Ranko Matasovi University of Zagreb E-mail: [email protected] Homepage: www.ffzg.

hr/~rmatasov

SUBSTRATUM WORDS IN BALTO-SLAVIC Copenhagen, September 2012 1. The Methodological Problem PSl. *bra shallow water expanse (Croat. bra, Slov.dial. barj, Bulg. bra, Russ. CSl. bra, Ukr. bar wet area between two hills, Pol. barzyna, dial. barzwka, Cz. dial. bara, baina, Polabian poro mud, Vasmer I: 53, Sawski I: 191, ESSJa I: 153). Considered as Illyrian by Gluhak 125. Not connected to Gr. brboros mud, mire, filth (Vasmer I: 53), since the Slavic forms are acute. Gr. brboros can be plausibly connected to Arm. kork dirt (Beekes 226f.). Rather, PSl. *bra < PIE *bheh2- shine (LIV. s. v.), cf. Ved. bh ti shines, Gr. phan shine. The deverbal adjective from this root would have been *bheh2-ro-, and the collective > feminine noun *bheh2reh2 shiny stuff > PSl. *bra. For the semantics, cf. PIE *bholHto- white (Lith. bltas) > PSl. *blto mud, swamp (Croat. blto, Russ. bolto, Cz. blto, Pol. boto). 2. Criteria for identifying substratum vocabulary (Matasovi 2012): A. Loanwords usually belong to semantic fields that are especially prone to borrowing (e.g. technological and cultural terminology, names of plants and animals). B. The loanwords should be identifiable by their unusual phonological characteristics (i.e. unusual root structure, the presence of the rare PIE vowel *a, etc.); they should be isolated in the lexicon and not easily derivable from verbal roots. C. We should avoid root etymologies. That is, reducing an attested form to a PIE root does not amount to a sound etymology, unless we can also explain its word-formation and relate it to established Indo-European patterns. If this cannot be done persuasively, the word is suspect of being a borrowing from some substratum language. Schrijver (1997: 297): [the] identification of substratum words by the fact that they show phonological and morphonological alternations which are regular in the sense that they recur in more than one etymon according to a certain pattern but irregular in the sense that they cannot be explained, for some reason or other, on the basis of Indo-European phonology and morphology. 3. Indo-European substratum in Balto-Slavic? Holzers (1989) Temematic Example: PSl. *svobod free, *svoboda freedom < PIE *swe-poti, *swe-poteh2, with the first element *swe- from the root of the reflexive pronoun (cf., e.g., OCS svoj own), and the second element from PIE *poti- master, lord (Skr. pti-, Gr. psis husband, etc.). Alternative: PSl. *svobod from PIE *swo-bho- (cf. OCS svobstvo person, OPr. subs -self, own, Asg. subban, Goth. sibja kin, Latv. at-svabint set free). The suffix is *-oda (cf. e.g.

OCS agoda fruit vs. Lith. oga id., Vasmer III: 596, or *lgoda lightness > Croat. lagoda, cf. *lgk light > Russ. lgkij). Only 11 out of Holzers 45 Temematic roots have cognates in both Baltic and Slavic. Of these, all except the adjective meaning orphaned are root etymologies. Examples: s widower, cf. Av. Tm. *k'ey-ro-: PSl. *sir orphaned (Russ. sryj, Cz. sir), Lith. eiry sa- orphaned, which points to PIE *k'ey-. Smoczyski (628) and EWA II: 615 adduce also Ved. ay- orphan < *k'eyu-. Holzer (134f.) relates the BSl. words to Skr. hyate is left, ros orphaned, empty, khra