1 In the work under review, Shevelov gives more than 300 bibliographical references.

2 Originally published in Paris, 1924. The second edition (Paris, 1934) has recently been reprinted (Paris, 1965) and has been translated into Russian by Kuznecov, P. S., under the title Obščeslavjanskij jazyk (Moscow, 1951)Google Scholar.

3 For example, with the suffix -tb (pp. 348-49): strastb ‘suffering’ from the stem of stradati ‘to suffer,’ věstb ‘knowledge’ from the stem of vědeti ‘to know,’ čbstb‘honor’ from the stem čisti (pres. čbto) ‘to honor,’ etc.

4 For example, the few forms with the suffix -oba (p. 355): zbloba ‘maliciousness’ from the stem of Zblt ‘evil, bad,’ Russian xudoba ‘leanness, thinness’ from the stem of xudb ‘needy, poor, slight,’ etc.

5 For example, nouns in -bba (pp. 355-56): službba ‘slavery’ from the stem of sluga ‘servant' (cf. also služiti ‘to serve’), družbba ‘friendship’ from the stem of drugb ‘friend,’ etc.

6 Originally published in Paris, 1902-5. The second part of this work has recently been reprinted in a phototype edition, Paris, 1961.

7 Note that it is not always the case that deverbal nouns have root vowel oj e.g., běgb ‘flight' from the stem of běžati ‘to flee,’ sbnbmb ‘council’ from the stem of snnsbneti s$ ‘to meet, come together.' For discussion and further examples, see Le slave commun, p. 340, and Études, pp. 214-15.

8 See note 3 above.

9 Prague, 1929; “Travaux du Cercle Linguistique de Prague,” No. 2. Reprinted in Jakobson, , Selected Writings, I : Phonological Studies (The Hague, 1962), 7–116 Google Scholar.

10 For further discussion see Jakobson, “Observations.” The most recent argument against this position is given in Paul Postal, Aspects of Phonological Theory (Harper and Row, to appear).

11 By v, Shevelov means w (pp. 283-84); it is not clear why he does not use w when he means w, and v when he means v. This is not an isolated comment, for Shevelov generally confuses the symbols he uses; 0a is used throughout the book to refer to a back, mid, rounded vowel (why doesn't he use 0?), „a to refer to a front, mid, unrounded vowel (why doesn't he use «?), etc. Not only symbols, but elementary linguistic terms are confused; thus “palatal” and “palatalized” are frequently interchanged, as though the two terms were synonymous.

12 The only example of x after I cited by Shevelov is the much disputed *vix- ‘all'; the other attested form (smisati s$—also disputed) is not cited at all. See Machek, V, “Zur Erklärung der sog. Baudouinschen Palatalisierung im Slavischen und im Baltischen” Cercetări de linguistică;, III (1958), 328 Google Scholar.

13 For the mutative palatalization of velars, for example, Jakobson suggests that velars did not shift directly to c 3(z) s / š , but that they passed through the intermediary stage of nonstrident palatals k g x. On page x Shevelov states that he intends to ignore intermediate stages which left no direct factual evidence. It is true that no Slavic language or manuscript shows k g x as a reflex of k g x; therefore no direct factual evidence can be found for this intermediary stage. But how else can one explain the disparate behavior of k g and *? Independent confirmation for the correctness of this postulated intermediary stage comes from the fact that in OCS, for example, vowels were fronted after č“ž Š j c 3 and after s derived from x; cf. nom. sg. dorm ‘house’ but othch ‘father.’ If one assumes an intermediary stage of non-strident palatals, then the vowel fronting rule applies after palatals (č ž Šj k g x). If one does not assume an intermediary stage, one cannot state this vowel fronting rule simply, because č ž Š j c 3 s do not form a natural class. Moreover, vowels are not, in general, fronted after s (cf. běsb‘demon, devil’ but vbsh, assuming, as Shevelov does, that vt>sh is derived from *vix-; cf., however, loc. pl. vbsě“xb> for expected vbsixband the reference to Machek in fn. 12 above).

14 I am indebted to Morris Halle for his help in formulating the following solution.

15 Shevelov actually makes the completely unfounded assertion that this form (lujQ) was the I sg. form in CS (p. 283). Typically, the R form lovlja is mentioned in this discussion. One wonders why. Of what relevance is it to a discussion of CS? Once the form is mentioned, however, it gives rise to a serious problem, for in R one finds ov^ovl in inf. lov'if, 1 sg. lovl'u, nom. sg. lovl'a, on the one hand, but ov∼uj∼ovl in inf. torgovat', 1 sg. torguju, nom. sg. torgovl’a, on the other hand. Are these forms relevant to a discussion of CS? If so, why? How does one explain them? And why is R lovl'a mentioned and not R torgovl'a?

16 I have given a different solution to the R problem in “O cikličeskix pravilax v russkom sprjaženii,” Voprosy jazykoznaniia, No. 2, 1965, and to the OCS problem in “On the phonology of Old Church Slavonic conjugation,” International Journal of Slavic Linguistics and Phonetics, X (1966).

17 Cf. the problem mentioned in §2 above. Within the feature system suggested by Jakobson and others, one can say that velars shift to strident palatals before any nonconsonantal, nongrave segment; see Preliminaries, mentioned in §1.3 above.

18 “K voprosu o kačestve plavnogo v kornjax vosxodjaščix k *turt *tirt *tult v drevnerusskom jazyke XII-XIII w.,” Istoričeskaja grammatika i leksikologija russkogo jazyka (Moscow, 1962), pp. 20-28.

19 The details of these processes have been discussed by Morris Halle and myself in two articles in the Quarterly Progress Report (QPR), Research Laboratory of Electronics, MIT: Halle and Lightner, “On the Phonology of tort tolt tert telt in Old Church Slavonic and Russian,“ QPR, No. 75, Oct. 15, 1964, pp. 121-23; and Lightner, “On the Development of turt tirt tult tilt in Russian,” QPR, No. 76, Jan. 15, 1965, pp. 274-76.

20 The most readily understandable and most reliable introduction for students is Meillet's Le slave commun; after this, students with a strong background in linguistics can turn to Jakobson's difficult-to-read but extremely rewarding Remarques.