Šiaurės vakarų dūnininkų tarmių kirtis

LIETUVIŲ KALBOTYROS KLAUSIMAI, I LIETUVOS TSR MOKSLŲ AKADEMIJA ACCENT IN THE NORTHWESTERN DŪNININKAI DIALECTS V. GRINAVECKIS I. Types of accent In Lithuanian dialects, the weakening or disappearance of accent in word endings and its occurrence at the beginning of a word or in syllables closer to the beginning (e.g.: pagaléis ~ pagalais (Šakyna), lauku,ū ~ laukų (Laukuva), malū-nus ~ malūnus (Lyduokiai)) is generally called retraction of accent. In the northwestern Dūnininkai dialects (NWD), accent is likewise retracted from the ending to the beginning of the word in certain cases. It is important to note here that in the NWD, as in other Samogitian dialects that retract accent, accent is not concentrated entirely in a single syllable but is distributed across several or all of the syllables of a word. In NWD words with retracted accent, all the syllables are accented, e.g.: aš,ės ~ ašis, dárb,ū ~ darbų, lá-u.k,ọ ~ laukų, lėd,ū ~ ledų, blėz-d,į-n.g,ą ~ blezdingą, par,ėn,õ.n.t,ę ~ pareinanti. The examples show that those syllables that are accented in the literary language (LL) also bear accent in the dialect. This accent, which occurs at the end of the word and corresponds to the accent in the LL, may be called ancient, because, as we shall see later, it is older than the other accents in the same word. Thus, “ancient accent” here means the accent as it existed before retraction. All the other accents in the examples above are generally called retracted, since, having newly arisen in syllables at the beginning of a word, they seem as if retracted from the ending. Thus, the terms themselves (ancient and retracted accent) indicate that the accents in words with retracted accent have different origins. Not all accents in the same word are pronounced with equal strength: some are pronounced more weakly, others more strongly. In words in which accent has been retracted, the first syllable is always pronounced more strongly than the other syllables of the same word; therefore, in this case the first syllable bears the primary accent. Since this primary accent, as already discussed, is1 1 The northwestern Dūnininkai dialects (NWD) occupy the southeastern part of central Samogitian, including the Šilalė and Varniai districts (with a few villages excluded), the northeastern corners of the Šilutė district and southern corner of the Priekulė district, the southeastern part of the Rietavas district, and the Paėzeris and Košiai localities in the Skaudvilė district. The settlements at the corners of the NWD area are Švėkšna, Pavandenė, Varsėdžiai, and Ž. Naumiestis. The most important phonetic feature distinguishing the NWD from other Dūnininkai dialects is the absence of affricates (č, dž < tj, dj) at the end of an unaccented word before Baltic ā, e.g.: jãute ~ jaučio, mė'dė ~ medžio, pã'tes ~ pačios, stã'tes ~ stačios, but pãč,ūos ~ pačios. 109 retracted, it will be called the primary-retracted accent. The remaining accents in words with retracted accent are pronounced more weakly and are therefore generally called secondary. However, secondary accents are not equal in strength to one another either; for example, in the word blėzd,ĩːng,ã ~ blezdingą, the accent on the final syllable, as noted above, is ancient, whereas the accent on the penultimate syllable is retracted. Moreover, these two secondary accents (as will become clear later) differ in strength, the vowels beneath them are lengthened to different degrees, and they have different intonations; they therefore need separate names as well. Since the secondary accent farther from the end of the word is retracted, the term secondary-retracted accent suits it best. The secondary accent at the end of a word is ancient, but it cannot be called a secondary-ancient accent, because in ancient times it was not secondary. The term secondary-final accent suits it best, since this secondary accent always occurs at the end of a word. In the NWD there are also words in which the accent is in the same position as in the LL, e.g.: bùte ~ būti, pavàde ~ pavādė. It was noted earlier that an accent