A common issue encountered by editors is overlap of text with an author’s own previously published work, particularly with the increasing use of plagiarism detection software. This practice is known as ‘text recycling’ (also sometimes referred to as ‘self-plagiarism’). Opinions on the acceptability of text recycling vary greatly and it can be a challenge for editors to know how to deal with it once it has been identified.
The topic for discussion at this Forum was ‘Text recycling’. Self-plagiarism, also referred to as ‘text recycling’, is a topical issue and is currently generating much discussion among editors. Opinions are divided as to how much text overlap with an author’s own previous publications is acceptable, and editors often find it hard to judge when action is required. In an attempt to get some consensus and consistency on the issue, editors at BioMed Central produced some guidelines.
An article was published in July. In October, a corrigendum was published to correct large sections of unattributed text. Two weeks later the journal and publisher received a complaint from a reader who accused the author of the published article of using text from an unpublished collaborative manuscript on which the published author was participating. This participation on the collaborative work was initiated and ongoing during the time that the manuscript was being prepared for publication at the journal. The unpublished collaborative work has not yet been published.
The reader requested retraction of the published article, with the possibility of a republication only when all collaborators of the unpublished work were in agreement with the article content.
The publisher and journal initiated the procedure outlined in the COPE flowchart 'Suspected plagiarism in a published manuscript'. The editor-in-chief has requested feedback from the published author on the reason for the large overlap in text.
Question(s) for the COPE Forum
• Should the publisher and journal publish an 'expression of concern' while continuing with the investigation? • Is this the correct approach in this situation?
The editor told the Forum that she had checked the articles and the degree of overlap of text was nearly 20%. The Forum warned that percentages can be misleading, and the editor needs to look carefully at where the overlap occurred (in the introduction and methods may be fine, but it can be more problematic in the results or conclusion sections).
The institution(s) may need to be involved in this case.
The Forum would advise against an expression of concern on the article as these are generally used for ongoing unresolved cases. In this case, there is nothing proven or finalised —it may ‘end’ in a correction or a retraction. Hence it may be premature to publish an expression of concern and the editor should wait for a response from the authors. There is also the issue of the negative connotations of an expression of concern and/or stigma for the author, which may be unwarranted
A suggestion was for a less permanent ‘Editor’s note’ on the article for now, written in neutral terms.
The journal’s review of the guidelines on text recycling led to the conclusion that the scientific content was not disputed, and in fact the article adds to the body of knowledge. Also, the text recycling was not in the discussion or conclusions but rather in the methodology. The journal decided not to publish an expression of concern or retract the publication. The editor considers the case closed.
Recently, our journal has introduced systematic analysis of all submitted manuscripts for plagiarised text, using anti-plagiarism software. We had noticed increased incidences of recycling of existing text which is why we introduced the systematic check. It turns out that a large proportion of the submitted manuscripts (an estimated 30–50%) yield positive results, with copy values of somewhere in the region of 25% to >35%. These are substantial values and certainly beyond fortuitous incidences.
However, in almost all cases it is difficult to suspect acts of conscious (self)plagiarism as the copied text (ranging from single sentences or fragments of sentences to passages of 2–3 sentences) can be attributed to a very large number of sources: often more than 60, and in one case 129 different sources. It looks as if copying text containing what is perceived as elegant expressions has become a means of improving lack of language skills.
In principle, there is no issue of scientific fraud or even plagiarism of ideas or concepts to be suspected. But also in principle, a text that consists of one-third of passages that can be attributed to other sources is not satisfactory and is not what we would consider good scientific writing practice. The question is how to deal with these cases that we see in a quickly growing number? It is not fair to authors who produce good science to penalise them for trying to cope with their limited language skills. It is not fair either, to give the advantage of facility to those authors who easily copy from existing work, over those authors who make the conscious effort to avoid such doubtful practice.
Presently, when significant proportions of text have been copied from a large number of sources (as mentioned above), I do not take this into account when making a decision based on the science of the paper but inform the authors that we consider this a doubtful practice that should be avoided in future manuscripts.
Question(s) for the COPE Forum • What kind of coherent policy should the journal have on this issue? • Have any COPE members had experience with similar situations?
The Forum advised that any policy or process should be based on the context of the content. Choosing a particular benchmark or cut-off for anti-plagiarism software is unreliable, and the a granular reading of the text is often needed to understand the level of copied text. Clearly duplication is more worrying in the results section of a paper compared with the introduction or materials and methods. If large sections of text are duplicated in the discussion/conclusions section of a paper, that would raise concerns.
In terms of technical issues related to the anti-plagiarism software, it is advisable to remove stock phases before running the software. The editor may wish to decide on what size of strings of words to exclude.
The Forum suggested that the editor’s current policy seems very reasonable. In addition, a suggestion was to look carefully at attribution. If large sections of consecutive sentences are not attributed, that could be problematic and the editor may wish to ask the author for an explanation. It is more serious when the words and ideas an author has reproduced are not their own.
The Forum also advised considering the type of paper—in a research paper, the editor may feel that borrowed phases can be overlooked, but this may be unacceptable in a review paper if the author is purporting the presentation of novel ideas.
Regarding a process, the Forum suggested the editor may want to clearly state the journal policy in the instructions to authors, to head off similar cases in the future. The editor should continue to check all submitted manuscripts for plagiarism and duplicated text using anti-plagiarism software; reject those with moderate/major overlap of text; if malicious intent is suspected, contact the author’s institution; if the authors are junior researchers, consider asking them to rewrite passages and re-submit.
There is a role for the institution in these cases as they govern the behaviour of their researchers. Institutions need to investigate any such cases, and educate and support those who are unaware of good practice. Hence the editor should contact the institution if he suspects misconduct or if he believes that good publication practices need to be reinforced. This is especially true if the editor sees patterns emerging within particular institutions or countries; it is up to the institution to investigate these practices. Collective awareness raising of the issue is needed among authors and institutions.
The editor believes the case is closed although he remarks that unfortunately the phenomenon has not gone away and new examples crop up almost every day.