LITUANUS LITHUANIAN QUARTERLY JOURNAL OF ARTS AND SCIENCES   Volume 37, No.4 - Winter 1991 Editor of this issue: Antanas Klimas, University of Rochester ISSN 0024-5089 Copyright © 1990 LITUANUS Foundation, Inc.

REFLEXES OF INDO-EUROPEAN SYLLABIC RESONANTS 
IN BALTIC, SLAVIC, AND ALBANIAN

HARVEY E. MAYER
Defense Language Institute

Pre-Slavic-Pre-Albanian differed from other dialects of Indo-European by an important morphological innovation — inactive verbs marked by the formant -ks-.1 This -ks- was the reversal of -sk- in active verbs.2 It later became -h- as in Slavic -x- in exati 'to ride' and Albanian -h- in dihet 'it is known'.3 Probably the last special common genetic feature of Pre-Slavic-Pre-Albanian were its first multiple reflexes of syllabic nasals n, m, long and short.

Any change of syllabic resonants in Indo-European was dangerous. It could cause morphological ambiguity. With the loss of laryngeals and schwa, their vocalic allophone, syllabic resonants tended to be replaced by vowels plus resonants. Probably a was the most neutral vowel in these new combinations. But in various phonetic environments this a could be palatalized to e, labialised to o, raised and palatalized to i, or raised and labialized to u, An e plus resonant could fall together with an existing e-grade allomorph. An o plus resonant could fall together with an existing o-grade allomorph.

Thus with, say, zero-grade *g(h)^wr-4 'to warm', e-grade *g(h)^wer-4 'hot', and o-grade *g(h)^wor-4 'to burn', a change of the syllabic r to e or o plus r might make it hard to understand *g(h)^wer- which now might mean 'hot' or 'to warm' or *g(h)^wor- which now might mean 'to burn' or 'to warm'.

A change of the syllabic r to a plus r might also not be safe. With any general falling together of o and a or e and a (or e, o, and a as in Indic and Iranian), we would have similar difficulties.

To avoid this, Proto-Germanic generalized the change of all syllabic resonants to u plus resonant.5 Baltic matched this, but, having satem palatalization,6 arose with dual reflexes, u plus resonant or i plus resonant.7

Albanian and Common Slavic, also with satem palatalization, more or less matched the Baltic pattern with syllabic liquids.8 Common Slavic shows br/br rl/bl from Proto-Slavic ur/ir, ul/il. In Albanian ri/ur, li/ul are normal reflexes of syllabic r,l.9

It is in the reflexes of syllabic nasals where Baltic differs intrinsically from Albanian and Slavic. Baltic shows a pattern there uniform with that of syllabic liquid reflexes, u or i plus resonant. Albanian and Slavic reflexes of syllabic nasals are more complicated. Albanian ones are normally a with or without the following n, m. But e, in un may also occur.10 Slavic data are far less clear. But I believe that they will be better understood if seen against the background of Albanian data than against that of Baltic data. Many Slavic matchings with Baltic most likely show Slavic calkings of Baltic forms. But Slavic multiplicity of reflexes of Indo-European syllabic nasals matching in a general way that of Albanian shows an older pattern dating back, I say, to Pre-Slavic.

Some say that the Slavic situation is undecipherable since i plus nasal and e plus nasal fell together to nasal vowel ę. But this did not happen everywhere. In final position e plus nasal becomes e as in imene 'in the name' while i plus nasal becomes b from earlier i in potb, mati 'pathway', 'mother'. Here i is the reflex of both original im and syllabic nasal m.

It is in the morpheme for 'ten', 'hundred', 'thousand' that Slavic shows what I believe are multiple reflexes of the same syllabic nasal m. From Indo-European syllabic m in *k'mt- we find ę in desęt- 'ten' and r in sito 'hundred'.11 But in the word for 'thousand', we find both ę and o: *tysętj-/*tysotj-.

Baltic uniformly shows im: Lithuanian dešimtis, dešimts "ten", šimtas 'hundred'; Latvian desmit 'ten' (with a later metathesis of im to mi), simts 'hundred'; Prussian tūsimtons 'thousands, dessimpts, dessimton, etc. 'ten'.12

Albanian has only dhjete 'ten' with je from r from syllabic m.

The expected Slavic reflex of Indo-European *k'mt- is *set-. The word for 'hundred', sito, seems to have arisen from an earlier *sumto with u plus nasal as a reflex of syllabic m. The alternation *tysetj-*tysotj- in the word for 'thousand' may seem ambiguous. But it is no more so than *sęt- all by itself.

