Atvirojo ir uždarojo skiemens balsių trukmė pietinių vakarų aukštaičių tarmėje

LIETUVIŲ KALBOTYROS KLAUSIMAI XL (1998) Asta KAZLAUSKIENĖ Vytautas Magnus University, Kaunas VOWEL DURATION IN OPEN AND CLOSED SYLLABLES IN THE SOUTHERN WESTERN HIGHLANDER DIALECT 1. V. Vaitkevičiūtė (1960: 207–208), describing the phonetic factors that determine vowel length in Standard Lithuanian, states that vowels and diphthongs in closed syllables are supposedly shorter than those in open syllables, because some of the energy is also expended on the consonant; in word-final closed syllables, she argues, vowels are longer than in other positions, but shorter than vowels in open word-final syllables. Thus, according to V. Vaitkevičiūtė, in Standard Lithuanian the longest vowels are in open word-final syllables, shorter ones in closed word-final syllables, and the shortest inside the word. However, more recent research into the duration of internal and open word-final vowels in the southern Kaunas subdialect of the Western Highlander dialect (hereafter PVA) (Kazlauskienė 1998) shows that the dialect on which Standard Lithuanian is based is characterized by quantitative reduction of word-final vowels. Thus, open word-final vowels in the PVA dialect are certainly not longer than internal vowels. It remains to determine whether in the PVA dialect1 vowel duration (especially at word endings) depends on syllable type (openness or closedness). 1.1. Uždarųjų ir atvirųjų galūnės balsių trukmė tirta žodžiuose2: renkà·s „renkąs“ – renkā, „renką“, rankàs „rankas“ – rankà „ranką“, rẽp.kas (sutrump. es. l. 3 asm. „renkasi“) „renkas“ – rẽj.ka „renka“, lenkẽ·s „lenkiąs“ – lenkẽ. „lenkią“, lẽj.kae.s „lenkės“ – lẽj.kae. (veik. r. būt. k. l. dalyvio dgs. vard.) „lenkė“, silkẽ·s „silkes“ – silkè (vns. įnag.) „silke“, lẽj.kæs (sutrump. es. l. 3 asm. „lenkiasi“) „lenkias“ – lẽn.kæe „lenkia“, ni.kšti·s „nykštės“ – ni.kšti· „nykštį“, ká.rti.s „kartys“ – ká.rti „kartį“, ru·kštis „rūgštis“ – ru·kšti „rūgštį“, bá·rtis „bartis“ – bá·rti „bartį“, rú·kštu.s 1 This study focuses only on those Western Highlander Kauniškiai who, according to the prevailing view (e.g., Salys 1946: 51–52; Zinkevičius 1966: 118 ff.; LKT 1970: 28–29; LKA 2 1982), do not shorten word-final vowels. 2 In transcribing the examples, the assumption was adopted that the front, low vowel [e] is more open before hard consonants, back vowels, and at the end of a word, e.g.: rẽ·sto. “of a rye field,” rẽ·sti “to a rye field,” rẽtù “rare,” sár.tæ. “sartę,” svíæste (singular locative) “in the world.” Of course, this is not so in all PVA subdialects. From approximately Lekėčiai, Višakio Rūda, Kazlų Rūda, and Marijampolė eastward, it is usually only the stressed long or half-long stem [e] that is opened (regardless of the softness of the following consonant). Since only two speakers came from these localities, the variant representing the pronunciation of the majority of the speakers was chosen. In presenting the words under study, only the softness of consonants occurring before back vowels was marked. Since it has already been established (Kazlauskienė 1998) that in the PVA dialect unstressed long word-final and even stem vowels are shortened (except for vowels occurring before a short ending, e.g.: dra·sa “courage,” ru·pi “worry,” ti·ki “faith,” tre·ši “to thresh,” ke·dæ “chair”), these vowels are marked as half-long. 113 “sour” – rú·kštu... “sour,” ru·kštùs “sour” – ru·kštù “sour,” bú·stus “dwellings” – bú·stu “in a dwelling.” The duration of internal vowels in closed and open