ACTA LINGUISTICA LITHUANICA

LII (2005), 1 – 21

Acoustic Features of the Lithuanian Standard Language’s Monophthong Prosodic Accents

Robertas Kudirka

Vilniaus universitetas

Kaunas Faculty of Humanities

The purpose of the article is to describe the acoustic features of Standard Lithuanian accented monophthongs and to reveal differences between circumflex and acute tones. Initially stressed words with monophthongs of the CV / CV type, were investigated. In all, 376 words of this type were selected. The analysis of the spectrograms was based on description and investigation of the following properties of monophthongs: fundamental frequency, intensity, duration, and basic formant.

1. INTRODUCTION

The prosodic accents of standard Lithuanian have already been studied instrumentally to a considerable extent. Reading descriptions of the results of these experiments, one is struck by the substantial differences in the data on the acoustic features of Lithuanian prosodic accents and in their assessment. Researchers of standard Lithuanian distinguish duration as one of the most important features. Experimental studies conclude that syllables bearing a circumflex accent are longer (Pakerys, Plakunova, Urbelienė 1970: 43–47; Svecevičius 1964; Vaitkevičiūtė 1960: 210–216; 2001: 31). Some researchers have obtained a different result: acute diphthongs can sometimes be longer than circumflex ones (Pakerys 1968a: 106; Pakerys, Plakunova, Urbelienė 1972: 7). It has also been observed that the duration of syllable components varies depending on the prosodic accent: the first components of acute diphthongs are longer, while the second components of circumflex diphthongs are longer (Pakerys, Plakunova, Urbelienė 1972: 34). The most recent study of duration in standard Lithuanian found that stressed short vowels are approximately 2.3 times shorter than their corresponding long vowels; however, the experiment does not take prosodic accents into account (Kaukėnienė 2003: 40–47).

Researchers assess the characteristics of fundamental frequency differently. It has been established that fundamental frequency depends chiefly on phrase intonation; it is sometimes even asserted that tone is merely a means of expressing phrase intonation (Pakerys 1982: 116). On the basis of experimental data from studies of diphthongs, there is a tendency to reject the existence of rising and falling tonal types, since these studies show that both acute and circumflex prosodic accents can have either a rising or a falling tone; therefore, tone movement cannot be a distinguishing feature of prosodic accent (Laigonaitė 1958: 71–100; cf. Svecevičius 1964: 20)¹.

¹ When assessing the conclusions of experiments, one must always bear in mind that the studies were conducted using different hardware and software, which may account for differing results. The same can be said of the experimental material: some study diphthongs, others monophthongs.

2 |

ROBERTAS KUDIRKA

Similar results were obtained in a study of the fundamental tone of diphthongs: 1) words pronounced with nominative phrasal intonation have a rising and falling tone, but diphthongs with the same pitch accent may, under the influence of other phrasal intonations, acquire either a falling or a rising tone; 2) in whispering, which has no F₀ at all, pitch accents are clearly distinguished (Pakerys 1967b: 14–15; 1968b: 83–92).

Intensity is also an important cue to pitch accent, and is less related to changes in phrasal intonation than fundamental tone is. In her conclusions, Adelė Laigonaitė, who conducted an experiment, states that intensity, rather than fundamental tone, is the most important cue to pitch accent (Laigonaitė 1958: 90–100; 1959). Experiments have established that circumflex long vowels and diphthongs in Standard Lithuanian are several tenths of a decibel more intense than the corresponding acute ones (Pakerys 1982: 175). The intensity of the first component of a diphthong or mixed diphthong also differs: the first component of acute mixed diphthongs is more intense than the corresponding components of circumflex diphthongs (Pakerys 1982: 178). There are also opposing views: using analysis and synthesis, Bronius Svecevičius varied the location of the intensity peak to test its effect on the perception of pitch accent, and found that it is a significant parameter in pitch-accent perception (Svecevičius 1978: 66). Valerija Vaitkevičiūtė, who studied pitch accents in Standard Lithuanian, maintains that pitch accents differ not in their horizontal contour, but in the vertical differences in fundamental tone, intensity, and other parameters of sounds bearing different pitch accents, concentrated in one of the three equal-duration portions of the sound (Vaitkevičiūtė 1995: 89–91).

Changes in qualitative as well as prosodic cues are important for distinguishing pitch accents. These cues depend particularly on the phonemic structure of the long syllable: the relationship between pitch accent and sound quality is clearly observable in syllables consisting of diphthongs and mixed diphthongs, whereas pitch accent has less influence on the quality of long monophthongs. It has been established that the first components of acute diphthongs and mixed diphthongs in Standard Lithuanian are more open, while those of circumflex ones are more closed (Pakerys 1995: 298). It has also been observed that qualitative differences are conveyed not by specific formant values, but by changes in the frequencies of those formants: these are pronounced and contrastive in acute vowels, and less noticeable in circumflex vowels (Girdenis 1974: 194; Girdenis, Pupkis 1974: 119). Studies of the North Žemaitian dialect have shown that formant values in acute long vowels also depend on articulation: closed acute vowels are more diffuse, while open ones are more compact. This feature is associated with the concentrated energy of the acute accent and the diffuse energy of the circumflex (Girdenis 1996: 78).

In recent studies, the prevailing view is that Lithuanian pitch accents are mixed in nature: they are distinguished by certain qualitative (timbre) features, syllable quantity, modulations in voice strength (intensity) and fundamental tone, and spectral structure (Ambrazas, ed., 1994: 39; Pakerys 1982: 189; Vaitkevičiūtė 1995: 9; Stundžia 1996: 27). It should also be added that more than one researcher has mentioned the neutralization of pitch accents: Northern and Eastern Aukštaitians no longer distinguish the falling-from-the-beginning accent from the falling-at-the-end accent in syllables with long vowels and the diphthongs [ie], [uo] (including when speaking Standard Lithuanian) (Ambrazas, ed., 1994: 39; Pakerys 1967a: 132; 1995: 303; Laigonaitė 1959: 20). Lilija Anusienė (1983: 5–12) also identified signs of neutralization; she argues that the difference in duration between falling-at-the-end and falling-from-the-beginning vowels has been neutralized in Standard Lithuanian.

