Ethnic group affiliation and second/foreign language accentedness in English and Mandarin among Hong Kong speakers

ABSTRACT

As part of a larger project that investigates the issue of identities in Hong Kong, this study anchored on the sociocognitive paradigm in second language acquisition (SLA) explores the potential relationship between one’s identity and perceived language accentedness. Our study set in Hong Kong (HK) aims to extend Gatbonton and colleagues’ works (e.g. [2005]. Learners’ ethnic group loyalty and L2 pronunciation accuracy: A sociolinguistic investigation. TESOL Quarterly, 39(3), 489–511. https://doi.org/10.2307/3588491; [2008]. The ethnic group affiliation and L2 proficiency link: Empirical evidence. Language Awareness, 17(3), 229–248. https://doi.org/10.1080/09658410802146867; [2011]. Ethnic group affiliation and patterns of development of a phonological variable. Modern Language Journal, 95(2), 188–204. https://doi.org/10.1111/j.1540-4781.2011.01177.x) that examine the relationship between ethnic group affiliation (EGA) and language proficiencies in diglossic contexts. HK is a multi-glossic context where Cantonese, English and Mandarin are the official languages, and they perform distinctive functions in various public and private domains. Through analysing participants’ (n = 65; born between 1970s–1990s) self-identification and their reported accentedness in English and Mandarin, we address the question of whether EGA as a set of social factors has a bearing on a person’s linguistic achievements. Findings indicate that participants’ identification with the Chinese/ HK identity is related to their perceived accentedness in the targeted languages in intricate ways that do not align completely with our predictions. We conclude by calling for further socio-cognitively informed research that investigates multiglossic situations where languages/ language varieties complement or compete with each other.

KEYWORDS:

• Ethnic group affiliation
• second language accentedness
• Hong Kong
• sociocognitive SLA
• identities

1. Introduction

The studies of second language acquisition (SLA) have evolved considerably over time. The focus on the cognitive and mentalistic factors (i.e. learner internal factors such as age, aptitude, etc.) among some earlier work has fuelled research that aims to better understand the nature of linguistic competence among second/ foreign language learners. However, researchers such as Firth and Wagner (1997, 2007) have argued that the heavy focus on such factors at the expense of understanding the social, interactional and contextual elements of language use has led to a skewed perspective of the field. As they see interactions as the site that engenders language development, Firth and Wagner contend that ‘the dichotomy of language use and acquisition cannot defensibly be maintained’ (2007, p. 800). They urged further explorations of the interrelationships between language use, language learning, and language acquisition in order to reach a more comprehensive understanding of the field. Such arguments have contributed to or even led to the ‘social turn’ in the field of SLA. Though it should also be noted that social factors and the notion of context have featured in some models and theorisation of SLA which predate Firth and Wagner’s widely cited paper in 1997. They include Spolsky’s general model of second language learning (1989) and Schumann’s acculturation model (1978). The language socialisation paradigm (Duff & Talmy, 2011), the socio-educational model (Gardner, 2006) and the socio-cultural model (Lantolf, 2011) are other examples which place a strong emphasis on the social elements.

With the social turn (Block, 2003, 2007) comes increasing attention towards the role that social factors and social context more generally play in the process of SLA (e.g. Leung & Young-Scholten, 2013). Most works in this paradigm, which often have a socio-cultural orientation, focus on analysing learners’ socio-linguistic experience and how learners engage and invest in the process of learning (see Darvin & Norton, 2015).¹ However, fewer studies have investigated how these social circumstances might influence linguistic outcome. One could argue that the pendulum has once again swung to the extreme, where this time investigations and interests in ‘the social’ came at the expense of concerns over linguistics/ cognitive factors and outcome. The unfortunate state of affairs where the cognitive/ linguistic and the social paradigms have remained entrenched and separated can perhaps be attributed to the perceived ontological and epistemological ‘dichotomy’ that researchers from the two paradigms hold (cf. Leung & Young-Scholten, 2013; Hulstijn et al., 2014). But to be fair, the linguistics/ cognitive factors are not the prime concerns of researchers who have a more social orientation (and vice versa); researchers in the social paradigm are often interested in documenting the complexity of experience and learners’ ability to negotiate in complex social settings as opposed to learners’ ability to supply the correct verbal inflections, for instance.

