(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



The manifold costs of being a non-native English speaker in science [1]

['Tatsuya Amano', 'School Of The Environment', 'The University Of Queensland', 'Brisbane', 'Queensland', 'Centre For Biodiversity', 'Conservation Science', 'Valeria Ramírez-Castañeda', 'Museum Of Vertebrate Zoology', 'University Of California']

Date: 2023-07

The use of English as the common language of science represents a major impediment to maximising the contribution of non-native English speakers to science. Yet few studies have quantified the consequences of language barriers on the career development of researchers who are non-native English speakers. By surveying 908 researchers in environmental sciences, this study estimates and compares the amount of effort required to conduct scientific activities in English between researchers from different countries and, thus, different linguistic and economic backgrounds. Our survey demonstrates that non-native English speakers, especially early in their careers, spend more effort than native English speakers in conducting scientific activities, from reading and writing papers and preparing presentations in English, to disseminating research in multiple languages. Language barriers can also cause them not to attend, or give oral presentations at, international conferences conducted in English. We urge scientific communities to recognise and tackle these disadvantages to release the untapped potential of non-native English speakers in science. This study also proposes potential solutions that can be implemented today by individuals, institutions, journals, funders, and conferences.

Data Availability: The data underlying Figs 2B, 3A, 3B, 4C, S1, S2, S4, S5, S6, S7, S8, and S9 can be found in S1 Data . The data underlying Figs 1A, 1B, 1C, 1D, 1E, 1F, 2A, 2C, 2D, 4A, 4B, and S3 are raw data directly from the survey questions, which our ethics approval prevents us from sharing to secure confidentiality of the respondents. Access to these raw data should be requested from the University of Queensland Ethics office, which can be contacted at [email protected] . All codes used in the analysis are available at: http://doi.org/10.17605/OSF.IO/Y94ZT .

Introduction

Unlocking the potential of disadvantaged communities is one of the urgent challenges in science. Collaboration involving a diverse group of people can better solve problems [1] and deliver higher levels of, and more relevant, scientific innovation [2] and impacts [3]. Today, the need to tap into a diversity of people, views, knowledge systems, and solutions in order to successfully address global challenges, such as the biodiversity and climate crises [4–6], is being increasingly recognised, and there is a critical need to do so across multiple disciplines [7–9].

Increasing the diversity within scientific communities requires breaking down the barriers that impede the career development of disadvantaged groups of researchers, and one such barrier is rooted in language. Although the use of English as the common language of science has no doubt contributed to the advance of science [10], this benefit comes with considerable costs for those whose first language is not English (hereafter, non-native English speakers). Non-native English speakers, who constitute the majority of the world’s population, face a number of challenges in conducting and communicating science in English, which inevitably impose an excessive burden on their career development in science. This issue is widely recognised [11,12], as English now plays a dominant role in the execution and communication of science, as well as the evaluation of scientists, in almost any scientific discipline [13]. For example, the United Nations Educational, Scientific, and Cultural Organization (UNESCO)’s recommendation on open science, adopted by 193 member states in 2021, highlights the need to overcome language barriers in order to achieve 4 of the open science core values and guiding principles (Equity and fairness, Diversity and inclusiveness, Equality of opportunities, and Collaboration, participation and inclusion) [14]. Yet scientific communities still desperately lack the concerted effort needed to reduce language barriers faced by non-native English speakers and promote equity in science.

The difficulties faced by non-native English speakers in conducting science, and how they translate to numerous disadvantages for career development, are still poorly understood. Earlier studies have reported the experience and perception of language barriers in speakers of a single non-English language [15] or to certain types of scientific activities, such as paper writing [16], paper publication [17], and research dissemination [18]. Attempts to assess the disadvantages of being non-native English speakers in science are emerging (e.g., [19,20]). Nevertheless, to date, no published study has quantified how multiple aspects of language barriers concurrently affect the career development of speakers of different non-English languages, compared to native English speakers.

This study addresses this knowledge gap by first estimating the amount of effort (e.g., time and financial cost) required by individual researchers in conducting a variety of scientific activities in English. We compare the estimated amount of effort between researchers from countries with different linguistic and economic backgrounds, with the aim to quantify the multiple disadvantages faced by non-native English speakers practising science.

