Free Science Blog

It is often observed that there is a danger that doctors may be corrupted by the pharmaceutical industry – the interest in selling drugs may not always be in harmony with the interest in curing patients, and thus commercial interests may get in the way of efficient medicine. Is there an analogous danger for pure science, e.g. for the field of linguistics? Which commercial companies do we work with closely, perhaps so closely that they may influence our decisions?

At the MPI for Evolutionary Anthropology, there was a rumor that some of the nice extra features of the building (such as a sauna and a climbing wall) were paid for by the companies that sold the expensive DNA sequencing machines to the scientists. We poor linguists do not spend so much on our machinery, but we often buy books (or at least our libraries do). So not surprisingly the commercial publishers are trying to treat us well. At the linguistics conferences that I attend, it is normal to find conference bags, name tags, pens and writing pads with the names of publishers on them (Benjamins, De Gruyter, Brill), and sometimes they even sponsor entire receptions. This is nice – one gets a small glimpse of the treats that the doctors (supposedly) get from the pharmaceutical industry.

But is there an analogous danger of undue influence of publishers on our science? I think so: There is one particular development over the last 15 years that finds its primary explanation in publishers’ interests, namely the enormous expansion of handbook publications. In the 20th century, De Gruyter’s HSK series had little competition, but now there is the Oxford Handbooks in Linguistics series, the Cambridge Handbooks in Language and Linguistics, the Routledge Handbooks in Linguistics, as well as the Brill Handbooks in Linguistics. To feed this suite of handbook suites, a large proportion of the active community of linguistics researchers must be engaged in summarizing and re-summarizing research results, rather than in producing new research results.

Clearly, this development is in the publishers’ interest, because while monographs are hardly profitable anymore, handbooks still sell fairly well. Their content may not be original, but it looks so general and relevant to librarians that they are likely to order them, even if their budget does not allow them big purchases. So what do publishers do to publish a handbook? It’s really very simple: Identify a well-established scholar with a track-record of publication and perhaps some editing, and invite them to edit a handbook on some topic (of their choice – this may be a very narrow topic, such as the syllable, or tense and aspect). Being entrusted with the task of editing a handbook seems prestigious (I did it myself some time ago), and if you are a social type, also seems an interesting social activity. Then if one is invited to write an article on a special topic, one feels flattered, because it confirms one’s status as the world’s leading expert on the topic, so one accepts. Writing this article then may get a bit tedious, but in the end one does it because one does not want to hold up the publication of the article or lose one’s face before the senior colleague.

Thus, the system works well, but is this what we as scientists actually want? Did we ask for someone to make these handbooks, because we need them for our research or teaching? It seems that nobody ever asks this question, because the division of labour seems clear: Publishers publish, and scholars do research and write. But if the publishers no longer publish what we submit, but make specific proposals about what kinds of contents we should produce, we are allowing the cart to be put before the horse. We are allowing commercial interests to set our agenda.

How did it get to be like this? In the 20th century, there were not many publishers, and scholars did not have a lot of pressure to publish. At that time, a publisher who approached scholars with a certain publication idea was able to make a useful contribution, because scholars may not have been sure who to turn to for publication. But as as publication options multiplied over the last 20 years, and the internet (and more and more conferences) made it much easier to communicate, authors became a scarce commodity. While finding a publisher was sometimes difficult 30 years ago, publishers now have to make efforts to find authors. Book publication is hardly selective anymore – publisher’s representatives are actively looking for junior authors to submit their dissertations in order to acquire more books. But once a scholar has a teaching job and little time for big new research efforts, handbook articles is something that publishers can still get out of them.

So the flood of handbooks is benefitting the commercial publishers. Is it also benefitting science? Maybe, because overview articles are useful for students, and the handbooks may make it easier to get a basic understanding of an adjacent (sub)discipline if one wants to. But because of the reader-pays model that the handbooks employ, most readers of these handbooks actually read pre-publication versions on social repositories (Academia.edu, ResearchGate), or copies of unclear legal status. The publication system thus helps keep the commercial publishers alive, but it does not really serve the scholars’ interests.

Scholar-owned publication of reference works would take the form of scholars getting together (perhaps at the LSA or the SLE) and discussing what the needs of the field are. The commercial publishers should be our service providers, and should not set our agendas.

Most linguists know that the way we publish our papers could be more efficient (and this of course extends to other disciplines). Many of us could tell stories about how long we waited for a paper in an edited volume to come out (just last week saw the appearance of a book with a paper of mine that was presented at a conference in the spring of 2008 and was hardly changed in the meantime).

