КАЛБŲ KONTAKTAI LIETUVOS TSR

LIETUVIŲ KALBOTYROS KLAUSIMAI XXIII (1983)

LAIMA GRUMADIENE

A SOCIOLINGUISTIC STUDY OF THE LITHUANIAN SPEECH OF NATIVE VILNIUS RESIDENTS

Introduction. Urban speech is variable, which complicates its analysis. It should be noted that variability is a constant property of language in general and is manifested in the capacity to vary linguistic resources (cf. Skvortsov L. I., 1978, p. 22), expressing the vitality of language. In the study of linguistic variation, the starting point is the idiolect (Stepanov G. V., 1979, pp. 13–14). Only after defining the limits of variation in an individual’s speech can one undertake the specific formulation of a linguistic problem such as the study of the language of a community as a whole (Grosse R., Neubert A., 1976, p. 316).

An individual’s speech may vary within the limits of the norm of a given group or collective invariant, within a particular social-communicative subsystem / A. D. Schweitzer’s term (1976, pp. 76–77)/, since it must conform to the laws of communication. However, no social-communicative systems are free of a certain degree of variation, since “the presence of variants and heterogeneous structures in speech communities already studied is, of course, a firmly established fact” (Weinreich U., Labov W., Herzog M., 1968, p. 101).

This position is explained by the operation of objective and subjective factors, both social and purely linguistic, which are activated under various circumstances (Desheriev Yu. D., 1977, pp. 127–187; Labov W., 1975a, pp. 96–181; Postovalova V. I., 1978, pp. 148–156; Serebrennikov B. A., 1974, pp. 7–63).

The concept of the sociolinguistic variable, whose rationale is presented in the works of W. Labov, brings together the linguistic and social dimensions of the material under study: sociolinguistic variables “are defined by a point of reference and a scale of values that change in a particular direction. Values within these scales are determined by social factors in much the same way as phonological environments condition the phonetic realization of allophones” (Schweitzer A. D., 1976, p. 35). The concept of the sociolinguistic variable is applied and successfully developed in many works on sociolinguistic topics.

The study of speech variation in a particular speech community using sociolinguistic methods presupposes a sociolinguistic

62

interpretation of language variation, expressed in the relative distinction between its stratificational and situational variability (Schweitzer A. D., Nikolsky L. B., 1978, pp. 76–85). The relative nature of the distinction between these types of variation should be emphasized, since they are closely interconnected. It is precisely the consideration of the mediating role of the aforementioned situational variability that makes it possible to avoid a vulgarizing interpretation of the influence of social factors on language. The distinction between these types of language variation also takes into account the relationship between individual and social factors within it (for more on this, see Makovsky M. M., 1976).

These two types of language variation correspond to two types of sociolinguistic variables: stratificational and situational (or stratificational-situational). Stratificational variables are identified in relation to the society’s class structure, whereas stratificational-situational variables change both in relation to the society’s social structure and according to the social situation. The study of the social situation presupposes the examination of two types of empirical material: the so-called data concerning the objective and subjective aspects of language. Taking account of the subjective aspect (i.e., the opinions, evaluations, judgments, and orientations of language users regarding their own speech or that of others, and their definitions of value orientations in the domain of language and language norms) prepares the ground for the productive application of sociolinguistic rules and, thereby, for identifying the main factors (both linguistic and social) that influence language evolution.

Only through a careful study of the subjective aspect of language and the correlation of data from this domain with data on the objective aspect of language does it become possible to interpret the transition of stratificational variables into stratificational-situational ones. The manifestation of the latter is connected with a person’s conscious activity in the domain of language, i.e., with value orientations and social attitudes.

Conscious activity in the domain of language is always associated with the concept of a language norm. The notion that these norms change very slowly has become commonplace. Yet as early as 1931, E. Polivanov wrote that this notion could be refined by observing the course of linguistic evolution “at close range” in individual components of the grammatical system, using a probabilistic-statistical method, of course, in cases where the facts are accessible to such observation (Polivanov E., 1931, pp. 75–76). Other linguists also write about the need to master the mechanism of language evolution when establishing norms (Kostomarov V. G., Shvarts-kopf B. S., 1965, p. 126). Only through a comprehensive study of the mechanism of linguistic change does it become possible to address the problem of norm-setting—that is, language codification—in a timely and more objective manner, taking into account a certain dependence of intra-systemic changes on the development of particular communicative functions of languages (Desheriev Yu. D., 1966, pp. 235–239).

63

Именно выявление социолингвистических норм (Никольский Л. Б., 1976, с. 129—130) предполагает учет взаимодействия в языковом развитии т.н. „внешних“ и „внутренних“ факторов, рассмотрение лингвистических фактов не только объективного, но и субъективного характера, установление их взаимосвязи. Поэтому социолингвистическое изучение разговорной речи важно не только для определения системы литовского языка жителей города Вильнюса, но и в определенной мере для дальнейшего процесса нормирования и кодификации литовского литературного языка.

An urgent problem in linguistics, as in science as a whole, is the development and understanding of procedures for substantiating scientific knowledge; in particular, the problem of the adequacy of substantiation when confirming hypotheses (Postovalova V. I., 1980, p. 89), and the selection, choice, and screening of empirical facts and the grounds for contrasting or comparing them (Bell R., 1978, p. 123—156).

When language is studied as a functional system, the presence of “weak points” in the language (a term coined by E. Coseriu, 1963, pp. 238—239), that is, a mismatch between the needs of expression and the available means of expression, signals impending changes. A statistical-probabilistic method may also be applied in the search for “weak points” in language. It is even suggested, in particular, that statistical analysis can be used to formulate and solve such problems as: 1) the problem of absolutely dating changes, 2) calculating the rates of change, 3) the problem of forecasting (Graudina L. K., 1980, p. 132).

The role of the statistical method in the study of language was noted as early as the last century, for example, in the works of I. Baudouin de Courtenay. However, the application of mathematical methods in the social sciences (which also applies to the study of language from a sociolinguistic perspective) encounters a number of difficulties arising from the complexity of social phenomena, as well as from the fact that it is not only the objective but also the subjective aspect of a phenomenon that is studied; expressing the facts of the latter in quantitative form requires the development of a special mathematical apparatus. The main difficulty is that one sometimes has to deal with short statistical series, and there is no certainty that a significant portion of what is observed has not been created by the observers’ own imagination (cf. Andreev E. P., Osipov G. V., 1973, p. 339).

The application of mathematical methods in linguistic research is assessed in different ways. This is natural, since they are also applied in different ways. Relatively few linguists express strongly negative views on the application of mathematical methods in linguistics, for example, N. S. Trubetskoy (1960, p. 18), R. A. Bulagov (1978, pp. 40—42), and others. Far more specialists hold the opposite view, according to which linguistics “cannot bind itself once and for all to predetermined research methods” (Zvegintsev V. A., 1977, p. 46).

1. Sociolinguistic characterization of the object.

The object of the present study is the social-communicative system (SCS) of Lithuanian speech in the community of native Lithuanian Vilnius residents (LRKLV) (Shvei-

64

цер А. Д., Никольский Л. Б., 1978, с. 67—76). Конкретным предметом исследования послужил один из компонентов данной СКС — литовская речь исследуемого коллектива. Так как между компонентами СКС существуют отношения функциональной дополнительности, которые задаются социальными связями между членами коллектива, сферами использования, социальными функциями и социальными ситуациями, в целом, в том или другом аспекте изучалась вся СКС речевого коллектива коренных литовцев вильнюсцев. С целью определения некоторых закономерностей и тенденций стратификационной и ситуативной вариативности ЛРКЛВ речевой коллектив был разбит на речевые общности на основе социальных и социально-демографических признаков (там же, с. 67—68).

In the present study, speech communities were distinguished on the basis of two characteristics: age and level of education. According to the research plan developed, the nature of work was also to be taken into account, since this characteristic is the fundamental criterion for the internal differentiation of social-class groups under the conditions of a developed socialist society (cf. Shkaratan O. I., Rukavishnikov V. O., 1974, p. 38). Differences by type of work (manual labor, mental labor of relatively low skill, and highly skilled mental labor), based on differences in the nature of work (Rutkevich M. N., 1972, p. 13), were initially taken into account in the present study as well, but ultimately a choice arose: either to disregard the nature-of-work characteristic or to increase the sample size substantially. The first option was chosen (cf. Zdravomyslov A. G., 1969, pp. 186—187). Three age groups are distinguished in the present study: 1) those born before 1932; 2) those born in 1932—1952; 3) those born in 1953 and later (cf. among the age scales recommended by the Institute for Social Research of the USSR Academy of Sciences is a three-level scale: 1) up to 19 years, 2) 20—39 years, 3) 40—79 years (“Ekspert-2,” 1974, p. 3)).

In the present study, two levels of education are considered relevant: 1) incomplete secondary, secondary, and specialized secondary education; 2) incomplete higher and higher education (a more fine-grained classification would also have required an increase in sample size).

The problem of sample size depends on the problem of the unit of analysis and the unit of counting (Yadov V. A., 1968, pp. 174—177). The units of analysis in this study are: 1) individual sociolinguistic variables, for example, the set of variants of unstressed ĕ, ē, ā, ū in final position, or the set of variants of unstressed ie, uo in interconsonantal position, etc.; 2) the opinions and judgments of members of the speech community under study about their own speech. The units of counting proved to be scaled values of sociolinguistic variables or judgments, the selection of alternatives, or the presence/absence of the unit under study as a whole. The sample was two-sided—linguistic units were selected and groups of people were surveyed—so the number of sample-size criteria was increased: priority was given to the criteria for the number of members of the LRKLV, i.e., the criteria

5. 949

65

sociological sample, and then the criteria of the linguistic sample were taken into account. This course of action was dictated by earlier sociolinguistic studies. In our view, the most successful study of urban speech is the work of W. Labov (Labov W., 1966). It devotes considerable attention to the selection of informants, which was the main key to its success.

In the present study, the target population meets the following requirements: participants were born in Vilnius and lived there without any substantial interruption, and both parents are Lithuanian (cf. similar requirements, e.g., in the book Russian Language According to the Findings of a Mass Survey, 1974, p. 20). The course of education and the succession of linguistic situations in Vilnius over the past 100 years has resulted in a very limited number of members of the older generation meeting the requirements stated above. There is a marked increase in the number of such members of the younger generation. Ultimately, some departures from the sampling requirements had to be made: members of the older generation could have been born outside Vilnius, provided they had settled there in childhood.

The sociolinguistic information obtained must meet the requirements of reliability, stability, and representativeness (cf. Shlyapentokh V. E., 1973, p. 17). To ensure the reliability of the information, the sample had to cover all generations of the speech community under study (in order to describe speech changes in real time), since data were needed to illustrate language shifts and changes in apparent time, i.e., across age groups in the population (Labov, 1975b, p. 203). Reliability, like stability, is ensured by the material preservation of primary sociolinguistic data (recordings of interviews on magnetic tape, questionnaire forms, spectrograms). Representativeness is characterized by the area to which the study’s findings can be generalized. Overall, the requirements of reliability, stability, and representativeness are met by drawing an unbiased, sufficiently large sample and conducting fieldwork correctly, provided that data are obtained through the use of a range of data-collection methods. To avoid bias in selecting informants, a random sample was drawn from a complete list of potential informants, represented by the 1970 All-Union Population Census questionnaires held in the archive of the Central Statistical Administration of the Lithuanian SSR. The list cannot be considered entirely complete, because indirect data concerning length of residence in Vilnius could be found in only 25% of the available questionnaires. To ensure randomness in the selection of informants, numbered folders containing questionnaires were selected according to a table of random numbers.

The greatest problems concern sample size. It depends on the variability of the characteristics of the objects in the target population and may be small if the target population is nearly homogeneous (Sociologist’s Handbook, 1976, p. 259). In our case, to obtain a sample of high-

66

representativeness, a very large sample is required, since the speech community under study is highly heterogeneous. The degree of precision of sample results also depends on sample size. On the basis of the propositions that “the sample size is almost always sufficiently large if it contains at least 30 elements” (Weinberg J., Shumaker J., 1979, p. 126), that “the baseline number of 25–30 units is the minimum for small samples” (Man and His Work, 1967, p. 61), that “highly regular types of stylistic and social stratification are established even when our individual cell contains only five speakers and we have no more than 5–10 examples of a given variable per speaker” (Labov W., 1975a, p. 115), that a sample of 1 ≤ n ≤ 50 is possible (Chesnokov S. V., 1977, p. 89), and that “a sample should be considered small only if the required precision and reliability cannot be achieved when processing it using methods based on grouping observations” (Gaskarov D. V., Shapovalov V. I., 1977, p. 19), we ultimately set the sample size for the study of LRKLV at 25 informants from each of the three age groups (N=75). After the primary sociolinguistic information had been processed with respect to the representativeness of the informants, the main corpus comprised 15 members of the older age group, 19 of the middle age group, and 26 of the younger age group, as well as 9 informants who were radio announcers and theatre actors.