MONOFTONGŲ PRIEGAIDŽIŲ POŽYMIAI

| 3

A considerable number of studies of pitch accents in Lithuanian dialects have been published in linguistic journals (Atkočaitytė 2000; Bukantis 1984; Eidukaitienė 1977; Kosienė 1982; Kačiūskienė, Girdenis 1997; Girdenis 1967, 1974; Murinienė 2000; Leskauskaitė 2000). A detailed study has been made of the pitch accents of the Western Aukštaitian Kaunas subdialect (Girdenis, Pupkis 1974: 111–120). Having analyzed prosodic features (voice duration, the location of the fundamental-tone peak, and the location of the intensity peak), the authors emphasize in their conclusions that duration is the most reliable cue and best distinguishes the pitch accents. Pitch accents also differ in the location of the fundamental-tone peak and the intensity peak: the intensity peak of acute vowels occurs earlier than that of circumflex vowels. Other features, such as fundamental-tone range and peak vowel intensity, are almost independent of pitch accent. This conclusion was later confirmed by another study of vowel duration in the southern Western Aukštaitian dialect: circumflex vowels are 1.1 times longer than acute vowels (Kazlauskienė 1998: 53). An instrumental study of the prosody of the Šakiai variety of the Western Aukštaitian Kaunas subdialect asserts that the most stable cue distinguishing its pitch accents is the timing of the fundamental-tone peak: peaks in acute syllables occur much earlier than those in circumflex syllables (Bacevičiūtė 2001: 78). Other prosodic features depend on syllable structure. Long vocalic syllables also differ in duration, with circumflex syllables being longer. In long diphthongal and mixed-diphthongal syllables, the pitch accents are most clearly distinguished by the range of fundamental-tone change: the range is wider for acute syllables. The location of the tone peak and the qualitative features of the first component distinguish syllables of this type less reliably.

It was previously asserted that North Žemaitian pitch accents are distinguished by abrupt or gradual changes in the acoustic “contour” and articulation: the acute is pronounced and abrupt, while the circumflex is indistinct and gradual (Girdenis 1974: 192). Later studies showed that the distinguishing features of North Žemaitian pitch accents are associated with changes in overall energy level: the acute has concentrated energy, whereas the circumflex has energy dispersed in the center of the syllable. Experiments also showed a clear correlation between fundamental tone and energy level, leading to the conclusion that fundamental tone is a highly reliable cue to North Žemaitian pitch accents (Girdenis 1996: 82). The conclusions of earlier experiments were confirmed by recent trials involving artificial modification of fundamental tone and intensity, which found that North Žemaitian pitch accents differ in the abruptness of fundamental-tone change: the acute tone changes abruptly, while the circumflex changes gradually and smoothly (Girdenis 1998: 38). The pitch accents of the South Žemaitian Varniškiai subdialect differ most clearly in quality and dynamic features: the acute has a shorter pronunciation duration, an abrupt onset, and a fall in intensity, in contrast to the more gradual and longer circumflex (Bukantis 1984: 22). The pitch accents of the South Žemaitian Raseiniai subdialect differ significantly in duration: acute syllables are longer than circumflex syllables. Another important cue to pitch accents in the Raseiniai subdialect is the trajectory of fundamental-tone movement: the acute tone starts high and descends for almost the entire duration of pronunciation, while the circumflex is fairly level and rises slightly (Atkočaitytė 2000: 114). Studies of Eastern Aukštaitian pitch accents have established that duration and intensity are the most important distinguishing cues (Kosienė 1982: 67; Eidukaitienė 1977: 23). A comparison of diphthongal syllables in North Žemaitian and Eastern Aukštaitian showed that the pitch accents of Žemaitian are distinguished by the range of fundamental-tone change, whereas those of Eastern Aukštaitian differ in the timbre of the first component (Kačiūskienė, Girdenis 1997: 32).

4 | ROBERTAS KUDIRKA

A review of studies on Lithuanian dialectal pitch accents shows that various cues can signal the pitch-accent opposition, and that the significance of each prosodic cue varies across dialects (Laigonaitė 1958: 7–90; Kazlauskas 1966: 119–121).

To determine which acoustic cues distinguish circumflex and acute monophthongs, we conducted an experiment. The acoustic properties of stressed monophthongs were examined: fundamental frequency, intensity, duration, and formant frequencies. The study used disyllabic (in exceptional cases, trisyllabic) words of Standard Lithuanian in which the stressed monophthong occurred in the first syllable of the CV / C’V type. All possible variants of this syllable type, with all vowels and consonants, were selected from dictionaries. The words were drawn from pronunciation and explanatory dictionaries (Vaitkevičiūtė 2001; Vitkauskas 2001; DLKŽ; TŽŽ). In total, 376 words with stressed initial syllables of the CV / C’V type were selected for the experiment; the isolated words were read with a nominative intonation. The Lithuanian experimental material was recorded in 2003 in the Experimental Phonetics Laboratory of the Kaunas Faculty of Humanities, Vilnius University. Professional actors were selected as speakers:

Evaldas Leskauskas (g. 1976), gimė ir gyvena Kaune, dirba Akademiniame Kauno dramos teatre, kalba bendrine lietuvių kalba;

Dainius Svobonas (g. 1968), gimė ir mokėsi Vilniuje, 21-erių persikėlė į Kauną ir čia gyvena jau trylika metų. Dirba Akademiniame Kauno dramos teatre, kalba bendrine lietuvių kalba.

The speakers’ intonational delivery was not specified in detail; they were asked to read at a normal pace, avoiding expressiveness and using as neutral a nominative intonation as possible.

