Proto-Balto-Slavic language

Proto-Balto-Slavic
Geographic distribution:	northeastern Europe
Linguistic classification:	Indo-European Proto-Balto-Slavic
Subdivisions:	Balto-Slavic languages —

Proto-Balto-Slavic is reconstructed proto-language descending from Proto-Indo-European (PIE) and out of which all later Balto-Slavic languages (represented by Baltic and Slavic branches) and dialects descended, such as modern Latvian, Lithuanian, Polish and Russian.

The Proto-Balto-Slavic language is not directly attested by any surviving texts but has been reconstructed using the comparative method. There are number of isoglosses that Baltic and Slavic languages share in phonology, morphology and accentology, which represent common innovation from Proto-Indo-European times, and which can be chronologically arranged.

Phonology

Proto-Indo-European phonological system has exhibited several significant changes in Balto-Slavic period:

• the three series of PIE stops were reduced to two series (voiced and unvoiced)
• PIE syllabic sonorants were substituted with sequences of a short vowel (*i or *u) and a non-syllabic sonorant
• the three PIE laryngeals merged into one (*H), which may have disappeared even during the Balto-Slavic period
• the complex system of PIE dorsals was simplified due to the delabialization of labiovelars and the change of PIE palatovelars into fricatives

Stops

PIE voiced and unvoiced stops were preserved in Balto-Slavic and Proto-Slavic, and aspirated series was deaspirated.

Winter's law was still operable when there was phonemic distinction between the series of plain and aspirated voiced stops. As a result of Winter's law, the distinction between those two series has been indirectly preserved in Proto-Balto-Slavic, because Balto-Slavic vowel would lengthen before a plain voiced stop, but not before an aspirated stop, this occurring probably only if the stop was in syllable coda (i.e. in closed syllable).

On the basis of relative chronology of sound changes it has been ascertained that Winter's law acted rather late, after some other less prominent Balto-Slavic changes occurred, such as after the disappearance of laryngeals in prevocalic position. Compare:

• PIE *eǵh₂om > PBSl. *eźHam (by Winter's law) *ēźHam > PSl. *jāzun (OCS azъ, Slovene jaz)

Therefore, the merger of PIE aspirated and plain velar stop series was one of the last common Balto-Slavic sound changes.

Dorsals

Three series of PIE dorsals (velars, palatovelars and labiovelars) merged to two series in Balto-Slavic: velars and palatovelars. PIE labiovelars lost their labialization in Balto-Slavic, just like they did in the Indo-Iranian, Armenian and Greek branches. Unlike some other Indo-European languages, Balto-Slavic labiovelars were delabialized unconditionally and at once, leaving no noticeable direct or indirect traces.

There are a number of words in Balto-Slavic which show Centum reflex of PIE patalalized dorsals. A number of these can be explained by regular sound laws, although some of these laws have been obscured by numerous analogical developments. Others are argued to be borrowings from Centum languages, e.g. Proto-Balto-Slavic *kárwā 'cow' (Lith. kárvė, OCS krava, Russ. koróva) was likely borrowed from Proto-Celtic *karawā, which in turn is a regular reflex of PIE *ḱerh₂weh₂.

PIE palatovelars could also depalatalize in Balto-Slavic. Several depalatalization rules for Balto-Slavic have been proposed.^[1] According to Matasović,^[2] the depalatalization of palatovelars occurred before sonorant followed by a back vowel: K' > K/_RV^back. That would explain Centum reflexes such as:

• Lithuanian akmuõ, Latvian akmens and OCS kamy would have regular /k/ as opposed to Sanskrit áśmā < PIE *h₂eḱmōn 'stone'
• OCS svekry < PIE *sweḱruh₂ 'mother-in-law'
• Old Prussian balgnan < PIE *bʰolǵʰno- 'saddle'

PIE palatovelars */ḱ/, */ǵ/, */ǵʰ/ turned to Balto-Slavic fricatives: */ś/, */ź/ and */źʰ/, this latter one becoming merged with */ź/ after the loss of contrastive aspiration. They probably had an intermediate stage of affricates */ć/, */đ/, */đʰ/,^[3] and it's conceivable that they may have still been affricates at the Balto-Slavic stage. However, by applying usual methods of reconstructions on Baltic and Slavic languages, fricatives */ś/ and */ź/ represent the most likely phonological interpretation of the reflexes of PIE palatovelars.^[4]

Laryngeals

Reflexes of PIE laryngeals */h₁/, */h₂/, */h₃/, which represented 3 different phonemes in PIE, became merged in Balto-Slavic to a single */H/. Laryngeals disappeared in the Balto-Slavic period over a very long period. No Balto-Slavic language has preserved them, but relative chronology of sound changes shows that they were not lost at once in all positions in a word.

