A practical primer on processing semantic property norm data

Area of Science:

• Cognitive Psychology
• Computational Linguistics
• Psycholinguistics

Background:

• Semantic property listing is vital for creating semantic property norms used in cognitive modeling and stimuli creation.
• Methodological choices in property listing tasks significantly impact the quality and nature of derived semantic measures.
• Existing research often overlooks the methodological details of property listing, hindering cross-study comparability.

Purpose of the Study:

• To provide a practical primer on collecting and processing semantic property norms.
• To discuss methods for eliciting semantic properties and deriving meaningful representations.
• To propose a transparent processing pipeline for enhanced comparability across studies.

Main Methods:

• Discussing the influence of instructions and test context on property elicitation.
• Detailing property preprocessing techniques such as lemmatization.
• Exploring property weighting and relationship encoding using ontologies.
• Demonstrating a processing pipeline with intrinsic and extrinsic measures.

Main Results:

• The proposed pipeline enhances transparency and comparability of semantic property norms.
• Methodological choices demonstrably impact measures of reliability, property count, categorization, and semantic similarity.
• The primer offers solutions to long-standing issues in semantic norm generation.

Conclusions:

• Standardizing the collection and processing of semantic property norms is essential for robust cognitive research.
• The presented methods and pipeline facilitate the development of high-quality, comparable semantic datasets.
• This work empowers researchers to overcome previous limitations in creating semantic property norms.