Kapovic Adjective Accentuation

7/31/2019 Kapovic Adjective Accentuation

1/136

Mate KAPOVIUnivsity of Zagb

HISTORICAL DEVELOPMENT OF ADJECTIVE

ACCENTUATION IN CROATIAN (SUFFIXLESS,

*-n AND *-kADJECTIVES) 1

Introduction2

The article deals with the historical development of the accentuation of

sufxless (root), *-n and *-k adjectives. Their development is analyzedin detail from their Proto-Slavic origin to their modern reexes in tokavian,akavian and Kajkavian dialects.3 The dialectal data is taken frompreviously published dialectal descriptions, but also from the authors ownextensive and previously unpublished eld data mostly from Posavina4 andVrgorska krajina. As we shall see, the analysis of the historical developmentof adjectival accentuation can provide us with interesting insights that go

beyond the scope of adjectives. A careful historical study of adjectivalaccentuation reveals many interesting early changes and isoglosses in tok/ak/Kajk., helps us understand the modern dialectal forms, which are then

1 The name Coatian is used because the article deals mostly with data from dia-

lects spoken in Croatia. Of course, it goes without saying that the analysis of the devel-opments in tokavian is valid not only for the dialects of tokavian spoken in Croatiaor by ethnic Croats but also for those spoken in Bosnia and Herzegovina, Serbia and

Montenegro.2 I would like to thank Tijmen Pronk, Dijana urkovi and especially Mikhail Os-

lon for proofreading of the article. I also owe my gratitude to Misha Oslon for letting

me use his Juraj Kriani database (material from Krianis texts that does not havean explicit reference stems from this database) and for helping me with it.

3 The separate treatment of tokavian, akavian and Kajkavian should be regard-ed as provisory. The whole South Slavic territory is a prototypical example of a dialectcontinuum (cf. Ve r m e e r 1982; K a p o v i 2011d, 150152).

4The eld material from Posavina often provides insights that cannot be found

in the classical work of Ivis (1913), where the information needed for this kind ofstudies is usually insufcient and sometimes even misguiding.

BALTISTICA VII PrIedAS 2011 103128, 339448


2/136

104 Mate Kapovifar more useful from a comparative Slavic perspective and shed light on

some important prosodic changes that occurred in the history of tok/ak/Kajk., such as the One mora law (cf. Kapovi 2011b). The rest of the

topics related to adjectives (i.e. adjectives with other sufxes, comparativeand adverbs) will be analyzed in future works.Unlike nouns, where the accentual paradigm, if it changes at all, changes

mostly in individual words as a result of which the descriptions of the

history of noun accentuation deal mostly with paradigmatic accentuation,5when dealing with adjectives the accentual paradigm is often changed in a

whole class of adjectives or large groups of adjectives at once. That is why

it is possible for all (or most) of the adjectives with a certain sufx and a

specic number of syllables to shift entirely to another accentual paradigm,e.g. for -an adjectives to shift from a.p. A to a.p. C.6 Thus, the history ofadjectival accentuation deals more with such interparadigmatic shifts than

with intraparadigmatic changes, unlike the history of noun accentuationthat deals mostly with intraparadigmatic changes (i.e. the changes of theaccent in various cases in different accentual paradigms). This is alsodue to the fact that indenite adjectives in Slavic mostly have the samedeclension as nouns, while the accentuation of the denite forms is constant

in all cases, i.e. identical to the accentuation of the nominative. Of course,such paradigmatic shifts need not be marked in lists of adjectives.7 Anotherspecic trait of adjective accentuation is the importance of analyzing allor most of the available examples. That is the only way to interpret the

5 Cf. for instance Ka p o v i 2010.6 We use a, b, c for Proto-Slavic accentual paradigms (a.p.) and A, B, C for modern,

synchronic accentual paradigms. The semicolon (:) is used to mark the long variant of

the synchronic accentual paradigms (like a.p. B: or C:). Other abbreviations include:sg. (singular), pl. (plural); nom. (nominative), gen. (genitive), dat. (dative), acc. (ac-cusative), voc. (vocative), loc. (locative), instr. (instrumental) (also N/G/D/A/V/L/I.and n/g/d/a/v/l/i. for sg. and pl. respectively in paradigms); m. (masculine), f. (femi-nine), n. (neuter); adj. (adjective); def. (denite), indef. (indenite); dial. (dialect),Stand. (Standard); ct. (century); Croat. (Croatian), tok. (tokavian), ak. (akavian),Kajk. (Kajkavian), Lith. (Lithuanian), Slv. (Slovene), OCS (Old Church Slavic), PSl.(Proto-Slavic), BSl (Balto-Slavic). The mark+ is used for (dialectal) forms that are notdirectly attested in a given source but can be supposed as such according to the rules

given in the description.

7 If these were marked, almost all adjectives would be A/B/C, which would makethe list practically useless.


3/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 105development of the accentual pattern of a certain class of adjectives, since itis very common for the original pattern to be preserved in just one or a couple

of words (for instance ingolnaked, bos barefoot, bolan painful, tanak

thin,gorakbitter, obagood). Thus, it is useless just to analyze theaccentual paradigms in general without paying attention to what happened

to the adjectives that belonged to certain accentual patterns.

In some cases, it is very difcult to establish a common or originalaccentual paradigm (even if the Proto-Slavic reconstruction is certain) sothe grouping in the lists of adjectives should be regarded as provisional.

Some of the changes are described in more than one place in the article forinstance, if there is an a.p. a > a.p. C shift, this can be analyzed both in the

a.p. a or a.p. c section of the text. The text should be read as a whole sincemany parts are not repeated or are not repeated in detail for instance, ifsome feature is analyzed in tokavian, the same or a similar feature will notbe analyzed in akavian or it will be analyzed in less detail; or if the sameprocess occurs in different adjectival groups (e.g. both in root and *-nadjectives), it will be dealt with when describing the accentuation of therst group and just be mentioned briey elsewhere.

The sufxless adjectives1. a.p. a

Proto-Slavic

indenite adjectives denite adjectivesm f n m f n

* st clean *sta *sto * stj * staja *stoje

The a.p. a root (sufxless) adjectives had a constant acute on the stem in

Proto-Slavic, either on the rst (* st) or on any other syllable (*boga t).The short syllable equivalent was the short neo-acute (*gotv). The stressremains in the same position in denite adjectives.

Cf. in Russian:

pa pa po

In some adjectives in Russian, the root stress is preserved (a.p. A). Butmonosyllabic adjectives usually shift to a.p. C in short forms,8 cf.

8 Cf. St a n k i e w i c z 1993, 202.


4/136

106 Mate Kapovi o (but both and in the plural). The originalstress position is preserved in def. forms: a oe.

tokavian


st sta (sta) sto (sto) st(st) st (st) st (st)

The original a.p. a adjectives have a constant on the stem in the clas-sical (Danii-Vuk) literary tokavian,9 i.e. their reex is a.p. A. This isalso the situation in many tokavian dialects. However, the same kind ofxed root stress, i.e. the synchronic a.p. A, is seen in original short vowela.p. b adjectives such as nv new, the one exception being the adjectivegl10 naked, which remains in a.p. B.11,12 Here, we can see a tendencyof adjectives splitting into two classes based on the quantity of the root

(something similar exists in *-n adjectives as well, see below). Thus, wehave adjectives with a short root vowel (whether they are ultimately derivedfrom the original a.p. a acute or from the original a.p. b short neo-acute)in a.p. A (except forgoland bos), while adjectives with a long root vowelbelong to a.p. B: or a.p. C: (the B: vs. C: opposition is lost in many dialects).Thus in the East, a number of tokavian dialects show a shift a.p. C: > B:,while a number of Western dialects, mostly akavian, have the oppositechange in most adjectives a.p. B: > C:. In classical literary tokavian,one nds the system with all short vowel monosyllabic sufxless adjectivesin a.p. A and all long vowel monosyllabic sufxless adjectives in a.p. B:,thus st sta sto satiated/ nv nva nvo new : t ta to yellow/ drg drga drgo dear.13 The merger of the original

9 Cf. for instance 1925, 213.10 The formglin standard Croatian is secondary compared tog. The length is

due to the vocalization of the nal -l#(gl>go >g gl).11 Similarly, the semantically close adjective bs barefoot remains the only

monosyllabic short vowel a.p. C adjective in many dialects.12 Of course, all of this concerns the sufxless adjectives alone. Cf. the preserva-Of course, all of this concerns the sufxless adjectives alone. Cf. the preserva-

tion of a.p. B in adjectives with sufxes likeirko wide,zlno green, blo fat,bo good, tplo warm.

13 This was the case in classical, Vuk-Danii, Serbo-Croatian. In the contempo-This was the case in classical, Vuk-Danii, Serbo-Croatian. In the contempo-

rary Croatian norm, the situation is a little different, because the Western tokavian ar-chaic distinction of a.p. B: and C: (to : drgo) is getting back into the standard norm.


5/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 107a.p. a and short vowel a.p. b (st= nv)encompasses both tokavian andakavian, which is a major accentual innovation. However, this merger isnot certain in the case of Kajkavian the situation there is inconclusive (see

below).In later versions of literary tokavian, as well as in contemporary Stan-dard Croatian, one nds both the original a.p. A (i.e. a constant in all indef.forms) but also the newer a.p. B (sto / nvo).14 This sort of shift is commonto many tokavian dialects. In a.p. A, the short falling accent is constant inall forms, while in a.p. B all cases except nom. (and acc. sg. if the noun isnon-animate) have a (i.e. original end stress): nom. sg. stclean, gen. sg.sta / sta, dat. sg. stu / stu, instr. sg. stm / stm, etc.

For the preservation of a.p. A, cf. aptinovac (Ivi 1907, 141142)with a.p. A in adjectives such as slb slba slbo weak (also st,zdrv healthy these are the only adjectives given) and Osijek (Beni2007, 13, 2425): svj svja svje fresh (also st, gnjo rotten,slb, str old, st; nv, prst rude, spr slow, strg strict, smbrown15but pn full (B)). For a partial preservation of a.p. A, cf. forinstance Baka (Sekuli 2005): slb slba slbo (the same in pn,nv) but also zrla ripe, nom. pl. f. zrle, acc. sg. f. ru. For Posavian

adjectives, see below as well.The above-mentioned a.p. B is, all things considered, secondary in

comparison to the older a.p. A, thus st / nv A > B. The merger of orig-inal a.p. a adjectives like slb < *slab and original short vowel a.p. badjectives like l < *l bad is not a recent change, considering thata.p. A in adjectives like these appears in akavian as well (see below).It seems that the distinction of the original a.p. a and short vowel a.p. b

adjectives is not preserved anywhere in Croatian (except, in a very limitedway, in a few adverbs16 and in Krianis dialect, see below). Theoreticallyspeaking, a.p. B (sto =psto) in certain dialects could also be explained

14 Doublets of that type appear in the standard language since PHKJ (1960). Cf.also for instance M a t e i 1970, 164; S t a n k i e w i c z 1993, 126, etc.