The Balto-Slavic laryngeal was especially durable in a position before a vowel; PIE *tn̥h₂u- 'thin' (Latin tenuis, Sanskrit tanú) was in Balto-Slavic reflected as *tunHu-, and only then as Proto-Slavic *tunu-ku/*tinu-ku (OCS tьnъkъ, Russ. tónkij, Pol. cienki), which shows that the loss of laryngeals in Balto-Slavic occurred after the development of vocalic prothesis in Balto-Slavic syllabic sonorants.

In a syllabic position (between consonants), laryngeal disappeared if it was in the second syllable, but in the first syllable it was preserved as */a/. Compare:

• PIE *(h₁)rh̥₃deh₂ 'heron, stork' > (Ancient Greek erōdiós, Latin ardea) Proto-Slavic *radā > Common Slavic *roda (Serbo-Croatian róda)
• PIE *sh̥₂l- (oblique case stem of *seh₂ls 'salt') > OPr. sal, Proto-Slavic *sali (OCS solь, Pol. sól, Russ. sol´)

Loss of laryngeals in syllabic position occurred probably in early Balto-Slavic period. Compare:

• PIE *dʰugh₂tḗr > PBSl. *duktēr > Lith. duktė, OPr. duckti, Proto-Slavic *duktī or *duktej (OCS dъšti, Russ. doč’)

The same phenomenon happened in Germanic and Celtic, which indicates that it might have been a dialectal isogloss in Late Proto-Indo-European.

Fricatives

PIE */s/ has been preserved in Balto-Slavic and Proto-Slavic in most of the positions; it changed to Balto-Slavic */š/ according to the RUKI law, and in Proto-Slavic it was probably lost word-finally. No attested Slavic language has word-final *-s preserved.

Before voiced stops */s/ manifested as */z/ in Balto-Slavic. That */z/ came to be phonologically distinctive in Slavic after the transition of Balto-Slavic */ź/ (a reflex of PIE */ǵ/ and */ǵʰ/) > Proto-Slavic */z/.

As a result of RUKI law, Proto-Slavic has */š/ before front vowels (*/e/, */i/), */x/ before back vowels and */s/ before consonants. That distribution is most probably a result of series of changes:

1. PIE */s/ > */š/ after */r/, */u/, */k/, */i/
2. */š/ > */s/ before consonants, */š/ > */x/ before vowels
3. */x/ > */š/ before front vowels (Slavic first palatalization of velars)

RUKI rule also operated if there was a laryngeal after */u/ or */i/, i.e. */s/ changes to */š/ after *uH and *iH, but it remains open to debate whether the laryngeal was already lost in that environment, i.e. are we dealing with the change of */s/ to */š/ after Balto-Slavic */ū/ and */ī/.

In Baltic languages the evidence of RUKI rule is recognizable only in Lithuanian, because in Latvian and Old Prussian a merger occurs of Balto-Slavic */š/ (< PIE */s/ by RUKI rule), */ś/ (< PIE */ḱ/) and */s/ (< PIE */s/). In Lithuanian, Balto-Slavic */š/ and */ś/ are merged to /š/, which remains distinct from /s/ so the effect of RUKI rule is still evident in Lithuanian.

Most handbooks, on the basis of Lithuanian material, state that in Baltic RUKI law has been applied only partially. The most common claim is that Balto-Slavic */s/ turned to */š/ in Baltic unconditionally only after */r/, while after */u/, */k/ and */i/ we have both */s/ and */š/. Compare:

• Lith. aušrà; Latv. ausma < PIE *h₂ewsro- (cf. Latin aurōra, Sanskrit uṣás) with RUKI applied vs.
• Lith. ausìs; Latv. auss < PIE *h₂ews- (cf. Latin auris, OCS uxo) with */s/ unchanged.

Similarly, Lith. maĩšas/Latv. maiss "sack" completely matches etymologically with OCS měxъ and Sanskrit meṣá, but in the word teisùs "correct" */s/ has been preserved while in Slavic there is */x/ < */š/ in accordance with RUKI rule (OCS tixъ, Russ. tíxij 'quiet, peaceful').

There is no simple solution to such double reflexes of PIE */s/ after */r/, */u/, */k/, */i/ in Baltic, and thus no simple answer to the question of whether RUKI law is a common Balto-Slavic isogloss or not. The most probable seems the assumption that PIE */s/ was changed to */š/ after */r/, */u/, */k/, */i/ completely regularly in Baltic, just like in Slavic, but the traces of the effect of RUKI law were erased by subsequent changes, such as the change of word-final *-š to *-s.

Generally it can be ascertained that Baltic shows the effect of RUKI law only in old words inherited from Balto-Slavic period, meaning that Lithuanian /š/ will come after /r/, /u/, /k/, /i/ in words that have complete formational and morphological correspondence in Slavic (ruling out the possibility of accidental, parallel formations).