Yet, recent research in the ‘socio-cognitive’ paradigm allows us to see the potential complementarity and synergy of the two seemingly distinct and incommensurable perspectives. In addition to the traditional focus on learner-internal factors, the growing body of research under the socio-cognitive perspective has begun to consider how social factors such as attitude and identity mediate the linguistic experience of language learners which may in turn affect acquisition outcome (see Batstone, 2010; Hulk & Marinis, 2011; Moyer, 2013). For example, a stronger affiliation to the target language group may enhance the extent and diversity of language exposure, the enhanced input in terms of both quantity and quality in turn creates a richer or more favourable environment for linguistic development to take place. In our view, this socio-cognitive perspective aligns well with the recent and renewed interest in transdisciplinarity and the exploration of the interplay among the micro, meso, and macro contexts and factors that the Douglas Fir Group have drawn our attention to (2016, 2019).

One realisation of such socio-cognitive perspective is the work by Gatbonton and her colleagues (e.g. 2005, 2008) which examine the relationship between people’s ethnic group affiliation (EGA) and their (self-perceived) language proficiencies in Quebec where French and English co-exist and function alongside one another (see also Trofimovich et al., 2013 and Trofimovich & Turuševa, 2015 for the Latvian context in relation to Russian and Latvian). The present study models after this line of research. Set in Hong Kong where Cantonese, English and Mandarin function alongside one another as official languages, our research study aims to investigate speakers’ EGA and their self-perceived accentedness in English and Mandarin. It is hoped that this study will add to current understanding of the relationship between social factors – more specifically EGA – and SLA/ foreign language learning (FLL) from an Asian perspective.

The remainder of the article first reviews some EGA studies. It then proceeds to provide some background information about the research setting. This is followed by the study design and findings from the present study. We conclude by calling for further investigations into contexts where different languages or language varieties function in a diglossic or multi-glossic manner.

2. Ethnic group affiliation (EGA) and second/foreign language learning

Although Gatbonton and her colleagues initially made no explicit reference to the sociocognitive orientation, their series of studies which explore the relationship between EGA, which is defined as ‘one’s sense of belonging to a primary ethnic group’ (Gatbonton et al., 2005, p. 489), and second/ foreign language proficiency can be viewed in such a light. EGA is constituted by a set of social factors, e.g. depth of involvement in the ethnic group; pride in, familiarity with, and feelings of comfort with the group; perception of the place of the ethnic group in relation to other groups; perceptions of the group's vitality; and views towards the socio-political concerns of the group. Hence studies that aim to explore the relationship between EGA and linguistic outcomes, often operationalised as global proficiency, intelligibility, or accent in the target language (see below), can be considered to fall under the sociocognitive framework.

In a series of pioneering studies, the potential relationship between EGA and (self-perceived) language proficiency and political affiliation in the context of Quebec was explored. For example, Gatbonton et al. (2005)’s seminal study on the topic investigated Francophone (n = 24) and Chinese (n = 84) L2 English learners’ perception of L2 English in Quebec. They found a tension between perceived efficiency/ proficiency in the L2 and perceived affiliation to the ethnic group. Speakers who spoke a ‘less-accented’ form of L2 English are perceived to be less affiliated by listeners of their L1 group. However, at the same time, listeners preferred to have less accented speakers as political leaders, hinting at the importance of ‘effectiveness’ (intelligibility) in speech.

