Translanguaging and Second-Language Reading Proficiency: A Systematic Review of Effects and Methodological Rigor

Abstract

Translanguaging has become a significant concept in applied linguistics, promoting inclusive education and equitable treatment of languages. However, despite its increasing prominence, the impact of translanguaging pedagogy on second-language (L2) reading proficiency and the methodological rigor of studies in this area remain insufficiently examined. This systematic review, conducted in alignment with the PRISMA guidelines, addresses these gaps by analyzing 21 studies comprising 25 samples that examine the relationship between translanguaging and L2 reading development. The findings indicate a small but statistically significant positive effect of translanguaging on L2 reading comprehension (g = 0.33, CI [0.21–0.45]), though considerable variation exists across studies. Qualitative research (k = 9) generally reports favorable outcomes, while quantitative studies (k = 16) present mixed findings—ten studies show positive effects, whereas six report no significant impact. The methodological assessment highlights several shortcomings, including the absence of a priori power analysis, inconsistencies in reporting instrument and coding reliability, insufficient transparency in data reporting, and vagueness in the implementation of translanguaging practices. Additionally, contextual trends reveal a need for more research in underrepresented regions and secondary school contexts. This review emphasizes the importance of conducting rigorous, contextually diverse research to validate translanguaging as an effective approach for enhancing L2 reading proficiency.

1. Introduction

Translanguaging—defined as the use of a multilingual speaker’s full linguistic repertoire as an integrated system for meaning-making (García, 2011)—has rapidly gained prominence in the field of applied linguistics. Unlike traditional models of bilingualism that conceptualize languages as separate, switchable systems, translanguaging views language use as fluid, dynamic, and shaped by speakers’ identities, communicative intent, and social context (Wei, 2018). This contrasts with multilingual practices such as code-switching or language alternation, which assume clear boundaries between named languages. This reconceptualization of bilingual language use has positioned translanguaging at the forefront of multilingual education research, where it is increasingly explored as a tool to raise awareness for marginalized language groups advocating for their language rights, equity, and inclusivity in education as well as a proposed pedagogical practice for enhancing learning content and improving the learning environment (Kim & Weng, 2022). As such, this topic has become a prominent theme at numerous conferences and has been extensively written about in prestigious journals. A simple Google search for the term “translanguaging” alone yields approximately 1.5 million hits as of 2 December 2024, not accounting for the various interchangeable terms frequently used in translanguaging literature. Considering this significant visibility and influence of translanguaging, calls to investigate the impact of translanguaging on substantive aspects of language learning, such as second-language (L2) reading—the focus of the current study—are recurrently made (Prilutskaya, 2021; Huang & Chalmers, 2023; Kim & Weng, 2022).

Moreover, there is a broader methodological shift within applied linguistics that underscores the need for a more rigorous examination of L2 research methods (Loewen et al., 2020; Plonsky, 2023). The extensive literature on translanguaging provides fertile grounds for scrutinizing research quality. To address these calls, this paper undertakes a comprehensive systematic review of translanguaging with two overarching objectives. First, it aims to investigate the effects of translanguaging pedagogy on a significant domain of language learning, specifically L2 reading proficiency. L2 reading proficiency was selected as the focal skill for this systematic review because reading plays a central role in academic development and is often considered the most critical language skill for academic success (Grabe, 2009; Grabe & Stoller, 2019). In educational contexts, learners are frequently required to engage with complex texts, conduct research, and demonstrate content knowledge through reading. Additionally, translanguaging practices are also increasingly documented in reading-focused instruction, where learners draw on their full linguistic repertoires to construct meaning from texts (e.g., Qureshi & Aljanadbah, 2022; Roussel et al., 2017). Therefore, examining L2 reading allows for both methodological rigor and pedagogical relevance within the scope of this review. Second, the review seeks to assess the methodological rigor and quality of the research evidence within this body of literature, thereby offering insights into future research directions.

2. Literature Review

2.1. Translanguaging and Second-Language Reading

Previous research on translanguaging and L2 reading development presents a mixed picture. Findings of these studies can be grouped into two categories: (a) studies that support the effects of translanguaging, and (b) studies that offer exceptions or indicate inconclusive findings. The studies in the first group emphasize that translanguaging provides additional cognitive support (Storch & Wigglesworth, 2003) in the generation and conversation of ideas (Hawras, 1996). The beneficial effects of translanguaging have been reported across various applications, including mental translations (Hawras, 1996; Kern, 1994), group-reading activities (Seng & Hashim, 2006), L2 reading assessments (Chu, 2017), and negotiation of meaning around the text (Hungwe, 2019). Together, these studies weave a coherent narrative that illustrates the integral role of translanguaging in enhancing L2 reading development through engagement with L1 resources.

In the second group of studies, the impact of translanguaging on second-language reading development is explored with varied conclusions. For young L2 learners in immersion contexts, there is no reported advantage of translanguaging for L2 reading comprehension; instead, a greater use of L2 is reported in non-translingual conditions (Cohen, 1974; Hopewell, 2013; Legarreta, 1977). In experiments involving simultaneous bilingual instructions, a mismatch between learners and teachers’ use of linguistic resources has been observed, where students’ reliance on L1 contrasts with the predominant use of English—the target L2—by teachers (Legarreta, 1977). The L1 preference of young learners diminishes in adult advanced biliterate readers with either a random preference for L1 and L2 resources (Kwon & Schallert, 2016) or a preference for the use of target L2 over L1 by those with an advanced proficiency in the L2 (Upton, 1997). This preference for the L2 use is heightened in Hungwe (2019), where a group of learners with increased exposure to L2 literacy refused to use their L1 in discussions, expressing a greater comfort in utilizing their L2. Similarly, in Allard (2017), a group of learners reported the use of translanguaging by teachers as a disempowering aspect of their learning process, which they believed led to a poor proficiency in L2 and low grades on the assessments, which is a finding affirmed in other research (i.e., Vaish, 2019). This perception of translanguaging as disempowering appears linked to learners’ desire for increased exposure to the target language and a belief that L1 use may hinder L2 mastery. These views contrast with those reported in studies where translanguaging was associated with identity affirmation and academic confidence through the strategic use of learners’ full linguistic repertoire. Overall, the results of previous research, at best, are inconclusive. While some studies support the positive effects of translanguaging on second-language reading proficiency, others either reject these claims or present mixed findings.

Several factors related to study quality can explain the opacity of the previous research findings. To account for these factors, systematic reviews of the literature on translanguaging and L2 development have started to emerge. These reviews not only suggest directing future research on more substantive areas of language learning (e.g., reading and writing) but also emphasize scrutinizing translanguaging literature to advance the methodological rigor—particularly through clearer operational definitions and more robust empirical designs (Chalmers & Murphy, 2021; Treffers-Daller, 2024). The following section presents a summary of previous research evaluating the methodological quality of translanguaging research.

2.2. Translanguaging and Methodological Rigor

Prilutskaya (2021), for example, in her review of 223 empirical studies on translanguaging pedagogy, observes that a major portion of the literature (83%) adopts a qualitative orientation, with only 3% of the studies employing a quantitative design. The author also points out the relatively smaller sample size in these studies and stresses the need for “conducting more controlled intervention and/or mixed methods studies in order to make more substantiated claims regarding the affordances of classroom translanguaging” (p. 15). In addition, Prilutskaya also points out a relative lack of translanguaging research with students from secondary schools, in teacher training courses, and in Central and South America, and in the Middle East. This study emphasizes a need to further explore the effects of translanguaging pedagogy on learning outcomes, including language learning.

Similarly, Kim and Weng (2022), who reviewed 20 empirical studies examining pedagogical translanguaging with reference to TESOL specifically, identify teachers’ lack of knowledge, poor implementation of the translanguaging pedagogy, and issues related to logistics as the major factors affecting the outcome of research. While they highlight that translanguaging contributes to equity, inclusivity in ESL contexts, and scaffolding content in EFL settings, they also point out that participants in some studies (e.g., Allard, 2017; Vaish, 2019) considered it inefficient and time-consuming, which the authors believed could be improved if teachers were trained in translanguaging pedagogy (see also Liu et al., 2020). Moreover, the study itself was mainly a summary review of the research papers included in the analysis, without stating any specific inclusion/exclusion criteria and without reporting the actual outcome of these studies or necessarily analyzing their methodological rigor, which indicates a need for a more methodologically rigorous review of the literature on the topic.

Another narrative review of pedagogical translanguaging by Huang and Chalmers (2023) specifically evaluates the relative trustworthiness of studies by conducting a ‘quality appraisal’, along with examining its impact on substantive learning outcomes (e.g., reading, vocabulary). Overall, the authors report no advantage of translanguaging pedagogy over English-only conditions for vocabulary, grammar, and metalinguistic awareness. For reading, writing, and overall language proficiency, they suggest that there might be some advantage; however, they point out that this is based on a very limited body of relatively low-quality studies. The authors highlight smaller sample sizes, clustering effects in data reporting, and vagueness in the description of the application of translanguaging pedagogy as major methodological weaknesses.

While the authors made a concerted effort to evaluate study quality, it is important to recognize that the limited number of eligible studies identified in each substantive area—(k = 2) for vocabulary, grammar, and metalinguistic awareness; (k = 3) for reading and writing; and (k = 2) for overall language proficiency—highlights a significant limitation in the current body of literature on the topic. This scarcity of evidence underscores the need for more rigorous and comprehensive research in these domains. Moreover, while the study employs a widely recognized tool for assessing study quality, much of the information explaining how specific scores were assigned—particularly for criteria such as ‘scale—sample size’ and ‘dropout’—is located in the Supplementary Materials rather than the main text. Although this is consistent with the standard practice in systematic reviews, including a clearer explanation in the main paper of how decisions were made could enhance clarity and help readers better understand the rationale behind the assigned scores. Despite these limitations, the study makes a concerted effort to closely analyze the quality of research published on translanguaging pedagogy and its effects on essential learning variables in L2 development.

To sum up, the reviews above highlight the need to systematically evaluate the methodological quality of the included studies in a systematic review. Without an explicit and transparent assessment of methodological quality (e.g., sample size adequacy, control for bias, clarity of intervention description), a synthesis may risk aggregating flawed evidence, potentially leading to unreliable or misleading conclusions about translanguaging’s effects. This is supported by Chalmers et al.’s (2024) review of the International Database of Education Systematic Reviews (IDESR), which highlights a neglect of transparent reporting of quality appraisal.

To address these methodological concerns, this paper undertakes a comprehensive systematic review of translanguaging with two overarching objectives. First, it aims to investigate the effects of translanguaging pedagogy on a significant domain of language learning, specifically L2 reading proficiency. Second, it seeks to assess the methodological rigor and quality of the research evidence within this body of literature, thereby offering insights into future research directions.

3. Materials and Methods

3.1. Domain and Scope of the Study

In a systematic review, defining the domain of interest is the first critical step (Plonsky & Oswald, 2012; Plonsky & Brown, 2015). Having a vaguely defined domain may not only result in confounded outcomes but may also lead to incorrect interpretations by stakeholders (Plonsky & Brown, 2015). Considering the significance of the domain definition, it is vital to define the two important terms—‘translanguaging’ and ‘reading comprehension’—that constitute the independent and dependent variables of the study, respectively.

Translanguaging has been defined in a variety of ways. Two major and distinct classifications are provided by Cenoz and Gorter (2020), who classify translanguaging in ‘spontaneous’ and ‘pedagogical’ terms. The former refers to how languages exist in a natural context where bilinguals may fluidly transition between languages, while the latter denotes “a pedagogic theory and practice that refers to instructional strategies which integrate two or more languages” (p. 18). The current study takes the latter approach to examine the use of “one language to reinforce the other to increase understanding and to augment the pupils’ ability in both languages” (Williams, 2002, p. 40). Another significant debate surrounding the term translanguaging concerns its ontological nature. In its stronger form, it is defined as “a single linguistic system” (García & Otheguy, 2020, p. 25), in contrast to the multilingual perspective on language learning, which views this repertoire as “multilingual/multimodal” (cf. Cenoz & Gorter, 2020; MacSwan, 2017). This distinction is important, as it can have serious implications for the survival and maintenance of minority languages (Hickey, 2001; Musk, 2010). However, the aim of the current study is to examine the impact of translanguaging on L2 reading proficiency, rather than to explore the conceptual or theoretical nature of the term itself.

