Age-related changes in the structures of accent prominence in the English language

Размещено на http://www.allbest.ru/

Moscow State Linguistic University

Age-related changes in the structures of accent prominence in the English language

Shevchenko T.I., Sokoreva T.V.

Abstract

The purpose of this study is to look into the age-related changes of the fundamental pitch value, the intensity and duration of syllables, and the degree of variability of these parameters for accentuation of stressed syllables in the speech of participants in telephone conversations. This paper focuses on the search for prosodic means of preserving accent models necessary for verbal communication rather than on the identification of a physical decrease. The analysis of 102 samples of American speech, examples of telephone conversations normalized by age and gender, showed that the maximum development of the person's voice characteristics in the middle age, while the elderly retain the prosodic contrast between accented and non-accented syllables due to compensation strategies.

Keywords: telephone conversations, accent prominence, prosody, age changes, American English.

Аннотация

Целью настоящего исследования является изучение возрастных изменений показателей частоты основного тона, интенсивности и длительности слогов, а также степени вариативности данных параметров для акцентного выделения ударных слогов в речи участников телефонных переговоров. В данной работе внимание направлено не на выявление физического спада, а на поиск просодических средств сохранения акцентных моделей, необходимых для речевой коммуникации. Анализ 102-х образцов американской речи, взятых из телефонных переговоров и нормированных по возрасту и гендеру, показал, что максимальное развитие голосовых характеристик человека происходит в среднем возрасте, в то время как пожилые люди сохраняют просодический контраст между акцентируемыми и неакцентируемыми слогами за счет стратегий компенсации.

Ключевые слова: телефонные переговоры, акцентная выделенность, просодия, возрастные изменения, американский вариант английского языка.

The paper is concerned with the issue of age-related changes in accent-based prominence patterns relevant for spoken communication. In the present study we accept the notion of accent as a property of utterances in spoken communication which may or may not enhance lexical stress which is a property of words [3], [6]. In the flow of speech accented syllables (AS) are contrasted to unaccented syllables (UAS).

Most of the researchers agree that English accentual prominence means are fundamental frequency (F0), intensity (Int) and time (duration). However, those phonetic means are given varying ranking in different studies [1], [4], [7], [9]. In the current study we find it imperative to explore the effects of all the prosodic parameters in accentual prominence of syllables in English spoken communication.

As far as age-related changes in prosody are concerned, most of the research works concentrate on physical decline, deterioration and perceptual losses. Accentual patterns are also expected to lose in pitch and intensity, while temporal components of speech production and perception tend to demonstrate consistent slowing down [5], [10], [12]. The problem can be formulated as follows: How are the losses in pitch, intensity and tempo compensated for? What are the speech habits of people who have more social practice in telephone conversations?

The facts of equivocal results on certain points of prosodic data [2], [10] call for methods which can register not only maximal and minimal values, as well as their range but also the extent to which they vary. A useful tool called pairwise variability index (PVI) is among the recent metrics in rhythm which reveals the degree of variance between the durations of adjacent syllables in different languages [11]. The PVI method can be likewise applied to different age groups data, since our aim is to find how distinctive AS and UAS in duration and other prosodic features are. The degree of variance estimated by means of PVI can be further correlated both with the speakers' ages annotated as their dates of birth (DOB) and with their group means. The PVI method can be normalized for tempo, which is another advantage of that method.

The main purpose of the research is to look for specific age-related changes in the ways English accent-based prominence is provided for.

Our objective, therefore, is to track the adult life-long trajectory of pitch, intensity and duration of syllables, as well as their variance employed for accent-based prominence patterns in speech communication. We hypothesize that there are ways of keeping up the accentual prominence in English which mature speakers resort to, in spite of obvious physical and perceptual losses. We assume that people might benefit from their constant linguistic practice in the opening phrases of telephone conversations.

Material and methods

The audio material for the research was taken from “Switchboard II” [8], a collection of recordings of spontaneous telephone conversations between speakers of American English. The Switchboard Corpus is annotated for participants' dates of birth and their educational background. For the present analysis we selected participants with `some graduate' annotation who represent 102 American English speakers balanced for gender (51 in each category) and equally distributed in three age groups (34 in each group): young (20-39), middle (40-59), old (60-69).

The following measurements were taken in Praat: F0_max, F0_min, F0_range; Int_max, Int_min, Int_range; Accented/Unaccented syllable duration, Foot duration.

Variance of prosodic parameters was tested with PVI normalized for tempo [11], to find the difference in successive syllables duration and in successive feet duration, as well as for F0_max and Int_max values:

where m - the number of syllables/feet in an utterance;

dk - duration/F0/intensity of the k_th syllable/foot.

