Indo-Hittite Syntax?

Workshop The Precursors of Proto-Indo-European: Indo-Hittite and Indo-Uralic Hypotheses Leiden University, 9 July 2015

Dag Haug, Andrei Sideltsev Indo-Hittite Syntax?1 1. Hittite vs Other Indo-European Languages. A well-known feature of Hittite is the use of relative/interrogative pronouns en lieu of indefinite pronouns in conditional clauses (1) as well as after našma “or” and, considerably more seldom, after negation marker (2)2: (1)

MH/MS (CTH 147) KUB 14.1+ rev. 45 nu=wa=mu mān idālu-n memia-n CONN=QUOT=me

if

kui-š

[mema-i]

evil-ACC.SG.C word-ACC.SG.C who-NOM.SG.C tell-3SG.PRS

“If anybody tells me a bad word”3.

(2)

MH/MS (CTH 199) ABoT 1.65 obv. 11 ammug=a āššul UL

kuit

me.DAT.SG=but

which.ACC.SG.N write-2SG.PST

greeting.ACC.SG.N

NEG

ḫa[tr]ā-eš

“To me you did not send any greetings”4. The use is instantly reminiscent of that attested in all other Indo-European languages (3)5: (3a) (3b)

Greek ἐάν τις περιπατῇ ἐν τῇ ἡµέρᾳ “If anyone walks in the daytime,…” (John 11.9). Avestan kat̰ mōi uruuā isē cahiiā auuaŋhō (Yasna 50.1) QUESTION my

(3c) (3d)

soul.NOM

command.3SG.PRS

“Does my soul command any help?” Latin si quis “if anyone” Gothic ni manna in analaugnein no

man.NOM

in

secret.DAT

WH.GEN help.GEN

ƕa

taujiþ (John 7.4)

WH.NOM

do.3SG.PRS

This is attested in all the ancient languages of all the branches save Armenian and Tocharian and is reconstructed for narrow PIE as the use of relative/interrogative pronouns en lieu of indefinite pronouns under specific licensing conditions, most commonly conditionals, questions and negations, to a much more restricted degree in modal contexts (Haug forthcoming). Mind that the reconstruction is that of the pattern – so it holds irrespectively of what interrogative/relative pronoun is reconstructed or whether several ones are. However, this is not quite so. As is well known, Hittite attests the use in (1-2) in postOH period (CHD sub mān). The oldest attested Hittite texts (OH/OS originals) have only indefinite pronouns in conditional clauses (introduced by takku “if”) and after negation markers, as in: (4)

OH/OS (CTH 291.I.a.A) KBo 6.2 obv. ii 33 (§ 44a) takku LÚ-an paḫḫuen-i kuiški if

man-ACC.SG.C

fire-LOC.SG

peššie-zzi

somebody.NOM.SG.C throw-3SG.PRS

                                                                                                                1

We thank the audience at the Workshop for suggestions. Our particular gratitude goes to I. Yakubovich for (Hoffner, Melchert 2008: 149) cite one case. NH/NS KUB 21.38 obv. 48 attests UL kuit. See fn. 3 for yet another. The only case in the OH corpus is provided by an OH/NS text, see below. 3 Following Beckman (1996: 146): “if a person speaks an evil word to me”. 4 Following Hoffner (2009: 243). 5 See (Haug forthcoming). Haug unambiguously shows that the use is in no way limited to these two contexts, but supposes that they are the core contexts. Hittite also attests sporadic use of bare interrogatives as indefinites in other contexts. 2

 

1  

“If anyone makes a man fall into a fire, ...”6. (5) Ù

OH/OS (CTH 1.A) KBo 3.22 obv. 7-8 DUMUMEŠ URUNeš[aš id]ālu natta kuedanikki

takkiš-ta7

and

sons

set.up-3SG.PST

Nesa

evil.ACC.SG.N NEG

someone.DAT.SG

“And he plotted no evil to any of the citizens of Nesa”8.

The count of OH/OS corpus revealed 62× takku kuiški/kuitki vs 0× *takku kuiš/kuit. According to (CHD sub mān), the latter is sporadically attested only in later copies of OH texts and is very likely to reflect MH/NH usage. One of the earliest examples of relative/interrogative for the indefinite pronoun is: (6)

OH-MH/MS (CTH 262) IBoT 1.36 obv. ii 63 mān =aš?=ši peran=ma kuwapi KASKAL–i-š 9

if=it?=him

ahead=but

where

road-NOM.SG.C

“If the road ahead is at some point too narrow for him”10.

ḫatku-š narrow-NOM.SG.C

As for negative pronouns natta/UL kuiš, it appears only once in the NH copy of the edict of Telipinu11, see (7), versus OH/OS natta/UL kuiški, for which see (5) above: (7)

OH/NS (CTH 19.II.A) KBo 3.1+ obv. ii 44 parkunu-ši=ma=za UL kuit purify-2SG.PRS=but=REFL

NEG

which.ACC.SG.N

“But you do not purify in any way”12.

They are never attested in NS copies and even later versions of OH Laws. The statistics is impressive enough not to be just a matter of coincidence. 2. Other Anatolian Languages. All the rest of Anatolian languages pattern with OH/OS usage, i.e. they attest indefinite pronouns and not relative/interrogative ones in conditional clauses and after negation markers, even though the data are severely limited13. According to the Luwian corpus of Yakubovich, Hieroglyphic Luwian14 attests only one case of indefinite in a conditional clause. This is an indefinite pronoun, and not a relative/interrogative: (8)

KARATEPE 1 Hu. § LIX 331-333 REX-ta-ti-i-pa-wa/i kwa/i+ra/i hantawatta-tti=ba=wa kwari

kwa/i-sa-há ¦¦ kwisha

rule.as.king-3SG.PRS=but=PRTC

someone.NOM.SG.C

if

“But if anyone rules as a king”15.