Since the object of study was primarily linguistic units—pronunciation variants—the minimum volume of texts on which conclusions were to be based also had to be determined. After all, the effectiveness and validity of applying methods of mathematical statistics in sociolinguistic research depend on the stability of frequencies and the homogeneity of comparison sets. “Substantial changes in frequencies or mathematical expectations make it possible to infer a transition to texts of another type” (Yakubaitis T. A., Sklyarevich A. N., 1978, p. 60). In light of this proposition, the postulate that linguistic units occur independently is rejected (ibid., p. 55; cf. Segal D. M., 1972, p. 7). Statistical sets are considered homogeneous if the main parameters of two arbitrary samples from the target population coincide. It proved expedient to divide the collected sociolinguistic data, considered from a linguistic point of view, into three sets grouped by style (cf. the discussions in Structural and Probabilistic Analysis of Language Based on Speech Data, 1979; Studies in the Phonetic Features of Speech Styles, 1979).

The present study considers three contextual styles, understood as shifts in certain sociolinguistic variables in informants’ speech when the social context and topic change (cf. Labov W., 1975b, pp. 199–220). Following W. Labov, the author of the present study considers the degree of attention speakers pay to their speech to be the main criterion for distinguishing contextual styles. The degree of attention to one’s own speech

5*

67

depends not only on the speaker’s conscious self-monitoring, but also on a whole range of sociological, psychological, and other extralinguistic factors. It has been established that, for example, “In a state of emotional tension, significant shifts occur in the performance of those speech operations that require conscious control over the quality of their execution” (Nosenko E. L., 1979, p. 16), while the emotional state itself also depends on the social situation. In sociolinguistics, the parameters of the social situation are considered to be the relations between communicants, the setting of the speech act, the communicative orientation, and so on (cf. in: Shveitser A. D., Nikolsky L. B., 1978, p. 66). In recent years, sociolinguistics has adopted the practice of subsuming the system for selecting linguistic means according to the social situation under the term “register” (ibid., p. 75). Three registers are usually distinguished: informal, neutral, and formal; sometimes they are designated as the two extreme points on a scale of situational features—informal and formal (see in: Russian Colloquial Speech, 1973, pp. 9–11).

When studying contextual styles, role relations between communicants, their status, and the setting of the communicative act are considered relevant. In the study of LRKLV, during the main stage of empirical data collection, the role relations between interviewer and interviewee can be viewed as semi-formal relations between strangers conversing on light everyday topics. The “interviewer—interviewee” relationship is treated as standard (the same for all interviewees), although in reality, naturally, it proved to vary to some extent. The setting of the communicative act was also standardized across all interviews: all conversations took place in the interviewees’ homes (except for actors and radio announcers, whose interviews were conducted at their workplaces).

Through standardization—or, more precisely, the elimination of features of the setting, status, and roles of communicants—the differences between contextual styles were ensured by applying relatively linguistic criteria: changes in topic and text type. In the study of LRKLV, three contextual styles were distinguished: 1) informal, non-directed (conducted in the form of a personal account and conversation with the interviewer); 2) formal, non-directed (conducted while reading a standard continuous text in which a teenager recounts his adventures and experiences); 3) formal, directed (conducted while reading a list of 20 minimal pairs of words) (see Appendix 1).

2. Methods of collecting sociolinguistic information about LRKLV.

In the study of LRKLV, sociolinguistic information was collected using several methods. This made it possible to assess its quality, adequacy, and representativeness as early as the initial stage of processing.

The semi-standardized interview method was adopted as the principal method for collecting sociolinguistic information about LRKLV (cf. Monina M. L.’s account of the standardized interview, 1974, pp. 40–83).

68

Several objectives were set: 1) to establish the speech facts of the group under study in particular sociolinguistic conditions; thus, the entire body of interview material served as a document for further analysis; 2) to determine the subjects’ attitudes toward the normativity of their own speech and toward the speech of representatives of normative speech (radio and television announcers, and drama-theater actors); 3) to establish opinions on whether or not there is a Vilnius speech community with its own speech norms (this included questions such as “Can you recognize a Vilnius resident solely on the basis of features of their speech?”, “Has anyone identified you as a resident of the city of Vilnius on the basis of your speech?”, and others); 4) to investigate certain aspects of the functioning of social-communicative systems, in particular, issues of bilingualism (multilingualism) and diglossia; to determine the domains in which Russian (and, in some cases, Polish) and Lithuanian dialects are used; to ascertain the degree of proficiency in other languages and dialects; and to find out how and when they were acquired (see Appendix 2). This article presents some results of addressing the first and, in part, the second objectives; accordingly, it describes only some of the methods used to collect the material.

The interview was termed semi-standardized because, in essence, the same questions were asked of each interviewee (see Appendix 2), but they were modified somewhat for each individual, taking into account that person’s capabilities and ability to answer.

The interview lasted 20–30 minutes. With minor variations, it proceeded as follows. After obtaining the interlocutor’s consent, the interviewer confirmed their year of birth and inquired about the dialectal background of their parents. Answers to these questions were usually brief. To elicit accounts from the interviewees and hear their spontaneous, connected speech, questions were asked about their travels, trips to their fathers’ and mothers’ home villages, and vacations / cf. the question used by W. Labov (1975a, pp. 121–122) about the “mortal danger” to which the respondent had been exposed /. Here, a natural shift in contextual style toward informality and spontaneity was observed. After 5–10 minutes of narration, the interviewer asked the interviewee, in passing, whether anyone had recognized them during the trip being described as a resident of the city of Vilnius. Next, the interviewee’s own opinion about the distinctive features of Vilnius speech was elicited, and questions were asked about the respondent’s attitude toward the correctness of their own speech, the speech of announcers and actors, and the teaching of Lithuanian speech in general-education schools. They were then asked to read aloud a continuous text of 1,800 printed characters—a teenager’s account of his adventures in the courtyard (see Appendix 1, Text No. 1) (cf. Labov W., 1975a, p. 124), followed, finally, by a list composed of 20 minimal pairs of words (see Appendix 1, Text No. 2) (cf. ibid.), most of which had already appeared in the continuous text but could not be recognized by the interviewees. After this, certain issues of bilingualism (multilingualism) and diglossia were explored. As a result of the interview, the three contextual styles mentioned above were elicited.

69

The interviews were supplemented by a questionnaire survey. Each respondent was asked to complete a questionnaire (see Appendix 3), which included 55 questions on pronunciation and accentuation, constructed on the principle of alternatives.

The third method used in our study was audio testing. It was used to examine only one feature of the speech of native Vilnius residents (LRKLV): the realization of the opposition between long and short vowels in various phonetic positions. The material under study—58 minimal word pairs (see Appendix 4.1)—was divided into 6 types of homonyms according to the following criteria: 1) the origin of the vowels in opposition; 2) the presence (or absence) of stress on the vowels under study. Since experience with audio testing (see the literature review in Garšva K.’s dissertation, 1977) showed that it is most convenient to examine the realization of 4–5 minimal pairs during a single audio-testing session, the entire set of pairs under study was further divided into 13 groups, whose words were grouped not only according to the two criteria mentioned above but also by morphological features. The procedure and methodology of the audio testing were standard (see Girdenis A., 1981, pp. 33–35).

The same linguistic material, i.e., 58 minimal word pairs, was examined using the method of spectral sound analysis. At this stage, the aim was to establish the duration (t) and the frequencies of the first formants (F₁ and F₂) of the sounds under study, in order to identify physical indicators of changes in pronunciation.

The 116 sounds under study occurred in 116 sentences arranged according to a table of random numbers. A total of 464 tokens of the sounds under study were obtained, since the text was read by 4 speakers. The speakers were men (S. Zh., D. K., R. P., and A. I.) aged 22–25, who met the general requirements for informants in the present study. They were unaware of the purpose of the study. In terms of vocal register and timbre, three had baritone voices and one a tenor voice. No speech defects were detected.

The recordings were made in April 1974 in the audiovisual laboratory booth of Vilnius State University, using a “Tembr” tape recorder at a speed of 19.5 m/min and a 19A-3 condenser microphone on type 6 ferromagnetic tape (for more details on this methodology, see: Statkevičienė J., 1972). The recording level was set separately for each speaker and was not changed while the equipment was operating. Words containing realizations of the sounds under study were cut from the recorded material and edited together (the technical work was carried out by V. Lakėnė, senior laboratory assistant at VSU’s Laboratory of Experimental Phonetics). The recordings were then processed on a KPG-69 spectrograph (in the same laboratory) with a tape speed of 63.6 mm/s; time marks were made every 20 ms (50 Hz) (the technical work was carried out by engineers from the same laboratory, P. Mateikis, A. Lipeika, and S. Kucas).

3. Methods for analyzing sociolinguistic data on the speech of native Vilnius residents.

In the present study, to examine the material obtained

70

by each of the specified methods of collecting sociolinguistic data, a number of different analytical methods were applied in order to control the objectivity of the results.

Correlations in which social indicators serve as independent variables and linguistic phenomena as dependent variables are considered typical in sociolinguistics. Correlation analysis, a method of mathematical statistics, can be applied to the study of linguistic phenomena only with great caution, since it is impossible, at least at the present stage in the study of sociolinguistic problems, to take into account the effects of even all the principal factors, and each “hidden” factor may significantly affect the results.

The material collected through interviews was analyzed primarily using the methodology proposed by W. Labov in his study of New York City speech (Labov W., 1966, pp. 41–64). Each sociolinguistic variable under study at the phonetic level was coded as follows. In the transcribed interview text, each pronunciation variant of a variable was assigned a code, e.g.: (ie—1) denotes the realization [ie], (ie—2) — [ie], (ie—3) — [ei]. The relative frequency of occurrence (in%) was then calculated.

Calculations were performed separately for each informant in each of the three contextual styles. The results of informants’ responses were then calculated (summed and averaged) for each of the three age groups, taking into account two nominal levels of education. These data were presented in diagrams (see Appendix 5), whose vertical axes show the corresponding values of the mean frequencies of occurrence of the variants of the sociolinguistic variable under study, while the horizontal axes indicate age groups by level of education. Numerical values for all contextual styles were presented on the same diagram.

According to the formula

m₁/m ± 1.96 · √((m₁/m) · (1 − m₁/m) / m)

(Bitinas B., 1974, p. 54), where m is the total number of occurrences of the sociolinguistic variable under study, and m₁ is the number of occurrences of a particular variant of the variable; for a confidence probability of 0.95, the confidence interval for the frequency distribution is determined. If the values of these intervals do not overlap, it is concluded that the relative frequencies being compared differ significantly.

The question of the possible influence on informants’ speech of the dialectal origins of their parents’ speech was examined using one-way analysis of variance (Zaks L., 1976, pp. 455–461). The possible influence of each parent’s speech was examined separately. The informants’ parents’ places of origin were combined into seven broad groups: 1 — Western Aukštaitians (southern flank), 2 — Southern Aukštaitians, 3 — natives of Vilnius and Kaunas, 4 — Samogitians, 5 — Eastern Aukštaitians from

71

the environs of Utena, 6 — the northern flank of the eastern and western Aukštaitians, 7 — eastern Aukštaitians from the environs of Vilnius. Since analysis of variance requires groups comprising at least five informants, speakers of closely related dialects had to be combined into larger groups, characterized by the presence/absence of a tendency to lengthen short stressed sounds and of a vowel system with three degrees of length. The method of analysis of variance (see Zaks L., 1976, pp. 448—514; Statistical Methods..., 1979, pp. 178—194) is very labor-intensive compared with the method described above; however, it was used not only to process the field data obtained, but also to determine whether already-developed programs for computer processing of similar sociolinguistic data could be applied. In the present study, all calculations were performed manually by one person, using a desktop computer, but the results showed that this material too could be programmed and processed on more powerful computers. The consultant on the application of analysis of variance in the present study was B. Kaminskene, Senior Researcher at the Institute of Mathematics of the Academy of Sciences of the Lithuanian SSR, Candidate of Physical and Mathematical Sciences.

The analysis of variance was based on comparing the mean frequencies of each of the studied variants realizing a sociolinguistic variable, provided that the requirement of equal variances among several samples was met and that the distribution was normal. A normal distribution is achieved by random sampling. Variances were tested using Bartlett’s method (Zaks L., 1976, pp. 453—455). The measure used was the χ²-criterion. The calculations, which are not presented here for lack of space, made it possible to assert that analysis of variance could be performed on the available material on the speech of native-born residents of Vilnius.

The one-way analysis of variance, the results of which are presented below, confirmed the existence of certain significant differences between the mean frequencies of use of variable variants realized in the speech of informants, grouped according to the dialectal origin of their mother’s and father’s speech. Therefore, the method of drawing conclusions about linear contrasts proposed by Scheffé was applied (Zaks L., 1976, pp. 461—464). This method makes it possible to identify atypical groups.

Processing of the sociolinguistic questionnaire results was limited to counting the indications of variants and determining their relative frequency (in%) with regard to the informants’ age and education level. The results are presented in Appendix 6.

The method for processing the results of the commutation tests (audio tests) was as follows: using the formula _n1/n 100%, where n₁ is the number of responses indicating the presence of an opposition between the sounds under study, and n is the number of potential responses, the frequency of responses indicating an opposition was determined for each group of words under study. The significance of the results, Р, was assessed using a special-

72

ной таблице (Girdenis A., 1981, p. 212), причем за критическую черту значимости принималась отметка — 67% (см. табл. 4.2).

The spectral analysis of sounds had two aims: to measure the duration (t) and the frequency of the principal formants (F₁ and F₂) of the sounds under study.