whose position coincides with that of the LL accent will be called ancient; since in this case the ancient accent is also primary, it may be called the primary-ancient accent. In words bearing primary-ancient accent, a secondary accent also occurs in certain cases, e.g.: dẽrb,ãu ~ dirbau, vỹr,ã ~ vyrai, vãl,ĩːte ~ valyti (valyti). In this case, as the examples show, the new accent has arisen not in pre-accentual syllables but in post-accentual syllables. An accent in a dialect that arises in post-accentual syllables is always pronounced more weakly than the primary-ancient accent and is therefore secondary. To distinguish it from the other secondary accents mentioned above, it may be called the secondary-shifted accent, since it is as if shifted from the beginning of the word to final syllables or syllables closer to the end. Thus, the NWD has five types of accent: two primary accents—the primary-retracted and primary-ancient—and three secondary accents—the secondary-retracted, secondary-final, and secondary-shifted. II. The Origin and Antiquity of Stress Retraction 1. The cause of stress retraction from the ending to the beginning of the word in the dialects of northern Lithuanian (including Samogitian) is most often considered to be the influence of other languages—Latvian or Curonian and Semigallian2. The first opinion at once seems attractive because of its simplicity, but it is contradicted by the fact that Lithuanians in the Zarasai region, living alongside Latvians, do not retract stress. The second opinion has one fairly weighty argument in its favor: the area of stress retraction in Lithuanian dialects almost coincides with the area of the Curonian and Semigallian tribes that lived in the territory of Lithuania3. However, the assimilated Curonians among the Samogitians themselves did not originally have stress at the beginning of the word4, so even if Curonian influence here is acknowledged, the question remains unresolved: it must still be demonstrated what caused the Curonians to shift stress to the beginning of the word. It is thought that this type of stress pattern came into the Curonian language from 2 Archívum Philologicum (toliau APh), 4, 33, Kaunas, 1933. 3 APh, 4, 33; Die Žemaitische Mundarten (toliau ŽM), 204, Kaunas, 1930. 4 J. Endzelīns, Latviešu valodas gramatika, 32, Rīgā, 1951. 110 the Ugric-Finnic tribe of the Livonians. It is possible that the Curonians passed on to the Samogitians the retraction of stress to the beginning of the word that they had inherited from the Livonians. However, the process of stress retraction itself must have proceeded gradually, according to certain rules. Curonian influence here can perhaps be regarded only as the initial impulse that gave rise to the tendency toward stress retraction. K. Jaunius believes that stress in Lithuanian dialects was retracted independently of the influence of other languages. He reasons as follows: “usually we say that … stress ‘moved’ or ‘was transferred’ from the ending to the root syllable. But in fact stress does not move at all and is never shifted; rather, the following process takes place: as the old primary stress weakens, the stress of the preceding syllable, which has… a rising intonation, grows stronger; this intonation, as it gradually strengthens, does not change its acoustic structure, and consequently secondary-retracted (передневоричное) stress always has a circumflex intonation. The gradual weakening of the old primary stress and strengthening of the secondary stress is demonstrated by the fact that if the old stress falls on a long vowel, then, when primary-secondary stress arises in the preceding syllable, in the Darbėnai dialect it does not fall to the level of weakly stressed syllables but retains part of its former value, becoming secondary stress: cf., for example, Darbėnai gė.i.sm,iė, žmūon,ūos⁵.” In his theory, K. Jaunius demonstrates that stress was retracted from the ending to the beginning of the word gradually, initially leaving part of the stress in its former position, at the end of the word. K. Jaunius’s claim is confirmed by many Lithuanian dialects that retract stress. First of all, almost all Samogitian dialects that retract stress (except for some localities along the Latvian border: Akmenė, Vegeriai, Klykholai) still retain part of the stress at the end of the word. Western Aukštaitians, who live alongside the Samogitians and also retract stress, likewise still retain some stress at the end of the word, e.g.: dienà ~ dienà (Krakės), šàka ~ šakà, vàikai ~ vaikai (Šiauliai). These facts clearly show that the shift of stress from the ending to the beginning of the word has not come to an end in the dialects under discussion. K. Jaunius’s theory of stress retraction becomes entirely convincing when we link stress retraction with the shortening of acute endings. Prof. J. Endzelynas commented on this argument by K. Jaunius as follows: “Jaunius has demonstrated that such stress on the first syllable can also arise entirely independently of foreign influence.” Prof. J. Endzelynas adds here, in his own words: “This (stress) could have arisen quite spontaneously in Proto-Latvian, and thus stress was first retracted—from short final syllables in disyllabic and polysyllabic oxytone words to the first syllable—exactly as in some Lithuanian dialects; later, stress on the first syllable could have been generalized under the influence of foreign languages”⁶. The theory of stress retraction proposed by Jaunius and Endzelynas can be briefly formulated as follows: a) first, stress was retracted from a short ending to a long first syllable; b) later, stress on the first syllable could have been generalized (most likely under the influence of other languages). ⁵ К. Явнисъ, Памятная книжка Ковенской губернии на 1898 годъ, 183, Ковна, 1897. ⁶ J. Endzelins, min. veik., 32. 111 2. The data from the North-Western Samogitian dialects confirm the first proposition of the theory set out here: a) In disyllabic words whose first syllable is long and second short, the basic-retracting accent is pronounced more firmly, as in all the remaining disyllabic words (e.g., bã.u.g,õs ~ baugus, rỹn,į ~ rynį (ryji)). b) In trisyllabic words whose first syllable is long and the others short, in certain cases the accent is retracted without leaving any trace of accent on the other syllables (e.g., klýgdama ~ klykdama). In both these cases, vowels and diphthongs bearing the basic-retracting accent are lengthened (including a and e when they occur as the first components of diphthongs) and have a secondary-continuing intonation. From this it can be inferred that the accent here must have been retracted before stressed a and e were lengthened. c) In polysyllabic words whose final syllable is short and whose penultimate syllable is long, the accent will be retracted to the penultimate syllable at the same time as in the first two cases (a, b). This is shown by the same length of the penultimate syllable bearing a secondary-retracting accent (e.g., kẽrmv,ãːr,p,à ~ kirmvarpà). 3. The first proposition in the accent-retraction theory of Jaunius and Endzelīns is the key to determining the subsequent development of accent retraction as well. The history of accent retraction in the North-Western Samogitian dialects cannot be examined separately from that in the other Samogitian dialects, nor from that in the dialects of Eastern Lithuania; therefore, here we will also examine some more general issues concerning accent retraction in Lithuanian dialects. As a rule in Lithuanian, if an accent occurs on the syllable before the old accent, the new accent, when it falls on a long syllable, always has a circumflex or medial intonation (which is circumflex in terms of pitch movement). This is the case in the North-Western Samogitian dialects and in the northern Aukštaitian dialects that retract the accent, e.g., bãltą ~ baltà, bãltos ~ baltõs, lãuku ~ laukù (Šakyna), gýva ~ gyvà, jaũna ~ jaunà (Obeliai). In the coastal Samogitian dialect, however, things are different here: if the accent is retracted to a circumflex root, it receives a continuing intonation (~ standard Lithuanian circumflex), e.g., ũːž,ė ~ užį (pres. 1st sg., 2nd person, užti), žõːs,ės ~ žąsis (Kretinga), or