Raising Awareness in the Nursing Community About ‘Tortured Phrases’

Nursing and other biomedical literature consists of many specializations, each of which relies on a linguistic “code”, referred to as jargon or technical terms, that allow specialists in the same field to communicate with others. Inaccurate scientific communication caused by using confusing phrases or incorrect jargon may lead to a misunderstanding of research findings, a treatment, or nursing care intervention.

The term ‘tortured phrases’ (TPs) was coined for the inaccurate representation of technical or scientific terms, or jargon, typically where one or more words (e.g., nouns) are synonymously replaced with other words (Cabanac et al., 2021). These phrases are incorrectly used, replacing accepted jargon, as exemplified by some generic cases in Table 1. Tortured phrases can arise when using artificial intelligence (AI)-driven translation or paraphrasing software, i.e., text synonymizers, which modify words and phrases, serving as one strategy for authors to avoid the detection of plagiarism or reduce the volume of textual similarity (Cabanac et al., 2021; Teixeira da Silva, 2025). The presence of TPs in nursing and other biomedical literature can impact the interpretation of scientific information and its message, relaying it inaccurately to readers and end-users (Teixeira da Silva, 2022). A large language model (ChatGPT) accurately reversed and corrected TPs to native jargon (Teixeira da Silva & Tsigaris, 2024).

This editorial serves to raise awareness among nurse researchers, who may be authors, peer reviewers, or editors, that articles they might be reading or relying upon, either to cite in academic papers or in search of evidence in both gray literature and peer-reviewed or indexed literature may contain TPs. TPs may result from the undeclared use of AI by authors (Oermann, 2024) or from translation or text-changing software to avoid plagiarism, and may indicate scientifically compromised literature that may have other integrity issues. In their role as authors, nurse researchers who encounter TPs in literature they are reading will need to decide whether they should cite such work in their own academic papers because those TP-tainted articles might at some point be retracted, thereby risking the information or scientific “integrity” of their own article. In their role as peer reviewers or editors, nurse researchers who encounter TPs (e.g., those in Table 1) in articles they are peer reviewing or planning to publish should query the authors about the origin of that imprecise terminology, while matching the authors’ responses with any ethical declarations regarding the use (or absence of use) of AI. If multiple TPs appear in the same article, and the authors declare that no AI was used, then an ethical investigation may be warranted.

AI is itself not a problem, but it only becomes one when it is abused or misused. TPs arise due to imperfect AI, or due to AI-based manipulations, e.g., the use of online synonymizers. Due to constraints of human resources in academic publishing, e.g., volunteer peer reviewers and editors who typically have other roles in nursing, journals and publishers could tap the power of AI to detect TPs during submission, as they would screen papers for plagiarism. Like other “ills” that are negatively impacting the integrity of the scientific literature, a human + AI interaction may be needed to find a sustainable solution.