The relationship between EGA and L2 proficiency is examined more directly in Gatbonton and Trofimovich (2008). 59 adult French–English bilinguals from Quebec who participated in the study read English text and completed an EGA questionnaire which assesses the pride, loyalty and support for their ethnic group and its language. Results revealed that EGA is related to certain aspects of accent in a complex way. Both positive and negative links exist between EGA and L2 English speaking ability (self-rated and judged by natives). Those who strongly supported Quebec’s independence sounded more accented, less comprehensible, less fluent and less proficient overall. While those who had double positive orientation (towards their own group and the L2 group) are judged to be the most proficient. The observed effect is mediated by language use. Doucerain (2019)’s exploration in the Quebec context further demonstrated that among recent multicultural immigrant participants, L2 experience, in terms of L2 use and L2 social contact (i.e. friendships in the ‘mainstream’ group), mediates their positive cultural orientation and self-assessed L2 competence.

Apart from global measures of proficiency, the relationship between EGA and a specific linguistic element has also been studied. Gatbonton, Trofimovich, and Segalowitz’s study (2011), for instance, investigated the production of the voiced interdental fricative, /ð/, by 45 Francophone L2 English learners. EGA is again found to play a role in participants’ L2 English. Findings suggested that the stronger the Francophone EGA participants held, the less native-like their L2 pronunciation accuracy was, as evaluated by native speaker judges.

EGA research has been extended to another research setting outside Quebec firstly by Trofimovich et al. (2013) and Trofimovich and Turuševa (2015). They examined the relationship between ethnic group identity and L2 speaking ability of ethnically Russian and Latvian speakers in Southern Latvia. Similar to the context of Quebec, where the majority French speaking community co-exist alongside the minority English speaking group, in Southern Latvia, the majority ethnic Latvian population co-exist with the minority ethnic Russian group. Their results indicated that for Latvians (n = 82), the stronger their EGA, the lower they rated their own L2 Russian ability. However, no such effect is observed for the L2 Latvian among ethnic Russian speakers (n = 26), which suggests that their ethnic identification is unrelated to their perceived Latvian ability. It is also interesting to note that Trofimovich and Turuševa (2015) have articulated their sociocognitive orientation more explicitly when they set out to ‘integrate evidence from social psychology and applied linguistics, by focusing on the identity–language link from the perspective most relevant to second language (L2) development, namely, by considering how ethnic identity might be implicated in L2 learning’ (234).

More recently, related studies set in other contexts have emerged. They include Peng and Patterson’s (2021) exploration of international students’ cultural identification and their L2 English proficiency in the US higher education context, Gu et al.’s (2023) study of ethnic minority immigrant parents in Hong Kong, and Banyanga et al.’s (2018) investigation of Finnish-Swedish highschoolers in Finland. Peng and Patterson (2021) found a negative relationship between their 77 participants’ ethnic identification and self-perceived English proficiency which is mediated by motivation in learning English. They also demonstrated that ‘American identification promoted English proficiency through motivation in language learning’ (67). Gu et al. (2023) found that their 655 ethnic minority participants’ cultural knowledge was linked to better spoken Cantonese, while greater cultural identification with their own ethnic community was linked to better English-speaking skills. They also observed a difference between males and females with females reporting inferior Cantonese- and English-speaking proficiencies. Banyanga et al. (2018) showed that language and Finland-Swedish culture are important to their 1012 ninth-graders’ self-identity.

By and large, this body of work have demonstrated the relationship between EGA and language proficiency and/or language learning, though that relationship is often complex and multifaceted (but see also Tekin, 2019 where no correlations were found between EGA and ratings of comprehensibility, intelligibility, accentedness, and acceptability of L2 English speakers in the US from a range of L1 backgrounds). However, much of the existing literature, as reviewed above, focused on contexts with two languages at play, less is known about settings where more than two languages (or language varieties) function alongside each other. Our study aims to extend this body of work and further enhance our understanding of the relationship between EGA and language learning/ accentedness by considering Hong Kong, a context where three main linguistic codes, namely, Cantonese, English and Mandarin, exist and function in the society.

3. Socio-political and linguistic background of Hong Kong

Shaped by its past as a British colony, HK people claim a dual/ mixed identity, seeing themselves as both ‘Hongkongers’ and ‘Chinese’ (Brewer, 1999; Ma & Fung, 2007). This allows them to claim their ethnic heritage with the wider Chinese population, but also to differentiate themselves from mainland Chinese outside HK, prizing their cosmopolitanism and command of the English language (Chan, 2002; Joseph, 2004). With tourism liberalism measures easing travel restrictions between mainland China and HK and the Closer Economic Partnership Arrangement since the sovereignty was returned to China in 1997, there has been an increase in contact between residents of the two places. These have significant ongoing effects and repercussions on political, cultural, and socio-economical dimensions of people’s life in HK, bringing the notion of identity to the fore. The complexity in identity and identification is once again under the spotlight during recent tension between part of the population and the local and central Chinese government. The social-political volatility has arguably led to a new wave of emigration and a renewed attempt by the HK government to attract talents both from mainland China and abroad. For instance, Ng (2022) reported that Hong Kong’s Chief Executive John Lee in July 2022 announced government plans to set aside 30 billion HK dollars to launch the ‘Top Talent Pass Scheme’ to entice talents outside HK to pursue their careers in the city. All of these factors contribute to the complexity in (and perhaps impossibility of) pinning down the notions of identity and affiliation in the context of Hong Kong (see Leung & Lee, 2016).

The language environment in Hong Kong is also rich and intricate with the vivid presence of Cantonese, English, and Mandarin which function as the official languages in various domains of public and private life. We can classify the linguistic context in HK as diglossic/ multiglossic, where two or more varieties/ languages are used for complementary purposes in the same society (Ferguson, 1959; Jaspers, 2017). Cantonese is the mother tongue and home language of around 96% of the population (Census and Statistics Department HK, 2022; Evans, 2013). It is also the medium of instruction in Chinese as a Medium of Instruction (CMI) primary and secondary schools. English is a second language and remains important in the city (Bolton et al., 2020), partly due to its status as a lingua franca for global as well as local communications (cf. Sung, 2023). English is the medium of instruction in English as a Medium of Instruction (EMI) primary and secondary education as well as the medium of instruction in many tertiary education programmes. While Mandarin is the national language which has seen an increasing presence especially in state media. The three main/ official languages are accompanied by languages of the ethnic minoritised groups in the population such as Tagalog and Indonesian, the languages of the foreign domestic helpers from the Philippines and Indonesia, as well as other Chinese ‘dialects’ such as Hakka, Fukien and Chiu Chau (Census and Statistics Department, Hong Kong, 2022).

Language planning and associated language policies in postcolonial HK have undergone multiple rounds of fine-tuning and modifications (Chan, 2014). This includes the proposed adoption of a compulsory mother-tongue policy at junior secondary level soon after the return of sovereignty to China in 1997. The policy is subsequently ‘fine-tuned’ partly due to outcry from parents who wish for their children to continue be educated in English. In some primary and secondary schools, Mandarin has replaced Cantonese as the medium of instruction in Chinese subjects (Evans, 2013). Although the three languages co-exist and perform distinct functions in a largely multiglossic manner, e.g. Cantonese for everyday communications, English for education, Mandarin for state ceremonies, some have discussed the potential tension among them and even threat that one language poses to the other. For example, the vitality of Cantonese might be under threat due to the introduction of Mandarin as a medium of instruction (Li, 2018). The nature of this article precludes an extended discussion of the linguistic situation of HK, but readers interested in the topic can refer to Ng and Cavallaro (2019) for a recent overview.

It is against this backdrop that we collected data to examine how native Cantonese-speaking informants identify themselves in such a complex geo-socio-political and linguistic environment and whether or not their identification is in any way related to their perceived accentedness in languages that represent the past coloniser, English, and the current governor, Mandarin.

