There is interest in appreciating if ‘tortured phrases’ (i.e., odd linguistic phrases in scientific literature that purportedly show technical explanations, but which actually are nonsensical or difficult to interpret) exist in the sport literature. To gain an appreciation of this phenomenon, the Tortured Phrases Detector of the Problematic Paper Screener (PPS) was consulted on 9 September 2023, revealing 160 results. After manual screening and filtering, 54 papers related to any aspect of sport (as assessed by papers’ titles) were examined, in consultation with their entries at PubPeer (if available) to appreciate the level and extent to which tortured phrases have infiltrated the sport literature. Of the 54 papers examined, 41 were retracted (or withdrawn) to date (11 July 2024), mostly from Elsevier’s Microprocessors and Microsystems, but none indicated tortured phrases as an explicit reason for retraction in their retraction notices. Even though the absolute volume of papers with tortured phrases is tiny relative to the wider body of sport-related literature, that argument is countered by noting that these 54 papers had already, at least until 6 October 2023, collectively been cited 449 times, suggesting that imperfect or fraudulent science tainted by tortured phrases has already begun to permeate the wider sport science literature.
communication, editorial oversight, ethics, plagiarism aversion, reproducibility
There is a very small, but dedicated, literature focused exclusively on the integrity of sport literature, which has some serious integrity issues, as exemplified by 237 or 908 items with retractions and/or expressions of concern in the topics “Sports Science” or “Sports and Recreation,” respectively.1 The very first analysis of retractions in the sport literature assessed 52 papers, eight deriving from the Journal of Applied Physiology, noting that 61% were associated with misconduct, while 39% were labelled as honest error (Kardeş et al., 2020). In kinesiology research, some of the questionable research practices include publication bias, exploratory research that is reported as confirmatory, post hoc hypothesizing (or HARKing), excessive self-citation, and data fabrication (Tiller & Ekkekakis, 2023). Sports researchers also need to reflect on proper experimental design and statistical analyses (Bernards et al., 2017). From 129 original research papers published in four top-ranked sports journals, (Büttner et al., 2020) found that the primary study hypothesis in about 26% of them was only partially supported by the results, raising the alarm about three questionable research practices (HARKing, P-hacking, and cherry-picking). Given the wide-ranging problems with statistical analyses in sports research (McLean et al., 2021), some have suggested the need to collaborate with statisticians when publishing sports research (Sainani et al., 2021). Some sports journals have seen a marked increase in submissions, with several authors not having noble publishing objectives, requiring journal editors to raise the bar and fortify ethical screening procedures (Abt et al., 2022). One of the important tasks that editors and peer reviewers have is being able to distinguish valid science from pseudoscience (Tiller et al., 2023). Such practices would involve the implementation of more transparent research practices (Caldwell et al., 2020; Schulz et al., 2022). Gaspar and Esteves (2021) advocate for the need to better appreciate misconduct within sport science. To meet that end, and given that one of the ultimate aims of sports researchers who publish is to have their work cited (Khatra et al., 2021), the objective of this short communication is to bring to the attention of the sport science community a relatively new phenomenon that straddles the line between poor scientific writing practices and, in some instances, misconduct.
