Qualitative research in the Arabic language. When should translations to English occur? A literature review

Qualitative studies are a valuable approach to exploratory research. Frequently, researchers are required to collect data in languages other than English, which requires a translation process for the results to be communicated to a wider audience. However, language-embedded meaning can be lost in the translation process, and there is no consensus on the optimum timing of translation during the analysis process. Thus, the aim of this paper was to review how researchers conduct qualitative research with Arabic-speaking participants and the timing of data translation. Three databases were searched (PubMed, Scopus, and Web of Science) for the period January 2010 to January 2020. Studies were excluded if the data collection was not in Arabic or the study was not qualitative or healthcare related. Thirty-one studies were included, 26 of which translated all transcripts into English and then analyzed the data in English. Five studies transcribed the data in Arabic, analyzed it in Arabic, and then translated the results to English or conducted a parallel analysis. The reason provided for translating the data into English before the analysis was to enable non-Arabic authors to access the data and assist with the analysis. The search results suggest that researchers prefer translating data before analyzing it and are aware of the possibility of losing meaning during the translation process, which might affect the results. A more thoughtful approach to the timing of translation should be undertaken to ensure the subtleties of language are not lost during the analysis of qualitative data.


Introduction
Qualitative studies are an effective and valuable method for exploratory research as they can identify and provide an in-depth understanding of the issues under study and lead to better intervention designs that may improve healthcare delivery and outcomes. 1 Qualitative studies attempt to answer many of the 'why' queries concerning patients, healthcare providers, and systems that cannot be answered by other research methods. 1 Qualitative research is meant to gather and analyze non-measurable data about meaning and words. 1 Thus, it is able to provide information about the ways participants experience the world. 1 In many instances, researchers must collect data in languages other than English, and using the native language of study participants is preferred. 2 Collecting data in languages other than English occurs either by conducting research with participants in non-English speaking countries or with foreigners in English-speaking countries. 2 The Arabic language is widely used when conducting qualitative research in the healthcare field. 3 Not only is it spoken in 25 countries, but 30% of foreigners in western countries are Arabic-speaking migrants. 4 However, conducting qualitative research with Arabic participants requires translation for the results to be shared with a wider audience. 3 Translation from Arabic to English can be conducted either before initiating the analysis (i.e., collecting the data in Arabic and then translating and analyzing it) or after the analysis (i.e., collecting the data in Arabic, analyzing it in Arabic, and then translating the results to English).
Translation involves interpretation of texts that convey the meaning and not solely a word-to-word translation, which overlaps with what the analysis of the qualitative research is about. 5 Conducting the analysis of the translated data rather than the original data might affect the accuracy of the analysis, as different languages involve different epistemological assumptions and positions. 6 It is surprising that there is no consensus regarding the optimum timing of translation during qualitative data analysis. 3,[5][6][7][8] Few attempts have been made to discuss the impact of translation in the context of qualitative research. [5][6][7]9 An approach to translate the results rather than the original data was recommended by some researchers. 5,6,9 As qualitative research aims to study meaning, this approach of translating the results and not the original data will assist in being closer to the meaning which participants experienced and shared. 3,5,10 Since the timing and the challenges of translation of qualitative research did not receive the required attention globally and particularly to the Arabic language, the aim of this study was to review how researchers conduct qualitative research with Arabic-speaking participants and the timing of the translations they make.

Objective
The objective of my study was 1. To map the existing literature of the Arabic-to-English translation for the purpose of sharing the results of a qualitative study with an Englishspeaking audience 2. To discuss researchers' provided justifications for the to inform practice, research and policymakers.
The research questions were: when does a researcher conduct an Arabic-to-English translation for the purpose of sharing the results of a qualitative study with an English-speaking audience? Do researchers provide a justification for the approach they used?

Methods
Studies eligible to be included in the review had to use a qualitative study design, be conducted in Arabic, and must have reported the results in English.

Electronic searches
Three databases were searched (PubMed, Scopus, and Web of science) for the period January 2010 to January 2020 using the Boolean operators of the following search terms: 'Arabic or non-English' and 'translation' and 'qualitative research or interviews or focus groups'. Subject headings for terms describing qualitative research were used. The full search strategy for one of the electronic databases is presented in Table 1 and Fig. 1.

Study selection
The search and screening were conducted by the author. Abstracts were classified as 'included' 'excluded' or 'possible'. Publications that matched the inclusion criteria were retrieved. Full text versions of all relevant and possible versions of studies were examined and decisions were then made for final inclusion.

Data extraction
Data were extracted from the full-text studies by the author. Data extraction was conducted using a predesigned Microsoft excel sheet that included the following items: author, year, country, setting, aim, use of theory, data collection method, topic guide development, time of translation, method of analysis, and software used. Further collection of data validity criteria was also applied. The concept of rigor in qualitative research was applied using different criteria 8,11 . Most criteria are complex and unclear. However, while Lincoln and Guba's (1985) trustworthiness model may not be a gold standard, it is the most explicit and agreed upon in the literature. 11 Therefore, terms were expanded to include studies if they had mentioned any of the following terms: trustworthiness, reliability, validity, or rigor.
The search terms were further expanded to include credibility, transferability, dependability, or confirmability. 8,11 Studies were excluded if the data was not in Arabic, not qualitative, or if it was not healthcare-related research. The PRISMA flow chart (Fig. 2) and the PRISMA checklist are presented.

Study selection
There were 168 search results. Of those, PubMed contributed 44 results, 59 were from Scopus, and 65 results were from the Web of Science. After removing duplicates, the number of unique articles was reduced to 103. In total, 31 studies were eligible for inclusion. Studies were excluded if the study was not qualitative, it was not health related or the collected Fig. 2. PRISMA 2020 flow diagram. *Consider, if feasible to do so, reporting the number of records identified from each database or register searched (rather than the total number across all databases/registers). **If automation tools were used, indicate how many records were excluded by a human and how many were excluded by automation tools.    Credibility (a sub-set were checked by bi-cultural co-authors)   Table 3.

Translation process
Twenty-six studies translated all of their transcripts to English and then analyzed them in the English language. The authors were aware of the limitations of this approach as the field work was conducted in Arabic while the analysis used the translated reports. [13][14][15][16]27,28,[30][31][32][33][34] The explanation given for using this method was that some of the authors were Englishspeaking and the translation enabled them to access the data. 13,14,[30][31][32][33]39 Authors mitigated any potential effects of translation on the quality of the analysis by involving bilingual researchers in the coding verification process, 14 retaining key words in Arabic, 36,37 or by having translators add notations to help ensure that the meaning of cultural idioms was not lost. 16 Authors also clearly mentioned the need to involve professionals when conducting analyses in the language that was spoken and not the translated reports. 13 Two studies transcribed their data in Arabic, analyzed it in Arabic, and then translated the results to English. 14,19,29 Three studies did parallel analyses 20,26,30 ; one in Arabic and one in English, 30 while the other 2 studies conducted some of the analysis in English and the rest in Arabic. 20,26 It should be mentioned that these 5 studies involved experts in qualitative analysis who were also Arabic native speakers. Characteristics of the included studies are presented in Table 2.
Topic guides were described in all of the included studies except for 2. 23,31 Topic guides were created by the authors either from the literature or from previous research. The language of the topic guide was described in only one study as having been developed in English and then verbally translated during the interview. 22 All of the studies used thematic content analysis. An evaluation of trustworthiness and rigor was undertaken in 12 studies, with some authors explicitly describing each term and how they used them. 15 Guba's criteria without describing how. Others described some aspects of data validation. 23,35,39 Thirteen studies used qualitative analysis computer packages for their data analysis, while 2 studies used pen and paper, Microsoft word, and excel.