syllables was also analyzed. The following minimal word pairs were selected for this purpose: sùktas (singular nominative) “twisted” – sùkas (abbrev. pres. tense 3rd person “twists”) “twists,” suktàs (plural genitive) “twisted” – sùkas (verb “to twist,” future tense 3rd person) “will twist,” kã·pta (impersonal gender) “baked” – kẽ·pa “bakes,” kepti (passive voice, past tense participle, plural nominative) “baked” – kẽpi “bakes.” For the experiment, examples were selected in which the quantity of vowels before a single stop and before a fricative-plus-stop consonant cluster was compared: suskì “with Suskys” – šùki “with Šukys,” sùskis “a suskis” – sùkis (verb “to twist,” imperative singular 2nd person) “twist.” In addition, the duration of vowels before a single fricative and before a fricative-plus-stop cluster was also to be compared. The words selected for this were: pẽ·štù “I comb” – pešù “I comb,” pešti (infinitive) “to comb” – peši (future tense 2nd person) “you will comb.” 1.2. Short sentences with clear meanings and similar rhythms were constructed using these words, with the target words placed in the middle. The sentences were read by 8 speakers (2 women, 6 men) of the Kaunas subdialect (4 kapsai and 4 zanavykai). Their mean age was 62 years. 1.3. Sound duration was measured using an IBM-type personal computer and the speech-signal processing program “KALBAM 43,” developed at Vilnius University by P. Kasparaitis (programming language TURBO-PASCAL, v. 6.0)3. The words read by the speakers were recorded in the computer’s memory, and the program-generated oscillogram was displayed on the screen. After listening to it, it can be read as separate sounds, whose durations are measured by the program itself. 1.4. Measurement results were processed using methods of mathematical statistics: the arithmetic mean (x̄), standard deviation (s), coefficient of variation (v), and 95% confidence interval (÷) were calculated. The significance of the results was also tested using Student’s criterion. Statistical data analysis was performed using the program “STUDENT,” developed by A. Girdenis (programming language TURBO-PASCAL, v. 7.0). The formulas used by the program are well known from other works in experimental phonetics4. 2. The results show that the quantitative ratios of vowels in open and closed word-final syllables are approximately as follows: [−]/[— #]5: [—]/[— s] = 1: 1.2; [≃]/[— #]: [≃]/[— s] = 1: 1.1; [−]/[— #]: [−]/[— s] = 1: 1.3; [˘]/[— #]: [˘]/[— s] = 1: 1.3. Thus, vowels in closed word-final syllables are on average 1.2 times longer than vowels in open word-final syllables. Comparing these ratios, one may assume that the duration difference between short vowels in open and closed word-final syllables is greater. 2.1. The quantitative ratios of circumflexed word-final vowels in the positions under study are as follows: [ã·]/[— #]: [ã·]/[— s] = 1: 1.1; [ẽ·]/[— #]: [ẽ·]/[— s] = 1: 1.1; [ĩ·]/[— #]: [ĩ·]/[— s] = 1: 1.2. Comparing the quantity ratios of circumflexed and other vowels (see Section 2), this is the smallest duration difference. The examples studied probably influenced this result (ręnkã·s – ręnkã·, leņkẽ·s – leņkẽ·, ni.kštĩ·s – ni.kštĩ·), since two pairs consisted of active-voice present-tense participles, which, unfortunately, are now used increasingly rarely, while the open word ending of the third pair 3 Trumpą jos aprašymą žr. (Girdenis ir kt. 1995: 15–21). 