Emoji-based semantic representations for abstract and concrete concepts

Area of Science:

• Cognitive Science
• Neuroscience
• Computational Linguistics

Background:

• The semantic system is increasingly understood as distributed across brain regions and multimodal.
• Traditional feature norms rely on verbal strings, potentially missing non-verbal semantic information.
• Abstract and concrete concepts may differ in their reliance on non-verbal features.

Purpose of the Study:

• To propose and validate a novel method for collecting non-verbal semantic features using emoji.
• To investigate how emoji representations differ for abstract versus concrete concepts.
• To analyze the cognitive strategies employed when using emoji for semantic representation.

Main Methods:

• A crowdsourcing task where participants used emoji to represent 300 English nouns.
• Content analysis by multiple annotators to classify cognitive strategies used in emoji representations.
• Comparison of emoji-based semantic features with traditional verbal feature norms.

Main Results:

• Abstract concepts, compared to concrete ones, were represented using more emoji, including more facial/emotional emoji.
• Emoji representations for abstract concepts were less stable and less shared among users.
• Abstract concepts showed increased use of figurative (metaphor) and linguistic (rebus) strategies.
• Emoji-based representations correlated less strongly with traditional verbal feature norms.

Conclusions:

• Emoji offer a valuable tool for capturing the multimodal and non-verbal aspects of semantic representations.
• Distinct representational characteristics exist for abstract versus concrete concepts when using emoji.
• This approach highlights limitations of purely verbal feature norms for understanding semantic memory.