Formal theories of grammar and traditional parsing models,insofar as they presuppose a categorical notion of grammar,face the challenge of accounting for gradient judgments ofacceptability. This challenge is traditionally met by explaininggradient effects in terms of extra-grammatical factors, positinga purely categorical core for the language system. We presenta new way of accounting for gradience in a self-organizedsentence processing (SOSP) model, which generates structureswith a continuous range of grammaticality values. We focuson islands, a family of syntactic domains out of whichmovement is generally prohibited. Islands are interestingbecause, although most linguistic theories treat them asfully ungrammatical and uninterpretable, experimental studieshave revealed gradient patterns of acceptability and evidencefor their interpretability. We report simulations in whichSOSP largely respects island constraints, but in certain cases,consistent with empirical data, coerces elements that blockdependencies into elements that allow them.

When making a reference, speakers must choose between nouns and pronouns. At what level of representation do speakersmake such a choice? The non-linguistic competition account predicts that the choice of using a pronoun occurs at thenon-linguistic level, so speakers should use fewer pronouns when the potential referents compete more strongly at thenon-linguistic level. By contrast, the linguistic competition account predicts that the pronoun choice occurs at the lexicallevel; speakers should use fewer pronouns when the potential antecedents are semantically or phonologically more similar.We show that regardless of whether the selection of a pronoun requires access to the antecedent (French pronouns) ornot (English pronouns, Italian null pronouns), speakers use fewer pronouns and more repeated nouns when the referentialcandidates compete more strongly in the non-linguistic context, whilst the similarities of their linguistic antecedents playno role. The finding provide support for the non-linguistic competition account.

We examined how speakers avoid gender ambiguous pronouns, exploiting cross-linguistic variations in French, Italian,and English. The gender congruence between two human referents led to fewer pronouns (more nouns) in both Frenchand English, whereas the grammatical gender congruence between two inanimate referents had no effect on the use ofpronouns in English, where grammatical gender is absent, as well as French, where grammatical gender is present. InItalian, gender congruence did not affect the use of null pronouns in all conditions, which are ambiguous regardless. Theresults are compatible with the non-linguistic competition account: Speakers avoid gender ambiguous pronouns only whenthe gender congruence increases their non-linguistic similarity.