Early Slavic short and long o and e
Tijmen Pronk
UniversiteitLeiden, P.N. van Eyckhof 2, NL - 2311 BVLeiden, T. C.Pronk@hum.leidenuniv. nl
- SCN IX/1 [2016], 5-33 -
Članek obravnava razvoj praslovanskih novoakutiranih vokalov *o in *e, ki se v sodobnih slovanskih jezikih odražata kot kratka vokala, dvoglasnika ali dolga vokala. Njun odraz je pogojen z izvorom novega akuta - če je nastal z naglasnim umikom s končnega jera, sta se novonaglašena *o oz. *e podaljšala, kasneje pa v številnih slovanskih narečjih diftongizirala; če je novoakutirani glas nastal na drug način, je kvantiteta novonaglašenega *o ali *e odvisna od narečja.
The article1 discusses the development of the Proto-Slavic vowels *o and *e with a neoacute accent. These vowels are reflected as short vowels, diphthongs or long vowels in the modern Slavic languages. Their outcome is conditioned by the origin of the neoacute: if it arose through retraction of the accent from a word-final jer, the newly accented *o or *e became long and was subsequently diphthongized in a number of Slavic dialects. If the neoacute accent arose in a different way, the quantity of the newly accented *o or *e depends on the dialect.
Ključne besede: vokal, novi akut, praslovanščina, slovanski jeziki
Key words: vowel, neoacute, Proto-Slavic, Slavic languages
1 *O and *e in Slavic
It is a well-known fact from the historical grammar of Slavic that the vowels *o and *e go back to historically short vowels. Nevertheless, late Proto-Slavic
1 Part of the research for this article was conducted within the project Documentation and Interpretation of the Earliest Croatian, financed by the Croatian Science Foundation.
I am much obliged to Frederik Kortlandt with whom I discussed many of the problems discussed in this article.
— 5 —
Tijmen Pronk
*e or *o could be long or short. Long *e and *o arose under the following circumstances:2
1.	through contraction after the deletion of intervocalic *j, first in posttonic syllables, e.g. Cz. nom.sg.n. nové, gen.sg. nového (Vaillant 1950: 195, Kort-landt 2011: 38).
2.	through assimilation of a preceding *j, e.g. *píse- < *písje- (Kortlandt 2015: 66).
3.	under a rising tone (a "neoacute"3) (Van Wijk 1916, Vaillant 1950: 265ff.).
It is the last category this paper is concerned with. When we collect all examples of neoacute *e or *o we can distinguish two categories:
1.	*e or *o which is reflected as a long vowel or diphthong in all Slavic languages which provide evidence about vowel length, with a few minor exceptions.
2.	*e or *o which is reflected as a long vowel or diphthong in some languages but as a short vowel in others.
The variation we find is connected with the fact that the rising accent referred to as "neoacute" has multiple origins (cf. Kortlandt 2015: 69). I will give a brief explanation of the origin of the different reflexes, after which the relevant material will be discussed. We can distinguish four processes which resulted in rising accents that can be referred to as a neoacute (in the examples below I will write the neoacute with a grave accent (*é, *ó) regardless of its length):
a)	neoacute that arose as a result of the retraction of the accent from a final jer onto a preceding syllable, which obtained a rising accent, e.g. 2sg.pres. *nesesb > *nesésb, l-ptc. m.sg. *nesly > *nésl\>.
b)	neoacute that arose as a result of Dybo's law, according to which a non-acute, non-falling accent shifted to the following syllable. If that syllable contained
2	There are more early lengthenings, but those are limited to smaller dialect areas, e.g. lengthening before voiced consonants in closed syllables (e.g., Polish bog < *bogb). Lengthening in monosyllabic forms with a falling accent (e.g. Sln. bog < *bogb) might be a common Slavic development (Kortlandt 2011: 305), but length is found in western South Slavic only. In this paper I have chosen to consistently use the term "Proto-Slavic" to refer to the language spoken until the last innovations common to all of Slavic no earlier than the first half of the 9th century, without distinguishing between "Proto-Slavic" and "Common Slavic", "Early Proto-Slavic" and "Late Proto-Slavic" or using similar formal periodizations because such a distinction would have no influence on the outcome of the present discussion.
3	In Slavic accentology, the term "neoacute", in its widest sense, refers to a subgroup of the rising accents that are reconstructed for Proto-Slavic. They contrast with the "acute", which is a short rising accent on the vowels *a, *i, *y, *u, or or a diphthong. The most precise definition of what a "neoacute" is, is the following: 1) any rising accent on the vowels *e, *o, *t or (e.g. *sedmb, gen.pl. *nogb, *dim>, nom.sg. *bobb, nom. sg.m. *desnb, gen.sg. *potoka); 2) a long rising accent on the vowels *a, *i, *y, *u,
or *9 or a diphthong (e.g. *susa, *storza, gen.pl. *zimb). In medial stressed syllables (not counting word-final jers), the difference between neoacute and acute cannot be seen directly, because Proto-Slavic had no distinctive length in medial stressed syllables.
— 6 —
tS^la-Via Centra lis 1/2016
	 Early Slavic short and long o and e 	
a short vowel, it obtained a rising accent, e.g. *pdtok-b > *potdk-b. Dybo's law did not shift the accent onto a word-final jer (Kortlandt 2011: 17-19, 170f.), with the possible exception of final jer preceded by a tautosyllabic resonant (*-CRb/b) in Czech and South Slavic, see below. Accordingly, the old nonfinal rising accent of nom.sg. *kdnb, *otbcb, gen.pl. *zenb was preserved. This rising accent is identical to the one in *potdkb.
c)	neoacute that arose as a result of Stang's law, according to which a long falling accent in a final syllable (not counting final jers) was shifted to the preceding syllable, which obtained a rising accent, e.g. 2sg.pres. *mozesb > *mdzesb.
d)	neoacute that arose as a result of the retraction of the accent from an internal jer onto a preceding syllable, which obtained a rising accent, e.g. *konbskbjb > *kdnbskbjb.
These four developments affected all of Slavic but they did not take place at the same time. The relative chronology of these changes accounts for the length differences of neoacute *o and *e in the individual Slavic languages we will now proceed to discuss. Note that a) affected words belonging to the Proto-Slavic accent paradigm c, while b), c) and d) affected words belonging to accent paradigm b (on these accentual paradigms see Stang 1957). There are a few innovations in the individual Slavic languages that need to be kept in mind. In Czech, Upper Sorbian and Slovene, short rising vowels were lengthened in non-final syllables (with different additional conditions in the individual languages, see Greenberg 2000: 128ff., Kortlandt 2011: 341f., Derksen 2008), while in Ukrainian *o was *o lengthened if jer (b or b) was lost in the following syllable and *e if a front jer (b) was lost in the following syllable (Shevelov 1979: 302ff., 318ff.).
In forms in which *e, *o is consistently long or a diphthong in the relevant Slavic languages, the accent was originally on a word-final jer and was retracted onto the preceding syllable (development a) above). The vowel that received the stress was apparently lengthened. This retraction is found exclusively in paradigms with mobile accentuation and can be dated before the other developments mentioned above (cf. Kortlandt 2011: 170, 338f.). Evidence for a long vowel comes from:
-	the gen.pl. of nouns with mobile accentuation (i.e. belonging to accent paradigm c): Cakavian (Vrgada) kos 'hair', nebes 'heaven', Slovene nog 'leg', Czech synu 'son'. The long vowel in the gen.pl. of mobile nouns is clearly of Proto-Slavic origin and it spread to non-mobile nouns in South Slavic (S, Cr., Sln. krav 'cow') and Slovak (os 'wasp', sil 'power', kopyt 'hoof') and in Polish in the a-stems (os 'wasp', mqk 'suffering'), but not in the neuter o-stems (pqt 'fetter', blot 'marsh', Kortlandt 2011: 54), and not at all not in Czech, where a long vowel is found in some mobile nouns in the older language and today only in relic forms (Verweij 1994: 507f.).
-	the nom.sg.m. of the possessive pronouns *mojb, *tvojb, *svojb: Ru. dial. moj, Cz. muj, Slk. moj, Sln. moj, Kajk. (Bednja) myej, Cak. (Brae) mo j (van Wijk
— 7 —
Tijmen Pronk
1916: 323ff.). In Slovak tvoj, svoj, the original o became o after the initial cluster eding in -v-, cf. Slk. dial. svoj, tvoj, Cz. svuj, tvuj (Pauliny 1990: 176). The paradigm of these pronouns has end-stress in Slavic. The nom.sg. form must have had end-stress at an earlier stage as well. If the stress had been on the root, it should have had a falling tone (*mojb etc.), and this tone should have been preserved in most of Slovene, Kajkavian and Čakavian. These forms are not due to early Slavic lengthening before *-jb, as van Wijk (ibidem: 325) suggested (cf. Nonnenmacher-Pribic 1961: 94).
-	the adverb *domovb 'home', an old dat.sg. to the M-stem *domb (Ru. dial. domoj, OCz. domov, Cz. domu, Sln. domov (secondary domov)). The short reflex in Slk. domov is due to the following *u (Verweij 1994: 515). Ukr. dial. domiv preserves the final accentuation but provides no information about the original length of the *o. The stress must have been on the final jer in early Proto-Slavic, because *domb belongs to the accentual type with stress alternating between the initial and final syllables. The attested stress on the medial syllable of *domovb can therefore only be due to a retraction of the stress from the final jer.
-	the imperatives *(ne) bojb sq '(do not) fear!', *stojb 'stand' (Čak., Kajk. stoj, boj se, Sln. stoj, boj se, Cz. stuj, dial. (Silezian) buj, Slk. boj, but stoj with an analogical short vowel. The present of these verbs has mobile accentuation. If the stress had originally been on the root in the 2sg. imperative, it should have had a falling tone (*bojb, *stojb), and this tone should have been preserved in most of Slovene, Kajkavian and Čakavian.
-	present forms of mobile verbs of the type Čakavian (Senj) pije, peče. The long rising accent originates in the 2sg. *nesešb < *nesešb and 3sg. *nese(tb) < *nesetb (on the latter ending see Kortlandt 2009: 157ff.). The long rising vowel resulted from the retraction of the accent from a final jer and can be compared to the long vowel in the gen.pl. The long vowel is found in some Čakavian dialects, e.g., apart from the Senj dialect mentioned above, Susak rest'e, klade, in various Štokavian dialects (cf. Kapovic 2009), e.g. Posavian peče, teče, Neoštok. dial. pečeš, in Prekmurje Slovene (Cankova) nase, zabe,4 and in Central Slovak, nesie,pečie. Kajkavian, most of Slovene and Čakavian and part of Štokavian generalized a short suffix vowel, as did Czech and part of Slovak (cf. Vermeer 1984: 380ff.). In much of Štokavian and Čakavian, we find a redistribution of vowel length depending on the length of the root vowel: the suffix is long when the root vowel is short and the suffix is short
4 Suffix-stress is also preserved in north-west Slovene: Gailtal nase, race, Resia nase, bode. The timbre of the vowel is surprising, because long *e otherwise became Gailtal is, Resia i e.g. liad, lit < *ledi>, siast < *sestb. Perhaps the vowel was originally short (like in the neighbouring Slovene dialects) and was lengthened analogically at an early stage, i.e. before the loss of post-tonic length in verbs like 3sg. *bryse, *bqde, *kupuje > Gailtal brise, kapuje, bode) but after the raising of *e. Alternatively, the timbre of the vowel may have been restored after the raising of *e to *e in analogy to verbs with a long root-vowel and a short stressed suffix, e.g. *raste, *trqse > Gailtal raste, trese, Resia raste ■
— 8 —
tS^la-Via Centra lis 1/2016
	 Early Slavic short and long o and e 	
when the root vowel is long (ibidem: 365f., the same distribution is found in north-western Slovene).
