Related Experiment Video
Updated: Aug 16, 2026

Transcranial Direct Current Stimulation (tDCS) of Wernicke's and Broca's Areas in Studies of Language Learning and Word Acquisition
Published on: July 13, 2019
Designing vocabulary multiple-choice tests for exploring word frequency effects in distributional semantics
Joseph P Levy1, John A Bullinaria2
1Independent Researcher, London, UK. joe.levy@cantab.net.
Designing specific vocabulary multiple-choice question (VMCQ) tests for distributional semantic vectors (DSVs) yields better results than human-centric tests. This study introduces improved VMCQ methods for evaluating DSVs, particularly word frequency effects.
Area of Science:
- Computational Linguistics
- Natural Language Processing
- Cognitive Science
Background:
- Distributional semantic vectors (DSVs) represent word meanings in vector spaces, crucial for AI and psychology.
- Vocabulary multiple-choice question (VMCQ) tests are common for evaluating DSVs, often adapted from human linguistic tests.
Purpose of the Study:
- To demonstrate that VMCQs designed specifically for DSVs outperform human-centric tests.
- To develop improved VMCQ scoring and evaluation methods for DSVs.
- To explore the impact of word frequency on DSV performance using novel VMCQ approaches.
Main Methods:
- Developed an improved VMCQ scoring method for statistically reliable DSV evaluation.
- Introduced new evaluation methods to analyze word frequency effects on DSVs.
- Tested the proposed VMCQ approach on nine diverse DSVs.
Main Results:
- DSVs benefit from VMCQ designs tailored to their specific properties.
- The new scoring method enhances the reliability of DSV performance assessment.
- The study identified an effective VMCQ strategy for exploring word frequency effects in DSVs.
Conclusions:
- Tailored VMCQs offer superior evaluation of distributional semantic vectors compared to adapted human tests.
- The developed methods provide more reliable insights into DSV performance and word frequency impacts.
- This research advances the methodology for assessing and understanding semantic vector representations.
Related Concept Videos
Expected Frequencies in Goodness-of-Fit Tests
Language and Cognition
Group Design
Components of Language
Frequency-dependent Selection
Choosing Between z and t Distribution

