Computational Modeling of Word Reading , Acquired Dyslexia , and Remediation

Many researchersassumethat the mostappropriateway to expressthesystematicaspectsof languageis in terms of a setof rules. For instance,thereis a systematicrelationship betweenthe written and spoken forms of most English words (e.g., GAVE ), and this relationship canbe expressedin termsof a fairly concisesetof grapheme-phoneme correspondence (GPC)rules(e.g.,G , A E , V ). In addition to beingable to generateaccuratepronunciationsof so-calledregular words,suchrulesalsoprovide a straightforwardaccount of how skilled readersapply their knowledgeto novel items—forexample,in pronouncingword-like nonwords (e.g., MAVE ). Most linguistic domains,however, areonly partially systematic.Thus,therearemany Englishwordswhosepronunciationsviolatethestandard GPCrules(e.g.,HAVE ). Giventhatskilledreaderscanpronouncesuchexceptionwordscorrectly, GPC rulesaloneareinsufficient. More generally, skilled languageperformanceat virtually every level of analysis— phonological,morphological,lexical, syntactic—requires botheffective handlingof exceptionalitemsandtheability to generalizeto novel forms.