What's in a Wordle?

I recently came across a rather interesting java applet at www.wordle.net that seemingly counts up how many times each word is used in a bit of text, and then presents a visual image of all the words - the more frequent the word, the larger the font.  There are other bells and whistles to change the colour scheme, layout etc.

Created by Jonathan Feinberg at IBM, I saw this as a bit of a gimmick that might be useful for putting together a pretty picture for a module handbook, or perhaps just a look at the popular words used in a journal paper or a bibliography.  It will even do RSS feeds - I bring you the words of Digital Directions:

Seems I have a facination with Google!

However, in his blog, Feinberg has created a few visual comparisons of inaugural presidential speeches, which made me think about comparing a few judicial speeches in HL cases.

I haven't got the colour scheme right, as I think the feature to compare texts is not yet available to the general public, but this is a visual representation of the Shogun Finance Ltd v Hudson (FC) [2003] UKHL 62 text:

Nothing special, one might expect to see words such as 'contract', 'person' and 'goods' appearing frequently.  But, compare now the individual speeches of the Lords.
Nicholls and Millet dissenting:

Hobhouse, Phillips and Walker affirming the Court of Appeal:

There are some differences - Hobhouse and Walker both mention the word 'contract' far less than the others.  Walker is the only one to mention 'principle' to any great level, and he also uses the word 'Lord' extensively, but then he does concur with and make many references to Hobhouse's speech.  Nicholls and Hobhouse both refer quite often to a 'crook' and a 'rogue' respectively, but only Hobhouse refers to 'Patel' the innocence third party in the case to any great extent.  Phillips makes more use of the word 'contract' than anything else, whereas the other Lords often make similar usage of other words such as 'person' and 'agreement'.  I could go on, but I'm not quite sure what this analysis tells me.  There may be some merit in looking at the words 'face' and 'written' but when the context is removed, it's not particularly analytical.

I may have some more thoughts on this for deeper analysis; but I thought it might be fun, and perhaps useful for the occasional seminar.

Comments

[Carole McCartney]

Michael - love this! In fact our dept. had a 'word cloud' made up for our RAE narrative submission to see what we mentioned the most and what stuck out (and we did for a few other universities) and we noticed right away that one Prof had stood out more than anything - a point of interest for when doing our REF submission! Esp. as the word cloud for LSE was v.different and had no names appearing in it. I've now done them for the minutes of some seminars I've been holding to see what comes out - apparently I've written 'also' a great deal so I'm going through to cut some of them out!

[Michael Hughes]

We ran the whole SIMPLE final report through this to generate our front cover (unfortunately this cover isn't on the published version on the site).