Lietuvių bendrinės kalbos nekirčiuotų balsių kiekybė

ACTA LINGUISTICA LITHUANICA XLVIII (2003), 35–47 The Quantity of Unstressed Vowels in Standard Lithuanian Lidija Kaukėnienė Vilniaus universitetas Kauno humanitarinis fakultetas, Kaunas The article deals with the quantity of unstressed vowels in Standard Lithuanian. Attention is paid to the first and to the second post-stressed and pre-stressed vowels and their relationship to stressed vowels. 1. It has long been known and is beyond doubt that stressed long vowels are longer than the corresponding short ones. This phenomenon in Standard Lithuanian has been studied, analyzed, and demonstrated (see Vaitkevičiūtė 1960: 207–217; Svecevičius 1964: 14–32; Dambrauskaitė-Urbelienė 1967: 17–25; Pakerys, Plakunova, Urbelienė 1970: 27–28, 30). One of the first to measure and describe vowel quantity in Lithuanian was Valerija Vaitkevičiūtė (1960: 207–217). Her experiments showed that vowel duration depends on stress and tone: stressed vowels are longer than unstressed ones, while acute-accented vowels are shorter than circumflex-accented ones. Her research shows that in Standard Lithuanian long vowels are 1.5–2 times longer than short ones. Bronius Svecevičius found (1964: 17) that stressed long vowels are 2.1–2.5 times longer than short ones. Antanas Pakerys (1982: 92–103) measured and statistically assessed the duration ratios of long and short vowels. His data showed that acute-accented vowels are on average 1.7 times longer, and circumflex-accented vowels more than 2 times longer, than the corresponding stressed short vowels. According to Pakerys (1982: 94), reduction in the quantity of unstressed pre-stress stem vowels is not characteristic of Standard Lithuanian; however, he notes that more pronounced signs of quantitative reduction may occur in other positions. In general, the effect of stress position on Lithuanian vowel duration has hardly been investigated, and analyses of quantity differences between stressed and unstressed vowels have usually relied on only one unstressed position. Only Vaitkevičiūtė mentions (1960: 207) that the farther an unstressed vowel is from the stressed one, the shorter it is said to be. In Janina Urbelienė’s view (1967: 461), long post-stress vowels in Standard Lithuanian are as much as 50% shorter than the corresponding stressed vowels. Lilija Anusienė (1984: 19), who studied unstressed vowels in connected Lithuanian speech, found that short post-stress vowels are reduced by 10%, and pre-stress vowels by 20%; long vowels shorten the most: the first pre-stress vowels by 50%, the second by 40%, and the second post-stress vowels by 40%. However, none of the aforementioned studies devoted to unstressed vocalism, except Pakerys’s research, statistically assesses the experimental quantity data. 1.1. This article describes and statistically assesses the quantity of unstressed vowels in Standard Lithuanian, comparing the duration of short and corresponding long vowels in stressed and unstressed positions. 36 | LIDIJA KAUKĖNIENĖ Trisyllabic words in Standard Lithuanian were selected for the analysis¹. Mean durations of stressed vowels were calculated from the first syllable of trisyllabic words². The following words were selected for the study: fikūsas, kilimas, sinusas, tikimas, dėmonas, ēpinis, bėmolis, kàsimas, ràšimas, àtneša, pàdeda, dò noras, kò mikas, lōtosas, gùmulas, mùzika, ùbagas, džėzelis, lỹdalas, lỹrika, ēdalas, žėralas, sėlenos, tēpalas, ēžeras, lės atalas, ādāta, gātavas, bātika, dōbilas, sōminis, pōsēdis, dūminė, ūkana, rūk alas. To obtain duration values for unstressed vowels, words were selected so that the vowel under study occurred in the following positions: first post-stress after a short and a long stressed syllable³: dūsinė, muzika, kilimas, didelis, dėbetas, kasdamas, rašdamas, čiokolis, dėmonas, rōbotas, gùmulas, rēbusu, sērumu, rùžyžė, pãcypė, alinė, aludės, ramūnės, nutūkės, kokētė, vėsinės, kolitą, komodą, vaišiną, išmoka, grobiko, merginų, saliūtų, mōtina, kūdikis, lỹrika, ēžeras, tēbelis, fākelas, ādāta, sāmaną, husāras, kárbolis, ārnotas, slālomas, ēpušė, sūnėnus, mėlynas, gydyti, daktyvi, kanāpė, kanāpės, būtybė, nutōlės, šakėlė, bokāla, pirāta, nutōlstas, pāmotė, rūk omas, bizūnų, medūzų; second post-stress after a short and a long stressed syllable: sifilis, didelis, ēpini, nėšimas, kàsimas, pàdeda, rēbusu, rēbusus, sērumu, būčinį, ēterį, nūlinė, ēpinės, nūlinės, fizinė, virinės, dėmong, òpera, būdino, rēbusų, rùporų, dželės, kūdikis, mōteris, dūminės, ēpušės, lỹrines, lōpeta, sāmaną, Jākobsonas, Sācharovas⁴, ām aru, āmarus, ādini, āvili, dūminė, kāpinės, pāmotė, māšinė, āšarą, lės alą, kālbingas, tōbulo, dōbilo, lāšino, būsenų; first pre-stress before a short and a long stressed syllable: ekipā, didikas, kikilis, debesis, aferā, meletā, makakā, panamā, melasā, komodā, obōjus, polōnis, dykumà, ruletė, pakylà, masyvūs, satyrā, kalėdā, ēdikas, tęstinis, tasiūkas, žąsidė, motušė, odinis, cezūrā, lagūnā, radinys, kūrinys, giminė, etāpas, esybė, čemerys, banālu, arābas, alēja, morālas, bokālas, ovālas, urānas, pupelė, rutulys, utelė, byloti, pažymys, tylėnis, šėtonas, ūmedė, grėžioti, grėžioja, tašyti, tašynės, dorōti, kokybė, sūnėnas; second pre-stress before a short and a long stressed syllable: citatā, gitarā, idilė, debesis, melasā, ekipā, aferā, banālus, pavijā, komodā, kolegā, omarū, ūlonū, rutinā, sumanūs, dykumà, gydyklā, erenā, vėsumā, tęstinių, lęšiuką, tasiukų, žąsinė, ašelė, dovanā, obelis, liūsenā, gūduma, ūdenė, giminė, židinys, birutės, čemerys, deginys, pasakys, sakinys, pašalė, kyšūkis, gyvulys, dėmesys, vėgelė, tęsinys, grėžinys, ražulys, obuolys, kosulys, būdinys, sūkurys, gonorėja, šokoladas, honorāras. 1.2. The Lithuanian words containing the vowels under study were read by actors P. V. (b. 1949) and E. L. (b. 1976) from the Lithuanian State Kaunas Academic Drama Theatre. The words were presented to the actors in a randomized order. Each was repeated three times. The actors were instructed to read at an ordinary conversational pace and, as far as possible, with a neutral intonation, making short pauses between words. 1.3. The mean duration of the vowels under study was measured using PRAAT 4.0.11. Measu- 1 It has been demonstrated that trisyllabic words are the most frequently used in Standard Lithuanian texts (see Karosienė, Girdenis 1990: 36–48). 2 Visų žodžių tartis tikrinta remiantis V. Vitkausko (2001) „Lietuvių kalbos tarties žodynu“. 3 Vowel duration before and after circumflex- and acute-accented syllables was not calculated separately. 4 No trisyllabic words with the short vowel [ɔ] in the required position were found, so vowels for the study were taken from four-syllable words. BENDRINĖS KALBOS NEKIRČIUOTŲ BALSIŲ KIEKYBĖ | 37 The results were processed using mathematical statistical methods: the arithmetic mean duration (x̄), standard deviation (s), coefficient of variation (v), and 95% confidence interval (÷) were calculated. The significance of the results was tested using Student’s t-test. The statistical analysis was carried out using the program “STUDENT,” developed by Aleksas Girdenis (programming language TURBO-PASCAL, v. 7.0). 2. The duration results obtained are presented in tables and also shown graphically. TABLE 1. THE QUANTITY OF LONG STRESSED AND UNSTRESSED VOWELS IN STANDARD LITHUANIAN AND ITS STATISTICAL EVALUATION5 Position | n | x̄ (ms) | s (ms) | v (%) | 95% conf. interval (ms) | tₚ > t₀,₀₀₁ stressed [iː] | 12 | 193 | 22 | 11,6 | 179 ÷ 207 | 2 pre-stress [iː] | 23 | 88 | 27 | 30,3 | 74 ÷ 101 | 11,23 >3,69 stressed [iː] | 12 | 193 | 22 | 11,6 | 179 ÷ 207 | 1 pre-stress [iː] | 19 | 106 | 19 | 17,9 | 97 ÷ 115 | 11,57 > 3,66 stressed [iː] | 12 | 193 | 22 | 11,6 | 179 ÷ 207 | 1 post-stress [iː] | 21 | 85 | 20 | 24,0 | 75 ÷ 94 | 14,23 > 3,63 stressed [iː] | 12 | 193 | 22 | 11,6 | 179 ÷ 207 | 2 post-stress [iː] | 16 | 118 | 16 | 13,4 | 109 ÷ 126 | 10,5 > 3,71 stressed [eː] | 16 | 229 | 30 | 13,2 | 210 ÷ 249 | 2 pre-stress [eː] | 19 | 76 | 25 | 33,3 | 62 ÷ 90 | 14,33 > 3,73 stressed [eː] | 16 | 229 | 30 | 13,2 | 210 ÷ 249 | 1 pre-stress [eː] | 16 | 104 | 26 | 25,3 | 88 ÷ 120 | 11,04 > 3,77 stressed [eː] | 16 | 229 | 30 | 13,2 | 210 ÷ 249 | 1 post-stress [eː] | 19 | 152 | 36 | 23 | 135 ÷ 169 | 6,21 > 3,66 stressed [eː] | 16 | 229 | 30 | 13,2 | 210 ÷ 249 | 2 post-stress [eː] | 24 | 152 | 25 | 16,4 | 142 ÷ 163 | 8,14 > 3,6 stressed [eː] | 14 | 246 | 29 | 11,9 | 229 ÷ 263 | 2 pre-stress [eː] | 27 | 119 | 25 | 21,1 | 103 ÷ 135 | 11,79 > 3,75 5 Meanings of the symbols used in the tables: n – number of measurements, x̄ – arithmetic mean, s – standard deviation, v – coefficient of variation, conf. interval – confidence interval, tₚ – calculated Student’s t-test value, tₐ – critical Student’s t-test value, α – significance level of the results. 