Slavic verbal accentuation

Slavic verbal accentuation. Miguel Carrasquer Vidal, September 2006 As I argued last year at IWoBA I in Zagreb, I believe that Proto-Balto-Slavic had three distinct accent paradigms: barytone (I), theme-stressed (II) and mobile (III). The novelty here is the postulation of a theme-stressed paradigm, reflecting: (1) PIE oxytone o-stem neuters, (2) compounds with dominant (i.e. stressed) suffixes such as PIE *-ikós, and (3) verbs with PIE stress on the thematic vowel. This latter category is especially abundant, in contrast with what is the case in nominal formations, where the theme-stressed a.p. was always marginal in comparison with the other two. A thorough study of the PIE background and further development of Balto-Slavic verbal accentuation, dare I say a Glagol’ oja ak e tua ija v altijsko i slavja sko , ould e e ui ed to put the 3-paradigm h pothesis o fi e g ou d a d so t out all the details. I do t ha e that fo ou o . What I ha e are a few suggestions about the origin and development of Slavic verbal accentuation, which I will outli e elo . Most of the “la i ate ial is d a f o the lists i )aliz jak s Ot praslavjanskoj akcentuacii k russkoj , a d is o pa ed ith PIE ate ial la gel f o LIV. Origin of the Slavic conjugations The first conjugation (e-verbs) consists of: 1. PIE bhárati- e s, hi h afte the o ki g of Pede se s la passed i to the o ile III) accent paradigm. 2. PIE tudáti-verbs, which remained in the theme-stressed (II) accent paradigm. 3. PIE -ské (> -sté) verbs, a conspicuous category in Baltic, but marginal in Slavic. We assume they remained theme-stressed (a.p. II) in Balto-Slavic. 4. dhé-verbs, a uniquely Slavic category: athematic verbs which were thematized by generalization of the 2sg. imperative form in *-dhí. They were of course theme-stressed. Many originally athematic verbs were subsequently thematized, presumably at different times, by simple substitution of thematic endings for the athematic endings, and passed into the first (or third) conjugation, resulting in the eventual reduction of the athematic category to just a handful of verbs in Common Slavic. The second conjugation (ne-verbs) continues the PIE n-infix verbs (sg. *bhu-né-dh-, pl. *bhu-n-dh-'). While Baltic generalized the plural form with n-infix, Slavic generalized the singular form, with theme-stressed -né- as a suffix (*bhudh-né-: innovation or archaism?). The third conjugation (je-verbs) consisted in PIE of verbs with theme-stressed -jé- and zero-grade root, which naturally passed into a.p. II, and verbs with full-grade root and unstressed suffix -je-. This latter category seems to have passed into a.p. I instead of the mobile accent paradigm (unlike the bhárati-verbs). The fourth conjugation (i-verbs) has its origin in two separate PIE verbal categories: the causativeiteratives in - -, infinitive in - ti (Lith. -õ-, inf. -ýti), and the essive-fientives in - -, infinitive in -eti (Lith. -i-, inf. - ti). In Sanskrit, the causative-iterative suffix -aya- alternates with -paya-, the latter after verbal stems ending in a synchronic vowel. If -aya- and -paya- are variants of the same entity, and if the p- is not a feature of the preceding verbal stem, then p- must be a prefix. The only PIE morpheme I know which fits is the preverb pe(:)-/po-, therefore the entity *-ei-e/o- must be a verb. The Hittite thematic verb iyami/iezzi can be derived from a root *(h1)ei(h1)- (variant *(h1)yeh1-), and the se a ti s do, ake a e i pe a le, as fa as the ausati e is concerned. I therefore conclude that the causative-iterative suffix is in origin an incorporated verb, added to a specific (o-infix) form of the verb. The Slavic infinitive in - ti can in principle be derived from *eiH, *iH, or it can have an analogical acute after the verbs in -ati, -eti, etc. The Lithuanian infinitive, however, can only come from *iH, the zero grade of the causative-iterative suffix *-eih1-. The Slavic present tense with circumflex - - is an irregular reduction of *-ei(h1)-e- > *-ije- > - -, while the Lithuanian