Estimating affective word covariates using word association data.

Area of Science:

• Psycholinguistics
• Computational Linguistics
• Affective Science

Background:

• Affective word ratings are crucial in psycholinguistic research.
• Traditional methods of obtaining these ratings are time-consuming.
• Computational approaches using semantic similarity offer a more efficient alternative.

Purpose of the Study:

• To computationally generate affective norms for 14,000 Dutch words.
• To explore the utility of word association data for estimating semantic similarity.
• To compare the effectiveness of Orientation towards Paradigm Words and k-Nearest Neighbors methods for predicting affective ratings.

Main Methods:

• Extracted semantic similarity from Dutch word association data.
• Utilized Orientation towards Paradigm Words and k-Nearest Neighbors algorithms.
• Predicted valence, arousal, and dominance for 14,000 Dutch words.

Main Results:

• Achieved very high correlations between predicted and human affective ratings using Orientation towards Paradigm Words.
• Observed even higher correlations when employing the k-Nearest Neighbors method.
• Demonstrated the efficacy of word association data in computational norm generation.

Conclusions:

• Word association data can be effectively used to computationally generate affective word norms.
• The k-Nearest Neighbors method shows superior performance in predicting affective dimensions.
• This approach offers a scalable and efficient alternative to traditional rating procedures.