Friday, October 17, 2014

Quality time with your LAD

THe NYT has a piece on language development and its relation to kids' lexicons (here). The reported study "debunks" an earlier one that tried to track vocal size and language proficiency at 3:

"It has been nearly 20 years since a landmark education study found that by age 3, children from low-income families have heard 30 million fewer words than more affluent children, putting them at an educational disadvantage before they even began school." 

The new study insists that quality is more important than quantity:

"The quality of the communication between children and their parents and caregivers, the researchers say, is of much greater importance than the number of words a child hears."

What's "quality" mean? Well it seems that language proficiency is better if adults talk to their kids meaningfully:

A study presented on Thursday at a White House conference on “bridging the word gap” found that among 2-year-olds from low-income families, quality interactions involving words — the use of shared symbols (“Look, a dog!”); rituals (“Want a bottle after your bath?”); and conversational fluency (“Yes, that is a bus!”) — were a far better predictor of language skills at age 3 than any other factor, including the quantity of words a child heard.

So, lots of words without "shared symbols," "rituals," or "conversational fluency" is not optimal.  Here's what Hirsch-Pasek says:

“It’s not just about shoving words in,” said Kathryn Hirsh-Pasek, a professor of psychology at Temple University and lead author of the study. “It’s about having these fluid conversations around shared rituals and objects, like pretending to have morning coffee together or using the banana as a phone. That is the stuff from which language is made.”

It's the last sentence that bothers me. Why? Because it suggests that there is some interesting linguistic relation between how words are presented to kids and the G that develops.  Let me be clear, I can imagine that interacting with your kid in "meaningful" ways could have an impact on G development. But this is not a fact about language or FL, but a fact about human interactions. Meaningful interactions matter all over the place: I also believe that students who think teachers care about them do better at learning than students who think they don't. Nor is it species specific. Chomsky in Aspects (p. 34) cites Richard Held as showing that "stimulation resulting from voluntary activity…is a prerequisite to the development of visual space, although it may not determine the character of this concept… [and] it has been observed (Lemmon and Patterson 1964) that depth perception in lambs is considerably facilitated by mother-neonate contact, although again there is no reason to suppose that the nature of the lamb's "theory of visual space" depends on this contact."

So too with "quality talk."

In fact, we suspect that such talk is not particularly efficacious as language growth can take place well without it. Indeed, in some cultures one doesn't talk to kids because they don't talk back (thanks Paul). Middle class americans talk to their kids, and their dogs and their cats and even their cars (especially if it has a Navi system). These don't develop anything language like despite all the quality discussion.  So whatever is going on here, this interaction is at best accidental and is not "the stuff from which language is made."

I have nothing against interacting nicely with our younger non-fluent conspecifics. I have indulged in this practice myself. However, I am pretty sure that it tells us nothing about our distinctive human capacity to acquire and use a G or a lexicon for that matter. It tells us that kids do better when emotionally supported (if it tells us even that). So talk to LADs, and nuzzle a lonely lamb, but don't be fooled into thinking that how you do this really tells us much about the inner workings of FL.


  1. I would not dismiss this so quickly, Norbert. I agree that it's tempting to think that it's just a matter of having an emotionally supportive environment, etc. And that's what I used to think, too. I also share your frustration that the NY Times is inexplicably failing to plunge its readers into the deeper learning issues. But this might not be as irrelevant as you think.

    The question here is: what is the role of having a sizable vocabulary (word forms and/or word meanings) in supporting other types of generalization? And are certain types of interaction more conducive to growing a vocabulary, especially in the earlier stages. I've seen more and more cases in the past few years where vocabulary development is a predictor of generalization in other domains. And it's perhaps not surprising that certain situations facilitate word learning.

    In learning generalizations about the native language sound system (~8 months), learning a handful of stable word forms seems to have a transformative effect on learners' ability to recognize sound distributions (cf. convergent work by Dan Swingley, Henny Yeung, and Naomi Feldman). And work by Pat Kuhl suggests a boost from live interaction in learning the sound distributions (she doesn't explain why, but the Swingley/Yeung/Feldman insight fills in that piece).

    In the second year of life, there seems to be a tight linkage between vocabulary size and important shifts in phonological development. [Note 1: this is about the abstract representation of the phonology, not some superficial aspect of motor control] [Note 2: in this instance I'm less certain of the cause-effect relation]

    In the third year of life, work by Jeff Lidz and his students shows an interesting link between vocabulary size and kids' ability to exhibit sensitivity to Principle C in preferential looking tasks. Again, the exact reason for this remains unresolved, though Jeff and friends have been hard at work on that. But the point again is the kinds of interaction that might foster vocabulary growth appear to promote other types of learning, of the type that you (and I) find more interesting.

    I used to think that vocabulary growth was dreadfully boring. I'm no longer so sure.

    1. If I implied that vocab size was unimportant then I misled. I have no idea if it is or isn't. What I find unsettling is the idea this is the heart of the matter. That language competence reduces to vocab quality. Note their view is not size, but quality of the linguistic interactions. This smells to me like Lang as communication shtick once again. I can see that learning 100 words is a big deal to grease the system. I'm not sure that I see how quality interactions would play a similar role. In fact, I don't believe that the work you allude to has anything to say about this. Does it? What's quality got to do with this other work. Numbers, sure. Nice interactions. How?

    2. Two important things to remember here. First, these folks are meting out advice of the "take two aspirin" variety, and they're not trying to lay out the biochemical effects of acetylsalicylic acid. And in doing so, they're going to appeal to some folk ideas that might rub you the wrong way. Second, the "size doesn't matter" point here is about the input to learning, not the output of learning. I.e., getting lots of language input via the TV might not be so useful. The mediating factor in what they're describing is probably simply that kids are learning more (words) from input that is pitched at a level that they can understand. If you talk over kids' heads, then they might not get so much out of it. This reminds me of work by Meredith Rowe that found that tried various interventions to promote vocab growth in low SES families, and found that the most useful thing was not to work with the kids directly on vocabulary, but simply to educate the parents about children's cognitive development.

      I suspect that all that's going on in the work behind the NY Times piece is that quality interactions facilitate vocab growth (there's lots of work that addresses why that might be), and that vocab growth greases the wheels for learning other things that we'd find more interesting. It allows learners to recognize things in the input that the would not pick up on otherwise.

      Does all of this matter at the level of the anxious NY Times reading parents who will eat up articles like this as they plot out junior's path to Stanford? Perhaps not. Those kids have so much going for them that things are going to work out just fine anyway. For those guys, the same researchers would have the message "don't sweat it, and don't waste your money on Baby Einstein" and suchlike. The effects that they're interested in the current article are more relevant for kids who are growing up in very different settings, where they're at far greater risk.

  2. Here is a link to an earlier study by some of my fellow co-conspirators:

    I haven't seen this current paper but it sounds very similar, down to the headline "quality matters more than quantity".

  3. One thing I always worry about in studies like this is genetic relatedness. They tend to show that parents who spend more time speaking to their children, or use more vocabulary, or aim particularly high 'quality' bits of speech at them, have children with higher verbal intelligence, better vocabulary, phonological awareness, etc. But if the measured outcomes, like most complex cognitive traits, are at least partially heritable, the studies don't really teach us anything about cause and effect, do they?

  4. A propos of the genetic issue, there is this blog post by Dorothy Bishop here .

  5. Thanks for the link, Alex. Very interesting. I had just been wondering whether there have been twin studies about this sort of thing.