The breakup of Common Slavic was gradual and many sound changes (such as the second regressive palatalization) still propagated throughout what must have been by then a
dialect continuum. However, several changes were more restricted, or had different outcomes. The end of the Common Slavic period occurred with the loss of the
yers (weak high vowels, derived from Proto-Balto-Slavic and ultimately Proto-Indo-European *i and *u). This ended the era of
syllabic synharmony (when most, originally all, syllables were
open) by creating large numbers of
closed syllables. The conditions for which yers were strong and which ones weak is the same across most or all Slavic languages, but the particular outcomes are drastically different. The clusters *tl and *dl were lost in all but West Slavic, being normally simplified to *l. Exceptions are some
Northern Russian dialects where they instead changed to *kl and *gl respectively (today only traces of this remain) and the
Gail Valley dialect of Slovene (with traces in other
Carinthian dialects). For many Common Slavic dialects—including most of West Slavic, all but the northernmost portions of East Slavic, and some western parts of South Slavic—Proto-Slavic
lenited from a
voiced velar plosive to a
voiced velar fricative ( → ). This remains in some modern languages: for example, Czech and Slovak , Belarusian , Ukrainian , which developed from Proto-Slavic . Because this change was not universal and because it did not occur for a number of East Slavic dialects (such as Belarusian and South Russian) until after the application of
Havlík's law, calls into question early projections of this change and postulates three independent instigations of lenition, dating the earliest to before 900 CE and the latest to the early thirteenth century.
Overview of languages The Slavic languages are generally divided into East Slavic, South Slavic and West Slavic. For most comparative purposes, however, South Slavic does not function as a unit. Bulgarian and Macedonian, while quite similar to each other, are radically different from the other South Slavic languages in phonology and grammar. The phonology of Bulgarian and Macedonian is similar to East Slavic rather than their nearest Slavic neighbor Serbo-Croatian (suggesting an early East–West divide across the whole Slavic territory, before South Slavic was separated from the rest of the Slavic languages by the spread of Hungarian and Romanian). In grammar, Bulgarian and Macedonian have developed distinctly from all other Slavic languages, eliminating nearly all case distinctions (strongly preserved elsewhere), but preserving and even strengthening the older Indo-European aspectual system consisting of synthetic aorist and imperfect tenses (largely eliminated elsewhere in favor of the new Slavic aspectual system).
Old Church Slavonic (OCS) data are especially important for the reconstruction of Late Common Slavic (LCS). The major exception is LCS accent, which can only be reconstructed from modern Slavic dialects.
Palatalization At least six separate sound changes involving palatalization can be identified in the history of the Slavic languages: •
Satemization, which converted
Proto-Indo-European (PIE) front velars *ḱ, *ǵ, *ǵh into Balto-Slavic *ś, *ź, *ź, and further into Slavic *s, *z, *z. • The
first regressive palatalization of velars. • The
second regressive palatalization of velars. • The progressive palatalization of velars. •
Iotation, which palatalized all consonants before *j. • General palatalization of all consonants before front vowels (not in all languages). The first palatalization (satemization) is reflected in all Balto-Slavic languages, while the rest are represented in nearly all Slavic languages. (The
Old Novgorod dialect did not undergo the second regressive palatalization, and underwent the progressive palatalization only partly.)
Velar palatalization outcomes The outcome of the first regressive palatalization is uniform across all Slavic languages, showing that it happened fairly early. The outcome of the second regressive palatalization shows more variety. It is possible, however, that this is a later development. Many authors reconstruct a uniform outcome *ś, which only later resolves into or . (According to Aleksandar Belić, the phonetic character of the palatalizations was uniform throughout Common Slavic and West Slavic languages developed later on by
analogy.) In all dialects (except for Lechitic), was deaffricated to , but is still found in a few of the earlier Old Church Slavonic texts, where it is represented by the special letter
Dze (Ѕ). The following table illustrates the differences between the different dialects as far as phonetic realization of the three velar palatalizations: Some dialects (in particular South Slavic), allowed the second regressive palatalization to occur across an intervening . The OCS and Bulgarian outcome is somewhat unusual as it is not an affricate but rather a fricative followed by a stop, having undergone metathesis. In Macedonian, the outcome is non-sibilant. In Proto-Slavic, iotated *ľ *ň *ř contrasted with non-iotated *l *n *r, including before front vowels. This distinction was still apparent in Old Church Slavonic, although they aren't always consistently marked (least for *ř, which may have already been merging with *r' at the time the Old Church Slavonic manuscripts were written or copied). In Southwest Slavic (modern Serbo-Croatian and Slovene), this contrast remains to this day. In the other Slavic variants, however, regular *l *n *r developed palatalised variants before front vowels, and these merged with the existing iotated *ľ *ň *ř.