In this paper, focus is placed on an issue that has not yet – to the author’s knowledge – been formally addressed in the sport literature or by the sport scientific community: a linguistic and ethics-related phenomenon, ‘tortured phrases’, which can broadly be described as odd phrases that purportedly show technical terms in the scientific literature, but that are not, and may have arisen from imperfect translation or reverse translation, one reason being the desire of the authors of those papers to avoid plagiarism being detected, but instead resulting in non-sensical text (Cabanac et al., 2021). In biomedicine, the presence of tortured phrases not only negatively impacts the specificity of writing. If these terms are used by early career researchers or others who might have limited experience, they might copy and/or cite such incorrect terms, thinking that they are accurate (Teixeira da Silva, 2022). Authors of papers in which tortured phrases appear are to blame, but so too are journals that claim peer-review and stringent quality control as they might not be completing proper peer review or other processes (e.g., copyediting) associated with careful screening of the content of papers they publish (Moradzadeh et al., 2023), to determine whether they contain tortured phrases (Teixeira da Silva, 2022b). No scientific field is immune to being “infected” by tortured phrases, such as neuroscience (Teixeira da Silva, 2023; Teixeira da Silva & Daly, 2023). While not always the case, the presence of tortured phrases in an article might reveal a deeper set of ethical issues (Else, 2021), including the undeclared use of revision or translation services (Kendall et al., 2016; Teixeira da Silva, 2021; Teixeira da Silva et al., 2024). For this reason, tortured phrases can serve as an epistemic marker (or identifier) of potentially wider ethical infractions (Teixeira da Silva, 2023b). Tortured phrases are not limited to peer-reviewed literature, and might also be prevalent in preprints, i.e., non-peer-reviewed literature (Teixeira da Silva, 2023c).
To determine whether any papers, preprints, congress papers, or book chapters might contain tortured phrases, the Tortured Phrases Detector of the Problematic Paper Screener (PPS) was searched on 9 September 2023 using “sport” as the keyword (Cabanac et al., n.d.). This search yielded a total of 160 results, which were manually screened to identify, from titles, whether the articles were truly related to sport, or not. Following this filtering process, a body of 54 candidate papers was identified (Table 1). Of these 54 documents published between 2020 and 2023, 87% were articles (the rest being book chapters, preprints and proceedings papers), while 41 (76%) were retracted (or withdrawn). The latter status and corresponding statistic was last assessed on 11 July 2024. The highest incidence of tortured phrases was in Elsevier journals, followed by Springer Nature, mostly in Microprocessors and Microsystems and Arabian Journal of Geosciences, with a history of association with manipulated peer review as well as incompetent guest editors of special issues, leading to mass retractions.2 Fake, improper or manipulated peer review, especially of special issues, is an issue that is plaguing the integrity of the scientific literature (Rivera & Teixeira da Silva, 2021). According to PPS, these 54 articles had accrued 449 citations until 9 September 2023. None of the journal titles are related directly to sport. Although the papers are on topics related to sport, mostly via an applied prism (e.g., detection of movement of sportspersons, image analysis, etc.; see column 1 of Table 1), the vast majority were published in journals or books covering other specialties (e.g., computer science).
Authors in the sport sciences need to appreciate that when using online thesauruses or other software that might be used for reverse translation (e.g., QuillBot3), the selected output text might not always have the same and desired technical meaning. The English term to substitute the word might be correct in a broad sense (sensu lato), but incorrect in a strict technical sense (sensu stricto), i.e., technically or scientifically. Related to this broad versus narrow sense of linguistic terms in scientific writing, it is the responsibility of editors (specifically editors-in-chief) or journal management (including copyeditors employed by the journal or publisher) to ensure that once papers are accepted for publication, they are properly copyedited and screened for nonsense text, such as tortured phrases. Even if such papers have already been accepted for publication, if tortured phrases are detected at the copyediting stage, editors are obliged to pause the processing and publication of that paper, and initiate an investigation, including of the peer reviewers, to assess fully why the authors employed such terms, and why peer review failed to detect them. This study only provides a very small window of appreciation of this topic (see limitations in Table 1 footer), limiting itself to a few dozen papers, but a large-scale study is merited.
