Constructivist views of language acquisition (as distinct from constructivist theories of general learning) take a general cognitive approach to the logical problem of language acquisition.
This contrasts with Chomskyan and generativist approaches which see language as essentially different and separate from other cognitive functions.
Emergentists assume that generic associative learning mechanisms underpin all aspects of language acquisition (Ellis, 2003).
Indeed, Ellis claims that:
simple learning mechanisms operating in and across human systems for perception, motor action and cognition while exposed to language data in a communicatively rich human social environment navigated by an organism eager to exploit the functionality of language are sufficient to drive the emergence of complex language representations.
He includes connectionist and emergentist accounts of language acquisition in his definition of constructivism in language acquisition, and contrasts these with generative SLA models.
Ellis and Schmidt's 1998 study of human and virtual network learning of an artificial language is based on Rumelhart and McClelland's influential 1986 connectionist past-tense experiment (neural networks); both studies support an emergentist view of language acquisition.