38 | LIDIJA KAUKĖNIENĖ CONTINUATION OF TABLE 1 Position n x̄ (ms) s (ms) v (%) 95% conf. interval tₚ > t₀.₀₀₁ (ms) stressed [eː] 14 246 29 11,9 229 ÷ 263 1 pre-stress [eː] 24 122 27 21,7 111 ÷ 133 13,38 > 3,58 stressed [eː] 14 246 29 11,9 229 ÷ 263 1 post-stress [eː] 24 171 47 27,6 150 ÷ 191 5,38 > 3,58 stressed [eː] 14 246 29 11,9 229 ÷ 263 2 post-stress [eː] 25 142 20 14,4 133 ÷ 150 13,06 > 3,58 stressed [aː] 12 245 35 14,3 223 ÷ 267 2 pre-stress [aː] 15 78 17 22,2 68 ÷ 87 16,27 > 3,73 stressed [aː] 12 245 35 14,3 223 ÷ 267 1 pre-stress [aː] 18 77 19 25,3 67 ÷ 86 16,98 > 3,67 stressed [aː] 12 245 35 14,3 223 ÷ 267 1 post-stress [aː] 23 132 39 29,7 115 ÷ 149 8,39 > 3,61 stressed [aː] 12 245 35 14,3 223 ÷ 267 2 post-stress [aː] 22 150 27 17,8 137 ÷ 161 8,84 > 3,63 stressed [oː] 13 203 22 10,9 189 ÷ 216 2 pre-stress [oː] 21 89 18 20,5 80 ÷ 98 15,43 > 3,67 stressed [oː] 13 203 22 10,9 189 ÷ 216 1 pre-stress [oː] 21 99 27 27,3 86 ÷ 111 11,71 > 3,62 stressed [oː] 13 203 22 10,9 189 ÷ 216 1 post-stress [oː] 19 107 36 33,6 90 ÷ 124 8,52 > 3,65 stressed [oː] 13 203 22 10,9 189 ÷ 216 2 post-stress [oː] 26 101 23 22,4 92 ÷ 110 13,27 > 3,58 stressed [uː] 15 182 47 25,6 153 ÷ 212 2 pre-stress [uː] 24 77 18 23,6 69 ÷ 86 9,09 > 3,65 stressed [uː] 15 182 47 25,6 153 ÷ 212 1 pre-stress [uː] 19 103 29 27,7 89 ÷ 117 5,90 > 3,66 BENDRINĖS KALBOS NEKIRČIUOTŲ BALSIŲ KIEKYBĖ | 39 CONTINUATION OF TABLE 1 Position | n | x̄ (ms) | s (ms) | v (%) | 95% conf. interval (ms) | tₚ > t₀,₀₀₁ stressed [u·] | 15 | 182 | 47 | 25,6 | 153 ÷ 212 1 post-stress [u·] | 18 | 133 | 23 | 17,3 | 121 ÷ 144 | 3,88 > 3,67 stressed [u·] | 15 | 182 | 47 | 25,6 | 153 ÷ 212 2 post-stress [u·] | 14 | 128 | 16 | 12,6 | 119 ÷ 137 | 4,11 > 3,75 Two graphs were plotted from the obtained quantity data (see Figs. 1 and 2); they show that, compared with stressed vowels, long unstressed vowels are shortened much more than short unstressed vowels. 2.1. Among the stressed long vowels, [e·], [a·], and [e·] stand out as having the greatest quantity (x̄ is 246 ms, 245 ms, and 229 ms, respectively). As can be seen, the mean duration of the stressed vowels [e·] and [a·] is nearly the same. The difference in the duration of syllable nuclei in stressed and unstressed syllables is statistically significant, as indicated by the confidence intervals, which are more or less distant from the confidence intervals for stressed vowels, and by the calculated Student’s t-test value, which is also considerably greater than the critical Student’s t-test value. Moreover, the significance level is the same and very high for all vowels: 0.001. FIG. 1. THE QUANTITY OF LONG VOWELS (STRESSED AND UNSTRESSED) 2.2. Our study showed that the unstressed long variants in Lithuanian undergo fairly substantial quantitative reduction. Comparing the quantity of stressed vowels with that of variants in unstressed positions revealed (see Table 2) that vowels in pre-stress syllables are reduced the most. In these positions, unstressed variants are 2–3 times shorter than the corresponding stressed ones. The second variants shorten the most 40 | LIDIJA KAUKĖNIENĖ pre-stress [aː] and [eː]. Their ratios to the stressed long vowels are approximately as follows: [aː]: [aː]⁶ = 1: 3.1; [eː]: [eː] = 1: 3. The first pre-stress [aː] is also greatly shortened; its ratio to the stressed vowel is: [aː]: [aː] = 1: 3.2. All the remaining unstressed long variants in pre-stress positions are shortened by a factor of 2–2.5. Quantitative reduction is smaller in post-stress positions. Here, vowels shorten by a factor of 1.3 to 2.2. In addition, the first pre-stress variants are shorter than the first post-stress variants, and