Lew-Levy and Amir argue that children’s peer cultures play an important role as drivers of innovation in cultural transmission. We examine how this proposal aligns with evidence about children’s role (or lack thereof) in the cultural transmission of language – arguably the most prominent human trait. While LL&A reference language in support of their proposal, we argue that current evidence for children’s role in language change is less compelling. Unlike other cultural practices discussed by LL&A, language presents a unique challenge for young children to exert influence on the wider community, thus making them unlikely agents of change. Moreover, in the context of novel communication systems, adopting, maintaining, and spreading structured and meaningful linguistic innovations through peer interaction requires cognitive and pragmatic maturity that young children in their first decade of life do not yet possess. These limitations may also have restricted the role of young children’s peer culture in language emergence.

Crucially, language differs from other cultural traits in its high degree of normativity and conventionalization, which renders most input-divergent variation produced by child learners errors rather than innovations that offer functional – and therefore adaptive – solutions for communication. Such errors are well documented in the domains of phonology (e.g., introduction of new phonetic variants, see Dodd et al., Reference Dodd, Holm, Hua and Crosbie2003), morpho-syntax (e.g., regularization and reinterpretation of grammatical structures, see Hudson-Kam & Newport, Reference Hudson Kam and Newport2005), and the lexicon (e.g., over-extensions and colexification, see Brochhagen et al., Reference Brochhagen, Boleda, Gualdoni and Xu2023). While some suggest that these acquisition errors can affect language change (Cournane, Reference Cournane2019), language development research shows that children typically recover from these errors before participating in vertical transmission as adults (Ambridge et al., Reference Ambridge, Bidgood, Twomey, Pine, Rowland and Freudenthal2015). Moreover, social learning biases such as the prestige and frequency bias (Henrich et al., Reference Henrich2001) do not favor children as models, and therefore make the adoption of children’s acquisition errors by the broader community highly unlikely. While we cannot exclude the possibility that children had greater prominence and social standing in preagricultural hunter-gatherer societies, there is currently no evidence to suggest that the acquisition errors of individual children propagate through wider linguistic communities.

In fairness, LL&A do not claim that individual children are agents of cultural adaptation and transmission, but rather that children’s innovations may propagate once they gain social or functional significance within their peer groups. For peer groups of older children and adolescents, this claim is well supported by sociolinguistic evidence: when linguistic variation (e.g., phonological shifts, morpho-syntactic innovations, or neologisms) acquires social or functional significance, it can trigger vernacular change (Kerswill, Reference Kerswill1996). Examples of adolescent vernacular change include the adoption of copula deletion (e.g., “He tall”), negative concord (e.g., “I don’t know nothing”), or quotative be like (e.g., “She’s like ‘Oh!’”) (Labov, Reference Labov1972; Martínez, Reference Martínez2011; Tagliamonte & D’Arcy, Reference Tagliamonte and D’Arcy2004). Such adolescent innovations carry sufficient social prestige and group relevance to be subsequently adopted by younger children (Tagliamonte & D’Arcy, Reference Tagliamonte and D’arcy2009). This transmission pathway aligns with cross-cultural observations showing that – as LL&A point out – in many societies, children receive most linguistic input from other children, predominantly older siblings (Cristia et al., Reference Cristia, Dupoux, Gurven and Stieglitz2019). Some adolescent linguistic innovations are also adopted by adults, albeit to a more limited extent (Buchstaller, Reference Buchstaller2015; Tagliamonte & D’Arcy, Reference Tagliamonte and D’arcy2009).

However, for younger children (i.e., 3–10-year-olds), there is no evidence that the ubiquitous phonological, morpho-syntactic, and lexical variation characteristic of their acquisition errors would be adopted by peer groups and diffuse from there into the wider community. Even though young children engage in improvised language play as an in-group affiliation device (Cekaite, Reference Cekaite2018), we are not aware of observations of these transient forms of language use spreading to wider peer groups, including older children. The lack of knowledge about what linguistic innovations are taken up by peer interactions of young children may be a consequence of the fact that the study of children’s acquisition errors has focussed primarily on infants’ and young children’s communication with adult caregivers. This focus is facilitated by long-standing data collection efforts centered around the TalkBank repository, including the CHILDES database (MacWhinney, Reference MacWhinney2025). As a result, it is unclear whether acquisition errors such as linguistic regularizations, overgeneralizations, and mispronunciations (e.g., *goed for went) are adopted by younger peer groups. However, even if they were, there is no evidence that such errors spread further into the community.

LL&A also invoke the argument that children can introduce systematic structure when linguistic systems emerge de novo, a claim that has been made for creoles (Bickerton, Reference Bickerton1984), emerging sign languages (Senghas & Coppola, Reference Senghas and Coppola2001), and artificial gestural systems (Bohn, Kachel, & Tomasello, Reference Bohn, Kachel and Tomasello2019). Thus, even if children do not contribute to change in established languages, they might drive the emergence of new ones (Senghas, Reference Senghas2021). However, adults are more likely agents of creolization (Arends & Bruyn, Reference Arends and Bruyn1994; Roberts, Reference Roberts1995) and work on Nicaraguan Sign Language (Senghas, Kita, & Ozyurek, Reference Senghas, Kita and Ozyurek2004) cannot disentangle whether structural innovations are specific to children or are attributable to a later position in an iterated learning chain (Kirby, Cornish, & Smith, Reference Kirby, Cornish and Smith2008). In fact, young children find it difficult to avoid ambiguity and are much less likely than adults to introduce meaningful structure into novel communication systems (Kempe et al., Reference Kempe, Gauvrit, Gibson and Jamieson2019; Raviv & Arnon, Reference Raviv and Arnon2018). These difficulties are partly due to children’s pragmatic limitations arising from an immature Theory of Mind, required to invent motivated signs, and partly due to their limited cognitive capacity, required to scan the semiotic affordances of referential domains and monitor alignment with an interlocutor. Young children require adult scaffolding (Matthews, Lieven, & Tomasello, Reference Matthews, Lieven and Tomasello2007) to reach communicative efficiency and do not achieve effective peer communication up until 10–11 years of age (Garrod & Clark, Reference Garrod and Clark1993; Girbau, Reference Girbau2001). Even though children can produce gestural innovations when communicating about familiar meanings (Bohn et al., Reference Bohn, Kachel and Tomasello2019), this ability improves with age, and there is no evidence that their innovations have greater functionality than those of adults.

Taken together, we argue that language is likely a special case: linguistic variation in the form of acquisition errors does not spread to the wider community, making young children unlikely agents of language change. Moreover, children’s pragmatic and cognitive limitations constrain their role in language emergence.