Korean Children’s Attitudes toward Varieties of English: The Role of Age and English Learning Environment

: This study investigates young Korean children’s attitudes toward three English varieties: American English (AmE), Singapore English (SiE), and Korean English (KoE). A total of 42 Korean children participated in this study. For data analysis purposes, the results were categorized according to the children’s age and their experience of exposure to formal English learning. In addition to this, 30 Singaporean children were also involved in the study, and their results were compared with the results of the younger group of Korean children. A mixed methodological approach, which included a modiﬁed verbal guise technique appropriate for use with children and semi-structured interviews, was also adopted. The results show that 5-year-old Korean and Singaporean children do not prefer one speciﬁc variety of English more than the other varieties of English. However, this was not the case for 12-year-old Korean children. These older Korean children preferred AmE and SiE more than KoE, and the “speaker’s pronunciation” was considered to be the critical feature in determining these attitudes. The ﬁndings suggest that Korean children’s developing attitudes toward a particular variety of English emerge sometime during their elementary school years. in their English class (Jo et al. 2017). This study aims to explore the emergence of Korean children’s attitudes toward three varieties of English, including American English (AmE), Singapore English (SiE), and Korean English (KoE). To do this, we employed a mixed methodological approach and compared the attitudes toward the three English varieties of 5-year-old Korean children to 5-year-old Singaporean and 12-year-old Korean children.


Introduction
English is the most widely spoken global lingua franca. It is used in more than 100 countries and considered one of the most hybridized and rapidly changing languages, reflecting various countries' languages and cultures (Jenkins 2014;Schneider 2003Schneider , 2014. Despite the wide global use of several English varieties, both English education providers and learners in Korea have prioritized teaching the 'standard' variety of English (i.e., American or British), referring to it as native English (Ahn et al. 2020). In general, Korean adults are most familiar with American English (AmE), perceiving it as the most potent variety of English and preferring to learn it over other forms of English (Ahn 2017;Byun 2016;Yoon 2007). Korean English teachers in elementary schools also have this paradoxical perception, agreeing that there is a need to recognize the importance of accepting different varieties of English but at the same time that it is more effective to focus on teaching one particular variety of English, such as AmE, for practical reasons (Seo 2016). Likewise, 11 and 12-year-old Korean students prefer 'American-accented English' in their English class (Jo et al. 2017). This study aims to explore the emergence of Korean children's attitudes toward three varieties of English, including American English (AmE), Singapore English (SiE), and Korean English (KoE). To do this, we employed a mixed methodological approach and compared the attitudes toward the three English varieties of 5-year-old Korean children to 5-year-old Singaporean and 12-year-old Korean children.

Language Attitudes
Language attitudes are influential and integral to our communicative competence, which deeply permeates our daily lives (Hymes 1972). The level of a learner's success can be determined by his/her attitude to the target language (Dörnyei et al. 2006). Despite the vital role of attitudes toward language learning, a definitive concept of attitudes has not been readily agreed upon. Definitions vary in the degree of elaboration and the weight given to different features of attitudes (Garrett 2010). In this paper, attitude is seen as an evaluative orientation to a social object of some sort, such as language (Sarnoff 1970;Thurstone 1931).
Language attitudes, in particular, are often understood by three components: cognitive (socialized thoughts and beliefs), affective (a person's feeling), and behavioral (the predisposition of a person's actions) (Agheyisi and Fishman 1970). Children influenced by a teacher to think of a particular language or culture in a certain way would be an example of the cognitive component. An example of the affective component would be South Koreans responding positively or negatively to an Indian English speaker because they felt 'pleasant' or 'unpleasant' towards the speaker's unrecognizable accent. Lastly, South Korean mothers saving money for their children's English education would be an example of the behavioral component. The questions used in our study to measure children's attitudes toward varieties of English included both the cognitive (e.g., Who do you think is smarter and would be good at studying?) and affective (e.g., How were the English speakers today?) components.

Measuring Language Attitudes
To date, language attitudes have been investigated through direct (e.g., interview, questionnaire) and indirect (e.g., verbal guise technique) methodological approaches. When the direct approach is employed, the respondents are explicitly asked to provide their opinions about the language (Galloway 2017). While this method allows researchers to measure respondents' expressed attitudes, which may be thought to provide more reliable responses, respondents may also choose not to provide honest answers. Such a limitation could be solved by using an indirect approach, which is a less explicit form. The respondents are not precisely aware of what is being rated and thus become less sensitive to 'reflection and social desirability biases' (Cargile et al. 1994).
The Verbal Guise Technique (VGT), used in this study, is one of the most widely used indirect approaches in language attitudes studies. It is a modified version of the Matched Guise Technique (MGT). In both techniques, the respondents are required to hear several versions (e.g., English spoken by different speakers) of (an) audio-tape recording(s) of the same text and then rate the extent to which they agree or disagree with the issue in question using the attitude-rating scales that are either five-point or seven-point scales. While the MGT involves a single speaker in recordings, the VGT involves multiple speakers (Galloway 2017). Thus, when the VGT is utilized, researchers cannot rule out the effect of individualistic properties of different people's voices (Bradac et al. 2001). Despite its limitation, it has the advantage of allowing researchers to examine respondents' language attitudes towards more than two or three varieties (Galloway 2017). Both the direct and indirect methods, also known as the mixed methodological approach, were employed in our study to mitigate the weakness and synergize both methods' strengths. model has received criticism for its limitations in depicting the reality of current Englishspeaking contexts in recent years (Marlina and Ahn 2011;McKay and Bokhorst-Heng 2017), many studies have continuously and repeatedly employed the concentric model as an adequate framework to categorize English varieties. Therefore, the terms IC, OC, and EC are employed in this study as defined by Kachru: English spoken by Americans was recognized as IC, English spoken by Singaporeans was recognized as OC, and English spoken by Koreans was recognized as EC.
Past studies have found that AmE and British English or BrE (referred to as IC English varieties) tend to receive higher evaluations than other English varieties in many countries (Ahn 2017;Zhang 2009). For example, Taiwanese university students attending an English intensive course in the Philippines regarded English spoken in North America as standard English (Kobayashi 2008). Japanese, South Koreans, and Malaysians favored IC English varieties more than their own varieties of English (Tokumoto and Shibata 2011). Japanese English teachers and students saw IC English varieties as the only correct English varieties. They perceived the Japanese variety of English either as incorrect English or an English that has strayed from the real English (Matsuda and Friedrich 2011).
Similar attitudes are also found in Korea. According to Ahn (2017), Korean English teachers believe that AmE is the only variety of English fit to be taught as it best represents the English language in the modern world. Korean adults also showed more positive attitudes toward IC varieties of English speech than EC or OC English varieties (Ahn 2017;Chung and Kim 2017). In a popular experiment conducted by the Educational Broadcasting System (EBS), Korean adults held negative attitudes towards Ban Ki-moon's (the UN secretary-general) speech when they did not know who the speaker was. Even without understanding the meaning of the speech, Korean adults described it as 'unstylish' and 'not fluent'. Meanwhile, IC speakers applauded the speech for the 'use of very highlevel vocabularies', 'the clear sentence structure', and 'the directness of message delivery' (EBS Production Team 2018). It seems clear that IC English varieties are significantly more favored than other varieties (Garrett 2009;Liou 2010;Matsuda and Friedrich 2011;Yoon 2007) in many EC countries, such as China, Japan, Korea, and Taiwan, where the norms of English varieties largely come from external sources (Kachru 1985, pp. 16-17).

English Education for Korean Children
Based on Kachru's (1982) 'Concentric Circles' model, Korea is an EC country, and English is considered to be a foreign language. However, the symbolic power of English in Korea carries more weight than any other foreign language. English is considered as a critical tool to make the dream of 'South Korea's ambition to be an economic world leader' come true (Bok 1998).
In 1997, English became part of the national curriculum in Korea starting at Grade 3 (9-year-old students) under the 'National English curriculum policy (NECP)'. Since then, the national pedagogical method has been transformed from a grammar-translation model to communicative language teaching. With the goal of increasing the opportunities for children to learn English from teachers who speak the "standard" variety of English, more English teachers, named 'guest teachers', were invited from IC countries through the 'Teach and Learn in Korea' (TaLK) English immersion plan that the government implemented with a budget of US$4 billion over five years. One of the TaLK program goals was to support a public education policy that aimed to help children communicate with foreigners without difficulty by the time they graduated from high school by placing at least one "native" English teacher in every school nationwide and at all school levels.
The obsession of Koreans with the English language has been comprehensively documented (Seth 2002;Park 2009), and this obsession begins with children's first contact with education. Although the Korean English education policy states that English education is not mandatory for children below Grade 3 (8 or 9-year-olds), having some English proficiency is recognized as essential once children enter elementary school (Lee and Lee 2018;Park 2009;Seth 2002;Shim and Park 2008). Most early childhood education institutions Languages 2021, 6, 133 4 of 18 recognize that providing an English education is an essential part of their curriculum to meet the demands of Korean parents for their children to be exposed to the English language as early as possible. It was also noted that enrolment numbers could depend on whether or not the institution offered English education (Jun 2009).
For instance, out of 159 kindergartens, more than 95.6% offered English education (Choi and Son 2011), and almost 80% of daycare centers were conducting English education classes (KICCE 2015). Meanwhile, the enrolment rates of 3 to 5-year-old Korean children in kindergarten and daycare programs (excluding other forms of institutions, such as hagwons) in 2016 were at 93% (OECD 2018). Thus, we can expect that most Korean children, by the age of five, attend an early childhood institution and that a majority of them start learning English before they go to elementary school. Okumura (2005) suggests that English learners' attitudes toward English may develop at an early stage because children in elementary school also show more positive attitudes towards AmE than other varieties of English. However, previous studies related to English learner attitudes toward English varieties are very much limited to adults. Very little is known about the attitudes of young children (below seven years) toward varieties of English.

