Implementation and Effects of Pedagogical Translanguaging in EFL Classrooms: A Systematic Review

: In response to calls for more research to assess the effects of translanguaging on substantive learning outcomes, this systematic review begins that process by synthesising existing research on pedagogical translanguaging approaches that have been formally assessed for effectiveness in English as a Foreign Language (EFL) classrooms. A systematic process of searching and selecting the literature found 10 eligible studies. Data were extracted for narrative synthesis and quality appraisal. Results showed that pedagogical translanguaging was most frequently used to teach reading and writing rather than other domains. Five studies favoured translanguaging over English-only approaches, four of which were rated as having a high risk of bias. The remaining studies either detected no statistically signiﬁcant differences between these approaches or favoured translanguaging in a small number of highly speciﬁc measures. We conclude that pedagogical translanguaging may be helpful in teaching in some circumstances, but that the low methodological quality of the literature on average makes it difﬁcult to draw ﬁrm causal inferences. This systematic review provides the relevant background on what is already known about the effects of pedagogical translanguaging in EFL contexts to inform the work of researchers wishing to act on recommendations for more intervention research to establish the effects of the approach on English language outcomes.


Background and Rationale
For decades, there has been extensive discussion about involving students' home languages in the classroom (Duarte 2019;Leung and Valdés 2019;Lin 2019).As multilingualism is common, linguistic diversity among teachers and students should be seen as a readily available pedagogical resource instead of a limitation (Chalmers and Murphy 2022).Translanguaging has been proposed as a practical theory of how multilingualism might be brought to bear on the educational experiences of multilingual children (Li 2018) against the backdrop of a paradigm shift away from monolingual standards in the understanding of language learning and use (Anderson 2022).Vogel and García (2017) posited three core theoretical foundations for translanguaging.First, multilingual individuals have only one unitary linguistic system, from which they select and use features for communication, rather than two (or more) internally differentiated linguistic systems-or languages, in the common understanding of the term.Second, translanguaging prioritises dynamic linguistic and semiotic practices over named languages in the understanding of multilingualism.Third, translanguaging asserts that named language categories and structuralist language ideologies have material effects, such as the ideas of language purity and verbal hygiene.Translanguaging does not reject the existence of different named languages (Arabic, Chinese, German, etc.) defined by nations and states.However, translanguaging suggests that these are not materially important for understanding multilingualism.In addition, much of the literature on translanguaging has positioned it as an approach to social justice for linguistic minority children, empowering them to see their multilingual practices as an asset, as an important characteristic of their personal identities, and a way to disrupt traditional monolingual hegemony.Alongside these, translanguaging has also been held up as a way to improve linguistic competence in general through related pedagogical approaches.
These assertions, especially as they relate to the internal and external representation of languages among multilingual people, have been criticised for ignoring (indeed contradicting) empirical evidence on the nature of bi/multilingual language practices (MacSwan 2022a; Auer 2022).These criticisms point to a wealth of empirical evidence, accrued over decades in the fields of sociolinguistics (see MacSwan 2017) and bilingual first language acquisition (see Genesee 2022), demonstrating that languages are discretely represented, both in the way they are held in the mind and in the ways in which they are used by multilingual individuals.As an approach to pedagogy, translanguaging has also been criticised for lacking a robust evidence base from which to make causal claims about its effect on learning outcomes (Chalmers and Murphy 2022).
The origins, theoretical underpinnings, and ideological positioning of translanguaging have been well rehearsed elsewhere (see García 2009;García and Li 2014;Conteh 2018;Cenoz and Gorter 2021; inter alia), as have criticisms of it (MacSwan 2022b).Therefore, we limit our focus to one aspect of this field of enquiry: pedagogical translanguaging and its effects on target language development.Cenoz and Gorter (2021) distinguished between spontaneous translanguaging and pedagogical translanguaging.The former refers to "the reality of bilingual usage in naturally occurring contexts where boundaries between languages are fluid and constantly shifting" (p.18).The latter is "a pedagogic theory and practice that refers to instructional strategies which integrate two or more languages" (p.18).One of the aims of related scholarship is to assess whether use of the first language (L1) of multilingual learners in pedagogical contexts is beneficial to second language (L2) teaching and learning (Macaro 2022).
From a translanguaging point of view, use of the whole linguistic system (that is, both L1 and L2) in pedagogical contexts is seen as beneficial as it helps encourage the creativity and criticality of multilingual learners and maximise their linguistic potential (Li 2022).There are also compelling theoretical bases for why attending to L1 might provide scaffolding by which developing competence in L2 can be supported (Cummins 1979(Cummins , 1980)).However, because of a paucity of robust intervention research, it remains unclear whether pedagogical translanguaging is beneficial in the sense that it helps to improve language proficiency, and in particular target language proficiency, in objectively demonstrable ways.This lack of direct evidence has not stopped people making strong claims that is does (Chumak-Horbatsch 2019;Celic and Seltzer 2012;Espinosa et al. 2016;Duarte 2020;inter alia).Clearly, if strong claims are to be made for the educative effects of translanguaging, these must be supported with strong empirical evidence.Prilutskaya (2021) conducted an extensive systematic review of research into the pedagogical use of translanguaging in English language teaching (ELT) classrooms across various settings.Her principal aim was "to shed light on the current state of knowledge about the affordances of translanguaging pedagogies . . .and to highlight possible avenues for future research" (p. 1).Her review acts as a comprehensive mapping of this field of enquiry, identifying research that has been conducted into the classroom use of translanguaging and outlining the principal foci in this literature.
The resulting synthesis of 233 publications demonstrated that translanguaging has been explored in a wide variety of contexts through the lenses of a variety of methodological and pedagogical approaches.The largest body of research (83% of the included studies) adopted qualitative designs, typically ethnographies, interviews, and observations, focusing on a very small number of learners.In the main these explored attitudes towards translanguaging, discussions of how to affect mindset change towards multilingual practices in schools, and explorations of affordances and barriers associated with enacting translanguaging pedagogies.Far fewer (3%) adopted designs suited to the collection of quantitative data.In only some of these (number not reported) did the researchers seek to objectively assess the educative effects of translanguaging.
On the basis of her findings, Prilutskaya (2021) made several recommendations for future research.Among these was that the "predominance of small-scale qualitative studies needs to be addressed by researchers by conducting more controlled intervention and/or mixed methods studies in order to make more substantiated claims regarding the affordances of classroom translanguaging" (p.15).Working from the principle that new research should be informed by existing evidence about an area of enquiry, this systematic review aims to establish what is already known about the effects of pedagogical translanguaging on substantive learning outcomes so that researchers can take that knowledge into account when planning new primary research in response to Prilutskaya's recommendation.
We have chosen to focus on synthesising research evidence of the effectiveness of pedagogical translanguaging when it is used in EFL contexts.As Prilutskaya's review demonstrates, pedagogical translanguaging has been explored in an extensive variety of contexts, with different and sometimes competing conceptualisations of what 'effective' means.Addressing the field of EFL specifically allows us to be more focussed in our conceptualisation of both context and effectiveness, and thus reinforce the way the included studies complement each other.We operationalise the former as pedagogical contexts where English is the focus of teaching and learning, in settings where English is not the language of the majority.We operationalise the latter as substantive measures of English language proficiency.Thus, our systematic review aims to provide focused evidence relevant to the practices of EFL teachers and learners, rather than those in other programme types.
Like Prilutskaya (2021), we include only studies that describe themselves as evaluating "translanguaging", specifically.This is because, notwithstanding valid criticism (MacSwan 2022b), translanguaging has been established as a term that differs in important ways from other terms common in bilingualism and bilingual education research (such as codeswitching).Although some interventions that use different terms might be perceived as translanguaging, usage of the term provides evidence that researchers have framed their interventions around translanguaging theory specifically, rather than other, competing, theories such as linguistic interdependence (Cummins 1979).Unlike Prilutskaya (2021), we also include quality appraisal of the literature in our synthesis.This element of the synthesis provides the reader with an understanding of the relative trustworthiness of the research and thus evidence upon which to base their interpretation of the overall findings.
We hope that our review will provide useful substantive evidence of what is known about the educative effects of pedagogical translanguaging in EFL contexts, inform classroom practitioners, and provide an authoritative starting point for researchers wishing to build on this body of evidence through new primary research.