4. Study design

Data were collected from 65 native Cantonese-speaking informants (born in HK) ranging from the birth years of 1970–1979 (n = 22), 1980–1989 (n = 21), and 1990–1999 (n = 22).² This grouping was designed to investigate potential generational differences among the population who have experienced British colonialism and post-colonialism in HK. We adopted this grouping/ categorisation initially as they correspond to local scholars’ classification of the HK population (Lui, 2007 in Cheung, 2014). That to us signals a degree of ecological validity, as these categorisations fall within the repertoire of the general public, commonly circulated and utilised in popular discourse in the current research context. Yet, we also wish to point out that categorising participants according to these age groups, which are hard-and-fast, pre-existing categories, can be problematic as we subscribe to the view that identities and identification are complex, fluid and multidimensional constructs (Norton, 2013; Trofimovich & Turuševa, 2015). In fact, as shall be seen below, participants in the various age groups do not actually differ significantly in their EGA scores. Participants who were born in HK and with Cantonese as their mother tongue were also asked to classify themselves into one of the following: Hong Konger (n = 28), Hong Kong Chinese (n = 4), Chinese Hong Konger (n = 18), Chinese (n = 13) or Others (n = 2).³

To tap into their sense of identification, we adopted the EGA questionnaire which was developed to examine how a varying degree of ethnic affiliation and identification in the Quebec context would affect learners’ level of attainment in L2 French/ English. Researchers were able to establish that EGA scores or a general sense of belonging to/ affiliation with the target language and its community positively correlate, to an extent, with how proficient a person is in the L2, as well as how accented they are deemed to be, e.g. the more they identify with the French ethnic group, the less accented they sound in L2 French (see literature review above). The EGA questionnaire, adopted from Gatbonton et al. (2014), contains 93 9-point Likert scale self-rating statements which pertain to five main themes (see 1–5 below) alongside some sub-themes.

1. depth of involvement in the ethnic group;

2. pride in, familiarity with, and feelings of comfort with the group;

3. perception of the place of the ethnic group in relation to other groups;

4. perceptions of the group’s vitality, and

5. views towards the socio-political concerns of the group.

The higher the rating, the stronger a participant identifies with the theme in question. To explore the potential relationship between EGA scores and participants’ self-reported proficiencies, which is operationalised as self-rated accentedness in this study, in English and Mandarin (9-point scale: 1 = heavily accented, 9 = no accent at all), we conducted correlation analyses according to self-categorisation groupings reported above. It should be noted that we have considered obtaining participants’ scores in standardised language tests such as the International English Language Testing System (IELTS) and the Hanyu Shuiping Kaoshi (HSK – Chinese Proficiency Test), however, we were unable to collect such data especially from those in the oldest group since such tests were not common nor compulsory at the time in which they completed their formal education. We have, therefore, followed previous studies in using self-ratings as a proxy for linguistic proficiencies (e.g. Gatbonton et al., 2005; Gatbonton & Trofimovich, 2008; Trofimovich et al., 2013). Informed by the findings from previous EGA studies, it is predicted that (a) participants who identify more with the Chinese identity will consider themselves to be less accented in Mandarin, which is the national language, (b) those who identify more with the Hong Kong identity will consider themselves to be more accented in Mandarin (due to the current antagonistic relationship between some in Hong Kong and China). Further to that, based on the assumption that English is seen as an integral part of HK identity (see Hansen-Edwards, 2015 for a discussion), it is anticipated that (c) informants who identify more with the HK identity will also see themselves as less accented in English, on the flip side (d) those who identify more with the Chinese identity will consider themselves to be more accented in English, the language of the past coloniser, which can be seen as an oppressing language that conflicts with or even undermines the development of local/ ethnic identity (cf. Le Page & Tabouret-Keller, 1985).

5. Findings

In order to explore the potential generational difference, we first conducted a Kruskal–Wallis test for the EGA means that participants assigned. This analysis revealed that EGA ratings assigned by each age group were not significantly different from one another (p = 0.128). Hence data were then analysed solely according to how participants categorised themselves, i.e. Hong Konger, Hong Kong Chinese, Chinese Hong Konger, Chinese or Others. To enhance the statistical power, we have opted to aggregate the data by collapsing sub-groups that logically belong to one another before conducting our inferential statistical analyses. Given the semantic prominence of ‘Hong Kong’ in the categories, Hong Konger and Hong Kong Chinese, they were aggregated into one group, resulting in a ‘Hong Kong’ group (n = 32); while Chinese Hong Konger and Chinese were combined to form another ‘Chinese’ group (n = 31) for analyses, as ‘Chinese’ appears to be more prominent in these latter two categories.⁴ Participants who identified as ‘Others’ were excluded from further analyses.