Present Study
Since a great number of Koreans begin learning English at a very young age, examining young Korean children's attitudes toward English varieties may provide valuable insights into the extent to which young Korean students reveal their favored attitudes towards AmE or "native" varieties of English. One way to do this would be to examine how young children perceive English varieties. Do young children's attitudes toward varieties of English differ according to their English learning experience? To what extent do young Korean children hold a similar attitude towards English varieties as adults before they enter elementary school? If young Korean children perceive IC (AmE) as 'better' English than OC (SiE) and EC (KoE), is it because they are in the EFL context where AmE is highly valued? If this is not the case, then what would be the possible reason for such a perception? To answer these questions, Singaporean children and older Korean children (12-year-olds) were recruited separately as a comparison group. Singaporean children were chosen because Singapore is one of the Asia Pacific regions and, thus, shares Asian characteristics with Korean children. However, because English is designated as an official language in Singapore, Singaporean children experience different English learning environments than Korean children.
This study addresses three research questions: (1) Do 5-year-old Korean children who attend after-school English lessons in kindergarten show a different preference for certain varieties of English than those who do not attend? (2) Do 5-year-old Korean and Singaporean children show a different pattern of choice for certain varieties of English? (3) Do 5-year-old and 12-year-old Korean children show any preference for different varieties of English? Why or why not?

Methodology
The study employed a mixed methodological approach. The widely used VGT was modified to be appropriate for young children, and an interview was also conducted. Direct and indirect measures of language attitudes were analyzed separately and then compared to derive conclusions about when and why Korean children developed a preference for certain English varieties. Languages 2021, 6, 133 5 of 18 girls = 44.1%) spoke Korean as the primary language at home and in their kindergarten. All participants were from one kindergarten located in Seoul, South Korea. Children spent between seven to nine hours per day at the kindergarten. Out of 34 Korean children, 19 children were also attending the after-school English lesson offered by the kindergarten. The English class ran three days a week for 40 min. Fifteen children out of the total number enrolled did not attend either the kindergarten's after-school English program or a private educational institute (hagwon) to learn English. All children were born in Korea, and none of them had lived in another country for any length of time. Singaporean children were from three different kindergartens in Singapore (boys = 50%; girls = 50%). The kindergartens' teachers spoke English, and all children spoke English fluently and had a high level of comprehension.

Participants
Interview. To address the third research question, four 5-year-old and four 12-year-old (6th grade elementary) Korean children were additionally recruited. All children had no reported history of speech, hearing, or language disorders. It should be noted that the interview was conducted only with these two groups of Korean children. Because our study's primary purpose was to investigate Korean children's emerging attitudes toward English varieties, we did not go further with interviewing the Singaporean children.

Data Collection Method
Modified verbal guise technique. The VGT was simplified to minimize the length considering participants' age and their ability to concentrate on the task. Rather than asking the children to rate each utterance on a five-or seven-point scale for each question, we played two recorded voices for children to hear first. Then, children were asked to choose one speaker that corresponded to the question (e.g., Who do you think is more friendly and is nice to a friend?). A similar method was employed in Rutland et al.'s (2005) study where racial bias was measured in children (aged between three and five years). In their study, children were presented with four images of children of different races and were asked to match diverse adjectives (i.e., kind, clever) by asking the question, "who is friendly? Is it the Black child, the White child, the Asian child? . . . ".
The three 'adjectives' used in our study were 'smart' (knows a lot), 'rich' (has a lot of money), and 'friendly' (nice to friends). All adjectives were positive. The terms 'smart' and 'friendly' have been used in adult studies (e.g., Ahn 2017; Chung and Kim 2017;McKenzie 2008). The adjective 'rich' was included because we believed that this term is directly related to measuring someone's socio-economical position .
Interview. The four 5-year-old and four 12-year-old Korean children were encouraged to express their opinions freely for the follow-up interview question in addition to the indirect questions mentioned above. While 5-year-olds were individually asked to respond to each item (i.e., "Who do you think is smarter and would be good at studying") followed with an additional question ("Why do you think so"), a group interview was conducted at the university lab for the 12-year-olds. For 12-year-olds, five additional questions were asked to investigate their thoughts more deeply. The questions were as follows: (1) What do you consider most important when someone speaks English?
(2) How were the English speakers today? (3) Why do you think you can recognize the difference between the speakers? (4) How do your other friends speak English? (5) What do you think of Korean English?

