This chapter identifies the central theoretical-empirical claim of MP, the Merge Hypothesis (MH). It rehearses the motivations for a simple combination operation that takes two objects, combines them in the simplest way possible, and treats the combination so constructed as capable of further combination. I review and explicate the claim that the simplest combination operation would do no more than combine its inputs. This means that the combination operation should not impose a serial order on what it combines, nor should it change the properties of what it combines in any way (as either would involve more than “mere” combination). So construing “simplicity” implies the No Tampering Condition (a principle that forbids changing the structures of the elements combined) and supports the idea that expressions so formed have set-like structure. I further provide a more technical specification of the combination operation by specifying its inductive definition. I then show how to derive a bunch of recognized properties of natural language Gs from this Merge conception of combination and review eight of these, again largely following and elaborating Chomsky’s earlier suggestions.

This chapter focuses on the syntax of wh-question patterns that have been referred to as long-distance or unbounded dependency constructions. Starting with core dependency properties of wh-constructions, this chapter reviews the main problems that movement approaches encounter when attempting to represent the link between the filler wh-phrase and its corresponding gap. It develops a declarative, feature-based analysis to capture the linkage between the filler and the gap, while resolving problems originating from movement analyses. The key mechanisms of analysis are the ARC (Argument Realization Constraint), which allows any argument to be realized as a GAP element, the HEAD-FILLER CONSTRUCTION, which licenses the combination of a filler and an incomplete sentence with a nonempty GAP value, and the NIP (Nonlocal Inheritance Principle), which regulates nonlocal features like GAP in relevant mother phrases. The interplay of these construction-based mechanisms allows us to license a wide variety of wh-constructions: main-clause nonsubject wh-questions, subject wh-questions, wh-indirect questions, non-wh indirect questions, infinitival indirect questions, and even adjunct wh-questions.