Dvibalsių ir mišriųjų dvigarsių priegaidės Eržvilko šnektoje: kiekybė ir spektrai

LIETUVIŲ KALBOTYROS KLAUSIMAI XL (1998) Daiva ATKOČAITYTĖ Institute of the Lithuanian Language, Vilnius ACCENTUAL PATTERNS OF DIPHTHONGS AND MIXED DIPHTHONGS IN THE ERŽVILKAS DIALECT: QUANTITY AND SPECTRA 1. The accentual patterns of diphthongs in the Southern Samogitian Raseiniai subdialect, their distribution, and their acoustic characteristics have not yet been studied and described in detail. Data are particularly scarce from the southern edge of the dialect area, where the Eržvilkas dialect under investigation is spoken. Authors of both monographic and survey studies generally rely on empirical observations, which are not always accurate and can describe the acoustic features of accentuation only approximately. There are no studies in which the accentual patterns of individual dialects or of the entire dialect have been investigated experimentally; moreover, it is nowhere discussed how the accentual patterns of the Raseiniai subdialect resemble the corresponding patterns of other Samogitian dialects, and how they are more similar to those of Standard Lithuanian. This article accordingly attempts to substantiate, through instrumental analysis, the distinction between two accentual patterns—acute and sustained—in diphthongs, to identify the more important distinguishing features, and to compare them with the findings of corresponding studies of neighboring Samogitian dialects and Standard Lithuanian1. 2. Although the acute and sustained accentual patterns are customarily distinguished in the Raseiniai dialect, dialect researchers sometimes use other terms as well. For example, M. Slušinskaitė calls the acute accent in the Lomiai dialect the “sudden acute,” which is supposedly similar to the acute accent of Standard Lithuanian, although there is no clear explanation of how this accent differs from that of Standard Lithuanian (Slušinskaitė 1981: 20). Z. Zinkevičius likewise does not use the term “acute accent.” In his view, at the southeastern edge of Samogitian (and thus also in the Eržvilkas dialect), the Samogitian broken accent is changing into the Aukštaitian pushed accent. When a word is pronounced emphatically, the pushed accent becomes a true broken accent (Zinkevičius 1966: 35). The sustained accent of circumflex origin differs more from its Standard Lithuanian counterpart: it is generally thought that its stress is distributed across both components of the diphthong. Yet although this accent is characteristic of the entire Raseiniai area, V. Tamošaitytė-Jokubauskienė uses the term “circumflex accent” in her description of the Pagramantis dialect (Tamošaitytė-Jokubauskienė 1974: 27). It remains unclear whether this is how it is used here 1 In this study, comparisons will be made primarily with data on the accentual patterns of diphthongs in the nearest Southern Samogitian Varniai dialect (Bagvilaitė 1975; Bukantis 1984), Northern Samogitian (Girdenis 1974), and Standard Lithuanian (Pakerys 1968; Pakerys ir kt. 1972; Pakerys 1974; Pakerys ir kt. 1974; Pakerys 1982). The methodology used in studies of accentuation in other dialects was also consulted (Girdenis, Pupkis 1974; Kosienė 1982; Girdenis 1996). 88 vadinama įprasta tęstine priegaide (ši terminą vartojo ir P. Jonikas (Jonikas 1939: 12)), ar Pagramančio šnektoje tariama priegaidė iš tikrųjų sutampa su bendrinės kalbos tvirtagale. Since there is no clear reason to use any new terms, this article follows the traditional view that the Raseiniai dialect distinguishes two accentual patterns—the acute and the sustained (cf. Grinaveckis 1973: 83; Zinkevičius 1994: 108)—while the other accents occasionally encountered are merely variants of these (see § 3). 3. Empirical observations have already shown that the dialect clearly distinguishes in diphthongs the opposition between accents of acute and circumflex origin, e.g.: lá.uk ‘láuk’ – lā.ū.k ‘laũk’, ká.lts ‘káltas’ – kā.l.ts ‘kaltas’. The accent of acute origin in the dialect is only acute. The broken accent can be heard very rarely. It occurs only sporadically, most often in a syllable bearing secondary stress, e.g.: ta_tas antrašis | valėi jų.˟kˡiˑ iš juó || tās tė.