2.1 Introduction
We assume that Proto-Slavic (or Balto-Slavic)Footnote 1 developed from an Indo-European dialect in which there were no longer the so-called laryngeal segments.Footnote 2 The merger of the short *o and *a into a, and of long *o: and *a: into a: was the first Slavic change to the original Indo-European dialect. As a result, a simple vocalic system emerged, which consisted of eight segments: four short ones and four long ones; four high ones and four low ones, with only two distinctive heights. The vowel system may have been as follows:
| *i | *u | *i: | *u: |
| *e | *ɑ | *e: | *ɑ: |
The vowel o emerged only later, in the initial period of the disintegration of the Slavic linguistic unity.
Proto-Slavic *e, both long and short, were relatively low vowels, which is supported by their further development, by the alternations of verbal stems, and, most importantly, by borrowings into other languages. Borrowings into other languages suggest also that the so-called jers (ь < short i, ъ < short u) emerged as late as after the sixth century.
The process of the delabialization of the long *u: began probably in the sixth century, which later resulted in the creation of the vowel *y: (initially a high back non-rounded vowel). If the Indo-European syllabic sonorants *r̩(:), *l̩(:),*n̩(:), *m̩(:) were also to be counted as functional vowels,Footnote 3 then their development in Proto-Slavic increased the occurrence of high vowels. They developed into *ir, *ur, *il, *ul, *in, *un, *im, *um, *i:r, *u:r, *i:l, *u:l, *i:n, *u:n, *i:m, *u:m. In the descriptive tradition of Slavic, these groups are referred to as diphthongs. The diphthongs *ɑi̯, *ɑu̯, *ei̯, *eu̯, *ɑ:i̯, *ɑ:u̯, *e:i̯, *e:u̯ existed in Proto-Slavic at least until the fifth century. From the systemic perspective, it is correct to interpret these groups as composed of two phonemes; and i̯ and u̯ are treated as combinatory non-syllabic allophones of the corresponding vowels. In later development, the long vowels in these diphthongs were shortened, similar to the case of syllabic sonorants. The long vowels in the diphthongs *e:r,*a:r, *e:l, *a:l were also shortened.
During the period of common development, diphthongs *ɑi̯, *ɑu̯, *ei̯, *eu̯ underwent monophthongization, producing long vowels. A new *u: emerged in this way (from *ɑu̯, *eu̯, *ou̯), filling in the systemic gap which appeared after the delabialization of the original long *u:. The lower vowels gradually changed their qualities, so that e and e: became lower, and a and a: became higher. Thus, around the seventh century, the late Proto-Slavic vowel system may have been as follows:
| *i | *u | *i: | *y: | *u |
| *ɑe|Footnote 4 | *ɑ° | *ɑe: | *ɑ°: |
The vowel *y: later changed into *ɨ. Certainly, there also existed the nasal vowels *ę and *ǫ, which are interpreted as variants of the groups of phonemes ‘oral vowel + nasal sonorant’ before a consonant and in the word-final position. If we take into consideration that only open syllables (i.e. syllables ending in a vowel) existed at that time (see also Chapter 4 in this volume), then such a context for the occurrence of a nasal sonorant would be uncommon, which supports the interpretation of *ę and *ǫ as independent phonemes.
The Indo-European word-final consonants were phased out in Proto-Slavic and a series of metatheses resulted in the elimination of consonantal codas (cf. *kɑr-vɑ > *krɑ(:)-vɑ ‘cow’, etc.). The emergence of nasal vowels and the monophthongization of diphthongs also contributed to the opening of syllables. Another restriction was the so-called syllabic synharmonism – a kind of vocalic harmony. It required the agreement of all segments in the syllable with respect to the feature +/− palatal.Footnote 5 This caused the replacement of the back vowels with the front ones after palatalized consonants.Footnote 6 Until the very early Middle Ages, Slavic vocalism developed in a uniform manner. After the seventh century, further changes took place, some of which were still shared: short *i and *u became lower, and in the ninth century (possibly even later in East-Slavic) turned into ‘extra-short’ or ‘reduced’ vowels (a kind of schwa, front and back, marked in Slavic source material by ь and ъ respectively; these are known in Slavic linguistics as front and back jers). At that period, jers differed in pronunciation according to their position in a word: weak jers occurred in word-final position and in a syllable which immediately preceded a syllable with a full (non-jer) vowel or a strong jer. Strong jers occurred in a syllable before a weak jer. Due to this arrangement, the jers soon transformed into full vowels (in strong position, a change termed vocalization) or disappeared (in weak position). In an example like *sъnъ ‘dream’ the first jer was strong and the second one weak; thus, in Polish it gave sen, in Russian sоn, in Serbian san, etc. Seeming exceptions, in which a weak jer seems to vocalize, are to be otherwise explained.
