The three accent paradigms of Proto-Balto-Slavic

The three accent paradigms of Proto-Balto-Slavic and the evolution of the three Slavic accent paradigms Miguel Carrasquer, July 2005 In the following presentation, I will try to outline a theory of how the three Common Slavic accent paradigms (a, b and c) can be derived from accentual patterns in Proto-Indo-European, for both nouns/adjectives and verbs. A central assumption will be that Balto-Slavic had three accent paradigms, not two, as is usually assumed. 0. Indo-European For Proto-Indo-European, I will assume the following: Athematic nouns could be acrostatic, with constant stress on the root, or mobile. There were a number of mobile patterns (proterodynamic, hysterodynamic, amphidynamic), but in ProtoBalto-Slavic these had all merged into a single laterally mobile accent class: PD h2ákmõ h2 AD póntoh2s pónth2 HD dʰugə2 dʰugə2 h2kménos th2ós dʰugə2trés Becoming something like: h2 h2ákmenim h2akmenés pantóh2s pánth2im pinth2és dʰugə2 dʰúgə2terim dʰugə2terés The resulting accent curve showed an opposition between end-stress in the nominative vs. begin-stress in the accusative singular, and begin-stress in the nominative/accusative dual and plural vs. end-stress in the dual and plural oblique. The singular oblique was split into baritone forms (dative, locative) and oxytone forms (genitive, instrumental). Vowel stems (o-, ah2-, i- and u-stems) were either barytone or theme-stressed. Nominal suffixes could be stressless or dominant (like for instance the diminutive suffix *-ikós, which always attracted the stress). The verbal system had more or less the same categories as the nominal system. There were root-stressed and mobile athematic verbs, and barytone and oxytone thematic verbs. The accent curve of the athematic mobile verbs showed root-stress in the singular, end-stress in the dual and plural: sg. h1ésmi h1és(s)i h1ésti pl. ə1smós ə1stés ə1sénti (> ə1sentí) du. ə1swáh2 ə1stáh2 ə1sté Simple thematic verbs belonged to the barytone category (except for a few verbs of the tudátitype), while the verbs with suffixes -jé-, -ské-, -dé- (itself from end-stressed imperative -dhí ), né- (alternating with -n-), denominatives in -ijé-, -ejé-, causatives-iteratives in -éje- were endstressed (better: theme-stressed). Statives in -éih1- (alternating with -h1i- in the plural and -eh1- in the infinitive) probably had a mixed paradigm, being theme-stressed in the singular, end-stressed in the plural. 1. Pedersen’s Law Pede se ’s la is the a alogi al t a sfe of o ilit f o the athe ati lasses to the o el stems. In the nominal system, the is d’ê of the analogy was to mimic in vowel-stem nouns and adjectives the prosodic distinction between nominative and accusative singular that existed in athematic nouns. It was therefore the oxytone thematics which shifted the stress back in the accusative singular, in the dative-locative singular (except the i- and u-stems, which retained their locatives in end-stressed - i and - u), and in the nominative and accusative dual and plural (except the o-stems, which retained end-stress in the nominative plural -áj ). In the ostems, the barytone ablative singular (-ã) took the place of the genitive. This resulted in (ignoring laryngeals and voiced aspirates): Nom Acc Voc Gen Dat Loc Ins Nom Acc Gen Dat Loc Ins NA GL DI athematic akm ákmenim (ákmen) akmenés ákmenei ákmeni ) ákmenes ákmenins ak m akmenimás akmenišú i š akmenáu i o-stems draugás dráugam dráuge d ug i dráugai d ug draugáj d ug ns draug m draugamás draugaišú draugájš d ug draugáu d ug ah2-stems s i s i m (stáina) s i (n)s s i i s i i (m) s i s i (n)s s i ns stain m s i más s i šú š s i stáinai staináu s i i-stems g íš ágnim (ágnei) g iš ágnejei i g ) ágnejes ágnins agne m agnimás agnišú g i š g agnejáu g i u-stems ledús lédum (lédau) l d uš lédawei u l du) lédawes léduns leda m ledumás ledušú l du š l duː ledauáu(š) l du Thematic oxytone neuter nouns had no accusative, so they remained oxytone (or, rather, theme-stressed): o-stems (neutra) pterám pter i pterái pter m pterámas pteráišu pterájš pteráu Barytonesis in the NA plural may have a PIE background (cf. Иллич-Свит ч, p. 53). Another category where theme-stress was maintained in Balto-Slavic is that of composite nouns with stressed suffix: barytone root dvarikás dvarikám dvarike d i d i i dvarikái d i dvarikáj d i ns dvarik m dvarikámas dvarikáišu dvarikájš d i dvarikáu d i mobile root maldikás maldikám maldike ldi maldik i maldikái maldik maldikáj maldik ns maldik m maldikámas maldikáišu maldikájš ldi maldikáu ldi In the present system of the verb, accent mobility in the athematic verbs varied between singular and non-singular. When the thematic verbs took over the mobility, it was the barytone verbs which moved the stress to the final syllable in the dual and plural: 1 2 3 1 2 3 1 2 3 athematic ésmi ésaj ésti esmás estés sentí s s esté e-verbs ši béreti beramás beretés berantí bereté Theme-stressed verbs remained theme-stressed: 1 2 3 1 2 3 -jés l s l ši steljéti steljámas steljétes steljánti -nése s ši seknéti seknámas seknétes seknánti (East Baltic) s sénksi sénkti senkmés senktés senkénti 1 2 3 1 2 3 -éjewadéi > wad d wadéiesi > wad ši d ši wadéieti > wad i d i wadéiamas > wad s d wadéiates > wad d wadéianti > wad i d i s -éih1baléih1mi > bal ː baléih1ši > bal ši baléih1ti > bal i balh1imás ~> bal balh1ités ~> bal balh1iénti ~> bal í l l ši l i s l l l s (East Baltic) ~ balí~ balí~ balíti balimés balités í The infinitive/aorist system in Slavic consisted largely of end-stressed (theme-stressed) forms. The infinitive is derived from *-táj. The aorist forms continue the PIE thematic root aorist (Cverbs only): dʰ-óm dʰ-és dʰ-ét dʰ-ómos dʰ-éte dʰ-ónt or a mix of thematic and athematic aorist forms (V-verbs, C-verbs [except in the 2/3 sg.]): noh3-šóm nóh3-s(s) nóh3-st noh3-šómos noh3-sté noh3-š The l-participle was theme-stressed in PIE, and therefore mobile in PBS (at least for verbs with a mobile present system): nom. acc. m. dʰl s dʰl n. dʰl -f. dʰl dʰl m In summary, we can reconstruct three accent paradigms for Balto-Slavic: I II III root-stressed theme-stressed mobile Lithuanian has merged I and II, Latvian II and III (only detectable in acute roots), Slavic shows clear traces of all three accent paradigms, as explained below. The three accent paradigms can be derived from PIE prototypes as follows: nouns verbs I ath. acrostatic them. barytone ath. acrostatic II them. oxytone (n) them. oxytone III ath. mobile them. oxytone (m/f) ath. mobile them. barytone 2. Hirt’s Law Hi t’s la aused a et a tio of the a e t if the s lla le efo e the i tus o tai ed i its oda a non- o alized la geal. Hi t’s la had o effe t if the a e t as al ead o the s lla le containing the laryngeal (accent class I), or if the ictus was two syllables or more removed from the syllable containing the laryngeal. This is the case in thematic mobile verbs of class III: páh2s páh2seši páh2seti pah2samás pah2setés pah2santí pah2sawáh2 pah2setáh2 pah2seté The aorist/infinitive system of these verbs, however, was su je t to Hi t’s la : inf. aor. 1sg. aor. 3sg. aor. 3pl. l-ptc m. l-ptc. f. l-ptc. n. pah2stéi pah2s(s)óm pah2sét pah2sónt pah2slós pah2sláh2 pah2slód Athe ati o ile e s e e of ou se su je t to Hi t’s la i infinitive/aorist systems: pres. 1sg. pres. 1pl. aor. 1sg. inf. l-ptc. m. l-ptc. f. l-ptc. n. léh1 leh1 leh1 leh1 leh1 leh1 leh1 ʰmi ʰmós ʰsóm ʰtáj ʰlós ʰláh2 ʰlóm > páh2stei > páh2sam > páh2set > páh2sant > páh2slas > páh2slah2 > páh2sla oth the p ese t a d the = léh1žmi > léh1žmas > léh1žsam > léh1žtaj > léh1žlas > léh1žlah2 > léh1žla(m) Mobile vowel-ste ou s e e a d la ge affe ted Hi t’s law, although exceptionally a mixed paradigm (with barytone singular and mobile plural/dual) may have emerged, perhaps in the word suh1nús so : súh1nus súh1num súh1nau súh1 uš súh1nawei súh1 u súh1numi súh1nawes súh1nuns suh1n m suh1numás suh1nušú suh1 u š súh1nuh1 suh1nawáu suh1 u In the DLI plural of the ah2-ste s, Hi t’s la š). s - šu vowel (- aused a et a tio of the a e t to the the e The laryngeal was vocalized in the sequences eRH, aRH, where R is one of m,n,r,l,i,u, except in the case of aiH e.g. Hi t’s la o ks i paiHláh2 pela, ut ot i ases like galəwáh2, tenəwás, leiHláh2, etc.). As shown by Francis and Normier, in both Greek and Tocharian the laryngeal was vocalized in the sequences ih2, ih3, uh2, uh3. The same must have happened in Balto-Slavic, as witnessed by Slavic bylá, žil , pilá, Latvian bût, dzît, he e Hi t’s la did ot ope ate, f o PIE *bhuh2-, *gwih3(w)-, *pih3-. 