Wednesday, September 04, 2002

As a follow-up to my previous posting, I remembered that Microsoft Word gives readability statistics, so I experimented with copying-and-pasting some of my favourite blogs into Word and then checking their readability. Halfway through the process, I realised how pointless this was, but I persevered. The results are completely flawed, but here they are:

There are six readability stats for each site, and they are stated in this order: (1) number of sentences per paragraph; (2) number of words per sentence, (3) number of characters per word; (4) number of passive sentences; (5) Flesch reading ease; and (6) the Flesch-Kincaid grade level.*
 (1)(2)(3)(4)(5)(6)
Here Inside3.722.05.012%44.8%12.0
Blogadoon1.425.64.64%55.5%11.7
Bboyblues1.321.24.00%72.4%8.4
Not So Soft1.6144.70%61.3%8.1
Not You2.316.24.41%67.2%7.9
Troubled Diva5.016.54.67%65.4%7.9
Swish Cottage4.414.54.31%70.9%7.1
So...3.815.84.24%76.7%6.1
Naked Blog4.511.44.31%78.0%5.2

*Flesch reading ease: the higher the number, the more readable the text.
Flesch-Kincaid grade level: the US school grade level a reader would need to be to understand the text. Seventh grade means twelve years old, I think.


Methodology: I chose bloggers known for writing large chunks rather than simple links, but - typically - Ian and Meg are currently writing untypically short paragraphs. I deleted stuff like the date and the 'comments' link, etc [and quotes from the London Review Of Books].

Spurious conclusions:
  • The bloggers with the most-educated readers: Charlie and Ian.
  • The most passive bloggers: Charlie and Mike.
  • Most rambling: Ian and Charlie.
  • Most concise: Peter and Meg.
  • Most accessible: Peter and Duncan.

    This last result makes me realise just how statistics can lie, as Peter's writing contains many brilliant stylistic quirks, such as his ear for local dialect, ignored by automatic analysis.
  • No comments: