Presentation

Yaqi Xu¹, Melissa Troyer², Rupali Limachya³, Suzanne R. Jongman, Katrien Segaert³, Kara D. Federmeier²; ¹UC Santa Cruz, ²University of Illinois, Urbana-Champaign, ³University of Birmingham

A growing body of literature shows that people engage in predictive behavior during ongoing language comprehension processes. Previous studies have shown facilitated word processing for predictable compared to unpredictable words as evidenced from online measures including reaction times, eye-tracking, and event-related brain potentials (ERPs). The current study aims to understand the effect of language production on prediction observed during sentence processing. Although there is an abundance of evidence supporting prediction while reading sentences, only a few studies have looked at production and comprehension together. In our blocked design study, we recorded EEG/ERPs while participants read sentences in one block (the comprehension block) and read sentences with a production task in another (the production block). In the production block, the critical last word was replaced with a line drawing and participants were required to name the picture out loud. In both blocks, the critical word/picture was one of three types: the expected ending (i.e., best completion), a within-category (i.e., related) anomaly, or a between-category anomaly. An example sentence would be: “They wanted to make the hotel look more like a tropical resort. So, along the driveway, they planted rows of …” For the example sentence, an expected word would be “palms,” which is the best fit given the sentence context. A related anomaly would be “pines,” which is considered to be in the same general category of “trees” and shares many similar features. A between-category anomaly would be “roses,” which is not a type of tree. These stimuli have previously been shown to elicit related-anomaly effects suggestive of predictive processing (Federmeier & Kutas, 1999; Federmeier & Kutas, 2001). Half of the participants completed the comprehension block first and the other half did the production block first. Our analysis focused on differences in the amplitude of the N400 component to critical words in the comprehension block. The N400 is a negative going waveform feature peaking around 400 ms after stimulus onset that has been associated with semantic processing. We compared the comprehension-first group to the production-first to investigate the possible influence of prior production on later comprehension trials. We hypothesized that if prediction and production mechanisms are linked, then we should observe relatively larger N400 prediction effects when participants had recently been producing language—i.e., a greater reduction in N400 amplitude to within-category words compared to between-category words in the production-first group compared to the comprehension-first group. Our preliminary results suggest that this is the case. Such a finding would support claims that neurobiological pathways involved in production are also active during prediction.

Topic Areas: Meaning: Combinatorial Semantics, Meaning: Lexical Semantics