Unlike Indo-Iranian, where the change */s/ > */š/ also occurred after the palatovelar */ḱ/, it is possible that palatovelars yielded fricatives in Balto-Slavic even before the effect of RUKI law. Compare:

• PSl. *desnu 'right' (OCS desnъ, Russ. désnyj, SCr. dèsnī, Lith. dešinys) < PIE *deḱs-no- (Lat. dexter, Skt. dákṣiṇas)
• PSl. *asi 'axle, axis' (OCS osь, Russ. os’, SCr. ȏs, Lith. ašis) < PIE *h₂eḱsi- (Lat. axis, Skt. ákṣas)

By satemization of PIE dorsals and the merger of PIE laryngeals, Balto-Slavic has significantly modified the system of PIE fricatives. After the merger of PIE voiced and aspirated stop series, Balto-Slavic system of fricatives had the following shape:

*/s/	*/z/
*/ś/	*/ź/
(*/š/)
*/H/

Phonological relevance of the consonants */š/ and */z/ is disputed; it cannot be known whether they were phonologically predictable allophonic variants of */s/ and */z/ in all environments.

Phonological interpretation of the laryngeal */H/ is also disputed; on the basis of typological considerations it can be ascertained that the Balto-Slavic laryngeal was probably a voiceless glottal fricative [h] or a glottal stop [ʔ].^[5]

Sonorants

Proto-Indo-European sonorants */w/, */j/, */l/, */r/, */m/, */n/ were preserved in Balto-Slavic and Proto-Slavic, as they were in most other Indo-European branches. From context-conditioned sound laws, notable is the disappearance of word-initial PIE */w/ before */r/ and */l/ (so-called Lidén's law).

PIE */w/ was retained in Balto-Slavic and Proto-Slavic as a bilabial semivowel (glide), but in Lithuanian and most Slavic languages it has eventually changed to labiodental fricative /v/.

PIE */m/ changes to */n/ word-finally in Balto-Slavic period; in Old Prussian there is a clear attestation of that change e.g. in nominoaccusative of neuters (cf. OPr. assaran 'lake' < PIE *eǵʰerom). In Slavic however that change of *-m > *-n is indirect because in Common Slavic period all word-final vowels were dropped. It becomes more clear in sentence sandhi conditions due to which earlier *kom emōj yielded Proto-Slavic *kan jemъ (OCS kъ n'emu), and not **ka memō.

Syllabic sonorants

PIE */i/ and */u/,^[6] syllabic allophones of PIE glides */y/ and */w/, have been preserved as vowels in Balto-Slavic. Before laryngeals they yielded long vowels *iH > */ī/, *uH > */ū/.

PIE */u/ has lengthened into Balto-Slavic */ū/ when followed by */n/ which was followed by a stop. In Slavic *-n- later drops regularly. Compare:

• PIE *Hunk 'to get used to' > PBSl. *ūnk > Lith. jùnkti, Latv. jûkt, OCS vyknǫti, Upp. Sorb. wuknyć

PIE */i/ did not exhibit lengthening in such conditions, as older literature often states.^[7]

In a syllabic position, PIE sonorants */l/, */r/, */m/, */n/ have twofold reflexes in Balto-Slavic, differing in a prothetic vowel: *iR and *uR (where symbol R denotes any of aforementioned sonorants). Analysis of their distribution has shown that the reflexes of type *iR are much more common. It has remained an unsolved mystery to this day which exact phonological conditions trigger which reflex.

Several theories have been proposed, most notable being one by André Vaillant.^[8] According to him, *uR reflexes arose after PIE labiovelars. If this was true, it would be the only trace of PIE labiovelars in Balto-Slavic.

Similarly, Jerzy Kuryłowicz thought that *uR reflexes arose after PIE velars, and also notable is also older opinion of Jānis Endzelīns and Reinhold Trautmann according to whom *uR reflexes are the result of zero-grade of morphemes that had PIE */o/ (> PBSl. */a/) in normal-grade. Matasović (2008) proposes following rules:

1. PIE syllabic R > PBSl. *əR
2. *ə > */i/ in a final syllable
3. *ə > */u/ after velars and before nasals
4. *ə > */i/ otherwise

Vowels

Balto-Slavic preserved Late PIE vowel system, after the effect of "laryngeal colouring". The only exception is the change of PIE */o/ > PBSl. */a/, which is an isogloss shared with the Germanic and Anatolian branches.

Proto-Slavic preserved Balto-Slavic system of short vowels unchanged, but merged PBSl. */ō/ and */ā/ yielding PSl. */ā/, while the difference between these long vowels was preserved in Baltic.

