) < *(s)kel- Lith. olis jump (Rieken 1997)

/ziralamma/i-/ fertile (H) < *krh1o- Lat. pr-crus (Rieken 2003: 45-50)

versus

kattawatnalli- spiteful (C) & Hitt. kadduw(i)- become aggrieved (Puhvel loc. cit.), + HLuv. kata- enmity in ka-ti-i CRUS (L. Younger, pers. comm. vs. Morpurgo Davies 1986: 132-3) ~ Grk. spite, anger, Skt. tru- enemy < *kot-. Separate from kattu- fight contra Morpurgo Davies (1986: 144), Melchert (1987: 189-90), Puhvel (1997: 140), Woudhuizen (2010: 44)6 Runtiya- < K(u)runtiya- < *kru-nt- horned (Watkins 1999: 15-18, Woudhuizen 2010: 44)7

kumma- sacralized (C/H) ~ Lyc. *kuma-

5

[Less likely but possible: kattu- talon ts are either in exclusively fronting environments or in paradigms with fronting environments Second, the number of examples cited for *k > ts is immaterial; the crucial example is that of ki(i)- to comb that shows *k > k in an exclusively fronting environment (more on this word below). Third, a dispassionate review of all the relevant evidence includes some very credible examples of Luvo-Lycian k < *k in non-fronting environments.

Current available evidence therefore points to a conditioned palatalization of *k before front vowel (*i, *e, and *), yod, and *w (not u!). For *w as a fronting environment note the virtual complete absence of PIE sequences in *Kw (versus the well attested *Kw) and the strong tendency of labiovelars to be palatalized (Hock 2009). HLuv. zurnid- horn < *kr ng-id- suggests that also *kR - > *kjR- > zVR- (see below on zanta down < *km/n t-).

Several points call for further comment. The existence of the stem ki- to

comb in Hittite (P3Sg kizi) suggests that Anatolian inherited a root present *ks-ti, whose weak stem *ks-(nti) might in principle have provided a non-fronting environment for Luvian ki(i)-. However, whereas Hittite trivially leveled the strong stem,10 one cannot possibly derive the Luvian verb in -(i)- starting from the root present. Prehistoric *ks-/ks- could have led only to a Luvian *kas+, leaving the i-vocalism of ki(i)- wholly unexplained. One can derive the latter with op (1970: 93) and Melchert (1994: 152) from a virtual *ksh2-ye/o- with lengthened grade like Latin clre, and this is supported by CLuvian il(i)- to wash, whose first vowel can only reflect a long *. Less likely but not to be entirely excluded is *kes-h2-ye/o- with a short root vowel (if Luvian really does share with Hittite the change of pretonic short *e to i, for which see Melchert 1994: 240, with due caution). In either case, the Luvian verb

My own attempt to take it as a personal name (2004a: 97) was an act of desperation also effectively refuted by the context. The position of keruti in its clause (following ql=ebi, that is, ql(a)=ebi this here precinct) suggests rather that it names a recipient of the sacrifice. Given the prominence of the more common name for the Stag-god in western Anatolia (reflexes of /Runtiya-/), such a recipient does not seem to me impossible. My hesitation rests mainly on the less than clear morphology. In view of the tendency of to be lost in Luvian in a sequence *-Rh2w- (Melchert 1994 : 258), it is not impossible that Lycian keru- reflects a virtual *korh2wV-, i.e. that its base contained a laryngeal like that of Karhuha (see note 6). 10 Since the Hittite verb is attested all of twice, only in New Hittite manuscripts, its i-vocalism before an s may easily be secondary and requires no special explanation (contra Kloekhorst 2008: 482). On the effective merger of /i/ and /e/ before /ss/ in NH verbs see already Melchert 1984: 147-9.

6

requires a preform with fixed mid-front vowel, and the non-affrication of the initial *k retains its force.

Contra Lipp (2009: 1.273-5, 278 with note 36, 298) the reading z for hieroglyphic sign 448 is assured. Despite the protestations of Hawkins (2000: 36) et al., the alleged interlocking evidence for a value s does not exist. The fact that sign 108 (CORNU), the drawing of a horn, has a phonetic value s proves absolutely nothing, since as Hawkins himself concedes, we find in KARATEPE the word CORNU+RA/Isu-ra/i- meaning plenty, abundance (where the use of horn as the determinative reflects the notion of a cornucopia). It is quite clear that the value s for sign 108 is acrophonic to this word, which is based on the verbal root su- fill and has nothing to do with the word for horn. If on the other hand the word for horn truly began with /su-/, it is completely inexplicable why it was not spelled with the transparent 108 sign for horn, but rather with the opaque 448: (CORNU)z+ra/i-ni. Further confirmation for the value z is furnished by z+ra/i-wa-ni-(URBS) of Tyre (i.e. r) (thus Lipiski 2004: 115 and Yakubovich 2010: 66-758), with a predictable rendering of Semitic emphatic as an affricate ts (see the references given by Lipp 2009: 1.310).

The three HLuvian words for horse, dog, and horn thus cannot be

loanwords from Indic via Hurrian, since they show an affricate /ts/, not a sibilant /s/. The morphological match of HLuvian /zurnid-/ and Hitt. karkid- in karkidant- horned also precludes borrowing. While the suffix -id- is very productive in Luvian, including in loanwords, it is decidedly not so in Hittite, and the word equation can only reflect a common preform *kng-id-, where a further suffix has been added to the base seen in Sanskrit ga- (see for this analysis of the Hittite already Puhvel 1997: 89). Hittite, which retained the stop in what was now a sequence *-arngi-, deleted the nasal. Loss of the stop in Luvian (see below) allowed the nasal to remain. As discomfiting as it may be, sporadic uR < * beside aR in Anatolian is a fact (see Melchert 1994: 260), comparable to the situation for uR and iR in Baltic as described by Lipp (2009: 1.24-7).11

New evidence presented by Kloekhorst (2008: 34-50) for u spelling /o/ in Hittite versus for /u/ offers welcome support for my rather tentative claim (Melchert 1994: 125) that the immediate result of * in Hittite was *oR: note an-tu-u-ri-ya- interior /ando:riya-/ < *andwriya- < *en-dhw-. If the same was true for Luvian, then the prehistoric result *oR would have normally lowered to merge with /a/, but in some words rose instead to merge with /u/.

11 Obviously, however, my consistent citing there of the root of gurta- as beginning with a front velar *g(h)- was wholly erroneous.

7

That Luvian did share the same development is suggested by kumma- sacralized. Rather than assume an unmotivated resyllabification as given in Table 1 above (*kwn -mo- >*kunmo-), we should suppose rather a sequence *kwn -mo- > *kwonmo- > *konmo- (with dissimilatory loss of the *w before *o) and read the word as /kommo-/. Lack of any plene spellings in CLuvian precludes proof of this analysis, but other use of u and in CLuvian confirms that Kloekhorsts basic generalization holds also for it. That is, we must assume that Luvian also had both phonemic /u/ and /o/.