15 Beni says that the formssta instead of the more frequentstra andsta in-stead of the much more frequentsta appear sporadically as well, but that such variantsare almost non-existent in other examples.

16 Cf. mlo few (andzmalo almost), rno early (andzana) from the origi-nal a.p. a,butskro recently; almost (the original a.p. b).


6/136

108 Mate Kapovias the result of the old merger of a.p. a and short vowel a.p. b adjectives

not in a.p. A but in a.p. B. That is to say that in some dialects the mergerof original sto / nov went in the direction ofsto / nvo (A) while in

others it went in the direction ofsto / nvo (B). However, this option doesnot seem very likely since the progress of the secondary shift A > B can beclearly observed in some dialects (cf. the Posavian data below).

The mix of the paradigmatic reexes of the original a.p. a and shortvowel a.p. b is most easily explained on the basis of the same form innom. sg. m. (st= nv), since both the old acute (* st) and the old shortneo-circumex (*nv) yield the same result. By analogy to these forms,the same accent can be generalized in other forms as well. However, it is

unusual for an analogy of this kind, i.e. short vowel a.p. b > a.p. A shift, tohave occured so early. In addition, it seems that the root xation of stressin the old short vowel a.p. b cannot be a result of the equation st= nv,considering the fact that the merger was already present in Krianislanguage, where the reexes of the old acute and the short neo-acute hadnot yet merged completely (see below).

Apart from the already mentioned shift a.p. A > B, there is another

possible shift, a.p. A > C', which occurs, for instance, in Western tokavianDalmatian dialects, cf. ststa sto there.17 In those dialects, all of theoriginal a.p. A adjectives (originally a.p. a and short vowel a.p. b adjectivesexcept forgol) shifted to a.p. C' with just a few exceptions (the common onebeing, from Imotski to Vrgorac and Neretva, the adjectivepn, which shiftsto a.p. B and not a.p. C').

imundi (1971, 126127), as usual, has alternative forms botha.p. A and C' for most adjectives such as stand nv but also a.p. B forpn

and a.p. A or B fortj quiet (and lklight, which is originally an *-kadjective).

In Prapatnice in Vrgorska krajina (my data), one nds the secondarya.p. C' with the pattern dg long (gen. sg. dga) f. ga n. dgo pl. m. dgi def. dg. The same pattern is seen in the adjectives: st, ktbrittle, f. mla n. mlo nice (def. ml), mkglum,st,slb, strold,

17 A.p. C' means that there is no lengthening in nom. sg. m. (as in bs), which is acharacteristic of the original a.p. C.


7/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 109stm steep,ktstingy (gen. sg. m.kta) f.krta (gen. sg. f.krt, acc.sg. f.ktu),zdrv healthy, as well as in: l la lo, lk, nv,prstrude, sprslow. The exceptions are tj tja tjo quiet (due to the

disappearance of -h-, the lengthening in front of -j#and analogy in otherforms) and the adjectives pn, vrja boiling and zrja ripe that shift toa.p. B:pn (gen. sg. m.pna) f.pna (gen. sg. f. pn) n.pno (also afrequent adverbpno) pl.pni.

In the Ijekavian tokavian dialect of Neretvanska krajina (Vidovi2007, 209), the adjectives zro, dg, l, strgbelong to a.p. C', pn toa.p. B, while bsbelongs to a.p. B' (bs f. bsa n. bso with the lengthin nom. sg. m. as a remnant of the original a.p. C).18

The shift to a.p. C' is seen in Dubrovnik as well, which is thus inconcordance with the Dalmatian Neo-tokavian dialects to the north of it.Cf. Dubrovniknv nva nvo, as well as st, ug, plah timid, tih(Budmani 1883, 172).

In Molise Croatian (Piccoli, Sammartino 2000), an emigranttokavian dialect in Italy (with its origins in the hinterland of the townof Makarska and near the river Cetina, which is dialectologically close to

the dialects just mentioned), one nds the following preserved a.p. A ina group of adjectives: st, -a, -o (the same pattern in nv, st,str,zdrv),while the other group shifts to a.p. C': dg dga19 dgo (the same patternin mk soft, pn, zrj, k bitter).20 This clearly represents the olderphase of a.p. A > C' shift in this area. At the time of their departure, theshift had obviously already taken place, but it was clearly still a change inprogress the process was not complete, as is the case in the dialects of thearea today.

In the Old tokavian Slavonian dialect, the situation is different. As wehave already seen, the original a.p. A is preserved in aptinovac. Ivi

18 Cf. bs bsa bso in Imotska ( i m u n d i 1971, 131) and Vrgorska Krajina(my data).

19 From the older *dug, since the dialect has kanovako lengthening (i.e. in-stead of `).

20 The formzrj seems to be a regular reex of the older *zrl (cf. also variants

gland gj and cj from *cl whole) even though the development is not quiteclear. The adjectives mkandkare not *-k adjectives originally (see below).


8/136

110 Mate Kapovi(1913 2, 4243) claims that the shift of a.p. A to a.p. C' is general in Posa- vina, as in the already discussed dialects of Dalmatinska Zagora (Imotskiand Vrgorac), cf. nom. sg. f. sta, la, nov nom. sg. n. lo acc. sg. f.

istu vdu into clean water),21

while Bao t i (1979, 196) notes a splitof the original a.p. A adjectives in Kostr in the Bosnian part of Posavinato a.p. A (such as o hale, dg,pn, trm inert,strg, etc.) and a.p. C'(such as st, tst fat, l, prst, etc.). In Southern Baranja22 Sekere(1977, 388) gives the pattern str stra stro (also prv straight),while mlnice and ltake a.p. B.

That Ivis claims of a general a.p. C' pattern in Posavina in these kindsof adjectives are wrong, is conrmed not only by the already mentioned

later studies but also by my own eld material from Posavina. Here, weshall present the data from seven Posavian villages in all of them, theoriginal a.p. A is preserved in many adjectives and in some of the dialects

in most of them. The most archaic dialects are those of Budrovci, BrodskiStupnik, (Slavonski) Koba and Sikerevci. In Budrovci, onlyspr(spa spro) shifts to a.p. C, with a.p. A perfectly preserved in all other adjectives,cf.slb slba slbo def.slb(thus also l la, st, trm trma,

l, mk, nv nva,pav,prst,pn pna,s t,sm,str stra,stm stma,strg, th, trl trla, vjt, vrl vrla,zdrv zdrva,zrl zrla).

In Brodski Stupnik, only stshifts to a.p. B (sta sto), while all otheradjectives (l, l, ml, mk, nv, prv, prst, pn, ra red haired, st,slb,sm,spr,str, stm,strg, th, trm, trlrotten, vrl,zdrv,zrl)preserve the original a.p. A. In Koba, most adjectives preserve the originala.p. A (ml, nv, prst(def. C), pn, st, slb, sm, spr, str, strg, th,zdrv), while only two adjectives (stand ttvain) shift to a.p. C'. In

Sikerevci, almost all adjectives remain in a.p. A (thusprst,pn, r, rmlame, slb,sm,st, st(and C),strg, th, trl, zdrv,zr l), while standspshift to a.p. C/C' (the adj.kthas a.p. C: pattern). In Babina Greda,a.p. A is preserved bypn (also C'), rdglad, r,slb,spr(also C),str,strg, th, zdrv, while nv shifts to the mixed a.p. A-B and st,prst,st

21 Ivi notes the monosyllabic adjectival a.p. B only in lak lak easy but thisis, as already mentioned, originally an *-k adjective.

22 This is actually the northern part of the present day Croatian Baranja, sinceNorthern Baranja is today a part of southern Hungary.


9/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 111to a.p. C'. In Slobodnica, the original a.p. A is preserved in l(la lo),pn,snn,str; mk,prv,st,slb,stm,strg, the shift to a.p. B is seenin nv nva nvo (and spr, vrl,zrl) and the innovative a.p. C(') in

st,prst, vjt; trl,zdrv.23

In Orubica, the situation is the most complexand innovative. There, a.p. A is preserved by pn,slb,sm,spr,strg,trl, zrl, while the other adjectives (st, nv, prv, prst, st, str, tt,th, zdrv,kbitter) shift to all kinds of combinations and mixes of theoriginal a.p. A and the newer a.p. B and C' with a great deal of vacillationand alternative forms. It is interesting that the adj. st loses the originala.p. A in all dialects except for aptinovac and Budrovci.

The shift to a.p. C', in spite of the preservation and further spread of the

original a.p. C in many Western tokavian (and akavian) dialects, mayseem unmotivated, since there is only one old short vowel a.p. c adjectivein Croatian: bs bsa bso and, even there, nom. sg. m. of a.p. C isdifferent from nom. sg. m. of a.p. A (bs st). Of course, adjectives likegldan of a.p. C type are much more numerous. The distinction in nom.sg. m. forms remains even when innovative forms such as sta appear.The situation in Neretvanska krajina, where adjectives like dg have

a.p. C' pattern, but bsbelongs to a.p. B' (see above), shows that the onlymonosyllabic a.p. C adjective (bs) did not take part in the a.p. A > C' shift.It must have been a part of a wider tendency of generalization of mobile

accentuation in adjectives.

This process could perhaps be explained as a result of the generalhegemony of the a.p. C type in adjectives in the dialects under discussion,i.e. considering -an and -akadjectives, etc. as well. It is also possible thatwe are not dealing with the shift of a.p. A > a.p. C' but instead with a more

complex shift of a.p. A > B > C', i.e. the pattern st sta sto (C')would be a result of the change from olderst sta sto (B), whichwould, in turn, be from olderst sta sto (A). This possible processof A > B > C', however, does not completely explain the appearance of thenew a.p. C type here, in spite of the bs st relation. In addition, it isperhaps too complicated to assume two phases of innovation (A > B > C)

23 The adj. ml shifted to a.p. B: (mla mlo) and lhas a vacillation, i.e.a.p. B/C (l la lo / lo).


10/136

112 Mate Kapoviinstead ofone, since these are in other aspects rather archaic dialects. Itseems that the lengthening in nom. sg. m., typical for the original a.p. C,was not considered an obstacle for the new accentual mobility to develop in

adjectives that were previously immobile.In most cases, the old accent of denite adjectives is preserved (e.g.ml, st,slb, smjlcourageous, zdrv, trl, st), but in some cases,the original form (st) can have a variant or can be completely replaced bythe secondary a.p. C type accent (st).24 Cf. also dglong (beside dg).It is interesting that the original a.p. C accent is lost in many dialects(thus one has bs, drgdear instead of the olderbs, drg) so in somedialects there can be a peculiar situation that the type exists only in orig-

inal a.p. a adjectives such as dg. The old short vowel a.p. b adjectivesmostly retain the original root accent in long forms (prst, nv,spr,l, etc.). Accentual type like stis not a result of the inuence of formslike bs / drg(original a.p. C) in all dialects. Since this accentual typeis lost in many dialects, that kind of inuence would be very unlikely insome of them. The spread of the new type in a.p. A can also be due toanalogy to younger (A > B) indenite forms like sta, sto or to sta, sto(A > C'). For instance, in the Posavian dialects of Koba and Sikerevci that

preserve the original a.p. A in nearly all adjectives, it is very indicative thatthe type is found only in the denite form ofst, which is practically theonly root adjective that has shifted to a.p. C' in those dialects.25 Of course,the rise of the innovative type in the original a.p. A may differ in variousdialects.

