Word Learning In Older Children And Adults: When Does Specialization Occur?

By the time you reach adulthood, you are estimated to know as many as 60,000 words. Despite this, vocabulary learning in your native language might feel effortless over the years. What kind of processes allow for the learning of so many words? One debate in word learning is whether we learn words by using domain-general or domain-specific learning processes. Domain-general processes are processes that are not unique to language, and that have developed to support a range of abilities: Examples are attention, memory, and the ability to detect statistical regularities in our environment.

In contrast, domain-specific processes are mental processes that are only used in a particular domain and/or have developed specifically to support that skill. In the context of word learning, it has been argued that our ability to learn words quickly relies on domain-specific processes.

Evidence for the latter account comes from symbolic flexibility: In word learning, symbolic flexibility refers to the ability to map a range of symbols as a label for a new object. For example, a symbolically flexible learner might accept any type of auditory sound to refer to a new meaning. In contrast, a symbolically inflexible learner might only accept what sounds like a word as a label. Studying symbolic flexibility thus can help us better understand to what extent and at what time point a child undergoes “specialization” during word learning.

Previous research has shown that toddlers that learn a spoken language quickly home in on preferring words over other sounds or hand gestures, and that they find it easier to learn a new word if the novel word follows sound patterns that are common in their native language. However, this research often has been limited to infants and toddlers. As a result, these experiments only include a small number of trials where words are taught explicitly by an adult, as children are too young to participate in longer sessions or to learn words more independently.  However, this misses something important about word learning—children must learn many words, and crucially they must do so in cluttered and under-constrained contexts.  For example, they may be confronted with a whole array of toys and asked to find the “slinky”, with no overt cues as to which it could be.

Thus, we were interested in exploring symbolic flexibility in older children’s word learning, and in situations where children must figure out the meanings without any obvious clues. In Experiment 1, children aged 6-to-8-years participated in a word learning experiment: On each trial, half of the children heard a novel word (e.g., GOBA) and saw three objects on a computer screen; these children were in the word condition. The other half of children—the ones in the sound condition—heard a non-linguistic sound (e.g., a beep) and saw three objects. Across trials, each auditory stimulus (either a word or a non-linguistic sound) consistently co-occurred with one object. This co-occurrence allowed children to extract the correct meanings of the auditory stimuli, even as they never received feedback as to whether they clicked the correct object.

Based on previous findings, one might predict that children in the word condition would have an easier time learning which word mapped onto which object than children in the sound condition, thus showing evidence of being symbolically inflexible.

Children were able to learn the labels in both conditions. Moreover, they learned labels for meanings equally well in each condition. This suggests that it was not easier for them to map words onto objects than non-linguistic sounds: Children acted like symbolically flexible learners.

In Experiment 2, we asked the same question, but this time in adults. As before, adults either learned to map a novel word or a non-linguistic sound onto a novel meaning. In contrast to children, adults had an easier time learning new symbols when they were novel words (e.g., GOBA) than when they were non-linguistic sounds (e.g., beeps). To be clear, adults were always able to learn that an auditory symbol (either a novel word or a beep) mapped onto a novel object, but it was harder doing so with a non-linguistic sound! Adults were less symbolically flexible than the children.

Overall, these two experiments suggest that symbolic flexibility decreases over development, but that this process might be both more extended than previously thought and partial at best. Why do we find that children are symbolically flexible whereas previous studies do not? It is possible that how word learning is tested influenced our interpretation of how symbolically flexible children are. In prior studies on younger children, symbolic flexibility is defined as the ability to map all types of labels onto a novel meaning. That is, a toddler is referred to as symbolically inflexible if they only learn to map a novel word onto a meaning (but no other types of symbols). However, in older children and adults, this definition does not apply, as people are generally able to form associations between any types of items. Thus, we defined symbolic inflexibility as the relative advantage of mapping a word onto an object versus mapping a different type of symbol.

Based on our data, we cannot know exactly when symbolic flexibility decreases. But what factors might contribute to this change in symbolic flexibility? For one, adults have had much more experience with hearing words and learning novel words; as a result, they might either have become better at encoding words or they might be less likely to accept other types of sounds. If it is true that people simply learn to be better at memorizing and storing novel words, this might suggest that word learning relies on a more domain-general ability, but that its efficiency can change as the result of perceptual learning.

Words are not automatically “special”; instead, they become special (or preferred) during development. This also leads to a decrease in symbolic flexibility, though people appear to retain some flexibility even in adulthood by which time they have been exposed to millions of words. Differences in word learning might be more driven by differences in processes such as attention or working memory that make it easier—or harder—to form an association between a word and a meaning.

These findings are described in the article entitled Symbolic flexibility during unsupervised word learning in children and adults, recently published in the Journal of Experimental Child Psychology.