Auxiliary selection and the typical properties of subjects

I present a semantic analysis of two types of auxiliary selection patterns. Italian, Dutch, French, and Old Japanese have a pattern in which auxiliary selection is conditioned by verb class, and there is also a pattern active in many Italo-Romance dialects where the distribution of the two auxiliaries is dependent upon the person and number of the subject. I present a unified analysis of these two patterns that is rooted in prototypical vs. atypical subject properties, particularly with respect to affectedness, person, and number. I show that auxiliary selection indicates a typicality distinction on the part of the subject with respect to prominence mismatches along various semantic scales.