Some scholars speak of ‘secondary jers’, but in fact these are vowels that do not come from an original jer but are inserted later to remedy a difficulty in syllable structure. The secondary inserted vowel need not be the same as the reflex of either one of the strong jers (cf. Mac. *mьgla > mgla > mаglа ‘fog’ with a and neither o nor e that are the reflexes of the strong jers; Bel. *krъvь > krоŭ ‘blood’ with o from strong back jer but kryvаvy ‘bloody’ with the secondary inserted vowel y between kr and v).
The most uniform were the vocalizations of strong jers in the East (including Bulgarian and Macedonian), where the front and back jers produced different reflexes. In the rest of the Slavic area, however, both jers developed in the same way (although they did not produce the same reflex in all languages); at most, in some West-Slavic languages, the softening of the consonant was preserved before the reflex of the front jer.
The short *ɑ changed into *o.
The low front long vowel *ɑe: (= Proto-Slavic *e:, later, the so-called jat: *ě) developed dialectal variants.
The tautosyllabic groups *ɑr, *er, *ur, *ir, *al, *el, *ul, *il started to follow different developmental paths.
The articulation of the back vowel *y shifted forward; it changed into *ɨ and then, in part of the Slavic area, into *i.
The system of long and short vowels was reformulated as a result of the replacement of the original length contrast with value contrast and a series of shortenings and lengthenings.
After the emergence of jers and *o, but before the delabialization of *y and, possibly, further changes in the quality of low vowels, the vowel system may have been as follows:
| *ь | *ъ | *i: | *y | *u: |
| *e | *o | *e: | *ɑ: |
The vowel referred to as jat (*ě) developed from the long *e:. Initially, it must have been relatively low (in Old Church Slavonic texts it was written in the same way as ɑ after a palatalized consonant and the group j+ɑ). Its quality soon started to differentiate. At present, it has various reflexes in different languages: higher or lower, monophthongal or diphthongal.
Most of the numerous later changes were geographically restricted and did not influence the phonemic inventory, but enriched allomorphism (e.g. numerous so-called dispalatalizations in the North – changes of front vowels into back vowels in specific contexts).
Towards the end of the Proto-Slavic period, further changes in the vowel duration took place: lengthenings of short vowels and shortenings of long vowels in certain positions (e.g. shortenings in final positions, lengthenings under the so-called new rising accent or under the new and old rising accent, depending on the dialect, and on the South of the Carpathians also lengthenings together with the metathesis in the Proto-Slavic groups *TarT,*TalT, *TerT, *TelT, *arT, *alT, *erT, *elT,Footnote 7 cf. Croatian mlijeko ‘milk’, krava ‘cow’, in Polish mleko, krowa|; Croatian [ije] and [ɑ] are the reflexes of long vowels, Polish [ɔ] is a reflex of the short *ɑ). As a result, the correlation of the length was reconstructed – only jers lacked their long equivalents.
In the ninth century, the vowel system may have been as follows:
| *i | (*ɨ) | *u | *i: | (*ɨ:) | *u: | ||
| *e | *ь | *ъ | *o | *e: | *o: | ||
| *ě | *ɑ | *ě: | *ɑ: |
In the ninth century, *ɨ(:) was still phonologically independent, but on the South of the Carpathians, it may already have merged with *i. Also *ě(:) may have sounded different, depending on the dialect. Further shifts had taken place by the end of the thirteenth century. They changed the shape of morphemes, but only a few of them led to qualitative changes in the vowel system. Such were the elimination and vocalization of jers. The jers in the strong positions produced full vowels which already existed in particular Slavic dialects. Thus, the change consisted in the elimination of these phonemes, with the exception of Bulgarian, where the back jer preserved the quality of a schwa type. Other changes were restricted territorially.
2.2 Differentiation
Due to enormous territorial dispersion and the fact that Slavic tribes inhabited areas where other non-Slavic tribes lived, or where the Slavs neighbored non-Slavs, Slavic vocalism began to diversify. In some areas, certain archaic features were retained, in other areas, innovations appeared. Initially, mostly because of the phonetic development, three major groups formed: East-Slavic, West-Slavic, and South-Slavic. At present, this differentiation is reflected mostly in morphophonology. Further phonetic development led to a change in the typological arrangement. Today, with regard to phonetics, Slavic languages are divided into North-Slavic and South-Slavic (South-West), with the border line along the Carpathians (on the territory of Bulgaria, it runs across the Balkan Massif, where it becomes blurred). Typologically, Bulgarian is today counted in the North-East group. The former inclusion of Bulgarian in the South-Slavic group is reflected today in its morphophonology and syllable structure. Both Bulgarian and Macedonian are typologically the most varied from the geographical vantage point. The Bulgarian-Macedonian area constitutes a continuum, in which it is difficult to draw a well-defined boundary. The typologically different centers (which have given rise to present-day standard Macedonian and Bulgarian) are located in the West of Macedonia and in the East of Bulgaria. The contrast is more pronounced in the consonantism of the languages than in the vowel systems. It should be emphasized that in the South of the Slavic area there are no well-defined boundaries between languages – the boundaries are always formed by extensive transition bands.