The need to assume that syllabic sonorants acted as a fronting environment

in a language where the anaptyctic vowel appears as either /a/ or /u/ is decidedly unwanted. However, we may be dealing with the kind of reciprocal effects seen in Greek set of wheels, chariot < *kwekwlh2, where the labiovelars rounded the anaptyctic vowel, which in turn caused delabialization of the preceding stop (see Eichner 1985: 139). Likewise, we are allowed to suppose that the front velar stop led to a palatal onset of the anaptyctic vowel, which then caused the palatalization (affrication) of the front velar, after which it was absorbed in the affricate, leaving only the non-front nuclear vowel: *kjr ng-id- > *kjjorng-id- > *tsjorng-id- > *tsorng-id- > zurnid- (again really /tsornid-/).

The recent demonstration by Goedegebuure (forthcoming) that CLuvian

zant/da means down and equates to Hittite katta likewise requires that a syllabic sonorant act as a fronting environment. Her analysis rests on comparisons such as CLuvian zanta...kiamman du with Hittite katta kin du let it be combed down and likewise zanda dpaimmi- with katta GUL-ant- struck down, afflicted. She cites further the functional equivalence of CLuvian zantalanuna with Hittite tepnumanzi belittle, persuasively interpreting the Luvian as erniedrigen, humiliate < *zantala/i- *low (cf. arla/i- high and arl(i)- exalt). As she correctly argues, there is no connection with Hittite katta(n), katti in the meaning with, beside (cognate with Latin cum with, OCS k to etc.), so the only sure cognate is Greek /. This permits us likewise to start from front velar *kjm/n tV and assume a development parallel to that sketched for horn above: *kjm/n tV > *kj jom/ntV > *tsjom/ntV > *tsontV > zanta.12 This reciprocal effect scenario is manifestly hypothetical, and I do not insist upon it in the form presented. I offer it only to show that such behavior of syllabic sonorants following a sound that was itself already a front velar is by no means preposterous.

Palatalization of *k before short * is undeniable. The attempt by Lipp

(2009: 1.284-5) to eliminate short * conditioning is impossible: there simply is 12 Lipp (2009: 1.53) assumes that a syllabic nasal served as a fronting environment in pre-PIE, in order to account for *dekm (t) ten with consistent front velar, but does not explain how it might have done so.

8

no conditioned change of posttonic *-es > Luvian -is, as he proposes. CLuvian ta-pi-a sky, heaven is to be read as ta-pax-a, as shown by tap-pa-a-a (for pax see Neu-Rster 1989: 216 w/refs.). There is thus no way to derive iter. -za- from a non-existent preform *-ski(C). Lycian tise someone, anyone < *kwis-ke also directly contradicts any alleged rule of posttonic * > i. We thus confront either a contrast of ki(i)- < *kesye/o- versus -za- < *-ske- with short * or of ki(i)- < *ksye/o- versus zr < *kr (with vocalism from weak stem for *zr) with long *. Luvian and Lycian must therefore have inherited a three-way contrast of */kj/, */k/ and */kw/, with conditioned palatalization of only*/kj/, but not */k/, in fronting environments. 4. Luvo-Lycian Evidence III (voiced stops) The picture for voiced dorsal stops (which include PIE voiced aspirated dorsal stops) is unfortunately much less clear, for two reasons. One is simply the lack of more than a handful of compelling etymologies for the velar set. Those available do not cover enough different environments to determine with confidence the conditioning for various reflexes. The second problem is that there is clear evidence for widespread medial loss of voiced stops (see Kimball 1994 and Melchert 2004b). This means that, for example, in /zurnid-/ horn < *kr ng-id- (NB with velar *g) it is impossible to determine whether the loss is due to a general loss of medial prevocalic *g(h) or results from palatalization to yod before *i and then absorption of the glide (as certainly is the case in /imra/i-/ open country < *ghemro- cognate with Hittite gimra-).

The evidence that we have available appears in Table 2.

Table 2: Summary of Evidence for the Voiced Series (compare in general Kimball 1994)

*g(h) *g(h) *gw(h) /kuttassra/i-/ orthostat (C/H) gurta- citadel (C) wawa/i- cow (H) ~ Lyc. < *gh(o)ut- wall < pour < *ghrdho- Skt. grha- wawa- < *gwo/ew- katmariya- defecate (C) t(u)watra/i- daughter (H) wna- woman (C) < * ghodm(V)r < *dhugh2tr- < *gwn- kallar(a)- unfavorable < *ghalH- pa- get a share (H) u- to drink (C/H) ON galli flaw OIr. galar illness < *bhag- Skt. bhaj- etc. < *egwh- Lat. brius etc. Lith. al damage, wound (Melchert, 2004b)

t(i)yamm(i)- earth (C) zurnid- horn (H) dakkuwa/i- dark (C) < *dhghm(i)- (delocatival) < *kr ng-id- Skt. rnga- < *dhngwo- Fris. diunk dark Welsh dew fog ra/i-, /istra/i-/ hand (C/H) qn- slay < *gwen-(?) < * g(h)esr- *kwanza- (REL-za-)

incise (H) < *gwh- s(k)e/o-

9

/imra/i-/ open country (C/H) nna/i- brother (C/H) ~ nne/i- < *ghemro- (< earth) < *ngno- ~ Hitt. negna- (?) (PES2)waza- drive (H) (CORNU)ki-pu-t/ra horn(?) (H) < *wgh-ye/o- < *geibh- bent, curved(?) zwa- food (C)

< *gyuHo- OHG kiuwan chew parray(a)- high (C) < *bherghoi- Arm. barjr etc. papparkuwa- purify (C) < *bhrh1gw- Av. brzaiti shines /taggam-/ land (H) < *dhghm Hitt. tkan etc. (?) ruwan formerly(H) < *ghrw+ OIce. grjandi dawn, Lith. rti gleam

Deocclusion of labiovelar *gw( h) > w except next to nasal seems assured. The case of Lycian qn- is still much debated. However, the recent demonstration by Kloekhorst (2006: 97-101) that the one sure source of Lycian q is *h2w effectively eliminates any comparison with Hittite anna- judge (thus still Puhvel 1991: 82, with references to predecessors). The occurrences of the Lycian verb in any case demand a sense destroy, kill. We would expect a singular *gwnti to lead to *wti with deocclusion, but in the plural *gwnnti the following n could have preserved the stop. What is much harder to account for is the a-vocalism. I can only suggest that the immediate result of the third plural *qnti led to a specifically Lycian anaptyxis, hence *qnti (by chance unattested), from which was back-formed a new singular qti. The number of required unverifiable steps in this derivation leaves the word of uncertain status.

However, if one accepts the plausible though not absolutely compelling

etymology of HLuvian (CAPERE+SCALPRUM) REL-za- incise from an iterative of *gwhen- strike, then one must likewise here assume that occlusion of the initial voiced labiovelar was blocked by the following nasal. Carruba (1978: 16921) assumes a reading *kuenza-, but in an e-grade form we would definitely predict deocclusion: compare CLuvian wi/enal- stick (*striking instrument) < virtual *gwhen- (thus with Starke 1990: 313-5). We should rather assume *kwanza- (thus Hawkins 2000: 70 and passim), based on an iterative with zero grade of the root *gwh- s(k)e/o-. It remains surprising to me that a syllabic nasal would have such a blocking effect on deocclusion.

