A new tool for equating lexical stimuli across experimental conditions

Area of Science:

• Cognitive Psychology
• Psycholinguistics
• Computational Linguistics

Background:

• Lexical characteristics of word stimuli can introduce confounds in cognitive psychology research.
• Controlling for these lexical properties is crucial for accurate experimental results.
• Existing methods like randomization may lead to unbalanced stimulus lists.

Purpose of the Study:

• To introduce LIBRA (Lexical Item Balancing & Resampling Algorithm), a MATLAB-based toolbox.
• To provide a method for generating closely equated stimulus lists based on user-defined lexical properties.
• To offer an efficient alternative to random assignment for controlling confounds.

Main Methods:

• LIBRA utilizes a genetic algorithm for inter-list balancing.
• It includes a tool for filtering and trimming word lists before balancing.
• The toolbox allows balancing on any number of user-specified lexical properties.

Main Results:

• LIBRA generates stimulus lists that are closely equated on specified lexical properties.
• It ensures minimal differences between experimental conditions, reducing bias and noise.
• The toolbox offers greater efficiency compared to manual balancing or pure randomization.

Conclusions:

• LIBRA effectively controls for lexical confounds in cognitive psychology research.
• The toolbox enhances experimental reliability by ensuring balanced stimulus sets.
• LIBRA provides a user-friendly and efficient solution for researchers in psycholinguistics and cognitive psychology.