Multilingual linguistic representations

This keynote article discusses the issue of crosslinguistic influence (CLI) in third language acquisition. I argue that CLI should be seen as co-activation of structures in previously acquired languages; it happens property by property, and its source can be either or both previously acquired languages. The subtractive language groups design, a relatively new methodological choice for experimental design and its advantages are presented. An extended example of how feature bundles become adjusted in the third language is explored, and new syntactic ideas (e.g., parametric hierarchies) are examined with a view to establishing whether they are supported by experimental findings thus far available. Some experimental evidence is reviewed with an emphasis of studies that examine Japanese. Other factors that can affect CLI, as predicted by the Scalpel Model, are also discussed. The overall message is that L3/Ln acquisition affords valuable insights to investigating language from a formal perspective