|
|
|
|---|---|---|
10.1016/j.micpro.2020.103490 (recognition of movement in sport)* | designs preparing unit | graphics processing unit |
facial acknowledgment | facial recognition | |
keen gadget | smart device | |
molecule swarm | particle swarm | |
uphold vector machine | support vector machine | |
profound learning | deep learning | |
10.1016/j.micpro.2020.103593 (recognition of movement in sport)* | clamor decrease | noise reduction |
picture recognition | image recognition | |
brutal conduct acknowledgement | recognizing rough play? | |
10.1016/j.micpro.2020.103583 (sport injuries)* | touchy factor | sensitive issue? |
X-beam | X-ray | |
large information | big data | |
blunder rates | error rates | |
10.1016/j.micpro.2020.103437 (recognition of movement in sport)* | administered learning | supervised learning |
concealed/shrouded Markov model | hidden Markov model | |
equal calculation | parallel computing | |
histogram levelling | histogram equalization | |
likelihood appropriation | probability distribution | |
picture division/handling | image segmentation/processing | |
10.1109/ICRTCCM.2017.57 (video-based sport event recognition and classification) | acknowledgment framework | recognition system |
characterization execution | classification performance | |
highlight vector | feature vector | |
Gaussian appropriations/dissemination | Gaussian components/distribution | |
information mining | data mining | |
likelihood appropriation/circulation | probability distribution | |
misfortune work | loss of function | |
picture grouping | image classification | |
speculation capacity | predictive ability | |
10.48550/arXiv.2209.07528 (sports analytics) | help vector machine | support vector machine |
K-closest neighbors | K-nearest neighbors | |
bolster vector | support vector | |
choice tree | decision tree | |
exactness rate | accuracy rate | |
grouping and relapse | classification and regression | |
grouping errand | classification task | |
prescient model | predictive model | |
10.1016/j.micpro.2020.103423 (detection of movement and a body’s physiological parameters in sport)* | center internal heat level | core body temperature |
cloud worker | cloud server | |
directing convention | routing protocol | |
far-off wellbeing checking | remote health assessment | |
inertial estimation unit | inertial measurement unit | |
irresistible ailment | contagious disease | |
mist registering | cloud computing | |
radio recurrence | radio frequency | |
sign to commotion | signal to noise | |
10.1016/j.micpro.2020.103753 (movement analysis in swimming)* | 10-crease cross validation | 10-fold cross validation |
choice emotionally supportive network | dynamic decision support system | |
clamor | noise | |
condition of-workmanship | state-of-the-art | |
creating nations | developing countries | |
discourse acknowledgement | speech recognition | |
enormous information | big data | |
human services conveyance | healthcare delivery | |
information gushing | information overload | |
nourish forward | feed forward | |
p esteem | p value | |
palatable execution | satisfactory performance | |
Parkinson's ailment/infection/malady/sickness | Parkinson's disease | |
PC vision | computer vision | |
shrouded layer | hidden layer | |
sigmoid capacity | sigmoid function | |
slope plunge | gradient descent | |
therapeutic gadgets | medical devices | |
unfriendly wellbeing | adverse health | |
10.1109/ICESIC53714.2022.9783514 (machine learning to predict outcome of cricket matches) | directed learning | machine learning |
AI calculation | machine learning algorithm | |
arrangement calculation | classification algorithm? | |
characterization relapse~5 | classification and regression | |
choice tree | decision tree | |
directed learning calculation | supervised learning algorithm | |
logistic relapse | logistic regression | |
man-made brainpower | artificial intelligence | |
regulated AI calculation | supervised machine learning algorithm | |
flickering | batting | |
10.1063/5.01143605 (unclear) | heat/warmhmove/movement/ | heat transfer |
limit layer | boundary layer | |
standard differential condition | ordinary differential equation | |
thick dissemination | viscous dissipation | |
warm conductivity | heat conductivity | |
warm radiation | heat dissipation | |
10.1155/2022/1061461 (sports training education management)*6 | R2/p/t esteem | R2/p/t value |
direct/straight relapse | linear regression | |