Hardware: Celeron 333 Mhz / 64 Mb RAM / 4 Mb SVGA / SB Audio PCI 64 V computer, Senheiser MD 421 microphone, Audiotechnica ART 27F Tube Mic amplifier. Recording segmentation and spectral analysis were performed in the Experimental Phonetics Laboratory of VU KHF using PRAAT 4.0.47. Statistical calculations were performed using Microsoft Excel, Statistica 5.5, and SPSS 12.0 for Windows.

For the analysis of the experimental data, more neutral pitch-accent terms were chosen: circumflex is used instead of the traditional Lithuanian term for the circumflex accent, and acute instead of the traditional term for the acute accent. In the diagrams, the abbreviations for the pitch-accent terms are: circumflex – “c”, acute – “a”.

2. D U R A T I O N

For each word, the absolute duration in milliseconds of the monophthong in the stressed first syllable of the CV / C’V type and the absolute duration of the word were measured; relative duration was calculated from these values [x = X_b: X_z (X_b – vowel duration; X_z – word duration)]. The data obtained were evaluated statistically (see Table 1).

The ratios of the absolute durations of circumflex and acute vowel allophones are shown in Table 1; the data are rounded to the nearest 1 ms. As can be seen, only one position differs statistically significantly: circumflex [ãː] is longer than acute [áː], and the confidence intervals overlap only slightly. Clearly, the small

MONOFTONGŲ PRIEGAIDŽIŲ POŽYMIAI

5

TABLE 1. ABSOLUTE DURATION OF VOWEL ALLOTONES AND ITS STATISTICAL EVALUATION²

Vowels n x̄ (ms) s(ms) v (%) 95% confidence interval (ms) tₚ > <tₐ [iː] 32 234 39,64 16,9 220 ÷ 248 [iː] 34 222 51,44 23,1 204 ÷ 240 1,02<t₀,₀₅=2,00 [ɛː] 31 242 35,09 14,5 229 ÷ 255 0,36<t₀,₀₅=2,00 [ɛː] 35 239 31,59 13,2 228 ÷ 250 [ɛː] 28 264 33,35 12,6 251 ÷ 277 [ɑː] 42 247 27,26 11,0 239 ÷ 256 [ɑː] 8 227 19,04 8,4 211 ÷ 243 2,10>t₀,₀₅=2,01 [ɔː] 38 246 36,42 14,8 234 ÷ 258 0,50<t₀,₀₅=1,99 [ɔː] 36 241 36,14 15,0 229 ÷ 254 [uː] 46 226 42,58 18,8 214 ÷ 239 [uː] 46 233 39,74 17,1 221 ÷ 245 0,75<t₀,₀₅=1,99

the sample of acute [áː]. The confidence intervals of all the other vowel allophones overlap, and Student’s t-test indicates no significant differences. However, comparing the confidence intervals and means reveals a longer duration for circumflex vowels (except for circumflex [ũː], whose mean duration is shorter). The mean durations of circumflex vowels are 3–20 ms greater than those of acute vowels; likewise, the lower bounds of the confidence intervals for circumflex vowels are always 1–28 ms higher, and the upper bounds 4–13 ms higher, than those for acute vowels. Let us examine the relative-duration data: according to the table (see Table 2), none of the positions differ statistically significantly. However, comparison of the confidence intervals shows that in four out of five positions, both the lower and upper bounds of the confidence intervals for circumflex vowels are higher than those for acute vowels. Analysis of duration data for all the vowels examined yielded no statistically significant results, but comparison of the confidence intervals revealed a general systematic pattern: the intervals for both absolute and relative duration indicate that vowels with the circumflex accent are longer than those with the acute accent. This pattern is apparent in four of the five pairs compared. The differences between the ranges of the confidence intervals are also supported by the mean durations: the absolute duration of circumflex vowels is 243 ms (relative duration – 0.35), and that of acute vowels is 232 ms (0.33).

² Meanings of the symbols used in the tables: n – number of measurements, x̄ – arithmetic mean, s – standard deviation, v – coefficient of variation; confidence interval – confidence interval, tₚ – calculated Student’s t-test value, tₐ – critical Student’s t-test value, α – significance level of the results.

6 |

ROBERTAS KUDIRKA

TABLE 2. RELATIVE DURATION OF VOWEL ALLOTONES AND ITS STATISTICAL EVALUATION

Vowels n x̄ (ms) s(ms) v (%) 95% confidence interval (ms) tₚ > < tₐ [íˑ] 32 0,33 0,07 21,0 0,31 ÷ 0,36 [íˑ] 34 0,33 0,07 20,8 0,30 ÷ 0,35 0,27 < t₀,₀₅ = 2,00 [ěˑ] 31 0,34 0,06 17,9 0,32 ÷ 0,36 [ěˑ] 35 0,35 0,06 18,2 0,33 ÷ 0,37 0,76 < t₀,₀₅ = 2,00 [ėˑ] 28 0,38 0,06 15,3 0,36 ÷ 0,40 [ãˑ] 42 0,33 0,05 14,1 0,32 ÷ 0,35 [áˑ] 8 0,30 0,03 10,3 0,28 ÷ 0,33 1,79 < t₀,₀₅ = 2,01 [õˑ] 38 0,35 0,07 19,2 0,33 ÷ 0,37 0,26 < t₀,₀₅ = 1,99 [óˑ] 36 0,35 0,07 19,7 0,32 ÷ 0,37 [ũˑ] 46 0,34 0,06 19,0 0,32 ÷ 0,36 [ũˑ] 46 0,33 0,07 21,6 0,31 ÷ 0,35 0,54 < t₀,₀₅ = 1,99

3. FUNDAMENTAL FREQUENCY

Fundamental frequency was measured at five points on the pitch contour, with the voice duration divided vertically into five equal parts; measurements were taken at these five points: at the beginning, end, and midpoint of the voice, and then at the midpoints of the segments from the beginning to the midpoint and from the midpoint to the end. Based on these data, the fundamental-frequency range was calculated as the interval between the peak and the lowest point of the pitch contour. The peak value of the fundamental frequency was also measured, along with the time of the peak in milliseconds (the distance from the beginning of phonation to the highest point). The results are presented in Table 3. Statistical testing of the range widths of the circumflex and acute vowels compared revealed no significant differences: the ranges are similar in width, with circumflex-vowel ranges more often being wider. Analysis of the confidence intervals likewise reveals no systematic differences; the intervals overlap and sometimes, one might say, even coincide.

