Our world is a multilingual one. 20% of the U.S. population can converse in two or more languages, while some experts estimate that at least half of the world’s population is bilingual (Mathews, 2019). With the growing demands in political, economic, and cultural interactions between the United States and China, bilingual English and Mandarin Chinese speakers in both countries are playing an increasingly important role in bridging the two peoples. Although they are fluent in both languages, different choices in language preference and usage may result in miscommunication, cultural misunderstandings, and group identity conflicts. In addition, the population expansion of bilingual English and Mandarin Chinese speakers in universities and workplaces in both the United States and China is confronted with new challenges to linguistic diversity and cultural maintenance. Therefore, it is critical to understand how macroscopic and microscopic sociolinguistic factors, including society’s language policy and communication norms and the communication participants’ background and language uses, influence these bilingual speakers’ individual language choices.
This research explores the individual language choices of balanced bilingual speakers of English and Mandarin Chinese. First, it reviews the existing studies in Chinese sociolinguistics and the topic of language choice and identifies a gap in the literature. Then, it formulates its research hypotheses and discusses its mixed-method study design. Next, both quantitative and qualitative data are analyzed and interpreted, and a few findings are discussed. The essay ends with a discussion of this research’s significance, implications, and limitations.
Literature Review
In the field of Chinese sociolinguistics, bilingualism has become a bourgeoning focus of research, especially against the backdrop of local language protection and the promotion of multiculturalism. The scholarly output on bilingualism has experienced a 300% growth over 2019-2021 in Chinese sociolinguistics. Existing studies explore the topic of bilingualism from different perspectives. Many discuss bilingualism and multilingualism within the context of language policy and planning in Mainland China, Hong Kong, Macau, and Taiwan (Harwood & Lai, 2017; He & Liao, 2020; Hopkins, 2006; Moody, 2021; Poon, 2010; Scott & Tiun, 2007; Wang & Phillion, 2009; Wu, 2011). The theories and models of linguistic landscapes are employed in some of this literature to extend the exploration of bilingualism to the public display of languages (Grey, 2021; Yao et al., 2022). Some studies examine the family language policy of bilingual and multilingual households in the Greater China and overseas Chinese diaspora (Leung & Uchikoshi, 2012; H. Li et al., 2022; Y. Li et al., 2019; F. Liang & Shin, 2021; L. Liang et al., 2022; S. Liang, 2020; C. Wei et al., 2022). Other researchers investigate how bilingualism and multilingualism construct individual and collective identities (Dong & Blommaert, 2009; Liu, 2012). Another important research topic on bilingualism is the English language education in China, where sociolinguists point out the inadequacy and undesirable impacts of current policies in China (Cheng & Li, 2021; Hu, 2008).
Language choice is a prominent topic in sociolinguistics that closely connects with bilingualism. The existing sociolinguistic literature across different languages is primarily concerned with a particular country’s language policy and planning as the entire community’s “language choice,” in the form of coherent sets of social norms, in response to political and social changes (Clayton, 2006). There are also many studies examining how macroscopic political institutions or social structures impact an individual’s language choices (Herman, 1961; Heller, 1992, 1995). In addition, some studies examine the neural mechanisms of language choices, especially among bilingual speakers but fail to point out how these mechanisms could respond to different social cues differently (Reverberi et al., 2018). However, only a few studies explore how people’s language choices are influenced by social or sociolinguistic factors within a microscopic domain, such as the characteristics of other participants in the conversation (Romaine, 1994, pp. 77–78). It is nevertheless critical to understand how “pressures of various kinds” within each domain “influence the bilingual towards the use of one language rather than the other” since these individual language choices are often institutionalized at the societal level in bilingual or multilingual communities (Romaine, 1994, p. 45).