the second pre-stress variants are shorter than the second post-stress variants. Calculating the overall average, we found that the second pre-stress variants shorten by an average factor of 2.5, and the first pre-stress variants by 2.1; the first and second post-stress variants by 1.6. Having calculated the average duration of all unstressed long vowels, we obtained the following results: in standard Lithuanian, the first pre-stress variants shorten by an average of 59%, the second pre-stress variants by 53%; the first post-stress variants by 42%, and the second post-stress variants by 41%. TABLE 2. QUANTITY RATIOS OF LONG VOWELS IN STANDARD LITHUANIAN IN STRESSED AND UNSTRESSED POSITIONS7 Positions / Vowels | 2 Pr: S | 1 Pr: S | 1 Po: S | 2 Po: S [iː]: [iː] | 1: 2,2 | 1: 1,8 | 1: 2,2 | 1: 1,6 [eː]: [eː] | 1: 3 | 1: 2,2 | 1: 1,5 | 1: 1,5 [ɛː]: [ɛː] | 1: 2 | 1: 2 | 1: 1,4 | 1: 1,7 [aː]: [aː] | 1: 3,1 | 1: 3,2 | 1: 1,8 | 1: 1,6 [oː]: [oː] | 1: 2,2 | 1: 2 | 1: 1,8 | 1: 2 [uː]: [uː] | 1: 2,3 | 1: 1,7 | 1: 1,3 | 1: 1,4 Mean values | 1: 2,5 | 1: 2,1 | 1: 1,6 | 1: 1,6 2.3. By duration, stressed vowels are ordered as follows: [eː] < [aː] < [ɛː] < [oː] < [iː] < [uː]. As already noted, the quantity of [ɛː] and [aː] does not differ (x̄ 246 ms and 245 ms, respectively). This vowel sequence shows that vowel-duration relations almost conform to the general rule: low vowels are the longest, mid vowels are shorter, and high vowels are the shortest. Experimental studies of many languages have also shown that sounds at each height are characterized by a duration that decreases as vowel height increases (see Malmberg 1963; Lehiste 1970; Mohr 1971; Liepa 1979; Ščerba 1983). As early as 1925, 6 Since we calculated the absolute duration of long vowels, that is, we did not separately distinguish vowel duration before or after syllables with a falling or rising tone, the bar above the long vowel indicates that the vowel is stressed. The notation [aː] should be read as “stressed long vowel”. 7 The symbols used in the tables are: S – stressed position; 2 Pr – second pre-stress variant, 1 Pr – first pre-stress variant, 1 Po – first post-stress variant, 2 Po – second post-stress variant. BENDRINĖS KALBOS NEKIRČIUOTŲ BALSIŲ KIEKYBĖ | 41 Richardas Ekblomas observed that the duration of sounds in the Eastern Aukštaitian dialect depends on the manner of articulation (Ekblomas 1925: 39). Discussing the work of many researchers who measured the duration of vowels in Standard Lithuanian, Pakerys, Plakunova, and Urbelienė (1970: 48) stated that low vowels, i.e. open vowels, generally have a longer physical duration than high vowels, i.e. close vowels. Thus, it can safely be stated that this prosodic regularity is universal (see Pakerys 1982: 43). Unstressed long vowels are distributed somewhat differently by duration: the first pre-accentual group – [eː] < [iː] < [eː] < [uː] < [oː] < [aː]; the second pre-accentual group – [eː] < [oː] < [iː] < [aː] < [uː] < [eː]; the first post-accentual group – [eː] < [eː] < [uː] < [aː] < [iː] < [oː]; the second post-accentual group – [eː] < [aː] < [eː] < [uː] < [iː] < [oː]. In almost all positions, [eː] retains its place—it is the longest, but the other vowels change their order. The duration of the first pre-accentual [aː] is the shortest. This probably also shows that this vowel undergoes substantial quantitative reduction. In addition, the first pre-accentual high vowels [uː], [iː] are the longest. Only the order of the second post-accentual long vowels almost corresponds to the universal relationship between vowel duration and height. Overall, the duration differences among unstressed long variants are smaller than those among stressed vowels, and they are therefore apparently more vulnerable (see Pakerys 1982: 47). TABLE 3. QUANTITY OF STRESSED AND UNSTRESSED SHORT VOWELS IN STANDARD LITHUANIAN AND ITS STATISTICAL EVALUATION Position | n | x̄ (ms) | s (ms) | v (%) | 95% confidence interval (ms) | tₚ >=<tₐ stressed [i] | 14 | 75 | 15 | 19,4 | 66 ÷ 83 | 2 pre-accentual [i] | 26 | 62 | 20 | 32,0 | 53 ÷ 70 | 2,08 > t₀,₀₅ = 2,04 stressed [i] | 14 | 75 | 15 | 19,4 | 66 ÷ 83 | 1 pre-accentual [i] | 