Examination of Table 2 shows that available assured etymologies permit assumption of a conditioned treatment of just *g(h) > [j] > /__Vfront, and *g (h) > ts /__*y, with preservation initially before back vowels, but a crucial

10

contrasting example to prove that the first two changes did not apply also to velars are lacking! Although the determinative obviously is quite suggestive, we cannot be sure that HLuvian (CORNU)ki-pu-t/ra- (/kibud(a)-/ or /kipud(a)-/) refers to the crescent horn of the moon (see Hawkins 2000: 470 on the problem and Puhvel 1997: 188 on the questionable Hittite-context SIkiputi-). Even if it does, the derivation suggested above from a root meaning bent, curved is a mere possibility and is not the sort of etymology on which one bases a sound law.

At least suggestive is the apparent contrast we find between palatalization of

*k in HLuvian zurnid- horn < *kr ng-id- and the lack thereof in CLuvian gurta- citadel < *ghrdho-. I must insist on the latter etymology, against that from *gw(H)to- (e.g. Bader 1991: 127, Kimball 1999: 250). The problem is not formal (against Melchert 1994: 260), since the labiovelar could have been unrounded by the round anaptyctic vowel before being deoccluded: *gw(H)to- > *gworto- > gurta- (read /gorta-/).13 The insurmountable obstacle is the semantics: all reflexes of the root *gwerH- in question mean mountain (Sanskrit giri-, Avestan gairi-, OCS gora) or refer to natural phenomena putatively associated with mountains, such as Greek north wind < wind off the mountains. However, it is clear from the use of CLuvian gurta- in the Deeds of Suppiluliuma that the word refers to a man-made fortified enclosure, which is not necessarily on a high place: KBo 5.6 iii 33: arzzi gurti into the upper citadel/walled city. If gurta- meant inherently acropolis, high place, it would be pointless to specify it as upper, which clearly implies the existence of a similar lower enclosure. The problem for our present inquiry is that the absence of palatalization in gurta- < *ghrdho- cannot prove that a sequence of voiced front velar *g(h) plus syllabic * did palatalize. It is only such a direct contrast within the voiced series that would be probative, because even if there was conditioned palatalization of only front velar *g (h) and not non-front velar *g(h), we cannot simply assume that the conditioning matched precisely that of the voiceless series (especially in non-obvious environments such as *).

As already noted above, the general treatment of medial voiced front velars and non-front velars is obscured by uncertainty about the precise conditions on the deletion or retention of both. Medial loss before vowel seems likely. See Melchert 2004b: 377 for HLuvian pa- allocate; receive a portion < *bhag-V-. In the word for daughter the unsyllabifiable *dhugh2tr- received an anaptyctic vowel colored by the laryngeal, which was then lost regularly before stop. Loss of the prevocalic *g and then glide insertion led regularly to HLuvian 13 This remains true whether one derives Hittite gul- and Luvian gulz- inscribe from *gws- prick (thus with Puhvel 1997: 244 following Carruba) or from *kws- (Oettinger 1979: 204 et al.). As Puhvel rightly insists, that choice must be made on other grounds.

11

/tuwatra/i-/: *dhugh2tr- > *dhu.gah2.tr- > *du.gah.tr- > *du.ga.tr- > *du.a.tr- > du.wa.tr-. Lycian syncope of the first vowel led further to kbatra-.14

Whether medial *g (h)/g(h) was lost medially before consonants is far less

clear.15 op (1970: 91) reads HLuvian dat.-loc. sg. ta-ka-mi in the country as /taggami/, reflecting generalization of the strong stem *dhg m, with the medial voiced *g being preserved due to its being geminated by ops eponymous law (followed in Melchert 1994: 232). However, ta-ka-mi may alternatively be read /tagmi/ (Lipp 2009: 1.297 et al.) and taken as a reflex of *dheghm-(e)i, with preserved voiced stop before nasal. Hittite negna- brother is generally equated with Luvian nna/i- and Lycian nne/i- (most recently by Kloekhorst 2008: 601). However, a clear etymology for the word remains to be found, and we cannot at present exclude that the Luvian and Lycian words are mere Lallwrter that resemble the Hittite purely by chance (see Puhvel 2007: 108).16 I must not fail to mention one last nagging complication. We have seen that *w acted as a fronting environment and led to affrication of *k to ts. However, in the verb papparkuwa- purify, most likely from reduplicated *bhr h1gw- to the root *bhreh1g - seen in Av. brzaiti shines, the *g is preserved as a velar stop. Is this different treatment due to the difference in voicing of the stop, or did the preceding non-syllabic r that developed in *bhrh1g- > *park- block the palatalization? Strictly speaking, of course, we do not even control the outcome of *g (h)w- more generally, for lack of evidence. 5. Conclusion The unconditioned three-way contrast I claimed in 1989 for the voiceless dorsals in Luvian and Lycian is to be rejected. Luvo-Lycian is rather the mirror-image of Albanian: before the merger of the front and non-front velars, the voiceless front velar, but not the non-front velar, underwent conditioned palatalization. A similar conditioned palatalization of just voiced front velars is also possible, but cannot at this point be affirmed. Thus Anatolian is in terms of the traditional classification centum: it does show an eventual phonological merger of the front and non-front velars.

14 Since we cannot know whether the HLuvian stem was /tuwatra/i-/ or /twatra/i-/, *duatr- may alternatively have led directly to *dwatr- with disyllabification of the *u, producing /twatra/i-/ and kbatra- directly. 15 For further consideration of this problem see Ofitsch 1998: 426 with references. 16 Unfortunately, the very attractive etymology of Neumann (1991: 63-4), *ni-gnh1-- inborn, runs afoul of the already OS spelling ne-eg-na in KBo 20.31 Ro 6, whose e-vocalism cannot be derived from *ni-. The Luvian and Lycian obviously also exclude *ni-, if they are in fact cognate!

12

Appendix: Evidence for Three Sets of Dorsals in Contrasting Environments

Initial before *r: *kremh2- become slack: Skt. ram- become weary, Grk. krmamai hang down *kreuh2- bloody flesh, gore: Skt. kravs-, Grk. kras flesh etc. *kwreih2- obtain by exchange, buy: Skt. krti, Grk. pramai buy etc. Initial before *l: *klei- lean, recline (intr.): Skt. ri- recline, Grk. kln lean, etc. *klep- steal: Lat. clep, Grk. klp, TochB klypi- steal Final after *i: *weik- enter: Skt. vi- enter, Grk. okos house, Lat. ucus clan etc. *weik- separate, select: Skt. vivkti separate, Lat. uictima victim, Gothic weihan