The duration of the sounds under study was measured to the nearest 0.01 s. The absolute duration of long vowels (x₁) and vowels short by origin (yᵢ) was measured, and their difference was calculated for each realization of the vowel opposition pairs under study: Δᵢ = xᵢ − yᵢ. The mean durations of “long” (x̄) and “short” sounds (ȳ) were also calculated:

x̄ = 1/n ∑xᵢ; ȳ = 1/n ∑yᵢ, where n is the number of realizations of each sound under study.

To test for the presence of an opposition between long and short vowels in the speech of native-born residents of Vilnius, two alternative hypotheses were proposed: H₀ — no opposition exists; H₁ — an opposition exists. These hypotheses were tested using Student’s t-test (t). For this purpose, the mean differences Δ̄ᵢ between the mean durations of each sound under study (Δ̄ᵢ = x̄ᵢ − ȳᵢ) were determined, and the variances s² were calculated using the formula:

s²ₓ = ∑ (xᵢ − x̄)² / (n − 1), s²ᵧ = ∑ (yᵢ − ȳ)² / (n − 1).

The mean (standard) deviation was determined: s = √s². The significance of the difference between mean durations was assessed using the t criterion according to the formula:

t = |x̄ − ȳ| / √[(∑(xᵢ − x̄)² + (yᵢ − ȳ)²)/(nₓ + nᵧ − 2) · (nₓ + nᵧ)/(nₓ · nᵧ)] ≥ tα.

where t̂ is the Student’s t-distribution statistic, the quantity nₓ + nᵧ − 2 = f, where f is the degrees of freedom, and tα is the quantile found in the tables of Student’s t-distribution quantiles according to the calculated f (see Tikimybių teorija ir matematinės statistikos taikymas, 1971, p. 274); nₓ and nᵧ are the numbers of realizations examined for each side of the opposition (see Table 4.3).

At the second stage of processing the spectrograms, the frequencies of formants F₁ and F₂ of the sounds under study were measured at the following points: at the beginning of the sound (F₁н and F₂н), in the middle (F₁с and F₂с), and at the end (F₁к and F₂к). The filter numbers were determined from the filter grid, and thereby the sound frequencies in hertz, which were converted into units of pitch—mels (a table and rulers compiled at the VSU Laboratory of Experimental Phonetics were used); according to G. Fant’s approximation, 1 mel = 1000 · logₑ (1 + F₀/1000), where a mel is a unit of pitch and F₀ is the sound frequency in Hz. A “Mikrofot” microfilm reader, which magnifies tenfold, was used during measurement. The arithmetic mean pitches at the three control points were then determined for each speaker and each group of words under study: F̄₁н = 1/n ∑ F₁н, F̄₁с = 1/n ∑ F₁с, F̄₁к = 1/n ∑ F₁к, and so on, where n is the number of realizations

73

of each speaker, F₁н, F₁с, F₁к, F₂н, F₂с, F₂к are the absolute formant frequencies in each speaker’s realizations.

A graphical presentation of the measurement and calculation results is not provided due to lack of space in the article. Some results are indicated directly in the text. For clarity, when comparing the tenseness and laxness of vowels, the model of the so-called neutral vowel [æ] was used (F₁ = 585 mel, F₂ = 1320 mel; see: Fant G. M., Jakobson R., Halle M., 1962, p. 176). The greatest deviation Δᵢ of the first formant of each vowel under study from the first formant of the neutral vowel was calculated: Δf₁ = F₁н – F₁; the corresponding data for the second formant were: Δf₂ = F₂н – F₂. The total deviation ΣΔf was determined using the formula: ΣΔf = =Σ |Δf₁| +Σ |Δf₂|. The greater the absolute value of the total deviation, the tenser the sound was considered.

The scientific consultant on the application of experimental phonetics methods was Associate Professor A. Girdenis of VSU.

4. Some results of the study of the vocalism of LRKLV.

The essence of the problem of LRKLV vocalism lies in the issue of long and short vowels. Processes are observed in which vowel realizations change, resulting in the (complete or partial) neutralization of the opposition between long and short vowels, reduction, monophthongization of the diphonemic combinations ie, uo, or diphthongization of the long vowels o, ē.

4.1. Long and short vowels i, e, a, u. The problem with the long and short vowels mentioned in LRKLV is as follows: these vowel groups comprise 8 distinct phonemes, which are always in opposition in Lithuanian, but do not contrast in every instance in LRKLV.

The realization of the opposition between long and short vowels (hereafter V̄/V̆) in LRKLV depends to a great extent on the presence of stress: in unstressed position, long vowels tend to be pronounced short or half-long, whereas short vowels are almost never pronounced long or half-long; in stressed position, long and half-long sounds are sometimes pronounced instead of short vowels, while instances of long vowels are relatively rare. Lithuanian has two syllable accents for long vowels—the circumflex and the acute—whereas a short vowel may be stressed, but remains short. In LRKLV, a relative leveling of all stressed sounds is observed: a short vowel may acquire the syllable accent of a long vowel (or one of its transitional forms, also observed in dialects), while a long vowel may receive stress similar to that of a short vowel.

The results of the study of LRKLV show that, in the realization of V̄/V̆, the position of vowels in a word is a highly relevant factor: word-final vowels undergo changes more often, and vowels in pre-stress and immediately post-stress

74

In syllables, a long vowel in a position following a long stressed syllable changes more often than after a short stressed one.

4.1.1. V̄/V̆ in final unstressed position. In the speech of native-born Vilnius residents, three pronunciation variants of the sociolinguistic variable ī, ē, ā, ū are observed in an unstressed final position before a pause or -s, hereafter (unstressed V̄ #):

Variant I — pronunciation of a short, lax sound, respectively [i], [e], [a], [u], e.g.: [i] – [vāl’t’is]¹ „vāltys“ (nom. pl.), [aū.s’i] „aūsi“ (acc. sg.), [skaī.tan’t’is] „skaitantys“ (nom. pl.), [l’ēk’en’t’i] „lēkianti“ (acc. sg.); [e] – [ei.l’ē] „eilė“ (acc. sg.), [būv’es] „būves“ (nom. sg.), [pasūk’æ] „pasūke“ (acc. pl.); [a] – [l’ova] „l’ova“ (acc. sg.), [grā ž’æ] „grāžią“ (acc. sg.); [u] – [daī.gu] „daīgu“ (acc. sg.), [v’īru] „vīru“ (gen. pl.), [sū-r’ū] „sūriu“ (gen. pl.), [ēi.tu] „eitu“ (subjunctive, 3rd person); Variant II — pronunciation of a half-long, somewhat tense sound, respectively [i.], [e.], [a.], [u.], e.g.: [i.] – [ā k’i:s] „ākys“ (nom. sg.), [nó s’i.] „noši“ (acc. sg.), [n’æšan’t’i:s] „nēšantys“ (nom. pl.), [tūr’in’t’i:] „tūrinti“ (acc. sg.); [e.] – [l’ēl’æ.] „lėlė“ (acc. sg.), [skaī.t’æ.s] „skaitęs“ (nom. sg.), [iš’ē jæ.] „išėję“ (nom. pl.); [a.] – [raī.ka.] „raīka“ (acc. sg.), [baī.k’s’c’æ.] „baī-kščia“ (acc. sg.); [u.] – [karā’ū.] „karalių“ (acc. sg. or gen. pl.), [mālku.] „mālkų“ (gen. pl.), [kāī’ū.] „kāilių“ (gen. pl.), [s’kr’is.tu.] „skris-tų“ (subjunctive, 3rd person); Variant III — pronunciation of a long, tense sound, respectively [īˑ], [ēˑ], [āˑ], [ūˑ], e.g.: [īˑ] – [žūv’is] „žūvys“ (nom. pl.), [m’ē-d’ī.] „mēdi“ (acc. sg.), [rāšan’t’is] „rāšantys“ (nom. pl.), [b’es’ikāi.t’inan’t’i] „besikaitinanti“ (acc. sg.); [ēˑ] – [k’ē d’æ] „kėdė“ (acc. sg.), [mā t’æ.s] „mātes“ (nom. sg.), [nun’ē š’æ] „nunėše“ (nom. sg.); [āˑ] – [gālva] „gālva“ (acc. pl.), [kōjæ] „kōją“ (acc. sg.); [ūˑ] – [smār.ku] „smārkų“ (acc. sg.), [šāūkštu.] „šāūkštų“ (gen. pl.), [vaī.’g’ū.] „vaīgių“ (gen. pl.), [p’ir-.ktu.] „pirktų“ (subjunctive, 3rd person).

The results of analyzing the material by means of one-way analysis of variance showed a certain influence of the dialect factor, although this is by no means confirmed in all dialect samples. The statistic exceeded the critical value of the F-distribution in only two samples: in the dialect groups along the paternal line, with respect to the use of the short variant (unstressed V̄ #) in contextual style II (F̂ paternal, 1, II)= =2,14 > 1,93=F₆; ₅₄; ₀,₁) and with respect to the use of the long variant (unstressed V̄ #) in contextual style III (F̂ paternal, 3, III=2,1 > 1,93=F₆; ₅₄; ₀,₁).

The Scheffé method of contrasts showed that, for the use of the short variant (unstressed V̄ #) in contextual style II, the mean frequency in dialect group 2—that is, in the speech of informants whose fathers originate from the territory of the southern Aukštaitians—deviates significantly from the overall mean frequency (F̂=3,43 > Fкр. =3,40). In this dialect group, the mean frequency of use of the short variant

¹ The variation of other variables is not indicated in the transcription.

75

(unstressed V̄ #) is the lowest (x̄₂ = 38,2; range of variation of the means 38,2—70).

Analysis of the use of the long variant (unstressed V̄ #) in contextual style III revealed that the mean frequency in dialect group 4 (Samogitians) is significantly below the overall mean frequency (F̂=4,93 > Fкр. = 3,4; x̄₄ = 30, range of variation of the means 30—82).

The data presented are difficult to explain. Regarding dialect group 2, the following can be said: relic features of the dialect of the southern Aukštaitians, who preserve vowel length in the position under study, are evident in the fact that informants in this group pronounced the long variant relatively often in all styles: the x̄₂ value in style I (paternal line) is 34, i.e. the second highest after x̄₁ = 36, which belongs to the dialect group of the western Aukštaitians, who also preserve vowel length in this position; the x̄₂ value in style II (paternal line) is the highest (48,6), and only in style III does it not stand out, although it is relatively high (58). A similar pattern is observed when comparing dialect groups along the maternal line, although mathematical processing of the data did not reveal significant contrasts: the values for use of the long variant by informants in group 2 in all styles, respectively, are: x̄₂, I = 53 (the highest of all groups), x̄₂, II = 63,8 (also the highest), x̄₂, III = 70 (the range of values for the other groups is 47—78). Use of the short variant in the speech of this group, including when the half-long variant is taken into account, shows the opposite pattern: paternal line x̄₂, I = 54 (range 53—80), x̄₂, II = 38,2 (the lowest value), x̄₂, III = 30 (range 16—61); maternal line x̄₂, I = 40 (the lowest), x̄₂, II = 36 (the lowest), x̄₂, III = 24 (range 18—37). The results for dialect group 1 are very similar to those for group 2, confirming the assumption that parents’ dialect influences their children’s speech: these regional dialects are precisely those that best preserve vowel length in word-final position. It should be noted that parents may already be speakers of a “semi-dialect,” since they often merely originate from an area within the dialect region but have lived in Vilnius for many years and do not use the dialect.

A less clear but perceptible correspondence in the use of long and short variants (unstressed V̄ #) is observed in the analysis of the results for dialect group 4, which is most prone to shortening (unstressed V̄ #). The x̄₄ values for use of short variants in all styles approach the highest values: paternal line x̄₄, I = 71 (range 53—80), x̄₄, II = 70 (range 38,2—70), x̄₄, III = 61 (range 16—61); maternal line x̄₄, I = 55 (range 40—77), x̄₄, II = 56 (range 36—64), only x̄₄, III = 20 (range 18—37). Conversely, the x̄₄ value for use of long variants (unstressed V̄ #) in all styles approaches the lowest values: paternal line x̄₄, I = 23 (range 11—36), x̄₄, II = 16 (range 10—48,6), x̄₄, III = 30

76

(range 30–82); along the maternal line x̄₁, I = 24 (range 11–53), x̄₄, II = 22.6 (range 20.57–63.8), x̄₄, III = 75 (range 47–78). These indicators can be explained by the presence of relics of Samogitian reduction of long final vowels in unstressed position, which nevertheless has certain specific features (see: Zinkevičius Z., 1966, p. 115–118). In dialect group 6, whose results did not prove statistically relevant, some relics of ending shortening characteristic of the Panevėžys dialect can also be traced (Zinkevičius Z., 1966, p. 119–120). The results for groups 3, 5, and 7 most often fall in the middle of the interval of mean frequencies, although the indicators for dialect groups 3 and 7 approach those of groups characterized by short variants of the variable under study (group 3—children of Vilnius and Kaunas residents; group 7—children of people born in the territory of the Vilnius dialect). It is interesting to note that although three degrees of length are characteristic of some dialects, no perceptible relics were found in the present study.

