Jeremy recently posted a question on who are the most stylish writers in ecology. Stylish is good – scientific writing can be beautiful. But, as I mentioned in the comments, my goal is more prosaic. I just want to be a clear writer. My PhD adviser, Mike Rosenzweig, was a leading inspiration for this. I tell every graduate student of mine to read the introduction to his book on species diversity where he expounds on why it is important to be a clear writer and gives a number of quick tips. I can’t quote the whole passage here (it is pages xv-xvi if you have access to a copy), but here are two quotations to start with:
On why to write clearly (and why most scientists don’t): “Here’s another warning. Clear writing brings a grave danger: People may begin to understand you! Then they will probably disagree with you.”
And on what it takes to write clearly: “But be warned. Writing more clearly takes hard work. The more effortless it seems, the more effort it took. It all depends on whether you have something to say. If you do, you’ll care to work hard to get it across.”
Writing a piece about how to write well is one of those really hubristic, setting yourself up for failure endeavors. So let me be clear. I don’t in anyway hold myself up as a writing expert nor a great writer, although I would like to think I am (and based on feedback I probably am) at least a clearer than average writer. I expect half the readers are better writers than I, so I offer these thoughts in the spirit of use what you like, ignore the rest, and please add your own thoughts in the comments. (Plus I’m feeling guilty on how long its been since my last post, and Jeremy’s post brought this topic to mind so I’m going to go with it)
Here are some of my favorite tips for clear scientific writing:
1) Audience, audience, audience. You may be sole author, but writing is still a dialogue – between you and the reader. There is no such thing as writing that is most clear in an absolute sense. Writing is only more or less clear in the context of a particular audience. Think about your audience before you start writing. One audience’s dense jargon is another audience’s shorthand. Regularly check in on your writing and whether you think it is reaching that audience.
2) Much of writing is convention. This follows immediately from point #1. Introduction/methods/results/discussion is not the only way nor even necessarily the best way to write a paper. But it is what a reader expects. So fulfill their expectation. They’ll spend less time trying to figure out the overall flow of the paper and more time trying to understand the details of what you’re saying. Points #3, #4, #5 and #6 are at least in part about conventions. An example of a convention that tripped me up for my first five papers – the methods are in the past tense, the discussion is in the present. The introduction is mixed depending on whether describing previous work or what we do and do not currently know. Partly this is logical, but partly it is just a convention and I had to find it out from reviewers.
3) Know how to emphasize. There are numerous rhetorical devices for emphasizing a point. Repetition is a big one – when writing papers it is tempting to think you should not be repetitious. Indeed it is a fatal flaw to repeat unimportant material. But repeating your points of emphasis is your only hope to get somebody to actually remember them. Position is also important – the beginning and end of anything are the most emphasized parts. This is true for a paper, for a section of a paper or for a sentence. For example, one should always have a paragraph in the discussion mentioning the limitations you know about your work (better you get them out than the reviewers think you are ignorant of them), but this paragraph should never be the first or last paragraph of the discussion section unless what you really want the reader to remember is the shortcomings of your work. Same thing holds true for a sentence. “While travelling by train, I saw a purple unicorn standing in a meadow chewing her cud.” Rather buries the main point of the sentence, doesn’t it? Purple unicorn is so novel it tries to swim to the top, but it is ultimately buried by the preferential treatment given to travelling and chewing. Breaking a grammatical rule also calls attention. One of my favorite rhetorical techniques (to the point of overuse by me), is to start a sentence with “But”. My 3rd grade teacher, Mrs Adlof, told me never to do this. But it really grabs your attention. In the quote above from Mike he starts a whole paragraph with “But”. This was honest signalling because it was his most important point on the whole page. All of this, of course, assumes you know what you want to emphasize. Emphasizing random points is like listening to a person who doesn’t modulate their voice to help you know when to pay attention. If you don’t know definitively what you want to emphasize, then you shouldn’t be writing yet. Your time is better spent thinking and talking to people about what your main points are.
