ON THE BURUSHASKI-INDO-EUROPEAN HYPOTHESIS BY I. ČAšULE Текст научной статьи по специальности «Языкознание и литературоведение»

John D. Bengtson Association for the Study of Language in Prehistory and Evolution of Human Language Project

Vaclav Blazek Masaryk University

On the Burushaski-Indo-European hypothesis by I. Casule*

The paper deals with a relatively recent hypothesis, put forward by the scholar I. Casule, according to which the Burushaski language, traditionally considered an isolate, actually belongs to the Indo-European linguistic stock. The authors approach Casule's hypothesis from the comparative side, evaluating phonological, morphological, and lexical arguments in its favour side by side with the corresponding arguments in favour of the Dene-Caucasian hypothesis, according to which Burushaski forms a separate one-language branch of the vast macrofamily that also includes Na-Dene, Sino-Tibetan, North Caucasian, Basque, and Yeniseian languages.

It is concluded that arguments for the Dene-Caucasian status of Burushaski quantitatively override the Indo-European-Burushaski hypothesis by a very large margin; suggested Indo-European connections are either highly unsystematic (when it comes to phonetic correspondences), sporadic and insufficient (in morphology), or practically non-existent (in basic lexicon). Consequently, all of the resemblances between Indo-European and Burushaski must be ascribed to (a) recent contacts between Burushaski and Indo-Aryan languages, (b) chance resemblances, or (c) in a very small number of cases, traces of «ultra-deep» relationship that do not represent exclusively «Indo-European-Burushaski» connections.

Keywords: Indo-European linguistics, Burushaski language, macrocomparative linguistics, Dene-Caucasian macrofamily, language isolates.

Over the last two decades, Ilija Casule has published a monograph (Casule 1998) and an article (Casule 2003) in which he attempts to show that the Burushaski language — traditionally considered an isolate — is a member of the Indo-European language family. One of the authors has already published a critique of the 1998 monograph (Bengtson 2000). In this article we shall mainly be dealing with the 2003 article in JIES, and all page number references will be to the latter work.

While we agree with Casule that there are some affinities between Burushaski (Bur) and Indo-European (IE), we do not consider Bur a part of the IE family, or even of the postulated deeper macro-family to which IE belongs (Nostratic or Eurasiatic), and we intend to show that

* We are deeply indebted to the work of the late Sergei A. Starostin, who, in the last few months of his life, worked intensively on the Burushaski language and its relationship with Dene-Caucasian languages. The results can be seen in his DC phonology and glossary, and EHL/ToB etymological databases (see References). Since his father's passing Georgiy (George) Starostin has continued to work with us and we are grateful to him. We are thankful for useful comments from Elena Bashir, Bertil Tikkanen, and Michael Witzel. We are also deeply thankful to the Evolution of Human Language Project, Santa Fe Institute, and Murray Gell-Mann, and the Centre for the Interdisciplinary Research of Ancient Languages and Older Stages of Modern Languages (MSM 0021622435), Masaryk University Brno, for their support.

Journal of Language Relationship • Вопросы языкового родства • 6 (2011) • Pp. 25-63 • © Bengtson J. D., Blazek V., 2011

a large part of the resemblances between Bur and IE can be explained as areal, i.e., the results of long-term contact and borrowing — in both directions — between Bur and surrounding IE languages.1

However, we shall not simply demolish Casule's hypothesis without providing what we consider a better, more plausible, and more probable alternative for the classification of this fascinating (Bur) language. We shall present evidence that Bur is more likely a member of the Dene-Caucasian (or Sino-Caucasian) macro-family. This is of course not a new idea: it was prefigured long ago by scholars such as Karl Bouda, O. G. Tailleur, V. N. Toporov, and others. Recently this hypothesis has been given a firmer grounding using traditional historical linguistic methods: see, e.g., Bengtson (1997a, 2001a, 2008a), Blazek & Bengtson (1995), Starostin (n.d., 2005a, 2005b). While it is not possible to present all the evidence for this latter view (see the references), we think some salient aspects of the phonology, morphology, and lexicon of Bur are enough to indicate the greater probability of its Dene-Caucasian (DC) affiliation.2

Phonology

At first glance Casule's comparison of IE and Burushaski phonology seems impressive. An ample number of examples is cited, and superficially it seems that Casule (henceforth "C") has made a good case for a correspondence between IE and Burushaski phonology. However, on closer examination a number of problems appear.

(a) Some "Bur" words cited for comparison are actually loanwords from Indo-Aryan or Iranian languages. Thus, dumas 'cloud of dust, smoke, water' (p. 31) is clearly borrowed from Old Indic3 dhumah 'smoke, vapor, mist'4 (even the accent is the same); purme 'beforehand, before the time' (p. 34) is isolated in the Bur lexicon and looks like a derivative of OI *purima- > Pali purima- 'earlier' (CDIAL 8286; cf. Eng. former); bada 'sole, step, pace' (p. 40) appears to be from OI padam 'step, pace, stride' (CDIAL 7747), and perhaps others.

(b) Some comparisons adduced in support of the correspondences are semantically tortuous if not utterly dubious. For example, IE *dheu- 'to die, to lose conscience (sic)' ~ Bur diu 'lynx' (p. 36); IE *h2erg-nt-om 'white (metal), silver' ~ Bur hargin 'dragon, ogre', etc.

(c) The proposed correspondences are not consistent and do not form a coherent system. For example, IE *g, *gh are said to correspond to Bur g (voiced velar stop) or g (voiced uvular fricative) (p. 39), apparently in free variation, but in Bur berkat 'summit, peak, crest; height' (pp. 30, 35) IE *gh is matched with Bur k (voiceless velar stop), in Bur buqheni 'a type of goat' (p. 31) IE *g is matched with Bur qh (aspirated uvular stop or affricate), and in Bur je, ja 'I' (p. 72) IE *gh is matched with Bur j [3 = dz]. IE *kw is said to correspond to Bur k (voiceless velar stop) (p. 38), but in Bur -sogut 'the side of the body under the arm; bosom' (p. 30) it is matched with Bur g (voiced uvular fricative), while in Bur waq 'open the mouth, talk' (p. 38) it is matched with Bur q (voiceless uvular stop). PIE *w (*u) becomes Bur w in waq 'open the mouth,

1 The authors accept Nostratic/Eurasiatic and Dene-Caucasian as working hypotheses that represent, in our opinion, the best available explanations for language classification in northern Eurasia (see, e.g., Bengtson 2008b, Blazek 2003, 2008).

2 For some history of the DC hypothesis see e.g. Bengtson (1994), Blazek & Bengtson (1995), Peiros (1988), Ruhlen (1996, 1998a, 2001).

3 Old Indic (OI) here encompasses Vedic and Classical (Sanskrit) forms of OI.

4 H. Berger (p.c. to author Bengtson) regarded Bur dumas as a loanword from Indic (CDIAL 6849). See Bengtson (2001b, p. 185).

talk' (p. 38),5 but b in budoo 'rinsing water; water that becomes warm in the sun' (p. 31).6 For C the Bur uvulars (q, qh, g) are merely variants of the velars and do not form an historical class of their own (but see [d.3] below).

(d) C totally overlooks (or minimizes) many distinctive features of the Burushaski phonological system. These features include (1) the retroflex stops, (2) the phoneme /y/, (3) the uvular consonants, (4) the tripartite sibilant contrast /s ~ s ~ s/, and (5) the cluster -It-, and the t---lt- alternation (corresponding, we think, to Dene-Caucasian lateral affricates). We reproduce below (with minor modifications) the table of Burushaski consonants presented by Berger (1998, I: 13):

uvular velar retroflex dental retroflex palatal laminal labial

S S s

qh kh th th ch ch ch Ph

q k t t c c c P

g g d d j j z b

r) n m

h r l y

Table 1

(1) The retroflex stops. C (pp. 26-27) claims "We do not know the genesis of the retroflex consonants in Bur ... we cannot know with certainty whether Bur originally possessed aspirates and cerebrals or whether these phonemes were acquired from IndoAryan." Although C does not discuss it, the DC hypothesis provides a ready explanation for at least some of the retroflex consonants in Bur: 7

• Bur *git 'anus; vulva; intestines with inner fat' < *girt or *gilt ~ Caucasian: PEC *kwiltV (Dargwa kulta 'belly, stomach', Agul gutul 'kidney', etc.)8 ~ PY *gi?d 'fat': Ket, Yug ki?t, Kott kir, Arin ki (NCED 711, CSCG 119)

• Bur *-phat 'gizzard, stomach of fowl' < *phart ~ Caucasian: PEC *pHVrtwV (Bezhta pirti 'lung, bladder', Archi parti 'large intestine', etc.)9 ~ Basque *e-purdi 'buttocks, rump' (NCED 871, CSCG 160)10

• Bur *git 'slime'11 < *girt ~ Caucasian: PEC *Xwirdi (Avar xwerd 'pus', Agul furd 'dung', etc.) ~ Basque *lirdi 'drivel, saliva' ~ PST *lVt 'mucus, phlegm' (Tibetan lud 'phlegm, mucus; manure, dung', etc.) (NCED 763, LDC 19, CSCG 132)

5 See CSCG (p. 8) for an alternative comparison with DC.

6 Cf. instead OI *budyati 'sinks', Marathi budbud 'sound of bubbling', etc. (CDIAL 9272).

7 It is important to note that *t in Nikolaev's & Starostin's Caucasian reconstructions does not denote a retroflex stop but rather a glottalized stop (similarly with other glottalized obstruents: p, c, c, c, X, k, q. On the other hand, in this paper t, th, d, s, c, ch, j, y in Burushaski words always denote retroflex obstruents.

8 Some Caucasian words, e.g. Udi gurdak 'kidney', Tabasaran gurdum id., seem to reflect influence of Persian gurde 'kidney'. Perhaps in some cases there is a blend of the Persian word with Proto-Lezgian *k:wirt- (k:wilt~?) (thanks to E. Bashir, pc.).

9 a represents a pharyngealized vowel, also (confusingly) written al, where I represents the palocka in the Cyrillic orthography of Caucasian languages (Catford 1977: 296).

10 Assuming a semantic development such as 'large intestine > colon > rectum > buttock' in Basque. Cf. OI guda- 'intestine, entrail, rectum, anus', Sindhi gu! 'anus, posterior', etc. (CDIAL 4194).

11 'Schlamm (feucht oder ausgetrocknet)' (Berger 1998). E. Bashir (pc.) suggests possible Indo-Aryan origin: cf. Panjabi gidd ~ gidd 'matter that accumulates in the corner of the eye'.

• Bur *chad-nm 'narrow'12 < * chard— Caucasian: PEC *cHVrdV 'narrow' (Avar c:edera-b, Dargwa Akushi carta, etc.) ~ PY *to?d- (~ *co?d-) 'shallow (of a river)' (NCED 387, CSCG 199)

• Bur *gatu 'clothes' < *gart- ~ Caucasian: PEC *gwirdwV 'a kind of clothing' (Avar gorde 'shirt', Dargwa Akushi gurdi 'dress', etc.) ~ PY *xo?t(ir 1) 'cloth, felt' > Arin qot, kot 'trousers', etc. (NCED 449, CSCG 223)

These examples suggest that the Proto-DC intervocalic clusters *-lt-, *-rt-, *-rd- regularly correspond to Bur retroflex consonants. While this process does not account for all occurrences of retroflex consonants in Burushaski, it does indicate a very old origin of the retroflex series that is analogous to the origin of retroflexes in Indo-Aryan.13 (See below for the development of a new cluster /lt/ in Bur.).

(2) The Bur phoneme /y/. C (p. 25) briefly mentions Bur /y/, but it has no real place in his IE-Bur phonology. As far as we can see, /y/ figures in only one of C's Bur-IE comparisons, that of Bur guy-aq 'hair' with IE *gour- 'hair' (p. 32). C provides no explanation of why IE *r becomes Bur /y/ in just this one case.14 This seems to us a very unsatisfactory treatment of this important Bur phoneme. Before presenting our view of the genesis of /y/, some further information is necessary:

Burushaski and Domaki (an Indo-Aryan language spoken in parts of the Burushaski-speaking area)15 have an unusual consonant [y], variously described as "a fricative r, pronounced with the tongue in the retroflex ('cerebral') position" (Morgenstierne 1945), "a kind of r | y and z" (Lorimer 1937: 72), "a voiced retroflex sibilant with simultaneous palatal-dorsal narrowing" (Berger 1998), "a curious sound whose phonetic realizations vary from a retroflex, spirantized glide, to a retroflex velarized spirant" (Anderson, ms.). Because of the elusive character of this sound, it has been transcribed in various ways; for example, the word for 'my father', transcribed here as aya, is found in the literature as aiyah, alya, agha, aya, or ara.

As noted by Morgenstierne (1945), Bur [y] in loanwords from Indo-Aryan derives from the retroflex sound *r, which in turn can come from *t, *d, *dh. Morgenstierne and Berger cite the examples:

• Bur (H,N) day 'fat, strong, robust' < OI drdha- (Beiträge 36, no. 3.35)

• Bur (H,N) bayum 'mare' < *vadam- = OI vadaba- (Beiträge, ibid.)

• Bur (H) paayo, (N) payo, (Y) palu 'wedge' < OI pataka- (Beiträge 24, no. 3.13)

• Bur (H, N) kilaay 'beesting curds' = Late OI kilata 'cheese' (but see further below)

Note also:

• [y] is heard in the Hunza and Nager dialects, but not in Yasin ("Werchikwar"), where [y] either corresponds to zero (as in ba for bay 'millet') or a different phoneme: Yasin palu 'wedge' ~ (H) paayo, (N) payo; Yasin khac '(stony) shore, bank' ~ (H, N) khay, etc.;

12 The variant (Y, H) c(h)an-um appears to be contaminated by the verb du-c(h)an-.

13 "The development *lt > retroflex is evident also from early Indo-Aryan, and later again in the Prakrits. Nostraticists explain Dravidian retroflexes in the same way. This areal tendency should probably not be attributed to influence of Dravidian (which is not seen in the early Rgveda), but as an areal feature of the Northwest (of Greater India), as seen in Bur, Pashto, Old Indic of the Rgveda, and later also Khotanese Saka." (M. Witzel, pc.)

14 /y/ is also seen in C's comparison of Bur biy 'butter' with IE *pi- 'fat' (p. 40), though no IE suffix corresponding to Bur -y is proffered.

15 Domaki, an endangered language, is spoken in the village of Mominabad (Hunza) and in a couple of villages in Nager (B. Tikkanen, p.c.).

• Berger (1998 I: 22, note 8) also finds [y] similar to the Tamil sound commonly transcribed as ];

• Place names confirm the ancient affinity of [y] with [l] or other laterals: Bur Namay = No-mal; Punyaay = Punial (Lorimer 1937: 73);

• The Bur word (H, N) kilaay 'Quark aus Biestmilch' is found in Vedic as kilala- 'beestings, a sweet drink' (Witzel 1999: 3), also in Khowar as kilal, kilari;

• Some Indo-Aryan dialects (including those of some Vedic texts) have/had a retroflex l corresponding to the d of Classical OI,16 as in Ved. nila- 'nest' = Skt. nida- < PIE *nizdo-.

With that background, we propose that Burushaski [y] — apart from loanwords — ultimately derives from laterals (*l,*t) and clusters involving laterals (e.g., *lc, *lc, *lx, *fil) in Proto-DC. The following examples support this interpretation:

• Bur *gay 'thread, strand (in weaving)' ~ Caucasian: Lezgi ml = gal 'thread', etc. < PEC *xatV 'sinew, thread' (NCED 1067) ~ Basque: *ha[l]i 'thread, yarn, filament, wire'

• Bur *khiy > (H,N) khiy 'leaf', (Y) khi-aq '(fallen) leaves' ~ Caucasian: Tindi koli, Abkhaz a-kdla 'sheaf', etc. < pNc *kawtV (NCED 690).

• Bur *qhiye > (H,N) qhiye '(single, small) stones, gravel' ~ Caucasian: Archi ¿¡wil 'rock, cliff', Abkhaz a-^a-ra 'rocky river bank', etc. < PNC *qwild (NCED 939)

• Bur *bay, (Y) ba '(small-grained) millet' ~ Caucasian: Chechen borc 'millet', etc. < PNC *bdlcwi (NCED 309, CSCG 15)

• Bur *huy- 'to dry'17 ~ Caucasian: Dargwa Urakhi =irc-/=uc- 'to roast, fry', etc. < PEC *=i[l]cwE 'to roast, fry, dry' (NCED 633, CSCG 103)

• Bur *huyoo > (H,N) huyoo 'wool animal, sheep' ~ Caucasian: Chechen 1axa-r 'lamb', Andi ixo 'sheep, ewe', etc. < PNC *TilxU (NCED 247, CSCG 265)

• Bur *guy 'hair'18 ~ Caucasian: Chechen exang 'woollen thread, yarn', Rutul arx 'spring wool', Tsakhur arx 'autumn wool', etc.19 < PEC *2alxV 'wool' (NCED 242) ~ Basque *ulhe 'hair, wool'

• Bur *gaqay(-um) 'bitter; unsweetened; sour' > gaqay(-um) (H,N), qaqam (Y) ~ Caucasian: Archi ¡ala 'bitter', Khinalug ¡ilez 'salty', Ubykh ¡aqd 'sweet', etc. < PNC *qeMV (~ -1-) (NCED 912) ~ PY *qVqVr 'gall; bitter' ~ Basque: *kerac 'bitter, sour; stench' (CSCG 236)20

The following examples indicate DC lateral suffixes (*-alV, *-ulV, *-ilV) with the reflexes /ay/, /uy/ in Bur:

• Bur *tumay 'shell of nut, fruit stone' ~ Caucasian: Archi tummul 'grape', Budukh tombul 'plum', etc. < Proto-Lezgian *tum(:)ul (beside suffixless Chechen, Ingush, Batsbi tum 'marrow; kernel of fruit, nut') < PNC *tumhV 'kernel, nut, fruit-stone; marrow' (NCED 1004, CSCG 205)

16 "The Rgveda originally did not have [retroflex l] but acquired it only during [oral] transmission, by c. 500 BCE. And Panini also does not have it in his grammar ... He does not even have the vowel l [1], just the vowel r [r]. The later Vedic (Post-Rgveda) record is quite checkered [in regard to retroflex l]. The Delhi area and some texts east and south of it had such a retroflex. ... [retroflex l] is now found in the mountain area of Indo-Aryan, from the Afghan border to the western Nepalese border." (M. Witzel, p.c.).

17 (H, N) b-uy-, (Y) b-u-, du-hu-.