In all other cases, neoacute *o and *e are reflected as short in some languages, but as long or diphthongal in others. Štokavian and Čakavian always have a short reflex in these categories (disregarding later lengthenings in specific environments). Russian dialects with a distinction between monophthongal and diphthongal reflexes of *o always have a diphthongal reflex of neoacute *o. Long/diphthongal reflexes are relatively frequent in monosyllabic forms of nominal paradigms belonging to accent paradigm b and in those cases in which *e or *o received the accent as a result of Stang's law or the retraction of the accent from a weak internal jer:
-	the nom.sg. of masculine (j)o-stems belonging to Proto-Slavic accent paradigm b: *košb, *konjb, *potb, *bobb etc. The root vowel is short in South Slavic, but long in Russian dialects and predominantly long in West Slavic. The diphthongal reflex in Ru. dial. stol, kon, snop etc. is in line with fact that any neoacute *o becomes Ru. dial. o. I agree with Kortlandt (2011: 345f.) that the reflex of a long vowel in, e.g., Czech kun, stul, nuž, Slovak kon, stol, nož is likely to be due to analogy with other case-forms with stem-stress, e.g. loc.sg. *koni, gen.pl. *kon, loc.pl. *konix, inst.pl. *koni. In South Slavic, except Kajkavian, only the gen.pl. had a long vowel (see below). Accordingly, there was no motive for introducing a long vowel in the nom.sg. form in South Slavic, as there was in Czech and Slovak.
-	the definite form of adjectives belonging to accent paradigm b: *novbjb, *golbjb, *dobr-bjb. The accent was originally on the final stem-syllable (*ndvbjb), after which post-tonic *-bjb was contracted to *-yj, the accent shifted to the following syllable according to Dybo's law (> *novyj), from which it was retracted again as a result of Stang's law (> *ndvyj/*novyj).
-	i-presents belonging to accent paradigm b: *nosišb, *volišb, *modlišb, *xodišb. Here, too, the neoacute results from Stang's law. We find a diphthong in Russian dialects, a long vowel in Kajkavian, but a short vowel in Slk. nosiš, Pl. nosisz because of the (originally) long vowel in the following syllable.
-	je-presents belonging to accent paradigm b: *stelešb, *glod'ešb. The thematic present *možešb also belongs here (cf. Vermeer 1984: 363). It has a diphthongal reflex in Russian dialects and Slovak, a long reflex in Kajkavian and an analogical short reflex in Pl. može.
-	individual case forms of nouns belonging to accent paradigm b, e.g. loc.sg. *vdle, *kdni, loc.pl. *vdlex, *kdnix, inst.pl. *vdly, *kdni.
-	the voja-type, which is much discussed in the literature, cf. Stang 1957: 57ff., Kapovic 2007, Fecht 2010, Kortlandt 2015. The neoacute in this category is the result of Stang's law, which caused the stress to shift from the ending (*vola) to the preceding syllable. Examples with *o are *vola, *koža, *xod'a, *vona and *noša. We find a reflex of a long vowel in Slovak, East Slavic and Kajkavian and reflexes of a short vowel in Čakavian and Štokavian. The volja-type was frequently influenced by ja-stems with different accentuation and by nouns in -bja, e.g. Slk. koža, noša instead of *koža, *noša,
— 9 —
Tijmen Pronk
USorb. wola, koža instead of *wola, *koža, Ru. dial. volja instead of *volja, Štokavian hdda instead of *hoda, Old Polish wola, wonia (with analogical long *-a < *-bja) instead of *wola, *wonia, S, Cr. hdda, tdnja instead of *hoda, *tonja etc. Besides ja-stems, a neoacute is sometimes also found in other a-stems, such as *kora 'bast' (in analogy to *koža 'skin') and the loanwords *škoda 'damage, pity' and *sgbota 'Saturday'.
-	words in which the accent was retracted from a medial weak jer onto a preceding vowel: *kdnjbskbjb, *ženbskbjb, *selbskbjb, *otrdčbjb, *pdsblješb, *ddjbdešb, *glezbn-.5 It is often difficult to determine the original length of the vowel in these cases. The short stem vowel of Cz. konsky, žensky, zeli, Slk. konsky, žensky, dial. zele can be connected to the following syllable which contains a long vowel, cf. the reflex of a long stem vowel in USorb. konski, Slnc. kounshi.
In all other cases short reflexes are dominant:
-	l-participle of mobile athematic verbs: *neslb, *peklb, *bddlb>. Short in South Slavic, long in Slovak (niesol), short in modern Czech (nesl), but cf. Old Czech vedl. The accent was originally on the final syllable, from which it was retracted causing a diphthongal reflex in Slovak, but not in Czech or South Slavic. Kortlandt (2014: 130) considers the short reflexes in South Slavic to be due to analogy.
-	the l-participle *mdglb (accent paradigm b). The accent was on the first syllable before the application of Dybo's law and shifted to the final jer in South Slavic (to be retracted again later), but remained on the first syllable in West Slavic in view of Slovak mohol. The latter was apparently accentually distinct from niesol, which has a diphthong due to a retraction of the accent from the final syllable. The reflexes of neoacute *o in *mdglb are consistently short, except in OCz. mohl, Slovak dial. mohol (Ukr. mih and Sln. mqgel are due to regular lengthenings in these languages). The long reflex in Old Czech and Slovak was probably introduced analogically (Kortlandt 2011: 345).
5 The type *zelbje, *perbje, *grobbje, *kozbjb is slightly more complicated because one expects final unstressed *-bje to have been contracted before Dybo's law (*zelbje > *zele etc., cf. Kortlandt 2011: 38). This should have become *zele as a result of Dybo's law and then *zele as a result of Stang's law. Apparently, the sequence *-b(j)e was restored on the basis of caseforms in which no contraction had taken place (e.g. the gen.sg. *-b(j)a, cf. the preserved cluster of Stok., Cak. braca 'brothers' < *bratbja) and on the basis of nouns in which the suffix was stressed *-i>(j)e (>*-bje after Dybo's law, reflected in, e.g., Cak. (Novi) voci, kameni, zeli). Later contraction of *-b(j)e produced the long final vowel of Vrgadaperje, zele, grozje, Noviperji, zelenji, kozji, Grobnikperji, veseji, grobji etc. The restoration of posttonic *-b(j)e in *perbje, *grobbje etc. probably took place before Dybo's law (which then shifted the accent to the jer: *zelb(j)e). If this is correct, the neoacute on the initial syllable of *zelbje in western South Slavic is due to the retraction of the accent from an internal jer (Kortlandt 2014: 129).
— 10 —
tS^la-Via Centra lis 1/2016
	 Early Slavic short and long o and e 	
Similar examples are found in the nom.sg.m. of nouns and adjectives belonging to accent paradigm b with a stem ending in a resonant: nom.sg. *odrb, *bobrh/*bebrh, *koprb, *kosmb, nom.sg.m. *dobrt, *mokrb, *ostr\>. *teplb/*toplb. These have a short vowel in South Slavic, but Old Czech has dobr. Note that Slovak has a long reflex in kopor and vodor, which may indicate that this is the regular reflex of *o in this type of words (cf. Slk. dial. mohol). However, in view of the secondary -o- in masculine o-stems like kon and stol, it is conceivable that we are dealing with an analogical -o- here as well. Masculine o-stems of the type *odrb etc. were sometimes subject to generalization of final stress from the oblique cases after the introduction of a secondary jer in the second syllable, e.g. Neostok. ddar. This generalization was fuelled by analogy with nouns with a disyllabic stem of which the second syllable contained a jer, e.g. *orblb, *ovbm>, *ovbSb, *osblb (Neostok. drao, dvan, Ru. os'ol, ov'os). The noun *ognb (m. i)/ *ognb (m. jo) 'fire' has final stress in the nom.sg. in almost all languages (Ru. ogon\ Sln. ogenj, Neostok. dganj).6 This final stress points to a relatively early introduction of a prop-vowel between *g and *n (compared to the type *odrb, *koprb). This may be because *ognb/*ognb shares its initial *o- with *orblb, *ovbm>, *ovbSb, *osblb or because the word-final cluster *-gn that arose after the loss of the final jer was not tolerated.
The cardinals *sedmb and *osmb also belong here. They have a diphthongal reflex in Ru. dial. vosem', but a short reflex in Cz. osm, Slk. osem, Pl. osiem, Slnc. vuesem, Stok., Cak. osam, Kajk. osem (Sln. qssm is ambiguous, USorb. has wosom next to wosom). In contrast, the ordinals *sedmy'b, *osmy'b are predominantly attested with a long vowel or diphthong, cf. Cz. dial. usmy, Slk. osmy, Pl. osmy, Slnc. vousmi, Polab. vasme, Cak. osmi, Kajk. osmi, Sln. dial. (Prekmurje) 'ousmi, although USorb. wosmy has the reflex of a short vowel. The diphthong in the East Slavic cardinal is the regular reflex of neoacute *o (cf. Vaillant 1950: 187). In both the cardinal and the ordinal the accent was on the initial syllable before Dybo's law (*dsmb, *dsmy'b), where it stayed in *dsmb in Slovak osem but shifted to the final jer and was retracted again in (Western) South Slavic (> *osmb > *dsmb, cf. Pronk 2013: 115f.). In the ordinal *dsmbjb, the initial stress is due to a combination of Dybo's and Stang's laws, see above on *novbjb etc. The Russian definite form vos'moj exists since Old Russian (Zaliznjak 2011: 168) next to a form with root stress, os'myj, Ru. dial. and Ukr. vos'myj. Similar doublets exist for *sedmy'b (ibidem: 167), cf. sedmy but sedmaja in the 14th century Cudovskij Novyj Zavet. The forms with suffix stress are probably secondary in analogy to sestyj 'sixth' (rarely sestyj, which is probably analogical to osmyj). South Slavic dialects that preserve traces of end stress in the definite form of mobile adjectives, such as the Cakavian dialect of Novi and the Stokavian dialects of the Posavina, preserve the old root stress in the ordinals 'sixth', 'seventh' and 'eighth': Novi sedmi, osmi (Belie
6 With the exception of some Cakavian dialects (e.g. Brae ôgon), in which we also frequently find unexpected initial stress in forms with a medial jer (e.g. Brae oca < *otbca, pismo < *pisbmo, oral < *orbh>, examples from Simunovic 2009).
— 11 —
Tijmen Pronk
2000: 143), Posavina šesti, sedmi, osmi, but prvi 'first' (Ivšic 1913: 56f.). The long root vowel in Čakavian and Štokavian is analogical after peti 'fifth', deveti 'ninth' etc., and the same is true for the long vowel of OCz. and Cz. dial. šesty, sedmy, Polish siodmy, osmy (Bulaxovskij 1983: 109, Kapovic 2007: 92, fn. 5).
-	the second syllable of nouns with a polysyllabic stem like *potok-, *narod-, *bbrlog-, *vinogord-, *e/olen-, *plemen- etc. It received the accent from the first syllable as a result of Dybo's law.
-	the final syllable of the nom.acc.sg. of neuter nouns and adjectives belonging to a.p. b: *dbno, *pisbmo, *lice, *novo. Here the vowel is generally short, but diphthongal in Ru. dial. dno, pis 'mo, šyroko. The long vowel of Sln. dial. dnQ, rojstvQ is due to more recent lengthening of *-d, probably originally only when followed by a clitic (Ramovš 1921: 235f.). There is also rojstvo, the timbre of which may be analogical to that of other neuter o-stems where it is regular, e.g. meso, oko, leto. The long ending is, however, reminiscent of that of Čak. (e.g. Novi, Vrgada) -stvo, -stvo (cf. Belic 2000: 142, 159): društvo, trojstvo, where the analogy proposed for Slovene would not work. Belic (ibidem: 142) explained the Čakavian long vowel from a merger of the suffixes *-bstvo and *-bstvbje (similarly already Škrabec for Slovene, see Valjavec 1897: 211). The latter would regularly have produced long endings, cf. Čak. (Novi) kameni, (Vrgada) korene, Sln. dial. (Nadiški) korania, (Prekmurje) zelenje < *-bje (Šekli 2013: 142f.). The neocircumflex of Sln. otrgštvo, žqnstvo < *otročbstvo, *ženbstvo is probably not due to the internal weak jer (cf. otroški, ženski < *otročbskbjb, *ženbskbjb, see Pronk 2013) and therefore also points to a long ending.
There is one category in which we find long *e, *o next to short *e, *o which is not connected with the neoacute:
-	the nom.sg. of masculine (j)o-stems with a monosyllabic stem belonging to Proto-Slavic accent paradigm c: *bogy, *most\>, *gnojb, *medb, *ledb etc. The root vowel is short in West Slavic, but long in western South Slavic, where vowels with a falling tone were regularly lengthened in monosyllabic words (not counting final jers).
As the first set of examples above shows, all languages have a long reflex of *o and *e when these vowels became stressed as a result of the accent-retraction from a final weak jer (except when preceded by an occlusive plus resonant). In Štokavian and Čakavian, the reflex of neoacute *o and *e is short in all other cases. We will now briefly discuss the situation in East Slavic, West Slavic and Kajkavian before treating the Slovene data in more detail.
2 East Slavic
In Russian dialects we find a reflex of neoacute *o that differs from that of *o with an originally falling accent, e.g. from Stang's law (gonit, lovit, xoces,
— 12 —
tS^la-Via Centra lis 1/2016
	 Early Slavic short and long o and e 	
poret, tonet, dobryj, golyj), from Dybo's law (potok, gotov, sbor, urok, mnogo, moloko, syroko), in masculine (j)o-stems belonging to a.p. b (stol, kdn, noz. kdl, topor, syrok), with retraction from a final jer (moj, domoj, synof gor), with retraction from an internal jer (nozka, dvornik, pomnit') and before *-CRb/b (mog, vosem').7 Forms with polnoglasie also have a stressed diphthong, e.g., with acute korova, borod, kolot', with neoacute kolotit, moldcnyj, golovka. Originally stressed *o in a penultimate syllable before a weak jer is reflected as o: gen.pl. por, sov (sová and porá belong to accent paradigm b).
In Ukrainian, neoacute *o and *e remained unchanged, e.g. mózet', nósyt', torzestvó, gréblja 'dam', unless they were followed by a weak jer. Every *o before a syllable in which a jer was lost, regardless of its original intonation, was lengthened and diphthongized and eventually became i (word-initial vi-) in the standard language (Shevelov 1979: 318f ). Every *e before a syllable in which a front jer was lost also became i. Only the reflex of *e before a syllable in which a back jer was lost depended on the original accentuation of the *e. According to Shevelov, the reflex of *e before a lost back jer is i "under a retracted stress" (o.c., 323), e.g., nis 'carried', gen.pl. kolís, sil, otherwise it is e. The original distribution is obscured by the changes *e > *o (after a palatal consonant) > i (if a jer was lost in the following syllable), interdialectal borrowing and analogies (Bulaxovs'kyj 1943, Shevelov 1979: 327). It seems plausible that the -i- of nis and kolís was originally a long *e that had been lengthened as a result of the retraction of the stress from the final jer (*nesl~b > *néslb, cf. Slk. niesol) and merged with the reflex of *e in Ukrainian.
A special case are instances in which the final syllable of a word contains stressed -ó- as a result of polnoglasie, e.g., with acute polón, moróz, horóx, gen. pl. vorón, kolód, but analogicalporíh 'threshold', with neoacute ohoród, xvoróst, storóza, horóza, voróta, molótys. The regular reflex is -ó-, not -i- in spite of the lost jer at the end of the word. We do, however, often find -i- in the gen.pl. of mobile nouns (holív, vorít, boríd, storín, cf. also ceríd, but derév), derivatives with the suffix -ka (holívka (but holóvka of cabbage, cheese, onions etc.), borídka, but storózka) and in the masculine singular form of the l-ptc. (volík, f. voloklá, beríh, f. berehlá). It has been argued that the opposition between Ukr. holív and moróz reflects an earlier tonal opposition between acute and neoacute accents (Bulaxovskij 1958: 87f., Zovtobrjux et al. 1980: 94f., Kortlandt 2011: 172), but this idea cannot be maintained, as is shown by isolated examples with -oró- from a neoacute like ohoród, xvoróst, storóza, horóza and voróta (Seliscev 1951: 170, Zaliznjak 1985: 161-163), and by the absence of forms with -oRi- or -eRi- in texts before the late 17th century (Shevelov 1979: 109, 607). The forms with -i- all contain analogical -i-, like in koróva, korív, korívka, beréza, beríz, berízka, bloxá, blix, blíska, voróta, vorít, vorítka. Forms like nis, kolís and sil mentioned above formed the model for the introduction of -i- for -e- in beríh, beríz, ceríd etc.
7 Data given here are from sources quoted by van Wijk (1916) and Vasil'ev (1929).
— 13 —
Tijmen Pronk
3 West Slavic
Slovak has a diphthongal reflex of neoacute *o (in a few examples also of neo-acute *e), except before a syllable containing a long vowel, e.g. in /'-presents of the type nosim and adjectives of the type konsky. A short reflex of the neoacute appears to the regular outcome in a number of disyllabic forms ending in a jer which had initial stress before Dybo's law, e.g. the numerals sedem, osem and the /-participle mohl (but dial. mdhol).
It has traditionally been assumed that Czech had the same reflexes as Slovak (cf. Nonnenmacher-Pribic 1961: 95), although the situation in Czech is difficult to assess. Czech lengthened acute and neoacute vowels in disyllabic words unless the following syllable contained a long vowel, which largely obscured the original situation, cf. the ambiguous muže, pujdeš, nosiš, dobry, zeli. The Czech loc. and ins.pl. of (j)o-stems belonging to a.p. b have a short reflex of a neoacute *o (e.g. stolech, stoly), but this can easily be due to analogy with other forms of the paradigm. The ins.pl. kožem/ (secondarily also kužem/) to kuže might point to an originally short reflex, but the *o is probably analogical to gen.pl. koži, dat.pl. kožim, loc.pl. kožich, all with a regular short reflex because of the long vowel in the final syllable (cf. Bulaxovskij 1983: 109). The short root vowel of je-presents like meleš, tešeš, češeš, steleš, pošleš is due to analogy, cf. the long reflex in kuleš (but OCz. koleš), OCz. uoreš (ibidem). The only evidence for an originally long or diphthongal reflex in Czech is indirect and comes from the nom.acc.sg. of the masculine (j)o-stems. The o-stem paradigm with length in the nom.sg. (stul) but a short vowel in the rest of the paradigm (stola) must be relatively old in view of the fact that Slovak shows the same pattern: stol, stola. In Czech, this pattern is found with a number of nouns with a root containing *-o- originally belonging to a.p. b, viz. kun, dvur, pust, stul, kul, vul, nuž, but also with nouns belonging to a.p. c: buh, vuz, dul, dum, luj, hnuj (Verweij 1994: 525ff.). In Slovak, (j)o-stems with -o- throughout the paradigm or in the nom.sg. only predominantly belong to a.p. b (ibidem: 530ff.). It can be observed that nom.sg. forms with a short root vowel replaced forms with a long root vowel within Czech (Nonnenmacher-Pribic 1961: 94). Van Wijk regarded the long reflex as the regular outcome of neoacute *o in kun etc. Kortlandt (2014: 128) rejects Van Wijk's view because of the large number of counterexamples and suggests that the long reflex was introduced from case forms in which the accent was retracted onto the root as a result of Stang's law, i.e. loc.sg. *konji, loc.pl. *konj/x, ins.pl. *konji. If this is correct, these cases would originally have had a long root vowel (or, in Kortlandt's formulation, a diphthong). Nonnenmacher-Pribic (1961: 95), following Kul'bakin, assumed that the long reflex in the nom.sg. was due to the retraction of the accent from a final jer. This is problematic because, as Kortlandt has shown (2011: 17ff.), the consistently short root vowel of S, Cr., Sln. konj and other masculine (j)o-stems belonging to accent paradigm b indicates that the final jer of *konjb etc. was never accented. Although this is evidently correct for South Slavic, the West Slavic data are less transparent. Nevertheless, the contrast between
— 14 —
tS^la-Via Centra lis 1/2016
	 Early Slavic short and long o and e 	
the short root vowel of Slk. mohol, osem < *mdgl~b, *dsmb (accent paradigm b) and niesol < *nesly (accent paradigm c) appears to indicate that Dybo's law had not shifted the accent to the final jer of *mdglb and *osmb.8 I therefore agree with Kortlandt that analogical origin for nominatives of the type stul, stol is most likely.
A factor that may have had some influence on the introduction of a long vowel in the nom.sg. of masculine (j)o-stems is the existence of a small group of nouns which originally (i.e. before Dybo's law) had final stress in the nom. sg. form. In late Proto-Slavic, the accent was retracted onto the stem syllable, which was lengthened as a result. An example of such a word is PSl. *dbzdjb 'rain' (Derksen 2008: 15f., cf. the long vowel of OCz. dešč, Slk. dažd', Stok. (Vuk) dažd, Čak. (Orbanici) daš, (Lumbarda) daž)9 Other examples are speculative, e.g. *nožb 'knife' (long in Neoštokavian nož, Old Stokavian, Čakavian nož, Cz. nuž, Slk. nož, short in Sln. nož, Kajkavian nož (e.g. Bednja než, but with a long vowel in Fertohomok nož), *ežb 'hedgehog' (long in Neoštokavian jež, ježa, Sln. jqž (cf. dial. (Gailtal) tež), perhaps also USorb. jež (but cf. Derksen 2008: 125f.), short in Cz. (dial.), Slk. jež) and *dvorb 'court, yard' (long in Neoštokavian dvor, gen.sg. dvora, Old Stokavian, Čakavian, Kajkavian dvor, Sln. dial. (Gailtal) duar, Cz. dvur, short in Sln. dvor; Slk. dvor is ambiguous).10
8	The situation is different in western South Slavic, where both *moglb and *nesh> are reflected with a short root-vowel (e.g., Kajkavian mogel, nesel), pointing to earlier *mogli>, *nesli>, both with end-stress. If the accent had been on the root the Kajkavian and Slovene forms should have had a neocircumflex. Apparently, in western South Slavic Dybo's law did shift the accent onto final *-CRb, to be retracted again later. This is confirmed by the consistently short root-vowel of other words with this structure, e.g., Kajkavian osem, dober (Pronk 2013, although Kortlandt (2014: 130) regards these forms as secondary). In Czech, words with the structure Ce/oCRb/b consistenly have a short root vowel: nesl, pekl, sedm, osm, mohl, dobr, bobr, odr, kopr. In Slovak, some of these have a diphthong: niesol, piekol, vodor, kopor, dial. mohol. Czech nesl, sedm etc. did not undergo the Czech lengthening of short vowels with a rising accent in penultimate syllables of disyllabic words (krava < *krava) because the forms were apparently monosyllabic at the time. An exception is OCz. mohl, which may have a secondary long reflex, like Slk. dial. mohol. The reflex of a long vowel in USorb. wosom is due to the regular Upper Sorbian lengthening of a short rising vowel.