The allophones also do not differ with respect to the second fundamental-frequency feature—the peak fundamental-frequency value: in every case, the critical Student’s t-test value was not exceeded. A review of the confidence intervals and means for the peak pitch values of the different allophones reveals no clear trends: sometimes the means and confidence-interval bounds are higher for circumflex vowels ([ãˑ] c 1.42 ht, confidence interval 0.80 ÷ 2.04 ht and [áˑ] 1.05 ht, confidence interval -0.18 ÷ 2.28 ht; [õˑ] c 1.51 ht, confidence interval 0.91 ÷ 2.11 ht and [óˑ] 1.50 ht, confidence interval 0.92 ÷ 2.08 ht; [ũˑ] c 2.59 ht, confidence interval 2.15 ÷ 3.02 ht and

MONOFTONGŲ PRIEGAIDŽIŲ POŽYMIAI

7

[űː] a 2.31 ht, confidence interval 1.84 ÷ 2.78 ht), but sometimes with the acute accent ([íː] a 2.35 ht, confidence interval 1.90 ÷ 2.80 ht and [íː] c 2.17 ht, confidence interval 1.75 ÷ 2.59 ht; [éː] a 1.96 ht, confidence interval 1.43 ÷ 2.49 ht and [ěː] c 1.62 ht, confidence interval 0.95 ÷ 2.29 ht).

The third analyzed fundamental-frequency feature is the timing of the fundamental-frequency peak. Statistical testing of the results showed no significant differences: the timing of the peak varies greatly, and the confidence intervals overlap throughout. Comparing the mean peak fundamental-frequency timings and confidence intervals of circumflex vowels with those of acute vowels revealed no systematic pattern: sometimes the means and confidence-interval bounds are higher for circumflex vowels ([ěː] 109.16 ms, confidence interval 77.15 ÷ 141.17 ms and [éː] 72.89 ms, confidence interval 46.80 ÷ 98.97 ms; [āː] 101 ms, confidence interval 69.62 ÷ 132.38 ms and [áː] 98.88 ms, confidence interval 8.26 ÷ 189.49 ms; [ōː] 130.55 ms, confidence interval 98.26 ÷ 162.84 ms and [óː] 92.31 ms, confidence interval 63.54 ÷ 121.08 ms), and sometimes for acute vowels ([íː] 90 ms, confidence interval 64.13 ÷ 115.87 ms and [ìː] 82.78 ms, confidence interval 53.48 ÷ 112.09 ms; [űː] 77.96 ms, confidence interval 56.01 ÷ 99.90 ms and [üː] 69.98 ms, confidence interval 48.01 ÷ 91.95 ms).

The diagrams (see Figs. 1, 2, 3, 4, 5) show the fundamental-frequency curves based on vowel measurements taken at five points along the curve. The fundamental-frequency curves of circumflex and acute tones can be compared. The frequency ranges of circumflex and acute vowels are similar; in some cases, it can be said, even the confidence-interval bounds coincide.

TABLE 3. FUNDAMENTAL-FREQUENCY FEATURES OF VOWEL ALLOTONES AND THEIR STATISTICAL EVALUATION

Feature under study

Vowels

n

x̄

s

v (%)

95% confidence interval

tₚ > <tₐ

Range (ht)

[iː]

32

1,54

1,28

83,1

1,08 ÷ 2,00

[íː]

34

1,47

1,12

76,4

1,08 ÷ 1,86

0.23 < t₀,₀₅ = 2.00

Peak fundamental frequency (ht)

[íː]

32

2,17

1,16

53,4

1,75 ÷ 2,59

[íː]

34

2,35

1,29

55,1

1,90 ÷ 2,80

0,59 < t₀,₀₅ = 2,00

Time of occurrence of the fundamental-frequency peak (ms)

[íː]

32

82,78

81,26

98,2

53,48 ÷ 112,09

[íː]

34

90,00

74,16

82,4

64,13 ÷ 115,87

0,38 < t₀,₀₅ = 2,00

Range (ht)

[ěː]

31

1,29

0,74

57,7

1,01 ÷ 1,56

[éː]

35

1,44

0,89

61,9

1,13 ÷ 1,74

0,74 < t₀,₀₅ = 2,00

8 | ROBERTAS KUDIRKA

TABLE 3 (CONTINUED)

Measured feature

Vowels

n

x̄

s

v (%)

95% confidence interval

tₚ > <tₐ

Peak fundamental frequency (ht)

[ėː]

31

1,62

1,82

112,3

0,95 ÷ 2,29

[ėː]

35

1,96

1,55

79,0

1,43 ÷ 2,49

0.82 < t₀,₀₅ = 2.00

Time of occurrence of the fundamental frequency peak (ms)

[ėː]

31

109,16

87,27

79,9

77,15 ÷ 141,17

[ėː]

35

72,89

75,93

104,2

46,80 ÷ 98,97

1.81 < t₀,₀₅ = 2.00

Range (ht)

[ėː]

28

1,85

1,39

75,2

1,31 ÷ 2,39

Peak fundamental tone (ht)

[ėː]

28

1,38

1,68

121,4

0,73 ÷ 2,03

Time of occurrence of the fundamental-tone peak (ms)

[ėː]

28

107,11

97,59

91,1

69,26 ÷ 144,95

Range (ht)

[āː]

42

1,58

1,07

68,1

1,24 ÷ 1,91

[āː]