18 (Y) goyarj, (H,N) guyarj 'hair' (both with ordinary /y/), (N) -thoguy 'fine hair of small children', also in (H) phul-guuy, (N) phur-guuy 'feather'.

19 denotes the Caucasian pharyngealized voiceless uvular affricate = NCED /xI/.

20 For semantics, cf. Albanian embel 'sweet', Armenian amokh 'sweet', maybe cognate with Latin amarus 'bitter', Old Swedish amper 'sauer, scharf, bitter', etc.

• Bur (N) -phaguy 'stick, walking-stick' (beside [H] -phago) ~ Caucasian: Andi moq':ol 'ceiling' (beside suffixless Avar moq': 'pole', Tsez maq 'short stick, rod',21 etc.) < PNC *bhmqV 'pole, post' ~ Basque *makita 'stick, cane' (beside Bizkaian mak-et 'club', with a different suffix)22 (NCED 295, CSCG 14)

• Bur qaruuyo (H), garuuyo (N) 'heron' ~ Basque *kurV(-lo) 'crane' (Bizkaian, Gipuzkoan kurrillo, kurlo, Zuberoan khurlo, vs. suffixless Low Navarrese kurru, Roncalese kurri);23 Caucasian words for 'crane' display a variety of suffixes and reduplications: cf. Chechen mrmli = garguli, Andi q:urru, Karata q:uru-n, Adyge q:araw 'crane', etc. < PNC *qardqwV beside the simplex *qwVrV (NCED 914-5, CSCG 237).

We believe we have shown that the Bur phoneme /y/ is an integral feature of the language, and that only the DC model provides a plausible explanation of its origin.

(3) The uvular consonants. The Bur uvular consonants, as a class, are totally ignored by C, to whom /q/, /qh/, and /g/ are simply erratically occurring variants of /k/, /kh/, and /g/. We intend to show that the Bur uvulars constitute a class of importance and long standing in the language, and can be derived from the DC uvulars.24

• Bur qaruuyo ~ garuuyo 'heron' ~ Basque *kurV(-lo) 'crane' ~ PNC *qardqwV / *qwVrV 'crane' (see above)

• Bur *qVt- > -qat (H), -qhat (N), -qet-araq (Y) 'armpit' ~ Caucasian: Avar me-hed 'brisket (chest of animal)', Bezhta mde = gade 'brisket' < PEC *qVdV (NCED 897) ~ PY *qot- (~xot-) 'in front, before' (cf. Eng. abreast, etc.) (CSCG 170)

• Bur *qorqor- > (H) qorqor 'soft porous stone', (N) qoqor 'small stones' ~ Caucasian: Dargwa q:arq:a 'stone', etc. < PEC *GorGV 25 ~ Basque *gogor 'hard'

• Bur *qus- > (Y) qus 'armpit (of clothing)' ~ Caucasian: PNC *qHwaci 'hole, hollow' > Chamalal q:uca 'vagina', Lezgi que 'armpit', etc. (NCED 922, CSCG 176)

• Bur. *qaq- 'dry, hungry' ~ PY *qV[(?)G]i- 'dry': Kott xujga, Arin qoija, etc. ~ PNC *GwiGwAr: Lak q'a-q'- 'dry', etc. (CSCG 223)

• Bur *qhas- > -qhasir/ (H,N) 'hind end, arse', -xasar/ (Y) 'female sex organ' ~ Caucasian: Udi qos 'behind', etc. < PEC *-VqV (NCED 1026)

• Bur *qhat- > -qhat (H,N), -xat, -xat (Y) mouth' ~ Caucasian: Lak qit (dial. qwit, qut) 'Adam's apple, beak', etc.26 < PEC *qwiti (NCED 905, CSCG 172)

• Bur *qhurc 'dust' ~ Caucasian: Tsez, Khwarshi qec 'dirt, mud, slush', Lezgi xanc' 'a layer of hardened dirt', etc. < PNC *qanVcwV (NCED 884, CSCG 169)

21 /a/ denotes a pharyngeal vowel = NCED /aI/.

22 The supposed derivation of *makila from Latin bacilla (pl.) 'sticks' (Trask 2008: 281) seems to us to be rather a case of chance resemblance. Lat. bacilla cannot account for the Bizk. form maket. Lat. bacillum, baculum are themselves suspect, having the rare PIE phoneme *b-, and reflexes of PIE *bak- (if it existed) are found only in western IE languages for which hypothetical DC-like substrata have been supposed.

23 One could suspect derivation of the Basque words from Romance (cf. Latin grüs, Italian gru, French grue, Spanish grúa, grulla), but the Basque words always have initial /k/ vs. Romance /g/, and in Romance a lateral suffix is found only in the Castilian variant grulla, where we can suspect Vasconic influence, or a blend of Romance grúa + Basque kurrillo. The Basque simplex forms Low Navarrese kurru, Roncalese kurri are parallel to the Caucasian simplex forms such as Andi q:urru, Karata q:uru-n 'crane' (NCED 915).

24 In Basque all DC uvulars become velars /k, g/ or the spirant /h/; in a few cases *Gw > *gw > /b/.

25 < *GorqV or *qorGV?

26 /i/, /u/ denote pharyngealized vowels = NCED /iI/, /uI/.

• Bur *qhdi 'revenge' ~ PY *xV(?)j- 'to be angry' ~ Caucasian: Udi xuj 'anger', Dargwa qa 'oath', etc.27 < PEC *qwejV (NCED 901, CSCG 171)

• Bur -qhurpat (H,N), -xorpet (Y) 'lung' ~ ? Cauc.: Tsez xotori, Lak hutru, etc. 'lung' < PEC *qwdlQV(rV) ~*xwdl6V(r9) (NCED 901) ~ ? Basque *hauspo 'bellows, lungs' (LDC 22)28

• Bur *qhVltd 'sack, pocket' > (H) qhiltd, (N) qhaltd, (Y) xalt(y)d ~ Caucasian: Akhwakh q:eAe 'sack, pillow', etc. < PEC *GHtr&wV (NCED 457, CSCG 55)

• Bur *gaqdy(-um) 'bitter; unsweetened; sour' ~ PNC *qeMV ~ PY *qVqVr 'gall; bitter' ~ Basque *kerac 'bitter, sour; stench', etc. (see above)

• Bur *gul 'grudge, enmity, hatred' ~ Caucasian: Avar Ewel = gwel 'gossip, rumor; abuse', Khinalug qol 'offence', etc. < PEC *Gwalho (NCED 465) ~ PY *qd(?)r- (x-) 'angry' ~ Basque *bVrhao / *bVraho 'curse, blasphemy' (CSCG 55)

• Bur *chdgur 'chest or box for grain or meal' ~ Caucasian: Avar caKur = cagur 'corn bin, barn', Chechen cxar 'penthouse', etc. < PEC *cVGVr- (NCED 328, CSCG 189)

• Bur gonderes, gondoles (Y) 'water that runs over many stones' ~ Cauc.: Botlikh mdaru = gadaru 'stream, brook', Lak atara 'mountain stream', etc. < PEC *GHwadVrV (NCED 478, CSCG 185)

• Bur *gorqu- > gurqun (H), gurquc (N), gorkun (Y) 'frog' ~ Caucasian: Tindi qorq:u, qoq:u, Khinalug qurqor, Kabardian handar-q:waq:wa, etc. 'frog' < PNC *qwVrVqV (NCED 942) ~ PY *X3?r- 'frog' > Ket, Yug n?l, Arin kere (CSCG 243)

• Bur *ltag > tag (Y) 'branch, shoot'29 ~ Caucasian: Avar X:ox: 'stubble', etc. < PEC *KhwaxV 'stick, chip' (NCED 778, CSCG 137)

• Bur *gay 'thread, strand (in weaving)' ~ PEC *jalV 'sinew, thread' ~ Basque *ha[l]i 'thread, yarn, filament, wire' (see above)

The Bur uvulars are thus far from being merely peripheral and erratic variations of the velars: they constitute an integral series in the Bur phonological system that cannot be understood apart from the DC context from which they arose.

(4) The tripartite sibilant (and sibilant affricate) contrast. A sibilant contrast with three points of articulation that carries through to sibilant affricates, though ignored by C, is a significant feature of Burushaski phonology that did not exist in Proto-IE,30 but is characteristic of Caucasian languages as well as of Basque. Below is the Burushaski system as outlined by Berger (1998, I: 13):

laminal palatal retroflex

s S S

ch ch ch

c c c

z j j

Table 2

27 /a/ denotes a pharyngeal vowel = NCED /aI/.

28 A questionable comparison. At the very least, there have been some irregular changes and/or contaminations, e.g. Basque *hauspo with *hauc 'dust', etc.

29 See below for the correspondence of Bur t- with Caucasian lateral affricates.

30 Unlike most IE languages, Old Indic had a triple contrast (s, s, s). We suggest that this was an areal feature acquired by early Indic as its speakers sojourned in the Hindu-Kush area. "A good point again about the three sibilants in IA: Iranian only has two (s and s). I agree with your assessment as an areal feature: again the NW [northwestern Greater India]. Note that many other forms result from the NW predilection for 'bending back the tongue': (PIE) *rek's > *racs > *rags > (Skt) rat (nom. 'the king')." (M. Witzel, p.c.).

This is very similar to the slightly more complex system reconstructed for Proto-Caucasian (NCED, p. 40; palatal = hissing-hushing):

hissing palatal hushing

s S S

z z z

c c c

3 3 3

c c c

Table 3

And cf. the more simplified system of Basque (Hualde 1991):

lamino-alveolar apico-alveolar palatal

s S S

c c c

Table 4

In the Basque orthographic system the sounds /s/, /s/, /s/, /c/, /c/, /c/ are denoted by the letters z, s, x, tz, ts, tx, respectively.

We think it interesting that this characteristic DC pattern has been maintained to the present day in widely separated descendant languages. Naturally, there have been extensive changes, but the systems as a whole have remained.

The following comparisons are typical of the Bur system of sibilants and affricates and their relationship to those of other DC languages. Note that some of the phonetic correspondences are complex, and CSCP (Starostin 2005b) should be consulted for the details.

• Bur *'-s 'heart, mind' ~ Caucasian: Ubykh p-sa 'soul, spirit', Bezhta, Hunzib has 'sky, cloud, fog', etc. < PNC *2amsa ~ Basque *haise 'wind' ~ Yeniseian: PY *?es 'God, sky' (NCED 243, CSCG 263)31

• Bur *'-so[m] 'kidney'32 ~ Caucasian: Chechen sam-g 'sausage (made from a large intestine)', Akhwakh s:e 'sinew, muscle', etc. < PEC *sehmV / *hemsV ~ Basque *sain 'vein, nerve, root' (NCED 959, CSCG 187)33

• Bur *-sVsVn 'elbow'34 ~ Caucasian: Udi sun 'elbow', Lak s:an 'foreleg, paw', etc. < PEC *sino ~ Basque *san-ko 'leg, calf, foot, paw', etc. (NCED 963, CSCG 187)

• Bur *san 'spleen' ~ Caucasian: Archi s:am 'gall', Dargwa *sumi 'gall, anger', etc. < PNC *cwajme ~ Basque *beHa-su[m] 'gall' (NCED 329, LDC 18, CSCG 22)

31 For semantics, cf. Rumanian inima 'heart, soul, mind,' etc. < Latin anima 'wind, air, breath, spirit, mind', etc.

32 Underlying m found in the plural form '-somuc.

33 Starostin (CSCG 187) adds the following Sino-Tibetan forms: PST *siam 'heart, soul' > Old Chinese *sam 'heart'; Tibetan sem(s) 'soul; think', b-sam 'thought'; Burmese simh 'to conceive, be in the charge of'; Lushai thiam 'to know'; Lepcha a-som 'spirit, breath', etc. For semantics, cf. e.g. Skt. hira- 'band, strip, fillet', hira 'vein, artery'; Gk. xop&q 'gut, cord, string'; Lat. hira 'empty gut'; Lith. zarna 'intestine, small intestine'; Ger. Garn 'yarn, thread, net', Eng. yarn, etc. (IEW I: 604); Turkish bobrek 'kidney'; Proto-Tungus-Manchu *pugi- / *puki- 'intestines, stomach' Proto-Japanese: *punkuri 'testicles' (ToB).

34 (Y) -sesen, (H, N) -susun.

• Bur *-su[m] 'umbilical cord, navel'35 ~ Caucasian: Dargwa zu 'navel', Khinalug c'um id., etc. < PEC *jdn?u (NCED 1096, CSCG 249)

• Bur *sa 'sun, day, month' ~ Caucasian: Lak s:aw 'sky', Botlikh ziwu 'day', etc. < PNC *3§wi (NCED 1092, CSCG 248)

• Bur *sum 'sprout, shoot; tail; spout (of a vessel)' ~ Caucasian: Lak c'un 'spout (of a vessel)', Chechen c'om 'trunk', etc. < PEC *cumV (~ *3umV) (NCED 367, CSCG 249)

• Bur *sesin- 'clear, clean' ~ Caucasian: Chechen c'ena 'clean, pure', Abaza b-zi 'good', etc. < PNC *HajEm- ~ Basque *susen 'right, correct, just' (NCED 552, LDC 189, CSCG 64)36

• Bur *-sqa '(on one's) back' ~ Caucasian: Proto-Abkhaz-Tapant *zdkwa 'back' ~ Basque *bi-ska-r 'back; crest, hill' ~ PY *suga / *?uska 'back, backwards' (ToB)

• Bur *bus 'sheaf (of grass, hay)' ~ Caucasian: Chechen buc 'grass', Adyge wdcd id., etc. < PNC *wicV (NCED 1053, CSCG 219)

• Bur *kus 'wonder, sorcery' ~ Caucasian: Ingush kust 'bearing, appearance, figure', Archi kus 'habit', etc. < PEC *kwijcV ~ Basque *hoc 'noise, sound; fame, reputation; longing, mania', etc. ~ Yeniseian: PY *k[u?u]s 'idol, ghost' (NCED 710, CSCG 118)

• Bur *bas 'wooden plow' ~ Caucasian: Karata bec:e 'wooden plow', Abkhaz a-pdza 'plowshare', etc. < PNC *pVrVcE (NCED 877, CSCG 164)

• Bur *mos 'mud avalanche' ~ Caucasian: Agul mes 'mould', etc. < PEC *maswV ~ PY *pu?s 'mould' (NCED 296 [note], CSCG 141)

• Bur *si 'fireplace, hearth' ~ Caucasian: Ingush c'i 'fire', Lak c'u id., Abkhaz a-m-ca id., etc. < PNC *caji ~ Basque *su 'fire' (NCED 354, CSCG 23)

• Bur *se[m] 'wool'37 ~ Caucasian: Lezgi r-cam 'eyebrow' (< *'eye-wool'), Chechen cocqam id., etc. < PEC *chweme ~ Basque *sama-r 'fleece, mane; chamarra', etc. ~ Yeniseian: PY *cdqe 'hair' ~ PST *cham 'hair (of head)' > Kanauri cam 'wool, fleece', etc. (NCED 364, CSCG 27)

• Bur *sulu 'driftwood'38 ~ Caucasian: Tindi c:ela 'rod', Abkhaz a-c'la 'tree', etc. < PNC *cdlV ~ *cdlV ~ PST *Cal ~ *Cdl 'wood' (NCED 362, CSCG 26)

• Bur *-sdr 'limbs, body parts' ~ Caucasian: Lezgi cum 'shin-bone', Bezhta oc 'knuckle-bone', etc. < PEC *Hcwejnd ~ Basque *soin 'shoulder, upper back', etc. (NCED 555, CSCG 66)

• Bur *son 'blind' ~ Caucasian: Lak can 'darkness', Ubykh ja 'black', etc. < PNC *cAwnV (NCED 352, CSCG 24)

• Bur *soq-um 'wide, broad' ~ Caucasian: Dargwa Chirag caqw- 'high', Kabardian -sxwa 'big', etc. < PNC *cHdqwV ~ Basque *asko 'much, many', *aski 'enough' ~ PST *cok ~ *jok 'enough, sufficient' (NCED 386, CSCG 36)

• Bur *susun '(child's) penis' ~ Caucasian: Lezgi cuc 'spout (of a tea-pot)', Kryz cic 'clitoris, ring-stone', etc. < PEC *cocV ~ Basque *soc 'spigot, faucet' (NCED 367, CSCG 28)

• Bur *so 'dried leaves, stalks, roots', etc. ~ Caucasian: Avar s:waji 'small chaff', Khinalug psa 'bread', etc. < PNC *swT?e ~ Basque *osi 'germ of grain, shoot that becomes a head of grain' ~ PST *sej 'seed, fruit' (NCED 977, CSCG 195)

• Bur *qus- 'armpit (of clothing)' ~ Lezgi que 'armpit', etc. (see above)

• Bur *ausi- 'guest'39 ~ Caucasian: Chechen hasa 'guest', Ubykh pea id., etc. < PNC *HVcwE ~ PY *?3ca (*ca-) 'guest' ~ Basque *hauso 'neighbor' (NCED 612, LDC 179, CSCG 83)

35 Underlying m found in the plural form -suimuc.

36 The semantic values in some languages apparently reflect the development: 'clean > pure > good > correct,

37 Underlying final m found in the plural form semirj.

38 "consider Kalasha [suia] 'firewood' ... with an IA etymology (T 12349 [< OI salaka f. 'any small stake or stick'])" (E. Bashir, p.c.).