9	Many Cakavian dialects have an analogical short vowel in this word. The falling tone of Brae doz, Hvar (Brusje) dorz shows that its long vowel is due to relatively recent secondary lengthening before a resonant as in lon 'linen cloth' < *lhnb. The lengthening probably took place in the unattested nom.acc.sg. *dar, which developed from *daz < *dbzdjb with rotacism as in more- < *moze-. The nominative was later adapted to the oblique stem *dazj-.
10	The remarkable long vowel of *dvorb was pointed out to me by Willem Vermeer. It may indicate that this word had end-stress in Slavic before Dybo's law, contrasting with the paradigm of Lith. dvaras, which reflects earlier initial stress, and Sanskrit dvaram, also with initial stress. The shift from a Proto-Indo-European neuter to a Balto-Slavic masculine o-stem also points to original initial stress (Illie-Svitye's law). It is difficult to derive the apparent end-stress of *dvori> from a w-stem paradigm. Such a paradigm would probably have to be due to analogy with a Proto-Slavic nom.sg. *domi>, but the
— 15 —
Tijmen Pronk
The word for 'hedgehog' has joined a.p. c in Cakavian and Kajkavian (jez, jezc). In *nozb and *ezb the long vowel may also be due to lengthening before -z, as prof. Kortlandt suggested to me. In either case, the words for 'rain' and 'knife' were inherited into Czech and Slovak with a long vowel in the nom. acc.sg. and a short vowel in the other forms. There may have been a few more nouns of this type (perhaps *odrb 'bed, couch' and *koprb 'dill', cf. Slk. vodor 'threshing floor', kopor).
4 Kajkavian
In his monumental Jezik Hrvata Kajkavaca (1936), Stjepan Ivsic gives the following overview of categories in which Kajkavian has a long rising accent on an etymologically short vowel:
1.	zelje, groblje, stolnjak
2.	nom.pl.n. sela, rebra, reseta, okna, jajca (cf. sg. jajce)
3.	gen.pl. lonec, konec
4.	loc. and ins.pl.m. and n. koni(h), voli(h), lonci(h), koli(h)
5.	the definite form of certain adj., dobri, novi, siroki, zeleni
6.	the ordinals tretji, sedmi, osmi
7.	adjectives like zenski, konski, peklenski, bozji1
There are also a few categories in which neoacute *o and *e are not reflected with a long rising accent:
1.	the volja-type: vdlja, kdza, stelja, with a circumflex. For the somewhat abber-ant Kajkavian dialect of Hidegség and Fertohomok in Hungary, Houtzagers (1999) provides forms with a short vowel: voja, koza.
2.	presents of i-verbs of the type nosim, hodim. The short vowel must be analogical.
evidence for such a form is limited to Cz. dum, while South Slavic dom and Ru. dial. dom (not *dom) point to Proto-Slavic *domi>. In theory, we could be dealing with traces of an old difference between nom.sg. *domi> and acc.sg. *domi>, corresponding to the mobile pattern refected in Lith. nom.sg. medus, acc.sg. medq, but it seems that the initial accentuation of the accusative was generalized in the u- and /-stems already in Proto-Slavic (cf. Cr. med, gost, pet, Cz. med, host, pet). The u-stem gen.sg. *dvoru in West Slavic (Cz. dvoru, Pl. dworu) is a recent innovation. There is no trace of a u-stem in OCS or East Slavic and Proto-Slavic derivatives like *dvori>m> and *dvorišče are derived from the older o-stem.
11 Houtzagers' description of the dialect of Hidegseg and Fertohomok (1999) gives short accents: žensko, konjsko, selski. I found hardly any other exceptions to Ivsic's categories with long o, e in descriptions of Kajkavian dialects. There are a few lexemes with unexpected reflexes which may be due to analogy or local phonetic changes, e.g. Mursko Središče st'oljak and k'ojnskj, where one would expect ou < *o, cf. gr'oubje, k'oula (data from Blažek & Grozdana Rob 2014).
— 16 —
tS^la-Via Centra lis 1/2016
- Early Slavic short and long o and e -
3.	internal and final *o and *e that received the stress as a result of Dybo's law: potoka, selo.
4.	originally disyllabic words ending in *-Cb/b or *-CRbA>: konj, vol, osem, nesel, dober.
5.	early borrowings with short stressed *o: sobota, skoda.
It follows that in Kajkavian, neoacute *o and *e are reflected as a long vowel if the neoacute resulted from Stang's law or the retraction of the accent from a final or internal jer. The unexpected short vowel in the present of /-verbs (nosim, hodim) must be analogical to regular short *o in forms without initial stress like 1sg. *nosu, 1pl. nosimo etc. It has traditionally been assumed that the falling accent of the voja-type in Kajkavian is a neocircumflex. If this is correct, the neocircumflex is probably due to a long vowel in the ending (*-ja etc.) which was restored on the basis of other 'a-stems. The orginal short ending was preserved in the dialect of Hidegseg and Fertohomok. Note the sharp contrast between the Kajkavian falling accent on the root and the accentual mobility found in neighbouring Slovene (Pannonian in the North and perhaps the Kozjansko-Bizeljsko dialects of Styrian in the West, see below), both being replacements of the original paradigm with a neoacute on the root.
5 Slovene
We will now proceed to discuss the Slovene evidence for the development of neoacute *o and *e. It is often not immediately clear whether we are dealing with a reflex of an original long or short vowel in Slovene because of the lengthening of short stressed vowels in non-final syllables ("¿rato-lengthening", on which see Rigler 2001: 302ff., Zorko 1998: 189ff., Greenberg 2000: 128ff.). The main evidence comes from:
-	eastern dialects that did not undergo ¿rata-lengthening;
-	dialects in which the timbre of the vowel that underwent ¿rato-lengthening is distinct from the timbre of older long *o and *e, especially the northern ones in which ¿rata-lengthening is a late phenomenon.
Ramovš (1921) discussed the development of neoacute *o in Slovene in detail and concluded that the reflex of neoacute *o was long in a) absolute initial position, b) closed syllables and c) absolute final position. A short reflex would be regular in other open syllables. Rigler (2001: 67) rightly pointed out the weaknesses of Ramovš's view. He argued on the basis of more dialectal data that, instead, the predominantly long reflex of Kajkavian was also the primary reflex in Slovene, at least "v pretežni večini" (ibidem: 418). Elsewhere, he more carefully stated that "so namreč tudi v slovenščini nastopila daljšanja v posameznih kategorijah, ki se precej ujemajo s kajkavskimi podaljšavami" (ibidem: 399). Rigler (ibidem: 25, 69ff., 313) distinguishes between early lengthening of neoacute *o and *e, which produced a long vowel that merged with existing
— 17 —
Tijmen Pronk
long *o and *e in individual words, often different in different dialects, and later lengthening which affected all remaining cases. The former category includes those cases in which other Slavic languages also have a long reflex, such the genitive plural, but also some other examples collected by Ramovš and Rigler himself. Rigler discussed the relevant historical Lower Carniolan (dolenski) data in his important article Pregled osnovnih razvojnih etap v slovenskem voka-lizmu and the Carinthian (koroški) data in his later Pripombe ("remarks") to his Pregled. The main objection against Rigler's scenario is that his occasional early lengthenings do not have the character of a sound law (with the exception of the pan-Slavic long vowel in words in which the accent was originally on a final jer). This suggests that they are at least partly due to analogy, and for most long reflexes adequate models for the secondary introduction of a long vowel can be indeed adduced, as will be shown below.
An important source for our purposes is the standard language as represented in Pleteršnik's dictionary and works based on it, e.g. Breznik's Slovenska slovnica za srednje šole. Here we find -q- and -q- for neoacute *o and *e in all categories (kQnjih, vQlja, nevQlja, mQrem, pQljem, nQsim, kQzji, kQnjski, potQka, mqljem, zqlje, žqnski, sqlski, debqla, jelqna) except in final syllables (bob; kdnj, dno, gumno, jajce) and those forms in which *o received the stress from a final jer (domov, moj, gor, but nebqs). The timbre of -Q-, the reflex of neoacute *o in non-final syllables, contrasts with the reflex of long *o that arose through lengthening of monosyllabic *o with a falling tone, e.g. bog, kost, and with that that arose as a result of the forward shift of the falling accent onto a following syllable, e.g. meso, sirota}2 The timbre of *o with a neocircumflex, however, is identical to that of neoacute *o: kQnju, kQnjih, kQnji, dobrgta, otrgštvo, gotovim (cf. inf. gotgviti). The reflex of *e is identical in all categories: lqd, poljq, gen. sg. imqna, plemqnski. It contrasts with e, which is the regular reflex of Proto-Slavic *e (jat').
In some Slovene dialects, inherited long *o merged with neoacute *o, e.g. in Upper Carniola into o, Črni Vrh ua (Tominec 1964: 13f.), Banjšice uo (Logar 1996: 15, 306f.). However, most dialects retain a distinction, e.g. Lower Carniolan, Inner Carniolan *o > u (Pleteršnik's o), but neoacute *o > uo in non-final syllables (Pleteršnik's g), Ter, Nadiža *o > uo, but neoacute *o > o in non-final syllables (cf. Logar 1996: 12), Tolmin *o > uo, but neoacute *o > o in non-final syllables, Gailtal *o > ua, but neoacute *o > o in non-final syllables (koža, patoka, olje) etc.
Forms in which *o received the stress from a final jer have the same reflex as circumflex long *o in all dialects, e.g. Lower Carniolan muj, nux, Gailtal muaj, nuag. This is not surpiring, because such examples have a long vowel in the rest of Slavic as well. In Lower Carniolan, we also find the reflex of old *o in loc.pl. kujnah, ins.pl. kujna, which is clearly analogical to the gen.pl. kujn (Škrabec 1917: 225, Ramovš 1921: 229f.). The same analogy took place elsewhere, e.g. Upper Savinja (Spodnje Kraše, Weiss 2001) gen.pl. ut'rok, loc.
12 Note that in the present day standard language g/o and g/e are not distinguished.
- 18 --Slavia Centralis 1/2016
	 Early Slavic short and long o and e 	
pl. ut'rocsh, Podjuna (Ojstrica, Zorko 1991) gen.pl. ot'rusk, ins.pl. ot'ruscmi, Gailtal (Potoce, Pronk 2009: 58) gen.pl. trusk, loc.pl. truscsh, ins.pl. truscmi 'child'. Another category in which Lower Carniolan sometimes has -u- for neoacute *o is the nom.pl. of some neuter nouns, e.g. ükna, küla. Here too, influence from the gen.pl. *oksn, *kol seems likely.13
In eastern dialects without brata-lengthening, the reflexes are partly identical to those of Kajkavian. In the 18th and 19th century literary language of Prekmurje long *o > ou (merging with long *ç) and *ë > ë (written é, merging with the reflexes of long jers and *f). Neocircumflex *o > ou: osnouviti, dobrouta, loc.sg. potouki. Neoacute *o and *e merged with the long vowels in some categories, but remain short in others.14
A long vowel is found, as expected, when the neoacute is due to the retraction from a final jer: gen.pl. noug, nebéfz, analogically also 'zén. There are also examples in which we find ou as the reflex of neoacute *o that arose as a result of Stang's law: koula (pl.) 'wagon', kouza, kouzo, kouzov, kouze, nouvi (cf. also ponouviti), zeléno, vefzéli, fzédmi, ousmi. We also find a long reflex in pérje, where the neoacute is due to retraction from an internal jer, but in some younger dialectal data this word has a short reflex: long Porabje pérdje but short Cankova ins.sg. perjem, Martinjep'ierd'e.
