LIETUVIŲ AREALINĖS LINGVISTIKOS KLAUSIMAI LIETUVOS TSR MOKSLŲ AKADEMIJA

K. GARŠVA

THE RELATIONSHIP BETWEEN ACCENTUATION AND VOCALISM IN THE NORTHWESTERN PANEVEZYS DIALECT

In many languages, vowels form the nucleus of the syllable, and they cannot be described accurately without first discussing stress, because some phonetic properties of an utterance are “superimposed” on a syllable, morpheme, or entire word (e.g., stress), while others belong to individual minimal sound units (phonemes and their variants). The four sections of this article (1. Word stress, 2. Vowel quantity, 3. Intonations, 4. Vowel phonemes and their variants) attempt to distinguish synchronically the suprasegmental (prosodic¹) and segmental phenomena of one East Aukštaitian dialect—the Northwestern Panevezhian—and to determine the functions they perform in language². The corresponding expressive means of other Lithuanian dialects (and of Standard Lithuanian) should be explained similarly. The article touches on more complex theoretical problems in phonology, about some of which linguists’ views differ greatly, and not all our conclusions are yet definitive.

The Northwestern Panevezhians are distinguished from neighboring dialects by three principal phonetic features³:

1. Characteristic lowering of the vowels i, u in at least three positions (see below), and the rendering of Standard Lithuanian vowels ę, ą and circumflex diphthongs ēm, ēn, ām, ān as the sounds ė, o and ēm, ēn, ōm, ōn⁴;

2. The so-called retraction of stress in all cases;

3. Reduction of former short vowel endings.

This area corresponds much more closely to A. Baranauskas’s “Second Easterners” and to the traditional dialects of the Mūša basin than to the Northern Panevezhians (according to

¹ In Lithuanian linguistics, two variants of the term are used in the same sense: prozodija and prosodija; cf. Tarptautinių žodžių žodynas, Vilnius, 1969, pp. 628, 624, etc. The latter should be preferred because this is how the word was used in its original source, Greek (in more recent borrowings from Greek, the consonant s is generally no longer replaced, e.g., Odisėjas, Poseidonas, etc.).

² For a similar analysis, see: Вопросы теории лингвистической географии, ed. R. I. Avanesov, Moscow, 1962, pp. 28–63; T. Hill, Phonemic and Prosodic Analysis in Linguistic Geography, “Orbis,” XII, No. 2, 1963, pp. 449–455. — Reprinted in: Readings in American Dialectology, New York, 1971, etc.

³ Since the printing house does not have all the necessary symbols, the article uses not a phonetic but an almost phonological transcription. (Phonetic transcription symbols are used only where phonetics is discussed specifically.) For a consistent phonetic transcription and the acoustic and articulatory properties of the sounds, see: K. Garšva, Šiaurės vakarų panevėžiškių prozodija ir vokalizmas, summary of diploma thesis (typewritten, Department of Lithuanian, VVU), 1973, pp. 24–25 (VVU Lithuanian Studies Reading Room, D 434).

⁴ When stressed with retracted stress, these diphthongs are the same as when stressed with primary stress; cf.: rɔ̀nkos “hands” — rɔ̀nkos “hands,” but the Pantininkai form lāngs “window (windows)” — lāngė “window” (not lɔ̀ngė), rānds — rāndė, žānds — žāndė, etc.

76

Girdenis–Zinkevičius’s classification). Therefore, the Northern Panevėžiškiai still need to be divided into Northwestern and Northeastern Panevėžiškiai (Biržiškiai). The eastern boundary of the Northwestern Panevėžiškiai dialect would roughly correspond to the pontininkai–puntininkai boundary in the “Atlas of the Lithuanian Language,” except that the fringe where the puntininkai feature begins should also extend as far as the Mūša (Namajūnai, Saločiai, Pasvalys). The acute diphthongs ám, án, which the pontininkai render as óm, ón, occur rarely in speech (moreover, perhaps more than half of them are exceptions⁵), and no other significant phonetic differences have been observed between the northern so-called pantininkai and the pontininkai (the Northwestern Panevėžiškiai).

1. WORD STRESS

Stress is most often defined as the highlighting of one (stressed) syllable in a word and its contrast with the other (unstressed) syllables. But separate monosyllabic words, or a sequence of them, can also be stressed⁶, so a contrast between stressed and unstressed syllables is not essential to the perception of stress (just as vowel quantity is not determined solely by comparison with adjacent vowels). It is impossible to pronounce an isolated monosyllabic word containing a long vowel (diphthong) without a pitch accent, and thus without stress, since, as many linguists believe, Lithuanian pitch accents are clearly differentiated only in stressed long syllables⁷ (in our view, this is the strong position for pitch accents).

The sentence màm kù.l’ grù.ds laũk’ “mother threshes the grain in the field” shows that monosyllabic words can also be stressed, and can contain vowels of varying quantity (short, half-long, long) and diphthongs. Thus, the duration (quantity) and fundamental frequency (pitch) of Lithuanian vowels, as well as the quantitative reduction of unstressed vowels in the Panevėžiškiai and Širvintiškiai dialects and the qualitative reduction of some diphthongs, would be only supplementary cues to stress, while the primary cue is sound intensity (loudness). The vowels of stressed and unstressed syllables differ somewhat in their place of articulation and, consequently, in their acoustic quality.

In works on Lithuanian linguistics, alongside the acute and circumflex pitch accents, “middle” and “cut” accents are still often mentioned; these can occur not only in long but also in half-long stressed syllables, as can the “short pitch accent,” which occurs only in short syllables. But in the case of the so-called short (and middle, cut) “pitch accents,” what is often being discussed is not a pitch accent—emphasis on the beginning or end of a syllable and a rapid or gradual change in intensity—but only the shortness or half-length of the syllable; therefore, the pitch-

⁵ Žr. Z. Zinkevičius, Lietuvių dialektologija, Vilnius, 1966, p. 97—98. Dar plg. pastabą p. 16 (5 išnaša).

⁶ In Russian, this view was defended by L. Shcherba and L. Zinder; see Л. Р. Зиндер, Общая фонетика, Ленинград, 1960, p. 298—300. For an attempt to criticize it, see А. А. Реформатский, Фонологические этюды, Москва, 1975, p. 39—41, 63—64. Cf. also Р. Якобсон и М. Халле, Фонология и её отношение к фонетике, — Новое в лингвистике, вып. 2, Москва, 1962, p. 250 et seq.