Ivi (1913 2, 49) notes the original type gl, nv but also theinnovative type nov, as well as stand ist (like bs / bs)for Posavina.For aptinovac Ivi (1907, 142) gives the preserved a.p. A (st sta

sto) but also a generalized innovative type of the def. ist ,slab ,zdrav .According to Ivi, the shift in def. adjectives occurred in all cases inaptinovac and not just in some of them, as is the situation elsewhere. Sincea.p. A is preserved in indef. adjectives in aptinovac, the accent type ofistshould probably be attributed to analogy to the original a.p. C forms suv

24 Cf. alsopvpvrst.25 In Slobodnica, most of the adjectives remain in a.p. A but a considerable num-In Slobodnica, most of the adjectives remain in a.p. A but a considerable num-

ber shifts to a.p. B and C. Still, only the form sthas such an innovative accent. Allother examples have the original accent, cf. l,stm,prst,snn,spr, etc.


11/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 113dry andutmad although it is not clear why -would be generalized inall a.p. A adjectives, at least according to Ivis not-so-explicit description,while it is attested in only two of the original a.p. c reexes.26As for Kostr

in Bosanska Posavina, Baoti (1979, 196197) does not explicitly saywhat kind of accent appears in a.p. A denite forms, but it seems that theytend to keep the stem accent. As for original a.p. a and short vowel a.p. badjectives that have shifted to a.p. C', their def. forms have either nalaccentuation only (i.e. - , -, -)or nal accentuation alongside the originalvariant (i.e. stem stress). The shift to the a.p. C def. stress pattern occurs inmost adjectives in Neretvanska krajina as well.27 imundi (1971, 137)has both older forms like st and younger like st but claims that the

latter are more frequent. In Prapatnice (Vrgorska krajina), like in Dobranjeand Vidonje, it seems that all or most adjectives behave like st,st, etc.except ml, nv, which preserve the original pattern. In Prapatnice, formslike st appear beside the new a.p. C' (st sta sto) so one cansuppose analogy of the indef. sta to the def. st, but original a.p. C formssuch as bs, bz,gst, etc. could also be an important source of analogy,since they are well preserved in the dialect. The innovative type also

appears in a.p. B, cf. the def. gl with the indef.gla.Bu dm an i (1883,173) notes stfor Dubrovnik and the same pattern for all other adjectiveswith in at least some indef. forms (except fornov similar to Prapatniceand Dobranje / Vidonje), cf. also dg(Ree t a r 1900, 129). In Dubrovnik,

26 The innovative -also appears in all adjectives with sufxes that belong to thesynchronic a.p. C mokrwet, toplwarm,sretnhappy, tankthin. The only ex-ample of those in which one would historically expect desinential stress in the denite

form is tank. This could be a case of analogy to examples like the latter, but there istoo little data to be certain. Synchronically, the accent of the def. f. mokr can hardlybe due to analogy to the indef. f. mkra. Such an analogy works perfectly in Neo-tokavian where the indef. mkra can easily inuence the def. mkr the result beingthe newer def. form mkr. However, in Old tokavian, with no stress retraction (andwith kanovako lengthening in aptinovac as well), such an analogy does not work.One could perhaps suppose the analogy of the def. form mokr to the old *mokr (the

proto-form of todays mkra).27 For the villages of Dobranje and Vidonje Vi d o v i (2007, 203) gives only th

(cf. also the oldgst, etc.in a.p. C) but such accentuation is present in other examples

as well (st, dg, l,prst, st,zdrv, etc.), except fornvand ml, which pre-serve the old accent (Domagoj Vidovi, private communication).


12/136

114 Mate Kapovithe A > C' shift in def. adjectives may have to do with the same sort ofpattern shift in the indef. forms but not necessarily in some dialects, formslike dg appear alongside dg dga dgo (B). In Molise (Piccoli,

Sammartino 2000), one nds older (def. nvi, pni, zdrvi, zrli, ki)but also younger forms (dg, gen. sg. m. dgga but f. dga, gen. sg. f.dge), as well as peculiar forms like the def. sti, -a, mki, -a soft (gen.sg. m. mkoga but gen. sg. f. mke).28

In four original a.p. a adjectives an unusual a.p. B: denite pattern isfound. Cf. the indef. forms prv, rn early, str, mo small (mla mlo, ARj) and the adverbsprvo right, rno early, mlo little but thedef. forms prv, rn,29ml,30str, which derive from the older forms

prv, rn, ml,str, attested with the neo-acute in Posavina (Iv i 19132, 50) and in other tokavian dialects that preserve the neo-acute (like inPoljica near Omi). These adjectives have neo-acute in akavian as well(see below). Thus, this is not a case of the old neo-circumex that appearsin Kajkavian or in North akavian (ml , etc.), as some wrongly think.It should also be emphasized that all tokavian (and South and Centralakavian) dialects seem to have such stress, which clearly points to anold innovation. Some tokavian dialects distinguish the forms prv and

prv / prv,31 while the secondary type B: can spread to other adjectivesas well, cf. zdrv / zdrvandslb / slb in Imotska krajina and Bekija( imundi 1971, 137).32 There is also the younger form ksnlate that

28 The forms sti, mki could be derived from *ist, *mek but the supposedshortening of the nal syllable is unexpected.

29 In some dialects, the variant with the expected exists as well (for instance inOsijek, cf. Be n i 2007, 24). This can either be original or a back-formation by anal-

ogy to the indef. forms.30 The length can be transferred to the indef. form as well, cf. ml. However, inmany dialects only the def. form mlexists, while in others, like Prapatnice, the def.and indef. forms are distinguished by accent alone cf. the indef. ml(but the def.ml), mla, mlo (a.p. B:).

31 Cf. in B u d m a n i 1867prv(and secondaryprv) dritto butprvvero, inKostr (B a o t i 1979, 196)prvright andpa vsuitable, as well asprvin Slo-

bodnica in Posavina (my data). Cf. also By 1940, 284 for Piva and Drobnjaci,and 1964, 279 for Srijem.

32 In Molise, in the example mli a usual, typically Molisean, shortening occurs in

gen. sg. m. form mloga but not instri gen. sg. m.stroga. Cf. also below for otherexamples of such a shortening in Molise.


13/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 115exists in some dialects beside the older ksn, which is probably due toanalogy with rnearly.

Although it is quite clear that in these cases we are dealing with

secondary forms and an analogy to the a.p. B: type, the reason for such ananalogy specically in these adjectives is not too clear. Perhaps it is notaccidental that all primary examples have a resonant (-v-, -n-, -l-, --) as thenal consonant of the stem. However, cf. alsozdrv(but dialectallyzdrvas well),pnand mlwith no lengthening. One may also surmise that thesame unusual lengthening before -v- is found in the formpvrst insteadofpv, although in this case this is just a variant that exists in some (but notall) tokavian dialects.

In compound adjectives with a linking -o- and zero sufx, there is atendency (relatively young) for the accent to shift from the middle -- (whichis the original place of stress in most cases) to the very rst syllable, whileleaving the syllable after -o- long. Cf. golglav bareheaded > gloglv,bosnogbarefoot > bsong. The same kind of change occurs in nounsthat are derived in the same manner (although the younger accent is morefrequent in nouns than in adjectives): sup dish-washer > sdopr,

kamenlom quarry > kmenolm, etc.akavian (Orbanii)33


s sta sto sti sta

In akavian, archaic dialects maintain a.p. A (< Proto-Slavic a.p. a andshort vowel a.p. b), while others show a shift (partial or general) to a.p. C'.There is no shift to a.p. B, as in some tokavian dialects, since a.p. C isdominant in akavian (cf. also the shift B > C below). Such a situation inthe akavian South is a continuation of a similar situation in DalmatianWest tokavian dialects. One other thing that distinguishes akavian fromtokavian is that in akavian a.p. A (or what becomes of it) the def. adj.type -accent is less frequent than in tokavian. This is hardly strange, sincethis type of accent is much less frequent in akavian in a.p. C as well, where

33 K a l s b e e k 1998, 429.


14/136

116 Mate Kapoviit would be expected historically, so it is no wonder that it could not haveinuenced a.p. A in that regard.

There are two basic types of developments in akavian. One group

of dialects preserves the original a.p. A, while the other shows partial orcomplete transfer of monosyllabic adjectives to a.p. C'. Denite forms usuallyhave the original stem stress.34 The rst, archaic, group is represented, forexample, by Hvar35st sta (and ist, there is no mention of such avariant for other adjectives36) sto, constant stem stress also in dg,st,th,slb (slba),37 l, nv, trm, def. sti, etc.;38 Filipjakov39st sta sto (the same inslb,st,prst,spr, while the length from nom. sg. m. isgeneralized in all forms in nv nva nvo); Preko40st sta sto

(the same inst, nv,prst,spr, whileslb slba slbo has generalizedthe lengthening from the f. and n.);Pag41brja f. with young (of animals),dg dga dgo def. dgi,lob lba lbo def.lbi weak,nuv nva nvo def. nvi,k ka ko def.ki bitter, etc.;42Susak43dl dla dlo def. dli long (nf, pn, slp, sy t, t,zdrfalso belong to a.p. A);Senj44lla lo (the same pattern inst,st,slb,pun, zrl,prst, nv), def. li etc; Orlec45st sta sto,

st,slp,st; Rijeka46

dh long, pn, st,slb,str, th,zdrv (constant34 Of course, there are exceptions. Cf. nov in Novi Vinodolski (the very name of

the town is alsoNov).35 H r a s t e 1935, 32; DL.36 DL gives only sta for the dialect of Brusje on the island of Hvar.37 In DL, the data given for Hvar in general isslb (cf.slb in Bra), -a, -o (i.e.

+slba, +slbo if not a mistake).38 Cf. Pitve (Hvar, my data):slb sla ba sla bo,st sta sto and Vrboska on

Hvar (M a t k o v i 2004): st, -a, -o, pn pna, -o, st, -a, -o, slb slba slbo,zdrv zdrva, -o, zrlzrla, -o butstr str stro old with a shift to a.p. C(because of the lengthening instrand the def. formsti).

