ACTA LINGUISTICA LITHUANICA

XLVIII (2003), 49–57

Notes on Lithuanian Speaking Rate

Asta Kazlauskienė, Kristina Veličkaitė

Vytauto Didžiojo universitetas, Kaunas

The article presents an investigation into the overall rate of Lithuanian speech, based on the speech of eight individuals (four men and four women, belonging to two different age groups) living in Kaunas. The average age of the younger group is 27, that of the older group is 55,5. Each respondent was asked to talk on a freely chosen topic for about 20 minutes. In each case, the investigated amount of text did not exceed 36 000 syllables. The average number of syllables per second is chosen as a speech indicator. The average rate of the respondents’ free narration was 3,8±0,6 syll./sec., the confidence interval was 3,2÷4,3 syll./sec. The average speech rate of women was 4±0,4 syll./sec. (3,4÷4,5 syll./sec.), that of men 3,6±0,9 syll./sec. (2,2÷4,9 syll./sec.). Internal differences in speech rate among women and men proved statistically insignificant: the calculated value of Student’s criterion is lower than the critical value: t(6)=0,910<t(0,95)=1,943; the confidence levels overlap (2,2÷4,9 syll./sec. for men, 3,4÷4,5 syll./sec. for women). The average speech rate of the younger age group was 4,2±0,4 syll./sec. (3,5÷4,9 syll./sec.), that of the older group 3,3±0,5 syll./sec. (2,5÷4,2 syll./sec.). Such a difference is statistically significant: t(6)=2,567>t(0,95)=1,943.

Speaking rate is not a straightforward matter. From the perspective of linguists, it is the speed of articulation of words, syllables, or sounds. Psychologists are interested in what rate reveals about a speaker’s personality and how a listener perceives speech rate. In rhetoric, speaking rate is an important means of intonation. Speaking speed serves many functions. It can be constitutional, situational, and individual (Ceplytis 1974: 136). Constitutional rate is characteristic of a language or its forms, such as a dialect. Situational rate depends on the circumstances of speaking: the number of listeners and the acoustic properties of the room. Individual rate is determined by a person’s personal traits, such as temperament. Intonational rate, which has a semantic function in speech, is also distinguished. Intonational rate can be divided into intellectual, emotional, and representational types. Intellectual rate refers to a pace at which speech is deliberately slowed down or sped up to show the importance of the text being spoken or, conversely, its insignificance. Emotional rate expresses the speaker’s emotions; in this case, speaking rate may be difficult to control. Representational rate conveys the speed of an action described by the speaker or imitates another person’s speaking rate (Ceplytis 1974: 136–139). Lithuanian standard speech also serves all these functions, but so far they have neither been described nor studied.

This paper attempts to analyze the constitutional rate of Lithuanian speech, while also addressing situational and individual rates; it seeks to determine whether rate may depend on sex, age, and education.

There are no studies that examine Lithuanian speaking rate in greater detail. The only one worth mentioning is a study of the speaking rate of Northern Samogitian, conducted in the early 1970s at the Phonetics Laboratory of Vilnius University (Girdenis, Lakienė 1975:

50 | ASTA KAZLAUSKIENĖ, KRISTINA VELIČKAITĖ

71–74). Rhetoric books (Koženiauskienė 1999; Bielienė 2000) provide information on how fast Lithuanians speak, but these are the authors’ personal observations rather than data established through experiments.

This paper examines the speaking rate of eight people currently living in Kaunas. Five of them have lived in Kaunas all their lives, while the other three moved here from Telšiai (A. G. m./j. ¹, 8 years ago), Prienai District (A. S. v./s., 20 years ago), and Lazdijai (A. P. m./s., 31 years ago). In selecting informants, attention was paid to their age and sex: the speech of four men and four women, and of four younger and four older people, was recorded. The average age of the younger group was 27, and that of the older group was 55.5. Four informants had higher education, two had post-secondary education, and two had secondary education. They all speak colloquial Lithuanian, which is not as correct as the standard written Lithuanian required in public life. The participants did not know the purpose of the experiment; they were only told that their speech was being studied.

Each informant was asked to speak for a certain period of time (about 20 minutes) on a topic of their choice. The main requirement was to tell a story rather than read. All informants met this requirement. Since they had to speak for a specific length of time, rather than for as long as they could naturally continue without stopping, they were allowed to take a break while speaking. Not all informants made use of this option. Breaks were not included in the total speaking time.