For statistical analysis and apparent time comparison of the three groups' data the mean values of prosodic parameters and PVI scores were computed using two-way ANOVA, the factors being age (three levels: young, middle-aged, old) and gender (two levels: male and female); the Spearman correlation coefficient was used to test the relationship between the obtained prosodic characteristics and the speakers' age (date of birth - DOB).

Results

ANOVA data

Foot duration in women's speech increases at the expense of accented syllable time. As one-way ANOVA showed there is a significant effect of Age factor on women's feet duration (DF=2; F=3.095; p=.054) which tends to rise with increasing age.

F0_max values in Accented and Unaccented syllables drop. The two-way ANOVA was conducted with F0_max values in AS as the dependent variable and Age and Gender as two factors. The obtained results reveal a significant effect of Age (DF=2, F=3.04, p=.052) and Gender (DF=1, F=193.54, p<.000). ANOVA of F0_max in UAS as the dependent variable showed the significant effect of both Age (DF=2, F=6.54, p<.05) and Gender (DF=1, F=206.10, p<.000). Thus we find decrease in F0_max values both in AS and UAS, thus we observe that the contrast `AS - UAS values' is sustained.

Accented/Unaccented syllable contrast in Int_max increases significantly with age. One-way ANOVA was performed with the ratio values of maximum intensity as in AS compared to ones in UAS as the dependent variable and Age as factor. The results reveal a significant effect of Age (DF=2, F=4.498, p<.05) (Fig. 1).

Fig. 1 - Intensity contrast in Americans' speech across three age groups

Correlation with DOB

Foot duration increases at the expense of accented syllable time. According to correlation analysis there is strong negative correlation between the speakers' DOB and their mean foot values (rho=-.303, p<.01).

Accented/Unaccented syllable contrast in F0_max is significantly increasing. The correlation analysis reveals a significant negative correlation between the speakers' age and F0_max values representing the ratio of F0_max in AS to that in UAS (rho=-.214, p<.05).

F0_min values in Accented and Unaccented syllables drop. As correlation analysis points there exists a significant positive correlation between the speakers' DOB and F0_min values in AS (rho=.237, p<.05) and F0_min values in UAS (rho=.246, p<.05).

Accented/Unaccented syllable contrast in Int_min increases significantly with age. The ratio values of minimum intensity as in AS compared to those in UAS were correlated with the speakers' ages as a result a significant negative correlation was found (rho=-.214, p<.05) (Fig. 1).

Accented syllable duration increases with age. Figure 2 shows the mean values of AS and UAS durations in the speech of male and female informants across three age groups (n=102). The ANOVA results reveal significant change in mean AS duration increasing with age (DF=2, F=7.12, p<.01) and dependent on the speaker's gender (DF=1, F=4.36, p<.05). The analysis of mean UAS duration reveals only significant effect of Gender (DF=1, F=4.24, p<.05) (Fig. 2).

The correlation analysis discloses a significant negative correlation between the mean AS duration and speakers' dates of birth, with high values of AS duration associated with lower values of their DOB (rho=-.319, p<.01).

Fig. 2 - AS and UAS durations in Americans' speech across three age groups (ms)

Pitch range widens. ANOVA results for F0_range values in AS as the dependent variable show the significant effect of Age solely (DF=2, F=3.92, p<.05). The ANOVA conducted with the F0_range values of the UAS as the dependent variable reveals the significant effect of both Age (DF=2, F=3.86, p<.05) and Gender (DF=1, F=4.06, p<.05) (Fig. 3).

The relationship between the speakers' DOB and each of the F0 parameters reveals itself in significant negative correlation between DOB and F0_range both in AS and UAS (rho=-.335, p<.01 and rho=-.295, p<.01 correspondingly). As a result we observe a significant increase in F0_range values with age.

Fig. 3 - F0_range in Americans' speech across three age groups

F0_max variance grows. The PVI of F0 and intensity values are supposed to possess more variance in more accent-based speech. This tendency is observed in F0_max PVI scores that increase with age and it is supported by the two-way ANOVA results showing the significant effect of Age (DF=2, F=7.80, p<.01) on the PVI values of F0_max in adjacent syllables. The correlation analysis also reveals the significant negative correlation between F0_max PVI values and DOB (rho=-.286, p<.01) (Fig. 4).

