Jezikoslovni zapiski 21 • 2015 • 2 _ 21 Frederik Kortlandt The development of vowel length in Slavic CoBiss: 1.01 Razvoj vokalne dolžine v slovanščini V prednaglasnih zlogih so dolgi vokali izšli iz Dybojevega zakona, medtem ko so v naglašenih in ponaglasnih zlogih nadaljevanje praindoevropskih vokalov s podaljšano stopnjo in narečnih indoevropskih kontrakcij in so izšli iz slovanske monofongizacije difongov ter so po nastanku novih barvnih razlik nasledek Van Wijkovega zakona in kontrakcij v ponaglasnih zlogih, v naglasni paradigmi (c) umika naglasa s končnih jerov in daljšanja v enozložnicah, v na-glasni paradigmi (b) Stangovega zakona, samo v ponaglasnih zlogih pa preminitve laringalov. Ključne besede: laringali, podaljšana stopnja, monoftongizacija, kontrakcije, daljšanje, krajšanje In pretonic syllables long vowels originated from Dybo's law while in stressed and posttonic syllables long vowels continue Proto-Indo-European lengthened grade vowels and dialectal Indo-European contractions and arose from the Slavic monophthongization of diphthongs, and after the rise of the new timbre distinctions resulted from Van Wijk's law and contractions in posttonic syllables, in accent paradigm (c) from the retraction of the stress from final jers and from lengthening in monosyllables, in accent paradigm (b) from Stang's law, and only in post-posttonic syllables from the loss of laryngeals. Keywords laryngeals, lengthened grade, monophthongization, contractions, lengthening, shortening Though the importance1 of Stang's classic monograph (1957) is generally recognized, the consequences of his findings have not yet been properly understood by the majority of scholars in the field. In the pre-Stang era, scholars tried to derive the accentuation of Slavic word forms from inherent tonal properties of their constituent morphemes. According to Hirt's law, an acute vowel attracted the accent from a following syllable. According to de Saussure's law, an acute vowel attracted the accent from a preceding non-acute syllable. Since it has now become clear that the acute was not a tonal, but a glottalic feature, the assumption of inherent tonal features of morphemes must be abandoned and replaced by the reconstruction of glottalized versus non-glottalized syllables. 1 This article was presented at the third Slavic studies conference, as part of which the Department of Slavic Languages at the University of Ljubljana's Faculty of Arts hosted the Tenth International Workshop on Balto-Slavic Accentology (10. mednarodna konferenca o baltoslovanskem naglasoslovju / X Mewdynapodnbiu ceMUHap no 6ajim0-cjiaBMHCK0u aK^eHmoMo^uu, IWoBA X) from October 16th to 18th, 2014. 22 _ Frederik Kortlandt • The development of vowel length in Slavic The most important result of Stang's analysis is that the Slavic accent patterns must not be derived from inherent tonal properties of their constituents but, conversely, that the tones must be derived from the accent patterns (1957: 179). Stang showed that the acute is characteristic of paradigms with fixed stress (a), that the neo-acute developed from a retraction of the stress in paradigm (b), and that the circumflex is characteristic of paradigms with mobile stress between initial and final syllables (c). Dybo has shown that paradigm (b) developed from a paradigm with fixed stress as a result of an accent shift from a non-acute vowel to a following syllable (1962, 1968). Since paradigms (a) and (b) are in complementary distribution, they can be identified with Lithuanian accent patterns (1) and (2). Since the acute was glottalization, not a tonal movement, it follows that the rise of distinctive tone was a more recent