• Bur *si / *si / *su 'to eat' ~ Caucasian: Tsez, Khwarshi =ac- 'to eat', Tindi c:a- 'to drink', etc. < PEC *=VcV ~ Basque *ausi-ki 'to bite' ~ Yeniseian: PY *si- 'to eat' ~ PST *jha id. (NCED 1017, CSCG 209)

• Bur *suqur 'sour, to sour' ~ Caucasian: Andi c:ik:u 'sour', etc. < PEC *cakwV 'sour, raw' ~ PST *sak 'bitter, pungent' (NCED 356, CSCG 24)

• Bur *suli 'tube, pipe'40 ~ Caucasian: Avar (dial.) sulu 'pipe', Hunzib selu 'horn', etc. < PEC *swoi(H)V ~ Basque *sullo 'hole, cave' (NCED 978, CSCG 195)

• Bur *siy 'milk' ~ Caucasian: Chechen sin 'udder', Andi s:iwu, s:imu 'milk', etc. < PNC *sam?V ~ Basque *e-Sene 'milk' ~ PY *de(?)n 'nipple, milk' (NCED 982, CSCG 196)

• Bur *ltis > *tis 'wind' ~ PEC *A[a]rcV 'movement of air' > Khwarshi Aaca 'wind', Tindi Aac:u 'voice, shout', etc. (NCED 767, CSCG 134)

• Bur *his 'breath'41 ~ Caucasian: Chechen hozu 'odor', Ingush haz, Batsbi haic < Proto-Nakh *hac 'odor' ~ Basque *hac 'breath; stench' (LDC 17)

• Bur *'-mes 'finger, toe' ~ Caucasian: Kryz micek 'nail, claw, hoof', etc. < PEC *(H)micV ~ *(H)micV ~ Yeniseian: Ket bes-taq5 'index finger' (NCED 819 [as *micV ~ *micV], LDC 38, CSCG 77)

• Bur *mus- > musk (H, N, Y) 'wood, thicket', mus-qu (H, N) 'branches with leaves' ~ Caucasian: Dargwa murc 'rod, stick, vine', Abkhaz a-mca 'wood, firewood', etc. < PNC *mucU / *cumU ~ Basque *moskor 'trunk of a tree' < *mos-ko-r (NCED 833, CSCG 147)

• Bur *'-ci- 'to kindle' ~ Caucasian: Abkhaz a-cd 'hot', Rutul =isa- 'to roast (grain)', etc. < PNC *=ercA ~ Basque *i-se-(ki) 'to set fire, kindle, burn', etc. ~ PST *cha 'hot' (NCED 415, CSCG 48)

• Bur *ca- 'to stand' ~ Caucasian: Lak =a-c'a- 'to stand', Akhwakh hec'- 'to stand up, raise', etc. < PEC *HercV- ~ Basque *e-asV (standard jaso, jasan) 'to lift, raise, support, bear', etc. ~ Yeniseian: PY *ta-, *pa-ta- 'to stand up' (NCED 562, CSCG 67)

• Bur *bdcin 'shank, hind leg above the hock' ~ Caucasian: Chamalal becw 'knee (of animal), thigh', Tsez besi 'fist', etc. < PEC *b[d]cV ~ Basque *borc 'five' (< *'hand') ~ Yeniseian: PY *ba?t- 'knee' ~ PST *put(-s) 'knee' (NCED 291, CSCG 19)

• Bur *bac 'small terrace between mountains, grown with grass' ~ Caucasian: Akhwakh beca, Tindi besa 'mountain', etc. < PEC *wice ~ Basque *baso 'forest, desert' (NCED 1053, CSCG 217)

• Bur *'-ncu 'paternal aunt' ~ Caucasian: Chechen neca 'maternal aunt or uncle', Abkhaz dca 'sister-in-law, daughter-in-law', etc. < PNC *nEcV ~ *cEnV ~ Basque *nes-ka 'girl, unmarried young woman' (NCED 322, CSCG 153)

• Bur *-jec- 'to see'42 ~ Caucasian: Hunzib =ac-- 'to see', Ubykh ca- 'to know', etc. < PNC *=amcE ~ Basque *e-ncu-n 'to hear' ~ Yeniseian: PY *7Vt- 'to know' ~ PST *sid(H) 'to know, think' (NCED 262, CSCG 4)

• Bur *phunc 'dew' ~ Caucasian: Lak pic 'dew, sweat', Dargwa penc 'resin', etc. < PNC *pincwA ~ Yeniseian: PY *pi?t 'glue' (< '*resin')

• Bur *qhurc 'dust' ~ Caucasian: Tsez qec 'dirt, mud, slush', etc. (see above)

39 (Y) aisen, ausin, pl. ausu, (H, N) oosin, pl. ooso. "The word is also present in Shina öoso 'guest', where it is most probably < Burushaski (despite highly dubious derivation in Turner 427 < Skt. *apadesya-)" (CSCG 83). "I think that this is probably an IA element. There are a considerable number of words in Khowar in which the initial aw-element is related to a meaning of 'separateness, distance', e.g. a(u)werik 'to take away' or awizä 'relative', which seem to show the IA apa- element. This again would seem to be more likely to be an old IA loan" (E. Bashir, p.c.).

40 'Gewehrlauf; Schnabel (an einem Gefäß); Rohr zum Anblasen des Feuers' (Berger 1998).

41 (Y, H, N) his 'breath', (Y) also -hes 'breath', (H, N) hns 'sigh' (with secondary nasalization).

42 "The reconstruction of Bur. 'to see' would probably be *-jeec-. The double vowel suggests that there may once have been a consonant (probably /g/ or /h/) between the vowels." (B. Tikkanen, pc.)

• Bur *chdgur 'chest or box for grain or meal' ~ Caucasian: Avar caKur = cagur 'corn bin, barn', etc. (see above)

• Bur *chigir 'goat' ~ Caucasian: Lak cuku 'goat', Andi c:ekir 'kid', etc. < PEC *3ikV / *ki3V ~ Basque *sikiro 'castrated ram' (NCED 1094, CSCG 187)

• Bur *chul- 'male breeding stock'43 ~ Caucasian: Andi cora 'heifer', Agul luc 'heifer', etc. < PEC *HcwilV ~ *HlicwV ~ Basque *cahal 'calf' (NCED 556)

• Bur *c(h)iki > (Y) ciki 'small' ~ Caucasian: Tabasaran ziq:i 'short', Chamalal cik:u-b 'small, short', etc. < PNC *3ikwA ~ Basque *ciki 'small' ~ Yeniseian: Kott thuki 'short' (NCED 1108, LDC 194, CSCG 197)

• Bur c(h)argV > (Y) carge 'flying squirrel' ~ Caucasian: Adyge cdKwa = cdgwa 'marten, mouse', Chechen satqa 'weasel', etc. < PNC *cargwV ~ Basque *sagu 'mouse' ~ Yeniseian: PY *sa?qa 'squirrel' ~ PST *srer(H) 'we asel, squirrel, mongoose,' etc. (NCED 322, CSCG 21)44

• Bur *micil / *bicil 'pomegranate' ~ Caucasian: Chechen hamc 'medlar', Khinalug mic 'apple', etc. < PNC *Samcd ~ Basque *mahanc 'grape' (NCED 237, CSCG 267)

• Bur *chap 'flesh, meat' ~ Caucasian: Bezhta sebo 'liver', Chechen zim 'kidney', etc. < PNC *JawV ~ Basque *sab-el 'belly' ~ Yeniseian: PY *tVpVi- 'spleen' (NCED 1106, CSCG 196)45

• Bur *chemil 'poison' ~ Caucasian: Tsakhur circima-n 'sour', Khinalug mic 'sour', etc. < PNC *hmVjcwA / *hcwVjmV ~ Basque *samin 'bitter, pungent, piquant; choleric' (NCED 521, CSCG 93)

• Bur *chad-um 'narrow' ~ Caucasian: Akushi carta, etc. (see above)

• Bur *chage-: (Y) cage 'jackdaw', (H) chagen 'crow with a red beak' ~ Caucasian: Chechen cemg = cegag 'magpie', Lezgi caK = cag 'jackdaw, rook', etc. < PEC *camm (NCED 381, CSCG 35)

• Bur *chis 'mountain' ~ Caucasian: Chechen ciz 'amulet (stone)', Lak cuca 'small stone', etc. < PEC *cacwV ~ Basque *cinca 'small stone, pebble' ~ Yeniseian: PY *ci?s 'stone' (NCED 382, LDC 114, ToB)

• Bur *chas 'thorn' ~ Caucasian: Akhwakh zaza 'thorn, prick', Ubykh caca 'spit', etc. < PNC *3d3e ~ Basque *sa(r)si 'bramble, thorn' (NCED 1090, CSCG 248)46

• Bur *jam 'distant relative' ~ Caucasian: Tabasaran jam 'bridegroom', Ingush zame 'best man', Lak mac:a 'kinsman', etc. < PEC *jamV / *majV (NCED1101, CSCG 251)

• Bur *jal- / *jal- '(long) hair'47 ~ Caucasian: Godoberi zali 'fringe, forelock', Bezhta zaro 'horse's mane', etc. < PEC **3alhi (NCED 1101, CSCG 251)

• Bur *muj-oq 'fringe, bunch of hair (on cow's tail)' ~ Caucasian: Chechen merz 'hair (in horse's tail)', Archi mocor 'beard', etc. < PEC *mecuri (NCED 800, CSCG 150)

• Bur *jo- 'to come'48 ~ Caucasian: Kabardian -id- '(to move) back', Avar =ac-in- 'to come', etc. < PNC *=icwE ~ Basque *e-uci 'to let, leave, permit' (NCED 627, CSCG 101)

43 (Y) cula 'fertile billy-goat', culdar 'bull', (H, N) chula 'billy-goat, drake', chindar 'bull'.

44 This etymon exhibits a wide range of semantic variation, though all pertaining to rodents or mustelids. Within the Caucasian family the meanings include 'weasel', 'marten' and 'mouse'. According to NCED (p. 322) Georgian ciq'wi 'squirrel is a loanword from East Caucasian. In Basque the stem *sagu or its variant *sat- (prob. from *sag-t-, with a fossilized oblique marker) is used for other animal names, such as *sagu-sahaf 'bat' (lit. 'mouse-old'), *sat-hor 'mole' (lit. 'mouse-dog'), *sat-icu 'field-mouse' (lit. 'mouse-blind').

45 This etymology may not hold together in all its parts, because of phonological difficulties. See the note in CSCG (p. 196).

46 This root, with two successive sibilant/affricates, has apparently been subject to various assimilations and dissimilations. Cf. also Spanish zarza 'bramble, blackberry bush' (OSp sarga), probably of Vasconic origin (the 17th c. Basque writer Oihenart had garci: Trask 2008: 337).

47 (Y) jalas 'hairy', (H) '-jal 'strip (of cloth)', jalei, jalii 'beard (of goat)', (H, N) jalali-miq 'long hair (of people)'.

48 (Y) jo-, (H, N) ju- (with retroflex /]/.

(5) The cluster /lt/, and the t---It- alternation. In the course of a thorough study of Bur

phonology one becomes aware of the cluster /lt/ and the fact that in certain verbs as well as nouns there is a frequent alternation of initial dental stops /t-, th-/ with medial lateral-dental clusters /-lt-/. The dental stops occur in both noun and verb stems in word-initial position, while the lateral-dental clusters occur in the same stems when they occur after a prefix. For example, in Bur (H, N) -ltur 'horn' is a bound morpheme and can only occur with a possessive prefix, such as a-ltur 'my horn', gu-ltur 'thy horn', while in the Yasin dialect 'horn' is simply tur, a free morpheme. The underlying form of all these is *-ltur 'horn' (thus Starostin, ToB). In a verb such as turu- 'fall apart, disintegrate' the cluster /lt/ appears in prefixed froms such as (absolutive or converb circumflex) nultur 'having fallen apart' (with analogical variants nutur, nuturu). The underlying root is thus *-ltur- 'to fall apart', etc. (Starostin, ToB).

It should be noted that Klimov & Edel'man (1972; see also Beiträge p. 80, no. 10.9) formulated an ingenious hypothesis that several of the words discussed here, and others that denote paired nouns (*-ltur 'horn', *-ltumal 'ear', *-lten 'bone', etc.) contain a prefix *-lt- derived from the numeral 'two' (see below under Numerals). While we admit this solution is inventive, we think it is an example of the dangers of relying solely on internal reconstruction. For example, the existence of external cognates to Bur *-ltur 'horn', namely Avar A:ar 'horn', Basque *adar 'horn', and others (see below), would require that this prefixing of the numeral 'two' must have taken place already in Proto-Dene-Caucasian. Furthermore, the existence of other Bur words with initial (or underlying) *lt-, and no semantic content of pairing, e.g. Bur *ltus 'grave', *ltap 'leaf',49 and of words for paired body parts such as Bur *qVt- 'armpit', -qhurpat ~ -xorpet (Y) 'lung', *'-so[m] 'kidney', *-sVsVn 'elbow' (see above) that lack the supposed *-lt- prefix, indicates to us that it is probably only fortuitous that some words with initial *lt- denote paired objects.

The following examples show both the internal Burushaski alternation of the initial dental stop t- with the medial clusters -lt-, and the regular correspondence of both with Caucasian lateral affricates. In the following comparisons /X/ denotes a voiceless lateral affricate = [tl], /X/ denotes a glottalized lateral affricate = [tl], and /t/ denotes a voiced lateral affricate = [dl]:

• Bur *-ltur 'horn' > (Y) tur / (H, N) -ltur 'horn' (bound form) ~ PEC *%wtrV 'horn; braid, mane' (Avar A:ar, Chechen kur, etc.)50 ~ Basque *adar 'horn' (< *a-rdar) (NCED 771, CSCG 134)

• Bur *-lten > (Y) ten 'bone' / (H, N) -ltin 'bone' (bound form); (Y) tanc, (H, N) -ltanc 'leg' ~ PEC *ÄwVn?V 'groin; part of leg' (Avar X:an 'groin', etc.) ~ PST *ldq 'shin, ankle' (NCED 785, CSCG 139-140)

• Bur *ltap > (Y) tap 'leaf', (H, N) tap 'petal, page' / (Y) du-ltapi-, (H, N) du-ltapu- 'to wither' ~ PNC *Mpi 'leaf' (Lak capi 'leaf', etc.) ~ Basque *lapar 'bramble'51 ~ PY *jäpe 'leaf' ~ PST *läp 'leaf' (NCED 774, CSCG 136)

• Bur *ltopo, *(l)tultopo > (H, N) topo, tultopo 'a kind of thin bread of leavened dough' ~ PEC *HärlöpV (Tsez Aepeli 'a pastry made of barley flour', Lak arc:ap 'a food made of barley flour, curds, butter, and rice', etc.) (NCED 546, CSCG 63)

• Bur *-ltur- > (H, N) turu- / nu-ltur / -tur(u), (Y) tur-, du-ltur- 'to fall apart, disintegrate, be cut into pieces', etc. ~ PEC *=eXwV(l) 'to burst, tear' (Hunzib =uÄ-, etc.) ~ Basque *leher 'to

49 The underlying form *ltap is indicated by the verb *du-ltapV- 'to wither'.

50 In Avar (and Andian and Tsezian languages, and Archi) Proto-Caucasian lateral affricates are, by and large, preserved as such. In Nakh, Lak, Dargwa, Khinalug, and Lezgian languages (except Archi, which has ve-larized lateral affricates) lateral affricates have largely been replaced by lateral resonants, velars, or uvulars (NCED); cf. Catford (1977), Starostin (2005b). However, under certain conditions there are velar reflexes in the first group of languages as well.

51 For the semantics, cf. the IE etymology that includes Skt. tfna- 'grass, herb, straw' and Eng. thorn, etc.

burst, smash' ~ PY *?il 'to break, split' ~ PST *rui ~*ruai 'to demolish, ruin' (NCED 413, CSCG 105)

• Bur *-ltâ- > (H,N) tâ- / -ltâ-, (Y) tâ- 'to follow,', etc ~ PEC *=VmXV 'to go, come' (Hunzib =ëX- 'to go, walk', etc.) ~ Basque *urten 'to go out, leave' (NCED 1026, CSCG 212)

• Bur *-ltâl- > (H, N) -Itâli-, (Y) -Itâli 'to wind, turn', tâlen- / -Itâlen- 'to go round', etc.52 ~ PNC *Awtri ~ *rtAwi 'wheel, vehicle' > Chechen lara 'oval cradle runners; fan of the mill wheel', Agul fur 'wheel', etc. ~ PST *r[ua]t 'round, roll, wheel' (CSCG 134)

• Bur *ltul- > (H, N)-ltul-, (Y) tul- / -ltul- 'to saddle', tiliha/, teléhari 'saddle', (H, N) tilia/ id. ~ PEC *Xwiië 'saddle' (Avar X:ili, Lak kili, etc.) (NCED 783, LDC 160, CSCG 139)53

• Bur *-ltân- > tan- (tân-) / -ltân- 'to pound (objects)' ~ PEC *=VXVw 'to beat, hit' (Avar A:ab-'to beat, hit; burst, shoot', Andi A:a-hun, A:a-tun to burst, shoot', etc.) ~ Basque *labur 'short' (< *'pounded down') (NCED 1023, ToB)

• Bur *-lté- > (Y) té- / -lté- 'to swear' / (H, N) te-s 'oath' ~ PEC *HilV 'to say' (Ingush le-, al- 'to say', Hunzib iA- 'to call', etc.) ~ PY *?V(?)Î- 'to speak' ~ PST *ld 'speak, speech' (NCED 572, CSCG 70)

• Bur *-ltâ- > tâ- / -ltâ- 'to put on (shoes, stockings)' ~ PEC *=omlV 'to put on (trousers, shoes)' (Andi =iA:in- 'to put on [shoes, footwear, trousers], etc.) (NCED 861, CSCG 130)

In the following examples the Burushaski initial dental stop t- corresponds with Proto-Caucasian lateral affricates:

• Bur. *(l)tam54 > (H, N) tam dél- 'to swim, bathe, wash' ~ PEC *AHwemV 'liquid' (adj.) > Avar A:ami-ja-, Archi A:ama-t:u- id., etc. ~ Basque *limuri 'moist, humid; slippery', etc. ~ PST *lidm 'to soak', etc. (CSCG 134)

• Bur *(l)tis > *tis 'wind' ~ PEC *A[a]rcV 'movement of air', etc. (see above)

• Bur *(l)tul > (Y) tul ~ (H) tol 'snake' ~ PEC *wHorAwVtV 'snake'55 (Avar borox 'snake', Lak Vikhli barcalu 'snail') ~ PY *?urol 'leech' ~ PST *rul ~ *rui 'snake' (NCED 1048, CSCG 218)

• Bur *(l)tal > tal 'palate; eyelid'56 ~ PEC *HMV 'mouth, jaw' (Tindi erA:i 'jaw', Tsakhur, Rutul yal 'mouth', etc.) ~ PY *jii- 'gills' (NCED 589, CSCG 75)

• Bur *(l)tal > *tal 'dove' ~ PEC *AeAë (Avar A:iA:i 'a kind of songbird', Lezgi kek 'cock', etc.) (NCED 776, ToB)

• Bur *(l)tal > (H) tal 'belly, stomach' ~ PEC *HlaXV / *HMV 'liver' (Avar tul, Tindi relaA:, Lak t:ilik, Lezgi leq, etc.) (NCED 586, CSCG 76)

52 With other derivatives: see Berger (1998).

53 This comparison raises interesting questions about the spread of horsemanship and the saddle, implying that this was prior to the diaspora of the western Dene-Caucasian languages. If the split between Vasco-Caucasian and Burusho-Yeniseian took place about 10 kya (see below: Postscript), and domestication of horses only ca. 6 kya, with the saddle even later, it is difficult to reconcile genetic transmission of the word in both Caucasian and Bur. Another, probably likelier, possibility is that an equestrian culture bequeathed a word such as *Xuli, *tluli 'saddle' to both Cauc and Bur separately, with subsequent usual developments in each language.

54 The notation *(l)t- means that the /l/ is only assumed from circumstantial evidence, since the correspondences are the same as in known Bur alternations of t- / -lt-.

55 This appears to be an old compound. Only the second element is compared with Bur *tul.

56 "Skr. talu- 'palate' [is] exactly matching Burushaski tal 'palate' — which is usually regarded as borrowed from Indian, but in fact also would be quite a regular reflex of [PDC] *HAtlV" (CSCG 75-76). The Sanskrit word, which has no clear Indo-European etymology, is thus probably one of the words adopted from Burushaski when Proto-Indic speakers entered the Indian subcontinent. See Witzel (1999).

• Bur *(l)tapi > (H,N) tapi 'stone terrace' ~ PEC *ÄepV (Chechen laba 'shed, peak of cap', Avar Äeb 'stone', etc.) ~ Basque *lape 'shelter under an eave'57 ~ PST *t-lep 'flat, tablet, etc. (NCED 777, LDC 32, CSCG 137)

• Bur *(l)tur 'cross-beam in door' > (H) tul, (N) tur ~ PEC *ÄwärHV (Tsez Xe 'bridge, stairs', Tindi, Karata X:eru 'bridge', etc.) (NCED 783, ToB)

• Bur *(l)tag > tag (Y) 'branch, shoot' ~ PEC *MwaxV 'stick, chip' (see above)

• Bur *(l)tharen- > (H, N) tharen-um 'narrow'58 ~ PNC *=iMV 'thin' (Avar iere'na-b, Agul kille-f, etc.) ~ Basque *lirain 'slender, svelte, lithe' (NCED 639, CSCG 105)

• Bur *(l)tan- > (H, N) tano 'colon (lower bowel of animal)', tano, taneelo 'bastard, of lowly birth'59 ~ PNC *HÄönü 'bottom' (Avar tinu 'bottom', Archi, Lezgi kan id., etc.) ~ PST *t-lär/ 'floor' (NCED 590, LDC 169)

• Bur *(l)tali > (H) tali 'slope (of a mountain)' ~ PEC *Mtü 'stone' (Avar talu 'rock, rocky plateau', Bezhta Mo 'stone', etc.) (NCED 773, CSCG 136)

• Bur *(l)tene > (Y) tene 'year before last', (H, N) te'n-dili 'last year' ~ PNC *HÄwinV 'winter, year' (Avar X:in 'winter', Bezhta Ai 'year', etc.) (NCED 591, CSCG 76)

• Bur *(l)tur- > (Y) tur-cun, (H, N) tur-Sun 'marmot' ~ PNC *lärV ~ *MrV 'hare' (Ingush lerg, Karata A:an-kala, etc.) (NCED 788, ToB)

• Bur *(l)ter > (H, N, Y) ter 'summer pasture, mountain pasture' ('Hochweide, auf die das Vieh im Sommer getrieben wird') ~ Avar lol 'open enclosure (for sheep)', Archi Aoli 'yard, place in front of the house', etc. < PEC *twelV (NCED 791) ~ Basque *lare 'pasture, meadow' ~ PST *räl 'fence, framework' (CVST II: 56, no. 204)

• Bur *(l)tar- > (H, N, Y) tar-ir 'skin bag' ~ PNC *töli 'color, skin' (Avar Ä:er 'color', Dargwa *k:uli '(sheep)skin', etc.) ~ Basque *laru 'skin, leather' (NCED 789, CSCG 130)

This development of initial *lt- > t- in Bur partially converges with that in one Caucasian language, Avar (specifically northern Avar: see NCED, pp. 52, 102), where the glottalized affricate PNC/PEC *X, *Äw yields t (glottalized dental stop). (The fuller forms of the following comparisons are found above.):60

• Bur *tano 'colon (of animal), bastard' ~ Avar tinu 'bottom' < PNC *HAönü

• Bur *tali 'slope (of a mountain)' ~ Avar talu 'rock, rocky plateau' < PEC *Äätü

• Bur *tapi 'stone terrace' ~ Avar (dial.) teb 'millstone, whetstone' < PEC *AepV

• Bur *tal 'belly, stomach' ~ Avar tul 'liver' < PEC *HMV

• Bur *tharen-um 'narrow' ~ Avar terena-b 'thin' < PNC *=iXtlV

Hermann Berger, the authority on Bur, ventured some Basque-Burushaski lexical comparisons in his early works (Berger 1956, 1959). In his last published work (Beiträge: 2008), Berger acknowledged this early interest, and reckoned that a relationship between Bur and other non-Indo-European remnant languages was thinkable but not demonstrable.61 Nevertheless, Berger (1959, p. 26, note 34) discovered the correspondence of Basque initial *l- = Bur

57 'refugio bajo el alero de un tejado / abri sous un avant-toit' (Azkue).

58 Aspirated /th/ is probably due to pretonal syllabic position. Note the similar -n- extension in Bur, Avar, and Basque.

59 S. A. Starostin preferred to compare this Bur word instead with PNC *Xana 'bottom' (CSCG 131).

60 But not the tense affricates %w, which remain in Avar as X: (or velarize to k: under certain conditions; see NCED pp. 52-54).

61 " ... eine Beziehung zum Baskischen und anderen nicht-indoarischen Restsprachen [ist] zwar denkbar, aber bei dem heutigen Entwicklungsstadium dieser Sprachen nicht mehr zu beweisen ist" (Beiträge, p. 1).

initial *t(h)-, which we consider valid (as developments of DC lateral affricates), based on the following examples:

• Bur *tdpi 'stone terrace' ~ Basque *lape 'shelter under eaves'62

• Bur *ter 'summer pasture' ~ Basque *lafe 'pasture, meadow'

• Bur *tar-ir 'skin bag' ~ Basque *lafu 'skin, leather'

• Bur *tap 'leaf; petal, page' (< *ltap) ~ Basque *lapaf 'bramble'

• Bur *tam del- 'to bathe', etc. ~ Basque *limuri 'moist, humid; slippery'

• Bur (H, N) turu-, (Y) tur- 'to fall apart', etc. (< *-ltur-) ~ Basque *lehef 'to burst, smash'

• Bur *(l)tharen- 'narrow' ~ Basque *lirain 'slender, svelte, lithe'

The following examples (in addition to several above) confirm the correspondence of Bu-rushaski medial -lt- with Caucasian lateral affricates. The reflex -lj- = [lj] occurs in a few words, apparently from *-lti, *-ltja-:

• Bur *diltar 'buttermilk'63 ~ PNC *rhaAwV 'milk' (Tsez riA 'butter', Avar rax 'milk', etc.) (NCED 949, LDC 153, CSCG 183)

• Bur *(y)alt- > (H, N) giydlt 'spoon, scoop'64 ~ PEC *ja[l]AwV 'wooden shovel' (Lezgi jirf, Bezhta ako, etc.) ~ Basque *saharde 'pitchfork; dinner fork; rake'65 ~ PST *jok 'scoop, ladle' (NCED 673, CSCG 113)

• Bur *yult > (H, N) yult 'time, (right) moment'66 ~ PNC *AajV 'time, day' (Akhwakh Aa-li-ge 'in the daytime', ze-Aa 'today', etc.) ~ Basque *ordu 'time, hour, occasion' (NCED 766, CSCG 133)

• Bur *ydltar > (H,N) ydltar 'upper leafy branches of a tree, crown of a tree', etc.67 ~ PEC *MlXVlV (Avar 1arA:el 'branch, bough', Tsez aAiru 'pod', etc.) ~ Basque *adaf 'branch' (< *ardaf)68 (CSCG 91)69

• Bur *-ltdltar- > (H) -ltdltar, (N) tdltar 'foreleg (of a quadruped), shoulder (of horse), 'human arm' (sometimes)70 ~ PNC *HluXE ~ *XulHV 'arm' (Avar ruA: 'arm, shoulder', Archi A:ol 'shoulder-blade, foreleg (of animal)', etc.) ~ PST *t-luH / *t-lu-k (?) 'hand, arm, wing' (NCED 588, CSCG 138)

• Bur *maltds 'butter' ~ PEC *nheAV (Chechen nalxa 'butter', Archi naA: 'milk', etc.) (NCED 849, CSCG 146)

62 See the complete DC etymology (CSCG 137) for semantic developments: original meaning probably something like 'flat slab of stone'. Chechen and Ingush also have the meaning 'shed', possibly originally a crude outbuilding with roff made of stone slabs.

63 Bur initial d- ~ Caucasian *r is the regular initial reflex: see CSCP, p. 41.

64 Bur giyalt appears to be a compound of the verb giy- 'pour', etc. + -yalt or -alt.

65 The Basque word appears to be an old compound: *sa- + *harde (with obscure first element).

66 In stem-final position we would expect *yul (see below). In this case there was probably a variation between *yul (in absolute final position) vs. *yult- (preceding inflectional suffixes), with analogical leveling to the latter.

67 Cf. also (H,N,Y) galtar 'small twig', (H,N) giltir 'pod, husk (of peas, beans, etc.)'.

68 In Basque this word has merged phonetically with *adaf 'horn' (see above).

69 The correspondence of Bur *y- = *j— PNC initial *fi- is recurrent: cf. Bur *yat-is 'head' ~ PEC *üwömdV 'brain, head' (below).

70 (Y) 'projecting breasts' ('hervorstehende Brüste').

• Bur *hardlt 'rain, rain cloud'71 ~ PEC *renXwV ~ *ranXwV 'cloud, fog' (Chechen doxk 'fog', Khinalug unk 'cloud', etc.)72 ~ Basque *lanbro 'fog, mist, drizzle'73 ~ PST *rer/ 'drop, rain' (NCED 947, CSCG 179)

• Bur *alt- 'two', *w-dlt- 'four' ~ PWC *p(:)aXa 'four' (Ubykh pXd, etc.)74 ~ PST *P-Kj 'four' ~ Basque *lau- 'four' (NCED 314, CSCG 212)

• Bur *balti 'front room of house, veranda' ~ PEC *bulXV 'house' (Hunzib buAi 'at home', Lak burca-lu 'threshold', Hurrian purli 'house', etc.) ~ Basque *borda 'cottage, cabin, stable' (NCED 312, LDC 158, CSCG 15)

• Bur *'-ltV-r 'to show' > (Y) '-ltar-, '-ltir-, (H, N) '-ltir- ~ PEC *?ilV 'to look' (Chamalal X:i-d, Tabasaran lig-, etc.) ~ PY *?V(?)l- ~ *?V(?)ri- > Kott. r)-al-iga 'I know' ~ PST *t-ld(H) 'to see, look' (NCED 209, CSCG 255)

• Bur *multur > (H,N) -multur 'nostril' ~ PEC *wenXV (Batsbi marAo 'nose', Bezhta moXo 'beak', etc.) ~ Basque *mutur 'snout, muzzle; end, edge' < *murtu-r ~ PST *luH 'head' (NCED 1041, CSCG 216)

• Bur *qhVltd 'sack, pocket' ~ < PEC *GHirXwV (see above)

• Bur *-hdlt- 'to wash' > (Y) (ba)-hdlt-, (H, N) -alt-/ -yalt- ~ PEC *=VXVn 'to wash, pour, weep' (Chechen =elx- 'to weep; to pour (of rain)', Archi e=X:in- 'to make an ablution', etc.) ~ PST *t-ler ~ *t-lar 'to wash, clean' (NCED 1023, CSCG 212)

• Bur *dalt- > (N) daltdn- 'to thresh'75 ~ PEC *=V-rlV < *rVLV 'to thresh' (Batsbi arl-, Bezhta =ol-, etc.; Andi loli 'threshing; threshing floor; Archi Aorom 'threshing board', etc.) ~ Basque *larain 'threshing floor' (NCED 1031, CSCG 182)

• Bur *-welji 'dream' > (Y) -welji, (H,N) -ulji ~ PNC *lemXA 'dream' (Dargwa hank 'sleep', Karata hanXu 'fog, cloud', etc.) ~ Basque *lainho 'cloud, mist, fog'76 (NCED 512, CSCG 93)

• Bur *-lji 'behind, backwards'77 ~ PEC *Xi 'below, down' (Bezhta Ai- 'down, below', Lak luw id., etc.) (NCED 778)

• Bur *-welji 'womb, afterbirth' ~ PEC *rVHVnXwV / *HVrVnXwV 'some internal organ': Tindi reX:a-(k:a riX:i) 'diaphragm', Rutul nixrd 'placenta', etc. (NCED 955, ToB)

• Bur *hulj- > (Y) huljd- 'to ride (a horse)' ~ PEC *?iXV 'to run, leap' (Avar X:u-r-d- 'to dance', Rutul hi=iga- 'to drive, urge', etc.) ~ PST *t-laj(H) 'to run, gallop' (NCED 209, CSCG 256)

The Burushaski reflex of all lateral affricates in stem-final position is simply /l/:78

• Bur *'-yal- 'to hear' ~ PNC *=eAu 'to hear': Andi anXi- 'to hear', Budukh ix- id., etc. (NCED 411, CSCG 46)

• Bur *w-el- / *b-el- 'to put on (clothes)' ~ PEC *=VXV 'to put clothes (on the upper body)': Chamalal, Tindi =al-, Khwarshi s-iX-, etc. ~ PY *?alVr 'trousers' (NCED 1024, CSCG 212)

• Bur *bal-, *-wdl- 1 'place between the shoulders', 2 'back of the shoulders, upper part of the back', 3 'back' > (H) bdlbal 1, bdl-giciq 2, -wdldas 3, (N) bdlbal 1, bdl-gicaq 2, -wdldas 3, (Y)

71 Initial *ha- may be influenced by hará- 'to urinate'. In stem-final position we would expect *(ha)rál (see below). See the note to *yult, above.

72 PEC % is reconstructed on the basis of circumstantial evidence.

73 The Basque word requires a metathesized protoform such as *XanwrV.

74 This is probably related to PEC *bünle 'eight' (Avar miX:-go, Hunzib beA-no, etc.).

75 Bur initial d- < *r: cf. Bur *díltar 'buttermilk', above.

76 "Andian languages demonstrate a non-trivial semantic development 'dream' > *'vision' > 'cloud'" (NCED). Likewise in Basque.

77 Starostin (ToB) prefers to compare Bur *-lji with PNC *Hl[a]XV 'breast, back', etc.

78 Apparent exceptions are probably the result of analogical leveling. (See the notes to *yult and *harált, above.)

wdldes 3 ~ PNC *buAV 'upper part of the body' (Batsbi bali 'shoulders', Adyge, Kabardian pAa 'upper part of the back', etc.) ~ Basque *sor-balda 'shoulder' < *sor-barda (NCED 313, LDC 32, CSCG 158)

• Bur *-hil 'lip, edge, shore' ~ PEC *HaKwV (~-e-,-t-) 'forehead' > Chechen haz, Tindi haA:a, etc.)79 (NCED 543, CSCG 84)

• Bur *bal 'marrow, brain, kernel (of walnut)' ~ PEC *bherXV '(large) intestines' > Bezhta baAa 'large intestine', Udi buq:un 'belly', etc.) ~ Basque *barda / *marda 'belly, abdomen, bowels, tripe, stomach, rennet' ~ PY *pi?il 'intestine(s)' ~ PST *bik 'bowels' (NCED 297, CSCG 13)

• Bur *el- > (Y) el-den 'year before year before last' (den 'year') ~ PEC *?VAwV 'last year' (Avar dial. uAi-sa, Tsez, Hinukh eAi, Bezhta iAe, etc.) ~ Basque *urte 'year' (NCED 225, CSCG 259)

• Bur *bel-is 'ewe that has already given birth' ~ PNC *bhaAwi 'small cattle' (Bezhta, Hunzib biA 'sheep', Andi belir 'deer', etc.) ~ Basque *bil-doc 'lamb (that has begun to feed itself)' (NCED 293, CSCG 12)

• Bur *(l)tal > tal 'dove' ~ PEC *AeAe (Avar A:iA:i 'songbird', etc.) (see above)

• Bur *-ul 'belly, abdomen' ~ PEC *=ir(a)lV 'stomach; rennet, abomasum' (Karata m-eA:u 'stomach', Hunzib b-eA 'rennet, abomasum', etc.) ~ Basque *urdail 'stomach, abomasum, womb' ~ PST *t-ldw 'belly, stomach' (NCED 670, CSCG 112)

One might have noted that in some forms above (*hardlt 'rain, rain cloud', *-hdlt- 'to wash') Burushaski has /lt/ in what appears to be final position, an apparent contradiction to the rule just cited. The restoration of /lt/ in these cases can be attributed to analogy, based on inflected forms such as hardltiq 'rainfall, rainclouds'. Likewise in the case of Bur *bel-is 'ewe' (see above) the development of > stem-final /l/ had already taken place before the addition of -is (a frequent Bur suffix).

For more details on DC lateral affricates and their reflexes, see Bengtson (2008a: 59-61).

Typological parallels of the change TL > LT: If we symbolize the postulated change of DC lateral affricates to Bur /lt/ (reduced in initial position to /t/ and in final position to /l/) as TL > LT, some typological parallels support the probability of this type of phonological change. The clearest and most familiar may be the change seen in Spanish:

• Lat. spatula > OSp. espadla ~ espalda > MSp. espalda 'back'

• Lat. capitulu > OSp. cabidlo ~ cabildo > MSp. cabildo 'town council'

• Lat. foliatile > OSp. hojadle ~ hojaldre > MSp. hojaldre 'puff pastry'

• Lat. titulu > (Catalan) title > OSp. tidle ~ tilde > MSp. tilde 'written accent'

In Old Spanish the /dl/ and /ld/ forms coexisted, while in the modern language the /ld/ forms have prevailed. In Judeo-Spanish the change has been extended to include imperative plural + clitic constructions (Bradley 2006: 80):

• JSp. traeldo = MSp. traedlo 'bring it' < Late Latin tra(h)ete + illu

• JSp. tomalda = MSp. tomadla 'take it'

• JSp. daldo = MSp. dadlo 'give it'

79 For semantics, cf. Hunzib bil 'lip', Tindi bala 'edge, end, corner', Lezgi p:el 'forehead', etc.; Basque *beia-f 'forehead'.