To summarize the analysis of the possible influence of parents’ dialect background, some influence can be noted: specifically, 1) the speech of informants whose parents were born in the territory of the southern and western Aukštaitian dialects stands out most clearly; it is characterized to a lesser extent by shortening of long final unstressed vowels; 2) the presence in parental dialects (e.g., Samogitian, Panevėžys, etc.) of certain regular patterns of shortening unstressed endings contributes to strengthening the tendency to shorten long vowels in unstressed word-final position, which occurs in the speech of native Vilnius residents.

Until now, our study has examined variation in the sociolinguistic variable (unstressed V̄#), considering its sources while giving somewhat less attention to the place of this variation in the linguistic system of native Vilnius residents as a whole. Therefore, below we present materials from auditory testing and spectral analysis of sounds, conducted using the methods of traditional experimental phonetics. Since auditory testing is the only experimental method in phonology (Girdenis A., 1981, p. 14), the results obtained by spectral analysis of sounds are considered here as auxiliary in explaining the results of auditory testing.

The material presented in the so-called full style, i.e., without taking contextual styles into account, was investigated using auditory testing. Eight experiments were conducted, examining V̄/V in an unstressed position at the absolute end of a word.

The material for auditory-testing experiments Nos. 5–9, 11, 12, and partly 13 consisted of quasi-minimal pairs differing in the presence of the above-mentioned opposition.

Experiments Nos. 5–8 examined quasi-minimal pairs with the i/ī opposition: 5) mìrti – mìrti, vỹti – výti, skìlti – skílti, kárti – kárti, skỹsti – skýsti; 6) kártys – kártis, ántys – ántis, méntys – méntis, páItys – páltis, kándys – kándis; 7) pasibaigusi – pasibaigusi, pỹlusi – pýlusi, dainãvusi – dainãvusi,

77

svarščiusį – svarščusi, sukaitusį – sukaitusi, ąsōty – ąsōti, šalty – šalti, medy – medi, ūosy – ūosi?). Processing of the auditory-testing data yielded a probability of 0.95–0.999 that the V/V̄ opposition exists. Spectral analysis of the sounds showed no quantitative differences in the realization of this opposition; the absolute durations of the sounds under study differ only slightly: x̄ᵢ=40 ms; ȳᵢ=38 ms; sₓ=13; sᵧ=21, since t=0.67>1.98 (P=95%). Analysis of the qualitative differences in this opposition confirmed the presence of certain differences in sound tenseness. The sum of the deviations Δf of formants F₁ and F₂ from the formants of the neutral vowel [æ] yielded the following results: (ΣΔf̄ᵢ=|260|)−(ΣΔf̄ᵢ=|210|)=|50|.

Experiment No. 9 examined the ē/ĕ opposition, realized in forms of ē-stem nouns in the singular instrumental and accusative cases (ver̄te – ver̄t(ē), silk̄e – silk̄(ē), lēle – lēl(ē), āukle – āukle, bjaur̄ybe – bjaur̄ybe). Calculations of the arithmetic means of the absolute duration of this opposition’s realizations showed that the so-called short vowel is longer than the so-called long one: x̄ē=59 ms, while ȳē=64 ms, sₓ=27; sᵧ=24; t=0.59 < 2.04 (P=95%). Analysis of the qualitative properties of the sounds under study (measurement of formants F₁ and F₂) revealed that there is no difference in the tenseness of ē and ĕ:

Σ Δf̄ē=|280| = Σ Δf̄ē=|280|.

In experiment No. 11, the opposition ā/ă was studied in the forms of a-stem adjectives and o-stem nouns in the nominative, accusative, and instrumental singular (kōra – kōra, līpra – līpra, dūona – dūona, kiāura – kiāura, lóva – lóva). Audio testing yielded a probability of 0.99 that the opposition exists. Spectral analysis showed a minimal difference in the absolute duration of ā and ă (statistically irrelevant): x̄ā=60 ms; ȳā=58 ms; sₓ=35; sᵧ=25; since t=0.29 < 1.99 (P=95%). Qualitative differences exist, but they are relatively small: (ΣΔf̄ā=|290|)−(ΣΔf̄ă=|270|)=|20|.

Experiment No. 13 examined two pairs of hypothetical quasi-minimal pairs (words with two oppositions). They exhibited the ā/ă and ū/ŭ oppositions: forms of verbs in the present indicative, third person, and o-stem nouns in the singular accusative case (sāla – sāl̄a, bāla – bāl̄a, rāštu – rāšt̄u, prakāštu – prakāšt̄u). Results analogous to those of experiment No. 11 were obtained.

Experiment No. 12 examined the ū/ŭ opposition, realized in forms of a- and ia-stem nouns in the plural genitive and singular instrumental cases. Auditors distinguished the vowels ū and ŭ with a probability of 0.999. Spectral analysis of the sounds revealed no statistically significant difference in their absolute duration: x̄ū=70 ms; ȳū=56 ms; sₓ=38; sᵧ=28; since t=1.48 < 2.01 (P=95%). Analysis of the qualitative properties of the ū/ŭ opposition showed that the so-called ŭ is pronounced with greater tenseness:

(Σ Δf̄ū=|240|)−(Σ Δf̄ŭ=|290|)=|50|.

78

Experimental phonetic methods made it possible to establish the following: there are only minimal qualitative and quantitative differences among the three variants (unstressed V̄#), but listeners distinguish the variants well. Context and linguistic expectations apparently play a major role: in connected text, realization of V̄/V̄ is less salient than in the pronunciation of individual words. The fact that listeners did not distinguish the words with 100% accuracy may serve as evidence for the existence of a general process of V̄/V̄ leveling.

The sociolinguistic study of the speech of native Vilnius residents (SVNR) had two aspects—objective and subjective. To date, the objective aspect of variation (unstressed V̄#) has been considered. In Section VII of the sociolinguistic questionnaire (see Appendix 3), informants were asked to indicate their own pronunciation of the variable (unstressed V̄#): 1) dvi valtis or dvi valtys, 2) dvi antis or dvi antys, 3) dvi ausis or dvi ausys (it was specified that these were plural nominative forms). These data characterize the value orientation adopted by members of the Vilnius speech community with regard to the choice of (unstressed V̄#), in other words, the subjective aspect of language.

The questionnaire results (see Fig. 6.1) show that 13 respondents, i.e., 18.57%, reported pronouncing the short variants (three, two, or one of them), while 8 (11.43%) reported short pronunciation of all three examples. Among the informants who indicated a short variant, none had parents originating from the West Aukštaitian dialect area, and only one informant's father was from the South Aukštaitian dialect area. The parents of these 13 respondents were mainly from the East Aukštaitian dialect area and the cities of Vilnius and Kaunas. As the analysis of variance showed, the speech of informants whose parents originate from these dialect areas does not stand out from the speech of the entire SVNR informant group; on the other hand, this speech apparently also shares a feature of SVNR—the tendency to shorten long unstressed vowels in word-final position.

A comparative analysis of the speech recordings of the 13 informants mentioned above and their subjective judgments, as reflected in their questionnaire responses, showed that, in general, all of them shorten long unstressed vowels in final position, but to a lesser extent than they themselves reported: a person uses the short variants (unstressed V̄#), yet believes that they are used more often than they actually are.

Although the sociolinguistic norm (unstressed V̄#) is still emerging, members of the Vilnius speech community are already beginning to become aware of it. Evidence for this can be found in the responses to the sociolinguistic questionnaire, as well as in the fact that members of the SVNR acknowledge that, in general, they tend to shorten long vowels.

4.1.2. Unstressed ī, ē, ā, ū in non-final position in the word.

In standard Lithuanian, long vowels are pronounced long and tense regardless of their position in the word and whether they are stressed.

79

In Lithuanian dialects, there are instances of ĭ, ē, ā, ū being pronounced unstressed in word-final position instead of ī, ē, ā, ū (in approximately one quarter of the territory; see Zinkevicius Z., 1966, pp. 109–110). In some Aukštaitian dialects, there is a clear three-way system of vowel length, so long unstressed vowels become half-long there. Some Samogitians also shorten long vowels in this position, but because of complexities in accentuation, this pattern has certain distinctive features (see ibid., p. 111). Overall, the shortening of long unstressed vowels operates across approximately half of Lithuania, with a number of exceptions and particular features in different dialects.

The sociolinguistic variable of unstressed ī, ē, ā, ū in word-final position (unstressed V̄C) is one of the most salient in SVNR: the short variant is used almost universally. The scale of (unstressed V̄C) is simplified in comparison with that of (unstressed V̄ #), since distinguishing the half-long variant proved difficult: in effect, short and half-long variants are used, but under emphasis, half-long vowels may be pronounced as long.

There are a number of positions in which (unstressed V̄C) occurs: 1) at the absolute beginning of a word, e.g.: [i] or [i.] ([iˑ]) — [ipat’ingas] or [i.paˑt’ingas] „ypatingas“, [is’in’ēš’ēˑ] or [i.s’in’ēš’ē.] „išėšė“ (colloquial past tense, 3rd person); [a] or [a.] ([aˑ]) — [ažuol’inas] or [a.žuol’inas] „ąžuolynas“; [u] or [u.] ([uˑ]) — [ukv’ēd’īs] or [u.kv’ēd’īs] „ūkvedys“; 2) in the second and subsequent pre-stress syllables, e.g.: [i] or [i.] ([iˑ]) — [r’iš’t’ingumas] or [r’i.š’t’ingumas] „ryžtingumas“, [n’eis’ibraus’i] or [n’e.is’ibraus’i] „neįsibrausi“ (future tense, singular, 2nd person); [e] or [e.] ([eˑ]) — [t’es’in’is] or [t’e.sin’is] „tesinys“; [a] or [a.] ([aˑ]) — [saž’in’ingas] or [sa.ž’in’ingas] „sąžiningas“; [u] or [u.] ([uˑ]) — [kud’ik’ēˑl’is] or [ku.d’ik’ēˑl’is] „kūdikėlis“; 3) in the first pre-stress syllable: a) if the vowel in the stressed syllable is long, e.g.: [i] or [i.] ([iˑ]) — [g’iv’ēna] or [g’i.v’ēna] „gyvena“ (present tense, 3rd person); [a] or [a.] ([aˑ]) — [raž’it’is] or [ra.ž’it’is] „ražytis“; [u] or [u.] ([uˑ]) — [ruk’īt’i] or [ru.k’īt’i] „rūkyti“; and b) if the vowel in the stressed syllable is short, e.g.: [i] or [i.] ([iˑ]) — [sugr’iz’imas] or [sugr’i.z’imas] „sugrįžimas“; [e] or [e.] ([eˑ]) — [prat’es’imas] or [prat’e.s’imas] „pratęsimas“; [a] or [a.] ([aˑ]) — [gražā] or [gra.žā] „grąža“; [u] or [u.] ([uˑ]) — [n’esurus] or [n’ēsu.rus] „nesūrus“; 4) in post-stress syllables, e.g.: [i] or [i.] ([iˑ]) — [pùs’r’ić’ē] or [pùs’r’i.ć’ē] „pūsyčiai“; [e] or [e.] ([eˑ]) — [iž’gr’ēž’ēˑ] or [iž’gr’e.ž’ēˑ] „išgrėžė“ (colloquial past tense, 3rd person); [a] or [a.] ([aˑ]) — [māžaja] or [mā-ža.ja] „mažaja“ (accusative singular); [u] or [u.] ([uˑ]) — [pósuk’is] or [pōsu.k’is] „posūkis“.

The analysis of the interview material for the study of (unstressed V̄C) was conducted without taking into account the positions in which this variable occurs in the word, since there was insufficient material for this purpose. The age (A₁, A₂, A₃) and education (C₁, C₂) groups were considered, as were two contextual styles (B₁ and B₂) (in style B₃, i.e., when reading quasi-homonyms, this variable occurred too rarely). The number of hours was calculated—

80

The proportion of the long and short variant (unstressed V̄C) (in%) was calculated, and a graph was plotted taking into account the factors of age (A), education level (C), and style (B). The speech of the group of actors and radio announcers (C₃) was also taken into account (see Fig. 5.1).

Variation in (unstressed V̄C) is generally slight when the speech of the actors and announcers is excluded. No substantial shift is observed; only a slightly greater number of long variants are used in the formal, non-directed style (B₂), and the confidence intervals overlap. The use of long variants differs statistically significantly only in the speech of the younger group with lower and higher education levels: the confidence intervals do not overlap—11 ÷ 17% and 0 ÷ 8%. In the group of actors and announcers, the frequency of the long variant, although considerably higher than in the other groups (where it ranges from 4–24%), is nevertheless relatively low: 66% in B₁ and 70 in B₂.

The results of the analysis of the distribution of (unstressed V̄C) in LRKLV support the conclusion that this variable has acquired linguistic status: it is almost impervious to both stratificational and situational variation, yet (unstressed V̄C) remains a sociolinguistic variable, since variation does exist: i, ē, ā, ū are not always pronounced as i, ē, ā, ū in a non-final unstressed position—only in approximately 4/5 of cases.

Analysis of the speech of informants who do not characteristically use (unstressed V̄C) in the manner described showed that the main factor explaining why their speech is “uncharacteristic” of LRKLV is the dialectal origin of their parents. Both parents (or one parent) of the informants in the group “uncharacteristic” in this respect came from areas of Lithuanian dialects in which long vowels are not shortened in non-final position.

Investigation of the subjective aspect of this question, i.e., analysis of the questionnaire responses, helped clarify the sociolinguistic norm of short pronunciation (unstressed V̄C) (the questionnaire included examples of (unstressed V̄C) in all the word positions mentioned above).