development. This offers a simple explanation for the fact that the normal reflex of the acute is falling while the circumflex is rising in standard Lithuanian, whereas the converse distribution is found in Latvian. It appears that the rise of distinctive tone was a development of the separate languages (cf. already Kortlandt 1977). It probably never reached Prussian (cf. Kortlandt 2009: 267). In Slavic, distinctive tone originated in initial syllables when the accent was retracted to a preposition or prefix in barytone forms of mobile paradigms (c) at stage 6.10 of my chronology (2011: 166, 301), for example, in Russian na vodu 'onto the water', ne byl 'was not', prodal 'sold', povod 'rein'. If the accent was a high tone, as it was in Sanskrit, this development is perhaps best understood as a generalization of the low tone of pretonic syllables in barytone forms of mobile paradigms, which received a falling tone movement on the initial syllable. The result may be compared with the system of standard Serbo-Croatian, which has a falling tone on initial syllables and a rising tone on non-final syllables. In non-initial syllables, distinctive tone on long vowels originated as a result of Dybo's law at stage 8.7 of my chronology (2011: 171, 305) because newly accented long vowels received a falling tone movement, as opposed to stressed long vowels of an earlier date. Long falling vowels in final syllables (not counting final jers) lost the stress to the preceding accentuable syllable according to Stang's law (stage 9.3 of my chronology) and were shortened, as were all other long falling vowels except in Slovene monosyllables and Serbo-Croatian monosyllabic and disyllabic word forms, for example, bog 'god', acc.sg. ruku 'hand' (stage 9.4 of my chronology). The combination of Dybo's law and Stang's law gave rise to accentual mobility between adjacent syllables in paradigm (b), with a rising tone in the first syllable and a short vowel in the second. Short rising vowels were lengthened under certain conditions in Russian, Czech, Upper Sorbian, and Slovene (cf. Kortlandt 2011: 173f., 307f.), for example, Czech vule, Slovene vQlja 'will'. Finally, Slovene developed new long falling vowels under certain conditions, for example, oko 'eye', mladost 'youth', bitka 'battle', leta 'years', osnqva 'base' (stages 10.7-10.9 of my chronology). Thus, we have an accent paradigm (a) with fixed stress on an acute vowel, an accent paradigm (b) with a rising tone (neo-acute) in some forms and a short jezikoslovni zapiski 21 • 2015 • 2 _ 23 accent on the following syllable in others, and an accent paradigm (c) with a falling tone (circumflex) on the initial syllable (or proclitic element) in some forms and a short or long rising tone on the ending (or enclitic element) in others. When final jers lost their stressability (stage 8.2 of my chronology), the preceding accentuable syllable received a long rising tone, for example, Slovene gen. pl. gor < *gori 'mountains', dan < *dbni> 'days', ovbc < *owbCb 'sheep', Polish rqk < *rQkb 'hands', Russian dat.pl. detjam < *detbmi> 'children', all (c). When the acute eventually lost its glottalic character (stage 9.2 of my chronology), it merged with the short rising tone, for example, Slovene dim 'smoke', gora < *gora 'mountain', Ukrainian moroz < *-oro- 'frost' as opposed to gen.pl. holiv < *-olo- 'heads'. As a result of the loss of glottalization, the almost universal shortening of long falling vowels, and the widespread lengthening of short rising vowels, distinctive tone was limited to Slovene and Serbo-Croatian while vowel length remained distinctive everywhere up to a later stage (cf. Kortlandt 2011: 111-115 on Bulgarian). The oldest type of long vowel in Balto-Slavic are Proto-Indo-European lengthened grade vowels, for example, Lith. dukte 'daughter', akmud 'stone', Greek dvyatnp, aK^mv, SCr. zerav 'crane', sigmatic aorist 1st sg. donijeh 'brought', umri-jeh 'died', root nouns Lith. gela 'pain', zole 'grass', mesa 'meat', all (4), SCr. rijec 'word', car 'magic', sam 'alone', Czech car, cara, sam (b). In principle, these long vowels were never shortened (cf. Kortlandt 1985; Vermeer 1992). The second-oldest type of long vowel in Balto-Slavic developed from the loss of a laryngeal between two full vowels (*e, *o), for example, Lith. gen.sg. algds 'salary' < *-as < *-aHes, Greek alyffc. This was a dialectal Indo-European development which Balto-Slavic shared with Indo-Iranian, but not with Greek, where the circumflex points to a disyllabic sequence at an earlier stage of the language. Other long vowels originated in the separate branches of Balto-Slavic. At that time, the remaining laryngeals had merged into a glottal stop, for example, Lith. alga (4) < *-a?, galva (3) 'head' < *gol?wa? < *golHuaH, and the Proto-Indo-European glottalic consonants had dissolved into a laryngeal and a buccal part (Winter's law, stage 4.3 of my chronology), for example, Latvianp§ds < *pe2dom 'footstep', nuogs < *no?gwos 'naked'. In Slavic, glottalization was lost in pretonic and post-posttonic syllables with compensatory lengthening of an adjacent vowel (stage 5.3 of my chronology), for example, *golwa? < *gol?wa? 'head', *pila? < *p?ila? '(she) drank', inst.sg. *sunumi < *su?numi 'son', *opsnowd < *dpsnowa? 'base', inst.pl. *gena?mis < *gena?mi?s 'women'. The long vowel in the final syllable of the latter words is reflected by the neo-circumflex tone of Slovene osnQva < *osnowa, zenami < *zenami, where the middle syllable received the stress as a result of Dybo's law. Glottalization was eliminated by analogy in barytone forms of mobile accent paradigms (Meillet's law, stage 5.4 of my chronology), for example, SCr. sin 'son', acc.sg. glavu, neuter pilo (cf. Lith. galvq, sumt). Glottalization was preserved in stressed and first posttonic syllables up to a later stage. 24 _ Frederik Kortlandt • The development of vowel length in Slavic New long vowels originated from the monophthongization of diphthongs: *e < *ai, *e < *ei, *o < *au (my stage 6.5). The rise of nasal vowels *iN, *eN, *aN, *on, *un can be dated around the same time. The same holds for the rise of glottalized vowels I, e, e, a, o, u, which had the timbre of the corresponding long vowels, as in the case of the Latvian broken tone in i, ie, e, a, uo, u. At a later stage (7.8), the rounded vowels *u, *u, *un and their glottalized counterparts were de-labialized to *y, *y, *yN, after palatalized consonants *i, *i, *iN, and the long mid vowels *e and *o were subsequently raised to *i and *u (stage 7.9). This resulted in the following vowel system (cf. Kortlandt 2011: 106): 1 V u eN ON i y e a aN e a Here, the long vowels and the nasal vowels could be either glottalized (acute) or not. In initial syllables, the non-acute vowels could be either falling (circumflex) or not. At this stage (7.13), the loss of glottalization in posttonic syllables gave rise to a series of new short vowels i, e, a, u, y, which were opposed to the older short vowels b, e, o, t by timbre and vowel height. The result is the following vowel system (cf. Kortlandt 2011: 107): i y u e b o eN ON e a aN In stressed syllables, the acute vowels were now half-long while the non-acute vowels could be either long or short. In pretonic syllables, long vowels were shortened and the opposition between long and short vowels was