Previous studies have hypothesized that Japanese sound-symbolic words with voiced initial consonants (SSWV; e.g.,boroboro) rather than those with voiceless initial consonants (SSWVL; e.g., horohoro) or semi-voiced initial consonants(SSWSV; e.g., poroporo) induce stronger evaluations of the quality of psycholinguistic features. To investigate this hy-pothesis, we asked 36 Japanese participants to evaluate 13 psycholinguistic features (familiarity, visual imagery, auditoryimagery, haptic imagery, arousal, preference, disgust, hardness, softness, heaviness, lightness, fastness, and slowness) withSSWV, SSWVL, and SSWSV using 5-point semantic differential scales. All the initial consonants involved h (f; SSWVL),p (SSWSV), or b (SSWV). The experimental results showed that SSWV included higher levels of visual imagery, auditoryimagery, haptic imagery, arousal, disgust, hardness, and heaviness over SSWVL or SSWSV (ps ¡ .05). Taken together,these findings suggest that SSWV induces psychological and physical quality evaluations more than SSWVL and SSWSV.