A Cyclic Optimization Approach to the A/A-bar Distinction in Syntax

• Applicant: Professor Dr. Gereon Müller
• Subject Area: General and Comparative Linguistics, Experimental Linguistics, Typology, Non-European Languages
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 439622645

Classical approaches to A and A-bar movement based on designated positions (Chomsky (1981; 1993), Deprez (1989), Mahajan (1990)) have been shown to be empirically problematic because they are at variance with movement types showing variable behaviour with respect to A and A-bar properties. They are also conceptually dubious because defining the relevant positions is far from simple, and because grammatical building blocks must explicitly refer to the relevant information (e.g., reflexivization demands A binding, not just binding, parasitic gaps require an A-bar binder, not just a binder, etc.). This is incompatible with basic minimalist tenets. Upon closer inspection, current approaches based on features rather than positions (van Urk (2015), Colley & Privoznov (2021), Scott (2021), Chen (2023), Lohninger (2024)) face exactly the same problems. Empirically, despite some claims to the contrary, a mixed behaviour of movement types (e.g., of scrambling) can only be accomodated in a very limited way, with absence of intervention effects emerging as basically the only A-bar property of otherwise A-like movement types that is easily accountable for (e.g., the approaches in van Urk (2015) and Lohninger (2024) do not permit a co-occurrence of parasitic gap licensing and weak crossover circumvention with a single movement type, but such a pattern is well established). Conceptually, grammatical building blocks regulating the interacting operations (reflexivization, bound variable pronouns licensing, parasitic gap licensing, reconstruction, improper movement, etc.) still need to make reference to the status of movement as A or A-bar. Furthermore, it is not quite clear what features should be identified for determining A and A-bar movement: phi features and information-structural features, although widely adopted for this purpose, can be shown not to suffice in many cases. The conceptual problem with all these recent approaches is addressed by the analyses in Safir (2019), Boskovic (2024), and Hewett (2024), which postulate that there is a categorical difference between A and A-bar movement that can be traced back to the counter-cyclic addition of a shell to A-bar moved items in the course of the derivation. Again, however, these approaches face problems with a mixed behaviour of movement types (and there is no independent evidence for the shells). In view of this state of affairs, the project pursues an approach in terms of derivational timing according to which the A/A-bar distinction reduces to the question whether c-command is possible or not at the relevant stage of the derivation: If movement precedes some interacting process, there is feeding and bleeding (``A movement''); if movement follows the other process, there is counter-feeding and counter-bleeding (``A-bar movement''). This approach can account for both categorical and mixed behaviour of movement types by relying on restrictive concepts of cyclicity and optimization.

• DFG Programme: Research Units
• Subproject of: FOR 5175: Cyclic Optimization