mandarine

Answering Lorelle

In a recent post, Lorelle on Wordpress has gathered interesting resources on readability test tools. She invited her readers to have a go with the tools and post the results.

So I’m curious. Take the test and see what it has to say about your blog and tell us about it.

In the flood of comments that followed, I picked up the very interesting reply by Wong Teck Jung, whose blog is written in Chinese. The readability scores were better than most other submissions in the other comments. From that, I understood that these readability tests were absolutely not concerned with meaning, spelling or grammar. Moreover, I was under the impression that these tests were US-biased, as they seemed to apply to the writing rules of the English language, and to relate the scores to US schoolgrades.

I therefore became very skeptical about the validity of those tests. I left a comment basically stating that they told us so little on readability that they were useless:

Let’s be honest - I believe these indicators are as worthless as psychology tests you take in women’s magazines. The only thing the Gunning Fox index tells about your [writing] is how much it scores on the Gunning Fox index - period.

Although Lorelle’s post had reached its goal in helping me think about readability, I had all reasons to distrust such tests as a writing aid.

Inside readability tests

This was my feeling, but I still had to prove it. Lorelle had suggested me to delve into this subject more thoroughly. The following is what I found out.

Readability tests are mathematical formulas, mostly developed for English language texts, that aggregate semantic statistics about a text. The most used characteristics are the number of syllables in each word and the number of words in each sentence. Reference [2] provides a fairly complete overview on the subject. The Wikipedia article [1] will point you at a variety of tests, for which the algorithms or formulas are explicitly written. Test scores seem to correlate quite well with how many (US) people understand a given text, or the (US) schooling level of those who understand the text (see [4]).

In any case, these remain simplistic statistical computations, where neither spelling, grammar, nor semantic field are considered. This explains Wong Teck Jung’s good results: Chinese sentences generally use few words, and words have few syllables (note that syllable-count of the test is highly questionable in this case, as it can only be based on the Unicode representation). In fact, the bulk of readability tests are not adapted for other languages, or should be used with the utmost care (see [3]).

Knowing all this, can we still consider readability tests as a tool to improve our writing ? After all, it was Lorelle’s intention to make us use these tools to think about readability. Now that you know the algorithms that crunch the figures under the hood, you can try to orient your writing to influence the scores. You try to reduce polysyllabic constructs and convolute sentences - or should I say: use short words; put lots of periods. This is not a bad idea, but who needs readability tests for this ? Then you can go further and try to write the shortest possible sentences with the simplest possible words. This seems to be what journalists do when they design headlines. In my opinion, such a habit hardly makes for easy reading (I often find myself wondering where the verb is), although short headlines may help to use larger fonts. Reference [5] gathers a list of hilariously ambiguous short sentences.

Reference [2] gives a good illustration that writing efforts dedicated to improve readability scores are pointless:

The notion of “writing to formula” has been condemned by formula designers from the beginning. They call it “cheating” and compare it to holding a match under a thermometer to warm a room.

I believe once we have made our best ‘common-sense’ effort to avoid undue complexity in our writing, there is no interest in trying to achieve higher ‘readability’ scores by refactoring our prose. If you have prepared a scientific publication on the ‘infrared refraction properties of amorphous zyrconium oxydes’ that scores low on the readability tests, hacking it into little words and little sentences will only make it unreadable to your peers.

Conclusion

I believe readability tests are not writing aids. They do not tell you whether (and even less how) you can improve the readability of your text. I do not deny that the scores are correlated to reading ease, but in my opinion, texts that can be written with short sentences and short words are easier to understand because their general meaning is simpler.

Read on

Why I started blogging

Read away

[1] The Wikipedia article

[2] An interesting article and resource

[3] A resource in French (Quebec)

[4] SMOG on Wikipedia

[5] Ambiguous headlines

This entry is filed under language, tools, reading/writing. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.