21 | 74 | 28 | 38,0 | 62 ÷ 87 | 0,04 < t₀,₀₅ = 2,03 stressed [i] | 14 | 75 | 15 | 19,4 | 66 ÷ 83 | 1 post-accentual [i] | 19 | 70 | 19 | 26,7 | 61 ÷ 79 | 0,79 < t₀,₀₅ = 2,04 stressed [i] | 14 | 75 | 15 | 19,4 | 66 ÷ 83 | 2 post-accentual [i] | 24 | 82 | 27 | 33,1 | 70 ÷ 95 | 0,93 < t₀,₀₅ = 2,04 stressed [e] | 13 | 110 | 27 | 24,2 | 92 ÷ 128 | 2 pre-accentual [e] | 26 | 75 | 18 | 23,8 | 67 ÷ 83 | 4,47 > t₀,₀₁ = 3,65 stressed [e] | 13 | 110 | 27 | 24,2 | 92 ÷ 128 | 1 pre-accentual [e] | 23 | 96 | 21 | 21,3 | 87 ÷ 105 | 1,69 < t₀,₀₅ = 2,04 stressed [e] | 13 | 110 | 27 | 24,2 | 92 ÷ 128 | 1 post-accentual [e] | 18 | 92 | 16 | 17,1 | 84 ÷ 100 | 2,29 > t₀,₀₅ = 2,05 42 | LIDIJA KAUKĖNIENĖ TABLE 3 (CONTINUED) Position | n | x̄ (ms) | s (ms) | v (%) | 95% confidence interval (ms) | tₚ >= < tₐ stressed [e] | 13 | 110 | 27 | 24,2 | 92 ÷ 128 | 2 post-accentual [e] | 26 | 102 | 20 | 19,7 | 94 ÷ 110 | 1,05 < t₀,₀₅ = 2,03 stressed [a] | 12 | 85 | 22 | 26,3 | 70 ÷ 100 | 2 pre-accentual [a] | 26 | 68 | 15 | 22,6 | 60 ÷ 76 | 2,35 > t₀,₀₅ = 2,06 stressed [a] | 12 | 85 | 22 | 26,3 | 70 ÷ 100 | 1 pre-accentual [a] | 21 | 70 | 18 | 25,9 | 62 ÷ 78 | 2,06 > t₀,₀₅ = 2,04 stressed [a] | 12 | 85 | 22 | 26,3 | 70 ÷ 100 | 1 post-accentual [a] | 15 | 79 | 18 | 23,4 | 68 ÷ 89 | 0,79 < t₀,₀₅ = 2,06 stressed [a] | 12 | 85 | 22 | 26,3 | 70 ÷ 100 | 2 post-accentual [a] | 27 | 90 | 21 | 23,2 | 82 ÷ 99 | 0,73 < t₀,₀₅ = 2,03 stressed [ɔ] | 13 | 100 | 17 | 16,6 | 90 ÷ 110 | 2 pre-accentual [ɔ] | 16 | 82 | 22 | 26,6 | 70 ÷ 94 | 2,40 > t₀,₀₅ = 2,06 stressed [ɔ] | 13 | 100 | 17 | 16,6 | 90 ÷ 110 | 1 pre-accentual [ɔ] | 20 | 93 | 19 | 20,5 | 85 ÷ 100 | 0,45 < t₀,₀₅ = 2,04 stressed [ɔ] | 13 | 100 | 17 | 16,6 | 90 ÷ 110 | 1 post-accentual [ɔ] | 21 | 78 | 15 | 19,2 | 71 ÷ 85 | 3,9 > t₀,₀₁ = 3,62 stressed [ɔ] | 13 | 100 | 17 | 16,6 | 90 ÷ 110 | 2 post-accentual [ɔ] | 16 | 78 | 18 | 22,6 | 67 ÷ 90 | 3,12 > t₀,₀₅ = 3,10 stressed [u] | 16 | 92 | 21 | 22,4 | 79 ÷ 106 | 2 pre-accentual [u] | 20 | 76 | 18 | 23,4 | 67 ÷ 85 | 2,21 > t₀,₀₅ = 2,06 stressed [u] | 16 | 92 | 21 | 22,4 | 79 ÷ 106 | 1 pre-accentual [u] | 17 | 90 | 36 | 39,7 | 71 ÷ 108 | 0,23 < t₀,₀₅ = 2,06 stressed [u] | 16 | 92 | 21 | 22,4 | 79 ÷ 106 | 1 post-accentual [u] | 16 | 75 | 10 | 13,1 | 70 ÷ 81 | 2,57 > t₀,₀₅ = 2,08 stressed [u] | 16 | 92 | 21 | 22,4 | 79 ÷ 106 | 2 post-accentual [u] | 22 | 88 | 23 | 26,6 | 76 ÷ 99 | 0,58 < t₀,₀₅ = 2,05 BENDRINĖS KALBOS NEKIRČIUOTŲ BALSIŲ KIEKYBĖ 43 FIG. 2. QUANTITY OF SHORT VOWELS (STRESSED AND UNSTRESSED) 3. Analysis of the data shows that unstressed short vowels undergo much less quantitative reduction than long vowels; their duration differences are very small and statistically insignificant. Fig. 2 shows that the duration gap between stressed and unstressed short vowels is not as large as that between long vowels. Even in the positions where we obtained statistically significant duration differences (see Table 3), the calculated Student’s t-value is only slightly greater than the critical Student’s t-value, and perhaps, if we increased the number of tokens further, that significant difference would disappear altogether. Moreover, the significance level for short vowels is much lower: 0.05 and 0.005, with the highest level—0.001—found in only a few positions. 3.1. The calculated quantity ratios of short vowels in stressed and unstressed positions showed (see Table 4) that unstressed short variants in Lithuanian undergo very little quantitative reduction. The second pre-accentual [e] shortens somewhat more. Compared with the stressed vowel, its ratio would be approximately as follows: [e]: [ė] = 1: 1,4. In the remaining positions, unstressed short vowels shorten by a factor of 1.1 to 1.2, or their duration is the same as that of the stressed vowel. The second pre-accentual vowels show somewhat greater quantitative reduction. Calculating the overall mean duration, we found that they are shortened by a factor of 1.2 on average. The first pre-accentual and post-accentual variants shorten only slightly, by about a factor of 1.1. TABLE 4. QUANTITY RATIOS OF SHORT VOWELS IN STRESSED AND UNSTRESSED POSITIONS IN STANDARD LITHUANIAN Positions / Vowels | 