Bibliometric Review of Bilingual Education Between 1983 and 2023 (40 Years)

The Meaning of Bilingual Education

Bilingual education uses two languages for instruction, ranging from assimilation-focused to diversity-promoting models (Baker, 2011; Baker & Wright, 2021). In the United States, it fosters biliteracy, cultural understanding, and academic success (Yeban, 2023). However, Cenoz and Genesee (1998) emphasised additive over subtractive bilingualism. Additive bilingualism is a model in which learning a second language enhances rather than replaces the first language (L1). In contrast, subtractive bilingualism is a model in which learning a second language (L2) leads to diminished proficiency in L1. Bilingualism is the ability to use two languages fluently, which can be classified as simultaneous or sequential bilingualism depending on acquisition patterns (Grosjean, 2010). Similarly, Genesee (2004) said that a sound bilingual education program should promote bilingual (or multilingual) competence and added that the program should consider including linguistic goals, pedagogical approaches, and levels of education.

Theories Used in Bilingual Education

Many seminal works have influenced how bilingual education operates. Many scholars have supported language separation in bilingual education as they focus on maintaining clear boundaries between languages in bilingual education. Cummins (1976, 1979) is known for his work on bilingual education, notably his Threshold Hypothesis and Interdependence Hypothesis, he also suggested that languages should be separated at different stages of learning. His early work emphasised that academic skills (Cognitive Academic Language Proficiency—CALP) should be developed in one language before transitioning to L2. He considered two critical dimensions of language proficiency, independent of whether in the L2 or L1. One proficiency dimension is Basic Interpersonal Communication Skills (BICS), which usually come into play in highly contextualised, everyday informal interactions (e.g., in a playground), which give participants many contextual clues to interpret what is being done and spoken. At the same time, the other, CALP, requires the speaker to comprehend, read, talk, and write about abstract and complex academic topics, for instance, at school. It might take one to two years to acquire BICS and 5 to 7 years to use CALP. Therefore, schools should deploy explicit instruction in acquiring CALP earlier, and teachers should constantly adjust it to learners’ cognitive and linguistic levels for academic success. Lambert’s (1974) work on additive and subtractive bilingualism contributed to the notion that certain forms of bilingual education should maintain separation between languages to avoid subtractive bilingualism, in which L1 is lost as L2 is learned. Skutnabb-Kangas (2013) focused on language rights and maintenance, arguing that keeping languages separate in specific contexts can preserve linguistic diversity and prevent language shifts, particularly in minority language communities. Genesee’s (2004) work in bilingual education often focuses on providing structured and distinct contexts for the two languages to avoid confusion and ensure that L2 is acquired effectively.

Fishman’s (1991) work elaborated further on the critical need for maintaining separate spheres for languages, especially when there is a dominant majority language, such as English in the United States. He warned that without language separation in specific domains (e.g., home language use, community support), minority languages would face significant risks of erosion and eventual extinction.

However, more recent discussions on the operation of bilingual education have begun to emerge. García and Wei (2014) and García and Lin (2017) proposed another inclusive approach to promoting bilingual education, translanguaging, which is the practice of blending multiple languages fluidly within communication and instruction. This approach views bilingualism as an integrated and dynamic process, so this allows students to integrate all languages and use them dynamically and fluidly to support their learning. Thus, this view challenges the assumption that languages must be separated. Creese and Blackledge (2010) used the term flexible bilingualism to highlight the dynamic ways students use their entire linguistic repertoire to boost learning. Several scholars have investigated this approach’s effectiveness. For instance, Fuster and Bardel (2024) found spontaneous translanguaging is much preferred in Sweden’s educational context, highlighting its role in enhancing student participation and content learning, especially for multilingual students. Prilutskaya (2021) found that pedagogical translanguaging can improve classroom learning, with teachers’ positive views on translanguaging as a facilitator. However, examining long-term effects and professional development for sustainable translanguaging practices is informative. Flores and Rosa (2023) criticised translanguaging for being adopted in neoliberal education policies without always benefiting minoritised communities. Ferguson et al. (1977) remarked that a comprehensive bilingual program should unify multilingual communities, equip students with language skills for employment and status, and preserve their identities. It should promote linguistic equality and foster an understanding of diverse languages and cultures rather than purposefully assimilating students into mainstream society.

Furthermore, more specifically, Valdés and Figueroa (1994) explain that age of acquisition can affect language learning outcomes. They also identified three levels of language ability for consideration: incipient (basic exposure with low proficiency), receptive (understanding the language through listening and reading but limited speaking and writing), and productive (full proficiency in understanding and using the language). Language proficiency levels in both languages need to be adequately assessed, and they can develop or decline over time, depending on the context in which the languages are used. Lastly, they defined two types of bilingualism: circumstantial bilingualism (when external pressures force individuals to learn L2, possibly causing the loss of their L1) and elective bilingualism (when individuals choose to learn L2 while maintaining proficiency in their L1).

Second Language Acquisition Theory (SLA)

SLA is the process through which individuals learn a non-native language, influenced by cognitive, social, and affective factors (Ellis, 2015). Vygotsky’s (1978) Sociocultural Theory highlights the Zone of Proximal Development (ZPD), where social interactions and cultural contexts enhance learning and promote collaboration in bilingual settings. Krashen’s Monitor Model proposed by Krashen (1982) explains five hypotheses for students to acquire L2, which can help bilingual teachers improve their L2 teaching in bilingual education. The Monitor Hypothesis suggests that L2 teachers provide input (i + 1) slightly above the students’ level, ensuring comprehension and acquisition through effective methods. The Affective Filter Hypothesis considers that students’ emotional factors, such as anxiety, can hinder students’ L2 acquisition. The Acquisition-Learning Hypothesis explains that students can acquire L2 like they acquire their L1, happening naturally in meaningful interaction (subconscious acquisition). Learning is about conscious raising—it is when students learn forms and rules of L2. The Monitor Hypothesis supports self-correction through conscious learning, while the Natural Order Hypothesis suggests L2 structures follow a predictable sequence.