Use of (unstressed V̄C) realized at the absolute beginning of a word was determined from responses about the pronunciation of the words ąsotelis, ąsėlė, ąžuolynas, išuotas. Overall, the frequency of reported use of the long variant in the first three words is relatively low (range 0–40%) (see Fig. 6.2), which, to some extent, almost corresponds to the actual use of (unstressed V̄C) shown in the graph analyzed above. The same pronunciation of ā is not reported for all three words: in the word ąžuolynas the long variant is used more often (22–40%), while in the word ąsėlė it is used less often (0–40%). Interestingly, some discrepancies are evident in respondents’ reports by education level, especially for the first two words: respondents with a higher education level, particularly in the older and middle age groups, report using the long variant (unstressed V̄C) in these words less often than respondents with a lower education level.

6. 949

81

It should be emphasized that the frequency of the long variant in the speech of the group of actors and announcers is quite low—33%.

The words ąsotėlis and ąžuolynas were included in a continuous text that all informants read during the interviews. Comparing actual use of the long variant in these words with respondents’ subjective judgments showed that it was precisely those respondents whose speech never or almost never features the long variant who overstated their rates in the questionnaires; for example, in the speech of the older group with a lower education level (A₁C₁), interview data show that the frequency of the long variant in the words ąsotėlis and ąžuolynas is only 4%, whereas the questionnaires report 40. This pattern was found in all groups, especially those with a lower education level, except for the group of actors and announcers: they underestimated their rate in the questionnaires (33%) relative to the actual rate (90% use of the long variant in the text).

Analysis of use of the long variant in the word išiuotas (although it belongs to this group of words) is presented below, in the discussion of the group of words with (unstressed V̄C) before a long stressed syllable, since it follows the rules of that group.

Comparing realization of the long variant (unstressed V̄C) in the formal, non-directed style with speakers’ subjective judgments leads to the conclusion that the rates of use of the long variant in speech are overstated: apparently, the written form of the word had an influence.

The long variant (unstressed V̄C) at the absolute beginning of a word is quite rare in LRKLV, apparently because vowel length in this position is often a historical relic and performs no distinctive function.

The distribution of frequencies of the long variant (unstressed V̄C) realized in the second and subsequent pre-stressed syllables (illustrated in the questionnaire by the words tęsinys, gręžinys) is somewhat different from that in the absolute initial position in a word (see Fig. 6.3). In both words, the reported frequency of use of the long variant ē is relatively uniform and not especially high, averaging 30–40%. The rate for the word gręžinys in the middle-aged group with a lower education level rises sharply, however—to 56%—but falls to 14% in the younger group with a higher education level. Here, as in the analysis of questionnaire responses on the distribution of variants of (unstressed V̄C) at the absolute beginning of a word, the frequency of the long variant declines in the group with a higher education level. In the group of actors and announcers, the frequency is high: 68% for the word gręžinys and 78% for the word tęsinys.

Comparison of the frequency of the long variant (unstressed V̄C) in the informants’ actual speech with their subjective judgments presents a picture similar to that found in the analysis of (unstressed V̄C) in the absolute initial position in a word: questionnaire rates are somewhat overstated, especially in groups with a lower education level, and understated in the group of actors and announcers.

82

In the use of the long variant (unstressed V̄C), realized in the II and subsequent pre-stress syllables, there is a tendency to shift to the use of the short variant, especially in words with historically long vowels—the nasal a and e.

This conclusion is confirmed by an analysis of the questionnaire responses, which was aimed at studying the distribution of the long variant (unstressed V̄C) realized in the I pre-stress syllable before a long stressed syllable. This position (unstressed V̄C) is illustrated in the questionnaire by the following examples: plytėlė, gyvėnti, vyriausias, drašuolis, žąselė, rūkýti, grūdai. The graph (see Fig. 6.4) clearly separates two groups of words: the frequency curves for drašuolis and rūkýti present a picture analogous to that seen in the examples for the position at the absolute beginning of the word (ąsotėlis, ąšelė, ąžuolynas; see Fig. 6.2), whereas the curves for the words plytėlė, gyvėnti, vyriausias, rūkýti, grūdai resemble the curve for ūsiuotas, that is, in the position at the absolute beginning of the word. In these 6 words, respondents report using the long variant (unstressed V̄C) much more often (range 33–100%, with the main frequencies distributed in the 60–90% range). The opposite pattern is observed here compared with the examples cited above: whereas at the absolute beginning of a word, representatives of the less educated groups more often reported using the long variant, in the I pre-stress syllable before a long stressed syllable, representatives of the more highly educated groups more often reported using it.

The words plytėlė and gyvėnti were included in a connected text read by all interviewees. Actual use of the long variant (unstressed V̄C) in these words does not correspond to respondents’ subjective judgments about their own speech: with the exception of announcers and actors, the long variant was used in no group in more than 20% of cases. With respect to age groups, the frequency figures for the long variant in the position under study stand out somewhat only in the older, less educated group—they are lower, in the 20–60% range. When reading the text, only one member of this group used the long variants; their parents came from an area in the territory of the Southern Aukštaitian dialects, which, as is known, preserve vowel length in this position.

The distribution of the long variant (unstressed V̄C) realized in the I pre-stress syllable before a short stressed syllable was studied using the examples sugrįžimas, Kęstutis, trešimas, sugrąžino, siųstuvas, included in the questionnaire. In these words, the variables under study are i, e, a, u. The frequency-distribution curves for the long variant in the words Kęstutis, trešimas, sugrąžino (see Fig. 6.5) resemble those for the other analyzed examples with nasal vowels: here, too, the less educated groups reported more frequent use of the long variant, and the overall frequency of reports of using it is relatively low (25–57%). The curves for the words sugrįžimas and siųstuvas seem to belong to another type: they are considerably higher—

6*

83

—reaching 90%—and are not symmetrical with respect to educational level; for example, the highest peaks belong to the older group with a high level of education and the younger group with a lower level. Apparently, the fact that related words bearing stress on the long vowels under study are frequently used in speech plays a role here: sugrįžo, siųsti, etc.

The words sugrįžimas, Kęstūtis, sugrąžino were included in a connected text read by the interviewees. Long vowels in all the words were pronounced only by radio announcers and two members of the middle generation of Vilnius residents whose parents came from areas in the territory of the Western and Southern Aukštaitian dialects. Although long pronunciation of nasal vowels is widespread in dialects, relics of this phenomenon are scarcely found in the speech of native Vilnius residents. Even in the speech of actors, long pronunciation of these vowels in the text occurred in only 40% of all recorded cases. Isolated instances of pronouncing a long vowel in one or two of the specified words were attested in 9% of all cases (excluding the announcer and actor group), more often in the word sugrįžimas. Not a single instance of pronouncing the long variant (unstressed V̄C) in the position under study occurred in the speech of representatives of the older generation.

The distribution of the long variant (unstressed V̄C) realized in a post-stress syllable was studied using the examples šlūostyti, kráustyti, džiáustyti, included in the questionnaire (see Fig. 6.6). Respondents reported relatively frequent use of the long variant (most frequency values fall in the 60–80% range). No clear stratification is observed along either the age or education factor. The middle-aged group with a high level of education stands out somewhat: its members assess their own speech more skeptically, but the frequency they report is nevertheless inflated—33–56% use of the long variant, whereas in reality this figure should be approximately half as high.

The word kráustytis was included in a connected text for the interview. Use of the long variant was recorded in 23% of cases (excluding the actor and announcer group), mainly in the speech of middle-generation informants whose parents came from areas of the Southern or Western Aukštaitian dialects.

Studying the subjective aspect of the use of one or another pronunciation variant allows us to conclude that two sociolinguistic norms exist. The first applies to words with nasal vowels, e.g., Kęstūtis, ažuolynas, sugrąžino, etc., and can be formulated as follows: pronouncing long nasal vowels in a non-final, unstressed position in a word is not obligatory, although it is possible. The choice of variant (unstressed V̄C) often depends on the formality of the situation (the long variant is used more often when reading a text than in casual speech), on subjective attitudes toward one’s own speech (in formal situations, some strive to pronounce (unstressed V̄C) more correctly), and, finally, on the dialectal origins of one’s parents (if a person has never heard the long variant of this vari—

84

… in his own speech, even with moderate efforts to speak correctly, long variants will be very rare). This type of variation in a sociolinguistic variable can be considered a type of intrapersonal variation (see Bell R., 1980, p. 58): variation is determined neither by linguistic nor by statistical demographic categories, but by dynamic aspects of the situational use of language. Nevertheless, the variable cannot be considered situational, since it does not lend itself to stylistic stratification. Variation in this variable can be characterized as occurrence-based (ibid., p. 61): it can never be predicted.

The second sociolinguistic usage norm (unstressed V̄C) applies to words with long, but not nasal, vowels, e.g.: gyvẽna, šypsōtis, plytēle, etc.: there is a norm favoring the selection of the long variant of the variable, which nevertheless allows the short variant to be used without any specific stratificational or situational restrictions. This is indicated by the fact that population groups with lower levels of education, when making subjective judgments about their own speech, choose the short variant of the variable more often than groups with higher levels of education, which attests to the absence of stylistic variation and to the long variant (unstressed V̄C) not being established as a prestigious form of speech.

Adopting the stricter norm, which is the first one, would signal a clear shift toward the loss of the opposition between long and short vowels as a whole and its replacement by an opposition between stressed and unstressed vowels. Respondents’ subjective judgments diverge so greatly from the actual situation because, at its present stage of development, LRKLV cannot fully replace the first opposition with the second: consciously acknowledging that in 4/5 of cases a long unstressed vowel is pronounced short would mean that the pronunciation norm is the short variant, i.e. the loss of V̄/V. Thus, the short pronunciation variant (unstressed V̄C) exists as a norm which, unlike the pronunciation norm for historical nasal vowels in non-final unstressed position, is unconscious.

4.2.1. Stressed i, ė, ã, ū in word-final position.

In the Lithuanian standard language, as in the dialects, stressed i, ė, ã, ū at the absolute end of a word or before -s are always pronounced short. In some dialects bordering Slavic-speaking areas and in the speech of some bilinguals, semi-long or long pronunciation of these vowels is observed (see Zinkevičius Z., 1966, p. 104, 107).

In Vilnius residents’ speech, instances of short, semi-long, and long (in emphasis) pronunciation of the vowels under study occur in the following positions: 1) after a syllable with a long vowel or diphthong, e.g.: [i] or [i.] ([iˑ]) — [suˑrˈi] or [surˈi.], „sūri“, [e] or [e.] ([eˑ]) — [kˈaelˈu̯osǣ] or [kˈaelˈu̯osˈae.], „keliuose“ (loc. pl.), [a] or [a.] ([aˑ]) — [ruoša] or [ruoša.], „ruoša“, [u] or [u.] ([uˑ]) — [ruˈsˈū] or [rusˈū.], „rūsiū“ (instr. sg.); 2) after a syllable with a short vowel, e.g.: [i] or [i.] ([iˑ]) — [nˈešˈi] or [nˈešˈi.], „neši“ (2nd pers. sg. pres.), [e] or [e.] ([eˑ]) — [katˈǣ] or [katˈǣ.], „katė“ (instr. sg.), [a] or [a.] ([aˑ]) —

85

[balà] or [balā.] „balā“ (nom. or instr. sg.), [u] or [u.] ([uˑ]) — [gʹarū] or [gʹarū.] „gerū“ (instr. sg.).

The scale for the sociolinguistic variable of short stressed vowels in word-final position (hereafter stressed V̌#) comprises two values: the short and long variants. A semi-long vowel in emphasis can attain the length of a genuinely long vowel. This variable is not characteristic of LRKLV, but since the Lithuanian language system precludes the emergence of a long variant (stressed V̌#), instances of its occurrence in speech are highly conspicuous and are regarded as slang or interference forms.

In the present study, this variable was investigated only on the basis of sociolinguistic interview material in styles B₁ and B₂.

Because of accentual variability, which is not considered here, the occurrence of words with (stressed V̌#) varies from informant to informant. If words such as balti (nom. pl.), namè (loc. sg.), kalbà (nom. or instr. sg.), kitùs (acc. pl.) are all pronounced by informants with (stressed V̌#), then words such as aštuoni, musè (instr. sg.), dienàs (acc. pl.), maloniùs (acc. pl.) frequently lose (stressed V̌#) in LRKLV: the stress shifts to the penultimate syllable. In rare cases, (stressed V̌#) appears through hypercorrection, that is, through shifting the stress to the final syllable, e.g., širdìs (acc. pl., = širdis), nendrės (acc. pl., = nénadres), galvàs (acc. pl., = gálvas), rūščius (acc. pl., = rūščius). For example, when reading connected text during interviews, the following cases of hypercorrection that give rise to (stressed V̌#) are found: devynìs (acc., = devýnīs), vienù (instr., = viēnu), guklè (instr. sg., = áukle), mergaitès (acc. pl., = mergáites), bjaurybè (instr. sg., = bjaurýbe).

