UDK 808.63-4:808.67-4 Frederik Kortlandl Državna univerza v Leidnu
THE ACCENTUATION OF NEUTER NOUNS IN SLOVENE AND WEST BULGARIAN
Zahodnobolgarska narečja krepijo zanesljivost slovenske razvidnosti za praslovansko kolikost v končnih zlogih. In narobe: slovenska gradiva nudijo podlago za razlago in razvoj naglasnih razredov v zahodnobolgarskih narečjih.
The West Bulgarian dialects corroborate the reliability of the Slovene evidence for Proto-Slavic vowel quantity in final syllables. Conversely, the Slovene material offers a basis for explaining the rise and development of accent classes in West Bulgarian dialects.
The Slovene neo-circumflex is our major source of information for the reconstruction of Proto-Slavic long vowels in posttonic syllables (cf. Kortlandt 1976). As I have pointed out earlier (1975: 27 and 1976: 3f.), trisyllabic word forms which have medial stress as a result of the Proto-Slavic progressive accent shift known as Dybo's law manifest a long reflex of originally acute vowels in final syllables, e.g. osnqva 'base', zabâva 'amusement', nosila '(she) carried', pisâla '(she) wrote', telęta 'calves', ženami 'women (inst.)', but gostila '(she) treated', kovala '(she) forged', also sedéla '(she) sat', želela '(she) wished', where the stress had never been on the initial syllable. The resulting metatony in the Slovene /-participle was extended by analogy in the dialects, e.g. želela, also mislila, videla beside mislila '(she) thought', videla '(she) saw' (cf. Rigler 1970 on the geographical distribution of these variants). Thus, we reconstruct a long final vowel in Proto-Slavic *osnövä, *zäbavä, *nosilâ, *pbsàla, *telęta, *ženami, but a short final vowel in *sedèla, *želela, also imenà 'names', *žena 'woman'. The rise of the long vowel must be attributed to an early loss of the acute in post-posttonic syllables, as I have argued earlier (1975: 11 and passim).
As a result of the development in post-posttonic syllables, the unstressed nom. acc. pi. ending of the neuter consonant stems was always long, e.g. *sèmenâ 'seeds', *plemènâ 'tribes', agnęta 'lambs', *teląta. This length spread to the unstressed nom. acc. pi. ending of the neuter o-stems, e.g. *lètâ, Sin. letä, 'years'. I have dated this analogical development to the period before Dybo's law (1975: 32 and 1976: 5). It must have been anterior to the loss of the acute in stressed syllables because it did not affect such forms as drvà 'firewood', which has original final stress. We may now ask the question: did the length spread to all neuter o-stems with an untressed nom. acc. pi. ending before Dybo's law? I think that it did, and this is substantially in agreement with Stang's view (1957: 83), but I will return to the question at the end of this article. Thus, I assume analogical lengthening in *sèlâ, after the Proto-Slavic progressive accent shift *selâ, then retraction of the stress according to Stang's law yielding *sèla with a neo-acute root vowel and a short ending, Sin. sęla 'settlements'. The accentuation of pôlja 'fields', which represents Proto-Slavic end-stressed *pol'à, must have been taken from the
singular, where it is regular after a preposition, e.g. loc. sg. pôlju (cf. Kortland 1976: 7). The neo-circumflex eventually spread to such forms as vina 'wines' (ibidem).
If this is correct, we have to reconstruct the following nom. acc. sg. and pi. forms of disyllabic neuters for the end of the Proto-Slavic period:
(a)	*lèto, *lètâ\
(b)	*selô, *sèla < *selâ;
(c)	*pol'e, *pol'à.
Neuter consonant stems had unstressed *-ä and stressed *-à in the plural form. The long ending was generalized in Slovak and Babina Greda Posavian, and the short ending in neoštokavian and Omišalj čakavian (Vermeer 1984: 374). Most Posavian dialects tend to have a long ending in all cases except drvâ, while Novi čakavian has a short ending in disyllabic neuters and a long ending in consonant stems and in the isolated expression na mestâ 'to the fields' (ibidem). The long ending of Slk. mestâ 'towns' and Varoš Posavian fît ä 'cereals' must go back to the Proto-Slavic period becouse it provides the model for the analogical extension of the long vowel. Vermeer has pointed out that at least in a part of the Proto-čakavian dialects this analogical extension must have taken place before Stang's law already in order to account for such instances as Novi jàja 'eggs', jelita 'entrails', pola 'fields', Omišalj jâja (1984: 375). These nouns belong to the mobile accent pattern (c), while the distinction between (b) and (c) nouns has been preserved in Omišalj loc. pi. selTh 'villages' versus jajth 'eggs' (with loss of the tonal contrast).