Other Related Theories and Frameworks

Ecological Systems Theory proposed by Bronfenbrenner (1979) can be brought into one of the bilingual education theories. The scenario includes Microsystem (immediate surroundings, e.g., their family and friends), Mesosystem (interactions between microsystems, e.g., the relationship between school and their parents), Exosystem (indirect environments, e.g., parents’ workplace, community services), Macrosystem (cultural and societal influences, e.g., laws, cultural norms), and Chronosystem (time, changes in students’ life course and societal changes). Moll et al. (1992) view students’ home languages and cultures as valuable sources. They proposed the Funds of Knowledge framework to integrate these elements into the classroom so students could achieve their best academic performance. Integrated instruction strengthens students’ identities by valuing their background, preventing bilingual deficits and classroom bias. Another theory that can be included in this review is Critical Language Awareness (CLA), initially proposed by Gee (1996). This theory encourages learners to recognise the connection between language, power, and social identity. Through this CLA, learners can learn about language power structures and recognise how some languages are privileged over others. Moreover, CLA helps them identify power imbalances in academic and daily communication and address linguistic bias.

Bilingual Education Models and Approaches

Bilingual education models vary, with Roberts (1995) describing several. The Submersion Model teaches non-native speakers English and promotes assimilation into North American society. The program model does not support using L1, so it is considered the subtractive bilingualism (Lambert, 1974). The second is the ESL (English as a second language) Pullout Model, which pulls learners out of their regular classes to take ESL courses. It is often applied in places where students’ L1 is varied, and finance is limited. However, as influenced by behaviourist approaches, this method focuses on teaching L2 grammar, vocabulary, and language drills instead of focusing on a more communicative approach (Richards & Rodgers, 2014). The third is the Transitional Bilingual Education Model, which uses the student’s L1 to support studying content subjects. At the same time, teachers teach the students English as a second language. This model reflects Cummins’ Interdependence Hypothesis (1979), advocating students’ use of L1 to learn L2. However, this program is criticised for its long-term implications as it tends to transition students to L2 instruction classes only, causing subtractive bilingualism. The Maintenance Bilingual Education Model transfers learners into English content classes and supports them in developing their L1 while learning the subject matter. This model reflects Additive bilingualism (Lambert, 1974) and Cummins’ Threshold Hypothesis (1976), attempting to make learners literate, bilingual, and multilingual. The Enrichment, Two-Way, Developmental Bilingual Model, and Dual-Language Immersions all attempt to help both native and non-native speakers of English to become proficient in both languages (Baker, 2011; Cummins, 2008). In the first stages, ESL or L1 content classes are segregated as the goal is to help students of both language backgrounds to study content classes in both languages. This model aims to enrich minority and majority students’ linguistic abilities so both groups can use both languages proficiently (Baker, 2011). However, Cummins (2008), García (2009), and Grosjean (2010) regard this model operates within the monoglossic framework (Monoglossic refers to a societal or educational approach that focuses on the dominance of one language—usually the majority or official language—while disregarding or devaluing the use of other languages).

Skutnabb-Kangas (2013) considers this enrichment model may be for elite bilingualism, not addressing the needs of linguistic minorities. Finally, using Immersion—The Canadian model in bilingual education is when most English speakers learn French. Immersion Bilingual Education often encourages addictive bilingualism or pluralism. However, when minority language speakers are immersed in the majority language, the minority language assimilates and becomes subtractive (Garza & Crawford, 2005). This model may go against Cummins’ Threshold Hypothesis and García’s translanguaging. Some studies have found that subtractive schooling might negatively affect learners’ full cognitive development and academic performance. Menken and Kleyn (2010) found that placing students in remedial classes without content and moving them to mainstream classes without language support leads to long-term failure for many immigrant students.

Regarding bilingual education promotion in Europe, the European Commission and the Council of Europe have strongly encouraged its citizens to become bilinguals/multilinguals. They use content-language integrated learning (CLIL) to promote it, and CLIL has become a prevalent approach in the classroom in this continent. Dalton-Puffer (2011, p. 183) helps clarify CLIL as follows:

CLIL is about using a foreign language or a lingua franca, not L2. The dominant CLIL language is English, reflecting that a command of English as an additional language is increasingly regarded as a critical literacy feature worldwide. CLIL also implies that teachers will generally be non-native speakers of the target language; CLIL lessons are usually timetabled as content lessons (e.g., biology, music, geography, mechanical engineering), while the target language continues typically as a subject in its own right in the shape of foreign language lessons taught by language specialists; Typically less than 50% of the curriculum is taught in the target language; and CLIL is usually implemented once learners have already acquired literacy skills in L1, which is more often at the secondary than the primary level.

According to May (2017), the amount of foreign language instruction varies by institution. However, CLIL remains unevenly successful because of the lack of teacher proficiency, educational resources, and policy implementation within member states.

When researching bilingual education, English as a Medium Instruction (EMI) has also emerged and recently been increasingly recognised and implemented, especially in nations with non-native English speakers, such as Asian countries (Galloway et al., 2017). This review includes EMI because it has not been clearly defined in terms of the percentage of the target language taught in the classroom by Dearden (2014), leaving a CLILised EMI (Moncada-Comas & Block, 2019). Dearden (2014, p. 2) defines it as “the use of the English language to teach academic subjects in countries or jurisdictions where L1 of the majority of the population is not English.”

Influence of Policy Orientations on Bilingual Education

The paper would like to mention some influential bilingual policies. Ruíz (1984) argues that language policy orientations—whether they view language as a problem (challenging to overcome language diversity, usually resulting in policies emphasising assimilation and submersion programs with the expense of native language), a right (the right to preserve native languages, usually leading to policies promoting bilingual education), or a resource (viewing language diversity as resources for individuals and the society, leading to policies promoting bilingual education) influence how bilingual education programs are developed and achieved. For example, according to Ovando (2003), in the United States, due to the civil rights movement seeking to rectify historical inequities faced by non-English-speaking students in schooling, the federal legislation Bilingual Education Act (BEA) of 1968 began to recognise language minority students’ diversified needs. This fair policy encouraged using these students’ linguistic and cultural backgrounds alongside English to ameliorate their school performance. Crawford (2004) agreed that BEA-funded bilingual education programs. However, it showed an absence of a uniform guideline for implementation, leaving much of the decisions made by local districts. Despite these challenges, Baker (2011) posits that BEA helps shape bilingualism as a valuable medium for language minority students to gain academic success and preserve their cultural identities, thereby being seen as a resource rather than a problem.

Another example is that when the United States experienced political changes, No Child Left Behind (NCLB) in 2001 was enacted, and it started to require English proficiency and standardised testing, leading to reducing subsequent federal funding support for bilingual education and moving toward English-only programs. Official Languages Act (1969) in Canada can also be the document that promotes bilingual education as it mandates that federal service and education be provided in both French and English. French and its speakers’ cultures were also preserved, strengthening community unity. However, some challenges were unavoidable, such as resource funding and resistance in dominating English-speaking regions (Government of Canada, 1969). In Europe, bilingual education is successfully implemented due to the strong commitment of the European Commission. For example, the Barcelona European Council states that every European must speak other languages in addition to their mother tongue (European Commission, 2006). To do this, they recognise the importance of bilingual teacher training, support, and resources to facilitate teachers’ use of a foreign language for instruction while maintaining content quality (Coyle et al., 2010).

