The role of dynamic schemas in online processing: implications for language attrition

The most powerful influence exerted on linguistic structures during processing is their current internal context. Crucial to defining what ‘internal context’ is, is the notion of the schema. A schema is a web of associations that links representations of different types. Example: auditory structures (AS) of a sound pattern and associated phonological structures which in turn are associated with particular syntactic structures and which in turn are associated with conceptual structures. Typically schemas are much more internally complex than the above chain of associated structures (AS<>PS<>SS<>CS). This is especially the case when representations that belong to the conceptual and/or the affective systems get involved.  The two last-mentioned systems in particular are ‘hubs’  having multiple connections with most of the other systems of which the mind is composed. Typically, then, the given coactivated internal context of a particular linguistic representation, say a PS, will involve schemas much more elaborate than the one above. It will have coactivated associations spreading across the whole mind. The equivalent neural patterns involved are,  in this sense at least,  no different. Wake up one representation and you wake up a whole army. This explains quite a lot. It includes, for example, why, in a non-stressful L2 exchange with, say, a nice native speaker or during an enjoyable L2 class,  certain (relatively)  weakly established L2 structures might show up in performance where normally they would be outcompeted by an L1 competitor. MCF architecture explains in some detail why and how activated schemas can shift and change during processing due to changes in the internal context. This can have knock-on effects since regular activation is crucial for maintaining or increasing the competitiveness of a representation and thereby preventing a slow decline in its ‘accessibility'. Once a representation has declined sufficiently that it can no longer participate in any schema on a regular basis. There will be longer term structural consequences to such schemas affecting the performance that they generate. Crosslinguistic influence is a typical feature of this online restructuring which is part of the explanation why language attrition is not just a question of individual representations failing to appear in performance. Understanding processing as the activation of schemas is important for explaining such phenomena. More details and examples will emerge in the course of the discussion.