We conducted an online survey of a total of 908 researchers in environmental sciences (particularly ecology, evolutionary biology, conservation biology, and related disciplines) who have published at least one first-authored peer-reviewed paper in English, with one of the following 8 nationalities: Bangladeshi (n = 106), Bolivian (100), British (112), Japanese (294), Nepali (82), Nigerian (40), Spanish (108), and Ukrainian (66) (see more details including their demographic information in S1 Table). These nationalities are stratified by the level of each country’s English proficiency (based on the English Proficiency Index [21]) and income (based on the World Bank list of economies [22]): Bangladeshi, Nepali (low English proficiency and lower-middle income), Japanese (low English proficiency and high income), Bolivian, Ukrainian (moderate English proficiency and lower-middle income), Spanish (moderate English proficiency and high income), Nigerian (English as an official language and lower-middle income), and British (English as an official language and high income). This is to distinguish the effect of language barriers from the effect of other types of barriers in science that are often confounded with language barriers, notably economic barriers to conference participations [23,24]. The survey asks participants about the amount of effort needed to conduct 5 categories of scientific activities: paper reading, writing, publication, and dissemination, and participation in conferences (see Materials and methods for more details and S1 Text for the survey itself).

The results unveiled profound disadvantages for non-native English speakers in conducting all scientific activities surveyed. First, non-native English speakers require more time to read an English-language paper⁠—a requisite for obtaining necessary, especially cutting-edge, knowledge in research (Fig 1A and S2 Table). In a comparison among researchers who have published only one English-language paper, non-native English speakers of moderate English proficiency nationalities spend a median of 46.6% (2.5 to 97.5 percentiles: 19.0% to 78.1%) more time, and those of low English proficiency nationalities spend a median of 90.8% (60.6% to 125.4%) more time reading an English-language paper than native English speakers do (Figs 1A and S1). This disadvantage is found even in mid- and late-career researchers, especially those of low English proficiency nationalities (Figs 1A and S1). Importantly, in a comparison of the estimated time needed to read a paper written in their first language, non-native English speakers were shown to need less time than native English speakers (Fig 1B and S3 Table), showing that the above disadvantage arises from the need to read in English, not in their first languages.

Similarly, non-native English speakers need more time to write a paper in English, than their native English speaker peers, at an early career stage (Fig 1C and S4 Table). In a comparison of researchers who have published only one English-language paper, non-native English speakers of moderate English proficiency nationalities spend a median 50.6% (2.5 to 97.5 percentiles: 31.1% to 52.6%) more time, and those of low English proficiency nationalities spend 29.8% (6.6% to 59.3%) more time writing a paper in English than native English speakers do (Figs 1C and S2). This disadvantage is not found in those at a later career stage (S2 Fig). Again, non-native English speakers need less time to write a paper in their first languages than native English speakers do (Fig 1D and S5 Table). This signifies that the need to write in English, not in their first languages, poses a disadvantage to non-native English speakers.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Language barriers in paper reading and writing. (A) Minutes taken to read and understand the content of the most recent English-language research article each participant read in their field. (B) Minutes it would take to fully read and understand the same paper in one’s first language. (C) Number of days (assuming 7 hours being spent per day) taken to write the first draft of each participant’s latest first-authored paper in English. (D) Number of days that would be taken to write the first draft of the same paper in their first language. (E) Percentage of papers where English writing was checked by someone as a favour. (F) Percentage of papers where English writing was checked by a professional service. The regression lines (with 95% confidence intervals as shaded areas) represent the estimated relationship with the number of English-language papers published (shown on the log 10 -transformed axis), based on the results shown in S2–S5 and S7–S8 Tables (income level was not significant in (C)). The data underlying this figure are raw data directly from the survey questions, which our ethics approval prevents us from sharing to secure confidentiality of the respondents. https://doi.org/10.1371/journal.pbio.3002184.g001

Non-native English speakers also require more effort than native English speakers for the English proofreading of their papers. Apart from late-career researchers of moderate English proficiency nationalities, non-native English speakers ask someone to proofread their English for, on average, 75% or more of their papers, while most native English speakers do this in less than half of their papers (S3 Fig and S6 Table). Non-native English speakers of moderate English proficiency nationalities tend to ask someone to proofread their English as a favour (Fig 1E and S7 Table), while those of a low English proficiency nationality and high income level (i.e., Japanese in our study sample) tend to use a professional English editing service (Fig 1F and S8 Table). Non-native English speakers of low English proficiency nationalities and lower-middle income level neither ask someone to proofread their English as a favour nor use a paid service for most of their papers (Fig 1E and 1F).

