Onomatopoeia and metonymy

3. Onomatopoeia and linguistic convention
Linguistic literature has had a tendency to marginalize the role of onomatopoeia in language. Saussure (1915/1959, p. 69), for instance, made note of the fact that onomatopoeia might serve as evidence against the arbitrariness of the linguistic sign, but refuted this position by arguing that onomatopoeic forms occur in relatively low proportions in any language (and thus are of marginal importance). Yet he also stated that the link between the signifier and the signified was far less natural in onomatopoeia than what it was assumed to be, since the sounds in onomatopoeic formations do not fully correspond to the imitated sounds – thus, such words do exhibit conventionality to some extent.⁷ Saussure illustrated his point with the example of pigeon, which was borrowed from the Vulgar Latin pipio – an onomatopoeic formation – to demonstrate that onomatopoeic words also undergo linguistic evolution, in the process of which they become increasingly similar to the unmotivated linguistic sign.

Hence, despite the misgivings that Saussure had about onomatopoeia in general, he made a very important point by emphasizing that an onomatopoeic form – just as any other word in the language – undergoes certain modifications (phonological and morphological) in order to fit into a particular linguistic system. Even the nonce formations in the Garfield comics that were mentioned above need to comply to certain impositions of English to be able to appear in the comic strip and convey the meaning that they have. Thus, it is a fact that onomatopoeia do need to conform to the restrictions of a particular language to be meaningful for speakers and thus do undergo modifications; it is simply the extent of these modifications that varies within the various examples of onomatopoeia.

Nevertheless, the natural link that these words have to their referents – whether this link is perceived or only imagined – can still be available for the language users. This feature of onomatopoeia was highlighted in a study by Assaneo et al. (2011), who examined the degree of imitation of sound in knock and click words across languages. They found that the vocal configurations that were judged by subjects to best represent the sound that was made when striking a blow on the door or when pressing on light switches were readily associated with co-articulated consonants, and these co-articulations were the most established and wide-spread sounds found in knock and click onomatopoeia across languages.
It is the great degree of variation among onomatopoeic words with respect to the extent of modification that they have undergone (to conform to the phonological and morphological rules of a language) that lies at the heart of the difficulty of succinctly defining what “onomatopoeia” are. Take, for example, the following definition from the OED as a point of departure: onomatopoeia, which is of Greek origin (from the combining forms onomato “of or relating to names or naming” and poeia “forming nouns denoting the making or creating of a thing (or things) expressed by the first element”), refers to the “formation of words imitative of sounds”.⁸ There are a multitude of problems with this definition. First, it fails to clarify that onomatopoeia do not just simply imitate a particular sound, as they can also denote the sound produced by a certain entity, as in the case of cuckoo, which denotes the bird that produces a sound imitative of its name. Second, the word “imitation” that appears in the definition glosses over the fact that similarity between the actual sound heard in nature and the onomatopoeic form is very much in the eye of the beholder. After all, there is a finite set of sounds and letters that represent these sounds in any particular language; thus, the precision of the translation of the sound into onomatopoeia will invariably depend on “how fine-grained are the sign-units of the target system” (Tsur 2001, n.p.).
Going back to the example of cuckoo, Tsur (2001) compared a recording of the bird’s song with the sounds in cuckoo, and while a sound that is very close to the English vowel [ʊ] can indeed be detected in the bird’s song, there is nothing that would suggest a [k] sound. This finding is quite peculiar in light of the fact that the [k] does appear regularly in the sound shape of the bird’s name in other languages too – see, for example, kakukk (Hungarian), coucou (French), koekoek (Dutch), cuculo (Italian), kukavica (Croatian), etc. Why is [k] then favoured in so many languages if it does not actually show up in the sound that is produced by the bird? The reason is probably co-articulation – it is easier to articulate the velar [k] with the back vowel [ʊ], as opposed to the dental [t] or the bilabial [p] (both [t] and [p] are stops, similarly to [k]).⁹ Third, the OED definition also introduces ambiguity by the notion of “word” – what degree of lexicalization is required for an onomatopoeic form to count as one? While achoo can be found in the OED,¹⁰ the sound that an oven makes when the food is ready (cf. ba-going) is not. Last but not least, not all sounds of an onomatopoeic word participate in the imitation – consider, for example, the word whisper, where it is the combination of the glide and the sibilant that brings about the sense of speaking very softly.