Bilingual education programs have also flourished in the R.O.C. Since 1949, the Chinese government has supported the development of education in minority languages, resulting in bilingual education programs for ethnic minorities (Zhang & Tsung, 2019). Different models of bilingual education have also been proposed in this country, where bilingual education usually refers to using minority languages alongside Chinese as a medium of instruction or some degree of teaching minority languages (Tsung & Cruickshank, 2009). Since there are many ethnic minority regions in R.O.C., the main goal of bilingual education models is usually cultural and linguistic assimilation. The standard models include Structured Immersion and Transition Models, which aim to fully integrate ethnic minority students into the Chinese language-based education system (Zhang & Tsung, 2019). Recently, Taiwan (R.O.C.) has proposed a Bilingual policy 2030, aiming to make its citizens bilingual in both Mandarin Chinese and English, using CLIL and EMI approaches depending on English proficiency and education levels (NDC, 2018, 2021).

Identification of Sources and Analytic Software

As this study does not include any human participants, obtaining informed consent was not necessary. The keyword “bilingual education” was entered into the Web of Science (WoS) database. The document type was limited to “articles,” excluding other sources. The search was further refined to include only documents categorised under “Education Educational Research” and indexed in the Social Sciences Citation Index (SSCI), Arts & Humanities Citation Index (A&HCI), and Science Citation Index Expanded (SCI-EXPANDED). The study sets the publication years from 1983 to 2023. There are 1.716 articles found, and they were exported and saved in four separate files for further processing, which can be provided upon request. VOSviewer version 1.6.20 by Van Eck and Waltman (2023) was employed to analyse three major types of data: co-occurrence, bibliographic coupling, and co-citation, leveraging its ability to create knowledge maps that visualise relationships among various bibliometric elements (Zupic & Čater, 2015). All clusters of data from all types of analysis were meticulously copied into a Microsoft Excel spreadsheet, with every entry being checked for accuracy, tables were created for easy reference, and all related articles were downloaded to ensure the correct attribution of authors and the precise naming of clusters in each type of analysis.

This study follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Moher et al., 2010) for screening and identifying reviewed documents, as illustrated in Figure 1. An initial search was conducted in the Web of Science (WoS) database to establish a dataset of peer-reviewed journal articles on bilingual education. Only peer-reviewed journal articles were selected for analysis in the dataset (Hsieh et al., 2023) to ensure the overall academic quality of the reviewed literature. Accordingly, a search was performed in WoS on July 7, 2023, with the document type limited to “articles” and the keyword “bilingual education” as the primary search term. The initial search yielded 2,624 relevant documents. Since this study only includes peer-reviewed research, exclusion tools were applied to remove 908 documents, including conference proceedings, book reviews, and editorial materials. Ultimately, 1,716 articles were retained for bibliometric analysis.

OPEN IN VIEWER

Figure 1.

PRISMA flow diagram detailing steps in the identification and screening of sources for this review of research on bilingual education.

Data Analysis

The main strategies used in this study are bibliometric and content analysis. The term “bibliometrics” was first introduced by Pritchard (1969) to describe the application of statistical methods to analyse scholarly outputs, including journal articles, citation counts, and journal impacts. The study of bibliometrics can help understand the development and progress of academic fields by systematically analysing aspects of written communication (Pritchard, 1969). Zupic and Čater (2015) outlined several essential bibliometric methods, such as citation analysis, co-citation analysis, bibliographic coupling, co-author analysis, and co-word analysis, which are instrumental in describing, evaluating, and monitoring published research. This study employs descriptive analysis, keyword co-occurrence, bibliographic coupling, reference co-citation, and content analysis to investigate the literature (Hsieh & Li, 2024; Hsieh et al., 2023). In addition, as the paper investigated the main findings of top influential papers, content analysis was also used to provide a subjective interpretation of the manuscripts through the systematic classification process of coding and locating themes and patterns (Hsieh & Shannon, 2005).

Keyword co-occurrence analysis facilitates exploring connections between keywords to elucidate a specific scientific field’s knowledge structure and mapping (Radhakrishnan et al., 2017). In visual representations of these co-occurrence networks, nodes represent keywords that encapsulate the accumulated knowledge of a domain. At the same time, links indicate the frequency with which pairs of keywords co-occur in documents. The frequency of co-occurrence determines the strength of these links. Keyword co-occurrence analysis, or co-word analysis, examines the co-occurrence of keywords or phrases which indicate the content within a research topic (Van den Besselaar & Heimeriks, 2006).

When two publications cite the same third publication, this is called bibliographic coupling (Martínez-López et al., 2018). The number of shared references determines the strength of their coupling. This method is beneficial for identifying current hot research topics by setting appropriate thresholds for shared citations and the strength of bibliographic links (Glänzel & Czerwon, 1996). However, because bibliographic coupling may yield delayed or static information, and given that two articles might reference an entirely unrelated subject, co-citation analysis often complements it by offering insights into subject similarity (Ferreira, 2018).

Small (1973) states that a reference co-citation occurs when two documents are frequently cited together. It is determined by comparing cited document lists in the Science Citation Index and counting the identical entries. Clusters of co-cited papers can be generated to explore the specialised structure of a scientific field. Bibliometric studies use co-citation analysis to map knowledge landscapes and identify influential research trends (Leydesdorff & Rafols, 2009).

Specifically, keyword co-occurrence analysis can be used to explore the relationships between keywords, facilitating an understanding of the knowledge components and knowledge mapping within a scientific field (Radhakrishnan et al., 2017). By analysing the frequency of keyword appearances, this method can further reveal research hotspots and key themes within a domain (Van Eck & Waltman, 2014). Meanwhile, bibliographic coupling occurs when two documents cite the same references, forming a relationship. Through bibliographic coupling analysis, researchers can identify hot research topics (Glänzel & Czerwon, 1996). Co-citation analysis has been widely used to explore the theoretical literature that scholars in a particular research field frequently cite, providing further insights for researchers to understand the theoretical foundations of that field (Zupic & Čater, 2015). All clusters of data from all types of analysis were meticulously copied into a Microsoft Excel spreadsheet, with every entry being checked for accuracy, tables were created for easy reference, and all related articles were downloaded to ensure the correct attribution of authors and the precise naming of clusters in each type of analysis.

