Adjunction, Movement and Primitive Syntactic Operations
Many descriptive generalisations have been established concerning the characteristic properties of adjuncts: for example, (1) adjuncts are optional and iterable, (2) they are typically islands (Huang 1982), (3) they are able to avoid reconstruction (Lebeaux 1988), and (4) they can be extraposed rightwards to a strictly local degree (Baltin 1981). This talk addresses the question of why these properties should cluster together, and proposes an analysis of adjunction which unifies the various characteristic properties that have been observed.
To do so, I develop a system where movement and adjunction are not independent phenomena, but rather are related byproducts of the same underlying grammatical machinery. This relies on the interaction of two crucial ingredients that I take from previous work. The first is the observation that in neo-Davidsonian semantics adjunction generally corresponds to a simple mode of semantic composition (namely predicate conjunction) that is not mediated by grammatical roles or thematic relations, in contrast to predicate-argument relations (Hornstein and Nunes 2008). The second is the intuition that syntactic movement might usefully be thought of as "re-merging", which I implement by drawing on insights from formal computational analyses of minimalist syntax (Stabler 2006). I show how together these ingredients yield a system in which adjunction and movement are revealed to be two sides of the same coin. The system therefore makes strong predictions about how adjunction and movement should interact, and these predictions unify the distinctive properties of adjunction listed in (1)-(4) above.