39Near Zadar, data by Nikola Vuleti.40 On the island of Ugljan, data by Nikola Vuleti.41 K u s t i 2002, 62.42 The nal closed syllable is lengthened on Pag, cf. also the nouns botbrother,

naudpeople (Kusti 2002, 5051).43 H a m m, H r a s t e, G u b e r i n a 1956, 113114.44 M o g u 1966, 76.

45 H o u t z a g e r s 1985, 117119.46 S t r o h a l 1895, 158.


15/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 117root stress) and zrl +zrla, nv +nva; Orbanii47zdrf zdrva zdrvo, nuofnva,pu n pna,slp slba,st sta, str stra, etc.In Grobnik,48 the original pattern is preserved in st(-a, -o, also dg, rd

glad, slb,stm steep),str stra stro (also ml, pn, trlrotten,vrlboiling, zdrv,zrlripe), but there is a shift in st ist sto(alsokt, t slim, empty).49 The original a.p. A occurs, as we have seen,in all of the akavian territory, from the South to the North.

On Vrgada (Jurii 1966, 8283; 1973) there is a transitional systemwith a group of adjectives preserving the original pattern (hrm, -a, -o,the same in l, th,prv, all a.p. A), a group with variant stress (zov zdrav / zdrva, k ka / uk ko, a.p. A/C') and a group with

complete shift to a.p. C' (st ist, acc. sg. f. stu sto, the samein dg, slb, st, stm, t and pu n, st). In denite forms, the originalpattern can be preserved (ml, nv), but there are also innovative formswith desinential stress (ist , dug ). On Bra (imunovi 2009, 44), asopposed to the more archaic Hvar, there is a vacillation of a.p. A/C' in theadjectives dg, th, st,50 trm, prst (the last adjective belongs to a.p. Aonly in the dictionary), while the other group of adjectives has the youngera.p. C' only: nv nov nvo, the same inzdrv zdrav zdrvo, vr l,

zdr l ripe, pn, slb (cf. the long forms: dgi, nvi, thi, trmi, prsti,zvi, vrli).51

For the younger - in the denite forms of adjectives in akaviana.p. A, cf. for instance Novi Vinodolski ist(next to st), Krasicaslab(Langs ton 2006, 184185).

As in tokavian, South akavian also shows the unexpected in a.p. Aof some adjectives, cf. Vrgada (Jurii 1973) mol, pov, stor, onand Bra ( imunov i 2009) mli,pvi,sti, ni (butzvi < *zdrv),

which is in complete agreement with tokavian.47 K a l s b e e k 1998, 143144.48 L u k e i , Z u b i 2007, 95, 100101, 105.49 As for the reex of the old short vowel a.p. b in Grobnik, cf. nv nov nvo

(C) andprst prost prsto (C) usual, simple as well asprst prost prsto(C' < *A) rude in a.p. C but trm trma trmo inert (A).

50 In the dictionary, the accentsta is noted as stemming from Bol andsit as fromDraevica.

51 For Puia on Bra (Domagoj Vidovi, p.c.) cf. dgin a.p. A/C' (dg / dg

dga / dug dgo) and only a.p. C' elsewhere (pn pun pno,prst prost prsto, etc.). Def. adj. have root stress (dgi,sti, trmi, nvi, etc.).


16/136

118 Mate KapoviIn Senj (Mogu 1966, 76), such an accent appears in several other

adjectives as well: nvi, pvi, zvi. In the North of the akavianterritory, these adjectives have neo-circumex (which appears in all a.p. A

adjectives in Kajkavian, Slovene and some North akavian dialects); cf.Novi Vinodolski (Langs ton 2006, 184): strbut dg, ml, st, slb,etc.,Orbanii (Ka l sbeek 1998, 144) mli,prvi, stri, rni (and the indef.rnby analogy) but not in other adjectives: dgi, mli, thi, etc. The Orbaniiform ksan late has the circumex probably by analogy to the older def.form *ksni (the present day def. form is the secondary ksni, shaped byanalogy to the new indef. form ksan). In Orlec (Ho utzager s 1985, 118),only two adjectives, which have secondary neo-acute in the South, have

neo-circumex: mli andprvi (cf.sti < *str, ni < *rn).52 As alreadypointed out regarding tokavian, the reason for the occurrence of such asecondary accent in these specic adjectives is not clear perhaps it is dueonly to their frequency, which was the reason for the (slightly paradoxical)analogical a.p. B: accent in the South, while in the North the old neo-circumex was preserved in these same adjectives (the neo-circumexmight have been the feature of all a.p. A adjectives originally as is still the

case in some dialects).In other dialects of the akavian North, the neo-circumex can be regularin all a.p. A adjective denite forms: cf. Kastav (Beli 1914, 19, 2829):dg, ml,pn,st,slb, th,zdrv, the same in polysyllabic adjectives:bogtrich, rjvrusty, etc. Cf. also Grobnik (Luke i, Zu b i 2007,95, 100101, 105): dg, str, vrl, ml, gotv, bogt (also in deniteforms of bradt beardy, glavt with a big head, kosmt hairy, otobvious, kamentstony, etc.), korsnuseful,pobnpious (the same

in the def. form siromn poor) but nonetheless +st, +slb, +velk,+ot, +ln, +srn, st(the indef. form being stC'), bstrclear, etc.

It is not certain what the original situation was like concerning the neo-circumex in denite a.p. A adjective forms in the North of the akavianterritory or if there was any unity there to begin with. Some dialects,like Kastav, have the length in all a.p. A denite forms, like Kajkavian

52 These two forms could also theoretically derive from the older *str and*rn.


17/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 119and Slovene, while others, like the dialect of Novi Vinodolski, have it intraces only. One possibility is that originally all (or most) of the Northernakavian dialects had the neo-circumex in a.p. A def. adj. forms a fea-

ture that was later lost in many dialects by analogy to indef. forms. The otherpossibility is that in some dialects the neo-circumex was never presentin all adjectives. The difference between akavian North and Slovene /Kajkavian is that akavian has the neo-circumex while at the same timepreserving posttonic length (of course, in dialects that preserve it in othercases as well). Here, there are once more two possibilities. One is that thisposttonic length was never lost during the process of the neo-circumexlengthening and the other is that it was lost but then restored by analogy to

a.p. B and a.p. C, where the preservation would be expected in any case.

kajkavian(Velika Rakovica)5354indenite adjectives denite adjectivesm f n m f n

st sta sto siti (Bednja)54

The reexes of the Proto-Slavic a.p. a in Kajkavian are clear. The reex

is a synchronic a.p. A in all dialects, which means the constant root stressin indef. forms and, typical for Kajkavian and Slovene, the neo-circumexin def. forms. Thus in most archaic Kajkavian dialects, the indef. formsare st sta sto and the def. ones are sti sta sto. The neo-circumex is attributed to the old contraction length in the ending, cf.tok. st sta st from PSl. * stj > * styj > *st, etc. Insome less archaic Kajk. dialects, the neo-circumex may be replaced byan analogical from the indef. forms, i.e. def. sti by analogy to the indef.st. Other changes (like the spread of the a.p. B: type in V. Rakovica) areless frequent.

Cf. in Bednja (Jedvaj 1956, 304305) st55 def. isti, the same inslb, st, str, zdrv, dg, ml, ml little. In polysyllabic adjectives, the

53 M a r c h 1981, 265.54 J e d v a j 1956, 305. The V. Rakovica formssti sta sto are secondary.

55 In Bednja, only nom. sg. m. exists as a separate indef. form. In all other cases,the def. forms are used (Je d v a j 1956, 303).


18/136

120 Mate Kapovineo-circumex appears in internal syllables, e.g. insrdtsediti angry,ksmt kesmoti hairy but not in rst syllable: hrpov(i) coarse,pov(i) lame, etc.56

In Velika Rakovica (March 1981, 265266):st sta sto,the sameinpun, mal, ug, mil,pav,sta,zav, tul,zland nov. The def. formshave an innovative neo-acute (def.sti, etc.), taken from a.p. B:. The samesituation is found in the polysyllabic adjectives: bogt bogta bogto,the same pattern ingotv nished,strplv patient, etc., cf. the secondarydef. forms bogti, etc. For the reexes of the old short a.p. b see below.

In Varadin (Lip l j in 2002), cf.zdrf(f. and n. formszdrva and zdrvoare irrelevant due to the retraction that occurs in the dialect, cf. nga leg,

vda water, e na woman), indef. nom. pl. f. zr l ripe def. nom.sg. m. zr li and gologlf with no hat def. gologlvi but also br j br ja br jo,slp slba slbo def.slbi (the same inst). In the lastthree adjectives, one sees curious cases of levelling, with secondary in theindef. nom. sg. m. due to the old def. forms, while the def. forms have thesecondary from the old indef. forms. The neo-circumex is well preservedin the adjectives that have def. forms only: sti, thi, sti, m li (adverb

mlo),prvi, rni, vr li, d gi, mli, etc.Roi (18931894 2, 141144, 147148, 152, 166) asserts that inPrigorje, unlike tok., it is impossible to distinguish def. and indef. formsby accent alone (i.e. that there is no pattern of indef.zdrv def.zdrvi). Inhis description of Prigorje, one nds the pattern:zdrv za va za vo(in oblique cases, older forms likezdrvega go together with newer formslike zavga).57 The following adjectives have this pattern as well: slb,st, dg,pn,st,str(but f.sta ra / stra here, the trace of the old neo-

circumex from the def. forms is preserved), tj quiet,zrl, brj. Cf. alsoin polysyllabic adjectives: bgat f. boga ta n. boga ti (-i instead of -o),sdit f. srdta. In some adjectives, the old from the old def. forms appears:rn f. rna n. rni (instead of-o), ml f. m la n. m lo.58 Here, the f.and n. forms were originally def. forms. Such a situation is found inpjan

56 In this position, the neo-circumex is probably not to be expected in any case.57 Rois accentual marks are adapted in this article. The mark is used for his

and for his .58 The form mlhas the secondary forms mla, mlo in f. and n.


19/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 121f.pija na / pijna n.pija ni / pijni (instead of -o) drunk next to the oldindef. forms, as well as in ksmat,gzdav gaudy,piv crummy. The oldneo-circumex has been generalized in all forms in kamen t f. kamen ta

n. kamen ti and is found in denominative adjectives liketni summer,krni bread, etc. that have def. forms only.In Ozalj (Teak 1981, 268, 270), Proto-Slavic a.p. a adjectives form

three groups now. The rst one preserves the original pattern: ml mla mlo def. mli. The other adjectives are dklong, ml,prfreal,str,zdrf, zrl and polysyllabic dugovrt long neck, rlenkst reddish,gologlf, etc. In the other group, the def. forms have the innovative fromthe indef. forms:st sta sto def.sti and the same inpn, trl, vrl,

vland brdovthilly, kastyellowish, mafskinny, etc. The third,smallest group, has transferred to a.p. C' and has the same pattern as gl gla (< *gol) glo naked. The adjectives areslp weak,plh and st(the latter preserves the original def. accent in the phrase sta srda AshWednesday). Def. forms in Ozalj have , but this can be either the originala.p. a neo-circumex or a reex of the old neo-acute from a.p. B: (thispattern spreads to a.p. C as well).

