In this thesis, I investigate the collective/singulative systems of Welsh and Arabic which are shown to have the same descriptive behaviours as the English object mass system and the Mandarin Chinese general number system. Explicitly, object mass, collective and general number are shown to be number neutral and uncountable, and the singulative marker is classificatory in nature. Based on this, I propose an analysis where object mass, collective and general number systems share the same underlying primitive features. Following this, I also account for the differences between these systems, including: the role of construed aggregation in the Welsh, Arabic and English object mass/collective systems that is absent in the Mandarin Chinese general number system; why the singulative marker in Welsh cannot pluralise, but the singulative marker in Arabic may do so; and the degree of nominal flexibility (or lack thereof) of collective/singulative terms in ‘universal grinder’ contexts.
Collective/singulative languages offer a curious paradigm for meaning/form correspondences: alongside a singular/plural system, they experience a minor collective/singulative system where the singular form is morphologically marked. There are two prominent theoretical approaches for collective/singulative systems which have complementary strengths and limitations. Grimm (2012b, 2018)’s approach prioritises incorporating perceptual influence of construed aggregation at the lexical level to account for the fact that the collective/singulative class typically refers to individuals that come in groups or aggregations. However, it is not clear how this system captures grammatical similarities between collective/singulative and general number systems, which are not associated with aggregation. On the other hand, Mathieu (2012a)’s (et. seq.) approach prioritises crosslinguistic parsimony in the syntax, where the function of singulatives and classifiers are subsumed under a single syntactic head. However, it is not clear how this system captures how or why aggregation influences collective/singulative but not general number systems.
The analysis I offer unites the strengths of Grimm and Mathieu’s approaches in the style of de Vries and Tsoulas (2021)’s SUBPREDICATIVE ICEBERG SEMANTICS. I expand Subpredicative Iceberg Semantics by incorporating a Grimm-inspired aggregation semantics to account for collective/singulative distribution as a minor number system, but also propose functions of number that unite object mass, collective and general number under the number neutral uncountable umbrella, while treating classifier and singulative semantics as akin. I offer a split analysis of the plural of the singulative where number features are distributed across the nominal spine in a Mathieu inspired way. The semantics of the account is designed such that the plural of the singulative is attested in Arabic, but is ruled out under economy in Welsh. Finally, this thesis goes one step further than previous explorations of collective/singulative systems: I explore the use of collective/singulative terms in ‘universal grinder’ contexts (see Pelletier, 1975; Borer, 2005a; Cheng et al., 2008; Rothstein, 2017). I show that grinding effects for collective terms in Welsh and Arabic loosely parallel what is seen for Mandarin Chinese general number terms, however the grinding effects for singulative terms is not uniform across Welsh or Arabic (Welsh singulatives can be ‘ground’, Arabic singulatives cannot). The data questions whether there exists a single ‘universal grinder’ within and across languages, and subsequently I propose a multi-source approach to grinder readings.