PIE */e/ changes to PBSl. */a/ before */w/ in heterosyllabic position, i.e. */e/ > */a/ / _wV. Compare:

• PIE *néwos 'new' > PBSl. *nawas > OCS novъ, Lith. naũjas

It appears that in some cases in Balto-Slavic period initial *(H)e- and *(H)a- were mixed. That change, called Rozwadowski's rule by some, is based on the cases where Balto-Slavic has initial *e- in etymons which in PIE had initial *(H)a-, *(H)o-, *h₂e-, *h₃e-. Slavic has preserved some relics of initial *e-, *a- alternations. Compare:

• PSl. *elawa, *alawa (Common Slavic *olovo) 'lead' > Middle Bulg. élav, Pol. ołów, Russ. ólovo as opposed to OPr. elwas 'tin'

Similar to vowels, PIE diphthongs were preserved in Balto-Slavic, with the exception of *ow (and also *h₂ew and *h₃ew), which yielded PBSl. *aw. Later in Proto-Slavic PBSl. *aw (< PIE *ow, *aw, *h₂ew, *h₃ew, *How) > PSl. */ō/, which was reflected as /u/ in all Slavic languages.

Relative chronology of sound changes

Austrian Balto-Slavist Georg Holzer has reconstructed a relative chronology of 50 Balto-Slavic sound changes (just phonology, no accentuation), from Proto-Balto-Slavic down to the modern daughter languages.^[9] However, only the first 12 are Common Balto-Slavic, and thus relevant for this article:

1. change of PIE */o/ > PBSl. */a/
2. RUKI law
3. PIE *CHC > PBSl. *CC
4. Winter's law
5. PIE. */Cʰ/ > PBSl. */C/
6. PIE */Cʷ/ > PBSl. */C/
7. Satemization
8. PIE *ewC > PBSl. *jawC (e.g. PIE *h₁lewdʰ- > OCS ljudьje, Lith. liáudis)^[10]
9. PIE *ewV > *awV
10. PIE syllabic. */R̥/ > PBSl. *iR (*uR)
11. PIE *#wr-, #wl- > *#r-, #l- (Lindemann's law, e.g. *wronkeh₂ > OCS rǫka)
12. PIE *sr > PBSl. *str (Slavic, Latvian, but not in Lithuanian)

Balto-Slavic accentual system

Balto-Slavic languages by the number of speakers (as of 1997).^[11]

The Proto-Indo-European accent was completely reworked in Balto-Slavic, with far-reaching consequences for accentual systems of the modern daughter languages. For the reconstruction of Balto-Slavic accent the most important are the Balto-Slavic languages that have retained tonal oppositions, these being Lithuanian, Latvian, (probably) Old Prussian and West South Slavic languages of Slovene and Serbo-Croatian. However, one should keep in mind that the prosodical systems of dialects in the aforementioned languages are sometimes very different from those of standard languages. For example, some Croatian dialects like Čakavian and Posavian dialects of Slavonian Štokavian are especially important for Balto-Slavic accentology as they retain more archaic and complex tonal accentual system than the Neoštokavian dialect on which modern standard varieties of Serbo-Croatian (Bosnian, Croatian and Serbian) are based. On the other hand, many dialects have completely lost tonal oppositions (e.g. some Kajkavian varieties, the Zagreb spoken nonstandard idiom).

To this day, there is no consensus among Balto-Slavists on the precise details of the development of Balto-Slavic accentual system. All modern research is based on the seminal study of Stang (1957), which basically instituted the field of comparative Balto-Slavic accentology. However, a number of laws and correspondences have been discovered that are nowadays held to be true by the majority of researchers, event though the exact details sometimes remain in dispute.

Notation

What follows is a short overview of the commonly used diacritical marks for Balto-Slavic accents, all used on the example letter a. In each case, there is a crude characterization of the pronunciation in terms of High, Mid, and Low-tone sequences.

• Lithuanian: "acute"/HL á, "circumflex"/H(L)H ã, "short"/H à
• Latvian: "falling"/HL à, "rising"/LH (or "lengthened") ã, "broken"/L'H â
• Slovenian: "falling"/HL â, "rising"/LH á, "short"/H ȁ (sometimes also à)
• Serbo-Croatian:^[12] "short falling"/HL ȁ, "long falling"/HML ȃ, "short rising"/LH à, "long rising"/LMH á, "posttonic length" ā
• Common Slavic: "short falling"/HL (short circumflex) ȁ, "long falling"/HML (long circumflex) ȃ, "acute"/LH (old acute, old rising") a̋

In Croatian dialects, especially Čakavian and Posavian, the "new acute" (neoacute, the "new rising") is usually markied with tilde, as ã. Short neoacute ("short new rising") is marked as à. Neoacutes represent post–Proto-Slavic development.

Balto-Slavic acute and circumflex

The development of the Proto-Indo-European accent in Proto-Balto-Slavic was conditioned by several delicate factors, such as the syllable length, presence of a laryngeal closing the syllable, and the position of PIE ictus.