Existing literature on the individual language choices of bilingual Chinese speakers is far more limited. Previous studies primarily rely on the speakers’ own identity, background, and experience to explain differences in their language choices (Lei & Jiang, 2019; L. Wei et al., 1992). Only a few extend the scope of research to examine variables external to the speakers themselves, such as occasions of communication (Saravanan, 1999). However, none of them focuses on the sociolinguistic characteristics of other participants in the communication.
This research identifies a critical gap within the existing literature on the language choice of bilingual speakers: how social and sociolinguistic factors within microscopic communication settings influence a bilingual English-Chinese speaker’s language choice requires further research. Other participants in a conversation often exert the most potent social pressures on people’s language choices within microscopic settings. Thus, it is essential to inquire how the sociolinguistic characteristics of these conversation participants influence bilingual speakers’ individual language choices. Consequently, as a first step, this research intends to identify factors influencing the individual language choice of balanced bilingual English-Chinese speakers when communicating with other balanced bilingual English-Chinese speakers.
The researcher formulates a set of hypotheses based on the accommodation theory discussed in Romaine (1994). In Language and Society, Romaine repeatedly examines how an individual often makes their language choice to accommodate other participants actively or passively in the conversations. She mentions how national backgrounds (p.13), ethnicities (p.61), and language norms (p.242) of other participants in the communication lead to using one language rather than the other for bilingual speakers (Romaine, 1994). Thus, the present research intends to explore national backgrounds and demonstrated language norms as potential factors influencing a bilingual speaker’s language choice. The research further operationalizes these two potential factors by focusing on a specific and easily detectable ramification of each factor: the other participants’ home country and the conversation’s initiating language. It hypothesizes that the other speaker’s home country and the language in which the other speaker initiates the conversation are factors that determine the language choice of a balanced bilingual speaker of English and Mandarin Chinese. In quantifiable terms, the research formulates the following hypotheses in line with the accommodation theory discussed in Romaine (1994):
- When responding to another balanced bilingual English-Chinese speaker, balanced bilingual English-Chinese speakers are more likely to respond in Chinese if the other speaker initiates the conversation in Chinese and English if the other speaker initiates the conversation in English.
- When responding to another balanced bilingual English-Chinese speaker, balanced bilingual English-Chinese speakers are more likely to respond in Chinese if they know that the other speaker was born and raised in China and English if they know that the other speaker was born and raised in the United States.
Methodology
To test these hypotheses, the research identifies its target population, adopts stratified sampling to recruit participants, and designs a mixed-method study to collect and analyze data.
Target Population
This research intends to target current undergraduate students in the United States who are balanced bilingual speakers of English and Mandarin Chinese. There are two reasons why the present research focuses on this population. First, undergraduate students learn and practice languages actively. For local American students, most undergraduate programs have foreign language requirements that compel the students to learn and practice a foreign or heritage language. The classroom setting and the bountiful academic and social resources in universities also contribute to the active learning experience. For many international students, English is a foreign language but a requirement for living and studying in the United States. Since most classes are taught in English in an average American university’s undergraduate program, international students are compelled to learn and practice English actively and frequently. Moreover, the extracurricular and social resources provide additional incentives for them to use English and develop a higher proficiency. These active learning experiences and environments make them more sensitive to nuances in various sociolinguistic contexts and signals and enable them to articulate their language attitudes and practices drawing from their personal experiences.
Sampling
This research uses stratified sampling to ensure its representativeness. It attempts to include approximately 45% of the participants identifying the United States as their home country, 45% identifying China as their home country, and 5% identifying other home countries. It attempts to collect 30 samples. Due to the limited time and budget, the researcher recruits participants from advanced-level Chinese language courses in two universities in a southern U.S. state. Then, the researcher identifies some bilingual speakers of English and Mandarin Chinese within each sampling stratum given their characteristics and actively approaches these potential participants. To make sure that this study includes only balanced bilingual speakers, the researcher establishes the following operational definition of balanced bilingual speakers of English and Mandarin Chinese: either (a) the normalized self-perceived language proficiency scores for both English and Mandarin Chinese are equal to or above 0.67, and the difference between the two scores is equal to or below 0.33 or (b) their years of learning or speaking both English and Mandarin Chinese are equal to or longer than 15 years and the difference between the two time periods is equal to or below 5 years. A high level of self-perceived language proficiency complemented by a small difference between the self-perceived proficiencies in two languages or a balanced exposure to and education in both English and Mandarin Chinese is necessary for participants to make language choices with sufficient linguistic competencies and resources in both languages.
