Volume 34, No. 4 - Winter 1989
Editor of this issue: Antanas Klimas
ISSN 0024-5089
Copyright 1989 LITUANUS Foundation, Inc.



In the latter stages of Proto-Indo-European there was delabialization of o to a and of laryngeal H3, the o-coloring one, to laryngeal H2, the a-coloring one, in some dialects. These included Pre-Hittite and Pre-Germanic. The known earliest descendants of these dialects offer no systematic body of evidence for long and shorts * separate from long and short * in their proto-languages. From Proto-Indo-European dialects, Pre-Hittite and Pre-Germanic, the very first stages as separate languages of Hittite and Germanic inherited a vowel system of long and short





Available evidence indicates that the same was true for Prussian. This means that the latter-stage Proto-Indo-European ancestors of the historic Prussians, as they were beginning to speak a dialect to become Prussian, like Pre-Hittites and Pre-Germanics, decided to delabialize o to a and o-coloring laryngeal H3 to a-coloring laryngeal H 2. Thus Prussian inherited the same vowel system of long and short





The antiquity of this feature is verified by its uniformity in all these languages, Hittite, Germanic and Prussian, as prototypes. Note also that it is, most likely, the oldest dialect-defining phonemic feature of Proto-Indo-European which has survived into the present. It precedes the total physical loss of the laryngeals as separate entities, a loss visible in all descendant Indo-European languages except Hittite. As the first delabialization, it precedes later ones including systematic kw to k and gw(h) to g(h) which is the first set of features separating satem languages from centum ones. Characteristically, this very first delabialization involving laryngeal H3 and the vowel o associated with it occurs in geographically the most northern and southern centum languages, Germanic to the North and Hittite to the South, and in geographically the most remote northwestern, what was to become, satem language, Prussian.

The significant feature of this early dual delabialization was its morphological character. Thus, these were not just simple phonetic changes making some dialects merely slightly harder to understand to those speaking dialects without them. These laryngeal H3 to laryngeal H2 and o to an involved morphology and could be expected to initiate trains of other morphological differences including deviations in lexicon.

The morphological element involved in these phonetic changes of delabialization of o to a and H3 to H2 concerned morphemes with o alternating with e and H, the e-coloring laryngeal. Surely morphs with o/e stood in opposition to those with a, while morphs with H3/H1, stood in opposition to those with H2. Now, with these delabializations of o to a and H3 to H2some confusion was caused. Morphs with a, or H2, no longer stood apart discretely from morphs with a/e from original o/e and morphs with H2/H1 from original H3/H1. Because of this, Pre-Hittite, Pre-Germanic, and Pre-Prussian, very likely, were especially prone to neologisms, including borrowings, to relieve the pressures of misunderstandings caused by homonyms.

Because of the above, it is not surprising that Hittite was so subject to foreign lexical influences. Germanic and Prussian retained original elements better because of their early relative isolation.

Another indication of the morphological element of the phonetic change of the laryngeal is the latter replacement of an intersyllabic laryngeal, that is, possible, a later schwa, either with i in Indie and Iranian or with a in Non-Indie and Non-Iranian.1

The antiquity of this first Proto-Indo-European delabialization involving o and its closely linked o-coloring laryngeal, H3, should be clear from the evidence in Hittite. Only there in this anciently attested Indo-European daughter language do we find some consistent, relatively direct physical traces, h*, h*h*, of the laryngeals. Yet, at the same time, there is no trace of o whatsoever in Hittite. Consequently, this delabialization of o to a and H3 to H2 had taken place in prehistoric, I say, Pre-Hittite dialectal Proto-Indo-European times. Since this seems fairly clear for Hittite, I suggest that the same was true for Germanic and Prussian.

As for Germanic, a language usually lumped together simplistically with Celtic, Italic, and Greek as Western Centum Indo-European, there is no internal evidence whatever for long and short o separate from long and short § in its earliest stages. Celtic shows later partial merging of long o with long a, but, apparently, no merger of short o with short a (which resembles Tokharian's total long o, a merger versus its total short o, a separation). But Germanic internal evidence does not permit specialists in any way to determine which merger came first, long 6 with long a, or short o with short a even when they believe in the existence of long and short o separate from long and short a in primitive Germanic. I say this indicates that this separation from the start failed to exist even there.

