Towards a unified account of the representation, processing and acquisition of second language knowledge

Abstract

This article argues for the need to reconcile symbolist and connectionist accounts of (second) language learning by propounding nine claims, aimed at integrating accounts of the representation, processing and acquisition of second language (L2) knowledge. Knowledge representation is claimed to be possible both in the form of symbols and rules and in the form of networks with layers of hidden units representing knowledge in a distributed, subsymbolic way. Implicit learning is the construction of knowledge in the form of such networks. The strength of association between the network nodes changes in the beginning stages of learning with accumulating exposure, following a power law (automatization). Network parts may attain the status equivalent to ‘symbols’. Explicit learning is the deliberate construction of verbalizable knowledge in the form of symbols (concepts) and rules. The article argues for a nonnativist, emergentist view of first language learning and adopts its own version of what could be called a non-interface position in L2 learning: although explicit knowledge cannot turn into implicit knowledge through practice, it is argued that explicit learning and practice often form efficient ways of mastering an L2 by creating opportunities for implicit learning.