General palatalization In most languages (but not Serbo-Croatian or Slovene), a general palatalization of consonants before front vowels (including the front yer ь), as well as of *r in *ьr, occurred at the end of the Common Slavic period, shortly before the loss of weak yers. The loss of the weak yers made these sounds phonemic, nearly doubling the number of phonemes present. The already palatal or palatalized sounds — the outcomes of the velar palatalizations and iotation — were unchanged. Newly palatalized sounds *l' *n' *r' merged with palatal *ľ *ň *ř from iotation. However, newly palatalized *t' *d' *s' *z' did not usually merge with existing *ť *ď (from iotation) or *č *š *(d)ž (from the first palatalization of velars). The new sounds were later depalatalized to varying degrees in all Slavic languages, merging back into the corresponding non-palatal sound. This has happened the least in Russian and Polish: before another consonant, except for l', which was always preserved, as in сколько ''skol'ko
"how many", and dentals before labials, as in тьма t'ma'' / ćma "darkness", and before a pause for labials. r' was depalatalized early before dentals, as in чёрт
čort / czart "devil", but otherwise has been preserved in Polish and in many Russian dialects, as well as for some older standard speakers, who pronounce верх as ''ver'h
(cf. Polish wierzch). In many cases palatalization was analogically restored later, particularly in Russian. Russian has also introduced an unusual four-way distinction between non-palatal C
, palatal C',
the sequence C'j
of palatal + (from Common Slavic *Cьj with weak ь), and the sequence Cj
of non-palatal + (only across a clear morpheme boundary, when a prefix is followed by a morpheme-initial ); however, only dentals show a clear contrast before j''. Czech underwent a general depalatalization in the 13th century. It might be argued that Czech never underwent palatalization at all in most cases, but the Czech sound
ř (an unusual
fricative trill) is found everywhere that *r followed by a front vowel is reconstructed in Late Common Slavic. This suggests that former *r' escaped depalatalization because it had evolved into a new sound — no longer paired with a corresponding non-palatal sound — by the time that depalatalization occurred. The same thing happened more broadly in Polish — paired palatalized sounds occur only before vowels, but original *r' *l' *t' *d' *s' *z' are reflected differently from *r *l *t *d *s *z even word-finally and before consonants, because all six pairs had diverged by the time any depalatalization occurred. *r' evolved as in Czech, later becoming , but still written
rz. *t' *d' *s' *z' evolved into
alveolopalatal consonants; and in the case of *l', non-palatal *l evolved into a back velar and then further into , still written
ł. In Bulgarian, distinctively palatalized consonants are found only before . Velars are allophonically palatalized before front vowels in standard Bulgarian; the same thing happens to all consonants in Eastern Bulgarian. Palatalization triggered a general merger of Common Slavic *y and *i. In East Slavic and Polish, the two sounds became allophones, with occurring after non-palatal sounds and after palatal or palatalized sounds. In Czech, Slovak and South Slavic, the two sounds merged entirely (although in Czech, *i triggered palatalization of
t d n prior to the merger, and in Slovak, it triggered palatalization of
t d n l). Researchers differ in whether the paired palatalized consonants should be analyzed as separate phonemes. Almost all analyses of Russian posit phonemic palatalized consonants due to their occurrence word-finally and before consonants, and due to the phonemic distinction between and . In Polish and Bulgarian, however, many researchers treat some or all paired palatalized consonants as underlying sequences of non-palatal consonant + . Researchers who do this in Polish also generally treat the sounds and as separate phonemes.