Similarly, in the satem languages systematic evidence, sometimes obscure, for the early separation of long and short * from long and short * exists everywhere except in Prussian. In Armenian it is relatively clear for the long and short vowels. In Albanian, Latvian and Lithuanian where the short vowels fell together completely to a, consistent evidence for long separate from long allows specialists via morpheme alternations with short a versus long , and short a versus long to reconstruct from short a either short a or short o. In other words, in ancient Indo-European daughter languages, the existence of a long separate from a long implies the existence of short counterparts, at least, in Late Dialectal Proto-Indo-European ancestors, as in Pre-Albanian, Pre-Latvian, and Pre-Lithuanian since the longs were based on the shorts, that is, via compensatory lengthening of short o or a followed by a disappeared laryngeal. Through somewhat similar processes the reverse can be done for Indie and Iranian. There through Brugmann's law ancient short o, separate from short a, can be reconstructed from long before l, r, m. n followed by a vowel, that is, in an open syllable (though this is verified by comparing cognate morphemes in other ancient Indo-European daughter languages which separate reflexes of Proto-Indo-European short o from short a), at least, for Late Dialectal Indo-European Pre-lndic and Pre-lranian. With this clearly demonstrable and thus certain short o, others can be derived in other morphemes through comparisons with cognates in short o-differentiating ancient Indo-European daughter languages. Then, via alternations of these reconstructed short o with long , original long o can be reconstructed from long . Thus, either existence of early separate long allows us to reconstruct early short o, or existence of early separate short o allows us to reconstruct early long in these languages. Reconstructing separate long and short * in Slavic can be done only very indirectly through the unique almost absolutely consistent fronting of long and short back vowels whereby one can reconstruct a system with j plus long and short high vowels *, * becoming, respectively, long and short j*, j* versus j plus isolated long remaining unchanged at one time in Early Common Slavic.2- (Though later j*> j*, j* > j* in Iranian and ja > je in Prussian occur, they are sporadic and, therefore, do not offer the same interpretation as the Slavic changes.)

A glance at the geographical positions of Hittite versus those of Germanic and Prussian would make an early genetic linguistic kinship involving all three seem unlikely. Though specialists might be willing to consider one between Prussian and Germanic, they would hardly be expected to embrace one which would also include Hittite. But geographically all these groups, indeed, do have something in common - their northern or southern peripherality with respect to the rest of Indo-European.

Specialists believe that with respect to centum versus satem the centum languages with reflexes of k, g, gh as opposed to those of k*, g*, g*h* of the satem ones are more conservative and are, at the same time, geographically peripheral. But the earlier delabializations of o to a and H3 to H2 characteristics of Pre-Hittite, Pre-Germanic, and Pre-Prussian demonstrate that, generally speaking, northern and southern peripheral dialects were originally innovative. The extreme northern versus southern peripherality of these dialects should suffice to indicate that, most likely, they never amalgamated to form any special genetic linguistic prototype apart from the rest of Indo-European. The relative paucity of special early lexical correspondences between all of them as a whole, or even, groups of them in particular, beyond a few concordances between Germanic and Prussian, supports this view. Yet I have reason to suspect that another special connection between these dialects developed, one of limited inter-dialect borrowing.

A possible cause of the location of the Pre-Hittite, Pre-Germanic, and Pre-Prussians on the northern and southern expulsion of the Indo-European area could have been their temporary expulsion from the general community. Either as a result or as a cause of this expulsion delabialization of o to a and of H3 to H2 occurred. The change of H3 to H2, as I already mentioned, had more special morphological than mere phonetic significance. In light of this, the inclusion of o to a following in its wake can also be seen as primarily morphological. If the pronunciation of certain ritualistic terms was viewed by most Indo-Europeans as "distorted", those doing it might well have been expelled from the community. Or, if expelled for other reasons, outcast Pre-Hittites, Pre-Germanics, and Pre-Prussians might have, in defiance, produced willfully, that is, quite consciously, these delabializing distortions not only on meeting other Indo-Europeans, but also on meeting one another. Though contacts between speakers of all three aboriginal a-dialects (which with the loss of laryngeals later became *-languages), Pre-Hittite, Pre-Germanic, and Pre-Prussian, no doubt, were inhibited by the geographical factor of remoteness from one another, still they tended to be encouraged by social factors of easier acceptance and common opponents in the overall Indo-European community. I imagine that as a sign of these social factors, the speakers of the aboriginal a-dialects tended to use these delabializations deliberately when they met for purposes of identification (partly, as if to imply that underdog groups were supporting underdog alternations, since any involving e with o were, so to speak, "top dog"). Thus, if not all, then at least the leaders of these a-speaking groups might well have belonged to a cult of some kind. This, rather than any other form of closeness, would then be the underlying, unifying cause of this linguistic change.