The reflexes that appear to point to a long reflex of neoacute *o or *e that arose as a result of Stang's law are suspect of being secondary. The definite adjectival forms nouvi and zeléno probably have a secondary neocircumflex, like in most other Slovene dialects. This also applies to the long vowel of fzédmi, ousmi, which may alternatively have obtained a long vowel (cf. séfzti) in analogy to péti, devéti like in Cakavian, Stokavian, dialectal Czech and Polish. The noun koza has the reflex of a long vowel throughout the paradigm and this long reflex is also found more to the West in Northern Styrian (see the data on two Pohorje dialects below). Because in Northern Styrian neoacute vowels
13	Ramovš also mentions a few other cases with Inner Carniolan -u- for neoacute *o. The long reflex in pujde 'goes' (but Pleteršnik pQjdem) cannot be separated from the long vowel of Čakavian, Old Stokavian dojde, Neoštokavian dode (Ivšic 1911: 146, cf. also Kajkavian dojde), and must therefore be old. The length cannot be due to recent lengthening before -j- because south-eastern Čakavian has an acute (Brač dojde, cf. kroj < *kraj with later lengthening). The forms can be explained by assuming that they received initial stress as a result of Stang's law after contraction of *-o(j)b- to *-oj-: *pobde > *pobde > *pojde > *pojde. The long reflex in Inner Carniolan puljski, gurski < *poljski, * gorski is secondary for older *gorski, *poljski (cf. Pleteršnik gorski, poljski, S, Cr. gorski, poljski). The vocalism is probably analogical to the gen.pl. *gor, *polj, cf. Carinthian (Rož) žignsqi for more archaic (Gailtal) ženski on the basis of the gen.pl. žign. Similarly, Lower Carniolan bužji replaces earlier *božji (cf. Pleteršnik božji, S, Cr. božji). The origin of the long reflex in Gailtal kugzji, Prekmurje kouzgi must perhaps also be sought in the gen.pl. *koz. The long vowel of Sln. kdo (Inner Carniolan gdu, Gailtal tug etc.) is due to an early irregular and probably originally syntactically conditioned lengthening of inherited *ki>td.
14	Data cited here stem from the works of Stevan Kuzmič, Mikloš Kuzmič and Jožef Košič, who consistently distinguish -o- from -ou-, as cited in Novak 2006.
— 19 —
Tijmen Pronk
consistently show reflexes of an earlier short vowel, the length in Prekmurje kouza cannot be taken as evidence for a long or diphthongal reflex of neoacute *o from Stang's law. One possibility is that kouza reflects *koža, like in Ka-jkavian (Rigler 2001: 70). Neuter plurals like koula, oukna may be compared to Lower Carniolan kúla, úkna and have an analogical long reflex from the gen.pl., or they have an analogical neocircumflex after the type sg. mesto, pl. mejsta. Carinthian (Rož) has a short reflex in these cases (ibidem: 68).
A short vowel is consistently found when the neoacute is due to Dybo's law, retraction from an internal jer (except pérje) and in disyllabic forms ending in a jer: dat.sg. potoki, skoda, fzobota, konyszki, senfzka, kozjo (acc.sg.f.),poslem, bob, boj, nefzao, tekao, mogao, oster, topeo, fzedem, ofzem. Like in Kajkavian, the present of the /-stems has a short root vowel: nosi, hodimo, vodi. The present morem, mores, more etc. also has a short root vowel.
A special case is *vola, which shows vowel alternation: nom.sg. vola, acc.sg. volo with a short vowel, but gen.sg. voule, dat.sg. vouli, ins.sg. voulov with a diphthong reflecting length. Similarly nevola, acc.sg. nevolo, gen.sg. nevoule, loc.sg. nevouli, gen.pl. nevoul, dat.pl. nevolám, loc.pl. nevol(j)áj 'misfortune'. In his Historical phonology of the Slovene language, Greenberg (2000: 128, fn. 24) cites a paradigm with mobile accentuation reflecting a generalized pattern from Prekmurje material: nom. 'vola, acc. vo'lou, gen. vo'le:, dat. po 'vouli, ins. z vo'louf. A similar paradigm is given in Mukic's 2005 Porabje dictionary: vóla, gen. židane volé, but dóbre váule, acc. volau, but za mójga/tvójga vólo, dat./loc. po vauli, pr vauli, cf. also nevóla, acc. nevólo, po nevóli. The Prekmurje forms can all be derived from a paradigm with mobile accentuation, forms with a long root vowel are due to the forward shift of the old circumflex from a prefix onto the root: *po voh > *po voli. The mobile paradigm can be compared to that of volja in Old Russian and is secondary. In the case of Prekmurje Slovene it was probably the mobile noun *želja 'wish' that was the source of the mobility, cf. zsela, v zséli in Pavel's grammar (2013).
The Prekmurje data above correspond to the data given in Pavel's normative Prekmurje grammar and in the dialectological literature. An extensive overview of the data is given in the appendix to this article. I will here only provide some additional data not attested or attested with a different reflex in the literary language: Pavel (2013) long gen.pl. lonec, nom.pl.n. domo, plécsa, bédra, péra, short svoj, gen.sg. jelena, 1sg.pres. orjem, posztelem, polem, kolem, Porabje (Mukič 2005) long rébra, domau, short mečen, tesen, Žetinci/ Sichelsdorf (Zorko 1998) long do'ma:u, short 'noša, 'zelje, Polana (Greenberg 1993) long gr'ouzdja, short m'ala, p'ala, s'admi, Martinje (ibidem) short z'isld'x.
In all of northern Slovene, including Pannonian, and in Kajkavian, a neocircumflex was retracted onto a long vowel, e.g. písala < *pisala. In some words this shift also took place onto a syllable containing *o, e.g. ótava < otava and nom.acc.pl. kóléna < *koléna. In those cases the shift was probably analogical (Pronk 2007). The newly stressed *o is reflected as long in Pannonian, e.g. Cankova kopita, kolina, Martinje 'aotxvx.
20	(Centra lis 1/2016
	 Early Slavic short and long o and e 	
The data available to me for other Pannonian Slovene dialects without brata-lengthening, i.e. Prlekija, eastern Haloze and eastern Slovenske Gorice, mainly shows short reflexes of neoacute *o and *e (for an exception see below).15 The categories that have a long vowel in Prekmurje (definite adjectives, neuter plurals, the word koža, some forms of volja) are, however, poorly represented in the literature. The only examples I found are Slovenske Gorice (Koletnik 2001) 'ko:uža and Cerkvenjak (Rajh 2002) 'dobre 'voile with a long vowel and Biserjane (Zorko 2009) pot 'kožoj and 'kola with a short vowel. Interesting examples with a short reflex are Slovenske Gorice 'osmi, 'meče, eastern Haloze (Zorko 1998) 'noša, 'xoja, 'zelje. The dialects more to the west all took part in brata-lengthening. The northern dialects of Slovene nevertheless almost always distinguish neoacute *o and *e from older long *e and *0. The reflexes in western Pannonian and northern Styrian are comparable to those in eastern Pannonian. Long reflexes are found in, e.g., western Haloze 'na:uvi, eastern Pohorje kouža, but in most other cases the reflex of neoacute *o and *e is a short vowel that was later lengthened as a result of brata-lengthening.
In some Styrian dialects, such as the Cental Styrian, Central Savinja and Kozjansko-Bizeljsko dialects, neoacute *o and *e with brata-lengthening merged with older long *0 and *e. The situation here is not identical to that in Kajkavian because long and lengthened *e merged with *e. The Kozjansko-Bizeljsko dialects show some exceptions to the generally long reflex of neoacute *o and *e in non-final syllables: Lesično 'xoija, 'no:ša, 'vo:ja, Pišece 'xo:ja, škoida, g'roijzdje (but 'vu:la), Kapele 'xo:ja, 'no:ša (but 'vu:la). These dialects may originally have had the same mobile paradigm for *xoja, *noša and, in Lesično, *volja that is found in Pannonian.16
The Carinthian dialects carried out brata-lengthening relatively recently (cf. Rigler 2001: 315). Some of the categories that show an analogical long reflex in Lower Carniolan or Pannonian have a short reflex in Carinthian that was later lengthened as a result of brato-lengthening, e.g. the noun koža and the nominative and accusative plural of neuter nouns like okno, selo etc. The Carinthian dialects are further interesting with regard to the question what the primary reflex of neoacute *o and *e was, because all Carinthian dialects with the exception of the Gailtal dialect have a twofold reflex in those examples in which Lower Carniolan and Pannonian point to a short reflex. Originally short *e and neoacute e have identical twofold reflexes, so the rise of twofold reflex can be dated after the merger of short *e and *e. According to Rigler (2001:
15	The nom. sg.m. moj, tvoj, svoj have an analogical short vowel in most of Pannonian (Rigler 2001: 65), but cf. Šafarski (southern Prlekija) 'moj, Biserjane (north-western Prlekija) 'moj, t'voj, s'voj. In the Western Haloze dialect of Zgornja Sveča (Zorko 2009: 197ff.), we find a long reflex in 'vo:la and 'xo:ja as opposed to the short reflex in 'nu:osim, p'ru:osim, šku:oda (cf. ku:oza, 'no:c). This is reminiscent of the long reflex in *koža in eastern Slovene and of the long falling vowel in the volja-type in Kajkavian. Data from other Haloze dialects show a short reflex in *volja and *xoja (see the appendix).
16	In the dialect of the village of Mostec, on the other hand, the forms vula and kuža reflect earlier *volja and *koža, similar to what is found in neighbouring Kajkavian.
— 21 —
Tijmen Pronk
69ff., 313), this twofold reflex is due to two independent waves of lengthening, first in individual cases, and subsequently in all non-final penultimate syllables. This seems extremely unlikely. The attested material is explained much better if one assumes that the conditioning factor for the twofold reflex are the following sounds (Logar 1996: 20, Zdovc 1972: 92, Zorko 1998: 194, Karničar 1990: 30).17 Before */ (perhaps also *n and *j) and syllables containing -i- (perhaps also *j), the reflex was a closed vowel, in most other cases the reflex is an open vowel or diphthong, e.g. Kapla cve'teeda vs. cve'tedi. Ojstrica ne'de:la vs. b'ree:za, cf. also *jagoda > je'goo:da, but *dqte/a > de'teda. Paradigmatic alternations that arose as a result of the phonetic split could be removed or reshuffled by analogies that obscured the picture. If we look at the individual lexemes, we can identify some words that are attested only with a closed vowel, e.g. *nosim, *prosim, *xodim, *meljem, *volja, *zelje and *sedm, some that are attested only with an open vowel or diphthong, e.g. *nesl, *tekl, *pekl, *koža, *škoda, *mečem and *morem, and some which are attested with both reflexes, e.g. Ojstrice 'no:ša and 'noo:ša, Ojstrice 'xo:ja, but Pernice 'xoo:ja, Rinkolach/Rinkole mokr, but Pernice moo:ker, Rinkolach/Rinkole topu, okna, but Grafenbach/Kneža toapu, oa?na, Suetschach/Sveče po:šlam, but Ebriach/ Obir po3:Slem. The details of the development remain unclear and additional data are required to give a complete acount of the development.
I conclude that the Slovene reflex of neoacute *o and *e is not identical to that in Kajkavian, as Rigler claimed, but rather to that in Čakavian and Štokavian: it was always short, except in those cases in which it arose through retraction of the accent from a final jer. This long vowel sometimes spread throughout paradigms or to derivatives from the same root. Some seemingly long reflexes in other positions in dialects without tonal opposition actually reflect a (analogical) neocircumflex accent. This is probably the case in the word koža, which has a reflex of a long root vowel in Pannonian and Styrian, like in Kajkavian. Other volja-type nouns (hoja, noša) reflect a short root vowel in these dialects, with the exception volja itself, which reflects a paradigm with mobile accentuation in eastern Slovene.
