In-between voice: Accent bias and identity among baluchi speakers in Iran

1. Introduction

Iran's considerable linguistic diversity presents a compelling context for linguistic research. Within this multilingual landscape, Persian, the official language, has long coexisted with numerous indigenous and minority languages, including Arabic, Azeri, Baluchi, and Kurdish. An essential method for investigating the status and ethnolinguistic vitality of these languages is the analysis of their speakers' attitudes (Rezaei & Bahrami, 2019). Such research is of considerable importance to sociolinguists and national policymakers as it contributes to the protection of global linguistic diversity and preservation of ethnic and linguistic balance (Rustmovna et al., 2018). While previous scholarship has often focused on the attitudes of majority linguistic groups towards minorities (e.g., Alshehri & AlShabeb, 2023; Dixon & Ergin, 2010; Murad, 2007; Ridwan et al., 2016), a noticeable gap exists in the literature concerning the attitudes of minority groups towards their own languages.

Language attitudes are typically understood through a tripartite framework, encompassing three core components: affective (emotions associated with the language), behavioural (actions influenced by the attitude), and cognitive (beliefs and knowledge about the language) (Loureiro-Rodriguez et al., 2013). This framework allows for a comprehensive analysis of the internal and external factors shaping an individual's relationship with a particular language.

The study of language attitudes is particularly crucial for the sustainability of minority languages as these attitudes directly impact their transmission, revitalization, and ultimate survival. The Baluchi language, for instance, has been significantly influenced by the dominant languages of the countries where it is spoken, such as Urdu in Pakistan and Persian in Iran. This has led to a noticeable language shift in some communities. For example, some younger generations of Baluchi speakers are adopting dominant languages in the domestic sphere, viewing it as a means of modernization. This trend poses a significant threat to the intergenerational transmission of Baluchi, potentially leading to its endangerment—a challenge faced by numerous minority languages across Iran (Gholami, 2020).

The present study aimed to investigate the attitudes of native Baluchi speakers in southeastern Iran towards three linguistic objects: their native Baluchi language, the national language of Persian, and Persian spoken with a Baluchi accent. It is anticipated that this research will contribute to a deeper understanding of the relationship between language attitudes and social identity, as well as the broader challenges of language maintenance. This is particularly relevant in a context where Baluchi speakers are in frequent contact with non-Baluchi speaking communities.

To achieve these objectives, the study was guided by the following research questions:

• RQ1: What are the attitudes of Iranian speakers of Southern Baluchi towards their mother tongue, Persian, and Persian spoken with a Baluchi accent?
• RQ2: Are there significant differences among the attitudes of Iranian speakers of Southern Baluchi towards their mother tongue, Persian, and Persian spoken with a Baluchi accent?

2. Theoretical Framework

The study of language attitudes gained significant traction in the mid‑twentieth century propelled by the foundational work of scholars like Gardner and Lambert (1959). Its establishment within sociolinguistics was gradual as the field itself was then in its nascent stages. Traditionally, attitude research is built upon a tripartite model, conceptualizing attitudes as an interplay of three components: cognitive (beliefs about the language), affective (feelings towards the language), and behavioural (actions related to the language) (Mantle‑Bromley, 1995). Subsequent developments have also incorporated a social‑constructivist perspective, analyzing attitudes in relation to concepts like ethnolinguistic vitality and socio‑structural factors like standardization (Adegbija, 2000; Ryan et al., 1982).

A language attitude can be defined as an evaluative stance—whether favorable or unfavorable—towards a language or its speakers. For decades, sociolinguists have sought to understand the mechanisms, through which individuals evaluate speech styles and how these attitudes can predict language behavior. In their influential work, Agheyisi and Fishman (1970) classified attitude studies into three major domains: those focusing on attitudes toward a language itself, those examining community‑wide impressions of language varieties, and those concerned with the functional use of different attitudes. While these domains are distinct, it is argued that a blended model incorporating all three yields more comprehensive results (Majidifard et al., 2019). Methodologically, language attitude studies employ a range of quantitative and qualitative techniques to collect data, including questionnaires, Matched‑Guise Tests (MGT), semi‑structured interviews, and focus groups (Kircher & Zipp, 2022). These methods often illuminate not only a speaker’s feelings towards a language (warmth), but also their assessments of its social standing (competence).

