MATLAB Toolbox for Reading Speed Tests Eliminates Memorization Bias

A new open-source MATLAB toolbox from researchers Daniel P. Spiegel and Romain Bachy tackles a persistent problem in vision science and psycholinguistics: accurate reading speed measurement. Traditional static reading charts suffer from memorization effects on repeated testing, and oral recitation introduces speech-motor confounds that mask true information processing speed. The toolbox implements and extends the Perrin, Paillé, and Baccino (2014) sentence generation paradigm for English corpora.

At its core, the system uses a semantic ontology and a 'proto-truth' logic to autonomously generate thousands of unique, grammatically simple sentences, each with an unambiguous truth value. This eliminates the need for repeated use of the same sentences, preventing memorization bias. The toolbox goes beyond the original Maximum Reading Speed (MRS) measurement scope by introducing band-pass psycholinguistic filtering and specific logic to resolve semantic ambiguities unique to English.

The software package is designed as an open platform for the scientific community to validate and refine. It includes ancillary files like example sentences, a standard operating procedure, and MATLAB source code. This allows researchers to generate custom sentence sets tailored to specific experiments, ensuring standardized and reproducible reading speed assessments across different labs and populations.