Study of the interview texts showed that the long variant (stressed V̌#) occurs rarely in LRKLV (see Fig. 5.2): the frequency range for the short variant of the variable is high—76—100%. Somewhat more long variants were recorded in the speech of young people, although only two frequencies in style B₂ differ statistically significantly: those of the older group with a high level of education and the younger group with a lower level (86÷96% and 70÷82%). Some corrections are needed here: in the speech of older-generation speakers, the long variant occurs even more frequently than in the speech of young people. Overall, however, (stressed V̌#) is observed very rarely in their speech—older-generation speakers more often shift the stress to the preceding syllable. Hypercorrect forms occur more often in the speech of young people, which overall contributes to lowering the frequency rate of the short variant.

The connected interview text contained 36 examples of (stressed V̌#) in multisyllabic words and 21 in monosyllabic words. All monosyllabic words in the speech stream, except for dvi, became proclitics or enclitics; only in a few cases was the long variant (stressed V̌#) observed in the words nèt, bèt, nès, čìa. Only 1/3 of the 36 multisyllabic words containing (stressed V̌#) consistently retained stress on the final syllable: manè (acc. sg.), tadà, kuris, tikrù (instr. sg.), niekadà, namè (loc. sg.), aplinkà

86

lietus, pilte (“verbal adverb”). Some words (out of the 36 examined) occurred with (stress V#) in only a few informants’ speech, e.g.: sklypū (instrumental singular), ievà (instrumental singular), vertė (instrumental singular), Džèkū (instrumental singular), Džīmīū (instrumental singular).

The sociolinguistic variable (stress V#) is at a stage of emergence; it is a characteristic example of inherent variation (Bell R., 1980, pp. 58–59): it is entirely unpredictable even in the speech of the same person in the same situation.

The question of the sociolinguistic norm (stress V#) in the speech of native Vilnius residents is complex: no member of this speech community consciously acknowledges the long variant (stress V#) as normative even in Vilnius speech; such use of the variable is considered objectionable in every case. In fact, in the speech of native Vilnius residents the long variant (stress V#) is systematic. This systematicity depends on how deeply the tendency to pronounce all stressed vowels long—both those historically long and those historically short—has taken root in an informant’s speech.

4.2.2. Stressed i, ē, ā, ū in non-final position in the word.

In standard Lithuanian, stressed i, ē, ā, ū are pronounced short. There are many words with stressed i and ū in standard Lithuanian, but only a limited number with stressed ē and ā, since these vowels underwent secondary lengthening (Zinkevičius Z., 1981, vol. 1, p. 57). Stressed ē and ā are never lengthened only in the western Samogitian dialect; in the other dialects, the occurrence of short vowels in various grammatical forms varies: some lengthen these vowels in certain forms, others in different forms (Zinkevičius Z., 1966, pp. 103–107). Stressed i and ū are lengthened less often in dialects than ē and ā: most Samogitians, western and southern Aukštaitians—that is, the population of almost half of Lithuania’s territory—do not lengthen stressed i and ū, but in some cases broaden them (ibid., pp. 107–108).

The vowels i, ē, ā, ū + r, l, m, n occur in tautosyllabic combinations i, ē, ā, ū + r, l, m, n, but only in combinations with i and ū, when stress falls on the first component, are i and ū pronounced short. The vowel ū occurs in the diphthong ui, where it is likewise pronounced short when stress falls on the first component.

In the dialects, the first component of ū, i + r, l, m, n and ui is pronounced short or lengthened only slightly by some western Aukštaitians, certain Samogitian border communities, and speakers of the Zietela dialect; that is, this feature is observed in only a small proportion of dialects (see Zinkevičius Z., 1966, pp. 108–109).

The scale of the sociolinguistic variable involving short stressed vowels in non-final position (hereafter stress VC) has three grades: short, half-long, and long variants. Ultimately, the half-long and long variants are considered together as indicators of vowel lengthening. The variable under study is highly characteristic of the speech of native Vilnius residents. Apparently, the fact that the use of long (half-long) variants has many parallels in the dialects plays an important role.

87

In the speech of Vilnius residents (stress ŬC), it occurs in the following positions: 1) at the absolute beginning of a word, e.g.: [i], [i.], [iˑ] — [lma], [l.ma], [lma] „lma“ (present tense, 3rd person), [ɪrˑtʲi], [ɪ.rˑtʲi], [ɪrˑtʲi] „irti“, [e], [e.], [eˑ] — (only in words of foreign origin) [estas], [e.stas], [estas] „estas“, [ɛmka], [e.mka], [ɛmka] „ɛmka“, an MK-brand motorcycle, [a], [a.], [aˑ] — [atʲnʲaesa], [a.tʲnʲaesa], [atʲnʲaesa] — „atnеса“ (present tense, 3rd person), [amtʲelʲeˑtʲi], [a.mtʲelʲeˑtʲi], [amtʲelʲeˑtʲi] „amtelėti“, [u], [u.], [uˑ] — [uškʲisa], [u.škʲisa], [uškʲisa] „užkiša“ (present tense, 3rd person), [uˑtʲi], [u.tʲi], [utʲi] „ūti“, [ūrmu], [ū.rmu], [ūrmu] „ūrmu“; 2) before a syllable containing a long vowel (or a diphthong), e.g., [i], [i.], [iˑ] — [kʲitaˑ], [kʲi.taˑ], [kʲitaˑ] „kitą“ (accusative singular), [pʲilnaˑ], [pʲi.lnaˑ], [pʲilnaˑ] „pilną“ (accusative singular), [e], [e.], [eˑ] — [balʲtʲesʲnʲiˑ], [balʲtʲe.sʲnʲiˑ], [balʲtʲesʲnʲiˑ] „baltėsni“ (accusative singular), [a], [a.], [aˑ] — [mānɔˑ], [mā.nɔˑ], [mānɔˑ], „māno“ (possessive pronoun), [u], [u.], [uˑ] — [būvɔˑ], [bū.vɔˑ], [būvɔˑ] „būvo“ (simple past, 3rd person), [žūikʲiˑ], [žū.ikʲiˑ], [žūikʲiˑ] „žūikį“ (accusative singular), [mūrʲgdʲitʲi], [mū.rʲgdʲitʲi], [mūrʲgdʲitʲi] „mūrkdyti“; 3) before a syllable containing a short vowel, e.g.: [i], [i.], [iˑ] — [vʲisas], [vʲi.sas], [vʲisas] „visas“, [pʲilʲtʲi], [pʲi.lʲtʲi], [pʲilʲtʲi] „pilti“, [e], [e.], [eˑ] — [žalʲesʲnʲis], [žalʲe.sʲnʲis], [žalʲesʲnʲis] „žalėsnis“, [a], [a.], [aˑ] — [jæunāsʲis], [jæunā.sʲis], [jæunāsʲis] „jaunasis“, [u], [u.], [uˑ], — [sūka], [sū.ka], [sūka] „sūka“ (present tense, 3rd person) [žūikʲis], [žū.ikʲis], [žūikʲis] „žūikis“, [pūlʲtʲi], [pū.lʲtʲi], [pūlʲtʲi] „pulti“; 4) after a syllable containing a long vowel (or a diphthong), e.g.: [i], [i.], [iˑ] — [rʲitʲinʲis], [rʲitʲi.nʲis], [rʲitʲinʲis] „rytinis“, [iˑpʲilʲtʲi], [iˑpʲi.lʲtʲi], [iˑpʲilʲtʲi] „ipilti“, [e], [e.], [eˑ] — [apkʊnʲesʲnʲis], [apkʊnʲe.sʲnʲis], [apkʊnʲesʲnʲis] „apkūnėsnis“, [a], [a.], [aˑ] — [gʲivāsʲis], [gʲivā.sʲis], [gʲivāsʲis] „gyvasis“, [u], [u.], [uˑ] — [nʲūrūsʲis], [nʲū.rū.sʲis], [nʲūrūsʲis] „niūrusis“, [iˑpūlʲtʲi], [ipū.lʲtʲi], [ipūlʲtʲi] „ipulti“; 5) after a syllable containing a short vowel, e.g.: [i], [i.], [iˑ] — [užʲmʲikʲtʲi], [užʲmʲi.kʲtʲi], [užʲmʲikʲtʲi] „užmigti“, [nutʲilʲtʲi], [nutʲi.lʲtʲi], [nutʲilʲtʲi] „nutilti“, [e], [e.], [eˑ] — [suvʲešʲtʲi], [suvʲe.sʲtʲi], [suvʲešʲtʲi] „suvėsti“, [a], [a.], [aˑ] — [nʲekāsʲū], [nʲekā.sʲū], [nʲekāsʲū] „nekąsiu“, [u], [u.], [uˑ] — [ušʲčʲūptʲi], [ušʲčʲū.ptʲi], [ušʲčʲū.ptʲi] „užčiūpti“, [ižgūitʲi], [ižgū.itʲi], [ižgūitʲi] „išgūti“, [prʲikūlʲtʲi], [prʲikū.lʲtʲi], [prʲikūlʲtʲi] „prikulti“.

In the present study (stress ŬC), the phenomenon was investigated using interview texts in styles B₁, B₂, and B₃ and questionnaire responses; in addition, audio testing and spectral analysis of the sounds were carried out.

Analysis of the interview texts, which did not take into account the position of (stress ŬC) in the word but did take into account the factors of age (A), contextual style (B), and education level (C), showed that the use of the variants of (stress ŬC) differs significantly across age groups (see Fig. 5.3). Determination of confidence intervals revealed that, with the exception of the oldest and youngest groups of informants with a high level of education, the variable is situational in nature, since clear variation occurs across different contextual styles. In the speech of the group of actors and announcers, instances of use of the long variant of (stress ŬC) are observed only when i, й form part of tautosyllabic sequences or the diphthong ui, and even then rela-

88

tively rarely: the rates for the short variant (stress УС) fall within the range of 91–100%.

The interview texts mainly exhibit short and half-long variants, while the low percentage (1–17) frequency of long variants is due to the fact that long variants can usually occur in two cases: 1) when the vowels under study form part of tautosyllabic sequences or the diphthong i; and 2) when the pronunciation of words containing this variable is subject both to substantial dialectal variation and to exceptions of a kind to the general rules of accentuation in the standard language. For this reason, a graph was prepared showing only the use of the short (and, correspondingly, long) variant (stress УС).

The qualitative and quantitative characteristics, as well as the presence of an opposition between short and long stressed vowels in homonymous words, were analyzed using audio testing and spectral analysis of the sounds.

Five audio-testing experiments were devoted to the study of this opposition. The material for the first three experiments consisted of minimal pairs distinguished by the presence of long ā or short ă in the stressed position in the word root. For Experiment No. 1, quasi-homonyms were selected, represented by different case forms of the a-stem noun rāstas and the masculine passive participle rāstas²; for Experiments Nos. 2 and 3, a series of verb forms of kāšti and kàsti was used (see Table 4.1). A probability of ≥0.95 was obtained that the forms differ. The listeners additionally indicated that the vowels differ in quality (but not quantity). Spectral analysis of the sounds showed that there is no significant difference between the absolute durations of the vowels under study: x̄ā = 146 ms, ȳā = 139 ms, sₓ = 33, sᵧ = 36, since t = 1.18 < 1.98 (P = 95%). Next, formants F₁ and F₂ were measured in spectrograms of the vowels under study, and the total deviations Δf from the corresponding formants of the neutral vowel [æ] were calculated in order to determine the difference in tenseness between long and short vowels. It was found that ā is pronounced with greater tenseness:

(Σ Δ f̄ₐ = |270|) − (Σ Δ f̄ₐ = |240|) = |30|.

For Experiment No. 4, the audio test presented minimal pairs distinguished by long ū and short ŭ in the word root in a stressed position: the material under study included different verb forms of skūsti, skùsti and sū̇šti, sùšti. The probability that the ū/ŭ opposition exists is 0.999. Spectral analysis of the sounds indicated no quantitative difference, i.e., no difference in the absolute duration of the sounds: x̄ᵤ = 111 ms, ȳᵤ = 96 ms, sₓ = 45, sᵧ = 36, since t = 1.02 < 2.04

² The participle rāstas, like the participle forms kāstas presented in Experiment No. 2, takes a rising intonation in the specified forms in codified Lithuanian (Laigonaitė A., 1978, p. 91), but this rule is not observed in the speech of indigenous Vilnius residents: the vowel a is interpreted as short and stressed.

89

(P=95%). When investigating qualitative differences between ā and ă, the minimal difference in the total deviations Δf of the formants F₁ and F₂ from the corresponding formants of the neutral sound [æ] was obtained:

(Σ Δf̄ā = |250|) − (Σ Δf̄ŭ = |240|) = |10|.

In the pronunciation of vowels of the u type in the speech of native Vilnius residents, they exhibit relative delabialization, which naturally also affects the acoustic properties of vowels of this type.

Experiment No. 13 investigated “double” minimal pairs, which can only provisionally be called that: these are words with an opposition between long and short vowels occurring both in the root and in the word ending. Root-position ā and ă occurred under stress in the verb forms sálti, bálti, prakāsti, among others, and in the α-stem noun rãstas (see Table 4.1). The auditors distinguished these words with a probability of 0.95. Spectral analysis of the sounds reproduced the results of Experiments Nos. 1, 2, and 3 (see Table 4.3).