widow, Lat. hrs heir. However, the hni- lack, Gr. khe PIE root can be safely reconstructed as *g'heh1-, cf. Skr. h- leave. Tm. *ponto-, *pont / *ponti-: PSl. *pto fetter (OCS [pl.] pta, Croat. pto, Russ. pto, Pol. pto, Derksen 417), Lith. pntis rope (for fettering horses), pnia id., OPr. panto fetter; Holzer derives this from PIE *bhendh- bind (= Eng. bind, Skr. bandh-, etc.), but a more a likely etymology starts with PIE *(s)penH- stretch, spin, weave (Arm. henowm weave, Lith. pin attach, OHG spannan, Gr. pnomai get tired, IEW 988). s, dial. te, Tm. *tel- calf: PSl. *tel calf (Croat. tle, Russ. telnok, Pol. ciel), Latv. tlen lis, te lias); derived by Holzer from PIE *dheh1-l- sucking (Gr. thlys Lith. dial. te feminine). Most likely, this Slavic word is of Turkic origin, cf. Tuvan tel calf, Kazakh tel, Yakut tl. Starostin (www.starling.ru) reconstructs Proto-Turkic *Tl a kid or calf sucking two milch-ewes or cows. Since there are no direct Turkic loanwords in Baltic, it is probable that the Baltic words for calf were borrowed from Slavic. 4. Non-IE substratum vocabulary in Balto-Slavic? The Northwestern Connections 4.1. PSl. *olxa alder (Russ. olxa, Pol. olcha, Bulg. elx, ESSJa VI: 23-25.), Lith. alksnis, elksnis, dial. alksnis alder (Smoczyski 11); OE alor, OHG elira < *alisa (Orel 15), Lat. (alnus alder, de Vaan 34-35). The variation in the Anlaut points to probable substratum origin (thus Derksen 370-1). 4.2. PSl. *bagno swamp (Russ. dial. bagn, Pol. bagno, Cz. bahno, ESSJa I: 125-127), Dutch bagger mud, OHG bah stream, OIc. bekkr brook, rivulet, OE bece (< PGerm. *bakiz, Orel 33), perhaps MIr. bal flowing water (if from *boglo-). Long *a in Slavic could be due to Winter's law, in which case this is a very early loan from some unknown source. 4.3. PSl. *balka ravine, pool (Pol. dial. baka pool, pond, Russ. dial. balka ravine, perhaps also Croat. dial. balka, bala Stipa pennata L, ESSJa I: 149), Lith. bal swamp, Latv. bala woodless valley, OE pl pool, OHG pfuol id. < PGerm. *plaz (Orel 292). 4.4. PSl. *belen, *bln henbane (Russ. dial. belen, OPol. ble, Cz. bln, Bulg. blen daydream, ESSJa I: 185-187), OE beolone henbane, OS bilene, perhaps Gaul. belenountan (Asg.), cf. also OHG bilisa (Germ. Bilsenkraut), Swedish Bolmrt, Kluge 111). 4.5. PSl. *bob bean (Russ. bob, Croat. bb, Pol. bb, ESSJa II: 148), OPr. babo bean(perhaps a Slavic loanword), OE bean, OHG bna (< PGerm. *baun < *bab-n), Lat.
2

faba bean. The usual comparison with Gr. phaks lentil and Alb. bath Vicia faba does not lead anywhere (Demiraj 94, see also Kuiper 1995). gas pole 4.6. *drg bar, pole (OCS drgy 'sticks', Pol. drg, Croat. drg 'rail'), Lith. dran (used as a lever), ESSJa V: 129-130; OIc. drangr detached pillar of rock, OIc. drengr heavy stick; a variant with *-k- is PSl. *drk bar, pole (Russ. druk, Cz. drouk, Croat. druk), ESSJa V: 130-131. The alternation of voiced and voiceless root-final stops might point to substratum origin (thus Derksen 121). 4.7. PSl. *golb p igeon (Croat. glb, OCS golb, Russ. glub, Pol. gob, ESSJa VI: 215-17), Lat. columba < *Kol-ombh-, with the suffix *-ombh-/*-embh- which is common in bird-names, cf. Lith. gelumb blue cloth, OPr. golimban blue vs. Russ. golubj pale blue, OPr. goby greyish, blue-grey (ESSJa VI: 217). 4.8. PSl. *grabr hornbeam (Croat. grb, Serb. dial. gbar, grbar, Sln. gber, grber, Rus. grab, Pol. grab, Cz. habr, ESSJa VII: 99-100). Cf. OPr. wosi-grabis Evonymus Europaeus. The appurtenance of Umbr. Grabovius (epithet of Jove on Iguvine Tablets) and Ancient Macedonian gr bin a kind of tree is uncertain. Cf. Lith. skrblas hornbeam (Smoczyski 568), perhaps Lat. carpinus id. < *(s)grp/bh-, a distinctively non-IE shape. 4.9. PSl. *grm d , *krmelj fester in the corners of the eyes, gramiae (Russ. CS grmd, Croat. krmlj, Slov. krmlj, ESSJa VII: 158-9). Cf. Lat. grmiae, gramiae viscous humour, rheum in the corner of the eyes and Goth. qrammia moisture (from substratum according to De Vaan 270). 4.10. OLith. kasulas hunting spear, Lat. corulus hazel-tree < *kosulo-, OIr. coll, W coll < PCelt. *koslo- hazeltree, OHG hasal, OIc. hasl < PGerm. *xaslaz id. (Orel 164). The alternation in the shape of the suffix (*-slo- / *-sulo-) points, perhaps, to non-IE origin. 4.11. PSl. *elbed, *olbd swan (Russ. lbed', Cz. labut', Croat. lbd, OPol. ab, Pol. abd, ESSja VI: 19, XXXII: 50-51.), OHG albiz, elbiz, Lat. olor, OIr. elu, W alarch, pl. eleirch. The forms with *a- may have been influenced by *h2elbhos white (Lat. albus etc.). 4.12. PSl. *xvoja needles of a coniferous tree, (Pol. choja, Russ. xvja, Pol. choja, ESSJa VIII: 125-6), Lith. skuj, Latv. skuja (Smoczyski 568, doubting the etymology), OIr. sc, scad [Gsg.] thorn bush (PCelt. *skwiyat-, EDPC 339). We may posit an original alternation *skwoy- / *skuy-, but the shape of the root is distinctively non-Indo-European. 4.13. *klati kneel (OCS klta [Ndu part. pres. act.] kneeling, Pol. klcze, Croat. klati, ESSJa X: 28-9), Lith. klnkiu walk with difficulty; Snoj (277) derives these words from PIE *kleng-, *klenk- (Lat. clingere, MHG lenken, MHG gelenke, OHG (h)lanca thigh, haunch, Kluge 310). The alternation of *k and *g may point to substratum origin. vas 4.14. PSl. *klen maple (Russ. kln, Pol. klon, Croat. kln, Vasmer I: 567), Lith. kle (Smoczyski 297), Latv. kavas, OE hln, OIc. hlynr < PGerm. *hluni. W kelyn holly, OCo. kelin, OIr. cuilenn (EDPC 213) < PCelt. *kolino- and need not be related (likewise OE holegn holly, OHG hulisboum. The connection with Macedonian klin(s)trokhos (Theophrast) is also doubtful.

4.15. PSl. *kobc, *kobuz, *kobz kite, hawk (Russ. kbec, , Pol. kobiec, Croat. kbac, Russ. dial. kbuz, Pol. kobuz, kobz, Vasmer I: 582-3), OHG habuh kite, OE hafoc hawk, OIc. haukr < PGerm. *habukaz (Orel 138). PSl. *kobc seems to be formed with the diminutive suffix *-c, and the form *kobuz may be directly comparable to the Germanic words for hawk. However, a reconstruction *kobug'o- does not look like a PIE word. 4.16. PSl. *klp swan Pol. dial. kieb swan, Russ. dial. kolp' spoonbill, Croat. dial. kp swan, ULus. ko, cf. Latv. glbis, Lith. gulb, gulbs (4), with initial g-, Derksen 261. The original form had g- to judge by OIr. gulban beak, sting, W gylfin beak, snout, if this etymology is correct (EDPC 168-9 does not accept it). 4.17. *PSl. *lntja lentil (CSl. lta, Russ. ljaa, Bulg. lta, Croat. la, Vasmer II: 84, ESSJa XV: 63-65), OHG linsa lentil (Kluge 521 claims that this is not a Latin loanword), Lat. lns lentil (De Vaan 238). Gr. lthyros pulse, Vicia sativa is probably unrelated. kas purse, PCelt. *makin (MW megin bellows, Co. mygen, EDPC 254), 4.18. Lith. ma OHG mago belly, OE maga id. < PGerm. *magn, Orel 253. The vowel *a may point to non-IE origin. 