4 All sound measurements and statistical calculations were carried out in the Department of General Linguistics at VU. 5 This symbol denotes the position occupied by a vowel: [−] – long unstressed, [˘] – short unstressed, [≃] – circumflexed, [˘] – short stressed vowel, [— #] – open word ending, [— s] – closed word ending, [— C] – word-internal position. 114 secondary (ni.kšti ‘dwarf’), it is presumably inherently longer than the primary one. On the other hand, the duration of circumflex ending vowels may depend less on syllable type. T a b l e 1 Duration of open and closed ending vowels (ms) and its statistical assessment6 Examples | n | x̄ (ms) | s (ms) | v (%) | 95% confidence interval (ms) | tₚ > tα [ræŋká:s] | 19 | 161,5 | 36,6 | 22,6 | 143,9÷179,2 | 0,88<t₀,₀₅=2,03 [ræŋká] | 20 | 150,9 | 38,7 | 25,6 | 132,8÷169,0 [leŋkė:s] | 14 | 186,5 | 39,8 | 21,3 | 163,5÷209,5 | 1,03<t₀,₀₅=2,06 [leŋkė] | 14 | 170,8 | 40,7 | 23,8 | 147,3÷194,3 [ni.kšti:s] | 18 | 123,5 | 34,8 | 28,2 | 106,2÷140,8 | 1,52<t₀,₀₅=2,03 [ni.kšti] | 18 | 105,4 | 36,6 | 34,7 | 87,3÷123,6 [≈]/[— s] | 51 | 155,0 | 44,2 | 28,5 | 142,5÷167,4 | 1,62<t₀,₀₅=2,01 [≈]/[— #] | 52 | 140,5 | 46,4 | 33,0 | 127,6÷153,4 [leŋ.kae:s] | 10 | 129,5 | 21,0 | 16,2 | 114,5÷144,5 | 0,78<t₀,₀₅=2,10 [leŋ.kae] | 10 | 121,8 | 23,1 | 18,9 | 105,3÷138,3 [ka:rti:s] | 18 | 83,2 | 15,8 | 19,0 | 75,3÷91,0 | 3,76>t₀,₀₀₁=3,61 [ka:rti] | 17 | 64,2 | 13,9 | 21,7 | 57,1÷71,4 [ru:kštu:s] | 18 | 94,2 | 17,2 | 18,2 | 85,6÷102,7 | 5,70>t₀,₀₀₁=3,60 [ru:kštu] | 18 | 64,7 | 13,6 | 21,1 | 57,8÷71,5 [—]/[— s] | 46 | 97,5 | 24,7 | 25,3 | 90,2÷104,9 | 3,52>t₀,₀₀₁=3,52 [—]/[— #] | 45 | 77,2 | 28,9 | 37,4 | 68,5÷85,9 [raŋkàs] | 23 | 89,9 | 18,3 | 20,3 | 82,0÷97,8 | 2,69>t₀,₀₁=2,69 [raŋkà] | 23 | 73,8 | 22,1 | 29,9 | 64,3÷83,4 [silkès] | 21 | 100,7 | 22,5 | 22,3 | 90,5÷111,0 | 2,40>t₀,₀₅=2,02 [silkė] | 21 | 83,7 | 23,4 | 27,9 | 73,1÷94,4 [ru:kštis] | 22 | 73,6 | 18,6 | 25,3 | 65,3÷81,9 | 4,04>t₀,₀₀₁=3,54 [ru:kšti] | 22 | 54,1 | 12,8 | 23,7 | 48,4÷59,8 [ru:kštùs] | 22 | 78,5 | 16,4 | 20,9 | 71,2÷85,7 | 4,85>t₀,₀₀₁=3,55 [ru:kštù] | 21 | 55,6 | 14,5 | 26,0 | 49,0÷62,2 [ˇ]/[—s] | 88 | 85,6 | 21,4 | 25,0 | 81,0÷90,1 | 5,67>t₀,₀₀₁=3,41 [ˇ]/[— #] | 87 | 66,8 | 22,3 | 33,3 | 62,1÷71,6 [reŋ.kas] | 19 | 78,0 | 19,7 | 25,3 | 68,5÷87,5 | 2,85>t₀,₀₁=2,72 [reŋ.ka] | 19 | 59,9 | 19,4 | 32,3 | 50,6÷69,2 [leŋ.kae:s] | 16 | 91,6 | 23,2 | 25,3 | 79,3÷104,0 | 2,11>t₀,₀₅=2,04 [leŋ.kae] | 16 | 76,3 | 17,5 | 22,9 | 67,0÷85,6 6 Meanings of the symbols used in the statistical assessment tables: n – number of measurements, x̄ – arithmetic mean, s – standard deviation, v – coefficient of variation, 95% pasikl. interv. – 95% confidence interval, tₚ – calculated Student’s t value, tα – critical Student’s t value, α – significance level of the results. 115 Examples n x̄ (ms) s (ms) v (%) 95% confidence interval (ms) tₚ < tₐ [bā́.r̥tis] 19 61,2 9,3 15,2 56,7÷65,7 5,80>t₀,₀₀₁=3,59 [bā́.r̥ti] 18 43,8 8,9 20,3 39,4÷48,3 [búː.stus] 20 61,7 16,5 26,7 54,0÷69,4 4,20>t₀,₀₀₁=3,57 [búː.stu] 20 42,9 11,3 26,3 37,6÷48,2 [—]/[—s] 74 72,2 21,3 29,4 67,3÷77,1 5,13>t₀,₀₀₁=3,43 [—]/[—#] 73 54,9 19,7 35,9 50,3÷59,5 As can be seen from the results of the statistical analysis presented in Table 1, the differences in the duration of circumflex open and closed ending vowels are not statistically significant. Nevertheless, the differences between the mean durations of the vowels being compared are not small ([āː]/[—s] − [āː]/[—#] = 10,6 ms; [ēː]/[—s] − [ēː]/[—#] = 15,7 ms, [īː]/[—s] − [īː]/[—#] = 18,1 ms), and would become apparent if the number of realizations were increased. 