10.1515/9783110790146-011 (augmented and virtual reality for sports) | expanded/increased reality | augmented reality |
PC upheld | computer-aided | |
PC vision | computer vision | |
face acknowledgment | face recognition | |
high-exactness | high-resolution | |
information picture | input image | |
10.2139/ssrn.3620017 (AI in sport) | distinguishing proof | identification |
Formula 1 tennis | ? | |
human-made consciousness/ | artificial intelligence | |
wellspring of income | source of income | |
profound figuring | deep learning | |
10.1007/978-981-15-5258-8_22 (soccer anthropometry and player attributes) | high pay nations | high-income countries |
onlooker base | fan base | |
10.1016/j.micpro.2021.103945 (mobile communication and aerobics)* | incorporated aloof gadget | integrated passive device |
relative blunder | relative error | |
global wandering | international roaming | |
yield layer | output layer | |
fake neural organization | artificial neural network | |
help vector machine | support vector machine | |
10.1016/j.micpro.2021.103924 (sports course management)* | Linux Working Framework | Linux Operating System |
information mining | data mining | |
recognizable proof | identification | |
10.1016/j.micpro.2020.103648 (sports dance movement)* | PC vision | computer vision |
info picture | input image | |
k-implies calculation | k-means algorithm | |
move learning | transfer learning | |
test arrangement | test set | |
10.1016/j.micpro.2020.103348 (sports movement detection)* | PC vision | computer vision |
distinguishing/recognizable proof | identification | |
image acknowledgment | image recognition | |
10.1007/s41133-019-0025-2 (virtual reality in sport) | choice help | decision support |
enlarged/increased reality | augmented reality | |
10.1016/j.micpro.2021.104083 (sports injury prediction)* | Gaussian dissemination | Gaussian distribution |
information mining | data mining | |
muscle shortcoming | muscle weakness | |
prescient model | predictive model | |
muscle gatherings | ? | |
muscle harm | muscle injury | |
muscle lopsidedness | ? | |
sports wounds | sports injuries | |
10.1016/j.micpro.2021.104063 (RoboCup Federation international soccer competition)* | computerized reasoning | artificial intelligence |
hereditary calculation | genetic algorithm | |
picture preparing | image processing | |
self-ruling vehicles | autonomous vehicles | |
fluffy regulator | fuzzy logic? | |
10.1016/j.micpro.2021.104019 (sports injuries)* | focal sensory system | central nervous system |
lactic corrosive | lactic acid | |
recurrence band | frequency band | |
worldwide situating | global positioning | |
10.1016/j.micpro.2021.104120 (sports information management)* | common language handling | natural language processing |
picture acknowledgment | image recognition | |
preparing information | training data | |
10.1016/j.micpro.2021.103927 (sports app to assess fitness and performance)* | versatile stages | mobile platforms |
arbitrary woods | random forest | |
informational index | dataset | |
10.1016/j.micpro.2021.103975 (sports injury simulation)* | increased reality | augmented reality |
PC vision | computer vision | |
picture acknowledgement | image recognition | |
(sports administration)* | man-made cognizance | artificial intelligence |
neural associations / organization | neural networks | |
10.1016/j.micpro.2020.103676 (badminton injuries)* | cloud worker | cloud server |
shrewd sensors | smart sensors | |
upper appendage | upper limb | |
10.1016/j.micpro.2021.104000 (sports industry in coastal cities)* | versatile correspondence | mobile communication |
colossal information | big data | |
monetary development | economic development | |
10.1016/j.matpr.2021.01.489 (physical activity in sports)* | acknowledgment execution | recognition performance |
inertial estimation unit | inertial measurement unit | |
10.1016/j.micpro.2021.104181 (big data in sports)* | National B-Milk Acceptance/National Board of Accreditation (NBA) | National Basketball Association (NBA) |
area of enthusiasm | region of interest | |
information mining | data mining | |
10.1016/j.micpro.2020.103792 (sports image segmentation)* | picture division | image segmentation |
10.1016/j.micpro.2021.104069 (imaging techniques to assess sorts injuries)* | PC vision | computer vision |
distinguishing proof | identification | |
10.1016/j.imavis.2021.104214 (automated detection of sports movements) | concealed state | hidden condition? |