⁷ Н. С. Трубецкой, Основы фонологии, Москва, 1960, p. 223; P. Garde, L’accent, Paris, 1968, p. 161; tą patį žr. P. Garde’as, Lietuvių kalbos kirčiavimo sistema (išvertė ir komentavo A. Girdenis), „Kalbotyra“, XXII(1), 1971, p. 94—96; J. Kazlauskas, Lietuvių kalbos istorinė gramatika, Vilnius, 1968, p. 5; M. J. Kenstowicz, Lithuanian Phonology, University of Illinois, 1969 (daktaro disertacija, rotaprintas), p. 84; О. С. Степанов, Основы общего языкознания, Москва, 1975, p. 99 ir kt.

77

The concept of pitch accent is therefore unnecessary here⁸. We mark stressed short and half-long syllables, which have no pitch accent, with the stress (grāvjo) mark.

R. Ekblom (1917 and later) was the first to begin demonstrating the “short pitch accent” in Lithuanian solely through experiments, followed by N. van Wijk, J. Gerulis, A. Laigonaitė, and V. Vaitkevičiūtė⁹. Short vowels were studied in varying phonetic contexts, without the use of mathematical statistics; consequently, some researchers consider the “short pitch accent” in stressed syllables to resemble both the acute and circumflex pitch accents (in different syllables)¹⁰, while others consider it to resemble only the circumflex pitch accent¹¹. Only what can be heard by an ordinary, unaided ear has linguistic significance; instruments reveal only objective physical properties of sound, and from these it is impossible to determine how the ear synthesizes all those properties¹².

Linguists who consider Lithuanian a moraic language also deny the existence of pitch accents on short vowels, and maintain that such vowels can only be stressed or unstressed¹³. The related Latvian language likewise has no “short pitch accent.”

In all Aukštaitian dialects, stress can occur in any position: to some extent, it is independent of a syllable’s position in the word and its open or closed status, as well as of the quality, quantity, and pitch accent of the vowel (diphthong) that forms the syllable. In disyllabic words, stress can produce all nine possible combinations of contrasts between short, half-long, and long syllables: I 1) sàk’əs “I would say” — sak’əs “I would say” [conditional], 2) and sàːk’əs — sak’əːs; II 3) and sàk’əs — sak’əːs, 4) i.šnėːs “he/she takes out” — išnėːs “he/she took out”; III 5) nom. sg. žãbos “of the žãbo” (dial. žabõs) — fut. 3rd person žabòːs “of the žabõ” (žabóti), 6) vėːžam “we carry” — vežàm “carried” [feminine participle], 7) rù.do “ore” — rudóː “brown” [feminine], 8) and vėːžam — vežàːm; IV 9) žãːbos “stakes” — žabòːs “of the žabõ.”

As these examples show, four types of opposition are distinguished in the dialect according to stress placement: I, words differ only in stress placement (when both syllables are short or both are half-long); II, words differ not only in stress placement but also in the length of the stressed vowels (when one syllable is short and the other half-long); III, words differ not only in stress placement and vowel length but also in the presence or absence of a pitch accent in stressed syllables (when one syllable is long and the other short or half-long); IV, words differ not only in stress placement but also in the pitch accents of the stressed vowels (when both syllables are long).

In monosyllabic words (in the weak position), there is no contrast based on stress placement (a phonological cue); stress exists here only phonetically, whereas in polysyllabic words (in a syntagma), stress can perform meaning-distinguishing (distinctive-

⁸ Plg. K. Būga, Kirčio ir priegaidės mokslas (1924), — Rinktiniai raštai, III, Vilnius, 1961, p. 21; J. Kazlauskas, min. veik., p. 5; A. Girdenis, V. Žulys, „Trumpinė priegaidė“, „Kalbotyra“, XV, 1967, p. 113—116; A. Pakerys, Apie kirtį, „Kalbotyra“, XVII, 1967, p. 131 ir kt.

⁹For the first four authors, see: A. Laigonaitė, Literatūrinės lietuvių kalbos kirčiavimas, Vilnius, 1959, p. 8, 10; Lietuvių kalbos gramatika, I, Vilnius, 1965, p. 128 et seq.

¹⁰A. Laigonaitė, cited work, p. 10.

¹¹ V. Grinaveckis, Žemaičių tarmių istorija, Vilnius, 1973, p. 98.

¹²Cf. I. Lehiste, Suprasegmentals, Cambridge, Massachusetts and London (M. I. T.), 1970, p. 5 et seq. — E. Fischer-Jørgensen and others believe that acoustic techniques are needed primarily to establish phonetically the speech features already identified by listening: E. Fischer—Jørgensen, What can the new techniques of acoustic phonetics contribute to linguistics?—Reports for the VIII International congress of linguists, I, Oslo, 1957, p. 88; Ю. С. Степанов, О зависимости понятия фонемы от понятия слога при синхронном описании и исторической реконструкции, „Вопросы языкознания“, 1974, № 5, p. 98; cf. also A. Girdenis, V. Žulys, cited article, p. 115.

¹³H. C. Трубецкой, cited work, p. 232; M. J. Kenstowicz, cited work, p. 84.

78

functions¹⁴. There is no independent word without phonemes or stress—the latter being a sign of the word’s independence. Nor do phonemes without stress or stress without phonemes perform their functions, although in both cases the primary significance belongs to the phonemes.

The function of distinguishing meanings in phonology must be understood more broadly. If a phoneme or the position of stress or accent is changed, the meaning of a word usually does not change, but the word may be distorted to the point of becoming completely unrecognizable. For example, on hearing the artificial word “samānos,” without context we would not understand what was meant: sąmanos or sama- nos. In words that can be stressed in two ways, the distinctive function of stress is neutralized, e.g.: noun pāk̃rate – noun and verb pakratė “you shook,” locative plural laukùos’ / laũkās’ “in the fields,” dative plural vaikám / vaikam “to the children,” etc.