In short syllables PIE tone became short accent in Balto-Slavic, and was preserved as such in both Baltic and Slavic branch, although its lengthening could be triggered by certain conditions. For example, in Lithuanian vowels /a/ and /e/ were lengthened when they initially bore short accent in open syllable, and rising tone emerged that is marked with tilde sign ã. Compare:

• PIE *kʷékʷlo- 'circle, wheel' > PBSl. *kákla- > Lith. kãklas 'neck', SCr. kȍlo
• PIE *déḱm̥t 'ten' > PBSl. déśimt > Lith. dẽšimt, SCr. dȅset

In long syllables, however, opposition between tones emerged, which are called acute and circumflex, i.e. acute (á) and circumflex (ã) accent. In this context, Balto-Slavic long syllables encompasses the following cases:

1. syllables with long (lengthened-grade) PIE vowels */ē/ and */ō/, and long vowels which emerged in Balto-Slavic (e.g. by means of Winter's law)
2. syllables with short vowels closed by a laryngeal (which merged to one Balto-Slavic laryngeal */H/)
3. syllables with PIE diphthongs (i.e. all clusters of short vowels followed by a sonorant, including */y/ and */w/)

Balto-Slavic acute emerges in the following cases:

1. in all syllables which were closed by a laryngeal in PIE, probably also when PIE laryngeal closed syllable with lengthened-grade vowel
2. in all syllables which were closed by a voiced stop in PIE, and were lengthened in PBSl. according to the Winter's law
3. in all cases of Balto-Slavic vrddhi, i.e. apophonical lengths (including new alternations *u/ū and *i/ī) which emerged only in Balto-Slavic period and have no PIE correspondences
4. on long PBSl. */ū/ which was lengthened before *nC (this can be considered a case of new Balto-Slavic length, and grouped under the preceding case)

Balto-Slavic circumflex emerges in all the other syllables, and these are:

1. PIE ablaut lengths^[13]
2. PIE diphthongs (which were not followed by a laryngeal), i.e. all sequences of PIE short vowels and the sonorants (*/m/, */n/, */l/, */r/, */y/, */w/)

As one can see, rules governing the development of Balto-Slavic acute and circumflex accents seem to be very complicated, when formulated within the framework of "classical" Proto-Indo-European laryngeal theory. Dutch Indo-Europeanist Frederik Kortlandt has proposed an alternative, more elegant and economic rule for the derivation of Balto-Slavic acute: acute is a reflex of a glottal stop, which has two sources - merger of PIE laryngeals and the dissolution of PIE pre-glottalized stop ("voiced stops" in traditional reconstruction) to glottal stop and voiced stop, according to the Winter's law. Kortlandt's formulation appears very elegant initially, and seems to be confirmed independently by a glottal stop in Latvian as a reflex of Balto-Slavic acute in words in which accent was retracted, and is in accordance with the typological universal according to which in most languages high tone is developed in syllables closed with a glottal stop.^[14] Rising tone can then be explained as a result of the development of high tone on the second mora of a long syllable.

Though elegant, Kortlandt's theory also has some problems. Glottalic theory of Proto-Indo-European reconstruction which was proposed in the 1970s is not generally accepted among linguists, and today only a small minority of linguists would consider it a reliable and self-supportive framework onto which base modern Indo-European research. Kortlandt's interpretation of intonational effects in post-PIE period of Balto-Slavic with glottal stops moreover presupposes independent development of PIE "preglottalized stops" to voiced stops in every PIE branch, which is a very unlikely scenario. Also, there is a number of Balto-Slavic lexemes which point to acute accent but that are provably not of PIE laryngeal origin, and some of which were are result of apophonical lengthenings occurring only in Balto-Slavic period.

Matasović (2008)^[15] lists the following scenario as the most probable origin of Balto-Slavic acute:

1. Rising tone, which we call Balto-Slavic acute, initially arose in the syllables closed by a laryngeal, partly due to the retraction of word-final accent onto such syllables which were phonologically long (Hirt's law). Other long syllables, if they bore the accent, were circumflexed (with falling tone).
2. In later period new Balto-Slavic lengths were acuted.
3. That younger acute has been largely eliminated in Slavic due to the effect of Meillet's law.

Reflexes in Balto-Slavic languages

In Lithuanian the acute becomes a falling tone (so-called "Lithuanian metatony"), and is marked with an acute accent. Word-finally the acute was regularly shortened: gerà 'good' (indefinite adjective) : geróji 'the good' (definite adjective). That rule is called Leskien's law, after the German neogrammarian August Leskien. Shortening operated according to Leskien's law after the Lithuanian metatony. In Žemaitian dialects the usual reflex of Balto-Slavic acute is the so-called "broken tone", with a glottal stop on a syllable carrying it (like in Latvian). In monosyllablic words the acute became circumflexed. Metatonical retraction of the accent from the final syllable to the penultimate syllable also created a circumflex automatically. In diphthongs, which had a sonorant as a second part, the acute has been preserved, but since the diacritical mark is put on the first part of such diphthong, the grave accent is used instead (e.g. Lith. pìlnas 'full' < PIE *plh₁nos).