8

1,10

0,91

83,2

0,33 ÷ 1,86

1,19 < t₀,₀₅ = 2,01

Peak fundamental tone (ht)

[āː]

42

1,42

1,99

140,4

0,80 ÷ 2,04

[āː]

8

1,05

1,47

139,8

-0,18 ÷ 2,28

0,50 < t₀,₀₅ = 2,01

Time of occurrence of the fundamental-tone peak (ms)

[āː]

42

101,00

100,69

99,7

69,62 ÷ 132,38

[āː]

8

98,88

108,38

109,6

8,26 ÷ 189,49

0,05 < t₀,₀₅ = 2,01

Range (ht)

[ōː]

38

1,36

0,93

68,6

1,05 ÷ 1,67

0,39 < t₀,₀₅ = 1,99

[ōː]

36

1,28

0,96

75,4

0,95 ÷ 1,60

MONOFTONGŲ PRIEGAIDŽIŲ POŽYMIAI

9

TABLE 3 (CONTINUED)

Measured feature | Vowels | n | x̄ | s | v (%) | 95% confidence interval | tₚ > < tₐ Peak fundamental frequency (ht) | [ōˑ] | 38 | 1,51 | 1,82 | 120,5 | 0,91 ÷ 2,11 | 0,02 < t₀,₀₅ = 1,99 | [óˑ] | 36 | 1,50 | 1,72 | 114,5 | 0,92 ÷ 2,08 | Time of peak fundamental frequency (ms) | [ōˑ] | 38 | 130,55 | 98,25 | 75,3 | 98,26 ÷ 162,84 | 1,79 < t₀,₀₅ = 1,99 | [óˑ] | 36 | 92,31 | 85,03 | 92,1 | 63,54 ÷ 121,08 | Range (ht) | [ūˑ] | 46 | 1,87 | 1,52 | 81,3 | 1,42 ÷ 2,32 | 0,26 < t₀,₀₅ = 1,99 | [úˑ] | 46 | 1,79 | 1,33 | 74,1 | 1,40 ÷ 2,18 | Peak fundamental frequency (ht) | [ūˑ] | 46 | 2,59 | 1,46 | 56,6 | 2,15 ÷ 3,02 | 0,87 < t₀,₀₅ = 1,99 | [úˑ] | 46 | 2,31 | 1,57 | 68,2 | 1,84 ÷ 2,78 | Time of peak fundamental frequency (ms) | [ūˑ] | 46 | 69,98 | 74,00 | 105,7 | 48,01 ÷ 91,95 | 0,52 < t₀,₀₅ = 1,99 | [úˑ] | 46 | 77,96 | 73,90 | 94,8 | 56,01 ÷ 99,90

10 |

ROBERTAS KUDIRKA

FIG. 1. FUNDAMENTAL FREQUENCY CURVES OF CIRCUMFLEX [iː] (–) AND ACUTE [íː] (–·) VOWELS

FIG. 2. FUNDAMENTAL FREQUENCY CURVES OF CIRCUMFLEX [ūː] (–) AND ACUTE [úː] (–·) VOWELS

MONOFTONGŲ PRIEGAIDŽIŲ POŽYMIAI

| 11

FIG. 3. FUNDAMENTAL-FREQUENCY CONTOURS OF CIRCUMFLEX [êː] (–) AND ACUTE [éː] (– ·) VOWELS

12 |

ROBERTAS KUDIRKA

FIG. 5. FUNDAMENTAL-FREQUENCY CONTOURS OF CIRCUMFLEX [õː] (−) AND ACUTE [óː] (−·) VOWELS

In summary, circumflex vowels do not differ statistically significantly from acute vowels in fundamental-frequency range, peak value, or the time at which the pitch peak occurs. The confidence intervals show no systematic differences.

4. INTENSITY

Intensity was measured using the same methodology as for fundamental frequency. The statistically evaluated results are presented in Table 4.

TABLE 4. INTENSITY FEATURES OF VOWEL ALLOTONES AND THEIR STATISTICAL EVALUATION

Measured feature | Vowels | n | x̄ | s | v (%) | 95% confidence interval | tᵦ < tₐ Range (dB) | [íː] | 32 | 6,47 | 2,60 | 40,2 | 5,53 ÷ 7,41 | | [íː] | 34 | 5,79 | 2,50 | 43,1 | 4,92 ÷ 6,66 | 1,07 > t₀,₀₅ = 2,00

MONOFTONGŲ PRIEGAIDŽIŲ POŽYMIAI

| 13

CONTINUATION OF TABLE 4

Measured feature | Vowels | n | x̄ | s | v (%) | 95% confidence interval | tₚ > < tα Peak intensity (dB) | [ĭ] | 32 | 69,56 | 3,21 | 4,6 | 68,40 ÷ 70,72 | 0,32 < t₀,₀₅ = 2,00 Peak intensity (dB) | [ĭ] | 34 | 69,32 | 2,94 | 4,2 | 68,30 ÷ 70,35 Time of peak intensity (ms) | [ĭ] | 32 | 100,09 | 56,75 | 56,7 | 79,63 ÷ 120,56 | 1,10 < t₀,₀₅ = 2,00 Time of peak intensity (ms) | [ĭ] | 34 | 85,79 | 48,27 | 56,3 | 68,96 ÷ 102,63 Range (dB) | [ė̌] | 31 | 5,48 | 2,39 | 43,6 | 4,61 ÷ 6,36 | 0,14 < t₀,₀₅ = 2,00 Range (dB) | [ė̌] | 35 | 5,57 | 2,59 | 46,5 | 4,68 ÷ 6,46 Peak intensity (dB) | [ė̌] | 31 | 70,68 | 3,65 | 5,2 | 69,34 ÷ 72,01 | 0,47 < t₀,₀₅ = 2,00 Peak intensity (dB) | [ė̌] | 35 | 71,09 | 3,42 | 4,8 | 69,91 ÷ 72,26 Time of peak intensity (ms) | [ė̌] | 31 | 101,61 | 48,79 | 48,0 | 83,72 ÷ 119,51 | 1,28 < t₀,₀₅ = 2,00 Time of peak intensity (ms) | [ė̌] | 35 | 86,34 | 47,73 | 55,3 | 69,95 ÷ 102,74 Range (dB) | [ė̌] | 28 | 4,61 | 1,91 | 41,5 | 3,87 ÷ 5,35 Peak intensity (dB) | [ė̌] | 28 | 70,57 | 4,10 | 5,8 | 68,98 ÷ 72,16 Time of peak intensity (ms) | [ė̌] | 28 | 76,36 | 57,77 | 75,7 | 53,96 ÷ 98,76 Range (dB) | [ã] | 42 | 5,93 | 2,48 | 41,9 | 5,15 ÷ 6,70 | 0,30 < t₀,₀₅ = 2,01 Range (dB) | [á] | 8 | 5,63 | 3,25 | 57,8 | 2,91 ÷ 8,34 Peak intensity (dB) | [ã] | 42 | 70,74 | 4,82 | 6,8 | 69,24 ÷ 72,24 | 0,01 < t₀,₀₅ = 2,01 Peak intensity (dB) | [á] | 8 | 70,75 | 4,77 | 6,7 | 66,76 ÷ 74,74