It should be noted that the informants’ speech was not entirely spontaneous, since they could think in advance about what they would say. However, this is unlikely to have affected their speaking rate, because extralinguistic parameters of spoken language (in this case, pauses and rate) are usually not consciously controlled (Ignatova 1999: 64).

The texts were recorded on a tape recorder. The recordings were transcribed (typed in WinWord format), and longer pauses were marked. Twenty minutes of speech from each informant was examined. The total volume of text examined for all informants was 2 hours and 40 minutes. This amounts to more than 36,000 syllables. We believe that this volume of text should be sufficient to identify the main trends in overall speaking rate². In addition, when the same person speaks for a long time (especially if unaccustomed to it and aware of being recorded), they may become tired, which could change their natural speaking speed.

The speaking-rate index chosen was the average number of syllables pronounced per second (syll./sec.). This choice was motivated by the fact that syllable fluctuation (i.e., random deviation from the average) is among the smallest in comparison with the sentence, word, sound, morpheme, or other unit (Ceplitys 1974: 127).

A computer program developed for this study counted all the vowels and diphthongs used by each informant³. Fillers were not counted as separate syllables:

¹ The first letter written after the informants’ initials indicates sex, and the second indicates generation (younger, older).

² We have considerably more recordings (dialects, dialogues, examples of various functional styles), but coding them takes a great deal of time because we aim to transcribe them as accurately as possible so they can also be used in our other studies. Therefore, the data from this study can be supplemented in the future, providing not only the main characteristics of speaking rate but also their relationship to various aspects of speech.

³ The program was written by T. Kazlauskas. In a text saved in a particular WinWord format, it finds vowels occurring after a palatalization marker, diphthongs, and plain unopposed vowels (their total is given at the end of the text).

PASTABOS DĖL LIETUVIŲ KALBĖJIMO TEMPO

51

aaa, eėė, mmm and others. However, syllables that do not form words were also counted, i.e., syllables of unfinished words were counted if the syllable had been fully pronounced, for example: su... sutvarkyt, viena prasčiau... sparčiausiai and so on. Pauses between words were retained, since overall speaking rate is determined by pauses, hesitations, and lengthenings as well as by the speed of articulation of sounds, syllables, or words. The resulting number was divided by the duration of the recording.

The data were subsequently processed using mathematical statistical methods. This was done using the STDEV, CONFIDENCE, and TDIST functions in the MS Excel program. The statistical parameters conventionally used in such studies were calculated: the arithmetic mean, standard deviation, confidence interval; the difference in tempo was assessed using Student’s t-test.

The data obtained show that the mean speaking rate of our informants is 3.8±0.6 syllables/sec., with a confidence interval of 3.2÷4.3 syllables/sec.

According to Girdenis and Lakienė, the speaking rate of Northern Samogitians is 3.4±0.6 syllables/sec. (confidence interval 3.2÷3.6 syllables/sec.). If we compare these data with the mean rate of standard Lithuanian that we established, we see that Girdenis and Lakienė’s supposition that Northern Samogitians speak more slowly than people who speak standard Lithuanian is confirmed. However, the age of the informants should be taken into account. Almost all the informants studied by Girdenis and Lakienė were over 60, whereas the mean age of the informants in this study was 41; thus, the mean ages of the participants in the two experiments differ by nearly twenty years. Therefore, it would be more accurate to compare the speaking rate of Northern Samogitians with that of older people. The older participants in our experiment pronounced an average of 3.3±0.5 syllables/sec. (2.5 ÷ 4.2 syllables/sec.). In this case, the speaking rates of people speaking either dialectally or in standard language are similar.

Ceplyčio tyrimo duomenimis (1974: 129), latviai per sekundę ištaria vidutiniškai 4,3 skiemenis. Galėtume manyti, kad latviai kalba greičiau. Tačiau reikia atkreipti dėmesį į tai, jog Ceplytis latvių kalbėjimo greitį tyrė kitaip: trys diktoriai pasakė vieną ir tą patį tekstą. Šiame tyrime kalba buvo spontaniška, todėl tempas priklausė ne tik nuo gebėjimo greitai ar lėtai artikuliuoti žodžius, bet ir nuo mąstymo galimybių, nes reikėjo patiems sugalvoti, ką pasakyti. Kai žmogus kalba spontaniškai, jis nereguliuoja savo kalbėjimo tempo, kalba palyginti natūraliai, jam patogiausiu greičiu. Kas kita, kai žmogui reikia pasakyti konkrečią frazę. Diktorius gali stengtis ją pasakyti aiškiai ir dėl to gali šiek tiek sutrikti natūralus kalbėjimo tempas. Žmogus, paprastai kalbantis greitai, gali sulėtinti tempą, o lėtakalbis – pagreitinti savo kalbėjimo tempą. Taigi spontaniškos ir nespontaniškos kalbos tempas gali skirtis.