form is an irregular reduction of the same suffix after -a- had been generalized as the thematic vowel (*-ei(h1)-e/o- > *-eja- > *-ea- > *-ã- > -õ-). The essive-fie ti e suffi is the lo g diphtho g *-eh1(i)-, which, according to the soundlaws laid down by Rasmussen, should have evolved to -éh1- before CC/C#, -éih1- before C, -éh1i- before V, and -h1i- (-ə1- before /t/) when unstressed. This points to a paradigm: -éih1-mi -éih1-si -éih1-ti -h1i-mós -h1i-té(s) [<ə1té(s) ] -h1i-énti, later thematized, with aorist -éh1-m, -éh1-s, -éh1-t, transferred to the infinitive as well (Slavic -eti, Lith. -eti). As was the case with the ne-verbs, Lithuanian in the present generalized the plural form in -i-, while Slavic continues the singular in -éih1-e/o- > - -, although maintaining the mobile accentuation of the PIE prototype. Finally, the fifth conjugation represents the PIE athematic verbs, mostly mobile (> a.p. III), a few with fixed stress on the root (a.p. I). Diagramatically: conjugation accent paradigm I I I I II III III IV IV V V III II II II II II I II1 III III I infinitives l-participle aorist 1st person aorist 2/3 persons II III II III presents: bhárati tudáti -ské-/-sté-dhí -né-/-n-jé'-jecaus/iter ess/fient ath. mob. ath. stat. 1 Causatives usually take the a.p. of their base word (I, II or III). Lengthened grade iteratives are a.p. I (see below). The development of Slavic verbal accentuation: the soundlaws Pedersen’s la causes barytone thematic verbs (bhárati-verbs) to become mobile. Hirt’s la draws the ictus one syllable back if the immediately pretonic syllable contains a consonantal laryngeal after the syllabic peak. This affects a.p. II verbs and some a.p. III verbs containing sequences VH (e.g. *léh1g-mi, *leh1g-mós > lez-), oRH (e.g. *poiH-láh2 > ela) and h1, provided the ictus fell on the immediately following syllable (this excludes thematic mobile forms). Hi t s la fails if the la geal follo s a e-diphthong (eRH = eRə) or if the laryngeal is *h2 or *h3 and follows a syllabic resonant (ih2/3 = iə, uh2/3 = uə, perhaps also h2/3 = irə, l 2/3 = ilə, 2/3 = iNə), as in *bhuh2-láh2 > bylá, *pih3-láh2 > pilá, *gwih3-láh2 > žilá. Winter’s la causes length and acute intonation on a vowel or diphthong preceding a PIE *b, *d, *g, *g or *gw. It had no immediate effect on the place of the ictus. Meillet’s la causes, among other things, acute intonation to be lost from the barytone forms of mobile paradigms. Meillet s law fails in closed syllables (Slaaby-Larsen’s la ). This explains the accentuation of the l-ptc. of C-verbs (nèslъ, neslá, nesló; e glъ, berglá, bergló, both by D o s la ; a d kladlъ, kla la, kla lo by the jablъko law). It also explains the accentuatio of the athe ati e to e èsmь, esí, èstь, èsmъ, esté, sotь , D o s la f o o igi al *és-mi, *é(s)-sei, *és-ti, *es-mú, *es-té, *s-entí, which failed to become enclinomenic due to Slaaby-La se s la . A u lea issue e ai s the accentuation of the acute athematic verbs (v mь, damь, mь). The eso-law causes a word with a pretonic circumflex in an open syllable to join the mobile accent pa adig . O iousl , this happe ed afte Meillet s la had ade i u fle i to atio o the first syllable a feature of the barytone forms of mobile paradigms. The working of the law is seen in theme-stressed né- and jé-verbs having a long circumflex vowel or diphthong and ending in an open syllable, such as v , , m , e , and d j , žj j , klj j , lej , s ej , sn j , o j , in part also je j and golgolj . A sté-verb like o st also seems to fit this pattern, if the syllabification was *a st . I cannot explain why the dhé-verb kl had acquired circumflex accentuation, while k a , e retained the acute. An interesting case are tudáti and jé-verbs ending in a syllabic resonant. In theory we would expect the following distribution: ‘ ‘H tudáti -i.R -iR.H jé -iR.j -i j tudáti -iR (b) -i - jé -i jo (c) -iRHj > -iRj (b) leading to: ‘ ‘H (c) That is to say, tudáti-verbs become a.p. c when they have a se -root, and remain a.p. b otherwise, while jé-verbs remain a.p. b when se , but become