Since the 1970’s, evidence has suggested that preschool children’s encoding of information differs from that of adults in some respects. In their comprehension, preschoolers have been shown to allow target-deviant interpretation of indefinites, which Krämer (2000, 2003) argues is due to insufficient discourse integration. In their production, they have been reported to over-use definite noun phrases in contexts where adults would use indefinites (Maratsos 1974; Karmiloff-Smith 1979; Emslie & Stevenson 1981). This has recently been argued to be due to an immature pragmatic system (Schaeffer & Matthewson 2005). Delayed pragmatic competence has also been invoked to explain various phenomena, from binding errors (Chien & Wexler 1991) to the use of null subjects with finite verbs (Wexler 1998; Schaeffer et al. 2002).
On the other hand, research has also shown that from their first year of life, infants are already able to distinguish new from old information and that language production even at the one-word stage is constrained by a principle of informativeness (Baker & Greenfield 1988). The degree of informativeness of referents, along with other discourse factors, has since been shown to play a significant role in the realisation of arguments in early acquisition (see e.g. Allen 2001, 2007). It has recently been proposed that target-deviance in this area is not caused by a delay in discourse competence, but by the excessive processing demands arising from the interaction of discourse and narrow syntax (Serratrice, Sorace & Paoli 2004).
This paper focuses on the acquisition of topic by monolingual children. Topic is a key discourse/pragmatic notion, whose mastery has often been argued to be deficient in early acquisition (see e.g. Chien & Wexler 1991; Wexler 1998; Schaeffer et al. 2002). It is also the interface phenomenon par excellence, as it requires the integration of up to three components of the language faculty: discourse/pragmatics, syntax and phonology (Author 2007b). Given preschool children’s processing limitations and their well-documented impact on language (Gathercole & Baddeley 1993; Reinhart 2004, 2006), one might expect that topic structures be problematic in early acquisition. However, the longitudinal study of early spontaneous production suggests otherwise: topic structures appear from the onset of expressive syntax and children do not make commission errors in the encoding of topics. In this paper, I present experimental evidence confirming these findings, showing that, at least as early as 2;6, children correctly identify and encode topics.