4.19. Lith. no tryn nettle, OPr. noatis, perhaps also PSl. *nat leafy top of a root vegetable (Ukr. dial. nat', Pol. na, Croat. nt ESSJa XXIII: 186f.), PGerm. *natn, *natiln (OE netele, OHG nazza, nezzila, Orel 281f.), PCelt. *ninati- (OIr. nenaid, MW dynat [pl], EDPC 290). The reflexes point to *ninati- and *nti-, a very un-Indo-European form. 4.20. PSl. *ovs oats (Russ. ovs, Pol. owies, Croat. vas, Derksen 384f.), Lith. avi, Latv. uza, OPr. wyse, Lat. avna < *aweksn; both the alternation of the voiced and voiceless palatalized velars in BSl. and the unusual shape of the suffix (*-eKs- / *-iKs-) point to the substratum origin (Oettinger 2003. 189). gas, 4.21. PSl. *rog horn (OCS rog, Croat. rg, Russ. rog, Pol. rg, Derksen 438), Lith. ra h Latv. rgs, OPr. ragis. Often connected to Lith. reg ti see, assuming a root *reg - to be visible, appear, Germ. sich regen, cf. also Lith. rogsti stick out, CS rogoz bull-rush, sedge, Russ. rogz. LIV reconstructs the root as *regh-, but there seems to be also a variant *rek- in PGerm. *rah stick, pole (ON r sailyard pole, OHG raha shuttle, Orel 293), Lith. r kls [pl.] scaffolding. 4.22. PSl. *ruxo clothes (Russ. rxo, Croat. rho, Pol. rucho) can be derived from *ruk-s-oand related to PGerm. *rukkaz (OHG rocko overcoat, OE rocc upper garment, Orel 308), PCelt. ?rowkk- / *ruk- (MIr. rucht tunic, W rhuch film, layer, jerkin, coat, EDPC 315). The alternation between the geminate and the simple *-k- (and *x in Slavic) looks non-IE. 4.23. Lith. serbent blackberry, redcurrant, dial. sarbent (Smoczyski 543), Russ. dial. sorbalna blackberry, Rubus fruticosus, serbalna, serbarna rose-hip, Byelorussian cerbaln blackberry (Vasmer II: 697), Lat. sorbus service-tree (De Vaan 576), perhaps also Swedish sarv rudd, redeye, Leuciscus eryhrophthalmus (a reddish fish). A Baltic loanword in Slavic? The verbal root seems to be attested in Lith. sir bti ripen (perhaps originally redden, since all of the berries denoted by this set of verbs are red).

 bras, Latv. 4.24. PSl. *srebro silver (OCS srebro, Russ. serebr, Pol. srebro), Lith. sida sidrabs (Smoczyski 546, who claims that the Baltic forms were borrowed from Slavic), Goth. silubr. The appurtenance of Celtib. ilapur is probable, but not quite certain (EDPC 41). 4.25. OPr. spurglis sparrow, OE spearwa sparrow, PCelt. *sfrawo- crow (Bret. fraw, Co. frau, EDPC 334), perhaps Lat. parra a kind of bird, Gr. (Hesych) sparsion sparrow-like bird, cf. Schrijver 1997: 304. 4.26. *(i)vlga oriole (CS vlga, Russ. volga, Pol. wilga, dial. wywioga, wiwielga, Croat. ludze, MHG vga, Bulg. avlga, ESSJa VIII: 251-252), Lith. volung oriole, Latv. va witewal oriole, Dutch wielewaal. Since wite- of MHG witewal means wood, it has been claimed that the apparent prefix *i- in Slavic is from *iwo-, i.e. that *ivlga is from *ivovlga (Derksen 217), where *ivo- is the word for willow (Russ. va, Croat. va, Lith. eva bird-cherry). 5. The Southeastern Connections 5.1. PSl. *kolyba, *koliba hut (Slov. kolba, Cz. dial. koliba tent, Bulg. kolba), Gr. kalb; possibly related to PSl. *xalupa hut, cottage' (Croat. halupa (Kastav), Slov. halpa, Russ. dial. xalpa, Pol. chaupa). 