ī.niˑ sā.u | rìsas || In emphatic speech, the broken accent can sometimes also be heard in a syllable bearing primary stress, e.g.: tai_gerai padā.rī | pliešìks dá. bū.tu plieše(s) žmúonis || kuokė būwa žūdˑnˑi || This rather rare and irregular variant of the acute accent, which most often depends on phrase intonation, is almost never found in diphthongs; therefore, its distribution and acoustic features will not be discussed here. The accent of circumflex origin is more difficult to characterize. At times, a sustained accent is pronounced quite clearly in the stem, with its stress spread across both components, but at other times it is difficult to determine whether both components are stressed or only one of them. As a positional variant in absolute word-final position, a medial accent is more often pronounced; its stress tends to be concentrated on the first component, while the second component is greatly reduced, e.g.: pavā.saris | tetervīnāi šā.ū.k || pakrúoviēm | ta_višta_i_tupieje gerāi || In this case, most of the vocal stress falls on the first component, while the second is pronounced much more weakly2. For this reason, when selecting the word pairs for investigation, greater attention was paid to the opposition between accents in the root, where both accents are pronounced clearly, although all diphthongs were also examined in absolute word-final position. The main objectives of the study are to determine: 1) how the duration of the entire diphthong and of its individual components depends on accent, 2) how the quality of the components is related to accent, 3) whether the duration of the entire diphthong and of its individual components depends on the quality of the components. 2 Hereafter in the article, this medial accent is not distinguished and is referred to as sustained. 89 4. Since the relationship between diphthong quality and quantity and the tones was of interest, only stressed a3, e+u, i, l, m, n, r were examined experimentally. The diphthong ui is not discussed at all, since it is used rather rarely in the dialect and it is difficult to find suitable word pairs for the experiment; moreover, the tones of this diphthong are poorly distinguished. The mixed diphthongs i, u+l, m, n, r are not discussed here either, since the relationship between the quality and quantity of their first components and the tones has already been addressed in an earlier publication (see Atkočaitytė 1997), while the relationship between the duration of the entire diphthong and that of its second component and the tone should not differ from the tendencies characteristic of the other mixed diphthongs examined here. The words selected for the experiment had stressed syllables with either an acute or a sustained tone. The minimal word pairs were read by speakers of this dialect: two informants (a man, born in 1932, and a woman, born in 1939), who live in the northern part of the dialect area, in the village of Varlaukis, and the article’s author, who also speaks this dialect. In total, 148 words, read with declarative intonation, were examined. The following groups of words were read: 1) diphthongs: gá.u.ti ‘gáuti’: gá.ū.sti ‘gaūsti’ – buvá.u ‘buvau’⁴; k’á.ura ‘kiáurą’: k’á.ū.le ‘kiaũlę’– jā.ū. ‘jaũ’: ajá.u ‘ėjau’; lé.istī ‘leĩsti’: sklē.ī.stī ‘skleĩsti’ – kvì.tē.i ‘kvietei’: kvì.tē.ī ‘kviečiai’⁵; lá.ida ‘láido (mėto)’: lā.ī.da ‘laidō’ – sakā.i ‘vksm. sakai’: sakā.ī ‘dktv. sakai’; 2) mixed diphthongs: lá.ŋks ‘lángas’: lā.ŋ.ks ‘lañkas’; ká.lts ‘káltas’: kā.l.ts ‘kaltàs’; vaiká.ms ‘vaikáms’: kā.m.(p)s⁶ ‘kãmpas’; ká.rs ‘kãras’: kā.r.sts ‘karstas’; pé.rs ‘pėras’: pė.r.št ‘peršti’; ké.lti ‘kėlti’: kē.l.ti ‘keļti’, lé.mpa ‘lém̄pa’: lē.m.ts ‘lėmtas’, lé.ŋks ‘lénkas’- lē.ŋ.kts ‘leñktas’. The word pairs read by the informants were recorded on magnetic tape and transferred to the computer’s memory, while the author’s examples were read directly into the computer. P. Kasparaitis’s “Kalbam44” program was used for acoustic analysis. Oscillograms, cross-sectional and dynamic spectrograms of the diphthongs under study were produced. The total duration of the diphthongs and the duration of their individual components were measured in milliseconds (ms). In diphthongs, especially au, ei, the two components merge and pass into one another without a marked break, so it was difficult to determine the boundaries and exact duration of the components of these diphthongs. The data for mixed diphthongs are more reliable, since their components are easier to distinguish—the boundary is indicated by a change in the spectrum. The aim of the spectral analysis of the first components of diphthongs and mixed diphthongs is to determine the effect of tones on vowel quality. The formant values of components a and e 3 In the root au, the sound was examined after both a hard and a palatalized consonant, since in the [C’–] position a vowel of a different quality, 'a, occurs and the diphthong 'au is formed. 