09:00 - 10:30
Parallel sessions 4
09:00 - 10:30
Room: HSZ - 7E02
Chair/s:
Susanne Dietrich
Language comprehension often proceeds with remarkable speed, yet successful communication depends on the ability to slow down, revise, and adapt when input is ambiguous, unexpected, or inconsistent. This symposium brings together perspectives from neuroscience, psycholinguistics, and developmental research to examine how temporal flexibility supports coherent comprehension. “Time to think” is not a failure of processing but an adaptive resource: when comprehension is challenged, listeners and readers adjust the pace of processing to integrate conflicting cues, resolve ambiguity, and update mental representations. The symposium opens with a neurobiological perspective, showing how electrophysiological activity supports the processing of acoustic and abstract temporal structures in auditory-verbal stimuli. Studies from sentence comprehension and communicative interaction demonstrate that brain activity not only synchronizes with current stimuli but also aids the management of upcoming input through temporal estimation and prediction. These processes rely on structures such as the basal ganglia and pre-SMA, providing neural scaffolding for adaptive control. Neuroimaging evidence further shows that presupposition failures engage these circuits, indicating that discourse-related reinterpretation depends on adaptive gating and slowing mechanisms. The third talk presents psycholinguistic evidence on negation and pragmatic inference. Negation is rarely purely logical; comprehenders use it as a cue for pragmatic reasoning, revising mental models and integrating contextually relevant alternatives. Less felicitous contexts increase processing time, reflecting the additional effort required to construct a coherent interpretation. Next, a developmental perspective uses eye-tracking data from children learning German and Czech. Younger children struggle to integrate multiple linguistic cues for thematic role assignment, and their ability to reanalyse heuristics depends on language-specific features, such as word order versus case marking. Finally, ambiguity resolution in discourse is examined, showing how verb causality and adjective semantics shape pronoun interpretation. Comprehenders dynamically reweight cues integrating earlier expectations. Together, the contributions illustrate how slowing down, revising, and flexibly reallocating processing resources are central to achieving robust and coherent comprehension under uncertainty.
Submission 147
Not so Fast: Cognitive and Pragmatic Demands of Negation
SymposiumTalk-03
Presented by: Francesca Capuano
Francesca Capuano
University of Tübingen, Germany
Negation is a central yet cognitively demanding feature of language that highlights the need for effortful slowing and reanalysis in comprehension. Understanding a sentence like “The door is not open” involves suspending a default affirmative interpretation and re-evaluating the state of affairs. According to the two-step model of negation processing, the initial stage involves activation of the affirmative meaning, followed by inhibitory control and reinterpretation. These inhibitory mechanisms appear to overlap with general cognitive control systems. The reinterpretation and selection of an appropriate alternative depend on pragmatic reasoning processes, which form the focus of the present work. Listeners infer why a speaker chose to use negation, considering communicative intent and discourse relevance. According to pragmatic principles, negation is most felicitous when it corrects a false presupposition or contrasts with an expected state of affairs, contexts in which comprehension has been shown to proceed more smoothly. Even without explicit context, listeners infer plausible alternatives rather than strict logical opposites, highlighting the pragmatic nature of negation. This study extends these pragmatic insights to large language models (LLMs). Like humans, LLMs often struggle with negated statements, appearing anchored to affirmative meanings, particularly when no salient pragmatic alternative is available. However, in offline cloze tasks, humans tend to complete negated sentences with particularly similar alternatives, reasoning pragmatically about what alternative state would make the negated statement relevant. Only conversationally trained chat models approximate human-like behaviour, indicating that effective negation processing might depend on pragmatic reasoning and discourse sensitivity acquired through dialogic experience.