I previously talked about how changes in the demography of learners can affect the cultural evolution of a language. The hypothesis is that language adapts to the balance between declarative and procedural memory users. Since alcohol consumption affects procedural but not declarative memory (Smith & Smith, 2003), we might expect to see communities that have a high alcohol consumption using less complex morphology.
I find that communities that have a morphologically marked future tense have significantly higher alcohol consumption than communities that have a lexically marked future tense (Alcohol consumption data from WHO, language structure data from World atlas of language structures, 198 languages, t = 14.8, p<0.0001). This statistic does not take into account many factors, but is meant as a motivation for further research into language structure and social structure.
Smith C, & Smith D (2003). Ingestion of ethanol just prior to sleep onset impairs memory for procedural but not declarative tasks. Sleep, 26 (2), 185-91 PMID: 12683478
This is a very interesting research question, but it’ll need to be linked to fluctuations in alcohol consumption too, right? Alcohol consumption was far higher 150 years ago in the USA. And the last 30 years have seen a particularly dramatic downtrend (see
here for reference).
Or, could there be a temporary effect? Would you expect to see a temporary rise in paraphrastic constructions with alcohol consumption? I wonder if you could do a corpus study of Jersey Shore to study this, haha!
I agree with Chris: interesting research question. Plus, I’m sure you could get some valuable data-collection done at the pub 🙂
it’ll need to be linked to fluctuations in alcohol consumption too, right?
I’d imagine there’d be a time lag between these cultural-demographic changes and the subsequent changes in the language. So the current results may represent demographic conditions where alcohol consumption was high, even if it has substantially dropped since then. Calculating the rate of change would be important, so we can see if short-term demographic changes have an effect.
Good points- I was assuming that current alcohol consumption reflected more long-term distinctions between fermenting and brewing cultures, but you’re quite right about sudden changes in trend. In fact, I did the analysis with the inflectional synthesis of the verb and found a correlation, but this dissapeared when languages from the united states were taken away (it’s robust for inflectional future). The WALS has data on the etymology of ‘tea’ which might be more suitable for long-term analysis.
Fittingly, I’m writing this on the way to the pub!
Figure this then. I had a regular weekly conversation in a pub thirty years ago. It concerned the weather and the height of the smoke coming out of a stack as a forecasting method.
Thirty years later I returned to the same pub and the same guy was standing at the same place, leaning on the bar and having the exact same conversation. What does this mean?
Tony – maybe we should do a study of your friend as a window into linguistic patterns of the 80s. What I had in mind was change over successive generations of speakers, not necessarily a change in the language of a single speaker over their lifetime. It’s also worth admitting that I’m thinking about European pre-industrial societies where everybody drank alcoholic drinks as a way of purifying water.
OK, I realise this post may not have been intended to be taken seriously, but… you’re basing your calculations of language speakers’ alcohol consumption on the WTO figures for the entire country that they’re centered in? I guess this might not be so bad if you’ve restricted yourself to the majority language of each country, but then you’d have to unify the figures for countries with the same language and drop countries without a majority language, so there’s no way you could have gotten 198 by doing that. Unless you did something clever I haven’t thought of, I can’t see how your figures for this can possibly be meaningful. Fun idea though.
Given the link between Islam, which has religious instruction regarding alcohol, and Semitic language use, which has to bias the issue one way or the other, it is hard to believe that corollation implies causation in this case.
These are good points, and I see that convincing people about the effect of social structure on language via statistical analyses of empirical data is going to be impossible. Instead, I’ll need experimental proof. I’ll look at the effects of extra loads on procedural or declarative memory on language transmission. A fun way to do this is makig people do an iterated learning task while drunk!
Yeah, I agree. Drunk diffusion chains are the only way to go.
I have just re-run the statistics for this by collapsing the morphological complexity measurement (MCM) within countries (taking the average). This means there’s only one MCM-alcohol data point per country. As some of the commentators suspected, doing this removes the significant relationship. Ah well, back to the drawing board …