«FREDERIK KORTLANDT I Professor Hamp has recently returned to the problem of PIE *eu in Balto- -Slavic (1976). I take the matter up again because his ...»
THREE PROBLEMS OF BALTO-SLAVIC PHONOLOGY
Professor Hamp has recently returned to the problem of PIE *eu in Balto-
-Slavic (1976). I take the matter up again because his analysis has certain implica-
tions for the relative chronology of sound laws.
After a detailed study of the earlier literature, Endzelin concludes that both
prevocalic and preconsonantal *eu have a twofold reflex in Balto-Slavic, viz. *ev and *jau (Slavic ju) if the following vowel is front, but *av (Slavic ov) and *au if the following vowcl is front, but *av (Slavic ov) and *au (Slavic u) if the follo- wing vowel is back (1911 : 78-104). This point of view is often repeated in the more recent literature (e. g., Vaillant 1950 : 110 and 123, Stang 1966 : 32 and 74).
I agree with Hamp that it cannol be correct. The Slavic dat. sg. synovi *-euei and nom. pl. synove *-eues suffice to show that prevocalic *eu yielded Slavic ov before front vowcls äs well. Since H. Pedersen's conclusive discussion of Lith. tau (1935), it can hardly be doubted that the only phonetic reflex of preconsonantal *eu was *jau in Balto-Slavic.
If the Balto-Slavic reflex of PIE *eu was *av (or rather *ov) before vowels and *jau (or tarher *jou) before consonants, the occurrence of ev requires an ex- planation, especially in Lith. devyni, Slavic devgtt. The Suggestion that de- was borrowed from desimtjdesgtb cannot be maintained. As Hamp points out, ev must have been reintroduced in the cardinal *dovin *Η^ neun on the model of the ordinal *deuno-, which was subsequently replaced by *devino- on the model of the 1 new cardinal *devin. It follows that preconsonantal *eu had becn preserved at a stage which was posterior to the phonetic elimination of prevocalic *eu and that the latter development was early Balto-Slavic.
This chronology is in contradiction with the one given by Zupitza, who dates the Slavic development of *ev to *ov after the first palatalization (1907 : 251). The latter chronology i s based on Czech navsteva,visit', Old Czech vscieviti,to visit',
which is derived from *(s)keu-, cf. Gothic usskaws, Latin caveo (Matzenauer 1884 :
179 and Mikkola 1904 : 96). Though Machek does not even mention this etymo- logy (1968 : 392), I think that it is correct. It is certainly preferable to the proposed connections wAh Lith. svecias and Slavic posetiti, which do not fit phonologically, or PIE *ueid- (Berneker), which cannot be identified without violating Winter's law (sce below). I assume that ev was restored in this word on the basis of precon- sonantal *eu, e. g. in cuti, in the same way äs in devgtb.
1 I think that the initial d- is the phonetic reflex of PIE *H1n-, I learn from Die Sprache 24 (1978), 239 that Hamp puts forward the same view in the CLS book of squibs (Chicago, 1977), which has not been ac
The Indo-European proto-language possessed two series of velar stops, viz.
a palatovelar and a labiovelar series.2 The „plain velars" resulted from the depalatalization of the palatovelars in some dialects and the delabialization of the labiovelars in others. As Steensland has shown (1973 : 30-35), the Opposition between the two velar series was neutralized after initial *s in Proto-Indo-European. The archiphoneme was palatovelar before *i and plain velar in other positions (Steensland 1973 : 34). This explains the double reflex of initial *sk in Balto-Slavic.
According to the two principal doctrines, PIE *SK appears äs Lith. s, Slavic s (e. g., Endzelin 1939), or äs Lith. sk, Slavic sk (Büga 1922 : 249-252 or 1959 : 284In his most recent discussion of the matter, where references to the earlier literature can be found, Stang agrees with Endzelin on the initial reflex and with Büga on the medial reflex of PIE *SK (1972 : 83-87). The main evidence is the
Sl. sovati, Lith. sauti, Old Norse skjota.
Sl. sbjatiy Gothic skeinan.
Sl. s£nb, Latvian sejs, Gr. skia.
Sl. iskati, Lith. 'ieskoti, Skt. icchati, OHG. eiscön.
Sl. voskz, Lith. väskas, OHG. wahs.
Sl. -bsk-b, Lith. -iskas, Gothic -isks.
Sl. jasno, Lith. aiskus.