2.2. The ratios between unstressed long⁷ vowels in open and closed endings are approximately as follows: [eː]/[—#]: [eː]/[—s] = 1: 1,1; [iː]/[—#]: [iː]/[—s] = 1: 1,3; [uː]/[—#]: [uː]/[—s] = 1: 1,5. The statistical significance of these vowels differs somewhat from that of circumflex vowels. The duration differences for the high vowels [iː, uː] in the positions under study are statistically significant: the confidence intervals established do not even overlap, and the calculated Student’s t values exceed the highest critical values for this criterion at the 0.001 level. Only the duration difference for the vowel [eː] is small ([eː]/[—s] − [eː]/[—#] = 7,7 ms) and statistically unreliable. 2.3. Stressed short vowels in open endings relate to the corresponding vowels in closed endings approximately as follows: [ă]/[—#]: [ă]/[—s] = 1: 1,2; [ĕ]/[—#]: [ĕ]/[—s] = 1: 1,2; [ĭ]/[—#]: [ĭ]/[—s] = 1: 1,4; [ŭ]/[—#]: [ŭ]/[—s] = 1: 1,4. The duration differences between all short stressed vowels in open and closed endings are statistically significant, since the confidence intervals for duration barely overlap and the calculated Student’s t values exceed the critical values for this criterion. 2.4. Unstressed short vowels in open endings are also shorter than those in closed endings. Their ratios are as follows: [a]/[—#]: [a]/[—s] = 1: 1,3; [e]/[—#]: [e]/[—s] = 1: 1,2; [i]/[—#]: [i]/[—s] = 1: 1,4; [u]/[—#]: [u]/[—s] = 1: 1,4. The duration differences for these vowels are statistically reliable: the confidence intervals for low vowels overlap slightly, those for high vowels do not overlap at all, and the calculated Student’s t values exceed the critical values for that criterion. 7 As already mentioned, long vowels are shortened in unstressed syllables; therefore, the term long is used here in a qualified sense. 116 3. Although circumflex- and short-vowelled (both stressed and unstressed) vowels in closed endings are longer than those in open endings, they are shorter than the corresponding stem vowels8 (see Table 2). Table 2 Duration ratios between certain stem vowels and vowels in open and closed endings9 Vowels | Position and duration ratios | P: U Open ending | Closed ending | Penultimate syllable [āˑ] | 1 | 1,1 | 1,3 | 1,3 [ēˑ] | 1 | 1,1 | 1,2 | 1,1 [īˑ] | 1 | 1,2 | 1,5 | 1,3 [≈] | 1 | 1,1 | 1,3 | 1,2 [à] | 1 | 1,2 | 1,4 | 1,2 [è] | 1 | 1,2 | 1,5 | 1,2 [ì] | 1 | 1,4 | 1,8 | 1,3 [ù] | 1 | 1,4 | 1,5 | 1,1 [ˋ] | 1 | 1,3 | 1,5 | 1,2 [a] | 1 | 1,3 | 1,5 | 1,2 [e] | 1 | 1,2 | 1,2 | 1 [ɪ] | 1 | 1,4 | 1,7 | 1,2 [u] | 1 | 1,4 | 1,8 | 1,2 [ˈ] | 1 | 1,3 | 1,5 | 1,2 Table 2 shows that in the compared positions, the duration ratios of high vowels are greater than those of low vowels (cf. [Vʲi]/[— #]: [Vʲi]/[— s]: [Vʲi]/[— C—] = 1: 1,3: 1,5 and [Vʲe]/[— #]: [Vʲe]/[— s]: [Vʲe]/[— C] = 1: 1,1: 1,2). Thus, the duration of high vowels depends more on position. This is not an unexpected result, since, as A. Pakeris’s study of vowel perception in the standard language shows (1982: 96–102), quality is more important for distinguishing the opposition between high long and short vowels, whereas duration is more important for low vowels. Therefore 8The duration of stem vowels was measured in the words: mā:sto ‘thinks’ ([āˑ] – 201,4 ms), grã:štu ‘would drill’ ([æˑ] – 208,7 ms), grì:štu ‘would return’ ([īˑ] – 154,0 ms); pė.gręž.tù ‘would redrill’ ([æ] – 92,0 ms), pė.rgri.stu ‘would return’ ([ì] – 88,1 ms), pãr.pu.stu ‘would blow through’ ([ù] – 80,8 ms); rãsti (standard) ‘to find’ ([à] – 103,0 ms), rùsti ‘to rust’ ([ù] – 83,2 ms), trešti ‘to crush’ ([æ] – 123,1 ms), rišti (standard) ‘to tie’ ([ì] – 98,4 ms); ratù ‘by wheel’ ([a] – 91,8 ms), krutù ‘I move’ ([u] – 76,4 ms), retù ‘rare’ ([e] – 93,3 ms), ritù ‘I roll’ ([ɪ] – 73,8 ms). The speakers, recording, and duration-measurement methodology were the same (see Sections 1.2 and 1.3). The mean vowel duration is given in parentheses. 