Materials
Utterances. The Speech Accent Archive website (http://accent.gmu.edu/ accessed on 25 October 2018) was used to retrieve the utterances. To control for potential confounding variables, all recordings were from the same gender (female) and a similar age range (the 20s-30s). All speakers of each recording had a typical English language learning experience as described in Kachru's Concentric Circles model: all speakers were born in their native country. Furthermore, speakers from the United States and Singapore learned English naturally since birth (non-EFL context), and Korean speakers learned English as a foreign language in constructed academic settings.
The authors listened to all AmE, SiE, and KoE utterances that met the criteria (Table 1) and selected two SiE and two AmE utterances. We could not find an appropriate KoE speaker in the archive, so we recruited two female Korean speakers and recorded the same statement. Among the six different English speakers, one SiE speaker spoke Mandarin and Hainanese, and another SiE speaker spoke Cantonese, Mandarin, and Japanese.
The original utterances were around 15 s long, but only the first five seconds of the sentence were used for the study. All the speakers said: "Please call Stella. Ask her to bring these things with her from the store: six spoons."

Procedure
Modified verbal guise technique. The experiment was conducted at the children's kindergarten. All children were individually tested in a quiet room, separate from their regular classroom. The location helped to ensure that the participating children felt comfortable in the environment. Children were told that their teacher would stay with them in the room if they wanted them to, but none chose to stay with their teacher. Sessions took approximately six minutes per child.
A researcher sat beside the child facing a laptop screen. First, we explained why we were conducting this study and their right not to participate. In addition to informed parental consent, all children orally agreed to participate in the study. Second, the study procedure was explained. Children were told that (1) they would hear two female adults saying something in English, that (2) they would be asked to choose either of the speakers as their response to the questions, that (3) it was ok if they did not understand what the speaker was saying, and (4) that there were no right answers to the questions presented. The practice question: Who is the male? was given before and after hearing a female and male speaker reading a sentence not related to the statement used.
On the screen, two blue cards with the numbers '1' or '2' appeared using PowerPoint. Once an utterance was played, either the number '1' or '2' card changed from blue to orange (Figure 1). For example, when the first utterance was played, the number '1' card changed to orange. Once the second utterance was played, the color of the number '2' card changed to orange. The question was asked before and after utterances were presented. Children were asked to choose one speaker based on the following three questions: Who do you think is smarter and would be good at studying? Who do you think is more friendly and is nice to a friend? Who do you think is richer and has a lot of money? This procedure was repeated until each child answered all the questions. For each question, children heard the following series of utterance combinations: AmE vs. KoE, AmE vs. SiE, and KoE vs. SiE. The order of different utterance combinations presented to children was counterbalanced in four different ways (Appendix A.1). friendly and is nice to a friend? Who do you think is richer and has a lot of money? procedure was repeated until each child answered all the questions. For each ques children heard the following series of utterance combinations: AmE vs. KoE, AmE vs and KoE vs. SiE. The order of different utterance combinations presented to children counterbalanced in four different ways (Appendix A.1). Interviews. The additionally recruited 5-year-old children for the interview als perienced the same procedure mentioned above (modified VGT). Afterward, they wer dividually asked to respond to each item (i.e., Who do you think is smarter and wou good at studying?) followed by an additional question (Why do you think so?). For 12-year-olds, a group interview was conducted. These children also responded to the ified VGT in addition to the interview questions. All children were encouraged to exp their opinions freely.

