摘要：Native speakers can reconstruct null arguments from a linear string of words where they are notoriously absent. How they solve this problem remains an unsolved and largely unaddressed issue. This article argues for a unified analysis of null arguments that relies on bare sensory input. The key assumption is that antecedent recovery takes place at the syntax-semantics interface if and only if the unvalued phi-features of a lexical element cannot be valued from the information available in the sensory input. The analysis unifies the theory of finite null subject pronouns and control and furthermore interprets null argument sentences without generating phrasal null pronouns. Evidence is considered from three languages with distinct null argument profiles: English (non-pro drop), Italian (consistent pro-drop) and Finnish (partial prodrop). Finally, the model was formalized and tested by means of a computer simulation.