4 No pair differing in the ending between á.u and ã.ū was found, but stress placement in verb endings is not consistent in the dialect, so the same informants may pronounce such an ending diphthong in two ways—once stressed with an acute tone, and another time with a sustained tone. 5 The first component 'a- of the diphthong corresponding to Standard Lithuanian -iai shifted toward the front long ago in the Samogitian dialect and became -e-, indistinguishable from the first element of the old diphthong -ei. 6 The labials p and b, located between the labial m and the voiceless s, are almost always deleted. 90 Frequencies were determined at three points: at the beginning, in the middle, and at the end. In the segments u and i, one measurement was made, since these sounds are shorter than a, e, making it difficult to identify their beginning, middle, and end. The formant frequencies of u are particularly difficult to determine. Therefore, sonorants were not investigated either. The measurement data were statistically processed using the “Student” program. The following statistical parameters were calculated: a) arithmetic mean (x), b) standard deviation (s), c) coefficient of variation (v), d) 95% confidence interval, e) Student’s t criterion (t) and its critical value (ta7. The indices of segment compactness, bemolization, and tonality were calculated using A. Girdenis’s “FORMANTES.PAS” program8. 5. Although there are some similarities in quantity and quality, diphthongs and mixed diphthongs are discussed separately in this article. 5.1. Diphthongs. 5.1.1. The data on overall duration show that, for diphthongs pronounced with declarative intonation whose first components are a, e, overall duration is a constant prosodic-accent feature. Circumflex diphthongs are longer than sustained-accent ones by a ratio of 1.2:1. This difference is fairly reliable, since each circumflex diphthong considered individually is longer than its sustained-accent counterpart (see Table 1); their confidence intervals do not overlap, except for ei and au (cf. ei – 316÷ 341 ms and 288 ÷ 316 ms; au – 297÷ 348 ms and 228 ÷ 322 ms)9, while the values of the t criterion (except for au) exceed the critical 95% significance threshold. Table 1. Overall duration of diphthongs (ms)10 Diphthong | n | x | s | v (%) | confidence interval (95%) | t ≷ tₐ á.i | 12 | 366 | 8,6 | 2,3 | 355 ÷ 376 | 8,2 > 5,0** á.ī. | 12 | 324 | 7,5 | 2,3 | 315 ÷ 333 é.i | 12 | 328 | 9,9 | 3,0 | 316 ÷ 341 | 3,9 >3,8* é.ī. | 12 | 302 | 1,4 | 3.8 | 288 ÷ 316 'á.u | 12 | 359 | 8,8 | 2,4 | 348 ÷ 370 | 5,2 >5,0** 'á.ū. | 12 | 317 | 15,8 | 5,0 | 297 ÷ 337 ă.u | 8 | 322 | 27,9 | 8,6 | 297 ÷ 348 | 2,3 > 2,2 ă.ū. | 8 | 275 | 44,7 | 16,2 | 228 ÷ 322 7 For more on mathematical statistical methodology, see Girdenis 1974; Kosienė 1982. 8 Skaičiavimo formules ir metodiką žr. Piotrovskis 1960. 9 Since the mean was calculated irrespective of position, these results may have been determined by the structure of the diphthongs. Because both components belong to the same class, in open word endings the second components of these diphthongs are more assimilated and shorter than in the root. The entire diphthong is also shortened as a result, so when calculating the overall mean, such values considerably widen the confidence interval and reduce the likelihood of distinguishing them. 10 In the tables, pairs marked ** differ at the 99.9% probability level, while those marked * differ at the 99% level. The mean x and confidence-interval figures are rounded to the nearest integer. 91 The overall mean duration of diphthongs in both stems and endings was calculated, although in open endings continuous and acute-accented diphthongs are shorter than in the root, since a tendency was observed in the dialect to reduce the stressed syllable in the ending slightly as well, e.g.: tuõ dvã.ra kùmetei | ten_šĩna pjúovi|| kùr_ten priẽ.ĩ.si| mĩškai didžã.us|| The beginning of the second component merges with the end of the first and affects it to such an extent that, by the end of the first component, it is already clear what the height of the second sound will be; thus, the second component need not be fully pronounced and can be shortened. Acute-accented diphthongs at the end of a word shorten less, because a greater proportion of their duration is concentrated in the first component, which does not undergo reduction. Nevertheless, the duration ratio between acute-accented and continuous diphthongs does not depend on position: both in the stem and in the ending, acute-accented diphthongs are longer. In this respect, the Eržvilkas dialect differs from the Samogitian dialects farther north, where circumflex diphthongs are considerably longer (cf. Bukantis 1984a: 72; Girdenis 1974: 175), and is more similar to the standard language, in which the lengths of falling diphthongs do not depend on accent: sometimes acute-accented diphthongs are longer, sometimes circumflex-accented ones (Pakerys 1968: 107; Pakerys et al. 1972: 10). 