consecrate *leikw- leave behind: Skt. rikti, Grk. lep, Lat. linqu leave behind etc. *leig- bind: Lat. ligre, Alb. lidhem bind, MLG lk bond *weig- give away: Skt. vij- flee, OE wcan yield, TochAB wik- disappear *neigw- wash: Skt. nij- wash, Grk. nniptai is washed etc. *leigh- lick: Arm. lizem, Grk. lekh, OIr. ligim lick etc. *steigh- stride (upward): Skt. stigh-, Grk. stekh advance, Goth. steigan climb etc. *(s)neigwh- to adhere: Skt. snhyati love, Grk. nephei, Lith. sniga snows etc. Final after *r: *derk- look at: Skt. d- see, Grk. drkomai look at etc. *perk- fill up: Skt. pk- fill, OIr. ercaid fills *terkw- turn (intr.): Hitt. tarku- dance, Lat. torque turn (tr.), TochB tetarku turned *werg- make: Av. vrziieiti does, makes, Grk. orge has carried out etc. *h2werg- turn: Skt. vj- turn, Hitt. urki- wheel etc. *tergw- threaten: Skt. tarj- threaten, Grk. trbos fright, Lat. toruus glaring etc. *bhergh- high: Av. brz- height, Hitt. parku- high etc. *bhergh- guard, watch over: Av. brjaiia- honor, Goth. bairgan preserve etc. Final after *l: *Hmelk- stroke: Skt. mti touches, Lat. mulce stroke *selk- pull, draw: TochB slk-, Grk. hlk pull, draw *h2melg- milk: Grk. amlg, Lith. mlu, OE melcan milk etc. *welg- roll: Skt. vlgati looms up, leaps, OE wealcan roll, press etc.

13

Final after *n: *denk- bite: Skt. da()-, Grk. dkn bite *kenk- hang (intr.): Skt. ak- be in doubt, Hitt. kank- hang, Lat. cnctor hesitate *bhengh- make firm, massive: Skt. bah- thick, many, Av. bzaiti fixes, Hitt. panku-

all, Grk. pakhs thick, massive etc. *ghengh- walk: Lith. engi, Goth. gangan walk etc. *h1lengwh- light, nimble: Skt. rhate hurries, runs, Grk. elaphrs light, quick, OHG gi-lingan succeed

14

References

Bader, Franoise. 1991. Problmatique du gnitif thmatique sigmatique. BSL 86/1.89-157.

Borchhardt, Jrgen, Heiner Eichner, et al. 1997-99. Archologisch-sprachwissenschaftliches Corpus der Denkmler mit lykischer Schrift. Anzeiger der philosophisch-historischen Klasse, sterreichische Akademie der Wissenschaften, 134. Jahrgang, 11-96. Vienna: AW.

Carruba, Onofrio. 1978. Il relativo e gli indefiniti in licio. Die Sprache 24.163-79.

op, Bojan. 1970. Eine luwische orthographisch-phonetisce Regel. IF 75.85-96. Fortson, Benjamin. 2004. Indo-European Language and Culture. An

Introduction. Malden MA: Blackwell. Goedegebuure, Petra. forthcoming. The Cuneiform Luwian adverb zanda

down, (along) with, together, jointly. 7th International Congress of Hittitology, orum, Turkey.

Gusmani, Roberto. 1961. Il suffisso -tjo- di aggettivi locali e la sua diffusione nelle lingue indoeuropee. AION-L 3.41-58.

Hajnal, Ivo. 1995. Der lykische Vokalismus. Graz: Leykam. . 2003. Jungluwisch eine Bestandsaufnahme. In M. Giorgieri et al.

(eds.): Licia e Lidia prima dell ellenizzazione, 187-205. Rome: CNR. Hawkins, J. David. 2000. Corpus of Hieroglyphic Luwian Inscriptions of the

Iron Age. Berlin: de Gruyter. Hock, Hans Henrich. 2009. Labiopalatalization in Indo-European Languages. In:

S. Jamison, H. C. Melchert and B. Vine (eds.), Proceedings of the 20th Annual UCLA Indo-European Conference, 69-78. Bremen: Hempen.

Jamison, Stephanie. 1983. Function and Form in the -ya-Formations of the Rig Veda and Atharva Veda. Gttingen: Vandenhoeck & Ruprecht.

Kimball, Sara. 1994. Loss and retention of voiced velars in Luvian: another look. IF 99.75-85.

. 1999. Hittite Historical Phonology. Innsbruck: Institut fr Sprachwissenschaft der Universitt Innsbruck.

Kloekhorst, Alwin. 2006. Initial Laryngeals in Anatolian. HS 119.77-108. . 2008. Etymological Dictionary of the Hittite Inherited Lexicon.

Leiden/Boston: Brill. Kmmel, Hans Martin. 2007. Konsonantenwandel. Bausteine zu einer Typologie

des Lautwandels und ihre Konsequenzen fr die vergleichende Rekonstruktion. Wiesbaden: Reichert.

Lipiski, Edward. 2004. Itineraria Phoenicia. Leuven: Peeters. Lipp, Reiner. 2009. Die indogermanischen und einzelsprachlichen Palatale im

Indoarischen. Heidelberg: Winter. Meillet, Antoine. 1934. Introduction ltude comparative des langues indo-

europennes.7 Paris: Hachette.

15

Melchert, H. Craig. 1984. Studies in Hittite Historical Phonology. Gttingen: Vandenhoeck and Ruprecht.

. 1987. PIE velars in Luvian. In: C. Watkins (ed.), Studies in Memory of Warren Cowgill, 182-204. Berlin/New York: de Gruyter.

. 1988. Luvian Lexical Notes. HS 101.211-43. . 1989. New Luvo-Lycian Isoglosses. HS 102.23-45. . 1994. Anatolian Historical Phonology. Amsterdam/Atlanta: Rodopi. . 2004a. A Dictionary of the Lycian Language. Ann Arbor/New York:

Beech Stave. . 2004b. A Luwian Dedication. In J. Penney (ed.), Indo-European

Perspectives. Studies in Honour of Anna Morpurgo Davies, 370-9. Oxford/New York: Oxford University Press.

Morpurgo Davies, Anna. 1986. Fighting, Ploughing and the Karkami Kings. In: A. Etter (ed.), O-o-pe-ro-si. Festschrift fr Ernst Risch zum 75. Geburtstag, 129-45. Berlin: de Gruyter.

Morpurgo Davies, Anna, and J. David Hawkins. 1988. A Luwian Heart. In: F. Imparati (ed.), Studi di storia e di filologia anatolica dedicate a Giovanni Pugliese Carratelli, 169-182. Firenze: Elite.

Neu, Erich, and Christel Rster. 1989. Hethitisches Zeichenlexikon. Wiesbaden: Harrassowitz.

Neumann, Gnter. 1991. Hethitisch negna- Bruder. HS 104.63-6. Oettinger, Norbert. 1979. Die Stammbildung des hethitischen Verbums.

Nrnberg: Carl. Ofitsch, Michaela. 1998. Indogermanischer Grundwortschatz in den

anatolischen Sprachen. Probleme der Rekonstruktion anhand der Verwandtschaftbezeichnungen. In: W. Meid (ed.), Sprache und Kultur der Indogermanen. Akten der X. Fachtagung der Indogermanischen Gesellschaft Innsbruck, 22.-28. September 1996, 421-36. Innsbruck: Institut fr Sprachwissenschaft der Universitt Innsbruck.

Puhvel, Jaan. 1991. Hittite Etymological Dictionary. Volume 3. Words beginning with H. Berlin/New York: Mouton de Gruyter.

. 1997. Hittite Etymological Dictionary. Volume 4. Words beginning with K. Berlin/New York: Mouton de Gruyter.

