Chapter 2 Literature Review
2.2 Linguistic Variation within Context
2.2.4 Language attitudes
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
expectation, the youngest group does not score the lowest in this vowel centralization process. Opposite to this finding, Labov (1994) proposes that older people would slightly be affected by sound change; while the younger ones are more easily be affected.
Cukor-Avila (2000) in his diachronic study of the use of zero third person
singular, zero copula, habitual be+v+ing, and had+past used as a simple past in a rural Texas community of Springville where people used AAVE, four subjects of different ages were interviewed twice in an interval of seven year. The two older subjects show no significant change, but the 9-year-old subject changes significantly, and the
6-year-old subject’s use of zero copula decreases in the second interview.
Cukor-Avila’s research shows that the older subjects change slightly, while the younger subjects show more obvious change. Chambers (2002), in his study of speakers use of [w], not [hw], in words like which and whine, in central Canada, finds that the younger groups give higher percentages of using the new form [w], less by older age group over 80 and the group between 70-79. Chamber’s findings support Cukor-Avila’s (2000) study. In this thesis, Hakka people of different generations will be examined in order to check whether age has any impacts to the phonetic realization of Mandarin /ɕ/.
2.2.4 Language attitudes
According to Bailey (1997), the course of sound change is influenced by the attitudes of the speech community toward that change. Before this statement is examined, it is important to know what it means by language attitude.
Two theories about the nature of attitudes compete with each other–the mentalist’s view vs. the behaviorist view. First, as Fasold (1987) points out, most language attitude work is based on a mentalist view, taking attitude as a state of readiness (i.e. an interviewing variable between a stimulus and the response to that
17
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
stimulus); while the other view of attitude, the behaviorist view, claims that attitudes can be found in people’s responses to social situation. Moreover, social psychologists, taking the behaviorist definition of attitude, consider attitude as single units. Whereas, mentalists claim attitudes have three components; cognitive component, conative component, and affective component, with these three components referring to knowledge, action, and feeling, respectively.
Language attitudes may extend to people’s attitudes toward a language, which is a reflection of people’s attitudes toward the speakers of that language. The study of language attitude can be applied to explain people’s motivations of using a specific language or a specific form in a language. For example, in Labov’s study (1966) on Martha’s Vineyard, the local residents’ use of the centralized vowels and their positive attitude toward vowel centralization are to mark local identity. Hoover (1978) reports that black Americans’ choice of Black English Vernacular (BEV) over standard English corresponds to their positive attitudes toward BEV, their mother tongue, which is used to symbolize their ethnic identity. Based on this conception, in this thesis, the relationship between Hakka speakers’ attitudes toward the Hakka ethnic group and their use of [s] as a phonetic manifestation of Mandarin /ɕ/ is examined.
18
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Chapter Three Methodology
In order to identify the patterns, and the linguistic and nonlinguistic causes of this phonetic variation of Mandarin /ɕ/, both quantitative analysis and qualitative analysis were conducted.
3.1 Design of Quantitative Analysis
In quantitative analysis, the impact of both linguistic and non-linguistic factors were examined. The former includes neighboring sounds within/beyond syllable boundary, syllable structure, and lexical frequency. The latter includes situational formality, geographical factor, and three social factors.
3.1.1 Linguistic factors 3.1.1.1 neighboring sounds
It was hypothesized that neighboring sounds would affect the phonetic
realization of /ɕ/. In order to describe the sounds that precede and those that follow /ɕ/, a conflation of those neighboring sounds of /ɕ/ is given below.
19
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
ɯ i u ia
o i ɛ y iou
e ## ɕ y (N) I y ɛ
a n ŋ
Within a syllable, Mandarin /ɕ/ is followed by a vowel. Since within a syllable, /ɕ/, except when followed by a monothong /i/, is always followed by an on-glide /i/
before the nucleus vowel, and since /y/, as a possible vowel following /ɕ/, is with roundedness, which is phonosemantically impossible to induce [s] without causing meaning problems, the impacts of the following vowels were excluded from this study.