The same stressed and unstressed vowels (diphthongs) are not separate phonemes (as the linguists of the Prague Linguistic Circle thought), since they always occur in different positions, and the corresponding homonyms (at the segmental level) are distinguished not by changing a phoneme, but by the position of stress¹⁵. Stress belongs to a unit longer than a phoneme—the syllable; that is, it belongs not to the segmental level, but to the suprasegmental level. Although the number of phonemes in the same context may be considerable (e.g.: sàk’ǫs “I would say,” sòk’ǫs “I would wind,” sù.k’ǫs “you will wind,” siek’ǫs “I would seek”), there could be no more stress prosodemes than there are syllables in a word (although in the Aukštaitian dialect, even in polysyllabic words, significant stress usually occurs in no more than two syllables). The inventory of prosodemes would include “stress on the first syllable of the word,” “stress on the second syllable,” “stress on the third syllable,” and so on. Since disyllabic and trisyllabic words are most numerous in the language, and the northwestern Panevėžys speakers most often retracted stress from the final syllable to the first or second syllable, the opposition between stress on the first and second syllables is the most frequent according to stress position in this dialect; less often there is an opposition between one of these and stress on the third syllable. (The number of prosodemes for the first syllables of words in the dialect increased at the expense of stress prosodemes for the second, third, and other syllables.) In nouns, where stress is usually retracted to the same syllable that bears primary stress, stress distinguishes only a small number of homophones belonging to different words, but no longer distinguishes forms of the same word, e.g.: dar̃žinės “granaries” – dar̃žinės’ “granaries.”

¹⁴ Some linguists believe that the primary function of stress in languages with free stress is distinctive (1); others, that it is culminative (2); and still others, that it is merely contrastive (or culminative) (3), e.g.: 1) П. С. Кузнецов, К вопросу о фонологии ударения (1948), — А. А. Реформатский, Из истории отечественной фонологии, Москва, 1970, p. 360, 366; Л. Р. Зиндер, min. veik., p. 299, 302–304; Р. И. Аванесов, Русская литературная и диалектная фонетика, Москва, 1974, p. 85, 100; А. А. Реформатский, min. veik., p. 31; P. Kratochvil, Tone in Chinese (1968), — ed. E. C. Fudge, Phonology, Penguin Books, 1973, p. 347, 348; I. Lehiste, min. veik., p. 147, 148; 2) С. К. Шаумян, Проблемы теоретической фонологии, Москва, 1962, p. 59; H. Pilch, Phonemtheorie, I, Basel–New–York, 1964, p. 101–102; in part also cf. Н. С. Трубецкой, min. veik., p. 38, 230–231, 238; 3) A. Martinet, Accent et tons, Miscellanea Phonetica, II, London, 1954, p. 13–24; P. Garde, Principles of the Synchronic Description of Stress (1967), — ed. E. C. Fudge, Phonology, 1973, p. 310–311.

¹⁵ R. Avanesov, V. Vaitkevičiūtė, and others prove the same point in a different way. According to them, for example, the words sàk’ǫs and sak’ǫs, which differ only in stress position, differ segmentally in two respects at once: not only in that the vowel a in the first word is stressed and in the second is unstressed, but also in that ǫ in the first word is unstressed and in the second is stressed; see: Р. И. Аванесов, min. veik., p. 101–102; V. Vaitkevičiūtė, Lietuvių literatūrinės kalbos balsinės ir dvibalsinės fonemos, LKK, IV, 1961, p. 21. But this analysis is possible only if we regard stress as a feature of individual vowel phonemes; cf.: П. С. Кузнецов, min. str., p. 361-362.

79

2. VOWEL QUANTITY

The dialect clearly distinguishes long, half-long, and short vowels (there is a three-grade length system), but any one vowel can have no more than two phonological degrees of length. In stressed (strong) position, high and low vowels form an opposition in quantity (I), whereas mid vowels do not form an opposition (II):

I 1. The vowels iˑ, iˑ, uˑ, u. occur as long and half-long (see examples below);

2. The vowels eˑ, e, (æˑ, æ), aˑ, a occur as long and short;

II 3. When stressed, the vowels eˑ (ɛˑ), oˑ occur only as long;

4. The vowels ɛ (ə), ɔ occur only as short.

The place of the short vowels ɛ, e, ɔ, a in the vowel system can be explained in two, or even three, ways: 1) on the basis of qualitative similarity, they could be linked with the long vowels eˑ, e, oˑ, aˑ; 2) on the basis of alternation within the same morpheme with the vowels iˑ, eˑ, uˑ, aˑ, the sounds ɛ, e, ɔ, a could be regarded as positional (retracted-stress) variants of these vowels; 3) or the short vowels ɛ, ɔ could be assigned to neither group and regarded as independent phonemes.

The vowels ɛ, ɔ alternate within the same morpheme not with the vowels eˑ, oˑ (as the long and short e, a alternate with one another), but with the vowels iˑ, uˑ; cf.: s’kėˑr’ “I distinguish, distinguish” – s’kiˑr’ “he/she distinguishes,” s’kėˑr’ “skėri”; kòˑr’ “I create, create” – kùˑr’ “he/she creates,” kõˑr’ “kõri”; therefore, only the vowels e, a can be linked with the long vowels.

The short vowels ɛ, e, ɔ, a could be regarded as positional variants of the vowels iˑ, eˑ, uˑ, aˑ only if we introduced into the dialect a category of two stress types¹⁶, largely corresponding to the diachronic category of primary and retracted stress. But, reasoning synchronically, we “do not know” which stress is primary and which is retracted, and, as auditory experiments show, these stresses are now barely distinguishable by ear; the differences observed can be marked solely by accents. Moreover, not only long but also short (half-long) vowels can bear either primary or retracted stress. In the position of primary stress, the half-long vowels iˑ, uˑ may occur in any syllable of the word, whereas the short stressed vowels e, e, o, a remain mostly in the final syllable¹⁷, but also occur in syllables farther from the end of the word (not only in verbs derived from onomatopoeic words, cf. trėp’telėje “he/she stamped,” or in words borrowed from other languages, cf. làt’vej “Latvia,” but also in some dialect areas—in the comparative adjective suffix -ėsnis, -ė, etc.). If short vowels under primary stress mostly remain only in the final syllable, under retracted stress these vowels occur not only in the final syllable but also in the penultimate syllable, while in syllables farther from the end their shortness is also often not preserved.

The decisive evidence concerning the status of the vowels ɛ, ɔ is that the unstressed vowels i – e, u – ɔ are free variants¹⁸, and therefore also in the stressed position (in which

¹⁶ A. Girdenis, V. Žulys, rec. J. Kazlauskas, Lietuvių kalbos istorinė gramatika, Vilnius, 1968, 414 p., „Baltistica“, VIII(2), 1972, p. 199.

