Are you a die-hard K-pop fan? Examining English Korean code mixing uttered by an American native speaker youtuber

Article history: Received: June 17, 2021 Revised: Sept 17, 2021 Accepted: Sept 25, 2021 This study analyzes the types and reasons for code-mixing used by the speaker with a concern for the interlocutors, who are a native speaker and a non-native speaker. The significance of comparing the speaker's code-mixing aims to know if changes in the interlocutors affect the types and reasons for code-mixing adopted. This analysis applied the qualitative descriptive method to analyze data taken from two YouTube videos of an American native speaker namely Edward Avila entitled Chopping all my friend’s hair off and giving him makeover with Korean native speaker as the interlocutor and catching up with Sorn + A soft glam lewk with Korean non-native speaker as the interlocutor. For Korean native speaker interlocutor, the most frequently used code-mixing type is alternation, while for Korean non-native speaker interlocutor is insertion. However, the data identified three specific reasons for codemixing with Korean native speaker interlocutor, notably clarification of the speech content, interjection, and quotation of someone else. Meanwhile, there are two specific causes for code-mixing revealed in the data of Korean non-native speaker interlocutor, namely discussion of a specific topic and repetition used for clarification. The varied types of interlocutors had an impact on the forms of code-mixing and the reasons for code-mixing in the dialogue.

