Pernilla Hallonsten Halling

Conclusion

As a prototypical category, adverbs cannot be defined as either lexical or functional. The semantic types attested among property words are closely related to other types that are less semantic and more grammatical in their meaning. Such a relation bridges the dis- tinction between lexical and functional, for instance, in the case of speed and time. Nor can the adverb category be determined by the structural coding or behavioral potential of grammatical categories. It may thus come as a surprise that a typological approach that emphasizes the universal tendencies of structural coding and behavioral potential can be used as basis for determining adverb as a part of speech. The characteristics of no structural coding (i.e. derivational morphemes) and very little behavioral potential (agreement, comparative and superlative constructions) for unmarked adverbs devi- ates from the characteristics of other unmarked parts of speech. However, unmarked adverbs are still attested as simple property lexemes used as modifiers within predicating expressions. This follows the tendency for unmarked items to appear in the prototypical combination of semantic class and pragmatic function. The semantics of these adverbs also follow strong tendencies. In this way, adverbs emerge as a clear prototype category, in spite of its lack of structural coding and rarely attested behavioral potential.


One objection to adverb as a part of speech may still be the relatively few simple property lexemes that occur in this function in many languages. A solid argument against so-called ‘splitting’ approaches (cf. Croft 2001: 78) to parts of speech is that single- member categories are not desirable. From such a perspective, languages with a single simple adverb denoting speed, such as Estonian ruttu ‘quickly’, are dubious candidates for having an adverb category.³ But such conclusions are drawn about individual languages, whose categories are necessarily language-specific, or even construction-specific. From a cross-linguistic perspective, the fact that Estonian has one simple speed lexeme used as a modifier within predicating expressions (or alternatively, is on its way to lexicalize such an item) falls into the unmarkedness pattern of the prototypical adverb category. In a cross-linguistic prototype category, languages with few simple adverbs must be considered just as well as languages with many. A parallel can be drawn to Dixon (1982 [1977], 2004), who refers to languages with very few adjectives. Against the background of adjective as a generally acknowledged part of speech, only four or five items seem perfectly acceptable as constituting an adjective class in certain languages.
Just as Croft’s map of parts of speech can be amended to include adverbs, so can Givón’s (1984: 52ff.) time-stability scale presented in (9.1) be expanded to include adverbs as they are defined in this thesis.

(9.5)
nouns - - - - - - - - - - adjectives - - adverbs verbs
———————————————————————————————
most time-stable intermediate states rapid change

Alternatively, general modifiers may occupy the middle region of intermediate states in (9.5). Adverbs then, are less time-stable than adjectives, but more so than the verbs that they modify. General modifiers can be expected to show different degrees of time stability reflecting this state of affairs. Both Givón’s time-stability scale and Croft’s table of parts of speech elucidate the intermediateness of adjectives and adverbs. Against the results of overlapping encoding presented in chapters 6 and 7, adjectives, adverbs, and general modifiers can be expected to display different kinds of overlaps to various extents, since they are intermediate by nature.