1 Word class universals and language-particular analysis MARTIN HASPELMATH Max Planck Institute for Evolutionary Anthropology (September 2021) The study of cross-linguistic word class variability and its limits needs to be based on clear comparative concepts because languages have different grammatical structures. Here I show that the notions of (i) function indicators and (ii) semantic root classes give us a way of formulating universals and of distinguishing cross-linguistic macro-classes in a rigorous way. The root classes are action, property, and object; function indicators are markers that signal an unusual propositional act function (e.g. a copula signals predicative use of an object root, and a relativizer signals modifying use of an action root). The paper argues for distiguishing clearly between comparison and language-particular analysis, because analysis must be based on language-particular constructions and is thus a rather different enterprise than the search for language universals. 1. Three word class universals This chapter considers both how word classes are analyzed or described in particular languages (in §3), and what we can say in general about major word classes. We will see that these are two questions are less directly related than is often thought. Let us start with some important word class universals in this first section, as well as some cross-linguistic macro-types of word class patterns in §2. Word class universals are not particularly well- known, although many comparative linguists are aware of them and will not be surprised by the Universals 1-3 given below, even if they have not read Croft (1991; 2000). Much of the comparative literature on word classes asks to what extent the distinction between nouns, verbs and adjectives is universal (e.g. Sasse 1993; Evans 2000; Baker 2003; Dixon 2010: Ch. 11). This question (which I refer to as the “distinctness question”) has no clear answer, so it will be discussed only later (§4), and I first focus on universals (§1) and cross-linguistic macro-types (§2). Universals 1-3 (due to Croft 1991: Ch. 3) are about the occurrence of function indicators, i.e. copulas, attributivizers, and termifiers, in three different propositional act functions: predication, modification, and reference, with three different semantic root classes: action roots, property roots, and object roots. All these terms will be explained and discussed further below, and the universals will become clearer after they are exemplified below. Universal 1 If a language has a copula, i.e. a special form that indicates predicative function, it is used with object roots and/or property roots. Universal 2 If a language has an attributivizer, i.e. a special form that indicates modifying function, it is used with action roots and/or object roots. Universal 3 If a language has a termifier (= nominalizer or substantivizer), i.e. a special form that indicates referential function, it is used with property roots and/or action roots.