Wednesday 15 April 2015

In the beginning was the Word


Words can be repeated easily, but are understood with more difficulty. The sense must be determined partly by context, and then the specific function of the word refined by closer examination of how it is used, and how it differs in meaning from other similarly functioning words. Listen, repeat, guess, venture a usage, test, refine, test, utilise. Search for meaning, not noise; reduce uncertainty rather than increase it. Concepts, not tropes.

Thus the word stores of individuals vary: not all people will be able to use precision instruments, or even see the need for them. 40,000 tools is a lot to carry in a brain. A passable job, rough and ready, can be achieved with 9,000. Vocabulary stores are good predictors of general ability, and their high storage cost must be worth it.

Looking across generations is harder. My brothers and I laughed at my grandmother for referring to a drinking glass as “a tumbler” and mocked her for insisting such vessels always be placed on a lace doily, so as to leave no ring marks on the dining table. Testing contemporaries on dated usages is unfair. Does that mean we cannot determine anything about the vocabularies of today compared with those of our grandparents?

I was pondering these weighty matters when once again the familiar urchin messenger boy arrived with yet another missive from the grand house of Woodley of Menie. I cannot be sure, but I think the miscreant boy calls on the under kitchen maid first, and only afterwards comes round to knock on the front door, which may account for his flushed countenance.

What Woodley and his dining companions Fernandes, Figueredo and Meisenberg have done is to tie together four disparate but crucially related aspects of vocabulary: the extent to which a word can be defined (intelligence loaded vocabulary test), the historical birth of the word (when it first shows up in dictionaries, which they call “word age”); levels of national literacy and the extent to which it is now used in written texts (Ngram usage measures 1850-2005). Crafty, this gang. Here are a few highlights before you look at the final corrections on the not yet finished proofs.

Michael A. Woodley of Menie, Heitor B. F. Fernandes, Aurelio José Figueredo and
Gerhard Meisenberg “By their words ye shall know them: Evidence of genetic selection against general intelligence and concurrent environmental enrichment in
vocabulary usage since the mid 19th century.” Frontiers in Psychology published: xx April 2015 doi: 10.3389/fpsyg.2015.00361

It has been theorized that declines in general intelligence (g) due to genetic selection
stemming from the inverse association between completed fertility and IQ and the Flynn effect co-occur, with the effects of the latter being concentrated on less heritable non-g sources of intelligence variance. Evidence for this comes from the observation that 19th century populations were more intellectually productive, and also exhibited faster simple reaction times than modern ones, suggesting greater information-processing ability and therefore higher g. This co-occurrence model is tested via examination of historical changes in the utilization frequencies of words from the highly g-loaded WORDSUM test across 5.9 million texts spanning the period 1850–2005. Consistent with predictions, words with higher difficulties (δ parameters from Item Response Theory) and stronger negative correlations between pass rates and completed fertility declined in use over time whereas less difficult and less strongly selected words, increased in use over time, consistent with a Flynn effect stemming in part from the vocabulary enriching effects of increases in population literacy. These findings persisted when explicitly controlled for word age, changing literacy rates and temporal autocorrelation. These trends constitute compelling evidence for the co-occurrence model.

The co-occurence model is the decidedly unsexy name given to the theory that Flynn effects and dysgenic effects co-occur. I call it “the leaky boats hypothesis” (Loehlin, 1997) : Flynn effect is the rising tide; Woodley effect the leaky boats.

In the West, up until the early to mid 19th century, those with high levels of socioeconomic status, wealth, and education (all of which are proxies for intelligence; Herrnstein and Murray, 1994) had higher numbers of surviving offspring relative to those with comparatively lower levels (Clark, 2007; Skirbekk, 2008), suggesting that higher intelligence may have conferred fitness advantages
on individuals having to cope with extremes of cold, disease outbreaks, and conflict (Woodley and Figueredo, 2013). Subsequent increases in global temperature, coinciding with the end of the Little Ice Age in the mid-19th century, reduced environmental harshness, boosting agricultural yields thus reducing ecological
stress and conflict (see: Zhang et al., 2007, 2011 for a demonstration of the inverse historical relationship between temperature and conflict). This would have substantially relaxed selection against those with lower intelligence (Woodley and Figueredo, 2013). This was coupled with advances in medicine (which would have included better means of fertility control, hygiene, nutrition, and medication; Lynn, 2011), and also social innovations such as welfare, mass schooling, and universal healthcare. The combined effect of these was a demographic transition characterized
by general reductions in fertility, which were most pronounced among those with higher intelligence (Lynn, 2011). This was mediated primarily by fertility control coupled with the increasing prevalence of opportunities to delay fertility (i.e., higher
education, increasing status competition, etc., which disproportionately attenuated the fertility of high-IQ.



As the authors say: More difficult words presented sharper historical declines in usage
over the 1850–2005 period, []words for which pass rates are more negatively
associated with fertility are decreasing in usage over time, which is consistent with those being the more difficult words []words that are both more difficult and for which pass rates are more negatively associated with fertility are decreasing over time.

It may be a minor matter, but I did my own little analysis, and find that word A is an outlier. It is very easy, and much more frequent than the other items: all the other words are rare by comparison. Every test needs to start with an easy item, but I wonder how much work this particular word does. However, even this very common word shows exactly the same trend as the other easy, but far less frequent words.

In sum, have Woodley et al. put another crucial piece into the dysgenic jigsaw? At the moment, it looks like it.

I sent the messenger boy off with his usual shilling, and noticed a certain impudence as he accepted it. Perhaps the under kitchen maid has been spreading false rumours about me, in her contumely way.


  1. Are "utilization frequencies" just "frequencies of use"? If so, why not say so? When I was at school, use of needlessly long words was taken to be a sign of the charlatan.