                                                                                                                6

Following Hoffner (1997: 52). Other attestations from the same text are OH/OS (CTH 1.A) KBo 3.22 obv. 24 lē kuiški (with damaged lē), 34 [l]ē kuiški. 8 Following (Neu 1974: 10-11). 9 The signs are damaged. 10 Following (Miller 2013: 112-3). 11 Against two attestations of the regular indefinite pronoun after the negation including one in the next line of the same text (obv. ii 45 lē kuinki), the other one is in obv. 35 lē kuiški, and one is fragmentary (obv. ii 14). So Hoffmann (1984: 32) probably had good reason to assess it as a copyist’s mistake and read as UL kuit. 12 Following (Hoffmann 1984: 32-33). 13 No subordinator “if” is attested for Lycian as me has a less specific meaning (Melchert 2004: 38), thus it cannot be ascertained whether it allowed for bare indefinites in conditional clauses. 14 Naturally, the distinction between Cuneiform and Hieroglyphic Luwian is a purely traditional one, it is based solely on the script and does not imply any linguistic grouping. 15 Following (Yakubovich 2015: 46). Cf. (Hawkins 2000: 56): “If anyone from (among) kings”. The indefinite pronoun translates Phoenician mlk “a king” (Hawkins 2000: 66). 7

 

2  

The usage of indefinite pronouns after “or” is more frequent (5×): (9)

MALPINAR § 20 ni-pa-wa/i-tà-’ POST+ra/i-i-sá

kwa/i-sà-ha-’

CAPUT-ti-sa

niba=wa=ada or=PRTC=them

kwisha someone.NOM.SG.C

x-s (noble)man-NOM.SG.C

appara-s lesser-NOM.SG.C

ARHA MALLEUS ahha away

wallaismash

“Or (if) any inferior man shall erase them, …”16.

Cuneiform Luwian attests more cases17, but all of them involve indefinite pronouns, e.g., (10)

preNH/NS (CTH 760.II.1.A) KUB 35.43+ obv. ii 7 mān=ata īššara-ti kuiḫa18 if=it

hand-ABL

something.NOM.SG.N

“If it is something from the hand”. Only indefinite pronouns are used after negations in both Cuneiform and Hieroglyphic Luwian (Melchert 2003: 204), as in KUB 35.79 rev. iv 13’ nawa kuiḫa. For Hieroglyphic Luwian see (11)

KULULU 1 6 § 14 wa/i-tà ¦¦ ¦ni-i ¦kwa/i-ti-i-ha

¦pi-i[a]-a-i

wa=ada PRTC=it

piya-i give-3SG.PRS

ni NEG

kwadiha someone.DAT.SG

“And let him not give it to anyone (else)”19.

This use actually dominates for Hieroglyphic Luwian. Out of 30 attestations of indefinite pronouns in Yakubovich’s corpus it is attested in half of the cases (14×). The same usage is attested for Lycian where indefinite pronouns are used both after negation markers and in conditional clauses: (12)

N 320 34-35 se=we=ne: χtta-di: and=PRTC=NEG

tike:

harm-3SG.PRS anyone

“And no one shall do harm”. (13)

TL 89 2-3 a-di=me=j20=ẽ:

do-3SG.PRS=CONN=him/her=if21

tike:

χttbã:

someone.NOM.SG

harm.ACC.SG any.ACC.SG

tisñke:

“If anyone does any harm to him/her” Alongside standard conditional clauses Lycian employs what is commonly understood as paratactic conditional, i.e. clauses which are contextually conditional, but which do not contain an overt subordinator: (14) 1.

TL 57 7-9 [s]e=[ije]

n[e

hrppi tã]ti

tike

kbi:

and=him

NEG

on

put.3PL.PRS

someone.ACC.SG

other

ta-di:

tike:

2. hrppi=(i)je=me=i:                                                                                                                

Following (Hawkins 2000: 342). See (Melchert 1993: 119). 18 Following (Starke 1989: 144). 19 Following (Hawkins 2000: 443). 20 Or =me(j)=. 21 Following (Garrett 1990: 241 fn. 1; Melchert 2004: 19). Cf. (Клукхорст 2013: 149). 16

17

 

3  

on=thereupon=CONN=him

3.

put-3SG.PRS

someone.ACC.SG

me=n=e:

tubei-ti

mãhãi

CONN=him=they

strike-3PL.PRS god.NOM.PL

huwedri:

itlehi:

trm͂ mili:

belonging.to.a.league.NOM.PL.C

Lycian.NOM.PL.C

se

all.NOM.PL.C and

“(1) And they will not put someone on top of him. (2) (If) one puts someone on top of him, (3) they will strike him, all the gods and the Lycian league”. (15)

TL 83 10-11 ñte=me=j22=epi: ta-di:

ti[k]e

[tik]e:

in=CONN=him=upon put-3SG.PRS

someone

someone

“(If) someone inters someone on top of him”. Part of the clauses can in principle be understood as generalizing, as (15) “whoever puts …”. However, others, illustrated here by (14) cannot, because in this case the resumptive pronoun would have the wrong reference. So, the traditional understanding of the clauses as paratactic conditionals, also attested in Hittite, looks to be descriptively right. Thus at first sight, the bulk of the data that come from other Anatolian languages supports Old Hittite usage vs Middle/New Hittite one and set the Anatolian languages apart from other Indo-European languages. 2.1. Deviating Data. However, some other uses of Luvo-Lycian indefinite pronouns are different from Hittite. They are regularly attested in the contexts where Hittite normally employs relative/interrogative pronouns. The first context is distributive (“some – other(s)”): (16) 1.

KARKAMIŠ A1a § 19 – § 20 wa/i-ta-’ kwa/i-i-ha

pa-sa-na-’

wa=tta CONN=PRTC=LOC

abasa-n that.POSS-DAT.SG

kwinha something.ACC.SG

(“PODIUM”)ta-ti

PRAE-i

¦(*466)zú-ní-ha

tad-i block-DAT.SG

pari before

zunni-ha x-1SG.PST

2.

kwa/i-i-ha-wa/i-ta

pa-sa-na-’

¦(“PODIUM”)ta-ti

kwinha=wa=tta something.ACC.SG=PRTC=LOC

abasa-n that.POSS-DAT.SG

tad-i block-DAT.SG

PRAE-i

¦(*466)zú-ní-ha

pari before

zunni-ha x-1SG.PST

“(1) And sometimes I …ed them (?) at the podium of one (2) and sometimes I …ed them (?) at the podium of the other”23. Still it must be admitted the analysis is not completely assured, other assessments have been proposed, see (Hawkins 2000: 90). A more philologically secure usage is that indefinite pronouns in Hieroglyphic Luwian and Lycian are sometimes used in the generalizing function (“whatever”, “anything”), completely foreign to Hittite. The first piece of evidence comes from Lycian indefinite pronoun tise which can be iterated, tise tise, and has in this case a generalizing function “whatever” (Garrett 1994: 50 fn. 15, 18; Melchert 2004: 67): (17)

TL 110 3

                                                                                                                22 23

 

Or =me(j)=. Following (Hawkins 2000: 88).