a medial intonation (which is simply a shorter continuing intonation), e.g., lãuk,ã ~ laukai, naũj,ũos ~ naujõs (Salantai); if the accent is retracted to an acute root, it receives an acute intonation (~ standard Lithuanian acute), e.g., bãːlt,à ~ baltà, bãːlt,ũos ~ baltõs (sg. gen., baltà) (Kretinga); pẽːrm,ũos ~ pirmõs (sg. gen., pirmà), vėžːž,ĩːs ~ vėžỹs (Salantai). This twofold pattern of intonation in the retracted accent of the coastal Samogitians shows that they must have retracted the accent at a time when they still pronounced two different intonations in pre-accentual syllables: circumflex and acute. We thus arrive at the conclusion that the coastal Samogitians retracted the accent before the intonations of unstressed pre-accentual syllables became uniform, whereas the other Samogitians did so only after this unification. Consequently, the coastal Samogitians retracted the accent earlier, and all the others later. By the time the acute intonation of pre-accentual syllables began to turn into a circumflex, the long pre-accentual syllables of the coastal Samogitians already bore a retracted accent and therefore could no longer undergo a change of intonation, since only the intonations of unstressed syllables 112 have become uniform: pre-stress syllables have generalized the circumflex accent, while post-stress syllables have the circumflex, e.g.: kalnai, but kalnus, vėgėlė, but vėgėlės7. That the accent in the coastal Samogitian dialect was retracted earlier than in the other donininkai and dūnininkai dialects is also indicated by the greater lengthening of vowels and diphthongs under the retracted accent. For example, in disyllabic dūnininkai words whose two syllables are both long, the intonation under the retracted accent is mid, while the length of vowels and diphthongs does not differ from that in unstressed syllables. For instance, in the words lãuk,ū̃ ~ laukų, dọn-t,ũ̃ ~ dantų, the length of the diphthongs au and on in the first syllables differs in no way from the length of the same diphthongs in unstressed syllables, e.g.: laukā·ms ~ laukāms, donti·ms ~ dantims; in coastal Samogitian, however, the first component of the diphthongs is half-long in this case: lᵃ·uk,ũ̃m ~ laukų, dọ·nt,ũ̃m ~ dantų, but laukā·ms ~ laukāms, donti·ms ~ dantims (Salantai). On the basis of the statements set out above, one may conclude that accent retraction began in the northwestern part of the Samogitian dialect area and then moved southward. This conclusion is further supported by the following data from the Samogitian dialects: a) the retracted accent in the donininkai dialects is pronounced more strongly than in the dūnininkai dialects; b) within the dūnininkai area, the retracted accent weakens from north to south, until it is completely merged at the southern edge of the dialect area; c) in the donininkai dialects, especially the northern ones, word-final vowels bearing the old accent are pronounced less distinctly than in švdt, and the old (secondary-final) accent is less audible, e.g.: gràž,õs ~ gražūs, tǭr,ǭ ~ turiu (Salantai). Since the strength of the retracted accent and the lengthening of vowels and diphthongs under it indicate how long ago accent retraction took place, these data suggest that, in accordance with the theory of K. Jaunius and Endzelynas set out above, Samogitian accent was retracted not all at once, but gradually. First of all, stress is retracted from a short ending to the preceding long syllable, then from a long ending to a long root and from a short ending to a short root, and finally—from a long ending to a short root. We find these two stages of stress retraction in present-day Lithuanian dialects as well: a) from a short ending to the preceding long syllable, e.g.: malū·nus ~ malūnus, vilkus ~ vilkus (Lyduokiai), laukus ~ laukus, but laukų ~ laukų (Obeliai), dienà ~ dieną, but dienõs ~ dienos (Krakės), gᶦ·va ~ gyvà, merga ~ mergà, ā̇mu ~ einù, but miškū· ~ miškų, rā·n,k,à ~ rankà, vā·i-k,ùs ~ vaikus, but šakà ~ šaką, vaikā· ~ vaikai (Kražiai); b) from a long circumflex and a short ending to the preceding long and short syllable, e.g.: lᶦ·g,uos ~ lygios, áukšt,uos ~ aukštos, šāk,uos ~ šakos (Darbėnai), ośn̄s ~ usnīs, šāk ~ šakà (Šeduva). 