Coding
Each child was asked a total of nine questions. Each question type (Who do you t is smarter?/more friendly?/richer?) was asked three times repetitively with different of utterances (AmE vs. KoE, AmE vs. SiE, and KoE vs. SiE). A score of ʹ1ʹ was given time a child chose the utterance/speaker (AmE, KoE, or SiE) for the three compari (AmE vs. KoE, AmE vs. SiE, and KoE vs. SiE). Each utterance received 0 to 3 point item, 0 to 9 points in total. An utterance with a higher score indicated that children r the specific utterance as made by a smarter, friendlier, or wealthier person than othe terances.

Statistical Analysis
Using SPSS, a descriptive analysis was conducted to determine the mean and st ard deviations for all the measurements. The Shapiro-Wilk test was performed to c the normality of data and Friedman's test and a repeated one-way ANOVA were ducted to investigate if Korean and Singaporean children showed a significantly diffe pattern in rating the speakers.

Results
(1) Comparison between two groups of 5-year-old Korean children's attitudes: Enr vs. not enrolled in an English lesson Interviews. The additionally recruited 5-year-old children for the interview also experienced the same procedure mentioned above (modified VGT). Afterward, they were individually asked to respond to each item (i.e., Who do you think is smarter and would be good at studying?) followed by an additional question (Why do you think so?). For four 12-year-olds, a group interview was conducted. These children also responded to the modified VGT in addition to the interview questions. All children were encouraged to express their opinions freely.

Coding
Each child was asked a total of nine questions. Each question type (Who do you think is smarter?/more friendly?/richer?) was asked three times repetitively with different pairs of utterances (AmE vs. KoE, AmE vs. SiE, and KoE vs. SiE). A score of '1' was given each time a child chose the utterance/speaker (AmE, KoE, or SiE) for the three comparisons (AmE vs. KoE, AmE vs. SiE, and KoE vs. SiE). Each utterance received 0 to 3 points per item, 0 to 9 points in total. An utterance with a higher score indicated that children rated the specific utterance as made by a smarter, friendlier, or wealthier person than other utterances.

Statistical Analysis
Using SPSS, a descriptive analysis was conducted to determine the mean and standard deviations for all the measurements. The Shapiro-Wilk test was performed to check the normality of data and Friedman's test and a repeated one-way ANOVA were conducted to investigate if Korean and Singaporean children showed a significantly different pattern in rating the speakers.

Results
(1) Comparison between two groups of 5-year-old Korean children's attitudes: Enrolled vs. not enrolled in an English lesson The mean and standard deviation of total responses are summarized in Table 2, together with the absolute value of the skewness and kurtosis of each observed variable. According to the Shapiro-Wilk test, English-lesson-enrolled Korean children's ratings for SiE and KoE were normally distributed but AmE was not (p < 0.01). Thus, the Friedman test was conducted to determine whether these children's rating scores for SiE, KoE, and AmE differed significantly. The test rendered a non-significant Chi-square value (p = 0.74).
For Korean children not enrolled in English lessons, their ratings for SiE, KoE, and AmE were normally distributed, so repeated one-way ANOVA was conducted. The main effect of 'utterance type' did not significantly violate the sphericity assumption (X 2 (2) = 1.02, p > 0.05), so the F-value for the main effect of the speaker's nationality did not need to be corrected. These children's rating scores for each English variety did not differ significantly (F(2,13) = 0.18, p = 0.20). The results reveal that Korean children who participated in this study, regardless of their enrolment in the English class, did not perceive one English variety as superior to other varieties of English (Figure 2). effect of 'utterance type' did not significantly violate the sphericity assumption ( (2) = 1.02, p > 0.05), so the F-value for the main effect of the speakerʹs nationality did not need to be corrected. These children's rating scores for each English variety did not differ significantly (F(2,13) = 0.18, p = 0.20). The results reveal that Korean children who participated in this study, regardless of their enrolment in the English class, did not perceive one English variety as superior to other varieties of English (Figure 2).  (2) Comparison between young Korean and Singaporean children's attitudes towards three varieties of English The mean and standard deviation of total responses are summarized in Table 3, together with each observed variable's absolute value for skewness and kurtosis. The Shapiro-Wilk test revealed that Korean children's rating for SiE was normally distributed but not KoE (p < 0.01) and AmE (p < 0.001). Singaporean children's ratings for KoE and (2) Comparison between young Korean and Singaporean children's attitudes towards three varieties of English The mean and standard deviation of total responses are summarized in Table 3, together with each observed variable's absolute value for skewness and kurtosis. The Shapiro-Wilk test revealed that Korean children's rating for SiE was normally distributed but not KoE (p < 0.01) and AmE (p < 0.001). Singaporean children's ratings for KoE and AmE were normally distributed but not SiE (p < 0.001). Thus, the Friedman test was conducted for both groups. Korean and Singaporean children's rating scores for the three varieties of English did not differ significantly (Korean: p = 0.16; Singaporean: p = 0.77). In other words, Korean and Singaporean children who participated in our study did not prefer a specific English type over another (Figure 3). Table 3. Mean score (SD) of children's ratings for the Singaporean, Korean, and American English speakers. (3) Comparison between two age groups of Korean children: 5-year-olds vs. 12-year-olds In this section, the difference between 5-and 12-year-old Korean children's attitudes towards AmE, SiE, and KoE is investigated. In addition to the modified VGT, we attempted to examine the essential criteria or standards children imposed when evaluating each utterance through the interview. The number and percentage of each response are summarized in Table 4 and Figure 4, respectively. We did not go further with the statistical analysis as it was clear from the percentage that 12-year-old Korean children, compared with 5-year-old Korean children, perceived SiE and AmE more positively than KoE (Table 3, Figure 3). Unsurprisingly, 5-year-old children who participated in the interview also rated the three types of English similarly in the modified VGT, not showing a preference for any particular English variety.  (3) Comparison between two age groups of Korean children: 5-year-olds vs. 12-year-olds In this section, the difference between 5-and 12-year-old Korean children's attitudes towards AmE, SiE, and KoE is investigated. In addition to the modified VGT, we attempted to examine the essential criteria or standards children imposed when evaluating each utterance through the interview. The number and percentage of each response are summarized in Table 4 and Figure 4, respectively. We did not go further with the statistical analysis as it was clear from the percentage that 12-year-old Korean children, compared with 5-year-old Korean children, perceived SiE and AmE more positively than KoE (Table 3, Figure 3). Unsurprisingly, 5-year-old children who participated in the interview also rated the three types of English similarly in the modified VGT, not showing a preference for any particular English variety.

