“Tortured phrases” in preprints

Abstract

This commentary takes note of the existence of “tortured phrases” (i.e. unspecific jargon or confusing alternative phrases), as indexed in the Tortured Phrases Detector of the Problematic Paper Screener (PPS) (sourced on January 10, 2023) in 213 preprints, 13 of which are related to COVID-19. Select “tortured phrases” in 11 preprints are highlighted, to offer readers an appreciation of this phenomenon. The incorrect representation of jargon in the medical and health literature may risk confusing readers by reducing the impact of effective and precise communication. Whereas some “tortured phrases” might represent simple mistranslations, in other cases, an abundance of such terms in a single preprint might reveal a more serious ethical issue, such as the undeclared use of a paper mill or an unprofessional editing service. This commentary is thus only a spring-board to introduce this linguistic phenomenon and to encourage interested academics to explore more cases, the practical implications of their existence, and even the weaknesses and strengths of PPS. Caution is needed about excessive extrapolation of the existence of “tortured phrases”, so as not to automatically associate them with ethical infractions or misconduct.

Keywords:

• Effective communication
• editorial oversight
• plagiarism aversion
• preliminary findings

Subramanian et al. noted that the ICMJE recommendations request that upon submission to a peer-reviewed journal, that the existence of a preprint be properly declared¹. This is one of the few mechanisms in place to ensure the quality of preprints, which are still loosely regulated, except for simple moderation and screening procedures. In fact, the policies in place for preprints, even among ranked and indexed journals, is still very heterogeneous, and many inconsistencies have been observed to occur, including in the transparency of reporting².

There is also a potential ethical loophole associated with preprints, namely the lack of clear policies related to conflict of interest statements in preprints, relative to peer-reviewed journals³. That assessment noted that at least one additional funder was added to the peer-reviewed versions of 25% of the preprints that were examined, as well as unsubstantiated changes (additions or removals) in authorship³. In peer-reviewed journals, the ethical standing of papers is challenged when such important information is omitted, or when there are undeclared or opaque changes in authorship. For this reason, given that preprints are being more widely accommodated into the academic publishing work-flow, it was recently argued that preprints should be treated as “ethical equals” with peer-reviewed papers, i.e. scrutinized with the same level of ethical standards⁴.

This commentary focuses on another emerging issue in preprints, namely of “tortured phrases”, which are non-standard derivatives of established technical terms or jargon that may have arisen – among other reasons – from reverse translation techniques to avoid the detection of plagiarism⁵. Students, early career researchers, non-specialists, members of the public, including both native and non-native English speakers, who might not be entirely familiar with technical jargon, especially in the biomedical sciences that employs a wide range of specialized and complex terminology, might accidentally employ “tortured phrases”, thereby reducing the specificity of their biomedical writing⁶. Papers in which “tortured phrases” appear also reveal that editing-related flaws or oversight might have occurred in that journal⁷. Finally, thematically-specific analyses of “tortured phrases”, such as for Parkinson’s disease⁸, are useful and necessary in order to reduce the spread of misinformation due to the misuse of technical language⁹.

To gain an initial appreciation of the incidence of “tortured phrases” in preprints, the Tortured Phrases Detector of the Problematic Paper Screener (PPS) was sourced on January 10, 2023¹⁰.

In PPS, from among 6378 entries related to any document type, 213 entries (3.3%) associated with preprints were identified (Supplementary file). The preprint servers or providers with the highest incidence of “tortured phrases” in PPS were Elsevier’s SSRN Electronic Journal (127 entries; 59.6%), Research Square Platform LLC’s Research Square (59 entries; 46.5%), and Cornell University’s arXiv (18 entries; 8.5%). The remaining nine entries were in Cold Spring Harbor Laboratory’s bioRxiv and medRxiv, ScienceOpen’s ScienceOpen Preprints, and MDPI’s Preprints.org. These 213 preprints have collectively been cited 258 times thus far, the most highly cited of these preprints (medRxiv; DOI: 10.1101/2020.05.01.20088211) having been cited 82 times. Twelve of the 213 preprints with “tortured phrases” were specifically related to COVID-19 (as assessed exclusively by the title) (Table 1).

Table 1. COVID-19-related preprints with “tortured phrases”, as indexed in PPS^a.