In Old Prussian the acute was reflected probably as a rising tone. The marks on long vowels and diphthongs in Abel Will's translation of Martin Luther's Enchiridion point to that conclusion, which is the only accented Old Prussian text preserved. Diphthongs that correspond to a reconstructable Balto-Slavic acute are generally long in the second part of that text.

In Latvian the acute is reflected in the first syllable as rising or lengthened intonation (stieptā), marked with a tilde. When the accent was retracted from word-final, or any other syllable, to a syllable that carried a Balto-Slavic acute, then the first syllable of a word in Latvian has a so-called "broken" (lauztā) tone.

In all Slavic languages the acute has been shortened. In Slovenian the shortened acute has again been lengthened in the first syllable of polysyllabic words.

Balto-Slavic circumflex yielded rising intonation in Lithuanian after the Lithuanian metatony, which is marked with a tilde. In Old Prussian, the Balto-Slavic circumflex corresponds to diphthongs with length on the first part (in Will's translation of Enchiridion). In Latvian, Balto-Slavic circumflex is reflected as falling intonation (krītošā).

In standard varieties of Serbo-Croatian (i.e. in Neoštokavian dialect), Balto-Slavic circumflex is reflected as long falling accent on the first syllable.^[16] In Slovene, the accent has shifted to the end of the word in syllables with short accent (originating from acute) and circumflex.

In Czech acuted syllables have been reflected as long, while the circumflexed syllables were shortened. Russian has lost almost all the traces of Proto-Slavic accentuation, except in the pleophonical reflexes of Proto-Slavic syllables closed by liquids (*/l/, */r/): stress is on the second part of disyllabic reflex if the Proto-Slavic vowel was acuted, on the first part if it was circumflexed.

Here is a table of basic accentual correspondences of the first syllable of a word:

	Balto-Slavic and Proto-Slavic	Lithuanian	Old Prussian	Latvian	Serbo-Croatian	Slovenian	Czech	Russian
acute	V́	V́	V̆V̄	Ṽ	V̏	V̏	V̄	VRV́
circumflex	Ṽ	Ṽ	V̄V̆	V̀	V̑	V̑	V̆	V́RV

Balto-Slavic fixed and mobile paradigms

Proto-Balto-Slavic had, just like Proto-Indo-European, a class of nouns with so called "mobile" accentuation in which accent alternated between the word stem and the ending. These classes of nouns are usually reconstructed on the basis of Vedic Sanskrit and Ancient Greek, which have retained the position of the original PIE accent almost unchanged. However, by comparing the Balto-Slavic evidence, it was discovered that the PIE rules on accent alternations, devised on the basis of Vedic and Greek, do not match. Moreover, nouns that belong to mobile paradigms in Balto-Slavic belong to declension classes that had strictly fixed accent in PIE paradigms, i.e. ā-stems and o-stems. So for a long time the exact relationships between the accentuation of nouns in Balto-Slavic and PIE was one of the most mysterious questions of Indo-European studies, and some parts of the puzzle are missing to this day.

Research conducted by Christian Stang, Ferdinand de Saussure, Vladislav Illich-Svitych and Vladimir Dybo has led to a conclusion that Balto-Slavic nouns, with regard to accentuation, could be reduced to two paradigms: fixed and mobile. Nouns of the fixed paradigm had accent on the root, and in the nouns of the mobile paradigm the accent alternated between the root and the ending.

As shown by the Illič-Svityč, Balto-Slavic nouns of the fixed paradigm correspond to the PIE nouns with accent on the root (PIE barytones), the only exception being nouns with the accent on the ending (PIE oxytones) which was shifted onto the root in Balto Slavic in accordance with Hirt's law: such nouns also have fixed accent in Balto-Slavic.

The origin of the Balto-Slavic nouns of the mobile paradigm has not been completely determined, but the most probable and prevalent possibility is that proposed by Illič-Svityč, according to whom they originate as an analogical development from fixed-accent PIE oxytones. Nevertheless, it remains unclear why PIE nouns with fixed accent on the ending would become mobile, as analogies usually lead to uniformity and regularity.

