1. Hornstein’s movement theory of obligatory control
The proper analysis of raising and control, illustrated in (1a) and (1b) respectively, has been one of the most important issues for some forty years within the framework of Generative Grammar.
In (1a), the matrix subject is thematically related to the infinitival complement in which it receives the Experiencer role from the embedded verb. This construction has been assumed to involve A-movement of the logical subject to the matrix [Spec, TP]. In (1b), however, the same subject is linked to both matrix and embedded verbs. This property motivates the standard analysis of Obligatory Control (oc) (Chomsky 1981), according to which the infinitival complement contains pro, controlled by the matrix subject. The pro has been motivated by the θ-Criterion at D-Structure, which bans movement into θ-positions. However, with the advent of the Minimalist Program (Chomsky 1995), which dispenses with all superfluous levels of representation including D-Structure, this option is a new analytic possibility. Following this minimalist logic, Hornstein (1999) proposes the Movement Theory of Control (mtc): a single dp receives a θ-role from the embedded verb and later moves into a matrix position to pick up another θ-role. Under the mtc, oc is a subspecies of A-movement on a par with raising; the grammatical operations underlying the two constructions are the same, with the only difference being whether the landing site of the movement is a θ-position or not.
The mtc has generated huge controversy. Some present empirical and conceptual evidence for it, while others strenuously argue against it. It is not my intention to review the large body of recent work in this area (Landau 2000, 2003; Culicover and Jackendoff 2001, 2003; Boeckx and Hornstein 2003, 2004, 2006; Bobaljik and Landau 2009; Boeckx, Hornstein, and Nunes 2010; among others). See Davies and [End Page 267] (2004, 2007) for two recent overviews. Instead, the purpose of this squib is to present new evidence against mtc from Standard Indonesian (si), based on the construction illustrated in (2).
2. Standard Indonesian:
Esti *(men)-coba memasak makanan Jepang.
Esti av-try cook food Japan
‘Esti tried to cook Japanese food.’
SI provides an excellent testing ground for the mtc because it possesses a unique morphosyntactic characteristic (Saddy 1991, Cole and Hermon 1998, Cole et al. 2008), namely the movement-sensitive distribution of the active voice (av) morpheme meN-, which helps diagnose the occurrence of movement of an np across oc verbs. It is widely known in the Indonesian/Malay literature that the movement of an np results in the obligatory deletion of the av morpheme within its extraction path. I provide evidence that the construction in (2) is an instance of oc in si. The mtc thus predicts that the av prefix must be deleted from the matrix control verb in (2), contrary to facts.
This squib is structured as follows. Section 2 reviews Cole and Hermon’s (1998) descriptive generalization that the movement of an np across a verb with meN- results in the obligatory omission of the prefix from the verb. Section 3 investigates the oc construction in si and provides evidence that meN- must remain prefixed with oc verbs. This distributional pattern of the av morpheme provides morphosyntactic evidence against the mtc.
2. The distribution of the AV prefix in SI
3. The obligatory omission of meng- with verbs that would otherwise permit meng-indicates the movement of an np argument over the meng- + verb.
(Cole and Hermon 1998:233)
I illustrate this generalization with A′-movement, A-movement, and the movement of an np vs. non-np in turn, using si examples. First, the examples in (4) show that wh-movement and relativization, two instances of A′-movement, cause av-deletion from the verb located within their extraction path. [End Page 268]
4. Standard Indonesian:2
In (4a), the movement of siapa ‘who’ crosses the matrix verb beritahu ‘tell’ but...