6 The rise of long *o and *e in Slavic
In 1916, van Wijk's posited "daß schon im Urslavischen das steigend betonte o etwas länger war als das fallend betonte". His thesis was accepted by Ramovš in his discussion of the reflexes of neoacute *o in Slovene and later by Kortlandt, who argued that "Stang's law yielded a Common Slavic quantitatively neutral rising diphthong *uo" (2011: 21). There is no evidence to support this point of view. The diphthongal reflex of *o that is found in a number of Slovene dia-
17 Zorko's idea "da je široki refleks koroški, ozki pa štajerski" (1989: 398) does not explain the attested distribution, nor the fact that the twofold reflex is found as far West as the Rož dialect.
— 22 —
tS^la-Via Centra lis 1/2016
	 Early Slavic short and long o and e 	
lects, e.g. in Lower Carniolan, cannot be equated directly with the Slovak and dialectal Russian diphthongal reflexes.18 In late Proto-Slavic, *o was at first always short. It obtained a new long counterpart with the rise of long *o as a result of various accent retractions. In Kajkavian and Slovak, *o that received the accent as a result of the accent retraction from a weak final or internal jer or as a result of Stang's law was lengthened. The evidence that the primary reflex in Slovak was a long vowel and not a diphthong is provided by the gen. pl., where we find a diphthong if the root-vowel was *o, but a long vowel in roots containing *a, *i etc. Also cf. Central Slovak dobruo < *dobro < *dobroje (Krajcovic 1971: 48, 78). In Kajkavian, the long vowel is preserved.
The new long *o eventually became the back counterpart of long *e in Slovak (but not of short *e!) and was diphthongized. The same happened in Kajkavian (cf. Vermeer 1983: 454) and Slovene (cf. Rigler 2001: 19, Vermeer 1982: 99, 102), but in Slovene long *o occurred in a largely different set of forms. Slovene, like the rest of western South Slavic, lengthened *o in monosyllables with a circumflex tone (*bog, *nos, *bos). Further instances of long *o arose as a result of the forward shift of the circumflex (*kolo, acc.sg. *sirotg, *boso). The fact that Slovene has a short reflex of neoacute *o and *e in all cases except those that arose through the relatively early retraction of the accent from final jers may be connected with the rise of long *o in other environments.
The diphthongization of short neoacute *e and *o in some Slovene dialects (e.g. Lower Carniolan, Styrian) is a relatively recent development that took place after brata-lengthening and cannot be connected directly to diphthongal reflexes elsewhere in Slavic. In Lower Carniolan, the diphthongization did not affect *e and *o that had not been lengthened, but it did affect neoacute *o that did not arise through accent retraction, e.g. Ribnica krop, gen.sg. kruopa 'boiling water' (Rigler 2001: 175) < *(vb?)krdpb, *(vb?)krdpa. It also affected neocircumflex *o. In Styrian, the diphthongization also affected originally short stressed *e, which had merged with short *e before brata-lengthening, e.g. Upper Savinja ziele < *-e-, sieme < *-e-.
7 Conclusion
Originally short *e and *o were lengthened as a result of the common Slavic retraction of the accent from a word-final jer and as a result of a number of later, dialectal Slavic lengthenings of short rising vowels. These lengthenings
18 Kortlandt (personal communication, cf. also 2011: 250) adduces the western Slovene dialect of Dreznica, which has a distinction between the rising diphthongs ie and i§, as evidence for a diphthongal reflex of neoacute *e that arose as a result of Stang's law. Kortlandt's interpretation of the Dreznica dialect is based on the unpublished field data collected by Logar and presented in tabular form by Greenberg (2000: 171). Greenberg's overview of the vowel system suggests that neither diphthong is in fact the reflex of neoacute *e, which is instead continued by a monophthong §.
— 23 —
Tijmen Pronk
sometimes affected only *o, sometimes both *e and *o and sometimes all short rising vowels. New long *o and *e were subsequently diphthongized in most Slavic dialects. These diphthongizations must be viewed as independent innovations of the dialects in question, although the structural motivation for the diphthongizations was often identical.
APPENDIX: NEOACUTE *O AND *E IN NORTHERN SLOVENE DIALECTS
The following is an overview of the reflexes of neoacute *o and *e in the northern dialects of Slovene that (partly) distinguish neoacute *o and *e from long *e and *o. For most dialects a distinction is made between long and short reflexes. For an overview of the develoment of the relevant vowel systems I refer to Greenberg 2000: 167ff.19
Pannonian
Porabje (Mukič 2005): long domau, pecén, rébra (cf. sg. rébro), káula/káule, kauža, acc. káuzo, pód káuzov, s káuzev, njégvi káuzi, nauvi, sédmi, ausmi, pérdje, short prósi, tócin, gónin, mój, tvój, svój, móren, mécen, skóda, tésen, dóber, móker, žénski, kónjski, séden, ósen, sobóta,potóka, "volja" vóla, židane volé, dóbre vaule, acc. volau, za mójga/tvójga vólo, po vauli, pr vauli, nevóla, nevólo, po nevóli.
^tevanovci/Apátistvánfalva, Hungary, Zorko 1998: 101ff., 2009: 286ff.): long
do'ma:u, zoVe:n, ža'nem, 'a:usmi, na:ui, short 'nieso, (cf. 'niesla), pieko, žienska, 'nuosim, 'nuoša, 'vuola, p'ruosim, 'duobar, 'duobri, 'muoj,' tvuoj, 'uosan, 'sieden.
Prekmurje (Pavel's 2013 normative Prekmurje grammar): long domó, gen. pl. lónec, neszém, treszém, paszém, zsivém, long nom.acc.pl.n. plécsa, bédra, péra, kóla, kópita, kólina, szédmi, ószmi, veszéli, short gen.sg. jelena, zsenszka, nebeszki, svoj, ins.sg. perjem, oszter, moker, dober, szedem, oszem, reko, teko, neszo, morem, hodim, noszim, gonim, orjem, posztelem, polem, kolem, "volja" vola, zavolo toga, but po voli/povdli.
Northern Prekmurje (Cankova, Zorko 2003): ' zeldže, 'nosi, p 'rosi, peko (but cf. pekli), po'vo:uli(k), 'dober, peče:n, peče:š etc. (moj after 'moja, 'mojega etc.), Greenberg 1993: z'elje (pages 468 and 474)/z 'e:lj£ (page 473), ms'e: etc., k'ouzá,
19 I would like to express my gratitude to Dr. Marko Jesensek, who kindly provided me with much of the dialect literature from which forms are cited here.
— 24 —
tS^la-Via Centra lis 1/2016
- Early Slavic short and long o and e -
(Martinje, Greenberg 1993): long ¿\r' sim, pid''siš < *pbješb, k' aožx, gen.pl. k 'aos, 'aotAVA, short z 'isldW, p Herd's, jxl'isna, p 'isbm, parn 'isso, m 'uomš, pr 'uosi, 'uodi < *xodi.
(Žetinci/Sichelsdorf, Austria, Zorko 1998: 87ff.): long 'ko:uža, short 'nosin, 'noša, p'rosin, š'koda, 'vola, 'moga, 'nesa, peka, 'zelje, 'ženska, 'seden.
Central Prekmurje (Polana, Greenberg 1993): long gr 'ouzdja, n'oug, na k'oulana, nas'ie, cvat'ie, živ'ie, short m'ala, p'ala, s'adan, s'admi, m'oramo, n'osimo, v 'ozi.
Southern Prekmurje (Turnišče & Žižki, Zorko 2006: 99f.): long 'nesi:en, ' zebi:e, short 'nosin, p 'rosin, š 'koda, 'zelje, 'nesu.
North-western Prlekija (Rajh 2010): long 'ko:ža, gen.sg. 'ko:že, 're:bra (sg. '^¿ro/a), 'okna, 'o:ken (sg. 'okno/a), 'ko:la, 'ko:rita, 'ko:pita, 'kodna, 'koden (sg. ko 'leno/a), 'o:smi, pe:rje, g 'roizdje, short 'xodin, 'nosin, 'osen, 'že n in, 'nesa, 'tešen, 'do¿er, 'moker, 'topel, z 'nošeni, raz 'ložin, raz 'loženi, "volja" 'vola, gen. 'vode, dat.loc. 'vodi, acc. 'volo, ins. 'voloi.
(Cerkvenjak, Rajh 2002: 11ff.): 'do¿er, 'moker, 'oster, 'topel, 'vola, 'toga 'volo, but ^¿re 'vode, 'nesa (but cf. 'nesla).
(Biserjane, Zorko 2009: 254ff.): long s ko:n, 'mo:j, t 'vo:j, s 'vo:j, short pot 'kožoj, 'zelje, 'xodin, 'nosin, p 'rosi, k 'met, 'nesa, po 'toka, te 'lesa, je 'lena (with raising before *n), 'moreš, 'kola (cf. 'no:č, b 'reza, 'koza, 'sestra).
Southern Prlekija (Šafarski, Zorko 2006: 100f.): mo:j, 'ženska, perje, p 'rosin, š 'koda ('reko < *reklb after * 'rekla because of its vocalism).
Eastern Haloze (Zorko 1998: 14ff., 2006: 103): 'nosin/m,p'rosin/m, 'vola, 'noša, 'xoja, 'd^r, 'zelje, k 'meta, 'ženska, 'ženix (cf. 'bo:uk, secondary 'muj)
(Videm pri Ptuju, Zorko 2009: 225ff.) 'koula (gen.sg. ko 'le:), 'vola, hodin, 'osen, acc.sg. 'motiko, ro¿ača, kjeden (secondary 'muj, ' tvuj, ' svuj, do 'mu).
Western Haloze (Žetale, Zorko 1998: 25ff.), with ¿rata-lengthening: 'nu:osim, 'mu:olim,p'ru:osim, 'xu:odim, 'nu:oša, 'vu:ola, 'xu:oja, š'ku:oda, 'du:obr, 'mu:okr, 'na:uvi (cf. 'ku:oš, 'ku:oza, 'ba:uk), secondary 'mu:oj, neoacute *e merged with the reflex of *e: pe:č like 'ne:so, pe:ko, 're:ko, 'se:dri, 'ze:le, že:nska, že:nix (but 'čalo, 'rakla with retraction of the accent from the final syllable).
(Zgornja Sveča, Zorko 2009: 197ff.) 'nu:osim, p'ru:osim, š'ku:oda, 'vo:la, 'xo:ja (cf. 'ku:oza, 'no:č).
— 25 —
Tijmen Pronk
Slovenske Gorice (Koletnik 2001). The eastern part of the dialect did not undergo brata-lengthening. Accordingly, we find short reflexes in \od/n, 'nos/n, p'ros/n, mol/n, 'vola, ne'vola, za'volo, 'osen, 'kople, 'koleš, 'olje, 'osmi, 'moga, so'bota, škoda, 'melen, 'nesa, odnesa, ' reka, peka,'zelje,'ženska,'meče,'seden, but long 'ko:uža, g'ro:uždje. Like most of Pannonian, 'moj, ' tvoj have a secondary short vowel. The western part of the dialect underwent brata-lengthening: 'nu:os/n, š'ku:oda, 'u:osn, š'ku:orja, 'vu:ola, 'xu:oja, 'du:ober (cf. k'ru:op, pu:ot), but 'ko:uža (cf. 'no:us); 'mie:len, 'sie:dn 'zie:lje (cf. k'mie:t, 'mie:ša, s'nie:xa). The m.sg. form of the participle has a secondary neocircumflex in pe:ka, 'ne:sa, 're:ka (f.sg. 'rekla etc.) < *nesl, *pekl, *rekl, cf. g'ri:izo (f.sg. g'ri:zla) < *grizl.
The dialects more to the west all took part in brata-lengthening. With regard to the reflexes of neoacute *o and *e, the northern dialects are of interest. Below, data from a number of those dialects are presented.