4) Be precise. Don’t say “productivity varied with temperature”. Say “as temperature goes up, productivity goes up”. Don’t say “we obtained samples” say “we captured birds in a mist net and drew 1 ml of blood”. And – one of my Achilles heels in writing – don’t overuse pronouns. “It”, “they” and “this” add very little value and should be used only when the antecedent is unmistakably clear. “This” at least allows the opportunity (which should be used) for some clarification like “this calculation” or “this result”.
5) Formatting is your friend. Modern writers have the benefit of easy access to many formatting conveniences. Journals require headings, but one should almost always use subheadings (and maybe even sub-sub-headings). Similarly, a good numbered or bulleted list is stylistically very inelegant but rates very high on the clarity scale. It is the ultimate tool for parallel construction. Bold and italics can be useful as long as they aren’t overused.
6) The battle for good writing is won sentence by sentence. A good sentence is: short, has the subject and verb together, has an active verb, has the points of emphasis at the beginning and end, and moves the reader along from a familiar launch point at the start to the new information at the end.
For understanding how to build a good sentence, I find a website on scientific writing at Duke to be very useful. The example below started from one of their examples. Compare the following 6 sentences:
a) The data was analyzed using multivariate statistics.
b) The data was analyzed using multivariate statistics by us.
c) We performed a multivariate statistical analysis on the data.
d) We analyzed the data using multivariate statistics.
e) We calculated a principle component analysis [on all seven of our variables].
f) We regressed productivity against temperature.
I hope you can see that there is a general progression towards better sentences. Specific comments follow.
a) Sentence a has no subject. Who analyzed the data?! Inquiring minds want to know. Seriously, our brains are hardwired to ask who did what from every sentence. It is jarring not to answer.
b) Sentence b has a subject but it is miles away from the verb. Sentence b (and a) are passive constructs. Experts tell you not to use passive constructs. (cf. with how I originally wrote this sentence – “You are often told not to use passive constructs in good writing”- it would have been ironic if it was intentional). This is not because passive voice is inherently evil (it has its places), but because the passive usually leads to other evils like separating subject and verb or using weak verbs. By the way, using the verb “to be” (is are were was am been) is not the same as a passive construction, but using “to be” as your main verb is also a warning sign of a weak sentence.
c) Sentence c pushes the action (analyzing) into a noun and uses a weak, vague verb (performed – this could mean anything from running a centrifuge to dancing on stage to in this case clicking on buttons in SAS – blech – terrible verb). Sentence c also has a multiple-noun/adjective collision: “multivariate statistical analysis”. Putting 3 nouns and adjectives together is barely legal if used infrequently. Using four or more in a row should get you arrested.
d) Sentence d is getting there. It now puts the continuity (the “we” who has presumably been mentioned in the 5 preceding sentences) at the start of the sentence, while also putting the novel information (multivariate stats) in the position of emphasis and novelty (the end of the sentence).
e) Probably sentence e is even more impactful and informative (analyze is better than perform but analyze is still pretty darn weak, calculate is stronger). Sentence e is probably about the best you can do for a multivariate statistical analysis (you can’t turn principle component analysis into a verb). Depending on context the phrase in brackets might be informative or excessively wordy. Or the phrase in brackets could be moved into the middle of sentence if it should be deemphasized (“We input seven variables into a principle component analysis”), or to the beginning of the sentence if it bridges to what went before (e.g. “Using all seven variables, we calculated a principle component analysis” when the previous 3 sentences were about the variables).
f) Notice that in a slightly different context we can still do much better than sentence e. In sentence f, we have a very specific verb (regressed) and we replace the annoyingly vague phrase data (or variables) with our specific instances.
