Enlarged arguments in Bantu: Evidence from Chichewa Silvester Ron Simango Abstract This paper re-examines two types of constructions that have featured in the discussion of possessor ascension in Bantu: one type – an applicative construction – is associated with alienable possession, and the other – non-applicative – is associated with inalienable possession. The study shows that the former expresses affectedness, and that the possessor reading arises only by construal; whereas the latter expresses a part–whole relationship. The paper argues that the two constructions differ in more significant ways than has previously been acknowledged; and that their distinct derivations cannot be captured by traditional possessor ascension (PA) analyses. The putative ‘‘alienable’’ possessor constructions belong to the class of (benefactive) applicative constructions and should be analyzed as such. The paper proposes that the socalled ‘‘inalienable’’ possessor constructions can best be accounted for by positing the existence of ‘‘enlarged arguments’’ wherein the possessum functions as a nominal predicate which more narrowly pinpoints the locus of the action described by the verb. 1. Introduction Many languages of the world provide for at least two ways of expressing the possessor relation: in one, both the possessor and the possessum appear in a single noun phrase headed by the possessum, as shown in (1a); and in the other, the two nominals are expressed as distinct constituents, as shown in (1b). (1) a. The horse kicked Penny’s shin (Levin, 1993:71) b. The horse kicked Penny in the shin In this pair (1a) is said to be the basic structure from which (1b) is derived through a processes known as possessor ascension (PA) whereby the possessor is ‘raised’ to assume the grammatical function of its host NP (see, for example, Baker, 1988a; Davies, 1997; Hyman, 1977; Hyman et al., 1970; Kim, 1989; Kimenyi, 1977, 1980; Perlmutter and Postal, 1983a; Scotton, 1981, and related works). In Bantu languages there are two distinct constructions which are said to be manifestations of PA: in the first type the verb is inflected with an applicative affix, but in the second the affix is absent from the verb. The following examples from Chichewa illustrate. 2. a. Tadala a- na- thyol -a ndodo ya-mwana SM-PST-break-FV stick ASSOC-child ‘Tadala broke the child’s stick’ [Lit: ‘Tadala broke the stick of the child’] b. Tadala a- na- thyol -er -a mwana ndodo SM-PST-break-APPL-FV child stick ‘Tadala broke the child’s stick’ (or ‘Tadala broke a stick for the child’) 3. a. Mphatso a- na- thyol -a mwendo wa-mwana SM-PST-break-FV leg ASSOC-child ‘Mphatso broke the child’s leg’ [Lit: ‘Mphatso broke the leg of the child’] b. Mphatso a- na- thyol -a mwana mwendo SM-PST-break-FV child leg ‘Mphatso broke the child’s leg’ In both (2) and (3), (a) represents the underlying structure from which (b) is derived: that is, the possessor is raised from its underlying position and becomes the surface direct object. Note that in each case the raising of the possessor results in the deletion of the prepositional element and the concomitant creation of a double object construction. Thus, (2b) and (3b) are supposedly similar in every respect except for the fact that theverb is inflected with an applicative suffix in (2b) whereas in (3b) it is not. Note, though, that (2b) has two potential meanings: one possible meaning is that the stick in question belonged to the child; and the other meaning is that the stick may not necessarily have belonged to the child, but that the child was nevertheless affected (either in a beneficiary or maleficiary sense) by its being broken. The significance of the ambiguity in (2b) will become clear in the sections that follow. The variation in verbal morphology is said to be directly linked to the alienability of the possession: it is postulated that PA triggers an