Movement and the semantic type of traces [PhD thesis]
Ethan Poole
September 2017

This dissertation argues that there are only two possible semantic representations of movement: (i) a bound variable, i.e. trace, ranging over an individual semantic type, such as entities and degrees, or (ii) reconstruction back into the launching site of movement. Even though natural language has expressions over higher types, these expressions cannot be represented as traces, which only range over individual types. I call this constraint the Trace Interpretation Constraint. The novel empirical motivation for this constraint comes from a detailed investigation of movement targeting DPs that denote properties, a kind of higher-type expression. I observe that such movement obligatorily reconstructs and argue that the absence of nonreconstructed readings entails that movement cannot map onto traces ranging over properties. This investigation is complemented by existing and novel arguments against traces ranging over generalized quantifiers, another kind of higher-type expression. A second core claim of this dissertation is that the Trace Interpretation Constraint cannot be circumvented by type shifting an individual-type trace into a higher type, which I call the Trace Rigidity Principle. I show that there is a class of expressions that cannot be type shifted into property denotations, namely anaphoric definite descriptions, and argue that this class of expressions properly includes traces under Trace Conversion, thereby providing independent support for the Trace Rigidity Principle. According to the Trace Interpretation Constraint and the Trace Rigidity Principle, movement is tightly restricted in how it can be semantically interpreted. This dissertation also explores the dichotomy between the two representations permitted by the Trace Interpretation Constraint: leaving an individual-type trace and reconstruction. I develop a syntax and semantics of movement under which this choice is not free, but deterministic and does not require special LF interpretation rules. Therefore, a given movement derivation maps onto one and only one semantic representation. I argue that a deterministic system of movement requires multidominant representations. I demonstrate that this system of movement accounts for a number of disparate reconstruction phenomena without further ado.
Format: [ pdf ]
Reference: lingbuzz/003652
(please use that when you cite this article)
Published in: University of Massachusetts Amherst Dissertation
keywords: movement, traces, semantic types, type shifting, properties, multidominance, semantics, syntax
Downloaded:223 times


[ edit this article | back to article list ]