2 Pre: Stressed | 1 Pre: Stressed | 1 Post: Stressed | 2 Post: Stressed [i]: [ī] | 1: 1,2 | 1: 1 | 1: 1,1 | 1: 0,91 [e]: [ė] | 1: 1,4 | 1: 1,1 | 1: 1,2 | 1: 1,1 44 | Lidija Kauķenienė TABLE 4 CONTINUED Positions / Vowels | 2 Unstressed: Stressed | 1 Unstressed: Stressed | 1 Post-stress: Stressed | 2 Post-stress: Stressed [a]: [ā] | 1: 1,2 | 1: 1,2 | 1: 1,1 | 1: 0,93 [ɔ]: [ɔ̄] | 1: 1,2 | 1: 1,1 | 1: 1,2 | 1: 1,2 [u]: [ū] | 1: 1,2 | 1: 1 | 1: 1,2 | 1: 1,1 Mean values | 1: 1,2 | 1: 1,1 | 1: 1,1 | 1: 1 It was surprising that in some unstressed positions, the short unstressed variants were slightly longer than their stressed counterparts. This was more characteristic of the second post-stress vowels. In this position, [i] and [a] were longer than their stressed counterparts. The mean duration of post-stress [u] was almost the same as that of the stressed vowel. Moreover, the overall ratio of the second post-stress vowels to the stressed vowels was 1: 1. This finding prompted a comparison of the first and second post-stress short vowels (see Table 5); there is no point in comparing the long post-stress vowels, since the data in the first table show that the long post-stress variants are somewhat longer than the second post-stress variants, or that their durations are very similar. Only the second post-stress [iˑ] and [aˑ] stand out, as they are longer than the first post-stress variants. TABLE 5. DURATION OF THE FIRST AND SECOND POST-STRESS SHORT VOWELS IN THREE-SYLLABLE WORDS AND ITS STATISTICAL EVALUATION Position | n | x̄ (ms) | s (ms) | v (%) | 95% confidence interval (ms) | tₚ > t₀,₀₅ 1st post-stress [i] | 19 | 70 | 19 | 26,7 | 61 ÷ 79 2nd post-stress [i] | 24 | 82 | 27 | 33,1 | 70 ÷ 95 | 1,62 < 2,03 1st post-stress [e] | 18 | 92 | 16 | 17,1 | 84 ÷ 100 2nd post-stress [e] | 26 | 102 | 20 | 19,7 | 94 ÷ 110 | 1,71 < 2,02 1st post-stress [a] | 15 | 79 | 18 | 23,4 | 68 ÷ 89 2nd post-stress [a] | 27 | 90 | 21 | 23,2 | 82 ÷ 99 | 1,81 < 2,03 1st post-stress [ɔ] | 21 | 78 | 15 | 19,2 | 71 ÷ 85 2 post-stress [ɔ] | 16 | 78 | 18 | 22,6 | 67 ÷ 90 | 0,02 < 2,04 1 post-stress [u] | 16 | 75 | 10 | 13,1 | 70 ÷ 82 2 post-stress [u] | 22 | 88 | 23 | 26,6 | 76 ÷ 99 | 1,71 < 2,05 BENDRINĖS KALBOS NEKIRČIUOTŲ BALSIŲ KIEKYBĖ | 45 3.2. The second post-stress vowels (see Table 5) are longer than the first in all positions; only the duration of vowel [ɔ] is the same in both positions. However, the obtained Student's criterion values are not greater than the critical values for this criterion, and the probability of a duration difference in these positions is 80%. These data allow us to conclude that there is a secondary rhythmic, or morpheme, stress in the second post-stress syllable, since the short vowel is not reduced; moreover, the second post-stress vowels are longer than both pre-stress vowels and the first post-stress short variants. The appearance of this type of stress is determined by word structure and the location of primary stress: highlighting this stress is the main function of secondary rhythmic stress. In the words examined here, it occurs in an odd-numbered, i.e. third, syllable of the word; apparently, it also affects the quantity of unstressed vowels. According to their position relative to primary stress, secondary stresses are divided into pre-stress and post-stress. In the case under consideration, the second pre-stress vowels are the shortest. On the other hand, these results suggest that in Standard Lithuanian post-stress secondary stresses are probably much more prominent and frequent than pre-stress ones (at least in three-syllable words). 