Analyse blog content with Textalyser

Ever wondered how your blog measures up to others in terms of readability, ego, concern for readers?

Well, using Textalyser, I took a tiny step down that path.

Reservations:

1) You could be quoting others (which could distort the use of 'I' or 'me' etc)
2) Your sidebars could be full of diluting text
3) The word counts don't seem to match my own counts (although this would affect all results equally)

I pointed Textalyser at my own main blog page and those of a few people I've met. I was looking for the balance between reader-referencing and self-referencing words (in the top ten word hits). The fog index of readability seemed like a good measure too: 6 is easy, 20 is hard. Congratulations Robert Scoble for being the most readable.

Update [4/4/06]: People didn't know what I was banging on about so I've added a couple more columns to make it clearer. I have also coloured the relevant columns. 'fog' is writing clarity, 'you(r)' is how much the blogs refer to their readers and self is how much blog writers refer to themselves. Frankly, I'm embarrassed at my own scores - my aim is to drop the fog value and reverse the positions of the other two coloured values. (In this particular post, I have failed miserably with a fog of 9.0, a you(r) of 2.2% and a self of 8.3%.)

Name	fog	you(r)	self	I	own name	my	me
David Tebbutt	7.0	2.0%	3.5%	3.5%
Dennis Howlett	5.2	1.0%	3.4%	1.7%	0.7%	0.5%	0.5%
Hugh Macleod	6.0	2.0%	5.1%	1.9%	2.6%		0.6%
JP Rangaswami	7.5	1.6%	5.7%	3.4%		1.3%	1.0%
Neville Hobson	6.1	2.2%	5.3%	2.2%		0.9%
Robert Scoble	4.3	1.8%	5.0%	3.1%		0.9%	1.0%
Shel Israel	7.1		4.0%	2.0%			2.0%

Make of it what you will. I'd be interested to develop a workable yardstick. Hopefully IBM's UIMA (Unstructured Information Management Architecture), which it is making Open Source, will provide all sorts of opportunities to analyse and interpret blog posts and other public information.

Comments

[Jason]

interesting .. i am trying it now on my blog : http://itblood.blogspot.com and http://todayfinancialworld.blogspot.com

[Jason]

interesting .. i am trying it now on my blog : http://itblood.blogspot.com and http://todayfinancialworld.blogspot.com

[David Tebbutt]

How odd. That didn't happen to me.

I only went for the top ten words but, even in the full list, I didn't notice much that was htmly.

Perhaps we need a better analyser. Anyone know of one?

[Gregory Kohs]

Interesting tool. However, when I ran it on my own blog, it seemed to pick up on a lot of the HTML mark-up formatting terminology in the "source" code of my page. Things like "margin" and "padding", which I don't really write about that much!

Still, I found it useful for the "Fog" and "Readability" measures.

Thanks for pointing out this resource!

Greg

[Dennis Howlett]

Thanks David - makes more sense now. As always these things seem to take on a subjective flavour.

The problem with all this stuff at a corporate level comes back to the 'metrics mantra.' I don't think we've got the right metric mindset for this stuff. Corporate bods are too focused on ROI when they should be thinking about what that means beyond a set of numbers. In other words, deconstructing ROI. For instance, if I've understaood the table correctly, Scoble is the most readable followed by myself. Yet I know that most people wouldn't understand my stuff without a reference framework.

Plenty of food for thought methinks.

[David Tebbutt]

I was looking at the proportion of words that refer to the blogger: "I", "me", "my", "David", "Dennis", whatever against the words that refer to the reader "you", "your".

I was using it as a kind of ego metric. If the reader is referred to more than the writer, then this would suggest helpfulness. If the inverse were true, then this might suggest an element of showing off.

As you can see from the caveats, the published figures can't be taken too seriously. But they probably home in on a tendency.

The FOG index was a bonus. It told me that Scoble was probably an easier read than anyone else on the table.

I'd actually started off with a different objective (using transactional analysis keywords, especially 'should', 'ought' and 'must' - SOM) but settled back to something more common and therefore more easily measured.

But you're a psychologist, I'm sure you know of much better measurement systems. Care to share them?

[Dennis Howlett]

Bit lost here David. Don't get it. Help me out please?

[David Tebbutt]

It's not my text analyser, anyone can use it by creating a URL like this:

http://textalyser.net?q=www.confusedofcalcutta.com/

This will deliver an analysis according to the default settings. To catch single characters like "I", you'd need to modify the parameters, which you can do on the same page.

So if you fancy analysing a bunch of thinkers/linkers/stinkers, we'd all be interested to see the results.

I think the value comes from the differences within a group more than the differences between the groups.

[JP]

Interesting analysis. You may have seen my post on thinkers/linkers/stinkers.I am beginning to wonder whether each of these yields a different pattern on your textanalyser, and what each pattern would mean.