Fundamentally, attitudes reflect an individual’s knowledge, feelings, and behavioral intentions regarding a language. Investigating the language attitudes of native speakers, particularly of less‑documented languages, allows researchers to make informed predictions about the trajectory of endangered languages. By observing the actions and inclinations of a speech community, attitude studies provide critical insight into the current state of global linguistic diversity and ethnolinguistic vitality of languages within a broader social justice framework.

Beyond the tripartite structure, recent cross‑disciplinary developments—especially in social psychology—have introduced the Stereotype Content Model (SCM) (Cuddy et al., 2008) as a complementary lens for understanding the evaluative dimensions of social perception. The SCM proposes that individuals judge social groups along two universal dimensions: warmth (perceived friendliness and trustworthiness) and competence (perceived ability and status). These constructs have increasingly informed sociolinguistic research, where warmth and competence align closely with affective and cognitive components of attitude, making SCM a valuable extension to traditional attitude models.

This theoretical foundation directly informed the design and objectives of the present study. The tripartite model guided the overall exploration of attitudes embedded in RQ1—which sought to map the affective, cognitive, and behavioral components of the listener’s stance towards differing speech styles—while the Stereotype Content Model (SCM), with its focus on the warmth–competence dichotomy, provided the specific lens, through which the data from the MGT were analyzed to address RQ2. Specifically, the social judgments elicited by the guise recordings were operationalized through these SCM dimensions. Thus, the research questions were not merely descriptive but were designed to test how these established psychological and sociolinguistic models manifested in the language attitudes of the Southern Baluchi community, showing the interplay between general attitude structures and content‑specific stereotypes.

3. Literature Review: Studies in the Iranian Context

Research from diverse multilingual settings consistently showed a complex interplay between minority language loyalty and the social and economic pressures exerted by dominant languages. Foundational studies on Catalan in Spain (Woolard, 1989), Welsh in the UK (Baker, 1992), and Scottish Gaelic (Dorian, 1981) established that while a heritage language is often cherished for identity and in-group solidarity (affective loyalty), the national language is typically favored for economic mobility and formal domains (instrumental motivation). This dichotomy remains a central theme in contemporary research, with recent studies on attitudes towards Irish in Ireland (Mhic Mhathúna & Mac Lochlainn, 2022) and Basque in Spain (Lasagabaster, 2017) continuing to explore these tensions. These global patterns find specific and complex manifestations in the sociolinguistic landscape of Iran, which provides a rich context for attitude research, especially as many of its linguistic varieties face the threat of endangerment.

Several studies have explored attitudes among Iran's diverse ethnolinguistic groups. Mirhosseini and Abazari (2016), for example, examined the attitudes of Persian-Azeri bilinguals in Tabriz towards Azeri. Using questionnaires and interviews, they found that despite a positive predisposition towards using Azeri, participants expressed reservations about its application in formal domains like education and mass media. Similarly, Fereidoni and Zardi (2024) explored the attitudes of 80 Iranian bilinguals towards standard Persian and Kurdish. While their findings revealed no statistically significant differences, participants generally showed more favorable attitudes towards Persian.

Research has also focused specifically on attitudes towards accented speech. Mirshahidi (2016) interviewed Iranian bilinguals (with L1s of Arabic, Azeri, and Kurdish) and found that speakers with nonstandard Persian accents were often perceived as having lower social and educational status. Interestingly, Gilaki and Azeri accents elicited more negative evaluations, whereas Arabic and Kurdish accents elicited more positive ones.

Other studies have concentrated on the attitudes of a single minority group towards their own heritage language. Rezaei et al. (2017) conducted a questionnaire-based study with 400 native Azeri speakers and confirmed they held highly positive attitudes towards their mother tongue, despite its exclusion from the national education system. In contrast, a study on language shift in Zanjan among Azeri-speaking families found a different outcome. Mirvahedi and Jafari (2018) interviewed parents who, despite expressing strong loyalty to their ethnic language, favored using Persian in the domestic sphere, resulting in decreased Azeri proficiency in both parents and children.

Finally, research has also extended to diaspora communities. Gharibi and Mirvahedi (2021) investigated Family Language Policy (FLP) among Iranian immigrants in the UK. Their work highlighted how factors, such as strong social networks, heritage language institutions, and frequent travel to Iran, created an environment conducive to promoting bilingualism and maintaining the heritage language abroad.

