As a cognitive tool the art of reading and writing has had a most profound impact on society and human cognitive functioning. Written language has the character of an external memory system where its spatial lay-out and its permanence permit inspection and repeated controls of the stream of thoughts, thereby considerably reducing the working load of the human memory. As all revolutionary inventions, however, written language has also had non-intended side effects. --Those who acquire it will cease to exercise their memory and become forgetful; they will rely on writing to bring things to their remembrance by external signs instead of on their own internal resources--, as the skeptical Egyptian king Thamus put it in Plato’s dialogue Phaedrus. And no doubt, nonliterate peoples are capable of astonishing feats of memory testified by ethnologists.

Although the art of verbatim long-term episodic memorization might have declined in literate societies, reading and writing seem to have put other demands on the memory system. In processing written text a reader must integrate moment-to-moment perceptions across time, rehearse them and combine them with simultaneous access to archival information about past experience, actions and knowledge. This is what working memory involves. In oral dialogues the working-memory load is normally considerably lower than in reading, as written discourse packs linguistic information differently. The absence of prosodic information further increases the working-memory load in reading. This point will be further elaborated in the paper. Even at the word level, especially in alphabetic scripts, and especially during the initial stages of reading acquisition, the decoding process involves considerable working-memory demands. A skilled reader processes many thousands of words each day, year after year. No doubt, this intense and extensive activity would be expected to have a profound impact on brain functions. And, in fact, recent studies have shown how literate adults have brain activity patterns as response to phonological memory tasks which are clearly different from the patterns of illiterate individuals from the same SES background (Ingvar, 1999). Also dyslexic individuals show lower activity or less integrated activity in brain areas of critical importance for phonological processing (Paulesu et al., 1998; Pugh et al. 1999).

A complex working-memory task is typically devised to mimic the competing cognitive demands involved in an activity such as reading. In the study to reported in this paper the subject is orally presented with a consonant letter followed by a simple sentence verification task also orally presented (--trees can walk-- yes-no). After two or more such presentations the subject is required to report the presented consonants in correct order. This is a task involving the phonological loop as well as the central executive (Baddeley & Gathercole, 1996).

This task and several other tasks were presented to 30 adults with a proven history of reading disability and to 30 normal readers matched on age, gender and educational level. The working memory capacity was significantly associated with a number of phonological tasks such as spoonerism, non-word reading, and phonological distinctness of vocabulary. Logistic regressions, however, demonstrated that working memory had the power of predicting group belonging over and above the other tasks. This finding supports the assumption of the critical role of complex working memory in reading. However, the causal direction is still an unsettled issue.