We now turn to the Bulgarian evidence. While case forms were lost in this language, the paradigm was enriched by the rise of the definite article. Like other enclitics, the article attracted the stress from barytone forms of nouns belonging to the mobile accent pattern (c), e.g. kosttà 'the bone'. The origin of this accent shift, which is known as Dolobko's law, can be dated to the period before Dybo's law (cf. Kortlandt 1975: 39). Slovene lahkega, lahkemu 'light (gen., dat.)' show that the stress shifted to the final syllable of the enclitic pronoun. Elsewhere I have argued that the stress was retracted from a short vowel in a final open syllable to a preceding open syllable in early Bulgarian (1982: 93), e.g vino 'wine', sèlo 'village', rešeto 'sieve', kostite 'the bones', MBulg. vlbkatôgo 'of the wolf, as opposed to rehrd 'rib', vlaknô 'fiber', mladosttà 'the youth', sin mù 'his son'. As a result of this retraction, the stress falls on the vowel before the article in neuter (c) nouns, e.g. senôto 'the hay', brasnöto 'the flour', polèto 'the field'. This accentuation spread by analogy to the indefinite form in many dialects and in the literary language, e.g. seno, brašnd, polè.
On the basis of all these developments, we can predict the following phonetic reflexes of the Proto-Slavic accent patterns established above:
(a)	*lèto, *lètoto, *lèta, *lètata;
(b)	*sèlo, *selôto, *sèla, *selàta;
(b') *rebrô, *rebrôto, *rèbra, *rebràta;
(c)	*pôle, * polèto, *pàl'a, *pol'àta;
(c') *brašno, *brašndto, *brašna, *brašnata.
Here we may add end-stressed seid, selà, rebrà, polè, pol'à, brašnd under the influence of the definite forms.
We now arrive at the central question of this article: are there traces of the long plural ending *-ä in Bulgarian? It goes without saying that there is no direct evidence in (a) nouns because vowel quantity was lost in this linguistic area, but it is possible that the long vowel spread to (c) nouns, as it did in Slovak and Posavian, and if this is actually the case, we expect pattern (c') instead of (c) because the stress was not retracted from long vowels (cf. Kortlandt 1982: 96). We must therefore examine the distribution of the accent patterns in the separate dialects, in the spirit of Rigler's work. If the accent patterns of sèlo and polè are kept distinct, as they are in the literary language, we have to reconstruct a long ending in *pol'â for the stage at which the Bulgarian accent retraction took place.
There are several complications which have to be taken into account. First of all, there is a tendency toward generalization of an accentual opposition between singular and plural forms, as in Russian. The opposition in indefinite (b') and (c') nouns can easily be extended to the corresponding definite forms. Secondly, the early Bulgarian accent retraction did not reach all dialects, as is clear from the Banat paradigms rakb, rakhta, rbce, rxetè '(the) hand(s)', sbrci, sarcito, sarcà, sarcàta '(the) heart(s)' (Stojkov 1967: 157f.). This is probably a southeastern feature. Thirdly, the original state of affairs has often been obscured by later retractions, especially in the western dialects. For example, plural forms have generalized retracted stress in the area of Blagoevgrad, giving rise to a paradigm vino, vindto, vina, vinata '(the) wine(s)'. West Bulgarian and Macedonian texts from the middle ages exhibit essentially the same accentual system as texts of eastern provenance, though the number of variants is larger (cf. Kortlandt 1982: 94). It may therefore be expedient to examine the distribution of accent patterns (c) and (c') in the West Bulgarian dialects.
Ronelle Alexander has recently studied the relevant material in some detail (1988). She lists nine major accent patterns:
(1)	sito, sitoto, sita, si tata;
(2)	vino, vindto, vina, vinata;
(3)	zbrno, zbrnoto, zbrna, zarnàta;
(4)	krilo, krilôto, krila, krilata;
(5)	pèro, pèroto, perà, peràta;
(6)	vlaknô, vlaknöto, vlàkna, vlaknata;
(7)	rebrö, rebrdto, rèbra, rebràta;
(8)	sèno, senôto, senà, senàta;
(9)	licè, licèto, licà, licàta.
