Syntactic movement and agreement production Julie Franck The major theoretical issue in psycholinguistic research on agreement is the modularity of sentence production, according to which some aspects of syntactic construction proceed in relative insulation from other components of language (semantics, phonology). However, this view of a syntactic module, characterized by informational encapsulation, leaves aside another important property of modules, that is, their domain-specificity. Domain-specificity supposes operating principles that cannot be reduced to general properties of cognition (at least as they are described to-date), but whose description requires particular assumptions that are specific to syntax. The work I will present links experimental psycholinguistics and theoretical syntax with the aim of identifying such syntactic principles at play in the realization of subject-verb agreement. I will report a series of sentence completion experiments in which the syntactic position of a potentially interfering word is manipulated, and so called 'attraction' errors are recorded (e.g., *The friend of the neighbors come). I will argue that broad concepts like 'encoding unit' or 'hierarchical structure' are insufficient to account for some of the fine-grained syntactic effects found. Rather, I suggest that principles identified by recent developments of generative grammar provide potential explanations for the experimental effects reported. These principles involve (1) precedence and dominance relations between the words in the hierarchical structure, and (2) specific intermediate syntactic positions that do no surface in the uttered sentence. I will briefly describe these principles, show how they account for the data, and how they generate interesting predictions for further research.