4. The antiquity of stress retraction in Švdt is attested not only by the sustained or mid tone on a long syllable bearing retracted stress, but also by other dialect data. First of all, by the shortening of the ending. As already mentioned, stress is not retracted from the second and subsequent syllables from the end of the word, but in those cases, 7 Plg. K. Būga, Lietuvių kalbos žodynas, XIX, Kaunas, 1924. 8. Kalbotyros klausimai 113 kai galinis ir antras nuo žodžio galo skiemenys dėl galūnės trumpėjimo yra susilieję į vieną, tai kirtis yra atitrauktas ir iš antrojo pagal kilmę nuo žodžio galo skiemens, pvz.: grįb,əks ~ grybukas, kėrt,ęks ~ kirtikas, viėž,əks ~ vėžiukas. Čia svarbu ne tiek pats kirčio atitraukimas, kiek atitrauktiniu kirčiu kirčiuotų balsių bei dvibalsių ilgumas ir priegaidė. Jeigu kirtis yra atitrauktas iš tarmėje nesutrumpintos žodžio galūnės, kurioje senovinis kirtis yra ant trumpo balsio, tai atitrauktasis į ilgą skiemenį kirtis visada turi tęstinę priegaidę, o balsiai ir dvibalsiai, kirčiuotieji šiuo atitrauktiniu kirčiu, yra pailgėję, pvz.: grīb,əš ~ grybus, lā.u.k,əš ~ laukus, bet aukščiau minėtu atveju, kur turime visai tokį pat skiemenų ilgumo santykį (— ∪), atitrauktinis kirtis turi vidurinę priegaidę, o juo kirčiuojamieji balsiai ir dvibalsiai yra tokio pat ilgumo, kaip ir nekirčiuotame skiemenyje (pvz.: grībō.ka ~ grybuką, kėrtē.ka ~ kirtiką, viėžčē.ms ~ vėžiams). Ką rodo šis atitrauktinio kirčio priegaidės ir balsių bei dvibalsių pailgėjimo nevienodumas žodžiuose su vienodu skiemenų ilgumo santykiu (— ∪)? Paėmę pavyzdžius su abiem ilgais skiemenimis, matome, kad čia turime visiškai tokį pat atitrauktiniu kirčiu kirčiuotų skiemenų ilgumą ir priegaidę, kaip ir nagrinėjamuose pavyzdžiuose su nutrupėjusia galūne (pvz.: lāuk,ū ~ laukų ir lāuk,ōks ~ laukiukas; kīm,ū ~ kiemų ir kīm,ōks ~ kiemūkas). Tiek pagrindinio-atitrauktinio, tiek ir šalutinio-galinio (senovinio) kirčio stiprumas šiuose pavyzdžiuose, nepaisant nevienodo skiemenų ilgumo santykio, visiškai sutampa. Iš viso to, kas čia pasakyta, paaiškėja, jog tuo metu, kai kirtis kėlėsi iš trumpo galinio skiemens į prieš einantį ilgą skiemenį, galinis skiemuo dar nebuvo susiliejęs į vieną skiemenį su prieš juo einančiuoju (kitaip sakant, laukiukas dar nebuvo išvirtęs į laukus). Jeigu kirčio kėlimosi iš trumpo galinio skiemens į prieš einantį ilgą skiemenį metu būtų jau buvęs laukus, tai šiandien jis būtų tariamas lā.u.k,ōks, visai taip pat kaip ir lā.u.k,əš ~ laukus. Vadinasi, galūnė sutrumpėjo jau po pirmojo kirčio kėlimosi iš galūnės į žodžio pradžią etapo. Užėjus antrajam kirčio kėlimosi į žodžio pradžią etapui, galūnė jau bus buvusi sutrumpėjusi, vadinasi, kirtis buvo atitrauktas vienu metu žodžiuose laukų ir laukiuks. Tuo pat metu kirtis bus atitrauktas ir iš trumpo galinio skiemens į trumpą prieš einantį skiemenį tiek iš tarmėje nesutrumpėjusio galinio skiemens, tiek ir iš naujai susiformavusio iš dviejų žodžio galo skiemenų dėl galūnės trumpėjimo (pvz.: šāk,à ~ šaką, klāb,āts ~ klebatas). Tai rodo atitrauktinio ir šalutinio-galinio (senovinio) kirčio stiprumo laipsnis ir kitų tarmių duomenys. 5. K. Būga writes about the antiquity of stress retraction: “The vowels a, e, i, u, which, when stressed, become half-long vowels in the Samogitian and eastern Aukštaitian (northern) dialects, are pronounced briefly by Lithuanians under retracted stress: a. šàkō, šnėkō, l’ęgō (Pnv.), l’og Dvd., gòdù || Samogitian šàkà, šnėkō ‘I speak’, lėgà, gòdù Skd. gudīn Mžk.” The shortness of the vowel shows that stress was retracted here quite recently68. However, in Samogitian dialects, during the lengthening of short vowels under stress, retracted stress (which was, of course, still weaker than it is now) may already have been present, but because of its weakness it may not have lengthened the short vowels. This is shown by 8 K. Būga, cited work, LI. 114 the fact that even under primary-original stress short vowels (e in most Dūnininkai dialects that retract stress, and a, e, ẹ, o in the northern