4. Methodology

This study utilized a quantitative approach to investigate language attitudes. The research design was centered on a survey that collected numerical data through participants' evaluations of audio stimuli, which were then analyzed by using statistical tests.

4.1 Participants

A total of 207 native speakers of Southern Baluchi residing in the southern region of Iran's Sistan and Baluchestan Province were recruited via convenience sampling. Participants' ages ranged from 17 to 61 with a mean age of 25.28 (SD=6.98). Detailed demographic information regarding gender and educational background is presented in Table 1.

 4.2 Instruments

The principal instrument for this study was a modified MGT. The MGT originally developed by Lambert et al. (1960) operates on the premise that speech patterns evoke social categorizations and trait associations (Giles & Coupland, 1991). While the classic MGT uses one speaker for all language guises, this study implemented a modified design often referred to as a speaker evaluation or verbal-guise technique (cf. Pishghadam & Sabouri, 2011; Williams, 1976). This approach uses different native speakers for each language variety (Persian, Southern Baluchi, and accented Persian), removing the need for a single speaker to create potentially artificial guises and thereby increasing the authenticity of the linguistic stimuli.

Participants rated the voices on 13 personality traits adapted from Lai (2007), which represented four dimensions: solidarity, competence, personal attractiveness, and aggressiveness. A 4-point Likert-type scale was used for each trait with the following descriptors: (1) Does not match at all, (2) Poor match, (3) Good match, and (4) Excellent match. While the list included more positive than negative traits, this design was not expected to introduce bias as participants could express negative evaluations by assigning low scores on positive traits.

To ensure the instrument's quality, a pilot study was conducted with 40 participants. The resulting Cronbach’s alpha values indicated high internal consistency for the Persian (α=.87), Southern Baluchi (α=.93), and accented Persian (α=.92) guises. Content validity was further established by three university professors of linguistics and translation studies, who reviewed the test items for accuracy and clarity.

4.3 Data Collection Procedure

For the audio stimuli, two male speakers in their forties with similar socioeconomic backgrounds were recruited. One was a native speaker of Southern Baluchi proficient in reading both Persian and Baluchi and the other was a monolingual native speaker of Persian. The speakers’ speeches were recorded reading a short, neutral text about nature to avoid triggering biased reactions. Three separate recordings were produced:

1. The Persian speaker reading the text in standard Persian
2. The Baluchi speaker reading the text in Southern Baluchi
3. The Baluchi speaker reading the same text in Persian with a discernible Baluchi accen.

The three recordings were compiled into a single MP3 audio file (total length: 1 minute, 16 seconds). This file was embedded in an online survey created using Google Forms, a secure platform that ensures participant anonymity (Derakhshan & Fathi, 2024). The survey included instructions and the rating scale with an estimated completion time of 10-15 minutes. Prior to participation, all individuals provided informed consent. Participation was voluntary and the participants were informed of their right to withdraw at any time without penalty.

4.4 Data Analysis

The collected data from 207 participants were screened, revealing no missing values or outliers. The dataset was then analyzed by using IBM SPSS Statistics, version 26.

An initial test for normality was conducted using the Kolmogorov-Smirnov test (see Table 2). Results showed that the data for the Persian guise were normally distributed (p>.05), but the data for the Baluchi and accented Persian guises were not (p<.05).

 Table 2- Tests of Normality for Different Guises

Although non-parametric tests (e.g., Friedman test) are often used when the assumption of normality is violated, repeated measures one-way ANOVA was chosen for the main analysis. This decision is justified on several grounds. First, ANOVA is widely considered robust to moderate violations of normality, particularly with a large sample size like the one in this study (N=207) (George & Paul, 2024; Pallant, 2020). Second, the Central Limit Theorem suggests that with a substantial sample size, the distribution of sample means will approximate normality, mitigating the impact of non-normal data (Islam, 2018; Kwak & Kim, 2017; Zaidi, 2024). Finally, the repeated measures ANOVA is statistically more powerful for a within-subjects design as it effectively manages variability between participants and its results are more widely and clearly interpretable within the academic community compared to those of non-parametric alternatives (Verma, 2015).

5. Results

This section presents the descriptive and inferential statistics from the MGT. The results were organized by research question, first detailing the overall attitudes toward the three guises (Persian, Baluchi, and Persian with a Baluchi accent) and then examining the statistical significance of the differences among them.

