A New Typology of Indefinite Pronouns, with a Focus on Negative Indefinites

This dissertation aims at providing a new typology of indefinite pronouns, and of negative indefinites in particular, based on a representative 179-language sample and a convenience sample of 20 languages. This work builds on Kahrel (1996) and Haspelmath (1997).

Negative indefinites have been proven difficult to define. In this dissertation, they are defined as indefinites with negation as their most important function. This definition includes negative indefinites with negation as their only function and negative indefinites with non-negative uses that still have negation as their most important function. With respect to the interaction of negative indefinites with clausal negation, two main patterns are distinguished: negative concord, involving the co-occurrence with clausal negation while yielding semantic negation only once, and the negative quantifier strategy, the pattern in which negative indefinites express negation independently. Negative indefinites involved in negative concord are called ‘n-words’ and negative indefinites involved in the negative quantifier strategy are called ‘negative quantifiers’.

In total, 28.5% languages from the large sample have negative indefinites, either n- words or negative quantifiers. The typological study of negative indefinites furthermore reveals that negative concord is not the most frequent pattern cross-linguistically, despite the claims that have been made in the recent literature. It occurs in 19% of the sample languages and shows geographical skewing towards Eurasian languages. In addition, languages are shown to differ considerably with respect to the nature of their negative concord systems. The negative quantifier strategy is predicted to be less frequent than negative concord and the reason is a functional one: clausal negation is predicted to be expressed clausally and not on the constituent. This is borne out by the data. 11.7% of the sample languages exhibit the negative quantifier strategy. There is geographical skewing towards Mesoamerican languages.

Morphologically, two types of negative indefinites are distinguished: morphologically negative ones and morphologically non-negative ones. Morphologically negative n- indefinites are assumed to arise via negative absorption, either of a clausal negator, a negative scalar focus particle, a non-verbal negator or a negative existential verb. Absorption of a clausal negator is by far the most frequent type. Morphologically non- negative n-indefinites arise via the quantifier cycle, which refers to the semantic change that indefinites can undergo from non-negative elements to negative ones. With respect to the interaction between clausal negation and the diachrony of negative indefinites, negative absorption leads to the dispreferred strategy, which is predicted to be restored. Restoration is more often found in languages with indefinites that have absorbed negative scalar focus particles than in languages with indefinites that have absorbed clausal negation. A low degree of lexicalization can be assumed for the indefinites in those languages in which restoration has not taken place. The other pattern leading to the dispreferred negative quantifier strategy, viz. the Jespersen Cycle, is predicted to be very rare and perhaps even restricted to Europe.

On the basis of similarities between the negative concord and double clausal negation, a relation has been claimed to exist between the two phenomena. Negative concord has been claimed to be a condition for double clausal negation. This claim is refuted. The sample languages show that negative concord and double clausal negation are basically unrelated phenomena, which can in certain cases exhibit an interesting interaction.

The sample languages that do not have negative indefinites use non-negative forms to convey negated indefiniteness. The typology suggested here is based on Kahrel (1996), but adds some finer distinctions. In particular, I added a strategy for languages that use the same forms for universal quantification and negative quantification. This type is found in African languages. I also tried to find out whether a language uses a negative polarity indefinite or a non-specific indefinite, in case it uses a ‘special indefinite’ in negation (terminology of Kahrel (1996)). If a language uses a polarity-sensitive indefinite, it mostly seems to concern a negative polarity item rather than a non-specific indefinite.

Apart from a typology of indefinites in negation, this dissertation proposes a new map to cover the cross-linguistic variation in the functional distribution of indefinite pronouns. The map remedies some of the problems of the map suggested in Haspelmath (1997). On this new map, I tried to consistently map meanings and contexts to yield functions that correspond to meanings-in-context. It is different in four main respects. Firstly, it takes into account the lexical semantics of indefinites. Widening and non-widening indefinites are distinguished. Widening indefinites are shown to functionally differ from non-widening indefinites when used in identical contexts. Secondly, unlike Haspelmath’s map, the map allows one to categorize an indefinite as a negative indefinite, despite possible non-negative uses. Thirdly, it adds a function for universal quantification. Fourthly, it maps indiscriminacy readings of free choice items.

The new map is taken to the test on the basis of a biased convenience sample of 20 languages.