replaced by the new timbre distinctions. In posttonic syllables, vowel length remained distinctive but final nasal vowels were shortened, for example, SCr. nom.acc.pl. glave with a short ending versus gen.sg. glave < *-q 'head', Slovene gen.sg. krave (a) 'cow' without the neo-circumflex versus gorq (c) 'mountain' with a long vowel, similarly Susak (Croatian) gen.sg. sestre (b) 'sister' versus vodie (c) 'water'. There is no trace of glottalization in final nasal vowels. The mid vowels e, b, t, o were always short, but that was to change very soon. According to Van Wijk's law (stage 7.15), clusters of consonant plus *j were shortened with compensatory lengthening of the following vowel, for example, *pise < *pisje 'writes', *wola < *wdlja 'will'. New *e did not merge with earlier *e, which had become e at this stage. Case endings could now have three different quantities. For example, the nom.sg. ending of the a-stems was short in *zena, long in *wdla and *osnowa, and half-long in *gora. The same distribution holds for the neuter nom.acc.pl. ending. At the next stage, several levelings took place. Endings jezikoslovni zapiski 21 • 2015 • 2 _ 25 which did not occur under the stress were shortened in the whole Slavic territory, for example, gen.sg. *koya, dat.sg. *ko%u, nom.pl. *ko%i 'horse', dat.sg. *zene 'woman', *poNti 'way'. Length was generalized in the unstressed nom.acc.pl. ending of Slovene leta < *-a 'years', but not under the stress, cf. drva < *-a 'firewood'. Other new long vowels originated from contractions in posttonic syllables (stage 8.1), for example, Cakavian (Novi) pita 'asks', Bulg. pita, cf. Cak. kopa < *kopa(j)e 'digs', Bulg. kopae, Old Polish kopaje. Here again, new *e did not merge with earlier *e, for example, Czech gen.sg. noveho 'new'. New long vowels under the stress arose when the accent was retracted from final jers in mobile accent paradigms (stage 8.2), for example, Slovene gen.pl. gor < *gori 'mountains', dan < *dbni> 'days', ovdc < *owbCb 'sheep', Polish rqk < *rQki> 'hands', Russian dat.pl. detjam < *detbmi> 'children'. The vowel length in the gen.pl. forms subsequently spread analogically to other accent paradigms. In pretonic syllables, vowel length became distinctive when in line with Dybo's law the accent shifted from non-acute vowels to the following syllable (stage 8.7), for example, *narodb 'people', *ONtrdba 'entrails', Slovene narod, vQtroba. Short falling vowels in monosyllables were lengthened (stage 8.8), for example, SCr. bog 'god', kost 'bone', dan 'day'. The final loss of glottalization in stressed syllables gave rise to new short rising vowels (stage 9.2), for example, Slovene dim 'smoke', gora < *gora 'mountain'. The retraction of the stress from long falling vowels in final syllables (Stang's law, stage 9.3) yielded new long rising vowels and short rising diphthongs 'e, uo, for example, *wuola < *wola < *wola (Dybo) < *wolja (Van Wijk), Czech vule, Slovak vol'a, Slovene volja, SCr. volja. These developments were followed by lengthening of short rising vowels and shortening of long falling vowels under certain conditions and by the rise of new long falling vowels in Slovene. The distinction between diphthongized jat ie < e and the new diphthong 'e < e has been preserved as ie versus i§ in the Slovene dialect of Soca (cf. Greenberg 2000: 171). Summarizing, we can say that in pretonic syllables long vowels originated from Dybo's law, while in stressed and posttonic syllables long vowels continue Proto-Indo-European lengthened grade vowels and dialectal Indo-European contractions and arose from the Slavic monophthongization of diphthongs, and after the rise of the new timbre