5.1 RQ1- Attitudes towards Language Guises (Descriptive Statistics)

To address the first research question concerning the attitudes of Iranian speakers of Southern Baluchi, descriptive statistics (means and standard deviations) were calculated for each trait and composite category across the three guises. The complete results are presented in Table 3 (for affective traits), Table 4 (for social-status traits), and Figure 1, which demonstrates a clustered bar chart showing the composite mean and overall attitude scores for Persian, Baluchi, and Persian with a Baluchi accent across five categories.

Fig 1- Composite Mean and Overall Attitude Scores for the Three Language Guises

A review of the composite mean scores in Table 3 reveals clear patterns. For the categories of Solidarity, Competence, and Personal Attractiveness, standard Persian consistently received the highest mean ratings, while Persian with a Baluchi accent received the lowest. This indicated that participants generally attributed more positive traits related to status and social appeal to the standard national language.

The only category that deviated from this pattern was Aggressiveness, where Baluchi (M=4.86) was rated slightly higher than Persian (M=4.76) and both were rated as more aggressive than the accented Persian guise (M=4.47).

Overall, the descriptive results suggested a hierarchy of preference. Standard Persian was evaluated most favorably followed by the Baluchi mother tongue, with the accented Persian guise being evaluated least favorably across most positive traits.

5.2 RQ2: Differences in Attitudes (Inferential Statistics)

To answer the second research question, a series of one-way repeated measures ANOVAs were conducted to determine if the observed differences in ratings among the three guises were statistically significant. Partial eta squared (ηp²) was used to measure the effect size. The results are summarized in Table 5 and detailed below.

A preliminary ANOVA on the overall attitude scores showed a significant and large main effect of guise: F (2, 205)=14.45, p<.001, and partial η²=.124. Post-hoc comparisons using the Bonferroni correction revealed that both Persian (M=37.77, SD=8.45) and Baluchi (M=37.20, SD=8.86) were rated significantly more favorably overall than Persian with a Baluchi accent (M=34.70, SD=9.44) (p<.001 for both comparisons). The difference between Persian and Baluchi was not statistically significant.

Further ANOVAs were run on each of the four trait categories:

• Solidarity: There was a significant main effect of guise: F (2, 205)=5.79, p=.004, and partial η²=.054. Post-hoc tests indicated that ratings for both Persian and Baluchi were significantly higher than those for accented Persian (p=.017 and p=.005). No significant difference was found between Persian and Baluchi.
• Competence: The analysis revealed a significant, large main effect of guise: F (2, 205)=15.79, p<.001, and partial η²=.134. Post-hoc tests confirmed that both Persian and Baluchi were rated significantly higher in competence compared to accented Persian (p<.001 and p<.001). Again, the difference between Persian and Baluchi was not significant.
• Personal Attractiveness: A significant main effect of guise was found: F (2, 205)=7.29, p=.001, and partial η²=.066. Post-hoc tests showed that Persian was rated significantly higher than accented Persian (p<.001 and p=.05). No other differences were statistically significant.

Aggressiveness: There was a significant main effect of guise: F (2, 205)=5.96, p=.003, and partial η²=.055. Post-hoc tests revealed that Baluchi was rated as significantly more aggressive than Persian with a Baluchi accent (p=.002). No other differences were statistically significant.

Finally, the practical significance of these differences was assessed using partial eta squared (ηp²) as the measure of effect size (see Table 4). Following Cohen's (1988) guidelines, the effect size for Competence and Overall Attitudes was large, while the effect sizes for Solidarity, Personal Attractiveness, and Aggressiveness were of a medium magnitude. While these effect sizes indicated significant differences, it is important to note that with a large sample, even small mean differences can become statistically significant (Pallant, 2020).

6. Discussion

The present study set out to investigate the complex hierarchy of language attitudes among Iranian speakers of Southern Baluchi. The findings revealed a nuanced landscape where the standard national language, Persian, was highly regarded, yet the ethnic mother tongue, Baluchi, commanded a strong, competing loyalty. This discussion will now interpret these findings, connect them to the broader academic literature, and explore their implications for both theory and practice.

