The Sound of Round: Evaluating the Sound-Symbolic Role of Consonants in the Classic Takete-Maluma Phenomenon Alan Nielsen and Drew Rendall University of Lethbridge Köhler (1929) famously reported a bias in people’s matching of nonsense words to novel object shapes, pointing to possible naı ¨ve expectations about language structure. The bias has been attributed to synesthesia-like coactivation of motor or somatosensory areas involved in vowel articulation and visual areas involved in perceiving object shape (Ramachandran & Hubbard, 2001). We report two experiments testing an alternative that emphasizes consonants and natural semantic distinctions flowing from the auditory perceptual quality of salient acoustic differences among them. Our experiments replicated previous studies using similar word and image materials but included additional conditions swapping the consonant and vowel contents of words; using novel, randomly generated words and images; and presenting words either visually or aurally. In both experiments, subjects’ image-matching responses showed evidence of tracking the consonant content of words. We discuss the possibility that vowels and consonants both play a role and consider some methodological factors that might influence their relative effects. Keywords: sound symbolism, language structure, bouba-kiki, name-shape biases Debate about the potential naturalness of words dates to at least Socratic times. In Plato’s Cratylus dialogue, Socrates argues for the natural relationship between the structure of words and the things they denote (i.e., their meaning), while his interlocutor Hermogenes argues that the relationship is purely conventional. The naturalist and conventionalist positions have been debated many times since, but following Saussure (1949) the dominant view in contemporary linguistics is that the form of the sign is arbitrary and established by social convention. This conclusion is not seriously disputed for the vast majority of words, but many possible exceptions have also been noted. Such exceptions are often considered together under the banner of sound-symbolism to indicate that the physical form of words (or of higher-order language constructions) might sometimes bear a non- arbitrary connection to the meanings instantiated (reviewed in Hinton et al., 1994). Among the best known examples is the phenomenon of vowel-specific marking of size diminution and augmentation where real or nonsense words containing high, front vowels (e.g., bit, chip, teeny) are judged more natural for denoting small size, while words containing low, back vowels (e.g., block, hunk, chunk) are judged more natural for denoting large size (Sapir, 1929). This effect has been traced to broader sound-size relationships obligatorily embodied in the physics of sound production generally and more specifically in the physics of voice production in humans and other animals (Morton, 1994; Ohala, 1994). For example, compared to small individuals, larger individuals also often have larger sound production apparatuses which naturally produce sounds of lower frequency. This basic relationship between body size and sound frequency holds across a variety of animal species and modes of sound production. It is important that it also applies among primate species, including humans, where sound produc- tion is primarily voice-based. In humans and other primates, the vocal folds of the larynx and the vocal-tract cavities that resonate sounds generated by the larynx are both generally larger in larger- bodied individuals. This pattern yields lower frequency voices (with both lower pitch and lower resonances) in, for example, adults compared to children and in men compared to women (reviewed in Fitch & Hauser, 1995; Fitch, 2000; Rendall, Kollias, Ney & Lloyd, 2005; Ghazanfar & Rendall, 2008). Consistent evidence that sound-symbolic relationships like this are actually manifest in the structure of real words is often lacking, controversial, or complicated (e.g., Bentley & Varon, 1933; Ultan, 1978; Diffloth, 1994). As a result, proposals of sound-symbolism are often viewed with skepticism and treated as marginalia in linguistics and psycholinguistics. However, the consistency of people’s behavior in experimental studies of sound-symbolism, even if they involve interpreting nonsense word material, suggests that people might operate with inherent perceptual-cognitive bi- ases that yield naı ¨ve expectations about language structure. Hence, it is important to investigate these possibilities further because perceptual-cognitive biases of this sort could serve to facilitate normative language processing, or even language acquisition, if the biases are commensurate with real language constructions. Alternatively, such biases could serve to frustrate or impede lan- guage processing or acquisition if they run counter to the relation- ships instantiated in real languages. Either way, the biases could Alan Nielsen and Drew Rendall, Behaviour and Evolution Research Group, Department of Psychology, University of Lethbridge. We thank Doug Mewhort and Daphne Maurer for comments that sig- nificantly improved the work. We thank the Natural Sciences and Engi- neering Research Council (NSERC) of Canada for grant support. Correspondence concerning this article should be addressed to Drew Rendall, Department of Psychology, University of Lethbridge, Lethbridge, Alberta, T1K 3M4, Canada. E-mail: d.rendall@uleth.ca Canadian Journal of Experimental Psychology © 2011 Canadian Psychological Association 2011, Vol. 65, No. 2, 115–124 1196-1961/11/$12.00 DOI: 10.1037/a0022268 115