In English a parallel can be seen in the popular name Sheltie for Shetland pony or Shetland sheepdog. In recent American English chipotle, the name for a dried chili pepper derived (through Mexican Spanish) from Nahuatl, is frequently pronounced /cip'olti/.80

It is interesting to note the derivation of Spanish alcalde 'judge' < Arabic al-qadi 'the judge' (Corominas 1990: 38), in which the Spanish cluster /ld/ substitutes for the Arabic "emphatic" d (which in turn comes from the Semitic lateral sibilant *s).

In Tibetan and other Bodic languages of the Sino-Tibetan family PST *t-l- may yield /lt/, /ld/, or /lc/, for example:

• Tib lto 'belly, stomach' < PST *t-ldw id. ~ PEC *=ir(a)LV, Bur *-ul, Basque *urdail, etc. (see above)

• Tib lte 'navel, center' < PST *t-laj 'center, middle' ~ PNC *=eXE 'middle, half', Basque *erdi id., PY *?a?l 'half' (CSCG 46)

• Tib lta 'look' < PST *t-la(H) 'to see, look' ~ PEC *?ilV 'to look', Bur *'-ltV-r- 'to show', etc. (see above)

• Tib ltag 'nape, back part of the neck' < PST *t-luak 'back' ~ PEC *Xarqwe 'forehead; cap', Basque *lok- 'temple; middle of forehead' (NCED 775, ToB)

• Tib ldeb 'leaf, sheet' < PST *(t-)lap 'leaf' ~ Burushaski *ltap- 'leaf; to wither', PNC *Xapi 'leaf', Basque *lapar 'bramble', PY *jape 'leaf' (see above)

• Tib ldeb-s 'side' < PST *t-lep 'border, side' (ToB) ~ (? Basque *lepo 'neck')

• Tib lcag 'rod, stick' < PST *t-ldk 'stake, stick' ~ Bur *ltag 'branch, shoot', Avar X:ox: 'stubble', etc. (see above)

• Tib lcag-s 'iron; lock' < PST *t-l[ia]k 'iron' (ToB) ~ Bur *ltik > tik 'earth, ground; rust'

The difference from Basque and Burushaski is that Bodic has the metathesized cluster only initially, not medially, as in the other languages. Since Burushaski is spoken in an area immediately adjacent to the Bodic dialects (Balti and Purik, archaic Bodic dialects, are spoken directly east of the Burushaski area), it is possible that at some time in the past, both families had lateral affricates, and that the change of *TL > /lt/ (etc.) was an areal phenomenon that affected Burushaski and Bodic, but not more distant Sino-Tibetan languages (such as Lushai, which frequently has /tl/ or /thl/ < PST *t-l-.

Morphology

Nouns

In the Burushaski nominal system the case endings, as admitted by C himself, are the same for both singular and plural. Bur therefore has an agglutinating morphology, not the inflected morphology typical of IE. We find the Bur case endings far more compatible with those of Basque and Caucasian, including the compound case endings found in all three families (Bengtson 2008a: 90-92).

Furthermore, though it is not mentioned by C, many (about 150) of the most basic nouns are bound forms, i.e., they cannot occur without a pronominal prefix (for example, Bur (H, N) -ltur 'horn' manifests as a-ltur 'my horn', gu-ltur 'thy horn', i-ltur 'his horn', mu-ltur 'her horn', etc.). Toporov (1971) pointed out these remarkable parallels between Bur and Yeniseian:

80 Chipotle is also the name of a restaurant chain. Evidence of the metathesis chipotle ~ chipolte can easily be found with an internet search of chipolte.

'my hand' 'thy hand'

Burushaski (H, N) fl-riiq gw-ríiq

Yeniseian (Ket) flb-íaq wg-íaq

Table 5

These prefixes can be reconstructed to something like *aya- 'my' / *uxGu- 'thy' (see the PDC pronoun stems, below), and the word 'hand' itself is reconstructed as *rVrjHV (by Starostin: ToB). This type of construction is totally alien to IE patterns, as is the enormous number of different plural suffixes: about 70, as noted by C (p. 23). So is the multiple class system of Bur, which is far more similar to class systems in Caucasian and Yeniseian than to gender in PIE.

Table 6. Burushaski noun classes81

Class type human non-human

Class description human-male human-female non-human animate (animals, countable objects) inanimate (uncountable objects, mass nouns, abstractions)

Class letter (Lorimer) hm hf x y

Class number I II III IV

Examples (Hunza-Nager) hir 'man' '-uy 'father' qhudáa 'God' gus 'woman' dasín 'girl' parí 'fairy' hagúr 'horse' báalt 'apple' '-l-cin 'eye' phu 'fire' ge 'snow' chap 'flesh'

Table 7. East Caucasian noun classes

Class type human non-human

Class description human-male human-female non-human animate inanimate

Class number I II III IV

Examples (Lak) cuw 'man' p:u 'father' ars 'son' s:ar 'wife' c:us:a 'female' ninu 'mother' cu 'horse' cimus 'onion' ja 'eye' c'u 'fire' s:in 'water' dik' 'flesh'

Bouda (1949); Catford (1977: 298-299).

Personal Pronouns

It is perhaps the personal pronouns that show most clearly the deep incompatibility of Bur and IE. IE, as is well known, is typified by the first and second-person pronouns *H\eg(H)-'I' / *(e)me- 'me' and *te-, *towe-, *tuHx- = *tu- 'thou, thee'. In Bur (Berger 1998: I, p. 80) the scheme is entirely different.

81 "The difference between class III and IV nouns is not as straightforward as [implied in the table.] Many class IV nouns are countable (and take class-specific plural endings), e.g. HN -riiy 'hand', -usis 'foot', -ltumal 'ear', -akin 'liver', ha 'house', tom 'tree', jame 'bow (made of horn)', while some abstract nouns are class III, e.g. cuti 'leisure, holiday', rupia 'money', cila 'the coldest period of the year', hariip 'melody'. Yet there is, of course, this strong tendency that objects and materials (incl. artifacts made from such materials) lacking a clearly defined or stable physical form are class IV. So 'trees' are IV, but their 'fruits' are III." (B. Tikkanen, pc.).

Table 8. Burushaski Personal pronouns

Person 1 sg. 2 sg. 1 pl. 2 pl.

Dial.\ Form dir. g.-e. v.p. dir. g.-e. v.p. dir. g.-e. v.p. dir. g.-e. v.p.

Hunza & Nagir je jáa áa- un uq N um gu- gú- gó{o)- -kó(o)- mi míi mée mi-mí-mé(e)- ma mamá-máa-

Yasin ja un gu-gú-gó(o)- -kó(o)- mi ma mamá-

Berger (1998); dir. = direct, g.-e. = genitive-ergative, v.p. = verbal prefix.

Here we see that the Bur system is suppletive, with different stems for direct forms and oblique forms, in both first and second person. C (p. 72) attempts to connect Bur je, ja with PIE *Hieg(H)- but he can do so only by violating the sound correspondence discussed above (PIE *g, *gh = Bur g, g)! He further tries to connect Bur un (~ um, utj) with PIE *tuHxom, emphatic form of *tuHx = *tu-, but again only by requiring another unprecedented change: t > d > 0!

For comparison, below we present the attested forms of personal pronouns in the Indo-Iranian languages that surround Burushaski:82 see tables 9 & 10.

Table 9. Personal pronouns in Nuristani & Dardic

Person 1 sg- 2 sg. 1 pl. 2 pl.

Lg. \ Case direct oblique direct oblique direct oblique direct oblique

Kati vuze, onc ía, ye t'u to, tu ema, imâ, yimo sa, so

Waigali aqa u tü tü ami ame wi wa

Askun ai yu tu ö ima wï yä

Prasun unzü andéis i/üyü ütyöis ase miü

Dameli ai mu, mo tu to ai amâ bi myâ

Gawar a mo tu to amo, ama- më

Wotapuri au ma- tu ta, tha- mü, mun thü

Sumasti â mo tu to âba ama wi ima

Pasai a ma- td, to to, tê- hama (h)ëmâ, momâ, myâ

Tirahi au, ao me tu, to te, të ao, mâ men tao tâ

Kalasa â mai tu / tü tai âbi homali âbi mïmi/e

Khowar awá ma tu ta ispá pisá

Torwali â, ai me, ma tu ta mo, moi mo, ma- to, tho to, ta-

Baskarik Va ma- tu tha- ma tha

Garwi yah ma- tu ta- mâ ta-

82 Thanks to E. Bashir for some corrections of Khowar forms.

Person 1 sg. 2 sg. 1 pl. 2 pl.

Lg. \ Case direct oblique direct oblique direct oblique direct oblique

Maiyan ma me tü te be za tus sa

Kanyawali ma ml, mi tu tl, ti be za tus ca

Phalura ma tu be as- tus

Sina ma(h) ma tu(h) tha, tü be as- tsho, co

Kasmiri ba(h) m'e cü(h) c'e as' as'e twah' twah'e

Vedic aham a. ma(m) tuvam a. tva(m) a. asman, d. asmabhyam a. va, g. yusmakam

Édel'man (1978, 289); a. = accusative, d. = dative, g. = genitive.

Table 10. Personal pronouns in Pamir languages

Person 1 sg. 2 sg. 1 pl. 2 pl.

Lg. \ Case direct oblique direct oblique direct oblique direct oblique

Yidgha zo, Z3 man, mun tu, ta tu/a/o/a max, mox maf, mof

Munjan za, za man, mun tu, ta to/a/a/aw mox mof

Sughni wuz mu tu mas tama

Rusan az mu tu ta mas tama

Khuf waz mu tu taw, ta mas tama

Bartangi az mun, mu tü ta mas tamas

Orosor waz mun, mu tu ta mas tamas

Sarykoli waz my, myn tew ta, ty mas tamas

Iskasim az(i) mak tï fak mïx(o) mïcïv(o) tïmïx tïmïx(ïv)

Yazghulam az mun, mon tow tu, ti- mox

Wakhi (w)uz, waz maz tu taw, tow sak sapo sa(y)ist sav

Avestan azam g. mana tuuam, tü g. tauua g. ahmakam OPers. g. amaxam g. yusmakam

Efimov & Édel'man (1978, 218); g. = genitive.

In spite of some formally similar forms in the contemporary languages, e. g. Yidgha mox, Munjan max, Iskasim mix 'we', vis-à-vis Bur mi id., deeper comparison shows that they have quite separate origins. Thanks to the archaic Indo-Iranian literary languages, Avestan, Old Persian and Vedic OI, we can project the Indo-Iranian forms into the past and derive them from the stem *asma-, from PIE *ns-mé-. Bur mi, on the other hand, maybe comes from PDC *mi(nV) 'self, (our)self', according to Starostin (CSCG 146: cf. ST: Lushai mi 'me, us, my, our', etc.).

We propose that comparison of the Bur personal pronouns with those of East Caucasian (and other DC languages) is more fruitful as well as more straightforward than comparison with IE. Both Burushaski and the reconstructed Proto-(North) Caucasian have suppletive pronoun stems in the first and second person singular. For the present purpose, let us compare Bur with two East Caucasian languages, Khinalug and Tsakhur. Khinalug is the highest

(2300 m. = 7546 ft.) and most remote village in Azerbaijan, where the inhabitants still speak a Caucasian language.83 Tsakhur is also spoken in Azerbaijan as well as in Dagestan. Both languages appear to have preserved remnants of old eastern Dagestanian suppletive paradigms: see table 11.

Table 11. Personal pronouns in Eastern Dagestanian languages

direct genitive dative

1st person sg. Khinalug zi (nom.) jä (erg.) i, e as

Tsakhur zu jiz-in za-

2nd person sg. Khinalug wi (nom.) wa (erg.) wi ox

Tsakhur84 wu ~ vu (= gu) j-iv- (= H£-) wa-

According to Nikolayev and Starostin (NCED, pp. 402, 483-84, 855, 1014-15, 1084-85), the original Proto-Caucasian pronominal paradigms were very complicated, and difficult to reconstruct with much certainty. In the first person singular West Caucasian and most East Caucasian languages have forms going back to PNC direct *zo(-n), ergative *?ez(V), genitive *?iz(V), oblique *za-, though Lak and Dargwa have instead a first person stem *nt (cf. Basque *ni 'I', PST *i]a- 'I, we', etc.). In the second person singular PEC had a "complicated suppletive paradigm" consisting of direct *uo(-n) / *kwV = *gwV, ergative *?okwV = *?ogwV, genitive *?euV / *?iuV, and dative *du.

Clearly a great deal of rearrangement has taken place in all of these languages since the original paradigms of thousands of years ago. West Caucasian abandoned most of the suppletive stems and kept only *sa 'I' (= *zo) and *wa 'thou' (= *uo). One East Caucasian language, Dargwa (Akushi and Urakhi dialects) has retained the stems *ni and *kwV = *gwV, resulting in a paradigm coinciding with that of Basque:85

'I' 'thou'

Dargwa nu hu

(Akushi, Urakhi)

Basque ni hi

Table 12

We can then summarize the genesis of the Burushaski first and second person singular pronouns as follows: see table 13.

Interrogative Pronouns

As stated correctly by C (p. 74), Bur interrogative pronouns are built on bases containing the labials /m/ and /b/: *me- 'who' and *be 'what', and he also quite correctly recognizes the Bur tendency to waver between /m/ and /b/. C connects the Bur interrogatives with the rare IE in-

83 http://www.xinaliq.com/; http://www.eki.ee/books/redbook/khinalugs.shtml.

84 Note that Tsakhur exhibits free variation between the two old second person stems: wu < *uo vs. gu < *kwV.

85 Note that some Dargwa dialects have instead retained the PEC stem *zo as du 'I'.

Proto-Burushaski Proposed cognates Proto-Dene-Caucasian86

1st pers. sg. direct *ja Khinalug zi Tsakhur zu Chechen so PWC *sa Yen. *?a3 *zV

1st pers. sg. oblique *a- (< *r/a-)87 Dargwa nu Basque *ni Kott *y]-l-y]ss *rv

2nd pers. sg. direct *u-n Archi un Khinalug wi Tsakhur wu (~ gu) PWC *wa Yen. *?aw l *?u *wV

2nd pers. sg. oblique *gu-1 *go- Tsakhur gu (~ wu) Chechen ho Dargwa hu Basque *hi Yen. *kV-l*?Vk- *xGwV

Table 13

terrogative stem *me/o-, attested only in Anatolian, Tocharian, and Celtic. We must point out, however, that the *mV- interrogative is much more richly attested in DC than in IE, and furthermore the m ~ b alternation is attested in DC, but not in IE:

• Caucasian: PEC *mV- > Chechen mi-la 'who', mi-ca 'where', ma-ca 'when' etc.; Andi emi-'who', Chamalal im id., Tind. ima-la 'who'; Lezgi, Agul mu-s 'when' / Archi ba-sa 'when'

• Basque: ba- conditional prefix, 'if-' (Trask 1997: 225)89

• Sino-Tibetan: PST *mV- > Karen *mV 'what', Serdukpen mu id., Bodo *ma? id., Ao Naga *mV id., Sichuan *mV id. (ToB) / PST *Pa 'what, which' > Burmese ba 'what, which', Jingpo pha1 'what', Bodo ba 'which one' (CSCG I: 92)

86 S.A. Starostin (ToB, 2004-2005a, 2004-2005b).

87 Loss of initial PDC *r in Bur (or replacement with /h/) is regular, per Starostin (CSCP 48).

88 According to Starostin, Ket b-/?ab- belongs here; but the development *b < *m < *r (CSCP 48) does not agree with the rules established by him earlier (Starostin 1982), while the Kott data agree excellently:

~aliga < räliga 'ich Weiss' = *'mein Wissen'

~aiteän (raiteän) 'ich will' = *'mein Wunsch'

~apear < rapear 'in; hinein' < *'mein Inneres'

~ani < rani 'mein Schwiegersohn' : Ket en 'Schwiegersohn'

~ama < rama 'mein Mutter'

~dp < rop 'mein Vater'.

See W. Werner, Vgl. Wörterbuch der Jenissej-Sprachen, Bd. 2, Wiesbaden 2002, 29-30, who has collected the Kott examples from Castren 1858. Concerning Ket ab- 'my', Arin b(i)-, Kott m-inso, and Ket & Yugh 1st person sg. verbal exponent ba-/bo-, a promising cognate appears in Hurrian -iffu-l-iffd-l-iffe- 'my', pl. -iff=az 'our'; and in the ergative suffix of the 1st person -aw, e.g. tad=aw 'I love [it]' (see Gernot Wilhelm, „Hurrian," In: The Cambridge Encyclopedia of the World's Ancient Languages, ed. by Roger D. Woodard, Cambridge: University Press 2004, 107, 112).

89 For semantic development, cf. Old Irish ma 'whether, if' < PIE interrogative stem *me/o-, cited by Casule (p. 74); German wenn 'if' < 'when'; Czech ci 'ob', Polish czy 'ob' < PIE interrogative stem *kwei-, etc .

• Yeniseian: PY *wi- / *we- 'interrogative pronoun'90 > Ket bistrj / bisay ( < bisairj3) 'where', bisse 'who' (masc.), be-sa 'who' (fem.), bi-Ia5,6 'how', bz-l'es / bilas 'whither'; Kott bi-li 'where', bilthur 'whither', bilcay 'whence', bi-I'ag 'which', etc.

Verb

In the verb the Bur variance from IE is just as pronounced as in the noun. The "typological similarity" claimed by C (p. 75) is only in regard to vaguely similar systems of aspects and tenses, without any material parallels pointing to common genetic origin. The verbal endings (C, pp. 75-77) are similar only in that both Bur and IE have endings containing n and m, though there are no real correspondences between them. Most striking is the existence of the Bur template verbal morphology with as many as four prefix positions preceding the verb stem.

Table 14. Burushaski verb template

prefix position -4 -3 -2 -1 0 +1 +2 +3 +4 +5 +6

AP/

NEG D PRON CAUS VERB PL.SBJ. DUR 1sgSBJ NON-FIN/ SBJ Q

e MODAL

u e 3 MH negative marker subject version pronominal prefix (person/ class) causivity/ valence verb stem plural marker

Tikkanen 1995, Berger 1998, Anderson, ms.

It is well known that Proto-IE had few verbal prefixes.91 The Bur prefixal template is far more compatible with languages such as those of the Yeniseian family, especially the well-documented verbal morphology of Ket, and of the extinct Kott; Basque, Caucasian (especially West Caucasian), and Na-Dene also seem to preserve distinctive features (multiple noun classes, polysynthesis, extensive verbal prefixing of pronominal and valence-changing gram-memes) of the postulated Dene-Caucasian proto-language: see, e.g. Bengtson (2008a, 2010a, 2010b), G. Starostin (2010a).

