Abstract
The relativization systems of most Slavic languages include relative pronouns that can be conventionally labelled as 'who' and 'which' and differ in a number of logically independent parameters (etymology, animacy, grammaticality of attributive contexts, and morphological distinction for number and gender). Prior research has shown that the choice between 'who' and 'which' in Slavic languages is largely dependent on the head type.
Some of the languages allow the 'who' pronouns to be used with pronominal heads, but not with nouns in the head, while in others, the pronominal heads in the plural are also ungrammatical with the pronoun 'who.' The present study aims to complement the available qualitative data on the distribution of the relativizers with quantitative data and to propose a unified account for all the observed tendencies. A corpus-based study was conducted in order to establish language-internal statistical tendencies comparable to the known grammaticality restrictions.
The results show much agreement between the qualitative and quantitative tendencies. Thus, the head 'those,' unlike the head 'that,' is incompatible with the relativizer 'who' in Slovak, Polish, Upper Sorbian, and Lower Sorbian languages, while the same tendency is quantitative in Czech, Slovene, Serbo-Croatian, Ukrainian, Belarusian, and the older varieties of Russian.
Corpus data suggest that there is also a stronger tendency for the relative pronoun 'who' to be avoided with the head 'those' than with the head 'all.' One more relevant parameter is the semantic type of the clause, maximalizing semantics being the preferred option for 'who.' I suggest that all these and some other tendencies can be subsumed under a macro-parameter of the extent to which the head is integrated into the relative clause.