Styrian
In some Styrian dialects, short *o in monosyllables produced a diphthong that is identical to the reflex of *o that received the accent through retraction from a final syllable, i.e. 'kuos = 'kuoza, k'miet = 'ziena.
Eastern Pohorje (Kopivnik, Zorko 2009: 140ff.) 'hu:oja, 'vu:ola, 'nu:osim, p'ru:osim, s'ku:oda, 'ni:eso, 'mi:ecem, 'si:edn, 'zi:ele, s'pi:eko (cf. 'ku:os and the different reflexes in 'bri:eza, da'mu:u, 'kuo:za, 'zie:na, dre'vejsa, 'le:jt, pe:jc, 'no:uc).
(Fram, Zorko 1998: 126ff.): long 'kouza, short 'xu:oja, 'nu:osim, 'vu:ola,p'ru:osim, s'ku:oda, po'tu:oka, 'ni:eso, 'ri:eko, 'zi:ele, jec'mi :ena.
Southern Pohorje (Oplotnica, Zorko 1998: 138ff.): long 'kauza (cf. 'bank 'bog'), short 'vu:la, 'nu:sim, pru:sim, 'du:ber, 'xu:ja, s'ku:da, s'ku:rja, 'nu:ret 'vinograd', po'tu:ka, z'lu:zeno, 'ri:ko, pi:ko, 'zi:le, jec'mi :na (cf. paic 'pec', 'riekla 'rekla', 'kuos, k'miet).
Sevnica-Krsko (Zemljak 2001): 'vu:ola, 'xu:oja, 'nu:osa, 'nu:osrn, p'ru:osrn, g'ru:ojzje, s'ku:oda, k'mi:eta, 'mi:ecem, pi:eku, 'ri:eku,psr'ti:eku, 'zi:ele, 'zi:enska (secondary 'duo:bdr, pi:rje).
Upper Savinja (Luce, Rigler 2001: 217ff.): short xuodi, nuosi, guonim, uossm, tuopu, uole, ziele, zienin, mesu, pieku, siedsm (cf. pec, bdk). Any *o that received the stress through the forward shift of the acute or the retraction of the neocircumflex has the same timbre as neoacute *o: jaguoda, bskuoica 'bukovica, bukev', kukuoica 'kukovica, kukavica', muotka 'motika', uotava 'otava'. The retractions of a short final syllable and from syllable with a long
— 26 —
tS^la-Via Centra lis 1/2016
- Early Slavic short and long o and e -
falling accent (but not a neocircumflex) yielded different diphthongs: buoga, vuoda, ni§ba, ži§na.
(Spodnje Kraše, Weiss 2001): 'ne:su, pe:ku, 'ze:le, pe:rja, 'že:nen, 'noise;m, Xo:de, 'do:bar, š'ko:da, š'ko:rja, 'ko:ža, Xo:ja, 'no:ša, also cf. 'vala, ja'go:da, 'mo:tka (cf. 'no:s, pe:č, 'kuo:za, 'sie:na, 'žie:na). The word 'že:nske has a 'long' reflex which is probably due to a local innovation.
In the other Styrian dialects neoacute *o and *e with brata-lengthening merged with older long *o and *e.
Kozjansko-Bizeljsko (Lesično, Zorko 2009: 184ff.) 'u:sn, 'nu:sim, p'ru:sim, š ku:da, g ru:jzdje, but xo:ja, no:ša, vo:ja.
(Pišece, Zorko 2009: 187ff.): 'vu:la, 'nu:sim,p'ru:srt, but Xo:ja, š'ko:da, g'rojzdje.
(Kapele, Zorko 2009: 190ff.): 'nu:srt, p'ru:sn, š'ku:da, 'vuda, but Xo:ja, 'no:ša.
(Mostec, Toporišič 1962): long damu, sinuf, nuk, nusi, muli, sela, melem, jelena, reku < *-o-, *-e-, vula, kuža < *-o-.
Central Styrian (Šmarje pri Jelšah, Povše 1988): dumav, utrauk, utraucmi, daubar, mauj, ausn, haudim, gaunim, vauzim, kauplem, maurem, maugu, saidrt, te traik(i), raik(u), draimlem,putaiplem, ječmaina, jelaina (cf. ukau, raič, kmiet, škuof žiena, nuoga).
(Žahenberc, Zorko 2009: 202): long Xa:uja, 'na:usim, p'ra:usim, short 'nu:oša (cf. me'sa:u, 'kuo:za).
Carinthian
In most of Carinthian except the Gailtal dialect, neoacute *o and *e and acute *e have twofold reflexes depending on the following sounds (see section 5 above).
Kozjak (Kapla, Zorko 1998: 194ff.): 'no:sim, p 'ro:sim, 'vo:zim, 'no o:šen, p 'rop o:šen, z'voo:žen, 'me:lem, 'kodem, po:lem, 'nee:so (cf. 'nee:swa, 'nesli), 'ree:ko (ree:kwa, 'rekli), 'pee:ko, ('pee:kwa, pekli), but secondary 'mu:arm.
Northern Pohorje-Remšnik (Remšnik, Zorko 2009: 83ff.) 'vo:la, b'ro:dim, go:nim, 'no:sim, Xo:dim, 'lo:mim, 'mo:dlim,p'ro:sim, 'to:čim, 'že:nim se,' se:dom, s'te:la, pog're:ba, t're:tki, sar'še:na, 'te:še, pe:lem se, klepe:če, o'tepa, but 'ne:e:so, pee:ko, 'ree:ko, 'tee:po, 'koo:ža, 'moo:kar, 'doo:bar, Xoo:ja, 'ngo:ša, po 'ko o:šen, po 'to o:ka, po 'do o:brt, pro o:šen, s Xo o:jen, š 'ko o:da, z'lo o:men (cf. ne 'de:la, ko 'le e:na < *-e -, 'nu:ak, 'ši:ast, 'se e:stra).
— 27 —
Tijmen Pronk
Mežica (Strojna, Zorko 2009: 31ff.) 'ze:le, Xo:je, 'noše, 'vode, 'že:ni sa,p'ro:sim, 'no:sim, 'xo:dim, 'se:dri, 'ne:su, 're:ku, but 'doo:ri < *ddbri.
Podjuna (Pernice, Zorko 1988): 'ze:le, 'se:dom, but 'nee:so (nee:sla, -o, -e but
'nesli), pee:ko, ječ'mee:na; 'no:sim, p'ro:sim, but Xoo:ja, 'moo:ker, š'koo:da.
(Ojstrica, Zorko 1991, 1998: 190ff.): 'wo:la, Xo:ja, 'no:ša/'n«o:ša, 'no:sim, 'mo:dlim, 'o:som, but 'koo:ža, š'k«o:da; 'me:lem, Ze:le, 'že:nska, 'že:nin, le'me:ža, but 'mee:čem, 'dee:sri, 'nee:so (nee:swa, 'nesli), 'pee:ko ('pee:kwa, pekli), 'ree:ko (ree:kwa/'rekwa, 'rekli).
(Rinkolach/Rinkole, Zdovc 1972: 92), with tonal opposition: gen.pl. nuax, koža, morš, more, sabota, škoda, perie, nesu, peku, teku, with a closed vowel hoja, uola, grozdi e, snopie, kozi a, ta noua, hodi, prosi, nosi, puhoien, naprošen, utroška, dobr, mokr, topu, okna, zele, sedm, sedmi, cf. also otoua, but ograda, motaka, hoduua, prosuua with a retracted neocircumflex (cf. muaia siastra, gen.sg. stoga < *stoga). Younger speakers of the dialect have a change *§ > e before a nasal: iačmena, ialena (ibidem: 96).
Ebriach/Obir (Karničar 1990), with tonal opposition: no:x, utro:q, mo:hu, qo:jsq, utro:šq, že:jsq, zeds, qo:zi, wo:la, xo:ja, qo:lem, moa:re,poa:šlem, qoa:ža, tea:šem, mea:čem, me:lem, no:sam, xo:dam, rea:qu, nea:su, do:bar, mo:qar, saboa:ta, škoa:da, woa:tawa, puhre:ba, ječmena, jele:na, sea:dam 'seventh', se:dam 'seven' (cf. bd:x, le:d).
Rož (Grafenbach/Kneža, Logar 1996: 292ff.), with tonal opposition: se:dm, ta tre:t?i, že:n(a) sa, ut?le:ne, neasu, pumeadu, pleadu, tea?u, seadm, puameana, (before /) zo:l(a), mo:lm; no:srn, uo:za, o:sm, uo:la, do:har, hro:zdi, ?oaža, oasma, oa?na, toapu, ?oapl, s?oapan, cf. gen.pl. žian, but žians?a, psri. Judging by the reflexes of acute *e in non-final syllables, the situation is more complicated here, e.g. de:u(a), measta, pe:na/peana, ure:ha, de:čla. Cf. further, with stress retractions, 'neha, 'sastra, 'zemua. The latter show two different outcomes of pretonic *e, also dependent on the following sounds (ibidem: 296).
(Suetschach/Sveče, Feinig 1985), with tonal opposition: se:dam, tre:tja, jale:na, že:nam se,puhre:ba, jačme:na,pe:lam, me:č(l)am, qle:plam, te:šam, wo:la, xo:ja, no:ša, sqo:rja, no:sam, wo:zam, xo:dam, po:šlam, qo:plam, hro:zdja, but qo:ža, ne:su, pe:qu, te:qu, re:qu, cf. f.sg., m.du. raqw'a, f.du., pl. raql'e, m.pl. raql'a. Twofold reflex of acute *e in non-final syllables: čwe:qa, sose:da, wre:ša/wre:xa, de:qla, but bre;:za, mre;:ža, de;:wo, le;:to, qole;:no, pule;:no, m^:st3, c^:sta, smm-qa, pe:na, stri;:xa, nawi;:sta. The examples are identical to those of Logar. Cf. also gen.pl. ž'e:n, nom.sg. m'e:d 'medica' (gen.sg. jm^:na (after nom.acc.sg. jm'e:?), but qole:sa, wrame:na), as opposed to gen.pl. n'u:x, q'u:jn. Cf. also the limited data for nearby FrieBnitz/Breznica (Rigler 2001: 279ff.): ?o:ža, no:sam, xo:djo, wo:la, pe:kou, me:lam.
— 28 —
tS^la-Via Centra lis 1/2016
	 Early Slavic short and long o and e 	
Gailtal (Potschach, Pronk 2009), with tonal opposition and a single reflex for short neoacute *o and *e. Vowels in closed syllables were regularly shortened:
more, ola, noša, koža, kople, pele, kleple, dobr, topu, nabese, patoka, ženšči, sabota, nesu, teku, reku, nosn, hodn, gonan, definite forms with a neocircum-flex: nobi, but zaliani, basuači, secondary kuazji. Long reflexes: muaj, nuag, guar.
LITERATURE
Aleksandar BELIC, 2000: Beleške o čakavskim govorima. O dijalektima. Beograd: Zavod za udžbenike i nastavna sredstva. (Izabrana dela A. Belica, 10). 136-186.
Buro BLAŽEKA, Grozdana ROB, 2014: RječnikMurskogSredišča. Zagreb: Učiteljski fakultet Sveučilišta u Zagrebu.
Anton BREZNIK, 1916: Slovenska slovnica za srednje šole. Klagenfurt: Druža sv. Mohorja.
Leonid Arsenijovyč BULAXOVS'KYJ, 1947: Sxidnoslov'jans'ki movy jak dzherelo vidbuduvannja spil'noslov'jans'koi akcentologičnoi systemy. Movoznanstvo 4-5, 7-17.
Leonid Arsen'evič BULAXOVSKIJ, 1958: Otraženija tak nazyvaemoj novoakutovoj intonacii drevnejšego slavjanskogo jazyka v vostočnoslavjanskix. Voprosy jazykoznanija 7/2, 87-92.
—, 1983: Izbrannye trudy v pjati tomax. Tom pjatyj: slavjanskaja akcentologija. Kiev: Naukova dumka.
Rick DERKSEN, 2008a: Quantity patterns in the Upper Sorbian noun. Evidence and counter-evidence. Essays in honour of Frederik Kortlandt. Volume I: Balto-Slavic and Indo-European linguistics. Amsterdam: Rodopi. (Studies in Slavic and General Linguistics, 32). 121-135.
—, 2008b: Slavic evidence for Balto-Slavic oxytona. Stressing the past: Papers from the Second International Workshop on Balto-Slavic Accentology (IWoBA II). University of Copenhagen, 1-3 September 2006. Amsterdam-New York: Rodopi. (Studies in Slavic and General Linguistics, 35). 15-19.
Rainer FECHT, 2010: Neoakut in der slavischen Wortbildung: Der volja-Typ. Roll. Dettelbach.
Tatjana FEINIG, 1985: Govor vasi Sveče v Rožu s težiščem na morfologiji. (Unpublished thesis). Klagenfurt.
Marc L. GREENBERG, 1993: Glasoslovni opis treh prekmurskih govorov in komentar k zgodovinskemu glasoslovju in oblikoglasju prekmurskega narečja. Slavistična revija 41/4, 465-487.
—, 2000: A Historical Phonology of the Slovene Language. Heidelberg: Winter.
Peter HOUTZAGERS, 1999: The Kajkavian dialect of Hidegseg and Fertohomok. Amsterdam: Rodopi. (Studies in Slavic and General Linguistics, 27).
— 29 —
Tijmen Pronk
Stjepan IVŠIC, 1911: Prilog za slavenski akcenat. Rad JAZU 187, 133-208. —, 1913: Današnji posavski govor (svršetak). Rad JAZU 197, 9-138. - -, 1936: Jezik Hrvata kajkavaca. Ljetopis JAZU 48, 47-88.
Mate KAPOVIC, 2007: The *vola-type accent in Slavic. Tones and Theories: Proceedings of the International Workshop on Balto-Slavic Accentology. Zagreb: Institut za hrvatski jezik i jezikoslovlje. 89-104.
—, 2009: O naglasku e-glagola u zapadnoj Posavini. Croatica et Slavica Iadertina 5, 107-121.
Ludwig KARNIČAR, 1990: Der Obir-Dialekt in Kärnten; die Mundart von Ebriach/ Obirsko im Vergleich mit den Nachbarmundarten von Zell/Sele und Trögern/Korte. Vienna.
Mihaela KOLETNIK, 2001: Slovenskogoriško narečje. Maribor: Slavistično društvo. (Zora, 12).
Frederik KORTLANDT, 2009: Baltic & Balto-Slavica. Amsterdam-New York: Rodopi.
--, 2011: Selected writings on Slavic and general linguistics. Amsterdam-New York,
NY: Rodopi.
—, 2014: On the accentuation of nesh> and related issues. Rasprave: Časopis Instituta za hrvatski jezik i jezikoslovlje 40/1, 127-131.
—, 2015: Proto-Slavic *j, Van Wijk's law, and e-stems. Rasprave: Časopis Instituta za hrvatski jezik i jezikoslovlje 41/1, 65-76.
Rudolf KRAJČOVIČ, 1971: Načrt dejin slovensheho jazyka. Bratislava: Slovenske pedagogicke nakladel'stvo.
Tine LOGAR, 1996: Dialektološke in jezikozgodovinske razprave. Ljubljana: ZRC SAZU, Inštitut za slovenski jezik Frana Ramovša.
Francek MUKIČ, 2005: Porabsko-knjižnoslovensko-madžarski slovar. Szombathely: Zveza Slovencev na Madžarskem.
Elisabeth NONNENMACHER-PRIBIC, 1961: Die baltoslavischen Akzent- und Intonationsverhältnisse und ihr quantitativer Reflex im Slovakischen. Wiesbaden: Harrassowitz.
Vilko NOVAK, 2006: Slovar stare knjižne prekmurščine. Ljubljana: ZRC SAZU.
Avgust PAVEL, 2013: Prekmurska slovenska slovnica/Vend nyelvtan. Maribor: Slavistično društvo. (Zora, 100).
Tijmen PRONK, 2007: The retraction of the neocircumflex in the Carinthian dialects of Slovene (on. Ivšic's retraction). Tones and Theories: Proceedings of the International Workshop on Balto-Slavic Accentology. Zagreb: Institut za hrvatski jezik i jezikoslovlje. 171-183.
—, 2009: The Slovene Dialect of Egg andPotschach in the Gailtal, Austria. Amsterdam-New York: Rodopi. (Studies in Slavic and General Linguistics, 36).
30	(Centra lis 1/2016
	 Early Slavic short and long o and e 	
—, 2013: On the accentuation of /-participles of the type nesh, in western South Slavic. Rasprave Instituta za hrvatski jezik i jezikos/ov/je 39/1, 105-131.
Bernhard RAJH, 2002: Od narečja do vzhodnoštajerskega knjižnega jezika. Maribor: Slavistično društvo. (Zora, 19).
—, 2010: Gučati po antujoško. Maribor: Slavistično društvo. (Zora, 73).
Fran RAMOVŠ, 1921: O slovenskem novoakutiranem à > o, q, o. Slavia 2, 227-239.
Jakob RIGLER, 2001: Zbrani spisi 1: Jezikozgodovinske in dia/ekto/oške razprave. Ljubljana: ZRC SAZU.
Afanasij Matveevič SELIŠČEV, 1951: Staros/avjanskij jazyk. Čast'pervaja: vvedenie fonetika. Moskva: Gosudarstvennoe učebno pedagogičeskoe izdatel'stvo ministerstva prosveščenija RSFSR.
George Yurii SHEVELOV, 1979: A historica/ phono/ogy of the Ukrainian /anguage. Heidelberg: Carl Winter.
Christian STANG, 1957: Slavonic accentuation. Oslo-Bergen-Troms0: Universitets-forlaget.
Matej ŠEKLI, 2013: Sklonidba i naglasak imenica srednjega roda o-sklonidbe u govoru sela Jevšček kraj Livka (nadiški dijalekt slovenskoga jezika). Rasprave Instituta za hrvatski jezik i jezikoslovlje 39/1, 133-145.
Petar ŠIMUNOVIC, 2009: Rječnik bračkih čakavskih govora. Drugo dopunjeno i popravljeno izdanje. Zagreb: Golden marketing-Tehnička knjiga.
Stanislav ŠKRABEC, 1917: Jezikoslovni spisi. I. zvezek, 2. snopič. Ljubljana: Leonova Družba.
Ivan TOMINEC, 1964: Črnovrški dia/ekt. Kratka monografija in s/ovar. Ljubljana: SAZU.
Jože TOPORIŠIČ, 1962: Ablösung des relevanten Wortintonationssystems durch den Quantitätsunterschied in einer slovenischen Mundart. Scando-S/avica 8, 239-254.
André VAILLANT, 1950: Grammaire comparée des /angues s/aves. Tome I: Phono/ogie. Paris-Lyon: C. Klincksieck.
Matija VALJAVEC, 1897: Glavne točke o naglasu kniževne slovenštine. Rad JAZU 132, 116-213.
Nicolaas VAN WIJK, 1916: Zur sekundären steigenden Intonation im Slavischen, vornehmlich in ursprünglich kurzen Silben. Archiv für s/avische Phi/o/ogie 36, 321-377.
Leonid Lazarevič VASIL'EV, 1929: O značenii kamory v nekotoryx drevnerusskix pamjatnikax XVI-XVII vekov. Leningrad: AN SSSR.
Willem VERMEER, 1982: Raising of *e and loss of the nasal feature in Slovene. Zbornik za filologiju i lingvistiku 25/1, 97-120.
—, 1983: The rise and fall of the kajkavian vowel system. Studies in Slavic and General Linguistics 3, 439-477.
— 31 —
Tijmen Pronk
—, 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.
Arno VERWEIJ, 1994: Quantity patterns of substantives in Czech and Slovak. Dutch contributions to the 11th international congress of Slavists: Linguistics. Amsterdam: Rodopi. (Studies in Slavic and General Linguistics, 22). 493-564.
Peter WEISS, 2001: Fonološki opis govora kraja Spodnje Kraše (SLA 314). Jezikoslovni zapiski 7/1-2, 321-347.
Andrej Anatol'evič ZALIZNJAK, 1985: Ot praslavjanskoj akcentuacii k russkoj. Moskva: Nauka.
-	-, 2011: Trudy po akcentologii. Tom II: drevnerusskij i starovelikorusskij akcentologičeskij slovar'-ukazatel' (xiv-xvii vv.). Moskva: Jazyki slavjanskix kul'tur.
Paul ZDOVC, 1972: Die Mundart des Südöstlichen Jauntales in Kärnten: Lautlehre und Akzent der Mundart der "Poljanci". Wien: Verlag der ÖAW.
Melita ZEMLJAK, 2001: Fonološki opis posavskega govora v Stržišču. Jezikoslovni zapiski 7/1-2, 349-363.
Zinka ZORKO, 1988: Perniški govor. Wiener slawistischer Almanach 22, 255-272.
-	-, 1989: Vzhodnokoroški govori v Dravski dolini, na severnem Pohorju in v Dravskem obmejnem hribovju. Zbornik razprav iz slovanskega jezikoslovja: Tinetu Logarju ob sedemdesetletnici. Ljubljana: SAZU. 395-405.
—, 1991: Oblikoslovje Ojstrice nad Dravogradom (z glasoslovnim orisom). Slavistična revija 39/1, 15-28.
—, 1998: Haloško narečje in druge dialektološke študije. Maribor: Slavistično društvo. (Zora, 6).
—, 2003: Oblikoslovje in leksika v govoru Cankove. Avgust Pavel. Maribor: Slavistično društvo. (Zora, 23). 73-94.
—, 2006: Samoglasniški sestavi v vzhodnih govorih panonske in štajerske narečne skupine. Diahronija in sinhronija v dialektoloških raziskavah. Maribor: Slavistično društvo. (Zora, 41). 98-107.
—, 2009: Narečjeslovne razprave o koroških, štajerskih in panonskih govorov. Maribor: Slavistično društvo. (Zora, 64).
Myxajlo Andrijovyč ŽOVTOBRJUX, Oleksij Tymofijovyč VOLOX, Stefan Pylypovyč SAMIJLENKO, Ilarion Ilarionovyč SLIN'KO, 1980: Istoryčnagramatyka ukraïns'koï movy. Kyïv: Holovne vydavnyctvo vydavnyčoho objednanyja "Vyšča škola".
32	i-Së—vt-— G^ntr— lis 1/2016
	 Early Slavic short and long o and e 	
ZGODNJESLOVANSKA KRATKA IN DOLGA O IN E
Prvotno kratka vokala *o in *e sta bila podaljšana kot rezultat skupnega slovanskega naglasnega umika s končnega jera in kot rezultat številnih kasnejših narečnih slovanskih podaljšav kratkih rastočih vokalov. Te podaljšave so včasih vplivale le na vokal *o, včasih na vokala *e in *o in včasih na vse kratke rastoče vokale. Nova dolga vokala *o in *e sta bila kasneje v večini slovanskih narečij diftongizirana. Ta pojav je sicer treba obravnavati kot neodvisne narečne inovacije, četudi je bila strukturna motivacija za diftongizacijo pogosto identična.
Prispevku je dodan izčrpen pregled novoakutiranih refleksov *o in *e v severnih slovenskih narečjih.
— 33 —