Fig. 4 - F0_max PVI in Americans' speech across three age groups

Discussion

The present study explored the ways speakers of American English changed their ways of bringing the most informative accented syllable in running speech to the fore. The speakers benefited from their conventional linguistic practice of telephone talks. They demonstrated their changing ways of achieving accentual prominence through the prosodic parameters of fundamental frequency, duration, intensity, including their maximal, minimal values and variance.

Comparison of a number of prosodic variables across the three age groups of 102 American English speakers gives evidence to the positive effect of communication experience that improves and optimizes linguistic habits against the background of physical decline.

Thus age-related communication gains in an accent-based language are as follows:

1. Accentual prominence patterns vital for word recognition and speech intelligence that depend on the contrast between the accented and the unaccented syllables is either sustained due to AS duration or increased due to pitch (F0_max) and intensity (Int_max) levels.

2. Although the maxima in pitch values tend to drop with age, their variance increases. Most of the age-related changes in the adult life-long perspective occur between the young age and the middle age, while in the older age variability level is kept up at a relatively comparable value. Middle-aged speakers' voices are normally favorably attested for their flexibility, expressivity and power. We can, therefore, assume that variance alone does not account for the older voices' instability; the overall effect of instability might be the result of accompanying features of voice quality characteristics, like shimmer and tremor, for example.

3. Our basic finding consists in the fact that despite the registered lowering of absolute pitch values in Hertz, the pitch range widens in semitones due to the use of the lower register. Here again it is the middle age that exhibits the most impressive increase of pitch range values which involves both AS and UAS. Older speakers manage to sustain a wide pitch range in AS only, which results in bringing the AS/UAS contrast to the fore.

Compensation strategies of older American English speakers are aimed, one can suggest, at keeping up the most important accentual prosodic patterns which provide for speech audibility efficiency, segmentation, word recognition and intelligibility.

Apparent time' methodology of comparing prosodic characteristics across at least three age groups means, when coupled with correlation analysis involving the speakers' dates of birth, appears to be productive in revealing different aspects of adult life-long trajectory of age-related prosody development and the role of prosody in shaping accent-based prominence patterns for spoken telephone communication.

accent syllable age telephone communication

References

1. Галочкина И. Е., Шевченко Т. И. Мелодика и ритм как составляющие социокультурной компетенции // Вестник Московского государственного лингвистического университета. Серия: Гуманитарные науки. - 2010. - № 580. - С. 42-54.

2. Шевченко Т. И. Актуальные проблемы социофонетического исследования просодии // Вестник Московского государственного лингвистического университета. Серия: Гуманитарные науки. - 2011. - № 607. - С. 273-282.

3. Шевченко Т. И., Садовникова Н. А., Сибилева Л. Н. Ритм и смысл просодии дискурса: когнитивный подход и статистика // Вестник Московского государственного лингвистического университета. Серия: Гуманитарные науки. - 2012. - № 1(634). - С. 175-187.

4. Шевченко Т. И. Тенденции, проверенные временем: гласные, согласные и просодия английского языка // Вестник Московского государственного лингвистического университета. Серия: Гуманитарные науки. - 2007. - № 523. - С.182-191.

5. Beck J. M. Organic variation of the vocal apparatus / W. J. Hardcastle, J. Laver, F. E. Gibbon // The Handbook of Phonetic Sciences, Second edition. - Wiley-Blackwell, 2010. - P. 155-201.

6. Cutler A. Lexical stress in English pronunciation / M. Reed, J. M. Levis // The Handbook of English Pronunciation . - Wiley-Blackwell, 2015. - Р. 106-124.

7. Fry D. B. Duration and intensity as physical correlates of linguistic stress / D. B. Fry // Journal of the Acoustical Society of America. - 1955. - Vol. 27. - Р. 765-769.

8. Godfrey J., Holliman E., McDaniel J. SWITCHBOARD: Telephone speech corpus for research and development. ICASSP'92, 1992.

9. Lieberman P. Some acoustic correlates of word stress in American English / P. Lieberman // Journal of the Acoustical Society of America. - 1960. Vol. 35. - P. 344-353.

10. Linville S. E. Vocal Aging / S. E. Linville. - San Diego: Singular Publishing Group, 2001. - 320 p.

11. Low E., Grabe E., Nolan F. Quantitative characterization of speech rhythm: syllable-timing in Singapore English / E. Low, E. Grabe, F. Nolan // Language and Speech. - 2000. Vol. 43. - P. 377-401.

12. Reubold U., Harrington J., Kleber F. Vocal aging effects on F0 and the first formant: a longitudinal analysis in adult speakers / U. Reubold, J. Harrington, F. Kleber // Speech Communication. - 2010. Vol. 52. - P. 638-651.

Размещено на Allbest.ru