4  

ep[ñt]e=me=i

ta-di:

me

thereafter=CONN=thereon

put-3SG.PRS

CONN something

tise

tise : something

prñnawa-ti build-3SG.PRS

“Furthermore, (if) one places (or) builds anything in there, …”. This is surprising from the Hittite viewpoint as in Hittite it is only relative/interrogative kui“which” that can be iterated with the generalizing function, never indefinite kuiški, as in: (18)

NH/NS (CTH 191) KUB 19.5+ obv. 9-10 (When Piyama-radu had humiliated me, set up Atpa over? me, and attacked (the country of) Lazpa,) 1. [… LÚ].MEŠṢARIPUTI ku-ēš ku-ēš ammel eš-er 2.

purple.dyers

who-NOM.PL.C who-NOM.PL.C

[nu=šši=kan

ḫ]ūmand-uš=pat

CONN=him=LOC

all-NOM.PL.C=EMPH in

anda

mine

be-3PL.PST

ḫanda-er join-3PL.PST

“(1) All of the purple-dyers (2) without exception (1) who were mine (2) joined with him.”24. Thus in this respect the Lycian indefinite pronoun tise behaves more like relative/interrogative one, even though it is an exact etymological match to Hittite indefinite pronoun kuiški (Melchert 1994: 302; 2004: 67; Клукхорст 2013: 145). This can most likely be explained by the fact that Lycian tise was beginning to lose its indefinite semantics and not analyzed any more as ti=se. Independent evidence for the reanalysis comes from (13) above where ti-se was optionally further marked by the redundant -ke. This was undoubtedly brought about by the fact that tise was no longer analyzed as inflected ti + enclitic -se. Reanalysis from the inflected wordform of ti + -se to one stem tise is demonstrated by tisñ in TL 65 11 instead of the expected tise in TL 5 8e (Melchert 2004: 67). Analogous evidence comes from Hieroglyphic Luwian indefinite pronouns, kuisha, which go back to a different model. These indefinite pronouns can be used along the same lines as Lycian tise above, i.e. they can be iterated to produce a generalizing meaning (“whoever”). However, as different from Lycian, in the iteration only one pronoun can be indefinite, the other has to be relative/interrogative, that is kuisha co-occurs with kuis in the iterated sequence kuisha kuis “whoever”, “whatever”, according to (Kloekhorst 2008: 489), as in: (19)

KULULU 1 § 7 – § 9 ¦wa/i-ti ¦¦ ¦za-ia ¦DOMUS-na-’

¦kwa/i-sá

wa=di PRTC=REFL

zaya parn-a this.ACC.PL.N house-ACC.PL.N

kwi-s which-NOM.SG.C

¦ni-pa-wa/i

¦á-ma-ta-’

¦tu-wa/i-ti-ia

¦wa/i-zi-ti-‹i›

niba=wa or=PRTC

amatt-a estate.component.ACC.PL.N

Tuwattiy-a T.DAT.SG

wazzi-di wish-3PRS.SG

¦ni-pa-wa/i

¦wa/i-ia-ni-[si?-]i

niba=wa or=PRTC

wiyani(ya)-asi vine-GEN.C

¦ni-pa-wa/i

¦la-hi?-zi-i

niba=wa or=PRTC

lah(i)-nzi estate.component-ACC.PL.C

¦tu-wa/i[+ra/i]-sà-za-’

¦kwa/i-sà-ha-wa/i-sa

¦kwa/i-sa-««pa»»

tuwarsan-za vineyard.ACC.SG.N-PRTC

kwisha=wa=as someone.NOM.SG.C=PRTC=he

kwi-s who-NOM.SG.C

“He who shall demand these houses from Tuwatis, or the amata or the lahi(zi) or the vineyard of vine(s), whosoever he be, …”25.                                                                                                                 24

25

 

Following (Hoffner 2009: 294). Following (Hawkins 2000: 443).

5  

(20)

EĞRİKÖY § 4 ¦kwa/i-ia-ha-wa/i-’

¦kwa/i-ia

¦(“CASTRUM”)ha+ra/i-n[í-sà]-’ ¦[...

kwayaha=wa some.N/ACC.PL.N=PRTC

kwaya which.N/ACC.PL.N

harnissa fortress.N/ACC.PL.N

“whatsoever fortresses …”26.

The corpus of Yakubovich brings us only two such cases. The rare use is attested alongside the expected iteration of two relative/interrogative pronouns, kuis kuis with the same function, identical to the Hittite usage above, as in (21)

KARAHÖYÜK (ELBİSTAN) 11 § 22 – § 23 kwa/i-i(a)-sa-pa-wa/i i(a)-ma kwa/i-i(a)-sa

¦CAPUT-ti-sa a-mi-i(a)

kwi-s=ba=wa imma which-NOM.SG.C=but=PRTC ever

kwi-s CAPUT–tti-s amiya which-NOM.SG.C (noble)man my

DOMUS-ní-i(a)

kwa/i+ra/i-i(a)-pa

¦URBS+MI-ní-i(a)

tara/i-pa-a-ti ·

parniya domestic.DAT.SG

kwari=ba or=but

URBS+MI town-DAT.SG

tarpa-di trample-3SG.PRS

kwa/i+ra/i-pa-tà

*293

*469

PONERE

kwari=ba=ada or=but=they

x x

x x

tuwaput

“Whatsoever person shall march against my house or city, or shall put *293, *469…”27. Actually, both (19) and (20) can be analyzed as regular relative pronoun + -ha “and”. Such cases are attested (22)

AKSARAY 5 § 7 kwa/i-sà-ha-wa/i-mu-u

za-¦¦ [...]

kwi-s=ha=wa=mu which=and=PRTC=me

za alant-za piya-i this.ACC.SG.N place.ACC.SG.N-PRTC give-3SG.PRS

“LOCUS”-lá/í-za

pi-ia-i

“And who gives this position to me?”28.

The analysis is explicitly proposed for (19) by Yakubovich in his glossing in the corpus. However, the following contexts are much more difficult to assess in this way: (23)

ASSUR letter f+g § 31 ¦wa/i-za ¦ka-mara/i-ra+a-na wa=anza PRTC=us

¦kwa/i-i-ha

kamara-n kwinha commodity-ACC.SG.C some.ACC.SG.C

¦VIA-wa/i-ni-i harwanni dispatch.2SG.IMP

“(Further, why did they(?) send us WASHASA? They(?) bought for us four KAPAPA’s. After us in person, one may … them, but before us no one may WAZASA them.) Send us any KAPAR(R)A, (or (if) to you there is not (any), and there remains no [Par]niwaris the AMURALURA, but it from him, and send it to us)”29. It appears from the translation that here the indefinite pronoun has the meaning “any”, which is cross-linguistically referred to the same group of meaning as “whatever” and are termed free choice, see, e.g., (Haspelmath 1997). The following context is still more curious. Here the indefinite pronoun is an attribute to iterated NPs “god god” with the generalizing function:                                                                                                                 26

Following (Hawkins 2000: 495). Following Yakubovich, pers.comm. Cf. (Hawkins 2000: 290). 28 Following (Hawkins 2000: 476). 29 Following (Hawkins 2000: 537). 27

 

6  

(24) a-wa/i

HAMA 4 § 5 DEUS-ni

a=wa massan-i CONN=PRTC god-DAT.SG

DEUS-ni

kwa/i-ti-ha

massan-i god-DAT.SG

kwadiha abissi-n some.DAT.SG that.POSS-ACC.SG.C

“SOLIUM”-sa-na

i-zi-i-ha

asa-n seat-ACC.SG.C

izzi-ha make-1SG.PST

á-pi-si-na

“And for every single god I made his own seat (but this seat for Bala’atis I made)”30. Finally, we find even not iterated indefinite pronouns which are still distributive: (25) 1. 2.

ASSUR letter e § 15 - § 17 ¦á-lá/í-wa/i-za ¦kwa/i-sà-ha

¦a-sa-ti

ara=wa=anza kwisha right=PRTC=us some.NOM.SG.C

as-ti be-3SG.PRS

NEG2-wa/i-ma-za-‘

[¦?]u-za5-za

¦á-lá/í-ha-wa/i-za

¦u-za5-za

na=wa=mmanza NEG=PRTC=you

3.

unzanza you.OBL

ara=ha=wa=anza unzanza right=and=PRTC=you you.OBL

“(1) Whoever is right for us, (2) is (he/she) not right for you? (3) (He/she) is also right for you!”31. Summing up the section on indefinite pronouns in the Anatolian languages other than Hittite it should be said that, cross-linguistically, the semantic sphere of indefinite pronouns is constituted by existential quantifiers (something), NPI ((not) … anything), free choice (whatever, anything), universal (everyone), wh-elements (what) (Haspelmath 1997). The semantic sphere is quite differently coded by various pronouns in Hittite and Luwian. In the former existential quantifiers and NPIs are marked by kuiški series. Universal quantifiers are marked by a specialized pronoun kuišš=a. Free choice is normally marked by either iterated relatives/interrogatives (kuiš kuiš) or by relatives (kuiš “whoever”). Hittite relative pronouns also very commonly mark the distributive function (“some – other(s)”). In Hieroglyphic Luwian32, on the contrary, the marking is different. What is normally termed indefinite pronouns mark existential quantifiers, NPIs and free choice. Free choice can also be marked by iterated relatives/interrogatives or by an indefinite + a relative. Indefinite pronouns also attest distributive function even when they are not iterated. Thus the distributive function in Hieroglyphic Luwian appears to be partly marked by indefinite pronouns. Thus a comparison between Hittite and Hieroglyphic Luwian reveals spread of indefinite pronouns onto the free choice sphere. This receives a very simple explanation: diacronically, Luvo-Lycian indefinite pronouns go back to the model which in Hittite functions as universal “each, every”. Now what do the Luvo-Lycian data say about the use of relative pronouns instead of indefinite ones? The use which is attested in Middle and New Hittite appears to be completely                                                                                                                 30

Following (Hawkins 2000: 405). Following (Rieken, Yakubovich 2010: 214). Cf. (Hawkins 2000: 536, 547). = ASSUR letter f+g, § 14 ¦á-lá/íwa/i-za ¦kwa/i-sà-ha ¦a-sa-ti. 32 The data on Cuneiform Luwian are very limited and Lycian tise tise is more likely to lose its indefinite semantics. 31

 

7  

unattested in Cuneiform, Hieroglyphic Luwian and Lycian. It is rather the indefinite pronouns which expand at the expense of relative ones in Hieroglyphic Luwian, as dramatically different from Hittite. And they do this in a different direction – spreading to mark freechoice quantifiers. Actually, if the etymology of the Hieroglyphic Luwian indefinite pronoun is considered, everything quickly falls into its place. The pronoun, as well as its Lycian and Cuneiform Luwian counterparts (Luw. kui=ḫa, Lyc. ti=ke) is built on a different model than Lycian tise and Hittite kuiški33, namely the declined form of the relative/interrogative pronoun + enclitic -ḫa “and”, the model which is attested in Hittite as distributive kuišš=a “each, every”, i.e., relative/interrogative kuiš + -ya “and” (Melchert 1993: 67; 1994: 285, 306; 2004: 119; Клукхорст 2013: 145). Etymologically, the Hittite forms kuišš=a, kuinn=a, kuitt=a, etc are commonly taken back to *kwis/n/d-Ho with the special development for the enclitic conjunction *-Ho “and”34 which surfaces in synchronic Hittite as -ya (Melchert 2004: 159, 164, 306; Kloekhorst 2008: 490, 378-9). It is assumed that the generalizing Hittite semantics is the original one and that the Luvo-Lycian indefinite function was a secondary development out of the generalizing one (Kloekhorst 2008: 490). This is certainly supported by Lydian qida “whatever” and Palaic kuiš-a “everybody” (ibid: 378, 489). In this case generalizing use of the Hieroglyphic Luwian indefinite pronoun simply retains the original distributive semantics. 3. Relatives/Interrogatives as Indefinites? Still there are two forms which have been expressly identified with the Middle/New Hittite use of relatives/interrogatives as indefinites and thus could refute the argument above. Now we will assess them in detail. 3.1. Lycian tihe. The first deviating form is Lycian tihe, formally genitive of the relative ti- (Melchert 2004: 66). In some contexts it is employed with indefinite tike and thus is interpreted as relative/interrogative functioning as indefinite (ibid), e.g., (26)

TL 59 2-3 me=i(j)=a-di

tike:

zum͂ m[ẽ/ã]

tihe

CONN=him/her=do-3SG.PRS someone-NOM.SG

who-GEN.SG harm.ACC.SG

“(If) anyone does harm of any kind35 to him/her”.

