<p>Research has shown that particular shapes and speech sounds have common higher-order emotional properties, which might mediate associating angular shapes with kiki-like words and round shapes with bouba-like words, resulting in the so-called kiki-bouba effect. However, research supporting this account has mostly recruited explicit association tests to investigate whether people link particular emotions with these shapes and pseudo-words. This study investigated whether the kiki-bouba effect, observed in the implicit association test, can be similarly based on these emotional mediation processes. We found that the explicit and implicit association tests robustly produced a link between angular shape and angry facial expressions, whereas the round shape was associated with happy and calm facial expressions. In contrast, aurally presented kiki and bouba-like words were associated with these facial expressions in the explicit association test but not in the implicit association test. These observations suggest that people process implicitly the emotional properties of angular/round shapes, while they do not automatically process the emotional properties of the perceived kiki/bouba-like words when the task emphasizes implicit association processes. Consequently, we propose that the kiki-bouba effect, which is observed in explicit association tests, can be partially based on emotional mediation processes. In contrast, the kiki-bouba effect, which is based on implicitly operating association processes, is not likely to be based on emotional mediation processes.</p>