Taking into consideration consonantal phonetic features, the North area seems more archaic, especially in the East, while the South-West languages have preserved more of the old vocalic features, especially vowel duration and accentual polytony, which is treated by some linguists as a vocalic segmental feature.
Apart from general differences of a systemic nature, smaller intersecting areas often emerged (characterized by specific features). This was caused by such factors as further changes in the vowel system, already geographically differentiated, phonotactic differences (especially in the structure of the syllable), the loss or preservation of assimilative palatality, and interaction with various non-Slavic dialects, especially in the area of the Balkan League.
In addition to the five mandatory so-called pure vocal segments, present-day Slavic languages have additional regional segments. Thus, we can distinguish at least the following: an area with long vowels, an area with a schwa-like vowel, an area with more middle vowels, an area with distinctive tonal features, an area with vowel reduction, an area with labiovelarization, an area with syllabic sonorants, an area with the so-called glottal stop, and areas with preserved nasalization.
The development of vocalism in North-West-Slavic languages is decidedly more complex. Here, the development of particular vowels was not only connected with the general developmental path of a given phone but was also conditioned by the length of the vowel and often by the consonantal context. As a result, several sources are reconstructed for most vowels. The most important consonantal features which influenced vowel changes are: +/− palatality, voicing, and, often, place of articulation.
Area with Distinctive Duration of Vowels
The northern boundary of this area is formed by the Carpathians, and the eastern and southern boundaries correspond to the borderline between Serbian and Bulgarian. In this way, long vowel phonemes occur in Czech, Slovak, Serbian, Croatian, Bosnian, Montenegrin, and Slovenian. Vowel length originates from the inherited length, and numerous lengthenings and contractions. Compared to Proto-Slavic, the length of vowels was repeatedly reformulated. In most Slavic languages which have preserved distinctive vowel duration, the occurrence of long vowels is significantly restricted distributionally. Only in Czech are there no restrictions on the occurrence of long vowels. In Slovak, two long vowels cannot occur in neighboring syllables in the same word. In the post-Shtokavian languages (Serbian, Croatian, Bosnian and Montenegrin), long vowels cannot occur before a stressed syllable. In Slovenian, long vowels occur practically only in stressed positions. In standard Macedonian, there are no long vowel phonemes, but in many dialects and in colloquial language long vowels are observed. They correspond to the vowel geminate of careful speech and are interpreted phonologically as such, for example, padnaa ‘(they) fell’ /pɑdnɑɑ/ [pɑdnɑɑ]/ [pɑdnɑ:]. On the other hand, Serbian dialects neighboring the Balkan Language League are characterized by a restricted occurrence of long vowels or they lack long vowels completely.
Area with a Schwa-Like Vowel Phoneme of Various Origins
This feature is characteristic of the North-East area.