5.2. PSl. *kos blackbird (Croat. ks, Russ. kos) probably unrelated to Gr. kpsikhos, kssyphos (also kttyphos, kssykos) id.. PSl. *kos < *kopso- or PIE *kos-, from the root *kes- to scratch, to comb (OCS esati, etc.). The semantic development would have been from the scratcher to blackbird. A parallel: Croat. ljugar (goldfinch, Carduelis caruelis), derived from ealj comb < PIE *kes- (Skok I: 311f.). 5.3. PSl. *mrky carrot (Russ. morkv, Croat. mrkva, Pol. marchew), OHG morha; often related to Gr. brkana [pl.] wild herbs (Vasmer II: 158-9). 5.4. PSl. *trst reed, cane (OCS trst, Russ. trost', Pol. tre, Croat. trst), Lith. triai, trus, Latv. trusis, Trautmann 330, Vasmer III: 141. The connection with Gr. thron reed, rush (IEW 1097, DELG 443) is doubtful. 6. Pan-European Substratum words 6.1. PSl. *brglz fink, Sitta syriaca (Russ. berglz, Croat. brglijez, Vasmer I: 75), Gr. fryglos, Lat. fringilla (Vasmer I: 75). 6.2. PSl. *emer 'Veratrum album, hellebore' (Russ. dial. mer, Pol. dial. czemier, Croat. meras, OHG hemera 'id.', Gr. kmaros. mr venom, anger, ESSJa IV: 52-3), Lith. ke 6.3. PSl. *eremuxa, *ermxa, *serma ramson, bird cherry (Croat. srjemua, crjema, Russ. dial. ermuxa, Pol. trzemcha, Slov. rmha, ESSJa IV: 66-68), Lith. dial. kermu tip of a drill, ramson, Gr. krmyon, krmyon onion, OIr. crem wild garlic, leek, W craf [Collective] garlic < PCelt. *kremu-, *kramu- (EDPC 222), OE hramsa ramson, ESSJa IV: 66-68. The forms with initial *s- < *k in Slavic show that this is a very early loan, since it exhibits Palatalwechsel. The comparison with Greek and Celtic forms does not allow the reconstruction of a PIE prototype.

6.4. PSl. *esn garlic (Russ. esnk, Croat. san, Pol. czosnek, ESSJa IV: 89-90), PCelt. *kasnin garlic (MIr. cainnenn, OW cennin [p] leeks, daffodils, OCo. kenin gl. allium, EDPC 193); the alternation *e/*a seems to point to substratum origin (Schrijver 1995: 495), but the e-vocalism of Slavic may be due to the influence of the verbal root *kes- to comb, (?) to peel (OCS esati). 6.5. PSl. *konop rope (Slov. konp, Bulg. konp, Russ. dial. konp, konp', OPol. kono, ULus. konop, Vasmer I: 615), OHG hanaf hemp, OE haenep < PGerm. *hanapa- (Kluge 354), Gr. knnabis. This is an old Wanderwort, perhaps of oriental origin, cf. Sumerian kunibu hemp. 6.6. PSl. *klk hip (Russ. dial. kolk, Croat. kk, Bulg. klka, Vasmer I: 600), Lith. kulkns, ankle, OPr. culczi hip, Latv. kulksnis tarsal joint, Lat. calx heel, hoof (De Vaan 86). 6.7. PSl. *mak poppy (Russ. mak, Croat. mk, Pol. mak, ESSJa XVII: 149-151), Lith. dial. guon, Latv. magune, OPr. moke. The Lithuanian and Latvian words were probably ma borrowed from Germanic, (Derksen 299-300), cf. OHG mago besides mho, and the OPr. word may be a loanword from Polish. PGerm. *makn- and *m kn, Gr. mkon, Dor. m kon poppy < *meh2ko-, pehaps alternating with *mh2k-. The vowel alternation in Germanic might be due to different adaptations of the same foreign sound (?//) either as *a, or as *. 6.8. PSl. *rpa turnip (Russ. rpa, Croat. rpa, Cz. epa), Lith. rp, OHG ruoba, ruoppa, Lat. rpum, Gr. rhpys, rhphys. The diverging vocalism of the words for turnip suggests this is a loanword from some non-IE source (thus also De Vaan 14). 