The form and style used in scientific articles have undergone transformations over the past 70 years, likely in response to the challenge of making academic writing clearer and more capable of captivating readers’ interest—whether they are scientists or laypeople. The conclusion is from a study published in June by Australian researchers in the journal Scientometrics, which analyzed nearly 21 million abstracts of scientific texts published in English between 1950 and 2021 available in the PubMed database. The study showed that the frequency with which papers in the fields of medicine and health now use features such as short sentences, verbs indicating movement, eloquent language, and first-person narratives has increased.
The change is attributed to researchers’ efforts to increase the visibility of their work during a period in which global scientific output multiplied rapidly. “Researchers know that for their work to be cited, it needs to be remembered. And to be remembered, it needs to be understood. That’s why there’s greater effort to write in an understandable and engaging way,” said the lead author of the study, ecologist Sean Connell, from the School of Biological Sciences of the University of Adelaide, in Australia, to Pesquisa FAPESP by email.
“We see authors being more assertive by saying ‘there was a greater risk of death when X was exposed to Y,’ instead of ‘the risk of death was different when X was exposed to Y.’ The first phrase is more informative and gets straight to the point,” Connell explained. It is also true that a set of changes was observed in the study, such as the increased use of numbers and expressions with three or more nouns, which moved in the opposite direction and, according to the authors, may have required extra effort to understand the texts.
Another point highlighted in the work is the increased use of first-person writing, which creates empathy with readers. This personal language—once disapproved of by journal editorial guidelines—began gaining acceptance in the 1950s and has been gaining space ever since, according to the study. “This change represents a central question,” comments linguist Carlos Vogt, researcher and former dean of the University of Campinas (UNICAMP). “According to the analysis, the style of scientific writing shifted toward a more personal narrative, revealing the researcher’s engagement and connection to what is being expressed,” he says. He observes that science has developed its own language and code since it became institutionalized, between the seventeenth and nineteenth centuries, which, to a certain extent, has excluded the nonspecialist public. “To understand, participate, and communicate within this world, it became necessary to be sufficiently trained and to master this technical and specialized code,” he explains.
The article suggests that science communication is undergoing a critical moment, marked by a growing movement toward greater openness, engagement, and accessibility in the way knowledge is shared, observes Vogt, who served as president of the FAPESP Board of Trustees between 2002 and 2007 and is currently president of the Conrado Wessel Foundation. “It’s moving away from the traditional canonical text, as if the text were narrating itself, and toward a more personal narrative,” he analyzes. In the article from Scientometrics itself, the Australian researchers use “we” to present the results. “We believe that these changes help reduce the cognitive load and increase the readability of scientific texts,” they wrote. The concept of cognitive load refers to the amount of mental effort that a reader needs to expend to understand a text.
In the assessment of Gilson Volpato, a retired professor from São Paulo State University (UNESP) and expert in scientific methodology and writing, the use of personal language can have an important effect: signaling to the reader that the results are not absolute truths and may be reassessed by other studies or reinterpreted. “When a study says ‘we, the study group, have come to these conclusions,’ the authors express the idea that the results are the current accepted understanding and that they are subject to reevaluation by others knowledgeable in the field, but any change must be supported by evidence,” he adds.
In his assessment, texts are using this more personal tone for that reason, and not necessarily because the authors want to produce more involving and engaging texts. Volpato is skeptical of the idea that it is possible to create scientific articles with sufficiently open language to be understood by the general public. For him, the complexity of the scientific topics covered in the studies is incompatible with this ambition in most cases.
The authors of the study developed software to count the frequency, in every 100-word sequence of the abstracts, with which different textual elements or characteristics appeared, such as the use of terms denoting uncertainty or multiword expressions that reduce reading fluidity. They also assessed changes in sentence length and how verbs and nouns were used. Some of these characteristics had the knack for making the texts more assertive, for example the decline in the use of hedge words. They are terms that denote uncertainty (such as “apparently,” “maybe,” “could”), which, when poorly used, can weaken good arguments. In articles from the 1950s, almost one in every 100 words of the abstracts was a hedge word—this frequency has nearly halved in current texts.