On the basis of the considerations set forth above I predict the following reflexes of the Proto-Slavic accent patterns:
— (a) yields (1). If the long plural ending *ä spread to (c) nouns, we expect that analogical introduction of accentual mobility in (a) nouns may lead to a merger with accent pattern (c) rather than (b);
-(b) and (b') yield (4) and (7). Furthermore, regularization of (7) may yield
either (6) or (9);
-(c) and (c') yield (4) and (8). Furthermore, regularization of (8) may yield
either (5) or (9). If the long plural ending *-a spread to (c) nouns before the Bulgarian accent retraction, all nouns of this class should belong to pattern (8), and subsequently perhaps to (5) or (9).
I shall now briefly discuss the material of the eight dialects from which Alexander adduces the largest number of examples. It must be noted that sarce will be classifield as (c), not (b), cf. Sin. srcç. Like Alexander, I shall treat the dialects anti-clockwise, starting from the southeast.
I.	Smoljan: (b) and (c) yield 20 x (8), 8 x (5), 3 x (1), and (4) in uho 'ear', while (a) yields 10 x (5), 4 x (8), 2 x (1). It appears that (b) nouns adopted pattern (c), which had a long plural ending *-ä, except in the word uho, which had the dual ending *-j; (a) nouns adopted accentual mobility at a stage when (b) and (c) nouns were already replacing (8) by (5). The nouns with fixed stress (1) are consonant stems.
II.	Sadovo: (c) yields (4) in uho, oko, darvo, variation between (4) and (8) in zlato, more, nebo, pole, vreme, (1) in ime 'name', further 8 x (8), and (9) in prase 'pig'; (b) yields 13 x (8), including mleko, which must be added to Alexander's list, 2 x (4), variation between (4) and (8) in srebro, (5) in dano and in runo, which must be added to Alexander's list, (9) in tele, (1) in lice (cf. Georgiev 1907: 433);
(а)	yields 5 x (5),4 x (8), 1 x (4), 1 x variation between (4) and(8),2 x (1). Itfollows that the short plural ending *-a was preserved in *drvà and probably in *mor'à, *pol'à, *nebesà, *vrëmenà, *imenà, while most other nouns may have had the long ending *-ä; the variation between (4) and (8) in zlato 'gold' and srebro 'silver' can easily be due to mutual influence. As in Smoljan, the reflex of pattern (c) was adopted by (b) nouns and later by (a) nouns.
III.	Ihtiman: (c) yields (4) in uho, oko, darvo, 6 x (8), 2 x (9), further (7) in červo 'intestine' and (5) in ime and vreme\ (b) yields (4) in selo and platno, (8) in mleko, lice, čelo, pismo, (7) in rebro, vedro, jajce, variation between (7) and (9) in vlakno, further 4 x (9), 1 x (1); (a) yields 3 x (1), 3 x (8), 2 x (5). Here we must add prase (9), tele (9), agne (5) and (9), which are missing in Alexander's list (cf. Mladenov 1966: 113). It appears that the short ending *-a was preserved in *drvà and the long ending generalized elsewhere, while the Bulgarian retraction of the stress is reflected in selo, mleko, lice, čelo, and its absence in rebro, vedro, jajce, vlakno, by analogy červo\ (a) nouns adopted the mobile pattern to a lesser extent than in Smoljan and Sadovo. In the dialect of Dobroslavci, which is structurally close to Ihtiman, pattern (b) is reflected as (4) in pero and 10 x as (8) or (9).
IV.	Gorno Pole (near Stanke Dimitrov): (c) yields 8 x (4), 3 x (2) or (4), further (1) in zlato, variation between (1) and (3) in ime, and (9) in Irebe 'foal'; (b) yields
(б)	and/or (9) in gnezdo, platno, rebro, sedlo, vlakno, vedro, jajce, srebro, sukno, dleto, lice, vino, rešeto, (4) in krilo, pero, gumno, pismo, (2) or (4) in mleko and runo, variation between (4) and (6) or (9) in selo and čelo, and (1) in dano-, (a) yields4 X (1), 3 x (4), 1 x (2) or (4), 1 x (3), and 2 x variation between (6) and (1) or (4). I conclude that there is no trace of the long plural ending *п, while the Bulgarian accent retraction is reflected in krilo, pero, mleko, runo, selo, čelo, dano, as opposed to gnezdo, platno, rebro, sedlo, vlakno, vedro, jajce, srebro, sukno.