5.1.2. An even clearer and more important cue to accent is the duration of the individual components (cf. Pakerys 1982: 158). The duration ratio shows that the first components of acute-accented diphthongs are considerably longer than the second components (by a ratio of 2.3:1), whereas the durations of the two components of continuous diphthongs differ less, although the first component is also longer than the second (by a ratio of 1.5:1) (cf. Tables 2 and 3). This is a rather unexpected result, since it is generally assumed that in a continuous diphthong, duration and vocal effort should be distributed equally between the two components. In the ending, the durations of both components are somewhat shorter in both acute-accented and continuous diphthongs, but the ratio between them remains similar. Table 2. Duration of first components (ms) Diphthong | n | x | s | v (%) | confidence interval (95%) | t ≫ tₐ á.(i) | 12 | 255 | 9.6 | 3.8 | 243 ÷ 267 | 13.5 > 5.0** ã.(ĩ.) | 12 | 188 | 5.3 | 2.8 | 182 ÷ 195 | é.(i) | 12 | 237 | 17.4 | 7.3 | 216 ÷ 259 | 5.1 > 5.0** ẽ.(ĩ.) | 12 | 197 | 2.7 | 1.4 | 194 ÷ 200 | 'á.(u) | 12 | 252 | 10.3 | 4.1 | 239 ÷ 265 | 12.9 > 5.0** 'ã.(ũ.) | 12 | 190 | 3.3 | 1.7 | 186 ÷ 194 | á.(u) | 8 | 211 | 30.5 | 14.4 | 183 ÷ 239 | 4.4 > 3.6* ã.(ũ.) | 8 | 146 | 12.7 | 8.7 | 130 ÷ 162 In both acute-accented and continuous diphthongs, the first components a and 'a are longer than e, because, as in the standard language, their inherent duration is greater (cf. Pakerys 1982: 48). This apparently also accounts for the greater overall duration: diphthongs with a or ‘a as their first component are considerably longer than ei (see Table 1). 92 Table 3. Duration of second components (ms) Diphthong | n | x | s | v (%) | confidence interval (95%) | t ≷ tₐ (á.)i | 12 | 100 | 3.8 | 3.8 | 96 ÷ 105 | 10.1 > 5.0** (ā.)ī. | 12 | 126 | 4.3 | 3.4 | 121 ÷ 131 (é.)i | 12 | 101 | 6.8 | 6.7 | 93 ÷ 110 | 1.7 < 2.3 (ē.)ī. | 12 | 109 | 6.7 | 6.1 | 101 ÷ 117 (á.)u | 12 | 103 | 3.6 | 3.5 | 99 ÷ 108 | 5.7 > 5.0** (ā.)ū. | 12 | 121 | 5.8 | 4.8 | 113 ÷ 128 (á.)u | 8 | 116 | 18.0 | 15.5 | 97 ÷ 135 | 1.5 < 2.3 (ā.)ū. | 8 | 132 | 16.7 | 12.6 | 112 ÷ 153 The second components of acute-accented and continuous diphthongs differ much less than the first components. This is most evident in the diphthongs ei and au: the confidence intervals for the second components overlap ((ē)i 93 ÷ 110 ms – (ē.)ī. 101 ÷ 117 ms; (ā.)u 97 ÷ 135 ms – (ā.)ū. 112 ÷ 153 ms), and t is smaller than the critical value of tₐ ((ē)i – 1.7 < 2.3, (a)u – 1.5 < 2.3). The lower reliability of the second-component duration is also confirmed by the higher coefficients of variation, especially in continuous diphthongs (cf. Tables 2 and 3). Thus, at least for ei and au, the duration of the second components is not an important distinguishing cue to accent. In the other diphthongs, too, the difference between the first components is greater than that between the second components. The measurement results show that although duration is distributed in a way similar to that in northern Samogitian dialects in continuous diphthongs, the first component is the more important indicator of accent. In this respect, the Eržvilkas dialect is closer to the standard language (cf. Pakerys 1968: 114) and differs substantially from the Varniai dialect, where the duration of the second component is more significant (see Bukantis 1984a: 73). 