Toward a scalable holographic word-form representation

Area of Science:

• Cognitive Science
• Psycholinguistics
• Computational Linguistics

Background:

• Verbal tasks like masked priming, lexical decision, and word naming are often explained by word-form similarity.
• Existing models lack a unified approach to representing word forms and their similarity measures across different tasks.

Purpose of the Study:

• To bridge the gap between word-form representations and similarity measures used in verbal tasks.
• To explore holographic reduced representations (HRRs) for visual word forms.
• To evaluate these representations' ability to explain masked priming, lexical decision, and word naming data.

Main Methods:

• Utilized holographic reduced representations (HRRs) to model visual word forms.
• Developed a representation encoding word-internal letter groups relative to word-terminal letter groups.
• Integrated the proposed representation with the BEAGLE model of lexical semantics.

Main Results:

• A novel word-form representation successfully predicted qualitative patterns in masked priming.
• This representation also accounted for observed data in lexical decision and word naming latencies.
• Integration with the BEAGLE model expanded its scope to encompass a broader range of verbal tasks.

Conclusions:

• A unified approach to word-form representation can account for diverse verbal task phenomena.
• The proposed HRR-based method offers a promising way to link visual form processing with lexical semantics.
• This research advances computational models of language processing by integrating form and meaning representations.