The reduction in the number of words per sentence—from 29 in 1960 to less than 15 in 2021—was considered positive by the authors. For Sigmar de Mello Rode, former president of the Brazilian Association of Scientific Editors (ABEC-BRASIL) and professor at UNESP’s São José dos Campos campus, the shortening of sentences is one of the article’s positive takeaways. “This has been one of ABEC’s concerns; editor training stresses the importance of concise and objective writing for greater clarity,” he says. However, he notes that the study in Scientometrics was limited to texts written in English. “If the research had looked at texts written in Spanish or Portuguese, the result might have been different, maybe even the opposite. We, as Portuguese speakers, tend to be more verbose.”
Other trends identified in the article were assessed as neutral, such as the increased use of nouns (from 29 to 31 for every 100 words) and verbs (from 9 to 11) since the 1980s, trends that began to accelerate from the 2000s. According to the researchers, this suggests “an increasing effort by authors to create more engagement, mental imagery, and narrative fluidity in their texts,” as they wrote in the article.
Volpato raises some caveats regarding the study’s design, such as the lack of selection criteria for the articles analyzed. “They indiscriminately assessed the evolution of writing in thousands of abstracts. Obviously, there are far more problematic publications than good quality ones. I would like to know what area of science the analyzed articles contain, and to assess the scientific writing separating good-quality articles from poor-quality ones,” he states. The fact that the study focused solely on the abstracts of the articles—short texts that summarize the content—may have limited the scope of the analysis. “The abstract may even work as a good showcase for the text, but from a stylistic analysis perspective, examining the full text is essential” Vogt notes.
Connell, author of the study in Scientometrics, observes that a large proportion of authors still adhere to what he calls the “official style” of scientific writing, marked by an impersonal and technical tone and heavy use of jargon. He argues, however, that researchers who write in a more appealing and empathetic manner may have an advantage over the others. In another article, which he published in July in the journal Biological Conservation with a colleague from the University of Adelaide, the ecologist analyzed the characteristics of scientific writing in the 500 most cited articles in the field of biodiversity conservation, according to the Web of Science database, and compared them against the 500 least cited articles in the same field. The main conclusion was that, in general, the heavily cited articles contained broad titles, used persuasive language, and included clear explanations about the context and relevance of their research, characteristics that were less present in the rarely cited articles.
Vitória Bas
Sensationalism
According to the study in Scientometrics, the use of language described as “sensationalist” has increased over the past 40 years. The term includes expressions that are not related to the clarity of the text, but are frequently used with the aim of convincing the readers of the importance of the work. The trend is most evident in the use of terms that emphasize importance, priority, and urgency (such as “convincing,” “fundamental,” and “emerging”). Although research that is innovative in nature justifies the use of these expressions, in many cases they overstate the value of the results, making the analysis inaccurate. “We see that sensationalism is on the rise and that the use of hyperbole has increased exponentially. Although it’s a way of grabbing readers’ attention, there is a cost in terms of credibility,” says Connell. “If scientists start to distrust the way their colleagues are describing their data, the public and the politicians who adjust their behavior based on that information could also feel less confident. This puts the reputation of science at risk,” he adds.
Psychologist Mônica Ramos Daltro from the Bahian School of Medicine and Public Health, where she teaches a course on scientific writing, states that the article by the Australian team highlights how researchers have communicated the results of their work and offers relevant historical analysis. “It gives relevant tips about how to write and reinforces the idea of objectivity and clarity in writing. I liked when the article emphasized the importance of short sentences, something I always tell my students,” she notes.
However, Daltro sees a risk in seeking language that reduces the text’s cognitive load and facilitates understanding. “By prioritizing information in detriment to reflection, there is a risk of emptying the text of its critical complexity and richness,” she warns. With colleagues, Daltro published an article in October 2024 in the journal Ensaio: Avaliação e Políticas Públicas em Educação, about authorship in scientific texts, in which she argues that scientific writing should aim to raise evolving questions and provoke debate, rather than just present data.
The story above was published with the title “The challenge of writing clearly” in issue 355 of September/2025.
Scientific articles
HOHMANN, M. H. et al. The evolution of scientific writing: An analysis of 20 million abstracts over 70 years in health and medical science. Scientometrics. v 130. Jun. 2025.
HOHMANN, M. H & CONNEL, S. D. Comparing the writing styles of highly and rarely cited papers in conservation biology. Biological Conservation. v. 307. Jul. 2025.
DALTRO, R. M. et al. Trilhas e desafios da experiência de autoria na escrita científica. Ensaio: Avaliação e Políticas Públicas em Educação. v. 32, n. 125. Oct. 2024.
Republish