14 |

ROBERTAS KUDIRKA

TABLE 4 (CONTINUED)

Measured feature | Vowels | n | x̄ | s | v (%) | 95% confidence interval | tₚ > tₐ. Time of peak intensity (ms) | [ãˇ] | 42 | 86,31 | 56,49 | 65,4 | 68,70 ÷ 103,92 | 0,44 < t₀,₀₅ = 2,01; [áˇ] | 8 | 95,63 | 45,38 | 47,5 | 57,68 ÷ 133,57. Range (dB) | [õˇ] | 38 | 6,82 | 2,52 | 37,0 | 5,99 ÷ 7,65 | 0,59 < t₀,₀₅ = 1,99; [óˇ] | 36 | 6,47 | 2,49 | 38,5 | 5,63 ÷ 7,31. Peak intensity (dB) | [õˇ] | 38 | 73,29 | 5,54 | 7,6 | 71,47 ÷ 75,11 | 0,78 < t₀,₀₅ = 1,99; [óˇ] | 36 | 72,36 | 4,64 | 6,4 | 70,79 ÷ 73,93. Time of peak intensity (ms) | [õˇ] | 38 | 104,32 | 46,05 | 44,1 | 89,18 ÷ 119,45 | 0,25 < t₀,₀₅ = 1,99; [óˇ] | 36 | 101,56 | 49,58 | 48,8 | 84,78 ÷ 118,33. Range (dB) | [ũˇ] | 46 | 5,24 | 2,61 | 49,8 | 4,46 ÷ 6,01; [úˇ] | 46 | 5,04 | 2,26 | 44,8 | 4,37 ÷ 5,71 | 0,38 < t₀,₀₅ = 1,99. Peak intensity (dB) | [ũˇ] | 46 | 70,83 | 4,10 | 5,8 | 69,61 ÷ 72,04; [úˇ] | 46 | 70,96 | 4,43 | 6,2 | 69,64 ÷ 72,27 | 0,15 < t₀,₀₅ = 1,99. Time of peak intensity (ms) | [ũˇ] | 46 | 76,52 | 64,58 | 84,4 | 57,35 ÷ 95,70; [úˇ] | 46 | 80,35 | 71,63 | 89,1 | 59,08 ÷ 101,62 | 0,27 < t₀,₀₅ = 1,99.

Statistical evaluation of the data shows that intensity ranges do not distinguish the accents; none of the positions compared differ statistically significantly. In four of the five positions, the mean range and confidence interval values for circumflex vowels are greater than those for acute vowels, meaning that the range of circumflex vowels is wider. Calculating the mean ranges makes it clear that this difference is very small: the mean range of circumflex vowels is 5.76 dB, and that of acute vowels is 5.70 dB.

Statistical testing of peak intensity values showed that none of the positions differ significantly. The confidence intervals of all allophones overlap or even coincide, and the calculated Student’s t-test value is usually very small.

Comparison of the timing of intensity peaks yielded no statistically significant results. In three positions, the peaks of circumflex vowels [ĩˇ], [ẽˇ], [õˇ] occur later

MONOFTONGŲ PRIEGAIDŽIŲ POŽYMIAI

15

than those of the corresponding acute vowels, whereas in the remaining two positions, the peaks of acute vowels [áː], [úː] occur later than those of the corresponding circumflex vowels.

The diagrams (see Figs. 6, 7, 8, 9, 10) show the intensity curves of circumflex and acute vowels. One can compare how the intensity of circumflex and acute vowels changes. In each diagram, the first curve represents circumflex vowels and the second acute vowels.

Summarizing all the acoustic intensity features, it can be concluded that circumflex vowels do not differ statistically significantly from acute vowels in terms of intensity range, peak value, or the timing of the peak. Only the slightly wider range of circumflex vowels stands out.

FIG. 6. INTENSITY CURVES OF CIRCUMFLEX [ĩː] (–) AND ACUTE [ĩˊ] (– ·) VOWELS

16 |

ROBERTAS KUDIRKA

FIG. 7. INTENSITY CURVES OF CIRCUMFLEX [ɛː] (−) AND ACUTE [ɛː] (−·) VOWELS

MONOFTONGŲ PRIEGAIDŽIŲ POŽYMIAI

17

FIG. 9. INTENSITY CURVES OF CIRCUMFLEX [õː] (−) AND ACUTE [óː] (−·) VOWELS

FIG. 10. INTENSITY CURVES OF CIRCUMFLEX [ùː] (−) AND ACUTE [úː] (−·) VOWELS

18 |

ROBERTAS KUDIRKA

5. FORMANT STRUCTURE

The midpoint frequencies of the first and second formants of circumflex and acute vowels were measured in the spectrograms. The formant data were statistically evaluated, and the means and confidence intervals were rounded to the nearest 10 Hz. The results are presented in Table 5.

