The train goes 'choo choo': A corpus analysis of onomatopoeic words in child-directed speech and early production

Research output: Contribution to conferencePosterpeer-review


Onomatopoeic words (OWs) like woof or choo-choo are lexical items with phonological forms that resemble sounds and objects in the real world. Although they are a common feature of child-directed speech across different languages, their role in early language development remains unclear. Some have proposed that the transparent sound-meaning association of OWs aids early word learning. The sound symbolism bootstrapping hypothesis, for example, suggests that the non-arbitrary nature of these items provides infants with a referential insight into sound-meaning mappings in words [1]. Others have proposed that the phonological characteristics of OWs, including the limited inventory of sounds and less complex syllables, may facilitate the initial stages of children’s word production [2]. A case study of a German-speaking child, for example, found that OWs (e.g., [vavaɯ]) functioned as a direct substitution for the conventional counterparts (e.g., [hʊnt]) from birth to two and a half years [2]. However, it is still not clear whether the tendency to use OWs indeed reflects the child’s active selection process as opposed to a reflection of their input frequency and contexts of use.

The data we analysed come from the speech of five typically developing children (age: 12 to 24 months) and their mothers from the Providence corpus [3]. We first compared the use of OWs (e.g., woof or woof-woof) with their conventional adult alternatives (AAs) (e.g., dog or doggy) in the children’s and mothers’ utterances. Secondly, we classified the use into sound effects versus referential forms. To illustrate, in the example ‘The train goes choo-choo’, train is a referential AA and choo-choo is OW functioning as sound effect. For each of these word types (OWs and AAs) and contexts (sound effects and referential forms), we then compared the frequencies of use in the children’s and mothers’ speech.

Our results show that three of the five children did use OWs at higher rates than their mothers [Table 1]. However, both groups of speakers used OWs as sound effects in the vast majority of cases (88% and 93% on average for children and mothers respectively). In these contexts, there are no adult alternatives, as OWs are the only forms of expression available for the function. Children, therefore, had no choice but select the OW. Furthermore, in referential contexts where either OWs or AAs were available (e.g., ‘There is a choo-choo/train’), the children overwhelmingly chose to produce the AAs [Table 2]. The proportion at which the children chose AAs (96%) closely resembled the proportion of the AAs produced by mothers (98%). Thus, our analysis yielded no evidence that children actively select to produce OWs over AAs when both options are contextually appropriate.

Taken together, these results show that while OWs may offer articulatorily easier alternatives, it is not necessarily the case that children preferentially use these words over adult forms available in the input. In fact, it appears that, in contexts when they have a choice to represent referents with different forms, children’s word productions reflect the patterns they encounter in the input.


Imai, , & Kita, S. (2014). The sound symbolism bootstrapping hypothesis for language acquisition and language evolution. Phil. Trans. R. Soc. B, 369(1651).
Laing, E. (2014). A phonological analysis of onomatopoeia in early word production. First Language, 34(5), 387-405.
Demuth, , Culbertson, J., & Alter, J. (2006). Word-minimality, epenthesis and coda licensing in the early acquisition of English. Language and Speech, 49(2), 137-173.
Original languageEnglish
Publication statusPublished - 2 Nov 2018
EventBUCLD 43 - Boston MA
Duration: 2 Nov 20184 Nov 2018


ConferenceBUCLD 43


Dive into the research topics of 'The train goes 'choo choo': A corpus analysis of onomatopoeic words in child-directed speech and early production'. Together they form a unique fingerprint.

Cite this