I examine some of the issues connected with the internalist/externalist distinction in work on the ontology of language. I note that Chomskyan radical internalism necessarily leads to a passive conception of child language acquisition. I reject that passive conception, and support current versions of constructivism [Tomasello, M., 2001. The item-based nature of children’s early syntactic development. In: Tomasello, M., Bates, E. (Eds.), Language Development: The Essential Readings. Blackwell, Oxford, pp. 169–186; Deuchar, M., Vihman, M.M., 2005. A radical approach to early mixed utterances. International Journal of Bilingualism 9 (2), 137–158], according to which the child actively constructs linguistic knowledge via a process of dynamic interaction with the environment, in which knowledge of social conventions is acquired. This approach to child language acquisition is, I believe, broadly in line with the view of Spurrett and Cowley [Spurrett, D., Cowley, S.J., 2004. How to do things without words: infants, utterance-activity and distributed cognition. Language Sciences 26, 443–466], who take utterance activity to be ‘jointly controlled by the embodied activity of interacting people.’ The view I adopt here, based on the Representational Hypothesis proposed by Burton-Roberts [Burton-Roberts, N., 2000. Where and what is phonology? In: Burton-Roberts et al. (2000), pp. 39–66], takes phonological knowledge to be knowledge of conventions of physical representation. It thus proposes an intersubjective ontology for the external physical representation of radically mind-internal conceptual content. However, I now seek to abandon radical internalism. I sketch Burton-Roberts’s Representational Hypothesis. I suggest that this offers a clear distinction between the roles played by nature and by culture in the range of phenomena broadly referred to as ‘language’. I attempt here to retain B-R’s conception of physical representation, but, having rejected his radical internalism, I seek to resolve the main problem thrown up by this attempt (namely, what it is that is being represented, if not elements from an innately-endowed, universal set of semantic primitives). In adopting the idea of physical representation, I reject the idea of specific languages as systems which encode conceptual content. I therefore support the view of Love (this issue) that languages are not codes. But I seek to counter Love’s extreme scepticism. I review some of the evidence for an alternative, constructivist, view of child language acquisition from empirical work on child data provided by Brulard and Carr [Brulard, I., Carr, P., 2003. French/English bilingual acquisition of phonology: one production system or two? International Journal of Bilingualism 7 (2), 177–202] and by Vihman [Vihman, M.M., 1996. Phonological Development. Blackwell, Oxford], which supports, I suggest, the ideas of emergent modularity [Karmiloff-Smith, A., 2001. Development itself is the key to understanding developmental disorders. In: Tomasello and Bates (2001), pp. 331–350] and usage-based models of language [Bybee, J., 2001. Phonology and Language Use. Cambridge University Press, Cambridge].