Experiments 1–3 relied on taste words rather than real tastants, which has been previously highlighted as a potential limitation in taste research (see Simner, Cuskley, & Kirby, 2010). Certainly, although our participants were instructed to imagine a food with the taste in each trial, taking into account the literature on sound symbolism (Marks, 1978; Parise & Spence, 2012; Sapir, 1929), one may wonder whether the sound of the words themselves may have influenced performance in the matching task. What is more, several studies have reported that people sometimes confuse the different basic tastes (e.g., Hettinger, Gent, Marks, & Frank, 1999; O’Mahony, Goldenberg, Stedmon, & Alford, 1979), which might help to explain why people gave similar roundness/angular- ity ratings to, for example, bitter and sour words. Although using real tastants may not resolve this issue, at least one can be sure that participants are responding to the actual tastes themselves. Real tastes can also be controlled for in terms of intensity, some- thing that is probably unachievable when working with imagined taste. In Experiment 4, real tastants were therefore used, including the taste of umami, which is widely considered to be the fifth basic taste (e.g., Kawamura & Kare, 1987; Kurihara, 2009; Mouritsen & Styrbaek, 2014).