Séminaire de Recherche en Linguistique

It is an old idea that the strength of syntactic categories may determine whether certain syntactic operations are possible; cf., e.g., Rizzi (1986) on pro-drop licensed by strong T, Pollock (1989), Roberts (1993), Vikner (1997) on verb movement licensed by strong T, or Chomsky (2013; 2015) on A-bar movement blocked by strong C. Syntactic strength is typically correlated with richness of morphological realization. However, as noted by Bobaljik (2002), if inflectional morphology is post-syntactic, strength cannot follow from morphological realization, which comes too late to license or block syntactic processes. The conclusion I would like to draw on this basis is that strength is an abstract inherent property of categories that

(a) determines whether or not syntactic operations can apply, and that

(b) determines post-syntactic morphological realization.  The task then is to make syntactic building blocks (operations, constraints,

rules) sensitive to different degrees of strength. Gradient Harmonic Grammar (Smolensky & Goldrick (2016)) is a new grammatical theory designed to implement effects of this type; here syntactic objects (features, categories) and constraints are associated with different weights capturing their strength.

In the talk, I will focus on a Gradient Harmonic Grammar approach to minimalist (feature-driven, phase-based) derivations that is based on three constraints, viz., the Phase Impenetrability Condition (PIC; Chomsky (2001)), a violable Merge Condition (Chomsky (1995)), and, crucially, a violable Anti-Locality Condition (Abels (2003), Grohmann (2011), Boskovic (2014), Erlewine (2016), Brillman & Hirsch (2016), Deal (2019), Pesetsky (2019)). Focussing on the movement types scrambling, wh-movement, and topicalization in German, I will show how a strength-based analysis accounts for systematic asymmetries between XP barriers (CP is always a stronger barrier than VP), between movement types (wh-movement can cross barriers more easily than scrambling, and topicalization can do so more easily than wh-movement), and between types of moved items (objects are more mobile than subjects, which in turn are more mobile than idiom part objects).

Finally, I show how the new approach may solve a classical conundrum arising given the substantial evidence that has been accumulated in cartographic studies for fine-grained systems of functional heads in each of the CP, TP, and vP fields: Typically, in any given sentence in any given language, many of these functional heads that are assumed to be always present in all structures of all languages seem to be inert (they do not license or block other syntactic processes and behave as if they were not really present); this can be captured straightforwardly in the new approach by attributing very little strength to them in these environments.