Discussion
This is a scoping review that sought to better understand qualitative research conducted in the Arabic language and the timing of data translation. Research has shown that meanings are embedded in spoken languages and discrepancies may occur during the translation process. 2,3,43 Translation is a vital step for communicating ideas and the results of investigations with the wider, English-speaking community. There is an overlap between translation and analyzing qualitative data as both involve interpretation of a meaning. [5][6][7]9 By understanding that the timing of a translation might affect the interpretation of the results, thoughtful consideration of when data should be translated is needed. 2,3,43 The data from 31 studies were reviewed with the aim of identifying when the translation of qualitative data for Arabic language participants occurred. The results indicated that most of the translation process was conducted at an early stage. Typically, researchers translated row data to English, analyzed it in English, and then shared the results. Currently (and unlike quantitative research), there are no guidelines for translating qualitative research. 3 Thus, it is likely that ambiguous translations would affect the data integrity and research results. 3,43,44 Studies have demonstrated that interpreters are inconsistent when interpreting the same interview at different times, 43,44 with differences such as augmentation, summarizing, and omitting information being identified. Hence, it is unclear why, given the added cost and effort, an analysis would be conducted on a translated report as opposed to the original if the primary researchers speak the language of the interviewees. 43 This review further found that researchers need to verify information in their topic guides. The language used to develop a topic guide, and its translation process, should be described in a study; however, only one study was identified that described the process. Most of the topic guides were developed from the English-language literature and then used with Arabic speakers. This is a step that requires both translation and the cultural adaption of questions to ensure they are appropriate for the participants and do not have explicit details that could weaken a guide's quality 44 .
Only one of the reviewed studies used an 18-month ethnographic approach to collect data (in addition to interviews and focus groups). All of the others used either interviews and focus groups or both. Furthermore, all of the included studies used thematic content analysis. The flexibility of this type of analysis, along with in-depth interviews, can provide rich and detailed results that justifies their use. 38 However, there are other methods of analysis that can be applied such as observations, grounded theory, and document study 45,46 , which could enrich qualitative data gathered from Arabic language speakers. 45,46 The overwhelming use of interviews, focus groups, and content thematic analyses may indicate that researchers-with the lack guidelines and/or best practices-model

2008
The study was not qualitative Jamal Ahmad (2017)   their work from the literature and not thoroughly consider the implications. 47 Evaluating the rigor and trustworthiness of qualitative research is crucial. While it is understood that results might not be generalizable to other settings and there are no agreed upon criteria, there remains the need to assess reliability and validity and to be explicit about how they are assessed. [47][48][49] In this review, the Guba and Lincoln criteria (with expanded terms) was used as it is currently the most accepted approach. [47][48][49] Yet, more than half of the included studies did not mention any reliability or validity terms in their publications. While the authors might have used these concepts, this was not explained in the publications, which could limit a reader's ability to trust the study. [47][48][49] Despite the limitations of not having an agreed upon criteria, it is essential to highlight the importance of documenting the rigorous steps involved in assessing reliability and validity and communicating these steps to improve the strength of the studies. [47][48][49] As my results have shown that most of the researcher translated the original data and conducted the analysis on the translated version instead of the original data. This approach might be justified if the authors were not an Arabic language speaker but if both, the participants and the authors are Arabic speakers, it deserves more attention 5-7,9 . Additionally, beside impacting the validity of the results translation is costly. 49,50 Conducting the analysis of the original data will help to narrow the gap between the meaning as experienced by the participants and the meaning interpreted in the finding. Furthermore, this approach will help reducing research cost.
The act of translating row data before an analysis could have been considered acceptable in the past given the lack of computer-aided, qualitative data analysis software (CAQDAS) compatible with Arabic languages. Currently, many of the languages supported by qualitative data software programs, including Atlas.ti 51 and MAXQDA 52 , are written from right to left (as Arabic is).

Strengths and limitations
Conducting qualitative research with non-English speakers is challenging. The challenges associated with translating qualitative data have also been discussed in relation to other languages such as Chinese, 5,9 German 7 and Spanish. 49 One limitation of this study is that it was not a systematic review and, therefore, might have missed other relevant publications. However, while the included studies provided sufficient data, further research that includes systematic reviews should be considered.
Encouraging researchers to analyze data in the language spoken by participants of a qualitative study is infrequent despite it being the original practice. Furthermore, conducting qualitative studies with foreigners speaking Arabic might require a different approach from those conducted in an Arabic country, which could help to validate translations occurring before the analysis. However, researchers should explain the effect of other researchers (authors who can understand Arabic) and build their decision regarding the timing of a translation on that.
Health-related qualitative studies are increasing in Arabic-speaking countries, and a discussion concerning the timing of data translation needs to occur. In addition, increasing the education and training of researchers in analyzing qualitative data is a fundamental step that could result in more qualified qualitative researchers who are able to analyze data in its original form.

Conclusion
This review explored the process of translation of qualitative healthrelated data conducted with Arabic-speaking participants. The results suggest that less than 20% of researchers perform an analysis on the original Arabic data. There is a need to thoughtfully decide when the best time to translate data to preserve its meaning occurs.

Impact of findings on practice statements
• Researchers prefer translating data before analyzing it and are aware of the possibility of losing meaning during the translation process • A more thoughtful approach to the timing of translation should be undertaken

Declaration of interests
The author declare that she has no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.