To determine whether Lithuanians really do speak more slowly than Latvians, we additionally conducted a reading experiment. Five speakers read the same two texts (excerpts from a work of fiction and a popular science article). The preliminary trial results show that Lithuanians read an average of 4.6 syllables per second. Thus, our reading rate is similar to that of Latvians. Interestingly, reading rate did not depend on the nature of the text being read: the speakers read both texts at a similar rate. During the experiment, we noticed one more thing. We asked the speakers to read the same text very slowly and very quickly. When reading slowly, the speakers pronounced an average of 3.3 syllables/sec. (the speaking time increased by an average of 20 secon-

52 | ASTA KAZLAUSKIENĖ, KRISTINA VELIČKAITĖ

ds); when reading quickly, they pronounced 5.7 syllables/sec. (the speaking time decreased by an average of only 12 seconds).

It should be noted that the speech rate perceived by a listener can differ greatly from that established experimentally, since perception of tempo depends on the phonetic and syntactic structure of the text, as well as on the semantic meaning of what is said (Ceplytis 1974: 135). For example, the tempo of the phrases “I speak quickly” and “I speak slowly” will be perceived differently, even though they are uttered at the same rate. The listener will be influenced by the meanings of the words lėtai and greitai. One can see that actual speaking rate differs from perceived rate by listening to the informants’ recordings. Lithuanian-language students who had not participated in the experiment, when asked to rank all the informants by speaking rate, found it fairly easy to identify the slowest and fastest speakers, but found it difficult to rank the other speakers by tempo, especially when the difference in rate was less than 0.5 syllables/sec.

Some researchers measure speaking rate not in syllables per second, but in words per minute. Koženiauskienė (1999: 357) states that a Lithuanian speaker utters 120–140 words per minute, while lectures are delivered somewhat more slowly—about 90–100 words per minute. The mean speaking rate of the informants in our study is 107 words per minute. It is unclear whose speech Koženiauskienė studied. Perhaps orators who were reading rather than speaking extemporaneously.

When speaking rate was determined using different tempo indices (syllables per second or words per minute), it was observed that the informants’ rankings did not coincide (see table). Two slow speakers (V. V. v./s. and A. S. v./s.) and four faster speakers (J. L. m./s., I. L. m./j., A. G. m./j., and A. Ž. v./j.) remained in the same positions, but speakers A. P. m./s. and A. B. v./j. switched places. This is related to the length of the words they used; for example, A. P. m./s. used many monosyllabic words, which in her speech were often stressed. This shows that measuring speaking rate in words per minute is not very accurate. It would hardly be accurate (and, most importantly, difficult to compare objectively) even if we counted phonological words (i.e., attached monosyllabic function words to polysyllabic words), since word length varies greatly in speech.

TABLE 1. LITHUANIAN SPEAKING RATE

Informants

Measured in words per minute

Measured in syllables per second

V. V. v./s.

68,5

2,6

A. S. v./s.

78,9

3,3

A. B. v./j.

99,0

3,6

A. P. syllables/sec.

110,3

3,5

J. L. syllables/sec.

114,7

3,8

PASTABOS DĖL LIETUVIŲ KALBĖJIMO TEMPO

53

CONTINUATION OF TABLE 1

Informants

Measured in words per minute

Measured in syllables per second

I. L. m./f. | 118.5 | 4.2

A. G. m./f. | 127.5 | 4.3

A. Ž. f./f. | 139.6 | 4.7

Average | 107.1 | 3.8

Another possible speech rate index is how many sounds a speaker pronounces per second. Since the average Lithuanian syllable consists of 2.4 phonemes (Karosienė, Girdenis 1994: 34), we can conclude that our informants pronounced an average of 9 phonemes per second.

We attempted to determine how the rate changes during speech, so each informant’s text was divided into four segments. Each segment lasted five minutes. The number of syllables in each segment was counted, and the average speech rate for that segment was determined (see Fig. 1).

Figure 1 shows that speech rate varies. At first glance, no obvious trend in the change of rate can be discerned. It may be that variation in speech rate is quite individual. This idea is also supported by the similarity of the overall averages for each segment: 3.7; 3.8; 3.7; 3.8 syllables/sec. It should also be borne in mind that speech rate may depend on the content of the narrative: we tell some things faster and others more slowly. For this reason, such substantial changes in the rates of particular speakers may occur during a single narrative.

FIG. 1. SPEECH RATE IN DIFFERENT TIME SEGMENTS

54 | ASTA KAZLAUSKIENĖ, KRISTINA VELIČKAITĖ

Another interesting finding: four of the eight informants spoke at a rate closest to their average in the second segment, three in the third, and only one at the end of the narrative. Thus, a person will begin speaking at their most natural rate not immediately, but after at least five minutes have elapsed from the beginning of the monologue.

There is a stereotype that women talk more than men. This assumption naturally leads to the conclusion that women also speak faster than men. These beliefs are entirely unfounded. Ignatova (1999: 64), who studied the relationship between parameters of monologic spoken language and individual psychological characteristics, observed that differences in speaking-rate capacity are associated with individual gender differences. The present study likewise shows that men’s and women’s speaking rates differ.

According to our data, the average speaking rate for women is 4±0.4 syllables/sec. (3.4÷4.5 syllables/sec.), and for men it is 3.6±0.9 syllables/sec. (2.2÷4.9 syllables/sec.). The average speaking rates of men and women differ by almost 0.5 syllables/sec. Nevertheless, the fastest speaker in this experiment was a man, not a woman. Informant A. Ž. v./j.’s speaking rate was 4.8 syllables/sec., while that of the fastest-speaking woman, A. G. m./j., was 4.3 syllables/sec. If we compare the average speaking rates of men and women, we see that rate variation was less pronounced in the women’s group (3.6; 3.8; 4.2; 4.3 syllables/sec.) than in the men’s group (2.6; 3.3; 3.6; 4.8 syllables/sec.).

Although men’s and women’s speaking rates differ slightly, our data do not allow us to state categorically that women speak faster than men, since statistical analysis shows that the difference between women’s and men’s speaking rates is not statistically significant: the calculated Student’s t-value is lower than the critical value: t(6)=0,910<t(0,95)=1,943; the confidence intervals overlap (men 2.2÷4.9 syllables/sec.; women 3.4÷4.5 syllables/sec.).

According to Girdenis and Lakienė, informants’ age has the greatest effect on variation in speaking rate (1975: 72). Similar trends are also observed in this study. The average speaking rate in the younger age group is 4.2±0.4 syllables/sec. (3.5÷4.9 syllables/sec.), while in the older age group it is 3.3±0.5 syllables/sec. (2.5÷4.2 syllables/sec.). The rates of these groups differ by almost one syllable (see Fig. 2). This is twice the difference between men’s and women’s speaking rates. Thus, age has a greater effect on speaking rate than the informant’s gender. Almost all older informants spoke more slowly than the younger speakers.

The results of the statistical analysis of speaking rates across different age groups show that the difference in rate is statistically significant: the calculated Student’s t-value is greater than the critical value for this test, t(6)=2,567>t(0,95)=1,943.

It is difficult to say whether a person’s speaking rate differs greatly in youth and old age. Nevertheless, the effects of aging may slow speech, as memory and orientation weaken, and older people have greater difficulty recalling names or less frequently used words (Neverauskas 1999: 10). For example, while recounting events from thirty years earlier, informant V. V. v./s. took longer pauses when trying to remember place names. Informant J. L. m./s. spoke about events from the same period and recalled names and other details with ease. Informant J. L. m./s. spoke 1.2 syllables/sec. faster than informant V. V. v./s. Although age is very important for speaking rate, a person’s individual characteristics also have a major influence.

Pastabos dėl lietuvių kalbėjimo tempo

55

FIG. 2. SPEAKING RATE OF THE YOUNGER AND OLDER INFORMANT GROUPS

There appears to be no evidence that speaking rate may depend on a person’s education. The number of words a person knows depends on their level of erudition. A moderately educated person can make do with 2,000–3,000 words. Writers’ vocabularies may reach 8,000–10,000 words (Bielienė 2000: 150). Thus, an indirect relationship may exist between education and speaking rate. Theoretically, a more highly educated person should pause less than a person of the same gender, age, and temperament who is less educated, because their vocabulary is larger and finding the right words is easier for them.