Most likely, the speakers of Pre-Germanic and Pre-Prussian of the aboriginal a-dialects had the closest, most sustained contacts with one another. An example of this, very likely, is the exclusive correspondence in the tendency to generalize the n-stem in the word for 'fire', fon, funins (Gothic)/ panno (Prussian) with a Grimm's Law correspondence, f/p to date it as, at least Pre-Germanic. (The Prussian word panno shows the delabialization of o to a in its root.) The evidence does not seem to show, though, that speakers of Pre-Hittite had such close contacts with those of Pre-Germanic or Pre-Prussian.

Eventually the Pre-Prussian re-established relatively early contacts with aboriginal o-dialect speaking Indo-Europeans. Phonetic evidence suggests that these included Pre-ltalics, Pre-Celts, Pre-Latvians, Pre-Lithuanians, and Pre-Albanians as well as Pre-Slavs. Presumably this happened when the laryngeals were disappearing as consonants. A sign of these contacts more or less direct is the agreement of these groups on simple deaspiration of pH, tH, kH, kwH (with H standing for any laryngeal) to, originally, p, t, k, kw. Pre-lranian deviated with, originally, f, Θ, x, xw. But its agreement with these groups on what I call the second Late Dialectal Indo-European full deaspiration3 of bh, dh, gh, gwh to b, d, g, g(w), a change which now did not include Pre-ltalic, suggests its later ties with, at least, some of them. Pre-Celtic with not only bh, dh, gh to b, d, g, but also gwh to g was, very likely, the first Indo-European dialect to start what I call the second Indo-European delabialization.4- Thus when Pre-Celtic deaspirated gwh, it simultaneously delabialized it. This is not apparent in any of the deaspirating satem dialects where, unlike Celtic, the reflexes of gwh fell together with those of gw into g.5- Other features in Celtic only of the centum languages like Indo-European syllabic r*, l* to ri, li, which make it resemble a satem one, show the extent of this first major Late Indo-European dialect cluster which included languages in geographically contiguous areas,6 but of somewhat deviaing earlier development (Pre-Prussian, an aboriginal a-dialect/the others, aboriginal o-dialects) and later developments and dialect cluster associations.

This new Indo-European dialect cluster united firmly by the second Late Dialectal Indo-European full deaspiration of bh.dh, gh, gwh to b, d, g, g(w), in addition to the first full one of pH, tH, kH, kwH to p, t, k, kw, essentially phonetic changes, was quite different from the old associations between Pre-Germanic, Pre-Prussian, and Pre-Hittite loosely connected by a from o and H2 from H3. This new cluster, because it consisted of neighboring dialects, was far more dynamic since it could provide far more opportunities for interdialectal exchanges of all sorts ranging from mutual support for the maintenance of old inherited elements to the spread via all kinds of borrowing, including calking, of innovations.

Since deaspiration and delabialization were closely associated processes leading to the dissolution of the Indo-European proto-language, both full deaspirations were bound to lead very soon, if not immediately, to delabialization tendencies. Pre-Celtic with its simultaneous deaspiration and delabialization of gwh to g behaved in a conservative manner by following the tradition set by Pre-Hittite, Pre-Germanic and Pre-Prussian in simultaneously deaspirating and delabializating. This means that by delabializing H3 to H2 they reduced their number of laryngeals and, as far as their sound systems were concerned, in effect, deaspirated. Beyond this occurrence, though, the association between delabialization and deaspiration generally weakened in the Indo-European dialects by the lime of the first full deaspiration. When H3 deaspirated, its labializing effect on e did not disappear immediately, if at all, in the o-aboriginal dialects. And even with g"h to g and an array of subsequent delabializations like Post-Indo-European ri, li, presumably from ur, ul from Indo-European syllabic r*, l*7 and the Post-Indo-European delabialization of long to long , Pre-Celtic, despite these drifts inherited from an earlier bond with what wast o become satem dialects, disassociated itself enough from its eastern neighbors to resist finally the second Indo-European delabialization and all the other changes caused by it so that Celtic now is known as a branch of the centum languages.

