What a difference a publisher makes

How important, how much, how good? Sometimes it’s needed and sometimes it’s not. Sometimes it's not even provided. Does it matter? Copy editing, that’s what I’m talking about. It’s a special little focus of interest at the moment because publishers claim it as an important area of added value and want to demonstrate how much they contribute to the integrity of scholarly literature through providing it, while the proponents of self-archiving counter-claim that the author’s final version of an article – the one which contains all the changes advised or required by the peer review process – is a perfectly adequate version to be deposited in a digital repository for open access purposes. Such a focus of interest is it that a couple of studies have been published recently that have examined the differences between published and author-final versions of articles. Wates and Campbell looked at copy editing changes carried out on a set of science, humanities and social science articles at Blackwell Publishing (as was) and reported that the biggest category of corrections by the publisher was concerned with the references (42.7% of all copy editing changes), the next biggest category (34.5%) was concerned with minor syntactical or grammatical changes and a small proportion (5.5%) of changes corrected author ‘errors that might otherwise have led to misunderstanding or misinterpretation’.

In the other study, Goodman, Dowson and Yaremchuk looked at journals in biology and social sciences and found that publishers had corrected numerous small errors that affected readability, that there were a number of differences between author-final and published versions that were ‘confusing’ and that sometimes the publisher version and sometimes the author version was the more confusing. They also found that in two cases the publisher had omitted data ‘necessary to evaluate the validity of the conclusions’: introduced an error during copy editing, in other words.

So, where do these studies leave us? Somewhat confused, I’d say. Does copy editing add significant value? Does it add insignificant value? Does it even plague an article by introducing errors that weren’t there before? Is the author’s final version adequate for scholarly purposes? 'Yes' seems to be a possible answer to all these questions. Which gets us, well, not very far when the issue of trusted repositories is at stake.

A previous JISC-funded project, IRRA (Institutional Repositories and Research Assessment) showed that trust is a very significant issue with respect to repository content. The published version of an article is seen by some parties – research assessors, for example – as more trustworthy than the author’s institutional repository. Clearly, if a funder or assessor or employer has intentions of using repository content for funding, assessing or employing decisions there must be a straight story.

There won’t ever be a general one. The difference between the author-final version and the published version will only ever be able to be described for any individual paper. Should we face up to doing this? Yes, since it seems that versions can differ quite markedly; since authors may deposit out-dated or inaccurate versions – or more than one version – of an article or even, in exceptional circumstances, manipulate the article to make it look more significant than the peer-reviewing process established; and since it also seems that on occasions publishers may actually introduce errors at the copy editing stage, it becomes even more important to know where the differences are, and what they are – absolutely precisely. If an author is to be judged in any way on the basis of what he or she has placed in a repository, the precise differences between the author’s final version and what was actually published must be clearly demonstrable.

All of the work for the two studies mentioned above was done by hand and involved meticulous checking of one version against another for each article studied. That is definitely not a scalable process. A solution is on the way, though. Under the current round of repository-related funding, the JISC has agreed to support work on what we have called the VALREC (VALidating REpository Content) project. Oooh yes, I know how irritating all these tortuous acronyms are but every project needs one, believe me.

VALREC will ask stakeholders what levels of validation they would like to see, and what broad categories of differences would be helpful, such as ‘editorial differences’ and ‘content differences’. The project will then develop the technology to measure differences and generate a digital certificate for any article detailing the differences. An example of such a certificate is on the VALREC website. Not only will there then be a means to itemise the exact differences between the author-final and published version, but between other, earlier, versions of an article too, perhaps those first exposed on blogs or wikis. This will permit better formalisation and monitoring of the scholarly record, especially as authors move to early-use of repositories and informal web tools as part of the communications process.

Note: thanks to the authors of the two studies cited above for self-archiving them somewhere so that I could read them in full.