5.1.4. It can already be heard that the components of acute-accented and continuous diphthongs differ not only in quantity but also in quality. The differences between the first components are more pronounced: acute-accented components are more like the corresponding monophthongs and assimilate less to the other component of the diphthong, as in the standard language (cf. Pakerys 1974: 149); they are lower in vowel height and have a somewhat more retracted articulation, whereas the first components of continuous diphthongs are higher and articulated more frontally (the first formant F₁ is lower, while the second F₂ is higher) (see Fig. 1). The lowest first component occurs in the diphthong áu: F₁ ≈ 700 Hz, while the highest, as expected, occurs in ēī.: F₁ ≈ 440 Hz (see Fig. 1b and c). In the continuous diphthong, the high vowel i also raises the height of e. A more diffuse, higher-pitched sound is produced (see Table 2), resembling a mid-height e. Similarly, i affects a in the diphthong ai: both continuous and acute-accented a(i) are higher and more frontally articulated than a in the diphthong au (cf. Fig. 1b and d), because u retracts the articulation of the first component somewhat and raises its height only slightly¹¹. In continuous diphthongs, the influence of the second component is even stronger than in acute-accented ones: ā. (ū.) and ā. (ū.) are more labialized and lower-pitched than the corresponding acute-accented forms (see Table 4). 11 In the standard language, too, the first components of diphthongs depend on the quality of the second components: á.(i) becomes more fronted, while á.(u) becomes more backed (Pakerys 1974: 150). 93 Fig. 1. Spectra of the first components of diphthongs12 Table 4. Qualitative indices of diphthongs Diphthongs Formants F₁ F₂ Compactness Tonality Flatness ’á. 740 1860 878 360 163 u 490 980 901 -521 176 ’ã. 650 1750 867 250 165 ü. 450 990 886 -510 177 á. 730 1350 915 -150 167 u 540 870 929 -628 176 ’ã. 640 1250 906 -250 169 ü. 440 920 892 -575 176 á. 690 1490 894 -10 166 i 330 2240 753 736 170 ã. 620 1640 868 135 167 i. 310 2250 744 755 171 é. 570 2120 829 620 164 i 370 2180 770 677 169 ė. 440 2150 793 650 165 i. 340 2290 754 790 170 12The formant values of the components with acute accent are marked by dotted (---), and those with circumflex accent by solid (—) lines. 94 The second components of continuous diphthongs are somewhat more diffuse and higher in pitch. Their formant values lie farther from the center of the spectrum; thus, the sounds are more tense. The formant dynamics of the diphthong as a whole show that the first components stressed with the continuous accent pass into the second components more evenly, without a major break (see Fig. 2). The formant-frequency trajectories of continuous ė.i. change little, which means that the two sounds are not only of the same frontness, but also of similar height, whereas the initial F₂ of initial-stressed ė.i is much lower, rising sharply only at the end, when i is pronounced. In the diphthong á.i the break is even more pronounced, because its components are of different frontness. A similar formant pattern for ai in the Varniai dialect is presented by J. Bukantis (see Bukantis 1984b: 63, Fig. 43), although, in his view, frequencies are not a sufficiently reliable indicator of accent (Bukantis 1984a: 79). a) ė̇.i. (---) – é.i (—) b) ã.i. (---) – á.i (—) Hz Hz Formant frequencies Formant frequencies c) ã.u. (---) – á.u (—) d) ã.u. (---) – á.u (—) Hz Hz Formant frequencies Formant frequencies Fig. 2. Formant dynamics of diphthongs13 The formant trajectories of diphthongs with u as the second component are less smooth: near the beginning of au and around the middle of 'au, a more or less pronounced peak is visible, after which the curves of both formants descend. However, in initial-stressed diphthongs both formants remain in the higher-frequency range, despite fluctuations in their frequency values; the more pronounced drop occurs only at the very end of the segment. In continuous 'ã.ũ. and á.ũ., the decline in both formants is somewhat more gradual. 13 Formants were measured in stem diphthongs. Nine measurement points were taken for each diphthong. 95 The results of the spectral analysis do not contradict the conclusion reached in § 5.1.2 that the properties of the first diphthong component are more important for accent. Although accent also changes the spectral characteristics of the second component, this is not of great importance for accent perception, because even in sustained diphthongs u, i are too short for their small differences to be noticeable. 5.2. Mixed diphthongs. 