Journal of Language, Identity & Education Gender ...

7 downloads 0 Views 193KB Size Report
Feb 16, 2010 - a Duquesne University, b University of Texas at Austin, ... patterns of Kurdish-origin students learning Turkish in Turkey based on students' identification .... range of different types of transactions with different individuals, of the ...
This article was downloaded by: [Polat, Nihat] On: 16 February 2010 Access details: Access Details: [subscription number 919307684] Publisher Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 3741 Mortimer Street, London W1T 3JH, UK

Journal of Language, Identity & Education

Publication details, including instructions for authors and subscription information: http://www.informaworld.com/smpp/title~content=t775653670

Gender Differences in Identity and Acculturation Patterns and L2 Accent Attainment Nihat Polat a; Laura J. Mahalingappa b a Duquesne University, b University of Texas at Austin, Online publication date: 16 February 2010

To cite this Article Polat, Nihat and Mahalingappa, Laura J.(2010) 'Gender Differences in Identity and Acculturation

Patterns and L2 Accent Attainment', Journal of Language, Identity & Education, 9: 1, 17 — 35 To link to this Article: DOI: 10.1080/15348450903476832 URL: http://dx.doi.org/10.1080/15348450903476832

PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

Journal of Language, Identity, and Education, 9: 17–35, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1534-8458 print / 1532-7701 online DOI: 10.1080/15348450903476832

Gender Differences in Identity and Acculturation Patterns and L2 Accent Attainment

1532-7701 1534-8458 HLIE Journal of Language, Identity, and Education, Education Vol. 9, No. 1, Dec 2009: pp. 0–0

Gender Polat and and Mahalingappa L2 Accent Attainment

Nihat Polat Duquesne University

Laura J. Mahalingappa Downloaded By: [Polat, Nihat] At: 16:35 16 February 2010

University of Texas at Austin

Addressing the influence of sociocultural theory, current views of second language acquisition situate language learning in a much broader context than the isolated box of the classroom. There is need to consider second language (L2) acquisition practices more broadly. This study addresses differences between girls and boys of Kurdish ethnic background as they acquired Turkish, the language of the dominant society. Data were collected from 56 girls and 65 boys at middle and high schools with variables including identification with the first and second language communities, acculturation and integration patterns via an investigation of social networks, and the attainment of a native-like Turkish accent. Quantitative analyses were conducted using multiple chi-square statistics, ANOVA, and MANOVA. Findings suggest that girls obtain much higher accent ratings than boys and that girls and boys are significantly different regarding their identification with the L2 community and their family and nonfamily acculturation and integration patterns. Key words: gender, L2 accent attainment, identity, social networks, acculturation, Kurdish

The reconceptualization of second language (L2) acquisition research from the sociocultural perspective has led to the idea that L2 attainment is a sociocultural process through which learners acquire a language through complex social interactions and power relations that engage the identities of L2 learners in many ways (Lantolf, 2000; Norton Peirce, 1995). Thus, when examining learners’ success in L2 attainment, it is important to consider many aspects of learners’ identities, the L2 learning setting, and the way that learners interact with various speakers in the language learning process. According to Schumann (1978), the degree of an individual learner’s success in L2 attainment depends in part upon the learner’s identification and degree of acculturation with the target language community. However, thus far, research has not adequately explored how gender, an important aspect of identity, in interaction with identification and acculturation, relates to L2 attainment. Situated in a unique setting, the current study examines gender differences in L2 attainment patterns of Kurdish-origin students learning Turkish in Turkey based on students’ identification Correspondence should be sent to Nihat Polat, Duquesne University, Department of Instruction and Leadership, 319 Fisher Hall, 600 Forbes Avenue, Pittsburgh, PA 15282. E-mail: [email protected]

18

POLAT AND MAHALINGAPPA

and acculturation patterns. We address similarities and differences in the success of Kurdish girls and boys in attaining a native-like Turkish accent and how their different rates of success relate to (1) their identification patterns with their ethnic L1 Kurdish and dominant L2 Turkishlanguage communities and (2) their degree of acculturation explored through an analysis of their social networks, investigating the differences in the content and nature of both family and nonfamily exchange, interactive, and passive networks.

