The exception of humour: Iconicity, Phonemic Surprisal, Memory Recall, and Emotional Associations (2502.01682v1)

Abstract: This meta-study explores the relationships between humor, phonemic bigram surprisal, emotional valence, and memory recall. Prior research indicates that words with higher phonemic surprisal are more readily remembered, suggesting that unpredictable phoneme sequences promote long-term memory recall. Emotional valence is another well-documented factor influencing memory, with negative experiences and stimuli typically being remembered more easily than positive ones. Building on existing findings, this study highlights that words with negative associations often exhibit greater surprisal and are easier to recall. Humor, however, presents an exception: while associated with positive emotions, humorous words also display heightened surprisal and enhanced memorability.

• The paper finds that humor, similar to negative stimuli, exhibits high phonemic surprisal and enhances memory recall, challenging traditional valence-based memory theories through analyses of psycholinguistic datasets.
• Results show humor strongly correlates with iconicity and suggest phonemic surprisal can be a critical feature for refining machine learning models for sentiment analysis.
• The study suggests future research directions include exploring cross-linguistic patterns and dissecting different types of humor to further understand language-related cognitive processing.

The Exception of Humor: Surprisal, Memory, and Emotional Associations

This paper by Kilpatrick and Flaksman examines the intricate interplay between phonemic surprisal, emotional valence, and memory recall, with a unique emphasis on humor as a cognitive exception. Drawing from a meta-paper approach, the research explores the hypothesis that humor, despite its positive emotional valence, shares characteristics with, and sometimes diverges from, negatively valenced stimuli in terms of phonological surprisal and memory recall.

Primarily, the paper underscores the established association between phonemic bigram surprisal and memory recall. Prior literature suggests that higher phonemic surprisal increases memorability due to unpredictable phonetic sequences eliciting stronger cognitive engagement. Additionally, emotional valence, especially negative experiences and stimuli, has been shown to enhance recall accuracy, attributed to evolutionary significance in threat detection and avoidance. Words associated with negative valence typically manifest increased phonemic surprisal, reinforcing their memorability. However, humor emerges as an exception in this framework.

Typically linked to positive emotions, humor surprisingly displays increased phonemic surprisal and is more memorable, akin to negative stimuli. This suggests that humor engages unique cognitive processes. Kilpatrick and Flaksman capitalize on existing psycholinguistic methodologies that measure amusiveness and phonological predictability to test their hypotheses, using datasets such as the SUBLEX-US corpus and the CMU Pronouncing Dictionary to conduct their analyses. They employ multiple linear regression models to evaluate associations among phonemic surprisal, emotional valence, iconicity, and humor scores.

The results reveal significant findings: negative valence corresponds with increased surprisal, but humor also demonstrates a notable positive correlation with surprisal. Interestingly, humor shows strong alignment with iconicity, where humorous words often possess distinct phonological cues, akin to iconic words. The paper highlights statistical models indicating robust relationships between negative valence, surprisal, and memory recall, corroborating the negativity bias paradigm. Moreover, humor displayed a significant role in enhancing memory recall, challenging traditional valence-based memory theories.

The authors suggest that these findings implicate humor as operating through cognitive mechanisms resembling those of negative stimuli. This aligns with theoretical models like Suls's Two-Stage Model for humor appreciation, wherein initial cognitive engagement due to the unexpected punchline transitions into a positive emotional resolution, thus unveiling a dual cognitive-emotional dynamic.

The implications of the paper extend to theoretical advancements in understanding the phonological markers of memory and emotion, indicating potential applications in enhancing machine learning algorithms for sentiment analysis. Phonemic surprisal, as an econometric measure of unpredictability, may serve as a critical feature in refining emotion and sentiment detection models.

Future research directions offered by the authors include exploring cross-linguistic patterns to ascertain the universality of these findings across different languages and cultural contexts. Additionally, dissecting types of humor could yield insights into their cognitive processing nuances relative to phonological and emotional elements. Such inquiries promise to enrich the comprehension of language-related cognitive processes, extending the domain's foundational knowledge.

In conclusion, this paper provides compelling evidence demonstrating the exceptionality of humor within the context of memory, plunging into the nuanced intersection of iconicity, emotion, and phonological structure, thereby offering a sophisticated understanding of cognitive and emotional processing in language.