recurring neural network | recurrent neural network | |
10.1016/j.micpro.2020.103345 (sports video and image analysis)* | help vectors | support vectors |
help vector machine | support vector machine | |
10.1016/j.micpro.2020.103389 (detection of sports motion and injuries)* | preparation information | training data |
programmable rationale gadgets | field programmable gate array (FPGA) | |
10.1016/j.micpro.2021.103837 (sports training management)* | man-made brainpower | artificial intelligence |
10.1016/j.micpro.2021.103918 (analysis of basketball game images)* | concealed Markov model | hidden Markov model |
information mining | data mining | |
10.1016/j.avb.2021.101587 (assessment of physical and psychological stress in sportspersons)* | Fourier change | Fourier transform |
exactness accuracy~5 | accuracy | |
10.1016/j.micpro.2020.103654 (prediction of sports injuries)* | info information | input data |
info picture | input image | |
10.1007/s12517-021-08077-0 (unclear)* | PC vision | computer vision |
fluffy induction | fuzzy induction | |
10.1016/j.micpro.2020.103335 (unclear)* | keen gadget | smart device |
shrewd home | smart home | |
10.1016/j.micpro.2021.103984 (promotion of public sports)* | shrewd gadget | smart device |
10.1007/s12517-021-07335-5 (unclear)* | image determination | image assessment |
10.1007/s12517-021-08185-x (unclear)* | information mining | data mining |
10.1016/j.jbusres.2021.03.031 (management of sport sponsorship) | distinguishing proof | identification |
10.1007/s12517-021-07198-w (unclear)* | information mining | data mining |
10.3390/ijerph18179049 (AI to improve students' physical quality and motor skills) | information mining | data mining |
10.3390/app12094429 (computer vision in sports) | highlight extraction | feature extraction |
10.1155/2022/9789933 (development of sports facilities)* | forbidden search | Tabu search |
10.1007/s00779-019-01242-z (development of the sports industry) | huge information | big data |
10.1016/j.micpro.2020.103445 (monitoring health of sportspersons)* | coronary illness | coronary artery disease |
10.1016/j.micpro.2021.103900 (prediction of sports injuries)* | information mining | data mining |
10.1016/j.micpro.2020.103331 (mobile communication for improved sports coverage)* | figuring asset | computing resource |
10.1016/j.micpro.2020.103584 (sport tourism)* | movement business AND tourism | tourism industry |
1As assessed by the Problematic Paper Screener (PPS), with additional verification on PubPeer using documents’ DOIs.
2Actual, presumed, or supposed topic based on the author’s perception; whenever the topic was unclear, these instances have been noted as “unclear.”
3Lists derived from PPS, PubPeer, and author’s determinations. Not listed in any specific order.
4Lists derived from PPS, PubPeer searches in unrelated entries, and author’s determinations; several possibilities are not entirely clear, and are denoted as a question mark (“?”).
5This entry was determined exclusively from the PPS entry since the full text could not be obtained.
6The retraction notice does not use the term “tortured phrases” but alludes to the integrity of the content more broadly, claiming that it was one of several indicators of “systematic manipulation of the publication process.”
*Retracted or withdrawn (although two terms are used, they are considered the same, i.e., literature whose scholarly record was removed for any reason).
Limitations and disclaimer: The lists of tortured phrases were drawn mainly from PPS and PubPeer, where errors or false positives might exist. Listed papers might carry more tortured phrases than are indicated. In several papers, the text is literally incomprehensible; therefore, tortured phrases may appear insignificant relative to the full body of text. Given that this is a nascent branch of publishing ethics, there are no absolute guarantees that the interpretation of the existence of these tortured phrases is correct, absent a confession by authors of the use, for example, of translation or reverse translation software, or other software that paraphrases sentences.
Editor-in-Chief:
Claudio R. Nigg, University of Bern, Switzerland
Section Editor:
Sebastian Ruin, University of Graz, Austria
The author has no funding or support to report.
The author has declared that no competing interests exist.
All relevant data are within the paper.