9 P: U – the ratio between the duration of vowels in the penultimate syllable and in the closed ending. 117 the duration of high vowels may depend more on position than that of low vowels. The results of comparing the duration of long unstressed stem vowels with that of vowels in open and closed endings are very interesting. Only the high stem vowel [i·] is longer than vowels of the [i·] type in open and closed endings (their duration ratios are approximately as follows: [i·]/[— #]: [i·]/[— s]: [i·]/[— C] = 1: 1,3: 1,4). The high stem vowel [u·] is longer than the [u·] in an open ending, but shorter than the [u·] ([u·]/[— #]: [u·]/[— C]: [u·]/[— s] = 1: 1,2: 1,5) in a closed ending. The low vowel [e·] in the post-stress penultimate syllable is shorter than the vowel in both the closed ending and even the open ending [e·] ([e·]/[— C]: [e·]/[— #]: [e·]/[— s] = 1: 1,3: 1,4). This shows that post-stress low vowels are considerably shortened in the PVA dialect. 4.0. It would be useful to compare the duration of ending and stem vowels in open and closed syllables. However, this is difficult to do precisely, because suitable pairs are hard to find. Moreover, when syllable boundaries are determined according to the phonological syllable model (STR)10, the number of closed stem syllables is further reduced. This is because closed endings occur only with the consonant [s], and this consonant within a word attaches to the following syllable. Three groups of examples were selected for the stem-vowel experiment: in the first, vowels occurring before one (V-TV) and before two plosive consonants (VT-TV) were compared; in the second, vowels before a fricative and a plosive (V-STV) were compared with those before one plosive consonant (V-TV); and in the third, vowels before a fricative and a plosive (V-STV) were compared with those before one fricative consonant (V-SV). 4.1. The results of duration measurements and their statistical assessment for examples in the first group (the vowel under study occurs before one and before two plosive consonants) show that syllable type (open or closed) may not have a substantial effect on the duration of stem vowels that are not part of diphthongs (see Table 3). Only in the words kã̄pta and kã̄pa is the difference in stem-vowel duration statistically significant: the confidence intervals overlap only slightly, and the calculated Student’s t-test value at the 0.05 level is slightly above the critical value. The mean stem-vowel durations in the other three pairs (sùktas – sùkas, suktȧs – sukàs, keptì – kepì) are very similar, and their differences are not statistically significant. It seems that stem vowels in open syllables should be longer than those in closed syllables. However, this can be observed only in suktàs – sukàs11. 10 For discussion of it, see (Girdenis 1985: 5–11; 1995: 115–121; Pakerys 1995: 286–288). 11 Incidentally, in the word sukàs, the syllable boundary coincides with the morpheme boundary (between the prefix su- and the future-tense stem kas of the verb kąsti). 