Fun with Ngrams

One of the pleasures of the British Poetry of the First World War conference (and the pleasures were many – I’ll be mentioning several over the next week or so) was Jay Winter’s talk on ‘Glory’.

He traced the decline of the word in English by use of  Google Ngrams. This is a neat bit of software, available on all computers at the press of a button, that searches the vast archive of Google Books to find all uses of a word or phrase in the books of a particular year. The results, in a graph, show something about the popularity of a word. So, plotting the words ‘glory’  and ‘glorious’ from 1900 to 2000, we find this decline in their popularity (click for a better view):


glory, glorious

You have to take these graphs with a certain amount of caution: they find references in books, not in newspapers or other documents, and they are based on the scanning of texts found in the mighty holdings of the Bodleian Library. These are very extensive, but not absolutely complete; there must be selection bias. Persons mentioned in academic texts are more likely to have their fame recorded than are the stars of the gossip columns.

With these provisos, however, ngrams are at the least very suggestive. Professor Winter made his point about the decline of ‘Glory’ in English by contrasting it with the staying power  of ‘gloire’ in French (which you can check out for yourself here).

I’ve mentioned Ngrams on this blog before, tracing the rise and fall in novelists’ reputations. Having just been to a poetry conference, I thought I’d try to find out about some poets. This chart tracks mentions of Wilfred Owen against mentions of Siegfried Sassoon:

Owen Sassoon

This is quite a neat graphic representation, I’d say, of the change in the poets’ reputations, with Sassoon leading through the twenties, and Owen’s fame steadily growing. Does the crossing of their lines show the change in what people wanted from war poetry, from satire to pity? One could maybe see how publication of the various editions of Owen’s poetry affected his fame.

Add Rupert Brooke to the equation, and you get this graph:

owen sassoon brooke

Of course, the Ngram program is crude in that it just counts references, with no account taken of whether the mentions are approving or negative.

I started wondering about the phrase ‘war poets’ itself (a specifically English term, as Professor Winter pointed out, with no real equivalent in other cultures), and plotted it for fun against ‘modernist poets’. Here’s what I got:

war poets modernist poets

Maybe this tells us something about fashions in criticism. The Ngrams program only goes as far as the year 2000. If it went farther, I bet we’d see a rise in ‘war poets’ over the past fifteen years, and maybe a slight decline in the stock of the friends of Ezra.

One last graph. I wondered what would come up if I plotted Wilfred Owen against Douglas Haig. The result is a graphic illustration of the change in the way in which the War has been popularly perceived:

wilfred douglas

As I said, these results prove nothing, and need to be taken with a large pinch of salt. But making these graphs is a great way of wasting an afternoon.

Try it for yourself.









  1. Bill
    Posted September 9, 2014 at 12:36 pm | Permalink

    It is also instructive to add Blunden to the equation. The graphs of him, Owen and Sassoon go up and down together, with Blunden ahead of both through 30s and 40s until Owen passes him in 1956 and Sassoon in 1964. Blunden then drops away (although still ahead of Rosenberg and Gurney).

    • Posted September 9, 2014 at 1:35 pm | Permalink

      Thanks for this, Bill. The Blunden graph is a very interesting one. Does it reflect his standing as a poet, or is it a result of his activity in propagandising for other war writers? Or a bit of both?

  2. Posted September 9, 2014 at 1:50 pm | Permalink

    Here’s what happens when you stick in a whole bunch of poets’ names, and look at their impact between 1914 and 1930.
    Notice the impact of Robert Nichols, the war poet nobody now mentions – though he was very big at the time.
    This is a thumbnail image. You’ll need to click it.

    • Jonathan Lighter
      Posted September 9, 2014 at 11:56 pm | Permalink

      Looks like “glory” and “glorious” were already on the down slide nearly fifteen years before the war began.

      And, if we can trust the graph at all, wartime writing did nothing to revitalize them.

      • Posted September 10, 2014 at 5:40 am | Permalink

        That’s actually the picture that Jay Winter drew. An earlier slow decline in the popularity of ‘glory’, accelerated by the War. It was a nuanced talk, going far beyond the simplicities if the Ngram.

  3. Jonathan Lighter
    Posted September 10, 2014 at 10:23 am | Permalink

    Of course, they can be used in religious writings too….

    As you say, “Fun with Ngrams.”

  4. Posted September 10, 2014 at 1:03 pm | Permalink

    One clarification: the Ngram Viewer doesn’t just use texts from the Bodleian, but from over 40 libraries worldwide.

    Jon (Ngram Viewer co-creator)

Post a Comment

%d bloggers like this: