Sound Symbolism

One of the traditional hallmarks of human language that distinguishes it from most other animal communication systems is its arbitrariness. The sounds of words and structure of sentences are presumed to bear an arbitrary relationship to their meaning. However, there are many examples across languages that violate the arbitrariness assumption. These nonarbitrary associations between sound and meaning have been termed sound symbolism.

Perhaps the most salient example of sound symbolism is onomatopoeia, in which words resemble the sounds to which they refer (e.g., meow, beep). There are additional within-language conventions that [Page 563]reliably link particular sound sequences with particular semantic domains. For example, phonesthemes are categories of sounds that tend to appear in semantically related words, such as sn-words relating to the nose (e.g., sniff, snort, sneeze, snore) and gl- words relating to shine (e.g., gleam, glow, glitter, glint). In several languages including Japanese, there is a class of words called mimetics that are believed to be particularly evocative of sensory experience across modalities and are, as a result, commonly used in poetry and in child-directed speech.

There are also sound symbolic cues to grammatical form class. Within a given language, for example, certain classes of phonemes tend to be more strongly associated with nouns than verbs and vice versa. None of these sound-to-meaning correspondences are entirely deterministic, of course, but the heightened bidirectional conditional probabilities (i.e., likelihood of a particular meaning given a particular phoneme inventory and of a particular phoneme inventory given a particular meaning) certainly challenge the arbitrariness principle.

The classic example of sound symbolism is the Maluma/Takete or Bouba/Kiki illustration. Across cultures and across development, people reliably associate words like Maluma and Bouba with rounded contours and words like Takete or Kiki with pointed contours. Consistent with this apparent universal agreement, there is cross-linguistic sound symbolism. That is, speakers of one language are able to select the correct meaning of words spoken in another language at rates that exceed chance performance. For example, native English speakers can reliably guess in a forced-choice task the meaning of Japanese antonym pairs and can distinguish bird names from fish names in Huambisa. It appears that common sound-to-meaning correspondences recur across languages and language families. That the same sound-to-meaning correspondences arose in distinct language families raises the possibility that sound symbolism is a result of more basic cross-modal mappings that are not specific to language.

A growing literature suggests that these sound-to-meaning correspondences facilitate word learning in children. The classic Bouba/Kiki phenomenon extends to children as young as 2½ years of age. More recent evidence indicates that novel mimetic-like Japanese nonsense verbs are easier for children to learn than those that lack transparent sound-to-meaning correspondences. This is true for both Japanese-speaking children who have previous exposure to mimetics and Japanese phonology and for English-speaking children to whom Japanese mimetics are wholly unfamiliar. These findings suggest that sensitivity to how sounds relate to meaning is available from early in development and may enable children to bootstrap their way into word learning via sound sensitivity.

Another less studied form of sound-to-meaning correspondence that may facilitate language learning is prosody. Prosodic cues to word boundaries, sentence structure, and pragmatics have been well established. Most recently there is evidence that both adults and children can also recruit prosodic information to infer word meaning. This is rather intuitive for valence-based tone of voice cues (i.e., words with happy/sad or good/bad connotations) but has also been established for other antonym pairs such as big/small and hot/cold, which do not readily align with valence. It is unclear whether this form of sound symbolism is governed by the same mechanisms as those involved in phonological sound symbolism. To the extent that sound-to-meaning correspondences are a product of general associations between acoustic features of the speech signal and semantic features, it is possible that prosodic and phonological sound symbolism are indeed examples of the same process.

...