For these reasons, in this study, neighboring vowels refer to the vowels in the open syllable preceding /ɕ/, and what was examined is whether the advancement of the preceding vowels is influential to the emergence of [s] through assimilation.
3.1.1.2 syllable structures
Syllable structure was also predicted to be related with phonetic variation. In Mandarin Chinese, all of the syllable structures containing /ɕ/ can be reduced into one syllable frame: /ɕ/ + V(N). In other words, what was examined is whether the syllable final nasal is influential to the emergence of [s] in the initial position of the same syllable. Table 1 shows the syllable frames and the individual syllable structures examined in this study.
20
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 1. Syllable frames and syllable structures containing /ɕ/
3.1.1.3 character frequency
It was anticipated that, the higher the lexical frequency is, the higher the frequency of the non-standard form [s] is. Therefore, in this study, both high frequency characters and low frequency characters3 are tested for the impact of character frequency on the presence of [s]. Although the character frequencies used in the study are based on written form, instead of spoken form, they are still used as a criterion to choose among characters for testing. Moreover, for convenience of contrast, the characters chosen to be tested in this study are from the two poles of the scale of character frequency. That is, for each syllable structure, two characters of the highest frequency and two characters of the lowest frequency were chosen and tested.
These characters chosen were tested in all three speech styles (namely, conversation, passage reading, and characters reading). In total, there are 17 high frequency
characters and 18 low frequency characters tested in this study.4 Table 2 presents the
3 The frequencies of these target characters are derived from Academic Sinica Balanced Corpus of Modern Chinese.
4 For syllable /ɕiou/, only one high frequency character is tested, because all of the other characters of this syllable structure, according to Academic Sinica Balanced Corpus of Modern Chinese, are of very
Syllable frame
Syllable structure
/ɕ/ +V
/ɕi/ /ɕiɛ//ɕia/
/ɕiau/
/ɕiou/
/ɕ/ +V +N
/ɕin/ /ɕi ŋ//ɕiɛn/
/ɕia ŋ/
21
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
related syllable structures, characters chosen for the nine syllable structures, and their character frequencies.
Table 2. Syllable structures and character frequency of / ɕ/ in the survey (SF=Syllable Frames, SS= Syllable Structures; numbers in parentheses are character frequencies).
low frequency.
Character frequency
SF SS
High Frequency Low Frequency
/ɕ/ +V
/ɕi/ 西(386) 戲(274) 息(35) 惜(20) /ɕiɛ/ 寫(1373) 些(2536) 謝(73) 洩(15) /ɕia/ 下(5000) 夏(151) 蝦(37) 瞎(13) /ɕiau/ 笑(774) 小(5000) 效(35) 孝(24) /ɕiou/ 修(319) ___ 休(29) 羞(17)
/ɕ/ +V +N
/ɕin/ 心(2598) 信(563) 辛(5) 星(123) /ɕi ŋ/ 行(945) 性(217) 幸(49) 形(83) /ɕiɛn/ 先(2500) 現(277) 險(54) 獻(21) /ɕia ŋ/ 向(3978) 像(4412) 享(55) 詳(27)
22
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
3.1.2 Non-linguistic factors
Nonlinguistic factors are also expected to cast impacts on the phonetic variation of /ɕ/. In this study, non-linguistic factors include situational formality, geographical variable, age, gender, and education level.
3.1.2.1 situational formality
Labov (1972), in his study of the stylistic variation and the social class variation of [t] and [ɵ] of /ɵ/ in New York City, discovered that speakers use more nonstandard variants in casual speech than in careful speech. In this thesis, three situational formalities (namely, conversation, reading passage, and reading characters) were examined in order to investigate whether the emergence of [s] is stylistically evoked.
It was anticipated that [s]% would increase with the decrease of situational formality, with [s]% being the highest in conversation (a style of lowest formality), lower in reading passages (a style of intermediate formality), and the lowest in reading characters (a style of highest formality).