than their own mother tongues. Because of the ease in language acquisition, many people are able to speak in two or more languages in their daily discussions. The conversations occur most commonly in informal contexts rather than formal circumstances, especially if the language is not changed. Bilingualism is a term for this sort of phenomenon, which refers to as the person's ability in speaking of more than one language on the same mutual intelligibility. Meanwhile, bilingual is a term for the person who is able to speak of more than one language. Recently, it is easy to find bilinguals in such traditionally monolingual countries as Korea as well on account of this globalization (Kim, 2006). Furthermore, Hamers & Blanc (2000) defined bilingualism as a situation in which there is a contact between two languages in a linguistic community and the use of the two codes occurs in the same interaction. The three types of bilingualism are compound bilingualism, coordinate bilingualism, and sub-coordinate bilingualism. Compound bilingualism occurs when a speaker develops two languages. Coordinate bilingualism occurs when a speaker learned two languages separately in different situations. Sub-coordinate bilingual refers to a circumstance in which the speaker learnt the secondary language by filtering it through the first language or mother tongue.
There is a phenomenon known as code-mixing and code-switching that occurs when bilingualism develops. According to Holmes (2013), codemixing is used by randomly altering the codes of the two languages in an utterance, potentially due to the speaker's inability to describe the code in only one language at the time. Wardhaugh (2006) also claims that code-mixing is a use of both languages. In addition, Ramadhaniarti et al. (2018) state that the mixing of the two languages can vary in terms of words, phrases, or sentences. According to Muysken (2000), there are three main types of code-mixing, namely insertion, alternation, and congruent lexicalization. Insertion occurs when material (lexical items or words and phrases) from one language inserted into a structure of the other language in an utterance. Alternation occurs when there is an alternation between structures from languages (clause) in an utterance. Congruent lexicalizations occurs when there is an influence of dialect within language use in an utterance.
In addition, according to Hoffmann (2014), code-mixing occurs for a variety of reasons, including interjection, quotation of someone else, expression of a specific group identity, statement of a lexical need, discussion of a specific topic, repetition used for clarification, empathy for something, request softly, clarification of the speech content, and exclusion for a limited audience. Interjection occurs when the speaker wants to represent their feelings. Quotation of someone else occurs when the speaker wants to quote the exact and original words of someone. Expression of a specific group identity occurs when the speaker wants to show their identity in a conversation. Statement of a lexical need occurs when the speaker needs to use words or vocabulary lexically. Discussion of a specific topic occurs when the speaker needs to use code-mixing in a conversation with a specific topic to make it more effective. Repetition used for clarification occurs when the speaker wants to clarify the words of the interlocutor. Empathy for something occurs when the speaker wants to emphasize a discussion. Request softly occurs when the speaker wants to request something by using one particular language to be more polite. Clarification of the speech content occurs when the speaker wants to make the interlocutor understand the conversation more easily using one particular language. Exclusion for a limited audience occurs when the speaker wants only a particular audience to know the conversation.
There have been eminent scholars discussing code-mixing issues (Ariffin & Husin, 2011;Barman et al., 2014;Das & Gambäck 2015;Yuliana et al., 2015;Chanda et al., 2016;Sutrisno & Ariesta, 2019) to name but a few. First, Ariffin & Husin (2011) with their study entitled Code-switching and Codemixing of English and Bahasa Malaysia in Content-Based Classrooms: Frequency and Attitudes aimed to learn the frequency of instructors employ code-switching and code-mixing in learning activities and to learn how students react to it. Meanwhile, Barman et al. (2014) with their study entitled Code Mixing: A Challenge for Language Identification in the Language of Social Media aimed to find out how the automatic language identification works in dealing with codemixing on social media, which in this case is in Indian. Then, Das & Gambäck (2015) with their study entitled Code-Mixing in Social Media Text: The Last Language Identification Frontier? attempted to analyze how the language identification detects code-mixing in particular context. The study focuses on the characteristics of code-mixing in English-Bengali and English-Hindi found from Facebook posts and messages. In addition, Yuliana et al. (2015) also conducted a study entitled Code-mixing and Code-switching of Indonesian Celebrities: A Comparative Study attempting to ascertain the most common sorts of code-mixing and code-switching use by Indonesian celebrities who have native-speaker parents and are fluent in two or more languages. Next, Chanda et al. (2016) conducted a study entitled Unraveling the English-Bengali Code-Mixing Phenomenon which aims to explore the phenomenon of English-Bengali code-mixing and to test the accuracy of predictor-corrector model in identifying the language of every-word form. Moreover, Sutrisno & Ariesta (2019) conducted a study entitled Beyond the use of Code Mixing by Social Media Influencers in Instagram aims to figure out the reason of the use of code-mixing by influencers in social media, specifically Instagram.
Among those aforementioned studies regarding code-mixing investigation, the use of code-mixing in the context of social media has gained increasing interest. Barman et al. (2014), Das & Gambäck (2015), and Sutrisno & Ariesta (2019) have discussed that topic through their studies on the occurrences of code-mixing phenomenon in the context of social media. However, the writers have been concerned that there is still lack of study addressing the code-mixing phenomenon occurring on YouTube. Therefore, the present study tries to fill in the gap that the previous studies offer by concentrating on the investigation on the types and reasons for the speaker's code-mixing, with regard for the speaker's interlocutors, including Korean native speaker (KNS) and Korean non-native speaker (KNNS). Both interlocutors are able to speak Korean. KNS is a native speaker as Korean is his first language (L1), while KNNS is a foreigner who has lived in Korea for a long time. The importance of comparing the speaker's code-mixing results in understanding whether differences in the interlocutors influence the types and reasons for code-mixing used. As a result, there are two research issues to address: 1. What types of code-mixing that are conducted by the speaker? 2. How different are the reasons for code-mixing performed by the speaker based on two distinct interlocutors (KNS and KNNS)?
The findings of this study may contribute to the body of knowledge particularly in the Sociolinguistics field. The results may give another perspective on the occurrence of code-mixing in the social media context. In addition, the result of this study also has societal significance, given that the widespread use of two or more languages in our daily conversations. Whether we realize it or not, technological advances encourage people to understand a language other than their native language, making communication with anyone much easier.

