Cognitive Daily

Two facts are true about young children: they sleep a lot more than adults, and they learn language at an astonishing rate. How can they learn so much when they’re sleeping so much of the time? Perhaps sleep itself enhances learning. In fact, a number of studies suggest that naps actually enhance learning in adults. What about kids?

A team led by Rebecca Gómez developed a clever test to see if 15-month-olds learn language faster when they’ve had a nap. At 15 months, most infants understand a lot of language, but don’t produce much. But of course, each baby learns at a slightly different rate. How is a researcher to know if the child is learning new words over the course of an experiment, or just reflecting previous knowledge? And if the child can’t say much at all, how does the experimenter determine that a word or concept is learned?

The first problem was solved in several previous studies led by Gómez: the researchers used an artificial language — or rather, just a small portion of a “language.” In English we might say “a bird quickly flies,” but we’d say “birds quickly fly.” The ending of the phrase changes depending on what’s said at the beginning of the phrase. The artificial language created an analogous situation:

pel wadim rud
vot kicey jic
pel puser rud
vot fengle jic
pel coomo rud
vot loga jic
pel gople rud
vot taspu jic

So phrases starting with pel always end with rud and phrases starting with vot always end with jic, while the nonsense words in the middle of the phrase can change. Can babies learn this pattern? And does napping make a difference?

Experimenters visited the babies’ homes and played with them quietly for about 15 minutes while a recording of the artificial language was played in the background. In all, each of the two patterns was heard by the babies 120 times, while 24 different nonsense words were used in the middle place in the phrase, so each of the nonsense words was heard 10 times. This was their language “training.”

The experimenter visits were carefully scheduled around the babies’ normal nap times — during the four hours after the language training, half the babies took naps and half did not. No baby was forced to stay awake when he or she would normally be napping.

When the experimenter returned, she placed the babies in a small booth designed to monitor their attention patterns using the head-turn preference procedure. First, a light illuminated in front of the baby to attract her attention. Then a phrase was played from a speaker to one side while another light nearby was turned on. Sometimes the phrase violated the pattern established in training (e.g. pel suleb jic), while sometimes it matched it (e.g. vot nilbo jic). The time looking in the direction of the speaker and light was measured by an observer. Here are the results:

The babies who had not taken naps looked towards the familiar phrase patterns significantly longer than they looked towards the unfamiliar phrases, while there was no difference in looking time among the nappers. Does this mean the non-nappers learned the pattern better than the nappers? Not necessarily. Consider this graph, which shows how long each baby looked at the pattern that matched the first pattern they saw during testing:

Regardless of whether that pattern was familiar or unfamiliar, babies who had taken naps looked significantly longer towards the speaker when that pattern was played again during the testing. The researchers argue that babies who had taken naps learned the artificial language better than non-nappers, because they could recognize the more general concept of consistency between linguistic patterns — even patterns they hadn’t heard before. The non-nappers simply preferred what they had heard during training, without learning how to apply it to new patterns.

Gómez’s team concludes that napping may actually help infants learn the general rules governing language.

Gomez, R.L., Bootzin, R.R., Nadel, L. (2006). Naps Promote Abstraction in Language-Learning Infants. Psychological Science, 17(8), 670-674. DOI: 10.1111/j.1467-9280.2006.01764.x

Comments

#1 kevin z
January 29, 2008

I don’t get how half of the babies stayed awake after 4 hours and half napped? What 15 month olds don’t nap after 4-6 hours of being awake?
#2 cds
January 29, 2008

Mine! Oh God, mine!
#3 Justin
January 30, 2008

My 19-month-old varies – sometimes he’ll get up at half six am and stay awake till eight pm.

Other times he’ll not wake till nearly eight am, take a two hour nap and then want to go to bed (he’ll take me by the hand to his bedroom and climb on his bed – very cute) by seven pm.
#4 Dave Munger
January 30, 2008

Kevin:

I’m guessing they scheduled this thing so that the no-nap group was tested right before their usual longest-awake period.

I do wonder if there might be a sort of self-selection going on: Did the no-nap group happen to be the babies who don’t sleep as well? Could there be a relationship between regular napping and language learning? In other words, this *particular* nap didn’t help the kids learn; rather, the general *pattern* of napping frequently did.

Also, they only had to be awake for four hours, not 4-6. The total experiment length was four hours, so it was less than four hours from the end of the initial play session to the test.
#5 kevin z
January 30, 2008

My 26 moth old gets up between 5:30 and 6:30am. naps for 1.5-2 horus around 12:30-1:00 and goes to bed between 7:30-8:30. Like Clockwork. He runs around like crazy so much I think he never has a problem going to bed.

Dave, I’m pretty skeptical that could control for nap time in this experiment. But, I haven’t read the paper. Do they analyze the data as one is a sample with several trials (i.e. nested)?
#6 Dave Munger
January 30, 2008

They just have two separate groups — nappers and non-nappers. I imagine when they recruited parents they asked for nap schedules. For your child, they could have showed up around 7 a.m. for the training and then did the testing at 11 a.m.

The data analysis was basically just a t-test (though they also did p-reps since this is Psychological Science).
#7 ScienceWoman
January 30, 2008

I’m skeptical about their ability to control for time of day. Most one year olds nap in the early afternoon. That means that the nap group was trained at lunch time and tested in late afternoon. Presumably most of the no nap group was trained after breakfast and tested at lunch time. How can they say that the babies didn’t just do better because it was later in the day? It takes lots of people a couple hours to be fully functioning in the morning.
#8 Josh
January 31, 2008

Your explanation of the artificial grammar in the 3rd paragraph pointed out something that really bugs me about such experiments. The pairings bird-flies and birds-fly is very different from the pairings pel-rud and vot-jic. The former are predictable; the latter are truly arbitrary. In fact, the salient feature of this artificial grammar has no correlate in English or any other language that I know.

So while the results of this and similar experiments do tell us something, it’s not clear what they tell us about learning language. We might as well train the kids on the xylophone, see if naps help, and then extrapolate to language.

This isn’t a criticism of your post, which was very clear. Many, if not most, artificial grammar experiments have exactly this type of limitation.
#9 Marc
February 4, 2008

Josh #8: I’m sympathetic to your point of view, but I think your reasoning is off. What do you mean “predictable”?
Perhaps “bird(s)quickly fly(s)” wasn’t the best natural language example because the two dependent elements are sometimes adjacent, but lots of languages have long distance dependencies just like in the artificial language.
So yes, pattern learning and language learning may be different and use different learning mechanisms, but that has nothing to do with predictability.