I pointed out above that there are two scenarios for what happens in the process of language attrition. The first assumes that attrition can have structural ramifications for first language knowledge: underlying rules can be eroded, the intuitive knowledge on how to use language can be lost, the lexical and grammatical repertoire can shrink. The attriter will therefore have an underlying representation of his or her language which deviates from the knowledge of the non-attrited control speaker. The second scenario assumes that this underlying knowledge is perfectly intact and unaffected by erosion or attrition. However, the attriter has more difficulty accessing it: the second language keeps getting in the way, and it becomes more effortful for the speaker to retrieve some of the words or grammatical processes from memory, particularly if he or she has not spoken the language for a long time. On this view, the attriter may use the language differently from the control speaker, simply because something has got to give: attriters have less cognitive resources to devote to getting out the linguistic message, because they need to divert some of these resources to inhibiting the L2 and to digging up those bits of knowledge that have become more difficult to access.