Uninterpretable features and EPP: a minimal account of language build up and break down

Like most L2 speakers, speakers of early L1, children with SLI, and Broca’s aphasics lack the or¬dinary speaker’s ability to produce utterances on line with almost no errors. Often they produce well-formed utterances, but quite frequently they produce errors that are almost non-existent in the speech of normal grown-up speakers. In my paper I will suggest, based on empirical material from Swedish, that this behavior can be understood as a performance problem: these speakers have the same knowledge of the target language as ordinary speakers, but cannot automatically adjust their produc¬tion to the language specific distribution of EPP, i.e. the demand to express a particular gram¬matical relation overtly. To obtain this I adopt an idea put forward by Pesetsky & Torrego (2001) that EPP is to be distin¬guished from the grammatical relation itself: EPP is connected to an uninter¬pretable feature, forcing Agree and Move to apply to avoid a violation of the interface condition.
The distribution of EPP is language specific, hence my account correctly predicts cross-lan¬guage variation for the four groups of speakers discussed here.