The yers ь and ъ Strong vs. weak yers The two vowels ь and ъ, known as (front and back)
yer, were originally pronounced as short high vowels. During the late Proto-Slavic period, a pattern emerged in these vowels which characterised a yer as either "strong" or "weak". This change is known as
Havlík's law. A yer at the end of a word, or preceding a strong yer or non-yer vowel was weak, and a yer followed by a weak yer became strong. The pattern created sequences of alternating strong and weak yers within each word: in a sequence of yers, every odd yer encountered was weak, every even yer was strong. The name *sъmolьnьskъ (the Russian city of
Smolensk) is shown here as an example, with strong yers in
bold and weak yers in
italics. • Nominative singular: *s
ъmol
ьn
ьsk
ъ • Genitive singular: *s
ъmol
ьn
ьska During the time immediately following the Common Slavic period, weak yers were gradually deleted. A deleted front yer ь often left palatalization of the preceding consonant as a trace. Strong yers underwent lowering and became mid vowels, but the outcomes differ somewhat across the various Slavic languages. Slovene in particular retains a distinct outcome that did not merge with any other vowels, albeit originally only in unstressed syllables, and Bulgarian has an outcome that merged only with nasal ǫ. Compare: • An apostrophe indicates palatalization of the preceding consonant. • The front and back strong yers merged in Serbo-Croatian, Slovene, Czech and Upper and Lower Sorbian. • In Slovene, arose from this merged result when stressed, otherwise. was later often replaced by analogically. • In Central (standard) Slovak, the normal outcomes of *ь *ъ are
e o, but various other sounds often appear, unpredictably. In East and West Slovak dialects, both yers merge and become
e, as in Czech.
Examples Clusters and fill vowels Deletion of weak yers created many new closed syllables as well as many of the unusual consonant clusters that characterize the Slavic languages today. Many cases of "spurious vowels" also appeared because a yer had been weak in one form of a word but strong in another, causing it to disappear in some forms of the word but not others. For example, the word for "dog" was *p
ьs
ъ in the nominative singular, but *p
ьsa in the genitive singular, with differing patterns of strong and weak yers. Following the deletion of weak yers and lowering of strong yers, this resulted in nominative Czech
pes, Polish
pies, Serbo-Croatian
pas, but genitive
psa (in all three). In some cases, however, deletion of weak yers would lead to an awkward consonant cluster such as word-initial
rt-,
ln- or
mx- (as in the example of *mъxъ "moss" above), with a
sonorant consonant on the outside of the cluster, a violation of the principle of rising sonority. These clusters were handled in various ways: • Allow them to exist unchanged. This happened especially in Russian and Polish. • Convert the weak yer into a strong one, thereby breaking up the consonant cluster. This happened most consistently in Serbo-Croatian. • Convert the sonorant into a syllabic sonorant. This happened with initial
r in Serbo-Croatian and Macedonian. • Insert a prothetic vowel before the cluster. This happened in some dialects of Belarusian, e.g. ''lënu ~ l'nu ~ il'nú'' "flax (gen. sg.)" (Common Slavic *lьnu). A similar problem occurred with awkward word-final clusters such as
-tr,
-gn or
-sm. These originated from words like *větrъ "wind" or *ognь "fire", where the cluster occurred syllable-initially and there was no sonority violation. Again various outcomes are found in different languages, largely parallel to the above outcomes for word-initial clusters. In this case, when a cluster needed to be broken up, a strong yer was inserted as a
fill vowel between the two consonants.
Tense yers Yers before are known as
tense yers and were handled specially. In languages other than Russian, they were sometimes raised, with *ьj *ъj becoming *ij *yj regardless of position. In Russian, the opposite sometimes happened, with *ij *yj sometimes lowering to *ьj *ъj, subsequently evolving normally as strong or weak yers. In languages other than Russian, resulting sequences of *ijV or *yjV may contract to a single vowel (especially in Czech). The outcomes are not consistent and depend on various factors. For example, *ъj in long adjectives becomes contracted
í in Czech, but stressed
oj, unstressed
yj (
ăj in the old literary pronunciation and some dialects) in Russian. In Russian, when the yer in *ьj was weak, the result was a sequence of palatal consonant + , which remained distinct from regular palatal consonants. In other languages, either the sequence compressed into a single palatal consonant or the palatal consonant was depalatalized. E.g. from Common Slavic *ustьje "estuary", when the yer was treated as weak the result is Russian ''úst'e
, Polish ujście
, Slovene ûstje
; when treated as strong, the result is Czech ústí
(with contraction of *ije), Bulgarian ústie'' .
