The Bouba Effect: Sound-Shape Iconicity in Iterated and Implicit Learning

The Bouba Effect: Sound-Shape Iconicity in Iterated and Implicit Learning Matthew Jones a (john.jones.09@ucl.ac.uk), David Vinson a (d.vinson@ucl.ac.uk) , Nourane Clostre a (n.clostre@ucl.ac.uk), Alex Lau Zhu a (a.zhu@ucl.ac.uk), Julio Santiago b (santiago@ugr.es), Gabriella Vigliocco a (g.vigliocco@ucl.ac.uk) a Division of Psychology and Language Sciences, University College London, London WC1E 6BT, UK b Dept. de Psicologia Experimental y Fisiologia del Comportamiento, Facultad de Psicologia, Universidad de Granada, Campus de Cartuja s/n 180179-Granada, SPAIN labels in adults (Perniss et al, 2010). In spoken languages: iconicity facilitates word processing in adults (Westbury, 2005); 3- to 4-month-old infants are sensitive to iconic mappings (Walker et al., 2010); and iconic words are easier for 3-year-olds to learn than non-iconic words (Imai et al., 2008). This is unsurprising given an embodied perspective: if the semantics of words depend on sensorimotor activation, then words with forms that generate appropriate activation will automatically be more learnable than arbitrary words, and will enjoy easier encoding, storage, and retrieval (Perniss and Vigliocco, in press). Abstract Although wordforms are often arbitrarily linked to their meaning, many exhibit iconicity (resemblance between form and meaning). This is especially visible in the lexica of non- Indo-European languages and signed languages. Iconicity has been argued to play a role in grounding linguistic form to real-world experience, rendering language more learnable (Perniss & Vigliocco, in press). Here we examine sound- shape iconicity, the ‘kiki-bouba’ effect, i.e. the tendency to associate bouba-type labels with round shapes, and kiki-type labels with spiky shapes. In a first experiment we show that this iconicity emerges in the course of iterated learning (presumably because it renders labels more learnable). However, it only emerges for the mapping between round shapes and bouba-type labels. In a second experiment (using cross-situational learning, see Monaghan et al., 2012) greater learnability is observed for mappings of the bouba-to-round type but not of the kiki-to-spiky type. We discuss possible mechanisms underlying this difference. Sound-Shape Iconicity Keywords: cross-modal; cultural evolution; iconicity; iterated learning; language evolution; sound symbolism. Introduction Iconicity as a Widespread Feature of Language The arbitrariness of wordform has often had the status of a truism (de Saussure, 1983). However, wordforms are often motivated by iconic relationships with meaning. In English, iconicity can be found in onomatopoeia, (e.g. bang, miaow). However, in languages outside the Indo-European family, iconicity is more pervasive. Large iconic or sound-symbolic lexica are reported for many unrelated languages (including sub-Saharan African languages, Australian Aboriginal languages, and Japanese and Korean; see Perniss Thompson, & Vigliocco, 2010). Iconicity is not limited to resemblance between sounds. In Japanese, reduplication of syllables often indicates repetition of an event, and voicing of an initial consonant can indicate object size (e.g. gorogoro – a heavy object rolling repeatedly; korokoro – a light object rolling repeatedly; Perniss et al., 2010). Given the hands’ potential for mimesis, it is unsurprising that signed languages are rich in iconicity (see Perniss et al., 2010 for a review). E.g. in British Sign Language, the sign for hammer is a hammering gesture, and the sign for lion uses the hands to mimic the pouncing cat’s paws. There is evidence, especially in sign languages, that iconic mappings facilitate learning and processing of novel A seemingly universal form of iconicity is the association between certain sounds (e.g. back vowels and voiced consonants) with heavy, slow, rounded objects; and others (e.g. front vowels and voiceless consonants) with small, quick, jagged objects (Ramachandran & Hubbard, 2001). In standard demonstrations, participants (being adult speakers of English, 4 month old infants, or people in non-literate, non-industrial societies) are given images of two 2- dimensional shapes, one round, the other spiky. The majority prefers to associate ‘kiki’ with a spiky shape, and ‘bouba’ with a round shape. (Maurer, Pathman, & Mondloch, 2006; Ozturk, Krehm, & Vouloumanos, 2013; Bremner et al., 2012). Similar shape-sound associations obtain when the methodology used is implicit learning (Monaghan, Mattock, & Walker, 2012). The origin of this sound-shape iconicity is, however, uncertain. It could be argued that the effect arises from multisensory statistics, with certain shapes co-occuring with certain sounds in the environment, but that such correlations exist is far from clear (see Spence & Deroy, 2012, for discussion). Ramachandran and Hubbard (2001) suggest that the effect is a reflection of cross-modal analogy between the articulatory gestures required to produce the labels and the visual properties of the shapes, implying that non-visual representation of articulation mediates between sound and shape (p. 19). Alternatively, they also suggest that ‘cross-wiring’ (p. 21) of auditory and visual brain maps may create an unmediated link (with associations depending on features of brain architecture). Both of these accounts predict that the effect is present for both ‘kiki’-spiky and ‘bouba’-round. There is, however, yet another possible explanation: associations between speech sounds and lip shape. Words