Methods
This systematic review is reported in line with PRISMA (Preferred Reporting Items for Systematic Review and Meta-Analysis) reporting guidelines (Page et al. 2021).

Review Questions
This systematic review addresses two questions: 1.
What approaches to translanguaging have been assessed for effectiveness in EFL classrooms? 2.
What are the effects of these approaches on English language proficiency?

Inclusion and Exclusion Criteria
We included studies based on

•
Publication date-They were published in or after 2009.We consider 2009 to be 'ground zero' for pedagogical translanguaging as this is when it was first popularised by García (2009).

•
Study type-experiments and quasi-experiments that report empirical data on the effects of pedagogical translanguaging.

•
Context-English language was the subject of teaching in settings where English is not the language of the majority or one of the formal languages of the country.Studies were included if the lessons were described as EFL, English for academic purposes (EAP), and English for specific purposes (ESP), and at least one of the interventions being compared was described as translanguaging.

•
Participants-The population was typically developing learners.That is, participants were not exclusively learners with specific learning difficulties or special educational needs.

•
Outcomes-They reported data on teachers' implementation of pedagogical translanguaging and its substantive effects on any measure of English language proficiency.
We excluded studies based on

•
Research context-English was the medium of instruction for curriculum learning and where English is the language of the majority or one of the formal languages of the country.Studies were excluded if the lessons were described as English as a second language (ESL), content-and language-integrated learning (CLIL), English medium instruction (EMI), or were part of a bilingual model of education.

•
Outcomes-Studies were excluded if they did not report data on the substantive effects of pedagogical translanguaging on English language proficiency.

Search Strategy
We searched the following databases: British Education Index, Education Resources Information Center (ERIC), Education Abstracts, MLA International Bibliography, Linguistics Collection, PsyINFO, ProQuest Dissertations and Theses, Web of Science Social Sciences Citation Index (SSCI), and Conference Proceedings Citation Index-Social Science and Humanities, Scopus, and Google Scholar.
The following search terms were used to search the databases for literature published from 2009 onwards: translanguag* AND (ELT OR "English language teach*" OR "English as a foreign language" OR EFL OR TESOL OR EAP OR ESP OR "teaching English to speakers of other languages" OR "English for academic purposes" OR "English for specific purposes") AND (effect* OR outcome* OR score* OR benefit* OR development* OR result*).
The search was conducted in February 2022.

Selection Process
We first screened the titles and abstracts of all the records returned by the search.Records that could not be excluded on the basis of the information contained in the title and abstract were retained for full-text screening.Any records that clearly violated any one of the inclusion criteria were excluded.After the screening of titles and abstracts, the full papers of all potentially eligible studies were obtained and screened.As a check and balance of the reliability of the screening process, a random sample of 5% (n = 224) of the titles and abstracts, and a random sample of 30% (n = 9) of the full texts were screened by two independent reviewers: the first author and a graduate student in educational applied linguistics with experience and training in systematic reviewing.Agreement in the title and abstract screening was 88%.Results were compared and disagreements resolved through discussion.Agreement was 100% for the sample of full texts.Reports that met all inclusion criteria were retained for data extraction and synthesis.

Data Extraction and Data Items
We recorded three categories of data: (1) general information: reference, source, language, publication type, and source of funding; (2) descriptive data: context, participants, methods; (3) analytical data: the pedagogical translanguaging approach(es), outcome measures, conclusions, and limitations.The outcome measures were quantitative measurements of students' English language proficiency, including means, standard deviations, and effect sizes (if reported).An example of the data extraction form can be found in the Supplementary Materials.Data were extracted by the first author and checked by the second author.

