On the origin of spines: Implications for an (emerging) Theory of Features

The concept of Spine (aka Extended Projection), a reworking of Chomsky’s (1986) Complete Functional Complex, appears to be analytically useful but remains theoretically elusive.

There have been several attempts to capture the uniformity of spines by invoking categorial uniformity of the heads making up a spine, typically along the lines of a categorial feature such as [N] or [V] (Grimshaw 1991 / 2003; Ouhalla 1991; Borsley & Kornfilt 2000; Panagiotidis 2011; 2015 among others). In these approaches a spine is what incidentally emerges when same-category projections are located in a sequence of c-command relations of the X [YP] (head-complement) type: spines are circumstantial or (at best) guaranteed by the ad-hoc relation of categorial Agree (Panagiotidis 2015).

At the same time, Platzack (2000), Grohmann (2003), and Wiltschko (2014) not only capitalise on the existence of spines but also identify separate domains within them, domains that are hardly co-extensive with phases. Wiltschko’s approach especially presents significant potential as an analytical compromise between the rigidity of cartographic approaches regarding the ordering of functional projections on the one hand, and the hand-waving of invoking vague ’interpretive factors’ on the other.

Based on joint work with Vitor Nóbrega, this talk will introduce an approach that incorporates both domains and spines within the actual definition of a formal feature, conceiving formal features as formative features [κ:Σ]δ where

κ corresponds to the universal domain, from which the formative was derived, e.g. Point-Of-View; Σ corresponds to the semantic unit individuating the formative, e.g. Imperfective; δ corresponds to the interpretive perspective that each formative feature attributes to its complement; if δ=[N], then its complement must be interpreted as a sortal, if δ=[V], then its complement must be interpreted as something extensible-into-time.

This approach renders the existence of spines and their domains not superfluous or accidental, but an actual outcrop of how features work as Borerian (2005) functors. The talk will demonstrate how different types of LIs (Lexical Items, the atoms of Merge) fall out from this approach to formal features.

The talk will conclude with some discussion, resulting from joint work with Antonio Fábregas, of how different κ’s (i.e. domains) appear to correlate with the actual granularity of the Σ in formal features.