Brendan Duong points me to another Google innovation, a new way of tracking mentions of words and terms in books, using the huge archive of scanned books in Google Books. It can be used to track ideological fashion and interest over time.
In the easy to use version (there is a complicated-looking raw data download option) some care has to be taken with interpreting the results. For single words it is calculated as a % of all ‘unigrams’ or single words, for two-word phrases it is a % of all ‘bigrams’ or two-words. So percentages will always be lower for bigrams than unigrams, and I won’t directly compare them.
And because we are looking at percentages of all words in the category, a term could be rising in absolute mentions but still declining relatively.
Over the 150 years to 2000, we can see the changing fortunes of the three main Western ideological forces: socialism, liberalism, and conservatism. The interest of intellectuals in socialism is very evident here. Despite socialism entering a long decline in the 1980s, in 2008 it was still more mentioned slightly more often than liberalism. And despite the apparent ideological revival of conservatism, it trended slightly down from the 1960s.
[19/12: graphs and text updated to take account of later data]
Though socialism entered a decline phase in the 1980s, it was replaced with ‘social democracy’ – socialism minus state ownership of commercial industries. But this too entired a decline phase after 2000. Comparing bigrams, classical liberalism rose in the second half of the 20th century but was still well behind social democracy.
Comparing newer political terms, we can see that ‘neoliberalism’ rapidly increased its popularity as a term after the early 1990s (though almost everyone who uses this term is hostile to it; this is picking up academic rather than real-world fashion), while ‘neoconservatism’
was fairly steady. rose more slowly. This may be partly because Neoliberalism’s numbers may be partly due to its multiple meanings in this period – the free market version, the modernising US Democrat version, and probably a few residual references to the German version. The slight increase in neoconservatism mentions in the last decade probably reflects the belief that it inspired some of America’s foreign adventures. It will be interesting to see trends in these terms over the last decade, with a lot of books on both subjects. For reasons I am not clear on, almost everything written on ‘neoliberalism’ is rubbish, but there are good books on ‘neoconservatism’.
20 thoughts on “A Google measure of ideological fashion”
I think this is a good example of how all the humanities hating people are wrong — a lot of the ideas behind much of the web searching and language tools that are used today was initially created by computational linguistics people living in universities (plus a few research centres like Bell labs). No doubt people complained about all the useless things they did at the time.
One of the other ironic things about this is that now there’s suddenly lots of commercial uses for these things, many of the linguistics and mathematics departments where you might have actually learned about the basis of them have been closed or drastically scaled down, and many of the best people have moved into private industry since they get paid much more. This means it’s harder than ever to learn about some of the most cutting edge things in this area — indeed, most students that do degrees in things like linguistics will have very little exposure to them at all.
Though linguistics, or indeed languages other than English, are not what humanities critics have in mind.
Given they myriad of differing interpretations of what ‘liberalism’ and ‘conservatism’ are, I would be wary of attributing much to these figures.
Also, there is no tracker as to whether the reference to the ideology is made favourably or unfavourably. References to socialism may be included in a polemic against the topic or references to conservatism may be used to indicate conservation policy if it can be assumed that the searches used boolean operators.
Food for thought though…
It would be interesting to know positive/negative/neutral but I don’t think it matters much for tracking general interest – people feeling the need to attack something is as much a signal of its importance as people writing in its favour.
As I read it, they search on exact terms. For example, I searched on liberalism and conservatism and rather than liberal and conservative, to avoid the non-ideological adjectivial meanings. Consistent with this reading, the latter are far more common than the former terms.
The words do not have exactly the same meaning in all contexts, I agree. But liberalism and conservatism are always opposed to socialism, so we can track that relativity fairly well over time.
Interesting find! The tool shows some dramatic rises and falls for the phrase (trigram?) “social democratic liberalism” since the 1950s.
You get a very different picture when you compare these:
It looks like all these ideological words have been in relative decline since 2000, except for the late starter “neoconservative”. Could this be due to Google having a smaller sample of recent books, or because there has been a publishing explosion?
One suspects the historically higher frequency of “liberal” is due to this word have non-political as well as political usage.
Alan – I’d simply assumed that the upper default year of 2000 that appeared on the initial screen was as far as they went – interesting that all the major ideological terms are in relative decline through to 2008 (as far as they will let me go). I’ll re-do the figures above later. Socialism is still the most common. Without the absolute numbers it is hard to say whether it is just more books on other subjects or lower absolute interest in ideology.