In Turopolje (ojat 1982, 400), def. forms lose the neo-circumex andthe pattern is dk d gi long (the same inzdrf, st). The adjectivesstrandslp weak can shift to a.p. C(str,slp), probably due to analogy tothe old def. forms that had the neo-circumex. An interesting developmentis seen in the secondary indef. form mli, which has the old def. accent,while the new def. form has the secondary neo-acute: mli (cf. in Prapatnicetok. indef. mlbut def. ml).

As we have seen, in tok/ak. the old short vowel a.p. b has the same

reex as the old a.p. a. However, the situation with the old short vowela.p. b is not as clear in Kajkavian, one of the reasons being the lack of data. In V. Rakovica, only three adjectives are attested. The adj. nov is a.p. A (likest, etc.), whilepostand loare a.p. C', next togl gol glo (originallyalso a.p. b). In Bednja, Jedvaj gives only the adj. ny v (def. ny vi) new,which is in a.p. C (next to a whole slew of the original a.p. b adjectives suchasutyellow andgy lnaked). In Turopolje, the one form nv no va(cf. also bs bosa / bsa, originally a.p. c) tells us nothing, just like Roistwo adjectives from Prigorje that belong to a.p. C: nv nva nvo (the


20/136

122 Mate Kapovisame inprst). In Teaks Ozalj, nfnew belongs to a.p. A together withml(def. mli); ljbad andsprare in a.p. C' (likegl gla glodef.gli) andprstis in a.p. C: (having the same pattern as lp lpa lpo

beautiful). In Varadin (Lip l j in 2002), one nds only the def. li (whichseems to be an innovative a.p. A form) and nvi, li (which do not point toa.p. A). According to what we have seen, it is not clear whether the processof merger of the old a.p. a and short vowel a.p. b, which is visible in tok/ak. (with the exception of *gol that remains in a.p. B), can be attributedto Kajkavian as well. In Slovene, there is no such merger (see below).However, the situation in Krianis dialect points to a process similar tothat in tok/ak. (setting the problem of the phonetic reex of the old acute

and short neo-acute aside).In Krianis dialect, the shift of the old short vowel a.p. b adjectives

to a.p. A is clear, unlike elsewhere in Kajk., cf. , H nom.pl. m. nom. dual n. nwa (cf. def.H), etc. However, one thingis interesting in Krianis dialect. According to the Russian accentologicalschool,59 the reex of the old acute ( in Krianis texts) and the old shortneo-acute ( in Krianis texts) is different in monosyllabic words in

prepausal position. Cf. the words brother, frost, whip, heaven, bird, bread, emperor, etc. (all fromthe old acute, a.p. a) but dream, roof, ox, catsh, rain, horse, basket, etc. (all from the old short neo-acute, a.p. b).This distinction is neutralized in non-prepausal positions andin polysyllabic words, cf. both and language, tongue. Thedistinction is seen in adjectives as well, albeit with much more vacillationthan in nouns. Cf. lglong, ,slb, Cp,zw (but lame

and / ) vs. evil,,H,Xp, (with variantswith :zl, chm,60, while is written with only).

Alternatively, this distinction might be explained as graphic only, butthe exact distribution of the accent marks is hard to explain in this way.The other possibility would be that these signs were intended to point to

different accentual patterns in other forms, for instance the distinction of

59 1968, 221; Os l o n 2011.

60 This adjective seems to belong to a synchronic a.p. B (cf. nom. pl. m. Xpo,gen. pl. chromch).


21/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 123 and . This suggestion might work for nouns but not so muchfor adjectives since there the old short vowel a.p. b has the same immobile

root stress as the original a.p. a, cf. nwo = Cpo. However, the variants of

/ ( / zl) that exist in almost all of the old short vowel a.p. b adjectivesand in some old a.p. a ones might be signicant. This could point to therelation of these graphic signs with the type of accentual paradigmatic

patterns even if these signs indeed represent different phonetic tones with

different historical background.61 The third option would be that these arejust some sort of allotones on different vowels since the old acute and old

short neo-acute do not usually appear on the same vowels. However, wordslike and speak against such a suggestion (even though such words

are expectedly rare).In any case, it seems reasonable to assume that the original distinction

of the acute and the short neo-acute could be maintained in monosyllabic

nouns in prepausal positions in Krianis texts (of course, this would pointto the fact that such a distinction had previously existed in other positionsas well). However, this is problematic when one considers the interpretationof the obvious fact that the original short vowel a.p. b shifts to a.p. A in all

tok/ak. dialects (and apparently in Krianis dialect as well). Such ashift (although unusual by itself considering it was quite early) can hardlybe explained if we take into account Krianis data where the equationof st = nv, unlike elsewhere, does not hold true. Perhaps this couldbe explained by the fact that the old acute and old short neo-acute havefunctioned in some sort of complementary distribution since the Proto- or

Common Slavic period. The old acute was present on long vowels onlyand the short neo-acute on short vowels only. Thus, the short neo-acute (or

some other prosodeme with the same reex in later dialects) functionedas a sort of replacement of the old acute on short vowels. We see this inprexal derivatives like *ss d neighbor they had the old acute if theroot was long but the short neo-acute (or some sort of tone that has the same

61 One could perhaps assume that Kriani found it easier to distinguish from if there was a distinction in other forms as well. Kriani might have understood theconnection of with constant root stress and the connection of with the post-stem

stress, even if these tones were indeed different in pronunciation. This may also pointto a gradual dissapearance of this distinction in Krianis dialect.


22/136

124 Mate Kapovireex) if the root was short, for instance in *potk brook. Therefore, it isperhaps not too strange that the phonetic distinction ofistand nov could bepreserved even if these two phonetic realizations (that usually appeared on

different vowels in any case) were functionally the same prosodeme. In thatcase, the analogy to *st *sta *sto may have caused the appearanceof the innovative *nva *nvo instead of the older *nov *nov inspite of the fact that *nv was phonetically different from *st. Besides,one should point out that the accent of the old a.p. a and short vowel a.p. b

was the same in the def. forms (see below), which may have had someinuence.

Cf. in Slovene:

st s ta sto def.s ti

Slovene is different from Kajkavian in that it has the neo-circumexin most adjectives in the indef. nom. sg. f. (due to the same kind of logicthat causes the neo-circumex to appear in some -stems derivatives and infeminine participle forms)and not just in def. forms. Cf. bgat(< *bogt) bogta bogto for polysyllabic adjectives. However, there is no change insome adjectives, cf. the indef.sta, a, sta (alsosva, which is a.p. b

originally).62 Unlike tok. and ak., the adjectives of the type ofnv andhrm (the old short vowel a.p. b) remain in a.p. B, i.e. there is no mergerwith a.p. A nvabutsta.

a.p. A:63br'ewith young (of animals),c'i,64'ilastir,'istclean (cf.the Stand. Croat. verb stiti clean),65'uglong, h'leager (> B:, cf. hlitirush joyfully), h'om lame (< *B), 'ist same (> ist' C, cf. stina truth),klj'astgame, lame, k'rtcrisp, l'akeasy, light (PSl. b, originally an *-kadj., dial. lagak), l'obad (< *B), m'alsmall (> B:, def. B:, cf. the adverbs

62 B r e z n i k 1924, 94; S t a n k i e w i c z 1993, 65.63 The words in word-lists are written in some sort of (tok.) prototype forms,

with for different reexes of theyatand with non-vocalized nal -l, thussmlandnotsmio, etc.

64 Usually only in the phrase c zma freezing cold but cf. also B a o t i 1979,196.

65 In the brackets, we give cognate words (in their Standard Croatian form) to

show if a certain adjective belongs to the original a.p. a (with cognate words havingconstant ) or not.


23/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 125mlo little, bit zmalo almost), m'eksoft (also C:, PSl. c,originallyan *-k adj., dial. mekak), m'ilpleasant (cf. mliti se feel like doingsomething), mn'og many (cf. mniti multiply), m'rk glum, n'ov new

(< *B, cf. obnviti renew), pl'ah timid (> C:, cf. pliti scare), p'av real(def. B:, cf. prvo right, ppaviti x), p'ostrude, simple(< *B, also C,PSl. b),66 p'rvrst(> C, B:), p'un full (cf.pniti ll), 'aglad (cf. rdogladly, rdovati rejoice), 'an early (> C:, def. B:, cf. rno early, rnitibeearly), r'ired haired (also C:), 'us red haired (also C:), s'itsatiated (cf.nsititi s satiate), skl'on apt(< *B, cf.sklnitiput away), sk'orsoon to be,recent (< *B, cf.skro soon, recently), sl'ab weak (cf.slbjeti grow weak),sm'ebrown (< *B, > B:), sm'lbold (cf.smjti may), sn'n sleepy, sp'o

slow (< *B, cf. uspiti slow down),67st'aold(def. B:, cf. strjeti growold), st'm stipe, st'ogstrict (< *B, cf. postriti stricten), sv' fresh (aloanword from Czech, cf. svjeiti freshen up),k'rtstingy (C: in Posavi-na),68 t'atvane (< *B), t'ih silent (> B:/C:),69t'om inert (< *B),t'ulrotten(cf. trliti rotten), ub'ogpoor (> ubog), v't accomplished (cf. vjticawitch), v'itslim, v'lhot (for water)(cf. vrti), v'lbrave, z'av healthy(def.A > B:, cf. zaviti get well / healthy),zn'an known (> B:/C:), z'l70ripe

(cf.zrti),'ukbitter(originally an *-k adj., cf. the variantuhak)Note: etymologically, the old a.p. a and the old short vowel a.p. b

adjectives are easy to distinguish since the latter always have one of the old

short vowels in the root (*e, *o, * or *), while the former have old longvowels (all other ones).

66 Cf. 2000, 217 and Slv.prst prsta for a.p. b.67 Since this is a prexal derivation, one could expect also *spr *spra

*spro with the constant a.p. a stress already in PSl. but examples like uspitipointto a.p. b (thus *spr *spora *sporo ).

68 Cf. for instance Sikerevci, Brodski Stupnik, Budrovci and Slobodnicakt.69 The length is probably always due to forms like tj with the new -j, with a

lengthening in front of -j (which is a regular phonetic rule in many dialects).

