Off the Chain

Minimalism has no agreed-upon approach to chains. The key problem, as noted by Collins & Groat (2018), is that an all-you-need-is-MERGE logic (Chomsky 2008, Chomsky et al. 2019) is not enough to distinguish repetitions (selected from the Lexicon) from copies (taken from the derivational workspace that carries the computation). Chomsky (1995, 2000, 2001), for instance, resorts to Numerations (NUM) to track the distinction, which has to assume the paraphernalia of indices. Alternatives presuppose, e.g., multi-dominance or an independent operation (like COPY) to implement the distinction (cf. Chomsky 2005, Epstein et al. 1998, Gärtner 2002, etc).

Chomsky (2008) pursues yet another approach where MERGE, plus phase-level memory, is supposed to capture the distinction. In this paper (part of a bigger project), we argue that chains could be understood as making use of a process of Case tokenization, which we take relates to the operation AGREE. Our approach has an obvious connection with locality, since it invokes phases by directly involving C and v* (assuming these are the loci for φ-features). From this, we will argue that this very phase involvement correlates with the well-known distinction between A and A’ chains, which we try to refine as well.