buenoA PhD thesis defended in late 2008 came up with unprecedented and consistent data regarding a recurrent debate in the scientific community: the disadvantage of Brazilian researchers in the area of academic publication because English, acknowledged as the language of science, is neither their first language nor even the country’s second language. The study, by Sonia Maria Ramos de Vasconcelos, from the Biosciences Education, Management and Dissemination Program of the Medical Biochemistry Institute of the Federal University of Rio de Janeiro (UFRJ), analyzed several aspects of the linguistic disadvantage, but it drew particular attention because it identified a statistical correlation between researchers’ productivity and their proficiency in writing English.
In 2005, Sonia Vasconcelos gained access to a database, made available by CNPq and derived from the Lattes Platform [database of Brazilian researchers CVs], with information about 51,223 Brazilian researchers, including their scientific production in domestic and international publications and their proficiency in foreign languages. The information on linguistic competence is based on a self-evaluation scheme that takes into account four skills (reading, speaking, understanding and writing), each of them classified, on the Lattes Platform, as “well”, “reasonable” or “poor”. The next step consisted of analyzing the relation between researchers’ competence in writing in English with indicators of scientific production by the researchers on the CNPq and BSI – Brazilian Science Indicators registries. These contain information about Brazilian authors in the ISI Web of Knowledge database, covering the 1945 to 2004 time span. It was found that authors with poor or only reasonable English writing skills were those who published less, whereas the skilled English writers were among the most productive.
Of those who had published one or two articles in English in international periodicals from 2001 to 2004, 53% stated they could write the language well, while only 7.8% said they had little proficiency in this area. Among those who published more than 50 papers, 91.8% said that they were totally proficient and not a single one indicated that he or she had limited writing skills in English. “Researchers who can write English well are considerably more productive in terms of publishing articles relative to those whose competence in this field is only reasonable or poor,” states Sonia, who did her master’s degree thesis in English language literature and who has been teaching courses on scientific communication in English to postgraduate students and researchers for seven years at least. “The data suggest that scientists’ written communication skills have an impact on the visibility of Brazilian science in international periodicals in English.”
Sonia cross referenced the same data taking into account, in this case, article citations. The correlation with linguistic competence repeated itself. Finally, she analyzed the so-called h index of the authors. This index combines productivity and impact and is defined as the number “h” of works that have at least the number “h” of citations each. A researcher with a 30 h index is one who has published 30 scientific articles, each of which has been cited at least 30 times in other works. Once again, high h indices were more common among those researchers who could write English well.
In the sample of individuals that was studied, only 33% of the Brazilian researchers declared they were totally proficient in English across the four skills. As the element of interest was their writing skills, Sonia focused on this competence, which was distributed as follows in the sample: 44.4% declared they could write English well, 35.2% said reasonably and 13% said poorly. The researcher warns us, however, about the data’s subjective bias, since the information was based on self-evaluation by the researchers. Nevertheless, it is valuable evidence, given the scarcity of studies about this theme. “This was the first study to measure the impact of proficiency on the lingua franca of science about productivity and scientific visibility in a Latin American country,” says Jacqueline Leta, a UFRJ professor and thesis advisor, along with Professor Martha Sorenson, of Sonia’s thesis. This part of the study was published last year in the journal Embo Reports, of the Nature group.
A study with a more generic scope, published in 2004, reached analogous results. Jonathan Man, a professor at the University of British Columbia, Canada, compared research financing data from several countries, the scores achieved on the Test of English as a Foreign Language (Toefl) and productivity in major periodicals in the medical field. Countries with high Toefl scores, such as Holland (616 points) and Denmark (606 points) published proportionately more than others, such as Sweden (589 points) and Japan (496 points). The latter were given more financing for R&D, but they posted weaker Toefl results. The striking influence of the language variable in this study was also observed in connection with the scientific production in other countries, such as Korea. Other than for Mexico, Latin American countries were not assessed in this study.
In 2004, while still in the pilot stage of her research, Sonia Vasconcelos tried to identify the specific problems involved in the evaluation of a scientific article written in poor English. Her intention was to find out to what extent the language barrier delayed the publication of articles, as a result of a protracted review process or of returning the manuscript to the author with queries about the text – this being the best known disadvantage of Brazilian researchers compared to those whose native language is English. She consulted by e-mail 40 editors of international publications in a range of fields of knowledge, asking them about the linguistic problems in texts submitted by researchers who were not native English speakers. The comments indicated that the losses were more complex than mere publication delays. “I spend a substantial part of my time in the editorial office correcting the grammar and improving the work of authors who don’t speak English. It is difficult to find good science in poorly written articles,” said Joan W. Bennett, a professor at Rutgers University, who was, at the time, the editor of Mycology and coeditor of Advances in Microbiology. The most striking response was that of Robert McMeeing, editor of the Journal of Applied Mechanics. “All articles written in bad English are turned down in the peer review process because, if the English is bad, knowledge cannot be understood at the level required for publication.” Harold H. Kung, editor of Applied Catalysis A, said that 90% of the articles written by researchers whose native language is not English are returned for reviewing. “And at least 50% require substantial revising,” he stated. Graeme Bonham-Carter, editor of Computers & Geosciences, also confirmed the impact of the language barrier in the process of evaluating scientific articles. “Undoubtedly, language is a substantial barrier to publication and I often feel bad about this. The scientific content is sometimes strong, but the language is so poor that it stands in the way of comprehension,” he said.