    Could it be that the use (nay, not the utilisation) of search algorithms inclines one to use crass, constipated English if one's colleagues do?

    P.S. Their punctuation could be improved.

    1. When you were in school you probably also learned that repeating the same word/term over and over and over is to be avoided otherwise it gets tiresome, so it is recommended to sometimes replace a repeated term with a synonym. "Use" and "usage" are each used dozens of times in the paper, so the authors used "utilization" here and there. Using jargon that no one understands or that has an obscure meaning is one thing (the practice of the charlatan), using a perfectly-correct synonym which is defined in all dictionaries is a different thing.

  2. I'm not sure WORDSUM is a good indicator of intelligence, as despite my great performance on free, online WORDSUM tests, it sounds to me very much as if co-occurence is a synonym for correlation.

    Alternate theories: 1) The historic record of genius tends to favor people who had access to capital, whether personal or that of a patron.

    The question of whether genius has increased or decreased is at heart a personal, philosophical valuation of whether Michelangelo's work is greater than that of the design of the iPhone or a better t-shirt. Before the machine age, any great artist would produce for the carriage trade. With the rise of mass consumption, great rewards fall to those producing for ordinary citizens. Are those producing for the masses less intelligent/talented/special than those producing for the educated few?

    There are many products produced today in secrecy. We may never know what industry has discovered, because there is a tactical advantage in keeping things secret until you go to market.

    2) The rise of electronic media (radio, tv, internet) decreased the amount of time spent reading print media. Mass acquisition of computers and smart phones saturated the population with devices able to broadcast every thought. (Note: the masses were not able to broadcast their thoughts in 1850.)

    Display of vocabulary is not necessarily proof of extent of vocabulary. I don't use the same vocabulary to talk to the plumber and the investment analyst. Our children are adept at choosing simpler words when interacting with peers, when warranted. People are adept at switching between registers. It is not a proof of shrinking intellect.

    I suppose it is possible that modern life is selecting against reaction times. If you react too quickly to a stimulus while driving a car, you may win a Darwin Award.

    1. Thanks for your comments, brief and quick replies for the moment. Wordsum reasonable predictor of intelligence. Co-occurence is not correlation: distinction important, happen at same time, but not for same reason. Ratings of genius always problematical, ditto innovation, but there have always been golden ages in particular disciplines and arts, see Murray on Human Achievement. All manner of circumstances can influence whether innovations are taken up, but some times are more conducive to them. Electronic media have increased volume of publication: quality will be judged later. Vocabulary tests are not about display, but about understanding.Quick reactions still relevant in modern life, including selecting best reaction.

  3. Good Sir, have you perchance written any poetry or fiction?

  4. On "The Simpsons," ancient robber baron Montgomery Burns (Yale Class of 1914) uses an extremely multi-syllabic vocabulary. For example, when attempting to drive for the first time in a century, he refers to his automobile's pedals as the "velocitator and deceleratrix."

    1. Which proves that Harvard alums write for the Simpsons.

  5. Dear Steve, Very impressed with the multi-syllabic descriptors of the foot operated pedals, but the real test of Mr Burns is whether he was able to advance the magneto on a Model T Ford just when required, using the spark advance lever, as I was able to do age 2.
    Any chance of your coming to the London Conference 8-10 May?

  6. You are quoted in the Washington Post:

  7. I blog about the Post article here:

  8. Ye Gods, journalist Lyndsey Layton bothered to track me down, listened politely and asked good questions, and then quoted me appropriately. This has made my day. (Perhaps this is a late consequence of the Steve Sailer effect on public discussion).

  9. Surely this is more about the fact that more morons are literate today than they were in the 19th century, where the only people doing any writing at all where almost exclusively high-g.

    1. Michael A. Woodley of Menie17 April 2015 at 22:24

      Literacy rates across the entire time series were controlled hierarchically in the analysis. The effects of selection on difficult word-frequency are therefore independent of this. Literacy did however predict increases in the use of easy words.

  10. When Michael Woodley first described this research to me, I was struck by its brilliance and originality as a method for investigating the question of long term intelligence trends. This specific work needs to be taken as part of a *pattern* (or nomological net) of consistent and coherent evidence which he (with collaborators) has discovered in the process of following up our original observation (first published on my blog in February 2012) of slowing simple reaction times (in Silverman's meta analysis) as objective evidence of declining general intelligence. There are now something like half a dozen studies on different data sets and using different forms of evidence, to confirm that for more than a century IQ scores have been rising while 'g' has been declining. Each of these studies represents an independent test of the original hypothesis - and so far the hypothesis has passed all the tests.

  11. I think that an avenue of research has been opened up, and that all researchers should now dig up any data sets which could confirm or disconfirm the hypothesis.

  12. Is it possible that the words themselves change in difficulty with the passing of time as some become archaic and others change in frequency relative to synonyms?

  13. Word "age" controls for that natural process.

  14. Apparently, there is a linguistic theory that holds language complexity is related to the characteristics of the society which speaks said language. See:

    Over the last century and a half, the English speaking world has changed from (relatively) small communities isolated from each other, connected by print and commercial travel, to, well, the world we have today.

    In between there were radio and television, of course, and now we have the internet. (It is interesting that many of the words changed trajectory on the chart just around 1900, when Marconi invented the telegraph, thus linking geographically far-flung communities through technology.)

    So the community is now world-wide and comprised of many, many second-language speakers. A more simple vocabulary is thus not necessarily proof of a dysgenic trend. It is a predictable consequence of a larger, heterogenous English-speaking culture.

    One way to test this would be to compare a modern language spoken by people who also speak English. Have the Dutch, the Finns and Romansch speakers faced declines in language complexity to the same extent as the English speakers? They benefit from all the advances of modern medicine, just as English speakers do.