5.2.1. The duration ratio of acute and sustained mixed diphthongs is similar to that of pure diphthongs: acute diphthongs are about 1.2 times longer than sustained ones. The total duration of a diphthong depends not only on accent but also on the inherent duration of its individual components, since the longest are those acute diphthongs whose second components are nasal sonorants (see Table 5). As in Standard Lithuanian, n lengthens the diphthong the most (Pakerys 1982: 51); m is slightly shorter, so diphthongs with the structures a, e+m are also shorter. The shortest of the sonorants is r, but acute diphthongs formed with it are quite long. Since the larger share of the intensity and duration of an acute diphthong falls on the first component, it lengthens more (a – up to 259 ms, e – up to 253 ms; see Table 6) and compensates for the shortness of r. Table 5. Duration of acute and sustained mixed diphthongs Diphthong | n | x̄ | s | v (%) | confidence interval (95%) | t ≷ tₐ á.l | 8 | 334 | 5,9 | 1,8 | 326 ÷ 341 | 8,0 >5,0** á.ĺ. | 8 | 272 | 16,3 | 6,0 | 251 ÷ 292 á.r | 8 | 342 | 5,6 | 1,6 | 336 ÷ 349 | 13,4 >5,0** á.ṙ. | 8 | 265 | 11,8 | 4,4 | 250 ÷ 279 á.m | 8 | 338 | 3,9 | 1,1 | 333 ÷ 342 | 10,9 > 5,0** á.ṃ. | 8 | 279 | 11,4 | 4,1 | 265 ÷ 293 á.n | 8 | 393 | 4,1 | 1,0 | 388 ÷ 398 | 8,4 > 5,0** á.ṇ. | 8 | 348 | 11,2 | 3,2 | 335 ÷ 362 é.l | 8 | 358 | 11,8 | 3,3 | 343 ÷ 373 | 4,5 >3,8* é.ĺ. | 8 | 334 | 1,0 | 0,3 | 333 ÷ 335 é.r | 8 | 342 | 16,7 | 4,9 | 316 ÷ 369 | 5,3 >4,3* é.ṙ. | 8 | 271 | 20,9 | 7,7 | 238 ÷ 304 é.m | 8 | 363 | 6,6 | 1,8 | 354 ÷ 371 | 6,1 >5,0** é.ṃ. | 8 | 340 | 4,7 | 1,4 | 335 ÷ 346 é.n | 8 | 397 | 1,9 | 0,5 | 395 ÷ 400 | 11,1>5,0** é.ṇ. | 8 | 358 | 7,6 | 2,1 | 349 ÷ 368 5.2.2. As with diphthongs, the duration of the components indicates the accent of mixed diphthongs fairly accurately. The component-duration ratio of acute diphthongs is similar to that of Standard Lithuanian: the first components are considerably longer than the second ones (about 1.8 times). The inherent duration of the first component is not so important; lengthening depends more on the sonorant: the first components of á.r and é.r lengthen the most, because 96 r is the shortest sonorant. n is inherently quite long and accounts for a larger share of the total duration of the diphthong; therefore, the first components of á.n and é.n are only slightly longer than the second ones (see Table 6). Table 6. Duration of the first components of diphthongs Diphthong | n | x | s | v (%) | confidence interval (95%) | t ≷ tₐ á.(l) | 8 | 192 | 11,6 | 6,0 | 178 ÷ 207 | 5,2 >5,0* ā.(ī.) | 8 | 147 | 15,9 | 10,8 | 127 ÷ 167 | á.(r) | 8 | 259 | 8,7 | 3,4 | 249 ÷ 270 | 13,3 > 5,0** ā.(r̄.) | 8 | 192 | 7,3 | 3,8 | 183 ÷ 201 | á.(m) | 8 | 229 | 1,9 | 0,8 | 227 ÷ 232 | 12,6 >5,0** ā.(m̄.) | 8 | 166 | 11,1 | 6,7 | 152 ÷ 180 | á.(n) | 8 | 269 | 5,7 | 2,1 | 262 ÷ 276 | 8,6>5,0** ā.(n̄.) | 8 | 204 | 16,0 | 7,8 | 184 ÷ 224 | é.(l) | 8 | 216 | 11,0 | 5,1 | 202 ÷ 230 | 8,2 >5,0** é.(ī.) | 8 | 159 | 11,1 | 6,9 | 145 ÷ 173 | é.(r) | 8 | 253 | 4,7 | 1,9 | 246 ÷ 261 | 13,8 >6,0** é.(r̄.) | 8 | 189 | 8,0 | 4,3 | 176 ÷ 202 | é.(m) | 8 | 248 | 6,7 | 2,7 | 240 ÷ 257 | 5,1 >5,0** é.(m̄.) | 8 | 215 | 13,0 | 6,0 | 199 ÷ 231 | é.(n) | 8 | 283 | 26,6 | 9,4 | 250 ÷ 316 | 4,7 >3,8* é.(n̄.) | 8 | 208 | 23,2 | 11,1 | 179 ÷ 237 The second components of sustained diphthongs are almost always shorter than the first. The duration of the components of diphthongs a, e+l, m, n differs less, whereas the component-duration ratio of a, e + r is more reminiscent of acute accent: duration tends to be concentrated in the first component. Although the intensity and duration of both accents are concentrated in the first components, the duration of the first components in sustained and acute diphthongs differs significantly: the probability of distinguishing all pairs exceeds 99%. The second components differ much less, and for some (ar, er and an) the probability of distinguishing them does not exceed 95%; the value of the t criterion is below the critical value (see Table 7). Table 7. Duration of the second components of diphthongs (ms) Diphthong | n | x | s | v (%) | confidence interval (95%) | t ≷ tₐ (á.)l | 8 | 124 | 4,2 | 3,3 | 119 ÷ 129 | 2,8 >2,3* (ā.)l̄. | 8 | 132 | 4,8 | 3,6 | 126 ÷ 138 | (á.)r | 8 | 83 | 7,9 | 9,4 | 74 ÷ 93 | 1,6 <2,3 (ā.)r̄. | 8 | 94 | 12,9 | 13,7 | 78 ÷ 111 | (á.)m | 8 | 111 | 2,4 | 2,1 | 100 ÷ 114 | 7,3 >5,0** (ā.)m̄. | 8 | 125 | 3,4 | 2,7 | 120 ÷ 129 | (á.)n | 8 | 123 | 3,1 | 2,5 | 119 ÷ 127 | 2,3 = 2,3 (ā.)n̄. | 8 | 131 | 7,1 | 5,4 | 122 ÷ 140 97 7 lentelės tęsinys Diphthong | n | x | s | v (%) | confidence interval (95%) | t ≫ tₐ (ė.)l | 8 | 131 | 2,9 | 2,2 | 127 ÷ 135 | 15,9 > 5,0** (ė.)