3.1.2.2 nature of the geographical area
All of the subjects of this study are native speakers of Hakka, and they are equally distributed to two cities—Taoyuan City and Chungli City. The former is the administration center of Taoyuan County and a city of commerce; the latter is a city of manufacturing industry. It was predicted that Chungli City, socioeconomically inferior to Taoyuan City, would yield higher percentages of [s].
3.1.2.3 social factors
In this study, the influences of age, gender, and education level are examined.
Age is divided into two groups, the old and the young; gender is categorized into male and female; education level is divided into higher education level and lower education level. The influences of these social/geographical factors were examined individually first, and then in combination. Since the sample size is small (only 32 subjects), the
23
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
interactions among the non-linguistic factors were forced to limit to those by any two of the four external factors.
3.2 The Tools
Three tools were adopted for quantitative survey of the use of [s] in three styles.
3.2.1 Conversation
Labov (1984) offers field methods for investigation of linguistic change and variation. For face-to-face conversation, Labov proposes the concept of module, which includes a group of questions focusing on a particular topic, to make the
face-to-face communication in a research as real as people in daily life. For this study, five major topics are included in the conversations: (1) personality and appearance, (2) education, (3) food safety, (4) economy, and (5) life style. Under these 5 topics, there are also some subtopics and subsubtopics to extend these topics into a natural
conversation. In this way, if subjects can not be elicited by target words in questions, subquestions are offered to provide more chances to secure enough quantity of
subjects’ production of the target characters. All of the target characters were designed to appear in a conversation for at least three times. In total, twenty-three questions were projected through the concept of module to link with each other.
Also, each conversation took about 50 minutes. During the conversations, in order to make them feel safe and comfortable to render natural data, the subjects were told that the purpose of the survey was not to check on their thinking or to probe for their privacy.
All of the conversations were recorded with a digital recorder and transcribed for further analyses. For details of the conversation topics and sub-topics, see
Appendix 1 (on p.95 of this thesis). In the appendix, those characters underlined are the target characters in this study.
24
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
3.2.2 Reading passages
In reading passages, subjects were asked to read 4 passages, which include popular issues in newspapers and magazines.5 To cope with cross-style comparison, the topics for passage reading, like those for conversation, are also about (1)
personality and filial obedience, (2) education, (3) food safety, and (4) fee of
electricity in summer. In passage reading, each target character appears 3 times in the passages. In total, 105 instances of the target words were examined. Again, the data were recorded and transcribed for analyses. For contents of these passages, see Appendix 2 (on p.100 of this thesis).
3.2.3 Reading characters
In reading characters, 35 target characters and 15 decoys were included in the test. Subjects were asked to read each word one by one. For details of these characters, see Appendix 3 (on p.101 of this thesis). In the table, those characters in shade are the targets, all of which were recorded and transcribed for analysis.
3.3 Design of Qualitative Analysis
After the quantitative tests, face-to-face interviews were conducted to elicit the subjects’ language attitudes, since quantitative study can only describe the patterns of linguistic variation, and it is qualitative analysis that would offer explanations to the causes of the patterns.
The qualitative design implemented in this study includes the following parts:
the subjects’ stereotypical impression of Hakka people, subjects’ subjective judgments of the target phonetic variation, their opinions toward the necessity of the association between ethnic identity, their proficiency in Hakka dialect, and their frequency of using Hakka dialect, and so on. For details of the interview questions, see Appendix 5
5 These passages were adapted from the newspaper Apple Daily and the magazine Common Wealth.
25
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
(on p.103 of this thesis).
Among the attitude tests, subjects’ attitudes toward Hakka dialect were
expected to offer explanations to the distribution of the target phonetic variation. Ten pairs of attributes on value dimension were given on 7-point semantic differential scales were used for the subjects to rate their stereotypical impression on the Hakka dialect. For details of these attributes, see Appendix 4 (p.102 of this thesis).
