In her classic paper "Oats and Wheat: Mass Nouns, Iconicity and Human Categorization," Anna Wierzbicka (1988) argued the case for the existence of numerous, subtly different, subclasses of mass nouns and postulated detailed explanatory links between underlying conceptualizations and grammatical behaviors. She also stressed the partly language-specific character of these subclasses and suggested that differences between languages are often related to culture (e.g., connected with different eating and food preparation practices). In this study, I aim to extend and improve on Wierzbicka's arguments and analyses, concentrating on concrete mass nouns in English. The two overriding points of the entire study are that the formal linguistic properties of mass nouns are systematically correlated with their conceptual content, and that this conceptual content can be identified with rigor and precision using appropriate methods of linguistic semantics. The analytical framework is the Natural Semantic Metalanguage (NSM) system of lexical semantic representation (Goddard and Wierzbicka 2002; Wierzbicka 1996).

In a global perspective, the language of mental state concepts displays a great deal of variation - much greater than imagined by most cognitive scientists. Almost all the words in the English lexicon of emotion and cognition are language and culture specific, i.e., they do not have exact meaning equivalents in many, perhaps most, other languages. This includes words for emotional and attitudinal states, such as sad, angry surprised, anxiety, and grief, words for epistemic states and cognitive processes, such as believe, doubt, and remember, and words for ethnopsychological constructs, such as mind, heart, psyche, and memory (Russell, 1991; Wierzbicka, 1999; Harkins & Wierzbicka, 2001; Palmer, Goddard, & Lee, 2003; Shweder, 2004; Amberber, 2007; Schalley & Khlentzos, 2007). Most cognitive scientists underestimate not only the scale of semantic variation across languages, but also the theoretical and methodological challenges it poses. In theorizing and discussing emotional states, they tend to take English for granted, effectively absolutizing the English lexicon of emotion and cognition (for example, assuming that words such as sadness, anger, and surprise represent natural psychological categories), while denying the same privilege to the lexical categories of other languages. At the methodological level, many researchers seem to regard the "problem of translation" as a mere nuisance that can easily be overcome by tagging indigenous concepts with English glosses. In the process, many cross-cultural studies are seriously flawed by inaccurate translations and concomitant "terminological ethnocentrism."

This study explores the conceptual semantics of numbers and counting, using the natural semantic metalanguage (NSM) technique of semantic analysis (Wierzbicka 1996; Goddard & Wierzbicka (eds.) 2002). It first argues that the concept of a number in one of its senses (number₁, roughly, "number word") and the meanings of low number words, such as one, two, and three, can be explicated directly in terms of semantic primes, without reference to any counting procedures or practices. It then argues, however, that the larger numbers, and the productivity of the number sequence, depend on the concept and practice of counting, in the intransitive sense of the verb. Both the intransitive and transitive senses of counting are explicated, and the semantic relationship between them is clarified. Finally, the study moves to the semantics of abstract numbers (number₂), roughly, numbers as represented by numerals, e.g. 5, 15, 27, 36, as opposed to number words. Though some reference is made to cross-linguistic data and cultural variation, the treatment is focused primarily on English.