Fairly often, I run into the claim that adjectives and especially adverbs should be avoided in scientific writing. I’ve been told, for example, that using “Surprisingly” (adverb) or “This important result” (adjective) is an attempt to manipulate the reader’s opinion about the data, and that, in scientific writing, the data ought to speak for themselves. I understand the thinking that leads to this belief – but I think it’s naïve. A new study by Ju Wen and Lei Lei, on adjective and adverb usage in scientific Abstracts, gives us an interesting look at the practice, and at arguments for and against. So despite that study’s limitations, let’s dig in a little.
Wen and Lei open their paper with a claim that I hope is uncontroversial: “Writing in clear and simple language is critical for effective scientific communication”* (all quotes are from the main text of Wen & Lei 2022). This is such an important proposition that most of The Scientist’s Guide to Writing deals with it! But Wen & Lei pivot quickly to the proposition that “clear and simple” means avoiding adjectives and adverbs:
“One feasible method to achieve that goal is to reduce the use of adjectives and adverbs because they are often redundant and meaningless and therefore make the writing cluttered and less comprehensible.”
Hmmm. I think we need to know more.
On to the data, and what it might say about how terrible adjectives and adverbs might be. Wen & Lei worked with a corpus of 700,000 life-sciences Abstracts from 1969-2019, using automated tools to count adjectives and adverbs, to separate those into “emotion” words (like “surprising”) and “nonemotion” words (like “purple”), and to score each Abstract for readability. Over the 50 years of Abstracts they found increasing frequencies of adjectives and adverbs and decreasing readability scores (and they aren’t shy about connecting those two trends).
Case closed, right? Well, no. The connection between adjectives/adverbs and readability is tenuous, in part because the former don’t start increasing much until about 2000, while readability decreases steadily over the study period. And Wen & Lei don’t make any attempt to consider other changes in our literature over the same time period. The remarkable growth in our use of acronyms, for example, seems like one strong contender as a driver of declining readability. If adverbs are ruining our scientific literature, I don’t think we have any evidence that it’s because they make papers harder to read.
Wen & Lei turn next to the possibility that scientific writers use adjectives and adverbs to manipulate readers. They note that much of the increase in frequency of adjectives and adverbs is driven by increases in “emotion” words, and they suggest that “authors may resort to rhetorical devices such as linguistic positivity to get their papers published in academic journals”.** This is, I think, the crux of the objection for a lot of people: the idea that writers use hyperbole to exaggerate the strength of patterns in their data and the novelty and importance of their work. Adjectives and adverbs like “surprising” and “importantly” (both “emotion words”) can certainly be used that way – for example, when authors claim their results are “surprising”, but that’s true only in comparison to a straw man they’ve set up.
But are “emotion” words bad in scientific writing? Those who like to pretend that science (and scientific writing) can or should be entirely objective, divorced from interpretation or passion, are likely to bristle at the very term “emotion words”. These are the folks most likely to say “the data speak for themselves”. (They don’t, of course – if they did, we’d simply publish tables of data rather than papers with Introductions and Discussions.) Things don’t work that way. Instead, scientific data are complex and need interpretation, and they exist in the context of readers’ expectations and beliefs.
Words like “surprising” help frame data, and in doing so they can accomplish either of two things. They can mark the authors’ position (implicitly, the authors are saying we were surprised because); and they can help the reader find their own position, interpreting data and relating it to context (implicitly, the authors are saying you might be surprised because).*** Both of these are helpful things for authors to do; in fact, for thousands of years we’ve understood that they’re central tasks of the art of rhetoric. When I write a paper, I want to communicate more than just the data: I want to communicate what I think the data mean, and what I think the reader should conclude from it. I don’t think that’s manipulating a reader, although I’m happy to concede that I’d like to shape the reader’s response. What separates scientific writing from advertising copy, for me, isn’t that we abstain from attempts at persuasion. It’s that we try to persuade only by saying things that are true (or at least plausibly true!), and that we have enough respect for our audience to presume that they may not actually take the position we’re trying to help them reach.
There’s another element to the pro-adverb case. Our scientific literature has a reputation, richly deserved, as being turgid and tedious. Anything we can do to make our scientific papers more engaging is, I would argue, a Good Thing™. So, for example, I appreciate attempts to inject some humour and beauty into scientific writing; and my colleagues and I have recently shown that humour in titles can increase impact of scientific papers. The judicious use of adjectives and adverbs can engage a reader in many ways. It might just be a matter of dressing up the prose a little, to avoid colourless, Hemingwayesque sentences. Or it might be that admitting to the possibility of surprise (for example) can humanize both authors and readers. Unfortunately, it’s hard to automate the study of engagingness, so any discussion of it necessarily has to lean hard on opinion.
The data are clear: our use of adjectives and adverbs in scientific writing has increased. Is that bad or good? That’s the hard part, and reasonable folk can differ here – especially when the data are so limited. I’m not convinced by Wen & Lei’s case against, and in my own writing I’ll continue to communicate my surprise (when that’s true) and my work’s importance (when I think that’s true). What about you? Please use the Replies.
© Stephen Heard September 13, 2022
Hat tip to Jeremy Fox for calling my attention to Wen & Lei’s paper, and for stimulating me to think about it.
*^This clause has 12 words, and 5 of them (42%) are adjectives. This is an interesting statistic, given Wen & Lei’s argument later in the same paragraph that adjectives and adverbs are redundant and meaningless, and that they clutter scientific writing and make it less readable. I’ll get to that.
**^Anyone else notice that this claim flatly contradicts their earlier one that adjective and adverbs are “redundant and meaningless”? Wen & Lei don’t try to resolve this contradiction. In fact, they don’t mention it at all.
***^To their credit, Wen & Lei understand this, offering in their Discussion “Hence, the use of adjectives and adverbs helps writers make compelling arguments and helps readers remember key points in the full text of an article”. It’s just odd that they don’t do much to integrate this pro-adverb point with their other, anti-adverb, arguments. A nuanced argument should bring up arguments from both sides – but really should do more than just list them.