Rieken, Elisabeth. 1997. Zu den Reflexen von uridg. *(s)k'el- springen, eilen. HS 110.167-175.

. 2003. Hieroglyphen-luwisch z+ra/i-la-mi-i (SCALPRUM. ARGENTUM)su-ha-pa-na-ti: ein Kompositum und eine neue luwisch-lateinische Isoglosse. HS 116.35-53.

Schaffner, Stefan. 2001. Das Vernersche Gesetz und der innerparadigmatische grammatische Wechsel des Urgermanischen im Nominalbereich. Innsbruck: Institut fr Sprachen und Literaturen der Universitt Innsbruck.

Sihler, Andrew L. 1995. New Comparative Grammar of Greek and Latin. New York/Oxford: Oxford University Press.

16

Starke, Frank. 1990. Untersuchung zur Stammbildung des keilschrift-luwischen Nomens. Wiesbaden: Harrassowitz.

Watkins, Calvert. 1987. Two Anatolian forms: Palaic akumua-, Cuneiform Luvian wa-a-ar-a. In: G. Cardona & N. Zide (eds.), Festschrift for Henry Hoenigswald, 399-404. Tbingen: Narr.

. 1999. A Celtic Miscellany. In: K. Jones-Bley et al. (eds.), Proceedings of the Tenth Annual Indo-European Conference. Los Angeles. May 21-23, 1998, 3-25. Washington DC: Institute for the Study of Man.

Woodhouse, Robert. 1998. On PIE. tectals. IF 103.40-69. Woudhuizen, Fred. 2010. Towards a Chronological Framework for Significant

Dialectal Tendencies in Indo-European. JIES 38.41-131. Yakubovich, Ilya. 2010. Sociolinguistics of the Luvian Language.

Leiden/Boston: Brill. Zehnder, Thomas. 2010. Die hethitischen Frauennamen. Wiesbaden:

Harrassowitz.

H. Craig Melchert ECIEC 29 Cornell University June 19, 2010

The Verbal Prefix u- and vs. Spellings in Anatolian Cuneiform

I. Hittite (and ostensibly Luvian) preverb u- her- per Melchert (1994: 104) < *au- (~ Lat. au-fer carry away/off) with semantic shift away from > towards (etymology already in Hrozn 1917: 1222). Similarly Kloekhorst (2008: 909) < *h2ou with special treatment in Anlaut vs. retention in eu. For semantic shift cf. Hackstein (2007: 138-42). Per Oettinger (1979: 126 & passim) < *h1/3ou- after Eichner (1973: 54 & passim) without explicit comparandum (PIE *au- scil. *h2au- seen rather in -u of eu come!), but Oettinger (2006: 43514) accepts derivation from *au-. Kimball (1999: 225) abandons derivation < *(e)u- and equation with alleged Arcado-Cypriot -/- and views PIE source as unclear. Kimball (2007:210) cites *(o)u- without comment.

II. Prefix spelled with in some verbs, with in others (see below). No problem with assumption of equivalence of two signs (Melchert 1994: 26 w/ refs.; Kimball 1999: 79-80, et al.). But as per Rieken (2005: 538-9 after Eichner), invariance in spellings must be given equal weight with variance. By this principle overwhelming consistency of vs. spellings in well-attested words argues for linguistically real contrast. E.g., 1x a-pu--un vs. 150+ a-pu-u-un argues for contrast, not equivalence. Similar arguments by Kloekhorst (2008: 35-60). Therefore, contrast exists between /u/ () and /o/ () in Hittite. Details of both synchronic and diachronic distribution remain to be determined. For very provisional formulation (as of 6/19/2010!) see Appendix 1. For evidence that contrast exists also in Palaic and CLuvian see Appendix 2.

III. Attested spelling of directional prefix u- in Hittite problematic for all formulations:

u-un-ni/a- (-)u-i-ya- up-pi/a- u-(i)-ya- -da- -wa-te/a- -e-/-wa-

Unmarked result is clearly /u-/ (). Others require explanation. For u-un-ni/a- drive hither laxing/lowering before tautosyllabic nasal is plausible (with Rieken 2005: 540-1; similar Kloekhorst 2008: 54, but with false generalization to position /_nV). No such solution for u-(i)-ya- (allegedly send here). Not credible Kloekhorst (2008: 52): lowering of /u/ to /o/ before i (actually yod, as per Kloekhorst 2008: 911). Rieken ignores word entirely. Likewise problematic is up-pi/a- with no plene spellings. Kloekhorst (2008: 36) assumes unattested OS spelling would be *-up-pi/a- (citing restriction of plene -u-i-ya- to OS). Possible but ad hoc (NB total of OS -u-i- 3x vs. post-OS u-i- 12x, but up-p/a- 100x, 22x in MS).

IV. Hittite uppi/a- send; bring

Since Hrozn (1917: 1222) near-universally assumed to contain prefix u- her-: Gtze (1933: 223 & 1938: 124); Friedrich (1952: 234 (her)schicken [sic!]); Oettinger (1979: 489 her-schicken [sic!]); Melchert (1994: 104&149); Yakubovich (2005: 245 [implicit in gloss to bring]); Kloekhorst (2008: 921-2 send (here) [sic!]). Objection of Pedersen (1938: 116-7) wrongly ignored: actual usage shows no directionality whatsoever!

2

(1) HKM 18:18-20 (MH/MS)

nu=mu k katti=mi RIN.ME KUR.UGU RIN.ME KUR URUIpitta kuiki n=an=ta uppai There are some troops of the Upper Land and I. here with me. Ill send them to you.

(2) KBo 5.4 Vo 22 (NH; Treaty with Targasnalli)

nu=mu atri RIN.ME=wa=mu ANE.KUR.RA.ME uppi nu=tta RIN.ME.E ANE.KUR.RA.ME.E uppai And you write me: Send me troops and horses! I will send you infantry and horses.

(3) KBo 3.6+ iii 77 (NH; Apology of Hattuili)

n=an ppn n=an=kan A.AB.BA tapua uppaun I seized him and sent him alongside the sea.

See also HKM 25:22-23, HKM 30:8-10. For likely OH existence see KBo 20.67 iv 33 (invocation to the mountains, CTH 591, OH/NS).

(4) KBo 17.65+ Ro 38-39 (MH/MS; Birth Ritual)

(But how the festival of birth (is)how they perform the festival when she gives birth,) [(n)=a GIkurta iy]anza n=a URUKizzuwatna nu=mu=kan EZEN KAxU-it [L karta n]=an apz uppai [It is m]ade [(as) a kurta-tablet], and it is in Kizzuwatna. I [do not know] the festival orally [by heart]. I will fetch it from there.

Restorations based on parallel passage ibid. Vo 45-46, which has rather udanzi they will bring here. See Beckman 1983: 136-7 and 161-4 with differing details.

(5) KUB 12.58 ii 36-42 (MH/NS; Ritual of Tunnawi)

(The ritual client goes to bathe,) n=ata MUNUSU.GI 9 GIGA.ZUM A GITG anda upp[i] GIn!an IM-a anda uppi The old woman fetches in nine combs of boxwood. She fetches in a figurine of clay.

Goetze (1938: 14-15) renders brings in, but the practitioner has been in the ritual space in the entire preceding context.