The liquid diphthongs Proto-Slavic had eliminated most diphthongs creating either long monophthongs or nasal vowels. But it still possessed sequences of a short vowel followed by *l or *r and another consonant, the so-called "liquid diphthongs". These sequences went counter to the law of open syllables and were eliminated by the end of the Proto-Slavic period, but differently in each dialect.
Mid vowels The situation for the mid vowels *e and *o is relatively straightforward. The South Slavic dialects used
metathesis: the liquid and vowel switched places, and the vowels were lengthened to *ě and *a respectively. The East Slavic languages instead underwent a process known as
pleophony: a copy of the vowel before the liquid consonant was inserted after it. However, *el became *olo rather than *ele. The situation in West Slavic is more mixed. Czech and Slovak follow the South Slavic pattern and have metathesis with lengthening. Polish and Sorbian underwent metathesis but without any lengthening, and the northwestern Lechitic languages (
Pomeranian,
Slovincian and
Polabian) retained *or without any metathesis at all. • The variants
le/lja, re/rja in Bulgarian, and
lije/le/li, rije/re/ri in Serbo-Croatian, are dialectal differences. • The variants
oli, ori in Ukrainian are due to a sound change
*ō > , where
*o was lengthened before a lost
yer under certain accentual conditions.
High vowels The evolution of the liquid diphthongs with high vowels in the various daughter languages is more diverse. In some West Slavic and South Slavic languages, syllabic sonorants appear, and in others (e.g. Polish), either vowel-consonant or consonant-vowel sequences appear depending on the context, which is most easily derived by assuming an earlier stage with syllabic sonorants (with the former occurrence of ь or ъ transferred into palatalization or lack thereof). East Slavic, however, consistently has vowel-consonant sequences with
e or
o as the vowel, which can be easily derived by assuming that the liquid diphthongs continued unchanged until the changes involving yers (assuming that the yers in these sequences were always treated as if strong). As a result, there is a divergence of opinion, with some scholars assuming that the high-vowel liquid diphthongs evolved into syllabic sonorants early in the Common Slavic period (even before the metathesis of the mid-vowel liquid diphthongs), while others assume that the change to syllabic sonorants was one of the last changes in the Common Slavic period and did not occur at all in many languages (e.g. East Slavic). Old Church Slavonic writes these as *lь, *lъ, *rь, *rъ, as if metathesis had occurred. However, various internal evidence indicates that these behaved differently from original Proto-Slavic *lь, *lъ, *rь, *rъ, and hence were probably actually pronounced as syllabic sonorants. (This is also consistent with evidence from later languages.) In the manuscripts, only a single vowel is found in this position, usually *ъ but also consistently *ь in a few manuscripts. This appears to indicate that the palatal(ized) syllabic sonorants had merged into the non-palatal ones. The syllabic sonorants are retained unchanged in Czech and Slovak. In Macedonian, Serbo-Croatian and Slovene, syllabic r is retained but an epenthetic vowel was inserted before syllabic l. Bulgarian inserted an epenthetic ǎ before both. Serbo-Croatian also underwent
l-vocalization. East Slavic reflects original *ьr and *ъr as
er and
or respectively, but merges *ьl and *ъl as
ol (Proto-Slavic *
vьlna > East Slavic > Russian
волна), similarly to the merger of *el and *ol as
olo.
L-vocalization later occurred in Belarusian and Ukrainian: for example, Proto-Slavic '''' > Old East Slavic > Ukrainian , Belarusian .