mobile when ani . This implies that at the time of the eso-law, laryngeals were still phonological in Proto-Slavic. In practice, we find: ani a.p. b (tudáti) jь o j tí *h1emь o tí *kenžь o ž tí *gem- se a.p. b (jé-verbs) dъ j o2 oti *dhmeHžь j o žeti *gwieh3žьr(j)o žь ti *gwerHtьr(j)o tь ti *terh1 a.p. c (jé-verbs) mь j mertí *merstь j stertí *ster-3 An apparent exception is mь o a.p. c (tudáti) žь že tí *gwerh3 nьr nertí *nerHpьr pertí *spherHklьn kl tí *klenHpьn tí *(s)penh1tьn t tí *temh1- eti (Lith. minù), probably originally an athematic verb (thus LIV). By the ablъko law the accent is retracted from theme-stressed paradigms (including infinitives) to the fi st a ute s lla le. This et a tio la ust e kept disti t f o Hi t s la , a d o ks i ci u sta es he e Hi t s la had p e iousl failed, o did ot appl , su h as i the ase of sequences *erH, * h2/3 a d a utes aused Wi te s la . E amples include presents such as j j , lj j , lj j (with *euH), infinitives such as ti, ž ti and ti (with *uh2 or *ih3) and presents such as vь g , e , egaj , etc. (with Winter acute). The phenomena traditionally attributed to Stang’s la are in my opinion best separated into at least two sets of soundlaws, one occurring before D o s la , the othe afte it. The fo e is Co o Slavic in extent, and generally results in neo-acute intonation, without lengthening of short vowels. Among the phenomena that belong here are (1) the fixing of the accent (old acute or neo-acute) to the first or last syllable of the word in compounds ( ol ъ-jь, elъ-jь, ь ь-sь, po-tò ъ, e - o ъ), (2) the retraction of non-acute stress from medial syllables in a.p. II verbs, with subsequent contrastive advancement of the stress in the 2/3 sg. of a.p. III verbs: lug-j lug-jési lug-jéti lug-jému lug-jéte lug-jánti = lъžjo > lъžješь > lъžjetь > lъžje ъ > lъžjete > lъžjotь bèresi bèreti beremú bereté berantí e ešь e etь ee ъ = bereté e tь The second set comprises post-Common Slavic phenomena, also resulting in neo-acute intonation, but usually with lengthening of short vowels, such as (1) stress retraction in volj -nouns (Cz. v le), (2) stress retraction in aje-verbs ( taje- > t - > pýta-), etc. The loss of stress on weak jers (I ši ’s la a d “ta g II . ust p o a l e pla ed so e he e i et ee “ta g I Barytone (a.p. I) verbs with non-acute intonation were subject to Dy o’s la . Old a.p. II verbs with et a tio due to “ta g s la e e ot, e ause the had isi g [ eo-acute] intonation. In an old a.p. I verb such as mogtjí, D o s la should ha e esulted i a pa adig like: ág mágesi 2 3 cf. Lith. dumiù. LIV (sub *sterh3- : Mögli he ogo ožéšь eise wird von dem slav. Verbum auch *ster- iede st e ke fo tgesetzt . mágeti mágemu mágete máganti ožétь ožé ъ ožéte ogotь As a result of the Stang II retraction, this was secondarily adapted to the standard a.p. b model, giving e.g. Czech mohu, žeš, že. The retraction must postdate the dialectal lengthening of the the ati o el, e ause i diale ts he e the st essed the ati o el e ai s sho t Štoka ia , e have žeš, like l teš, while in dialects where the stressed thematic vowel is lengthened (Old Czech) we have ožé e like vedéme. As noticed by Dybo, the distribution of je-verbs with lengthened root vowel is the following: ii > , uu > are in a.p. a (s ati, kati, s kati, t kati, s sati, skati, gati, st gati), while ee > , aa > are in a.p. b (skakati, a ati, a ati, akati, ka ati, ati). The lengthening of the root vowel in these verbs must therefore be relatively ancient, as it follows the PIE distribution, where the only long / / and / / were acute (from /iH/ and /uH/), while , and could be either acute or circumflex. In particular, the formation of these verbs must predate the development *ei > a d Meillet s la (which created new s a d s > s . The e is a atego of e s ith le gthe ed oot o el hi h could be even older than the s pati/st gati/skakati/ ati-group, at least it is claimed to be of already PIE origin by the makers of LIV. These are causative/iteratives (Slavic i/i-verbs) with a lengthened root vowel, LIV category 4b ( )-je-). If