Pearson correlations were employed to determine the potential relationship between participants’ self-rated language accentedness scores and the mean EGA scores they assigned under each of the five sub-themes. Below, we first describe the findings obtained for the HK group before moving on to the Chinese group.

The descriptive and correlation statistics for the HK group can be found in Table 1. Analyses indicate that aspects of the EGA scores are significantly correlated with participants’ self-rated Mandarin accentedness. Specifically, theme 2, ‘pride in, familiarity with, and feelings of comfort with the group’, and theme 3, ‘perception of the place of the ethnic group in relation to other groups’, of the EGA scores negatively correlated with self-accent rating in Mandarin with a medium correlation (r = –.401) and a small-medium correlation (r = –.375), respectively (see Plonsky & Oswald, 2014). Concurring with our prediction above (see (b)), the negative correlations suggest that participants in this HK group who had higher EGA ratings in the two themes consider themselves to be more accented in Mandarin (note back the scaling of the accentedness scores where 1 = heavily accented, 9 = no accent at all). In other words, the stronger participants feel pride, familiarity and comfort with their ethnic group (the Hong Kong identity), which relates to theme 2 in the EGA questionnaire, the more accented they see themselves in Mandarin. Moreover, participants in the HK group who have a stronger perception of the place of their ethnic group (i.e. HK group), which is related to theme 3 in the EGA questionnaire, see themselves as more accented in Mandarin. These findings mirror those reviewed above from the Quebec context, to an extent, in that learners who have a stronger Francophone identification were judged to be more accented in L2 English (Gatbonton et al., 2011). Similarly, our findings also resonate with that found in the Latvian context where Latvians’ stronger identification with the Latvian identity is associated with weaker self-perceived L2 Russian proficiency (Trofimovich et al., 2013). Given the nature of questions in these two themes, it is plausible that informants view Mandarin antagonistically, potentially treating it as a possible threat to their HK identity (cf. Evans, 2013 and Li, 2018).

Table 1. Means and correlations between 5 EGA themes and self-rating in languages for the HK group.

Means in Mandarin	Theme	Means	Correlation with rating in L2 Mandarin	Correlation with rating in L2 English
5.00 (1.97)	1	5.07 (1.22)	No sig. correlation	No sig. correlation
	2	5.77 (1.20)	r = –.401*	No sig. correlation
Mean in English	3	4.84 (1.17)	r = –.375*	No sig. correlation
	4	3.77 (1.03)	No sig. correlation	No sig. correlation
5.66 (1.84)	5	5.65 (0.96)	No sig. correlation	No sig. correlation

Notes: () = standard deviation; *p ≤ 0.05.

On the other hand, going against our prediction in (c), no correlations were uncovered between EGA ratings of this group and the self-rating of their L2 English, which is perhaps surprising given the discussion in the background section about the importance of English and identity construction in HK. Though it should also be noted that the lack of observed relationships between EGA and language ability has indeed been reported in some contexts as mentioned above (Tekin, 2019). We will explore this particular finding further below once we have also presented the findings for the Chinese group.

Table 2 details the findings for the Chinese group. As seen, for this group, EGA scores did not seem to be associated with their self-rated Mandarin ability. This contradicts our prediction in (a), where we anticipate a positive link between the Chinese identity and the proficiency of Mandarin, the national language. However, a significant positive correlation is found between EGA scores in theme 3 and English (r = .431; medium correlation). This indicates that the more securely participants feel about the Chinese group, as theme 3 pertains to perception of the place of the ethnic group in relation to other groups, the less accented they consider themselves to be in English, which runs contrary to our prediction (see (d) above). One possibility is that this finding reflects participants’ confidence of the position of the Chinese group in the HK society, so much so that being proficient in English is not seen as a threat (cf. post-colonial contexts (Le Page & Tabouret-Keller, 1985)).