Yearly Quantitative Distribution of Literature

Figure 2 shows two axes: publication quantity is on the vertical axis, while years are on the horizontal axis. Only 17 papers were seen in 1990, but the number nearly tripled in 2001, up to 48 publications. During this initial period (1983–2007), the number of publications was relatively even and slightly fluctuated, and it showed a surge in 2008, up to 95, the highest in this period. The number of publications receiving 95 in 2008 started a Growth Stage. The figure even more than doubled at the end of the period in 2011, with 219. The period from 2012 to 2016 can be seen as an Adjustment Stage. The number of publications dropped to 179 in 2013, reducing 40 publications compared to 2011. Nevertheless, the figure surged to gain 296 publications in 2014 but decreased to 238 in 2015 and 2016. Next, from 2017 to 2023, it can be seen as a Maturation Stage. Compared to other stages, the publications in this stage experienced the highest number of publications, up to 317 in 2021, within 40 years, although it showed a minor drop in 2023 with 317 publications.

OPEN IN VIEWER

Figure 2.

Number of publications on bilingual education between 1983 and 2023.

Overall, the number of publications in bilingual education research has increased over time, although it shows some fluctuations, which suggests that more scholars have shown interest in the field. Table 1 depicts the most influential scholars in each stage in the field.

OPEN IN VIEWER

Table 1.

Significant Publications on Bilingual Education in Different Development Stages.

Stages	Authors (year)	Citations
Initial stage (1983–2007)	Ruíz (1984)	102
	Valdés (1997)	84
	Thomas and Collier (2022)	84
Growth stage (2008–2011)	García (2009)	196
	Creese and Blackledge (2010)	97
	Baker (2011)	81
Adjustment stage (2012–2016)	García and Wei (2014)	79
	Flores and Rosa (2015)	95
	Otheguy et al. (2015)	63
Maturation stage (2017–2023)	Cervantes-Soon et al. (2017)	74
	García et al. (2017)	64
	Flores and García (2017)	50

Significant Publications in Different Development Stages

Table 1 presents an overview of the key authors and the citations their works have received across different stages of publication within the field of bilingual education.

Table 1 reveals that García (2009) dominated the Growth Stage with 196 citation counts, while Ruíz (1984) dominated the Initial Stage; Flores and Rosa (2015) were the most prevalent in the Adjustment Stage, while Cervantes-Soon et al. (2017) prevailed other scholars in the Maturation Stage. If looking at the publication totality, the Growth Stage most contributed to the field with 374 citation counts, the Initial Stage (270 counts), the Adjustment Stage (237 counts), and the Maturation Stage (188 counts). These results suggest that bilingual education has gained much interest since its onset, followed by a burst of citations in the Growth and Adjustment Stages. However, the Maturation Stage has humble citations, possibly due to their recent release publications and topical/thematic saturation in the field.

At the beginning of development, Ruíz (1984) introduced the concept of language orientations, focusing on language-as-problem, language-as-right, and language-as-resource, which are guidelines for later research. Another influential author in this stage is Valdés (1997), who asked for cautionary language planning in dual-language immersion programs, which must pay attention to minority students’ literacy development by bringing language complexities and sociocultural aspects into consideration. Thomas and Collier (2002) found that language minority students benefit from a well-planned bilingual education program. For instance, bilingual students in these programs academically outperformed those in ESL and English-only programs. The study highlights that early bilingual education helps bilinguals excel at school.

In the Growth Stage, García (2009), the most influential in the field, confirmed that in the 21st century, a standard language approach (e.g., the Western monologic view) must be reconsidered to allow space for language balance. Instead, it should integrate languages and skills (e.g., translanguaging) to address multilingual students’ needs for their academic performance. Creese and Blackledge (2010) argue that if students’ L1 can facilitate their learning, why should not we encourage it? Similarly, Baker (2011) advocated additive bilingualism in bilingual programs.

Regarding the Adjustment Stage, García and Wei (2014) significantly influenced the field by introducing translanguaging, which proved effective in educating Latino emergent bilingual youth. Likewise, Otheguy et al. (2015) advocate the translanguaging approach to allow language minority students to use their mixed language repertoires to excel at schools, so schools should not impose rigid named languages, such as national or state languages, on minorities in schools. In the same vein, Flores and Rosa (2015) criticised the appropriateness-based approaches in bilingual education because they go against promoting language minority students, causing inequalities while having to conform with the white norms and calling them sociolinguistic ideologies.

With the Maturation Stage, Cervantes-Soon et al. (2017) contend that TWI programs have brought positive results. Nonetheless, program design should consider social-political contexts (e.g., race) to avoid privileges for any social group. Second, teachers should use their students’ linguistic and cultural diversity. Third, translanguaging should be employed to avoid language segregation or dominance like English. Fourth, schools should direct teachers and students to think critically about social relations and powers, language ideologies, and other related issues.

Similarly, García et al. (2017) advocate translanguaging in K-12 classrooms, including stance (positive attitude towards bilingualism), lesson plans and assessment types (allowing students to use any language resources during learning and assessment, not just focusing on one language), and shift (being adaptive/flexible to their teaching methodology to meet students’ needs). Like Flores and Rosa (2015), after investing in the history of bilingual education in the United States, Flores & García (2017) found that TWI programs benefit white middle-class families, not low-income Latinx families, while these programs prioritise language minority students. Therefore, they asked to dismantle this racial hierarchy.

Keyword Co-occurrence Analysis (Research Question 1)

According to Dong et al. (2022), the co-occurrence method involves calculating the total strength of the co-occurrence links with other keywords and then selecting keywords with the greatest total link strength. As Hsieh et al. (2023) have pointed out, the large number of clusters generated would imply that cluster analysis has little meaning or significance. Following this guidance, our study carefully used the “full counting” method and “all keywords” as the unit of analysis, setting the minimum number of occurrences for a keyword at 10, resulting in 206 out of 4,584 keywords meeting this threshold, which were divided into six clusters. Figure 3 presents the network diagram generated from the co-occurrence analysis. Table 2 presents the top three publications in each cluster.

OPEN IN VIEWER

Figure 3.

Network visualisation of keyword co-occurrence analysis on bilingaul education.

OPEN IN VIEWER

Table 2.

Significant Keywords from Keyword Co-occurrence Analysis.