3.4 Measurements
In quantitative analysis, both percentages and frequencies are presented.
However, since the sample size is too small for statistic tests, only patterns and regularities were described for cross comparisons among internal factors and those among external factors.
In qualitative analysis, statistical tests were also expected to be applied to examine the patterns found. In the analysis of the interviews, subjects’answers to the causes of their subjective judgments were categorized first, and then analyzed and discussed.
3.5 Sampling
3.5.1 Source of subjects
Milroy& Milroy (1978) used the way of “a friend of a friend” to enter a
community. Labov (1963), in his work of vowel centralization on Martha’s Vineyard, chose locally born and raised adults and teenagers. In this study, subjects recruited are all locally born and raised either in Chungli City or in Taoyuan City. All of the
subjects are either friends and relatives of this researcher’s or informants recruited by the way of “a friend of a friend,” starting from this researcher’s classmates in high school and in college, extending to their friends and relatives.
26
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
or the qualitative interviews, only 29 of the thirty-two subjects for the quantitative tests were interviewed; the other three went abroad and could not be reached after the quantitative tests.
3.5.2 Social distribution of the subjects
In this survey, 32 subjects were recruited equally distributed to 2 geographical areas (Chungli City vs. Taoyuan City), 2 genders ( male vs. female), 2 age groups (the old vs.the young), and 2 education levels (higher education level vs. lower education level).That is, half of the subjects ranging between 20 and 40 years old were grouped into one age group, while the other half were between 45 to 60 years old. Moreover, half of the subjects, who have received college degree, were classified as higher education levels, and the other half, without college degree, were grouped into lower education level. The social distribution of these subjects is presented in Table 3.
Table 3: Social distributions of the subjects (for genders, M=Male, F=Female; for education levels, HE=higher education level, LE=lower education level) Geographical
Location
Age Gender Education Level
Taoyuan City
Chungli City
Total 1 Total 2 Total 3
Old
M HE 2 2 4
8
16
LE 2 2 4
F HE 2 2 4
8
LE 2 2 4
Young
M HE 2 2 4
8
16
LE 2 2 4
F HE 2 2 4
8
LE 2 2 4
Total 16 16 32 32 32
27
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
3.6 The Procedure of Data Collection
In this study, each subject was tested twice on separate days, with quantitative tests coming prior to qualitative.
Also, stylistic tests start with a conversation of casual style, through a passage reading of intermediate formality, and to a character reading of high formality.
Afterwards, 29 of the 32 subjects were interviewed for their language attitudes, starting with an evaluation of their stereotypical impression toward the Hakka dialect, moving on to their descriptions and explanations for their subjective judgments on the Hakka people and the target phonetic variant, and their opinions about the necessity of the association between Hakka ethnic identity, and their Hakka proficiency and their frequency of using Hakka.
28
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Chapter Four Data Analysis
In this chapter, the phonetic variation of Mandarin /ɕ/ by linguistic and non-linguistic factors are described and analyzed.
4.1 General distributions of the Two Variants of / ɕ/
Table 4 shows the distribution of the two variants of /ɕ/ in general.
Table 4. General distributions of the two variants of /ɕ/
(numbers in parentheses are frequencies)
According to Table 4, the nonstandard form [s] takes nearly one-fifth of the subjects’ phonetic realization of /ɕ/.
4.1.1 Distributions of the two variants of / ɕ/ by formality
It was predicted that the use of the nonstandard form [s] would decrease with the increase of formality. Table 5 shows the distribution of [s] in the three different styles:
conversation (CON, hereafter in this chapter), reading passages (RP, hereafter in this chapter), and reading characters (RC, hereafter in this chapter).