Table 2. Means and correlations between 5 EGA themes and self-rating in languages for the Chinese group.

Means in Mandarin	Theme	Means	Correlation with rating in L2 Mandarin	Correlation with rating in L2 English
4.90 (1.90)	1	4.86 (1.29)	No sig. correlation	No sig. correlation
	2	5.47 (1.37)	No sig. correlation	No sig. correlation
Mean in English	3	4.82 (1.46)	No sig. correlation	r = .431*
	4	3.71 (1.26)	No sig. correlation	No sig. correlation
5.35 (1.78)	5	4.82 (1.29)	No sig. correlation	No sig. correlation

Notes: () = standard deviation; *p ≤ 0.05.

6. Discussion, conclusion and future directions

Broadly anchored on the sociocognitive paradigm (Batstone, 2010) we conducted our study in Hong Kong. Modelling after the series of investigations pertaining to EGA and language proficiencies by Gatbonton and her colleagues (e.g. 2005, 2008, 2011, 2014), we aim to explore whether social factors, in particular, ethnic group affiliations are related to self-perceived language accentedness in English and Mandarin among three generations of Cantonese-speaking and HK-born participants in our research context. As one of the few studies that examine a multiglossic context where there is a complex interplay among the three official languages (i.e. Cantonese, English, and Mandarin), we hope to extend our understanding of the potential relevance of EGA in accentedness and language learning beyond what was previously established in the literature vis-à-vis binary, diglossic situations. By conducting correlation analyses, we were able to identify potential relationships between aspects of the EGA and informants’ self-reported accentedness rating in the two target languages. How participants identify themselves and their EGA scores seem to be related to how proficient they perceive themselves to be in languages targeted in this study, at least to an extent. However, the nature of the relationship between the two constructs is more nuanced than predicted, especially in comparison to some previous studies which are conducted in mostly binary contexts where two languages exist and function in the societies (e.g. French–English; Latvian–Russian).

In agreement with existing literature, we were able to establish some links between aspects of the EGA and participants’ self-evaluated language proficiency, which was opertaionalised as accentedness rating in this study. For those in the HK group, their EGA scores in theme 2, ‘pride in, familiarity with, and feelings of comfort with the group’, and theme 3, ‘perception of the place of the ethnic group in relation to other groups’, significantly correlated with self-reported accent rating in L2 Mandarin negatively. This concurs with our prediction that the stronger one feels affiliated to an HK identity, the more accented they deem themselves to be in Mandarin. Despite being the national language since the sovereignty was returned to China, some of our participants might not have learnt Mandarin formally since it was not a compulsory subject in HK before the handover (e.g. for the oldest group). For those who have learnt it, they would most likely have learnt it as a foreign language with limited classroom input of a few hours per week at maximum for a few years in late primary education or early secondary education. This low level of exposure compounded by the potentially antagonistic view that some participants might hold against the ‘imposed’ national language, which had not been part of their identity until the sovereignty was returned could be the reason why we have found a negative correlation between aspects of EGA and their L2 Mandarin self-rating. It may also be relevant to acknowledge the fact that our data was collected during a period marked by socio-political instability and turmoil, coinciding with the massive protests for more local autonomy when anti-mainland sentiment is arguably rife (e.g. Lowe & Tsang, 2018). Therefore, one could postulate that the negative correlation observed is attributable to a depreciated evaluation of participants’ Mandarin ability as a defence mechanism for the sake of preserving/ bolstering their Hong Kong identity (see also Hansen-Edwards, 2020 for a recent examination of the construction of linguistic identities in this time of significant political tension).

The only other significant correlation observed happens to be in an unexpected direction. Namely, for the Chinese group in this study, the more secure they feel about the ethnic Chinese group they are, the less accented they consider themselves to be in L2 English. According to existing literature, the more secure one feels about their identity the more likely they would consider themselves inferior in ‘other’ languages, in this case, English, which was the language of the past coloniser. As discussed above, this could be signalling that the group identity is so secure that they do not feel threatened by the language of ‘the other’. Alternatively, one could also speculate that, given the importance of English as an ASEAN and global lingua franca (see Kirkpatrick, 2020), English has become part and parcel of a cosmopolitan Chinese identity where proficiency in English is seen as an important asset.