(a) Origin from the Old Slavic *ɨ (Proto-Slavic *u:). This phone occurs in all East-Slavic and West-Slavic languages, with the exception of Czech and Slovak. It is written as y (Polish and Sorbian), ы (Russian, Belarusian), and и (Ukrainian). It is articulated as [ɨ] or a vowel of a similar quality. It is the only vowel which occurs exclusively after non-palatalized consonants. Therefore, after the termination of synharmonism and the phonologization of a number of palatalized consonants, Slavic languages have retained the complementary distribution of [i] and [ɨ]. However, the segment which conditions the quality of the whole CV group with respect to the opposition palatal/non-palatal was no longer the vowel (as in Old Slavic), but the consonant. The quality of the consonant determined whether it was followed by [i] or [ɨ]. The effects of palatalization (which are the base for morphophonological alternations) correspond at present to phones which are often significantly different from their non-palatalized counterparts, such as Pol. <t ~ ʨ> (e.g. płot [pwɔt] ‘fence’, loc.sg na płocie [nɑ pwɔʨɛ]), <s ~ ɕ> (kasa ‘cash register’, loc.sg w kasie [f kɑɕe]).Footnote 8
In all languages in which [ɨ] occurs, it was treated as a combinatory variant of the phoneme /i/. In the last 150 years, as a result of the intensification of contacts with the languages of Western Europe, a huge amount of foreign lexis has entered Slavic languages. Because the contexts ‘non-palatalized consonant + [i]’ occurred in the foreign lexis, in time the complementary distribution was phased out. This process is most advanced in Polish and in the Sorbian languages. In Polish, most users of the cultured variety of Polish pronunciation no longer use the exchange of [i] into [ɨ] in such words. There have even emerged minimal pairs, for example trik [trjik] ‘trick’ vs. tryk [trɨk] ‘ram’, plastik ‘plastic’ vs. plastyk ‘artist’. Although consonants become palatalized before /i/, the assimilation is not strong. The new palatalized counterparts of the non-palatalized phones before i in borrowings are phonetically much closer to one another. Consequently, with regard to most languages of this area, a change is being considered in the phonological status of [ɨ]. In the phonological system of Polish, /ɨ/ has been treated as an independent phoneme for a long time. Thus, the former palatalization correlation is reflected today only in morphophonological alternations, and the palatalizations which occur at present before /i/ result in most contexts in new allophony. This solution is not yet available for Russian, in which the former phonotactic rules still obtain, at least with regard to [i]/ [ɨ], and in which alternation still takes place even on word boundaries, for example brat Iriny [brɑt ɨrjinɨ] ‘Irina’s brother’.
(b) The second kind of centralized vowel is a type of schwa [ǝ] – lower and more central than [ɨ]. The phone [ǝ] takes its origins from the Old Slavic back jer *ъ and the back nasal vowel *õ, and also from the secondary vocalismFootnote 9. This type of centralized vowel occurs only in Bulgarian and a number of Macedonian dialects. It is written as ъ.
Generally, the area with the phonological schwa encompasses the whole eastern part of the Slavic language area. It is an archaic feature, whose preservation is probably a result of the long-lasting relations with various Turkic dialects, in which a similar sound occurs. It is one of the features which characterize the eastern area (the Eurasian area).Footnote 10
Various kinds of schwa occur also in the extreme western periphery of the Slavic area. The Slovenian [ǝ] has various origins, for example from reduction, from jers, from secondary vocalism, but also from short vowels, often even stressed ones, such as in the central dialect: brat [brǝt] ‘brother’, kruh [krǝx] ‘bread’ (Reference Tivadar and SawickaTivadar 2007). The phone [ǝ] occurs in Sorbian (apart from /ɨ/); however, its phonological status is unclear (according to the most recent descriptions – Reference Wornar and SawickaWornar 2007 and Reference JoczJocz 2011 – it is a variant of the unstressed /ɨ/ and /ɛ/).
Area with Additional Phonemes (Apart from the Five Cardinal Ones) and Diphthongs
This area also encompasses the western periphery (Sorbian and Slovenian), where there are additional mid vowels, and Slovak, where there are additional diphthongs.
(a) In the western periphery, there are two levels of mid vowels – in the Sorbian languages and in Slovenian. In these languages, the opposition occurs between /e/, /o/ vs. /ɛ/, /ɔ/. Moreover, Upper Sorbian has [ʊ] which is a kind of centralized mid-high back vowel. In Upper Sorbian, it is the reflex of the old long *o from the compensatory lengthening and the length connected with the old tonal accents. The high [o] in Sorbian languages, on the other hand, originates mostly from the combinations of *ow, *oł. In Lower Sorbian, [ʊ] occurs only in some dialects and originates from *o after labial and velar sounds.
In Slovenian, e and o is pronounced open and long [ɛ:], [ɔ:] where the accent has shifted on to it from the subsequent syllable (cf. žen||a [ʒen|ɑ] > ž|ena [ʒ|ɛ:nɑ] ‘wife, woman’, koz||a [koz|ɑ] > k||oza [k|ɔ:zɑ] ‘goat’). Short open o [ɔ] occurs only in final syllables and monosyllables. Open and short e [ɛ] is pronounced in some monosyllables.
o is pronounced long closed [o:] when derived from *ǫ and when derived from *o when the accent has not shifted. Short closed o [o] occurs only before the final l [w] (cf. vòl [vow] ‘ox’).
Slovenian e is pronounced closed long [e:] when derived from: *e with no accent shift, from *ę and from *ě (jat). Unstressed e is pronounced either [ɛ] or [ə].
The so-called raised vowels (mid-high [e] and [o] and [ɑ°]) occur vestigially in Polish dialects and are an effect of replacing the vowel length with the raised articulation. They disappeared from standard Polish in the nineteenth century. At present, they occur in Kashubian.