6.9. PSl. *roda heron (OCS roda, Croat. rda, Skok III: 163), Gr. erdis, rhdis, Lat. ardea, perhaps also OIc. rta a kind of duck. The diverging vocalism of the words for heron cannot be reconciled with a PIE reconstruction. This word was probably as migratory as the bird it denotes. 6.10. PSl. *vers, *versk heather (Russ. vresk, Cz. ves, Croat. vrjes, Derksen 516), Lith. vris, Latv. vir zis, vrsis; the Slavic form points to *werk'-, and the Baltic forms to *wrk'- or *wrg'-. T. Pronk (p. c.): Baltic -- due to influence from Lith. ver ti string, tighten, squeeze < *uerh -, cf. Gr. erek (< *wereyk'-o-), OIr. froech, W grug (< *wroyk'o-), but no common prototype can be posited. 7. Balto-Slavic words without PIE etymology gurs 7.1. BSl. *bauKura- > PSl. *bugor, *bugr hillock (Russ. and Ukr. bugr), Latv. bau b(u)ras hillock (with metathesis and unexplained k-); a id., cf. perhaps also Lith. kau connection with the PIE root *bhewgh- to bow, twist (Goth. biugan, Skr. bhuj-, ESSJa III: 79) is formally difficult (because of the acute in Latvian) and semantically unattractive. The E Slavic words may have been borrowed from Baltic. 7.2. BSl. *Purna- > PSl. *brna snout (Slov. dial. bna carnival mask depicting an animal, Croat. brna, Maced. brna nose-ring of animals, Lith. burn (3) mouth, face, but Latv. puns snout. The etymological connection with Arm. beran is conjectural (Derksen 69), as well as with the root *bherH- to bore (Lat. formen 'opening'), and the comparison with MIr. bern [ f] gap, breach, pass is a mere speculation.
6

7.3. BSl.*e/a-mela- > PSl. *emela,*jmela mistletoe (Russ. omla, Pol. jemioa, Croat. malas, e malas, Latv. amuols, amuls, OPr. emelno; cf. also (with mela, dial. mela), Lith. a Ablaut) Latv. muls, mulis id. Often derived from PIE *h1em- hold, take (Lat. emo, OCS ti, Smoczyski 13), because mistletoe is used in the production of glue. imati, Lith. im, im More likely, only Slavic forms with initial *j- (e.g. Croat. imela) were influenced by this root. The variations in the vocalism in BSl. are unexplained. 7.4. BSl. *graSa- threatening > PSl. *groziti threaten (Russ. grozt, gro, Pol. grozi, Croat. grziti, ESSJa VII: 143), *groza horror (OCS groza, Russ. groz, Pol. groza, Croat. grza, ESSJa VII: 141-142), Lith. grati threaten, gras beautiful, OLatv. grzns beautiful, luxurious. With s- we also have Lith. grasnti threaten, grass threatening. The connection with OIr. gargg wild, Gr. Gorg Gorgon, gorgs terrible (Snoj 193, IEW 353) is very dubious. The alternation of Slavic *s and Lith might point to a substratum origin. 7.5. BSl. *Kl(a)ua- > PSl. *glux deaf (OCS glux, Russ. glxyj, Croat. glh, Cz. hluch, ESSJa VI: 146-7), Lith. gluas dumb. Vasmer (I: 277) compares Lith. glusns obedient, klusns id., dial. klsas dumbish. The alternation of *k- and *g- may be a sign that these words are of substratum origin. 7.6. BSl. *Kraua- > PSl. *gra pear (Russ. gra, Cz. hruka, Pol. grusza), PSl. *krua (Bulg. kra, Croat. krka, dial. krva, Pol. dial. krusza, ULus. kruva, LLus. krua; in Baltic only with *k-, cf. Lith. kriu, OPr. [pl.] crausios, Vasmer I: 314. ESSJa (VII: 156) connects these words with the verbal root found in PSl. *gruiti / *kruiti crush, but this is hard to believe on semantic grounds. 