Study of the subjective aspect of usage (stressed VC) based on the questionnaire showed that certain consciously recognized sociolinguistic norms exist with respect to (stressed VC). This variable is not homogeneous, as respondents’ subjective judgments clearly indicate. First, respondents point to different usage of the variable when the short stressed vowel is i, ǐ, as opposed to ā, ě (cf. Figures 6.8 and 6.9 with 6.7 and 6.10, Appendix 6). In Figures 6.7 and 6.10, the frequencies of reports of using the short variant in words with ā, ě (excluding the frequencies for the words būvo and snīgo (Figure 6.7), which contain i, ū) lie somewhat below the frequencies for words with i, ǐ in Figures 6.8 and 6.9, indicating that short pronunciation of ā, ě is less obligatory. Second, short pronunciation in tautosyllabic combinations of stressed vowels with sonorants is rated relatively lower than short pronunciation of vowels i, ǐ; for example, the frequency lines for the short variant in the words snīgo, būvo (Figure 6.7), kìtas, vìsas (Figure 6.8), and ilgūmas, pabūsti (Figure 6.9) lie very high, in the 80–100% range (with some deviations, discussed below), indicating that virtually all native Vilnius residents are oriented toward short pronunciation (stressed VC) under the linguistic circumstances mentioned.

The orientation toward different usage of the variable under different linguistic circumstances is especially pronounced in the analysis of responses from the older group: frequencies for i, ǐ are recorded in the 67–100% range, while those for i, ǐ+R are 20–50%. In the responses from the other two age groups, this distinction is not as clear; the frequencies for some words (e.g., pìlti, pùmpuras) do not even lend themselves to this kind of interpretation. However, analysis of the responses reveals another pattern that may generally apply to all responses concerning usage (stressed VC): frequencies of short variants in the responses of the middle-aged group, and especially the younger group with lower educational attainment, rise sharply and reach the upper end of the scale, around 70–100%. In all cases

90

the indicators for the younger age group with higher educational attainment behave differently: they clearly decline, although in some cases a difference in usage (stressed V̆C) is also observed in tautosyllabic combinations.

Analysis of Figures 6.7, 6.8, 6.9, and 6.10, and comparison with actual speech data from native Vilnius residents (see Figure 5.3), support the conclusion that there is a consciously recognized sociolinguistic norm governing the use of (stressed V̆C), which varies among age groups: in the older age group, long variants (stressed V̆C) predominate; this is also recognized by the informants themselves, except when the short vowels in the variable are ĭ, ŭ. In the speech of the middle-aged group, long and short variants occur in almost equal numbers, as the informants’ subjective judgments also indicate, although the frequencies of short variants are overestimated—which is natural, since subjective assessments are almost always more favorable than the actual figures. The younger generation of Vilnius residents appears to have two norms. Members of this generation with lower educational attainment, mainly secondary-school students, use the greatest number of short variants (stressed V̆C), and are clearly oriented toward short variants. Those with higher educational attainment use slightly more than half short variants, but, unlike the middle-aged informants, do not overestimate their figures at all when subjectively evaluating their own speech: they are fully aware that they use long variants (stressed V̆C) when tautosyllabic combinations occur in their speech, which apparently is considered their sociolinguistic norm. Actors and announcers usually realize the short variant of the variable (with some exceptions).

Non-statistical analysis of the interview texts showed that the position of the variable in the word also has a certain bearing on the choice of variants (stressed V̆C): if long vowels or diphthongs occur before or after it and are realized in their long variants, (stressed V̆C) is usually realized as a short variant; but if the neighboring sounds are realized in their short variants, (stressed V̆C) is much more often represented by a long variant.

Study of the use of (stressed V̆C) in relation to the dialectal origins of the parents confirmed the existence of a general pattern also found in the study of other variables: the short variant is used by those interviewees whose parents come from localities within dialect areas where short vowels under stress are not lengthened. This pattern is especially evident when examining short vowels in combinations with sonorants.

The study of the pronunciation of short stressed vowels showed that there are cases in which, to avoid pronouncing a long vowel in V̄+R combinations, stress shifts from the vowel to the sonorant; from the perspective of the Lithuanian language system, this is considered a greater deviation than using a long vowel.

91

Conclusions. 1. The sociolinguistic study of the speech of native Vilnius residents showed that, although it is a heterogeneous structure, it is nevertheless a systemic phenomenon. An intrinsic property of the speech of native Vilnius residents—variation—depends on the influence of a number of factors, both linguistic and sociological, the study of which is inconceivable without the use of comprehensive research methods. 2. One of the causes of variation in the speech of native Vilnius residents is the varying strength and range of the influence of the dialect features of their parents’ speech on the speech of its speakers. 3. The differing use of variants of some variables conditions the use of other variables; however, the variants are interchangeable depending on the social situation, generating an impulse for further variation. 4. The study of the speech of native Vilnius residents showed that all its phonetic features have parallels in Lithuanian dialects or are observed in the speech of speakers of borderland dialects who are active contact bilinguals. 5. The most characteristic features of the vowel system of the speech of native Vilnius residents should be considered the tendency to shorten long tense vowels in unstressed position and to lengthen short lax vowels under stress.

СПИСОК ЛИТЕРАТУРЫ

Андреев Э. П., Осипов Г. В. Послесловие. – В кн.: Математические методы в социальных науках / Пер. с англ. М., 1973.

Анкета „Эксперт-2“ / ИСИ АН СССР. Экспертная служба. – М., 1974.

Белл Р. Социолингвистика. Цели, методы, проблемы. – М., 1980.

Буданов Р. А. Борьба идей и направлений в языкознании нашего времени. – М., 1978.

Вайнберг Дж., Шумекер Дж. Статистика / Пер. с англ. – М., 1979.

Гаршия К. Слоговые акценты в фонологической системе (на материале литовского языка): Канд. дис. – М., 1977.

Таскаров Д. В., Шаповалов В. И. Малая выборка. – М., 1977.

Траудина Л. К. Вопросы нормализации русского языка. Грамматика и варианты. – М., 1980.

Троссе Р., Нойберт А. Тезисы к марксистской социолингвистике. – В кн.: Теория и история языкознания. Проблемы зарубежной социолингвистики: Реф. сб. / Отв. ред. А. Д. Швейцер. – М., 1976.

Дешериев Ю. Д. Закономерности развития и взаимодействия языков в советском обществе. – М., 1966.

Дешериев Ю. Д. Социальная лингвистика. – М., 1977.

Закс Л. Статистическое оценивание / Пер. с нем. – М., 1976.

Звегинцев В. А. Лингвистика в системе наук и исследовательских методов. – В кн.: Научно-техническая революция и функционирование языков мира / Отв. ред. И. К. Белодед, Ю. Д. Дешериев. М., 1977.

Здравомыслов А. Г. Методология и процедуры социологических исследований. – М., 1969.

Исследования фонетических особенностей речевых стилей / Под ред. В. Д. Аракина. – М., 1979.

Косериу Э. Синхрония, диахрония и история. – Новое в лингвистике. М., 1963, вып. 3.

Костомаров В. Г., Шварцкопф Б. С. Работы по вопросам культуры речи (1962–1965 гг.). – ВЯ, 1965, № 4.

92

Лабов У. Исследование языка в его социальном контексте. – Новое в лингвистике. М., 1975 (в ссылке: а), вып. 7.

Лабов У. О механизме языковых изменений. – Новое в лингвистике. М., 1975 (в ссылке: б), вып. 7.

Маковский М. М. Соотношение индивидуальных и социальных факторов в языке. – ВЯ, 1976, № 1.

Монина М. Л. Критический очерк методов и техники социологических исследований за рубежом. – Инф. бюл. Сер. Материалы и сообщения / Отв. ред. В. В. Колбановский. М., 1974, № 1.

Никольский Л. Б. Синхронная социолингвистика (теория и проблемы). – М., 1976.

Носенко Э. Л. Специфика проявления в речи состояния эмоциональной напряженности: Автореф. докт. дис. – М., 1979.

Поливанов Е. Русский язык как предмет грамматического описания. – В кн.: За марксистское языкознание. М., 1931.

Постовалова В. И. Историческая фонология и ее основания. Опыт логико-методологического анализа. – М., 1978.

Постовалова В. И. Специфические черты гипотезы в лингвистике. – В кн.: Гипотеза в современной лингвистике / Отв. ред. Ю. С. Степанов. М., 1980.

Рабочая книга социолога / Отв. ред. Г. В. Осипов. – М., 1976.

Русская разговорная речь / Под ред. Е. А. Земской. – М., 1973.

Русский язык по данным массового обследования. Опыт социально-лингвистического изучения / Под ред. Л. П. Крысина. – М., 1974.

Руткевич М. Н. Методологические проблемы изучения социальной структуры социалистического общества. – Свердловск, 1972.

Сегал Д. М. Основы фонологической статистики (на материале польского языка). – М., 1972.

Серебренников Б. А. Вероятностные обоснования в компаративистике. – М., 1974.

Скворцов Л. И. Теоретические основы культуры речи: Автореф. докт. дис. – М., 1978.

Статистические методы анализа информации в социологических исследованиях / Отв. ред. Г. В. Осипов. – М., 1979.

Степанов Г. В. К проблеме языкового варьирования. – М., 1979.

Структурно-вероятностный анализ языка по данным речи: Всесоюз. сб. Тез. докл. конф. – Харьков, 1979.

Трубецкой Н. С. Основы фонологии. – М., 1960.

Фант Г. М., Якобсон Р., Халле М. Введение в анализ речи. – Новое в лингвистике. М., 1962, вып. 2.

Человек и его работа. – М., 1967.

Чесноков С. В. О представительности результатов социологических исследований, проведенных по малым неслучайным выборкам. – В кн.: Проектирование и организация выборочного социологического исследования. М., 1977.

Швейцер А. Д. Современная социолингвистика. Теория, проблемы, методы. – М., 1976.

Швейцер А. Д., Никольский Л. Б. Введение в социолингвистику. – М., 1978.

Шкаратан О. И., Рукавишников В. О. Социальная структура населения советского города и тенденции ее развития. – Социологические исследования, 1974, № 2.

Шляпентох В. Э. Проблемы достоверности статистической информации в социологических исследованиях. – М., 1973.

Ядов В. А. Методология и процедуры социологических исследований. – Тарту, 1968.

Якубайтис Т. А., Скляревич А. Н. Вероятностные характеристики связных текстов. – Рига, 1978.

Bell R. Language samples.– In: Universals of Human Language: Method and Theory / Ed. by J. H. Greenberg. Stanford, 1978, vol. 1.

93

Bitinas B. Statistiniai metodai pedagogikoje ir psichologijoje. – V., 1974.

Girdenis A. Fonologija. – V., 1981.

Labov W. The Social Stratification of English in New York City. – Columbia University Press, 1966.

Laigonaitė A. Lietuvių akcentologija. – V., 1978.

Statkevičienė J. Vienarūšių ir nevienarūšių pažyminių trukmė. – Garsai, priegaidė, intonacija (Eksperimentinės fonetikos ir kalbos psichologijos kolokviumo medžiaga). V., 1972, t. 5.

Tikimybių teorija ir matematinės statistikos taikymas. – V., 1971.

Weinreich U., Labov W., Herzog M. Empirical foundations for a theory of language change. — In: Direction for Historical Linguistics / Ed. by W. P. Lohman and Y. Malkiel. Austin, 1968.

Zinkevičius Z. Lietuvių dialektologija. – V., 1966.

Zinkevičius Z. Istorinė lietuvių kalbos gramatika. – V., 1981, d. 1.

Приложение 1

Текст № 1

Dabar daug žmonių Vilniuje gyvena naujuose namuose. Aš irgi. Mūsų name net septyni įėjimai po devynis butus. Centras tolokai, bet važinėja troleibusai – nėra ko skųstis. Tiesa, buvo tokių, kurie nenorėjo čia kraustytis, bet dabar apie sugrįžimą net nebekalba. Čia daug geresnė aplinka negu kitur: arti ąžuolynas ir tvenkinys. Ten yra net dvi valtys. Aplink yra nemaža privatininkų sklypų. Sausomis dienomis jie baigia išsemti tvenkinį. Reikalai pagerėja, kai lietus ima pilte pilti. Viena valtis priklauso mūsų klubui. Pernai buvo ją pavogę, bet milicija greit sugrąžino.

Kartą, kasdamas ties vienu sklypu, po ieva radau ąsotėlį. Parašiau skelbimą: „Rastas ąsotėlis. Kreiptis į trisdešimto namo pirmą butą“. „Netrukus net sukaitusi atlėkė tokia Violeta ir ėmė mane pulti dėl to savo ąsotėlio, kad aš jį pagavojau. Tada aš tą biaurybę išvariau. O ji atsiuntė savo šaukiančią auklę. Su aukle išsiaiškinti buvo sunku, nes teko nuo jos skusti kuo greičiau. Tiek triukšmo dėl to ąsotėlio, kurio verte aš labai abejoju. Išbėgęs į kiemą, sutikau sukaitusį Zenį, kuris paklausė, kam erzinu mergaites. Mat Violeta, patempusi apatinę lūpą, atsisėdusi ant rąsto, verkė. Aš norėjau būti tikru vyru ir parodyti jai savo vertę. Priėjau prie jos ir pasakiau: „Negražiai išėjo. Supranti, aš tikrai radau tavo ąsotėlį prie ievos, kai ieškojau sliekų“. Ji nusisypsojo ir davė man obuolį, todėl nebepasirodė biaurybė. Aš pakviečiau ją žaisti tenisą. Viskas būtų buvę gerai, jeigu aš, kasdamas obuolį, nebūčiau prasikandęs lūpos, nes užkliuvau už nusikulusios šaligatvio plytelės. Nesiseka tai nesiseka: tenisą jau žaidė Kęstutis su Džeku, ten eilėje dar laukė Ryžikovas su Džimiu, kuris niekada nesiskiria su savo gitara. Tada nutarėme eiti į filmą „Vyrų vasara“. Namuose dar laukė algebra. Lūpa ištino. Bet man kažkodėl norėjosi šypsotis!..