3. Winter’s Law Also to the Balto-“la i pe iod elo gs Wi te ’s la . This did ot ause a i ediate ha ges in the accent, but it did cause vowel lengthening c.q. acute intonation in the position before a PIE (unaspirated) voiced stop (*b, *d, *g, * , *gw). The exact conditions on the operation of Wi te ’s la e ai i dispute, ut gi e that i the ast ajo it of ases, Wi te ’s le gthe i g shows up in Latvian as broken tone, there must be something to Shintani’s suggestio that Wi te ’s le gthe i g of full o els only took place in the pretonic position (no such restriction seems to apply to the acute intonation of diphthongs in sequences eRD, aRD . If so, Wi te ’s lengthening would have failed to work in words of accent class I, and in thematic verbs of accent class III. ei et’s Law 4. s -law After the breakup of Balto-Slavic, the three Balto-Slavic accent paradigms underwent a series of accentual laws, which resulted in the three Slavic accent paradigms. The principal developments were:    Meillet’s law (III > c) “ta g’ s la II D o’s la a d its o e se hi h I all the jabloko-la I a, ; a Meillet’s la affe ts the a to e fo s of o ile pa adig s. I Ras usse ’s fo ulatio (Rasmussen 1992 : 475), the law is a further extension of lateral mobility: if a preposition or preverb preceeds, it takes the stress. Otherwise, if the stress is acute (ictus on the second mora), it e o es i u fle i tus o the fi st o a . I the e al s ste , Meillet’s law behaved unexpectedly in one way: the monosyllabic 2/3sg. forms of the aorist of (V-) verbs with a mobile present were treated as if they were were barytone forms of a mobile paradigm. This caused the elimination of acute intonation in the aorist 2/3 sg. of mobile verbs, as well as in the lparticiple (and sometimes even in the infinitive [žertí, pętí]). A late de elop e t, ut lea l a o se ue e of Meillet’s la [ asso iatio of i itial i u fle ith o ilit ], is hat I all the męso-law : o -mobile paradigms with a pretonic (i.e. a.p. II) circumflex vowel, but probably only in an open syllable, become mobile. This happens with a.p. II neuters like m só > ęs and jé > ; with né-verbs like vȋnǫ, ǫ, ǫ sę, s ǫ, and dé-verbs like mȋnǫ and ę ǫ; with jé-verbs like d ǫ, ž ǫ, l ǫ, ǫ, l ǫ, s l dǫ. I can find no similar transfer to the mobile paradigm in words that would have become a.p. afte the o ki g of D o’s law, so that would be a strong argument in favour of the existence of a Proto-Slavic accent paradigm II, besides traditional I and III. 5. Stang’s Law Although the usual defi itio of “ta g’s la is u h oade , I ould like to est i t it he e to the following formulation: in a.p. II forms with medial accent, stress is retracted to the root syllable, except, probably, when the stressed syllable is acute. In the verb, this immediately explains the accent curve of old a.p. II verbs: 1 2 3 1 2 3 -jéstelj stélješi stéljeti stéljemu stéljete stéljanti -nés sékneši sékneti séknemu séknete séknanti In the noun, the a.p. II neuters also acquire new mobility: -éjed d ši wád ti d mu d te d nti NA G D L I NA G D L I NA GL DI perá perúi perái pérami (> peramí) p peru (> pér N) péramas péraišu per š pér š) peráu (> ) As can be seen, the paradigm was analogically remodeled to look like the mirror image of the a.p. c mobile paradigm. The reverse happened in the verb, where the a.p. c verbal paradigms were reshaped to look like mirror images of the neo-mobile a.p. b paradigm: N bereší beretí beremúš bereté berantí bereté 6. ’s Law a ъk -law D o’s la affe ts the Balto-Slavic