Numerals

C (p. 75) makes some ingenious Burushaski-IE comparisons of the numerals 'one', 'two' (actually Bur 'two' + IE *Hal- 'other'), 'eight', and 'nine'. Before commenting on these attempts, let us first provide some background information on the complete numeral systems of Bur and its IE neighbors:

90 Yeniseian *w- is the regular reflex of PDC *m- (CSCP 35).

91 Concerning verbal prefixes in IE, the situation is rather complex. Most of the historically attested IE languages use prefixes, which represent the prepositions, sometimes "frozen," as in Hittite. The verbal augment is another example, different from usual prefixes. Its existence is attested in Indo-Iranian, Armenian, Greek. E. Hamp (1997, 127) tried to demonstrate that it is not excluded that it was known in other languages too, e.g. in the Latin form enos 'we' instead of nos in the Carmen Arvale. This means that this "prefix" would be free and not dependent only on the verb. There could also be some old prefixes of the type "s-mobile" in Indo-European, maybe corresponding with the Afroasiatic s-causative.

Table 15. Burushaski numerals

Dial. \ Num. 1 2 3 4 5

Hunza & Nagir hin han alto altâc iské(n) usko wâlto wâlti chundo chindi

hik altâ(n) iski

hen alto iské wâltu/ cendo

Yasin han hek altâc altâ(n) isko iski wâlte cindo, -i

Comments H 18: hun w + *alt- 2

Dial. \ Num. 6 7 8 9 10

Hunza & Nagir misindo misindi mao thalo thalé altâmbo altâmbi hunco hunti toorumo toorimi

Yasin bicindu bisinde thalo thalé altâmbu altâmbe huçô huti torum

Comments -mis, pl. -mianc Y -mes, pl. -mac 'finger' + '5' maybe cf. Khaling târ 7 (Hd 361) *altan be 2 without *hun- 1 minus *Cu 10? or from Y -cu-'take away' (Bl 328) toorum Y. taurum so many; cf. Khaling tadham 10 (Hd 361)

Dial. \ Num. 10 20 30 40 50

Hunza & Nagir toorumo âltar N âlthar âltar-toorumo /-toorimi alto-âltar N -âlthar alto-âltar toorumo

Yasin torum âltar

Comments < *alt- + *tarum-(B 16) 20 + 10 2 x 20 (2 x 20) + 10

Dial. \ Num. 60 70 80 90 100

Hunza & Nagir iski-âltar iski-âltar toorumo wâlti-âltar wâlti-âltar toorumo tha

Yasin iski-âltar walte-âltar tha

Comments 3 x 20 (3 x 20) + 10 4 x 20 (4 x 20) + 10

Berger 1998.

Table 16. Nuristani & Dardic numerals

Language 1 2 3 4 s б 7 8 9 1С

Kati ev d'u tre st(a)vo puc so, su sut ost, ust noh, nu due

Waigali ew, ek dü tre cata püc, puc su sot ost m dos

Askun ac do, du trä, tre eata ponc, pone su, sdu sut ost no, nu dus

Prasun i/upün lü est, cï c'pu wucu wusu sëtë astë m, nüyu läze

Dameli ek du trâ cor pac so sat ast no das

Gawar yak, yok du Ae/e eur po(n)e suo, sou sot, sat ost m dos, das

Wotapuri yek, yaks du ta, Aa e/sawur pan^/e so, se sat, sätd at, ata nau das (a)

Sumasti yäk du Aye, Aïe eöuur pon soo sat, sat ast nï däs

Pasai ï do trä, Arn car, eor pan] sa sat ast no de

Tirahi ek do tre eawor pane xo sat axt nab dah

Kalasa ek du tre cau pon,pans so sat ast no das

Khowar ï ]u troi cor pon] choi sot ost neoh ]os

Torwali e(k), e du, du, do ca, ça cau pan] so, so sat at nom das

Baskarik ak du tha cor pan] so sat ath num das

Maiyan ak du ca saur panz soh sat ath num das

Kanyawali ek du ça eour pas so sat ath nau däs

Phalura ak du tro cur panz soh sat ast m das

Sina ek du çe car pot sa sät äs nau daï

Kasmiri akh zü(h) tr'ü(h) eor pane sah sath ath naw da(h)

sc

Table 16 (cont.)

Language 20 30 40 50 60 70 80 90 100 1000

Kati vici, V3C3 vica-duc d'u-vdcd du vaca-duc tra vaca puc vaca

Waigali wist wisi-dosi du-wisi du-wisie-dosi trew(i)si catta-wisi puc-wisi

Askun wist wisi-a-dus do-wisi duisa-dos tre-wisi trewisi-dos cata-bisi, catto-isi cata-wisi-dos punc-wisi

Prasun 3U, zu laja(i)z lyogju lejebiz scog^u cpagju wucegju

Dameli bisi bisio-das du-bisi paz-bisi

Gawar isi isi-o-dos du-isi Ae-isi cur-isi painsi

Wotapuri bis(a) bis-o-das du-bis du-bis-o-das ta-bis cawur-bis panybis

Sumasti ist isi-das du-isi Aye-isi cour-isi pon-isi

Pasai wast wast-o- dai triw du-wya du-wya-u-dai tra-wya, Ae-wya car-wiya car-wea-de panjawia

Tirahi biau, byeh biau-dah do-be do-biau-dah panz-be

Kalasa bisi bisi-je-das du-bisi du-bisi-je-das tre-bisi(r) trebisi-das cau-bisi(r) caubisi-das pon-bisi

Khowar bisr bisr-jos ju-bisr ju- bisiro-ce-jos troi-isir troibisir o-ce-jos cor-bisr corbisiro -ce-jos ponj- bisr, sor

Torwali bis das-obis du-bis das-o-du-bis ca-bis co-bis panj-bis, soh

Baskarik bis das-obis du-bis das-o-du-bis tha-bis cor-bis panj-bis

Maiyan bis das-obis du-bis das-o-du-bis ca-bis saur-bis sal

Kanyawali bis du-bis ca-bis cour-bis sal

Phalura bhis bhis-e-das du-bhisa tro-bhisa cur-bhisa panz-bhisa

Sina bi(h) bi-ga-dai dibyo dibyo-ga-dai cebyo cebyo-ga-dai carbyo carbyo-ga-dai sal sas, sas

Kasmiri wuh truh catujuh pancah sath satat suth namat hath sas, sosu

Edel'man 1978, 285-87.

Table 17. Numerals of the Pamir languages

Language 1 2 3 4 5 6 7 8 9 10

Yidgha yü loh xuroi csir pans, on] úxso ávdo ásco nöu los

Munjan yü lu x'roi cfür ponz óxso óvdo oskyo nau da

Sughni yiw, yi bu, biyün aráy cavor pinj xÖf (w)üvd waxt nöw bis

Rusan/Khuf yiw, yi baw aráy cavür pinj xüw (w)üvd waxt naw/nöw bos

Bartang yiw, yi baw aray cavór pinj xöw üvd waxt naw bus

Sarykoli iw, i bew, ba aroy cavúr pinj xel uvd woxt new bes

Yazghulam w%) bow cuy cer pen] xu uvd uxt nu(w) bus

Iskasim uk, uk db(w) ru(y) cbfur pünj xul(l) uvd ot naw, nu dl dust

Wakhi yi(w) bu(y) tm{y) cúbur panj sab rnb at naw bas

Language 20 30 40 50 60 70 80 90 100 1000

Yidgha wisto xüroiwist panzwist

Munjan bist < Pers.

Sughni bu bis ara bis cavör bis pinj bis xöybis bis bis azo-r

Rusan/Khuf baw bos aray bos cavür bos pinj bos xüw bos bos bos-uk hazo-r

Bartang baw bus aray bus cavör bus cil pinj bus xöw bus bus bus-ak azör

Sarykoli

Yazghulam wast wast-a bus bow wast bow wast-a bus cuy wast cuywást -at bus cer wast cer wást-at bus pen] bist (h)azór

Iskasim bist, Sang dbwist

Wakhi wist / bist < Tajik bist-dt bas bu-bist bu-bist-dt-bas truy-bist tru-bist-d(t) bas cúbur bist cúbur bist-d(t) bas panybist bas-bas sad < S.

Payne 1989, 435; Efimov & Édel'man 1978, 226-28.

The first serious analysis of the Burushaski numerals was proposed by Tomaschek (1880, 823-24). He recognized the role of the numeral '2' in '4' and '8',92 and the vigesimal character of the higher numerals '30', '40', '50', '60', '70', '80', '90'. Also remarkable are his external comparisons, *Cu '10' (extracted from '9') with Yeniseian (PY *tu?-q; Starostin 1995, 289) and Ti-

92 Let us mention that an even stricter binary system appears in Haida, one of the Na-Dene languages: see Blazek (1999: 327).

betan bcu (PST *[jh]Vj; CVST IV, 144-45), and toorumo '10' with (Sino-Tibetan) Khaling tadham, tar am '10'. It seems very probable that a Burushic substratum is responsible for the existence of vigesimal systems in the Nuristani and Dardic and Pamir languages (Lorimer 1937: 83), rarely also in Pasto (dwah-silah '40', dre-silah '60', tsalor-silah '80'), Baluci (do-gist '20', si-gist '60', cyar-gist '80'), and Asiatic Romani (turrum-wist '60', turrum-wist-das '70')93 — see Tomaschek (1880: 826) — much as the vigesimal systems in Ossetic and Georgian are likely due to Caucasian substratum, and those of Romance and Celtic due to the Basque/Aquitanian substratum.94 Now as to C's proposed material correspondences between Bur and IE numerals: the first, comparing PIE *Hioi-no-s 'one' with Bur hen / hin (class I, II) ~ han (class II, IV) ~ hek / hik (counting form) 'one' is almost plausible, except that the form *Hioi-no-s is characteristic of western IE (Italic, Celtic, Germanic, Balto-Slavic), while forms with different suffixes *H\oi-ko-s and *Hioi-uo-s gave rise to the Indic and Iranian words for 'one' shown above. The late Sergei Starostin derived Bur *he- 'one' from Proto-DC *HVcd / *cHd 'one', a root that produced the word for 'one' in all DC languages (except Basque):95 Caucasian: PNC *cHd (Chechen cha?, Khwarshi has, Ubykh za, etc.), Yeniseian: PY *xu-sa, and Sino-Tibetan: PST *?it (Old Chinese *?it, Burmese ac, etc.). The phonetic development in Bur is regular, as also seen in the word for 'fox', e.g.:96

• Bur *he- 'one' : Chechen cha? 'one' < PNC *cHd

• Bur *hal 'fox'97 : Chechen chogal 'fox' < Proto-Nakh *chokal < PNC chwolV-kV

For Bur *alto 'two' C suggests comparison with IE *Hal- 'other' + ordinal suffix *-to-, in spite of the fact that this is not an ordinal but a cardinal number, and that the "suffix" -to- appears nowhere else in the Bur numerals. As we have shown above, Bur /lt/ is a distinctive cluster that can be traced back to PDC lateral affricates, and thus we prefer the comparison of Bur *alto '2' (and *w-alt- '4', *altamb- '8', and *altar '20') with PDC *=VnLe, whose other reflexes include PWC *p(:)dXd '4', PEC *bunle '8' (Chechen barh, Avar mzX'-go, Lezgi muzu-d, etc.), Basque *lau '4', and PST *(p-)lij '4' (Tibetan bzi, Burmese lijh, Kaling 'bhal, etc.). Note that only Bur retains this stem for 2, 22, 23, 2^10, while Basque, West Caucasian, and Sino-Tibetan use it only for 22, and East Caucasian only for 23, and that several of the languages cited have a labial prefix before the stem:

Table 18

Bur *w-alt- 22

PEC *bunle 23

PWC *p(:)aXa 22

PST *(p-)lij 22

93 Berger (1959) detailed Burushaski influences on Romani. E. Bashir (pc.) adds that the vigesimal system is also found in Panjabi.

94 Blazek (1999: 333-334) discusses in more detail the vigesimal systems in various IE languages and their probable origins from DC substrata.

95 S. A. Starostin suggested derivation of Basque bat 'one' from the PDC root *=itV 'to cut, divide, break', with a fossilized class prefix as in Avar b-uta 'part', Lak b-atu-l- 'separate', and Dargwa Chirag b-ita-l 'part' (NCED 660-661).

96 According to Starostin (CSCP 60-67) the PDC initial sibilant-laryngeal clusters *cH-, *jH-, *sH- regularly yield Bur *h-.

97 There is a certain resemblance to Indo-Aryan words for 'jackal, fox': Skt. srgala- > Hindi siyal, siyar, sal, 'jackal', Oriya siyala, siala, etc. (CDIAL 729), though Berger (1998 III: 186) makes no reference to this as a source of Bur hal.

Next, C attempts to derive Bur altambo '8' from PIE *oktö(u) '8', "with a change of ak > al under the influence of the Bur numerals for 2 and 4" (p. 75). In view of the holistic relationship of the Bur words for 2, 22, and 23, as shown above, it seems highly unlikely to us that all the other IE lower numerals would be discarded and only '8' retained, with this odd change.

Finally C (p. 75) tries to connect Bur hunti '9' with PIE *Hineun '9', "with dissimilation," presumably to eliminate the first nasal. However, the non-counting forms contain sibilant affricates: (H, N) hunco, (Y) huco, and we saw (above) Tomaschek's hypothesis of 'one' (hun-) away from 'ten' (-co, -co). Besides Yeniseian *tu1- '10'and Sino-Tibetan *[3h]Vj '10', Starostin and Nikolayev (NCED 245) have posited PNC *2encE '10' (Andi hoco-go, Lezgi cu-d, Abkhaz za-ba, etc.), and some have suggested that a cognate element *-ci is found in the Basque numerals *sor-ci '8' and *bedera-ci '9' (thus 10 — 2, 10 — 1, respectively), though this latter hypothesis has been criticized by Trask (1995: 64-65). One of the authors (Bl 328) has suggested another possibility: *hun- '1' + (Y) -cu- 'take away', i.e. '(10) take away 1'.98 Berger (Beiträge 79) derives hunco and huco < *huncio < *hun-tr-io ('1' + '10' + plural, i.e. '10 — 1', like Finnish yhdeksan) Finally, Starostin (CSCG 255, CSCP 81) compares Bur *hunco '9' with PEC *1ilcwi '9' (Andi hoco-, Khwarshi üci-n, Lak urc, etc.), though this does not account very well for the Bur counting form hunti.

In spite of C's ingenious (though, we think, erroneous) attempts, it is apparent that there is nothing in common between the Bur and IE numeral systems. The kinship of the Bur numeral system with those of DC languages is most clearly seen in the words for 2, 22, and 23.

Lexicon

If Burushaski is an IE language, one would expect it to have something in common with the inherited IE lexicon. We have already seen above that large segments of Bur basic vocabulary, including pronouns and numerals, have cognates in Dene-Caucasian languages. Here we compare some of the core vocabulary in both languages according to basic semantic fields.

Kinship terms

PIE *pa2ter- (*pHter-) 'father' / *mater (*meHter-) 'mother' : Bur *-uy 'father', *-mi 'mother'. The Bur word for 'mother', like the initial element of PIE *mater, is a variation of the universal stem *mA, cf. Basque *eme 'female', *ama 'mother', Yeniseian *?ama 'mother', etc.

Bur *-uy 'father'99 is clearly unrelated to PIE *pHter-, or to anything else in IE, for that matter. In any case the Bur words lack the characteristic IE structure ending in *-ter.100

98 "Interestingly, we have a similar situation in Vedic and later OI, where 19 = 20 minus 1. The minus is expressed by una 'gap': thus: eka-una-vimsati [> ekonavimsati] 20 - 1 = 19, [likewise] for 29, etc. Again areal influence? The Iranians of course do not do it." (M. Witzel, pc.).

99 A highly speculative hypothesis for the origin of Bur *-uy 'father' < *'foster-father' could involve the PDC verb *=i2wVl- 'to eat' (PNC *=i2wV 'to feed on, to eat; to bite', PY *?i?r- 'to eat', Basque *alha- 'to graze, feed': CSCG 111). See above for the proposed lateral origin of Bur /y/. A semantic analogy may be found in Old Irish al-tru 'foster-father' < al- to feed, nourish' < PIE *al- 'to raise, to feed'.

100 Elsewhere one of the authors has tried to demonstrate that the IE kinship terms in *-ter should be segmented as *p-H2-ter- 'father', *m-eH2-ter- 'mother', *bhr-eH2-ter- 'brother', *dhug-H2-ter- 'daughter', *gem-H2-ter 'son-in-law'. The suffixal complex *-(e)H2-ter- corresponds to Hittite -adar / Luwian -attar, which bear a function similar to English -hood or German -heit. Hence these IE kinship names probably reflect an abstract meaning which can be expressed as 'fatherhood, motherhood, brotherhood, daughterhood', etc. (Blazek 2001, 24-33).

PIE *bhra-t-er- (*bhreH2-ter-) 'brother' / *swes-er- 'sister' : Bur has instead one stem *'-cu that serves as both 'brother of male' and 'sister of female', and two others, *-hulVs 'brother of female, and *-yast 'sister of male'.101 All of these Bur words are bound morphemes — they can only occur with a possessive prefix — and all of them have parallels in DC languages.

Bur *'-cu closely resembles the Caucasian stem *=ici that serves as 'brother' and 'sister', often with changing class prefixes (e.g., Agul cu 'brother', ci 'sister', Chechen wa-sa 'brother', ja-sa 'sister', Dargwa u-zi 'brother', ru-zi 'sister', etc.); cf. Basque *an-his-ba 'sister (of a woman)'; PST *cdjH 'elder sister or brother'; Yeniseian *b-[i](?)s 'brother, sister' (CSCG 112).

Bur *-hulVs 'brother (of female), husband's brother' resembles PEC *xal?V / *2VxalV, a word root that gives rise to Lak ahal-cu 'bridegroom's kinsman' and ahal-s:ar 'bride's kinsman', along with cognates that mean 'guest' (probably a semantic development from *'wedding guest' < *'kinsman invited to a wedding'): Dargwa Akusha ahal, Tabasaranxalu-zv, etc. (NCED 1067).102

Bur *-yast 'sister of male, wife's sister' can be compared with PEC *cHVdV 'woman' (Chechen zuda 'woman', Dargwa Chirag cade 'female', Hunzib cutula 'bride', etc.), Urartian asti 'woman, wife, bride-groom', PY *cVt- 'husband', Basque *(ema-)ste 'married woman, wife' (CSCG 26).