Quality Assessment
We used Gorard's Sieve (Gorard 2015), a quality appraisal tool recommended for use with educational research, to evaluate the trustworthiness of each study in relation to design, scale, dropout, outcomes, fidelity, and validity.Each study was assigned an overall star-rating from 4*, meaning the study has provided the most trustworthy type of evidence, to 0, meaning the study cannot be said to have addressed sources of bias adequately, or at all.Per the guidance for use of this tool, "an evaluation will be judged to be as good as the lowest classification it has achieved for each of the six categories" (Gorard 2015, p. 6).
Design was rated based on the extent to which the groups could be considered comparable at baseline.Scale was rated in terms of sample size per group.Dropout was rated based on the level of attrition.Outcomes was rated based on the validity of the way outcomes were measured (for example, whether standardised tests were used, or whether researcher-designed assessment tools had been validated).Fidelity was rated according to the level of clarity of intervention and delivery.Validity was rated based on how well the results could be representative of the similar EFL learners that were not involved in the studies.Per the tool's guidance, where relevant information was not provided, the corresponding item was rated as 0.More detailed information about the method by which these items were assessed, and how ratings were informed, can be found in the tool's supporting materials (Gorard 2015(Gorard , 2021)).
Quality assessment was carried out by both authors independently.Agreement was 90%.The assessment of the one study over which there was disagreement was resolved through discussion.

Synthesis
Based on our understanding of the literature before starting this review and the nature of our review questions, we expected eligible studies to be heterogenous in terms of the characteristics of the interventions and the type of outcomes measured.Therefore, we chose to adopt a narrative method of synthesis (Popay et al. 2006) rather than a statistical method, such as meta-analysis.To address RQ1, we summarised the general approaches to translanguaging adopted across the body of the eligible literature.To address RQ2, we classified each study by its pedagogical focus(es) and outcome measures, then narratively summarised the findings of the studies by focus and outcome.For example, studies that assessed the effects of pedagogical translanguaging on English writing were considered together, studies that assessed the effects of pedagogical translanguaging on English reading were considered together, and so on.Due to the heterogenous nature of the literature, we were unable to weight individual studies for the purposes of the synthesis; therefore, vote counting was adopted.Vote counting is a method for synthesising evidence in which the number of studies showing a benefit is compared to the number of studies showing no benefit to provide an estimate of the overall direction of the findings of the body of evidence.Vote counting alone does not take into account the quality of the studies, the size of the sample, or the size of the effect (as is the case in meta-analysis).We recognise the inherent limitation of vote counting and thus provide a running commentary on the trustworthiness of each study (see Section 2.7) in reporting our findings, to help the reader infer the relative strength of those findings.

Results
After deduplication, the titles and abstracts of 4479 records were screened.A total of 4450 records were excluded on the basis of the information contained therein.This left 29 potentially eligible records, all of which were located for full-text screening.Following full-text screening, 10 studies were assessed as having met all eligibility criteria and were included in the synthesis (see PRISMA flow diagram in Figure 1).

Records identified from:
Databases

Summary Characteristics of Included Studies
Table 1 summarises the characteristics of the included studies.The earliest eligible study we located was published in 2018.Two studies were published in 2019, three in 2020, two in 2021, and two in 2022.Studies were conducted in geographically relatively diverse contexts (East Asia, North Africa, Mainland Europe, and the Middle East), with participants representing a variety of L1s, mostly the majority languages of the setting (e.g., Arabic, German, and Japanese), but in some cases, a small number of non-indigenous languages (e.g., Albanian, Greek, and Kurdish).* for items marked as unclear, the reports did not provide sufficient information to understand how they operationalised translanguaging.** The symbol refers to the 'star rating' for trustworthiness, per Gorard (2015).4 being the most trustworthy.
Writing was the most frequent assessment focus, followed by reading and grammar.Two studies assessed overall English proficiency.One study assessed vocabulary, and one study focused on speaking and writing.No studies focused on listening exclusively, that is, disaggregated from a measure of overall English proficiency.
Studies ranged in duration from a single classroom session to 23 weeks.Only one study conducted a delayed post-test (three months after the end of the intervention period).
The final column of Table 1 reports the results of our quality assessment.Two studies were rated 3*, meaning that their overall quality was high.Two studies were rated 2* (moderate).The remaining studies were rated 1* or 0 (low).The most frequent causes of moderate-and low-quality ratings were study designs that insufficiently accounted for potential biases (and therefore from which confident causal inferences cannot be made), scale, non-standardised outcome measures, and failure to report relevant information.The full breakdown of how these ratings were derived can be found in the Supplementary Materials.

