Wednesday, April 02, 2014

Open Access Linguistics: You're Doing It Wrong

If you're a linguist - any kind of linguist - then you, like me, will probably have received an email from the Open Journal of Modern Linguistics, inviting you to submit your work.

I'm extremely committed to open access in linguistics, and in academia more broadly; here's why. But OJML is doing it wrong, and the rest of this post aims to explain why. The tl;dr list version of this post is as follows:
  • Don't ever submit your work to OJML.
  • Tell your friends never to submit to OJML.
  • If you know someone who's on the editorial board, gently ask them not to be.
So, what's so very wrong with OJML? The short answer is that it is run by the wrong people and threatens to bring the entire, very promising, open access movement into disrepute by charging stupidly high APCs and skimping on quality both in terms of typesetting and intellectually.

The "costs" of progress: predatory publishers

Let's take a look at OJML's guidelines on Article Processing Charges (APCs). It's $600 per article, but only if that article is within ten printed pages: in linguistics, that's barely out of squib status. For each additional page above ten, an extra $50 is whacked on.

This may not seem like much, given that Elsevier charge up to $5000. But for a 20-page article, which is still short by linguistics standards, we're talking $1100. Moreover, this kind of incremental model penalizes thorough argumentation and, in particular, proper referencing. It might even not be so bad if what you paid for was worth it - but I'll argue below that it isn't even close.

The open access community has a name for this kind of publishing practice: "predatory". Jeffrey Beall maintains a list of predatory publishers on his website, along with criteria for inclusion. Surprise, surprise: "Scientific Research Publishing" (SCIRP), the publishers of OJML, are on the list at number 206.

What's in it for them? Large amounts of money, made from academics' naivety. Last year, journalist John Bohannon conducted a "sting" operation by submitting a series of 304 deliberately deeply flawed manuscripts by fictional authors to gold open access journals, many of them ostensibly peer-reviewed. More than half of them accepted the papers, including many that apparently sent the paper out for review, and 16 journals accepted the papers despite the reviewers spotting their damning flaws.

The journal Science, who hosted Bohannon's piece, were keen to trumpet the failure of open access (unsurprisingly, as they represent the status quo that open access threatens). However, there are a lot of problems with Bohannon's approach, which have been ably summarized elsewhere. In particular, since Bohannon didn't include a "control group" of traditional subscription journals, there's no evidence that open access peer review practices are any worse than those. And even if they were, the existence of exploitative behaviour within open access of course doesn't entail that open access itself is a bad thing. But it's clear from Bohannon's experiences and those of others that, where there are new ways of making shady money, there will be crooks who leap to seize them, and that gold open access (and OJML) simply illustrates one instance of this general principle.

Bad production standards

One of the areas where any publisher can claim to add value is in ensuring the formal quality of their published submissions: typesetting, copy-editing, proofreading, redrawing complex diagrams or illustrations, etc. If a publisher does this well, they may merit at least some of the fees that they typically charge for open access. However, OJML's performance in this area shows that they hardly even look at the papers they publish. Here are some examples from Muriungi, Mutegi & Karuri's 2014 paper on the syntax of wh-questions in Gichuka (which, at 23 pages, must have cost them a pretty penny):
  • Glosses are not aligned (e.g. in (6) on p2).
  • The header refers to the authors, ridiculously, as "M. K. Peter et al".
  • There are clauses which contain clear typographical errors, e.g. "the particle ni which in Bantu, which is referred to as the focus marker", on p3.
  • In (17), the proper name "jakob" is not capitalized.
  • There are spelling errors: "Intermadiate", in table 1, p8.
  • The tree on p14 has been brutally mangled.
  • Some of the references are incomprehensible garbage: "Norberto (2004). Wh-Movement. http.www.quiben.org/wp.content/uploads"
A quick glance through any OJML paper will reveal that these aren't isolated occurrences, and little of this is likely to be the fault of the authors: at least, any linguistically-informed copy-editor or proofreader should have picked up on all of these points instantly, and any proofreader at all should have picked up on most of them.

Low quality papers

What about the academic quality of the papers accepted? I don't want to pick on any particular paper: in fact, I'm sure that there are nuggets of gold in there (the Muriungi et al. paper mentioned above, for instance, is a valuable syntactic description of an aspect of an understudied language). But I invite you to skim some of the papers and draw your own conclusions.

In particular, the dates of acceptance and revision of the papers aren't exactly indicative of a thorough review process. For instance, the paper by Muriungi et al. was "Received 7 June 2013; revised 9 July 2013; accepted 18 July 2013". Again, this isn't unusual for the papers in this journal. It's certainly not impossible for quality peer review to take place at this speed - and it's certainly desirable to move away from the unacceptable slowness of some of the big-name journals - but it is at least doubtful. And one thing that is extremely eerie is how many of the articles are dated as having been revised exactly one month after receipt, suggesting that the process may have been even shorter and that SCIRP is trying to cover itself, by means of outright lies, against exactly the kind of allegation I'm making.

The fields of linguistics given under their Aims & Scope don't inspire confidence, either, with "Cosmic Linguistics" and "Paralinguistics" among them.

Why is this important?

OJML is symptomatic of exactly the wrong approach to open access. Open access, to me, is about disintermediation, about putting power back into the hands of academics. There are several good open access operators out there: Language Science Press is a prime example in the domain of books, the e-journal Semantics and Pragmatics has been performing a valuable no-fees open access service for years, and the Linguistic Society of America recently took a step in the right direction by making papers in its flagship journal Language openly accessible after a one-year embargo period. These initiatives are all run by researchers, for researchers.

In contrast, OJML is about opportunistic money-making. Here's a quote from SCIRP's About page, in relation to why their base of operations is in China while they're registered as a corporation in Delaware: "What SCIRP does is to seize the current global trade possibilities to ensure its legitimate freedom with regard to where to do what." If this sort of creepy graspingness doesn't put you off submitting to OJML, and the problems outlined in the previous sections don't either, then I don't know what will.

Unless we nip this problem in the bud, then it threatens to damage the reputation of the Open Access movement more generally. Time to boycott OJML, and to spread the word.