It is important to contextualize the scope of these findings regarding participant heterogeneity. The present study focused on generating overarching attitudes towards the three language guises (Standard Southern Baluchi, Persianized Baluchi, and Standard Persian) spoken by male speakers. Consequently, demographic variables, such as participants’ gender and educational background, were not analyzed as independent factors to detect potential interaction effects. The scale of the current participant pool did not permit a fully stratified analysis of these variables. The findings should therefore be interpreted as representing general trends within the participant group and future research could productively explore how these sociodemographic factors might modulate language attitudes, potentially using a larger sample size that allows for power analyses across subgroups.

The results clearly indicated that while Persian held a privileged position, particularly in domains related to social status and competence, its overall attitudinal rating was not statistically superior to that of Baluchi. This suggested a powerful dynamic at play: participants simultaneously acknowledged the instrumental value and prestige of Persian while maintaining a deep affective loyalty to their heritage language. This “prestige vs. solidarity” dichotomy is a classic finding in sociolinguistics, echoing patterns observed in diverse multilingual contexts, such as those with Catalan speakers in Spain (Woolard, 1989) or Welsh speakers in the UK (Baker, 1992). The present study thus provides strong evidence that this global pattern is being replicated in the Iranian context, where long‑standing language policies have elevated Persian’s status (Spolsky, 2004), yet have not erased the cultural resilience of Baluchi as a core marker of ethnic identity.

Perhaps the most compelling finding of this study was the significantly negative evaluation of Persian spoken with a Baluchi accent. The accented guise was rated lower than both the “prestigious” standard and the “solidary” mother tongue. This suggested that the accented speaker was perceived as sociolinguistically “inauthentic” — failing to legitimately claim membership in either the high‑status Persian‑speaking group or the authentic Baluchi‑speaking in‑group. This finding aligns with research on accent stigmatization (Dragojevic et al., 2021; Lippi‑Green, 2012) while adding a crucial layer of understanding. The low ratings across both competence and solidarity traits suggested that the accented guise signaled a speaker who was “in between”, potentially triggering negative judgments about their social and ethnic identity. This strongly supported the theoretical framework of the Stereotype Content Model (Cuddy et al., 2008), demonstrating that the accented speaker was perceived as lacking both the “competence” of the dominant group and the “warmth” of the ethnic in‑group. The primary contribution of this study, therefore, is its demonstration that an ethnically‑congruent accent on a high‑status language can paradoxically create a new social category that is judged more harshly than either the standard language or the minority mother tongue.

These findings hold significant practical implications. Given that Persian is already perceived as the language of competence, policies aimed at linguistic equity should focus not merely on preserving Baluchi as a heritage language, but on actively elevating its status and demonstrating its utility in modern, formal contexts. Furthermore, the clear evidence of “accent bias” — even among speakers of the accent themselves — points to a pressing need for public awareness initiatives in educational and professional settings to combat the marginalization that can result from linguistic prejudice. In addition, these findings have implications for media representation, suggesting that the portrayal of regional or accented speakers on screen and in broadcast can either significantly reinforce or actively challenge these harmful stereotypes.

The conclusions drawn here should be considered in light of the study’s specific design and methodological scope. While the sample size provided a solid foundation for exploratory analysis, future research could build on these findings by employing probability sampling to enhance generalizability to all Southern Baluchi speakers. Similarly, by focusing on male speakers for the audio stimuli, this study offers a crucial baseline; a fruitful next step would be to examine how these attitudinal dynamics operate with female speakers. Furthermore, while the quantitative data provide a robust overview, future qualitative studies — such as interviews — are needed to explore the rich, personal reasons why these attitudes are held. Finally, our analysis did not extend to participant demographics (age, gender, education), which offers another promising avenue for future investigation. Proposing these future directions is not to diminish the current findings but to place them in a continuous chain of scientific inquiry.

7. Conclusion

This study confirmed a central paradox of multilingual identity: while an official language may win prestige, the mother tongue often retains the heart. The primary contribution of this research, however, is the revelation that the perceived middle ground—an ethnically-accented version of the standard language—can be judged most harshly of all, suggesting it represents a form of social inauthenticity that belongs fully to neither world. This insight is not merely academic; it is fundamental to the work of fostering linguistic justice in any society where accent bias serves as a proxy for social prejudice. For the Baluch community, these findings affirm that the future vitality of their language is being actively negotiated in the attitudes of its speakers. By mapping the contours of these attitudes, this research provides an essential foundation for policies that treat linguistic diversity not as a complication to be managed, but as a core component of cultural wealth.