Monday, July 26, 2010

Mischief with words

I was reading a review of Sherlock Holmes that suggested that it telescoped together fifty years of english slang into one film, because the yoof of today can't distinguish Victorian speech vs Edwardian vs the Swinging Sixties, and also that films of yesteryear often look odd to us now because only one example will survive. Roger Moore running down the street like he has a broomstick up his rear end may seem daft to us now, but when Live And Let Die was released, there were ... plenty of films with a well-preserved chap from Streatham running badly?

Perhaps not, but there were lots of films like the Roger Moore Bond, so odd as it seems to us now, at the time it was quite normal.

Not quite sure of that, but hindsight can be a kind mistress. We probably all remember the Moore Bond films as high camp and a man in an alligator suit/submarine car/spacesuit, and forget how it was acceptable for James Bond to slap some sense into silly women. Ah, the good old days. Or at least, the Seventies.  (By strange syncronicity, there's a very interesting entry on Charles Stross' blog about the minimum number of people required to sustain a technological civilisation, with a side-explanation of why Microsoft needed to pay people to make operating systems even when nobody wanted to buy them, that sort of links up with this.  Or rather, shows that you need some way to sustain clusters of people that can build operating systems / can appreciate Roger Moore films unironically.)

From there, I travelled on through the internet to a literary criticism site where somebody was feeding classic books that had been summarised by Microsoft Word into I Write Like, a web page that purports to analyse your writing style and tell you who you're most like.

I'm either like Dan Brown or Cory Doctorow, neither of which struck me as the greatest of conclusions (I have no steampunk anti-DRM tirades involving made up descendents of Christ and a famous man looking at a red cup, for goodness' sake), but I found that if I put the word "sausages" into the web form again and again, I was apparently writing in the style of Neil Gaiman. Hmm.

Reading up on the tool, it seems to just do some Bayesian analysis on your writing, vs some texts it's been taught by the programmer: so really not much more than a word frequency tool, and thus less exciting than I first thought. It doesn't appear to care much about grammar, for example.

What I was a little startled by was how angry some people got that out of the 40 authors you could be like, almost all were dead white males. I read a very angry rant where a woman felt that the programmer was saying there have never been any female writers of any note, which had led her to send him an angry email, he'd got defensive, and she'd rallied all her friends round on LiveJournal to tell him off.

You should never attribute to malice what can be explained by incompetence. It seems likely that he just grabbed the first writers he could think of (maybe calculated to appeal to his prospective audience of geeky males - hence Doctorow, Lovecraft, Gaiman - leavened with some respectable old people to give some provenance of sophistication - Tolstoy, Twain, Dickens - and topped off with some randoms that luckily included a few women.

It doesn't seem that he was building his toy just to make the point that only men can write books, and it is his toy, so why he should subject his choice of writers to peer review seems a bit daft. But having had somebody have a go at him for not including any women or any non-white people just left him saying that he didn't arbitrarily choose on the basis of race, which made these people even angrier.

I'd tend to assume that if you were a Russian computer programmer living in Montenegro and English wasn't your first language, there'd be no reason you would have heard of anyone who wasn't on a list of Great English Writers that (for better and worse) tended to be English and male.

Ok, that is reinforcing a rather biassed view of who produces literature. However, I can't help feeling a more gentle approach would be better than the cybernetic equivalent of wailing 'aargh, you've invalidated my entire existence, you insensitive clod' when somebody didn't immediately conduct a statistical analysis of your favourite authors just so you can ensure that you had a little result from a free gadget on the internet that would tell you that your writing bore some specious similarity to someone else.

Which was rather a long sentence. Phewph. I'd better go to bed, and plan my own language-gobbling assault on the internet. They aint seen nuffink yet.

2 comments:

Anonymous said...

Well, I read that Holmes propositions Watson. History doesn't relate whether in Edwardian, Victorian, Swinging Sixties or plain am I bovvered modern English, but honestly. What sacrilege!

Mr Cushtie said...

Scandalous indeed! I'm sure it's all a misunderstanding and Holmes was doing nothing more salacious than getting Watson to take a good puff on his pipe...

Post a Comment