Non-native English speakers, especially those of low English proficiency nationalities, are more likely to have their papers rejected by journals due to English writing, compared to native English speakers (Fig 2A and S9 Table). For example, in a comparison of those who have published one English-language paper, 38.1% (31.6% to 44.5%) and 35.9% (30.5% to 41.3%) of the non-native English speakers of moderate and low English proficiency nationalities, respectively, have experienced paper rejection due to English writing, while only 14.4% of the native English speakers have, meaning that the frequency of language-related paper rejection is 2.5 to 2.6 times higher for non-native speakers. This result also supports the findings of recent papers that journals are less likely to accept papers by researchers in countries where English is not a primary language [25–27]. Similarly, non-native English speakers are more likely to be requested to improve their English writing during paper revision (Fig 2B and S10 Table). For example, 42.5% and 42.6% of the non-native English speakers of moderate and low English proficiency nationalities, respectively, compared to only 3.4% of the native English speaker population, report that they are often/most of the time/always requested to improve their English writing during paper revision. This equates to a 12.5 times higher frequency of language-related revisions for non-native English speakers.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Language barriers to paper publication and dissemination. (A) Proportion of researchers who have experienced rejection of a first-authored English-language paper due to English writing. (B) Frequency of being requested to improve English writing during the revision of first-authored English-language papers. (C) Proportion of researchers who have provided non-English-language abstracts of English-language papers. (D) Proportion of researchers who have disseminated English-language papers in other languages as well as English. The regression lines (with 95% confidence intervals as shaded areas) in (A), (C), and (D) represent the estimated relationship with the number of English-language papers published (shown on the log 10 -transformed axis), based on the results shown in S9, S11 and S12 Tables. Income level (solid line: high; dotted line: lower-middle) was only significant and thus shown in (C). The data underlying (A), (C), and (D) are raw data directly from the survey questions, which our ethics approval prevents us from sharing to secure confidentiality of the respondents. The data underlying (B) can be found in S1 Data. https://doi.org/10.1371/journal.pbio.3002184.g002

Non-native English speakers spend more effort disseminating their research in multiple languages than native English speakers do, may it be through the publication of their work in non-English-language journals (S4 Fig), preparation of non-English-language abstracts of English-language papers (Fig 2C and S11 Table), or outreach activities in 2 or more languages (Fig 2D and S12 Table).

Language can also be a major barrier to non-native English speakers attending conferences. Approximately 30% of the early-career (defined as those who have published 5 or fewer English-language papers) non-native English speakers of high income nationalities (i.e., Japanese and Spanish combined) report that they often or always decide not to attend an English-language conference due to language barriers (Fig 3A and S13 Table). Similarly, about half of the early-career non-native English speakers of high income nationalities (Japanese and Spanish combined) often or always avoid oral presentations due to language barriers (Fig 3B and S14 Table).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Language barriers to participation in conferences. The frequency of (A) not attending an English-language conference and (B) avoiding oral presentations at an English-language conference due to the lack of confidence in English-language communication. An ECR (early-career researcher) was defined as someone with 5 or fewer English-language papers. The numbers on the right of each bar represent the sample size. The data underlying this figure can be found in S1 Data. https://doi.org/10.1371/journal.pbio.3002184.g003

Even if they decide to give an oral presentation in English, non-native English speakers need much more time to prepare the presentation than native English speakers do; those of moderate and low English proficiency nationalities spend a median 93.7% (2.5 to 97.5 percentiles: 54.7% to 145.2%) and 38.0% (10.8% to 69.6%) more time, respectively, preparing an oral presentation in English than native English speakers do (Fig 4A and S15 Table). This disadvantage does not change with one’s career level (S5 Fig) and, yet again, does not apply when preparing a presentation in one’s first language. For example, non-native English speakers of low English proficiency nationalities even spend less time preparing a presentation in their first language than native English speakers (Fig 4B and S16 Table). At conferences, non-native English speakers often struggle to explain their work in English. This tendency is particularly noticeable in early-career non-native English speakers of low English proficiency nationalities, with over 65% reporting that they often or always find it difficult to explain their work confidently in English (Fig 4C and S17 Table).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. Language barriers to preparing and conducting presentations in English. (A) Number of hours needed to prepare and practice an oral presentation in English. (B) Number of hours that would be needed to prepare and practice the same oral presentation in one’s first language. (C) Frequency of not being able to explain research confidently during a presentation due to English-language barriers. The regression lines (with 95% confidence intervals as shaded areas) in (A) and (B) represent the estimated relationship with the number of English-language papers published (shown on the log 10 -transformed axis), based on the results shown in S15 and S16 Tables. In (C), an ECR (early-career researcher) was defined as someone with 5 or fewer English-language papers published so far. The numbers on the right of each bar represent the sample size. The data underlying (A) and (B) are raw data directly from the survey questions, which our ethics approval prevents us from sharing to secure confidentiality of the respondents. The data underlying (C) can be found in S1 Data. https://doi.org/10.1371/journal.pbio.3002184.g004

This study illustrates how a series of language barriers to conducting different scientific activities multiply to pose a profound disadvantage to non-native English speakers in the development of their scientific careers (Fig 5). Imagine being a PhD student whose first language is not English. Compared to a fellow student who is a native English speaker, you would need considerably more time, or financial cost, to understand every single English-language paper you read (causing you to spend up to 19.1 more working days per year on this activity. See S1 Fig for the calculation), to write your thesis chapters in English, and to polish the English writing before submitting your manuscripts to journals. You would also struggle with paper publication, as your papers will be rejected more often and be subject to revisions based on the written English. Following the publication of your papers, you would need to make an extra effort for dissemination, as you will be doing this in English as well as your own language(s). You will also find yourself hesitating to attend an international conference, or give an oral presentation, ending up losing opportunities to develop an international network. When you do decide to give an oral presentation, you would again need more time than native English speakers for its preparation, after which you would still be frustrated as you are unable to present your work as effectively in English as you would in your first language. What is more, all of these barriers will continue to get in your way as long as you remain in a research career.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. Estimated disadvantages for non-native English speakers when conducting different scientific activities. The height of hurdles indicates the relative length of time taken to read an English-language paper (Reading), to write a paper in English (Writing), and to prepare an oral presentation in English (Presentation), and the relative frequency of an English-language paper being rejected (Paper rejection) or requested to revise (Paper revision) due to English writing, for non-native English speakers (Non-native), compared to native English speakers (Native). The values are for non-native English speakers who have published only one English-language paper (higher value from moderate and low English proficiency nationalities), compared to the values for native English speakers. This figure is not intended to suggest that science is a race. https://doi.org/10.1371/journal.pbio.3002184.g005

Given all of these disadvantages, all else being equal, the apparent scientific productivity of non-native English speakers would undoubtedly be much lower than that of native English speakers. These disadvantages inevitably lead to a tremendous inequality in the development of scientific careers between native and non-native English speakers and the severe underrepresentation of research from countries where English is not a primary language in English-language publications [28]. Furthermore, at a bigger scale, one clear consequence of this inequality is the loss of opportunity for scientific communities to incorporate many researchers and associated knowledge in the early stages of their careers, partly because their first language happens to be one other than English. This may be reflected in our observation that some disadvantages seemed to disappear in late-career researchers (S1 and S2 Figs) We suspect this could be due to survivorship bias; only those non-native English speakers who have managed to conduct science in English as efficiently as native English speakers may have remained in a research career and thus been the dominant group among the experienced researchers who participated in this survey.

