A number of affective word covariates was computationally estimated using word association data.
In short, the procedure first obtains semantic similarity indices from overlap in associations; it then predicts affective values using a word's similarity to words for which these values are already known. Accuracy of obtained estimates is calculated through correspondence with human ratings.
The first datafile, `valence_arousal_dominance_estimates.xlsx`, contains valence, arousal, and dominance estimates for 14,000 Dutch words.
The second file, `big_5_estimates.xlsx`, contains the correspondence towards the personality traits openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism, for 16,000 Dutch words.
Both files contain two sheets:
- The first sheet simply lists the 'best' estimate for each word on each dimension; in both cases, this refers to estimates obtained at *k* = 10 using the *k*-Nearest Neighbors method (see the accompanying papers for more info), as for this combination, correlations with human ratings were the highest.
- The second sheet contains estimates for all values of *k* and for both extrapolation methods.
All ratings are released under a [CC BY-NC-SA][1] license.
For more information about the procedure used to obtain these estimates, see the accompanying papers (A link to the papers will be added once available.):
- Van Rensbergen, B., De Deyne, S., & Storms, G. (in press). *Estimating affective word covariates using word association data*. Behavior Research Methods.
- Van Rensbergen, B., Kuppens, P., Storms, G., & De Deyne, S. (2015). *Computationally coding responses of a free-format self-description personality test using word association data*. Manuscript submitted for publication.
[1]: https://creativecommons.org/licenses/by-nc-sa/4.0/