This chapter analyses the relation between mass noun phrases and measure readings within Landman’s theory of Iceberg semantics. Landman not only applies Iceberg semantics to different noun classes, but also generalizes the theory to apply to DPs and different interpretations of measure phrases. These developments are then used to address the main puzzle in the paper: ‘when mass counts’. Based on data mostly from Dutch and German, Landman proposes an explanation for why not only neat mass nouns, such as furniture, admit of cardinality comparison readings with quantifiers, such as most, and are felicitous with stubbornly distributive adjectives, such as big, but why it is also the case that mess mass nouns, such as meat, can also get cardinality comparison readings in the same contexts. For the latter case, the key to the analysis is to make use of contextual portioning, a process that is independently motivated in the portion readings of measure phrases.

In this chapter, we will offer a new way of analysing the syntax and semantics of the mass/count distinction at the syntax-semantics interface, by synthesising the constructionist framework originating in Borer (2005) with the (lexicalist) `iceberg semantics' proposed in Landman (2011, 2016. We believe that this synthesis has several conceptual and empirical advantages over existing frameworks. It combines the flexibility and morphosyntax-driven nature of constructivism with an explicit role for human conceptual categories such as INDIVIDUAL and SUBSTANCE. It allows us to distinguish between different kinds of mass/count shifts and makes explicit why some of them are harder than others. Furthermore, it clarifies the distinction between stuff-reference, number neutrality and non-countability, three distinct (although interdependent) nominal properties that are often explicitly or implicitly conflated in existing accounts of the mass/count distinction.