Learners of a newlanguage have to extractwords and the rules from speech. Learners are
endowedwith the capacity to extract statistical regularities from their environment allowing
them to extract words from continuous speech in the absence of other cues. However,
it has been proposed that natural languages have an intrinsic cue: prosodic information.
This cue seems to trigger the application of different computational resources that allows
the extraction of rules. This review summarizes work indicating that attention and
working memory are critical in the early stages of language acquisition, in the absence
of semantic information. Event-related potentials while participants learned artificial
languages with embedded morphological rules show a dissociation between the brain
responses associated toword and rule learning. The results indicate that salient cues such
as prosody help to direct attention biasing perception to ignore irrelevant information
and attend to the relevant segments containing the rule, shifting from word acquisition to
rule extraction. Finally, data from individual differences in brain connectivity related to
phonological working memory and data from brain-lesioned patients point to the basal ganglia as a coordinator structure among language, working memory, and attention
through its rich connections with brain areas responsible for these functions.