(b) In the descriptive tradition of Czech and Slovak, some diphthongs are treated as independent phonemes, although, in fact, they are morphophonemes. The following are distinguished in Slovak /ie/, /iɑ/, /iu/, /uo/, with the high non-syllabic vocoid (Reference Král‘ and SabolKrál’ & Sabol 1989); in Czech, there is /ou/ (Reference PalkováPalková 1994: 205). Additionally in Slovak, there is also a vestigial front low phoneme marked as ä, whose articulation is similar to [a]. It originates from the former nasal vowel after labial consonants. The most frequent present pronunciation of this letter is [e].
Numerous diphthongs occur in Kashubian, mostly as a result of frequent labialization and the decomposition of back vowels, for example for *o: [kwɔzɑ]/[kwɛzɑ] ‘goat’. Other reflexes of *o include [wɛ], [wi], [øw]. Similarly reflexes for *u: are [wu], [wɨ], [wi], for example [kwira] ‘hen’. Similar diphthongs may also (infrequently) take their origins from the pochylone á (mid-low back vowel), for example [ɑw], [ʌw].
(c) For some time the diphthong marked as /ÿe/ was posited in Croatian phonology (Reference BrozovićBrozović 1968 and other works by this author). This is the reflex of the Proto-Slavic long jat: *ě:. In fact, today this reflex is pronounced as [je:] or (in monosyllabic words) as [ije] and such a phonological interpretation is accepted at present (Reference Škarić, Horga and SawickaŠkarić & Horga 2007).
Area with Distinctive Tonal Features
(See also Chapter 1 in this volume.) This is an archaic, gradually withdrawing feature. Accentual polytony occurs only in Slovenian and all languages derived from the Shtokavian dialect, that is, Serbian, Croatian, Montenegrin, Bosnian, with the exception of the southernmost Balkanized Serbian dialects (Kosovo-Resavian and Torlak). Although it is accepted that tonal accent still occurs in Kosovo-Resavian, in fact, what matters is the difference in length. In all languages with polytony, there are significant distributional restrictions on particular tones. In ‘Shtokavian’ languages, rising accent requires two syllables for its expression. Thus, it does not occur in monosyllabic words or on the last syllable of polysyllabic words. Falling stress no longer has any restrictions at present, although in native words it does not occur in final syllables, and according to the ‘classic’ accentual system, which today occurs only in some regions, only the initial syllable can take falling stress.Footnote 11 The classic version of the accentual system, always appearing in phonological descriptions of Serbian and Croatian, practically does not exist in standard varieties of these languages. Apart from the cessation of shifting stress onto proclitics and the elimination of restrictions on the occurrence of falling stress, there is a clear tendency to neutralize the tonal opposition in short syllables. Slovenian has already passed all these stages – Slovenian polytony is in the state of atrophy: it is realized optionally, exclusively in long stressed syllables.
The issue is raised in this chapter because some linguists treat tonal markedness as a distinctive feature of phonemes. Reference BrozovićBrozović (1968), for instance, posits phonemes /V̀/ (short rising tone) and /V́/ (long rising tone) – the remaining types of stress result from distributional conditions. Reference JakobsonJakobson (1931) posits high and low vowel phonemes, which corresponds to tonal markedness. Reference TrubetzkoyTrubetzkoy (1958) analyzes falling tones as combinatory variants of the lack of tonal markedness. The most recent (and most intricate) interpretation is provided by Reference GvozdanovićJadranka Gvozdanović (1980). All these interpretations are based on distributional restrictions on stress in the ‘classical’ accentual system and on the separation of the distinguishing function of stress from its culminative function. In many cases, this resulted in shifting stress to another syllable while the tonal markedness remained connected with a given phoneme. There have also been developed systems locating stress placement between two syllables, which made it possible to eliminate the distinguishing function of tonal markedness (Reference Ivić, Drewniak and HeinzIvić 1965)Footnote 12 (cf. kȍren (short falling tone) [|kore:n] ‘root’, kòran (short raising tone) [k|orɑ:n] ‘Quran’).
Also Polish word stress is most often realized as falling tone, but the tone does not have a phonological function in Polish (cf. Reference DemenkoDemenko 1999; most grammar text books repeat the traditional opinion about the expiratory stress in Polish, which is not confirmed by research).
Area with Morphologically Regulated Place of Stress vs. Area with Phonetically Regulated Stress
(See also Chapter 1 in this volume.) Stress regulated phonetically, that is, falling on a particular syllable, is generally typical of West-Slavic languages: Polish (with the dominant penultimate stress), Czech, Slovak, Upper Sorbian, some southern dialects of Polish and southern Kashubian (with initial stress). Stress of this type occurs also in the extreme southern periphery of the Slavic area – with the dominant antepenultimate stress, which competes with penultimate stress – in standard Macedonian and western dialects of Macedonian. In Lower Sorbian, initial stress competes with penultimate stress. The remaining Slavic languages and dialects (East-Slavic, South-Slavic, except Macedonian and North-Kashubian) have morphologically regulated stress (the so-called free stress), thanks to which stress has a distinguishing function and may be interpreted as a segmental feature.