Clusters	Keywords	Occurrences
Cluster 1 (Red): Bilingualism and children in English education	English	197
	bilingualism	172
	children	118
Cluster 2 (Dark green): Language policy and identity	policy	109
	language policy	102
	identity	92
Cluster 3 (Blue): Bilingual students’ achievement	students	190
	achievement	74
	bilingual	68
Cluster 4 (Yellow): CLIL in bilingual education	bilingual education	527
	language	266
	clil	94
Cluster 5 (Purple): Literacy education and translanguaging approach	education	332
	literacy	132
	translanguaging	120
Cluster 6 (Light green): Inclusive education in schools	schools	40
	inclusion	13
	inclusive education	10

According to Hsieh and Li (2024), keyword co-occurrence is carried out by looking at the size of network nodes; it represents each important keyword; the more prominent a keyword is, the closer it is to the research hotspot. As seen in Figure 3, the high-frequency words in the bilingual education journal occur in all clusters, for example, ‘English’ in Cluster 1, ‘identity’ in Cluster 2, ‘students’ in Cluster 3, ‘bilingual education’ in Cluster 4, ‘education’ in Cluster 5, and ‘schools’ in Cluster 6. Table 2 presents the top three highest-frequency words in each cluster.

As shown in Table 2, different dominant keywords represent different focuses in bilingual education. Cluster 1 (Red): English can be seen as a means of instruction in educating children in the program. Keywords in Cluster 2 suggest a focus on students’ identities and language policies. Cluster 3 emphasises bilingual students’ achievement in the program. Cluster 4 emphasises bilingual students’ achievement and appropriate approaches, such as CLIL. Cluster 5 concerns bilingual children’s literacy development and ways to help them improve their literacy skills, such as using translanguaging. Cluster 6 shifts its focus to inclusive education in schools, for example, introducing sign bilingual education for deaf and hard-of-hearing students.

In summary, the study has identified several hot topics in bilingual education. They include EMI, the connection between students’ identities and language policies, bilingual students’ achievement, CLIL for improving academic outcomes for bilingual students, literacy development in bilingual children mainly through translanguaging strategies, inclusive education focusing on the importance of introducing sign bilingual education for deaf and hard-of-hearing students. These topics reflect ongoing debates and developments in bilingual education, focusing on identity, academic achievement, and inclusivity.

Document Bibliographic Coupling Analysis (Research Question 2)

With bibliographic coupling analysis, the study used all extracted data to prevent citation bias and locate emerging research fronts (Hsieh et al., 2023). A bibliographic coupling link is established between two items that cite the same document, and a document’s total link strength is the cumulative strength of its connections with other documents (Van Eck & Waltman, 2020). This study used the “full counting” method and “documents” as the unit of analysis; the minimum number of citations for a document was set to 20, with 403 out of the 1,716 documents meeting the threshold. As Hsieh et al. (2023) suggested, it is more effective to concentrate on fewer clusters than many as it allows more specific theme identification. The total strength of bibliographic coupling links was calculated, and the documents demonstrating the greatest total link strength were selected for further examination to reveal significant research fronts. Seven clusters were identified, as shown in Figure 4 and Table 3.

OPEN IN VIEWER

Figure 4.

Document bibliographic coupling network on bilingual education.

OPEN IN VIEWER

Table 3.

Significant Publications in Each Cluster of Document Bibliographic Coupling Network on Bilingual Education.

Clusters	Count	Authors (years)	Total link strength
Cluster 1 (Red): Bilingual education policies challenged	116 articles	Reljić et al. (2015)	362
		Tong et al. (2008)	340
		Combs et al. (2005)	315
Cluster 2 (Dark green): Culturally responsive pedagogy and ethnographic approach in bilingual education	104 articles	Johnson (2012)	305
		De los Ríos (2018)	293
		López (2016)	281
Cluster 3 (Blue): Translanguaging approach in bilingual education	70 articles	Menken (2013)	637
		Palmer et al. (2014)	630
		Yilmaz (2021)	556
Cluster 4 (Yellow): CLIL and EMI approaches in bilingual education	60 articles	Pérez-Cañado (2012)	328
		Li (2018)	265
		Lo and Lo (2014)	248
Cluster 5 (Purple): Issues of equality in bilingual education	40 articles	Cervantes-soon et al. (2017)	495
		Jong and Howard (2009)	430
		Zúñiga et al. (2018)	427
Cluster 6 (Light green): Sign bilingual education and using peers’ support in bilingual education	7 articles	Carhill-Poza (2015)	192
		Wolbers et al. (2014)	126
		Swanwick (2010)	44
Cluster 7 (Orange): Benefits from bilingual education	5 articles	R. M. Cho (2012)	88
		Agirdag (2014)	55
		Patrinos and Velez (2009)	8

Cluster 1: Bilingual Education Policies Challenged

The most influential publications in Cluster 1 are about challenges in bilingual education. Combs et al. (2005) investigated the influence of Proposition 203 on bilingual education in an elementary school in Arizona, where many Spanish-speaking students are enrolling. Spanish in the program, a structured English immersion, was restricted in the classroom so students could maximise their English. Most researchers did not support this program because it devalued students’ L1, and this program caused stress and fear among these minority students, and they asked for civil rights for these students. Tong et al. (2008) evaluated a 2-year oral English intervention for Mexican students in transitional bilingual education and structured English immersion programs in the United States. It was found that students’ native language helped improve English learning. Reljić et al. (2015) carried out a meta-analysis study to find out minority children’s academic achievement in Europe and found that many of the cases advocate bilingual education. Nevertheless, more empirical studies with larger effect sizes are needed.

Cluster 2: Culturally Responsive Pedagogy and Ethnographic Approach

Cluster 2 investigates bilingual education policies and culturally responsive pedagogies. Johnson (2012) used an ethnographic perspective to unpack the complexities of Children’s English Law (Proposition 203) in Arizona in 2000, influencing migrants’ academic performance. English-only instruction did not address immigrant students’ linguistic and cultural needs, so arbiters should consider this detrimental influence. López (2016) found that integrating culturally responsive pedagogies into bilingual programs helped improve Latino students’ learning achievement in Arizona. Such approaches align the educational content with students’ cultural backgrounds and experiences. De los Ríos (2018) investigated the case of an American Mexican bilingual student named Joaquin, enrolling in a bilingual school in the United States. It was found that integrating this student’s meaningful background, such as cultural and political values, into the classroom helped him develop literacy skills.