70 The adjectives trm,smo, vro,zro are originally participles of the verbs tticrush,smjti may, vrti boil,zrti ripen.


24/136

126 Mate Kapovi2. a.p. b

Proto-Slavic

short vowelindenite adjectives denite adjectivesm f n m f n

*gl *gola *golo *glj *glaja *glojenaked

long vowelindenite adjectives denite adjectives

m f n m f n*lt *lta *lto *ltj *ltaja *ltojeyellow

The accentual pattern of the indef. a.p. b adj. in PSl. is simple. The

stress is always on the syllable immediately following the stem (*gola ,*golo ), except in the cases where nalys cannot be stressed (*gl), i.e.the situation is the same as in nouns (*snp bundle, gen. sg. *snopa ,dat. sg. *snopu ). Long forms are more problematic all Slavic languagespoint to the neo-acute on the stem: *glj, *ltj (Croat.gl, dial.t).However, what is not clear is the origin of the neo-acute in this position.There are two possibilities retraction by Ivis Law (i.e. the retractionof the stress after the operation of Dybos Law) or the absence of DybosLaw. The supposed retraction by Ivis (Stangs) Law is often attributed tothe contraction of the ending (*-j, *-aja, *-oje) after Dybos Law, whichwould have caused the appearance of the non-initial falling accent (*-y , *-,

*-). This accent would have then retracted to the preceding syllable as aneo-acute (the same as in *ptajete you are asking > *ptte > Croat. dial.ptte).71 However, there are a number of problems with this explanation(Kapovi 2005b, 34; Langston 2006, 270f). Firstly, the neo-acute onthe stem is Pan-Slavic, which means that this pattern must be quite an earlydevelopment. However, vowel contractions are anything but early and Pan-Slavic. The vowel contraction is a rather late development in Slavic, it isoften absent in East Slavic and in other Slavic languages it occurs in various

71 For such an explanation, cf. Stang 1957, 101102.


25/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 127ways and to various extent (cf. e.g. Croat. bjati s be afraid but Czech/Slv. bti s).72 To set the said theory into the realm of impossibilities evenmore, in Russian the nom/acc. sg/pl. endings are not contracted even today,

cf. Russian nom/acc. sg. m. white, nom. sg. f. a, acc. sg. f.y, nom/acc. sg. n. oe, nom/acc. pl. e. What is more, in OldRussian the contraction was not complete even in oblique cases. The latedate of the contraction in def. adj. forms is clear from its results as well,which are different in various languages / dialects, cf. OCS -ajgo / -aago /-ago, Croat. -g(a) (with the vocalism by analogy to the pronominal -oga *go (vocalizationof nal -l) >g (contraction) >gl(analogical reintroduction of nal -lin some dia-lects). The older form / accentgl(beside the younger variantg)is preserved in thedialect of Vidonje in Neretvanska Krajina (Vidovi 2009, 292).

76 Such examples can also be a.p. A synchronically if one looks at adjectives alonebut other forms can point to a.p. B (cf.popiti generalize).

77 Cf. in Dubrovnik: g gla globut bs bsa bso (R e et a r 1900,113).


27/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 339here the shift to a.p. A was not possible because of the specic stem typethat has mobile a (i.e. the reex of the y), which is the reason why thisadjective preserves end stress (zl,zl) even in Neo-tokavian. Of course,

in the dialects where the original a.p. C adjectives have shifted to a.p. B(and a.p. C is no more), a.p. B is much larger (there, one nds adjectiveslike mld mlda mldo young and bs bsa bso in a.p. B). Thepreservation of a.p. C is typical for the Western tokavian dialects (this isan isogloss connecting Western tokavian to akavian), while a.p. C is notpresent in the Eastern tokavian dialects.

In a.p. B def. forms, one usually nds / (i.e. / in Neo-tok.).However, in short vowel adj. the secondary ` also appears, for instance the

innovative gl instead of the oldergl (as already said, gol is the onlymonosyllabic adjective in a.p. B, but a.p. B appears in adjectives with othersufxes as well and the original def. pattern is the same there). This is acase of a mixture with the old a.p. C forms and / or the inuence of indef.forms. Since the original desinential stress in a.p. C is quite rare (bsbeingthe only real short vowel monosyllabic adjective and the type drgdearbeing frequently lost), it is much more probable that secondary forms like

the def.gl gl instead of the oldergl gl are due to analogy tothe indef. formsgla glo and not to analogy to original a.p. C forms likebs bs ordrg drg. See above for a similar problem with thesecondary accent of a.p. A def. forms.

Ivi (1913 2, 42, 44, 49) gives the following indef. adjectives for thePosavian a.p. B: gl78, vr, jk strong (in some dial.), bl white, clwhole, gl rotten, mlk lukewarm, t, c n, and the following def.ones: gl, c n, bl, vr. Baoti (1979, 198) gives the following a.p.

B adj. for Kostr: bld pale, cl, c n, fn ne (a loanword), gl, ml(secondarily in a.p. B),plv blue,sv grey,stn foreign, vn black(see below), t, while vr, jkand mlk79 are in a.p. C. From the shortvowels adj., there is onlyg (gla,glo). In the def. forms, he gives bl,gl,etc. In aptinovac, according to Ivi (1907, 140), a.p. B in monosyllabic

78 The circumex in gl is due to the pre-resonant lengthening (from the older*gl).

79 Cf. Ivis mlk and my data where mlk is a.p. B: in Budrovci and a.p. C:(mlk) in Slobodnica.


28/136

340 Mate Kapovi(sufxless) adjectives is absent, with all adjectives shifting to a.p. C (cf.there cn cna cno).

My data from Posavina (from the villages of Sikerevci, Orubica, Koba,

Babina Greda, Slobodnica, Budrovci) show the following for a.p. B. As inIvis description, the adjectives meaning colors bl, cn,plv,t area.p. B everywhere.In the same semantic group, there is alsosv in Orubica,Koba, Budrovci and Slobodnica (cf. also Kostr above) but sv (C:) inSikerevci and Babina Greda80 (in Babina Greda, Budrovci and Slobodnica,sd grey-haired remains in a.p. C: as well). An a.p. C > B shift is seenin vn as well (attested as B in Sikerevci and Babina Greda, cf. a.p. B inNeretvanska krajina and Kostr). The adj. bldhasalso shifted to a.p. B: in

all dialects in Posavina81 (the same in Ivi and Baoti, as well as in Imotskaand Vrgorska krajina and Dubrovnik but not in Neretvanska krajina). Thisis a case of a tokavian innovation (cf. Czech bl for the original a.p.c), the same as jk(in Sikerevci and Budrovci, the same in Ivi, B alsoin Vrgorska krajina and Dubrovnik but not in Kostr and Neretvanskakrajina), cf. Czechjak for the original a.p. c. Such a tokavian innovationis seen in the adj. cilas well(I have clo attested for Orubica,82 cf. Czechcl for the original a.p. c) in Posavina, the problem is that today mostlyjust the def. form cl is used. In Sikerevci and Koba, an a.p. C > B shift isseen in the adj. bl gmild (cf. also B in Prapatnice below but Czech blahfor the original a.p. c). For paradigmatic shifts in adjectives in general, seebelow. The short vowel stem gl remains a.p. B everywhere in Posavinaand vr has shifted to a.p. C in all mentioned dialects in my data, which isin accord with Baotis data but not with Ivis. As in Kostr, the adj. mlis a.p. B: in all mentioned dialects (the original a.p. A is seen in the adverb

mlo). The neo-acute in the indef. form is taken from the frequent def. formml secondary forms ml / mla and ml / mlo (B) are results ofanalogy to ml.

80 Originally a.p. a (see below). A.p. C: is due to analogy to the formsv, which isthe expected reex of *s v with a pre-resonant lengthening. The formsv (B:) is dueto analogy to other adjectives meaning colors.

81 The adj.glcan also be a.p. B: in Posavina (thus in Slobodnica and Budrovcifor instance). Cf. also a.p. B: in Prapatnice as well.

82 It is interesting that the def. form cjlis common in Neo-tokavian Slavonia,with the accent probably due to analogy to the a.p. B indef. form.


29/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 341In Prapatnice (Vrgorska krajina, my data), a.p. B: is found in the usual

adjectives: cn (gen. sg. cna, dat. sg. cnu) cna cno nom. pl. m.cni def. cn, the same in bja (bla blo) white, r, sd,sv, vr (gen.

sg. vra nom. sg. f. vra nom. sg. n. vro), t. The adj. cl(gen. sg.cla, dat. sg. clu) is also in a.p. B, as well as the short vowel adj.g (gen. sg.gla) gla glo nom. pl. m.gli def.gl (gen. sg. m.glga). Otheradjectives that belong to this a.p. are bld(cf. Posavina bld), blg,gl,jk(cf. the adverbjko),83mld(with traces of the original a.p. C, see below),mlk, vl (vla vlo) soft (cf. a.p. B: for this adjective in Imotskakrajina as well). As we see, a.p. B mainly consists of adjectives meaningcolor (but not ofvrn, which is C). Besides these adjectives, pn, vrja hot

andzrja ripe shift to a.p. B from a.p. A and adjectives like bld, blg,jk,etc. are also in a.p. B, as usual in (at least some) tok. dialects (see the a.p.C list below). All of these adjectives are in opposition to the preserved a.p.C pattern one nds in cases like nom. pl. m. bzi quick, vsti hard, drgidear,glvi deaf,svi dry, etc. (see below).

For Imotski and Bekija, imundi (1971, 130131) gives a simplea.p. B forvr vra vre only. The mixed a.p. B/C consists ofbl

blda bldo / bl o, cil / cijo, gil / ijo, jak, mla, si, vla, ut. Forthe adjectives blg,jk,plv and r imundi gives the pattern ofblga /blga blgo / blgo (B/C), with variant accent even in nom. sg. f. form.This mixed B/C type is in accord with a.p. B: in Vrgorska krajina in thesevery adjectives.84 In spite of the fact that all a.p. B adjectives have a.p. Cvariants, the distinction of the original a.p. B and a.p. C is well preservedsince a.p. C always has the pattern of lp lpa lpo with no variants.In this dialect, there is no B : C but B/C : C opposition, because of the

inuence of a.p. C on the original a.p. B.In Vidonje and Dobranje in Neretvanska krajina,85 a.p. B, among other

adjectives, includes: bjl, cn,s v,t, co / cjl(but a.p. C when used withthe word vino wine),sjd, vr, vrn, bz, vst(the last three secondarily),while bljd, blg, ljn,sm,gl / g / gl,jk,g lare in a.p. C.

83 Cf. the adverbjko in Dubrovnik as well (Reetar 1900, 135).84 Interesting correspondences like these are just hinted at here. There has been no

accentological dialect geography research in Croatia.85 Domagoj Vidovi (p.c.).