LITERATURE REVIEW

Downloaded By: [Polat, Nihat] At: 16:35 16 February 2010

Language and Gender, Identity, and SLA Research on language and gender has recently undergone some noteworthy theoretical shifts from envisioning gender as an individual variable that can be generalized to viewing it as a less generalizable culture-specific phenomenon that is socially constructed (Cameron, 1990; Eckert & McConnell-Ginet, 2003; Norton & Toohey, 2004; Willett, 1995). A large volume of research exists on gender differences in language use, and although results have been inconclusive, some relationship between gender and language variation has been reported to exist, suggesting many context-dependent reasons why men and women may use various language features differently. Situated within a sociocultural framework, gender may have an impact on learners’ ethnic identification and acculturation patterns, thereby affecting both their L1 and L2 development. As understood in this study, based on poststructuralist and sociocultural theories, identity concerns both the individual and social levels and is diverse, complex, dynamic, and fluid (Hall, 1996). Attempts to define identity are generally centered around self and coherent manifestations of self in broader social settings (Bruner, 1990). Taylor (1998) uses Bakhtin’s notion of the dialogical self to represent the self as constructed within social interactions with others. Language is a way to construct identities as well as transmit knowledge and organize and impose class structures (Bourdieu, 1991). Diverging from research that maintained female superiority in learning languages (Ehrman & Oxford, 1990; Ellis, 1994), recent studies in SLA have been inconclusive as to whether women learn a second language better than men (Ehrlich, 1997). Nevertheless, several studies have found that men and women apply different learning styles and strategies in learning foreign/ second languages (Oxford, 1994). Several studies have found that women outperformed men in learning a second language because they had different motivations and attitudes about the target language (Gardner & Lambert, 1972; Spolsky, 1989). In interpreting these findings, Ellis (1994) suggests that when girls outperform boys in L2 learning, it may be because women deal with the inherent threat imposed on their identities more readily than men. Gendered differences in language use, which has been seen as pervasive throughout different settings, may stem from early socialization patterns, when boys and girls imitate adult gender roles and experiment with adult speech styles (Luykx, 2003; Collins, 2000). Jule (2005) addresses how language is used to construct gender identities, which may lead to differences in L2 attainment. Recently, Connell (2002) has argued that there is also a shift toward feminist-critical paradigms (Norton & Toohey, 2004) and poststructuralist approaches (Fairclough, 2001) that are centered on gendered socialization patterns shaped by ethnicity and class-related power relations that limit access to the context of language use and thus comprehensible input opportunities for L2 women