The o in *tysotj- may be said to come from o-grade allomorph *k'omt-. But then the 'expected' ę in *sęt- as in deset-, *tysętj- may be said to come from an 3-grade allomorph *k'omt-. But then the 'expected' ę in *set- as in desęt, *tysętj- may be said to come from an e-grade allomorph *k'emt-. Both *k'omt and *k'emt- reflexes exist in Indo-European. In Gothic we find -hinpan from *k'emt- 'to catch' and handus from *k'omt- 'hand'. But this morpheme is clearly attested only in zero-grade when it means a number except for Greek, apparently, o-grade - konta (as in triakonta 'thirty').13

Analogy and other morphological processes surely were at work here. The o in *tysotj- may represent a final selection of an o-grade form.14 But, in, at least Pre-Slavic days there may have been merely phonetically caused fluctuations in the reflex of Indo-European zero-grade *k'mt- with a syllabic nasal yielding *k'amt-/ *k'emt-/ *k'omt-/ *k'imt-/ *k'umt, a pattern reflected in a general way in Albanian reflexes of Indo-European syllabic nasals, and also reflected here.

Where reflexes of non-syllabic and syllabic nasals, etc. are expected to occur in Slavic alternations, we find similar problems. Specialists give conflicting interpretations for dimo/doti "I blow/to blow". Is the o in doti from a syllabic nasal m? Or is it an o-grade form?15 Yet nobody doubts a zero-grade origin for the ę in vrzęti 'to take' which alternates with i in vizimo 'I take'. Could this ę not be an e-grade form? Note also the alternation in/i in Old Church Slavonic viniti 'to enter' /višbl- 'entered' with vin- before a vowel but vi- before a consonant. So we find multiple alternations iN: ę, iN: o, iN; r which in some cases most likely represent Slavic multiple reflexes of Indo-European syllabic nasals.

I do not doubt that Baltic influence, starting from Pre-Baltic times when Pre-Slavs first migrated into Pre-Baltic territory from the Carpathians,16 continually influenced Slavic to do away with most of its multiple reflexes of Indo-European syllabic nasals. Having made i or u plus nasal along with i or u plus liquid its more or less regular reflexes of Indo-European syllabic resonants, Slavic, in effect, became a sort of Balticized Albanian. All the same, traces of multiple reflexes of syllabic nasals plus other features, lexical, phonological,17 and morphological (the x, that is, h from -ks- in exati 'to ride'), show an original close Slavic tie with the ancestor of Albanian rather than with Baltic.18