Translanguaging Approaches That Have Been Assessed for Effectiveness in EFL Classrooms
Our first review question asked what approaches to translanguaging have been assessed for effectiveness.Translanguaging has been interpreted differently by different people, with some making the distinction between translanguaging as a theoretical or philosophical construct related to the way multilinguals use language, and translanguaging to reify this construct through classroom practice.We noted in the introduction to this paper that while reified translanguaging has been explored empirically, it was unclear to us the extent to which this reification has been evaluated as a pedagogical approach to language teaching.In keeping with the disparate definitions of classroom translanguaging we witnessed in the literature in general, we found a similar variation among the included studies.Hopp et al. (2021) and Hopp and Thoma (2021) used the term pedagogical translanguaging to describe their interventions.Abdulaal (2020) described his intervention as stego-translanguaging.In addition to translanguaging, the terms heteroglossic, plurilingual, or translingual pedagogies were used variously in the reports by Abdulaal (2020), Hopp and Thoma (2021), Llanes and Cots (2020), and Qureshi and Aljanadbah (2022).
In general, the way these different terms were operationalised in the included literature can be divided into two overarching forms: strong and weak (García and Lin 2017).Strong forms were when fluid use of both/all languages to engage in classroom tasks was encouraged by the teacher.Weak forms were when participants were directed to use one or the other of their languages by the teacher for specific tasks.In some studies, we saw exclusive use of weak forms; in some studies, we saw exclusive use of strong forms; and in some studies, we saw the use of both.
Exclusive use of weak forms was implemented by Abdulaal (2020), Hopp et al. (2021), and Hopp and Thoma (2021).Abdulaal (2020) gave participants an Arabic language pretask reading assignment, which informed their engagement in writing task conducted in English.Hopp et al. (2021) and Hopp and Thoma (2021) used a suite of activities designed to draw their participants' attention to grammatical and lexical similarities and differences between English and German (the majority language of the setting of their study) and other languages represented in the classroom.These activities occupied 20% of instructional time; otherwise, English was used.Exclusive use of strong forms was implemented by Llanes and Cots (2020) and Qureshi and Aljanadbah (2022).Llanes and Cots (2020) encouraged participants to use both or either of their languages for peer discussions, note taking, and talking with the teacher.They gave participants exemplar texts in L1, and they asked them to think about how they would respond to a writing prompt in their L1 as they planned their response in English.Qureshi and Aljanadbah (2022) provided reading materials that included Arabic (L1) translations of key words, asked participants to summarise their reading of an English text in both or either Arabic or English, encouraged participants to discuss their learning in both languages, and allowed the use of Google Translate throughout the task.Both strong and weak forms were evaluated separately in the studies by Sano (2018), Turnbull (2019), and Yüzlü and Dikilitaş (2022).Sano (2018) and Turnbull (2019) adopted very similar approaches, directing participants to discuss and plan a response to an English writing prompt using only Japanese (L1 in both studies) before writing in English (weak), and encouraging the use of both languages for those discussions (strong).Yüzlü and Dikilitaş (2022) adopted a mix of strong and weak forms in the same lessons.For example, deliberately switching between Turkish (L1) and English for input and output, comparing Turkish and English versions of the same texts (weak), and encouraging paired and group discussions using both languages, providing bilingual dictionaries, translation apps, and bilingual word walls (strong).Tang and Peng (2019) and Nur et al. ( 2020) did not provide sufficient information to understand how they implemented translanguaging in their studies.Tang and Peng (2019) said that they used a "2 + 1 + 1" (p.147) approach to teaching phonics.They defined this as 1 /2 of the teaching being in English, 1 /4 in Chinese, and 1 /4 through the use of realia, pictures, and videos, but they did not elaborate on what this looked like in the classroom.Nur et al. (2020) gave no description of the intervention used in their study.

The Effects of These Approaches on Substantive Educational Outcomes
Our second review question asked what the effects of the interventions identified in the included literature were on substantive educational outcomes related to learning English.To assess this, we first identified the primary outcomes in each study.Two studies (Tang and Peng 2019;Yüzlü and Dikilitaş 2022) assessed all four skills (reading, writing, speaking, and listening) but collapsed these into one figure for overall English proficiency in their analysis (in the case of Yüzlü and Dikilitaş, these were divided into receptive and productive proficiency).Specific English language skills constituted the primary outcomes in the other studies included in the review.These were vocabulary, grammar, and metalinguistic awareness (Hopp et al. 2021;Hopp and Thoma 2021), reading (Abdulaal 2020;Nur et al. 2020;Qureshi and Aljanadbah 2022), writing (Abdulaal 2020;Llanes and Cots 2020;Sano 2018;Turnbull 2019), and speaking (Llanes and Cots 2020).Below, we summarise each study and their findings by outcome measure, and provide commentary on their methodological characteristics.

Overall English Proficiency
Tang and Peng (2019) examined the effects of using translanguaging during phonics instruction on the overall development of English language proficiency.The participants were six Chinese primary school students.The pedagogical translanguaging group (n = 3) was taught using English, Chinese, and visual or audio aids, while the control group (n = 3) was instructed only in English.The results showed that the translanguaging group scored statistically significantly higher than the control group on the combined scores of reading, writing, speaking, and listening.The trivial sample size suggests that a type one error cannot be ruled out.Additionally, the researchers did not provide details about the tests used to assess proficiency, and the validity and reliability of these tests were not reported.Given these limitations, this study was rated 0 for trustworthiness.
Yüzlü and Dikilitaş (2022) conducted two parallel experiments, one that compared translanguaging with a Grammar Translation method (focusing on analysing language), and one that compared translanguaging with Communicative Language Teaching (focusing on using language).The setting of the study was a high school in Turkey.Participants (n = 120) were high-achieving teenagers aged between 14 and 17 years.They were allocated to classes based on their age and English proficiency: two classes of 30 at pre-intermediate level and two classes of 30 at upper-intermediate level.Intact classes were then randomly allocated to either the translanguaging condition or to English-only controls, making these cluster randomised trials with one case per arm.The pre-intermediate control group was taught with a Grammar Translation method, while the upper-intermediate control group was taught with a Communicative Language Teaching method.The results showed that the translanguaging groups outperformed their respective controls statistically significantly on measures of productive and receptive English: pre-intermediate/Grammar Translation (F(1, 57) = 74.364;p < 0.05, η 2 = 0.56) and upper-intermediate/Communicative Language Teaching (F(1, 57) = 16.853;p < 0.05, η 2 = 0.22).However, the analysis was compromised as the authors did not appear to have accounted for clustering when calculating the statistic.This study was rated 2* for trustworthiness.
In summary, the evidence of the effects of translanguaging approaches to EFL teaching on general English language proficiency suggests that it may be advantageous.However, this evidence is very limited and of relatively low quality.

