LITUANUS LITHUANIAN QUARTERLY JOURNAL OF ARTS AND SCIENCES   Volume 18, No.2 - Summer 1972 Editors of this issue: Antanas Klimas, Ignas K. Skrupskelis, Thomas Remeikis Copyright © 1972 LITUANUS Foundation, Inc.

DONUM BALTICUM. Edited by Velia Rūke-Draving. Stockholm, Almqvist and Wiksell, 1970. Pp. xiv, 598, Sw. Cr. 85 (Hard Cover, Sw. Cr. 105).

This volume, dedicated to the renowned Baltic scholar Christian S. Stang on the occasion of his seventieth birthday, contains seventy-five articles covering a wide range of linguistic topics. There are papers concerned with the phonological, morphological, syntactic, and semantic structure of each of the Baltic languages from both synchronic and diachronic points of view, as well as a number of papers on the relation of Baltic to Slavic and Indo-European in general. The broad scope of the papers is a fitting tribute to Professor Stang, whose pioneering work in each of these areas of Baltic scholarship has furthered immeasurably our knowledge of this most interesting family of languages.

The book itself is beautifully printed with surprisingly few typographical errors. The majority of the articles are in English and German with a few in Lithuanian, Russian, and French. Each paper not written in English is provided with an English summary.

Limitations of space prevent me from reviewing the vast majority of the articles. I shall therefore make a few brief remarks about several of the more interesting papers that deal with topics I am competent to comment on.

Henning Andersen ("On some Old Baltic-Slavic Isoglosses") has an interesting discussion of some issues related to the interaction between Pedersen's Rule (the so-called "Ruki" rule treating the outcome of IE *s in position after [i, u ,r, k] ) and the Satem palatalization of *k. He argues that both Slavic and Baltic underwent Pedersen's Rule in essentially the same fashion. The many discrepencies between Baltic and Slavic with regard to this sound change are attributed to different morphophonemic responses to the subsequent palatalization of *k.

More specifically, Andersen suggests that both Baltic and Slavic developed a "compact" allophone, call it s₂, of basic IE *s in position after i, u, r, k. Phonetically, of course, s₂ appears as the velar x in Slavic and the palatal š in Baltic. Subsequent to the introduction of Pedersen's Rule, *k merged with the s₁ variant in Slavic, but with s₂ in Baltic. This then led to different reinterpretations of the s₁ -s₂ alternations left from Pedersen's Rule. Since *k merged with s everywhere, Slavic developed a contrast between s₁(<*k) and s₂ (<*s) in position after [i, u, r, k]. Morphophonemically, this required that the s₂ variant of the alternation be set up as basic and the s₁ variant be generated by a quite complex and unnatural rule taking x to s except after i, u, r, k. It is not surprising, therefore, that this rule was lost, thereby generalizing the prophophonemically basic x to all the suffixes which exhibited the s₁ -s₂ alternation (eg. the loc. pl., sg. pres., etc.).

In Baltic, on the other hand, the situation was reversed. Since *k became s₂, this meant that only s₂ (from both *s and *k) occurred after [i, u, r, k]. Consequently, in the morphophonemic system some s₂ (<*s) alternated with s₁ , while those from *k did not, requiring that the s₁ variant of the alternation be taken as basic. With the elimination of Pedersen's Rule from the grammar of Early Baltic, the s₁ variant of the alternation was generalized, explaining why we find e instead of š in the suffixes mentioned above and hence so many apparent exceptions to the rule in Baltic.

In addition to the different outcomes of IE *k, Andersen discusses two other isoglosses separating early Baltic and Slavic. One of these concerns the reflexes of initial *s+g(h) clusters, which appear as sk in Baltic, but x in Slavic. The author argues that the Slavic change of this cluster to x historically preceded the rule which neutralizes the opposition between voiced and voiceless stops after initial s; and since the latter rule is a general Indo-European phenomenon, this Slavic change is very early. This argument depends upon the assumption that the synchronic ordering of the rules reflects their historical, chronological order — a contention which cannot always be maintained. For example, suppose it could be shown that for each form with an initial "movable s", there was a corresponding form lacking the s, so that the underlying voicedness of a stop in a s + stop sequence could be recovered. Then it could be maintained that a historically later rule changing s+g(h) to x was inserted into the list of ordered rules in the grammar of early Baltic immediately before the neutralization rule. However, since it is unlikely that each movable s form had a correspondent without the s, the claim that the Slavic change of *s+g (h) to x chronologically preceded the neutralization of voicing in stops is probably valid. Nevertheless, the methodological point is clear: synchronic ordering need not (and frequently does not) reflect chronological order. Hence, the validity of an inference from synchronic ordering to historical order must always be judged on the merits of each individual case.

The latter point is important, for the argument for the earliness of the second isogloss is not nearly as strong. This isogloss concerns the Baltic metathesis of sibilant + velar stop clusters in position before a consonant: Lith. megzti, mezga, čirkšti, čirškia; etc. To explain those examples such as blokšti, bloškia in which the sibilant is an š and the preceding segment is not [i, u, r, k], Andersen accepts the traditional account which sees the š of blokšti arising from an application of Pedersen's Rule to the metathesized s (ie. blosk-ti > bloks-ti > blokš-ti). The š of the nonmetathesized cluster in bloškia might be explained as arising by analogy with forms where the preceding sound was [i, u, r, k]. Forms with sk — ks alternations would then have to be attributed to lexical innovations developing after Pedersen's Rule was eliminated from the morphophonemic system. Whatever the merits of this particular account, we cannot accept Andersen's conclusion that Metathesis must therefore have chronologically preceded Pedersen's Rule. For it is quite possible that Metathesis was chronologically (much) later and was then synchronically ordered in the grammar of Early Baltic in position before Pedersen's Rule. The latter interpretation receives some support from the fact that Pedersen's Rule operated in both Baltic and Slavic, since we might expect a purely Baltic innovation to have originated after an innovation common to both languages.