There are opposite tendencies in both these groups. In the languages in which stress is regulated phonetically, stress was initially determined within a phonetic (prosodic) unit. This resulted in modification of stress placement depending on the clitics attached to the stress-bearing unit. This feature is gradually retreating. Even in Macedonian, in which the principle of modifying stress placement depending on the composition of the prosodic unit is applied to the highest degree,Footnote 13 stress is no longer shifted to prepositions. In all languages in which stress placement is determined on the basis of the count starting from the end of the stress unit, stress shift in accentual units composed of a clitic and host has ceased to be implemented and is in decline (with the exception of the units composed by two clitics, as a preposition and short form of personal pronoun). In Macedonian, stress placement is still modified in stress units containing certain types of proclitics. The stress shift principle is best observed in the languages with initial stress.
Enclitics do not cause shifts of stress placement, although in Polish there are expressions which indicate that they also used to influence stress placement (e.g. powiedzmy ‘let us say’ [povjj|eʦmɨ]/ [povjj|eʣmɨ], where the original enclitic –my caused stress shift; the former status of –my as an enclitic is confirmed by the type of sandhi).Footnote 14 Until recently, stress shifting related to enclitics was claimed with regard to Macedonian, but even in this language the phenomenon is no longer relevant.
In languages with morphological regulation of stress, movability is disappearing and stress stabilizes on a particular morpheme. The process is very advanced in the South-Slavic languages (we claim that this has already happened in Slovenian, cf. Slovenian megl||a, vod||a ‘water’ > m||egla ‘fog’, v||oda, because gen.pl vôd, gôr). In that way the place of stress loses its distinctive value. In the south-western part of the Slavic area, there occur influences originating from the phonetic plane, for example shifting stress onto proclitics (which occurs only in some regions) or restrictions on stressing certain syllables (see above).
Area of the So-Called Reduction of Unstressed Vowels
This area breaks down into two subareas.
(a) With the so-called akanie (Russian, Belarusian). In Russian, the unstressed vowels which are not high and are preceded by a non-palatalized consonant undergo centralization. Different degrees of reduction are postulated depending on the distance between the unstressed syllable and the stressed syllable and the quality of the preceding consonant. The reduction of unstressed vowels preceded by a palatalized consonant is compiled with the result of combinatory palatalization (heightening) (cf. bеrёzа ‘birch’ [bjirjozə]). As a result of the reduction, [o] does not occur at all in unstressed positions. Mutatis mutandis, the situation is similar in Belarusian, except that the reduction of mid vowels consists in the lowering of articulation towards [ɑ], which is reflected in written forms, for example bjarоzа. The principles of reduction in the so-called ‘taraškievica’ differ in details from the common ‘narkamoŭka’,Footnote 15 which results from the differences in the adaptation of borrowings (differences in spelling).
(b) With the southern type of reduction, which occurs in Bulgarian and in peripheral Macedonian dialects in Bulgaria and in eastern Aegean Macedonia (in Greece). In this type, the vowels which are not high undergo raising. In standard Bulgarian, the principles of reduction are also strictly regulated. The degree of reduction depends also on the distance between a given syllable and the stressed syllable. In the peripheral east dialects, the reduction of mid vowels leads to the change of [o] into [u] and [e] into [i]. This type of reduction is determined geographically – the same type of reduction occurs in all northern Greek dialects, in which the reduction is actually stronger (mid vowels change into [u], [i], and unstressed high vowels disappear). The reduction is relatively the weakest in the case of [ɑ]. There are several types of reduction in Bulgarian (the so-called full reduction and partial reduction, cf. Reference StojkovStojkov 2002). In Macedonian dialects (in Pirin Macedonia and in the neighboring dialects in Greece, and also in the neighboring Bulgarian dialects) typically [o] and [ɑ] undergo reduction, while [e] is pronounced without any significant changes.