Therefore, circumventing the epistemological debate over what constitutes translanguaging, this study accepts translanguaging as the use of one language in developing another (Lewis et al., 2012). It is worth noting that the literature refers to this phenomenon using various terms—such as bilingualism, hybrid language use, multilingualism, and polylanguaging. However, to avoid ambiguity in scope and ensure conceptual consistency, this review includes only studies that explicitly employ the term translanguaging to describe the use of prior linguistic resources to develop reading proficiency in the target language. Terms like code-switching or language alternation were excluded, as they imply a separation or duality of language systems, which contradicts the unified linguistic system view underpinning the stronger conception of translanguaging.

The second salient variable explored in this study is ‘reading comprehension.’ While research on L2 reading is extensive, covering a range of aspects and processes, our current investigation centers on scrutinizing the improvements in learners’ reading comprehension resulting from translanguaging intervention. Studies may document improvements in reading comprehension through various means, such as changes in learners’ reading scores based on formal assessments, evaluations or observations by teachers or researchers of learners’ reading abilities, or self-reports from students. To ensure a comprehensive analysis, a study using any of these specified methods for reporting reading comprehension in L2 was included in the analysis.

3.2. Literature Search

After the domains of interest for the current study were determined, following the standard procedures for a systematic review (e.g., Plonsky & Oswald, 2012) in line with PRISMA 2020 (Page et al., 2021), a detailed search and review of the literature was carried out based on the following inclusion and exclusion criteria.

3.2.1. Inclusion Criteria

Studies were included if they

• Examined the impact of translanguaging—planned or explicitly claimed;
• Involved second-language (L2) learners as participants;
• Reported outcomes related to reading comprehension, reading proficiency, or reading development;
• Employed an empirical research design;
• In the case of quantitative studies, provided sufficient statistical data for effect size calculation (e.g., means, standard deviations, t values, F values, or correlation coefficients);
• Were published between 1980 and 2024, corresponding to the period after the term translanguaging was first coined by Cen Williams.

However, at a later stage, during the screening process, studies published before 1980 were also examined for inclusion if they were referenced in studies already included, to support or contest the impact of L1 use as a resource. These studies used terms like “L1 as a resource” or “L1 as an asset” to represent the core concept of translanguaging—the use of a learner’s existing linguistic resources in learning new languages. It is worth noting that these alternative terms were not part of the original search strategy. Rather, studies using them were examined individually only when the terms appeared in the primary literature already included in the current review. This process led to the inclusion of only one additional study (i.e., Cohen, 1974).

3.2.2. Exclusion Criteria

Similar to the inclusion criteria, we rigorously applied specific exclusion criteria. Studies were disqualified if they met any of the following conditions:

• Investigated aspects other than reading development (e.g., motivation, identity affirmation, writing, speaking, etc.).
• Examined the impact of prior language resources (e.g., L1 glosses) on L2 reading but did not adopt a holistic approach (i.e., did not use the term translanguaging, entire linguistics repertoire, asset, or resource).
• Addressed multiple components of L2 literacy (e.g., fluency, content knowledge, or transfer), but only one component was related to reading comprehension; in such cases, only the relevant reading-related data were retained and the rest were excluded.
• Did not provide sufficient empirical data to support the quantitative analysis (e.g., lacked results, effect sizes, or clear methodological descriptions).
• Theoretical or conceptual papers without empirical investigation.
• Published in journals lacking peer-review transparency or inclusion in recognized academic indexes.

3.2.3. Search Strategy

The search process started with exploring the key terminology related to translanguaging and second-language reading in Google Scholar and various educational databases. As noted, the body of literature on translanguaging is extensive, and navigating the vast number of available sources can be challenging. To ensure that we retrieved data specifically relevant to our study and to enhance the efficiency of the search process, we used targeted phrase searches. Each phrase (e.g., “translanguaging and reading”, “translanguaging and second-language reading”, “translanguaging and reading”) was entered individually into the database search bar using quotation marks to prompt exact phrase matching. The databases queried for these terms included, but were not limited to, Academic Search Ultimate, Education Full Text, ERIC, Google Scholar, JSTOR, Linguistics and Language Behavior Abstracts (LLBA), ProQuest, PsycINFO, Scopus, Social Science Database, and Web of Science.

The key phrases used in the search process are provided below:

“Translanguaging AND reading”.

“Translanguaging AND reading assessment”.

“Translanguaging AND reading comprehension”.

“Translanguaging AND reading competence”.

“Translanguaging AND reading development”.

“translanguage AND reading engagement”.

“translanguage AND reading levels”.

“translanguage AND reading proficiency”.

“translanguage AND reading gains”.

“translanguage AND reading ability”.

“translanguage AND L2” (followed by all the terms stated above).

“translanguaging AND target language” (followed by all the terms stated above).

“translanguage AND second language” (followed by all the terms stated above).

Details regarding the number of studies retrieved through this process are reported in Section 4.

3.3. Selection Process

After a collection of potential studies was gathered, each study’s title and abstract were skimmed, and the full body of the text was scanned for the variables of interest. While skimming, each study was probed for the existence of any of the following terms ‘translanguaging’ and ‘translanguage’ in tandem with the phrases ‘reading comprehension’, ‘reading proficiency’, ’reading development’, ‘reading levels’, and ‘reading competency’. If these terms were identified in a study, the study was considered qualified for the next stage. This process resulted in a total of 115 potential studies that were thoroughly examined in the next stage.

3.4. Screening

During this phase, two reviewers independently combed through the entire collection of sources, paying special attention to the research questions and the methodology sections of each study within the corpus. The primary objective at this stage was to make a final selection of studies that specifically investigated the influence of translanguaging on reading comprehension in the target language. It is worth noting that, during this phase, we also examined in-text citations and the references cited in the primary studies, leading to the examination of additional studies. These additional studies were referred to in the primary research on translanguaging while making strong claims for or against its impact, even when they did not directly use the term ‘translanguaging’, but adopted a holistic approach and used other similar terms such as ‘entire linguistics repertoire’, ‘L1 as an asset’, or ‘L1 as a resource’, which closely aligned with the concept of ‘translanguaging’ as defined in the domain. The rationale for screening these studies came from the fact that these were already cited in the literature on translanguaging and fell in the domain defined for the current study. In some cases, these studies preceded the emergence of the term ‘translanguaging’ (e.g., Cohen, 1974)—the only study that was included as a result of additional screening, resulting in the total number of potential studies to 116 for a detailed review. It is important to note that, as the main purpose of this research was to assess the quality of the translanguaging research pertaining to L2 reading, we included all the studies that claimed the impact of ‘translanguaging’ or its various manifestations stated above without necessarily—at this stage—assessing how effectively translanguaging was implemented. The specific criteria adopted for inclusion are provided below.

3.5. Data Extraction and Data Items

Since this study combines a systematic review of the qualitative and quantitative studies, specific criteria for analyzing data from these two different methodological approaches were adopted. For quantitative studies, our analysis included examining whether the studies reported

• An a priori power analysis—specifically, whether they indicated the number of participants required for the planned analysis, the statistical tests being powered, the number of outcomes assessed, and other relevant details;
• The actual number of participants in these studies;
• The presence/absence of a comparison group;
• Instrument reliability;
• An assumption check for the selected analysis;
• Details of the study outcome, such as mean values, standard deviations, effect sizes, and t and F values, among other relevant parameters.

For the qualitative studies, adopting a narrative approach, we followed the guidelines outlined in various journals and articles for maintaining rigor in the qualitative research (c.f. Mahboob et al., 2016; Johnson et al., 2020; Stenfors et al., 2020). These recommendations come from prestigious journals in the field of language, education, and medicine. For the current analysis, along with the duration and the actual application of translanguaging in the sample studies, data for the following specific criteria was recorded to evaluate the methodological rigor in the qualitative studies:

• Depth/triangulation: More than one source of data collection is reported.
• Transparency: All the data collected is analyzed/reported.
• Coding details are provided.
• Coder reliability: Coding/re-coding is performed or more than one coder codes the data.
• Instrument reliability: If an instrument (questionnaire/structured interview) is used, its piloting/reliability is reported.
• Inter-rater reliability: Member/participant checking is reported, and the reliability index or percentage agreement is provided.

In addition to the substantive and methodological variables, contextual variables (e.g., setting, country, educational level) were also coded. Data coding was conducted in two stages. Initially, all data were coded by the first author. Subsequently, 33% of the dataset was independently re-coded by the second author, resulting in 100% agreement. A detailed description of all coding variables is provided in

Table 1. Coding variables.

Study Type
Research paper	Dissert/thesis	Conf. proceeding	Book chapter
Setting
Second language	Foreign language	Multilingual
Location
Research Methodology
Quantitative	Qualitative	Mixed
Method
First Language/s
Target language/s
Sample
Number of groups	Power analysis	N/groups	Comparison
Instrument type—quantitative
Data collection—qualitative
Trans-application
Analysis
Assumption checks	Reliability	Coding details	Coder reliability
Result Reporting
Quantitative	M (SD)	t, f, p, etc., values	Effect size
Qualitative	Triangulation	Transparency

.

3.6. Synthesis

To address the first research question, which examined the effects of translanguaging on L2 reading proficiency, two types of outcomes were analyzed. First, we coded the reported effects of translanguaging on L2 reading development in each included study. In this initial stage, we recorded whether individual findings supported, challenged, or reported no clear impact of translanguaging. For studies employing a qualitative design, we adopted a narrative synthesis approach and categorized the reported outcomes as positive, neutral, or negative based on the authors’ interpretations. Second, for studies with a quantitative design, we extracted statistical data (e.g., means, standard deviations, t values, F values) and calculated effect sizes—Hedges’ g—using an inverse variance weighting approach, which assigns greater weight to studies with larger samples and smaller standard deviations to ensure more precise estimates (Oswald & Plonsky, 2010).

To address the second research question, which focused on assessing the methodological quality of the included studies, various design and reporting features—outlined in Section 3.5—were evaluated. Each variable was coded independently by the researchers using a three-point system: 1 = reported, 0 = not reported, and 99 = not applicable. Complete inter-coder agreement (100%) was achieved through multiple rounds of review and discussion.

4. Results

The terms reported in the search strategy were explored in 11 databases, provided in

Table 2. Educational databases consulted to find sources.

	Database	Search Results
1	Academic Search Ultimate	58
2	Education Full Text	22
3	ERIC	77
4	Google Scholar	31
5	JSTOR	329
6	Linguistics and Language Behavior Abstracts (LLBA)	5
7	ProQuest	292
8	PsycINFO	44
9	Scopus	74
10	Social Science Database	87
11	Web of Sciences	164
	Total	1183

. Searching through these databases returned a total of 1183 sources. However, after de-duplication and removing 154 reports, 1029 papers were left for abstract screening, which led us to remove 913 more studies as these did not seem to focus specifically on the relationship between translanguaging and L2 reading. This process left 116 reports to be assessed for eligibility for the full-text screening.

In the next stage, each of the 116 studies was individually screened by both authors, with particular attention given to the method section. This process initially led to the exclusion of 87 studies, with a 93% agreement rate between the two reviewers. Following a re-evaluation of the remaining sources and further discussion, an additional 8 studies were excluded, with 100% agreement. This screening process resulted in 21 studies being retained for detailed full-text coding. The search and screening procedures are outlined in Figure 1.

Summary Characteristics of Included Studies

The final dataset for the current study yielded a corpus of 21 studies published between 1974 and 2022 (the dataset can be accessed at https://data.mendeley.com/preview/zkj24bm8pw?a=ccb68df9-115a-4c2e-8e0c-f86055c5facc, accessed on 23 April 2025).

These studies predominantly comprised research papers while including one dissertation and a paper published in a conference proceeding. These included 9 samples following qualitative design and 16 that adopted a quantitative one. Overall, 25 samples were extracted from these studies for analysis. Three studies included multiple experiments or distinct participant groups. Roussel et al. (2017) involved three separate participant groups; Makalela (2015) reported mean differences across both English and Sepedi; and Cohen (1974) assessed participants at two different time points, one year apart, using separate control groups for each assessment, resulting in a total of 25 samples. Details about the nature of the corpus are provided in

Table 3. Corpus composition: Study types, numbers, and research methods.