Downloaded By: [Polat, Nihat] At: 16:35 16 February 2010

GENDER AND L2 ACCENT ATTAINMENT

19

learners (Hill, 1987; Norton, 2000; Norton Peirce, 1995). Women’s access to an L2 may be restricted because, traditionally, the learning of an L2 was prerequisite for men to take care of their families, whereas the designated area of work for women was the home (Rockhill, 1993); that restriction may lead women to perceive L2 learning as unimportant or not useful. In addition to issues of access, in many situations, women have been found to show greater loyalty to their native ethnic language and community (Phinney, 1990; Ullah, 1985; Zentella, 1987). Finally, unlike studies that show that women have greater loyalty to their L1, some have suggested that in many situations, women tend to adhere to more standard, prestige forms than men, who alternately use more vernacular language. Gal (1978) suggested that Hungarian women in Austria chose to use the more prestigious German (the L2) more than men for social advancement, distancing themselves from the symbolic value of peasant status attached to their native Hungarian. Alternately, Trudgill (1972) used the idea of covert prestige to explain why men may have the tendency to use more vernacular (nonstandard) forms. Covert prestige refers to speakers’ positive attitudes toward the vernacular, suggesting that “there are hidden values associated with non-standard speech” (Trudgill, 1972, p. 23). Nonstandard forms are often associated with masculinity, thus, they have value for men. In many multilingual situations, the native, ethnic language often carries covert prestige as a symbol of ethnic identity, if not as a means of economic or social advancement. However, as pointed out by Eckert and McConnell-Ginet (2003), the relationship between gender and language is one that is difficult to unravel because gender cannot be easily separated from other aspects of social class and identity, and the relationship will likely vary across different social communities. Thus, language use by women and men can be understood only if viewed as the result of strategic and socially meaningful linguistic choices. Kurds in Turkey: L2 Learning Setting and Social Networks Any investigation of the differences in Kurdish girls’ and boys’ identification patterns with the L1 and L2 communities and their degree of success in attaining a native-like Turkish accent needs to attend to the greater sociopolitical context represented by the current situation of the Kurdish-speaking community in Turkey. To varying degrees over the past 80 years, it has been illegal to essentially communicate in any language other than Turkish in government offices in Turkey, including schools (Skutnabb-Kangas & Bucak, 1995). Although recently the situation has improved quite remarkably as a result of EU sanctions, Kurds have been living in a situation where they are obliged to learn Turkish for any hope of educational, social, or economic advancement, but still consider the Kurdish language a symbol of ethnic identity. A setting where an ethnic minority, who perceives its native language as a symbol of ethnic identity, is obliged to learn the language of a dominant group trying to establish national linguistic and cultural hegemony presents a number of challenges for L2 learners. Taylor, Meynard, and Rheault (1977) suggest that when minority groups are involved in a struggle for cultural and linguistic survival in the face of threatened assimilation by more dominant groups, for some individuals, anticipated rewards of learning an L2 do not balance out the perceived costs in terms of loss of ethnic or cultural identity. The development of self is dependent upon that of language, because self is something that is socially constructed in the processes of social and cognitive development, while self as opposed to the other can act as a self-conscious badge indexing allegiance with or betrayal to a language community (Mead, 1934; Woolard, 1991).

Downloaded By: [Polat, Nihat] At: 16:35 16 February 2010

20

POLAT AND MAHALINGAPPA

Ethnic identification is constructed based on interactions with both distant and proximal environmental factors (Umaña-Taylor & Fine, 2004), which include parental cultural background, parental gender, and adolescents’ gender (González, Umaña-Taylor, & Bámaca, 2006). Research suggests that learners who have internalized the identity of a particular speech community will engage in positive identity practices (Bucholtz, Liang, & Sutton, 1999). Lamarre and Paredes (2003) suggest that native-language use is associated with attachment to identity, loyalty to family, and membership in a community by L2 learners, and children were likely to continue to speak to parents or older relatives in their L1. Triantafillidou and Hedgcock (2007) report that parental pressure and negative experiences of L2 use in home or academic settings are obstacles to L2 learning. In examining the attitudes of parents toward language use in bilingual Paraguayan families, García (2005) reports that while some parents support children’s use of Spanish (L2) for communication and education, others were more concerned about the maintenance of native culture and identity through L2 use. However, Paugh (2005) indicates that children’s active participation in their own language patterns is evidenced by their code switching and use of forms that differ from the cultural and linguistic practices of their parents. Learners may use an L2 to identify with the majority, while other times they use it to separate themselves from the L2 community, indicating that identity can range from pure identification with the ethnic culture to a blended bicultural identification (Bash & Zezlina-Phillips, 2006). As Freed (1995) suggests, all these varying perspectives underscore the potential variation in patterns related to gender, ethnicity, power, and identity in various sociocultural settings. Strong identification with the ethnic L1 community may hinder L2 learning as much as a high degree of social and psychological distance from the L2 community. SLA research has not only suggested that interacting with the L2 community increases exposure to comprehensible input (Krashen, 1987), studies have also found that L2 learners who are able to engage in social interactions with native speakers of the L2 will experience less social and psychological distance from the L2 community than learners who do not have native speakers in their social networks (Gullestad, 1991; Lybeck, 2002; Schumann, 1978). This lack of distance increases L2 learners’ degree of integration and acculturation into the L2 community, lowering affective barriers to learning and, thereby, improving their L2 proficiency to a greater degree. However, Berry (1998) asserts that the continuum between integration and separation depends on whether the learners want to preserve their identities or whether they value relationship with the target language group. Thus an L2 learner’s degree of success in L2 attainment may depend upon the strength of his/her identification with the L1 community versus the degree of integration and acculturation with the L2 community. Ethnic Kurds in Turkey can fall into three different categories according to language use: monolingual in Turkish, monolingual in Kurdish, or bilingual in Kurdish and Turkish. For Kurds who live in urban areas outside the southeast Turkey, it is almost impossible to remain Kurdish monolingual. In this case, often the first generation after migration is bilingual and afterwards can be assimilated and become fully Turkish monolinguals, depending on socialization patterns and other factors (Mutlu, 1996). Monolingual Kurds are those living in monolingual villages who have had little access to formal education (in Turkish); this group includes women and older individuals (Smits & Gündüz-Hosgör, 2003). Bilinguals can be considered native bilinguals, with equal proficiency levels in both languages, or consecutive bilinguals, monolingual in Kurdish until they began formal education in Turkish, as discussed in the current study. In order to understand fully the relationships between language and social categories with respect to native speakers’ patterned use of language, Milroy (1987) drew on the concept of social