V.	Kjustendil: (c) yields (4) in meso, variation between (4) and (8) or (9) in darvo and oko, between (7) and (9) in uho, further (3) in ime, (6) in krosno 'beam', (8) in more and vreme, (9) in prase and irebe\ (b) yields (7) in platno, sedlo, (6) in jajce, tele, (4) in dano, čelo, pismo, further 6 x (2), 1 x (1), 1 x (9); (a) yields 8 x (1), 1 x (2), 1 x (8). This points to a short plural ending *-a in *drvà and *imenà and a long ending *-ä in *prasqtä and *irębątd, while the Bulgarian accent retraction is reflected in dano, čelo, krilo, pero, selo, as opposed to platno, sedlo, jajce.
VI.	Šiškovci (Kjustendilsko Pole): (c) yields (4) in darvo, oko, telo, (2) in meso, sarce, perhaps (I) in vreme, (8) in more, nebo, pole, (9) in uho, žrebe\ (b) yields 7 x (2) and/or (4), 1 x (1), further (6) in jajce 'egg', (9) in vate 'rope'; (a) yields 7x (1), lx (2), lx (5),2x variation between (1) and (5), including stado, to be corrected in Aleksander's list (cf. Bojadžieva 1931: 255f). This points to a long plural ending *-ä in *nebesâ and *źrębętd and a short ending in most other nouns. The Bulgarian accent retraction was generalized in (b) nouns, except for jajce.
VII.	Leško (near Blagoevgrad): (c) yields (4) in oko, (2) in darvo, seno, meso, brašno, (1) in ime, more, nebo, vreme, proso, sarce, (6) in krosno, testo, zlato, prase, uho, trebe: (b) yields (6) in gnezdo, platno, rebro, sedlo, vlakno, vedro, srebro, sukno, lice, dano, (2) in selo, vino, (1) in krilo, mleko, čelo, runo, pismo, gumno, (9) in vaze; (a) yields 15x (1) only. This dialect generalized retracted stress in all plural forms except očite 'the eyes'. It suggests a short ending in *drvà, *imenà, *vremenà, *nebesà, and a long ending in *prasqtà, *trębętó; the accentuation of krosno and zlato may have been taken from platno and srebro, respectively. It testifies to the Bulgarian accent retraction in selo, vino, krilo, mleko, čelo, runo, and to its absence in gnezdo, platno, rebro, sedlo, vlakno, vedro, srebro, sukno. It displays no analogical mobility of the stress in (a) nouns.
VIII.	Goce Delčev: (c) yields (8) in meso, darvo, pole, sarce, (5) in zlato, vreme, telo, variation between (1) and (5) in ime; (b) yields 8x (5), including vino, which is missing in Alexander's list, further (8) in rebro, (1) in tele; (a) yields 7x (5), but (1) in agne 'lamb'; Alexander mistakenly writes (4) instead of (5) for this dialect (cf. Mirčev 1936: 58f.). This suggests that neuter o-stems had a long plural ending *-ä; the Bulgarian accent retraction was apparently generalized in (b) nouns, as in Šiškovci, and eventually (a) and (b) nouns adopted accentual mobility at a stage when (c) nouns already replacing (8) by (5), as in Smoljan. The remaining nouns with fixed stress (1) are consonant stems. There are traces of pattern (4) in öko, oköto, oči, očite, uši, ušite, which Alexander does not mention.
We may now try to put the evidence together in order to arrive at a tentative reconstruction of the original distribution of the long and short plural endings in West Bulgarian. It appears that all dialects may have had a short ending in *drvà, *imenà, *vremenà, and a long ending in *prasęta, *žrebqtd, and at least some of the o-stems. The long ending may to a large extent have been generalized in the northern and eastern dialects (I, II, III, VIII), but not in the western dialects (IV, V, VI, VII). There is direct evidence for the Bulgarian accent retraction in the north (III) and the west (IV, V, VI, VII), while mobility was generalized in the Rhodope dialects (I, II, VIII), as was retracted stress in plural forms in the southwest (VII).