Survey Design
The research designs an online survey on Qualtrics to collect data. The questionnaire consists of fourteen questions divided into three sections: introductions and instructions, scenarios and situational questions, and demographics.
During the scenarios and situational questions section, participants are first instructed to read general descriptions of a hypothetical scenario. This hypothetical scenario includes conversing with a balanced bilingual speaker of English and Mandarin Chinese. This balanced bilingual speaker is named “Taylor,” a gender-neutral name, to avoid gender associations with language choice. The hypothetical conversation is constructed outside of any specific location, community, or situation where an official language may exist (such as the official medium of instruction in some schools). Participants are instructed to presume that Taylor is proficient in both English and Mandarin Chinese. Finally, by instructing participants to presume that they never converse with Taylor before, the hypothetical scenario rules out the potential influence of a “default” language in conversations between two known people.
After these general descriptions, participants read additional conditions when answering each question. In each question, participants are instructed to presume that Taylor initiates a conversation. Other conditions include the language in which Taylor initiates the conversation and the knowledge of Taylor’s home country (where Taylor was born and raised). These conditions vary from question to question. To answer each question, participants are instructed to determine their likelihood of responding to Taylor in English and Mandarin Chinese, respectively, on a five-point Likert scale from “extremely unlikely” to “extremely likely.” Following is a table illustrating the setup of six Likert scale questions using different conditions:
At the end of each scenario, the participants are asked to share their rationales in a free-response question. They are requested to type their reasons for determining the likelihood of responding in different languages in a text box.
Finally, the demographic section asks for the participants’ gender, age, home country, years of formal education, perceived language proficiency in English and Mandarin Chinese, and years of learning and/or speaking the two languages. The question on the years of formal education is further divided into four different media of instruction (MOI): (1) English as the only MOI, (2) Mandarin Chinese as the only MOI, (3) both English and Mandarin Chinese as the MOIs, and (4) other MOIs. These demographic questions help describe the sample, detect subgroups in the sample, and identify their similarities and differences compared to each other. These questions can help ensure the representativeness of the sample in this research. They might also provide useful contexts to understand how different factors work in the target population.
The survey uses randomization to minimize the question order effects. As Table 1 indicates, these questions are grouped into Scenario A and Scenario B. The order of Scenario A and B is randomized, and the order of the last two questions within each scenario (Taylor’s home country being the United States or China) is also randomized. The display order of “English” and “Mandarin Chinese” within each question is also randomized consistently throughout the six questions.
Data Analysis
After collecting the data through the survey, the researcher conducts an ANOVA analysis of the quantitative data to test the hypotheses. The researcher also conducts thematic coding to analyze the qualitative data and explore potential explanations for the quantitative findings. The researcher uses online software to extract repeating keyword strings from the qualitative survey responses automatically, identifies prominent themes in these keyword strings, and codes the responses based on the themes.
Pilot Test
As a first step of this research, a pilot test is run to test the validity and reliability of the research’s methodology. The researcher invited 12 people to participate in this pilot test, and 10 of them completed the survey. The pilot test ensured that (1) participants were given sufficient information and clear instructions to complete the survey questionnaire, (2) all questions are comprehended as intended, and (3) the data collected was valid, complete, reliable, and could be analyzed efficiently.