At face value the Lycian example is identical to the following Hittite one and seems to finally provide a parallel from other Anatolian languages for the Middle/New Hittite use of relatives/interrogatives as indefinites. (27) mān

NH/NS (CTH 106.B.2) KBo 4.10+ rev. 18 URULUM kui-š našma AŠRU kuitki

if

city

m

which-NOM.SG.C

or

Ulmi–DU–upLUGAL KURURUDU–tašša

Ulmi-Tessup

king

land

Tarhuntassa

place

ANA

some.NOM.SG.N

to

piy-anna

UL

ZI–anza

give-INF

NEG

soul.NOM.SG.C

“If he does not wish to give some city or some locality to Ulmi-Tessup, …”36.

Actually, as acknowledged, the Lycian form is not a proper genitive synchronically, and it never functions as genitive of relative/interrogative ti-, it rather functions as an alternative                                                                                                                 33

But cf. (Yakubovich 2012 with ref.) who dissociated Lycian ti=ke from Luwian forms and directly compared it with Hittite kuiš-ki. 34 Most likely *-h3e (Kloekhorst 2008: 379). 35 Following the suggestion of I. Yakubovich, pers.comm. 36 Following (van den Hout 1995: 46-7; Beckman 1996: 107).

 

8  

accusative singular form of tike alongside tike37. Thus its usage is identical to ACC.SG tike, as, e.g., in (28)

TL 56 3 se=ije ti and=him/her

e-di:

tike:

who.NOM.SG do-3SG.PRS

mẽtẽ:

some-ACC.SG harm.ACC.SG

“who does any harm to him/her, …”38.

Another strange thing is that the absolute majority of contexts (besides the cited TL 59 2-3, these are TL 91 3; TL 95 2; TL 135 2; N 314b 2-3; possibly also with negation in TL 45B 78, only in one context, TL 44C 17, is does not co-occur with tike) consistently display both indefinite tike and interrogative/relative ti- in the form of tihe, not normally attested in Hittite39. The use is distantly reminiscent of Hieroglyphic Luwian where kuisha coccurs with kuis in the iterated sequence kuisha kuis “whoever”, “whatever”, according to (Kloekhorst 2008: 489) alongside the expected kuis kuis with the same function, see above for examples. In this light, and in the light of complete lack of other case forms of ti- functioning as indefinites as well as in the light of unattested genitive forms of tike it might be attractive to attempt to assess tihe as genitive of the indefinite tike, and not as genitive of the relative/interrogative ti-. In view of many allomorphs of the genitive morpheme, the original genitive of tike could be of at least two types: (a) *tise < *tis-ke < *tis-he with the genitive case marker *-s (Adiego 2010; Melchert 2012) and the simplification of the clause internal cluster along the same lines as *st, sH > s (Melchert 1989: 28-29; 1994: 288). Actually, (Melchert 1994: 304) supposes that *s is preserved in any internal cluster with the following obstruent. This form would be omonymous to nom.sg. ti-se “who/whatever/anyone” (Melchert 1994: 302; 2004: 67). But in view of the fact that *s was lost in tike < *tis-he and thus developed along the lines for -s at the absolute end of the word and not along the lines of word internal clusters it is quite likely that the same should hold for the genitive form and thus the likeliest outcome would be tike, the form omonymous to the nominative-accusative tike. (b) *tiheke < *tiseke < *tiseke with the genitive case marker *-(o)so (Adiego 2010; Melchert 2012). *tiheke could then get syncopated into *tihke. Due to the regular loss of h between vowels (Melchert 1994: 317), the form might have developed into *tike, homonymous to the nom.sg. form. Possibly, the homonymity should be held responsible for the attested tihe, although genitive forms do universally preserve intervocalic h (Melchert 1994: 317-8; 2012). Thus neither scenario derives the attested tihe out of the paradigm of tike. So in the logic of Melchert (2004: 66) it provides the only case outside Middle and New Hittite for the interrogative/relative use as indefinite and is equated with Latin se quis. However, the form is distinctly different from Hittite. First, in Lycian in the absolute majority of contexts tihe functions as a form of the indefinite tike, not of relative ti. Moreover, tihe does not function as a genitive form of tike. It is rather a frozen form, synchronically                                                                                                                 37

The strangeness is responsible for the fact that tihe is not included in the paradigm of either ti or tike in (Клукхорст 2013: 145). Tihe is not isolated, however. Other forms with a similar mismatch of form and fuction include ebehi, which, according to Melchert (2004: 11), originally accented possessive adjective of ebe- “of this”, but synchronically equivalwnt to simply ebe- “this”. It is also curious that no dat.-loc. forms of ebe- “this” are attested (Melchert 2004: 11). 38 Similar to fragmentary TL 72. 39 Hittite commonly attests either two indefinite or two relative/interrogative pronouns. In very rare cases like (27) above it does attest an indefinite pronoun and a relative one functioning as indefinite in the same context, but the use is never systematic.

 

9  

identical in function to acc.sg. tike, even seemingly tautonymous to it. Still at the very least in one context tihe is employed in the same clause not only with indefinite pronoun tike, but also with indeterminate relative one ti: (29)

N 314b 1-5 χupa ebe-hi

1.

tomb

2.

ti(j)=j=a-di:

this-GEN.ADJ who=it=do-3SG.PRS

me=n=e

tubi-di

eti

tike

zum͂ mẽ: tihe

some

harm

some/whatever

ebe-hẽ

CONN=him=he strike-3SG.PRS father this-GEN.PL

“(1) The tomb of this place40, whoever does any harm to it, (2) he will strike him, the Father of these”. The context is easy to translate as “whoever does any harm whatsoever to it”. The use is identical to the Hittite one in (30) and Luwian one in (31) (30)

MH/MS (CTH 257.4.A) KUB 31.100 rev. 11`-13` namma kui-ēš kui-ēš kueluwan-eš

1.

further what-NOM.PL.C what-NOM.PL.C

2. 3.