PIE *su-nu-, *su-yo- (*suH-nu-, *suH-yo-) 'son', *dhug(h)a-t-er- (*dhug-H-ter-) 'daughter' : Bur has one stem, *-i, for both 'son' and 'daughter'. Starostin (CSCG 156) connected this with PST *ye(j) 'child, young', with the regular Bur loss of initial 103 Cf. also Basque *nini 'child, doll'.

Bur also has the word *'-s (Yasin -is, Hunza, Nager '-sk) 'human child, animal's young', probably cognate with Caucasian *=tswE 'son, daughter' (Avar w-as 'son', j-as 'daughter', Kabardian sa-wa 'son', etc.); Basque *sV (in *se-me 'son', *o-sa-ba 'uncle', *alha-ba-so 'granddaughter', *a-sa-ba 'ancestor', etc.); PST *su 'grandchild' (CSCG 113).

IE *suH-nu-, *suH-yo- 'son' are derivatives of the verb *seuH- 'to give birth' (IEW 913-14; Rix et al. 2001: 538). Probably related are Kartvelian *sew-/*sw- 'to give birth': Georgian svili "son" (Klimov 1998, 248, 251) III Afro-Asiatic: Cushitic: (East) Somali was, Konso os 'to have sexual intercourse' I I Omotic: Shinasha, Mocha suw-, Kafa sii-, Anfillo suy- 'to give birth' (Lamberti 1993: 384) III Uralic: Mari sdwa 'to give birth' (Illic-Svityc 1967: 361: IE + Kartv. + Mari). There is likely a remote ('Borean') connection between PDC *=iswE and the other words in this paragraph, but the morphological features are entirely different: IE stem + suffix vs. Bur (and DC) prefix + stem.

In sum, there is no resemblance whatsoever, whether in overall kinship structure or lexemes, between Bur and IE kinship terms, apart from some possibly very remote ('Borean') cognates (PIE *su- ~ Bur *'-s, PIE *ma-t-er- ~ Bur *'-mi).

Body part words

PIE *kerd- 'heart' : Bur *'-s 'heart, mind'. The Bur word has been compared with Caucasian: PNC *2dmsa 'sky, cloud; soul, breath; god' (Akhwakh as:i 'breath', Ubykh p-sa 'soul, spirit', etc.), Basque *haise 'wind', etc. (CSCG 263);104 another possibility is comparison with

101 These words have extended meanings in the Burusho kinship system: *'-cu also serves as 'husband of a man's sister', *-hulVs as 'husband's brother', and *-jast as 'wife's sister'. The typology of the Bur sibling terms is similar to Basque: *anaie 'brother of male'/ *ne-ba 'brother of female'; *an-his-ba 'sister of female' / *are-ba 'sister of male'.

102 In these words /a/ denotes a pharyngealized vowel, and a voiceless pharyngealized uvular fricative, otherwise written (more awkwardly) with the palocka as /aI/ and /xI/, respectively.

103 Seen also in Bur *a- '1st person singular pronominal prefix' ~ PST *ya- 'I, we', PEC *m 'I', Basque *ni 'I', etc. (CSCG 156, CSCP 48).

104 For semantics, cf. Rumanian inima 'heart, soul, mind,' etc. < Latin anima 'wind, air, breath, spirit, mind', etc.

Basque *bi-si 'life; alive', PNC *siHwV 'breath, breathe': Chechen sa 'soul', oblique base si-na-, etc.) (CSCG 188).

The IE word is, we think, cognate with Kartvelian *mkerd- 'chest, breast' (Klimov 1998, 123; Illic-Svityc 1971, #200: IE+Kartv.) and, in Afro-Asiatic: Chadic: Hausa kirji, pl. kiraaza 'chest', Gwandara gariji id. (Skinner 1996).

PIE *okw- (*H3ekw-) 'eye' : Bur *'-l-ci / *il- (the latter in compounds). The Bur word is clearly comparable with Caucasian: PNC *2wil2i 'eye' (cf. especially Dargwa *huli, Tabasaran, Agul, Rutul ul) and Yeniseian: PY *de-s (Ket des, Kott tis, Pump. dat, where *d- is a regular initial reflex of PDC *l-: CSCG 266, CSCP 68).

The IE word *H3ekw- has, we think, external cognates in Altaic: PA *uk'u 'to understand, look into' (Old Turkic uq- 'to understand', Old Japanese uka-kap- 'to look into, inquire', etc.); cf. also Semitic: Ugaritic 'aq 'eyeball'; Hebrew 'aqa id. (Koehler & Baumgartner 2001 I: 873); Geez 'oqa 'to know, understand, observe', Amharic awwaqa 'to know', Harari aqa id. (Leslau 1987: 78-79); Cushitic: (Central) *-aq 'to know' > Kemant ax-, Kunfal ah-, Awngi -aq- id.; (East) Somali -aq id. (Appleyard 2006, 89-90).

PIE *o(w)as- 'mouth'105 : Bur *qhat. The latter is comparable with Caucasian: PEC *qwiti 'Adam's apple, uvula' > Lak qwit ~ qit ~ qut 'Adam's apple, beak',106 Kryz xulut 'larynx' (< *xut-ul), etc. (CSCG 172).107

PIE *kara-, *keras- 'head'108 : Bur *yat-is. Cf. Caucasian: PEC *lwomdV 'brain, head': Avar fada- 'head', Tsez, Hinukh ata 'brain', Archi ont 'head (of woman or animal)',109 etc. (CSCG 98).110

PIE *nas-, *nas- 'nose' : Bur *mus 'nose', *-mus 'snot'. Cf. Caucasian: PNC *mlace 'edge' (Ingush mfiz-arg 'snout', etc.: NCED 813); or PEC *mHarcwV 'pus; mucus, snot' (Chechen mars 'snot', Tsakhur mas 'pus', etc.: CSCG 144); Basque *mosu 'nose, face, kiss, point, beak'.

PIE *ost(h)- 'bone'111 : Bur *-lten 'bone', *-ltan-c 'leg'. Cf. Caucasian: PEC *XwVn?V 'groin; part of leg': Avar X:an 'groin', Archi X:on-t'ol 'fingernail', Kryz kin 'ankle', etc.; PST *ldq 'shin, ankle' (CSCG 140).

PIE *ped- 'foot' : Bur *-hut- 'foot'. Cf. Caucasian: Avar het / hete 'foot', Dargwa Kaitag tah 'foot, hoof', etc. < PEC *hitwV / *twifiV; PST *ttH ~ *dtH 'heel, ankle' (CSCG 207).

PIE *yekw- (*(H)iekwr(t)) 'liver' : Bur *'-ken 'liver'. Cf. Caucasian: PEC *kunHV> Chamalal ku 'liver', Bezhta, Hunzib koma 'kidney', etc. (NCED 728); cf. PST *kjVnH 'kidney' (CVST V: 58, no. 214).

105 According to D. Q. Adams (EIEC, p. 387), the form *o(w)as- 'mouth' should be reinterpreted as two distinct stems: (i) *Hi/4oHi(e)s-, gen. *HiCeHsds; (ii) *Hxoust-a.

106 /i/, /u/ represent pharyngealized vowels, also (awkwardly) written ii, ul, where I represents the palocka in the Cyrillic orthography of Caucasian languages.

107 Alternatively, cf. PNC *GwetV ~*GetwV 'crop, craw; beak, Adam's apple' > Lak. q:iti 'uvula', etc. (CSCG 172).

108 The IE word for 'head' should be reconstructed as *krreH2, gen. *krH2os, singulative *korH2sr, collective *kerH2or (Adams, EIEC 260). The meaning 'brain' developed in Latin cerebrum and Old High German hirni.

109 /o/ represents a pharyngealized vowel = NCED /oI/ (cf. note to 'mouth', etc.)

110 The correspondence of Bur *y- = *j— PNC *fi- is recurrent. Cf. Bur *yaltar 'leafy branches', etc. ~ PEC *hal$VlV 'branch, pod' (above in the discussion of Bur -lt-).

111 The IE word 'bone' should be reconstructed as *H3est(H)-.

C (p. 38) attempts to connect the Bur word with PIE *(H)iekwr(t) (a heteroclitic -r/-n stem), ignoring the root syllable *(H)iekw- = *yekw-,112 while another originally heteroclitic word, PIE *wed- 'water', is compared with Bur budoo 'rinsing water', which has no trace of either heteroclitic suffix -r or -n. (Cf. instead OI *budyati 'sinks', Marathi budbud 'sound of bubbling', etc.: CDIAL 9272.)

PIE *(o)nAbh- 'navel'113 : Bur *-su[m] 'umbilical cord, navel'114 ~ Cf. Caucasian: Chamalal suj, Lak cun, Dargwa zu, Khinalug cum 'navel', etc. < PEC *3on?u (CSCG 249).

Basic verbal roots

PIE *klewe- 'to hear' : Bur *-yal- 'to hear' ~ cf. Caucasian: PNC *=eAu 'to hear': Andi anXi-'to hear', Budukh ix- id., etc. (NCED 411, CSCG 46)

PIE *ed- 'to eat' : Bur *si (with class I, II, III singular object) / *su (with class I, II, III plural object) / *si 'to eat' (with class IV object) ~ cf. Yeniseian: PY *si- 'to eat' ~ PST *jha id. ~ Caucasian: Tsez, Khwarshi =ac- 'to eat', Tindi c:a- 'to drink', etc. < PEC *=VcV ~ Basque *ausi-ki 'to bite' (NCED 1017, CSCG 209)

PIE *do(w)- 'to give'115 : Bur (1) *-u- 'to give' (only with class I, II, III object), (2) *-chi- 'to give' (only with class IV singular object); (3) *-gun- 'to give' (only with class IV plural object).

The three class-determined Bur verb stems have distinct DC origins:

(1) cf. PNC *mVxwV; PST *rjaH 'to give, borrow, rent' (CSCG 156); 116

(2) cf. Caucasian: Chamalal ic- 'to sell, give', Bezhta =is- 'to sell', Khinalug ce=kwi 'to sell', etc. < PEC *=icV (NCED 626);

(3) ? cf. PEC *HVqVn- 'to take, snatch' (NCED 615); PST *gon 'to collect' (CVST V: no. 56); Basque *(e)-ken- 'to take away', etc.117

Here the verb used in Bur is determined by the class of the object. (Cf. the preceding example, 'to eat'.) This is a totally un-Indo-European feature, but it appears to be a deep-seated trait of Dene-Caucasian, with manifestations at least in Basque and Na-Dene.118

Other basic words

PIE *(e)nomen- 'name'119 : Bur *yek 'name, reputation': (Y) -yek, pl. -yekir, -yekicir, (H, N) -ik, pl. -ikicir . Cf. Yeniseian: PY *?iG > Ket i 'name', pl. e?r, Kott ix, ix, pl. ikr / ekr / eakr. This is

112 Lorimer (1935) considered Burushaski -akin, pl. -akimi], -aki-ni] 'liver' a borrowing from Indo-Iranian: OI yakrt, gen. yakna 'liver', Pashto yma, Yidgha yegan id. etc. (IEW 504; Bailey 1979: 108).

113 The IE word 'navel' should be reconstructed as *H3nobh- (Adams, EIEC 391).

114 Underlying *m found in the plural form -suimuc.

115 In LIV 105-07 reconstructed as *deH3- & *deHu.

116 According to Starostin < PDC *]VxwV 'to give, borrow', with regular loss of initial *] in Bur (CSCP 48).

117 Assuming the common semantic relationship of 'give' and 'take' (as in PIE *ghab(h)-, etc.).

118 This trait is highly developed in Na-Dene: Athapaskan: e.g. Navajo -tj 'handle animate singular object', -kq 'handle a rigid container with contents', -zddz 'handle a set of parallel long rigid objects' (each representing a different class). And at the far western extreme we find remnants of similar tendencies in Basque: the dialects have different words to express the concept 'dry', e.g. Zuberoan agor pertains to sources and streams of water, utsal to aliments and terrain, eihar to the human body, fauna and flora, and idor to dryness in general.

119 The IE etymon 'name' has been reconstructed as *H1ndmn (Polome & Mallory, EIEC 390).

one of the remarkable parallels between Bur and Yeniseian (cf. Toporov 1971), extending even to the inanimate plural endings with velar nasals.120

We can see from these examples that Bur really shares almost no basic vocabulary with IE.

Conclusions

It is impossible to disprove relationship. We agree with Casule that there may be some kind of very deep-level relationship between Burushaski and IE. However, we propose, and we believe we have shown, that Burushaski is much closer genetically to the Dene-Caucasian languages than it is to Indo-European.

Much of the similarity between Bur and IE can be attributed to a long period of symbiosis and language contact between Bur and its Indo-Iranian neighbors. There is evidence that early Indo-Aryan was influenced by Bur (or perhaps a wider-ranging Burushic family) as its speakers entered the Indian subcontinent by way of the Hindukush and Pamir regions (see, e.g., Lorimer 1937, Tikkanen 1988, Witzel 1999). We noted above such features as the vigesimal numeral system (discussed above) in Nuristani, Dardic, Pamir, Pasto, Baluci, and Asiatic Romani. There are also lexical borrowings from Bur that have penetrated into the basic lexicon, e.g. in Sina: birdi 'earth', phurgu 'feather', chis 'mountain', tam doiki 'to swim'; and in Khowar: tip 'full', phur 'hair', buk 'neck', etc. (Kogan 2005: 173). These parallels reflect only areal, not genetic relations, and so they are the results of secondary convergence. The areal parallels indicate the existence of a much wider expanse of the Burushic stratum in the past, but there are no direct Burushaski-Indo-Iranian/Indo-European genetic links, only some very old elements that represent archaic residue from a remote ancestor (Borean) common to the ancestor of Indo-European (Nostratic or Eurasiatic) and the ancestor of Burushaski (Dene-Caucasian).121

Postscript

Since this article was originally written (around mid-2007) there have been some new developments in the Dene-Caucasian hypothesis. A consensus has been developing that the eastern members, Sino-Tibetan and Na-Dene, probably result from an early split of the DC proto-language, leaving the western branches (Basque, Caucasian, Burushaski, and Yeniseian) to a period of common development in which some grammatical and lexical features (e.g., suppletive pronominal paradigms [see above]; words such as western *ZwilZi 'eye' [see above] vs. eastern *wemqV 'eye'122) crystalized.

In a recent lexicostatistical study by George Starostin (p.c.), using the 50 most generally stable items on the Swadesh 100-word list (G. Starostin 2010b), a tentative subgrouping has emerged in which the eastern branches (Sino-Tibetan and Na-Dene) are indeed opposed to the western group (Basque, Caucasian, Burushaski, and Yeniseian), thus confirming the old „Sino-

120 Besides 'name', Bur and Yeniseian share several important basic lexical isoglosses, e.g. 'eat' (B *si / *si / *su ~ Y *sï-), 'egg' (B *tir/- ~ Y *je?r/ / *jo?r), 'eye' (B *-l-ci ~ Y *de-s-), 'hand' (B *-rer ~ Y *for/), 'leaf' (B *ltap ~ Y *jâpe), 'root' (B *cherés ~ Y *cïj-), etc., as well as the pronominal and numeral words discussed above.

121 For example, the case of Bur *'-s 'child, young' ~ PIE *suH-(-nu-,-yo-) 'son', cited above.

122 PST *myVk (Old Chinese @ *muk, Tibetan mig, Lepcha mik, a-mik, etc. ,'eye'); Tlingit wàc, Athabaskan *-ne-wëG-e? 'eye'. See CSCG 216: this word was preserved with other semantic developments in the western DC languages.

Dene" idea of Edward Sapir (Bengtson 1994). Within the western group G. Starostin finds a split between a Basque-Caucasian branch on the one hand and a Burusho-Yeniseian branch on the other (Bengtson 2010a, 2010b; G. Starostin 2010a).

As to the recently developed „Dene-Yeniseian"123 idea initiated by Ruhlen (1998b) and continued by Vajda (e.g., 2008, 2009, 2010), it now appears that the Yeniseian languages have much more in common with Burushaski than (directly) with the Na-Dene languages. In other words, there is indeed a "relationship" between Yeniseian and Na-Dene, in the sense that both ultimately belong to different branches of the Dene-Caucasian macrofamily, but in our view they do not by themselves form a valid taxon.124 Likewise, Na-Dene seems to form a taxon with Sino-Tibetan and is thus closer to the latter than to Yeniseian.

Abbreviations of languages and dialects

Bur Burushaski

DC Dene-Caucasian (Sino-Caucasian)

H Hunza (Burushaski)

JSp Judaeo-Spanish

Lat Latin

MSp Middle Spanish

N Nager, Nagar (Burushaski)

OSp Old Spanish

PDC Proto-Dene-Caucasian (Proto-Sino-Caucasian)

PEC Proto-East Caucasian

PNC Proto-(North) Caucasian

PST Proto-Sino-Tibetan

PWC Proto-West Caucasian

PY Proto-Yeniseian

Tib Tibetan (Classical)

Y Yasin (Burushaski) = Werchikwar

Abbreviations of sources cited

Beiträge Berger (2008)

Bl Blazek (1999)

CDIAL Comparative Dictionary of Indo-Aryan Languages (Turner 1966)

CLI Compendium Linguarum Iranicarum (Schmitt 1989)

CSCG Comparative Sino-Caucasian Glossary (Starostin 2005a)

CSCP Comparative Sino-Caucasian Phonology (Starostin 2005b)

CVST A Comparative Vocabulary of Five Sino-Tibetan Languages (Peiros & Starostin 1996)

EIEC Encyclopedia of Indo-European Culture (Mallory & Adams 1997)

H Hayward (1871)

Hd Hodgson (1857)

IEW Pokorny (1959)

LDC Lexica Dene-Caucasica (Blazek & Bengtson 1995)

NCED North Caucasian Etymological Dictionary (Nikolaev & Starostin 1994)

SSEJ Sravnitel'nyj slovar enisejskix jazykov (Starostin 1995)

123 See http://www.uaf.edu/anlc/dy/.

124 For an Indo-European analogy, there is a "relationship" between, say, North Germanic and Western Iranian, in the sense that both are subgroups of IE, but they do not form any kind of taxon by themselves.

ToB Tower of Babel databases: http://starling.rinet.ru/main.html W Werner (2002)

X Xelimskij (1982)

References

Anderson, Gregory D. S. Ms. Burushaski.

Appleyard, David L. 2006. A Comparative Dictionary of the Agaw Languages. Köln: Köppe.

Bailey, H. 1979. Dictionary of Khotan-Saka. Cambridge University Press.

Bashir, Elena. 2000. A Thematic Survey of Burushaski Research. History of Language 6/1: 1-14.