Sl. räsnt, Lith. raiskus.
As to Lith. sokti, Slavic skociti, I think that these words are not related (cf.
Fraenkel 1965 : 1022). The initial 5 in sbjati and sem, (which replaces earlier *sy'a, cf. also Alb. hie and Toch. B skiyo) continues the palatal variant of initial *sk before */. The same development could be assumed for Lith. sauti, Slavic sovati, if the rise of *iou (or */a«) from preconsonantal *eu were anterior to the rise of new initial *sk before *i. I do not think that this chronology can be upheld, however. It follows from the preceding section that the development of preconsonantal *eu was posterior to the elimination of the syllabic resonants, which reintroduced initial *sk before *i, e. g. Lith. skirti, skilti, skinti. Thus, I subscribe to the traditional view that the initial fricative of Lith. sauti and Slavic sovati continues an initial palatovelar and that there is a mobile *s in the Germanic cognates.
The neutralization of the Opposition between the velar series after initial *s in the Indo-European proto-language suggests the possibility that the Opposition was also neutralized after non-initial *s. It has long been recognized that IndoIranian does not offer evidence for a distinction between palatovelars and plain velars after *s.s Moreover, there is an important piece of evidence which has not
received due attention in the literature on the subject (cf. Von Patrubany 1902 : 124):
Cf. Meillet 1894, Steensland 1973, Kortlandt 1978.
Cf. Zubaty 1892 : 9, Meillet 1894 : 295, Wackernagel 1896 :155.
THREE PROBLEMS OF ΒΑΙ/ΓΟ-SLAVIC PHONOLOGY 59Sl. mozg-b, Avestan mazga-, Old Norse mergr *mosgho-.
Lith. mäzgas, Gr. moskhos *mosgho-.
Αϊτή. mozi, Gr. moskhion *mosghio-.
Though the relevant material is small, I assume that after non-initial *s, too, the Opposition between the velar series was neutralized, and that the archiphonerne 4 was palatovelar before *i and plain velar elsewhere.
Thus, the expected reflex of medial *sk is Balto-Slavic sk in the words listed above. The appearance of s in Lith. ieskoti, -iskas, aiskus, raiskus must be attributed to the preceding *i. For Lith. väskas and OHG. wahs, which is Stang's main argunient for subscribing to Büga's derivation of sk from *SK, I reconstruct *uoKsko-, which is the only form that explains both the Gerrnanic and the Balto-Slavic material. The Baltic inchoative suffix -sta-, which cannot be separated from PIE *-ske-, requires special attention. Its historical relationship has been clarified by Van Wijk, whose point of view is unjustly disregarded by later investigators :5 „Ich halte das baltische Formans -sta- für identisch mit dem in ändern indogermanischen Sprachen häufigen -sqo-, und zwar nehme ich an, dass -sqo- zunächst bei denjenigen Verben durch -sto- ersetzt worden ist, deren urzel ein k oder g enthielt, und dass dann analogische Übertragung auf die Verba mit anderm Konsonantismus stattgefunden hat. Die Bedeutung des -sia-Präsens stimmt schön zu derjenigen der -sgo-Präsentia anderer Sprachen" (1933 : 58). Compare in this connection the Substitution of -utas for -ukas after stems in k and g in Lithuanian dialects (Hasiuk 1970).
It has been argued that the suffix -sta- represents the phonetic development of PIE *-SKB- (Leumann 1942 : 118-126). This position, which can no longer be maintained, is apparently supported by Lith. tukstantis, Prussian tüsimtons, Slavic tysgsti (Ru. tysjaca), tysgsti (SCr. tisuca). The agreement between the Hast Baltic and the South Slavic vocalism on the one hand, and between the West Baltic and the North Slavic vocalism on the other, suggests that the latter branches took the vowel from *simto (Lith. simtas) in late Balto-Slavic.6 The older vocalism is reminiscent of *-konta in Gr. triakonta and Breton tregont, cf. Arm. eresun *-sonta. The main problem is the presence of s, not s, in Lith. tukstantis, which cannot be derived from *tüs- or *tüks-. I would suggest that the cluster -kst- is due to metathesis of earlier *-tsk-, which is compatible with the Gerrnanic and Slavic material. If this is correct, Hast Baltic *-sk- betrays that the word dates from a period when the Opposition between the velar series was still neutralized after *s, while Slavic and West Baltic -5- point to compounding or reanalysis at a later stage.
