All Publications
All Publications
Supplementary Materials: The Typology of External Splits Greville G .Corbett University of Surrey Materials here elucidate and support the typology laid out in the main paper. Section 1 extends the account of agreement, and underpins the overall typology by showing that it is minimal (each element is essential). Section 2 analyses three significant sets of data, all relevant to the scope and validity of the typology; for each the issue is whether the best analysis involves postulating a split or whether there is a more insightful analysis. Finally, Section 3 makes clear the relation of the typology to work on differential argument marking. SECTION 1. CASE STUDY: THE AGREEMENT HIERARCHY. We turn here to a constraint on agreement systems, the Agreement Hierarchy. This will prove valuable since it shows how our typology allows us to separate out the component parts of complex splits. And it demonstrates that each element of our typology is indeed essential: for each of the four types of split, there are agreement examples which are split in that way, and others with no split. By demonstrating that each distinction is necessary, this section confirms that our typology is minimal. The Agreement Hierarchy has these positions (Corbett 1979, 2006:206–233). (1) The Agreement Hierarchy attributive > predicate > relative pronoun > personal pronoun The constraint is: For any controller that permits alternative agreements, as we move rightwards along the Agreement Hierarchy, the likelihood of agreement with greater semantic justification will increase monotonically (that is, with no intervening decrease). Supplementary materials for ‘The typology of external splits’, by Greville G. Corbett. Language 99(1).108–53, 2023. 1 There is a good deal of confirming data (Corbett 2006:206–237, 2022b), including luxuriant detail on Modern Hebrew (heb) in Landau 2016, and on English in Lakaw 2017. While the underlying logic is clear, instances often involve multiple splits and so are harder to spot. Take first a simple illustration of the constraint of the hierarchy: (2) that family are always making music together, and they evidently enjoy it a lot With hybrid nouns like family, attributive position allows only singular agreement, while in other positions there is also the possibility of plural (semantically justified agreement), as in the predicate and personal pronoun in 2. Primary lexemes are partitioned, separating hybrids like family and committee from nouns like spatula. Note that the partitioning goes further, in that different hybrids take different proportions of semantically justified agreement, thus family takes semantically justified agreement more frequently than does committee (Corbett 2015b:195–196).1 Clearly the Agreement Hierarchy constrains splits, categorical and gradient, and our typology of splits should be relevant. And indeed, we find that for agreement there can be a split — or no split — for each of the four possibilities of our typology. For the primary lexeme (controller) there can be lexemic splits or featural splits, or none; similarly for the secondary lexeme (target) there can be lexemic splits or featural splits, or none. Demonstrating both the existence and the absence of all four types of split will be clear evidence that our typology is minimal. The data are fully documented elsewhere, so this section will give the argument (summarized in Table 1) and point to the sources. I address the cells of Table 1 in turn. 1 The term ‘hybrid’ is established for nouns like family, which induce splits in agreement (they do not induce a consistent agreement pattern). And the term is used whether the split is categorical or gradient. But there is no generally accepted term for lexemes whose government, selection or anti-government requirements are inconsistent. Given the term ‘hybrid’, the term ‘split’ is less often used for agreement than for government or selection, but Sauerland (2004), for instance, talks of split agreement with relation to examples akin to 2. 2 evidence for split undifferentiated phenomenon primary lexemic split English committee Norwegian ‘pancake sentences’ featural split SC gazda ‘boss, landlord’ OHG wīb ‘woman, wife’ secondary lexemic split main Agreement Hierarchy effect real distance featural split SC deca ‘children’ normal situation TABLE 1. Summary of agreement evidence for the four-way typology Starting with the primary lexeme (the controller), we noted there is a lexemic split here…