šer É.G[AL? ]

k.-NOM.PL.C up pa[lace

kui-ēš

namma

kui-ēš

what-NOM.PL.C

further

what-NOM.PL.C

kuwapi where

n=uš

ḫūmand-u[š

wanalli-škandu

ištalki-škand[u]

CONN=them

all-ACC.PL.C

w.-IMPF.3PL.IMP

i.-IMPF.3PL.IMP

“(1) Further, whatever k.`s are up in the pa[lace(?)] (2) (and) whatever other k.`s are (any)where, (3) all those let them w. and i.”41 As follows from the translation of CHD, the context is assessed as relative pronoun functioning as a relative pronoun + relative functioning as an indefinite pronoun. However, nothing – apart from the would-be awkward English translation – prevents us from assessing such examples as double relatives “whoever … wherever”, which are perfectly possible, e.g., in Russian “какие бы ни к. где бы ни были”. (31)

BABYLON 1§ 8 kwa/i-a-za kwa/i-i-ta

PES-i

kwanza which.ACC.SG.N

x-i be.located-3SG.PRS

kwitta where

“(that) which comes (any)where”42 with the analysis of the second relative as “wherever” (Hawkins 2000: 393). The following case is also commonly translated as a relative pronoun + indefinite pronoun «But for whom (there is) some rite up in Ḫattusa»43. However, similarly, the first clause could mean «for whom it is which duty in Hattusa» with double relative pronouns: (32) 1.

URU

MH/NS (CTH 264.A) KUB 13.4 rev. iii 21-23 Ḫatš-i=ma=kan kuedani kui-š

Hattusa-LOC.SG=but=LOC

who.DAT.SG

which-NOM.SG.C

šaklāi-š

šer

duty-NOM.SG.C up

2. mān LÚSANGA LÚGUDU12 LÚ.MEŠḫaliyattallēš 3. kui[šš=a=aš] tarniškezzi “(1) He who is responsible for letting in someone who has some duty up in Ḫatsa, though – (2) be he a priest, an anointed one (or) the watchmen – (3) he must let only them in”44.                                                                                                                 40

Following (Melchert 2004: 11). Cf. (Kloekhorst 2011). Following (CHD L-N: 388). 42 Following (Hawkins 2000: 392). 43 See (Miller 2013: 394 fn. 544, 548) for an overview of earlier proposals. 44 Following (Miller 2013: 256-7). 41

 

10  

Thus from contexts like (29) it is easy to see how tihe came to be used together with indefinite pronouns. Ti adi zum͂ mẽ tihe “whoever does harm of whatever (kind)”45 was used alongside Ti adi tike zum͂ mẽ “whoever does any harm”. It is important to observe that by itself the use of indefinite pronoun with the relative one shows nothing about the semantics of either. Then hybrid and attested Ti adi tike zum͂ mẽ tihe “whoever does any harm of whatever (kind)” came into being. Finally, due to the synonymity of “whoever does harm” and “somebody does harm”, the use of tihe spread to sentences with indefinite pronoun proper: adi tike zum͂ mẽ tihe “someone does whatever harm”. It is important to bear in mind that the spread must have occurred after the form tihe was fossilized and lost its connection with the paradigm of relative/interrogative ti-. As was already observed, as different from Middle and New Hittite, tihe does not function as the genitive of relative/interrogative ti-. Thus it is not a synchronic parallel to Hittite use of relatives/interrogatives as indefinites. Diachronically, there is no reason either to assess the development as parallel to the Middle/New Hittite one. 3.2. Yet another Anatolian form compared to the post OH/OS Hittite usage is Hieroglyphic Luwian kwa/i-i-ia in the following context: (33)

KARKAMIŠ A6 § 8 - 4 za-a-pa-wa/i (“MENSA.SOLIUM”)á-sa-na-’

¦ku-ma-na

I

za-n=ba=wa asa-n this-ACC.SG.C=PRTC seat-ACC.SG.C

kuman when

Kamaniya K.DAT.SG

ka-ma-ni-i-ia

¦á-mi-i-’

(DOMINUS)na-ni

¦INFANS.NÍ ¦kwa/i-i-ia · AEDIFICARE+MI-ha

amiya my.DAT.SG

nanni master.DAT.SG

nimuwizza son.DAT.SG

kwiya tama-ha when build.1SG.PST

“And when I built this seat for Kamanis my lord’s child, …”46.

Hawkins (2000: 126) qualifies kwa/i-i-ia as obscure, but supposes that it may function inside a subordinate clause introduced by kuman “when” as indefinite. However, this still does not produce any sence and thus we follow Yakubovich in his analysis of the form as “when”. Double kuman “when” is attested in HLuwian in Potoroo 7b. Kwiya is attested as “when” in the same text (Karkamiš, KARKAMIŠ A6) in § 18. 4. Anatolian vs Narrow-IE Relatives/Interrogatives for Indefinite Pronouns. In this light the post-OH/OS use of relatives/interrogatives in conditional clauses and after negation marker for indefinite pronouns in Hittite cannot be directly equated with the seemingly identical usage in narrow IE languages and has to be an independent innovation within historical Hittite. Thus narrow IE languages attest relative/interrogative pronouns in conditional clauses and after negation47 whereas Proto-Anatolian48 attests only indefinite pronouns in these contexts, another feature which might provide further evidence for the Indo-Hittite hypothesis. Relative/interrogative pronouns en lieu of indefinite pronouns have to be an independent innovation in historical Hittite and in narrow IE. 4.1. Hiitite Contexts Other than Conditionals. Yet another argument in favor of our hypothesis is the fact that even the post-OH/OS use of interrogatives/relatives is practically restricted to contexts after mān “if”, they are only sporadically attested after negations and virtually unattested in other contexts. The rare cases when bare interrogatives are attested in contexts other than conditionals introduced by mān “if” paradoxically show all the more clearly how limited to conditional contexts they are. The first context where interrogatives/relatives are attested instead of indefinite                                                                                                                 45

We follow suggestions of I.Yakubovich for the translation of tihe. Following (Hawkins 2000: 124). 47 As well as in some other contexts, see (Haug forthcoming). 48 Reflected in OH/OS texts and other Anatolian languages. 46

 

11  

pronouns is temporal clauses, as in (34-5): (34) 1. 2.  