What can we do to improve the situation? In this post, I have two concrete suggestions which will surprise some of my colleagues, but which I think could make a real difference. In a wider perspective, these suggestions are not radical at all, because some people are now suggesting that the very principle of selective publication should be abandoned. I think that one can make publication significantly more efficient without such a radical (and in my view undesirable) step.

Proposal (1): Enforce deadlines for reviewers very strictly, e.g. a deadline (for linguistics papers) of six weeks. How can this be achieved? Here’s my proposal: If the editor does not get two reviews within six weeks, the paper counts as rejected. This may sound unfair – after all, it’s not the author’s fault that the editor didn’t find at least two reviewers who were able to send their report within six weeks.

But quick rejection is not about fairness – it’s about efficiency. Waiting for months and months for tardy reviewers is hopelessly inefficient, and thus unfair to the author (and to the readers, thus to the entire discipline). If the author gets a quick rejection plus the admission that two reviewers were not found, then she can submit the paper to a different journal and recommend to her colleagues that they should avoid this journal in the future. Thus, editors will make great efforts to find at least two reviewers within the specified time frame, because they want to continue to publish the best research.

Proposal (2): After the specified time frame (as mentioned, I’d suggest six week for linguistics), deliver an accept or reject verdict. Revise-and-resubmit is inefficient. It creates a lot of extra work for authors and editors. It is the journal editor’s task to select the best papers from among those that were submitted. Revise-and-resubmit makes sense only on the view that it’s the editor’s and reviewers’ task to help the author “improve” the paper. But are they omniscient? All too often, authors modify their paper to please the editor and the reviewers, and not because they believe that the change constitutes a true improvement. But changing one’s paper to please two or three colleagues is not only humiliating, it is also inefficient, because nobody knows how representative those reviewers are. Comments from colleagues are often useful, but I don’t see how it should help science if one of my peers forces me to make a change to my paper that I don’t see as an improvement. Moreover, I suspect that revise-and-resubmit is often chosen not so much because the paper needs to be improved in a specific way, but because the editor finds it difficult to reject the paper, hoping that the author will not resubmit it. And indeed, authors often wonder whether it makes sense to invest the energy to change the paper to fulfill the editor’s demands. But this serves no good purpose. A quick rejection is better for the authors, because they can resubmit the paper to a different journal, making only those suggested changes that they see as real improvements. And if the reviewing time is not too long, then this is not a major problem.

A third way to make our publication system more efficient would be to avoid edited volumes. These often take a long time because the publication has to wait for the slowest author (or the slowest reviewer, if strict reviewing deadlines are not enforced). However, edited volumes are quite deeply entrenched in the practice of linguists (at least in Europe), and one can argue that a lot of interesting work would not be published if authors had to find a journal that accepts it. But edited volumes can be made more efficient as well, by accepting the fact that “peer selection” has a different meaning in edited volumes, so that external “peer review” becomes unnecessary.

Finally, we can make publication more efficient by adopting a uniform text-structure style for the entire discipline. Different style rules for different journals are inefficient. Here’s a proposal for a set of Generic Style Rules for Linguistics. If these were adopted widely (also in M.A. theses and dissertations, i.e. learned by students), then a lot of boring work would simply disappear.

The following is an interview with John Benjamins Publishing Company (JB) concerning the future of linguistics publishing. Martin Haspelmath thanks Anke de Looper and Kees Vaes for contributing.

Question: Do you agree that the increasing trend toward self-archiving, whether on Academia.edu/ResearchGate or on institutional repositories, means that the status quo is unlikely to persist for long? Is your current business model affected if the trend gets stronger?

JB: There is concern among publishers, including us, that this “green open access” – self-archiving of earlier versions of articles – may not be sustainable in the long run, because it still depends on the publishers recovering their costs by selling subscriptions to libraries. As long as there is a need for an official “version of record” in the academic process, it would continue to be important for libraries to subscribe to a journal even if the pre-final versions are accessible for free. But this may change. It may also be the case that libraries no longer subscribe to a whole journal (volume), but rather set aside budget to buy single articles if and when their researchers need them. Those changes would have a considerable impact on publishers’ earnings, but as long as we’re able to recover the costs we make in some way – which means that somewhere, someone has to pay us to enable access to content – we should be able to adapt and stay in business.

Question: Does Benjamins currently offer the option of open-access publication when the costs are covered by author-side fees? For journals and/or books?

JB: We are not advertising this option widely yet, and so far the demand has been limited, but we have published one Open Access book and another to become available very soon. Depending on the subject and the type of book, we decide whether or not we will also do a print edition for purchase.