METHOD
This analysis used qualitative descriptive as the method. According to Gall et al. (2003), qualitative research and descriptive research are phrases used to describe an analysis that involves observation and survey instruments to collect data. The data sources for this research are two videos collected from Edward Avila's YouTube channel in which he speaks with code-mixing in both English and Korean alternately. Because of the disparities in interlocutors and the use of code-mixing in the videos, these two videos were selected as the source of data. Edward Avila is an American-Filipino YouTuber who was born and raised in California, USA, and is therefore an English native speaker. He currently resides in Seoul, South Korea. His YouTube videos are primarily on make-up, but they are also on Korean-pop or K-pop. For this study, the writers selected two specific videos from the YouTuber because of the discovery of code-mixing usage with types namely insertion, alternation, congruent lexicalization, and various reasons such as interjection, discussion of a specific topic, statement of a lexical need, and so forth. In addition, the similarity of themes from data sources helps the analysis of the different types and reasons of code-mixing with different interlocutors more precisely. The first video, titled Chopping all my friend's hair off and giving him makeover, the interlocutor is Yoonpic, a Korean native speaker (KNS). In addition, the second video entitled Catching up with Sorn + A soft glam lewk, the interlocutor is Sorn, a Korean non-native speaker (KNNS) who comes from Thailand and is a member of a Korean girl group called CLC.

Instruments
The data collection process was through document analysis, where the authors gathered YouTube videos. As a result, it would entail the process of collecting and interpreting non-numerical data to determine the notion. According to research aims, the two data sources selected based on their respective interlocutors (KNS and KNNS) have been analyzed regarding the types of code-mixing and reasons for code-mixing, two tables for each. The data code consists of six parts, namely indicators for datum numbers, data sources, interlocutors, code-mixing types, code-mixing reasons, and time markers.

Data analysis procedures
The data analysis processes consist of five sections. First, the writers collected data from two distinct sets of interlocutors, namely KNS and KNNS. Second, the writers submitted the data into the table to organize it. Third, the writers examined the types and causes of data from the first and second groups, including total data and individual datum. The writers sorted and divided the data by using codes, for example (1-YT1-KNS-A-Q, 0.10-0.21). Number 1 is for datum number indicator, YT1 as a description of the data source for the first YouTube video, and KNS as an interlocutor description for Korean Native Speaker. Also, A as a description of the code-mixing type which is used for alternation, Q as a description of code-mixing reason which is used for quotation of someone else, and 0.10-0.21 as a time marker of conversation in video from the data source. Fourth, the writers contrasted the sorts of reasoning of the two groups or interlocutors (KNS and KNNS). Fifth, the writers summarized the findings that will answer the research questions.

FINDINGS
This study aims at answering two prominent issues concerning types of codemixing and the reasons of doing code-mixing by American native speaker YouTuber.

Types of code-mixing performed by American native speaker YouTuber
The finding concerning the types of code-mixing performed by American native speaker YouTuber is explained in two sections, namely those interaction with KNS and KNNS. Types of code-mixing performed by the speaker when communicating with KNS Table: 1 Types of Code-mixing Found in video entitled Chopping all my friend's hair off and giving him makeover Utterances of Code-mixing Types We met a while ago actually, so we friends since that, but i probably oh you know if we can have time maybe we can go make video together, 근대 맨날 불러봐 언제하냐고? (geundae maennal bulleobwa eonjehanyago = everyday he asks, when we doing it?).