The lack of relationship observed in many other instances for both groups in the direction predicted is also worth commenting on. Given previous literature on the importance of English in HK and arguably to HK identity, the lack of relationship observed between English and EGA scores among the HK group is perhaps surprising (cf. prediction (c)). It is possible that this finding suggests that English ability is regarded as independent of one’s ethnic identification, hinting at the possibility that English is learnt for reasons other than integrative ones, which are normally associated with the desire to integrate into or identify with the target culture/ population. On the other hand, the lack of relationship between EGA scores and self-perceived accentedness in L2 Mandarin among the Chinese group is equally puzzling (cf. prediction (a)). It could also be the case that Mandarin like English is seen as a second or foreign language that has functional values, e.g. for enhancing career prospects, rather than integrative ones, e.g. signalling identity. In fact, we find support for this line of reasoning in a large-scale study of over 500 tertiary students in HK by Humphreys and Spratt (2008). Their study reported a strong instrumental orientation to the learning of both English and Mandarin (Putonghua) in the context of Hong Kong when compared to other foreign languages on offer such as French, German and Japanese.

As explained in our study design section, we were unable to obtain standardised language test scores from some of our participants, which might have impacted our results. Future studies should try to gather such information or actual linguistic performance data as far as practically possible to verify our findings. In addition, it will be useful to conduct follow-up studies, perhaps of a larger scale with more participants, over time. Since identity and by extension EGA is a dynamic and fluid construct that interacts with the socio-political-economic context (cf. The Douglas Fir Group, 2016, 2019), our findings are therefore likely a manifestation of the specific circumstances around which the data was collected. As the social political milieu evolve, how participants in the research context identify themselves may also reasonably evolve.⁵ Hence, it will be useful to track such changes and investigate whether the relationship we reported in our study between EGA and language proficiency will change over time as well. We very much hope that our work as one of the earlier studies that looked into a multiglossic context in Asia has helped further knowledge in the study of EGA. Future studies could continue to examine settings where more than two languages are at work or in competition to enhance our understanding of the relationship between EGA and S/FLA. We conclude by calling for other researchers to join us in trying to tackle the challenge of bringing together the social and cognitive dimensions into their research so as to further augment our understanding of the field of SLA.

Acknowledgements

We would like to sincerely thank the late Prof. Liz Gatbonton for generously sharing her EGA questionnaire with us, which has enabled the conduct of the present research reported in this paper. We hope that our work will help continue her legacy in this line of pioneering work around EGA. We would also like to thank the audience of the Japanese Society for Language Sciences 2017 conference and the audience of the Language, identity and education in multilingual contexts (LIEMC19) conference for their constructive feedback. We also like to thank the editor and reviewers of Language, Culture and Curriculum for their input on the earlier version of this paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 It is noteworthy that studies in related disciplines such as sociolinguistics have also discussed the relevance of ‘the social’ in language use. For example, whether one identifies with their interlocutor might affect how they would diverge from or converge with their interlocutor’s speech as stipulated in the communication accommodation theory (Giles & Ogay, 2007). There is also evidence from sociophonetics that speech perception can be affected by as small an artefact as a stuff animal that represents a specific country (e.g. stuffed toy kangaroos and koalas which are associated with Australia or stuffed toy kiwis which is associated with New Zealand) (Hay & Drager, 2010).

2 The project received ethical approval from the first author’s institution. Participants were given a brief participant information sheet that explains the purpose of the project. Their consent was sought prior to the commencement of the project. They were also given the opportunity to withdraw at any time, no one withdrew.

3 Other self-nominated categories were Han dynasty person and Asian.

4 The group aggregations were further confirmed by 10 randomly selected participants who consider them to be legitimate.

5 In fact, the relevance of the socio-political milieu and the temporal element is so salient that one of our participants explicitly commented on it in the larger study about identity (Leung & Lee, 2016) without being prompted.