Psycho-computational modelling of lexical access and organisation: what can we learn from Arabic?

Arabic morphology raises a formidable challenge to Markovian computational models of word processing, based on fixed-order memory chaining. Computational models of human short-term and long-term memory can help us considerably to shed light on the issues involved at the level of peripheral word processing (access lexical representations). According to this view, word processing (short-term activation) and lexical representations (entrenchment of habitual activation chains) are two sides of the same coin, as they involve the same levels of brain circuitry on a different time scale. Such an "integrative" view of the lexicon as a dynamic system will be possible only we are able to foster an increasing synergy of perspectives and scientific domains of inquiry: neurosciences, (psycho)linguistics and computing. Conventions of Arabic script are no accident (maliciously intended to trip up computer algorithms)! They rather reflect some fundamental dynamics of the way human brain processes language.