(6) KBo 15.37 ii 49-59 (MH/NS; ()iuwa-Festival) (When it dawns on the eighth day, the large loaf of cheese and the leavened bread of three parsu-measures of flour which (are) broken for the gods of the fathers and which are placed back on the sacrificial tables,) n=a ara parulnzi namma ANA PANI DINGIR.ME manda kuwapiya 1 parullin GA.KIN.AG 1 parullin NINDA=ya uppiyanzi n=u PANI DINGIR.ME zikkanzi They crumble them and then bring one morsel of cheese and one morsel of bread before each of all the gods (lit. before all the gods in each place) and place them before the gods.

Contra Carruba (1966: 2020), Oettinger (1979: 489) and Nakamura (2002: 159) no basis for separate verb lift, raise, but send (Gterbock & Hoffner 1995: 193) also unlikely.

3

Given total lack of evidence for presence of prefix her- but strong goal orientation (all 80+ examples with send in sufficient context have explicit or implied goal), perhaps deadverbial < *po as in Skt. pa up to (anticipated by Oettinger 1979: 489-90, but with false semantics for non-existent homonym raise, lift). For process see Melchert (2009) and compare rare transitive use of English near = bring near (to) (Ruskin & Wilde). For unexpected stem uppi- compare opposite discrepancy in Hitt. anna- conceal < *s(h1)o vs. Latin sine and OIr. sain- < *s(h1)i. V. CLuv. (u-)up-pa- carry and HLuv. (CAPERE)u-pa- carry (off)

Pace Kloekhorst (2008: 922) basic sense carry of CLuvian not in doubt, but his rejection of equation with Hittite uppi/a- is probably correct:

(7) KBo 13.260 iii 13-15 (Birth Ritual)

=tta dduwan=za pariyan adduwaliyan wattaniyan uppannandu Let them carry the evil over to an evil land.

Thus with Yakubovich (2010: 237).

(8) KBo 13.260 iii 16-20 (Birth Ritual)

zla=pa=tta za-x-x-ti DUMU.L.LULU-ni UM-U alziuidumar=a ppannandu wayai=a addulai=a annarumi=a Henceforth let them bring to this(?) humanone calls out his namelife, w., health (and) vigor.

(9) KUB 35.88 iii 11-12 (Birth Ritual)

upatta=pa=wa=du arriyanin 2-u 9-u[(n-)za] anta=wa=a=ta wallunaan wni uppanta She furnished to her twice nine s., and they carried them in to the midwife (lit. woman of lifting).

See also KUB 35.107 iii 21 (anda uppanta) and KUB 25.39 iv 16 (ppadda).

NB spelling with u-!

(10) KARKAMI A 11b+c 13

(I devastated those countries,) *a-wa/i-ta (SCALPRUM.CAPERE2)u-pa-n-zi a-t (CAPERE2)u-pa-ha I brought in/collected trophies/spoils (and I came up from those countries in glory.)

(11) KARKAMI A2+3 7

*a-wa/i-ta *a-mi-ya-za LITUUS+AVIS(-)ta-ni-ya-za REGIO-ni-i a-t (DEUS)BONUS-na (DEUS)VITIS(-)ti-PRAE-ya-ha ARHA (CAPERE2)u-pa-ta He (Tarhunza of Carchemish) in my days carried off the grain-god and the wine-god into the country.

(12) SKENDERUN 3-4

a-wa/i za-na-i (*255)ka-ru-na-na (CAPERE)u-pa-ha-i wa/i-tu-u-ta-i 4xMILLE 4xCENTUM a-ta (CAPERE)u-pa-ha zi-i-na (*256)zi-pa-ta-na-ti I carried off this granary and I collected for it 4,400 with this z-measure.

4

For (CAPERE)u-pa- as bring vs. (PES)u-pa- dedicate, furnish see Melchert 2004 contra Yakubovich 2005, but directionality implied there is dubious. Note further figura etymologica /upani-/ /upa-/ carry (off) spoils/trophy. Contra Melchert (2004: 372) spoils are not inherently brought (in), but rather carried off. This figure and the otherwise unexplained determinative CAPERE(2) take argues for original meaning carry off, remove, bleached to simply carry.

Spelling of CLuvian cognate with u-up- points to /o-/, thus combined evidence argues that this is the reflex of *au- off, away. Root *pa- is probably *(s)peh2- set in (violent) motion, draw > Grk. tear/pull, Arm. hanem pull; take away; lead: see Garca Ramn 2009. /o:ppa-/ *take away, remove whence carry < virtual *au-(s)p(e)h2-. NB inflectional class unknown, but third plural *au-(s)ph2nti would give /o:ppanti/ in either mi-or i-verb. VI. Hittite uye/a- send, drive

Universally assumed to be u-ye/a- her-schicken vs. pe/i-ye/a- hin-schicken, but again no evidence for speaker-oriented directionality of first. Pedersen (1938: 198); Friedrich (1952: 232) schicken (eigentlich herschicken); jagen; Kloekhorst (2008: 910) send (here).

(13) KBo 3.40b:9-10 (OH/NS; narrative)

am[(mug=a dU-a)] DINGIR piene [LUG]AL-i uyr t=wa L.MEGAL.GAL wemiy[(a)] But me the male gods of the Storm-god sent to the king (saying): go and find the great men!

(14) KUB 33.5 ii 4 (OH/NS; Telipinu, 2nd version) dMA-a NIM.LL-an uyt t=za x[]x dTelipinun zik ana Hannahanna sent the bee (saying): Go [ ] and you search for Telipinu!

Duplicate KUB 33.4 ii 17 has piyt! KUB 33.8 ii 22 in similar context has uyt.

NB both uiye/a- and pe/iye/a- are securely OH: x-x-x[ ]iz LUGAL-i atti=mi u-i-x[ ] (KUB 26.35:3) and [-a]n ara imma piyzzi (KUB 36.106 Ro' 5). (15) KUB 23.72+ Vo 22-23 (MH/MS; Mida of Pahhuwa)

A L.KR=ma=za=kan L EMI INA K=UNU L tarnanzi kui=a=ma L.KR L EMI uyzzi a-p-[n]=an ANA dUTU-I uppianzi They shall not let the messenger of an enemy in their gate. Whatever enemy sends them a messenger, [] and they shall send him to His Majesty.

(16) KUB 14.3 i 6-9 (NH; Tawagalawa letter)

nu=m[u U]N-an IGI-anda uiyat ARAD-anni=wa=mu d nu=wa=mu Ltukantin uiya nu=wa=mu ITTI dUTU-I uwatezzi nu=i LTARTENU uiyanun He sent a person to meet me (saying): Take me into servitude! Send the crown-prince to me, and he shall bring me to Your Majesty. So I sent the crown-prince to him.

(17) KBo 4.8 ii 13-14 (NH; Tawananna Affair)

k=ya=an 1-an dammeanunun ITU .GAL-LIM=pat=kan kuit katta uiyanun And I also did her this one harm, that I sent her down from the palace.

5

(18) KUB 1.1+ iii 10-11 (NH; Apology of Hattuili) URUakpia=ma kururiyata [nu] L.ME GagaI.A uiyanun n=an ITU N.TE=YA SIG5-aun But Hakpis became hostile, [so] I drove out the Kaskeans and on my own put it (the city) in order.