The nasal vowels ę and ǫ Nasal vowels were initially retained in most Slavic dialects, but soon underwent further changes. Nasality is preserved in modern Polish, as well as in some peripheral dialects of
Slovene (e.g. the
Carinthian dialect group) and
Bulgarian/
Macedonian (e.g. around
Thessaloniki and
Kastoria). In other Slavic languages, however, the nasal vowels lost their nasality and merged with other vowels. The outcomes are as follows: • Long and short nasal vowels developed primarily from accentual differences. The neoacute accent always produced long vowels, but the outcome of the other accents (circumflex and old acute) depended on the dialect. See above for more details. • The two outcomes listed in Czech occurred in hard and soft environments, respectively. "Hard environment" means preceding a hard (neither palatal nor palatalized) alveolar consonant. • In Slovak, short *ę >
ä after labials, else
a. • In Polish, original *ę and *ǫ can only be distinguished because the former palatalized the preceding consonant. • The two outcomes listed in Macedonian occurred in initial and non-initial environments, respectively.
The yat vowel ě The phonetic realization of *ě was also subject to phonetic variation across different dialects. In Early Proto-Slavic, *ě was originally distinguished from *e primarily by length. Later on, it appears that initially it was lowered to a low-front vowel and then diphthongized to something like . This is still reflected as
ia or
ja (i.e. with palatalization of the previous consonant) in certain contexts before hard consonants in Bulgarian and Polish; but in most areas it was raised to . This generally proceeded further in one of three directions: • Remain as a diphthong. • Simplify to . • Simplify to . All three possibilities are found as variants within the Serbo-Croatian area, respectively known as the
ijekavian,
ekavian and
ikavian dialects. An ijekavian dialect served as the basis of almost all the literary Serbo-Croatian forms (all except literary Serbian as used specifically within Serbia itself, which is ekavian). These dialects have short
je, long
ije (often pronounced as ). The ijekavian dialects of Serbo-Croatian are in fact the only Slavic languages that consistently preserve a reflex of *ě distinct from all other Common Slavic sounds. (Elsewhere, at most only some cases of *ě, e.g. those in stressed syllables, have a distinct reflex.) In cases where the reflex has remained as a diphthong, it has most commonly developed to , often followed by merger of the with a previous consonant to form a palatal or palatalized consonant. In Czech, for example, the reflex of *ě is sometimes still spelled
ě, but this in fact indicates after labials, and after
t d n, which become pronounced as palatal sounds ; in other cases the reflex is simply
e. In
Old Russian, the reflex of *ě simplified to , but this did not cause a merger with *e in stressed syllables, which was pronounced . Later, this (also including reflexes of the strong front yer) changed into (i.e. with palatalization of the preceding consonant) when not followed by a palatalized consonant: cf. modern Russian лёд 'ice' (loans from Church Slavonic do not display this change: небо "sky", крест "cross", перст "finger" in elevated style). The result of the sound change may be expressed in the present-day spelling by means of a diaeresis over the letter
e (
ё), but generally isn't. In contrast, the sound change did not affect the reflex of original
yat, which continued to be pronounced as , eventually merging with the surviving unaffected instances of as late as the 1700s (seen, respectively, in the words хлеб 'bread' and печь 'oven'). Original
yat continued to be represented distinctly from resulting from other sources in spelling until the
spelling reforms of 1918, and is still distinguished in some
Northern Russian dialects. Similarly, in Ukrainian, the reflex of *ě simplified to
i , but this did not cause a merger with either *e or *i in stressed syllables, because both sounds developed to a phoneme
y . (However, in some instances, former *o is also reflected as
i.) The following table shows the development of *ě in various languages: • Bulgarian (apart from the Western Bulgarian dialects) has
ja only when stressed and before a (formerly) hard consonant,
e otherwise (e.g. *tělo "body" produces singular тя́ло
tjálo and plural тела́
telá). • Macedonian (and the Western Bulgarian dialects) has only
e. The reflex between the Bulgarian and Macedonian versions forms an important
isogloss known as the
jat' border, running approximately from
Nikopol on the
Danube to Solun (
Thessaloniki) on the
Aegean Sea. • Serbo-Croatian shows great dialectal diversity; see above. • Polish has
ia before a (formerly) hard dental,
ie otherwise. • Some
Northern Russian dialects have *ě >
i, as in Ukrainian. • Some Ukrainian dialects, as well as some Northern Russian sub-dialects, preserve an earlier form . • Slovak has short
e, long
ie.