MS/MH (CTH 261.II) KUB 26.17 obv. i 4’-5’ mān DUTU–Š=I=ma kuwapi apāšila

laḫḫiyai-zzi  

when

Majesty=My=but

go.on.campaign-3SG.PRS  

nu

[KUR-e]

ŠA

CONN [land-LOC.SG] of

when LÚ

himself

K[ÚR]

e[nemy]

ḫūda-š

ēš-tu

preparedness-NOM.SG.C

be-3SG.IMP

apiya=ya there=and

“(1) When His Majesty himself, though, at any time goes on a campaign (2) [in the land] of the e[nemy] there too, preparedness must obtain (and you must fight the enemy unreservedly)”49. (35)

MH/NS (CTH 264.A) KUB 13.4 rev. iii 68 mānn=a=za MUNUS-i kui-š

GAM-an šeš-zi

when=and=REFL

down

woman

who-NOM.SG.C

“Also, when someone goes to sleep with a woman, …”50.

sleep-3SG.PRS

They are commonly introduced by the same subordinator which in Middle and New Hittite dominates in conditional clauses – mān, even though it is used in a purely temporal meaning here – “when”. Still, use of relatives/interrogatives is attested with other temporal subordinators, such as māḫḫan, obviously by analogy after mān “when”, see the following three attestations from the same text: (36a)

NH/NS (CTH 255.2.A) KUB 26.1+ rev. iii 50-51 [(š)]ummaš=ma kuit GIM-an

išdamaš-ten #

you=but

hear-2PL.PST

which-ACC.SG.N

when

“But when you have heard something, …”51. (36b) nu=za

NH/NS (CTH 255.2.A) KBo 26.1+ rev. iii 16 kuit GIM-an kiš-ari #

CONN=REFL which-ACC.SG.N

when

hear-2PL.PST

“And should something happen …”52. (36c)

NH/NS (CTH 255.2.B) KBo 26.8 obv. ii 5' […]x=ma kuit GIM-an *u-š*ke-tteni # x=but

which-ACC.SG.N

when

see-IMPF-2PL.PRS

“But when you observe something, …”53. The analogy is all the more expected, as, e.g., the context of (36c) largely employs the mān “if” with very similar meaning. It is also curious that in (36b) the context requires that the meaning of māḫḫan should be not the common “when”, but quite unexpected and otherwise unattested54 “if”! See for confusion of mān and māḫḫan in other meanings (CHD L-N: 146) and (ibid: 159-160) for the difficulty in distinguishing between temporal and conditional mān. The other context where interrogatives/relatives are attested instead of indefinite pronouns is seen in (37) 1.

MS/MH (CTH 261.II) KUB 26.17 obv. i 5’-7’ LÚ māḫḫan=[ma] KÚR-aš ak-i

                                                                                                                49

Following (Miller 2013: 130-1). Following (Miller 2013: 260-1). 51 Following (Miller 2013: 302-3). 52 Following (Miller 2013: 300-1). 53 Following (Miller 2013: 298-9). 54 According to (CHD L-N: 100). 50

 

12  

2. 3.

when=but

enemy-NOM.SG.C

die-3SG.PRS

kūrur

kui-š

ḫar-zi

hostility.ACC.SG.N

which-NOM.SG.C

nu

kui-š    

 

CONN which-NOM.SG.C

4.

ÉRIN

MEŠ

have-3SG.PRS  

ašandula occupation  

troops

n=aš=kan

anda ašandulaš    

CONN=them=LOC

in

 

occupation.GEN.SG

dāla-ḫḫi  

leave-1SG.PRS  

“(1) As soon as the enemy has been vanquished, [though] (2) (if) some (enemy) retains hostility (3) then the occupation contingent that (is to be left behind) (4) I will leave behind for the occupation”55. It is unlikely that in (37) the force of the subordinator is carried over from the previous clause56. The clause is rather a paratactic conditional, i.e. a conditional clause without mān “if” or any other overt subordinator, see for such cases (Hoffner, Melchert 2008: 423). The following case from the same text may belong to the same category, but here the force of mān “if” can be carried over from the previous clause: (38)

MS/MH (CTH 261.II) KUB 26.17 obv. i 9'–11' [m]ān LÚKÚR=ma kuwatka zaluknu-zi

1.

if

2. D

enemy=but

somehow

linger-3SG.PRS

kūrur

kui-š

[ḫar-z]i

hostility.ACC.SG

which-NOM.SG.C

have-3SG.PRS

3. UTU–Š=I=ma

[EGIR]-an

ANA

DINGIRMEŠ=ŠU iya-uwanzi ue-zzi

Majesty=My=but

back

to

gods=his

4.

našma=šši

[kuwap]i

ā[ššu]

or=him

where

good

[n=aš]

apadda

pai-zzi

CONN=he

there

go-3SG.PRS

5.

do-INF

come-3SG.PRS

“(1) [I]f an enemy, though, somehow perseveres, (2) (if) some (enemy) [retai]ns hostility. (3) My Majesty, however, comes [bac]k in order to venerate his gods (4) or [whereve]r it seems b[est] to him (5) there he goes”57. Thus the two cases deviating from the common pattern (conditional clauses introduced by mān “if”), paradoxically, show only clearer that the only real pattern is conditional clauses introduced by mān “if”. The two other contexts are obviously derived from mān “if”. One is paratactic conditional clauses (the same semantics as mān “if”). The other – temporal clauses, usually introduced by the same subordinator as conditional clauses, mān, but in a temporal function “when”. Actually, it is very clear why the distribution is exactly as attested. As is well known, in Old Hittite texts conditional clauses were introduced by the subordinator takku “if” whereas temporal clauses were introduced, among other subordinators, by mān “when”. The state of affairs in Middle and New Hittite was different: the conditional subordinator was mān “if” whereas the main temporal subordinator was by māḫḫan “when”. The transitional period occurred during late OH time: “Temporal clauses in Old Hittite (OH) are frequently marked by mān “when”, which in OH was only beginning to gain the secondary meaning “if” that it acquired in NH” (Hoffner, Melchert 2008: 416). Now, at the time when mān was taking over the conditional function of takku “if” it was used simultaneously58 both in temporal and conditional clauses. Thus it was highly likely to “borrow” the syntax characteristic of mān                                                                                                                 55

Following (Miller 2013: 130-1). See for such cases (Hoffner, Melchert 2008: 421). 57 Following (Miller 2013: 130-1). 58 Among other functions, for which see, e.g., (Hoffner, Melchert 2008: 416). 56

 

13  

“if”. So, (35) above coccurs with the following prototypical conditional use of mān in (39): (39)

MH/NS (CTH 264.A) KUB 13.4 rev. iii 74 mān=ma=za ITTI MUNUS–TI kui-š

šeš-zi

when=but=REFL

sleep-3SG.PRS

with

woman

who-NOM.SG.C

“If, however, someone sleeps with a woman, …”59.