Data
Quantitative Data
The researcher reached out to 35 people and received 31 responses. The total response rate is 88.57%. Out of the 35 people, the research identified and contacted 15 people from the United States, 16 from China, and 4 from Canada. 13 people (86.67%) from the United States, 14 (87.5%) from China, and 4 (100%) from Canada filled out the survey. Answers to the demographic questions are summarized in Table 2 below. The researcher applies the operational definition of balanced bilingual speakers of English and Mandarin Chinese to determine the eligibility of all 31 participants. Two participants are removed from the sample because they fail to satisfy either condition in the operational definition. The finalized participant sample is generally representative of the target population for age and gender, but participants that identify China as at least one of their home countries slightly dominate the sample.
To analyze the responses to the scenarios and situational questions, the researcher normalizes the responses to a 0-1 scale, where 0.00 indicates “extremely unlikely” to respond in that language, and 1.00 indicates “extremely likely” to respond in that language. A descriptive summary of the 29 participants’ responses is presented in Table 3 and illustrated by a set of plots in Graph 1.
The ANOVA analysis demonstrates significant patterns. For both languages, the initial language of Taylor’s conversation plays a much larger role in determining the probability, whereas Taylor’s home country has a much smaller influence. Taylor’s home country has a slightly greater effect on the participants’ probability of responding in Mandarin Chinese than in English. These findings are statistically significant, illustrated by the low p-values. Table 4 provides detailed results of the ANOVA analysis. The Findings section elaborates on the descriptive and analytical findings of the quantitative data.
Qualitative Data
In addition to these quantitative data, the survey also collects qualitative data on how the participants determine their likelihood of responding in either language. Out of the 29 recorded participants, the qualitative question received 23 responses in Scenario A (when Taylor initiates the conversation in English) and 25 in Scenario B (when Taylor initiates the conversation in Mandarin Chinese). Full texts of these responses can be found in Appendix A.
To begin the thematic coding, the researcher conducts a phrase frequency analysis to extract repeated keywords among these qualitative responses. The software automatically extracts 237 phrases containing three or more words that appear at least two times among the 48 responses. The researcher then focuses on the 90 phrases containing four and five words, reads through these phrases, and successfully identifies 7 prominent themes reflected in these phrases. A complete list of all 90 phrases and their corresponding themes can be found in Appendix B.
Next, the researcher conducts another read-through of all qualitative responses to identify further themes that are not reflected by the extracted keywords and establish four additional themes. All 11 themes are then organized into two categories – “principles” and “types of evidence.” While “principles” illustrate the participants’ underlying concerns or goals when deciding to respond in either English or Mandarin Chinese, “types of evidence” demonstrate what evidence the participants look for to make decisions according to their principles.
The researcher then codes the 48 qualitative responses with the 11 prominent themes and summarizes the frequencies and patterns of these themes in the participants’ responses. “Reciprocity” is the most common theme in the “principles” category, and the “initial language in the conversation” is the most common in the “types of evidence” category. A summary of all themes and their frequencies is provided in Table 5. The Findings section discusses these themes in greater detail.
Findings
Quantitative Findings
The ANOVA analysis of the quantitative data suggests that it is confident to reject the null hypothesis and confirm our two research hypotheses: (1) When responding to another balanced bilingual English-Chinese speaker, balanced bilingual English-Chinese speakers are more likely to respond in Chinese if the other speaker initiates the conversation in Chinese and English if the other speaker initiates the conversation in English; (2) When responding to another Chinese-English bilingual speaker, Chinese-English bilingual speakers are more likely to respond in Chinese if they know that the other speaker was born and raised in China and English if they know that the other speaker was born and raised in the United States. Qualitative data also support these findings since the “initial language in the conversation” and the “initiator’s personal background” are the two most common themes within the “types of evidence” categories in the participants’ qualitative responses. It corroborates that the conversation initiator’s personal background, such as their home country, and the initial language in the conversation influence the language choices of balanced bilingual English-Chinese speakers.
In addition, the research draws an additional finding by interpreting the quantitative data. In the following discussion, “initiator” refers to people who start the conversation, and “responder” refers to people who join the conversation and respond to the initiator.
