Visiting Scholar Manuel Bohn Presents “Beyond What Is Said: Pragmatic Inference in Young Children”

By Mischa Rosenberg, Teacher

At my snack table last summer, a child on the cusp of 3 years old, newly arrived from China, sat across from me. She was in the early stages of English language acquisition, and her communication at school was largely nonverbal, rooted in gestures. This child was particularly determined to overcome our spoken language barriers. That afternoon, after fruit and cheese had been distributed, she looked me square in the eyes and pointed to an empty spot on the table. I struggled to infer the meaning of her gesture. “Milk?” I held up the carton. She shook her head emphatically. “Apple?” I offered her the plate of slices. Another no. She grew more frustrated and I grew more flummoxed until I finally deciphered her intention. She was pointing to a spot on the table where I had previously placed a plate of graham crackers. Though the crackers were still on the snack tray, she recalled their presence on a plate of their own from earlier. I picked up the graham crackers. She gave me an exaggerated nod. As the summer session continued, she rapidly acquired English vocabulary and grammar, and I, to a much lesser extent, learned to recognize snack-time-specific words in Chinese. Our nonverbal communication became more fluid as well. And I did not again forget to place extra graham crackers on the table.

I remembered this anecdote last fall when psychologist Manuel Bohn gave a research presentation to Bing staff about how infants and young children use gestures, such as pointing, in social contexts. Such nonverbal signals can have a complex meaning that is specific to the relationship between the speaker and listener, said Bohn, who at the time was a Marie Skodowska-Curie Fellow at the Stanford Department of Psychology and has conducted some of his research at Bing.

What does pointing indicate in a conversation? What does “she” or “it” refer to in a sentence? Determining this requires going beyond what is said. The listener must make an inference about what the speaker means. The words and gestures as well as the particular social context provide clues to that meaning. For example, if a speaker points at a bicycle and says, “It belongs to Bob,” the listener must understand that “it” refers to the bicycle.  If the speaker nods toward a chair where a woman was recently sitting and says, “she has a red bicycle,” the listener must infer that the speaker means that the woman—“she”—has a red bicycle. The chair does not. In the case of the child at my snack table, pointing referred to the absent plate of graham crackers. Importantly, it only did so because we both knew (and knew that we both knew) that the plate had crackers on it before. Psychologists refer to this shared knowledge as common ground.

As Bohn explains it, “In general, common ground could be defined as the part of the social context that is relevant given the ongoing communicative interaction. To figure out what is relevant requires some reasoning about what the other person has experienced and/or knows.” In order to communicate effectively, a speaker must share common ground with a listener.

In a previous study, Bohn and his colleagues examined how 12-month-old infants used pointing to communicate about absent objects, potentially using common ground. In one experimental condition, a researcher showed an infant two plates containing different objects—perhaps one plate held balls and the other held blocks. For the infant, the balls tended to be the “highly desired objects,” whereas the blocks were less desirable. The researcher handed a ball to the infant whenever the infant pointed to the plate until the plate was empty. The researcher then briefly left the room. Upon the researcher’s return, if the infant again pointed toward the empty plate, the researcher left and returned with more balls. The infant’s pointing may have indicated a shared understanding that the empty plate used to hold balls. The infant displayed a rudimentary understanding of common ground.

In a second trial, after the researcher left the room, a new researcher entered who had not seen the contents of the plates. Would the infant continue pointing to the empty plate to request balls? Results showed that more infants pointed to the empty plate when the familiar researcher was in the room than when the new researcher was there instead. This suggests that infants are indeed considering shared experience in their encounters and rely on some form of common ground for communication.

In his research at Bing, Bohn examined how 3- to 4-year-olds and 4- to 5-year-olds integrated sources of information to determine a speaker’s intent. In one experimental condition, a picture showed a cartoon dog standing between two tables. The table to the dog’s right held an unfamiliar purple object composed of six small balls. The table to the dog’s left contained the same purple object as well as a pink, vaguely papaya-shaped object. The dog turned toward the table holding both objects, pointed, and said, “dax.” Children inferred at this time that the dog “is trying to tell me something.” If the dog had been referring to the purple object as the “dax,” it would not make sense for her to point to the table with both objects when she could have simply pointed to the table that had only the purple object. A statistically significant number of both 3- to 4-year-olds and 4- to 5-year-olds were able to correctly infer that the “dax” was the pink object.

In the next experimental condition, children were prompted to think about a speaker’s preferences and how this may establish common ground. Children in the study viewed a picture of a cartoon monkey. The monkey, like the dog, stood between two tables. One table held a yellow hexagon and the second table held an orange oval. Then the monkey pointed to an object—in this case, the orange oval—and said, “Oh, I like that!” Now, the child and the monkey shared knowledge—and common ground—about the monkey’s preference. When the monkey later asked for the “dax,” the child could use this knowledge to infer that the monkey was talking about the orange oval, because this was the one that he liked. On the other hand, what happened when a new character (a cartoon sheep) appeared on the screen and asked for the “dax”?  In this case, if children understood that what they learned about the monkey’s preference was not relevant for the sheep, they should have no clue which shape the sheep was asking for. Bohn has found that 3-year-olds begin to distinguish between what they experienced with the monkey and what they experienced with the sheep, and most 4-year-olds can make this distinction.

This study with preschool-age children builds upon Bohn’s earlier research with 12-month-old infants interacting with the known and new researcher. But whereas the infants seem to have a rudimentary sense of common ground, the children in the study were building a more sophisticated understanding.

Through examining how children make inferences based on ambiguous information, we can learn more about how a burgeoning sense of common ground is tied to successful communication, even if it is simply to request graham crackers from a befuddled teacher.