(c) The reductions discussed above occur in the eastern part of the Slavic area and they are well described and formalized (except for Ukrainian, in which reduction practically does not exist, although slight raisings are sometimes mentioned in the literature). Raising is more noticeable, on the other hand, with regard to the western periphery of the Slavic area – Sorbian, Slovenian, and northern Kashubian dialects. Generally, in Sorbian reductions are observed both at present and in the diachronic perspective. They are, however, so irregular that precise rules of reduction have never been formulated. What is known is that unstressed /ɨ/ and /ɛ/ after a non-palatalized consonant are pronounced predominantly as [ǝ]. In Slovenian, some vowels are pronounced as [ǝ], however, it is disputable whether this is a case of reduction, because it happens also in stressed syllables. Short mid-high vowels are pronounced as [ǝ] independent of the position relative to word stress. The vowel [ǝ] may also take its origins from the old jer. Consequently, two levels of mid vowels are distinguished in Slovenian only in the case of long vowels and these occur only in stressed positions (Reference ToporišičToporišič 2000, Reference Tivadar and SawickaTivadar 2007).
Area with Syllabic Sonorants (South-Western Area)
(See also Chapter 4 in this volume.) This area encompasses the whole south-western part of the Slavic area: Czech, Slovak, Slovenian, Macedonian, and Shtokavian languages. The syllabic [r̩] occurs in all of them; in Czech and Slovak the syllabic [l̩] also occurs. In the remaining languages enumerated above, all sonorants can be syllabic in less sonorous environments – this happens mostly in foreign names, except for [r̩], which occurs in native words. In all these languages, syllabic sonorants have the status of combinatory allophones, except for Croatian, where the syllabic pronunciation of [r̩] is vestigially preserved in the position next to a vowel, for example tȑo [tr̩o] ‘he rubbed’. In Slovenian, the syllabic [r̩] is interpreted as the group /ǝr/ due to the presence of the independent phoneme /ǝ/. In Macedonian, [r̩] is, in fact, more often pronounced as [ǝr] than as a strengthened rhotic. Syllabic sonorants derive from Proto-Slavic syllabic sonorants, from the combinations of sonorants with jers between consonants or between a pause (juncture) and a consonant or from borrowings. Proto-Slavic syllabic sonorants in the remaining Slavic languages have undergone decomposition into groups ‘sonorant + vowel’ or ‘vowel + sonorant’ or changed into a vowel. The syllabic l in native words changed into a vowel also in Shtokavian, namely [u, u:], for example *vḷ:k > [vu:k] ‘wolf’.
Area with the So-Called Glottal Stop
In some languages (especially Czech and optionally in Polish) a new quasi-consonant prosthesis is established – this is a kind of laryngeal occlusion or friction, usually referred to as ‘glottal stop’. This is a relatively new phenomenon and should be treated as a part of the combinatory realization of vowels. The occurrence of glottal stop is a progressive tendency, but we can still divide Slavic languages into those in which the transition between two vowels or between a consonant and a vowel on strong morphemic boundaries is abrupt (glottal stop is produced) (Czech and Polish) and the remaining ones in which the connection is smooth (or even a glide is formed between them). Laryngeal stop is pronounced between two vowels, between a consonant and a vowel on strong morphemic boundaries, and before initial vowels (cf. Cze. okno [ˀokno] ‘window’, Pol. aorta [ˀɑˀɔrtɑ], nauczać [nɑˀuʧɑʨ] ‘to teach’).
Peripheral Areas with Preserved Traces of Old Slavic Nasality (Polish, Disappearing Macedonian Dialects)
This is not a compact area. Nasality is preserved in Polish and Kashubian, on the one hand, and in the peripheral Macedonian dialect spoken mainly in Greece, on the other hand. Nasal vowels have been preserved also in Slovenian dialects in the Podjunska Valley (Reference StieberStieber 1969: 26, Reference RamovšRamovš 1936: 122–123). The development of nasal vowels in Slavic languages has passed through several important stages. Firstly, in Proto-Slavic, the rule of the open syllable caused the elimination of the contexts ‘vowel + nasal consonant + consonant or end of word’, which resulted in shifting nasality to the vowel. Consequently, nasal vowels emerged, whose phonological status was unclear. On the one hand, in the absence of the contexts ‘vowel + nasal consonant + consonant or end of word’ they could be interpreted as a realization of the group of phonemes ‘vowel + nasal consonant’; on the other hand, such a solution would allow closed syllables, which otherwise did not occur in Common Slavic. Phonologization could occur only after the loss of weak jers, when the contexts of the type ‘oral vowel + nasal sonorant’ emerged in final positions and opposed nasal vowels in the same position (cf. *tъnъ ‘this’ > Pol. ten vs. tę ‘this, fem.acc’).