30/136

342 Mate KapoviIn Dubrovnik (Ree ta r 1900, 114), cf. a.p. B: vr vra vre, pl.

vri vre vra, the same inut,jak, mlak, tu, cn,skur(an Italianloanword a.p. B because of the color semantics), bo (bijla bijlo),

bljedandgo.In Molise (Pi cco l i, Sammar t ino 2000), the only remnant of a.p. B isthe adverb vro, while all other a.p. b adj. shift to a.p. C:gl(gj) gla(gen. sg. m.gla, f.gl), cn cna cno (the same int).86 The adj.svhas generalized in all forms. The complete disappearance of a.p. B is theresult of the same tendency that is often seen in akavian.

All indef. oblique forms (both old preserved PSl. indef. forms and newones with def. endings, see below) have the rising accent (end stress in Old

tok.), while def. forms have the stem stress: indef. dat. sg. f. tj def.dat. sg. f.tj, indef. instr. sg. m/n.tm def. instr. sg. m/n.tm, etc.The original opposition of the indef. gen. sg. m/n.ta def. gen. sg. m/n.tg(a) can be replaced by the innovative indef. gen. sg. m/n.tg(a) def. gen. sg. m/n.tg(a). More on this below in the a.p. C part.

akavian(gl Senj,87bl Vrgada)88

short vowel


gl gol gol gli89 gla glo

long vowelindenite adjectives denite adjectivesm f n m f n

b(l) bl bl bl blo

Concerning the sufxless monosyllabic adjectives, a.p. B is far worsepreserved in akavian than in tokavian. In all akavian dialects, at leastsome of the original a.p. b adjectives shift to a.p. C and, in some of the

86 Cf. def.ti with a typical Molisean shortening in gen. sg.toga. Cf. alsosvi gen. sg.svoga.

87 M o g u 1966, 77.88 J ur i i 1973.

89 Instead of the expected *gli by analogy to gl, where the neo-acute is due topre-resonant lengthening, and by analogy to other (long vowel) adjectives.


31/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 343dialects, this occurs in all or almost all of the adjectives. On the other hand,most ak. dialects preserve at least a few adjectives of the original a.p. B.The shift to a.p. C is not strange since the number of a.p. C adjectives was

much greater to begin with.One should also note that in ak., unlike Neo-tok., t is usually

different from drg, but this does not obstruct the a.p. B: > C: shift. Thepivotal point for the B > C shift is probably constituted by feminine formslike t = drg, as well as by the def. forms where the original a.p. bpattern was generalized in ak.(t = drg). As in tokavian, golis anexception since it maintains its a.p. B in some dialects, unlike all other shortvowel sufxless a.p. b adjectives, which have merged with the original a.p.

a adjectives (see above).90On Hvar, Hras t e (1935, 32) attests only a.p. C: ut t to (def.

ti), gl gol glo (def. gli). However, there are some remnants ofa.p. B: even there. Cf. the ran rn rn black in DL (this isobviously a Hvar form, cf. the Bra n and the Vrboska form below) andthe transitional a.p. B:-C: form from Pitve (my data): nom. sg. n. t nom. pl.t(B:) butt nom. pl. f.te (C:)(and nom. pl. n.t that canbe both B and C). The preservation of the original a.p. B: is seen in Vrboska

as well (Matkovi 2004): bl bl bl, ran rn rn (B:) butt t to andgol gol glo with the shift to a.p. C:.

Cf. the Bra forms ( imunovi 2009, 44):t, vr, b l(def. bli), n,s v (all a.p. C) butgo lgol gol (def.gli) and, of course,zl(zl,zl)remaining in a.p. B.

The Vrgada (Jur i i 1966, 82) adjectives vr,t(andsd)91belong toa.p. C butg gol gol (def.gol with the secondary accent), b bl bl (def. bl), cn crn crn (def. cn)remain in a.p. B.

The case of the dialect of Okruk on the island of iovo92 is veryinteresting. In this dialect, the distinction of the old a.p. B: and a.p. C: is

90 That the adj.golis a special case in ak. as well is seen even in cases where a.p.B is not preserved, because thengolshifts to a.p. C (like all other a.p. b adjectives inak.), i.e. it does not merge with the old a.p. a like nov and the other old short vowela.p. b adjectives.

91 The adj.sijis originally a.p. c (see below) but we list it here since it is often

a.p. B: in tokavian.92 Data by Ante Juri.


32/136

344 Mate Kapoviwell preserved, as in some Western tok. dialects, but the inuence of a.p.C is present nevertheless. Cf. for a.p. B the pattern bl bla blo (< *bl,*bl by a regular phonetic retraction) and the same fort, cn, vr,sv,

sdand lp (which shifted to B from the old C). Opposed to this pattern, wesee the a.p. C pattern in adjectives like drg drga (< *drg) drgo(the same in mld, bld, jt, etc.). The distinction is preserved in neuterforms only, but it is nonetheless stabile. The circumex in the nom. sg. m. isapparently the result of the inuence of a.p. C,93 present here, as elsewherein akavian, but in Okruk only nom. sg. m. form merged with a.p. C, whileneuter forms (and thus the whole pattern) remained distinct. This is a uniquecase in our ak. data presented here. This preservation of a.p. B: can be

seen as an isogloss connecting this dialect with Western tok., and the samegoes for the fact that the adj.sv andsdbelong to a.p. B:, which is also atypical tok. feature. This is just one of the instances pointing to a tok.-ak. dialectal continuum.

In the dialect of Filipjakov,94 a.p. B: is completely gone. Cf. bl bla blo ( is generalized in a.p. C as well, see below), the same in cn,t(andsd). Even the adj.golhas shifted to a.p. C: gl / gja gla / gl glo(the latter by analogy to bs, see below, and by generalization of length).

In Preko on the island of Ugljan,95 all adjectives have shifted to a.p. Cas in the near-by Filipjakov: bi l bi la bi lo, the same in vr,t(andsi d; in cn cna cno the syllabic is shortened). Cf. alsogu gla glo (by analogy to bu s, see below).

On Rab (Kuar 1894, 3334), the adj. bl, vr,t(andsd) are in a.p.C, butgl gol gol pl.gol gol remains in a.p. B (bs hasshiftedto a.p. B by analogy, but in the def. form we havegolby analogy to the old

def. form bs).In Senj (Mo gu 1966), bl, vr,t(ands v) are in a.p. C, butglandcn remain a.p. B.

In Grobnik (Lu ke i, Zub i 2007, 106), cf. the preservation of a.p.B ingl gol gol (gen. sg. m. gol) as opposed to bs bos bso

93 It is important to note that the dialect preserves the distinction of the neo-acute

and circumex perfectly in all positions.

94 Data by Nikola Vuleti.95 The forms recorded by Nikola Vuleti.


33/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 345(C). In the long vowel group, a.p. B: is preserved in: n / n r n /rn r n / rn but not in bl bl blo,t t to. The adj.svbelongs to a.p. C: (sv svo).

A.p. B is attested by b l bl (def. b l) in Novi Vinodolski (1909, 185), for Orlec (Houtzagers 1985, 120121) cf. bel (def. b li),sln slan slan (def. slni), gl (f. and n. are not attested), and inOrbanii (Kalsbeek 1998, 145) gulgol gol (B) but bil, n,twith a shift to a.p. C.

In Rijeka (Strohal 1895, 157158), a.p. B is attested in the typezt zt zt def.zti. Here, one nds the adj. blas well but also manyold a.p. c adjectives (sh, glh, lp, ln lazy, etc.), which is strange in

ak. The indef. gen. sg. m/n. zt has the alternative form zteg (cf. thedef. gen. sg. m/n.ztega), while the indef. dat. sg. m/n. zt has the variantztem (cf. the def. gen. sg. m/n.ztemu). The end-stressed -og / -eg and-om / -em are found (in a.p. B and C) in Grobnik as well. See below forthe discussion of these forms.

kajkavian (gl Velika Rakovica,96t Prigorje)97


gl gol glo gli gla glo


t ta ti

(instead of -o)

The hegemony of a.p. C is typical for Kajkavian as well, even moreradically than in akavian.98 In most of modern Kajk. dialects, a.p. B is

96 M a r c h 1981, 265.97 R o i 18931894 2, 144.98 Generally speaking, the East (Eastern tok.) is prone to a.p. B, while the West

(Western tok., ak., Kajk.) is prone to a.p. C, with Western tok. being more moder-ate and Kajk. and ak. more a.p. C radical since they are spoken farther to the west.


34/136


35/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 347As to the old short vowel a.p. b, the adjectives from Bednja are too few

(only two although they do point to different outcomes of the old a.p.a and the old short vowel a.p. b). As for V. Rakovica, the data shows the

following. The original a.p. a adjectives(joined by nv, see above) yielda.p. A, while the old short vowel a.p. b reects as a.p. C' (i.e. a.p. C withno in nom. sg. m.). Despite the fact that there are only three a.p. C'adjectives derived from the original short vowel a.p. b, it is clear that therewas no merger of the old short vowel a.p. b and the old a.p. a here (with theV. Rakovica a.p. C' being a result of the older unchanged short vowel a.p.B),102 unlike in tok/ak. where the two groups merge into a.p. A (or thereex of it).

In Varadin (Lipljin 2002), one nds the def. forms only: b li, rni, ti,svi andg li (with a secondary ), which show the typical Kajkavian in the original short vowel stems as well (Iv i 1936, 72).103 Cf. also nvi,psti, ni. These forms, together with those from V. Rakovica, conrmthat at least in part of Kajk. the old short vowel a.p. b did not merge with theold a.p. a, unlike in tok/ak.

Krianis language preserves the old A/B/C distinction in short vowel

stems: (A), gen. sg. nom. pl., etc.(B), nom. pl. bsi (C). However, the distinction is gone in the def.forms in a typically Kajkavian manner: , , (there is nodistinction of and in Krianis texts). Cf. Oc 2011, 110 for the def.form. In long vowel stems, the distinction between B: and C: is in recession,as it seems, although data is scarce for B:. Cf. forC:, while the originally a.p. b adjectives czern / czrno and bl bl / blo (should be *blo) nom. pl. bjel apparently show a

vacillating B/C paradigm. Krianis data obviously attests the beginningof the general tendency of a B: > C: shift in Kajkavian.

Cf. Slovene:b l b la b lo / blo (def. b li / beli)

102 Had the adjectivesgl,prst, l(b)already merged with old a.p. a adjectiveslikest, it would not have been possible for them to shift to a.p. C' secondarily, withthe old a.p. a adjectives remaining in a.p. A.