Alternation
But we did Jiyook's hair, you already see 근대 (geundae = but) for the reveal.
(8-YT1-KNS-AI-SLN, 2.57-3.02) Alternation, Insertion I think Jiwook, I think also because 머리도옥 같이 했으니까 (meorido-ok gatchi haesseunikka = since we also did his hair), so the 변신 (byeonshin = change) was more, right? According to Table 1, there are roughly 15 utterances with code-mixing recognized from the video over a period of 10 minutes and 43 seconds for the first group data with a total of six utterances with alternation, five utterances with insertion, three utterances with combination of insertion and alternation, and three utterances combination with congruent lexicalization. The first type of code-mixing used by the speaker to the KNS interlocutor is insertion, which occurs when material (lexical items or words and phrases) from one language is inserted into a structure of the other language in an utterance, for example '근대' (geundae = however/but) and '이거봐' (igeobwa = look at this) . Another type found in the data is alternation since there is an alternation between structures from languages (clause) in an utterance, for example "maybe we can go make video together, 근대 맨날 불러봐 언제하냐고?" (geundae maennal bulleobwa eonjehanyago = everyday he asks, when we doing it?). Also, there are congruent lexicalizations, an influence of dialect within language use in an utterance, for example "유튜브 튜토리얼" (yutyubeu tyutorieol = YouTube Tutorial), '트리시이' (teurisi-i = 3CE / a brand name for a cosmetic), and '브랜드' (beuraendeu = brand).
The writers discuss extensive study of the aforementioned data with KNS as the interlocutor in the following part to clarify its essence.

Datum 3
: Oh and also Hairmon is back, hi Hairmon 인사할래 (insahallae = do the introduction/ introduce yourself) (3-YT1-KNS-I-RS, 1.19-1.23) The speaker uses the word '인사할래' (insahallae) in Korean instead of 'do the introduction or introduce yourself' in English, indicating the use of insertion.

Datum 4
: 아니 이거 립토너이거든 (ani igeo liptoneo-igeodeun = so this is a lip toner right) 근대 (geundae = but) i guess people are like "what is lip toner?" (4-YT1-KNS-C-CSC, 1.35-1.39) The first phrase uses Korean as the alternation due to the next phrase in English in a single utterance. Therefore, the speaker uses the word '근대' (geundae) in Korean instead of 'but' in English, indicating the use of insertion. Also, the speaker uses congruent lexicalization of 'lip toner' in English as '립토너' (lip-toneo) in Korean with the differences on the pronunciation of letter 'l' as 'r' in word 'lip', emphasizing of letter 't' as 'th' and elimination of letter 'r' in word 'toner'.
(11-YT1-KNS-C-SLN, 4.39-4.47) The speaker uses alternation between Korean utterances and English utterances. Also, the speaker uses congruent lexicalization of '3CE' in English, a Korean cosmetic brand, as '트리시이' (teurisi-i) in Korean with the differences on the pronunciation of word 'three' as 'teuri', emphasizing of letter 'th' as 'teu'. Also, the speaker use the word 'brand' in Korean '브랜드' (beuraendeu) rather than in English, with the differences on the pronunciation of letter 'b' as 'beu' and 'd' as 'deu'.
The following section describes another finding depicting the American native speaker YouTuber when communicating with Korean non-native speaker (KNNS) interlocutor. There are some distinctive characteristics differentiating the way he communicates with different interlocutor by different linguistics repertoire. According to Table 2, there are roughly 15 utterances with code-mixing recognized from the movie with a runtime of 26 minutes 29 seconds for the second group data with a total of seven insertion utterances, five alternation utterances, and three combinations with congruent lexicalization utterances. The first type of code-mixing used by the speaker to the KNNS interlocutor is insertion, for example '느낌' (neukkim = nuance), '그러며는' (geureomyeoneun = then), and '먹방' (meokbang = eating show). Another type observed in the data is alternation, such as "But, I was frightened that they would look like that on you, 근대 너무 잘나온데" (geundae neomu jalnaonde = but it suits you perfectly). Also, there are congruent lexicalizations, for example '포인트' (pointeu = point) and '메이크업' (meikeueop = makeup).

Types of code-mixing performed by the speaker when communicating with KNNS
The following section describes the analysis of the use of code-mixing performed by the speaker when interacting with KNNS in detail. '먹방' (meokbang = eating show) itself is a blending word of '먹는' (meokneun = eating) and '방송' (bangsong = broadcast or show), besides it is a term popularized by Korean streamers for a content of eating food as the trend nowadays in many platform for example AfreecaTV, YouTube, Twitch, and Douyin (TikTok). The reasons of doing code-mixing performed by the speaker when communicating with KNS Table: 3 Reasons of Code-mixing Found in video entitled Chopping all my friend's hair off and giving him makeover

Utterances of Code-mixing
Reasons of Code-mixing We met a while ago actually, so we friends since that, but i probably oh you know if we can have time maybe we can go make video together, 근대 맨날 불러봐 언제하냐고? (geundae maennal bulleobwa eonjehanyago = everyday he asks, when we doing it?).