Vocabulary, Grammar, and Metalinguistic Awareness
We found two reports (Hopp and Thoma 2021;Hopp et al. 2021) that assessed the effects of pedagogical translanguaging on vocabulary, grammar, and metalinguistic awareness.
Hopp and Thoma (2021) conducted a cluster randomised trial among 258 fourthgrade students in Germany, to investigate the effects of pedagogical translanguaging on understanding English whquestions and the active and passive voices.One intact class in each of four participating schools was allocated at random to the translanguaging group and one to the business-as-usual control.The interventions were conducted in two 45 min lessons across two weeks.The translanguaging group spent approximately 20% of their time engaged in language comparison exercises, where the grammatical constructions under investigation were systematically compared to constructions that served the same functions in German.Participants with knowledge of other languages were asked to say how these functions are served in their languages as well.Otherwise, all instruction was in English.In the control group, the same constructions were taught using only English, i.e., without the language comparison exercises.Primary outcomes were gains between students' pre-and post-test scores relating to answering subject whquestions (e.g., Which animal bites the horse?), object whquestions (e.g., Which animal does the horse bite?), and understanding the active and passive voices.A third outcome, subject and object relative clauses, was measured to assess whether learning transferred to other grammatical conventions that were not directly addressed in the lessons (i.e., whether the intervention had an effect on meta-linguistic awareness).The results showed that the translanguaging group scored statistically significantly higher in the acquisition of object whquestions between pre-and post-tests (β = 1.06,SE = 0.49, z = 2.187, p = 0.029, d = 0.466).On all other measures, both groups made comparable gains.Hopp et al. (2021) examined the effects of pedagogical translanguaging on English language learning outcomes of 122 primary school students from four schools in Germany.The way participants were allocated to interventions was the same as in Hopp and Thoma (2021), i.e., cluster randomisation.Those in the intervention group spent 20% of the lesson time on translanguaging activities across 35 lessons for six months, while those in the control group did the same activities using only English.Outcomes were English receptive and productive vocabulary, English receptive grammar, phonological awareness, and metalinguistic awareness.Although both intervention and control groups improved on all measures over time, no statistically significant differences were found between the two groups on any measure.There was also no statistically significant difference between students who spoke the majority language (German) and those who spoke the minority languages in these learning gains in either of Hopp's and colleagues' studies.As with Yüzlü and Dikilitaş (2022), however, Hopp and colleagues did not appear to have accounted for clustering in their analyses.Both of Hopp's and colleagues' studies were rated 3* for trustworthiness.
In summary, the small body of evidence on the effects of pedagogical translanguaging approaches to EFL teaching on measures of vocabulary, grammar, and metalinguistic awareness suggests that it is generally no more advantageous than using only English.The one exception to this general finding relates to the similarity between English and German grammatical conventions.Hopp and Thoma (2021) attributed the larger gains by the translanguaging group on object whquestions to the difference in how these are constructed in German and English.They suggested that for conventions that differ substantially between languages, translanguaging may help to draw attention to these differences, making them more salient and thus better remembered.

Reading
We identified three studies that explored the effects of pedagogical translanguaging on English reading attainment (Abdulaal 2020;Nur et al. 2020;Qureshi and Aljanadbah 2022).
EFL students at Port Said University, aged between 19 and 24, participated in Abdulaal's (2020) study.They were allocated either to a translanguaging group (n = 33) or an English-only group (n = 30).The method of allocation was not reported.Both groups took part in a four-week course that involved reading and writing activities intended to prepare them for the International English Language Testing System (IELTS) assessments.The translanguaging group was "asked to read first in Arabic about some of the topics involved" (p.466).The English-only group "was taught in a monoglossic way; no intervention of Arabic support" (p.466).Both groups were then assessed on IELTS reading comprehension tests.The translanguaging group statistically significantly outperformed the English-only group, with a mean difference of approximately 1.5 of an IELTS band (IELTS exams are scored from 0 to 9) (p < 0.001).
Nur et al. ( 2020) did not provide a description of their intervention or comparator, saying only that they assessed the effects of a translanguaging strategy on a reading comprehension test (again, not described) among eighth-grade students in Indonesia.The authors say that the translanguaging group statistically significantly outperformed the comparator (p = 0.05).The extremely poor quality of reporting of this trial makes it impossible to draw any inferences about how this contributes to our knowledge.
Qureshi and Aljanadbah (2022) investigated the effects of translanguaging on Arabic university students' English reading comprehension.The participants were all male, majoring in business and IT.The texts used in the reading task were different for the translanguaging and English-only groups.Both groups read a paragraph of text, summarised it, discussed their summaries with their peers, and then answered comprehension questions about the text.The translanguaging group was provided with the Arabic translation of 14 key terms in the text and was allowed to use any languages while completing all aspects of the task, as well as being given access to Google Translate.The English-only group was instructed to use only English and was not provided the Arabic language support (key words, Google Translate).The results showed no statistically significant differences between the two groups in any of the reading measures assessed (t = −1.84,p = 0.85, d = 0.04).Abdulaal (2020) and Qureshi and Aljanadbah (2022) were rated 1* for trustworthiness.Nur et al. ( 2020) was rated 0.
In summary, a limited and very low-quality body of evidence suggests that translanguaging approaches may be advantageous on measures of reading comprehension.

