Like clockwork every few months I feel prompted to write about The Nurture Assumption. In this case it is due to The New York Times reporting that the American Academy of Pediatrics is now recommending that parents start reading to their newborns. As noted in the piece in The New York Times a major reason for this recommendation is the research which shows that higher socioeconomic status families tend to talk a lot more to their offspring than lower socioeconomic status families, and provide them with a richer vocabulary. The assumption is that this head start allows higher socioeconomic status children to outpace their peers cognitively. Naturally they reference Meaningful Differences in the Everyday Experience of Young American Children. This work surveying 42 families was published in 1995, and concluded that children raised in professional households will hear 8 million more words in a year than children raised in a household on welfare.
But that’s old news. There is also a reference to a 2013 study, SES differences in language processing skill and vocabulary are evident at 18 months. The study here looks at 48 individuals. You can see the major result at the top of this post. To me one thing that strikes me are the rather modest correlations. The children in higher socioeconomic circumstances have larger vocabularies, but it’s not an incredibly big difference. There is a huge amount of variation within socioeconomic brackets, just as there is within families. The standard deviation of IQ in groups of siblings is nearly the same as the standard deviation of IQ in the general population. There’s only so much control families and genes have (shared environment + heritable component).
Naturally the first thing that comes to mind is that socioeconomic status and intelligence are not totally uncorrelated, and intelligence is at least somewhat heritable. Smart parents might simply talk more to their children, and those children will tend to be smarter than you would expect by chance. The authors are aware of the behavior genetic literature, but they tend to argue that it can be interpreted in a way which leaves open the possibility for the large effect of shared environment. In general my prior is to be skeptical of this, because the overall body of research suggests that for many behavioral traits the variation within the population which isn’t genetic (on the order of half) is simply unaccounted for.
This doesn’t mean that there aren’t environmental effects which might result in changes in outcome on the margin. But we just don’t know enough about non-genetic component to assume that a silver bullet policy prescription can be formulated out of a few studies. A few years ago Jim Manzi came out with the book Uncontrolled: The Surprising Payoff of Trial-and-Error for Business, Politics, and Society, where he unveiled the term “high causal density.” In other words, there are lots of causes for some effects, and it is difficult to tease apart the variables so as to engineer appropriate responses. This is clear even in the case of the genetically heritable component, which is often polygenic and difficult to assign to any given gene of large effect. And it is also likely in many cases for numerous environmental variables, some of which may simply be stochastic (which can explain differences between identical twins raised together).
Unfortunately a major human cognitive bias seems to be the need to think that we can control things, and effect change. This results in the adherence to fads and fashions such as Freudianism and attachment parenting as the years come and go. The single biggest thing you as an expectant parent can do to have a child with a large vocabulary is to select a mate with a large vocabulary. This won’t guarantee anything, because there is going to be lots of variation on individual outcomes, but in a developed world context this is probably the lowest hanging fruit in terms of ‘return on investment.’ Think of it as ‘loading the die.’ That doesn’t address the issue of inequality, which is really what’s bothering people in this particular case, but I strongly suspect that reading to newborns is going to be a waste of everyone’s time here, though it may make people feel as if they are doing something. Young parents have a finite amount of time, and it seems pragmatic for them to starting reading to children when children can actually start understanding the structure of narratives!