An amorphous model for morphological processing in visual comprehension based on naive discriminative learning.

Area of Science:

• Cognitive Science
• Computational Linguistics
• Psycholinguistics

Background:

• Inflectional and frequency effects are key phenomena in morphological processing.
• Existing models often require complex representations for inflections and word forms.

Purpose of the Study:

• To propose and validate a 2-layer symbolic network model based on Rescorla-Wagner equations.
• To demonstrate the model's ability to account for various morphological processing effects without specialized representations.

Main Methods:

• Developed a 2-layer symbolic network model.
• Conducted experiments in Serbian on sentential reading and lexical decision.
• Tested the model against established findings in morphological processing literature, including frequency, family size, and affix productivity effects.

Main Results:

• The model successfully replicates inflectional paradigmatic effects in Serbian without assuming separate inflectional representations.
• Frequency effects for complex words and phrases emerge without whole-word representations.
• Family size effects are observed for simple, derived, and compound words.
• The model accounts for affix productivity effects and predicts faster processing for new words with productive affixes.
• English phrasal paradigmatic effects are also modeled, demonstrating crosslinguistic scope.

Conclusions:

• A simple, naive discriminative learning approach can explain a wide range of morphological processing phenomena.
• The proposed model offers a parsimonious explanation for word processing complexities, challenging the need for intricate representational assumptions.
• The findings suggest that inflectional and phrasal paradigmatic effects have crosslinguistic relevance.