a.p. I: if the stressed syllable is not acute, the stress moves o e s lla le to the ight. The o e se of D o’s la I all it the jabloko-la , affe ts the Balto-Slavic a.p. II: if a word contains an acute syllable before the stress, the stress shifts to that syllable (this can span multiple syllables). The effe ts of D o’s la a e eadil see i o i al fo s, he e o -acute a.p. I words (from PIE athematic static and thematic barytone nouns and adjectives) are affected by it. As noted by Illich-Svitych, a.p. I neuters become a.p. b masculines in the process (e.g. dʰ => d ). There was no merger of the dvò -group with the peró-group, because the latter had become o ile hat I ha e alled “ta g’s la a o e , a d D o’s la does ot o k o o ile pa adig s. D o’s la also did ot o k i the si gula of as uli e o-stem barytona, where the retraction/loss of the end-stressed nom.sg. and ins.sg. in the mobile forms had brought about a merger of a.p. I and III in the whole singular. When the non-acute a.p. I oblique plural forms did u de go the effe t of D o’s la , the hole pa adig e ged ith a.p. c, both in ictus and (falling) intonation. I the e , D o’s la should ha e ei t odu ed the e-stressed paradigms, but this is rarely the case: the old athematic barytona fell together either with a.p. c (e.g. dǫ, bostí ) or with a.p. b (mogǫ ž š mogtí; meljǫ, mèlješ lti, etc.). Only the recessive i-stem causatives and denominatives formed from non-acute barytone roots acquired, according to Dybo (1990 : 36), a theme-stressed paradigm, (ložjǫ, lož t l ž ti). Accent paradigm II words with an acute root underwent the opposite development: the stress was pulled back to the acute, and they became a.p. a. In the nouns, examples of this are rare, as are a.p. II nominals themselves. The oxytone neuters (peró-group) had become mobile by d Wi te ’s la f o *wedróm > drá) is not “ta g’s la , hi h is perhaps why v d affected, although the close- or openness of the syllable may also have played a role here: the law does affect words like yugóm > gá > go. A word like l was mobile in PIE (*h2 ls > PBS *h2 ls, Wi te ’s la l- (oblique bul-?). In Slavic, it was rendered immobile and theme-stressed by addition of the dominant suffix *-kó-, and subsequently * bl kó > bl ko. In the verbs, where a.p. II was much more common, the jabloko-law retracts the stress from a.p. b present tense 1st person singular (the only form that was left with non-initial stress in the p ese t afte “ta g’s law). The effect on infinitives of all accent paradigms is more interesting: with the exception of a handful of infinitives with *erH (pertí, žertí, stertí, dertí ) and *eNH (pętí, tętí ) from verbs which are mobile in the present, all infinitives with an acute root have retracted the a e t. I pa t, this had al ead happe ed as a o se ue e of Hi t’s la e.g. all e s i *-ah2táj, *-eh1táj), but a separate retraction law is required to explain infinitives like s s i s i si s gi i i s i i ž i, et ., all ith Wi te ’s le gthe i g o VRH-sequences, , g > hi h e e e e su je t to Hi t’s la . Co t ast also a.p. a g > g g , from originally a.p. II verbs, with mobile žiwhere the acute had already been eli i ated Meillet’s la . The contrast seen in stative verbs ( /i) between a.p. a d i, s i, d ž i, sl š i vs. a.p. c ž i, s d i, š d i, is probably due to the same factors as the contrast between a.p. a s ǫ, ǫ, s ǫ, ǫ, s s ǫ, s ǫ, ž ǫ, s ž ǫ vs. a.p. b s ǫ, ǫ, ǫ, ǫ, ǫ, d ǫ (Dybo 1981 : 209-210). At the relevant time, there were no circumflex / / or /ũ/, only acute / /, / /, this in contrast with inherited /ã/, /õ/, / / besides / /, / /, / /. References Dybo, V.A., 1981, Slavjanskaja akcentologija, Moskva: Nauka. Dybo, V.A., Nikolaev, S.L., Zamjatina, G.I., 1990, Osnovy slavjanskoj akcentologii, Moskva: Nauka. Illi -S it , V.M., 1963, Imennaja akcentuacija v baltijskom i slavjanskom, Moskva: AN SSSR Rasmussen, J.E., 1992. Die Vorgeschichte der baltoslavische Akzentuierung. Selected Papers on IndoEuropean Linguistics (2). Copenhagen: Museum Tusculanum Press.