Who’s this war against? Data from June
In previous posts I’ve tried to use unsophisticated quantitative methods to understand the political rhetoric surrounding the Russian invasion of Ukraine. Here, and very briefly, I want to take a quick look to see whether using the some sort of approach as I took last time to the media of June can help us understand how the war is (mis)portrayed.
In those previous posts, I concentrated on the Russian language TASS news agency and the Ukrainian language Ukrinform agency. The conclusion I wound up last time was that the Ukrainian media places much more emphasis on various places in Ukraine; by contrast, and arguably suggestively, the Russian agency places just as much attention, it seems, on ‘Western’ entities, notably the US, Nato, Europe, and the EU. I suggested that this gives us a quantitative way of determining which of Putin’s twin casus belli — ‘denazification’ of Ukraine and the supposed advance of NATO — was the true one. I believe the June data (for details about collecting it, see the previous post; all this is easily reproducible and I’ll share my janky python script with anyone interested) bears this out. I also think attending to it can yield some more theoretical conclusions about political rhetoric. I’ll begin with that.
Okupanti vs D|LNR
Recall one of the very first bruited claims that people thought Putin would be after (it’s since become clear he wants much more): the recognition of the so-called Donetsk and Luhansk People’s Republics, two breakaway entities in eastern Ukraine where Russia has had control and support, informal and informal, particularly since the events of 2014 (I wrote, so-so-ly, about a good book about the topic here).
The thing I want to note is just this idea of seeking recognition. Ukraine and the international community at large doesn’t recognize it, and it’s noteworthy how this fact shows up in the respective data sets. If you’ll excuse the slovenly data viz, here’s what we see. Here is first the Russian stories, with a search for ‘DNR’:
477 results! That’s quite a lot, it seems. And here is Ukraine’s feed:
Many fewer results (33), and those that occur are often surrounded in quotation marks, as you can maybe see if you open the image in a new window. Here is a very concrete example of how the war is presented by two different sides. (‘LNR’ has a similar distribution.)
We can see the same sort of phenomenon looking at the Ukrainian data. A popular word to describe the Russian invaders is the word ‘окупанти’ which is just ‘occupants’:
If you read the Ukrainian media, you’ll very often see the Russians soldiers described in that way, something which doesn’t have parallel in the Russian media.
This is, I think, a simple but powerful point that in a longer post I would talk more about. The linguist George Lakoff famously talked about ‘framing’ in political rhetoric: this is the way that two sides talking about the same thing describe it in different ways. The classic example is that Republicans speak or spoke of ‘tax relief’ as if taxation were a burden to be relieved like haemorrhoids. ‘Pro-life’ vs ‘anti-choice’ is another such example. Recently philosophers of language have paid more attention to how these sort of lexical choice shape how we think about the world, and I think we see here a sort of data-y way of making the point: the Russian and Ukrainian media frame the war in their own terms, as — in part — the search for recognition of the breakaway Russophile entities or again as the fight against the occupants of the Ukrainians’ country.
Continuity
Last time, I suggested that the data bore out the claim that for Russia, the real enemy was less Ukraine than the west. I think we see the same this week, with one particularly interesting case study. First, last time I colour-coded the most frequently occurring name-like words into one of two categories: ‘Western’ or Ukrainian. I suggested looking at the data suggested the West was much more the focus of the Russian media than places in Ukraine were, and that on the contrary places in Ukraine were the focus of the Ukrainian wire. I think we see this again this week. Here’s the first page of Ukraine (only the first page because of laziness and theoretically I’m on holiday):
The yellow here means that the word stands for a place in Ukraine; the bluey colour for a ‘Western’ entity. The non-coloured ones are almost exclusively words I forgot to sift out (like the word ‘Ukraine’ or the word ‘Ukrinform’). (Also, the last four entries in the bottom right should be yellow.)
And here’s Russia (as last time, note the preponderance of США, USA):
It’s worth noting that many of the unhighlighted words are ДНР or ДНР, the Donetsk and Luhansk people’s republics, or again phrases that I forgot to weed out like the acronym for the Russian army or war ministry or the president’s press secretary. It’s fair to say, I think, that if one had another colour for Russian entities, this would be mostly blue and the other colour, and so while I think it’s not conclusive, my credence has increased in the idea that the direction of Russia’s attention is primarily westwards.
A case study: Severodonetsk
A key story of the month of June was fighting in and around the town of Severodonetsk: even causal observers (including me, because of the aforementioned holiday), readers of the Guardian or New York Times, would probably have heard about the fighting going on there. Arguably, then, it should feature heavily in the respective feeds.
Does it? Well, arguably for the Ukranian feed it does. I’m not going to present enough data to make the case, but it occurred around 260 times:
By contrast, our standard ‘control’ test, the US, occurred 151 times. And that’s surely what we’d expect!
What about TASS? Well:
And our ‘control’?
600 occurrences! While people were dying — and indeed while the Russians were making gains — in Severodonetsk, TASS’s attention was drawn overwhelmingly across the Atlantic.
The probative force of this is weak. As mentioned in previous posts, it’s subject to methodological mistakes or simple blunders (and this post was written super quickly). But it does seem to me a some sort of story reveals itself in the data, and that moreover looking at political rhetoric in this way is potentially useful.