Dounininkai dialects) are not lengthened in cases where secondary-postposed stress has arisen and weakened the primary-original stress. The shortness of the vowel only shows that stress shifted onto it later than from the short ending to a long syllable, where the short vowels forming the components of diphthongs have been lengthened. Consequently, in long syllables to which stress was retracted from a short syllable, retracted stress had already become primary during the lengthening of short vowels under stress. In short syllables, retracted stress may have been present (weakly) during the lengthening of short vowels, or it may not have been present. Vowel length does not show this. To demonstrate the latest shifting of stress from a long final syllable to the short penultimate syllable, one can rely only on the weakest realization of this retracted stress and on data from other pronunciations. In polysyllabic words, retracted stress may be thought to have arisen in the third and subsequent syllables from the end later than during the first stage of stress retraction. This is indicated by the circumflex accent in long syllables and by vowels and diphthongs that have not been lengthened. 6. Stress drawn away from the end of a word is probably, in general, inextricably connected with the weakening, disappearance, and shortening of the ending. This is shown first of all by the fact that dialects that retract stress also shorten endings; moreover, dialects whose stress is retracted farther toward the beginning of the word shorten endings more (e.g., Mūsų upynas), while those whose stress is retracted less have also shortened their endings less (e.g., Šiauliai, Krakės). This partly explains why stress first disappears from short endings. After all, in dialects it is precisely short endings that first weaken and disappear, and have disappeared. The fact that stress shifts from the ending toward the beginning of the word is connected with the disappearance of the ending is also shown by the fact that stress has shifted toward the beginning of the word only from an ending. Stress has not shifted from a non-final syllable, even a short one, to the preceding long syllable, e.g.: druobó.le ~ drobùle, augė.na ~ augino. In K. Jaunius’s argument about stress retraction cited above, the following statement is given: “depending on the weakening of the old primary stress, the stress of the preceding syllable grows stronger.” The weakening of word-final stress is clearly shown by the shortening of acute endings. As we know, some acute endings have shortened, while others have changed their acute accent to circumflex. Acute endings have not shortened equally in all Lithuanian dialects. Western and Central Aukštaitians, Dzūkai, much of the eastern dialect area, and southeastern Samogitians have short vowels instead of long acute endings, as in the standard language, e.g.: baltà (<*baltā, cf. baltóji), sausà (<*sausā), visà (<*visā). Most Samogitian Dounininkai, as well as the northern Dūnininkai, pronounce final vowels derived from long acute vowels more weakly, with some reduction, e.g.: bá.lt,á ~ baltà, rõ.n.kã ~ rankà, sã.u.s,á ~ sausà (Kretinga). Among Samogitian Donininkai these vowels are barely audible, e.g.: sã.us ~ sausà, rãs ~ rasà, ʒẽs ~ visà (Priekulė). The eastern pontininkai around Šeduva no longer have acute final vowels at all, e.g.: bált ~ baltà, rãs ~ rasà, šák ~ šakà. 115 Thus, the shortening of long acute endings (stressed and unstressed) is common to the entire Lithuanian language. In some Lithuanian dialects, a long acute final vowel, once it had become short, did not cease shortening but continued to disappear, so that baltà > bá·l̥t,a (Kretinga). Here, in K. Jaunius’s words, as the old final stress weakened, the stress of the preceding syllable grew stronger. Left unstressed, the vowel had to disappear even more rapidly, and thus from baltà only balt remained (Šeduva). Consequently, the shortening of acute endings must have contributed to the loss of stress at the end of the word; in other words, as they shortened, stressed acute endings caused the stress to