TABLE 5. FREQUENCIES OF THE FIRST AND SECOND FORMANTS (F₁ AND F₂) OF VOWEL ALLOTONES AND THEIR STATISTICAL EVALUATION

Formants | Vowels | n | x̄ (Hz) | s (Hz) | v (%) | 95% confidence interval (Hz) | tₚ < tₐ F₁ | [íː] | 32 | 230 | 15,83 | 6,8 | 230 ÷ 240 | [íː] | 34 | 240 | 13,19 | 5,6 | 230 ÷ 240 | 0,75 < t₀,₀₅ = 2,00 F₂ | [íː] | 32 | 2230 | 88,08 | 4,0 | 2200 ÷ 2260 | [íː] | 34 | 2220 | 91,08 | 4,1 | 2190 ÷ 2250 | 0,35 < t₀,₀₅ = 2,00 F₁ | [ėː] | 31 | 340 | 19,14 | 5,6 | 330 ÷ 350 | [ėː] | 35 | 350 | 16,97 | 4,9 | 340 ÷ 350 | 1,33 < t₀,₀₅ = 2,00 F₂ | [ėː] | 31 | 2090 | 112,19 | 5,4 | 2040 ÷ 2130 | [ėː] | 35 | 2090 | 98,15 | 4,7 | 2060 ÷ 2120 | 0,13 < t₀,₀₅ = 2,00 F₁ | [ėː] | 28 | 650 | 30,83 | 4,8 | 640 ÷ 660 F₂ | [ėː] | 28 | 1710 | 235,25 | 13,7 | 1620 ÷ 1800 F₁ | [áː] | 42 | 690 | 43,38 | 6,3 | 680 ÷ 700 | [áː] | 8 | 690 | 47,43 | 6,9 | 650 ÷ 730 | 0,09 < t₀,₀₅ = 2,01 F₂ | [áː] | 42 | 1220 | 128,57 | 10,5 | 1180 ÷ 1260 | [áː] | 8 | 1260 | 19,09 | 1,5 | 1240 ÷ 1270 | 0,80 < t₀,₀₅ = 2,01 F₁ | [õː] | 38 | 440 | 46,76 | 10,6 | 430 ÷ 460 | 0,43 < t₀,₀₅ = 1,99 | [õː] | 36 | 440 | 35,25 | 7,9 | 430 ÷ 460 F₂ | [õː] | 38 | 780 | 182,63 | 23,3 | 720 ÷ 840 | 1,56 < t₀,₀₅ = 1,99 | [õː] | 36 | 730 | 74,20 | 10,1 | 710 ÷ 760 F₁ | [ūː] | 46 | 300 | 29,11 | 9,8 | 290 ÷ 310 | [ūː] | 46 | 290 | 25,58 | 8,7 | 280 ÷ 300 | 0,84 < t₀,₀₅ = 1,99 F₂ | [ūː] | 46 | 660 | 90,47 | 13,7 | 630 ÷ 690 | [ūː] | 46 | 670 | 110,00 | 16,5 | 630 ÷ 700 | 0,28 < t₀,₀₅ = 1,99

MONOFTONGŲ PRIEGAIDŽIŲ POŽYMIAI

19

A comparison of the mean frequencies of the first and second formants of circumflex and acute vowels revealed no statistically significant differences: the confidence interv-

FIG. 11. SPECTRAL CHARACTERISTICS OF CIRCUMFLEX (−) AND ACUTE (−·) VOWELS

FIG. 12. FIRST (F₁) AND SECOND (F₂) FORMANTS OF CIRCUMFLEX AND ACUTE VOWELS

20 |

ROBERTAS KUDIRKA

Because the intervals overlap substantially or even coincide, the calculated Student’s t-test values are usually very small, and no position exceeds the critical value. The formant values most often differ by 10 Hz or coincide, except for the vowels [aˑ] and [oˑ], for which the second-formant values of circumflex vowels differ from those of acute vowels by 40–50 Hz. In summary, it can be concluded that the accents do not differ in their formant-structure characteristics. This can be clearly seen by examining Schemes 11 and 12.

6. CONCLUSIONS

Analysis of the experimental data on circumflex and acute vowels leads to the following conclusions:

a. In duration, circumflex vowels do not differ statistically significantly from acute vowels; however, a comparison of the confidence intervals reveals a general systematic pattern: the intervals for both absolute and relative durations of circumflex vowels indicate that vowels with this accent are longer than acute vowels. This pattern is evident in four of the five pairs compared.

b. Circumflex and acute accents do not differ statistically significantly in fundamental-tone and intensity characteristics; the confidence intervals show no systematic differences, except for the intensity-range intervals, which indicate a somewhat wider range for circumflex vowels.

c. No statistically significant differences were found when comparing the midpoint frequencies of the first and second formants of circumflex and acute vowels: the confidence intervals overlap substantially or even coincide.

ABBREVIATIONS OF SOURCES

DLKŽ – Dictionary of Contemporary Lithuanian, 4th ed., editor-in-chief S. Keinys. Vilnius: Institute of Science and Encyclopaedia Publishing, 2000.

TŽŽ – Tarptautinių žodžių žodynas, Vilnius: Alma littera, 2001.

LITERATŪRA

AMBRAZAS, V., red., 1994: Dabartinės lietuvių kalbos gramatika, Vilnius: Mokslo ir enciklopedijų leidykla.

ANUSIENĖ, L. 1983: Kirčiuotų ilgųjų balsių trukmė lietuvių bendrinės kalbos frazėse. Kalbotyra 34 (1), 5–13.

ATKOČAITYTĖ, D. 2000: Pietų žemaičių raseiniškių fonologinė sistema: prozodija ir vokalizmas. Daktaro disertacija, Vilnius: VPU.