Downloaded By: [Polat, Nihat] At: 16:35 16 February 2010

GENDER AND L2 ACCENT ATTAINMENT

21

network theory, a method of modeling behaviors from sociological theory. Milroy considered the most significant relationships to be those of kin, work, neighborhood, and friendship (Milroy, 1987). At varying degrees, these four types of relationships likely have an impact on density, whether members of a community network are in touch with one another, and multiplexity, the range of different types of transactions with different individuals, of the networks of Kurds in Turkey. For instance, Kurds historically have been organized according to a tribal structure, where kin is the most important tie influencing life decisions. In modern times, the tribal structures have loosened but not disappeared. In rural areas, relatives live in the same or neighboring villages, most of the work is agricultural, one’s neighbors are also friends and/or coworkers (multiplexity), and every person knows one another in some capacity (density). Despite institutional pressure from the government, people still speak Kurdish in their homes and in the street. In contrast, in urban areas, the content and nature of networks depend on a number of factors. If living in a predominantly Kurdish urban area (i.e., cities in the southeast), in addition to family, co-workers, neighbors, and friends are also likely to be Kurds, reinforcing the use of Kurdish. In urban areas outside the predominantly Kurdish regions, Kurds still live in pockets or neighborhoods where Kurdish is the main language used in the neighborhood, but outside that pocket, Turkish is the main language. Networks are less dense and multiplex than before, but more so than in areas where Turkish is the main source of communication, and neighbors and friends can be Turkish speakers. Turkish dominant networks have the capacity to impose their linguistic and cultural norms on Kurds who try to maintain their linguistic and cultural identity in these networks that provide them security, jobs, education, and so on. If Kurds do not have jobs, education, or security expectations in their networks with Turks, then they are likely to use their strong ties and hold onto their Kurdish norms. The motivations for acquiring Turkish fluently are great. Therefore, any pressure that would keep a Kurdish student from attaining absolute fluency in the Turkish language would also have to be great. As suggested previously with social network theory, strong network ties help to sustain a variety that is not seen as legitimate in the face of enormous institutional pressure as there is in Turkey, and weak ties will aid in language shift away from one variety to another. Therefore, Kurds with stronger Kurdish-speaking ties have more incentive to maintain their use of Kurdish and resist attaining a native-like Turkish accent. This is especially true in situations where teachers fail to maximize opportunities of exposure to Turkish for these children. Considering gender differences in language use and identity and Kurds’ social networks, it is therefore interesting to explore how this situation is manifested in identity and acculturation/ integration patterns of boys and girls differently. In this study, we examine how boys and girls may differ in identification with both L1 and L2 communities and acculturation/integration (through social networks) with both communities and attainment of a native-like Turkish accent.

METHODS Participants and Setting Participants, who were ethnically Kurdish and had lived in the same city all their lives, included 65 boys and 56 girls (121 students) at different public middle and high schools in Eastern Turkey.

Downloaded By: [Polat, Nihat] At: 16:35 16 February 2010

22

POLAT AND MAHALINGAPPA

Ages ranged between 13 and 14 for the 60 middle school participants (49.5% of total) and between 16 and 18 for the 61 high school participants (50.5% of total). Having grown up speaking Kurdish with varying degrees of exposure to the Turkish language, the participants had received all their education in Turkish. Most of them had been monolingual in Kurdish until they had begun formal education in Turkish, around 6 or 7 years of age. Due to the fact that the youngest participants in this study were at least in the 7th grade, all participants had a fairly good spoken and written command of standard Turkish; they had all passed their grade-level Turkish courses, the same Turkish language courses offered to native Turkish speakers at middle and high schools. Participants were mostly from low socioeconomic backgrounds representing the overall poor economic situation of the city. They were selected on the basis of random sampling. Data were collected in a major city in Eastern Turkey, where the Kurdish population is reported to be approximately 16%. Apart from personal contacts that helped the researchers gain access to the schools and participants, the most important reason for choosing this particular setting is in the nature of the relationships between its Kurdish and Turkish population; this city does not have a history of extremism and long-lasting ethnic clashes between Kurdish separatists and the Turkish government, despite the significant Kurdish population in the city, which still tends to live in ethnic enclaves. Given the study objectives, this project could not have been conducted in settings in which inter-cultural interactions were either sociopsychologically denied to people or one culture had entirely assimilated into the other. Research Questions In order to address the potentially different identification, acculturation and integration, and the L2 attainment patterns of the Kurdish girls and boys, the following research questions were addressed: 1. How different are girls and boys in producing a native-like Turkish accent when speaking Turkish as rated on a global scale? 2. How different are their identification patterns with the Turkish-speaking and Kurdishspeaking communities; how do these patterns relate to success in attaining a native-like Turkish accent? 3. Regarding the density and multiplexity of exchange, interactive, and passive networks, how different are girls’ and boys’ family and non-family integration and acculturation patterns, and how do these patterns relate to success in attaining a native-like Turkish accent?

INSTRUMENTATION Judge Training and Accent Measurement Accent native-likeness was measured by judges rating recordings of participants’ reading of a paragraph (Bongaets, 1999; Birdsong, 2004). Native-likeness was rated in terms of “passing as a native speaker” on a scale of 1 to 5; a score of 1 stood for “definitely nonnative; very strong foreign accent,” whereas 5 stood for “definitely native; no foreign accent.” Moreover, to utilize judges’ full sensitivity to L2 speech and thereby make the accent measurements most accurate, they