The one thing I haven’t talked about here in sentence construction is brevity. Like all rules it is not absolute. I’ve written some really good sentences that are 25-40 words long. But most of my really good sentences are in the 8-15 word range. This is another of my Achilles heels in writing. I love the good parenthetical phrase, the nuance, the sentence complexity that reflects the complexity of the real world. But I fight it on a regular basis. Subordinate clauses, adjectives and adverbs, parenthetical expressions and prepositional phrases should all be examined with severe scrutiny to see if they justify their existence. And if they do justify their existence, do they deserve a sentence of their own? There are many phrases that sound erudite and thus are often used in scientific writing but are really quite vacuous and wordy. Lesson 3 (on simplicity) in the aforementioned link to the Duke Science Writing website has excellent material on vacuous phrases.
This is a lot of work to rewrite a long sentence into a short one and then further work to rewrite with the issues highlighted in sentences a-f above in mind! Do you have to do this for every single sentence for every single paper? No. It does, however, mean that at points in your life when you are actively working on improving your writing, you need to go through your paper sentence by sentence, thinking about how you can improve it. Its probably also a good idea to do this for the abstract and conclusion of every paper. I have a Word macro that goes through and highlights sentences in yellow for longish and red for really long sentences, and a shadow on sentences that use forms of the verb “to be”. This quickly helps draw my attention to potential problems. Now, I don’t slavishly try to get my writing so it gets no flags. Robotic writing is not good writing either. But I find it a very helpful way to call my attention to the sentences I need to think about. I did this a lot when I was a graduate student writing my first few papers. Hopefully some of the lessons learned sunk in and carried forward, because I don’t have time to do this for every paper I write today (and certainly not every blog post or comment!). But I am going to start writing a book shortly, and I expect to spend time doing this again. Everybody needs to return to this level of attention once in a while.
My bottom line is this. The goal of writing a paper is to communicate. Communication involves things like brevity, following convention, good sentence structure, emphasizing important points, etc. In short it means saying what you’re trying to say as clearly as possible. Don’t be beguiled by the apparent erudition of complex, meandering phrases into being what is ultimately mealy-mouthed and unclear.
So, I have now pontificated on a subject on which I am not an expert and only a middling practitioner. On this blog alone, I am easily only the 4th best writer :-). I can write this only because these ideas are not particularly novel. One can find them in most good books on writing. Aside from the Duke Science Writing website I linked to above, and pages xv-xvi of Rosenzweig’s book, the paper by Gopen and Swan (1990 -The Science of Scientific Writing – just google it you’ll find PDFs everywhere) is also good.
Note that I’ve stayed away from the larger structural issues (i.e. how to organize a paper, etc.). Maybe I will come back to that someday. For now, I look forward to hearing others’ thoughts on the nuts and bolts of what makes clear writing.
For those who are curious here is a link to the Word macro I mentioned. I give it out AS IS. It should work in anything from Word 1997 to Word 2010 and on Mac and PC. It is written in VBA which is not officially supported since Word 2003, but it worked fine in Word 2010 for me today. Unfortunately, I cannot provide technical support on how to get it working on all these different platforms, but it should basically look like:
1) Download the file (you will probably get warnings as Word macros often carry viruses – you’ll just have to accept my word this is clean and click OK).
2) Open the file (WritingMacros.dot). In Word 2010 it automatically disables macros unless you click on the yellow bar at the top to enable them.
3) Open the file (word document) you want to analyze and make it the front/active document.
4) Run the macro VerboseCheck (how to do this is the part that changes a lot between versions of Word). My macro will flag sentences longer than 20, 30 and 40 words by colors (green, yellow, red if I remember). You can (and I sometimes do) change the lower limit to 15 words. It will also put any sentence with forms of the verb “to be” in shadow font.
5) Edit your document correcting hopefully most but not all flagged sentences
6) Run the macro UndoVerboseCheck to remove all the weird formatting (note that I intentionally used odd formatting like text color and shadowing so it doesn’t overwrite all of your own formatting).
People who figure out how to run this macro on their particular operating system/word version are welcome to post more detailed instructions here. And if you’re really stuck, you can post a question and see if somebody other than I can help you.