5-year-olds.
A summary of children's responses to the question: Why do you think so? is presented in Table 5 (for more details, please refer to Appendix A.2). The majority of 5-year-olds could not provide a specific reason for their choice of the speaker in answer to each question. Frequently, children failed to provide a clear standard or specific reason  Table 5 (for more details, please refer to Appendix A.2). The majority of 5-year-olds could not provide a specific reason for their choice of the speaker in answer to each question. Frequently, children failed to provide a clear standard or specific reason for their choice of answer (i.e., I just think . . . ) or they chose to simply re-state what was given in the question as a reason for their choice (i.e., because her voice is more friendly in answer to the question: Who do you think is more friendly and is nice to a friend?). Other times, the tone of the voice, not the pronunciation, was mentioned as the reason for their choice. Some responses were very creative but illogical (e.g., because it sounds like Dr. Smart went home to consult (someone's) sickness and provide medicine for the illness). 12-year-olds. The majority of 12-year-olds sensitively differentiated the three speakers' pronunciation and gave negative evaluations expressing disregard and skeptical attitudes toward KoE (Table 5, Appendix A.2). One child (12-year-old Boy B) distinguished the 'native English' by using the expressions 'British' and 'American'. Children perceived AmE as accurate (Girl B) and precise like their English teacher (Boy A/BoyB: she sounds like a teacher. Like the one in an English test). KoE was not seen as proper English (Girl A: Is it also English? Boy B: Pronunciation is weird; Boy A: She did not learn English well. Otherwise, it may be that she has a short tongue) and was seen as unprofessional (Boy A: She sounds like a wise wife and good mother). Meanwhile, SiE was perceived as a BrE (Boy B: I think she is British) and was described with terms such as 'stylish' (Girl A/Boy B: She is more stylish), 'professional' (Boy B: I think she will work in major companies like Samsung), and 'friendly with a good personality' (Girl B: I think she will have a good personality). Table 6 presents the 12-year-old children's responses to the five additional questions. Their responses showed clearly that all children valued pronunciation when speaking English. For them, fluency was also related to speed (for more details, please refer to Appendix A.3). Children's responses also reflected that the difference in their attitudes toward the varieties of English (compared with younger children) developed from their experience in a British or American-oriented English learning atmosphere. One child (Boy B), who had experience living in an English-speaking country and had had the opportunity to experience various English types, directly mentioned that his experience in Western countries and watching movies influenced his English pronunciation perception. Moreover, another boy (Boy A) said that the AmE speaker sounded like a teacher, which he heard on the school's standard English listening test.

