On November 3rd 2010, author Charles Stross wrote a wonderful piece pointing out a bizarre exchange in the British House of Lords. "Did somebody just try to buy the British government?" he asked.
Over the course of the day, comments flowed in and discussion happened around the web: as I write this, the prevailing theory is that the noble Lord James of Blackheath may have been fooled by the most audacious prank, performance art piece, or advance fee fraud scam in recent memory: the bizarre Office of International Treasury Control. That is, of course, just a theory.
One of the comments on Stross' article wondered whether, since Lords are appointed for life, there was a chance that a few of them might have lost their mental faculties. That comment linked to a Guardian article about Lancashire and Hurst (2009), a study titled "Vocabulary Changes in Agatha Christie's Mysteries as an Indication of Dementia: A Case Study".
(For legal reasons, I should make clear: no-one is asserting anything about the faculties of any particular noble Lord.)
According to that study, Agatha Christie's later novels suffer from a more restricted vocabulary and a significant increase in the use of the indefinite nouns thing, something and anything: Lancashire and Hirst provide references to show that those can be indicators of dementia.
Why not do the same analysis for the Lords?
So, I downloaded the complete text of every House of Lords speech since 1999 from the wonderful They Work For You datastore. That's over 700MB of data, and somewhere in the region of 100,000,000 words.
Then, I wrote a script that analysed each Lord's contributions, split into the annual October-to-July sessions of Parliament. If a Lord spoke less than 10,000 words during a session, they were excluded from that year's analysis. (As a comparison, Lancashire and Hirst used 50,000 words from each book in their study of Christie.)
The script then analysed three things for each Lord and each year:
The above motion chart tracks the speeches given by each of the Lords since the reform of 1999. Here's the raw Google Doc with the statistics.
Each dot on the motion chart represents a noble Lord. You can use the play button and time slider at the bottom to run from 1999 to 2009.
On that chart, we'd be looking for any Lord whose dot takes a steady decline towards the bottom right of the chart while maintaining the size of their dot (their word count).
The heartening but slightly anticlimactic result: no, they're not losing it as far as I can tell.
There are occasional outliers, but you'd expect that with a sample size like this. For example, the Earl of Erroll and Lord Addington both use a high number of indefinite nouns — but they do that consistently; it's not a steady decline over years.
And as for Lord James of Blackheath? He does use a few more indefinite nouns over time, true, but the difference is hardly statistically significant: one could not impugn the performance of his noggin without risking a libel case.
Needless to say, there are many caveats to this:
Still, here's one for democracy and for open data. Good news everyone: your lawmakers are probably not going senile.