A sentiment analysis of Marxist polemics

So, first off I’ll disclaim a lot of what appears below. I’m doing this analysis primarily because it’s frivolous exercise and I’m interested in the results rather than because I think that they reveal anything substantive about Marxism or politics. I’ve read and enjoyed most of these texts and since my doctoral research to some degree orbits the Marxism and stylistics I thought I’d see what sort of emotive words turn up in Marxian polemics, because a lot of what draws us to these texts are their sarcasm, rudeness and nastiness, as long as these are enlisted in a legitimate spirit of criticism. Though also sometimes also when it isn’t.

I wasn’t systematic about my approach here, I’d be interested to see what someone who scraped New Left Review or marxists.org turned up, instead I copy and pasted associatively through marxists.org until I got a list which included works written by Engels, Lenin, Luxemburg, Marx, Trotsky and Thompson, about 25 texts in total. I know Thompson is sort of the odd one out here, but if you’ve read any of his polemics you’ll have some sense of why I’m including him, sadly Kolakowski’s ‘My Correct Views on Everything’ is not available in plain text format anywhere.

As I’ve already suggested I took a bit of a rough and ready approach, using text mining to remove numbers, punctuation, white space, turning everything into lower case, removing stopwords and stemming our words. We then turn it into tidydata to render it more easily manipulated. We are using a lexicon approach to sentiment analysis, meaning that we have a pre-annotated set of words which are all taken to signify or indicate a particular emotion, you don’t have to be a literary critic to know this has its problems and sentiment is something which depends on context, but, as I said, rough and ready is the order of the day here. I opted for ‘nrc’ because it returned the most number of words, broken down into categories such as ‘trust’, ‘positive’, ‘joy’, ‘anticipation’, ‘surprise’, ‘anger’, ‘positive’, ‘negative’, ‘fear’, ‘sadness’ and ‘disgust’.

I then grouped the mean frequency each of these authors use words belonging to each of these categories. This left us with each six authors and eleven values. I decided I wanted a principal components analysis of these stylistic profiles as they provide some of the most readily interpretable visualisations, and the result appears below

So as we see, the majority of language used by Marxists is in fact not polemic or emotive at all, but objective enough to not be read in any significant way by a sentiment analysis library. Trotsky and Engels are by far the outliers here in terms of the degree to which their works are composed of emotive terms and we can see there is a distinct difference between even these two, with Trotsky tending more towards the negative end of the spectrum, Engels being more aspirational.

What else could be going on here, why is Marx more like Engels than everyone else if as we might hypothesise, scientific or economic terminology underpins this divergence? I went back to our categories and pulled out the top five most frequently occurring words associated with each category. These appear below. ‘Trust’ incorporates words like nation, law, labor, fact and money, ‘positive’ (civil, actual, labor, capitalist, develop), ‘joy’ (labor, money, organ, present, content), ‘anticipation’ (time, labour, develop, dictatorship, organ) ‘surprise’ (labour, betray, money, present, good) ‘anger’ (strike, dictatorship, socialist, betray, money) ‘positive’, (civil, actual, labour, capitalist, develop) ‘negative’ (strike, dictatorship, communist, socialist, war) ‘fear’ (dictatorship, socialist, war, case, rule), ‘sadness’ (socialist, dictatorship, betray, case), and ‘disgust’ (dictatorship, socialist, betray, bureaucrat).

It will not have gone unnoticed that there are some pretty severe value judgements in place here, words like capitalism and development are read as unambiguously positive and words associated with socialism and communism are read as unambiguously negative, most of Trotsky’s ouevre which appears here addresses itself directly to issues associated with actually existing socialism, as well as bureaucracy, so this is probably why he scores so highly for all these disgusting things right-minded people never consider for very long. Marx and Engels probably score so highly for positive, more utopian values because of the degree to which these works are concerned with things as they actually are and the attempt to move beyond appearance, (‘actual’, ‘content’) not to mention how much more concerned their writings are in general with spelling out capitalism’s laws as opposed to speculating on the nature of some other mode of production.

It’s a shame that these categories are so categorical, I understand that they do serve a purpose, but a reading of ‘labour’ as unambiguously positive (hard labour? forced labour?) or money as signifying trust makes such macro analyses difficult to justify, on a micro level they are utterly laughable.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s