buenoSonia Vasconcelos’s study received some repercussion in academia. Her thesis project, won an award, in 2007, from the Eugene Garfield Foundation in conjunction with the Chemical Heritage Foundation for the proposal’s originality and potential contribution to the scientometry field. The results were discussed at a recent workshop of Abec, the Brazilian Association of Scientific Editors, held in Gramado (state of Rio Grande do Sul). “Everyone thought that this correlation between proficiency and productivity existed, but the study helped to put the issue on a solid basis. We’re outside the axis of English, but the scientific and academic world speak in English,” says Benedito Barraviera, A professor of infectology at the Botucatu Medical School of Paulista State University (Unesp) and president of Abec. “Capes should demand that postgraduate programs impose stronger requirements regarding their students, proficiency in this language. A while back, it used to be said that the science paradigm was “publish or perish.” Now it’s changed. Publishing is not enough; one must publish and be cited in journals that have impact and those who only publish in Portuguese don’t get read, let alone cited,” states Barraviera.
Biologist Márcia Triunfol has restrictions regarding the link between proficiency and productivity that this study has established. “If the problem were merely this, it could be solved through a good translation service,” states Marcia, who has worked as an assistant editor of the journal Science and who currently runs a scientific communication firm and travels throughout the country holding workshops that provide guidance for researchers who are writing scientific papers – in English. Her experience suggests that the problem is broader. “What I see in the workshops is that Brazilian researchers lack the knowledge and training to understand what a scientific article is and the type of approach that is appropriate for publication in international journals,” she states. “An exercise that I always propose to my students is to identify the question in the article. Many are unable to do this. Often, the problem is that the scientific paper is merely repeating something that has already been done. I observe that there is a lack of understanding of scientific language, besides creativity and the courage to produce original contributions. When this objective is attained, finding a service to review the English properly is a minor problem.”
Barraviera, from Abec, agrees with this diagnosis. “I’m an editor of a scientific journal and it is common to receive articles by researchers who did not know how to plan their research nor how to conduct their experiments, with problems in the methodology adopted. The fact that they are also unable to write in English is just one detail in a chain of problems,” states Barraviera, whose association organized a long distance course on methodology for students last year. “More than one thousand people registered for the 350 places. Unfortunately, we lacked the resources to repeat the experience this year,” he states.
Sonia Vasconcelos agrees that the role of proficiency in English regarding the country’s academic production may be secondary. “The data presented are just one of the relevant factors that concern researchers’ academic productivity. It doesn’t mean that it is a determinant, as there are several other elements that are known, such as the percentage of GDP invested in research, the number of PhDs involved in research and development, and the researchers’ international collaboration networks. However, what is essential is not the relative size of the problem, but the fact that it exists and that, at present, Brazil lacks strategies to deal with it, contrary to other countries,” she states. As examples, she mentions China and South Korea, which have aggressive incentive policies when it comes to the teaching of English; this is also true even of the United States. “Americans, the leaders of world scientific production, have a tradition of investing in writing centers in a large number of US universities; they have assistance specifically for language editing services and they encourage people to use the editorial offices in several research centers,” she states.
The linguistic environment in Brazil does not even begin to face these challenges. The teaching of English in Brazil, comments Sonia, basically centers on students acquiring colloquial language. Though there are several projects geared toward teaching English for specific purposes, there is no broad and strategic approach capable of meeting the demand for scientific communication in English. The role of linguistic competence to write science in the training of Brazilian researchers is peripheral, she says. The country’s public universities curriculums generally focus on developing reading rather than writing skills. “We all know that being able to read English is extremely important in academia, but what about scientific writing? How can one train young researchers capable of developing their own voice in the lingua franca of science?” In Brazil and in most of Latin America, English scientific communication disciplines are not part of the tradition of scientific postgraduate programs. Moreover, there are no language editing offices established in Brazilian research institutions to help researchers with their writing.
In an article published in 2007 in the journal Embo Reports, Rogério Meneghini and Abel Packer, from Bireme, the Latin American and Caribbean Health Science Information Center, linked the issue of proficiency in written English with the concept of the “lost science of the Third World,” established by W. Wayt Gibbs in 1995 to describe the research of regional interest conducted by researchers from peripheral countries, but which remains unknown in the central countries. For these two researchers, it seems reasonable to assume that part of this “lost science” is produced by researchers that prefer to publish in their native language because of their difficulty writing the lingua franca of science. “It has become essential for researchers to understand and write English. Without it, they will be limited in their capacity to work, because they will be unable, for instance, to contribute to international networks,” says Packer, who is Bireme’s director.
Packer, however, objects to the notion that there is to be no scientific communication other than in English. “There are fields of knowledge whose traditions or characteristics demand that they be divulged in other languages and there’s nothing wrong with this,” he states, referring to the humanities and to the agricultural, social and health sciences, whose results may have an impact on society, yet do not reach their audience, which has little familiarity with English. “Multilingualism is one of the complex phenomena of globalization and it must be regarded as just one dimension of scientific communication,” states Packer, who criticizes the “autocratic” view of the hard areas of knowledge and of agencies, which find international visibility, as measured by impact factors, more important than the possible regional impact that an advance in knowledge may produce. He mentions the electronic library SciELO, managed by Bireme and financed by FAPESP, which has been encouraging scientific publications in Portuguese to produce bilingual editions – or at least collections of the best articles in the two languages.Republish