Thus with Otten (1981: 17) contra van den Hout (2003: 199). Confirmed by parallel KUB 1.7 ii 6, which adds ara. NB: no space in KUB 1.1 iii 11 for any restoration beyond [nu]. For sense banish, drive away for ara uiya- see also KUB 14.8 Vo 17-18 and Plague Prayers passim.

NB that sense banish, drive (away) shows that original sense did not include control by subject of object. Thus meaning send is secondary. Preverb ara also not required for sense drive away, banish. Argues for original *au-(Hi)Hyeh1- *cast away. Bleached generally to drive, send, hence creation of new univerbation with *pe- for send (off).

VII. Reconsideration of prefix u-

1. Contrast of CLuvian /o:-/ in u-up-pa-carry (off) vs. /u:-/ in -pa- dedicate, furnish (see Melchert 2004), also borrowed into pre-Hittite in -ba-ti- land-grant and of Hittite u-(i)-ya- send, drive (away) vs. - her- in -da- bring, -wa-te/a-bring (people /animals), and -e-/-wa- come argues for *au- away in the former, but something else in the latter.

2. Luvian awi- come beside Hitt. -e-/-wa- seems to exclude zero-grade *u- and spelling with - makes *ou- unlikelyassumption by Kloekhorst of different outcome of *h2ou in isolation vs. usual *ou > /o/ () totally ad hoc. Thus *eu-, but from what?

A. Rehabilitate Kimballs comparison with Arcado-Cypriot -/-? Difficult in view of Strunk 1986. Compare very cautious remarks of Widmer 2008: 623 with note 17.

B. Derive from far-deictic *e/ow(o)- with same assumed semantic reversal as cited in I above? NB vs. Av. auua-, OP ava- that, jener mixed deixis in Slavic: Serbo-Croatian vj, Macedonian ovoj this, but Polish w that. Note especially USorbian jow/LSorbian how here, hither. And/or the particle of Skt. as, amm yonder with similar semantic development?

6

Appendix 1

Provisional Analysis of /u/ vs. /o/ as of 6/19/2010

1. For earlier literature see Rieken 2005: 537 with refs. to Weidner, Held & Schmalstieg, Eichner, and Hart. The following represents blend of Rieken 2005 and Kloekhorst 2008: 35-60 with personal views. Note following premises: (1) if contrast is valid, also exists in Palaic and CLuvian (see Appendix 2); (2) in poorly attested words with alternate and spellings subjective treatment is unavoidable (each scholar will analyze in terms of personal overall schema); (3) even in cases of secure root etymologies proving that given word must be reflex of specific ablaut grade *eu, *ou, *u often difficult; (4) by any analysis some data remains recalcitrant.

2. Prehistoric *u (NB * = PIE */uh1, not *uh2/3) regularly > /u(:)/ spelled : a-tu-()-ga/i- frightful < *h2tgo- (but only 1x with --) i--kn yoke < *yugm ku--na-a dog (GSg.) < *k nos (regular contra Kloekhorst 2008: 54) mi-(i)-- soft, mild < virtual *mih1-u- u--e- push away, reject < *suh1-ye/o- u-u- full (N-ASgNt) < *suh3-u (last vowel) -ug I after *t (NB: cannot reflect *th2!) -me-(e)-ni we see and -wa-an-zi they see < *u-wni and *u-nti -na-at-ta-al-la- merchant pre-Hittite loanword < Luvian *us-no+ -nu- adorn < *u-n()u- -ra-a-ni burns < *ur-ri a-u--ul, im-mi--ul, i-i--ul, tk-u--ul (1x), wa-a-du--li (1x) syncopated from *-lom as per Rieken (2008: 252-3) contra Rieken (2005: 543-4)

3. *eu/u > /u:/ spelled :

i--uk brace, pair < *yug (Kloekhorst 2008: 423 after Rieken 1999: 61f.) ku--a- daughter/son-in-law; bride < *guso- *chosen one (Rieken 1999: 257) u-e-nu--ut, wa-a-nu--mi < *-nu- i-(i)--(n)- god < *dyu- (but u-up-zi rises (of sun) then cannot reflect *(h1)up-ti, as per Oettinger 1979: 233, Melchert 1994: 104; ku--uz-za wall also problematic, since as animate t-stem, hardly < *ghut-s with Kloekhorst 2008: 57) 4. Prehistoric *u secondarily lowered to /o(:)/: A. Before or after *h2 and *h3:

a-a-u-u goods < *h1suh2 u-u-ma-an-t all < *h2u-h1m-ent- (etym. with Kimball 2007: 10) nu-u-wa still, yet < *n-h2o u-u- full < *suh3-

7

B. Before r except word-initially (see -ra-a-ni above!)

i-nu-u-ra- kneading pan < *is-n-ro- (vs. Palaic ta-u--ra-!) -e/i-u-u-ri-ya- twist < *weis--ro- ku-u-ur-ka- foal (cf. Grk. ) and u-u-ur-ka/i- root (cf. Latin surculus) 5. *ou/au > /o(:)/ spelled

mu-u-ga-i- incite; goad < *muko- (Melchert, FS Archi, revising Kloekhorst) u-(i)-e-/ya- drive (away), send < *au-Hyeh1- 6. PIE *w between consonants and word boundary > prehistoric *woR > /o(:)R/ spelled

a-ni-u-ur, a-ni-u-r ritual with postconsonantal /-o:r/ from enkur etc. an-tu-u-ri-ya- interior < *en-dhw- pd-du-u-la- loop hypostasized from *peth2w *spreading out u-ur-ki- track < *wgi- u-ur-re-er < *wh1r [But sequence *wCC > warCC with two consonants as in d(u)warne/a- < *dhwnh1-nti; Kloekhorst 2007: 456-7.]

7. In word-final *-m(s)# the *o did not lower to merge with /a/ and appears preserved as /-o:n/ in ku-u-un and a-pu-u-un and the plurals ku-u-u and a-pu-u-u (cf. Melchert 1994: 186-7 and Kloekhorst 2008: 54&56). The vocalism of uni has been remade after the other two demonstratives, just like *si ai after k and ap (Kloekhorst 2008: 220-21).

8. Open Issues: (1) quality of anaptyctic u; (2) further lowerings in NH before n and r; (3) non-pronominal acc. pl. in --u; (4) given problems for *eu > /u:/ under 3. above, rather give up direct equation of Hitt. -e- with Luvian awi- and assume zero-grade for Hittite, thus returning to view that all *Vu > /o(:)/? ________________________________________________________________________

Appendix 2

Spellings with and in Palaic

1. (/u:/) < *u nu- now < *nu tu- to thee < *t mu--i you are satiated (P2Sg with Yakubovich, JANER 5.117 contra Melchert, KZ 97.35) < *ms-si, to imitative root of Grk. be satiated, in one gulp (Eichner) u--na-at has filled, u--na fill! < *s-ne-h3(-t) with accent retraction ta-u--ra- offering table (loc. sg.) *dhh1s--ro-; cf. Hittite inura- kneading pan,

but latter regularly i-nu-u-ra- (including OS!)