Preprint DOI	“Tortured phrase(s)” (verbatim)	Likely or presumed jargon
https://doi.org/10.21203/rs.3.rs-1725511/v1	chest X-beam picture tests graphic preparing unit (GPU); pharmaceutical units (GPUs) serious intense respiratory disorder (SARS) X-beam	chest X-ray film diagnosis graphics processing unit (GPU) severe acute respiratory syndrome (SARS) X-ray
https://doi.org/10.1101/2020.05.27.20100297^b^,^c https://doi.org/10.48550/arXiv.2110.09384^b^,^c	convolutional neural system figured tomography (CT) man-made consciousness profound neural system x-beam	convolutional neural network (CNN) computed tomography (CT) artificial intelligence deep neural network X-ray
https://doi.org/10.1101/2020.07.08.20148924^b^,^d	hack converse interpretation polymerase chain response (RT-PCR) spewing yield layer	cough reverse transcribed polymerase chain reaction (RT-PCR) vomiting output layer
https://doi.org/10.2139/ssrn.3852865	convolutional neural organizations crude picture pixels multi-facet perceptron (MLPs)	convolutional neural networks raw image pixels multi-layer perceptron (MLP)
https://doi.org/10.21203/rs.3.rs-867959/v1	biting the dust Covid contamination 2019 (COVID-19) extreme acute respiration syndrome Covid 2 (SARS-CoV-2) neural association top notch picture	dying/deceasing coronavirus disease 2019 (COVID-19) severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) neural network high-resolution image
https://doi.org/10.21203/rs.3.rs-765835/v1	Center East Respiratory Condition Extreme/Serious Intense Respiratory Disorder (SARS) non-drug mediations (NPIs) social separating tainted people World Wellbeing Association (WHO)	Middle East Respiratory Syndrome (MERS) severe acute respiratory syndrome (SARS) non-pharmaceutical interventions (NPIs) social distancing infected individuals World Health Association (WHO)
https://doi.org/10.21203/rs.3.rs-1294582/v1	blunder rate blunder worth chest congestion	error rate error value pulmonary congestion
https://doi.org/10.21203/rs.3.rs-2095290/v1	arbitrary woods COVID-19 plague/pestilence	random forests COVID-19 pandemic
https://doi.org/10.1101/2020.05.01.20088211^b^,^e	ailment fringe limitations	failure/disease border control
https://doi.org/10.1101/2020.07.15.205567^b	bleakness Chi-square metric lymph hub	morbidity Chi-squared test lymph node
https://doi.org/10.21203/rs.3.rs-100956/v3^b^,^f	extreme intense respiration syndrome-coronavirus 2 (SARS-CoV-2)	severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)

^aOnly select examples are noted (January 10, 2023); where possible, examples related to health or COVID-19 are indicated; sample derived from the mention of COVID in the preprint’s title.

^bClues drawn from PubPeer (searchable using the preprint’s DOI).

^cDuplicate preprints (albeit with cosmetic differences) on different preprint servers, and with different authorship.

^dIdentical errors propagated into the final published paper: https://doi.org/10.30538/psrp-easl2020.0054.

^eIdentical errors propagated into the final published paper: https://doi.org/10.1109/ICCC51575.2020.9344870.

^fWithdrawn/retracted.

These cases suggest that more stringent ethical and screening policies are needed for preprints, not unlike peer-reviewed journals, prior to their online publication by adding an additional screening mechanism for “tortured phrases”, rejecting such preprints with imprecise terms to make them more acceptable to the biomedical community, and to fortify the precision of health communication. In fact, the issue of “tortured phrases” has, especially in cases where it occurs abundantly, a clear and distinct ethical component to it, not only potentially reflecting the lack of a paper’s integrity of information, but in more deviant cases, its potential link to other ethical infractions such as plagiarism⁵, improper peer review, or to commercial entities that produce papers at a cost, i.e. “paper mills”¹¹. The general rule according to both COPE guidelines and ICMJE recommendations is that the use of any third party service, including those that are related to language improvement or editing, must be explicitly acknowledged¹². In some extreme cases, when a paper with multiple issues (sensu lato) is discovered at the post-publication stage to have included “tortured phrases” or the undeclared use of paper mills, it risks being retracted. As one example, a 2020 conference proceedings published by IOP Publishing was retracted in late 2022 due to the existence of “tortured phrases” as well as the existence of plagiarism, and even though plagiarism was not explicitly mentioned in the retraction notice, the phrase “masking overlap of other work” suggested it¹³.

This commentary has two main limitations due to space restrictions. The 213 entries were only noted, but not examined in detail, except for the 11 cases in Table 1. As a new and developing field of ethics-related research¹⁴ and bibliometrics, many of these entries at PPS need to be independently verified for errors or inconsistencies while new cases are likely to be added over time. PPS, although a useful database because it is open to the public, is still an experimental platform, much like “tortured phrases” themselves, which implies that additional validation of entries is needed and detailed paper-by-paper scrutiny, not only to fortify the accuracy of PPS, but to give greater precision to the boundaries that define a “tortured phrase”.

Finally, a few brief words of advice. For authors: not all words in a thesaurus to represent a desired word have the same (and correct) technical meaning; formally acknowledge the use of editing services in your paper. For editors: ensure that a professional copyeditor screens paper prior to publication in order to detect the presence of linguistic oddities, like “tortured phrases”.

Transparency

Declaration of funding

This paper was not funded.

Declaration of financial/other relationships

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgements

None.