Cluster 3: Translanguaging Approach

Menken (2013) explored the influence of bilingual policies on emerging bilingual students in the United States (e.g., state-level anti-bilingual education ordinances and the No Child Left Behind law). These policies did not advocate bilingual education programs. Students’ performance became weakened with English-only instruction and limited home language use. Palmer et al. (2014) argued that when teachers use appropriate teaching strategies, such as modelling bilingualism and conditioning students to use their language resources, students can improve their cognitive abilities. Yilmaz (2021) said translanguaging could promote educational equity for minority students by encouraging them to validate their identities and shortening the educational equity gap. It aligns with the theory of bilingualism.

Cluster 4: CLIL and EMI Approaches

Pérez-Cañado (2012) considered the potential of CLIL as a successful educational model that fosters multilingualism in Europe. Nonetheless, it highlights the critical need for further empirical research to inform and enhance its implementation in various educational systems. Lo and Lo (2014) found that students who received the EMI approach gained more success in English ability than in disciplined subjects. However, influential factors must be identified for the program’s success, such as time allocation and assessment appropriateness. Besides, the program must control students’ English ability prior to program enrollment. Similarly, Li (2018) found that social science students’ English proficiency at university in R.O.C. improved after enrolling in a bilingual program. However, they perceived that they had to put more effort into understanding content delivered in English.

Cluster 5: Issues of Equality in Bilingual Education

Studies in Cluster 5 concentrate on equity in bilingual education. Jong and Howard (2009) posit that TWI programs are effective because they balance the language benefits for both minority and majority students. However, if the programs are not carefully planned, minority languages would be limited to access. Cervantes-Soon et al. (2017) recognised the benefits of TWI programs. However, they criticised them as generating educational inequality because educators and students lack critical awareness of influential factors like social relations and power within educational systems. Zúñiga et al. (2018) investigated how two bilingual teachers implement a bilingual language policy. While one teacher did not completely adhere to the bilingual education plan, the other was loyal to it, although both had the same goals.

Cluster 6: Sign Bilingual Education and Using Peers’ Support

Data gathering and analysis in Cluster 6’s publications concern sign language and literacy development in deaf and hard-of-hearing students’ education. Wolbers et al. (2014) investigated how American Sign Language and bilingual English instruction (Strategic and Interactive Writing Guide methodology) affected deaf and hard-of-hearing students’ English writing ability. The study showed that this strategy improved their English writing, reduced their use of American Sign Language when writing, and indicated that bilingual literacy programs that emphasise implicit language competence and metalinguistic knowledge benefit them. Regarding sign bilingual education for deaf and hard-of-hearing students, Swanwick (2010) suggested researching the implementation of sign bilingual education, practical approaches for teaching literacy, and the effects of sign language on their academic and linguistic outcomes. Carhill-Poza (2015) found that peer-support in learning is a powerful tool in bilingual education. Also, a multilingual academic classroom must be encouraged to help immigrant students develop their English abilities, such as speaking the spoken language.

Cluster 7: Benefits From Bilingual Education

Patrinos and Velez (2009) found that minority students benefited from bilingual programs. For example, the program helps reduce students’ dropouts and repetition rates, which in turn helps save the educational budget. Also, they could perform well in school and become fluent speakers of their mother tongue. R. M. Cho (2012) found that non-English language learners (ELL) reading scores, but not math scores, decreased when studying together in the same class as their ELL counterparts. The study hypothesised that this negative influence might be due to grouping ability employment, non-ELL’s gender and socioeconomic background. Agirdag (2014) explored the long-term effect of bilingualism on immigrant children in the United States. They examined three linguistic minority groups (limited bilinguals, balanced bilinguals, and English-dominant groups). Balanced bilinguals could earn much more than minority students who are proficient in English only. They also suggested that balanced bilinguals tend to retain their schooling.

In summary, the dominant documents from this analysis examined bilingual education in various contexts, including immigrant and minority students and deaf and hard-of-hearing students. They also examined the impact of educational policies, such as English-only instruction, on these groups. Some of the mentioned studies emphasise the role of effective teaching strategies to enhance language performance and language proficiency, such as peer support and multilingual classroom approaches, in promoting language development and academic success. In addition, they examined the long-term benefits of bilingualism, for example, its effect on students’ academic persistence and future economic opportunities. Finally, many of the mentioned documents underscore the importance of equitable bilingual programs that support both native languages and English, fostering both educational achievement and cultural identity.

Reference Co-citations (Research Question 3)

According to Hernández-Lara (2020), co-citation analysis is a study of cited documents to examine the citation frequency of two earlier documents together. The current study investigates the evolution of themes in bilingual education research. It proposes to provide information about the current state of this field. The study analysed the co-citation data by selecting the “full counting” method and “cited references” as the unit of analysis. Additionally, using a threshold of 20 for the minimum number of citations for a reference, 184 out of 58,423 cited references met this criterion. The total strength of co-citation links was calculated, and the content analysis was used to analyse the three most cited references in each cluster, as shown in Figure 5 and Table 4.

OPEN IN VIEWER

Figure 5.

Reference co-citation on bilingual education.

OPEN IN VIEWER

Table 4.

Reference Co-citation Analysis.

Clusters	Count	Authors (years)	Citations
Cluster 1 (Red): The role of first language and culturally responsive assessment	48 articles	Cummins (1979)	84
		Thomas and Collier (2002)	84
		August and Shanahan (2007)	56
Cluster 2 (Green): The role of multilingual approaches in bilingual education	46 articles	García (2009)	196
		Creese and Blackledge (2010)	97
		Baker (2011)	81
Cluster 3 (Blue): Critical perspectives on language planning and equity in bilingual education	39 articles	Ruíz (1984)	102
		Valdés (1997)	84
		Cervantes-Soon et al. (2017)	74
Cluster 4 (Yellow): Racial bias and encouragement of multilingualism in educational development	31 articles	Flores and Rosa (2015)	95
		Moll et al. (1992)	81
		Vygotsky (1978)	79
Cluster 5 (Purple): Evaluating CLIL effectiveness and the role of research rigor in educational approaches	20 articles	Coyle et al. (2010)	42
		Cenoz et al. (2014)	41
		Cohen (1988)	40