On the other hand, Pre-lndic and Pre-Armenian, dialects to the East of the bh, dh, gh, gwh to b, d, g, g(w) deaspirating ones, now entered into closer association with them so that they also participated in the second Indo-European delabialization of k", gw to k, g. But since they had resisted, or, simply, not experienced the second Indo-European full aspiration of bh, dh, gh, gwh to b, d, g, g(w) , which had, in the dialects that had undergone it, given impetus to this delabialization in the first place, they, of course, show gwh to gh. These dialects added from the East seem to have participated somewhat more fully in all the later changes connected with this delabialization including a more consistent adherence to palatalization of velars k, g, gh to k*, g*, g*h with more extensive subsequent assibilation to s', z', z'h, and, in the case of Indie, more consistent application of the ruki law of i, u, r, k, + s to i, u, r, k +7- so that they, rather than most of the original western dialects like Pre-Albanian, Pre-Prussian, Pre-Latvian, Pre-Lithuanian, seem more typically "satem".

Considering all the above shifts in the history of dialectal Indo-European affinities, it seems very dangerous to adhere to any proposal involving any sort of genetically related Balto-Slavic, or Common Baltic protolanguage, or even any smaller common Indo-European dialect consisting of any combination with Pre-Prussian, or Pre-Latvian, or Pre-Lithuanian, or Pre-Slavic, such as "Slavo-Prussian",8 as a separate unit.

Especially since most of the similarities between these types are of such relatively late origin.9-

Consider merely the following facts, the first two of which are real linguistic results, the last two of which emphasize striking membership features of the very composition of the fully deaspirating Indo-European dialect cluster: 1. The significantly larger list of Albanian native cognates in the Russian etymological dictionary than in the Lithuanian one,10 an important indication that the Pre-Slavs had somewhat closer contacts with the Pre-Albanians than the Pre-Lithuanians, Pre-Latvians, and Pre-Prussians did, especially when seen against the background of the reverse situation with Hittite and Tokharian lists which are presumably of a vintage equal to the Albanian ones. 2. Pre-Prussian's, Pre-Latvian's and Pre-Lithuanian's essentially complete merger (which converted these Indo-European dialects to Prussian, Latvian, and Lithuanian, independent daughter languages) of their respective reflexes of the ruki law with those of assibilating Indo-European palatals k*, g* (from k, g from k, g, gh) which neither Slavic, nor Albanian, nor any other satem language shows. This, though with different quality of merger, namely, Lithuanian , versus Latvian and Prussian s, z, presumably from original s', z' for Latvian and Prussian.11 3. The evidence of Pre-Celtic's having been a member, even a pioneering one, of the cluster of fully deaspirating (kH to k, gh to g, g wh to gw, etc.) Indo-European dialects at its inception, and its subsequent departure from it. And finally, 4. Pre-Prussian's, an aboriginal a-dialect's differing in this very feature from aboriginal o-dialects from the very beginning of the documentable onset of the break-up of Proto-Indo-European into recognizable dialects and yet becoming and remaining a member of this cluster of fully deaspirating dialects.

Thus we find the permanent entrance into this cluster of a dialect which from the documentable very beginning of the break-up of Proto-Indo-European into recognizable dialects was noticeably different from the others. Unlike Pre-Celtic, Pre-Armenian, and Pre-lndic which either wandered in and out or joined relatively later, Pre-Prussian joined this cluster early, probably at its beginning, stayed, and formed special later closer clustering associations with Pre-Latvian, and Pre-Lithuanian. Though Pre-Prussian seems to have been extremely heavily influenced by the other dialects of this cluster over the centuries, it possibly also had some influence on them, too. Thus it might well have influenced the others, including, indirectly, Pre-lranian and Pre-lndic, to lean toward delabializing, if not always long o to long a, then later, more consistently, short o to short a, a process Pre-Prussian had surely kept productive.