1 This eventually may have become -k's- temporarily.
2 This verbal suffix is usually represented as -sk'-. By active verbs, I mean those that do not indicate status or motion. Verbs of perception like 'hear,' 'see,' 'know' can be active. Thus, the suffix -sk- may appear in these as in Greek gignosko "I know". But a verb like 'send' is fundamentally inactive. It includes the idea of 'going'. And 'go' is an inactive verb normally classified as intransitive. For discussion of active/inactive verbs see Shaban Demiraj, Gramatike historike e ghuhes shquipe, Kombinati Poligrafik Shtypshkronja "Mihal Duri," Tirane, 1986, 723-729.
3 Demiraj (pp. 662, 728) explains this -h- as a hiatus breaker arising under the influence of active verbs like njoh 'to recognize'. But -h- there developed later. First it developed in verbs like dihet from -ks- metathesized from the active verb suffix -sk- to indicate inactive verbs. This new suffix, -ks- as the sole marker of inactive verbs, eventually had a much larger distribution than -sk-. Since -sk- was only a secondary, redundant marker of active verbs, it could metathesize to -ks- without causing confusion, especially after original inactive marker -ks- had changed to -h- for phonetic reasons (assimilation of -ks- to -kh- to -hh- and simplification of -hh- to -h-).
4 Slavic, with allophonic (h0 glottalized away, offers greti, žar from * ger-, and gor-.
5 Note the Germanic delabialization of o to a and tautosyllabic umlaut of e to i.
6 Indic, Iranian, and Armenian also show satem palatalization of k(h), g(h) to, originally k-(h), g'(h).
7> Baltic, unlike Germanic, shows no tautosyllabic e o i. Thus, it would have no confusion of the reflexes of syllabic resonants with those of e plus resonant. Germanic, a centum language like Celtic, which as ri, li from syllabic r, l, could have developed i plus resonant as reflexes of Indo-European syllabic resonants. It developed only u plus resonant to avoid confusion with reflexes of e plus resonant.
8 Similar dual reflexes, ir, or ur, ir or ūr, occur also in Indic under specific conditions. See Manfred Mayrhofer, Sanskrit-Grammatik, Sammlung Goschen 1158/1158a, Walter de Gruyter & Co., Berlin 1965, p. 19.
9 Eqrem Cabej thinks the Albanian reflex of syllabic long r was ar as in pare 'first', Hyrje ne historine e gjuhes shqipe (Fonetika historike e shqipes), Universiteti i Tiranes, Falkulteti i historise dhe i filologjise, katedra e ghuhes shqipe, Tirane, 1975, p. 299. So do Martin E. Huld, Basic Albanian Etymologies, Slavica Publishers, Inc. 1984, p. 68, and Julius Pokorny, Indogermanisches Worterbuch, Francke Verlag, Bern-Munchen, 1959, p. 815. But Gustav Meyer thinks that pare is an o-grade form, Etymologisches Worterbuch der albanesischen Sprache, Verlag von Karl J. Trubner, Strassburg, 1891, p. 321. If the reflex of Indo-European long syllabic r, indeed, is ar in the Albanian word pare, and if Slavic and Albanian are essentially from the same Indo-European dialect, then the Slavic reflex pbrvb from *pirw- for 'first' might have been conditioned by (Pre-) Baltic influence. Thus from possible reflex doublets, etc., *parw-, *pirw-, etc., (Pre-) Slavic, guided by (Pre-) Baltic pirm-, finally settled on *pirw-.
10 See Cabej, Hyrje..., p. 300, Stuart E. Mann, An Albanian Historical Grammar, Helmut Buske Verlag, Hamburg 1977, pp. 25, 37, Karl Brugmann and Berthold Delbruk, Grundiss der vergleichenden Grammatik der indogermanischen Sprachen, Verlag vonWalther de Gruyter & Co., Berlin Leipsig, 1930, pp. 406-407,421.
11 See Maks Fasmer, Etimologičkeskij slovar' russkogo jazyka, Vol. III, Izdatel'stvo "Progress", Moscow 1971, p. 762 where Pisani's explanation is given of *sbto from "k'utom from *dk'u-to-, the reduced grade of *dek'u- 'ten' as in Latin decuria, Gothic tigjus 'ten'. This I consider unlikely. Note that in Latin and Germanic the stem *dek'u-, which seems to occur only there, never appears with a reflex of the suffix -to-. This suffix is, likewise, unmotivated in Slavic to serve better than an -s- as used in Latin and Germanic. Besides, the stem *dek'u- occurs only with the meaning "ten".
12 In Prussian the spelling dessempts also occurs. Universally, though, the form is considered to be *desimts.
13 See Pokorny Indogermanisches..., p. 192 "... (d)k'omt- (plur.) in den Zehnerzahlen (nur die Bildungen als idg. nachweisbar_...griech. triakontan (aus *-konta...), gallo-lat. Abl.Pl. tricontis...bret. tregont (*tri-k'omt-es)..."
14 Roman O. Jakobson during one of his class lectures at Harvard University in the early 1960s mentioned that Slavic often used o-grade morphemes to portray bigness versus their zero-grade allomorphs used to portray small-ness as in the adjectival formants -ok-/rk-: vysokr 'high' /nizrkrM/i> 'low' globokr 'deep' melrkr 'shallow'. Perhaps this morphological marking device influenced the appearance of an o-grade from *tysotj-.
15 The appearance of an o-grade form for the infinitive seems to reflect a tendency to mark nouns that way: bojb 'battle' /brjo 'l 'I beat' /biti 'to beat'. Here the infinitive doti might reflect its being felt more as a noun than a verb.
16 I believe the Carpathians were the final home territory common to Pre-Slavs-Albanians. From there Pre-Slavs migrated north to territory occupied by Pre-Balts while Pre-Albanians migrated south and eventually reached the Balkans.
17 These include at least initial ks- to h- (via kh- to hh-) and more cases of Late Dialectal Indo-European k'(h), g'(h) to k, g than in other satem dialects as well as multiple reflexes of at least Indo-European syllabic nasals and possibly of all Indo-European syllabic resonants at first. Baltic, incidentally, shows no initial ks-, kš- in native words. See Christian S. Stang, Vergleichende Grammatik der Baltischen Sprachen, Universitetsforlaget, Oslo-Gergen-Tromso, 1966, p. 95. Nor does it show h in native forms.
18 In Maks Fasmer, Etimologičeskij..., Vol. II, 1967, p. 10, we find the formant -x- in exati explained as the development of the Indo-European extension -s- which J. Zubaty (LF 28, 35) connects with the sigmatic aorist. Unfortunately, this explanation does not motivate the appearance of this suffix, and, therefore, is, most likely, incorrect.