Figure 5 visualises the co-citation analysis. Table 4 provides detailed information, revealing five main clusters. Cluster 1 (Red) centres on the crucial role of students’ mother tongue and culture in increasing learning outcomes. This cluster has 48 articles mainly investigating the critical role of students’ L1 in assisting minority students in understanding the content subjects. This cluster also emphasises the importance of bringing minority students’ culture to the classroom in improving their learning outcomes. The dominant scholars found in this cluster are Cummins (1979), Thomas and Collier (2002), and August and Shanahan (2007). Cluster 2 (Green) discusses possibly effective bilingual and multilingual models, translanguaging, techniques and strategies to help minority students achieve their academic goals. Authors García (2009), Creese and Blackledge (2010), and Baker (2011) all attempted to promote addictive bilingualism. Cluster 3 (Blue) mentions language planning and inequity in bilingual education. Thirty-nine articles were found, mainly requesting careful analysis of minority students’ needs and making use of language diversity in bilingual education. Renowned researchers in this cluster, Ruíz (1984), Valdez (1997), and Cervantes-Soon et al. (2017), are concerned about educational equity for minority students. Cluster 4 (Yellow), with 31 articles, concerns prejudice against minority students in schools and the encouragement of multilingualism. Prevalent scholars Flores and Rosa (2015) and Moll et al. (1992) are concerned about bilingual programs that limit students’ use of their mother tongue in schools and thus ask for changing traditional educational models to prevent prejudice in school. Vygotsky (1978) emphasised that a successful education integrates students’ culture into the curriculum and that learning should be socialised and contextualised. Finally, Cluster 5 (Purple) is about the CLIL approach and using rigorous research, consisting of 20 articles, asking to innovate new teaching approaches like CLIL to help increase minority students’ learning ability and employ research rigour to improve teaching and learning. Dominant researchers in this cluster are Coyle et al. (2010) and Cenoz et al. (2014), emphasising the CLIL approach, while Cohen (1988) emphasised the essence of research rigour in behavioural and social sciences research as a good research design helps detect problems based on meaningful impacts and interpretations.

In summary, the study reveals many influential authors. For example, Cummins, Thomas and Collier, and August and Shanahan argue that proficiency in a student’s L1 supports their academic success. In addition, other influentials like García, Creese and Blackledge, and Baker promote additive bilingualism and advocate for translanguaging as a strategy to empower minority students, helping them use both their native language and English to achieve academic goals. Ruiz, Valdez, and Cervantes-Soon et al. advocate for policies that ensure educational equity for minority students. Scholars like Flores and Rosa, Moll et al., and Vygotsky asked to integrate cultural contexts into the curriculum to overcome language biases and support multilingualism. Lastly, Coyle et al. and Cenoz et al. promote innovative, rigorous teaching methods for improving minority students’ learning. Cohen stresses the importance of research rigour to ensure adequate educational strategies.

Generally, these scholars collectively advocate for an equitable, inclusive, and context-sensitive approach to bilingual education.

Interpretation of the Findings

In nearly half a century, the number of bilingual education publications began humbling and fluctuated with minor fluctuations. It then began to surge in 2008 and peaked in 2021, albeit showing a slight decrease in 2023. This significant evolution shows that scholars are concerned about the field.

Regarding the first research question, “What are the hot topics of bilingual education?” The findings highlight several key hot topics in bilingual education that have shaped the discourse over the past four decades. For instance, Translanguaging has been revealed to be the most significant theme. This bilingual education approach aims to bring all students’ language repertoire into the classroom, so it shifts away from traditional models that separate students’ first and target languages. This reflects Ruíz’s (1984) concept, which shifts from seeing language as right or a problem to seeing it as a resource. Translanguaging encourages students to use all their language skills, allowing for a more natural and fluid interaction between languages in the classroom. Translanguaging fosters academic success among bilingual students encouraged to use their first and second languages to enhance understanding and expression. For instance, De los Ríos (2018) found that integrating language minority students’ meaningful backgrounds into the classroom was productive. Fuster and Bardel (2024) found spontaneous translanguaging enhanced multilingual students’ participation and content learning. Through a review analysis, Prilutskaya (2021) found translanguaging is an approach that can improve classroom learning.

However, educational policies at the site can influence how bilingual schooling operates, such as an English-only policy or the language of the elite (Skutnabb-Kangas, 2013), an international lingua franca (Montgomery, 2013). In this case, students’ L1 is regarded as a barrier to achieving L2, not as a resource for students to use their L1 to learn and make meaning in their class. This policy has received criticism as it devalues all students’ linguistic backgrounds and cultures, which are seen as valuable for students to master academic subjects. Many scholars (e.g., Bialystok, 2007; Cummins, 1976, 2000) have argued that if students do not understand the content of a strange language, how can they develop sufficient cognition to master it? In addition, through analysis, any models or policies that harm bilingual education have been critiqued. For example, Proposition 203—Children’s English Law, was found to harm bilingual education (Combs et al., 2005; Johnson, 2012) because, in the long term, students were transferred to English-only classes.

Because of this linguistic choice bias and limited exposure to students’ valuable linguistic and cultural background, bilingual schooling has shifted away from monolingual ideologies toward extra flexible language practices that permit bilingual students to use their linguistic resources completely to empower their school performance and for long term improvement of bilingualism (Creese & Blackledge, 2010; García, 2009).

CLIL and EMI are examples that encourage schooling to use both students’ L1 and the target language in the instruction of academic subjects. CLIL has become prevalent, particularly in Europe, where it has been widely adopted to teach academic subjects in L2. Europeans are encouraged to speak multiple languages besides their mother tongue (European Commission, 2006). CLIL is believed to give students a dual benefit as students can gain foreign language proficiency and simultaneously master subject-specific knowledge, for example, math. However, CLIL has faced challenges related to teacher training and the need for context-specific adaptations. Another related approach is the increasing use of EMI, which involves teaching academic courses in English, especially in non-English-speaking countries. However, this approach plays a very critical role in English for global higher education, so it has encountered many challenges, such as limited English proficiency among content teachers and students and resources (Wu & Tsai, 2022). Besides, this approach is not always welcome due to the risk of cultural marginalisation (Lin, 2018). Together, these topics (Translanguaging, CLIL, EMI) reflect the ongoing evolution of bilingual education in response to globalisation, multiculturalism, and the changing needs of diverse student populations. They are driving forces shaping bilingual education today.

Finally, the analysis unveiled that bilingual education is not just for mass conditions but also for students with special needs. For example, Sign Bilingual Education emerged to help improve literacy and writing skills for deaf and hard-of-hearing students (Swanwick, 2010; Wolbers et al., 2014). Hence, bilingual education is now more inclusive and includes disadvantaged students.

