Subject positions in Old and Middle Irish

I argue that Old and Middle Irish (spoken: 7th-12th c.) had two subject positions: subject-1 and subject-2, as well as the post-posed position, identified by Mac Giolla Easpaig (1980). I use the presence of demarcating adverbs (e.g. danó 'also', íarum 'then', trá 'so', didiu 'moreover', etc.) to distinguish these two positions. It is shown that all types of subjects can occupy both positions, however, there are certain semantic restrictions on indefinites. Indefinites and quantifiers in subject-1 (pre-adverbial) have wide scope interpretations, while those in subject-2 (post-adverbial) have narrow scope interpretations. I show that this is especially true of indefinites/quantifiers/NPIs in the scope of negation, which must occupy subject-2. This can be understood as the effects of the Mapping Principle (Diesing, 1992). However, the presence of definites in both positions suggests that it is information structure that plays a major role in regulating the placement of subjects, since definites are not amenable to a Mapping Principle account. I show that subject-1 is reserved for old information and subject-2 is reserved for new information.