118 Table 3 Duration of vowels in internal open and closed syllables (ms) and its statistical assessment Examples | n | x̄ (ms) | s (ms) | v (%) | 95% confidence interval (ms) | tₚ > tₐ [sùktas] | 15 | 65,9 | 12,8 | 19,4 | 58,8 ÷ 73,0 | 0,02 < t₀,₀₅ = 2,05 [sùkas] | 15 | 65,7 | 16,8 | 25,5 | 56,4 ÷ 75,0 [suktás] | 15 | 51,6 | 9,3 | 17,9 | 46,5 ÷ 56,7 | 1,73 < t₀,₀₅ = 2,05 [sukás] | 15 | 58,5 | 12,3 | 21,0 | 51,7 ÷ 65,3 [kǽ:pta] | 12 | 159,8 | 17,1 | 10,7 | 149,0 ÷ 170,7 | 2,26 > t₀,₀₅ = 2,07 [kǽ:pa] | 12 | 177,1 | 20,1 | 11,4 | 164,3 ÷ 189,9 [kepˈti] | 15 | 72,3 | 25,9 | 35,9 | 57,9 ÷ 86,6 | 0,18 < t₀,₀₅ = 2,05 [kepˈi] | 15 | 74,1 | 28,2 | 38,0 | 58,5 ÷ 89,7 [sùskis] | 12 | 83,8 | 9,9 | 11,8 | 77,6 ÷ 90,1 | 3,32 > t₀,₀₀₅ = 3,19 [sùkis] | 12 | 68,4 | 12,7 | 18,6 | 60,3 ÷ 76,5 [suškù] | 6 | 72,8 | 5,4 | 7,4 | 67,2 ÷ 78,5 | 3,93 > t₀,₀₀₅ = 3,58 [šukù] | 6 | 61,8 | 4,3 | 6,9 | 57,4 ÷ 66,3 [pẽšti] | 12 | 102,7 | 11,3 | 11,0 | 95,5 ÷ 109,8 | 1,49 < t₀,₀₅ = 2,07 [pẽši] | 12 | 95,8 | 11,5 | 12,0 | 88,4 ÷ 103,1 [pešti] | 9 | 79,4 | 11,1 | 13,9 | 70,9 ÷ 88,0 | 0,95 < t₀,₀₅ = 2,11 [peši] | 9 | 74,4 | 11,3 | 15,1 | 65,8 ÷ 83,1 I think these contradictory results (one duration difference is statistically significant, three are insignificant, and in the latter cases a tendency toward longer open-syllable vowels is barely discernible) can be explained only by the possibility that stressed vowels are longer in open than in closed syllables, whereas syllable type does not affect the quantity of short vowels¹². On the other hand, according to A. Girdenis (DLKG 1994: 36–37), Lithuanian Standard Language (and thus probably PVA as well) is characterized by a tendency toward open syllables, so in connected speech closed syllables may become open. The duration analysis of these examples seems to suggest that here we either have phonetically open syllables, or syllable type does not affect the quantity of short vowels. 4.2. The duration of internal vowels before a fricative and a stop (V-STV) and before a single stop consonant (V-TV) differs somewhat more, as the statistical analysis of their length differences shows. 12 The word kǽpta is, according to some speakers themselves, used very rarely or not at all. Therefore, the vowel in it may have been pronounced shorter than is usual in this position. Perhaps the very frequent word keptà, “baked,” influenced this. 119 In the examples suškù – šukù, suškis – šukis, the duration differences between internal vowels are statistically significant: the confidence intervals calculated do not overlap, and the calculated Student’s t values exceed the critical values at the 0.005 level. The vowels are longer in these examples before a fricative and a stop consonant (V-STV). If the duration of vowels in internal open and closed syllables (before one and before two stop consonants) differed, one could argue that in words of the suškù type the syllable boundary falls between the fricative and the stop consonant. In that case, open internal syllables would be shorter, and closed ones longer (as are word-final syllables). However, the duration of vowels in internal open and closed syllables does not differ. Therefore, other reasons must be sought for the fact that a vowel before a fricative and a stop (V-STV) is longer than one before a single stop (V-TV). Most likely, in this case the greater vowel duration was due to the following fricative, since, as the research results show, vowels are longer in that position. When a fricative is articulated, a constriction is formed in the oral cavity, whereas when a stop is articulated, a closure is formed (LKG 1965: 68–88; Mikalauskaitė 1975: 38–50; Pakerys 1995: 88). More energy is needed to overcome the latter than the constriction; consequently, adjacent vowels may be shorter13. 