distinctions resulted from Van Wijk's law and contractions in posttonic syllables, in accent paradigm (c) from the retraction of the stress from final jers and from lengthening in monosyllables, in accent paradigm (b) from Stang's law, and only in post-posttonic syllables from the loss of laryngeals. It may now be useful to see how these developments are reflected in nominal case endings. Here I give the paradigms of krava (a) 'cow', konjb (b) 'horse',pQtb (b) 'way', volja (a/b) 'will', igo (c) 'yoke' and detq (b/c) 'child' in the pre-Slovene dialect of Slavic at stages 8.0 (after Van Wijk's law), 9.0 (after lengthening in monosyllables) and 10.0 (after merger of the jers; for a fuller account, see Kortlandt 2011: 277-309). 26 _ Frederik Kortlandt • The development of vowel length in Slavic Slovene (8.0, 9.0, 10.0) nom.sg. krawa krawa krawa gen.sg. krawy krawy krawy dat.sg. krawa krawe krawe acc.sg. krawoN krawoN krawoN inst.sg. krawooN krawoN krawoN loc.sg. krawa krawe krawe nom.pl. krawy krawy krawy gen.pl. krawb krawt krawb dat.pl. krawamt krawamt krawamb acc.pl. krawy krawy krawy inst.pl. krawami krawami krawami loc.pl. krawaxb krawaxb krawaxb nom.sg. koye kd%b kd%b gen.sg. koya koya koya dat.sg. koyu koyu koyu acc.sg. koyb kd%b kd%b inst.sg. koyemb koyemb koyemb loc.sg. koyi koyi kud%i nom.pl. koyi koyi koyi gen.pl. koyb kd%b ko%b dat.pl. koyemb koyemt koyemb acc.pl. koyeN koyeN koyeN inst.pl. koyi koyi kud%i loc.pl. kdyixb koyixb kud%ixb nom.sg. pONtb pONtb pONtb gen.sg. pONti pONti pONti dat.sg. pONti pONti pONti acc.sg. pONtb pONtb pONtb inst.sg. pONtbmb pONtbmb pONtbmb loc.sg. pONti pONti pONti nom.pl. pONtbe poNtbe pONtbje gen.pl. pONtib pONtib pONti dat.pl. pONtbmb pONtbmb pONtbmb acc.pl. pONti pONti pONti inst.pl. pONtbmi pONtbmi pONtbmi loc.pl. pONtbXb pONtbXb pONtbXb Jezikoslovni zapiski 21 • 2015 • 2 _ 27 Slovene (8.0, 9.0, 10.0) nom.sg. wola wola wuola gen.sg. wdleN woleN wuoleN dat.sg. wolT woli wuoli acc.sg. woloN woloN WuoloN inst.sg. woleoN woloN WuČlČN loc.sg. wolT woli wuoli nom.pl. woleN woleN wuoleN gen.pl. WČlb wolb wolb dat.pl. wdlamb wolamb wudlamb acc.pl. woleN woleN wuoleN inst.pl. wdlamT wolamT wudlamT loc.pl. wolaxb wolaxb wuolaxb nom.sg. iyo iyo iyo gen.sg. iya iya iya dat.sg. iyu iyu iyu acc.sg. iyo iyo iyo inst.sg. iyomb iyomb iyomb loc.sg. iza ize ize nom.pl. iya iya iya gen.pl. iyb iyb iyb dat.pl. iyomb iyomb iyomb acc.pl. iya iya iya inst.pl. iyy iyy iyy loc.pl. izaxb izexb izexb nom.sg. dateN deteN deteN gen.sg. dateNte deteNte deteNte dat.sg. dateNtT deteNti deteNti acc.sg. dateN deteN deteN inst.sg. dateNtbmb deteNtbmb deteNtbmb loc.sg. dateNte deteNte deteNte nom.pl. dati deti deti gen.pl. datib detib deti dat.pl. datbmb detbmb detbmb acc.pl. dati deti deti inst.pl. datbmi detbmi detbmi loc.pl. datbxb detbxb detbxb 28 _ Frederik Kortlandt • The development of vowel length in Slavic In accent paradigm (a), e.g., krava, we have fixed stress throughout and loss of glottalization toward the end of the prehistoric period. In accent paradigm (b), e.g., konjb, pQtb, volja, detq, the accent shifted from the first to the second syllable (but not to a final jer) as a result of Dybo's law and was retracted in accordance with Stang's law in the loc.sg., inst.pl., and loc.pl. forms of konjb and in the majority of case forms of volja (and analogically in the other forms of this paradigm). In accent paradigm (c), e.g., igo and deti, there is a falling tone (which was eventually shortened) on the initial syllable in some forms and final stress (which was retracted from final jers) elsewhere. The accent pattern of mobile nouns (c) is identical with that of Lithuanian (3) except in the inst.sg. form of