In most Slavic languages, nasal vowels have lost nasality and became oral vowels. Nasality has been preserved in Lekhitic languages, but nasal vowels have disintegrated at an indeterminate time, and in most contexts they have been replaced by groups of two segments, the second of which was a nasal stop or a nasal approximant, depending on the following context. Although the Polish descriptive tradition maintained for many years that synchronic nasality of vowels had been preserved before fricatives, this was rather an incorrect phonetic interpretation and not a phonetic fact. At present, it often happens that before a non-labial fricative, the second segment – high back non-labial approximant [ɯ̃] – completely loses nasality, of which the recipients are unaware. Before a labial fricative, [ɱ] is realized. Another frequent pronunciation, less normative, represents a return to the group ‘vowel + nasal sonorant’, for example wąski ‘narrow’ [voɯ̃sci]/[vonsci] – the nasal sonorant in this context is usually articulated without closure. One way or another, firstly, nasal vowels occur at the morphophonological level; secondly, in the pronunciation of most Poles, these contexts have equalized with the pronunciation of the groups ‘vowel + nasal sonorant + consonant’ from borrowings, like infekcja, emfaza, finansowy. Before stops, old nasal vowels have decomposed into the groups V + N ([m, n, ɲ, ŋ]).
Nasality has been preserved also on the opposing pole of the Slavic area in Macedonian dialects located in northern Greece. Generally, nasality is preserved only before stops, which is connected with the Greek functional equivalence of the groups ‘nasal stop + voiced stop’ with voiced stops. Nasality here is also determined by a Slavic factor, namely, nasality has been preserved if the nasal vowel had a quality similar to schwa.Footnote 16 This is evidenced by the non-etymological nasality which developed with the secondary vocalism before a stop consonant (e.g. *mьgla ‘fog’ > *mgla > məgla > [mɑŋgɬɑ]/[mǝŋgɬɑ], cf. Reference Sawicka and CychnerskaSawicka & Cychnerska 2018). If nasality was sometimes preserved before a fricative, it was only because a stop was inserted between a nasal sonorant and a fricative, which changed the context – a nasal vowel occurred before a stop, for example *gęs ‘goose’ > [gǝns] > [gǝnts] > [gǝnʦ].Footnote 17 Unmotivated nasality before a stop occurs in the same dialects, too, for example [fɑmbrikɑ] ‘factory’, [bɑrɑŋgɑ] ‘barrack’.
2.3 Quantitative Relations
The above differences affected the quantitative relations. Phonological systems are usually divided into vocalic and consonantal ones (Reference IssatchenkoIssatchenko 1939–1940).
In general, all Slavic languages are consonantal languages. Nevertheless, certain differences are significant. Most often, a modest vowel system is accompanied by more elaborate consonantism, extended by additional palatal phonemes. Conversely, rich vocalism is accompanied by a more modest consonant system. For instance, the ratio of vowels to consonants in Serbian and Croatian phonemic inventories is approximately 1:2, in Czech 1:2.5, in Slovak 3:4, whereas in Russian and Bulgarian it is approximately 1:6. Exceptions to this apply to the periphery: in Macedonian, both subsystems are modest, in Kashubian both are relatively rich. Because all Slavic languages inherited mostly the same morphemes, the frequency ratio of vowels to consonants in text is not so varied. Nevertheless, it can be said that the south of the Slavic area is characterized by a greater proportion of vowels in text. On the basis of comparable samples, it has been determined that texts in the South-West-Slavic languages contain more than 45 percent vowels, whereas texts in the languages spoken North of the Carpathians and in Bulgarian and Slovenian include from 37 percent to 42 percent vowels (cf. Reference Korytowska, Sawicka and SawickaKorytowska & Sawicka 2007: 202). However, very significant differences relate to the occurrence of vowel groups. The languages with the lowest frequency of vowel groups are those that have partly preserved the Proto-Slavic consonant prostheses or have developed new prostheses (Sorbian, Belarusian, and Ukrainian). Balkan languages, on the other hand, have several times higher frequency of vowel clusters. Vowel groups in Slavic languages occur mostly in borrowings and on morphemic boundaries (this is why they are infrequent in the languages in which there are still consonant prostheses before morpheme-initial vowels). Sources of vowel groups are more numerous in the south of the Slavic area: in the post-Shtokavian languages the change of l into o at the end of the syllable, for example Ser. video ‘saw’; inflection of foreign words ending in a vowel, for example Ser. kupe ‘train compartment’, gen.sg kupea, dat.sg kupeu, etc.; in Macedonian and Bulgarian, there is a frequent elision of j before a front vowel in word-initial position and between two vowels. These processes have contributed to the exceptionally high frequency of vowel groups in all South-Slavic languages, except for Slovenian. For instance, in three pages of text in Serbian, there were 93 vowel groups, in Croatian 100, and in Macedonian as many as 173, whereas in Polish and Russian there were 30 vowel groups in each language, in Czech 25, in Slovak 17, in Ukrainian 6, and in Upper Sorbian only 4 (for details, see Reference SawickaSawicka 2007).