BACEVIČIŪTĖ, R. 2001: Šakių šnektos fonologinė sistema: prozodija ir vokalizmas. Daktaro disertacija, Vilnius: VPU.

BUKANTIS, J. 1984: Фонология южножемайтского Варняйского говора. Daktaro disertacija, Vilnius: VU.

EIDUKAITIENĖ, E. 1977: Kupiškėnų monoftongų priegaidės. Kalbotyra 28 (1), 18–23.

GIRDENIS, A. 1967: Mažeikių tarmės priegaidžių fonetinės ypatybės. Kalbotyra 15, 26–27.

GIRDENIS, A. 1974: Prozodinės priegaidžių ypatybės šiaurės žemaičių tarmėje. Eksperimentinė ir praktinė fonetika 6, 160–198.

GIRDENIS, A. 1996: Energetinė šiaurės žemaičių tarmės priegaidžių fonetinės prigimties interpretacija. Baltistica 31 (1), 71–84.

GIRDENIS, A. 1998: Šiaurės žemaičių priegaidės skiriamieji požymiai. Lietuvių kalba: tyrėjai ir tyrimai. Konferencija Kazimiero Jauniaus 150-osioms gimimo ir 90-osioms mirties metinėms paminėti. Pranešimų tezės, Vilnius: VPU, LKI, 37–38.

MONOFTONGŲ PRIEGAIDŽIŲ POŽYMIAI

| 21

Girdенis, A., Pupkis, A. 1974: Pietinių vakarų aukštaičių priegaidės (prozodiniai požymiai). Eksperimentinė ir praktinė fonetika 6, 107–125.

Kačiukienė, G., Girdenis, A. 1997: Rytų aukštaičių ir šiaurės žemaičių priegaidės: bendrybės ir skirtumai. Kalbotyra 46 (1), 31–36.

Kaukėnienė, L. 2003: Lietuvių bendrinės kalbos nekirčiuotų balsių kiekybė. Lietuvių kalbotyros klausimai 48, 35–47.

Kazlauskienė, A. 1998: Pietinių vakarų aukštaičių tarmės balsių kiekybė. Daktaro disertacija, Kaunas: VDU.

Kazlauskas, J. 1966: Fonologinė kirčio raidos baltų kalbose interpretacija. Baltistica 2 (2), 119–132.

Kosienė, O. 1982: Rytų aukštaičių uteniškių monoftongų priegaidės. Kalbotyra 33 (1), 61–71.

Laigonaitė, A. 1959: Dėl lietuvių kalbos kirčio ir priegaidės supratimo. Kalbotyra 1, 71–100.

Laigonaitė, A. 1958: Literatūrinės lietuvių kalbos kirčiavimas, Vilnius: Mokslas.

Leskauskaitė, A. 2000: Seirijų šnektos mišriųjų dvigarsių priegaidės. Lietuvių kalbotyros klausimai 43, 153–164.

Murinienė, L. 2000: Akmenės šnektos fonologinė sistema: vokalizmas ir prozodija. Daktaro disertacija, Vilnius: VU.

Pakerys, A. 1967a: Apie kirtį. Kalbotyra 18, 129–134.

Pakerys, A. 1967b: Spektriniai tvirtapradžių ir tvirtagalių dvibalsių skirtumai. Spalio revoliucija ir visuomeniniai mokslai Lietuvoje: konferencijos medžiaga, Vilnius.

Pakerys, A. 1968a: Dabartinės lietuvių literatūrinės kalbos sudėtinių dvibalsių au, ai, ei akustiniai požymiai. Eksperimentinės fonetikos ir kalbos psichologijos koliokviumo medžiaga 3, 97–118.

Pakerys, A. 1968b: Lietuvių literatūrinės kalbos dvibalsių au, ai, ei akustiniai požymiai: eksperimentinis fonetinis tyrimas. Kandidato disertacija, Vilnius: VPI.

Pakerys, A. 1982: Lietuvių bendrinės kalbos prozodija, Vilnius: Mokslas.

Pakerys, A. 1995: Lietuvių bendrinės kalbos fonetika, Vilnius: Žara.

Pakerys, A., Plakunova, T., Urbelienė, J. 1970: Относительная длительность гласных литовского языка. Eksperimentinės fonetikos ir kalbos psichologijos koliokviumo medžiaga: kalbos garsai ir intonacija, 4, 30–53.

Pakerys, A., Plakunova, T., Urbelienė, J. 1972: Относительная длительность дифтонгов литовского языка. Материалы коллоквиума лаборатории экспериментальной фонетики и психологии речи 5, 3–36.

Stundžia, B. 1996: Lietuvių kalbos kirčiavimas, Vilnius: Baltos lankos.

Svecevičius, B. 1964: Простые гласные современного литовского литературного языка. Канд. диссертация, Вильнюс: ВПИ.

Svecevičius, B. 1978: Bendrinės kalbos priegaidžių intensyvumo tyrimas analizės-sintezės-analizės metodu. Kalbos garsai ir melodika, Vilnius: Mokslas, 64–67.

Vaitkevičiūtė, V. 1960: Lietuvių kalbos balsių ir dvibalsių ilgumas, arba kiekybė. Lietuvių kalbotyros klausimai 3, 207–217.

Vaitkevičiūtė, V. 1995: Bendrinės lietuvių kalbos priegaidės, Vilnius: Enciklopedija.

Vaitkevičiūtė, V. 2001: Lietuvių kalbos tarties pagrindai ir žodynas, Vilnius: Pradai.

Vitkauskas, V. 2001: Lietuvių kalbos tarties žodynas, Vilnius: Mokslo ir enciklopedijų leidybos institutas.

Robertas Kudirka Vilniaus universitetas KHF Užsienio kalbų katedra Muitinės g. 8, LT-44280, Kaunas, Lietuva [email protected]

Gauta 2005 05 12