the aH-stems, where *-ojg was taken from the pronoun, and the nom.pl. form of the o-stems, where end-stressed Lith. -ai replaced the original neuter ending (cf. Kortlandt 1993). We can now identify the origin of long vowels in Slavic nominal case endings as follows. Proto-Indo-European lengthened grade vowels can be reconstructed for the loc.sg. endings *-ei and *-eu, which appear as long -T and -u after monophthongization. Long vowels from dialectal Indo-European contractions were shortened in gen.sg. -a < *-od and dat.sg. -u < *-oi and -e < *-ai, Lith. -o, -ui, -ai, which were never stressed, but length was preserved in inst.pl. -y < *-ois, Lith. -ais, where it is reflected as length (c) and by Stang's law (b) and Slovene neo-circumflex (a), for example, stzbri 'pillars', konji 'horses', Qkni 'windows', raki 'crabs', leti 'years'. The Early Slavic loss of glottalization in post-posttonic syllables yielded long vowels, which are reflected by the Slovene neo-circumflex in trisyllabic word forms where Dybo's law shifted the accent to the middle syllable, for example, osnQva < *opsnowa 'base', zabava < *zaba?wa 'amusement', inst.pl. zenami < *gena?mTs 'women', inst.du. zenama, nom.pl. telqta < *teleNta 'calves', fem.sg. nosila < *ndsi?la 'carried'. Length spread analogically in the neuter pl. ending, for example leta (a) 'years',polja (c) 'fields', also Slovak mesta 'cities', srdcia 'hearts', Cakavian and Posavian vrimena 'times', imena 'names', ramena 'shoulders', telesa 'bodies', and to a limited extent in other categories, for example, Slovene inst. gorami, gorama (c) 'mountains', kostmi, kostema (c) 'bones', and in a limited area also zelela 'wished', mislila 'thought', videla 'saw' beside regular zelela, mislila, videla (cf. Rigler 1970). New long vowels from the monophthongization of diphthongs were shortened in dat.sg. -i < *-ei, loc.sg. -e < *-oi, and nom.pl. -i < *-oi-s (cf. Kortlandt 2011: 128), which were never stressed, but preserved to a limited extent in gen. sg. *-T < *-e/ois, *-u < *-e/ous, loc.sg. *-T < *-ei, *-u < *-eu, and loc.pl. Slovene -eh < *-oisu. Long *-T is reflected by the neo-circumflex in the oblique form niti of nit (a) 'thread' and long *-u as Slovincian -u and in the Slovene locative by the neo-circumflex in orehu (a) 'nut' and the retraction according to Stang's law in kgnju (b) 'horse' (which has an analogical neo-circumflex). The long vowel of the loc.pl. ending is also reflected by the neo-circumflex in rakih (a) 'crabs' and the retraction according to Stang's law in konjih (b). In Kajkavian, the long vowel of Jezikoslovni zapiski 21 • 2015 • 2 _ 29 the loc.sg. ending is found in noči 'night',peči 'stove', kosti 'bone' (cf. Vermeer 1984: 380). The acute loc.sg. ending *-e < *-aHi is always short. While the nasal vowels of acc.sg. -g and nom.acc.pl. -q are always short because they were never stressed, the original distribution of long and short reflexes is preserved in gen.sg. Slovene krave (a) 'cow' (without the neo-circumflex) versus gorq (c) 'mountain' and Susak (Croatian) sestre (b) 'sister' versus vodie (c) 'water', and in inst.sg. Slovene kostjQ (c) 'bone', where the neo-circumflex of nitjo (a) 'thread' is due to the lost jer, not to the following nasal vowel. Van Wijk's law gave rise to new long vowels in endings in the paradigms of konjb and volja. These were subsequently shortened in the gen.sg. -a, dat.sg. -u, nom.pl. -i, and acc.pl. -q forms of konjb because these endings were never stressed, and analogically in inst.sg. -emb and dat.pl. -emt. Length was preserved in loc.sg. *-T, inst.pl. *-T, and loc.pl. *-Txt, which were later shortened in accordance with Stang's law. In the