the claim were false, and these formations were of later, early (Balto-)Slavic age, one would expect these verbs (all with root vowel /a/) to fall into a.p. b, like the skakati-verbs. This is not the case. Instead, we find the verbs in question scattered over all three accent paradigms: ga iti, a iti and vaditi are a.p. a av ti, al ti and t av ti are a.p. b sa ti and gas ti are a.p. c I ot su e a out the a.p. of kaniti, a iti and mariti, but they appear to be a.p. b or c (SCr. kániti ь ). Since these verbs are not uniformly a.p. b, they cannot k , máriti, m, Russ. be lengthened /a/ s of the skakati-type (/aa/ > / /). But a lengthened vowel of PIE origin (Dehnstufe) should be reflected in Balto-Slavic as an acute long vowel (/ / > /a/), and the verbs are not uniformly a.p. a either. The pattern is in fact similar to the one we found above in the tudáti and jé-verbs with syllabic resonant, except that the starting point here was the root-stressed a.p.: roots of the structure / /, / R/ remain a.p. a, while roots of the structure / RH/ become a.p. b. We have: *g(w gh-ei 1-e- g - - * s r-ei 1-e- -- * dhh1-ei 1-e- -- as opposed to: *dh uH-ei 1-e- - - *t uH-ei 1-e- > t - - *k nh1-ei 1-e- k -- In the first set, the syllabification is *g(w -ghei-h1e-, * s -rei-h1e-, * -dhh1ei-h1e- with a long rising vowel in the first syllable, which remains in a.p. I (a). In the second set, it is *dh u-Hei-h1e-, *t u-Heih1e-, *k -h1ei-h1e-, with a long falling diphthong in the first syllable, and therefore a circumflex (cf. the circumflex in * u-jóm > * je > v je/j je, or * s > eso). When the laryngeal fell away, the semivowel/resonant was pulled to the next syllable, but the circumflex accentuation stayed. The st ess as su se ue tl ad a ed D o s la . PIE barytone je-verbs did not join the mobile a.p. as did barytone e-verbs, but instead remained in the barytone a.p. Short- o elled e s u de e t st ess ad a e e t D o s la a d the retraction + lengthening by Stang II. This can be seen in cases such as kólje- > koljé- > kó(:)lje- (Cz. k le), and presumably the other verbs in this same category (koljo (Lith. kalù, Latv. kalu), o jo (Lith. barù, Latv. a u), eljo (Lith. malù, Latv. alu), oljo ~ eljo, o jo). It can also be seen in an o-grade je-verb like sto ati, sto jo, Cz. sto u, st š. The Old Russian verb sь áti, se o follows a.p. b, which is surprising for an e-verb. We are in fact dealing with an old je-verb (cf. Slov. sérjem), with loss of palatalization of /r'/, as in a number of other cases. If the verb had been end-stressed *ker(H)-jé-, the reconstruction (which LIV leaves at *ker(H)Se -Wurzel nicht ausgeschlosse , ust e essa il e *kerH-, as a form *se -jé- would have passed into a.p. c by the eso-law. On the other hand, if the verb was barytone in PIE, a form *kér(H)-je(either se or ani ) would have developed into se jo, se jéšь D o s la the la geal is lost i this environment), and then with retraction se j o, sé : j ešь, as is indeed suggested by Czech seru, sé eš, sé e. Summary Original a.p. I II III → Dybo + Stang2 Resulting a.p. a b e.g. ogo → Hirt, jablъko eso b a c e.g. aj , vь g e.g. ь → Hirt, Slaaby-Larsen Slaaby-Larsen + Dybo c a b e.g. lez , adla e.g. nesló References Carrasquer Vidal, M. The three accent paradigms of Balto-Slavic, i : Kapo ić, M., Tones and Theories, Zagreb 2007. Dybo, V.A., 1981, Slavjanskaja akcentologija, Moscow. Dybo, V.A., Nikolaev, S.L., Zamjatina, G.I., 1990, Osnovy slavjanskoj akcentologii, Moscow. Illič-“ it č V.M., 1963, Imennaja akcentuacija v baltijskom i slavjanskom, Moscow. LIV – Lexicon der indogermanischen Verben. Die Wurzeln und ihre Primärstammbildungen, unter Leitung von Helmut Rix und der Mitarbeit vieler anderer bearbeitet von Martin Kümmel, Thomas Zehnder, Reiner Lipp, Brigitte Schirmer. Zweite, erweiterte und verbesserte Auflage bearbeitet von Martin Kümmel und Helmut Rix. Wiesbaden: Reichert, 2001. Rasmussen, J. E., 1989, Studien zur Morphophonemik der indogermanischen Grundsprache, Innsbruck. Zaliznjak, A.A., 1985, Ot praslavjanskoj akcentuacii k russkoj, Moskva.