5-year-olds. A summary of children's responses to the question: Why do you think so? is presented in
An unfavorable evaluation of Korean pronunciation was apparent. For instance, two boys disregarded their Korean friend's accents. However, children were still unsure as to whether or not Korean English should be accepted as English. While one child thought Korean English was inappropriate for international communication (Boy B), another child found it useful in communication (Boy A). Table 6. 12-year-old responses to the five additional questions.

Discussion
This study investigated the attitudes of children studying English in an EFL setting toward different varieties of the language by comparing the responses of 5-year-old Korean children with those of children from different English learning environments and different ages. First, the responses of two groups of 5-year-old Korean children were compared; those attending with those not attending the kindergarten's English class. The combined results of these children's responses were then compared with Singaporean children in the same age group, for whom English is the main language of communication. Finally, an interview was conducted to examine the differences in responses between 5-year-old and 12-year-old Korean children's attitudes and thoughts about varieties of English. Overall, our results demonstrate that Korean children's developing attitudes toward a particular variety of English emerge between 5 and 12 years of age.
It is interesting to note that 5-year-old Korean and Singaporean children with different English learning experiences and English language abilities held similar attitudes towards AmE, SiE, and KoE. This result is somewhat inconsistent with previous literature on older Asian children (Matsuda and Friedrich 2011;Okumura 2005) and adults (Chung and Kim 2017;Tokumoto and Shibata 2011) who preferred IC varieties of English speech. Unlike the adults, both Korean and Singaporean children did not perceive IC English as 'more friendly', 'powerful', or 'intelligent' than OC and EC English varieties. Moreover, children perceived OC and EC English speakers as friendly, rich, and intelligent to a similar extent. When young Korean children were asked to give a reason each time they chose a speaker, they were not able to provide clear, logical, or detailed reasons for their choice of answer. In other words, our findings reveal that young children in both Korea and Singapore have not yet formed an image of 'how English should sound like'. This result highlights the sobering fact that young children at the age of five have not yet been socialized to think like adults.
However, one could argue that the reason why Korean children could not provide explanations as to why they rated the speakers in a certain way may be due to the methodological error. More specifically, children may have not understood the adjectives (smart, rich, friendly) used in the study or did not clearly understand the rating task they were asked to do. However, this is very unlikely firstly because additional explanations for each term (e.g., nice to friends) were presented. Secondly, the three adjectives have been used in prior studies (Thomas 2004;Ramsey 1991;Rutland et al. 2005) with preschool children. No evidence of issues related to children failing to understand the meaning of the adjectives was reported. As far as the procedure is concerned, one may think that 5-year-olds are not yet capable of expressing their opinions consistently on abstract matters such as linking voices and accents towards characteristics of speakers. However, it has been reported that 5-year-old children's social preference was influenced by speakers' language and accent (Kinzler et al. 2009;Souza et al. 2013) and that children at this age were successful in matching a happy-sounding and a sad-sounding vocal affect to a corresponding emotional face (Berman et al. 2016). These studies provide evidence that 5-year-olds are capable of evaluating speakers based on the way they speak (i.e., accent, tone, voices). Furthermore, the researcher could not detect any signs of children's failure in understanding the task. During the experiment, all children listened to the recordings and chose the speaker carefully. On the way back to the classroom, another researcher, who escorted children back to their classrooms, asked some children what they did in the room (with the experimenter). No children showed difficulty in describing what they were asked to do (e.g., I listened to a English speaking voice and had to choose which one is more rich, friendly, or smart).
Another intriguing result obtained in this study is that Korean children by the age of 12 already held similar attitudes toward AmE, SiE, and KoE as those of Korean adults. They preferred KoE less than AmE and SiE. KoE was seen as improper and unprofessional English. The 'speaker's pronunciation' was considered the critical feature in determining their attitudes toward certain English varieties. Such responses are in line with the previous study that found that 11-year-old Korean children considered attaining native-like pronunciation as important (Kim and Ko 2014). It also corresponds to the findings that Korean teachers and parents perceived pronunciation as a vital part of English education (Kim 2008;Seo 2016). Moreover, it was clear from the interview that older children's attitudes toward different English varieties developed from their American-oriented English learning environments. One boy perceived AmE as an English that he heard in class. With the experience of living in America for a short time only, another boy felt that the British variety was more prestigious and looked down on Korean-accented English. He saw AmE as 'standard' English, BrE as 'good English' (assuming that SiE was BrE), and KoE as 'wrong English'. These views are similar to those of English teachers in Korea, as found in Ahn's (2017) study.