The underuse of professional English editing services by those of lower income nationalities, presumably due to the lack of funding, indicates that disadvantages for non-native English speakers could be amplified by a country’s and individual’s low income level. Language barriers to some scientific activities, such as reading papers (Fig 1A), preparing oral presentations (Fig 4A), and attending and presenting at conferences (Figs 3A, 3B and 4C), appear to be less severe for those of lower income nationalities. This might again be explained by survivorship bias. Apart from those languages spoken in high-income countries, such as Spanish and Japanese, few non-English languages have an up-to-date lexicon of scientific terms, creating a much higher need for their speakers to receive scientific education in English [29]. In the low-income countries, only those who can afford to receive such English-language education may have been able to become researchers and participate in our survey.

This study is still likely to have underestimated the severity of the disadvantages faced by non-native English speakers. For example, we did not quantify the immense mental stress associated with all the extra time, cost, effort, and lost opportunities caused by language barriers, which could further exacerbate the already high risk of mental health issues in students and early-career researchers [30]. Non-native English speakers could also face the dilemma of adapting to conducting and communicating science in English or maintaining their skills in conducting and communicating science in their first languages [29]. The survey participants are most likely to be those who are currently active in research, and thus the survey has likely excluded those who have dropped out due to language barriers. Other biases in survey participants may also exist (see Limitations in Materials and methods for discussion). Although the survey was designed to isolate the disadvantages associated solely with language barriers, we cannot dismiss the possibility that the cost we have quantified may incorporate, at least partly, the cost associated with other barriers in science, such as economic, social, identity, and immigration barriers, which many scholars from countries where English is not widely spoken often experience [23,24,31]. While this may be a potential limitation of this study, what this implies in practice is that the disadvantages faced by non-native English speakers could be even bigger and more multifaceted. The level of disadvantages for non-native English speakers could vary among disciplines, presumably depending on, for example, the history of English-based education and the need for international collaboration. Therefore, while we believe that the issue of language barriers to the career development of non-native English speakers is pervasive, the findings of this study may not be quantitatively applicable to all disciplines.

To date, the task of overcoming language barriers has largely been left to non-native English speakers’ efforts and their investment in ways of improving their English skills. However, the magnitude of the disadvantage, quantified in this study, seems far beyond the level that can be overcome with individuals’ efforts. We urgently need a concerted effort, at institutional and societal levels, to minimise the disadvantages for non-native English speakers. We argue that every sector in science, from supervisors and collaborators to universities, institutions, journals, funders, and conferences, should take immediate action to provide language-related support to non-native English speakers and explicitly take into account those disadvantages when evaluating their scientific outcomes (see Fig 6 for proposed solutions). A key aspect of those solutions is to embrace linguistic diversity in science and encourage the multilingualization of science and its communication, as this can help to improve equity, diversity, and inclusiveness in science [14] and maximise the contribution of science to addressing some of the global challenges [32,33]. Our survey showed the relatively low use of machine translation by researchers in all countries (S6 Fig). However, emerging artificial intelligence (AI) tools, such as ChatGPT (https://chat.openai.com/) and DeepL (https://www.deepl.com/), could help non-native English speakers, especially those in low-income countries, reduce the amount of effort and the cost needed to do some of the scientific activities, by providing free, or affordable English proofreading/translation [34,35]. Although discussions are still ongoing about the use of generative AI in science [36,37], we believe that journals and universities should consider and allow the appropriate use of AI tools for English proofreading to reduce language barriers and improve equity in science.

The inequality faced by non-native English speakers due to language barriers can be a major reason for the current underrepresentation of non-native English speakers in global scientific activities [40]. One comment from a survey participant caught our eyes:

If it wasn’t for the language barrier, I could have made a much greater contribution to the advance of ecology and biodiversity conservation. (female participant from Japan in the 40 to 50 age bracket).

Non-native English speakers constitute 95% of the world population [41]. Imagine how many non-native English speakers around the world and over time have been frustrated, just like this participant, because they are unable to contribute to the advance of science to the best of their abilities. Think how many potential contributors scientific communities have failed to bring onboard due to language barriers. Given the multitude of pressing challenges facing humanity and this planet, surely, we cannot afford to miss contributions from such a promising, much needed, yet currently untapped source of researchers.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002184

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.