¹⁷ For more examples, see E. Grinaveckienė, “Some Features of Lithuanian Dialects,” LKK, V, pp. 148–150.

¹⁸ In most of the dialect area, the vowels i, u in any pre-stress or non-final post-stress syllable (whether closed or open), before any vowel (whether at the absolute or non-absolute beginning of a word), may be either widened or unwidened, even within the same word (and as pronounced by the same person).

80

Since the opposition between these vowels is no longer neutralized, the vowels e, o must be regarded as short variants of the vowels i, u. The vowels e, a, like the long vowels e˙, a˙, are independent phonemes.

In the dialect, in the same stressed position, there is an opposition between long—half-long vowels and long—short vowels (neutralized only in a final open syllable), and vowel quantity would be a distinctive feature belonging to the phonemes¹⁹. The vowels that can be stressed only when long (e˙, o˙) or only when short (e, o) do not form a correlation of length versus shortness. (On the basis of only the vowels e˙, o˙, which shorten in an unstressed syllable, quantity could also be assigned to the suprasegmental level.)

When long—half-long i˙, i., u˙, u. are contrasted, minimal pairs distinguish words with different meanings, e.g.: ski˙la ‘hole’ — ski.la ‘split’, mu˙š ‘the Mūša’ — mu.š ‘hits’, etc.; thus, when stressed, these vowels retain the same quantity. The long and short phonemes e˙, e, a˙, a, by contrast, most often distinguish forms of the same word (of which there are the most in the language), and much less often different words, e.g.: kė˙l’ ‘he/she lifts, you lift, a road’ — kėl’ ‘with a road, roads’.

When stressed, all short vowels can occasionally be lengthened in the same position to almost half-long, especially when the syllable is emphasized intonationally, but this does not change the word’s meaning. Such lengthened vowels may be regarded as optional variants of short phonemes, sometimes also having stylistic significance. (Similarly, formerly long vowels i˙, e˙, u˙, o˙, when very rarely pronounced half-long rather than short in an unstressed syllable, are optional variants of short positional forms, e.g.: gi.vy˙bės ‘life (genitive)’; bu˙vo.m ‘we were’, etc.)

In unstressed syllables, all the dialect’s vowels are generally short only: the opposition in quantity is neutralized.

In linguistics, the long and short vowels of Lithuanian have hitherto been explained in two ways: some linguists regard them as separate phonemes, while others hold that long vowels consist of doubled sequences of two identical short phonemes (moras), and thus assign quantity differences not to the vowels but to the syllable (regarding them chiefly as prosodic features). The first view is found in nearly all linguistic works written in Lithuania (although their authors, with the exception of V. Vaitkevičiūtė, A. Girdenis, A. Pakerys, and others, do not yet discuss phonemes); the second view has been held by Western structuralists since the time of N. Trubetzkoy²⁰.

The claim that long vowels consist of two parts is supported by the fact that in Lithuanian two stress types (pitch accents) are distinguished by quality not only in diphthongs and mixed diphthongs, but also in long vowels: the beginning and end of the syllable-forming long sound differ in a phonologically significant way. For this reason, long vowels, like diph-

¹⁹ Some propose using the concept of tenseness to encompass both qualitative and quantitative vowel differences; see: A. Pakerys, “The Phonetic Basis of the Opposition between Long and Short Vowels in Standard Lithuanian,” — III Union Conference on Baltic Linguistics, 25–27 September 1975, Abstracts, Vilnius, p. 42, etc.

In Standard Lithuanian, quantity belongs to certain morphemes; see P. Garde, cited work, p. 160, or P. Garde’as, cited translation, p. 94.

²⁰ Н. С. Трубецкой (1939), min. veik., p. 213, 214, 223, 224, 240; G. L. Trager, The Theory of Accentual Systems, — L. Spier, A. I. Hallowell, and S. S. Newman (Eds.), Language, Culture, and Personality, Menasha, 1941, p. 139, 140; C. F. E. Heeschen, Einführung in die Grundprobleme der generativen Phonologie mit besonderer Berücksichtigung der litauischen Phonologie (daktaro disertacija), Bonn, 1968, p. 234, 252; M. J. Kenstowicz, min. veik., p. 84, 85, 107, 132 ir kt.

6. Užš. Nr. 1026

81

Diphthongs consist of two parts (moras). But nowadays less is known among foreigners, and it is known that in the East Aukštaitian dialects of Lithuanian (as in the literary language²¹), pitch accents in long vowels do not have the same phonological significance as in diphthongs; therefore, long vowels cannot be equated fully with diphthongs. In stressed position, the dialect has no short vowels i, ė, u, o (ė and o occur only in international words), and it cannot be explained from which two components the long vowels iː, eː, uː, oː are made up. In our dialects, “in compound and prefixed words, as well as when a consonant disappears in the middle of a word, vowels that come to stand next to one another may remain intact”²², rather than coalescing into a single vowel (more often a long one than a short one); according to P. Kuznetsov, this entirely proves that long vowels can be independent phonemes²³.

3. PITCH ACCENTS

Lithuanian stress is not tonal (musical), as, for example, in Chinese, but dynamic (quantity plays a subsidiary role). In many dialects, in stressed circumflex diphthongs the first component, and in acute diphthongs the second component, is lengthened to approximately half-long, while the other component is short; even in the literary language, “the sounds constituting unstressed syllables are somewhat shorter than the sounds constituting stressed syllables”²⁴. Vowels can change in quality and quantity (especially when unstressed) only as a result of dynamic stress, not tonal stress²⁵. By ear, speakers of the northwestern Panevėžys dialect distinguish pitch accents (as they themselves indicated) chiefly by sound duration and intensity. In polytone languages, pitch accents may be prosodies not only of tone but also of intensity; in that case, their distinguishing features include abrupt versus gradual changes in intensity, and so on²⁶. Recent instrumental experiments have demonstrated that fundamental frequency (the fall and rise of pitch and intensity within a syllable) is not a distinctive feature of Lithuanian pitch accents²⁷.

²¹ A. Laigonaitė, op. cit., p. 20; Grammar of the Lithuanian Language, I, p. 128.

²² Z. Zinkevičius, op. cit., p. 124 et seq.