Accent Modern prosodic phenomena The modern Slavic languages differ greatly in the occurrence of the prosodic phenomena of phonemic
vowel length,
accent and
tone, all of which existed in Common Slavic (CS), ranging from total preservation (Serbo-Croatian) to total loss (Polish). However, the surface occurrence of length, accent and/or tone in a given language does not necessarily correspond with the extent to which the corresponding CS phenomena can be reconstructed. For example, although all of the standard Serbo-Croatian literary forms have phonemic tone, they cannot be used to reconstruct Late CS tone; only some of the non-standard dialects (e.g.
Chakavian) are useful in this regard. Similarly, although Macedonian has (marginal) phonemic accent, this does not continue the CS accent position. Contrariwise, although modern Polish lacks vowel length, some vowel quality differences (e.g. in nasal vowels) reflect former length differences. Phonemic tone is found only in western South Slavic languages — Serbo-Croatian and some Slovene dialects (including one of the two literary standards). Phonemic length is found in Serbo-Croatian, Slovene, Czech and Slovak. Phonemic accent is found in Serbo-Croatian, Slovene, the East Slavic languages, Bulgarian, the northern
Kashubian dialects, marginally in Macedonian. In terms of which modern languages preserve which CS prosodic features, it is important to understand that length, tone and accent are intricately tied together. Middle CS did not have phonemic length, and Late CS length evolved largely from certain tonal and accentual changes. (In addition, some long vowels evolved from contraction of vowels across or
compensatory lengthening before a lost yer, especially in Czech and Slovak.) Hence length distinctions in some languages (e.g. Czech) may correspond to tonal distinctions in other languages (e.g. Serbo-Croatian).
Development from Common Slavic As mentioned above, Middle Common Slavic (MCS) had a three-way tonal/length distinction on accented syllables (long rising, long falling, short). Long rising and falling tones continue Balto-Slavic acute and circumflex, respectively. Late Common Slavic (LCS) developed at first a four-way distinction, where rising and falling tones could occur in both short and long syllables, as in modern Serbo-Croatian. Later changes of a complex nature produced the prosodic phenomena found in the various modern languages. In general, the history of Slavic accentuation is extremely complex and still incompletely understood. The following is a summary of the most important changes in LCS: • Short-accented syllables develop into specifically short falling syllables. • Long rising (acute) syllables are shortened, becoming short rising. • The accent is retracted (moved a syllable towards the beginning) in certain cases, e.g. when it fell on a weak yer (
Ivšić's law). The new syllables developed a rising accent, termed the
neoacute. When this accent fell on short *e and *o, they were lengthened, except in Serbo-Croatian and Slovene. At this stage, most neoacute syllables remained separate from original acute syllables because of the difference in length (long vs. short, respectively). • Initial short falling syllables followed by a final weak yer (i.e. words which will be monosyllabic upon loss of the yer and which in MCS had a short accent on the initial syllable) are lengthened. Such syllables become long falling (although this doesn't cause a merger with original long falling syllables because the two differ in vowel quality, i.e. *e *o *ь *ъ vs. other qualities). This is hypothesized to be pan-Slavic, but only visible in Serbo-Croatian and Slovene because of the following step. • Long falling syllables are shortened everywhere except in Serbo-Croatian and Slovene. This undoes the previous step (if it occurred at all) and is responsible for MCS circumflex accent appearing as a short vowel in Czech, Slovak, Old Polish, etc. • Compensatory lengthening of some short syllables occurs in some languages when immediately followed by a weak yer. This does not occur in South Slavic, nor in Russian. It is most common in words that will become monosyllabic after the loss of the yer. In Ukrainian, it is general in this position, while in Czech and Polish it is common but inconsistent. It results in a Czech and Polish pattern in masculine nouns in which long vowels in the nominative singular alternate with short vowels in the other case/number forms. This pattern is then often analogically extended to other words. • Weak yers are lost. • Short rising syllables (arising mostly from MCS acute accent) are relengthened in East Slavic, Bulgarian and Macedonian. It also occurs in Czech and Slovene in the initial syllable of disyllabic words, under certain conditions. This causes a general merger of MCS acute and neoacute in the East Slavic and eastern South Slavic languages, leading to a two-way distinction of short falling vs. long rising. (This distinction is later lost, but revealed in some traces; see below.) Note that steps 3, 4 and 6 can all be viewed as types of