Linguistic Variants
% (freq)
[
ɕ]
[s] TotalAll 81.0% (5860) 19.0% (1384) 100% (7244)
29
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 5. Distributions of the two variants of /ɕ/ by formality (numbers in parentheses are frequencies)
According to Table 5, the subjects’ use of [s], as predicted, decreases as formality increase. To be specific, [s]% is the highest in CON (27.0%), which is of lowest formality; in contrast, when the formality increase s in RP and in RC, [s]% decreases nearly 12.0% and 16.0%, respectively. In other words, although the gap between the [s]% in CON is not particular large, it seems to be true that formality is as influential factor to subjects’ use of the non-standard form [s].
4.1.2 D
istributions of the two variants of /ɕ/ by social/geographical factors It was predicted that the subjects’ use of [s] is controlled by geographical features of the place in which they live and their social characteristics.Variants of / ɕ/
Formality [
ɕ]
[s] TotalConversation 73.0% (2020) 27.0% (744) 100% (2764) Reading passages 84.9% (2850) 15.1% (510) 100% (3360) Reading characters 88.6% (990) 11.4% (130) 100% (1120)
30
‧
Table 6. Distributions of the two variants of /ɕ/ by social/geographical factors (numbers in parentheses are frequencies)
Linguistic Variants Social/
Geographical factors
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
According to Table 6, subjects of different social/geographical backgrounds show different patterns in using [s]. First, when geographical area is taken into consideration, subjects of Chungli City use [s] significantly more often than those of Taoyuan City (30.7% vs. 7.7%). This result seems to be derived from the causes that Chungli City, in comparison with Taoyuan City, is a place with larger population size and higher population density of Hakka people, which would invite its Hakka
residents to use their mother tongue more frequently, which in turn would offer a higher possibility for Hakka dialect to intrude and influence Hakka people’s use of Mandarin. Moreover, as described in Chapter One, since Chungli City is a city of manufacturing industry, it is more likely that its residents take related occupations in this vocational field, which demands more frequent use of the Low Language, Hakka, instead of the High Language, Mandarin, which in turn may leave traces of Hakka dialect in those Hakka people’s use of Mandarin. In contrast, Taoyuan City, is the administration center of Taoyuan County and a place of service and financial industry, which would invite more use of Mandarin and less use of Hakka dialect, and thus leave less Hakka trace in their Mandarin performance.
Next, when comparing the two age groups’ use of [s], the older group scores significantly higher [s]% (31.6%) than the younger group (4.9%). This phenomenon may be because those older subjects use Hakka as their mother tongue, but younger subjects use Mandarin Chinese as their major communication tool in most domains of communication. Besides, older age group for example, may hold more positive language attitude than the younger ones. These younger subjects may use Mandarin Chinese frequently in school, and thus their frequency of using Hakka decreases; and their low frequencies of using Hakka may induce lower competence of Hakka, therefore, [s]%, as a trace of Hakka competence, is decreased.
32
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Unlike , the influences of geographical area and age, gender shows only moderate impact on the subjects’ use of [s], with the males, against this researcher’s expectation, using less [s] than the females (14.4% vs. 23.6%, respectively). This result may be because most Hakka females in these cities work in their hometown, where they have more chances to use Hakka, but males usually work in big cities, such as Taipei, in which, Mandarin Chinese is the major language. Therefore, more Hakka traces in Mandarin are left by females than by males.
As for the effect of the subjects’ education level, the two education groups resemble each other with almost identical [s]% (19.1% of HE and 19.0% of LE). In other words, education level is not a valid social factor to the subjects’ use of [s]. A possible reason why the effect of education level is blurry is due to the fact that many of the lower education subjects recruited for this study are with high school degree, and only a few of them are from education level lower than high school. Besides, since colleges and universities are excessive, with many of them offering academic trainings not too different from that in high school in Taiwan, it is hard to ensure that subjects with bachelor degree actually behave differently from those with lower degrees.
4.1.3 Distributions of the two variants of / ɕ/ by social/geographical factors and formality
It was anticipated that the subjects of different social/geographical backgrounds would still comply to the demands of situational formality, with the use of [s]
It was anticipated that the subjects of different social/geographical backgrounds would still comply to the demands of situational formality, with the use of [s]