Achievement

The new field of *processing linguistics* developed by this IGERT uses methods of theoretical linguistics to address problems in psycholinguistics, such as "is grammatical knowledge in the language production system categorical or gradient"? Now compound words such as 'book case' are not believed to be stored in memory as such: they are produced by using the mental grammar to combine 'book' and 'case'. Is 'book case' harder to produce than, say, 'book rate'? No such gradient difficulty should exist if grammatical knowledge is categorical; it could not arise from statistics of stored items because compounds are not stored. But in his dissertation work experimentally studying both brain-damaged and intact speakers, trainee Ariel Goldberg shows that such gradience does exist - and can be explained by a linguistic theory in which grammatical constraints (e.g. "avoiding consecutive 'k' sounds") are "soft" rather than "hard": violating them produces gradient not categorical ill-formedness.