Rapid or automatic recognition of words allows the L2 reader to free up attention to apply simultaneously his or her knowledge of the subject, general background knowledge and expectations regarding the likely development of clauses, sentences, paragraphs and whole texts. However, the interaction of these higher-order comprehension processes posited by schema theory requires a lot of the mind's processing capacity. Consequently, if a reader has to devote conscious attention to word recognition or decoding, the processing of higher-level elements in the construction of meaning is impaired. In beginning readers, comprehension can be even more severely disrupted as readers engage in the slow, cumbersome process of alternating between non-automatic, conscious decoding and the construction of meaning (Samuels, 1994: 821-2).

It is therefore important to recognize that providing background information on a text and previewing its content are even more crucially important for less able language learners than for proficient L2 readers. Although the former's problem tends to be that their reading breaks down at the word level, to deprive them of content preview makes their task doubly difficult, since it is very demanding to have to tackle both linguistic and cultural codes at the same time (Carrell and Eisterhold, 1983). If, on the other hand, the semantic content is conveyed to them, they are free to concentrate on the vocabulary and structures that express that content.

In the same way that low-level readers can benefit from higher-level strategies, fluent readers rely to a large extent on low-level identification of forms to support their top-down semantic predictions. As Eskey (1986:14) points out, it is precisely the fluent reader's ability to process language automatically and in fairly large amounts that allows the conscious mind to devote itself to the overall question of meaning. Good decoding in a reader is therefore a cause, not a result of fluent reading (Eskey, 1988:94), as top-down theorists used to assume. Proponents of this 'interactivist' view believe, moreover, that as the proficiency of good readers increases, they do not become less concerned with identification of language forms but increasingly efficient at it, at the same time as their interpretative skills develop.

To sum up, then, a tentative consensus has emerged in L2 reading theory. The meaning of a text is reconstructed through a constant interaction between the information obtained through bottom-up decoding and that derived from top-down analysis, both of which rely on various types of prior knowledge and information-processing skills, and at all levels of language proficiency these skills are interactively available to process and interpret texts.

A distinction is sometimes made between reading 'skills' and 'strategies'. The former relate to the motor skills of early reading, to decoding, or the (bottom-up) ability to recognize word shapes and patterns; the latter denote (top-down) approaches for getting at meaning through context, which are influenced both by the reason for reading and the type of text being read. In sections 7.2-7.6 we shall explore a number of strategies that promote more effective L2 reading.