Image credit: “Piled Higher and Deeper” by Jorge Chan www.phdcomics.com
As scientists, we spend a remarkably large fraction of our time writing. An obvious question immediately arises: do we apply science to our scientific writing? That is, can we (and do we) apply our scientific ways of thinking to make our scientific writing better? The question, as I say, is obvious; but the answer is not.
There are actually three different things I might mean by “is there science in scientific writing”, corresponding to three different meanings of phrase “scientific writing”. We might ask our obvious question of the product of our writing: the style and content of the text itself. We might ask it of the cultural practices that scientific writing represents: for instance, our conventions for written form (IMRaD and all that) or for review and publication. And finally, we might ask it of the process by which we write, with the focus on the behaviour of the writer producing (or struggling to produce) text.
(1) “Scientific writing” the product. Papers (and other scientific texts) vary in content and writing style, and we might wonder whether this variation might predict a paper’s success. By “success” I might mean reader comprehension (for instance, what kinds of sentences are most easily understood by readers with a given level of scientific training?), or I might mean impact (for instance, is there a correlation between a paper’s syntactic complexity and its citation rate?).
We know quite a bit about comprehension issues from studies in linguistics and cognitive psychology. There are data, for example, on depth and speed of comprehension for texts of different construction. We know of factors that improve clarity universally and factors that depend on the reader’s native language (for instance, English speakers prefer the familiar subject-verb-object order, but other languages use other orderings and their speakers have corresponding preferences). Little of this, though, is specific to scientific writing, and so if there are ways to make scientific writing in particular clearer to scientific readers in particular I’m not sure we know about them. This seems a bit odd, given that a paper titled “The Science of Scientific Writing” (Gopen and Swan 1990) is well known and widely read. But while that paper dispenses what’s surely good advice, none of it is particular to scientific writing, and if it’s supported by data and analysis the reader isn’t told (no experiments are reported and no scientific literature is cited)*.
In contrast, we know very little about impact issues. Despite the fetishization of citation metrics, how writing (as opposed to content or journal placement) shapes readership and citation rates seems virtually unexplored. What little has been done lies mostly in the domain of science studies and genre analysis (see, for instance Gross et al. 2002 or Swales 1990), and this literature is poorly known by working scientists. The only study I can think of that’s had some influence on the way we write is Sagi and Yechiam (2008), which claims to demonstrate that papers with humorous titles are cited less often. Despite the weakness of its analysis, this study is used by at least one major journal in ecology and evolution to discourage authors’ attempts at levity. (I’ve discussed the possible function of humour and beauty in scientific writing in this paper, but I haven’t done proper scientific analysis either.)
(2) “Scientific writing” the cultural practice. We’re all pretty familiar with the way we structure scientific papers and with our systems for reviewing and publishing them. What’s perhaps surprising is that many scientists aren’t very familiar with the history of these practices – the reasons we’ve come to do things the way we do – and I would claim that we don’t know much, empirically, about whether our actual practices outperform alternatives. For an example to do with history, consider my recent post about the history of the Methods section. I wrote about the reasons that we seldom provide enough detail to allow exact replication of our experiments, and I argued that knowing something of that history gives us useful perspective on what today’s so-called “reproducibility crisis”. But questions about our cultural practices could go far beyond that. Empirically, does frequent replication of studies accelerate a field’s progress (by detecting error and fraud faster) or retard it (by diverting resources from original work)? Does online commenting on published papers increase or decrease their influence? Do granting agencies’ requirements for open-access publication increase or decrease researchers’ impacts? While these questions are all open to empirical investigation, most of the discussion I’ve seen of them is philosophical (sometimes distinctly moralistic) instead.
(3) “Scientific writing” the process. It’s this meaning of “scientific writing” that may connect us to the most actual science, because there’s lots of research in psychology and neuroscience that can be applied to a writer’s habits. For example, fMRI studies suggest that resolutions to change behaviour (e.g., to write more) are more easily kept with conscious self-reflection; TMT theory of procrastination suggests techniques for achieving writing discipline; and the phenomenon of context-dependent memory can be exploited (in reverse) to disrupt the writer’s familiarity with their own text during revisions. Granted, little of this knowledge has to do directly with scientific writing, or even with writing at all – most psychologists study tasks that are a lot easier to quantify, or that have broader social implications, than scientific writing. That means it’s an open question whether techniques that work for increasing discipline at exercise (say) also work for increasing discipline at writing. I strongly suspect that they do, but as a scientist, I itch for data to test my suspicion.
I think the process of writing is mysterious to most scientific writers, though. We know we have to do it, and we may have systems in place (like word quotas or writing groups) to help us get it done; but we don’t know the psychology or neuroscience literature and how it can help us**. In other words, we tend not to think about our writing selves as agents that we can study scientifically, and we don’t think about how we can harness that kind of study to manipulate (improve!) our behaviour and thus the quantity and quality of our writing.
So in short: is there science in scientific writing? There could and should be, but there isn’t much, and I don’t think most scientific writers think much about it. This seems like a missed opportunity, and a big one.
© Stephen Heard (firstname.lastname@example.org) March 26 2015
This post is based (very loosely) on material from The Scientist’s Guide to Writing, my guidebook for scientific writers. You can learn more about it here.
*There’s also a book called “The Science of Scientific Writing” (Monroe et al. 1977), which might have something to say on this, although it’s out of print and I’ haven’t yet tracked down a copy.
**An obvious hypothesis is that psychologists and neuroscientists should be more aware of this literature, and should write better or faster as a consequence. This is also an empirical question that science could answer!