UNITED KINGDOM / AGILITYPR.NEWS / November 08, 2021 / It is not uncommon for non-native English-speaking authors submitting their research papers to journals to be rejected because of language, rather than the quality of their scientific content. Understandably, these authors are often left confused or angry, not knowing how they can overcome this hurdle. Here Abhishek Goel, Co-Founder and CEO at Cactus Communications, a technology company accelerating scientific advancement, discusses ways that we can overcome bias against non-native English speakers in science publishing, from adopting new peer review models to introducing smart technology into the review process.

CACTUS was built on the premise that it would have experts improve the language used in manuscripts written by non-native English-speaking researchers to improve their chances of being successfully published in English journals. We have identified that authors of papers rejected because of language and technical errors often have distinctly Asian names. We would regularly see papers authored by researchers from Japan, South Korea and China being desk rejected by the journal editor with the suggestion to have them edited for language.

Even when papers made it to peer review, oddly enough, there were reviewers’ comments that often focused on the language and not focusing upon the science presented in the paper. Typically, protocols to eliminate bias in the peer-review process are one-sided. Most journals adopt the single-blind model of peer-review, where the reviewers’ names are not shared with the authors to avoid any attempts to influence their verdicts. However, the authors’ names are visible to the reviewers meaning there is a risk of bias against them. This problem is lessened in the double-blind model, where neither party is privy to the other’s name. However, as outlined in Nature, even using the gold standard of double-blind peer review model cannot eliminate the chance of a paper being desk-rejected by the journal editor because of bias against the authors.

The increased adoption of the double-blind model should be commended as a step in the right direction. However, as long as humans are involved in deciding which research papers are published, unconscious bias will continue to be a hurdle for non-native English-speaking researchers to overcome.

Smart technology will be vital in eliminating this problem. Processes and platforms powered by artificial intelligence (AI) and machine learning will eliminate reliance on human decision making, selecting papers based purely on submission readiness rather than irrelevant factors like the authors’ institution, location or ethnicity. Tools like R Pubsure, UNSILO Evaluate, scite, AIRA, and Ripeta all leverage AI to facilitate the objective assessment of research manuscripts.

For example, researcher.life’s R Pubsure tool allows users to upload a file and download a report covering language, structure, references, and other metrics that help identify how publication-ready their manuscript is. It offers specific advice for users, highlighting specific errors within the text, and provides an overall R Pubsure score that can be used as a measure of overall quality. The tool can also be used to recommend suitable journals based on the content in the manuscript. The challenge that AI faces now is making sure that human bias, whether it be conscious or unconscious, does not get incorporated into machine models and taint future output.

To find out more on how AI can help overcome bias in science publishing, visit www.cactusglobal.com.