Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Sandbox submission: The effect of expectation management elements on subsequent predictions in sentence processing
Poster Session B, Friday, September 12, 4:30 - 6:00 pm, Field House
This poster is part of the Sandbox Series.
Benjamin Menashe1, Oren Kadosh1, Yoram Bonneh1, Michal Ben-Shachar1; 1The Gonda Multidisciplinary Brain Research Center, Bar-Ilan University
Expectation management, where speakers provide listeners with cues about upcoming sentential events, is a key component in human conversations. Expectation management often employs mirativity markers (e.g., “surprisingly” or “unexpectedly”), linguistic elements that encode unexpectedness (DeLancey, 1997). Unexpected sentential events are typically associated with a processing cost, which can manifest in greater N400 amplitude (Kutas and Hillyard, 1980). In theory, effective expectation management can alert listeners to upcoming unexpected sentential events, allowing them to modify their expectations and potentially mitigate processing costs otherwise incurred. Supporting this, Menashe, Drake and Ben-Shachar (2025) recently demonstrated that the inclusion of mirativity markers in a sentence-completion task increased the proportion of lesser-expected responses. The present study investigates whether the inclusion of mirativity markers modulates predictions during sentence reading, leading to an attenuated processing cost for subsequent unexpected words. Prior research shows conflicting evidence: Xiang and Kuperberg (2015) found that expectation-disconfirming markers attenuate N400 responses to unexpected words, while Rasenberg et al. (2020) reported no effect, indicating limited conclusive evidence on how mirativity impacts subsequent processing. In this study, we employ a dual eye-tracking and EEG setup, which allows us to measure three dependent variables that quantify processing cost: microsaccade rate, pupil dilation and N400 amplitude. We will recruit 60 native Hebrew-speaking participants, instructed to read Hebrew sentences presented word-by-word. The stimuli include 32 target words, each appearing in expected (e.g., “the king placed upon his head a crown”) and unexpected contexts (e.g., “the baker wore on his head a crown”). In pretest ratings (N=22), target words in the expected condition were rated as highly expected (<3/7) and target words in the unexpected condition were rated as highly surprising (>4/7). To control for wrap-up and spillover effects, target words appear in the seventh position followed by a sentence-final word, and the word preceding it (in the sixth position) is identical across contexts. Each sentence is preceded by either a mirativity marker (e.g., “I was surprised that”) or a neutral marker (e.g., “I saw that”). Thus, each target word appears four times overall, in a 2×2 (expectedness × mirativity) within-subject design. Additionally, 128 filler sentences of various lengths are included (comprising 50% of presented sentences). Each word is presented for 200ms with 600ms ISI. A comprehension question follows 25% of sentences. For our analysis, we will test whether the presence of a mirative marker reduces the processing cost (as measured by microsaccade rate, pupil dilation, and N400 amplitude) for unexpected target words, using a separate linear mixed-effects model for each dependent variable. We anticipate increased processing costs for target words in unexpected contexts compared to expected contexts. Importantly, we hypothesize that mirativity markers function as cues for upcoming expectation violations, enabling participants to modify their predictions and consequently reducing processing costs for target words in unexpected conditions. Overall, this study investigates whether expectation management serves as an effective strategy that allows interlocutors to adjust their expectations during online sentence processing, thereby decreasing processing costs associated with unexpected sentential events.
Topic Areas: Meaning: Discourse and Pragmatics, Reading