retract toward the beginning of the word. Long syllables are pronounced with greater force, and a syllable pronounced with greater force is, after all, called a stressed syllable in a word. Since a long syllable is pronounced more forcefully, stress could pass more easily from a short syllable to a long one (e.g., baltà > bā·l̥t,ā, than from a short syllable to a short one (e.g., šakà > šàk,à). With this in mind, it is easy to understand why stress was first retracted from a short ending to the preceding long syllable, and only later from a short ending to a short syllable and from a long ending to the preceding long syllable, and, latest of all, from a long ending to the preceding short syllable. When discussing the later retraction of stress (from a short ending to a short syllable and from a long ending to the preceding long syllable, as well as from a long ending to the preceding short syllable), one must also bear in mind the analogy of the earlier stress retraction (from a short ending to the preceding long syllable), which, undoubtedly, must also have contributed to the generalization of stress on the first syllable9. Here it is also necessary to note the effect of the shortening and disappearance of unstressed endings on stressed ones. After all, unstressed endings undoubtedly shortened more rapidly, while stressed ones did so more slowly. Dialect data also show this; for example, among the Kretinga Dounininkai, the vowels of unstressed acute endings have been reduced more than those of stressed ones (e.g., bá·l̥t,a ~ baltà and šārk,a ~ šárka). The more rapid disappearance of unstressed endings could also have accelerated the shortening of stressed ones and thereby contributed to stress retraction. Consequently, the shortening of acute endings must have been one of the most important factors conditioning stress retraction. The Curonians must have left the Samogitians only the tendency toward stress retraction itself. Some Lithuanian dialects that do not retract stress convincingly show how the pronunciation of the second syllable from the end of the word has grown stronger when the final syllable is short. For example, around Ignalina, one says vienà, but ve.nõ˙s, dienà, but de.nõ˙s. What does the diphthong’s failure to shorten show when the ending is short? Clearly, K. Jaunius’s law is at work here too. When the vowel of the stressed ending shortened and its stress consequently weakened, the pronunciation of the preceding long syllable grew stronger, so ie did not monophthongize. In all other cases, ie, being unstressed, has monophthongized and shortened in the dialect under discussion. III. The Antiquity of Stress Retraction Secondary retracted stress in the Švdt dialect, as in other Samogitian dialects, occurs only in long syllables (if it has arisen on a short vowel, the latter has usually lengthened). To explain the emergence of this type of stress, the previously 9 Plg. J. Endzelīns, min. veik., 32. 116 the aforementioned K, Jaunius’s theory concerning the emergence of retracted stress; in the case under discussion, however, the process of stress shifting from its position proceeded in the opposite direction from retraction. Retracted stress, as we know, arose in pretonic syllables (that is, stress shifted from the end of the word to its beginning: ←—), whereas transferred stress, by contrast, arose in posttonic syllables (here stress shifted from the beginning of the word to the end: —→). It may be assumed that transferred stress, like retracted stress, first arose in those words in which the primary ancient stress fell on short syllables. This is indicated by the fact that in this case secondary-transferred stress is stronger, and the vowels and diphthongs bearing it are more lengthened than in words where the ancient stress falls on long syllables, e.g.: (1) lēp̄ām ~ lipama (we are licking), d̄ē̄sės ~ didžiasis, rē̄zd̄āms ~ risdamas, but (2) nē̄š̄t̄ī ~ nešant, auḡām ~ augama (we are growing), d̄ūb̄ēls ~ dōbilas. In the first case (1), the length of vowels and diphthongs stressed with secondary transferred stress is the same as that of vowels and