‘Radical’, ‘conservative’, ‘liberal’ and ‘progressive’ all have non-ideological uses so I don’t think that search is hepful in tracking ideological interest.
‘Social democratic liberalism’ is getting so few hits that a ‘dramatic’ rise or fall may only be very small absolute numbers.
Interesting to track different words against “God” (make sure it’s in capitals- capitalisation DOES matter).
The only word I found that came close to competing with God’s popularity is “me”.
I initially thought that most words seemed to be trending up over time until I tried “God”. That’s definitely a word that has gone down in usage over time (but has seen a resurgence since the mid 90s).
Look for a massive rise in the mention of Socialism since Obama’s election.
I think this is a good example of how all the humanities hating people are wrong — a lot of the ideas behind much of the web searching and language tools that are used today was initially created by computational linguistics people living in universities (plus a few research centres like Bell labs).
Not sure who these humanities-haters are, but I’d guess that they would see daylight between computational linguistics and Media/Cultural/Whiteness Studies.
It looks like all these ideological words have been in relative decline since 2000, except for the late starter “neoconservative”.
I wouldn’t be surprised if there has been a sharp drop in the use of “Neoconservative” over the past 3 or so years, as that word’s users have gone berko with their new discursive toy; Neoliberalism.
PP – I think ‘neoliberal’ is usually used to denounce those who want economic freedom, and ‘neoconservative’ to denounce those who want to export democracy. But you are right that there is a random, according-to-fashion element to which term is used.
Why are, sorry, why were so many people mentioning socialism vis a vie liberalism.
Taking a step back, I think it just shows how silly books are.
Perhaps I am not as generous as you are, but my keen eyeball tracking of all the ideological nomenclature Pick-Up-Stix over the past decade and more, strongly suggests a much less coherent use of the two terms. My reading tends to their use being more incoherent and contradictory. One thing is for sure, both terms are used overwhelmingly – 99% anyone? – by leftists as terms of abuse. While there is an historically identifiable and cogent self-identified “Neoconservative” movement, the same is not true for “Neoliberalism”.
Now, as a long-time blogger, you are in a perfect position to test my hypothesis, depending on the quality of your blog’s “Search” function.
But I am willing to bet good money, that if you were to plot all the uses of “Neoconservatism” over the past decade, and more, on one axis, and “Neoliberalism” on the other, the plot would look remarkably like that for perfectly substitutable goods.
My hypothesis is that leftists – particularly the ginormous rump of the less cognitively advantaged, which makes up the Left – have been using these terms to mean “poo poo head” or their own version of “I don’t like it”.
I would posit an inflection point around 2008. I mean seriously, when was the last time anybody mentioned “Neoconservatives”? If you search back a bit, John Howard was a “Neocon” to a huge chunk of the media – including the blogosphere – whereas today, you never see “Howard” and “Neoconservative” in the same sentence” but boy will you see “Neoliberal”.
The later uses of the two terms do suggest a slight rise of neoconservatism about 2006 and a decline in neoliberalism. The data source of books is probably a lagging indicator- most authors will use terms that are already established, and publisher are more likely to go for books on topics which there is established interest.
I think neoconservatism enjoyed a surge due to its association with the Iraq war, but will now decline relative to neoliberalism – which should get quite a few mentions in the more ideological GFC books.
Here’s why you need “humanities people”: the frequency of the occurrence of these terms – over thousands of published books – tells you practically nothing. Intellectual historians would place all this in context – absent of that context, you can’t go much further than the insinuations that this post trades in. In the last couple of years, the people I’ve heard using the word “socialism” most often are those who imagine that it’s arrived in the White House, and have allowed this delusion to make them angry. Simply counting mentions of particular words misses subtleties like this, as anyone who’s done the most elementary content analysis would know.
Jason – I think that’s a straw man point. Who is saying that a word count can replace analysis of ideas?
If that’s not what you’re saying, what’s the post driving at? And these “humanities critics” you mention – what are they criticising, precisely?
Jason – Conrad raised that issue, and while I did not know exactly what he was talking about I presumed it was a reference to those who deride the humanities as not leading to jobs/studying trivia/being too left wing/encouraging bad writing/generally being useless. In the context of the point he was making, I thought those who make these criticism probably had English, history, politics, sociology and the various ‘studies’ in mind rather than linguistics.
The post should be taken at face value. I think this is an interesting data source. It does not replace other sources or ways of analysing ideologies and their importance.