Writing
We identified four studies that assessed the effects of translanguaging on English writing (Abdulaal 2020;Llanes and Cots 2020;Sano 2018;Turnbull 2019).
The methods used by Abdulaal (2020) are reported in Section 3.3.3.In addition to assessing reading comprehension, IELTS writing tests were also used to assess participants' writing proficiency.No inferential statistical analyses were reported, though box and whisker plots comparing the distribution of scores for each group show the median score of the translanguaging group to be higher than the median score of the English-only group by approximately 1.5 of an IELTS band.There is considerable overlap in the whiskers of both groups.This study was rated 1* for trustworthiness.Llanes and Cots (2020) examined the effects of translanguaging on writing skills in business English classes at a university in Spain.The participants were 54 Catalan/Spanish bilingual students learning English as L3.Intact classes were allocated to either the translanguaging group (n = 35) or the English-only group (n = 19) and were taught by two different teachers (meaning teacher effect and other systematic differences between groups at baseline cannot be ruled out as responsible for any differences observed at outcome).Over the course of 16 weeks, for three hours per week, students engaged in the same curriculum content.Students in the translanguaging group were encouraged to use Catalan, Spanish, and English while engaging with that content.For example, when learning to write a business letter, they were asked to think about how they would write one in Spanish, review the structure of a Spanish exemplar, and then write one in English.The English-only group was strongly discouraged from using any language other than English.
Writing was assessed holistically by "an expert examiner with more than 30 years of experience with Cambridge Assessment English exams" (p.8) who graded them from 1 to 5 on each of four domains: language, communicative achievement, content, and organisation.In addition, measures of written lexical complexity (Guiraud's Index of Lexical Complexity), fluency (average number words per T-unit), accuracy (errors per T-unit), and grammatical complexity (clauses per T-unit ratio) were calculated.The translanguaging group made statistically significantly greater gains on the holistic assessment.Both groups made comparable gains on all other measures.This study was rated 1* for trustworthiness.Sano (2018) studied how four different types of pre-writing discussion affected the written production of two Japanese university students.The four conditions were no prewriting discussion, prewriting discussion in English, prewriting discussion in Japanese, and prewriting discussion using translanguaging (i.e., both English and Japanese).The effect of each mode of pre-writing discussion was assessed by asking the participants to say which parts of their writing they felt had been influenced by the pre-writing discussion, then calculating the type token ratio of content words in those sections of their scripts.No inferential statistics were calculated, which is unsurprising given the sample size of two.From the descriptive data, the reviewers found no patterns of advantage associated with translanguaging.This study was rated 0 for trustworthiness.
Turnbull (2019) also investigated prewriting discussions among Japanese university students, aged between 18 and 20 years.Participants were in two classes of 30 students each.One class was studying in the Department of Literature, while the other was in the Department of Economics.Participants in each class were randomly allocated to engage in either an academic writing task or a creative writing task.They were further allocated (method of allocation not reported) into one of three conditions: English-only (pre-writing discussion conducted in English), weak-form translanguaging group (Japanese for the discussion, English for the writing task), and strong-from translanguaging group (any languages for discussion, English for the writing).The resulting essays were scored based on content, organisation, vocabulary, language use, and mechanics.The results showed that the strong-form translanguaging group scored statistically significantly higher than the English-only group in both writing tasks in both classes.The strong-form translanguaging group also statistically significantly outperformed the weak-form translanguaging group, but only in the Department of Economics class.The effect sizes for the Department of Literature class (η p 2 = 0.535 for academic writing and η p 2 = 0.592 for creative writing) were slightly larger than those for the Department of Economics class (η p 2 = 0.520 for academic writing and η p 2 = 0.530 for creative writing).This study was rated 1* for trustworthiness.In summary, the evidence from a small body of low-quality studies suggests that there may be an advantage associated with using translanguaging approaches, such as providing L1 reading materials and discussing the task in L1, prior to writing in English.

Speaking
Only one study included speaking as an outcome (Llanes and Cots 2020).See Section 3.3.4for the methods used in this 1* study.Outcomes were assessed in two ways.
An 'expert examiner' assessed spoken proficiency using a holistic grading rubric, and objective measures of fluency, accuracy, and lexical grammatical complexity were taken.The translanguaging group demonstrated an advantage in the overall score given by the examiner (F(1, 55) = 10,878, p = 0.002).No statistically significant differences in any of the objective measures were found.

Listening
No included studies focused on the effects of pedagogical translanguaging on listening comprehension specifically, although listening skills were assessed as part of the overall proficiency tests in two studies (Tang and Peng 2019;Yüzlü and Dikilitaş 2022).There is, therefore, a notable gap in our understanding of the effects of translanguaging on teaching EFL listening.

Discussion
The aim of this systematic review was to identify and describe the approaches to pedagogical translanguaging in EFL classrooms that have been formally assessed for effectiveness, and to report the effects of these approaches relative to alternatives on substantive measures of English language proficiency.Despite the popularity of pedagogical translanguaging as a concept, and the apparent faith in the approach as a mechanism by which to improve English language outcomes, we found very little research that would help us to understand whether that faith is well placed.In total, we located only ten eligible studies.Moreover, collectively, the methodological/reporting quality of these studies was such that we are not confident about drawing causal inferences on the basis of it.Nonetheless, the assembled literature provides a window into the way that pedagogical translanguaging has been framed within EFL teaching, and some starting points for considering how it might be explored further in better-designed and better-reported research.

How Has Pedagogical Translanguaging Been Operationalised in EFL Classrooms?
It is worth prefacing this part of the discussion with reminder that, for the purposes of this review, we were only interested in reports of research that both described and formally evaluated approaches to pedagogical translanguaging.While previous research exists that describes how pedagogical translanguaging has been operationalised in EFL classrooms (e.g., Cenoz and Gorter 2022), in the absence of formal evaluations of those approaches relative to clearly defined linguistic outcomes, they are limited in their usefulness for practitioners wanting to understand both the questions of 'how' and 'to what ends' pedagogical translanguaging might be operationalised in these contexts.
We found that pedagogical translanguaging can be categorised into two principal forms: strong and weak, in accordance with the taxonomy provided by García and Lin (2017).Strong-form translanguaging is where conditions are created in which students are encouraged to use both/all their languages fluidly and interchangeably throughout the learning process.For example, we found that for the duration of a lesson, students were encouraged to use all their languages for group and paired discussions, note taking, talking with the teacher, and constructing an English language text.Weak-form translanguaging is where the use of one or other language is prescribed for particular tasks within the larger activity.For example, we found that students were given an exemplar text in their L1 to read in preparation for a writing task that was conducted entirely in English.In some cases, we found that a combination of strong and weak forms was used over the course of a learning programme.For example, we found studies where an exemplar text in L1 was provided in preparation for a discussion task conducted in both L1 and English, and then followed up by a writing task conducted only in English.

What Are the Effects of Pedagogical Translanguaging on English Language Proficiency?
Findings were equivocal as to whether strong or weak forms of translanguaging were more or less likely to improve English language outcomes.Of the two studies that adopted strong forms exclusively, one did not detect a difference in outcomes compared to the instruction that used only English, and one found mixed results, favouring translanguaging on some outcomes but not on others.Of the three studies that used weak forms exclusively, one study favoured translanguaging, one study found that translanguaging was no more or less effective than English-only approaches, and one found that it was helpful for one of their two outcome measures, but not the other.Of the three studies that incorporated both strong and weak forms, two favoured the translanguaging approaches, and one found that translanguaging appeared effective for some measures but not others.The studies that did not provide sufficient detail to understand how they operationalised translanguaging both argued that they were more effective than English-only approaches.
Study quality did not moderate these findings.The only studies to receive a relatively high grading for trustworthiness (3*) adopted weak forms and found either that it made no difference (Hopp et al. 2021) or that an advantage was seen only in one of the four outcomes measured (Hopp and Thoma 2021).Arguably, the remaining studies are too compromised by poor design choices/incomplete reporting to allow for any confident assessment of their contribution to addressing this question.
To the question of whether different domains respond differently to pedagogical translanguaging approaches, again, the assembled literature is equivocal, compromised by limitations in their designs and reporting.Collectively, the two studies that adopted measures of overall English proficiency (rated 0 and 2* for trustworthiness) found an advantage associated with translanguaging approaches.Similarly, the three studies that focused on reading proficiency (rated either 0 or 1* for trustworthiness) generally reported an advantage associated with translanguaging.Writing was addressed in four studies (rated either 0, 1*, or 2* for trustworthiness).Where subjective measures of writing were used, translanguaging was considered to have provided an advantage, especially when pre-writing tasks were conducted in L1 or in both languages.Where objective measures were used, no such advantage was detected.In the one study that assessed speaking proficiency, the results mirrored those of writing-when objective measures were used, no statistically significant differences were found between translanguaging and English-only groups.When subjective measures were used, translanguaging was argued to have resulted in improved scores.No studies reported listening proficiency alone, that is, disaggregated from holistic scores of overall language proficiency.
The most illuminating studies in the assembled literature were those by Hopp and Thoma (2021).Methodologically, these were the most robust (both rated 3*), allocating participants to comparison groups on the basis of chance; recruiting a sufficient number of participants to allow for meaningful inferential statistics to be calculated; adopting objective measures of vocabulary, grammar, and meta-linguistic awareness; and running for a sufficient period of time to allow outcomes to emerge in the famously lengthy process of second language acquisition.In one (Hopp et al. 2021), no statistically significant differences between the translanguaging group and the English-only group were detected.In the other (Hopp and Thoma 2021), translanguaging was associated with improved understanding of an English grammatical convention that differed in form to the same convention in L1 of most of the participants.Where grammatical conventions were similar, no such advantage was detected.

The Effects of L1: Comparison with Previous Review Findings
This systematic review found that the overall picture of the included studies had some different characteristics to studies on translanguaging more broadly.Prilutskaya's (2021) systematic review found that the majority of empirical studies meeting her inclusion criteria were conducted in primary and secondary schools, and that 15% of the studies were in pre-schools.Our systematic review, with its focus solely on EFL classrooms, found that half of the ten included studies were conducted in universities, a third (n = 3) in primary schools, and a fifth (n = 2) in secondary schools.No study was conducted in pre-schools.
Thus, there remains little empirical evidence on the effects of pedagogical translanguaging in EFL contexts, at all levels of education.
In an echo of a related systematic review of L1-mediated pedagogy among primary aged learners (Chalmers 2019), our systematic review did not find sufficient robust evidence of the effects of pedagogical translanguaging on English language proficiency to form a confident judgement.However, despite the preponderance of the evidence in Chalmers' review being equivocal, some studies provided relatively robust evidence that using the L1 of students helps them to acquire new L2 vocabulary.This finding was reinforced by a review by Shin et al. (2020), who also found evidence that providing L1 translations and glosses of unfamiliar vocabulary was more effective than L2-only explanations.The only study that assessed vocabulary knowledge in our systematic review did not find an advantage associated with translanguaging, The other primary outcomes explored in Shin et al.'s (2020) review were grammatical accuracy and writing.Their findings relating to grammar were equivocal.They located two relevant studies, one of which found an advantage to a translanguaging approach compared to an L2-only approach, the other of which found the opposite.The findings relating to writing were generally supportive of a translanguaging approach.Importantly, Shin et al. did not assess the methodological quality of their assembled literature, leaving an uncertainty about how robust these findings are.On measures of grammar, our systematic review found that a positive role for translanguaging was limited based on the grammatical construct being studied.In studies assessing writing, we found that pedagogical translanguaging led to higher scores on subjective assessments by examiners but not in objective measures of lexical or grammatical choices.This suggests that pedagogical translanguaging might be more effective in teaching EFL learners to convey their ideas with better communicative achievement, content, and organisation but not on the language choices they make.In other words, the vocabulary and grammatical features in their written production were similar in terms of complexity and diversity as that of their peers taught with an English-only approach.Although more research is needed to confirm this assumption, such a finding implies that positive learning outcomes of pedagogical translanguaging may be limited to specific outcomes and the way in which those outcomes are measured.

