Linguistic consequences of event segmentation in visual narratives: implications for prominence

ABSTRACT

This experiment uses comic-type visual narratives to investigate effects of event segmentation on the representation of causal events (with an agent and a patient) in discourse. Prior work leaves open the question of whether the prominence of entities and events in mental models of discourse can be dissociated. By presenting participants with one-panel vs. two-panel segmentations of the same event, we tested whether differences in event segmentation boost the prominence of the patient, the consequence event, or both, as reflected in people’s choices about what is mentioned next. The results show participants are more likely to mention consequences in two-panel than one-panel conditions, indicating that panel segmentation influences the cognitive prominence of event-level representations. However, we find no clear evidence for segmentation influencing the cognitive prominence of entities. This suggests language users separately track expectations about who is mentioned next and expectations about what kind of event is described next.

KEYWORDS:

• Discourse structure
• event segmentation
• visual narrative
• prominence
• causality

Acknowledgements

Many thanks to David Cheng-Huan Li for drawing the stimuli. Thanks are also due to Emily Fedele, David Li, Josephine Lim, Tyler Anne Isaman, Michal Meyers and Iris Ouyang for help with data collection and analysis. The content is solely the responsibility of the author and does not necessarily represent the official views of the National Institutes of Health.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 The “other” category includes subsequent events not directly causally connected to the depicted event (e.g. “and then we went to go eat dinner”), events occurring before the depicted event (e.g. “and then Jen decided to play a trick on Jane”), etc. Continuations analyzed as “unclear” are also in this category. Recall that the task was intentionally open-ended, to avoid biasing participants” continuations. An unsurprising consequence is that both conditions yield approximately 25% “other” continuations. The proportion of “other” continuations in the two conditions does not differ (see Results) and thus does not impact the main claims of this paper.

2 Use of pronouns vs. NPs is not relevant: Prior work indicates this is driven by antecedent grammatical role (with pronouns preferring subjects, e.g. Fukumura & Van Gompel, 2010; Kehler & Rohde, 2013) – which we do not manipulate. Indeed, as expected, more pronouns were produced in both conditions when referring to the agent rather than the patient.

3 We excluded as uncodable those continuations where the first-mentioned referent was something other than a singular third-person expression referring to the agent/patient, because the purpose of the entity-based analysis is to assess the relative prominence of the agent and the patient.

Additional information

Funding

Research reported in this publication was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development of the National Institutes of Health [R01HD061457].