ī. | 8 | 163 | 3,4 | 2,1 | 158 ÷ 167 (ė.)r | 8 | 91 | 11,9 | 13,0 | 72 ÷ 110 | 0,6 < 2,5 (ė.)r. | 8 | 85 | 17,3 | 20,4 | 57 ÷ 113 (ė.)m | 8 | 116 | 1,1 | 1,0 | 115 ÷ 117 | 13,7 > 5,0** (ė.)m. | 8 | 148 | 2,9 | 1,9 | 145 ÷ 152 (ė.)n | 8 | 130 | 4,8 | 3,7 | 124 ÷ 136 | 11,1 > 5,0** (ė.)n. | 8 | 156 | 2,3 | 1,5 | 154 ÷ 159 5.2.3. Qualitative differences between acute and sustained diphthongs usually arise from the properties of their first components. It is commonly thought that sonorant quality has little influence on them (Pakerys 1982: 64), and therefore only vowels were examined here as well. In stressed acute diphthongs, both a formants are higher (see Fig. 3c and d). Thus, this sound is pronounced with a more open mouth and has a lower height and a more front articulation than a in a sustained diphthong. The sustained a is more back, closer, and lower in pitch (cf. ā – -120, and ā – -220). Fig. 3. Spectra of the first components of mixed diphthongs14 14 Only a few spectra of the first components are presented here. In diphthongs of a different structure, the formant arrangement is similar. 98 The vowel e in an acute diphthong is also lower in height and more open, resembling an æ-type sound in the middle of the segment. Its articulation is more back than that of the corresponding sustained sound, which is higher in height and slightly narrowed. The qualitative indices of the components also differ: e in sustained diphthongs is more diffuse (cf. the compactness index of ė, 900, and ē, 858), more tense, and higher in pitch (cf. é, 400, and ē, 470). In Standard Lithuanian, the first components of acute diphthongs are more tense, since in circumflex diphthongs the vocal effort and tension fall on the second component (cf. Pakerys 1974: 150-151). In the Eržvilkas dialect, the right-hand part of the duration of the sustained accent and, as can be seen, of the tension falls on the first component, while the quality of the second is not so important. 6. The experiment showed that both duration and sound quality are important indicators for distinguishing accents, complementing one another. Moreover, measurements of these parameters refine the definition of the sustained accent previously given for the southern fringe of the Raseiniai dialect area: vocal effort is not divided between both components of the diphthong (cf. Zinkevičius 1994: 98), but most of it is assigned to the first component. This is clearly demonstrated by the ratio of the components’ durations (see § 5.1.2.). Of course, the distribution of vocal effort in the sustained accent of Eržvilkas differs considerably from the circumflex of the northern Samogitian dialects: although the first component accounts for a greater proportion of the diphthong’s duration, it is considerably shorter than the corresponding acute component and remains an important indicator of accent. By contrast, the duration of the first components in other Samogitian dialects is said to be almost independent of accent: for the Varniai dialect speakers, the duration of the second component is a better indicator of accent, while for northern Samogitian speakers it is the duration of the entire diphthong (Bukantis 1984a: 73; Girdenis 1974: 182). The qualitative characteristics of the first components are no less important, although they are likewise not as significant in other Samogitian dialects (cf. Bukantis 1984a: 79; Girdenis 1974: 171). In this respect, the Eržvilkas dialect is more like Standard Lithuanian (cf. Pakerys 1974: 151). Qualitative differences between the components improve the possibility of recognizing the accents: the first component of a sustained diphthong is not only shorter but also somewhat higher in height, usually more front, and more prone to assimilate to the second component. In summary, it can be concluded that the accents of the Eržvilkas dialect are in many respects similar to those of Standard Lithuanian and are intermediate between Samogitian and Aukštaitian accents15. The sustained accent has retained more Samogitian features, while the acute accent differs little from Standard Lithuanian. Received 1998 04 06 15 It would be interesting to compare the accents of the Raseiniai dialect, especially the sustained accent, with those of the neighboring Western Aukštaitian dialects, since by ear they are almost indistinguishable. Unfortunately, a more detailed comparison is hindered by the lack of an experimental study of the accents of the northern part of the Kaunas dialect. 99 LITERATŪRA Atkočaitytė D. 1997: Dėl į ir u atvirumo pietų žemaičių raseiniųkių tarmėje.– Kalbotyra 46 (1), Vilnius, 5–12. Bagvilaitė R. 1975: Varnių tarmės fonetika: Diplominis darbas. Vilnius: Vilniaus universitetas. Bukantis 1984a: Б у к а н т и с Й. Фонология южножемайтского Варняйского говора: Дис. канд. филол. наук, Т.1, Вильнюс: Вильнюсский государственный университет. Bukantis 1984b: Б у к а н т и с Й. Фонология южножемайтского Варняйского говора: Дис. канд. филол. наук, Т.2, Вильнюс: Вильнюсский государственный университет. Girdenis A. 1974: Prozodinės priegaidžių ypatybės šiaurės žemaičių tarmėje.