Turning back to the other Slavic languages, we see that the Bulgarian evidence for a long plural ending *-ä is in agreement with the material from Slovene and its neighbors. The analogical extension of the long vowel to *prasętd and *irębęta parallels the transfer of these words from class (c) to class (b) in Slovene (cf. Stang 1957: 93). The preservation of the short ending in *drvà is also found in Slovene and Posavian. The preservation of the short vowel in *imenà, *vremenà and the generalization of the long vowel in *prasątó, *irebetâ are reminiscent of the preservation and generalization of accent pattern (c) in modern Russian imend, vremend, plemenâ, semenà, and the generalization of pattern (b) in porosjâta, žerebjdta, jagnjâta, cf. teljâta. In the o-stems, the long ending *-â must have spread to (b) nouns before Stang's law already because these would otherwise have joined pattern (c) as a result of the early Bulgarian accent retraction, yielding patterns (4), (8), (5), (9), but never giving rise to (7) and (6). After Stang's law, the long plural ending in the o-stems was limited to accent pattern (a), from where it spread to (c) before the Bulgarian accent retraction because these nouns would otherwise have joined pattern (b), which yielded (4), (7), (6), (9) but did not give rise to (8) and (5). There can be no doubt that the accent retraction affected the West Bulgarian dialects, not only because it is directly reflected in the accentual distribution of the material (see especially dialects IV and VII above), but also because it gave rise to accent pattern (4), which would not otherwise have come into existence. An exception must be made for the dialect of Smoljan (I), where the material is inconclusive.
Thus, the West Bulgarian dialects corroborate the reliability of the Slovene evidence for Proto-Slavic vowel quantity in final syllables. Conversely, the Slovene material offers a basis for explaining the rise and development of accent classes in West Bulgarian dialects.
References
Alexander, R.
1988 The accentuation of neuter nouns in Balkan Slavic, American contributions to the tenth international congress of slavists (Columbus: Slavica), 7-34. Bojadžieva, J.
1931 Kjustendilskitč polčani i tèhnijat govor, Izvëstija na Seminara po slavjanska filologija pri Universiteta v Sofija 7, 181-329. Georgiev, S. pop
1907 Po govora v s. Češnegir Nova-Mahala (Stanimaško), Izvèstija na Seminara po slavjanska filologija pri Universiteta v Sofija 2, 411-444. Kortlandt, F.
1975	Slavic accentuation: A study in relative chronology (Lisse: Peter de Ridder).
1976	The Slovene neo-circumflex, The Slavonic and East European Review 54/134, 1-10.
1982 Sravnitel'no-istoričeskie kommentarii k bolgarskomu udareniju. Zbornik za filologiju i lingvi-stiku 25/1, 91-96. Mirče v, K .
1936 Nevropskijat govor (Sofija: Pridvorna pečatnica). Mladenov, M.
1966 Ihtimanskijat govor (Sofija: Izd. na BAN). Rigler, J.
1970 Akcentske variante, Slavistična Revija 18, 5-15.
Stang, C. S.
1957 Slavonic accentuation (Oslo: Universitetsforlaget). Stojkov, S.
1967 Banatskijat govor (Sofija: Izd. na BAN). Vermeer, W . R .
1984 On clarifying some points of Slavonic accentology: The quantity of the thematic vowel in the present tense and related issues, Folia Linguistica Historica 5/2, 331-395.
POVZETEK
Zahodnobolgarska narečja krepijo zanesljivost slovenske razvidnosti za praslovansko kolikost v končnih zlogih. In narobe: slovenska gradiva nudijo podlago za razlago in razvoj naglasnih razredov v zahodnobolgarskih narečjih.
Slovenski novi cirkumfleks je močan vir obvestilnosti za vzpostavitev praslovanskih dolgih samoglasnikov v ponaglasnih zlogih. Trizložne besedne oblike s sredinskim naglasom kot posledico praslo-vanskega progresivnega naglasnega premika imajo, kakor se kaže po Dybojevem zakonu, dolgi odraz prvotno akutiranih samoglasnikov v zadnjem zlogu, npr. osnçva, zabava, nosila, pisala, telila, ženami proti gostila, kovala, sedéla, ieléla (po analogiji želela ter mislila). Zato za praslovanščino vzpostavljamo *osnàvâ, *zäbävä itd. proti sedela itd. Nastanek dolgega samoglasnika je posledica zgodnje izgube akuta v poponaglasnih zlogih. To stanje se na svoj način odraža v zahodnih bolgarskih govorih, ki so tu zajeti v 8 skupinah.