Dr. Jennifer Arnold’s lab seeks to understand the cognitive mechanisms behind using language in context. The ability to use language is a cognitive feat that defines what it means to be human. In addition to managing extensive vocabularies, sound systems and grammatical systems, people also manage to flexibly choose words and structures appropriate for the context. For example, I could say she, the cat, or Greta (among other options) to refer to my cat. People tend to use shorter terms like pronouns when the context makes the reference expected or otherwise easy to understand. However, this also creates an ambiguity that listeners must resolve in order for communication to be successful. How do people do this?

In a recent line of studies, Dr. Arnold and her team have shown that people pay attention to both the linguistic context and social cues like gazing or pointing (Nappa & Arnold, 2014). Subjects watched videos of a woman (see figure on right) telling a story about two puppets, e.g. Panda Bear is having lunch with Puppy. He… Since both characters are boys, the pronoun ‘he’ is ambiguous. Listeners have a preference to link the pronoun to the first character (Panda Bear), but also use gaze as a cue to reference. Both gaze and linguistic context are partial cues.

However, people differ in how much weight they place on each cue. Some people follow the linguistic context more consistently, and it turns out that this is correlated with how much people read – even though the task itself involved spoken language. This effect occurs for native English-speaking adults, second-language speakers of English, and children. This suggests that reading exposure affects discourse processing strategies even for spoken language. In an ongoing National Science Foundation funded project (2017-2020), Dr. Arnold is designing new cartoon stimuli to test this effect more extensively (see figure on left). A core question is whether these linguistic and gazing effects are related to listeners predictions about which character is most likely to be mentioned. Current theories suggest that predictability guides pronoun comprehension, but this hypothesis has not been broadly tested.

A related question is whether speakers consider predictability when they decide whether to use a pronoun (he, she) or a more specific form, like a name. For example, after a sentence like The Duchess gave a painting to the Duke, people expect the speaker to go on to talk about the Duke. Does this also make the speaker more likely to use a pronoun? To test this question, Elise Rosa (Ph.D. 2015) designed a naturalistic story-telling task about the characters shown (see figure on right), which has been used in multiple lab projects. This NSF-funded project (2013-2017) revealed that predictability does affect the use of pronouns, but only for some verbs. Corpus analyses suggest that the verb effect is due to differences in the frequency with which people tend to continue talking about one character or the other. Overall, these results support models where language users use predictions for both production and comprehension.


Dr. Jennifer Arnold is a Professor in the Cognitive Psychology Program within the Department of Psychology and Neuroscience at UNC Chapel Hill. Her work examines the psychology of language, with the goal of understanding the mental steps that underlie our ability to speak and understand. Learn more about her research online.


Comments are closed.