paradigm of volja there is no evidence for analogical shortening, which may or may not have taken place before the general phonetic shortening according to Stang's law. A new long nasal vowel developed from contraction in the posttonic inst.sg. ending -ojg; cf. Polish acc.sg. rybq (a) 'fish', inst.sg. rybq,, Slovene ribo, ribo (with the neo-circumflex reflecting a long ending). The long vowel of gen.pl. kosti (c) 'bones' < *-ib < *-bjb developed from the retraction of the accent from the final jer. References Dybo 1962 = Vladimir A. Dybo, O rekonstrukcii udarenija v praslavjanskom glagole, Voprosy slav- janskogo jazykoznanija 6 (1962), 3—27. Dybo 1968 = Vladimir A. Dybo, Akcentologija i slovoobrazovanie v slavjanskom, in: Slavjanskoe jazykoznanie: VImezdunarodnyjstezdslavistov, Praga, 1968, Moskva: Nauka, 1968, 148—224. Greenberg 2000 = Marc L. Greenberg, A historical phonology of the Slovene language, Wiesbaden: Harrassowitz, 2000. Kortlandt 1977 = Frederik Kortlandt, Historical laws of Baltic accentuation, Baltistica 13 (1977), no. 2, 319-330. Kortlandt 1985 = Frederik Kortlandt, Long vowels in Balto-Slavic, Baltistica 21 (1985), no. 2, 112-124. Kortlandt 1993 = Frederik Kortlandt, Tokie šalti rytai, Baltistica 28 (1993), no. 1, 45-48. Kortlandt 2009 = Frederik Kortlandt, Baltica & Balto-Slavica, Amsterdam: Rodopi, 2009. Kortlandt 2011 = Frederik Kortlandt, Selected writings on Slavic and general linguistics, Amsterdam: Rodopi, 2011. Rigler 1970 = Jakob Rigler, Akcentske variante, Slavistična Revija 18 (1970), 5-15. Stang 1957 = Christian S. Stang, Slavonic accentuation, Oslo: Universitetsforlaget, 1957. Vermeer 1984 = Willem Vermeer, On clarifying some points of Slavonic accentology: The quantity of the thematic vowel in the present tense and related issues, Folia Linguistica Historica 5 (1984), no. 2, 331-395. Vermeer 1992 = Willem Vermeer, In the beginning was the lengthened grade: On the continuity of Proto-Indo-European vowel quantity in Slavic, in: Rekonstruktion und relative Chronologie: Akten der VIII. Fachtagung der Indogermanischen Gesellschaft, Leiden, 1987, Innsbruck: Institut ffir Sprachwissenschaft, 1992, 115-136. 30 _ Frederik Kortlandt • The development of vowel length in Slavic pOVZETEK Razvoj vokalne dolžine v slovanščini Obstaja naglasna paradigma (a) s stalnim naglasom na akutiranem (glotaliziranem) vokalu, naglasna paradigma (b) z rastočim tonom (novim akutom) v nekaterih oblikah in kratkim naglasom na naslednjem zlogu v drugih in naglasna paradigma (c) s padajočim tonom (cirkumfleksom) na začetnem zlogu (ali proklitiki) v nekaterih oblikah in kratkim ali dolgim rastočim tonom na končnici (ali enklitiki) v drugih. Ko so končni jeri izgubili naglasljivost, je predhodni naglasljivi zlog dobil dolg rastoč ton. Ko je akut tik pred koncem predhistorične dobe izgubil glotalno naravo, se je zlil z rastočim tonom. Zavoljo izgube glotalizacije, skoraj splošnega krajšanja dolgih padajočih vokalov in široko razširjenega daljšanja kratkih rastočih vokalov se je ločevalni ton zamejil na slovenščino in srbohrvaščino, medtem ko je vokalna dolžina ostala ločevalna povsod še pozneje. V prednaglasnih zlogih so dolgi vokali izšli iz Dybojevega zakona, medtem ko so v naglašenih in ponaglasnih zlogih nadaljevanje praindoevropskih vokalov s podaljšano stopnjo in narečnih indoevropskih kontrakcij in so izšli iz slovanske monoftongizacije diftongov. Po nastanku novih barvnih razlik so nasledek Van Wijkovega zakona in kontrakcij v ponaglasnih zlogih, v naglasni paradigmi (c) umika naglasa s končnih jerov in daljšanja v enozložnicah, v naglasni paradigmi (b) Stangovega zakona, samo v ponaglas-nih zlogih pa preminitve laringalov.