“Great prose or not”

I am reading the books recommended by Arnold Bennett in his self-help guide Literary Taste: How To Form It, first published in 1909 and reissued in 1938. Can following a prescribed reading list from over a hundred years ago lead to forming a literary taste? A graph is normally included. This week, the statement, backed by facts, “famous writers have more readers.”

It has been a while since I crunched some numbers. The idea came this time from an article written by Mikhail Simkin in the Journal of Quantitative Linguistics. Using the results of a blind test where readers were presented with anonymous excerpts from the novels of Charles Dickens and Edwin Bulwer-Lytton, Simkin showed that:

  • the results were on the level of random guessing – that is up against a giant panda using a specially adapted panda-friendly keyboard, you would in all likelihood lose.

He then argued that:

  • famous writers are different from their obscure colleagues because they have more readers – the corollary being that the quality of writing does not differentiate between them.

As you can imagine this kicked up a bit of a stooshie, which Simkin summarised in another article in the journal of the American Statistical Society, Significance.

They sit around all day, eating bamboo and could probably tell the difference between Dickens and Bulwer-Lytton better than you. Source: Wikipedia.

They sit around all day, eating bamboo and could probably tell the difference between Dickens and Bulwer-Lytton better than you. Source: Wikipedia.

Simkin then went on to develop his argument using data from the Goodreads website. Looking at the top ratings given by readers to the works of Dickens and Bulwer-Lytton, he argued that given the proximity of average top ratings for both writers, once again, what differentiated them was solely the number of readers. All of which got to me  to thinking, what would a similar study show when looking at arch-cultural-rivals Arnold Bennett and Virginia Woolf. Funnily enough, it would look like this:

Watch those numbers stack up.

Which, when turned into a handy bar chart, looks like this:

The mighty bar chart

 

The results are very much in line with those that arose from the study carried out by Simkin, namely that:

  • the difference between the number of ratings for the two writers is noticeable: 2,725 – 91,010 for Virginia Woolf and 41 – 1,788 for Arnold Bennett;
  • the average listed rating is very close for both writers: 3.76 for Virginia Woolf and 3.77 for Arnold Bennett;
  •  Virginia Woolf’s advantage as regards 5 star ratings is small: Virginia Woolf’s average being 26.9% and Arnold Bennett’s being 23.8%,

and of course:

  • “…famous writers just have more readers.”

Objections to all of the above are, like the demons of the man from Gadara, legion. Not the least being, I did not carry out a blind test to distinguish extracts from the works of the two writers. But if nothing else, it has made me consider with a little more depth the validity of any notion of an accepted literary canon. I now have it in mind to repeat the same experiment with Joseph Conrad and H.G.Wells.

Coming up next, The Worst Journey in the World by Apsley Cherry-Garrard.

If you were wondering, I scored 8 out of 12 on the test. 

Note: I made corrections 23rd May 2014 to the title of the chart, Virginia Woolf’s average and range of ratings for Arnold Bennett following suggestions made by Mikhail Simkin.

 

Advertisements

‘Journey all over the universe in a map…’

I am reading the books recommended by Arnold Bennett in his self-help guide Literary Taste: How To Form It, first published in 1909 and reissued in 1938. Can following a prescribed reading list from over a hundred years ago lead to forming a literary taste? A graph is normally included. This week, the Mindly application for the iPhone. 

Regular readers will know that I occasionally like to point out apps that might help throw new perspectives on novels and novelists. Or at least pass a happy hour when you know you should be working. The Mindly app meets both those requirements. It allows you to create mind maps on any subject, include images, build up sub-categories and impress your colleagues. It is, in a word, a stonker. Here’s one I made earlier:

Arnold Bennett mind map.

It is a work in progress, which is not to say I will go back and refine it. But, as someone averse to organising my thoughts in any way, it has made me think it might not be a bad idea. It allowed me, if nothing else, to see what it is that I regard as important in the life and work of Bennett. It also takes all the hard work out of the process, namely making it look neat. I do not know enough about the life and works of Virginia Woolf, but I would be intrigued to see a mind map with her name in the middle.

No graph is included only the assurance that books filled with facts are soon to be ordered.

Calculations in search of an explanation.

I am reading the books recommended by Arnold Bennett in his self-help guide Literary Taste: How To Form It, first published in 1909 and reissued in 1938. Can following a prescribed reading list from over a hundred years ago lead to forming a literary taste? A graph is normally included. This week, just how much literary taste do I have? 

Time for a time out and ask the question, how much literary taste have I accumulated so far from reading the books recommended by Arnold Bennett?  Below in the first chart is the raw data:

The graph!

This can be broadly understood as:

When I'm good I'm very, very good.

Turn it into a pie chart and it looks like this (ignoring any points that have fallen on the dotted line, giving them a value of zero):

Fancy some pie?

Thus giving me a literary taste rating of 59%. However, and I’m sure you saw that coming, what does Not Literary Taste mean? In this calculation it all seems very passive whereas you would think it should be taking a much more active role in determining my literary taste. It is not an absence of literary taste, rather it is a collection of data in its own right which demands to be considered.

So, if 100 equals an average literary taste then the books I read will provide variations, plus and minus, from that mean. This can be expressed as 100 + Literary Taste – Not Literary Taste, in this case (100+59)-41, giving me a result of 118. Thus, according to this calculation, my literary taste is 18 points above the mean. Which, to me at least, sounds better than a literary taste reading of 59%.

All of the above assumes every coordinate is of equal value, i.e. 1. But, and I’m sure you saw that coming too, the position of the coordinate in relation to the dotted line is an indication of the strength, or otherwise, of the literary taste given to me through the reading of a book. Should that not be reflected in any calculation? For example, good coordinates nearer the dotted line should have lesser value than those further away, i.e. the closer to Not Literary Taste the less impact the book has had on me, and the reverse for bad coordinates, as in:

Now for the big sums.

Weighted like this, the coordinates now give a pie chart which looks like this:

Damn, that's good!Using the same weighted data, I come out with a literary taste 84 points above the mean. All of which seems to point out previously unknown depths to my literary taste, and which would doubtless come as a surprise to anyone who knows me. I know what you are going to say. Literary taste is composed of a huge range of ideas, beliefs, prejudices, all of which in turn draw on historical contexts, class backgrounds, that it is pointless to quantify such a nebulous and subjective concept. To that objection I say that we live in a age of wonder and change, where technology strides across the world like a behemoth. It is only a matter of time before someone develops an app for the iPhone.

Meanwhile, I will take refuge in the bucolic idyll of Mary Russell Mitford’s Village Walks and plant a coordinate in the graph once I have read it.

%d bloggers like this: