Premium
From Data to Theory: the Over‐Representation of Linguistic Knowledge in SLA
Author(s) -
Myles Florence
Publication year - 2004
Publication title -
transactions of the philological society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.333
H-Index - 26
eISSN - 1467-968X
pISSN - 0079-1636
DOI - 10.1111/j.0079-1636.2004.00133.x
Subject(s) - linguistics , rule based machine translation , computer science , natural language processing , phenomenon , second language acquisition , competence (human resources) , representation (politics) , verb , grammar , artificial intelligence , psychology , philosophy , epistemology , social psychology , politics , political science , law
Abstract This article deals with the over‐representation of syntactic knowledge in L2 learner data. In early L2 grammars, multimorphemic sequences which go well beyond learners’ grammatical competence are very common. Influential theories of the L2 initial state have failed to take proper account of this phenomenon, claiming the apparent presence of, e.g., finite verb forms, wh questions and clitics, as evidence for functional projections from the start of L2 acquisition. However, data from early learners show that these properties are not present outside chunks initially, and that early grammars consist of lexical projections and formulaic sequences, showing no evidence of functional categories.