diphthongs stressed with retracted stress, where stress has been retracted from a short to a long syllable; in the second case (2), the vowels are as long as unstressed vowels, while the first element of diphthongs is slightly shorter than that of diphthongs stressed with primary ancient stress and bearing the broken tone, but longer than half-long vowels. Other Samogitian dialects help establish the antiquity of the emergence of secondary transferred stress. In Dounininkai, words with secondary transferred stress in which the ancient stress falls on a, e are pronounced as follows: nākt̄ė̄i ~ naktie (for the night), vėst̄ė̄s ~ vesties (to lead) (Kretinga), whereas in the northwestern Samogitian dialect (švdt) the same words are pronounced nākt̄į̄ ~ naktie, vēst̄į̄s ~ vesties. Thus, in Dounininkai, the a, e pronounced here with primary ancient stress are short, whereas in švdt they are long. On this basis, it may be assumed that secondary transferred stress in Dounininkai arose before short a, e lengthened under stress. Once secondary transferred stress arose, the primary ancient stress, weakened as a result, could no longer lengthen these vowels. In švdt, secondary transferred stress must have arisen only after a, e had lengthened. The vowels ę, ǫ (~ į, u) are lengthened under primary ancient stress in both Dounininkai and švdt (e.g., kė̄ta ~ kitą, s̄ōva ~ siūvo), but in words with secondary transferred stress these vowels have remained short (e.g., švdt: stė̄pr̄ē ~ stipriai, sōkr̄ē ~ šūkriai; Kretinga: lėp̄ām ~ lipama (we are licking), b̄ūv̄uom ~ būvoma (we were)). Thus, secondary transferred stress in both Dounininkai and švdt must have arisen before ę, ǫ lengthened. The lengthening of vowels a, e under ancient primary stress in words with secondary transferred stress, and the unequal strength of these stresses across the švdt area (as well as in Dounininkai dialects), indicate that transferred stress, like retracted stress, first arose in the northern Samogitian dialects and later spread to the southern ones. STRESS IN THE NORTHWESTERN DOUNININKAI DIALECTS V. Z. GRINAVECKIS Summary The dialects of the Northwestern Dounininkai (NW Dounininkai) differ from those of other Dounininkai in that they do not have affricates in unstressed endings before ancient long ā. In NW Dounininkai, as in other Samogitian dialects in which stress shifts from the end of a word to its beginning, there are five distinct types of stress: primary ancient (vėsims ~ visiems), primary retracted (ā.u.kst,à ~ aukštà), secondary retracted (blėzd,įn.g,à ~ blezdingà), secondary transferred (vīr,ā ~ výrai), and secondary final (àc,ā.r.g,à ~ atsargà). Stress types are determined by the origin of the stress and the force with which the stressed syllable is pronounced. Two stages of stress retraction can be traced. Initially, stress retraction occurred only in words with a short or historically long strong-initial ending, in which the preceding syllable is long; later, it occurred in all other cases. One of the most important factors that led to the disappearance of stress at the end of a word was the reduction of endings (especially strong-initial ones). Stress retraction began in the northwestern part of the area where Samogitian dialects are spoken and spread from there southward and eastward. This is indicated by the fact that among the coastal Samogitians, retracted stress retained the ancient strong-initial accent (bál.t,à ~ baltà, Kretinga). This means that the coastal Samogitians shifted stress to the beginning of the word at a time when both strong-initial and strong-final intonations were still pronounced in pretonic long syllables. The other Samogitians shifted stress to the beginning of the word only after the strong-final intonation had become generalized in pretonic long syllables (bā.l.t,à ~ baltà, Laukuva). The transfer of stress to posttonic syllables also began in the northwestern Samogitian dialects. This is evidenced by the length of vowels a, e under primary ancient stress in words with secondary transferred stress (vā.k,ā.rs ~ vākaras, Kretinga, and vā.k,ā.rs ~ vākaras, Laukuva).