4.3. In the examples pešti – peši, pešti – peši, the differences in vowel duration are small, and there is no statistical basis for claiming that the quantity of these vowels differs. Nevertheless, one can observe a tendency for vowels to be longer before a fricative and a stop consonant than before a single fricative. As the preceding study shows, open-syllable (V-TV) vowels are slightly longer than closed-syllable (VT-TV) vowels. Therefore, in the examples pešti, pešti the syllable boundary cannot fall between the fricative and the stop consonant, because then the vowel of the closed syllable would be longer, contradicting the results of the duration study of open and closed syllables whose boundaries are not in doubt. Thus, it can be concluded that these words contain only open internal syllables, and the syllable boundaries are: pe-šti, pe-ši, pe-šti, pe-ši. The experiment therefore clearly supports the claims of phonological theory (Girdenis 1995: 115–121), according to which the ST cluster does not close a syllable. 4.4. Such statistically insignificant differences in the duration of vowels in open and closed stem syllables would indicate that A. Girdenis (1985: 6; 1995: 115–16) and A. Pakerys (1995: 282–283) were probably right in arguing that there are no phonetic cues signaling the beginning and end of a syllable14. This was also confirmed by 13 On the other hand, a comparison of the words examined shows that suškù “suskù” and suškis “suskis” are not entirely neutral. Although the speakers were asked to read the texts with as neutral an intonation as possible, these words may have been pronounced more emotionally than the comparison forms šukù “šuku” and šukis “sukis”. 14Perhaps for the same reason, syllable boundaries in Lithuanian have not been studied experimentally. There have, admittedly, been attempts to discuss them (LKG 1965: 122–126; Mikalauskaitė 1975: 71–75; Pakerys 1995: 281–288), but the specific nature of those works made it necessary to do so. More extensive research has been conducted only on Standard Lithuanian (Strimaitienė, Girdenis 1978: 61–68) and Northern Samogitian (Kliukienė 1983: 50–60; 1992: 38–51) 120 A study of the quantity of vowels in internal open and closed syllables in the PVA dialect. For this reason, in their view (DLKG 1994: 36–37; Pakerys 1995: 282–283), distinguishing open and closed syllables in Standard Lithuanian is not expedient. 5.1. In summary, the following conclusions can be drawn. 1. In the PVA dialect, vowels in closed endings are longer than vowels in open endings. Perhaps the final consonant of a closed ending protects the syllable vowel from shortening. On the other hand, the vowel-length effect in open endings may also arise from a smoother transition from the vowel to the pause. 2. The duration of high vowels depends more on position. 3. The fact that vowels are shortened in PVA dialect endings is once again confirmed. In almost all cases, vowels in endings (both open and even closed) are shorter than the corresponding internal vowels. 4. Syllable closure has no substantial effect on the duration of internal vowels that are not part of diphthongs. In open syllables (V-TV), vowels are only slightly longer than in closed syllables (VT-TV). 5. Internal vowels before a fricative and a stop (V-STV) are longer than those before a single stop (V-TV). Since vowel duration differs very little in unequivocally closed (VT-TV) and open (V-TV) syllables, the slightly greater duration of the internal vowel before a fricative and a stop in this case can be attributed only to the effect of the fricative. 