Year	N	Articles	Dissert.	Conf. Proc.	K	Qual.	Quant.
1974-22	21	19	1	1	25	9	16

, followed by the summary characteristics of the studies in

Table 4. Summary characteristics of the included studies.

Study ID	Context	Location	TransL	TL	Sample Size	Edu. Level	Design	Results
Qualitative Studies
Carroll and Sambolín Morales (2016)	SL	Puerto Rico	Spanish	English	29	Tertiary	Case Study—Descriptive	Favor
Caruso (2018)	ML	Portugal	Multiple Ls	English	15	Tertiary	Case Study—Intervention	Favor
Ceprano et al. (2018)	FL	Italy	Italian	English	22	Elementary	Case Study—Descriptive	Favor
Hopewell (2013)	SL	U.S.	Spanish/English	English	49	Tertiary	Qualitative (Reflective Practitioner Research)	Favor
Hungwe (2019)	ML	S. Africa	Multiple Ls	English	36	Tertiary	Case Study—Intervention	Favor
Martin-Beltrán et al. (2019)	SL	US	Spanish	English	9	Elementary	Mixed Methods—Action Research	Favor
Martínez-Roldán and Sayer (2006)	SL	US	Spanish/English	English	8	Elementary	Ethnographic—Classroom Study	Favor
Mbirimi-Hungwe (2016)	FL	S. Africa	Multiple Ls	English	181	Tertiary	Qualitative Intervention	Favor
Yang et al. (2021)	SL	US	Chinese	English	1	Elementary	Case Study—Descriptive	Favor
Quantitative Studies
Abdulaal (2020)	FL	Egypt	Arabic	English	63	Tertiary	RCT (pretest–posttest)	Favor
Bin-Tahir et al. (2018)	FL	Indonesia	Not provided	Arabic	64	Tertiary	Quasi-Experimental (pretest–posttest)	Favor
Cohen (1974)	IM	U.S.	English/Spanish	English	30	Elementary	Quasi-Experimental (pre/post longitudinal)	No Difference
Cohen (1974)	IM	U.S.	English/Spanish	Spanish	21	Elementary	----	No Difference
Chu (2017)	FL	Taiwan	Chinese	English	123	Secondary	Quasi-Experimental (posttest-only)	Favor
Iranmanesh and Golshan (2018)	FL	Iran	Persian	English	46	Tertiary	Quasi-Experimental (pretest–posttest)	Favor
Ledwaba (2020)	SL	S. Africa	English/Sepedi	English/Sepedi	70	Elementary	Quasi-Experimental (pretest–posttest)	Favor
Makalela (2015)	SL	S. Africa	Sepedi/English	Sepedi/English	60	Secondary	Quasi-Experimental (pretest–posttest)	Favor
Makalela (2015)	SL	S. Africa	Sepedi/English	Sepedi/English	60	Secondary	----	Favor
Nur et al. (2020)	FL	Indonesia	Not provided	English	35	Secondary	Quasi-Experimental (pretest–posttest)	No Difference
Qureshi and Aljanadbah (2022)	FL	UAE	Arabic	English	65	Tertiary	Quasi-Experimental (pretest–posttest)	No Difference
Roussel et al. (2017)	FL	France	French	German	102	Tertiary	RCT (pretest–posttest)	Favor
Roussel et al. (2017)	FL	France	French	English	84	Tertiary	----	Favor
Roussel et al. (2017)	FL	France	French	English	108	Tertiary	----	No Difference
Sefotho and Makalela (2017)	ML	S. Africa	ML	English	60	Elementary	Quasi-Experimental (pretest–posttest)	No Difference
Yafele (2021)	FL	S. Africa	ML	English	25	Tertiary	Quasi-Experimental (pretest–posttest)	Favor

.

Since these studies met the inclusion criteria of the current research paper, these were further analyzed for the contextual, substantive, and methodological variables. For the contextual variables, five major aspects were coded from the 25 samples extracted from the corpus. These variables included information about settings in which a study was conducted, the name of the country, the translanguaging orientation, the approach to the target-language development—L2 or multiple languages, and educational level of the participants in these samples. Details about the contextual variables are displayed in Figure 2, Figure 3, Figure 4, Figure 5 and Figure 6.

Based on this analysis, as shown in Figure 2, the majority of the samples (k = 12) came from studies conducted in foreign-language (FL) settings, followed by 8 studies in second-language (SL) contexts. Only three samples in the dataset operationalized the setting as a multilingual (ML) context, while two samples from a single study reported the context as immersion (IM).

In terms of the countries where these studies were conducted (see Figure 3), South Africa and the US reported seven and six samples, respectively, followed by France and Indonesia, reporting three and two studies, respectively. The rest of the seven samples were conducted in seven different countries. It is worth noting that a mere two countries, South Africa and the US, reported more than half of the studies (52%), while the remaining 48% of the studies (k = 12) were carried out in nine different countries.

For the translanguaging orientation (see Figure 4), most of the studies (k = 12) adopted a multilingual orientation as a support language where participants were allowed to use all of their linguistic resources while working around a text in the target language. However, a considerable number of studies (k = 11) adopted a monolingual—L1—approach, and participants were supported with L1 resources to negotiate meaning around the text in L2. There were two studies that did not report the translanguaging orientation.

In terms of target-language development (see Figure 5), most of the studies (k = 19) examined English as the target language. Only two studies targeted the development of participants’ home and target languages (in this case, Sepedi and English), whereas three studies targeted the development of Arabic, German, and Spanish as an L2, respectively.

Regarding the educational context (see Figure 6), most of these studies were conducted with participants who were attending tertiary education (k = 13), followed by those attending elementary schools (k = 8). Surprisingly, only four studies (k = 4) were conducted with participants enrolled in secondary schools, indicating a need for a future focus on this population. The smaller number of studies at the secondary level may reflect limited access to this population or a research focus skewed toward early literacy development and post-secondary academic contexts. This underrepresentation points to the need for further research involving secondary school learners.

To address the first research question—What is the effect of translanguaging on L2 reading proficiency?—two types of outcomes were analyzed. First, for all the studies included in the analysis, we coded for the effects of translanguaging on L2 reading development as reported in each sample. This analysis revealed that all the studies with a qualitative design (k = 9) supported the positive impact of translanguaging on L2 reading comprehension. For the samples with a quantitative design, the majority of studies (k = 10) reported positive effects of translanguaging on L2 reading comprehension, while some studies (k = 6) reported no impact of translanguaging. Second, for the samples with a quantitative design, we extracted data (i.e., M, SD, t, F values) needed for computing effect sizes. The results of the two analyses—the outcome as reported in each study and those obtained through computing effect sizes—are provided in

Table 5. Effect size comparison: Translanguaging and non-translanguaging conditions.

Study	Outcome as Reported in the Samples	Hedge’s g	CI Lower	CI Upper
Abdulaal (2020)	Yes	0.31	2.80	4.40
Bin-Tahir et al. (2018)	Yes	2.61	1.94	3.28
Cohen (1974), 1 *	No	0.23	−0.43	0.90
Cohen (1974), 2	No	0.13	−0.53	0.80
Chu (2017)	Yes	1.15	0.76	1.53
Iranmanesh and Golshan (2018)	Yes	−0.63	−1.23	−0.04
Ledwaba (2020)	Yes	−0.75	−1.23	−0.26
Nur et al. (2020)	No	0.72	0.03	1.40
Qureshi and Aljanadbah (2022)	No	0.05	−0.44	0.54
Roussel et al. (2017), 1 *	Yes	1.22	0.09	1.54
Roussel et al. (2017), 2	Yes	1.13	0.78	1.48
Roussel et al. (2017), 3	No	0.92	0.63	1.22
Sefotho and Makalela (2017)	No	0.30	−0.21	0.81
Makalela (2015), 1 *	Yes	−2.50	−2.98	−2.02
Makalela (2015), 2	Yes	−1.82	−2.25	−1.39
Overall weighted effect size	6/10	0.33	0.21	0.45
Yafele (2021)	Yes	--	--	--

* Roussel et al. (2017), Cohen (1974), and Makalela (2015) represent multiple samples from the same study.

.

The individual effect sizes for each sample indicate a greater variability in the effect of translanguaging on second-language reading comprehension. These results reveal three patterns:

• Studies showing significant positive effects (e.g., Bin-Tahir et al., 2018; Chu, 2017; Roussel et al., 2017, 1 and 2).
• Studies exhibiting strong negative effects (e.g., Makalela, 2015, 1 and 2; Ledwaba, 2020; Iranmanesh & Golshan, 2018).
• Samples showing no significant effects (e.g., Cohen, 1974; Qureshi & Aljanadbah, 2022; Sefotho & Makalela, 2017).

The overall weighted effect size suggests a small but significant positive effect g = 0.33, CI [0.21–0.45], revealing that, on average, the impact across these samples leans towards a favorable impact of translanguaging on L2 reading. However, the heterogeneity, as reflected across these studies in Table 5, underscores a need to examine the methodological rigor across these studies.

The heterogeneity of results in Table 5 clearly indicates a need for examining the second research question of the current study, which seeks to assess the methodological rigor and quality of the research evidence within this body of literature, thereby offering insights into future research directions. To address the methodological quality of the studies, various aspects of the samples—as discussed in the analysis section—were assessed.

The data for the qualitative studies reveals a rich picture when it comes to the length of treatment and application of translanguaging pedagogy. While one-shot studies do exist, the majority of samples in this set spanned from four treatment sessions to a semester of observation. Out of the total nine samples in this set, two used literacy circles, two applied story retelling or bilingual use when reading a story, while the rest of the five studies reported allowing students to use any of their semiotic resources to negotiate meaning while discussing a text either with peers or the course instructor. When it comes to attending to the criteria of depth/triangulation, the majority of studies reported collecting data from more than one source. However, for transparency and coding details, five out of nine studies met the criteria, whereas the remaining four samples either reported selective results or provided limited information about coding details. In most of the cases (k = 6), only a single coder coded all the data. The inter-rater or coding/re-coding reliability details were reported in only two samples included in the analysis. Summary results pertaining to the methodological quality of the studies in the qualitative dataset are presented in

Table 6. Methodological characteristics of the studies with a qualitative design.

Study ID	TL Application	Duration	Depth/Triangulation	Transparency	Coding Details	Coder Reliability	Inter-Rater Reliability	Other Observations
Carroll and Sambolín Morales (2016)	Literacy circles	1 month	1	1	1	1	0	- TAMZ Analyzer not explained. - Only 18% of data showing TL were analyzed. - Coder reliability: discussed. - No details about agreement or issue resolution provided.
Caruso (2018)	Any L in discussion; presentation in 3 Ls	17 h	1	0	0	0	0	- No details about the analysis provided. - 10/13 completed opinion survey. - 6/10 supported using translanguaging. - Remaining 4 make up 40% of students.
Ceprano et al. (2018)	Bilingual read-aloud and guided readings	10 days	1	1	0	0	0	- No coding details. - Only 3 out of 22 students were interviewed. - No comparison group involved.
Hopewell (2013)	Literacy circles	4 sessions	1	0	1	0	0	- Analyzed group dialogues only. - Written recalls were not analyzed. - No member-checking/second coder reported.
Hungwe (2019)	Discuss text: any L Half-class paraphrase: English Half in home language	1 session	0	0	0	0	0	- Selective results reported. - The decision is based on teachers’ observation. - No formal assessment was carried out.
Martin-Beltrán et al. (2019)	Buddies discussed Qs with younger buddies: L1	15 sessions, 45 min	1	1	1	1	1	- Coded collaboratively. - Details provided. - 10/24 recordings analyzed.
Martínez-Roldán and Sayer (2006)	24 retelling sessions in alternate languages 40 min each 1-week observation 2 discussion sessions	1 semester	1	1	1	0	0	- Used a holistic rubric. - Did not provide coder reliability.
Mbirimi-Hungwe (2016)	TL: Read/discuss: any L CG: Read/discuss: English	1 session	1	0	0	0	0	- Used a rubric. - Mentioned inter-rater reliability. - No index for reliability provided. - No details about the rubric are provided. - Data analyzed ‘qualitatively’; no details provided.
Yang et al. (2021)	Bilingual use reading story books at home	3 months	1	1	1	1	1	- Detailed information about data collection and coding. - Results are based on interaction analysis. - No specific measure of comprehension was used.
			8	5	5	3	2

K = 09; 1 = Yes; 0 = No: Depth/triangulation: More than 1 source of data was collected; Transparency: All the data collected is analyzed/reported; Coding details: Provided; Coder reliability: More than 1 coder coded the data; Inter-rater reliability: Member/participant checking reported along with the reliability index or percentage agreement.