2. /u(:)/ < *eu (probably shortened when unaccented)

i- come! per Melchert (1994: 212) ~ Hitt. eu, but required totally ad hoc loss of --. Rather with Oettinger (1979: 536) = reverse of Hitt. -e- and Luvian awi-, thus virtual *h1ei-eu (but see remark under Appendix 1, 8.).

8

3. /o:/ < *uh2

wa-u-u-(a) good ones < *wsuh2 (with shift of accent with added clitic) 4. /o:/ < *ou (?)

a-ru-u-na-am-p per Melchert KZ 97.39-40 acc. sg. high to aru-, thus probably generalized weak stem *h1r-w- (cf. CLuvian)

5. /o:/ < secondary u < *gwhu a-u-u-na to drink

Spellings with and in CLuvian

1. (/u:/) < *u a-ru--da-ti abl.-inst. to arut(i)- wing < *ar--ti- *joint (Melchert, HS 101.224f.) i--na-i-it- ability to walk; mobile wealth; built on infinitive i--na to go < *h1y-un- eh2 [but also pa-a-u-u-na to swallow!] ma-ad-du--wa-ti with wine < *mdhu- du--p-/du--pa-i- strike < *(s)tubhye/o- (must assume zero-grade iterative type, contra Melchert, FS Puhvel 135) du--ur, du--n urine < virtual *sh2-n- and zero-grade generalized to nom.-acc. NB *kwrV- leads to ku--rV-: ku--ru-na to cut, ku--ra-am-mi cutting (dat.-loc. sg.), ku--ri cut! (to denominative kri-/krai-)

2. (/u:/) < contracted *(u)wa

i-(i)-ru--un and i-(i)-ru--t oath (1x with -u- vs. several dozen with --); long vowel contracted from *-uwan, as per Melchert, GsForrer, after Watkins, FS Rix) lu--un-ni we take; -unni contracted from *-wanni < *-wni by ops Law [but also pi-u-un-ni we give!] -la-an-ti- dead (2x) syncopated from walant(i)-; also in -la-an-ta-al-li-ya- (1x, but latter also 3x with u-la-!) -li-ip-na/i- wolf (2x) syncopated from wa-li-ip-na- [but NB 1x u-na-at-ti-i woman syncopated from wa-na-at-ti-!]

3. Diphthong /aw/ spelled -Ca-(a)-- (cf. for Hittite Kloekhorst 2008: 42-44, 59)

i-ik-ku--na--na-a-i- of anointing *kknawar/*kknaun- gul-za-a--na to draw kur-a--na-an-ti-in-zi islands (erg. pl.) pa-tal-a--na to fetter [vs. unexpected la-la-u-na to take in same ms.interference from putative NH change claimed by Kloekhorst 2008:44?]

also in secondary diphthongs /a(:)w/

i-iz-za--un-ni we fetch; -unni contracted from *-wanni < *-wni by ops Law la--na-i- wash (cf. la-u-ni-)

9

and unexplained Pret3Pl in /-a:wnta/

i-ik-ku-na-a--un-ta anointed na-ak-ku-u-a-a--un-ta furnished a scapegoat wa-a-ar-ma-a--un-ta ?

[vs. unexpected tar-ra-wa-u-un-ta ?!]

4. /u:/ < *eu (or are first and third examples from zero-grade??)

nu--(t)- assent, approval < *nu- (Rieken 2006: 294) -ut-ti-i you (sg.) drink < *egwtis via secondary diphthong *e u -pa- dedicate (and -ba-ti-it- land-grant) < *eu-bhag- (Melchert 2004 contra Yakubovich 2005)

5. /o:/ < *ou/au

*a-ru-u-wa high (N-APlNt as adverb highly, very) (5x vs. 2x a-ru--wa!) likely from generalized weak stem *h1ru- mu-u-wa- overpower < *muh1/3- (hi-verb!) (and derivatives) zu-u-wa- food < *gyuh1/3-o- (tmos noun to root of chew) 6. /o:/ < *u next to *h2/3 u-u-t/da- alacrity < *h2h1-to- (with Eichner & Starke to root of run *h2weh1-; with accent retraction in substantivization) u-u-tar-la-a- slave, servant (from preceding) NB: no spellings u-- anywhere in CLuvian vs. 20+ lexemes with sequence u-u. 7. /o:/ < *u before tautosyllabic nasal(?)

u-un-za-~u-za-a you (pl.) generally taken as *us- with secondary *uns- after first plural *ans- < *s- (but NB very problematic hapax na-a-nu--un(-pa) where plene of second syllable is unexpected in view of na-a-!)

8. Open issues: popular derivation of *DUGu--wa-at-ra/u--wa-at-n (vessel) from u-u-wa- fill < *s(e)uh3- now dubious. NB CLuvian spelling of verb matches Hittite; spelling of pu--wa formerly contrasts with pu-u-wa- crush!

References

Beckman, Gary. 1983. Hittite Birth Rituals (StBoT 29). Wiesbaden: Harrassowitz. Carruba, Onofrio. 1966. Die Beschwrungsritual fr die Gttin Wiurijanza (StBoT 2). Wiesbaden:

Harrassowitz. Eichner, Heiner. 1973. Die Etymologie von heth. mehur. MSS 31.53-107. Friedrich, Johannes. 1952. Kurzgefates hethitisches Wrterbuch. Heidelberg: Winter. Garc-Ramn Jos Luis. 2009. Idg. *(s)peh2-, in (heftige) Bewegung setzen, ziehen: Ved. p3, heth.

pipp(a)-i und gr. , arm. hanem. In Rosemarie Lhr and Sabine Ziegler (eds.), Protolanguage and Prehistory. Akten der XII. Fachtagung der Indogermanischen Gesellschaft, vom 11. bis 15. Oktober 2004 in Krakau, 134-48. Wiesbaden: Reichert.

Gtze, Albrecht. 1933. ber die Partikeln -za, -kan und -an der hethitischen Satzverbindung. ArOr 5.1-38. . 1938. The Hittite Ritual of Tunnawi. New Haven: American Oriental Society. Gterbock, Hans G. and Harry A. Hoffner Jr. 1995. The Hittite Dictionary of the Oriental Institute of the

University of Chicago. Volume P. Fascicle 2. Chicago: The Oriental Institute.

10

Hackstein, Olav. 2007. Ablative Formations. In Alan Nussbaum (ed.), Verba Docenti. Studies in historical and Indo-European linguistics presented to Jay Jasanoff by students, colleagues, and friends, 131-53. Ann Arbor/New York: Beech Stave.

van den Hout, Theo. 2003. Apology of attuili III (1.77). In William Hallo and K. Lawson Younger (eds.), The Context of Scripture. Volume I. Canonical Compositions from the Biblical World, 199-204. Leiden/Boston: Brill.

Hrozn, Bedich. 1917. Die Sprache der Hethither. Leipzig: Hinrichs. Kimball, Sara. 1999. Hittite Historical Phonology. Innsbruck: Institut fr Sprach-wissenschaft der

Universitt Innsbrck. . 2007. Hittite mant- all, entire, each. In Alan Nussbaum (ed.), Verba Docenti. Studies in historical

and Indo-Europ