Although the four informants with higher education spoke faster (4.1±0.6 syllables/sec.; 3.2÷5.1 syllables/sec.) than those with secondary or post-secondary education (3.4±0.5 syllables/sec.; 2.6÷4.3 syllables/sec.), and t(6)=1,804<t(0,95)=1,943, we cannot fully rely on these results, since the three fastest-speaking informants are from the younger generation (see Fig. 3), and the difference in speaking rate may have been caused by age rather than education.

FIG. 3. RELATIONSHIP BETWEEN SPEAKING RATE AND EDUCATION

56 | Asta Kazlauskienė, Kristina Veličkaitė

Education, apparently, cannot be a decisive factor in speech rate simply because speech rate does not reflect the quality of language or thought in any way. Education is related to language quality. Moreover, speech depends greatly on what kind of education a person has acquired—humanities, technical, or some other kind.

Each person’s speech rate is individual. There is no doubt that it is related to a person’s psychophysiological characteristics, for example, their temperament, the speed of neurophysiological processes, and so on (Ceplytis 1974: 137). When the speed of neurophysiological processes changes for some reason (due to aging, depression, or trauma), speech rate changes as well. A person’s spontaneous speech rate is innate. Changes in tempo, the nature of pauses, intonation, and other aspects of spoken language are often related to the speaker’s state (Ignatova 1999: 52). Perhaps this is why some informants spoke at a different rate than expected. Their speaking rate may have been affected by the fact that their speech was being recorded—which caused some of them to experience stress, accelerating the thought process in some and slowing it down in others.

Some believe that certain moods and emotions affect a person’s speech rate. For example, anger, boredom, and joy speed it up, while sadness, solemnity, and surprise slow it down (Ceplytis 1974: 138). The informants’ moods remained constant while they were speaking. For some, the recording caused tension, so their emotions were suppressed. Joy and laughter most often slowed the overall rate, because pauses arose at those moments, while the joke itself was delivered faster than a neutrally spoken phrase. It can therefore be surmised that a person’s mood or emotions affect speech rate only when logical thinking is suppressed. This, however, is a subject for further research.

Having analyzed the results of the preliminary study of Lithuanian speech rate, the following main conclusions can be drawn.

1. Speech rate depends most on age: younger people speak faster than older people.

2. Speech rate depends little on the speaker’s sex or education. It can be tentatively suggested that the range of speech rates is greater among men than among women.

3. Speech rate apparently depends most on a person’s personal characteristics, which are the most stable and change the most slowly. A person’s character traits determine what is known as their overall speech rate, according to which people are classified as fast or slow speakers.

Speech rate is a complex, multifaceted linguistic phenomenon. Its parameters are determined simultaneously by many linguistic and non-linguistic factors (temperament, age, the speech situation, etc.), which should be investigated in greater depth.

LITERATŪRA

Bielienė, J. 2000: Iškalbos menas, Vilnius: Dailės akademijos leidykla.

Ceplytis, L. 1974: Анализ речевой интонации, Рига: Зинатне.

Ignatova, N. 1999: Monologinės sakytinės kalbos parametrų ir individualių psichologinių ypatumų ryšiai. Psichologija 20, 51–66.

Girdenis, A., Lakienė, A. 1975: Šiaurės žemaičių kalbėjimo tempas. Kalbotyra 27 (1), 71–74.

PASTABOS DĖL LIETUVIŲ KALBĖJIMO TEMPO

57

Karosienė, V., Girdenis, A. 1994: Lietuvių bendrinės kalbos skiemens tipų dažnumai. Kalbotyra 43 (1), 34–42.

Koženiauskienė, R. 1999: Retorika, Vilnius: Mokslo ir enciklopedijų leidybos institutas.

Neverauskas, J. 1999: Senėjimas. Medicina ir dar kai kas visiems 11, 10.

Asta Kazlauskienė, Kristina Veličkaitė

Vytauto Didžiojo universitetas

S. Daukanto g. 28, 3000 Kaunas, Lietuva

[email protected]

Gauta 2003 03 31