Besides, if one still assesses the use of relative pronouns in (30-32) as relatives–for– indefinites, the use will also easily fall into the extentions of the original conditional usage: generalizing relatives are very close to conditionals, see the argument above. So, Hittite very clearly attests the bulk of usage of relative/interrogative pronouns instead of indefinite pronouns only in conditional clauses and in cases closely related to them. 4.2. Now, if despite the evidence provided so far, one still traces the Hittite use of interrogatives/relatives to the Proto-Indo-European one, the development from Proto-IndoEuropean would be narrowing of the sphere of usage of original interrogatives/relatives because the usage of interrogatives/relatives in narrow IE is considerably wider than that of Hittite. As shown by (Haug forthcoming), narrow PIE had several other contexts where bare relatives were used for indefinite pronouns. The most common of them was questions, less common modal contexts. This would contradict Haspelmath’s (1997) generalization that interrogatives/relatives become more general in use over time as indefinite pronouns. Thus even in this case it agrees better with the material to speak of independent innovations in historical Hittite and in narrow PIE. It is all the more likely as the innovation of Hittite is in some important details different from the innovation of narrow PIE and is tied in with the peculiarity of Hittite syntax (mān both in temporal and conditional functions “if” and “when”, paratactic conditionals). 5. Conclusion. Thus the post-OH/OS use of relative/interrogative pronouns in conditional clauses and after negation marker in Hittite cannot be directly equated with the seemingly identical usage in narrow IE languages and has to be an independent innovation within historical Hittite and narrow IE. Narrow IE languages attest relative/interrogative pronouns in conditional clauses, questions and after negation, but Proto-Anatolian attests only indefinite pronouns in these contexts. References Adiego, I. 2010, R. Kim, N. Oettinger, E. Rieken, M. Weiss (eds.), Ex Anatolia Lux. Anatolian and Indo-European Studies in Honor of H. Craig Melchert on the Occasion of his Sixty-Fifth Birthday. Ann Arbor New York: Beech Stave Press. 1-8. Adiego, I. 2015, Lycian Nasalized Preterites, IF forthcoming. Beckman, G. 1996, Hittite Diplomatic texts, SBL Writings from the Ancient World 7, Atlanta. Garrett, A. 1990, The Syntax of Anatolian Pronominal Clitics. Ph.D. Diss., Harvard University, 1990. Garrett, A. 1992, Topics in Lycian Syntax, HS 105, 200-212. Garrett, A. 1994, Relative Clause Syntax in Lycian and Hittite, Die Sprache, Bd. 36, 29-69. Haspelmath, M. 1997, Indefinite Pronouns. Clarendon Press: Oxford. Haug, D. forthcoming, PIE *kwi-/kwo: Interrogative, indefinite or both? To appear in a Festschrift. Hawkins, J. D. 2000, Corpus of Hieroglyphic Luwian Inscriptions, Vol. 1, Studies in IndoEuropean Language and Culture NS 8.1, Berlin-New York: Walter de Gruyter.                                                                                                                 59

 

Following (Miller 2013: 260-1).

14  

Hoffmann, I. 1984, Der Erlass Telipinus, THeth 11. Hoffner, H. A. Jr. and Melchert, C. 2008, A Grammar of the Hittite Language, Part 1: Reference Grammar, Winona Lake, Indiana: Eisenbrauns. Kloekhorst, A. 2011, The Opening Formula of Lycian Funerary Inscriptions: mẽti vs. mẽne, JNES 70, 13-23. Melchert, H. Craig 1989, New Luvo-Lycian Isoglosses, Historische Sprachforschung 102/1, 1989, 23-45. Melchert, H. Craig 1993, Cuneiform Luvian Lexicon, Lexica Anatolica vol. 2, Chapel Hill, 1993. Melchert, H. Craig 1994, Anatolian Historian Phonology, Amsterdam-Atlanta, 1994. Melchert, H. Craig 2003, Language, in: C.Melchert (ed.), The Luwians, HdO 68, 170-210. Melchert, H. Craig 2004, A Dictionary of the Lycian Language, Ann Arbor – New York: Beech Stave Press, 2004. Melchert, H. Craig 2012, Genitive Case and Possessive Adjective in Anatolian, V. Orioles (ed.), Per Roberto Gusmani. Studi in ricordo. Linguistica storica e teorica. Vol. II, tomo 1, 273-286. Miller, J. 2013, Royal Hittite Instructions, SBL Writings from the Ancient World 31, Atlanta. Neu, E. 1974, Der Anitta-Text, StBoT 18, 1974. Rieken, E. And I. Yakubovich 2010, The New Values of Luwian Signs L 319 and L 172, in: I. Singer (ed.), Ipamati kistamati pari tumatimis. Luwian and Hittite Studies Presented to J. David Hawkins on the Occasion of his 70th Birthday. Tel Aviv, 2010. van den Hout, Th. 1995, Der Ulmitešub-Vertrag, StBoT 38, Wiesbaden. Yakubovich, I. 2012, The Reading of Luwian ARHA and Related Problems, AoF 39/2, 2012, 321-339. Yakubovich, I. 2015, Phoenician and Luwian in Early Iron Age Cilicia, AnSt 65: 35-53. Клукхорст, А. 2013, Ликийский язык, в: Языки мира. Реликтовые индоевропейские языки Передней и Центральной Азии. М., 2013, 131-154.

 

15