3.3. Having calculated the mean duration of unstressed short vowels, we found that second pre-stress short vowels are shortened by an average of 21%, first pre-stress and second post-stress vowels by 7%, and first post-stress vowels by 14%. Short vowels in first pre-stress and second post-stress syllables are shortened equally; this is a small and insignificant reduction. The greatest quantitative reduction is experienced by second pre-stress short vowels—a tendency identical to that for long vowels, where the second pre-stress variants are also shortened the most. 3.4. In terms of duration, stressed short vowels can be listed as follows: [e] < [ɔ] < [u] < [a] < [i]. The ranking of stressed short vowels does not entirely correspond to the universal duration–height model. The mid [ɔ] and high [u] vowels are longer than the low vowel [a]. The unstressed variants are ranked by duration as follows: first pre-stress—[ɔ] < [e] < [u] < [i] < [a]; second pre-stress—[ɔ] < [u] < [e] < [a] < [i]; first post-stress—[e] < [a] < [ɔ] < [u] < [i]; second post-stress—[e] < [a] < [u] < [i] < [ɔ]. The duration rankings of post-stress vowels almost correspond to the universal duration–height model. 4. Having calculated the values, we compared the relative values of stressed long and short vowels and unstressed long and short vowels in the same positions (see Table 6). The results show that stressed short vowels are approximately 2.3 times shorter than the corresponding long vowels. In general, stressed long vowels are 2–2.9 times longer than short vowels. The long stressed [aː] is as much as 2.9 times longer than stressed short [a], while [iː] is 2.5 times longer than [i]. 4.1. In unstressed positions, the large duration difference between long and short vowels is neutralized, and in pre-stress positions it disappears altogether, becoming almost minimal: long pre-stress vowels (first and second) are on average only 1.1 times longer than the corresponding short vowels. Only the long [iː] in both pre-stress positions retain a somewhat greater duration ratio: they are 1.4 times longer than [i], and, as the statistical analysis showed, this duration difference is statistically significant. Statistical analysis of the remaining long and short pre-stress variants showed no significant duration differences. All this indicates that long pre-stress vowels in Lithuanian undergo substantial quantitative reduction, 46 LIDIJA KAUKĖNIENĖ TABLE 6. QUANTITY RATIOS OF SHORT AND LONG VOWELS IN STANDARD LITHUANIAN IN STRESSED AND UNSTRESSED POSITIONS Positions / Vowels | S: S | 2 Pre-stress: 2 Pre-stress | 1 Pre-stress: 1 Pre-stress | 1 Post-stress: 1 Post-stress | 2 Post-stress: 2 Post-stress [i]: [iː] | 1: 2,5 | 1: 1,4 | 1: 1,4 | 1: 1,2 | 1: 1,4 [e]: [eː] | 1: 2,1 | 1: 1,1 | 1: 1,1 | 1: 1,6 | 1: 1,5 [a]: [aː] | 1: 2,9 | 1: 1,1 | 1: 1,1 | 1: 1,7 | 1: 1,6 [ɔ]: [oː] | 1: 2 | 1: 1,1 | 1: 1,1 | 1: 1,4 | 1: 1,3 [u]: [uː] | 1: 2 | 1: 1 | 1: 1,1 | 1: 1,8 | 1: 1,5 Mean values | 1: 2,3 | 1: 1,1 | 1: 1,1 | 1: 1,5 | 1: 1,5 some of their durations differ little from those of the corresponding short vowels.8 Post-stress long variants retain a greater quantitative contrast: on average, they are 1.5 times longer than the corresponding short variants. Overall, post-stress long vowels are 1.2 to 1.8 times longer than short vowels. In all post-stress positions, the duration ranges of the long variants are separate from those of the short variants, and the values obtained using Student’s criterion are greater than the critical values. Thus, long vowels in Lithuanian occur only in stressed syllables, while half-long vowels occur in unstressed syllables. Short vowels can occur in both stressed and unstressed syllables. Short vowels in stressed position contrast with long vowels, while those in unstressed position contrast with half-long vowels. Half-long vowels are allophones of long vowels, since they are in complementary distribution: long vowels occur in stressed positions, while half-long vowels occur in unstressed positions. 5. The results of the experiment can be summarized as follows: 1. Unstressed vowels in standard Lithuanian undoubtedly undergo quantitative reduction. Long variants are especially reduced: the second pre-stress variants are shortened by an average of 2.5 times, the first pre-stress variants by 2.1 times, and the first and second post-stress variants by 1.6 times. 2. Among all unstressed long vowels, the first and second pre-stress variants are shortened more than the corresponding post-stress variants. In addition, the second pre-stress variants are shorter than the first. 3. Unstressed short vowels in standard Lithuanian are shortened very little. 