The Balto-Slavic accentual system was further reworked during the Proto-Slavic and Common Slavic period (Dybo's law, Meillet's law, Ivšić's law etc.), resulting in 3 Common Slavic accentual paradigms (conveniently marked with letters as A, B, C) to correspond to 4 Lithuanian accentual paradigms (marked with numbers 1, 2, 3, 4) in a simple scheme:

		Balto-Slavic acute on the root
		yes	no
fixed accent	yes	a.p. 1/a.p. A	a.p. 2/a.p. B
	no	a.p. 3/a.p. C	a.p. 4/a.p. C

Common Slavic accentual paradigm a

The simplest accentuation is that of nouns which were acuted on the root in Balto-Slavic. They remain accented on the root^[17] throughout the paradigm in Baltic (Lithuanian first accentual paradigm) and Slavic (accent paradigm a). In the same time, Both Baltic and Slavic have expected reflexes of Balto-Slavic acute:

		Lithuanian	Russian	Serbo-Croatian
sg	N	várna	voróna	vrȁna
	V	várna	-	vrȁno
	A	várną	vorónu	vrȁnu
	G	várnos	voróny	vrȁnē
	D	várnai	voróne	vrȁni
	L	várnoje	voróne	vrȁni
	I	várna	vorónoj	vrȁnōm
pl	NV	várnos	voróny	vrȁne
	A	várnas	vorón	vrȁne
	G	várnų	vorón	vrȃnā
	D	várnoms	vorónam	vrȁnama
	L	várnose	vorónax	vrȁnama
	I	várnomis	vorónami	vrȁnama

In Russian the Balto-Slavic acute yielded expected reflex with "polnoglasie". In Serbo-Croatian the short falling accent in genitive plural has been substituted with long falling due to the loss of the yer.

Common Slavic accentual paradigm b

In the nouns with non-mobile initial accent, which did not have an acuted root syllable, in both Lithuanian and Slavic an independent accent shift occurred from the root to the ending. In Lithuanian these are the nouns of the second accent paradigm, and in the Slavic of accent paradigm b.

Lithuanian noun rankà 'hand' etymologically corresponds to Russian ruká and Serbo-Croatian rúka, but both of these became mobile in a later Common Slavic development. So the reflexes of the Proto-Slavic noun *jōxā́ 'soup' are listed instead.

		Lithuanian	Russian	Serbo-Croatian
sg	N	rankà	uxá	júha
	V	rankà	-	jȗho
	A	rañkai	uxú	júhu
	G	rañkos	uxí	júhē
	D	rañkai	uxé	júsi/juhi
	L	rañkoje	uxé	júsi/juhi
	I	rankà	uxój	júhōm
pl	N	rañkos	uxí	júhe
	V	rañkos	-	jȗhe
	A	rankàs	uxí	júhe
	G	rañkų	úx	júhā
	D	rañkoms	uxám	júhama
	L	rañkose	uxáx	júhama
	I	rañkomis	uxámi	júhama

In Lithuanian the initial accent was preserved in all cases in which ending did not contain syllable with Balto-Slavic acute. In these cases (NVI sg, A pl) accent shifted onto the acuted ending, in accordance with the rule discovered by F. de Saussure. Later that acuted syllable was shortened due to the Leskien's law.

In Slavic the accent shifted from the root onto the ending in accordance with the Dybo's law, regardless of the syllable nature (i.e. whether it contained Balto-Slavic acute or not), so the nouns of the a.p. b are consistently accented on the ending (oxytonic, except in the I pl). In Neoštokavian dialects, which is used as a basis for standard Bosnian, Croatian and Serbian, accent was retracted from the ending onto the root syllable and became rising (so called "Neoštokavian retraction"). Old Štokavian and Čakavian dialects preserved the original ending-stressed paradigm.

Common Slavic accentual paradigm c

Nouns with mobile accent had in some cases accented root, on the others the ending.

Lithuanian distinguishes two accent paradigms of these nouns: if they had acuted root, they belong to the third accent paradigm, and if the root was not acuted, by the operation of de Saussure's law the accent shifted onto the all acuted endings in the paradigm, so these nouns belong to the fourth accent paradigm.

In Proto-Slavic the acute has been eliminated in the nouns with mobile accentuation by the operation of Meillet's law, so therefore all the nouns with mobile accentuation belong to one accent paradigm, the so called accent paradigm c.