Regarding the second research question, “What are the significant documents of bilingual education?” we found many influential documents shaping bilingual education practices and policies. Among these, the works of scholars such as Jim Cummins, Ofelia García, and Colin Baker mainly contribute to defining key principles in bilingual education. Cummins’ work has offered foundational work for understanding how bilingual students develop cognitive and academic skills in students’ language and L2. His work on language proficiency and interdependence has influenced educational theory and practical applications in bilingual classrooms, aiming to promote additive bilingualism. García’s concept of the Translanguaging approach has helped reshape how educators view bilingual students’ language abilities, using any students’ linguistic and cultural background for academic improvement. To have a more fluid understanding of language use, García argued that bilinguals’ language backgrounds are crucial to supporting meaning-building and academic learning. This concept has led to a broader acceptance of multilingualism in educational settings, where students’ languages are treated as resources rather than barriers (Ruíz,1984). Baker’s influential work considers the diverse language experiences of bilingual students. His work on the bilingual continuum and language acquisition offers a framework for designing a bilingual education model that supports students’ linguistic development while ensuring equal language use.

These foundational works have provided a theoretical and practical framework for promoting bilingual education. They have helped shape educational policies, curriculum designs, and teaching methods considering language rights and allowing multilingualism to spread in schools.

Regarding the third research question, “Who are the significant scholars in the field of bilingual education?” we found that several scholars’ works have influenced both theory and practice in bilingual education. Jim Cummins, Ofelia García, and Colin Baker are consistently recognised as leading scholars. Jim Cummins is well known for his work on the interdependence language proficiency hypothesis. He considers that any methods encouraging students to use their first and second languages are recommended to boost their academic success. This work has prompted educators to think of instructional practices that can foster additive bilingualism rather than subtractive approaches.

Another leading scholar is Ofelia García, who introduced a new perspective on understanding bilingualism called Translanguaging, which supports more inclusive and flexible bilingual education practices. She encourages educators to allow students to use both languages in tandem to improve meaning-making and allow for deeper cognitive processing.

Another influential author, Colin Baker, has helped bilingual educators better understand bilingualism and language acquisition. For example, his work on the bilingual continuum and the advantages of bilingual education has prompted educators to better approach language teaching in bilingual programs. It is worth emphasising bilingualism’s cognitive and social benefits while teaching students bilingually.

Together, these scholars have shaped the theoretical foundation of bilingual education and continue to influence the policies, practices, and pedagogical approaches used in bilingual classrooms worldwide. Their work underscores the importance of valuing students’ linguistic diversity and creating educational environments that support multilingual development.

Implications for the Findings

As defined by many renowned scholars (e.g., Baker, 2011; Cenoz & Genesee, 1998; García, 2009; Genesee, 2004), bilingual education aims to use minority students’ mother tongue and the target language (can be L2, the foreign language, the majority language, or the national language depending on the context) as instruction to improve students’ school performance and achieve Additive Bilingualism. Translanguaging can help promote multilingualism and students’ sociocultural backgrounds (García, 2009). Hence, related stakeholders should adhere to those aims.

For policymakers, the trends in bilingual education emphasise the importance of policies that support additive bilingualism, where both languages are valued equally. Thus, they need to consider policies that encourage multilingualism rather than focus solely on the dominance of one language, especially in multicultural societies (Flores & Rosa, 2023). Adopting Translanguaging and CLIL models is an opportunity for policies to promote flexibility in language use within educational systems. Ensuring equitable access to bilingual education programs can also help reduce language-related disparities. Finally, policymakers should make more inclusive policies, as bilingual education is not just for mass education but should be inclusive; they should invest in innovating strategies to improve academic performance and language skills, such as writing and reading skills, for students with special needs, such as deaf and hard-of-hearing students.

For educators, the shift towards a more inclusive bilingual education system, including approaches like Translanguaging and Sign Bilingual Education for students with hearing impairments, suggests that educators should be equipped with strategies that integrate students’ full linguistic repertoires into their learning process. To make this happen, classroom teachers must use students’ home languages to improve knowledge comprehension and academic success. In addition, as found in the review, incorporating culturally responsive pedagogy can support marginalised communities and ensure students’ cultural identities are recognised and valued. Training programs should also focus on methods to integrate bilingualism into different subject areas, particularly with models like CLIL, which emphasises the dual role of language learning and content mastery. Professional development for teachers on these methods is critical for successful implementation. Training programs should also focus on methods to integrate bilingualism into different subject areas, particularly with models like CLIL, which emphasises the dual role of language learning and content mastery. Professional development for teachers on these methods is critical for successful implementation.

For curriculum designers: As CLIL and EMI models have shown potential for fostering bilingual proficiency while teaching academic content, curriculum designers. These approaches should be adapted to specific educational contexts, with considerations for teacher proficiency, resources, and students’ academic demands. Additionally, curriculum designers should create assessment tools to assess students’ abilities based on translanguaging principles without being penalised for their bilingualism and multilingualism when applying CLIL and EMI. In addition, Figure 6 below can help curriculum designers write a study plan based on the purpose of the training or program.

OPEN IN VIEWER

Figure 6.

The EMI continuum in practice (Thompson and Mckinley, 2018).

This spectrum shows a progression from models where content is taught in the target language to those where the primary focus is on learning the language itself.

Limitations and Future Research Directions

The study has some limitations and research directions to be considered. An essential hindrance of this bibliometric study is its reliance on the Web of Science (WOS) database, which might also exclude applicable research from different databases like Scopus or Google Scholar, mainly due to incomplete coverage (Hsieh & Li, 2024; Hsieh et al., 2023). Therefore, to improve similar research methods, future studies should use databases from other sources, such as Scopus, EBSCO, and ERIC, and include other relevant analyses for more information, such as co-authorship and full-text analysis of papers. Another limitation is that the current study did not synthesise 1.716 articles found after inclusion. It only emphasises the bibliometric features of the papers to reveal the features of intellectual development in bilingual education. Therefore, the analysis might pass over important tendencies because of the time lag in quotation accumulation, specifically for more modern publications. Thus, a more inclusive method in a shorter timeline can yield more focused themes. Furthermore, the choice of thresholds for keyword co-occurrence, reference co-citation, and document bibliographic coupling (e.g., minimum citation or occurrence thresholds) can significantly impact the results (Martínez-López et al., 2018). In addition, bibliometric analysis with VOSviewer often emphasises highly cited documents, potentially overlooking emerging or under-researched areas (Reyes-Gonzalez et al., 2016). This bias may lead to an incomplete view of the topic. Future studies could benefit from using other software like Histcite, Sci2 Tool, or Bibliometrix for better knowledge domain mapping. Finally, geographic and language biases in WOS should restrict the illustration of non-English or region-precise studies, affecting the worldwide scope of the analysis. This limitation can be the most challenging because researchers must make much effort to collect and understand publications in other languages.