.

In the quantitative dataset, a limited duration for the treatment was observed. While there are longitudinal studies reported in a minority of samples (k = 5), ranging from 17 h to a year-long treatment, the majority (k = 8) report a limited length of treatment, ranging from 15 min to 1 class session only. Three samples in the analysis do not report information about the length of treatment.

When it comes to the application of translanguaging, five major patterns emerge. These are depicted in Figure 7.

In the majority of cases (k = 9), multilingual support is provided or allowed in various ways, as depicted in

Table 7. Application of translanguaging pedagogy.

Explanation/answers to students’ queries in L1	Reading text in one language—answering in another
Sight words in L1	Text glossed in L1
Discussion in L1/L2	Summarizing in L1/L2 or any language
Retelling a story in L1/L2	Multilingual explanations
Matching keywords in L1 with those in L2	Note-taking/annotations in any language
Reading a story in two languages	Brainstorming/outlining in any language
Alternation of languages in vocabulary, silent reading, and reading aloud exercises	Translingual collaboration
Print environment in two languages	Using multilingual dictionaries and Google Translate

. Out of the remaining seven samples, two report using simultaneous bilingual instruction; two provide an assessment text and test questions in L1; one provides a pre-reading text in L1; and two samples do not provide any information about the treatment conditions.

A detailed summary of the methodological quality of the studies in the quantitative dataset is presented in

Table 8. Methodological characteristics of the studies with a quantitative design.

Study ID	TL Application	Duration	A Priori Sample	Comparison Group	Coding/Scoring	Reliability Check	Assumption Check	Statistical Analysis	Observations
Abdulaal (2020)	Pre-reading in L1	4 weeks	1	1	0	1	1	Kruskal–Wallis	- Did not state that r was for reading/writing task - Low instrument reliability (0.67)
Bin-Tahir et al. (2018)	Not provided	Not stated	0	1	0	0	0	t-test	- Most methodological criteria not reported
Cohen (1974), 1	Simultaneous bilingual instruction	1 year	0	1	0	0	0	Not reported	- Details about data collection or analysis not reported
Cohen (1974), 2	----	1 year	0	1	0	0	0	RM-ANOVA	- No difference between the TG and CG
Chu (2017)	Qs in L1 Chinese	1 session	0	1	1	1	0	t-test correlations	- Group receiving Qs in L1 sig. outperformed the L2-only option
Iranmanesh and Golshan (2018)	Explanation/answer to queries in L1	1 semester	0	1	0	0	0	t-test	- Trans. group outperformed the L2-only group
Ledwaba (2020)	Sight words—L1 Discussion—L1/L2 Retelling story—L1/L2 Matching L1 words with L2	30 min	0	1	0	0	0	t-test	- Reliability not provided for reading task - Significant difference favoring trans. group
Makalela (2015), 1	A story and activities in 2 Ls Print environment in 2 Ls Reading text in 1 L and answers in another	1 session	0	0	0	0	0	t-test	- No statistically significant difference - Mean gains in posttest in both languages: English/Sepedi
Makalela (2015), 2	----	1 session	0	0	0	0	0	t-test	- No statistically significant difference - Mean gains in posttest in English and Sepedi
Nur et al. (2020)	Text and MCQs: English and Indonesian	Not stated	0	1	0	0	0	t-test	- Results: mean scores only - Mean scores for both groups dropped in the posttest after the treatment
Qureshi and Aljanadbah (2022)	Text glosses in L1 Discussion in L1/L2 Summary in L1/L2	1 session	1	1	1	1	1	t-test	- No difference between the two groups
Roussel et al. (2017), 1	Reading in L1 Reading in L2 Reading in L2 with L1 translation	15 min	0	1	1	0	0	ANOVA	- Condition with L2 German and L1 - French translation outperformed L2 (German)
Roussel et al. (2017), 2	----	15 min	0	1	1	0	0	ANOVA	- Condition with L2 English and L1 - French translation outperformed L2 (English)
Roussel et al. (2017), 3	----	15 min	0	1	1	0	0	ANOVA	- No significant effect of language condition on the content knowledge
Sefotho and Makalela (2017)	Not provided	Not stated	0	1	0	0	0	t-test	- No significant difference between the 2 groups
Yafele (2021)	Reading activities in any L Translingual collaboration Multilingual dictionaries and Google Translate	1 session	0	1	0	0	0	t-test	- Instrument is not described - Results: t-test is not reported - M shows decline for the CG in the posttest - Qualitative data: coding not provided - Findings: not reported for either of the (2) studies
			1	14	5	3	2

K = 16; 1 = Yes; 0 = No; L = Language; TG = Treatment Group; CG = Control Group.

.

Overall, these studies show there is room for significant improvements. In terms of sample size, only two studies reported conducting a priori power analysis to determine the number of participants needed for the study, while the remaining studies (k = 14) did not report this information. To determine the effects of the translanguaging condition on learners’ reading comprehension, most samples in the quantitative set (k = 14) reported involving a comparison group. There were two samples, both from the same study, which assessed learning gain in students’ reading ability compared to their initial state. While this reflects the impact of translanguaging on learners’ reading ability in general, this may not be sufficient for making claims, either positive or negative, about the effects of translanguaging pedagogy in comparison to the typical target-language-only condition.

For the remaining three variables—item scoring details, reliability checks, and assumption checks—only five, three, and two studies, respectively, provide the required information. The predominant method applied for the statistical analysis was the use of a t-test (k = 9), followed by one-way ANOVA (k = 3), and RM-ANOVA and Kruskal–Wallis, 1 each. There was one sample in the study which did not provide details about the statistical analysis and another study which reported using correlation analysis along with t-test analysis.