²³ П. С. Кузнецов, О фонологической системе сербохорватского языка (1948), — А. А. Реформатский, Из истории отечественной фонологии, p. 344. — Bet tai dabar, gal būt, būtų galima aiškinti sujungimu (juncture). Lietuvių kalbą ne morine, bet skiemenine laiko A. Martinė, P. Gardas, žr.: A. Martinet, min. veik., p. 13; P. Garde, min. veik., p. 165, arba P. Garde’as, min. vertimas, p. 96.

²⁴ Grammar of the Lithuanian Language, I, pp. 129—137; А. Пакерис, Т. Плакунова, Я. Урбелене, Relative Duration of Lithuanian Diphthongs, — Garsai, priegaidė, intonacija, Vilnius, 1972, pp. 14—17, 36; J. Kazlauskas, cited work, pp. 8—9; Н. С. Трубецкой, cited work, p. 240, etc.

²⁵ А. Мейе, Введение в сравнительное изучение индоевропейских языков, Москва—Ленинград, 1938, p. 164.

²⁶ А. С. Либерман, Слоговая акцентуация кетского языка, — Лингвистические исследования, Ленинград, 1970, p. 378.

²⁷ A. Laigonaitė, On the Understanding of Lithuanian Stress and Pitch Accent, “Kalbotyra”, I, 1958, p. 71—100; A. Pakerys, Modulation of the Fundamental Tone (F₀) of Diphthongs in Contemporary Standard Lithuanian, “Kalbotyra”, XVII, 1967, p. 15; A. Girdenis, A. Pupkis, Pitch Accents of the Southern Western Aukštaitian Dialect (Prosodic Features), — Experimental and Practical Phonetics, Vilnius, 1974, p. 125 (ibid., see the survey of the twofold understanding of pitch accent dating from K. Jaunius and F. Kuršaitis, pp. 107—108); A. Girdenis, Prosodic Features of Pitch Accents in the Northern Samogitian Dialect (Duration, Fundamental Tone, Intensity), ibid., p. 198.

82

In the northwestern Panevėžys subdialect, as many as six tones can apparently be distinguished by ear, which, in order of decreasing sound quantity and, it seems, increasing articulatory effort, could be listed approximately as follows: circumflex (~), secondary medial (ˈ), medial (˙), acute (´), secondary checked (ˋ), checked (ˆ)²⁸. The acute, circumflex, secondary checked, and secondary medial are tones of primary stress, while the checked and medial are tones of retracted stress.

Both checked tones occur in a syllable that had an acute tone, while both medial tones occur in a syllable that had a circumflex tone. Secondary tones occur only in a diachronically final syllable (cf.: v’ɔlkdava “used to drive,” v’ɔlka “drove,” v’ɔlks “will drive”), and are therefore variants (allophones) of the circumflex tone. The tones of primary and retracted stress, like certain vowels of different quantities, may be in complementary distribution diachronically.

In diphthongs, tones can mark either the first or the second component: only the acute tone (acute accent) occurs on the first component, whereas all the other tones—variants of the circumflex—mark the second component more strongly. The strong position for tones is in stressed sounds of different quality (diphthongs, except ie, uo); the weak position is in stressed sounds of the same quality (vowels, including the second component of diphthongs) and in all long unstressed syllables: here the tonal contrast may potentially differ (and have additional phonological significance), but it may also be entirely absent. Lithuanian has only two strong tones—the acute and the circumflex²⁹.

The number of minimal pairs involving individual tones varies greatly. The dialect has the most homonyms contrasting circumflex and medial tones (about 1,200 pairs, nouns of accent paradigms II and IV), e.g.: sg. nom. p’ɔršts “finger” — pl. acc. p’ɔršts “fingers.” There are about 500 minimal pairs with an acute–checked tone contrast (nouns of accent paradigm III), e.g.: pẽšč’os “barefoot (fem. pl.)” — pėš’č’os “barefoot (fem. pl.).” Thus, tones most often distinguish forms of the same word.

Contrasts between all the other tones in the same context are scarce, because they distinguish only individual words (with different meanings). There would be about 60 pairs contrasting acute and circumflex tones, while the other possible tonal contrasts have only about 10 homonyms (mostly with diphthongs); cf.: nekáltos — nėkal-tos — nėkaltos “innocent” — nekaltos “innocent”; láuks “fields” — laũks “field” — laũks “fields” — láuks “fields,” etc.

Some linguists classify the tones of Lithuanian or Serbo-Croatian as phonemes (J. Dambrauskaitė; N. Trubetzkoy, A. Isachenko, J. Treiger), while others tend to

^Ne Linguists living in Lithuania have, to this day, mostly continued to believe that the stress of the modern language has preserved the Indo-European tone, e.g.: P. Garde, cited work, pp. 160–165; P. Skardžius, Lietuvių kalbos kirčiavimas, Chicago, 1968, pp. 11–12; C. F. E. Heeschen, cited work, pp. 190–283; M. J. Kenstowicz, cited work, p. 82 ff.; T. M. Lightner, Problemes in the theory of Phonology, I, Edmonton, 1972, pp. 390–391; П. И. Абаев, cited work, p. 83; А. А. Реформатский, cited work, p. 43; Ю. С. Степанов, cited work, p. 99, etc.

²⁸ In the absence of diacritics, the checked tone is assigned a different phonetic realization here. For more on tones, see also: A. Girdenis, A. Rosinas, review of V. Grinaveckis, Žemaičių tarmių istorija (Fonetika), Vilnius, 1973, 370 pp., „Baltistica“, X(2), 1974, p. 192; G. Gerullis, Litauische Dialektstudien, Leipzig, 1930, p. XLVII, 55 ff.; К. Гаршва, Восприятие слоговых акцентов литовского языка в поднаречии северо-западных паневежцев, „Lietuvos TSR Mokslų akademijos darbai“, A serija, 2(59), Vilnius, 1977, pp. 119–128.

²⁹ A. Girdenis, Prozodinės priegaidžių ypatybės šiaurės žemaičių tarmėje (Trukmė, pagrindinis tonas, intensyvumas), ibid., p. 160; J. Kazlauskas, cited work, p. 6.

6*

83

to be regarded as a property of a morpheme or word (L. Jelmslevas, V. Vaitkevičiūtė, P. Gardas; P. Kuznecovas, A. Reformatskis³⁰. Pitch accents, like stress, are assigned to the word as a whole mainly because it is thought that the opposition between pitch accents is preserved in only one (stressed) syllable of the word. But, first, unstressed syllables may be a weak position for pitch accents (just as for vowel quantity); second, in present-day Lithuanian dialects, unstressed syllables also have a certain pitch-accent distinction³¹. What is required for the pitch-accent opposition of the Lithuanian language to exist is not stress, but a long syllable; therefore, vowel quantity and the pitch-accent oppositions dependent on it could be assigned to the same—segmental or suprasegmental—level. If we were to regard quantity as a suprasegmental phoneme (according to the customary American analysis), or to divide diphthongs and long vowels that form the nucleus of one syllable into two short vowels (as the New Prague School and the glossematicians do), then pitch accents, too, would have to be assigned to the suprasegmental level³².

The assignment of pitch accents to the syllable is especially supported by the fact that, apparently, in all polytonic languages they extend onto the consonant that closes the syllable³³ (in a diphthong). Moreover, Lithuanian pitch accents depend not only on syllable length but also on its openness or closedness (prosodic properties of the word). For example, in an open stressed syllable, the North-Western Panevėžys dialect vowels i., u. have no pitch accent at all (A. Baranauskas and others marked these vowels with the circumflex accent), whereas the vowels e˙, a˙ have only an accompanying circumflex accent. The pitch-accent opposition arises only when these vowels form a sequence with another vowel (a diphthong) or with the consonants l, r, m, n (a mixed diphthong). In terms of quantity and pitch accent, diphthongs are sounds different from their component parts; therefore, both diphthongs and mixed diphthongs could be explained phonemically in the same way, although the latter, like the diphthongs ui, ai, abi, not only have both their components replaceable by other sounds (cf. lk kilti, kūlti, kélti, kálti — kūrti, kárti, etc.), but may also occur in different syllables (before a vowel, a diphthong breaks up into its constituent phonetic elements).

4. VOWEL PHONEMES AND THEIR VARIANTS

In a stressed syllable there may be all diphthongs and the following vowels: i˙, i., e˙, ẹ, e˙, e, ę˙, ę, u˙, u., ų˙, o˙, o, ø˙, ø, a˙, a. The vowels æ˙, æ occur only between a soft and a hard

³⁰ P. Garde, cited translation, p. 94; V. Vaitkevičiūtė, cited article, p. 22; П. С. Кузнецов, О фонологической системе сербохорватского языка, — ibid., pp. 338—342; А. А. Реформатский, cited work, pp. 43—46 (the works mentioned also survey other authors). For the second opinion, see also K. L. Pike, Tone languages, Ann Arbor, 1957, pp. 3—5 et al.

³¹ Plg. A. Pakerys, Apie kirtį, „Kalbotyra“, XVII, 1967, p. 133; А. Гирденис, Западнолитовские и древнепрусские слоговые акценты, — Балтийские языки и их взаимосвязи со славянскими, финно-угорскими и германскими языками. Тезисы докладов научной конференции, Рига, 1973, p. 71—75. — Didesnė kalbininkų dalis mano, kad nekirčiuoti aukštaičių tarmės skiemenys dažniausiai turi tik tvirtagalę priegaidę, žr.: A. Pakerys, Tvirtapradžių ir tvirtagalių dvigarsių spektras, — ten pat, p. 149 (8 išnaša).

³² For a review of the phonemic problems of diphthongs, quantity, and pitch accents, see: K. Togeby, Phonotactics (1971), — ed. E. C. Fudge, Phonology, 1973, pp. 291—296; I. Lehiste, cited work, pp. 43—53, 84—95. Cf. also W. Smoczyński, Some Problems of Lithuanian Phonology, — III Union Conference on Baltic Linguistics, Vilnius, 1975, pp. 58—60.

³³ Л. Р. Зиндер, min. veik., p. 304; taip mano ir doc. A. Girdenis.

84

consonant, the vowel εː occurs only after a hard consonant, and the vowel ə only before a hard consonant³⁴. They are positional variants of the vowels e (a), ė, ę; cf.: gėr’ “the good one” — gèr “good,” rėːk “shout” — l’ėːk “fly,” k’ėːt’s’ “cross” — k’òr’s “crouch.”

An unstressed syllable may contain all diphthongs (except ie, uo; in word-final position, diphthongs are monophthongized) and the vowels 1) i, ė, e, e, a, u, ė, o, ə, ɑ; 2) i₁, ē₁, ē₂, u₁, u₂, ε₁, ε₂³⁵. The first group of short vowels consists of variants of the corresponding long, half-long, and short vowels that occur in unstressed position. The second group of vowels occurs only in an unstressed syllable and, on the basis of their qualitative similarity, they should be regarded as variants of the vowels i, ē, u.

Thus, the vowel system of the North-Western Panevėžys dialect consists of 10 phonemes: |iː|, |iː.|, |eː|, |eː.|, |e|, |uː|, |uː.|, |oː|, |aː|, |a|. In addition, the dialect has sequences of two sounds: a) monophthongal diphthongs ie, uo (variants — ˈie, ˈuo), b) compound diphthongs eĩ (e₁ĩ), uĩ (õĩ), aĩ (ε₂ĩ), aũ (æu, əu), and c) mixed diphthongs i (ε, j), e (e), u (ə), a+ł, r, m, n. Variants of the first component of diphthongs occur in unstressed position (and in a circumflex-stressed diphthong, except ie, uo).

In an unstressed syllable, vowel quantity and intensity are neutralized, but quality is generally preserved, e.g.: vi(e)ri.ms “boiling” — vi₁(ē)ri.ms “threading,” aŭses’ “ear” — aŭses’ “ears” (ausiēs), šo(u)nėł’ “puppy” — šo(u₁)nėł’ “side,” gražō “beautiful (acc. sg.)” — gražo “beautiful (acc. sg.),” klaũsos “I listen” — klaũsos “you listen,” o(u)s’óks “little moustaches” — (u)₁(u₂)s’óks “little ash trees,” etc. (54–90% of untrained pupils distinguish such pairs by ear). In minimal pairs, it is not only the syllable containing the vowel under investigation that differs, but also the adjacent syllable. The lower the vowel’s height, the slightly longer are both that vowel and the syllable next to it (whether before or after it). We assume that prosodic differences in adjacent syllables (of a word) arise from the influence of different vowels.

