Well this is extracted from the highly recommended blog of Mike Kaspari so we can call it a steal, but it will be a useful theft anyway.
- Get rid of every adjective modifying a relationship. Was x larger than y? Just say so. Saying it was much larger, or especially tiny, or amazingly huge adds no information.
- Replace long words with short words. Good writing maximizes the content of the message per number of letters used. Replace long words with short words of equal meaning. Replace utilization with use.
- Replace every “differed” or “was different” with the actual, quantitative relationship. Compare the content per letters used for the following two sentences:
Plants fertilized with nitrogen differed in height from controls.
Not only have you conveyed that nitrogen increased growth, you’ve given a vivid word picture as to how much. In fewer words!
Plants fertilized with nitrogen were 2.5 x taller than controls.
- Make sure your Discussion has a caveat paragraph. Every study is flawed or makes simplifying assumptions; every study has a method or result that may be misinterpreted. Grad students often attempt to hide these flaws. But, like an untreated cut, such problems can fester in the mind of a reviewer. Consider inserting a caveat paragraph somewhere in the middle of the Discussion that thoughtfully addresses at least two topics.
The first is a plausible mistake a harried reader might make and why it is incorrect (look for patterns in your friendly reviews to identify likely candidates). Good writing is good teaching, and good teachers anticipate the problems of their students.
The second should confront the biggest weakness of the study, how you tried to ameliorate it, and perhaps how future work could better tackle it (in other words, ending on a positive note). Do you want to be the first person to raise this issue, or would you rather your reviewers do so?A caveat paragraph depicts a thoughtful author who is after the truth, not someone who is trying to sell something.
- If your Discussion is more than 2x longer than your results, cut it down. Discussions are not brain dumps, nor are they opportunities to lay claim in print to every idea you have on the subject. Careful topical reviewers, by the time they reach the Discussion, want to know how your results relate to your hypotheses, the strengths and weaknesses of your results, and perhaps one or two implications of your results. Focus on these three tasks, and leave your reviewer wanting more, not flipping ahead to see when the bibliography begins.
- Market test your title and abstract. More and more editors are rejecting papers before they send them out for review. Reviewers typically accept or decline to review papers on the basis of the title and abstract. The title and abstract are the front door to your study. They are the most important parts of the paper. Craft them carefully and show them to your friendly reviewers.
- Spell check everything. Natch.
- Even your bibliography. Your scientific career is built on a bedrock of trust. Reviewers want to believe that you have carefully collected and analyzed your data. However, to a large degree, your reviewer’s ability to see just how meticulous you are is limited. This is why typos in the manuscript loom far larger than many beginning scientists think. And, similarly, why care in constructing your bibliography–that Latinate names are italicized, that the journal’s formatting is followed to the letter, that authors names are spelled correctly–also reflects your ability to conscientiously manage detail. Will reviewers give you the benefit of the doubt? Often it’s the little things that decide.
- Read it aloud. There is no better way to gauge the flow and logic of a manuscript than to read it aloud, effectively using your whole brain in the enterprise. Beginning scientists should do this in three steps:
Read the first sentence of every paragraph, in sequence, from the Introduction through the Discussion. If the paper is well written, it should sound like you are explaining the study to a colleague, albeit in a rather stilted way. If the paper doesn’t make much sense, it needs work on its paragraph structure.
Next do it again, but this time read the first and last sentence of every paragraph. This should result in greater logical flow–the final sentence of one paragraph leading into, and often introducing, the first sentence of the next paragraph. If you find little difference between this reading and the previous, spend a day or two fixing the ends of your paragraphs.
Now, after a cup of coffee, a long walk, or a nice bout of screaming into a pillow, read the whole paper aloud. Listen for any awkward phrasing, which will sound like your car engine misfiring. For some reason, reading the whole manuscript aloud allows you to see it in a new light, or, more aptly, with fresh ears.
- The following is from the comentaries of the original post:
Arrange the tables and figures on your desk in the order they appear in your manuscript, but don’t look at the text. Do they tell your story, even with the text removed? Does the progression of figures have a natural feel to it, or do they need to be rearranged to flow better? Are there any figures that don’t seem to belong and potentially could be removed to streamline the paper? Is there consistency in style? If the figures are shrunk by 50%, can you still read the text? (Ben)
- Here there are more: "Fourteen steps to a clearly written paper"
- UPDATE: a great post with more writing advice: some well known tricks for clear writing