We must always keep in mind that this ancient fully deaspirating Indo-European dialect cluster was by no means a monolith. Various of its members or groups of its members were at different times subjected to influences from and shared developments with non-cluster dialects. Often such influences could well have extended over long periods and not been direct on all of those affected. An example of this are the special developments of affinities of Lithuanian, Latvian, Prussian, and Slavic with Germanic like -m- rather than -b- (from -bh-) as the dative plural marker. At the very earliest, this particular morphological feature dates from the Pre-Germanic and Pre-Prussian association when they were the northern aboriginal a-dialects of Indo-European. Its subsequent spread from Pre-Prussian and possibly later Pre-Germanic contacts to Pre-Latvian, Pre-Lithuanian, and Pre-Slavic, at the very earliest, dates from the Pre-Prussian entry into the fully deaspirating Indo-European dialect cluster. Pre-Slavic, most likely, was the last dialect to acquire it. Slavic's special native cognate lexical ties with Albanian (which does not have this dative plural -m-) which suggest the southernmost position for Pre-Slavic, that is, the one farthest from Pre-Germanic, give that impression. Later phonetic developments closer to Albanian ones show that Pre-Slavic as a linguistic unit must have been developing its own drifts before coming under this Pre-Baltic-Germanic influence, drifts, no doubt, having been receiving sustained support from more ancient, deeper associations with Pre-Albanian which seems, from the earliest times till then, to have been its nearest neighbor. Very likely, the spread of this dative plural -m- to Pre-Slavic marks the beginning of the intensive neighborly ties between the Slavic and Baltic languages resulting in extensive solely morphological, that is, not phonetic, exchanges between them.12 Thus today Baltic (a term I use with no protolanguage inferences to refer to Lithuanian, Latvian, and Prussian) and Slavic to many seem to have emerged from a special exclusive protolanguage with look-alike formations like Lithuanian gyvastis 'life' versus Old Church Slavonic ivost "livelines' with matching suffixes -astis = -osti, Latvian iziet versus Old Church Slavonic iziti 'to emerge' with matching prefixes iz- = iz-, Prussian sinda(n)ts versus Middle Common Slavic *sendan(t)s 'sitting' with matching infixes -n- = -n- in the roots si-n-d-: *se-n-d-, and even Lithuanian vãrnas: várna versus Middle Common Slavic *vamas : *várn (with várna = *várn from *vrn from earlier *va-a-rn) 'raven : crow' with matching phonemic pitch oppositions circumflex : acute = circumflex : acute stemming from matching acutes in a shortened tautosyllabic long vowel, which had arisen from contraction of the originally short root vowel -a-with a like-sounding infix -a- affixed after it in the word for 'crow', várna = *várn, in both languages. All these developments, though, were really only lexical borrowings via calking.13-

Thus the many look-alike forms in Slavic, Prussian, Latvian, and Lithuanian are extremely unreliable as indications of any sort of special common ancestor. This becomes clear when seen in the context of the over-all historic picture of shifting Indo-European dialect affinities. Just following the case of Prussian from beginning to end alone, we see a succession of shifting associations from aboriginal-a times with Pre-Germanic to early fully deaspirating times with Pre-Celtic to later closer associations with Pre-Latvian and Pre-Lithuanian, especially noteworthy at the end of the second delabialization period (kw to k, k to k*, k to /s', and the ruki law) with similar changes on a phonetic level and none shared with Pre-Slavic alone when Pre-Slavic participation was involved. In light of all this, it seems extremely unwarranted to accept the concept of any sort of genetic Common Baltic or, especially, Balto-Slavic linguistic unity.



