Towards a label-less grammar

The goal of this dissertation is to introduce a novel label-less grammar, namely a generative approach in which all labels and equivalent devices are absent from all components of the grammar (i.e. the lexicon, narrow syntax, and the Sensorimotor (SM) and C(onceptual)-I(ntentional) interfaces). More concretely, this label-less grammar aims to oust four types of (sub)labels which have hitherto been prominent in mainstream generative approaches, namely lexical category labels, functional category labels, X-bar theoretic sublabels, and uninterpretable features. A precise model of a label-less grammar is proposed as follows. First, a single, pre-syntactic lexicon is assumed (e.g. Chomsky [1995] 2015), in which there is crucially no division between ‘lexical’ and ‘functional’ items, and all listed items can only be specified for phonological content and either conceptual content and/or an entity denotation. Second, narrow syntax is argued to be centered on a modified form of Zwart’s (2004, 2012) Unary Merge – crucially a form of asymmetric Pair-Merge – as the sole form of Merge. Third, regarding the SM interface, Zwart (2004, 2012) and Kayne (2013, 2023) are followed in assuming that linear order overwhelmingly matches Merge order, and a novel Linear Order is Merge Order (LOMO) inference principle is introduced. Fourth, and arguably most importantly, a novel C-I interface condition – Merge as Specify – is proposed, which posits an asymmetric semantic relation between the two items involved in all applications of (non-First) Merge, based on the semantic part-of relation. This label-less grammar model is then applied to the analysis of a large range of both ‘nominal’ and ‘verbal’ phenomena, with – given both the number of and the widely studied nature of these phenomena – a primary focus restricted to data from English. Chapters 1 and 2 set the scene for this label-less grammar account. Chapter 1 shows that, whilst all four of the aforementioned types of (sub)labels have played a central role in previous mainstream generative approaches, the modern generative enterprise – the Minimalist Program (Chomsky [1995] 2015, et seq.) – is to a certain degree at odds with the positing of some if not all of them. Further, the chapter critically reviews various recent attempts to reduce the burden on these (sub)labels (e.g. Boeckx 2009, 2011, 2013, 2014; Borer 2005a,b; Chomsky 1995, [1995] 2015; Collins 2002; Collins and Seely 2020; De Belder 2011; Seely 2006). Chapter 2 then introduces in detail the label-less grammar model outlined above. Next, Chapters 3 to 6 turn to the first of the two empirical foci, namely that on the ‘nominal’ domain. Chapter 3 provides general background on previous, broader generative approaches to ‘nouns’. Chapters 4 to 6 then introduce a label-less grammar account of ‘nouns’: Chapter 4 focuses on the nature of divisions, Chapter 5 on quantity, and Chapter 6 on definiteness. Taking these chapters on ‘nominals’ as a foundation, Chapters 7 to 11 then proceed to provide a label-less grammar analysis of core phenomena in the ‘verbal’ domain, and the ‘nominal’-‘verbal’ interactions at their heart. More specifically, Chapter 7 begins with an overarching proposal about the broader nature of ‘verbs’, Chapters 8 to 10 then provide a detailed examination of Aktionsart – particularly but not only the (a)telicity distinction –, before Chapter 11 proposes a novel analysis of present tense ‘verbal’ agreement in English. Finally, Chapter 12 summarises the findings of the preceding chapters, and highlights ideas for future research, including tentatively proposing how a label-less grammar analysis might be extended to account for cross-linguistic variation. More specifically, I claim that the single, pre-syntactic lexicon posited in a label-less grammar is the core locus of cross-linguistic variation in it.