As the sounds continue to shorten³⁶, in pre-stress and non-final post-stress syllables (cf. the vowels i, u, which may be retained), the sounds ē, ʼe (ē₁), ε and o, ʼuo (ɔ) were already sometimes reduced to the sounds i₁, (i₂), bi and u₁ in A. Baranauskas’s texts of the previous century,

³⁴ In Lithuanian, the correlational features of palatalization (palatalization) and “bemollity” (velarization), which extend beyond a single phoneme, are most appropriately assigned not to vowels or even consonants, but to the suprasegmental level. Cf. also: A. Girdenis, Mažeikių tarmės fonologinės sistemos apžvalga, “Baltistica,” VII (1), 1971, p. 28; Ю. С. Степанов, мин. стр., p. 98—103 (on Lithuanian, see pp. 100—101); but W. Smoczyński, min. tezės, p. 58—61. In the North-Western Panevėžys dialect, hard and soft consonant phonemes cannot be posited because consonants depend more on vowels: at the end of a word, consonants are assimilated to palatalized consonants not only before front vowels, but also after them (with some exceptions), e.g., kárves’.

³⁵ As regards the dialectal counterparts of standard Lithuanian sounds iː, i, ie, ē and uː, u, uo, oː in unstressed syllables, there have so far been three views (for sources, see K. Garšva, min. dipl. d. santrauka, pp. 26—27). It has long been most commonly maintained (or interpreted as such) that in the dialect the aforementioned front vowels merge when unstressed into a single vowel ε, while the back vowels merge into the vowel ə (as do the word-final diphthongs ei, ai, and au). Another group of linguists writes that in an unstressed syllable only the standard Lithuanian sounds ie, ē and uo, oː turn into ε and ə, whereas the high vowels i, u are occasionally, in many areas, or always retained without lowering. Finally, around 1965 yet another view appeared, contrary to the first: except in the final post-stress syllable, in all other unstressed positions the dialect is said to have only the vowels i and u in place of the aforementioned standard Lithuanian vowels.

(The article’s author verified vowel differences not only by the traditional method—chiefly using a few examples and introspection. The study draws on approximately 8 km of recordings on magnetic tape from the dialect area, responses from 22 informants (students), auditory experiments conducted in the secondary schools of Linkuvõs, Pȧkrūjo, Pāsvālio, Salõčių, and Vaškų, printed dialect texts, etc.).

³⁶ Plg. A. Mežė, min. veik., p. 164.

85

(u₂) (this was perhaps also influenced by the southern Panevėžiškiai dialect). When examining a phoneme in the context of a word (following the Leningrad phonological school), the new sounds must be assigned to the phonemes /i/, /u/ on the basis of qualitative similarity, since, if the morpheme is taken as the basic unit (following the Moscow phonological school), sounds that are almost indistinguishable, such as i and i₁ (i₂), should be regarded as variants of three different phonemes, /i/, /ie/, /ė/³⁷; cf.: li(ė)pi.ms “climbing” — li₁ (i₂, ė)pi.ms “ordering” — li₁(iė)pi.ms “hiding”.

The vowels ē₁, ē₂, ε₁, ε₂ not only cannot occur in the same stressed morpheme, but also occur in different phonetic contexts (in addition, ē₁, ē₂ occur after palatalized consonants, and ε₁, ε₂ after velarized consonants); therefore, all of them should be considered positional variants of the vowel ė, although the sounds differ somewhat phonetically; cf.: lápės — šlàpė₁s “with the wet ones” — š’vėpe₂l’s “lisping person, lispers” — šlápės’ “you will wet”; kaũles “pigs” — kaũle₁s “bone(s)” — ar’kėl’s “horse.” The vowels ē₂, ε₂ are inserted in the final post-tonic syllable between certain plosive and fricative consonants z, s, š on the one hand and following consonants l, r, m, n, v on the other (in some localities around Pašvitinys, Vaškai, and Saločiai, a vowel resembling ǝ is inserted, chiefly between velarized consonants).

The phonetic classification of the dialect’s vowels is shown in Table 1³⁸ (the less frequent and newer variant of the same vowel is indicated in parentheses).

Front | Central | Back High | iː, i., (i₁), i, i₁ | ʊ | uː, u., (u₁), u, u₁ Mid | ēː, ē, ē₁, ē₂, (e.), ę | εː, ε, ε₁, ε₂ | oː, o, (ɔ.), ɔ, (ɔ.), ǝ Low | eː, æː, (e.), (æ.), e, æ | — | aː, (a.), a

Some linguists, following K. Jaunius (and partly J. Gerulis), argue that in the northwestern Panevėžiškiai dialect, instead of the formerly reduced short ending vowels i, e, æ, u, a, there are still murmured vowels³⁹ (more precisely, a single unclear…

³⁷ See also M. Hale’s review: O. Akhmanova, review of T. A. Sebeok (ed.), Current Trends in Linguistics, vol. 1, The Hague–Mouton, 1963, — Word, vol. 21, no. 1, 1965, p. 168.

³⁸ The classification of vowels is approximate, since unstressed short vowels cannot be “lengthened” (and their articulation thereby clarified); the acoustic impression of sounds is affected by the different neighboring consonants. E. Grinaveckienė and J. Šliavas also distinguished the vowel ė (ė̇) with a similar meaning, the vowel ε₁ was distinguished by E. Grinaveckienė and I. Jašinskaitė, and the vowel ɔ (ɔ) by K. Jaunius, J. Gerulis, E. Grinaveckienė, J. Šukys, and others; see: E. Grinaveckienė, “The Relationship between Consonant Palatalization and the Shortening of Endings in the Stegvilai Dialect,” LKK IV 174—175, 180; J. Šliavas, Žeimelis (VVU MB manuscript), 1972, p. 406; E. Grinaveckienė, “Some Features of Lithuanian Dialects,” LKK V 147; I. Jašinskaitė, “Stress, Accent, and Their Effect on Vocalism in the Biržai Dialect,” LKK I 190; K. Jaunius, “The Panevėžiškiai Dialect,” — V. Drotvinas, V. Grinaveckis, Kalbininkas Kazimieras Jaunius, Vilnius, 1970, p. 136; G. Gerullis, cited work, p. 60; J. Šukys, “Characteristic Phonetic Features of the Eastern Aukštaitian Pontininkai and Western Puntininkai Dialects,” LKK V 178—179; LKT 14,237, etc.