This finding supports Okumura's (2005) suggestion that English learners' attitudes toward English may develop at an early age. Looking at our study results, we can expect that the critical period of Korean children developing attitudes toward a particular variety of English is during the elementary school years when they begin to place value on 'appropriate English pronunciation (native)'. One possible reason is that children become more exposed to AmE as they enter school. They probably learn AmE not only in school (Grade 3 in public schools) but also in the afternoon classes, private institutions (hagwon), and also through the media. For instance, all schools offer 2 h/week of English education for third and fourth-grade students and 3 h/week for fifth and sixth-grade students. As children get older, they are exposed to more varieties of English when they travel abroad and/or attend overseas programs. This increases the likelihood of children being influenced by the attitudes of others (i.e., parents, teachers) and becoming socialized to think like their elders (prefer native English).
Another possible reason could be that younger children's English learning environment is different from those of older children. Because the purpose of young children's English education is to introduce English enjoyably, the pressure to speak in a native way or to construct a perfect sentence is absent. For instance, young children are highly encouraged to sing, chant, and listen to stories in English class (Kim and Yu 2012), but their performances are not graded. However, this kind of learning environment gradually changes as English becomes one of their school subjects. In the end, English proficiency is seen as a means to attaining status within South Korea's hierarchical society, and the focus of English education is on preparing students for English assessments (Song 2011). It is an undeniable fact that English education in Korea is still very AmE oriented. Therefore, as children get older, learning English switches from having fun and having an enjoyable experience of learning a foreign language to obtaining a higher American-oriented English assessment score.
For this reason, there needs to be a re-examination of the fundamental goals of elementary English education in South Korea. For instance, Korean English teachers recognize the importance of accepting varieties of English but at the same time emphasize the importance of focusing on one particular variety of English (e.g., AmE) in their teaching for practical reasons (Seo 2016). While the teachers are aware of the need to present an understanding of the world's varieties of English to their classes, the current English education atmosphere in Korea does not permit them to do so. Thus, it is necessary to enhance educational policymakers, teachers, and parents' awareness to develop positive attitudes toward the world's varieties of English.
This could be done through training workshops that introduce World Englishes perspectives and the relevant pedagogical implications. Teachers should learn about the poly-modal approach to implementing this awareness into their pedagogy rather than persisting with the mono-modal approach, emphasizing AmE as the only pedagogical model (Ahn 2017). By doing so, teachers will be exposed to varieties of teaching models and methodologies from countries of all three circles. This is essential as teachers' and parents' awareness and knowledge of English varieties will eventually be transmitted to children who would then learn about and have positive attitudes toward English varieties.
As we highlight our conclusions, we would like to point out that only a small number of children were involved in the comparison between two groups of 5-year-old Korean children's attitudes. The number of children in each group was less than 20 children. Thus, we believe that our result should be interpreted with caution and that there is a need for future study. Furthermore, it should be noted that the interview was only conducted with Korean children. Because our study's primary purpose was to investigate when Korean children's attitudes toward English varieties began to emerge, we did not go further with interviewing Singaporean children. However, it would be interesting to hear Singaporean children's voices in future studies. Another point the future research could encounter is considering young Korean children with an extensive amount of English education. Subjects in our study were children who attend a typical Korean kindergarten. However, some children attend other forms of institutions, such as English Kindergartens, where a full-day curriculum is instructed in English, specifically AmE. Perhaps children attending an English Kindergarten may have different attitudes toward English varieties than the children we observed in this study.
Despite the drawbacks, the present research provides the teaching community and society at large with critical educational implications for young children's English education. To our knowledge, this study is among the first to investigate the emergence of young children's attitudes toward varieties of English. Although our study's main subjects were Korean children, we believe our findings can contribute to the understanding of children in other countries within the Asia Pacific region for two reasons. Firstly, it is essential to note that Singaporean children also participated in our study as a comparison group. We purposely compared Korean children to Singaporean children because Singapore is one of the Asia Pacific regions where English is designated as an official language. Similar patterns of responses gathered from children in the two countries may imply that the difference in young children's learning English context and English language abilities before entering a school may not be critical predictors related to the formation of young children's attitudes toward varieties of English. Secondly, studies conducted in countries such as Japan and China with adults have shown a similar pattern of AmE preference over Asian English. Thus, there is the possibility that young children of these countries may show a similar pattern of attitudes toward English varieties as Korean children. Therefore, we believe that our study results can bring new insights to experts, policymakers, and teachers in the field of young children's English education in other EFL countries that share identical English educational policies and similar cultures with Korea. Informed Consent Statement: Informed consent was obtained from a parent of all children involved in the study. On the day of the data collection, verbal consent was also obtained from all children.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A.
Appendix A.