For journals, we have an author-pays option. Here also, the demand has been limited so far – most of the requests we have had, were driven by the RCUK policy and the block grants that the UK government made available for that purpose.

So all of our journals are, effectively, hybrid; we do not see how we could make them full Open Access (OA) at this point in time, as most authors in linguistics have very little funding and would find it difficult to pay publication fees.

The “hybrid” situation is not very satisfactory, though. It increases our costs because we have to maintain two payment models and have additional administration in the production process. And libraries are – rightfully – concerned about “double-dipping” by publishers charging both author fees and subscription fees for the same content. We’re still struggling with the best solution to prevent that: with the low volume of OA articles, it is difficult to decrease the subscription rates, because in one year a journal may have one article – say, 10% of the total – with an author fee, and then the next year, it may be down to zero OA again, and libraries would not like prices to go down and up again that significantly from one year to the next. Another option we’ve seen publishers use, and one that we may consider, is to give a discount on the author fee if they are at a subscribing institution, or the other way around, reduce the subscription rate for a paying authors institution.

Question: Are there plans to expand author-fee-based open-access publication in the near future?

JB: As mentioned above, we offer the possibility of publishing author-fee-based open-access books and journal articles in all of our journals. Our policy at this moment is mainly demand-driven.

Question: Some people fear that open-access fees will correlate much more with the prestige of a publication brand rather than with the efficiency of the publisher’s operations. What is your response to this?

JB: At John Benjamins we now have one rate for all our journals. I don’t think we would differentiate by “prestige” – if we’d even figure out how to do that across the various sub-disciplines of linguistics. What we might consider are introductory discounts on fees for new journals.

In general, it is conceivable that journals with high rejection rates – usually journals with a lot of prestige – have higher costs, because of the volume of articles dealt with, and need to ask higher author fees for that reason.

How author fees will develop in the future remains to be seen. At John Benjamins, we do not think that there is a gold-mine anywhere that allows us to ask whatever we want in order to make a bigger profit; we only hope that there will not be a gap between what authors can afford to pay, and what we need to earn in order to cover our costs and keep (a specific journal) going as efficiently as possible.

From the perspective of science funders (such as governments and charities), why would you argue that scientists should publish with Benjamins or other commercial publishers? What are the advantages over the alternative of subsidizing scholar-owned publishers?

JB: Publishing takes work, and in our perception, the most efficient system is one where scientists can spend their time doing research, and outsource the work involved with the publishing to people who specialize in that. It is likely that there will be a shift in how this outsourced work is paid for: from a situation where publishers had to make sure they recovered any costs afterwards by selling enough subscriptions/copies, to a situation where – in addition or solely – authors (or their institutions or funders) pay for that work through up-front publication fees. Of course the authors/institutions/funders will look for the “best buy”, taking into account the journal(s) that they would want to publish in because that is where the colleagues in their field look for information, where they get the best exposure, the most prestige, and the best assistance in publishing their paper.

But the playing field should be level: as a commercial company, if we want to set up a server or a platform to distribute e-books, we need to pay all of those costs out of our own revenues, and if we need to take out a loan from the bank in order to do this, they want every cent back with a healthy interest; that makes it hard to compete with “not-for-profit” businesses who have received ‘free’ money from funders or governments to set up their business and as a result already start out with lower costs.

The “mega-journal” trend, which has arrived in the humanities (including linguistics) may turn out to seriously disadvantage junior researchers, independent researchers, and researchers from low-income countries. This is not good for science.

In the 20th century, scientific publication served two purposes simultaneously: dissemination and reputation-building. The first is important for scientific progress, the second is important for the careers of scientists. Both are needed by the system, and they happily coexisted because there was no conflict.

Publishers needed paying customers, so they adopted restrictive, exclusive, selective policies: Only the best works were published in their journals and book imprints. High-profile publishers were able to sell more copies and charge less for their books (and journals), so authors had a double incentive to aim for a good brand.

In the 21st century, two things happened that destroyed this system: (1) Globalization (and the expansion of science budgets) turned scientific publication from a niche occupation into a global market with a few major international players, and (2) the internet decoupled dissemination from reputation-building. The first meant that publishers tried to extract the maximum amount of money from science budget, and the libraries revolted (leading to pro-Open Access declarations). The internet allows anyone to put their papers on the web, either on their personal web page, or using “blue open access” (Academia.edu, ResearchGate). Dissemination is now effectively possible at zero cost.

But what about reputation-building? Most open-access advocates seem to have forgotten about this aspect of (20th-century style) scientific publication. But can one simply transform high-profile journals into author-pays journals? Will a high-profile publisher like Oxford University Press charge authors for publishing freely available books, and everything else will remain the same?

I think that the answer is no. If the author pays for publication, there is no particular reason why publishers should adopt an exclusive, selective approach. The more papers and books they publish, the greater their income. The number of readers is irrelevant for publishers, so they should attract authors in great numbers. And indeed, there is an increasing tendency to abandon selectiveness, and to publish whatever is submitted – this trend goes by the name of “mega-journals”.

The best-known mega-journal is PLOS ONE, which publishes tens of thousands of articles each year (and charges USD 1350 per article). But more and more such journals are springing up, also in areas relevant to linguistics:

Brill Open Humanities (Brill)
Modern Languages Open (Liverpool University Press, GBP 500)
Open Linguistics (De Gruyter Open)
Open Library of the Humanities (not yet launched)
Frontiers in (50+ journals, e.g. Frontiers in Psychology with linguistics papers, EUR 575-2000)

The journal websites do not always say explicitly that they have abandoned selectiveness, but in the absence of a limit on the number of papers they will publish, they are clearly moving in the direction of “significance-neutral peer review”.

With the established mega-journals, the idea that evaluation is not for significance (or “impact”) but merely for “soundness” is an explicit policy. PLOS ONE says on its website that it does not have “subjective” acceptance criteria, and advocates of mega-journals argue that this is good, because it also allows the publication of negative results.

There seems to be a strong trend in many fields of science toward mega-journals (see this article by Peter Binfield, a former editor of PLOS ONE). Suppose that it also takes off in linguistics, the field where I work – what would it mean?

It would mean that linguists from low-income countries or early-career researchers would no longer have good access to mainstream publication modes, because they do not have the funds to pay the publication fees. The rich would get richer. But of course, if a journal publishes without any selectiveness, then publishing in it would not be a sign of scientific excellence (only of “soundness”). So how can scientific excellence be measured? Advocates of mega-journals commonly cite Article-Level Metrics (ALM) or “altmetrics”, i.e. social-network impact.

Yes, it is indeed possible to imagine that we may no longer associate the places of publication with any kind of reputation in the future. “Blue open access” is not restricted, and it measures a scholar’s impact. My ResearchGate profile tells everyone that my RG score is 16.47. But I have two questions:

(1) Why do we need specific mega-journals if we are going to be evaluated by article-level metrics anyway? Isn’t it sufficient to upload everything to Academia.edu and count the views and downloads? Does “peer-review for soundness” really matter that much?

(2) If reputation is based on article-level metrics, then what will scholars do in order to enhance their reputation?

I have no answer to the first question, but I can imagine some creative solutions in the second case. For instance, if an altmetrics system counts the tweets in which a paper is mentioned, then I might set up a couple of new Twitter accounts, or order the tweets from a specialized company (as is well-known, it’s relatively cheap to buy a couple of hundred Facebook “likes” for your company’s Facebook page).

More seriously: People attach reputation to labels, i.e. names – that’s an anthropological fact that is not going to change. Just as nowadays a paper published in a high-profile journal will get more tweets, the number of tweets will in the future be associated not only with a paper’s content, but with names. Hence, if journal names are no longer associated with reputation because one can buy oneself into a journal, other names will take up that space. Reputation may instead be attached to names of universities, or names of individuals. If you’re at a prestigious university, your chances of being read will increase greatly, and if you’re at a low-prestige university, nobody will read your papers. If you already have an established name, you will have many readers, regardless of how good it is (it will take readers a while to realize that a senior author has become lazy and no longer produces excellent work). By contrast, the work of junior scholars will hardly be read. At present, linguists do not put their names on papers written by their students, but in the future, students may be begging their renowned professors to become their coauthors. Even the names of cities and countries may become much more relevant than they are now, because they are associated with the publication.

Thus, if we give up the selectiveness/exclusiveness that is associated with traditional publication models, a completely new dynamic will probably unfold. Mega-journals are typically praised by open-access advocates, and one can indeed hope that they will bring the costs of the system down. But they may also damage science, because they remove a key ingredient in the old system, which has to be replaced by something else. Exclusive journal and book publication creates a kind of level field among all researchers – senior researchers and researchers from high-profile institutions need work hard to get into the good labels. (Thus, I’m not so sure that the current system is “a grading method from hell“, as Colin Phillips has recently called it.)

Thus, before we all jump happily on the new bandwagon, we should think hard about the possible consequences. My own favourite model is still the publisher-pays model, where universities pay for journals because they want to profit from the prestige generated by the journal. Such publisher-pays journals (and book imprints) would still be exclusive, but they would be open-access at the same time. (And in fact, most open-access journals are publisher-pays journals, as Stuart Shieber has observed.)