Quotation of someone else
But we did Jiyook's hair, you already see 근대 (geundae = but) for the reveal.

Repetition used for clarification
There have been like, you know, around the time that we did our video together, super meme, lots of memes, 근대 요즘해 (geundae yojeumhae = but nowadays) i feel like just like really chill, talking videos or like fun to watch. According to table 3 and 4, the same two reasons for code-mixing were discovered in both interlocutors, including statement of a lexical need (six utterances found in the first group and seven utterances found in the second group) and request or command softly (two utterances found in the first group and one utterance found in the second group). However, the writers found three particular reasons for code-mixing in the data of the first group or the interlocutor of KNS, which are clarification of the speech content (four utterances), interjection (two utterances), and quotation of someone else (one utterance). While the data from the second group or the interlocutor of KNNS reveal two distinct reasons for code-mixing, which are discussion of a specific topic (six utterances) and repetition used for clarification (one utterance).

DISCUSSION
This study demonstrates that speakers use three types of code-mixing in general, namely insertion, alternation, and congruent lexicalization. For KNS, the most frequently used code-mixing type is alternation. The speaker changes the structures with phrases between languages helping the interlocutor to understand the conversation better. While for KNNS, the most frequently used is insertion because of the need to insert lexical items. Occasionally, words are easier to understand by the speaker and interlocutor using one particular language. For example, the speaker prefers to use the word '근대' (geundae) in Korean instead of 'but' in English. The combination with congruent lexicalization is the least common code-mixing type for both interlocutors. Furthermore, the speaker rarely uses this type for KNS because terms from English influenced by Korean dialect are more difficult to understand.
However, the reasons for using code-mixing are seven that the speaker uses, including statement of a lexical need, request softly, clarification of the speech content, interjection, quotation of someone else, discussion of a specific topic, and repetition used for clarification. The most frequently used codemixing reason for both interlocutors is statement of a lexical need since particular words are easier to understand in one language. Then, the least used reason for KNS is quotation of someone else since it is conditionally on some particular occasion. On the other hand, the least frequently used reason for KNNS is request softly and repetition used for clarification because these two reasons used only in certain conditions. For example, when asking for something politely such as "재미는 얘기를해줘" (jaemineun yaegireulhaejwo = tell me a fun story) and repeating word to emphasize something such as the speaker use the word 'really?' both in English and Korean '진짜?' (jinjja). The similarities between the two interlocutors are the speaker used the same codemixing reasons, namely statement of a lexical need and request softly. The speaker uses code-mixing without forgetting to respect the culture of the interlocutor, in this case Korean cultures have polite language rules in a conversation. However, the differences of reason used by the speaker for KNS are clarification of the speech content, interjection, and quotation of someone else. The main points of the speaker used those code-mixing reasons are to make the interlocutors feel comfortable and understand the context better. Moreover, the differences of reason for KNNS are discussion of a specific topic and repetition used for clarification because it is only to emphasize several words or phrase in a conversation to make it clearer for the interlocutor.
The result of the present study is in line with the previous studies by Barman et al. (2014). The results from Barman show that there are three most recommended approaches in the use for code-mixing in social media, namely character n-gram features, contextual information, and information from dictionaries. Since abundant social media contexts require certain linguistic features, code-mixing is unavoidable. The present study also shows that the urge to explain the lexical need of a word found in social media is needed to support the clarity of the context brought. In the same vein, Das & Gambäck (2015) also showed that code-mixing in English-Bengali and English-Hindi found from Facebook posts and messages is only able to provide 75-80% performance due to the complexity of language identification that must be carried out at word-level. Therefore, code-mixing is chosen as the alternative to ease the communication. In addition, Sutrisno & Ariesta (2019) also found that there are four main reasons for the use of code-mixing by influencers, which are the desire to increase the English skills for the audiences, habit caused by influence of their family or friends, sponsorship requirements, and the needed to gain their followers feedbacks. In relation to what the present study has found, the same as what is stated in the previous study by Sutrisno & Ariesta (2019), a YouTuber also need to clarify for certain content or intention by using a code-mixing otherwise the message intention could not be delivered.
The role of code-mixing as a form of bilingualism phenomenon is very large to create meaningful and smooth communication. Therefore, the use of code-mixing cannot be avoided considering that the agreement supports efforts to minimize misunderstandings in communication. The codeswitching phenomenon is only a small part of the bilingualism phenomenon that we encounter in our daily lives as a sign that our society is growing with the increasing number of those who have more than one linguistics repertoire. A set of linguistic codes possessed by a person is called as linguistic repertoire as mentioned by Perdhani et al. (2021). In spite of the fact that the YouTuber is an American native speaker, the competence in mastering Korean shows his diverse linguistics repertoire that supports his career as a YouTuber needs to know more about other people languages as he is going to get mingled with other people out there having different codes.
In sum, bilingualism has in most cases a positive effect on the acquisition of other languages other than L1 but language acquisition is a complex phenomenon that is also influenced by many other factors (Cenoz, 2003). As a result, the speaker is able to switch languages easily from English to Korean as he masters those two codes with the same mutual intelligibility. Code-mixing or code-switching will help him run his expertise as a YouTuber since he needs to master more than one language to ease him when meeting a different person. Furthermore, it also supports him for the sake of providing various contents for his program.

