Infants perceive language sounds differently by age 6 months

Blogging on Peer-Reviewed Research

A study doesn't have to be brand-new to be interesting. Consider the situation in 1992: It was known that adults are much better at distinguishing between sounds used in their own language compared to other languages. Take the R and L sounds in English. In Japanese, both of these sounds belong in the same category of sounds: both sounds have the same meaning, which is why it's difficult for native Japanese speakers to learn the difference between the sounds in English. In 1992, it was thought that this linguistic specialization occurred at about the age of 1, when infants learn their first words. But a new experiment by a team led by Patricia A. Kuhl changed that.

Kuhl's team worked with 64 six-month-old babies: 32 from Sweden and 32 from the U.S. They focused on two sounds that are pronounced slightly differently in Sweden and the U.S. I've found an example online of the Swedish sound: it's the sound the y makes in this word, "fyra":

The closet way to spell the sound out in English would probably be ee: "feera." Here's a recording of me saying it:

As you can see, it sounds different. But native Swedish speakers would probably give me the benefit of the doubt, especially since there's no "ee" sound in Swedish [correction: actually there is. See comments]. In fact, native speakers of a language will accept a wide variety of vowel sounds similar to the "true" sound, as long as it doesn't sound like another vowel in the language. This is called a "perceptual magnet," and it helps us understand language spoken in a wide variety of voices and accents.

The researchers used a computer to generate 32 variants of both the y sound and the ee sound. These variants became progressively more different from the prototypical sound, in four "rings" around each sound. This graphic might help you see how the sounds were created:


The bullseye of each target is the protoypical sound, and each ring of sounds becomes progressively more different from the original sound. The two bullseyes are each about four rings away from each other.

The babies sat on a parent's lap and listened to one of the two prototypical sounds, which was repeated every two seconds. Then occasionally a sound four rings different from the original sound was played, and if the babies looked in the direction of the sound, they got to see a toy bear bang on a drum. They quickly learned to look when the sound changed! This was the training phase.

During actual testing, the sound would often change more subtly: the new sound could vary by one, two, three, or four rings.

Now, take a look at this graph showing how often the American babies turned and looked during testing:


Babies who had been trained and tested with the Swedish y sound noticed changes much more frequently than those trained and tested with the English ee sound. The authors argue that the babies are perceiving the sounds just like adults--they perceive many sounds similar to ee as the same. In just 6 months of life, and without speaking a word, they've learned that ee is a part of English, and y isn't. The Swedish babies showed exactly the reverse pattern.

Thus, the authors reason, learning the sounds particular to a language is one of the fundamental steps of learning that language, and it's done before the child can even produce words!

Kuhl, P.K., Williams, K.A., Lacerda, F., Stevens, K.N., & Lindblom, B. (1992). Linguistic experience alters phonetic perception in infants by 6 months of age. Science, 255, 606-608.

More like this

When Jim was about 13 months old, I happened to be enrolled in a graduate level developmental psychology class. Our big term paper assignment involved observing two children at different developmental stages. I decided it would be cool to do a "longitudinal study" of Jim's language development…
The fact that infants are able to learn language without any help from adults can sometimes seem almost miraculous. Not only do children learn to speak and understand language completely on their own, active teaching of language skills seems to make almost no difference in their ability to talk.…
Discriminating against people who do not speak your language is a big problem. A new study suggests that the preferences that lead to these problems are hard-wired at a very young age. Even five-month-old infants, who can't speak themselves, have preferences for native speakers and native accents…
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…

I heard about a similar study on a TV show a while back. It used, I think, American babies and two Chinese sounds (spelled, IIRC, she and shi) which sound the same to non-Mandarin speakers. The sound was repeated over and over again and periodically changed from one to the other. The babies would perk up when the sound was changed and then get bored again as it repeated. At about 6 months they stopped hearing the difference.

It's really interesting how we learn language and can understand different accents.

By marciepooh (not verified) on 03 Oct 2007 #permalink

Here's a good time to encourage new parents to get an infant's hearing (and vision) checked at the earliest opportunity.

Yeah, but is "fyra" plural?
(sorry; couldn't resist)

By Sven DiMilo (not verified) on 03 Oct 2007 #permalink

But native Swedish speakers would probably give me the benefit of the doubt, especially since there's no "ee" sound in Swedish.

Yes, there is, in e.g. the word fira [fi:ra], which means 'celebrate'. Fyra means 'four', by the way, so there isn't much room for confusion in this case, though there is plenty in many other cases.

Also, visual information is probably more important than auditory information for distinguishing between /i/ and /y/. In a study where the participants heard [i] but saw a person producing [y], about 60 % of men and 90 % of women relied more heavily on the visual information, perceiving the sound as [y:]. Additionally, I've never heard of a language that has /y/ but not /i/, and there are theoretical reasons to expect them to be rare, as well.

Thanks for setting me straight, brtkrbzhnv. I've added a note to the text....

Here's a link to a video demonstrating the McGurk effect, which is what you're describing. The same site also demonstrates it in Norwegian.

I couldn't find it in Swedish after a few minutes of looking.

Re comment #4:
Khirgiz has /y/ (and /I/) but no /i/ as does Tzeltal.

Also, the above has been called the perceptual magnet effect...