Do you know the Flesch score of your papers?

We just had a very interesting discussion in the Ask the Nature Editor Forum about scientific writing. Most people agreed that the quality of the writing in the end doesn't really influence the decision to accept or reject a paper. But good writing, especially in the first paragraph, certainly helps.

But what is good scientific writing? Two weeks ago I suggested a few good books on the subject in a blog post. Once you have written the paper using the advice in these books, you can use a number of tools to measure the readability of your paper.

Flesch Reading Ease Score:

bq. 206.835 – (1.015 x ASL) – (84.6 x ASW)

ASL is average sentence length and ASW is average number of syllables per word. The Flesch Reading Ease Score can be between 0 and 100, 100 being the most difficult.

The Flesch-Kincaid Grade Level Test is similar, but rates the text on a U.S. school grade level. Both Flesch scores are built into recent versions of Microsoft Word or you could use the Open Source application Flesh. Flesh is much easier to use than the Microsoft Word tool and will also open PDF files. Try to rewrite your manuscript if your Flesch Reading Ease Score is too low, e.g. below 30.

Readability is important not only for manuscripts. Informed consent forms for patients wishing to participate in clinical trials in Medicine are often difficult to read. And a perspective article by Celeste Condit in the current Nature Reviews Genetics talks about How geneticists can help reporters to get their story right. She points out the importance of readability to communicate often highly technical material to lay people.

Related Posts Plugin for WordPress, Blogger...
This entry was posted in Snippets. Bookmark the permalink.

9 Responses to Do you know the Flesch score of your papers?

  1. Bronwen Dekker says:

    I thought it would be fun to analyse my “last blog post”:http://network.nature.com/blogs/user/bron/2007/10/14/a-post-a-day-sunday so I downloaded the Flesh program.
    These were the results:
    Flesch-Kincaid Grade Level: 13.54
    Flesch Reading Ease Level: 45.02
    I also found another web-resource: “TxReadability”:http://www.lib.utexas.edu:8080/TxReadability/app?service=page/Home.
    For exactly the same text that I used for the “Flesh” analysis, I got:
    Flesch Reading Ease: 44.84
    Flesch-Kincaid Grade level: 13.29
    It also does a test for webpages (using the same form, but the result is only related to the “number of words-of-one-syllable”:http://www.utexas.edu/research/accessibility/resource/readability/manual/forcast-versus-flesch-English.html, as far as I can see, though the linked page contains quite an interesting table showing the proportion of people in the US estimated to be able to understand the content.
    When I plugged in the webaddress I got:
    Forcast Grade Level: 9.61

  2. Matt Brown says:

    It’s worth noting that Google Documents (free) also has these measures built in, in their latest release. In addition, you’re also given an “automated readibility index”:http://en.wikipedia.org/wiki/Automated_Readability_Index
    To see this information, just run a word count.

  3. Bronwen Dekker says:

    I popped it into Google Docs, and this is what I got:
    !http://protocols.nature.com/image/show/719!
    I think I am ready to move over to Google Docs for my own use now.
    *Is there any reason why I shouldn’t use Google Docs to create a place where both “my authors” and I can edit documents?*

  4. Bronwen Dekker says:
  5. Martin Fenner says:

    I have often seen the argument that the readability of scientific papers is worsening. So I did a quick check on two papers, written almost 50 years apart, that helped their authors receive a Nobel Price in Physiology and Medicine:
    *1962 Francis Crick, James Watson and Maurice Wilkins*
    Nature 1953: “A Structure for Deoxyribose Nucleic Acid”:http://www.nature.com/nature/dna50/watsoncrick.pdf.
    Flesch Reading Ease: 55
    *2006 Andrew Fire and Craig Mello*
    Nature 1998: “Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans”:http://dx.doi.org/10.1038/35888.
    Flesch Reading Ease: 32
    If I find the time, I’ll do a more detailed analysis. But a score of 55 is really good, most biomedical research published today will probably not come close to 50.

  6. Martin Fenner says:

    Bronwen,
    I like Google Docs – and didn’t know about the built-in readability tool. My biggest complaint is the lack of support for the Symbol font.

  7. Bronwen Dekker says:

    Hmmm. Really sorry. Must be overtired. The numbers are actually quite different in the Google Docs. I wonder why this is? Perhaps the text was not long enough or there is some bug…?
    *The TxReadability gives these complete stats*:
    Total words: 406
    Total words sampled: 392 (Single digit numbers are included in ‘Total words’ but not in ‘Total words sampled’. It also ignore text that does not have “sentence-ending” punctuation.)
    Number of sentences: 20
    Average number of syllables per 100 words: 1.61
    Average sentence length (number of words per sentence): 25.29
    1-syllable words sampled: 266
    2-syllable words sampled: 48
    3+-syllable words sampled: 78
    I messed about for like 30 minutes trying to get a version of this text that would give the same word and sentence count in the two. You can find this text “here”:http://docs.google.com/Doc?id=ddzhz2tp_2g9g24g
    The scores were still different though and I really cannot be bothered now to work out what the problem is. Perhaps I need to write to Google. :)
    |…|TxReadability|Google|
    |Reading Ease|41.58|47.70|
    |Grade Level|14.96|13.00|

  8. Nicolau Werneck says:

    I wonder: could there be any relation between the readability score of an author, and his “h-index”:doi:10.1038/436900a?… :)
    As for the discrepancy in counts, I believe some programs might get the syllabe counting wrong… I think this is a bit difficult to do, specially in English.
    Now, I also wonder: what happens to the index of a text when you translate it??…
    I just ran the word count in my incomplete sci-fi tale, written in portuguese… Here are the figures if anyone is interested :)
    Counts
    Words: 753
    Readability
    Flesch Reading Ease: 16.25
    Flesch-Kincaid Grade Level: 15.00
    Automated Readability Index: 11.00

  9. Martin Fenner says:

    Bronwen,
    another readibility tool is *Simple Measure of Gobbledegook (SMOG)*. You can use the tool from the “Natural Literacy Trust”:http://www.literacytrust.org.uk/campaign/SMOG.html website. Your text from above would receive a SMOG Grade of 14.7, or college education level.
    *Gobbledegook* is a perfect word for this discussion. It means language without meaning and has an interesting “history”:http://en.wikipedia.org/wiki/Gobbledygook. Higher SMOG Grades mean more Gobbledegook, something we should avoid.