Bengtson, John D.1993. The Macro-Caucasic hypothesis. Dhumbadji! 1/2: 3-6.

Bengtson, John D. 1994. Edward Sapir and the 'Sino-Dene' Hypothesis. Anthropological Science 102/3: 207-230.

Bengtson, John D. 1997a. Ein Vergleich von Buruschaski und Nordkaukasisch. Georgica 20: 88-94.

Bengtson, John D. 1997b. Basque and the other Dene-Caucasic languages. In Alan K. Melby (ed.) The Twenty-Third LACUS Forum, 63-74. Chapel Hill, NC: LACUS.

Bengtson, John D. 1998. Dene-Caucasian 'Navel': some proposed etymologies. Dhumbadji! 4.1: 86-90.

Bengtson, John D. 2000. Review of Casule 1998. History of Language 6.1: 22-26.

Bengtson, John D. 2001a. Genetic and Cultural Linguistic Links between Burushaski and the Caucasian Languages and Basque. (Paper presented at the 3rd Harvard Round Table on Ethnogenesis of South and Central Asia, Harvard University, May 13, 2001.)

Bengtson, John D. 2001b. Review of H. Berger, Die Burushaski-Sprache von Hunza und Nager. In Mother Tongue 6: 184-187.

Bengtson, John D. 2002. The Dene-Caucasian noun prefix *s-. In: F. Cavoto (ed.) The Linguist's Linguist: A Collection of Papers in Honour of Alexis Manaster Ramer, 53-57. Munich: LINCOM Europa.

Bengtson, John D. 2008a. Materials for a Comparative Grammar of the Dene-Caucasian (Sino-Caucasian) Languages. In: Aspects of Comparative Linguistics, v. 3, pp. 45-118. Moscow: RSUH Publishers. Online at: http://starling.rinet.ru/Texts/books.php?lan=en

Bengtson, John D. 2008b. The languages of northern Eurasia: Inference to the best explanation. In In Hot Pursuit of Language in Prehistory: Essays in the four fields of anthropology in honor of Harold Crane Fleming, ed. by J. D. Bengtson, pp. 241-262. Amsterdam/Philadelphia: John Benjamins.

Bengtson, John D. 2008c. Linguistic Fossils: Studies in Historical Linguistics and Paleolinguistics. Calgary: Theophania.

BENGTSON, John D. 2010a. "Dene-Yeniseian" and the Rest of Dene-Caucasian: Part 3: The Burusho-Yeniseian (Karasuk) Hypothesis; Part 4: Burusho-Dene. In Working Papers in Athabaskan Languages (Alaska Native Language Center Working Papers No. 8), ed. by Siri Tuttle & Justin Spence, pp. 1-18. Fairbanks: Alaska Native Language Center.

Bengtson, John D. 2010b. Burushaski, Yeniseian, and the Karasuk Culture. Paper presented at the 14th Harvard Round Table on the Ethnogenesis of South and Central Asia. Harvard University, Oct. 4-5, 2010.

Berger, Hermann. 1956. Mittelmeerische Kulturpflanzennamen aus dem Burusaski. Münchener Studien zur Sprachwissenschaft 9: 4-33.

Berger, Hermann. 1959. Die Burusaski-Lehnwörter in der Zigeunersprache. Indo-Iranian Journal 3.1: 17-43.

Berger, Hermann. 1974. Das Yasin-Burushaski (Werchikwar). Neuindische Studien 3. Wiesbaden: Otto Harrassowitz.

Berger, Hermann. 1998. Die Burushaski-Sprache von Hunza und Nager. Teil I. Grammatik. Teil II. Texte. Teil III. Wörterbuch. Burushaski-Deutsch; Deutsch-Burushaski. Neuindische Studien 13. Wiesbaden: Otto Haras-sowitz.

Berger, Hermann. 2008. Beiträge zur historischen Laut- und Formenlehre des Burushaski. Neuindische Studien 15. Wiesbaden: Otto Harassowitz.

BLAZEK, Vaclav. 1995. Yenisseian Etymology of the North Kazaxstan Toponym Selety? In Central Asiatic Journal 39: 9-10.

Blazek, Vaclav. 1999. Numerals. Comparative-etymological analyses and their implications. Brno: Masarykova univerzita.

Blazek, Vaclav. 2001. Indo-European kinship terms in *-ter. In: O. Sefcik & B. Vykypel (eds.) Grammaticus. Studia linguistica A. Erharto. Brno: Masarykova univerzita, 24-33.

Blazek, Václav. 2003. Lexica Nostratica Addenda. Mother Tongue 8: 11-22.

Blazek, Václav. 2008. Chukcho-Kamchatkan and Uralic: Lexical Evidence of Their Genetic Relationship. Mother Tongue 13: 209-225.

Blazek, V., and J. D. Bengtson. 1995. Lexica Dene-Caucasica. Central Asiatic Journal 39/1: 11-50, 39(2): 161-164.

Bouda, Karl. 1949. Lakkische Studien. Heidelberg: Carl Winter.

Bradley, Travis G. Metathesis in Judeo-Spanish Consonant Clusters. In: N. Sagarra and A. J. Toribio (eds.) Selected Proceedings of the 9th Hispanic Linguistics Symposium, 79-90. Somerville, Mass.: Cascadilla Proceedings Project.

Casule, Ilija. 1998. Basic Burushaski Etymologies: The Indo-European and Paleo-Balkanic Affinities of Burushaski. LIN-COM Etymological Studies 01. Munich: LINCOM Europa.

Casule, Ilija. 2003. Evidence for the Indo-European Laryngeals in Burushaski and Its Genetic Affiliation with Indo-European. Journal of Indo-European Studies 31.1/2: 21-86.

Catford, J. C. 1977. Mountain of Tongues: The Languages of the Caucasus. Annual Review of Anthropology 6: 283314.

Corominas, Joan. 2001. Breve diccionario etimológico de la lengua castellana, Madrid: Gredos.

van Driem, George. 2001. Languages of Himalayas. Leiden-Boston-Brill.

Dul'zon, A. P. 1959. Ketskie toponimy Zapadnoj Sibiri. Ucenye zapiski Tomskogo gosudarstvennogo pedagogiceskogo instituta 18, 91-111.

Dul'zon, A. P. 1961. Slovarnye materialy XVIII v. po ketskim narecijam. Ucenye zapiski Tomskogo gosudarstvennogo pedagogiceskogo instituta, Tom XIX, Vypusk 2: Lingvisticeskie nauki, 152-189.

Dul'zon, A. P. 1963. Étniceskij sostav drevnego naselenija Zapadnoj Sibiri po dannym toponimii. In: Trudy 25. Mezdunarodnogo kongressa vostokovedov, Tom 3. Moskva: Nauka, 289-295.

ÉDELMAN, Dzoj I. 1978. Dardskie jazyki. In: M. S. Andronov, N. A. Dvorjakov, A. M. Djakov (eds.) Jazyki Azii i Afriki II: Indoevropejskie jazyki (Iranskie jazyki & Dardskie jazyki); Dravidijskie jazyki, 254-315. Moskva: Nauka.

Efimov, V. A., ÉDELMAN, D. I. 1998. Novoiranskie jazyki: Vostocnaja gruppa. In: M. S. Andronov, N. A. Dvorjakov, A. M. Djakov (eds.) Jazyki Azii i Afriki II: Indoevropejskie jazyki (Iranskie jazyki & Dardskie jazyki); Dravidij-skie jazyki, 198-253. Moskva: Nauka.

Hamp, Eric P. 1997. Intensive and Perfective pro- in Latin. In: Studies in Honor of Jaan Puhvel, Part I: Ancient Languages and Philology, ed. by D. Disterheft, M. Huld, J. Greppin, 123-129. Washington: Institute for the Study of Man.

Hayward, G. W. 1871. Explorations in Gilgit and Yassin. Journal of the Royal Geographic Society 41: 1-46.

Hodgson, B. H. 1857. Comparative Vocabulary of the Languages of the broken Tribes of Népál. Journal of the Asiatic Society of Bengal 264: 317-427.

Hoffmann, Karl, and Bernhard Forssman. 1996. Avestische Laut- und Flexionlehre. Innsbruck: IBS 84.

Hualde, José Ignacio. 2001. Basque Phonology. London/New York: Routledge.

Illic-Svityc, V. M. 1967. Materialy k sravnitel'nomu slovarju nostraticeskix jazykov. Ètimologija 1965: 321-373.

Illic-Svityc, V. M. 1971. Opyt sravnenija nostraticeskix jazykov, I. Moskva: Nauka.

Klimov, G. A. 1998. Etymological Dictionary of the Kartvelian Languages. Berlin-New York: Mouton de Gruyter.

Klimov, G. A., and D. I. Édelman. 1972. K nazvanijam parnyx castej tela v jazyke burusaski. Etimologija 1972: 160162.

Koehler, Ludwig, and Walter Baumgartner (eds.). 2001. The Hebrew and Aramaic Lexicon of the Old Testament. Transl. by M.E.J. Richardson. Leiden-Boston-Koln: Brill.

Kogan, Anton I. 2005. Dardskie jazyki. Geneticeskaja xarakteristika. Moskva: Vostocnaja literatura.

KOSTJAKOV, M. M. 1979. Vremja rasxozdenija ketskogo i kottskogo jazykov po dannym leksikostatistiki. In: Voprosy stroja enisejskix jazykov, ed. E.I. Ubrjatova et al. Novosibirsk: Akademija nauk SSSR, Sibirskoe otdelenie, 118127.

Lamberti, M. 1993. Die Schinascha Sprache. Heidelberg: Winter.

Leslau, W. 1987. Comparative Lexicon of GeCez. Wiesbaden: Harrassowitz.

Lorimer, D. L. R.1935-1938. The Burushaski Language (3 vols.). Oslo: Instituttet for Sammenlignende Kulturfor-skning.

Lorimer, D. L. R. 1937. Burushaski and Its Alien Neighbours: Problems in Linguistic Contagion. Transactions of the Philological Society 1937: 63-98.

Mallory, J. P., and D. Q. Adams (eds.). 1997. Encyclopedia of Indo-European Culture. Chicago: Dearborn.

Morgenstierne, Georg. 1942. Notes on Burushaski Phonology. Norsk Tidsskrift for Sprogvidenskap 13: 61-95.

Nikolaev, S. L., and S. A. Starostin. 1994. North Caucasian Etymological Dictionary. Moscow: Asterisk.

Paxalina, T. N. 1975. Vaxanskij jazyk. Moskva: Nauka.

Payne, John. 1989. Pamir Languages. In: CLI 417-444.

Peiros, Ilija. 1988. Sinokavkazskaja teorija i burusaskij jazyk. Problemy izucenija sravnitelnogo-istoriceskogo jazyko-znanija i lingvisticeskoj istorii o vostoke i juge Azii: 214-227. Moscow: Institute of Far Eastern Studies, Academy of Sciences of the USSR.

Peiros, Ilia [Ilija], and Sergei A. Starostin. 1996. A Comparative Vocabulary of Five Sino-Tibetan Languages, I-VI. Melbourne: Department of Linguistics and Applied Linguistics.

Pokorny, Julius. 1959. Indogermanisches etymologisches Wörterbuch. Bern: Francke Verlag.

Ramstedt, Gustav J. 1907. Über den ursprung der sog. Jenisej-ostjaken. Journal de la Société Finno-Ougrienne 24/2, 1-6.

Rix, Helmut, et al. (eds.). 2001. Lexicon der indogermanischen Verben2. Wiesbaden: Reichert.

Ruhlen, Merritt. 1996. Une nouvelle famille de langues: le déné-caucasien. Pour la Science (Dossier, Octobre): 6873.

Ruhlen, Merritt. 1998a. Dene-Caucasian: A New Linguistic Family. In: K. Omoto and P. V. TOBIAS (eds.) The Origins and Past of Modern Humans — Towards Reconciliation, 231-246. Singapore: World Scientific.

Ruhlen, M. 1998b. The Origin of the Na-Dene. Proceedings of the National Academy of Sciences 95: 13994-13996. Available online at http://www.merrittruhlen.com/files/1998c.pdf

Ruhlen, Merritt. 2001. Il Dene-caucasico: una nuova famiglia linguistica. Pluriverso 2: 76-85.

Schmitt, Rüdiger (ed.). 1989. Compendium Linguarum Iranicarum. Wiesbaden: Ludwig Reichert.

Skinner, N. 1996. Hausa Comparative Dictionary. Köln: Köppe.

Sköld, Hannes. 1936. Materialien zu den iranischen Pamirsprachen. Lund: Gleerup.

Starostin, George. 2010a. Dene-Yeniseian and Dene-Caucasian: Pronouns and Other Thoughts. In Tuttle & Spence (2010), pp. 107-117.

Starostin, George. 2010b. Preliminary Lexicostatistics as a Basis for Language Classification: a New Approach. Journal of Language Relationship / Voprosy jazykovogo rodstva 3: 79-116.

Starostin, Sergei A. 1995. Sravnitel'nyj slovar enisejskix jazykov. In: S. A. Starostin (ed.) Ketskij sbornik — Lin-gvistika, 176-315. Moskva: Skola Jazyki Russkoj Kul'tury.

Starostin, Sergei A. 2006. Comments on the Basque-Dene-Caucasian Comparisons. Mother Tongue 2: 101-109.

Starostin, Sergei A. nd. Burushaski Etymological Database. EHL/Tower of Babel. http://starling.rinet.ru/

Starostin, Sergei A. 2005a. Comparative Sino-Caucasian Glossary. http://starling.rinet.ru/Texts/glossary.pdf

Starostin, Sergei A. 2005b. Comparative Sino-Caucasian Phonology. http://starling.rinet.ru/Texts/scc.pdf

Starostin, Sergei A. & George S. Starostin. 2005-2011. Yeniseian & Burushaski Etymological Databases. http://starling.rinet.ru

Tikkanen, Bertil. 1988. On Burushaski and Other Ancient Substrata in Northwestern South Asia. Studia Orientalia 64: 303-325.

Tikkanen, Bertil. 1995. Burushaski converbs in their areal context. In Converbs in Cross-Linguistic Perspective: Structure and Meaning of Adverbial Verb Forms - Adverbial Participles, Gerunds, ed. by M. Haspelmath & E. König, pp. 487-528. Berlin: Mouton de Gruyter.

Tomaschek, Wilhelm. 1880. Centralasiatischen Studien II: Die Pamir-Dialekte. Sitzungsberichte der philosophischhistorischen Classe der kaiserlichen Akademie der Wissenschaften 96: 735-900.

Toporov, V. N. 1971. Burushaski and Yeniseian Languages: Some Parallels. Travaux linguistiques de Prague 4: 107125.

TRASK, R. L. 1995. Basque and Dene-Caucasian: A Critique from the Basque Side. Mother Tongue (Journal) 1: 3-82.

Trask, R. L. 1997. The History of Basque. London/New York: Routledge.

TRASK, R. L. 2008. Etymological Dictionary of Basque. Ed. by Max W. Wheeler. University of Sussex. Linguist List Publications: http://www.sussex.ac.uk/linguistics/1-4-1-2.html

Turner, R. L. 1966. A Comparative Dictionary of Indo-Aryan Languages. London: Oxford University Press.

Tuttle, Siri, and Justin Spence (Ed.). 2010. Working Papers in Athabaskan Languages: Alaska Native Language Center Working Papers, No. 8. Fairbanks, Alaska: ANLC.

Vajda, Edward. 2008. The Siberian origins of Na-Dene languages. Paper presented at the Dene-Yeniseic Work Session, Fairbanks, Feb. 26-27, 2008, and at the Annual Meeting of the Alaskan Anthropological Society, Anchorage, Feb. 28, 2008.

Vajda, Edward. 2009. A Siberian link with Na-Dene languages. Ms. Draft Feb. 1, 2009, of a paper presented at the

Dene-Yeniseic Symposium, University of Alaska, Feb. 29, 2008. Vajda, Edward. 2010. A Siberian Link with Na-Dene Languages. In The Dene-Yeniseian Connection. Ed. by James Kari and B. A. POTTER. Anthropological Papers of the University of Alaska, New Series, no. 5. Anchorage: UAF Department of Anthropology /Alaska Native Language Center. Werner, Heinrich. 2002. Vergleichendes Wörterbuch der Jenissej-Sprachen, I—III. Wiesbaden: Harrassowitz. Witzel, Michael. 1999. Early Sources for South Asian Substrate Languages. Mother Tongue (Special Issue, October 1999): 1-70.

XELIMSKIJ, Evgenij A. 1982. Keto-Uralica. In: Ketskij sbornik, ed. by E.A. Alekseenko. Leningrad: Nauka, 238-251. XELIMSKIJ, Evgenij A. 1986. Arxivnye materialy XVIII veka po enisejskim jazykam. In: Paleo-aziatskie jazyki, ed. by

P.Ja. Skorik. Leningrad: Nauka, 179-213. Zaxarjin, B. A., and D. I. Zaxarjin. 1971. Jazyk kasmiri. Moskva: Nauka.

Статья посвящена относительно недавней гипотезе, выдвинутой И. Чашуле, согласно которой язык бурушаски, традиционно считавшийся изолятом, на самом деле входит в состав индоевропейской семьи. Авторы прибегают к сравнительному анализу, сопоставляя гипотезу Чашуле и те конкретные фонетические, морфологические и лексические аргументы, которые он приводит в ее поддержку, с соответствующими аргументами в пользу т. н. «дене-кавказской» гипотезы, которая утверждает, что бурушаски на правах отдельной ветви входит в обширную макросемью, включающую языки семьи на-дене, а также сино-тибетские, севернокавказские, баскский и енисейские языки.

Анализ данных показывает, что аргументы в пользу дене-кавказского происхождения бурушаски в количественном отношении значительно превышают аргументы в пользу индоевропейско-бурушаскской гипотезы. Связи бурушаски с индоевропейской семьей оказываются либо чересчур бессистемными (в области фонетических соответствий), либо спорадическими и явно недостаточными (в области морфологии), либо вообще практически отсутствуют (в области базисной лексики). Таким образом, все случаи схождений между индоевропейскими и бурушаскскими элементами следует объяснять либо как (а) следы недавних контактов между бурушаски и индоарийскими языками, либо как (б) случайные сходства, либо, в очень немногочисленных случаях, как (в) следы «сверхглубокого» родства, которые никоим образом не представляют собой эксклюзивных «индоевропейско-бурушаскских» связей.

Ключевые слова: индоевропеистика, язык бурушаски, макрокомпаративистика, дене-кавказская макросемья, языки-изоляты.