1 See Thomas Burrow, The Sanskrit Language. London, 1959.
2 Harvey E. Mayer, "Kann das Baltische als Muster fur das Slavische gelten?", Zeitschrift fur slavische Philologie. 39 (1976), 32-42.
3 The Aboriginal-a delabialization of H3 to H2, an Early Dialectal Indo-European feature, dialectally reduced the number of laryngeals and, therefore, counts as, systemically speaking, a partial deaspiration. I consider the general loss of laryngeals as consonants the first Late Dialectal Indo-European full deaspiration.
4 The changes o to a, H3 to H2 count to me as the First Indo-European delabialization.
5 Pre-Celtic had b, not g, as the reflex of gw. Since an almost equal number of Indo-European dialects deaspirated and/or delabialized, and Pre-Celtic with g"h to g demonstrably did so simultaneously, we may conclude that the second Indo-European delabialization followed the second Indo-European full deaspiration very soon. The second Indo-European full deaspiration, representing one of several adjustments to the first, extends far beyond the total deaspiration of bh, dh, gh, g"h to b, d, g, gw and the dialect cluster which did this. It includes Grassman's Law of Th - Th to T - Th (T = any consonant) for Greek and Indie, Grimm's Law of p to f, b to p, bh to b, etc. for Germanic and similar phenomena in Armenian, fluctuations in aspirations in several stems (like Latin faber 'artist', Armenian darbin 'smith' from 'dhabhro - versus German tapfer 'brave', English 'dapper' from *dhabro-), and the reduction and simplification of stops' series from p, b, bh to p, etc. in Tokharian. Because of the much greater extent of all this than of the features and dialects connected with the delabialization of the labiovelars including the palatalization of k, g, gh to k*, g*, g*h and the ruki law (i, u, r, k + s to i, u, r, k + /s'), it seems clear that the second Indo-European full deaspiration preceded the second Indo-European delabialization.
6 The Pre-Celts very likely at this time occupied territory south of the Pre-Germanic one and west of the Pre-Prussian, Pre-Latvian, Pre-Lithuanian, Pre-Slavic, and Pre-Albanian ones.
7 I suggest this in "Zur fruhen Sonderstellung des Slavischen." Zeitschrift fur slavische Philologie. 42, 2(1981), 300-314.
8 Mayer, "Was Slavic a Prussian Dialect?" Lituanus. 32, 2(1987), 5-21.
9 With the lack of special phonetic developments of the class of Grimm's Law to unite these linguistic types, matching derivational processes should be discounted as proof of special genetic relationships. Whether they include flexion or any other affixes in common roots, they may merely represent calking or other interlingual or inter-dialectal borrowing processes that do not indicate a prototype origin. Even roots in the same grade are suspect as proof of special kinships since they might be borrowings.
10 See Maks Fasmer Etimoloaicaskij slovar' russkogo jazyka. Moscow. 1964-1973 and Ernst Fraenkel, Litauisches etymologisches Worterbuch. Heidelberg-Gottingen, 1962-1965. A careful study of these lists shows that native Slavic-Albanian cognate root comparisons with no Baltic representation outnumber Baltic-Albanian ones with no Slavic representation, at least, by 24 to 19, and at most, by 29 to 19. Thus if the 6 conditional Slavic-Albanian-minus-Baltic comparisons prove correct, then Slavic-Albanian-minus-Baltic native cognate roots outnumber Baltic-Albanian-minus-Slavic ones by approximately 1/3.
11 Mayer, "Was Slavic...'
12 Shared changes on a phonetic level are more reliable for proving linguistic genetic kinship than those on a morphological level which includes lexicon, something readily borrowed from language to language even though those languages may not be related at all.
13 In my article "Zum Lexikon und der Frage des Balto-Slavischen", to appear in the Zeitschrift fur slavische Philologie. which I produced from my paper "On Lexicon and the Balto-Slavic Question' presented at the Eight Conference on Baltic Studies of the AABS (Association for the Advancement of Baltic Studies) in 1982 I derived many of the matching Baltic and Slavic pitch oppositions from calks of this sort where like-sounding vowel infixes were injected into morphemes. It seems that originally, if the morphemes were themselves affixes like verbal suffixes, etc., the long vowel resulting from contraction received a circumflex as it does in Lithuanian nete 'you carried' with unstressed -- indicating an originally circumflex pitch on it, a form matched by -- (jat) in Old Church Slavonic nesaste "you were carrying'. But, originally, in a root the long vowel from contraction was arbitrarily assigned an acute pitch following the pattern where Indo-European long vowels from laryngeal loss normally received an acute as in Lithuanian duoti (with Go from Indo-European long o) 'to give' versus dav 'he gave'. These acutes became phonemic only after tautosyllabic long vowels bearing them were shortened so that we get genitive plural varn: varn 'of crows : of ravens'. Since affixation (here, more specifically, infixation) was the ultimate cause of pitch oppositions in Baltic and Slavic, and since affixation involves the positioning of morphological elements, the function of syntax, it is clear that these pitch oppositions were syntactic in origin. Unlike phonetics, syntax, like lexicon is easily calked from language to language. Therefore, Baltic and Slavic accentuation comes not from a Post-Indo-European Balto-Slavic proto-language, but from a language (or, at least, dialects)-in-contact situation.