8Of course, the experiment we conducted may not have been very methodologically rigorous, since vowel quantity was measured in words with different phonological structures. As we know, vowel duration depends on various factors: syllable structure and position in the word, the quality of surrounding consonants, stress position, and so on (for the full range of factors and their influence on vowel quantity in Lithuanian, see Vaitkevičiūtė 1960: 207–208). If these factors are not taken into account when measuring absolute duration, the data obtained can sometimes be highly contradictory. On the other hand, it is hardly possible to find entirely ideal minimal pairs when examining the duration of vowels in different positions. Moreover, each person’s speech rate and other individual characteristics vary considerably. I believe that this comparison of the length of stressed and unstressed vowels at least reveals the main tendencies in quantitative reduction in trisyllabic words in standard Lithuanian. BENDRINĖS KALBOS NEKIRČIUOTŲ BALSIŲ KIEKYBĖ | 47 The second pre-stress variants undergo a somewhat greater quantitative reduction, shortening by an average of 1.2 times; the first post-stress and pre-stress variants shorten by 1.1 times, while the second post-stress variants show no tendency to shorten at all. Nor did most of the unstressed short-vowel positions we compared show statistically significant differences in duration. Thus, stress has much less influence on the duration of short vowels than on that of long vowels. In addition, the data obtained support the conclusion that the second post-stress syllable bears a secondary rhythmic, or morpheme, stress, since the short vowel there is almost unreduced. 4. It was observed that the short second pre-stress variants are shorter than the first pre-stress variants, whereas the second post-stress variants are longer than the first. 5. Comparing the mean durations of stressed long and short vowels, we found that stressed short vowels are approximately 2.3 times shorter than their corresponding long vowels. In unstressed positions, the substantial difference in duration between long and short variants is neutralized: long pre-stress variants are, on average, only 1.1 times longer than short ones, whereas long post-stress variants retain a greater quantitative contrast, being on average 1.5 times longer than their corresponding short variants. LITERATŪRA ANUSIENĖ [АНУСЕНЕ], L. 1984: Количественное соотношение слогоносителей как фактор ритмической структуры слов в литовском литературном языке. Автореферат канд. диссертации, Вильнюс: ВГУ. DAMbrAUSKAITĖ-URBELIENĖ [УРБЕЛЕНЕ], J. 1967: Долгие гласные современного литовского литературного языка. Автореф. канд. диссертации, Вильнюс: ВГУ. EKBLOM, R. 1925: Quantität und Intonation im zentralen Hochlitauischen. Uppsala: Lundequistiska Bokhandeln. KAROSIENĖ, V., GIRDENIS, A. 1990: Bendrinės lietuvių kalbos žodžio ir skiemens statistinė struktūra. Kalbotyra 41 (1), 36–48. LEHISTE, I. 1970: Suprasegmentals, Cambridge (Mass.) – London: M. I. T. Press. LIEPA, E. 1979: Vokālizma un zilbju kvantitāte latviešu literārajā valodā, Riga: Zinātne. MALMBERG, B. 1963: Phonetics, New York: Dover Publications. MOHR, B. 1971: Intrinsic variations in the speech signal. Phonetica 23 (2), 65–93. PAKERYS, A. 1982: Lietuvių bendrinės kalbos prozodija, Vilnius: Mokslas. PAKERYS, A., PLAKUNOVA, T., URBELIENĖ, J. 1970: Относительная длительность гласных литовского языка. Kalbos garsai ir intonacija 4, 30–53. SVECEVIČIUS, B. 1964: Простые гласные (монофтонги) современного литовского литературного языка: экспериментально-фонетическое исследование. Автореф. канд. диссертации, Вильнюс: ВГУ. ŠČERBA, L. V. 1983: Русские гласные в качественном и количественном отношении, Ленинград: Наука. VAITKEVIČIŪTĖ, V. 1960: Lietuvių kalbos balsių ir dvibalsių ilgumas, arba kiekybė. Lietuvių kalbotyros klausimai 3, 207–217. VITKAUSKAS, V. 2001: Lietuvių kalbos tarties žodynas. Antrasis papildytas leidimas, Vilnius: Mokslo ir enciklopedijų leidybos institutas. Lidija Kaučėnienė Vilniaus universitetas Kauno humanitarinis fakultetas Muitinės g. 8, 3000 Kaunas, Lietuva [email protected] Gauta 2003 03 27