4 Gr. aspis is probably a loan-word and does not constitute a counter-example (cf. Frisk 1973 : I 169).
6 Generally, Van Wijk's contnbutions to Baltic and Slavic linguistics are not sufficiently appreciated by the scholarly Community: too many of his valuable insights remain unknown to those who could benefit from his ideas. Stang does not even mention Van Wijk's opinion on the suffix -sta- (1966 : 343 and 1972 : 83). Endzelin's objections against Van Wijk's view are not convincing (1937:428-430).
6 Following Trautmann (1923 : 4), I assume that *simto was replaced with *sumto in early Slavic on the basis of its apophonic relations (cf. also Vaillant 1950 : 172). The reason for the replacement was the absence of e-grade alternants, while the o-grade had evidently been preserved in the decades in *-Komt-. The new form developed phonetically into szto (cf. Kortlandt 1980, section 3.13).
60 FREDERIK KORTLANDT
One of the most important discoveries of recent years is thc following: „In Baltic and Slavic languages, the Proto-Indo-European sequence of shoit vowel plus voiced stop was reflected by lengthened vowel plus voiced stop, while short
vowel plus aspirate developed into short vowel plus voiced stop" (Winter 1978 :
439). I have called this rule,Winter's law' in my chronological account of Baltic accentuation (1977). Here I shall discuss the main exceptions to the rule.
Some of the exceptions were explained by Winter hirnseif already. Thus, Lith. pädas and Slavic podz have nothing to do with PIE *ped-,foot', but must be derived from *ρο-άΚΆ-ί-ο- for both formal and semantic reasons. Slavic sedüo was probably borrowed from Gothic sitls. Lith. segti and Slavic xodi, have no certain etyniology. Some other exceptions are explained by the relative chronology of sound changes. The short vowel of Lith. duktS and Slavic *di,kti which must be derived from PIE *dhugH2ter in view of Gr. thugater, is regulär bccause the loss of the laryngeal and the assimilation of *g to the following *t, which Balto-Slavic shared with Germanic and Armenian, was anterior to Winter's law, which must be dated to the end of the Balto-Slavic period (cf. Kortlandt 1977 : 322).
The semantic identity of Slavic bogt, and Iranian baga- and the absence of the word from Baltic suggest that the Slavic word was borrowed from Iranian.
The semantic argument has been refuted by Meillet, who adduces „d'autres termes fondamentaux du vocabulaire religieux slave oü l'hypothese d'un emprunt est exclue" (1926 : 168). Moreover, the words bogati,, ubogz, Czech zbozi derive from
an earlier meaning,riches\ which is in perfect correspondence with Skt. bhdgah:
the latter word means both,fortune' and.distributor' (epithct of gods). It now turns out that Winter's law excludes the derivation of Slavic bogt from *bhogwos on formal grounds 7 I conclude that the word was borrowed from Iranian at an early stage, not only with the meaning ;god', but also in the sense of,fortune'. It seerns probable that other correspondences between Slavic and Iranian can also be attributed to very early influence of the latter on the former, e. g. the meaning of the word slovo, Avestan sravö* Slavic ognb, Lith. ugnis, Skt. agnih, Latin ignis can all be derived from *ygwnis.i The labialization of the original labiovelar, which accounts for the initial u (not z)
of Balto-Slavic *ungnis, was lost before the following n in Latin (cf. Meillet 1894 :
279). It appears that the medial cluster *-ngn- blocked the Operation of Winter's law.10 The first *n was subsequently eliminated in Baltic. The Slavic development of *un- to o- must be viewed in the chronological perspective of other developments.
7 The connection with Gr. phagein *bhH2g- (Frisk 1973 : II 980) cannot be maintained_ 8,,Le caractere religieux de sravö dans l'Avesta est manifeste; sravö y est une expression plus specifiquement religieuse de ce qui est ordinairement indique par vacö 'parole'." (Meillet 1926 : 169) The semantic change of Slavic slovo apparently ousted the original Balto-Slavic word, which has been preserved in Prussian wirds. The original meaning of slovo has been preserved in the verb sluti and its derivative slava.
" Cf. Hamp 1970. The view that the initial vowel derives from a syllabic nasal was already propagated by Meillet in Memoires de la Societe de Linguistique 8 (1894), 236.
10 This rule offers a clue for the relative chronology of Winter's law with respect to the loss of the syllabic resonants (cf. Kortlandt 1977 : 322).