compensatory lengthening before a lost (or about-to-be-lost) yer. Numerous further developments occur in individual languages. Some of the most notable ones are: • In East Slavic, Bulgarian and Macedonian, the pitch accent is converted into a
stress accent (as in English), and vowel length and tone are lost. Traces of these distinctions exist in a few circumstances: • Vowel length in early borrowings of Slavic words, e.g. into Finnish. • The position of the accent in original liquid diphthongs in East Slavic, when the vowel of the diphthong was
o or
e. Such sequences develop into bisyllabic sequences with
-oro-,
-ere- or
-olo-. A short falling accent (MCS circumflex) is reflected as
-óro- etc., while a long rising accent (MCS acute and LCS neoacute) produces
-oró- etc. • Words with a short falling vowel (MCS circumflex) tend to lose the accent to attached prefixes or
clitics (e.g. the
definite article added onto the end of Bulgarian and Macedonian words). • In East Slavic, stressed long *ō was raised to (notated
ô), while all other *o remained as . This is still reflected in some
Northern Russian dialects. • In some dialects of Macedonian, stress occurring on suffixes is moved onto the stem, but may otherwise appear on any syllable, while in others, including standard Macedonian, lexical stress accent is lost and replaced with fixed stress. • Phonemic tone and accent are lost in West Slavic (although some dialects of the
Kashubian language maintain phonemic stress accent). Phonemic length is eventually lost in Polish, although still present in
Old Polish. In Polish, some former long/short pairs have evolved to different sounds; e.g. *ō >
ó . Similarly, nasal
ę reflects a former short nasal, while
ą reflects a long nasal. (The two original nasals *ę and *ǫ merged in Polish.) • In the original eastern Serbo-Croatian dialects, phonemic tone is lost, with all accented syllables essentially gaining a falling tone. Later on, in a subset of these dialects (the
neoshtokavian dialects, the basis of all standard Serbo-Croatian registers), the stress is retracted one syllable when possible, producing a rising tone in the process (cf. the neoacute retraction). This reintroduces phonemic tone on initial syllables. Only some conservative Serbo-Croatian dialects (e.g.
Chakavian) maintain the original accentual system unchanged. Some Slovene dialects (see below) maintain all original properties of the accentual system, but with various changes in multisyllabic words. Slovene shows large dialectal diversity for its relatively small area of distribution. For example, only the central dialects and one of the two literary standards maintain tone, and some of the northwest dialects maintain original nasality. In the dialects maintaining tone, the prosody of monosyllables agrees closely with the most conservative Serbo-Croatian dialects (e.g. Chakavian). In multisyllabic words, all non-final stressed vowels were lengthened (acute and neoacute becoming long rising, while circumflex and original short become long falling), and all non-final unstressed vowels were shortened, which produced a prosodic pattern not unlike that found in modern
Italian. Length remained distinctive in final syllables only. But prior to this, various shifts happened: • Original acute became circumflex (long falling) in certain cases, e.g. prior to a lengthened syllable (the
neo-circumflex). • With non-final original circumflex and short syllables, the accent shifts to the right, becoming circumflex (long falling) (the
progressive shift). • With non-initial original acute, or with any original final-accented syllable in a multisyllabic word, the accent shifts left onto original long syllables, becoming acute (long rising). • In some dialects, a further leftward shift happens from original final-accented syllables to original short syllables. In the standard language, this happens specifically with *e *o, which become acute (long rising) with a low-mid quality (whereas other long mid vowels are normally reflected as high-mid). In some non-standard dialects, this also happens with *ǝ < strong yers (although it remains short). In West Slavic, esp. in Czech, a number of originally short vowels in monosyllables are lengthened. The conditions for this lengthening are incompletely understood and seem to involve good deal of analogy and dialect mixing. Note that the overall effect of all these changes is that either the MCS acute, MCS circumflex or both have ended up shortened in various languages in various circumstances, while the LCS neoacute has generally remained long. Example: == Loanwords ==