A pre-registered analysis of ratings of rides showed that longer names were perceived as riskier, t(230.3) = 5.18, p < .001, b = 0.71, 95% CI = [0.44, 0.98]. There was no effect of pronounceability on riskiness ratings, t(198.1) = -0.28, p = .78, b = -0.01, 95% CI = [-0.09, 0.07]. Participants gave higher ratings in the scenario presented as second, t(947.7) = 3.82, p < .001, b = 0.16, 95% CI = [0.08, 0.24], and somewhat lower ratings in the undesirable risk scenario, t(949.4) = -2.02, p = .04, b = -0.05, 95% CI = [-0.10, -0.00]. These effects were qualified by their interaction, t(953.3) = -4.31, p < .001, b = -0.21, 95% CI = [-0.31, -0.12], showing that the difference between adventurousness and riskiness ratings was higher for the first presented scenario. The effect of pronounceability did not differ based on name length, t(197.9) = 0.42, p = .67, b = 0.04, 95% CI = [-0.13, 0.20], but it depended on the scenario, t(18749.2) = -5.53, p < .001, b = -0.11, 95% CI = [-0.14, -0.07]; that is, it was stronger for the undesirable risk scenario. However, we did not include the interaction between name length and scenario in the model, so the interaction between pronounceability and scenario may be due to the association of name length and pronounceability. When the interaction between name length and scenario is included in the model, the interaction between pronounceability and scenario disappears, t(19053.4) = -0.70, p = .48, b = -0.02, 95% CI = [-0.08, 0.04]. <br><br> Given the scenario and order effects and the strong confounding association between pronounceability and name length, r(204) = -.79, 95% CI = [-.84, -.74], p < .001, we next analyzed the data only from the first presented scenario and for both scenarios separately. There was no association between pronounceability and risk for either the desirable risk scenario, t(213.2) = -0.38, p = .70, b = -0.02, 95% CI = [-0.13, 0.09], or the undesirable risk scenario, t(180.4) = -0.29, p = .77, b = -0.01, 95% CI = [-0.11, 0.08], when name length was taken into account, while name length still significantly predicted the ratings for both the desirable risk scenario, t(240.1) = 4.98, p < .001, b = 0.89, 95% CI = [0.54, 1.25], and undesirable risk scenario, t(222.3) = 5.76, p < .001, b = 0.90, 95% CI = [0.59, 1.21]. Furthermore, adding pronounceability as a predictor to a model with name length did not improve the model for either adventurousness, Χ2(1) = 0.74, p = .39, or riskiness ratings, Χ2(1) = 0.15, p = .70, while adding name length to a model with pronounceability resulted in a significantly better fit for both adventurousness, Χ2(1) = 28.47, p < .001, and riskiness, Χ2(1) = 39.93, p < .001. Therefore, the pronounceability effect that is seen without inclusion of name length as a predictor for both riskiness, t(275.0) = -7.03, p < .001, b = -0.23, 95% CI = [-0.29, -0.17], and adventurousness ratings, t(286.0) = -6.73, p < .001, b = -0.25, 95% CI = [-0.32, -0.17], can be entirely driven by the association of pronounceability with name length. Visual inspection of the data (see left graphs in Figure 4) suggests that the relationship of pronounceability and risk may be present for names 7 and 13 letters long. However, analysis including the pronounceability effect separately for all name lengths does not yield any significant pronounceability effect (see [https://osf.io/eg9yv/][1]). In summary, pronounceability of names of amusement-park rides was associated with their predicted riskiness and adventurousness; however, this effect disappeared when length of the names was taken into account. <br><br> Study 3 by S&S used only 3 fluent and 3 disfluent names and all disfluent names were longer than the fluent names. It was therefore not reasonably possible to evaluate the difference between the original and newly constructed names. However, it is noteworthy that we found no association between pronounceability and riskiness ratings for the original names for either of the scenarios. Descriptive statistics for the original names can be found on [https://osf.io/nbcqg/][2]. Curiously, while we were able to replicate the results of S&S in studies 5 and 6 only using the original names from their study, Study 7 obtained the opposite pattern of results – we found the association of pronounceability and perceived risk with newly sampled items, but not with the original items used by S&S. [1]: https://osf.io/eg9yv/ [2]: https://osf.io/nbcqg/