The Findings and the Translanguaging Literature
The findings relating to RQ1 (how has translanguaging been operationalised?)showed that interventions tended to follow translanguaging theory closely vis-a-vis utilising all linguistic and multimodal resources to help language learning.As Li and Lin (2019) argued, translanguaging as pedagogy should not only integrate the use of different languages but also enable all students to actively participate in the co-construction of knowledge.In the included studies, all interventions were described as promoting teachers' and students' use of their L1 in class.Some of these approaches might be demonstrations of knowledge comaking.For example, students may have learned from each other during peer discussion using translanguaging in ways that might not have been possible if only English had been used.
What's in a Name?However, some interventions may be difficult to interpret in the translanguaging space because of the ambiguity around the term and how it is operationalised.Some of the principal voices in this area of study are clear in differentiating translanguaging from other related concepts such as code-switching, code-mixing, and cross-linguistic pedagogy (see, for example, García 2021; García and Li 2014;Vogel and García 2017;García 2019).While we understand the philosophical motivation for this differentiation, empirically, what occurs in the classroom might reasonably be assumed not to care about this distinction.Nonetheless, in preparing this review, we took these voices at their word that there is something unique about translanguaging and included only studies where the intervention was labelled as such.The most notable difficulty we encountered related to this methodological choice was how to define strong and weak forms of pedagogical translanguaging, and what the implications of this distinction are for research and practice.While researchers generally agree that strong-form translanguaging allows bilinguals to use all their linguistic endowments, the characteristics of weak-form translanguaging remain unclear.García and Lin (2017) proposed that weak-form translanguaging acknowledges language boundaries but calls for softening of these boundaries.What this softening means in practice is ambiguous.One included study, Turnbull (2019), explicitly described the interventions as either 'weak' or 'strong'.The former was operationalised as using L1 for preparation and English for the task.The latter was operationalised as integrating both languages for preparation and then English for the task.However, a more recent theoretical paper by Cenoz and Gorter (2022) offers an interpretation of weak-form translanguaging as the inclusion of all languages across the curriculum, but where individual lessons use only one of those languages, which appears to be more similar to Turnbull's classification of strong-form translanguaging.The inconsistency around these terms, and whether the distinction is empirically important, would benefit from attention in the scholarship.
The results of RQ2 (what are the effects of these approaches?) showed that in EFL contexts, there is limited reliable evidence of the effects of the identified approaches to pedagogical translanguaging on substantive measures of English language proficiency.This finding may seem to contradict many studies (e.g., Barahona 2020;Fang and Liu 2020;Liu and Fang 2020) that report teachers' and students' perceptions about the effects of translanguaging for ELT.However, it should be noted (as Prilutskaya (2021) amply demonstrates) that many suggestions for incorporating translanguaging into ELT have been built on evidence from qualitative data through questionnaires, interviews, and classroom observations on other benefits and roles of translanguaging, not on the objective evidence that pedagogical translanguaging can improve English language proficiency.It was not the purpose of this systematic review to assess the effects of translanguaging on outcomes such as building a multilingual identity or relieving anxiety, important though these are.However, the actual or potential benefits for target language proficiency are an important factor to consider in any discussion of the effects of pedagogical translanguaging.Indeed, strong claims have been made about these effects, as we have already noted.As we have demonstrated, however, many of these claims are not well supported by empirical evidence.
This review focused on EFL settings only, but the discussion of the extent to which translanguaging is helpful in general is not new.According to Hickey et al. (2014), Williams (2000Williams ( , 2002) ) claimed that translanguaging would be most beneficial to those who have developed reasonable proficiency.He believed translanguaging was most suitable for helping to retain bilingual proficiency rather than to teach L2.Although two decades have passed since Williams proposed this, neither Williams' claim nor related claims-for example, that "monolingual practices do not work in linguistically diverse classrooms" (Chumak-Horbatsch 2019, p. 16, emphasis added)-have been adequately addressed in carefully designed comparisons with alternative approaches, for which L2 (or specifically, in this case, English) proficiency is the primary outcome.
Taking together the results of the two review questions, our systematic review found (1) consistencies and inconsistencies between translanguaging theory and the actual pedagogical translanguaging approaches and (2) a lack of trustworthy evidence on the effects of the approaches on English language proficiency.Overall, further work on translanguaging is needed.As Cummins (2021) has suggested, the legitimacy of theoretical constructs and claims of translanguaging can be problematic if the claims are inconsistent with relevant empirical evidence, internally contradictory, and do not help promote effective pedagogy and policies.

Implications
This systematic review found reports of a variety of approaches to pedagogical translanguaging.They all challenged traditional monolingual English-only approaches to ELT.These findings can help EFL teachers revisit their attitudes towards the use of learners' languages in the classroom, reflect on their teaching, and be inspired to explore how to approach leveraging students' L1 in their learning of English.
Translanguaging in EFL contexts is an emerging research area.We recommend that teachers and policymakers continue to pay attention to the evidence of the effects of pedagogical translanguaging, so that they can make well-supported pedagogical decisions in their classrooms.Teachers can try to adjust their teaching if pedagogical translanguaging is suitable for their lessons when taking into consideration other factors as well.These factors may include teacher and student attitudes, the teacher's teaching style, the purpose of the lesson, and school policies.This systematic review found some potential benefits of pedagogical translanguaging in teaching English grammar and writing.These two teaching goals can be the starting point if teachers would like to implement pedagogical translanguaging.
However, teachers are advised not to overestimate the benefits of pedagogical translanguaging on English language proficiency.Due to a paucity of robust evidence, it is difficult to make any firm conclusion about the effects of pedagogical translanguaging on English language proficiency in EFL contexts.

Conclusions, Confidence in Cumulative Evidence, and Further Research
This systematic review examined the effects of pedagogical translanguaging in EFL classrooms.The included studies demonstrated a variety of approaches, activities, tasks, and interactions considered to fit under the banner of pedagogical translanguaging.Most approaches were used to teach and assess reading and writing, while other aspects of EFL learning were less well researched.The results showed that no negative influences of pedagogical translanguaging were detected (that is, translanguaging approaches did not appear to be any less effective than monolingual approaches).However, most of the studies were not sufficiently well-designed and well-conducted comparisons to support claims that pedagogical translanguaging is beneficial for improving English language proficiency.
As with all research, there are limitations to our review.One limitation was that our search was confined to English-language databases.As a result, some relevant studies published in languages other than English may have been missed.We are confident that we conducted an expansive search for English-language publications.Nonetheless, we did not hand-search, nor did we conduct citation chaining.Any future replications or updates to this review should consider expanding the search strategy accordingly.
Another limitation may be our decision to include studies with participants of all ages (from primary to tertiary).There are sound theoretical reasons to believe that the effects of translanguaging pedagogy may vary depending on age, cognitive maturity, and proficiency in L1 and L2.However, given that we expected the eligible literature to be scarce, we took the decision to include all age phases to provide as comprehensive an overview of the topic as allowed by the extant literature.In future reviews, comparing findings within age phases and with reference to L1 and/or L2 proficiency would be instructive, assuming sufficient data from which to do so.
In addition, as our review focused specifically on translanguaging; it included only studies that described the interventions as such.We have already remarked that we take the most vocal proponents of translanguaging at their word when they say that there is something that differentiates translanguaging from other related terms and the way they are operationalised in the classroom.Nonetheless, as we have also observed, there are studies that might contribute to this literature that do not use the term (e.g., Ong and Zhang 2018) but which, pedagogically speaking, look very similar.
We urge the field to work on agreeing on a definition of pedagogical translanguaging that is clear and consistent and, importantly, falsifiable; and that the research community works to supplement the plentiful qualitative and observational evidence (Prilutskaya 2021) on this matter with research designed such that it is capable of producing reliable

Table 1 .
Summary of the characteristics of included studies.