CONCLUSION
After analysing the data, the writers discovered that the interlocutor of utterances that utilizes code-mixing, whether native or non-native speaker, had an impact on the types of code-mixing. The speaker uses more alternation with Korean native speaker and insertion for Korean non-native speaker. The similarities of the speaker's reasons for code-mixing that were discovered in both interlocutors includes statement of a lexical need and request softly. However, there are several differences found, including three specific reasons for code-mixing found in the data of the first group or the interlocutor of KNS, that are clarification of the speech content, interjection, and quotation of someone else. Also, two specific reasons for code-mixing revealed in the data of the second group or the interlocutor of KNNS, which are discussion of a specific topic and repetition used for clarification. With the results of this study, the analysis of code-mixing requires further research to continue the project in the future.
The next researchers are encouraged to do a study on code-mixing in another social media platform other than YouTube, for example Podcast, to see if the tendency is the same or not. In addition, it is also beneficial to see if different social media platform video may yield different phenomenon. Along with bilingualism, the code-mixing phenomenon is a diverse topic to explore. Therefore, having another hint of bilingualism such as code-switching will be beneficial to see if specific characteristics may emerge.

ACKNOWLEDGMENTS
The researchers would like to thank Brawijaya University for facilitating the journal publication program as a substitute for the final project through the Sociolinguistics course. Also, we are very grateful to have been given the opportunity by both the editors and anonymous reviewers of JOALL to publish our article.

THE AUTHORS
Khadijah is an undergraduate student majoring in English Literature at the Department of Languages and Literature, Faculty of Cultural Studies, Universitas Brawijaya since 2018 until now. Her interest in Sociolinguistics motivated her to conduct analytical research, which resulted in the publication of this journal paper.
Emy Sudarwati is a lecturer in the English Department at the Faculty of Cultural Studies, Universitas Brawijaya since 2006 until now. In 2001, she graduated from the Department of English Literature at the State University of Malang. In 2008, she advanced her education to the Master's level in English Education. She is currently pursuing her doctorate at the same university and has 14 journal publications to her credit.