Downloaded By: [Polat, Nihat] At: 16:35 16 February 2010

GENDER AND L2 ACCENT ATTAINMENT

23

were trained to use increments of half of a point to make their judgments more precise (Piske, MacKay, & Flege, 2001; Southwood & Flege, 1999). In this way, the scale became a 9-point scale decreasing ceiling effects and allowing for more specificity in determining participants’ Turkish accent levels. The read-aloud paragraph (Appendix A) was created based on criteria including literacy issues, level of difficulty, topic, rating blocks, age, and gender. Also note that to ensure authenticity and avoid a possible reading ability and pronunciation mix-up, the paragraph topic was carefully chosen to be about the weather in their city, including most basic vocabulary and sentence structure. Some representative distinguishing features of the regional Turkish accent were highlighted in the paragraph given to the judges to help them make their judgments more precise in distinguishing between native-like levels. Furthermore, judges were encouraged to use these segmental linguistics features only in cases where global parameters such as pronunciation, intonation, junctures, and so forth were (Piske et al., 2001) inadequate to help their perceptions. Seven judges, who were ethnic Turks and native speakers of the regional Turkish accent, underwent a 50-minute training session, which involved listening to sample recordings and explanations for each native-like level. Nonetheless, ultimately the scores of two judges were eliminated, and only the ratings of five judges were used. The judges were three college-graduate men and two college-graduate women who were born and raised in the same city, including three English as a Foreign Language (EFL) teachers, one doctoral candidate in EFL, and one professor in applied linguistics. Sample recordings of Turkish native speakers (n = 10) were also intermixed with those of Kurdish students (n = 25) to act as controls to ensure that judges were able to distinguish between native and nonnative accents during the training process; however they were excluded from the actual rating process. Following the exact same data collection procedures, sample recordings were collected from students from the same grade levels at the same schools with the participants. The ratings for Turkish controls ranged between 4.86 and 5.00 (M = 4.97). Accent ratings of the seven judges for both nonnative and the control samples used in training were compared to ensure the establishment of native-likeness norms (Piske et al., 2001). Two judges consistently rated the speech samples of the controls as accented, which indicated that there were intervening factors that hindered their sensitivity to establishing native speaker norms in accent measurement. An analysis of their backgrounds indicated that they were different from the rest of the judges regarding several aspects, including familiarity with the regional Turkish accent, lack of knowledge of foreign languages, and lack of contact with Kurdish speakers of Turkish. Therefore, considering concerns raised by previous research (Thompson, 1991) regarding uniformity and reliability of raters, these two judges were eliminated from the study. Finally, a mean score of all five judges was computed for each participant, and these means were then used as native-like accent ratings. L2 Identification and Acculturation Measurement Participants filled out three questionnaires in order to measure identification and acculturation and integration patterns. The Background Information Questionnaire (BIQ) aimed to collect information regarding participants’ general socioeconomic, ethnic, and linguistic background.

Downloaded By: [Polat, Nihat] At: 16:35 16 February 2010

24

POLAT AND MAHALINGAPPA

The Identity Questionnaire (IQ) consisted of two sections of a list of 30 positive words/phrases that could be used to describe Kurdish-speaking people and Turkish-speaking people. Participants rated how well these words/phrases described both groups on a scale from 1 (Not at all) to 5 (Very well). This questionnaire was based on the semantic differential technique used by Spolsky (1969) and Oller, Hudson, and Liu (1977). Using Milroy’s (1987) framework, the Social Networks Questionnaire (SNQ) aimed to identify the content and structure of participants’ exchange, interactive, and passive networks, as well as their density and multiplexity. Exchange networks include family and close friends, and interactive networks include acquaintances that are present in first-order network ties. Passive networks, in contrast, are more distant and present in second-order network ties. The questionnaire items included both family networks (121*5 = 605), such as parents and siblings, and nonfamily networks (121*7 = 847), such as classmates, teammates, neighbors, teachers, coaches, and acquaintances; each participant’s family networks were restricted to five contacts and nonfamily networks, to seven, for consistency purposes across rating levels. For each network, participants also reported which language (Turkish, Kurdish, Both) they used and how frequently, how long they had known each person, why and how often they met, and how close they thought they were.

DATA COLLECTION AND ANALYSIS Once eligible participants (ethnic Kurds) from three different middle and high schools were identified based on their answers from the BIQ, the researchers recorded them reading aloud the paragraph in Turkish. Finally, the participant filled out the identity and social networks questionnaires. Data were gathered during regular class periods on school premises; speech recordings took approximately one minute while questionnaires took approximately two hours in two sessions during the same week. Some qualitative data were also gathered through observations, semistructured interviews, and field notes from a stratified random sample of 20 participants (17%) across accent ratings for triangulation purposes. Social networks data were analyzed using percentages, chi-square statistics and descriptions while accent ratings and identity data were analyzed through analysis of variance (ANOVAs) and multiple regressions.