Текст № 2

vyru – vyrų, duona – duoną, sukaitusį – sukaitusį, gėlė – gėlę, antis – antį, biaurybė – biaurybę, lova – lovą, skysti – skysti, lūpa – lūpą, siūsdamas – siūsdamas, vertė – vertę, lėlė – lėlę, sklypų – sklypų, auklė – auklę, mirti – mirtį, valtis – valtį, spjoviu – spjoviu, kasdamas – kasdamas, skusti – skusti, rastas – rąštas.

Приложение 2

Вопросник полустандартизованного интервью

1. Откуда родом Ваш отец?

2. Откуда родом Ваша мать?

3. Навещали ли Вы те края?

4. Как Вы провели свой последний отпуск? Пришлось ли путешествовать?

5. Случалось ли так, что кто-нибудь узнавал в Вас вильнюсца на основании Вашей речи, произношения?

6. Как Вы считаете, отличается ли речь вильнюсцев от литературной? (Если да, то чем именно).

7. Кто вообще разговаривает на литературном языке?

94

8. Следите ли Вы за правильностью своей речи?

9. Прочитайте, пожалуйста, этот текст (№ 1). Представьте себе, что читаете его в кругу семьи или друзей.

10. Как часто и где Вам приходится пользоваться русским языком находясь в г. Вильнюсе?

11. Владеете ли Вы польским языком?

12. (Если ответ № 11 положительный.) Где и когда Вы научились говорить по-польски?

13. Прочитайте, пожалуйста, еще один текст (№ 2).

14. Будьте добры заполнить анкету (см. прил. 3).

Приложение 3

Часть социолингвистической анкеты

(Вопросы о реализации долгих и кратких гласных)

VII. snigo — snygo; pilti — pylti; plitelė — plytelė; šluostiti — šluostyti; dvi valtis — dvi valtys; dvi antis — dvi antys; dvi ausis — dvi ausys; krausiti — kraustyti; gyventi — gyventi; kitas — kytas; viriausias — vyr iausias; visas — vysas; džiaustiti — džiaustyti; skirti — skyrti; Vilnius — Vyln ius.

IX. tavo — tavo; savo — savo; geresnė — gerešnė; baltasis — baltašis; žalesnis — žalešnis; mesti — mešti; vesti — vešti; lešti — lešti; paima — paima; atneša — atneša.

nekasa — nėkasa; nebekalba — nebeḳalba.

X. drasuolis — drąsuolis; žaselė — žąselė; Kęstutis — Kęstutis; asotelis — ašelėlis; aselė — ašelė; ažuolynas — ąžuolynas; siustuvas — siųstuvas; tesinys — tęsinys; trešimas — tręšimas; sugrįžimas — sugrįžimas; sugrąžino — sugrąžino; grėžinys — gręžinys.

XI. ūsuotas — usuotas; rūkyti — rukyti; ilgūmas — ilgumas; pūlti — pulti; pabūsti — pabusti; kūrmis — kurmis; grūdai — grudai; būvo — buvo; pūmpuras — pumpuras.

Приложение 4

I. Список квазимонимов для аудиотестов

Эксперимент № 1: rąstas — rastas, rąsto — rasto, rąsta — rasta, rastu — rastu.

Эксперимент № 2: kąstas — kastas, kąsto — kasto, nekąsiu — nekasiu, kąstą — kastą, kąsti — kasti.

Эксперимент № 3: kąsdamas — kasdamas, kąsdami — kasdami, kąsdama — kasdama, kąsdamos — kasdamos.

Эксперимент № 4: skųsdamas — skusdamas, siųsdamas — siusdamas, skųsti — skusti, siųsti — siusti.

Эксперимент № 5: mirti — mirti, Vyti — vyti, skilti — skilti, karti — karti, skysti — skysti.

Эксперимент № 6: kartys — kartis, antys — antis, mentys — mentis, paltys — paltis, kandys — kandis.

Эксперимент № 7: pasibaigusių — pasibaigusi, pylusi — pylusi, dainavusi — dainavusi, sv ar sčius i — svaršč iusi, sukaitusi — sukaitusi.

95

Эксперимент № 8: ąsoty – ąsoti!, šalty – Šalti!, medy – medi!, uosy – Uosi!

Эксперимент № 9: verte – verte, silke – silke, lėlė – lėlė, auklė – auklė, biaurybė – biaurybė.

Эксперимент № 10: iškaštų – iškaštų, iškaštam – iškaštam, pakaštais – pakaštais, užkaštuos – užkaštuos.

Эксперимент № 11: kopą – kopą, lūpą – lūpą, duoną – duoną, kiaurą – kiaurą, lovą – lovą.

Эксперимент № 12: vyrų – vyrų, sūrių – sūriu, mūrų – mūru, spīvių – spīviu, sklypų – sklypu.

Эксперимент № 13: salą – salą, balą – balą, rastu – rastų, prakastu – prakastų.

Таблица 4.2 Результаты аудиотестирования

Слог | Гласный | Тип, № | Эксперимент, № | n |% | P Ударный | a | I | 1. | 180 | 85 | 0,999 Ударный | a | I | 2. | 200 | 74,5 | 0,95 Ударный | a | I | 3. | 180 | 77,8 | 0,99 Ударный | u | II | 4. | 200 | 81 | 0,999 Безударный | i | III | 5. | 200 | 80 | 0,999 Безударный | i | III | 6. | 180 | 74,4 | 0,95 Безударный | i | III | 7. | 200 | 75 | 0,95 Безударный | i | III | 8. | 200 | 89 | 0,999 Безударный | e | IV | 9. | 180 | 88,8 | 0,999 Безударный | a | V | 10. | 160 | 78,8 | 0,99 Безударный | a | V | 11. | 200 | 76 | 0,99 Безударный | u | VI | 12. | 180 | 91,7 | 0,999 Безударный | a/u | I/VI, V | 13. | 200 | 75 | 0,95

n – суммарное число ответов;% – процент опознанных оппозиций, P – вероятность различения оппозиций.

Таблица 4.3 Статистическая оценка различий абсолютной продолжительности обоих членов оппозиции V̄/V̄

Эксперименты, № | x̄V̄, мс | ȳV̄, мс | s̄x | s̄y | t | tα (P=95%) 1, 2, 3, 13 | 146 | 139 | 33 | 36 | 1,18 | 1,98 4 | 111 | 96 | 45 | 36 | 1,02 | 2,04 5, 6, 7, 8 | 40 | 38 | 13 | 21 | 0,67 | 1,98 9 | 59 | 64 | 27 | 24 | 0,59 | 2,04 10, 11, 13 | 60 | 58 | 35 | 25 | 0,29 | 1,99 12, 13 | 70 | 56 | 38 | 28 | 1,48 | 2,01

x̄V̄ = 81 ȳV̄ = 75

96

Приложение 5

Результаты социолингвистического интервью

Рис. 5.1. Распределение долгого варианта (неуд. V̄C) с учетом факторов возраста (A), контекстуального стиля (B) и уровня образования (C),%

Рис. 5.2. Распределение краткого варианта (уд. V̄#) с учетом факторов возраста (A), контекстуального стиля (B) и уровня образования (C),%

Рис. 5.3. Распределение краткого варианта (уд. V̄C) с учетом факторов возраста (A), контекстуального стиля (B) и уровня образования (C),%

7. 949

97

Приложение 6

Результаты социолингвистической анкеты

Рис. 6.1. Распределение долгого варианта (неуд. V#) по отношению к возрастным (A₁, A₂, A₃) и образовательным (C₁, C₂, C₃) группам в словах valtys (1), antys (2) и ausys (3),%

Рис. 6.2. Распределение долгого варианта (неуд. УС), реализуемого в абсолютном начале слова, по отношению к возрастным (A₁, A₂, A₃) и образовательным (C₁, C₂, C₃) группам в словах ąsotelis (1), ąselė (2), ąžuolynas (3) и išiuotas (4),%

Рис. 6.3. Распределение долгого варианта (неуд. VC), реализуемой во II и далее предударных слогах, по отношению к возрастным (A₁, A₂, A₃) и образовательным (C₁, C₂, C₃) группам в словах tęsiny s (1) и greižynės (2),%

98

Рис. 6.4. Распределение долгого варианта (неуд. УС), реализуемой в I предударном слоге перед долгим ударным слогом, по отношению к возрастным (А₁, А₂, А₃) и образовательным (С₁, С₂, С₃) группам в словах phytele (1), gyventi (2), vyriausias (3), drąsulis (4), žąselė (5), rūkyti (6) и grūdai (7),%

Рис. 6.5. Распределение долгого варианта (неуд. УС), реализуемой в I предударном слоге перед кратким ударным слогом, по отношению к возрастным (А₁, А₂, А₃) и образовательным (С₁, С₂, С₃) группам в словах sugrįžimas (1), Kęstutis (2), trešimas (3), sugrąžino (4) и siūtuvas (5),%

Рис. 6.6. Распределение долгого варианта (неуд. УС), реализуемой в последуарном слоге, по отношению к возрастным (А₁, А₂, А₃) и образовательным (С₁, С₂, С₃) группам в словах šluostyti (1), kraustyti (2) и džiaustyti (3),%

7*

99

Рис. 6.7. Распределение краевого варианта (уд. VC), реализуемой в абсолютном начале слова atneša (1) и перед слогом с долгим гласным в словах geresnė (2), nebekalba (3), tavo (4), savo (5), smigo (6), buvo (7), по отношению к возрастным (A₁, A₂, A₃) и образовательным (C₁, C₂, C₃) группам,%.

Рис. 6.8. Распределение краткого варианта (уд. VC), где V̄ — i, реализуемой перед слогом с кратким гласным, по отношению к возрастным (A₁, A₂, A₃) и образовательным (C₁, C₂, C₃) группам в словах kitas (1), visas (2), pilti (3), skirti (4) Vilnius (5),%

Рис. 6.9. Распределение краткого варианта (уд. VC), где V̄ — u, реализуемой перед слогом с кратким гласным, по отношению к возрастным (A₁, A₂, A₃) и образовательным (C₁, C₂, C₃) группам в словах ilgumas (1), pabusti (2), pulti (3), kurtis (4), purpuras (5),%

100

Рис. 6.10. Распределение краткого варианта (уд. VČ), где V̇ — а, е, реализуемой перед слогом с кратким гласным, по отношению к возрастным (A₁, A₂, A₃) и образовательным (C₁, C₂, C₃) группам в словах žalesnis (1), mesti, vesti, leisti (2), nekasa (3), baltais (4) и paima (5),%

L. GRUMADIENĖ

SOCIOLINGVISTINIS VILNIEČIŲ LIETUVIŲ KALBOS TYRIMAS

Reziumė

Straipsnyje analizuojamos kelios sociolingvistiniais metodais tirtos vilniečių lietuvių kalbos vokalizmo ypatybės, ieškoma jų įvairavimo priežasčių.

Nustatyta, kad vilniečių kalbai yra būdinga tendencija trumpinti nekirčiuotus ilguosius balsius ir ilginti kirčiuotus trumpuosius balsius.

Beveik visi vilniečių lietuvių kalboje vartojami balsių variantai turi atitikmenis ir tarmėse. Tik viena ypatybė (kirčiuotų galūninių trumpųjų ilginimas) tėra būdinga kai kurių dvikalbių paribio tarmių lietuvių kalbai.

Statistiniais metodais įrodyta, kad vilniečių kalbos nevienalytiškumui turi įtakos tiek lingvistiniai, tiek socialiniai faktoriai. Pvz., vieni balsių variantai žodyje sąlygoja kitų balsių variantų vartojimą, bet kitoje sociolingvistinėje situacijoje ištartas tas pats žodis jau gali iš esmės skirtis. Ieškant vilniečių lietuvių kalbos dėsningumų, buvo pasinaudota statistiniais metodais ir įrodyta, kad kalbos įvairavimui didelę įtaką turi tėvų tarmės ir sąmoninga vertybinė kalbos orientacija, taip pat amžius ir išsilavinimas.

L. GRUMADIENĖ

A SOCIOLINGUISTIC STUDY OF LITHUANIAN SPOKEN IN VILNIUS

Summary

The sociolinguistic variances of Lithuanian spoken in Vilnius were studied. Special attention was paid to the vocalism. Over 70 informants were examined and their speech was thoroughly investigated using a complex of methods and technique. The average value of sound frequencies was calculated and estimated in order to evaluate the role of some social and linguistic factors. 13 graphs and 2 tables are introduced to simplify the discussion. The existence of some systematically appearing sociolinguistic variances and their close connection with sociolinguistic norms have been established.

It might be concluded from the data obtained that the great variance of the urban language is due to the influence of the social factors, and that the main linguistic characteristics of Lithuanian spoken in Vilnius are similar to those known in the Lithuanian dialects.