³⁹ Z. Zinkevičius, cited work, p. 119; A. Girdenis, review of Lietuvių kalbos tarmės (Chrestomatija), Vilnius, 1970, 518 p. + 1 map, “Baltistica,” VII(2), 1971, p. 207. — For an overview of the authors holding the opposing view, see: Z. Zinkevičius, “Notes (XII),” — “Baltistica,” XI(1), 1975, p. 85—86; see also K. Būga, Lietuvių kalbos žodynas (1924), — Rinktiniai raštai, III, Vilnius,

86

a vowel of murmured quality (after a palatalized consonant, a front vowel; after a hard consonant, a back vowel). The so-called murmured vowel is “often not heard at all,” but its presence is “indicated” by certain other word-final sound features: 1) voiced consonants before -s are usually not devoiced, cf. blõgs “bad” (– blõks “block”), 2) the consonants s, š, ž do not merge with -s into one sound, cf. saũss “dry” (– saũs “dry”) 3) consonant clusters d, t, +s are “not assimilated” for voicing (mėds “honey” – mėts “year”), 4) when a vowel comes into contact with the consonants l, m, n, r, v, j, no secondary diphthong is formed (vėj “wind,” mąl “I grind”), etc. The examples cited (1, partly 2 and 3) involve differences in consonant voicing–devoicing (this opposition is not neutralized in the final syllable in the Livonian and southern central dialects of Latvian, either; in absolute word-final position, it is also found in Serbian-Croatian, English, Swedish, French, Erzya Mordvin, Ossetian, and other languages), consonant quantity⁴⁰ (2, possibly also 4); the formation of secondary diphthongs (4) is prevented by short stress and the circumflex accent, so there is no need to posit so-called murmured vowels⁴¹: not only do they have no linguistic significance, they are also usually not pronounced phonetically⁴².

К. К. ГАРШВА

THE INTERRELATION OF ACCENTUATION AND VOCALISM IN THE SUBDIALECT OF THE NORTHWESTERN PANEVEŽIŠKIAI

Abstract

This article undertakes, within a strictly synchronic framework, to determine the functions of the following four expressive means in one (East Aukštaitian) dialect of Lithuanian: 1) word stress, 2) vowel quantity, 3) syllabic accents, 4) vowel phonemes and their variants.

1961, pp. 82–83; P. Būtėnas, “Boundaries of the Aukštaitian dialects’ okuojančiosios pašnektės,” APh III 169, 174; Ž. Urbonavičiūtė, “Consonants l, r, s, z before word-final e in the Aukštaitian dialects,” “Kalbotyra,” XXV(1), 1974, pp. 73, 74, at least several diploma theses at the university, etc.

⁴⁰ M. I. Lekomtseva, “On the typological characterization of the phonological systems of Latvian dialects,” — Baltic-Slavic Studies, Moscow, 1974, pp. 231–241.

⁴¹ They would also have to be written differently in the words d aug, kà(d), sù.lig, tis’òg, ù.ž, etc. (since here too the consonant is not devoiced), and after epenthetic vowels, although a consonant cannot separate two vowels within one syllable. If we did not know the literary language, we would not know that, for example, in the word lèp’s’ “lipės” (“tails”), the murmured vowel “should” be written before the consonant -s, whereas in the homonym lèp’s’ “I will climb, you will climb” it should be written after the consonant s (after voiceless consonants, no “syntagmatic arguments” indicate the presence of murmured sounds). In the 1975 auditory experiments at the schools mentioned, out of 173 listeners, 90, or 52% of the audience, correctly distinguished the pair kas(.) “I dig” — kas “who/what” (the speakers were local); when the same speaker (from Linkuva) pronounced the present-tense forms with a slightly longer a, 77 out of 177 listeners, or only 43% of the pupils, distinguished them correctly (which merely indicates guessing). G. Ručinskaitė’s spectrograms made at the university show not the murmured sound itself, but only that the short vowels in “kas”-type pairs bearing retracted stress might be slightly tenser (and perceived as longer).

These consonant features persisted because in the nineteenth century the dialect still had many short vowels in inflectional endings (cf. K. Jaunius, cited work, p. 127; see also A. Baranauskas’s texts). During the transitional period of vowel reduction, the murmured vowel may have been: especially on the periphery of the dialect, distinct remnants of short vowels can still be heard; around Krikai, some speakers pronounce counterparts resembling i(e) or (after a hard consonant) bl (ɛ) instead of front and back vowels.

⁴² I am grateful for advice and comments to my teachers of phonology and experimental phonetics, Prof. J. S. Stepanov and Assoc. Prof. A. Girdenis.

87

Words may be monosyllabic and may also bear stress: even in them, syllabic accents are distinguished, as in stressed position. The stress inventory may be considered to comprise stress on the first syllable, stress on the second syllable, and so on. The dialect has four types of contrast according to stress placement: stressed syllables may also differ in vowel quantity and syllabic accent.

The dialect’s phonological vowel system consists of 10 monophthongs (without considering combinations of two sounds): /iː/, /i/, /ɪː/, /ɪ/, /eː/, /e/, /uː/, /u/, /oː/, /aː/, /a/. Vowels i, u may be long or half-long (their short variants are e, ə); vowels e, a may be long or short, while vowels ē, ō are long only. The quantitative distinctions between high and low vowels, as well as syllabic accents, are opposed in the very same strong (stressed) position; therefore, quantity is a feature at the segmental level. Syllabic accents occur both in biphonemic vowel–consonant combinations and, for this reason alone, may be considered a suprasegmental property.

Syllabic accents are distinguished only in long syllables. Short and half-long vowels in Lithuanian are not opposed by the contrast between the beginning and end of the syllable nucleus or by the course of intensity; therefore, short (ˈ), mid (ˌ), and mid-interrupted (ˇ) syllabic accents do not occur in them: only dynamic (not musical) stress is present. The strong position for syllabic accents is on different components of stressed diphthongs; the weak position is on long stressed monophthongs, on the same (second) component of a diphthong, and on all long unstressed (diphthongal) syllables. In the weak position, syllabic accents (the acute and five variants of the circumflex) represent potential syllable prominence, which is an additional or, more rarely, accompanying means (the latter occurs only on the vowels /eː/, /aː/).

In unstressed syllables, quantitative vowel distinctions are neutralized, while qualitative distinctions are mostly preserved. In any position in a pre-stress or non-final post-stress syllable, the short vowel variants i, e, u, o are preserved or change into the vowels ę, į, ǫ, ų. In endings, formerly short vowels i, e, a, u, a are reduced not only quantitatively but also qualitatively: words are then differentiated only by consonants in voicelessness–voicing and length.