Pernilla Hallonsten Halling

Parts of speech in general

The comparative concepts for attributive adjectives, predicative adjectives, and adverbs introduced in section 4.2 rest on Croft’s typological approach to parts of speech. In chapter 3, I showed that Croft’s approach allows for an expansion that includes adverbs (cf. table
3.1 in section 3.2), although they are not overtly included in his original version (see table
2.2 in section 2.3.4). The present section will be devoted to an in-depth discussion of parts of speech generally, and Croft’s approach specifically.
Within the generative tradition, at least nouns and verbs are assumed to be universal categories. When Chomsky (1981: 48) introduced the features [+N] and [+V] for the determination of lexical categories, the scope of these features excluded adverbs. Nouns were defined as [+N][–V], verbs as [–N][+V], and adjectives as [+N][+V]. Outside of the realm of lexical categories, prepositions were defined as [–N][–V] (1981: 48). The option of defining also some types of adverbs as [–N][–V] has been proposed (Emonds 1985; Huddleston & Pullum 2002). However, this does not include adverbs denoting properties, but rather those that are close to prepositions. Another alternative is to treat adverbs as “a special class of adjectives” (Radford 1988: 138). Adverbs have received much attention in certain generative accounts, notably in the works of Cinque (1999; cf. section 2.2). However, adverbs appear to be dealt with in different ways, with no clear consensus on their category status.
Among functionalists, opinions diverge on whether nouns and verbs (and other cate- gories) are universal or language-specific. The views also vary on what type of categories parts of speech are as such. As discussed in chapter 2, one prominent perspective bases parts of speech on discourse functions (also propositional act functions, discourse func- tions, Dixon 1982 [1977]; Hopper & Thompson 1984; Croft 1991, 2001, 2003). A different approach is that of Functional Grammar (Hengeveld 1992; Hengeveld & Rijkhoff 2005; Ri- jkhoff & van Lier 2013, cf. section 2.3.3), where universal functions are considered as the basis of parts of speech, although the latter cannot be identified in every language. The debate-provoking account of Mundari by Evans & Osada (2005) posits language-specific word classes, by both distributional and semantic criteria. From a categorial grammar perspective, Gil (2000, 2008) sees part of speech categories as purely syntactic categories. In contrast, Nau (2016) argues for parts of speech as entities not of language, but of linguistics.
Although typologists take different perspectives on parts of speech, there are two ap- proaches that specifically aim to account for such categories in a way that is typologically valid. Such validity requires a definition of parts of speech that allows for comparison of these categories across languages. The two approaches are the discourse function-based approach most elaborated upon by Croft (1991, 2001, 2003) and the Functional Grammar approach by, among others, Hengeveld (1992), Hengeveld & Rijkhoff (2005), and Rijkhoff & van Lier (2013). These typologically oriented approaches, which were introduced in chapters 2 and 3, will be discussed accordingly in more detail here. The two approaches differ in important ways. According to Croft, parts of speech are prototypes, in the sense of Rosch (1978), that can be identified when certain semantic classes are used for specific discourse functions, with unmarked lexical items as a result. As illustrated in table 9.1 (repeated from table 2.2 in chapter 2), these prototypical parts of speech pattern with

3.