Quantitative data suggests that responders may construct a default character of the initiator’s personal background when their home country is unknown. Knowing that the initiator was born and raised in the United States has only a negligible influence on the responder’s language choice. This is true in both the responder’s likelihood to use English and that to use Mandarin Chinese. The maximum difference between the normalized average probability when the initiator’s home country is unknown and when the initiator’s home country is the United States is only 0.11. However, knowing that the initiator was born and raised in China has a much more significant impact on the responder’s language choice. The normalized average probability of responding in English decreases by 0.19 when the initial language is English and by 0.13 when the initial language is Mandarin Chinese, and that in Mandarin Chinese increases by 0.27 when the initial language is English and by 0.05 when the initial language is Mandarin Chinese. Therefore, if the initiator’s home country is unknown, the responder tends to construct a default character of the initiator’s personal background so that they can make language choices accordingly. This default character is more similar to being born and raised in the United States than in China. The only exception seems to be deciding the probability of responding in Mandarin Chinese when the initial language is Mandarin Chinese, where this default character is withheld.
Qualitative Findings
Demonstrated by the prominent themes and the texts of individual responses, the qualitative data offer more insights into how these factors influence a balanced bilingual speaker’s language choice.
Reciprocity. 17 out of 48 qualitative responses mention reciprocating others’ language choice as an important reason to respond in the language the initiator uses. Some consider it a “basic principle” to accommodate others in a conversation using languages others are using. Some participants point out that they prioritize reciprocating others’ language choices before considering others’ personal backgrounds and their language proficiency.
Comfort. Nine responses discuss the comfort of the initiator, the responder, or both associated with the language choices. This differs from reciprocating the initiator’s language choice because it may not always be the language they are most comfortable with. One participant contends that they would infer the initiator’s “natural mode of conversation” from their home country in order to make the initiator more comfortable. Another participant points out that they would simply ask “do you prefer English or Chinese” to ensure that others’ preferences are respected and accommodated. Moreover, a few participants argue that their comfort levels with either language also influence their language choices in real life. The findings related to “reciprocity” and “comfort” combined are largely consistent with the accommodation theory discussed in Romaine (1994).
Communication norms. Nine responses also mention that certain communication norms influence their probability of using any specific language in a conversation. Many participants regard English as the default language spoken in the United States. Similarly, one participant considers speaking Chinese a communication norm among people born and raised in China, finding it slightly abnormal when people born and raised in China speak English to another person born and raised in China. Another participant mentions that they find it “weird” to switch languages in the middle of the conversation, establishing the reciprocity principle as a more salient communication norm.
Associations and disassociation between personal background and language proficiency. The general descriptions in the survey explicitly state that “Taylor speaks both English and Mandarin Chinese fluently.” However, a few participants make additional assumptions about Taylor’s, thus the conversation initiator’s, comparative language proficiency based on their home country. They indicate that those born and raised in the United States are generally considered more proficient in English, and those in China are considered more proficient in Mandarin Chinese. Corroborated by the participants’ answers to the quantitative questions, these assumptions play more salient roles in the participants’ language choice when the conversation is initiated in Mandarin Chinese. In contrast, one participant specifically contends the disassociation between one’s home country and language proficiency. They argue that “born and raised in China doesn’t mean one’s Chinese is very good.”
Other insights.Other insightful answers include responding in a particular language because the responder is “committed to being a lifelong learner of the language.” Some participants point out that switching languages within a conversation often occurs when communicating with other balanced bilingual English-Chinese speakers. A couple of responses consider the context of the conversation as another factor influencing their language choice. One participant also discusses how Mandarin Chinese can consolidate the Chinese-speaking communities in the United States, which the participant considers a reason why many Chinese Americans initiate conversations in Mandarin Chinese.
Significance and Implications
This research promotes a better understanding of language policy and planning in bilingual and multilingual Chinese-speaking communities in the United States. It is critical to understand how “pressures of various kinds” within each domain “influence the bilingual towards the use of one language rather than the other” since these individual language choices are often institutionalized at the societal level in bilingual or multilingual communities (Romaine, 1994, p. 45). The findings of this research illustrate that individual bilingual speakers internalize and employ language planning and norms in the United States, such as speaking English as the default language and associating one’s language proficiency with their home country when making language choices during individual conversations. Meanwhile, these findings also suggest that accommodating the audience or other communication participants through reciprocal or comfort-promoting language choices might be institutionalized at the community or societal level as the new communication norms in the future.
Moreover, this research has practical implications for cross-cultural communication, education, and formulating language policy especially in university settings. As both English and Chinese are increasingly used in parallel in academic, business, and professional settings, understanding factors that influence people’s language choices can be valuable in facilitating effective communication in multilingual contexts. Findings in this research suggest that while an explicit identification of one’s home country and clear guidelines of default languages within a specific context can be helpful, stating one’s language preferences directly might contribute to more comfortable and efficient communications among bilingual English-Chinese speakers. Educators can also use the findings in this research to initiate pedagogical improvements in their language instructions to encourage their students’ practice and use of English or Chinese languages. For instance, teachers can incorporate translanguaging in their class designs to create a more inclusive and comfortable learning environment. School administrators and other policymaking officials can also devise language policies that are more inclusive and culturally sensitive by promoting equity of different language uses in different academic and social domains and encouraging the use of minority languages as part of the institutions’ multicultural commitment.
Limitations
The present research’s methodology and design have some limitations. It relies on the participant’s self-reports on their perceived language proficiencies and their years of learning or speaking each language. However, these self-perceptions or recollections may not accurately reflect their subjective and objective linguistic resource and language competence because their frames of reference may considerably differ from one another.
For the quantitative data and analysis, the present research’s survey design attempts to mitigate the question-order bias by employing randomization across different groupings of questions and labels. However, even after randomization, the question order may still impact an individual participant’s answers because they may realize a pattern in their answers to the first scenario and attempt to confirm or deny this pattern in their answers to the second scenario. Moreover, the question’s wording may present biases to the survey results and research findings. Since Qualtrics does not support randomization in question texts, the descriptions of all scenarios and situational questions mention “English” before “Mandarin Chinese” and the “United States” before “China” to ensure consistent wording. This might impact participants’ answers to some questions.
The design of the qualitative questions may also introduce biases to the survey results and research findings. Participants’ responses might be subject to confirmation biases, where they simply confirm their selections in the quantitative questions instead of explaining their real reasons or considerations when making the language choice. Moreover, there may be insufficient space for participants to elaborate on their answers, so it is challenging for the researcher to determine if the participants have the same understanding of some concepts or expressions. Moreover, participants might simply list their considerations without ranking them by importance or other logical sequences within the limited space, making the data incomplete.
Conclusion
This research explores the language choice of balanced bilingual speakers of English and Mandarin Chinese. Through a mixed-method study, the researcher collects quantitative and qualitative data from 31 participants via an online survey and analyzes these data with ANOVA analysis and thematic coding. The findings confirm the research’s hypotheses and conclude that the initial language of the conversation and the initiator’s home country are factors influencing a balanced bilingual English-Chinese speaker’s probability of choosing English and Mandarin Chinese to respond within individual conversations. The research also illustrates that reciprocity is the most important principle for balanced bilingual English-Chinese speakers when making these language choices, while the comfort of both the initiators and the responders and the communication norms play large roles as well. Finally, this research has broader significance in understanding language planning and policy in the United States and practical implications on cross-cultural communication, education, and language policymaking. Further research is needed to test these hypotheses using a larger sample size and explore additional factors that influence the language choices of balanced bilingual speakers. It is equally important to further examine the connections between individual bilingual speakers’ language choices and the linguistic ecology, norms, and policies at the community or societal level.
This paper has 2 appendices and references to 31 external sources. Please see the original document to view the full list of references and the appendices.
