Psycho-linguistic Experiment on Universal Semantic Components of Verbal Humor: System Description and Annotation

TL;DR

The paper addresses the challenge of identifying universal semantic components of verbal humor by proposing an empirical method to locate humor triggers. It introduces SPReadAH, a Python/PsychoPy-based self-paced reading system that logs reader decisions and word-by-word progression to differentiate humorous from non-humorous texts. An online psycho-linguistic experiment with 27 participants across 120 texts (metaphor, irony, pun, and none) provides initial data, while prior work suggests that triggers may be better captured as word groups rather than single words. This approach aims to bridge theory and data in humor research and to inform computational humor detection, with future plans including eye-tracking integration for finer trigger localization.

Abstract

Objective criteria for universal semantic components that distinguish a humorous utterance from a non-humorous one are presently under debate. In this article, we give an in-depth observation of our system of self-paced reading for annotation of humor, that collects readers' annotations while they open a text word by word. The system registers keys that readers press to open the next word, choose a class (humorous versus non-humorous texts), change their choice. We also touch upon our psycho-linguistic experiment conducted with the system and the data collected during it.

Figures (4)

• Figure 1: Navigation to the next word and class assignment.
• Figure 2: When the text is fully shown and class chosen.
• Figure 3: Option for texts with no class chosen.
• Figure 4: Evaluation of "funniness".