5. Discussion and Conclusions

The current systematic review was conducted with two main objectives: first, to explore the effects of translanguaging pedagogy on L2 reading proficiency, and second, to examine the methodological rigor of studies exploring this relationship. To address these objectives, 21 studies comprising 25 samples were thoroughly analyzed using the standard practices outlined in the PRISMA 2020 guidelines for conducting systematic reviews, as well as recommendations made by experts in the field (e.g., Page et al., 2021; Plonsky & Oswald, 2012). Contextual analysis revealed that most studies were conducted in South Africa and the United States, with 52% concentrated in these two countries. The studies predominantly focused on the effects of prior languages on the development of English as the target language (k = 19) and involved participants at the tertiary (k = 13) and elementary levels (k = 8). These patterns highlight the need for further exploration of translanguaging’s impact in other contexts, particularly in developing languages other than English or both languages involved in the treatment. Additionally, there is a significant gap in research involving participants at the secondary education level, highlighting the need for future studies focusing on this population.

The analysis of outcomes showed a generally positive impact of translanguaging on L2 reading comprehension. The overall weighted effect size, Hedge’s g = 0.33, [0.21–0.45], indicated a small but significant positive effect. However, considerable heterogeneity was observed across the studies. Qualitative samples predominantly reported favorable outcomes, highlighting the pedagogical value of translanguaging in promoting learner engagement and comprehension. In contrast, quantitative studies presented mixed results, with 10 studies showing positive effects while 6 reported no significant impact.

These differences in reported outcomes may, in part, reflect both the differing epistemological orientations and the methodological designs of qualitative and quantitative research. Qualitative studies often emphasize learner perspectives and classroom interactions, which may more readily capture the affective, social, and contextual benefits of translanguaging. However, it is important to note that qualitative studies, such as case studies and ethnographies, while they may offer deep insights into the processes learners undergo during their learning journey, are not designed to establish causal relationships. Hence, caution should be exercised in interpreting these findings. In contrast, quantitative studies typically rely on standardized assessments of reading proficiency, which may not be sensitive to short-term gains or the more nuanced, indirect benefits of translanguaging pedagogy. In addition, the comparative analysis of the studies reviewed revealed important methodological distinctions. Qualitative studies generally involved longer treatment durations and more open-ended pedagogical designs but often lacked coding transparency and inter-rater reliability. Quantitative studies, while more likely to use control groups, frequently implemented brief treatments and underreported key methodological elements such as power analysis, reliability checks, and statistical assumptions. This combination of assessment focus and design limitations may contribute to the more favorable findings in qualitative research. Future studies adopting mixed-methods approaches that combine the depth of qualitative inquiry with the rigor of quantitative analysis are essential to developing a more comprehensive understanding of translanguaging’s impact on L2 reading proficiency.

Beyond the comparative differences between qualitative and quantitative studies, a broader methodological review of all included studies revealed several recurring weaknesses. Many lacked a priori power analysis, failed to consistently report instrument and coding reliability, and provided limited transparency in data reporting. Inconsistencies were also noted in the application of translanguaging practices, which may have contributed to the variability in findings. Overall, these findings emphasize the potential of translanguaging in enhancing L2 reading proficiency, but underscore the need for robust methodological practices in future research.

Addressing these methodological challenges aligns with the growing emphasis on improving research rigor in applied linguistics. The recent critical appraisals propose several steps to advance the field (see Al-Hoorie & Vitta, 2019; Loewen et al., 2020). For quantitative studies, these recommendations suggest improvements in existing practices, such as recruiting underexplored participants (i.e., very high/low performers), conducting research in marginalized settings, creating detailed instruments, reporting instrument reliability, providing evidence for validity, and ensuring assumption checks for the inferential statistics (Al-Hoorie & Vitta, 2019; Loewen & Hui, 2021; Plonsky, 2023). Additionally, these recommendations emphasize enhancing research rigor and transparency by offering statistical training for researchers, fully reporting significant and non-significant results, and adopting open science practices, such as sharing datasets for replication (Gass et al., 2021; Plonsky & Brown, 2015). To facilitate these improvements, researchers can access a range of tools and platforms that support methodological transparency and statistical rigor. These include the Center for Open Science (n.d.), which offers resources for open and reproducible research; OpenStatsLab, which provides guided statistical tutorials using applied linguistics datasets; and JASP Team (n.d.), an open-source software that enables accessible and replicable statistical analysis. Similarly, for qualitative studies, several suggestions—mainly derived from health and education research—advocate for stating a clear purpose; providing detailed context; and offering thorough descriptions of data sources, examples of data coding, and details of analyses (Johnson et al., 2020; Yadav, 2022).

These recommendations have started gaining attention within the field of SL research. For instance,

• Prominent journals in the field now require a complete reporting of descriptive and inferential statistics, including confidence intervals, exact p-values, and effect sizes (see author guidelines for Language Learning, TESOL Quarterly, and SSLA).
• Researchers are encouraged to share datasets, questionnaires, coding schemes, instruments, etc., on publicly accessible platforms, including the Instruments for Research into Second Languages (IRIS: https://www.iris-database.org) and the Open Science Framework (OSF: https://osf.io).
• A new journal Research Methods in Applied Linguistics exclusively focusing on research accuracy and rigor has been launched.
• The journal Language Learning has introduced “registered reports”, a new category for publication that requires authors to submit the rationale, proposed methods, and analytical procedures for review before data collection.

In conclusion, while translanguaging pedagogy is making a promising contribution in developing L2 reading proficiency, the variability in findings suggests a need for improvements in the methodological rigor of the studies. Employing more robust methodological practices, including a priori power analyses, detailed reporting of reliability and validity measures, and adherence to open science principles would further improve the quality and the credibility of this research. Future research can follow the guidelines proposed by prestigious journals, such as Language Learning and TESOL Quarterly, or by professional organizations such as the Standard Protocol Items: Recommendations for Interventional Trials, SPIRIT (SPIRIT Group, n.d.) and CONsolidated Standards of Reporting Trials (CONSORT Group, n.d.), to ensure a complete data reporting and transparency. Additionally, expanding the scope of research to underrepresented populations, non-English target languages, and diverse educational settings will further clarify the efficacy and applicability of translanguaging. By prioritizing these areas, future research can contribute to a more comprehensive and nuanced understanding of the impact of translanguaging on L2 reading proficiency than has hitherto been the case.