6. The duration of internal vowels before a fricative and a stop (V-STV) and before a single fricative (V-SV) differs little. It can therefore be argued that the consonant cluster ST does not close the syllable15. Received 1998 05 18 LITERATŪRA DLKG – Dabartinės lietuvių kalbos gramatika, Vilnius: Mokslo ir enciklopedijų leidykla, 1994. G i r d e n i s A. 1985: Fonologinio skiemens riba: konstruktas ar realybė? – Kalbotyra 36 (1), 5–11. G i r d e n i s A. 1995: Teoriniai fonologijos pagrindai, Vilnius: Petro ofsetas. G i r d e n i s A. ir kt. 1995: Lietuvių kalbos ir jos tarmių prozodinių reiškinių ir fonemų alofonų analizė. Mokslinė ataskaita, Vilnius: VU. K a z l a u s k i e n ė A. 1998: Balsių kiekybės santykiai pietinių vakarų aukštaičių tarmėje. – Kalbotyra 47 (1) (spausdinamas). vidinės atvirosios sandūros. Matyt, skiemens ribas galima nustatyti tik remiantis fonologine analize. Šiuo aspektu lietuvių kalbos skiemenys išsamiai aprašyti (Girdenis 1995: 113–120; dar plg. Girdenis 1985: 5–11). 15 Baigdama noriu nuoširdžiai padėkoti prof. habil dr. A. Girdeniui už vertingus patarimus ir skatinimą atlikti šį tyrimą. 121 Kliukienė R. 1983: Atviroji sandūra ir jos fonetinė realizacija šiaurės žemaičių tarmėje. – Kalbotyra 34 (1), 50–60. LKA 2 – Lietuvių kalbos atlasas 2, Fonetika, Vilnius: Mokslas, 1982. LKG 1 – Lietuvių kalbos gramatika 1, Fonetika ir morfologija, Vilnius: Mintis, 1965. LKT – Lietuvių kalbos tarmės. Chrestomatija, Vilnius: Mintis, 1970. Mikalauskaitė E. 1975: Lietuvių kalbos fonetikos darbai, Vilnius: Mokslas. Pakerys A. 1982: Lietuvių bendrinės kalbos prozodija, Vilnius: Mokslas. Pakerys A. 1995: Lietuvių bendrinės kalbos fonetika, Vilnius: Žara. Strimaitienė M., Girdenis A. 1978: Priebalsių junginių trukmė kaip atvirosios sandūros indikatorius bendrinėje lietuvių kalboje. – Kalbotyra 29 (1), 61–68. Vaitkevičiūtė V. 1960: Lietuvių kalbos balsių ir dvibalsių ilgumas arba kiekybė. – Lietuvių kalbotyros klausimai 3, 207–217. Salys A. 1946: Lietuvių kalbos tarmės. Tübingenas. Zinkevičius Z. 1966: Lietuvių dialektologija, Vilnius: Mintis. THE LENGTH OF THE OPEN AND CLOSED VOWELS IN THE SOUTH-WESTERN AUKŠTAITIAN DIALECT Summary The results of the experiment show, that the vowels of the closed word-endings are longer than those of the open word-endings in the south-western Aukštaitian dialect. Both closed and open vowels in the word-endings, however, are shorter than the vowels in the word-middle. The final consonants in the closed word-endings of the dialect investigated might be preserving the stem vowel from the latter being shortened. On the other hand, the effect of the length of vowels in the open word-endings might happen due to a more smooth vowelpause shift. The influence of the closed syllable on the length of word-middle vowels (those not constituting diphthongs) is the least. The word-middle vowels are longer in the position preceding both blown and explosive consonants (V-STV) than in the position preceding one explosive consonant alone (V-TV). However, in this case, the greater length of the word-middle vowel followed by both blown and explosive consonants is explained by the influence of the blown consonant only, because the cases with vowels followed by both blown and explosive consonants or blown consonant alone hasn’t shown big difference in vowel lengths and accordingly this is statistically insignificant. So, in the south-western Aukštaitian dialect the consonant ST group doesn’t close the stem. 122