Discrimination in lexical decision

Area of Science:

• Psycholinguistics
• Cognitive Science
• Computational Linguistics

Background:

• Lexical decision tasks are crucial for understanding word recognition.
• Classical measures like frequency and form similarity have limitations in predicting response times.
• Naive Discriminative Learning (NDL) theory offers an alternative framework for language processing.

Purpose of the Study:

• To introduce novel discrimination-based indicators for language processing derived from NDL theory.
• To compare the predictive power of these new measures against traditional lexical-distributional measures.
• To investigate the role of morphological segmentation in masked priming using lexical decision tasks.

Main Methods:

• Re-analysis of a large dataset from the English Lexicon Project.
• Conducting two new masked priming studies utilizing lexical decision tasks.
• Comparing discrimination-based predictors with frequency and form similarity measures.

Main Results:

• Discrimination-based predictors significantly outperformed lexical-distributional predictors in predicting lexical decision latencies.
• Comparable priming effects for 'corner' and 'cornea' masked primes challenge early obligatory morpheme segmentation.
• Results align with NDL and Word and Paradigm theories, which do not consider morphemes as primary units.

Conclusions:

• Discrimination-based measures provide superior predictions for lexical decision tasks compared to traditional methods.
• The findings do not support obligatory early morpheme segmentation in reading.
• Higher spelling proficiency and vocabulary size correlate with more efficient use of orthographic information and lexical competition handling.