		Lithuanian	Russian	Neoštokavian Serbo-Croatian	Čakavian Serbo-Croatian	Common Slavic
sg	N	galv-à	golov-á	gláv-a	glāv-ȁ	*golv-à
	V	galv-à	-	gláv-o	glȃv-o	-
	A	gálv-ą	gólov-u	glȃv-u	glȃv-u	*gȏlv-ǫ
	G	galv-õs	golov-ý	gláv-e	glāv-é	*golv-ỳ
	D	gálv-ai	golov-é (OESl. gólov-ě)	glȃv-i	glāv-ȉ	*gȏlv-ě → *golv-ě̀
	L	galv-ojè	golov-é	glȃv-i	glāv-ȉ	*golv-ě̀
	I	gálv-a	golov-ój	gláv-ōm	glāv-ún	(*golv-ojǫ̀)
pl	NV	gálv-os	gólov-y	glȃv-e	glȃv-e	*gȏlv-y
	A	gálv-as	gólov-y	glȃv-e	glȃv-e	*gȏlv-y
	G	galv-ų̃	golóv-	gláv-ā	gláv-	*gólv-ъ
	D	galv-óms	golov-ám	gláv-ama	glāv-án	*golv-a̋mъ
	L	galv-osè	golov-áx	gláv-ama	glāv-ȁx	*golv-a̋xъ
	I	galv-omìs	golov-ámi	gláv-ama	glāv-ȁmi	*golv-a̋mi

Lithuanian has preserved the best Balto-Slavic mobile paradigm. In Neoštokavian the final accent has been retracted and gained rising intonation, and the Proto-Slavic initial accent is preserved as a circumflex.

Balto-Slavic apophony

Indo-European ablaut has been significantly reworked in Balto-Slavic. Prominence of lengthened-grade has been significantly increased, as opposed to PIE in which it was used only for rare vrddhi-formations, nominative singulars of some consonant-stem nouns and sigmatic aorist.

Proto-Slavic abundantly used lengthened-grade in morphology. For example:

• PSl. *slāwā 'fame, glory' (OCS slava) vs. PSl. *slawa 'word' (OCS slovo)
• PSl. *twāri 'substance' (OCS tvarь) vs. PSl. *twarītej 'to form, create' (OCS tvoriti)

Similarly, in Lithuanian:

• Lith. prõtas 'intellect, mind' (< *prāt) vs. Lith. pràsti 'to understand'
• Lith. gė̃ris 'goodness' (< *gēr-) vs. Lith. gẽras 'good'

On the basis of already-present apophonic oppositions beween Balto-Slavic the long */ā/, */ē/, */ō/ and the short */a/ and */e/, new oppositions in arose between the long */ī/, */ū/ and the short */i/, */u/. This latter type of apophony was not productive in PIE. Compare:

• Lith. mū̃šis 'battle' vs. mùšti 'to kill, hit'
• Lith. lỹkis 'remainder' vs. lìkti 'to stay, keep'

This new type of apophonic length was especially used in Proto-Slavic in the formation of durative, iterative and imperfective verbs. Compare:

• PSl. *dirātej > OCS dьrati vs. PSl. *arz-dīrātej 'to tear' > OCS razdirati
• PSl. *birātej 'to pick' > OCS bьrati vs. PSl. *bīrātej 'to choose' > OCS birati

See also

• Proto-Indo-European language
• Proto-Slavic language
• Balto-Slavic languages

Notes

1. ^ For an alternative formulation, see Kortlandt (1978:12–24)
2. ^ Matasović (2008:86). For a more precise formulation of the rule, see Matasović (2005)
3. ^ Kortlandt (2002:8), rule 5.8.
4. ^ Matasović (2008:87)
5. ^ Matasović (2008:96)
6. ^ These were phonetically really vowels, but phonologically sonorants, as the syllabicity of the PIE */j/ and */w/ was predictable by a rule.
7. ^ Matasović (2008:109)
8. ^ Vaillant, André, "Grammaire comparée des langues slaves", (I-IV), IAC, Lyon 1950-77
9. ^ Holzer 2001, 2007
10. ^ This change was common for East Baltic and Slavic, but not for West Baltic.
11. ^ According to the data taken from Anatole V. Lyovin, An Introduction to the Languages of the World, Oxford University Press, New York - Oxford, 1997.
12. ^ All examples given for Serbo-Croatian are based on the standard language, i.e. stylised Neoštokavian dialect, and are accented accorcding to the Rječnik hrvatskoga jezika, F. Broz and I. Iveković, Zagreb 1901 and Akademijin Rječnik hrvatskoga ili srpskoga jezika, XXIII volumes, 1880-1976
13. ^ Some obsolete literature claims that Balto-Slavic acute is on all PIE long vowels, but this has been proven false.
14. ^ E.g. in Vietnamese and Old Chinese. There are, however, cases when syllables closed with glottal stop yielded low tone, e.g. in Tibetan and Halkomelem.
15. ^ Matasović (2008:136)
16. ^ But this is not the only source of long falling syllable in Štokavian: it can also originate from neoacute.
17. ^ root is here understood in Proto-Slavic, not PIE sense

References

• Frederik Kortlandt (2002). "From Proto-Indo-European to Slavic". http://www.kortlandt.nl/publications/art066e.pdf. 
• Matasović, Ranko (2008) (in Croatian), Poredbenopovijesna gramatika hrvatskoga jezika, Zagreb: Matica hrvatska, ISBN 978-953-150-840-7