Currently, I am writing papers with rather short sentences: About half of them contain around 20 words, about 10% to 20% even less. I am doing this for three reasons:

  • I find short sentences easier to read,
  • prefer to show relationships using other means, such as colons and conjunctions (e.g. thus, because of, but),
  • and thought the scientific community agreed upon short sentences being better.

Now I did some research and came upon contrary statements, for instance:

So: Should I use long or short sentences, or a mixture, in scientific writing? Why?


I would say that the sentences need to be the appropriate length to what you are saying, which is liable to be, on average, shorter than novel writing. One of the reasons for using long sentences is to convey a mood, to put a lot of ideas together in one, to build and build the picture you are drawing. In scientific writing there is no need for this, so this sort of long-winded sentence should be avoided.

However, it is also important to make your sentences make sense. If you need to put a lot of information together. Then put it in one sentence. Don't chop them just for style.

  • What, if chopping is really easy? Simple example: The solution is foo that has to do bar and therefore goes baz. -> The solution is foo that has to do bar. Therefore, it goes baz. Which one is better? Up to which length of the sentences?
    – DaveBall
    Mar 9 '12 at 12:39
  • 3
    I would say that the former is better, because it has a logical flow of a sentence - it is not overly long, and there is nothing gained by cutting it shorter - I don't feel there is greater clarity in a chopped version. Mar 9 '12 at 13:24
  • Yes, I see it the same way. I'd say chopping makes sense if sentences become longer than roughly 20 words (highly dependent on the sentence, though).
    – DaveBall
    Mar 9 '12 at 14:23

You want to do whatever makes the text easiest to understand. For me, that means a mix of long and short sentences.

Scientific writing is already going to be dense and complex. There are times when you have to write long sentences because you have to string a lot of information together, and separating the ideas will make them less clear. When you can, break up those long sentences with shorter ones. It will make the material more digestible, and give the reader someplace to rest between thoughts, if that makes sense.

You don't have to talk down to your audience, but there's no reason to put them to sleep, either. Paragraph-long sentences become a soporific drone to the reader's inner narrator. Short ones keep folks up.

  • 4
    @Dave: I would suggest a mix too. The most eloquent scientific papers I have read do not seem to be forcing anything, concise points are delivered concisely, complicated points might be expressed in more structured sentences. You only need to interfere in your natural writing process if you are doing too much of one thing. Forcing everything to be short might make your points seem staccato.
    – M.A
    Mar 6 '12 at 13:01
  • +1 for "making the text easiest to understand". But after doing some research, I am afraid that too many too short sentences might make the text choppy, therefore not easier to understand.
    – DaveBall
    Mar 6 '12 at 13:06
  • 1
    @M.A: I like your comment (+1). Why don't you write it as an answer?
    – DaveBall
    Mar 6 '12 at 13:07
  • @DaveBall: Great, so you've identified what not to do: Don't use too many choppy sentences. You should use long sentences when appropriate. Mar 6 '12 at 13:56
  • @DaveBall: I feel my comment is just a small addition to what Lauren said, glad I could help.
    – M.A
    Mar 7 '12 at 17:11

As always, writing style is only a means to an end (in technical writing, at least), so there exist no rules that must always be followed. Still, there are some general ideas that lead to good writing. I'll provide an analogy from Software Engineering.

In Software Engineering, we have the concepts of cohesion and complexity of a piece of code, let's say a function. A function is said to have high cohesion if all relevant parts to some task are handled by this function. This is considered good, as the reader only needs to read this function to understand the execution of a task.

A function has high complexity if it is big and/or nasty, i.e. difficult to read. This is considered bad.

When creating a function, you must sometimes be careful to make the function not too complex while remaining cohesion. The same can be said of writing: your piece of text is most understandable if all relevant information is in one place (in this case a sentence, but we can also consider this on the paragraph level), as long as it isn't too complicated to read.

I see two reasonable approaches here:

  1. Maximize cohesion while keeping complexity under a fixed 'limit'. This means writing long sentences when you can, i.e. when they can be understood. This is the approach that I personally seem to employ. (I don't consciously look at the length of my sentences unless they are way too big)

  2. Minimizing complexity while keeping the cohesion above a fixed 'limit'. This means writing sentences as short as you can afford. A risk of this approach is that you sentences get too much 'communication overhead': if you have a lot short sentences, you must indicate the relation between those sentences. If that 'glue' has the size of half the length of your sentence, readability is at risk.


In the broader technical writing field, we often consider the Flesch Reading Ease metric, and the Flesch-Kincaid Grade Level metric, and the underlying assumption tends to be that one should write for the lowest grade level and highest reading ease score that one can and still retain the requisite technical meaning and detail. That is due to the generalized and often untypified audience we face in writing manuals, procedures, conceptual descriptions and technique descriptors; in this instance, you presumably know the general reading level of your intended audience is high, so I'd tend to balance your writing for that target audience in specific.

This would support the "mixed length" approach previously suggested, but I would add that there are a number of online analyzers which can help you assess the Flesch and Flesch-Kincaid scores of a given piece of text: it might behoove you to test your approaches to see how well you can match the expected reading level of your audience.

I'd also add the caveat that your goal in all cases is to serve your audience, and their task in acquiring and comprehending the information and concepts being presented: where longer, more complex sentence structures better support the information being presented, use them. Where shorter sentences get the point across, use them. If several linked concepts are essential to present together to aide in comprehension of a meta-concept, then perhaps semicolon splices and comma splices are appropriate.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.