7.7. BSl. *ni- > PSl. *inj hoar-frost (OCS inii, Russ. nej, Cz. jn, Croat. nje, ESSJa VIII. 235-6), Lith. nis. Often related to Germ. Eis < Germ. *saz (Orel 204), Av. axacold, isu- icy, but *-sn- would be preserved in Balto-Slavic. das, Latv. ldus, OPr. ladis, ESSJa 14: 91-2. 7.8. BSl. *ledu- > PSl. *led ice, Lith. le Vasmer (25) adduces MIr. ladg snow (MoIr. laogh), but this the MIr. word is poorly attested (G sg. ladga or laide according to DIL). 7.9. BSl. *pausta- > PSl. *pust empty, deserted (OCS pust, Russ. pustj, Pol. pusty, Croat. pst, Derksen 424), OPr. pausto; perhaps from *powH-dh-to- cleansed (LIV *pewH- (1), cf. Ved. pun ti, OHG fouwen sieve)? 7.10. BSl. *(a)rayHa/u- > PSl. *orx nut, walnut (Russ. orx, Pol. orzech, Croat. rah, ksts, OPr. buccareisis beech-nuts; Snoj 476f. connects Derksen 374), Lith. reutas, Latv. rie also Alb. arr, Gr. pl. rya. The alternation of initial *o- (< *a-) in Slavic and the vowelless form in Baltic is similar to the alternation Schrijver (1997) detected in words from NW European substratum. 7.11. BSl. *seyHl > PSl. *sila force (OCS sila, Russ. sla, Pol. sia, Croat. sla, Derksen 451), Lith. sela 'soul', OPr. seilin diligence; perhaps to *seh2i- bind, Lith. dial. sien, Latv. set, Skr. s-, Hitt. ihiya-, The connection with the root *seh2i- is possible only if *sila is from *sih2leh2, while Lith. sela < *seh2ileh2 (which is quite improbable).

7.12. BSl. *ama- > PSl. *som sheat-fish (Russ. som, Pol. sum, Croat. sm, Derksen 461), mas, Latv. sams; the usual connection with Gr. kamase nes a kind of fish, [pl.] is Lith. a doubtful. Smoczyski 624 connects these words to Gr. kmaks shaft, pole, but the semantic connection is to weak (? fish as long as a pole). 7.13. BSl. *traupa- > *trup dead body, log, OPr. trupis log; Snoj (789) compares Lith. trup ti crumble and Gr. trp bore, pierce trough. Vasmer (III: 143-4) adduces Latv. trupt rot, decay, Lith. trups crushed to pieces, Latv. trupe soil, earth, Smoczyski 6923 accepts this and adduces forms with initial d- such as Latv. drupt crumble, Latv. dial. drapt crumble. Beekes (1513) doubts the appurtenance of Gr. trp and Gr. trp hole because these point to a laryngeal root. The alternation between initial *t- and *d- in Baltic might point to substratum origin. 7.14. BSl. *tuli- > PSl. *tlst thick (Russ. tlstyj, Pol. tusty, Croat. tst), Lith. tulti, tult swell, perhaps Latv. tulzis gall, Lith. tuls (4), Trautmann 332, Vasmer III: 117. The alternation between st and might point to a substratum origin. There may be a connection to Germ. *tulguz firm, steadfast (Goth. tulgus, Orel 411) and *talg tallow (Germ. Talg, OIc. talg, Orel 400), if we start from a substratum *TolK'- / *TlK'-. 8. Conclusions I. Baltic and Slavic share probable substratum loanwords much more often with the western and northern European languages (Germanic, Italic and Celtic) than with the southern ones (Greek and Albanian). II. Words attested only in European branches of IE, but lacking in Anatolian, Armenian, Tocharian and Indo-Iranian, are not often shared by both Baltic and Slavic groups of languages. III. The number of words that may be of substratum origin, and that are preserved only in Balto-Slavic, is very limited.

