Infants Update Their Guess About the Rules of Language as They Get More Information
11-month-olds are influenced by the order in which they hear words that reflect rules about how sounds are combined in their language. If the words occur in an order where many adjacent words can mislead infants about the rule (e.g., 2 words starting with p accidentally occur next to each other), infants don’t learn the rule. This finding tells us that infants are updating their best guess about the rules of their language as each new piece of information comes in.
It Only Takes 4 Words of Input for Infants to Learn a New Language Rule
11-month-olds are able to learn a new rule about how sounds are combined in their language from just four words of input. A question that arises from this research is, if infants learn so quickly, why aren’t they done learning language by the time they are age two years? Current research in the Tweety and Tigger Labs aims to tease apart what infants learn from a brief learning session and what they remember and use as they are becoming more mature language users.
Infants Pay More Attention to Language Patterns When They Think They Can Learn Those Patterns
17-month-olds were allowed to listen for as long as they wanted to lists of masculine and feminine words from a foreign language. For some of the infants, the fact that there were two types of words (masculine and feminine) with different endings was more apparent, and for other infants the pattern was more obscure. Infants who had the more transparent pattern listened longer, suggesting that infants devote more attention to language patterns when they sense that they are able to learn these patterns.
Younger Babies Can Detect Patterns in Music that Older Babies Cannot
Research from another laboratory recently showed that 7-month-olds were unable to learn an AAB or ABB pattern (with repetition at the beginning or end) when the pattern was produced with musical notes, but they could learn the pattern when it was produced with syllables. Does this finding mean that there is something special about language? Or could it be that experience with music makes repetition less noticeable. For example, most music is "smooth", so that if you're hearing one note, the next note is likely to be either the same or just one step up or down. If babies have learned about the smoothness of music, they may come to believe that there is nothing special about repetition, because it is just part of the musical system. This view suggests that babies who have not yet heard enough music to infer the smooth structure may find repetition more interesting and therefore learnable. To test this suggestion, we tested 4-month-olds on the AAB and ABB musical patterns that had eluded 7-month-olds in the earlier study. Counter to the claim that there is something special about language, they learned these patterns easily!
Language Patterns and Rule Learning in 9-month-old Babies
A number of researchers, including the famous Noam Chomsky, have suggested that babies might be biased to learn patterns that occur in real languages. Languages of the world have many rules about which syllables in words should be stressed. Many languages have a rule that syllables ending in a consonant should be stressed. No languages have a rule that says that syllables starting with 't' should be stressed. Consistent with that suggestion, research from our lab has shown that 9-month-olds are able to quickly learn the 'ends in consonant' rule, but not the other rule. However, we have also found that younger babies (7-month-olds) are able to learn the rule that does not occur in human languages. Therefore, it appears that the forms languages take are shaped by what is easy for language users to perceive and produce, and do not come "prewired" in babies.
Babies Know That They're Learning, Even Before They've Learned
One mystery about how babies learn is how they choose the learning problems to tackle in the first place. In any given situation, a baby could focus on any number of aspects of the environment to explore. How do they avoid wasting their time by focusing on aspects where there is nothing to learn, or learning problems that they are not yet ready to solve? Recent research from our lab suggests that infants are able to detect whether or not they are learning and pay more attention to aspects of their environment that are learnable. We let 17-month-olds listen to a language that other babies had learned or to one that other babies (and adults) found impossible to learn. When babies showed that they were no longer interested, the experiment ended. The babies who heard the learnable language listened much longer than the other group and began showing increased interest after they had heard only about 45 seconds of the language. Infants in the earlier study who were tested on their learning, not on their interest level, took about 3 minutes to learn the language. Therefore, it seems that babies know they're making progress on a learning problem in much less time than it takes them to actually solve the problem.
Babies Detect Multiple Patterns in the Same Input
Nine-month-old infants listened for 2 min. to 3-syllable 'words' with an AAB pattern, in which the first syllable repeats (e.g., laylaydee, weeweejay, etc.). The words either had another pattern embedded or not. The embedded pattern was that all of the words ended in the syllable 'dee'. Infants who heard the words without the embedded pattern were able to recognize as familiar new AAB words when we tested their interest. But infants who heard words with the embedded pattern were only able to recognize new words with the embedded pattern, not the more general AAB pattern (Gerken, 2006). In a later study, just 3 examples of words that had just the more general AAB pattern mixed in with the embedded pattern were enough to cause infants to recognize new words with the more general AAB pattern. This result suggests that they were considering both patterns all along and just needed a little evidence to tip them from favoring the embedded pattern to the more general pattern.