Tamara Buch ("Zur Akzentuierung des Futurums im Litauischen") relates the Western (and therefore standard) Lithuanian innovation preventing de Saussure's Law from displacing accent to the desinence in the 1 and 2 singular future with several other properties which differentiate this tense from the present and past, where the rule operates regularly:

One of these properties is that the future tense is built on the "infinitive stem" — that is to say, the initial suffix in the desinence begins with a consonant, while the present and past stems have vowel initial desinences: /sek-si-/ versus /sek-a-/ and /sek-e-/. This is significant because all of the other forms of the verbal paradigm which have consonant-initial desinences do not and, to the best of my knowledge, never did attract the ictus via de Saussure's Law: cf. infin. sèkti, imper. sèki (from /sek-ki/), imperf. sèkdavau, and subj. sèkčiau. Hence, Western Lithuanian has undergone an analogical innovation prohibiting de Saussure's Law from placing accent on a consonant-initial desinence.

Another relevant difference (not mentioned by Buch) between the present and past tense forms versus those built on the infinitive stem is that the regular morphophonemic rule lengthening accented a and e in non-final syllables fails to operate when these vowels are the final vowel of a verb stem and are followed by a consonant-initial desinence. Thus, in sèksiu, sèksi, sèkti, etc., the e remains short, while it is lengthened in the present and past forms seka and sekė (cf. seku, sekiau, etc. where the underlying short root vowel surfaces phonetically). Since this lengthening rule also fails to apply to prefixal a and e (cf. àt-meta, bè-kruta), this might be described by setting up an internal word boundary between not only a prefix and a following root, but also between a verb stem and a following consonant-initial desinence: /àt- met-a/, /sek si-u/, but /sek-a/. The rule will fail to lengthen the accented vowels in the first two forms because these vowels are now interpreted as being in the final syllable.1

David G. Guild ("The Development of the Concept of Definiteness in Baltic and Slavic") compares the implementations of the grammatical opposition Definite-Nondefinite in Bulgarian and Macedonian, Old Church Slavic, and Latvian and Lithuanian. In Bulgarian and Macedonian definiteness is realized as an enclitic, historically derived from a deictic, on the first member of a Noun Phrase: cf. Bulg. kniga 'book', knigata 'the book', novata zelena kniga 'the new green book'. In Old Church Slavic this opposition was limited to adjectives and was marked by the inflected pronominal element *ji which was enclitic on the adjective, a structure which is virtually identical with the historical source of the opposition in Baltic. However, in the contemporary Baltic languages this historical source is much more transparent in Lithuanian than Latvian: cf. Lith. N.M.S. mažas /mažasis (*mažas-jis) daržas versus Latv. mazs/mazais darzs; G.M.S. Lith. mažo/mažojo (*mazaja) daržo versus Latv. maza/maza darza, 'a/the small garden'. The author points out the interesting fact that while in Latvian there are constructions in which the "long form" definite adjective can be combined with deictics šis or tas or with possessives like mans, tavs, etc., these constructions do not occur in Lithuanian. I think that their absence might be related to the possibility that the definite marker is still felt to be synchronically derived from a pronominal/deictic *jis in Lithuanian, while this is most likely not the case in Latvian. In Bulgarian the definite clitic occurs with possessives, but not with deictics. This in turn might be explained by synchronically classifying the definite clitic as a deictic. The absence of constructions containing both a deictic and a clitic could then be predicted by a plausable syntactic constraint generating at most only one deictic per Noun Phrase.

In his interesting paper ("Aus dem Gebiet der Intonationen im Litauischen") Czeslaw Kudzinowski reminds us that one of the main reasons for the lack of a really satisfactory analysis of Lithuanian Accentuation is the failure to base descriptions upon the same dialect, since the accents frequently vary from region to region. This is especially important with regard to Metatony, of which Kudzinowski distinguishes three types: 1) the same intonation on different syllables (dabar, dabar); 2) different intonations on different syllables (baltuoti, baltuoti); and 3) different intonations on the same syllables (kaltas, kaltas). In connection with the latter kind of Metatony, the author has examined the way in which German and Slavic borrowings are rendered in the Kurschat dialect. One surprizing result of his investigation is that in those cases in which the borrowed word is segmentally identical with a word that already exists in Lithuanian, the borrowed word is assigned an "opposite" intonation. For example, márszas 'march' and márkė 'postage stamp' are given an acute intonation in order to distinguish them from the native Lithuanian words maršas 'oblivion' and markė 'he soaked'. The borrowed words margas 'acre' (cf. German Morgen) and rymo, gen. sg. of rymas 'ream' have the circumflex tone because of the forms margas 'colorful' and rymo 'he remains leaning'. This phenomenon is appropriately termed "disintonation" by the author and disconfirms the traditional assumption that borrowed words are usually given a circumflex accent. Although such phenomena are always difficult to evaluate, this appears to be another example in which considerations homophony influence what would seem to be a purely phonological process.

The volume contains many more papers just as interesting as the four discussed above covering the entire range of Baltic linguistic scholarship. It is sure to be one which the specialist will want to refer to time and again in the course of future research.

Michael J. Kenstowicz
University of Illinois

1 For further discussion of this rule and related issues, see my Lithuanian Phonology. Unpublished University of Illinois Ph.D. thesis, 1971. Available from University Microfilms.
2 For discussion of another case of Metatony in Lithuanian which might be interpreted along similar lines, see my "Lithuanian Third Person Future." Studies Presented to Robert B. Lees by his Students. J. Sadock and A. Vanek (eds.). 1970. Linguistic Research, Inc., Edmonton.