Comparing word frequency, semantic diversity, and semantic distinctiveness in lexical organization

Area of Science:

• Cognitive Psychology
• Psycholinguistics
• Computational Linguistics

Background:

• Word frequency (WF) has been a traditional predictor of lexical behavior.
• Previous research indicated contextual and semantic diversity measures outperform WF.
• Recent findings suggested WF might account for greater variance than diversity measures.

Purpose of the Study:

• To re-evaluate the predictive power of WF against updated versions of the semantic distinctiveness model (SDM).
• To address limitations in prior studies, specifically the exclusion of recent SDM developments.
• To compare context-based versus repetition-based accounts of lexical organization.

Main Methods:

• Compared the predictive accuracy of WF with various versions of the SDM on lexical decision and naming data.
• Utilized data derived from a consistent corpus to avoid confounding factors.
• Focused on the predictive variance explained by later iterations of the SDM.

Main Results:

• Early SDM versions showed lower predictive power than WF when derived from different corpora.
• Later versions of the SDM significantly outperformed WF in accounting for unique variance.
• The SDM demonstrated superior explanatory power for lexical decision and naming tasks.

Conclusions:

• Context-based models, particularly advanced SDM versions, provide a more robust explanation of lexical organization.
• Repetition-based accounts like WF are less effective in explaining complex lexical behaviors.
• The findings support the theoretical advantage of semantic and contextual diversity over simple word frequency.