103 For a discussion on the phonetic development of Proto-Slavic *` > Kajka- vian , see Ka p o v i (forthcoming).


36/136

348 Mate KapoviIn Slovene, a.p. B is preserved in both short (like nv)and long vowel

stems, but there is a variant a.p. C form in the neuter gender, while in def.forms a neo-circumex can appear by analogy to a.p. A.

a.p. B: b l'white, c n'black,gol'naked, jed 'nonly, l 'vleft, mog'(and m'og) possible, 'opgeneral, plv'blue (PSl.c),104 snj '(marine)blue (PSl. a),105 sv'grey(also C: < A,106 PSl. a),107vr'hot (> C:), zal'108evil, t'yellow

Three most frequent Proto-Slavic long vowel a.p. b adjectives were

color terms *bl, *rn and *lt. This is preserved in Croatian andhas inuenced other adjectives denoting color to shift to a.p. B:, since a.p.B: has become a salient marker of the adjectives used for color terms.109

Thus, *s n and *s v shift to a.p. B: from a.p. a, while *bl d, *s d and*plv shift to a.p. B: from a.p. c.A.p. B: shift tendencies are not necessarilygeneral and early (at least not in all of the adjectives),110 since the originala.p. is preserved in some dialects (cf.sv, blijdo andsdo above). As part ofthis tendency, the adj. *vrn black also shifted to a.p. B: in some dialects,and the adj. *mrk glum shifted from a.p. A to a.p. B: in some dialectsas well.

104 For the PSl. a.p. c cf. a.p. C in Zaliznyaks data ( 1985, 138) andCzech / Slovakplav. Lithuanian has bothpalvas andplvas. Snoj (in his dictionary)claims that Sloveneplv plva (a.p. A)is a loanword from other Slavic languages.

105 For the PSl. a.p. a cf. a.p. A in Zaliznyaks data ( 1985, 133) andCzech / Slovaksin.

106 Cf. ARj for the formsv in umberak.107 For PSl. *s v (a.p. a)cf. e.g. Slv.sv sva, Czech / Slovaksiv and Lithu-

anianvas.108

The adj.zo zl zl, because of its morphonological structure (mobile a,i.e. the reex of the y in the root) remains in a.p. B and preserves the end stress(the ending being the only syllable) even in dialects that have experienced retraction.Here, beside the expected def. formzl zl zl (with end stress, like in a.p. C, butagain due to the specic structure of this word) the secondary def. formzl zl zlwith the shortened ending is also attested in some dialects (probably by analogy to theindef. forms).

109 At least in tok., for ak. and Kajk. it is difcult to say since a.p. B in sufxlessadjectives is generally moribund or marginal there (but cf. above the Prigorje data forKajk.).

110 The exact territorial and dialectal extent of these changes is still to be examinedin details.


37/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 3493. a.p. c

Proto-Slavic


*bs barefoot *bosa *bso *bosj *bosaja *bosoje


*sx dry *sxa *sxo *suxj111 *suxaja *suxoje

In PSl., the indef. forms of adj. were morphologically exactly the sameas nouns ofo-stems (masculine and neuter) and -stems (feminine). Theaccentual pattern was the same as well. Thus, *bs and *sx have thesame kind of accentual paradigm as e.g. *bg god and *grd town,*bosa and *sxa as *voda water and *rka arm (cf. e.g. acc. sg. *bs,*sx) and *bso, *sxo like *zvno bell, *zlto gold.

In def. adj., the stress was always on the last syllable or on the one

before that. The exact position depended on the accentological propertiesof certain endings, i.e. on accentual valences. In i ndef. adjectives, like innouns, some forms had absolute initial stress (if the ending had () valence),while others had end stress (if the ending was (+)). Thus, forms like *bs,nom. sg. n. *bso, gen. sg. m/n. *bsa, dat. sg. m/n. *bsu, dat. sg. f. *bs,acc. sg. f. *bs, etc. are stress-initial, but forms like gen. pl. *bos , nom.sg. f. *bosa , gen. sg. f. *bosy , etc. have nal accentuation. Denite adjectiveforms are made by adding the forms of the demonstrative pronoun *j , *ja ,

*j (cf. 1981, 36 for the reconstruction) on the indef. forms (or, later,on the stem *bos- / bosy- in some cases). If an indef. form had its ownictus, i.e. if it was not unaccented (= with absolute initial stress), but had astrong (+) ending, like the forms *bosa (i.e. *bosa ), gen. sg. f. *bosy , gen.pl. *bos, the stress remained in the same position in the complex def. formas well: *bosaja, def. gen. sg. f. *bosyj, def. gen. pl. *bos j > *bosyj,etc. But if the form was unaccented (an nclinomnon, i.e. a form with the() ending), the stress was on the nal syllable, according to the rule of

111 From the older *bosj , *suxj .


38/136

350 Mate KapoviVasiev and Dolobko (cf. *dn day but *dns > Croat. nas today).If the second part of the complex adj. (i.e. the form of the pronoun *j)was monosyllabic, the stress was on the last syllable, of course, e.g. def.

nom. sg. m. *bosj , def. nom. sg. n. *bosoje . If the pronoun forms weredisyllabic (in oblique cases), where they were end-stressed (gen. sg. m/n.*jego , dat. sg. m/n. *jemu , loc. pl. *jx, instr. pl. *jm ),112 the stress of thedef. adj. form was also on the last syllable (i.e. on the second syllable ofthe second part of the complex adj. form): def. gen. sg. m/n. *bosajego , def.dat. sg. m/n. *bosajemu , 113 etc. In this way, for instance, the distinction ofthe def. acc. sg. f. *bosj (< *bs + *j ) and the def. instr. sg. f. *bosj(< *bosoj + *(je)j ) was created. The whole paradigm is as follows:114115116117118

m. n. f.sg.N. *bosj114 *bosoje *bosajaG. *bosajego *bosyjD. *bosujemu *bosjA. *bosj *bosoje *bosjL. *bosjm *bosjI. *bosyjm115 *bosj

pl.n. *bosij *bosaja *bosyjg. *bosj116d. *bosyjim117a. *bosyj *bosaja *bosyjl. *bosyjix118i. *bosyjimi

112 For the reconstruction, cf. 1981, 34, 36 and the oldest tok. accentnjga, njmu (tok. njma is secondary compared to PSl. *jma ).

113 As a parallel for *bsa + *jego > *bosajego cf. *p t + (gen. pl.) *des t >*ptdes t (Croat.ps t>pedst).

114 From the older *bos j.115 The nominative form (*bos + *jim) was taken instead of the expected

*bosom + *jim. *- then yields *-y- in front of *-j- and the form *bosyjim (OCSbosyim) appears. Cf. Ha m m 1970, 140.

116 From the older *bos j.117 Instead of *bosom + *jim. *-y- in the middle is either from the nominative

form *bos (which lengthens the ending *-- to *-y- in front of *-j-), by analogy togen. pl. (where the indef. ending is also *-- that lengthens to *-y- in front of *-j-) or

from instr. pl. (where the indef. ending is *-y-).118 Instead of *bosx + *jix.


39/136

Histoical vlopmnt of ajctiv accntuation in Coatian... 351Note. In dat., loc. and instr. pl. the reconstruction is not completely clear.

The accent of the def. form would differ depending on the derivation. If

the stem *bosy- is to be derived from the orthotonic instr. pl. *bosy (with

(+) ending), this would yield the accent *bosyjimi in the complex formas well. If *bosy- is to be derived from nom. sg. *bs (nclinomnon),one would expect the forms: dat. pl. *bosyjm < *bosyjm, loc. pl.*bosyjx > *bosyjx . These forms could have perhaps inuenced the instr.pl. *bosyjimi to become *bosyjm .119 Since these endings were contractedlater in Slavic and since one would expect levellings of all sorts in thesecases, it is impossible to tell what the original accent of these forms was.

It is clear that such a paradigm was quite complex as regards stress

position, i.e. whether it was on the ultimate or the penultimate syllable,upon which the intonation of the contracted vowel depended. There was a

tendency to generalize the stress position, e.g. *bosoje by analogy to *bosyjand *bosaja instead of the older *bosoje . It is understood that this means thata part of the presented reconstructions is actually not corroborated by the

later data because of different levellings it is rather based on a structuralanalysis of the system.120

Cf. Russian:o oo oo (def. oo)In Russian, a.p. C is quite well preserved and most of the root adjectives

belong to it. Many of the original a.p. a (e.g. c) and a.p. b (e.g. o)adjectives have secondarily shifted to a.p. C. Except for the mobile stress ofthe indef. forms, the end stress of the def. forms is also well preserved.

An important role in the reconstruction of PSl. adj. accentual types

is played by West Slavic languages. There, the original a.p. c adjectivespreserve the shortened vowel in the root,121 cf. Czech blah, bl,such.

119 Cf. Slv. -imin a.p. C.120 Stangs (1957, 103) reconstruction and analysis of a.p. c def. adj. accentua-

tion is not correct. The accent typesu h(instead ofsuh), which he wishes to explainfrom a Proto- or Common Slavic perspective, due to the ending contractions (whichwas a later phenomenon), does not have anything to do with PSl. Forms likeshare

much younger forms and are due to analogy to a.p. B (see below).121 Cf. Ka p o v i 2005a, 97100.


40/136

352 Mate KapoviThe brevity appears in the reexes of the original a.p. a as well, cf.sta,zav, but not in the original a.p. b, where the root remains long: hloupstupid, mou clever, bl white.

tokavian


bs bsa bso (bso) bs(bs) bs (bs) bs (bs)


sh sha sho (sho) sh sh sh

The accentuation of indef. declension of a.p. C adjectives is the sameas in nouns of course, in those cases where the original endings arepreserved:

N.sh sho sha (cf.grd zlto da soul)

G.sha sh(cf.grda / zlta d)D.shu (cf.grdu / zltu)A.sh / sha sho shu (cf.grd / vka wolf zlto du)L.shu122(cf.grdu zltu)123

n.shi sha shea.she sha she

122 Since many dialects do not have indef. adjectival declension, the opposition ofdat/loc. sg. in a.p. C (cf. K a p o v i 2010, 7981) or adjectival a.p. C, it is not clear ifa dialect exists in which indef. dat. sg. m/n.shu and indef. loc. sg. m/n. *shu would

be different or if these forms are always the same.Looking at the data from the dialectspreserving indef. adj. declension and a.p. C, it seems that indef. dat. sg. m/n. is identi-cal to indef. loc. sg. m/n., just like in neuter and masculine animate o-stems. Cf. dat/loc. sg.shu in Dubrovnik (Re e t a r 1900, 121), in Prapatnice (Vrgorska Krajina, mydata): svu krvu (living) on old bread, nlpu mstu in a nice place, nkrvumstu in a wrong place, in Posavina (Ivi 1913 2, 44) tu slu in a foreign

village, nlpu mjstu.123 For the accentuation of neutero-stems, cf. Ka p o v i 2011a.


41/136

Histoical

Documents

Kapovic Adjective Accentuation