Words shared with Words shared with Words with cognates NW European SE European in most European branches of IE 26 (12 attested in 4 (2 attested in both 10 (6 attested in both both Baltic and Baltic and Slavic) Baltic and Slavic) Slavic) Table I: The distribution of substratum vocabulary in Baltic and Slavic

REFERENCES Andersen, Henning 2003. Slavic and the Indo-European Migrations. Language Contacts in Prehistory. Studies in Stratigraphy (Current Issues in Linguistic Theory 239) ed. by Henning Andersen, AmsterdamPhiladelphia: John Benjamins, 4576. Anreiter, Peter 2001. Die vorrmischen Namen Pannoniens, Budapest: Archaeolingua. Anthony, David 2007. The Horse, the Wheel, and Language, Baltimore: Johns Hopkins University Press. Beekes, Robert S. P. 1996. Ancient European Loanwords, Historische Sprachforschung 109: 215-236. Boutkan, Dirk 1998. On the form of North European substratum words in Germanic, Historische Sprachforschung 111: 102-133. Boutkan, Dirk 2003 On Gothic MAGA ~ Old Frisian MEGITH and the form of some North European substratum words in Germanic, Amsterdamer Beitrge zur lteren Germanistik 58: 11-27. Brozovi, Dalibor 1992. Review of Holzer 1989. Wiener Slavistisches Jahrbuch 35: 223-225. Endzeln, Janis 1971. Jnis Endzelns' Comparative Phonology and Morphology of the Baltic Languages, The Hague: Mouton. Gob, Zbigniew 1972. Kentum Elements in Slavic, Lingua Posnaniensis 16: 53-81. Gob, Zbigniew 1990. The Origins of the Slavs, Columbus: Slavica. Holzer, Georg 1989. Entlehnungen aus einer bisher unbekannten indogermanischen Sprache im Urslavischen und Urbaltischen, Vienna: Verlag der sterreichischen Akademie der Wissenschaften. Kallio, Petri 2005. A Uralic Substratum in Balto-Slavic Revisited, in: Sprachkontakt und Sprachwandel: Akten der XI. Fachtagung der Indogermanischen Gesellschaft, ed. by Gerhard Meiser & Olav Hackstein, Wiesbaden: Reichert, 275-283. Katii, Radoslav 1976. Ancient Languages of the Balkans, The Hague: Mouton. Kortlandt, Frederik 2003. An Indo-European substratum in Slavic, in: A. Bammesberger (ed.) Languages in Prehistoric Europe, Heidelberg: Winter, 253-260. Kuiper, Franciscus B. 1995. Gothic bagms and Old Icelandic ylgr, NOWELE 25: 63-88. Mallory, James 1989. In Search of the Indo-Europeans, London: Thames and Hudson. Matasovi, Ranko 2005. The Centum Elements in Balto-Slavic, in: Sprachkontakt und Sprachwandel: Akten der XI. Fachtagung der Indogermanischen Gesellschaft, ed. by Gerhard Meiser & Olav Hackstein, Wiesbaden: Reichert, 363-374. Matasovi, Ranko 2011. The etymology of Latin focus and the devoicing of root-final stops before *s in PIE, Historische Sprachforschung 123: 212-216. Matasovi, Ranko 2012. Areal Typology of Proto-Indo-European: The Case for Caucasian Connections, Transactions of the Philological Society 110: 1-28. Moszyski, Leszek 1992. Review of Holzer 1989. Rocznik Slawistyczny 48/1: 88-95. Oettinger, Norbert 2003. Neuerungen in Lexikon und Wortbildung des NordwestIndogermanischen, in: A. Bammesberger (ed.) Languages in Prehistoric Europe, Heidelberg: Winter, 183-194. Schrijver, Peter 1997. Animal, Vegetable and Mineral: some Western European Substratum Words, in: Sound Law and Analogy, Papers in Honor of Robert S. P. Beekes on the Occasion of his 60th Birthday, ed. by A. Lubotsky, Amsterdam and Atlanta: Rodopi, 293-316. Stang, Christian 1972. Lexikalische Sonderbereinstimmungen zwischen dem Slavischen, Baltischen und Germanischen, Oslo-Bergen-Troms: Universitetsforlaget.