– Eksperimentinė ir praktinė fonetika, Vilnius, 160–198. Girdenis A. 1996: Energetinė šiaurės žemaičių tarmės priegaidžių fonetinės prigimties interpretacija.– Baltistica 31(1), Vilnius, 71–84. Girdenis A., Pupkis A. 1974: Pietinių vakarų aukštaičių priegaidės: (Prozodiniai požymiai).– Eksperimentinė ir praktinė fonetika, Vilnius, 107–125. Grinaveckis V. 1973: Žemaičių tarmių istorija (fonetika), Vilnius: Mintis. Jonikas P. 1939: Pagramančio tarmė, Kaunas: Spindulys. Kosienė O. 1982: Rytų aukštaičių uteniškių monoftongų priegaidės.– Kalbotyra 33 (1), Vilnius, 61–71. Pakerys A. 1968: Lietuvių literatūrinės kalbos sudėtinių dvibalsių au, ai, ei, akustiniai požymiai.– Eksperimentinės fonetikos ir kalbos psichologijos kolokviumo medžiaga 3, Vilnius, 7–118. Pakerys ir kt. 1972: П а к е р и с А., П л а к у н о в а Т., У р б е л е н е Я. Относительная длительность дифтонгов литовского языка.– Garsai, priegaidė, intonacija, Vilnius, 3–36. Pakerys A. 1974: Tvirtapradžių ir tvirtagalių dvigarsių spektras.– Eksperimentinė ir praktinė fonetika, Vilnius, 142–155. Pakerys ir kt. 1974: P a k e r y s A., P l a k u n o v a T., U r b e l i e n ė J. Lietuvių kalbos mišriųjų dvigarsių santykinė trukmė.– Eksperimentinė ir praktinė fonetika, Vilnius, 3–47. Pakerys A. 1982: Lietuvių bendrinės kalbos prozodija, Vilnius: Mokslas. Piotrovskis 1960: П и о т р о в с к и й Р.Г. Ещё раз о дифференциальных признаках фонемы.– Вопросы языкознания 6, Москва, 24–38. Slušinskaitė M. 1981: Lominių šnektos fonetika: Diplominis darbas. Vilnius: Vilniaus universitetas. Tamošaitytė-Jokubauskienė V. 1981: Pagramančio šnektos fonetika: Diplominis darbas. Vilnius: Vilniaus universitetas. Zinkevičius Z. 1966: Lietuvių dialektologija, Vilnius: Mintis. Zinkevičius Z. 1994: Lietuvių kalbos dialektologija, Vilnius: Mokslo ir enciklopedijų leidykla. 100 AKZENTE DER DIPHTHONGE UND DER GEMISCHTEN ZWIELAUTE IM DIALEKT VON ERŽVILKAS: DAUER UND SPEKTREN Zusammenfassung Im südžemaitischen Dialekt von Raseiniai können die betonten Zwielaute einen Gravis- oder Zirkumflexakzent haben. Im vorliegenden Beitrage sollen die Akzente der Diphthonge ei, ai, au und der gemischten Zwielaute a, e + l, m, n, r einer Mundart erörtert werden: ihre Dauer und qualitative Charakteristik sowie die wichtigsten unterschiedlichen Kennzeichen der Akzente. Die Unterschiede zwischen den Akzenten wurden experimentell untersucht: es wurde die Dauer des ganzen Zwielautes und seiner einzelnen Komponenten gemessen, die Spektralanalyse angewendet, die qualitativen Kennzeichen der Komponenten (Kompaktheit, Tonalität, u. a.) Die Resultate des Experiments beweisen, daß sowohl die Dauer als auch die Qualität der Komponenten wichtige, einander ergänzende Indikatoren zur Unterscheidung der Akzente sind. Außerdem präzisieren die Messungen dieser Parameter die bisherige Charakteristik des Zirkumflexakzentes: der Atemdruck verteilt sich nicht gleichmäßig auf die beiden Komponenten des Zwielautes (vgl.: Zinkevičius 1994: 98) - der größere Teil des Druckes bestimmt die erste Komponente. Die erste Komponente jedoch, obwohl ihr der größere Teil der Dauer des Zwielautes zukommt ist viel kürzer als die entsprechende Zwielautkomponente mit Gravisakzent. Das ist sehr wichtig für die Unterscheidung der Akzente, da die Dauer und die Qualität der zweiten Komponente von dem Akzent wenig abhängt. Die Länge der ersten Komponente bestimmt auch die Gesamtdauer des Zwielautes: der Zwielaut mit Gravisakzent ist 1,2-mal länger als der mit Zirkumflexakzent. Wichtig ist auch die qualitative Charakteristik der ersten Komponente Die qualitativen Unterschiede der Komponenten erlauben das Wesen der Akzente besser zu erkennen: beim Zirkumflexakzent des Zwielautes ist die erste Komponente nicht allein kürzer, sondern auch dem Ton nach ein wenig höher und gewöhnlich als vordere geneigt sich der nachfolgenden anzugleichen. Die Resultate des Experiments zeigen, daß für die Akzentanalyse im Dialekt von Eržvilkas die Gesamtheit der die Dauer und Qualität kennzeichnenden Charakteristika von Bedeutung ist. Die Akzente in diesem Dialekt nehmen eine Stellung ein zwischen dem Žemaitischen und dem Aukštaitischen: der Zirkumflexakzent hat eher žemaitische Züge, während sich der Gravisakzent von dem entsprechenden in der Gemeinsprache beinahe überhaupt nicht unterscheidet. 101