Appendix

Spanish words used as stimuli [English translation]. Critical words in bold.

Below: Statistical criteria for matching the three lists, Summary Table (below), and indices of valence, arousal, frequency, concreteness, and associative strength of each word.

Our objective in creating the lists was that the five lists with a positive valence would have a significantly higher average valence than the five lists with a neutral valence, which in turn would have a significantly higher average valence than the five lists with a negative valence. The 15 lists were matched on arousal, frequency, concreteness, and associative strength (and this scheme of differentiating the valence and matching the rest of the variables would also be applicable to the critical words, the study words, and the unrelated words).

With this objective in mind, we began by carrying out five one-way analyses of variance (ANOVA), jointly considering the study words, critical words, and unrelated words, comparing the three valence groups (containing 60 words each), and using valence, arousal, frequency, concreteness, and associative strength as dependent variables. These analyses confirmed (p < .05) that the three groups of 60 words each were equal in arousal, (F (2, 179) = 2.46, p = .09; with an average arousal of 5.40 points, which means that our words had a medium arousal), frequency (F (2, 179) = 2.54, p = .08; with a frequency of average use of 77.29 points), concreteness (F (2, 179) = 1.63, p = .20; with an average concreteness of 4.83 points), and associative strength (F (2, 143) = 1.70, p = .19; with an average associative strength of 4.06 points; which means that our words had a high associative strength), but they differed significantly on their valence (F (2, 179) = 1046.55, p < .0001). Scheffe's post-hoc tests (p < .05) showed that positive valence words had a significantly higher average valence (M = 7.44) than neutral valence words (M = 5.56), which in turn had a significantly higher average valence than negative valence words (M = 2.77).

Next, we analyzed the critical words by means of five ANOVAs that compared the 12 critical words with a positive valence, the 12 with neutral valence, and the 12 with a negative valence on their valence, arousal, frequency, concreteness, and associative strength (see Summary Table, below). This analysis again confirmed that these three groups of critical words were equal in arousal (F (2, 35) < 1), frequency (F (2, 35) = 3.08, p = .06), concreteness (F (2, 35) < 1), and associative strength (F (2, 35) < 1), but they differed significantly on their valence (F (2, 35) = 206.65, p < .0001). Scheffe's post-hoc tests showed that positive valence critical words had a significantly higher average valence (M = 7.37) than neutral valence critical words (M = 5.46), which in turn had a significantly higher average valence than negative valence critical words (M = 2.57).

We then analyzed the study words by performing five ANOVAs that compared the 36 positive valence, 36 neutral valence, and 36 negative valence study words on their valence, arousal, frequency, concreteness, and associative strength (see Summary Table, below). This analysis again confirmed that these three groups of study words were matched in arousal (F (2, 107) = 2.85, p = .06), frequency (F (2, 107) = 2.06, p = .13), concreteness (F (2, 107) = 2.89, p = .06), and associative strength (F (2, 107) = 1.29, p = .28), but they differed significantly on their valence (F (2, 107) = 511.05, p < .0001). Scheffe's post-hoc tests showed that the positive valence study words had a significantly higher average valence (M = 7.48) than the neutral valence critical words (M = 5.63), which in turn had a significantly higher average valence than the negative valence critical words (M = 2.83).

Finally, we performed four ANOVAs to compare the three unrelated word lists (see Summary Table, below), again showing that these three lists were equal in arousal (F (2, 35) < 1), frequency (F (2, 35) < 1), and concreteness (F (2, 35) < 1), but they differed significantly on their valence (F (2, 35) = 706.39, p < .0001). Scheffe's post-hoc tests showed that unrelated words with a positive valence had a significantly higher average valence (M = 7.36) than critical words with a neutral valence (M = 5.45), which in turn had a significantly higher average valence than critical words with a negative valence (M = 2.78).

Together, these statistical analyses confirm the suitability of the materials used here, in that the three lists of positive valence words, the three lists of neutral valence words, and the three lists of negative valence words, are matched on arousal, frequency, concreteness, and associative strength, but they differ significantly on valence, as we proposed in our objective.

Summary Table. Means (and SD) of the five analyzed variables according to type of word (study, critical or unrelated) and type of valence (VAL + = positive valence; VAL - = negative valence; VAL neutral = neutral valence).

Indices of valence, arousal, frequency, concreteness, and associative strength of each word.