FINDINGS Accent The high degree of uniformity among the ratings of native Turkish speakers used during the training of judges indicated the rating patterns of the judges. Therefore, based on the inter-rater reliability coefficients (Table 1) on a global accent-rating procedure, data suggest that judges were in agreement in their ratings of the Kurdish participants’ accents as they read aloud in Turkish. Results reveal that female participants obtain much higher accent ratings with a group mean of 3.48 (SD = .9), while the mean for boys is only 2.19 (SD = .9). An independent T-test

GENDER AND L2 ACCENT ATTAINMENT

25

TABLE 1 Inter-Rater Reliability Coefficients Among Judges Judges

All Samples

Downloaded By: [Polat, Nihat] At: 16:35 16 February 2010

1–2 1–3 1–4 1–5 2–3 2–4 2–5 3–4 3–5 4–5

.88 .87 .89 .86 .85 .89 .87 .89 .86 .86

analysis was performed to determine if in fact the observed mean difference between boys’ and girls’ accent ratings was significant. Using SPSS, participants’ ratings were entered as “Test Variables” and gender as “Grouping Variable” in this model. Results suggest that boys’ and girls’ accent native-likeness ratings were significantly different from each other (t = 7.722, df = 119, p < .05). Concerning the gender group structures, namely the number of boys and girls represented in each accent native-likeness level, data suggest remarkable percentile differences. Figure 1 shows that the majority of participants in levels 1 “Definitely nonnative” and 2 were boys (90% of level 1 and 82% of level 2). Although the percentile difference between boys and girls seems the smallest in level 3, there are more girls (58%) than boys. Similar to the large difference between boys and girls in levels 1 and 2, the difference between the groups in levels 4 and 5 is also notable; 79% of participants in level 4 were girls, and 82% in level 5, “Definitely native,” are girls, the same percentage as male participants in level 2. To address whether the percentile differences in the number of boys and girls in each accent level were significant, a chi-square test was performed. The chi-square statistics results suggest

25% 20% 15%

Males Females

10% 5% 0% Level 1

Level 2

Level 3

Level 4

Level 5

FIGURE 1 Distribution of participants across gender and accent ratings.

26

POLAT AND MAHALINGAPPA

that concerning the distribution of the number of boys and girls across the five native-like accent ratings, there is a significant difference (χ2 = 42.6, df = 4, p < .01).

Downloaded By: [Polat, Nihat] At: 16:35 16 February 2010

Identification Patterns Both MANOVAs (overall multivariate analysis of variance) and multiple regressions were used to examine the relationship between accent native-likeness and identification patterns of the Kurdish students separately for boys and girls. Table 2 shows that for the outcomes of identification with the Turkish-speaking community, girls’ means are significantly higher than boys’ means. In contrast, boys’ means are significantly higher than girls’ means regarding the outcome of identification with the Kurdish-speaking community. An overall one-way MANOVA was performed to compare boys and girls in terms of the four variables. In this model, participants’ accent ratings, Turkish identification means, and Kurdish identification means were entered as dependent variables and gender as the independent variable. Table 3 reveals that as far as these variables are concerned, the differences in the means between boys and girls are significant F(1, 117) = 23,703, p < .001. Moreover, boys’ and girls’ mean scores were also compared concerning the two identification variables using an ANOVA test. Table 4 indicates that boys and girls significantly differ on identification both with the Kurdish-speaking community F(1, 119) = 51.67, p < .001, and the Turkish-speaking community F(1, 119) = 37.27, p < .001. Simultaneously entered into two regression analysis models, the relationship between the two identity variables and accent native-likeness for boys and girls was calculated separately. An r-squared of .52 (p < .01) was obtained for girls and an r-squared of .51 (p < .01) was obtained for boys. Findings suggest significant differences regarding the relationship between girls’ and boys’ identification patterns and the attainment of a native-like accent. Note that the effect sizes

TABLE 2 Means (SD) for Boys and Girls on Identification Variables Variables Identification KUR Identification TUR

Boys

Girls

3.4 (.60) 2.9 (.80)

3.7 (.70) 4.1 (.40)

TABLE 3 MANOVA Results: Means of Boys and Girls regarding the Four Variables MANOVA table for gender

Value

F-Value

Num. DF

Den. DF

P-Value

S M N Wilks’ Lambda Roy’s Greatest Root Hotelling-Lawley Trace Pillai Trace

1.000 .500 57.500 .622 .608 .608 .378

23.703 23.703 23.703 23.703

3 3 3 3

117 117 117 117