Agreement Without A-Positions: Another Look at Algonquian

1 Introduction

Branigan and MacKenzie (2000, 2002)argue that cross-clausal agreement (CCA)in Innu-aimûn (Central Algonquian) is an example of Ā-agreement. In this squib, we propose that all agreement in Innu-aimûn and other Algonquian languages is in fact Ā-agreement. As is well known, agreement targets are selected on the basis of relative animacy rather than grammatical relations in these languages. Following Bruening (2001), we assume that they have a designated position for the argument that is highest on the animacy hierarchy.¹ However, we argue that this is an Ā-position, given its function, position, and interaction with other syntactic processes.

We propose that the absence of A-agreement derives from a property that characterizes all Algonquian languages: either they lack TP altogether, or T is not specified for Case or D-features. The evidence for our claim comes from the lack of other A-syntax phenomena related to Spec,T, including Case, Case-motivated A-movement, and A-binding.² On our proposal, arguments are projected into the vP, licensed by familiar θ-relations. They do not move for Case agreement, but only for discourse purposes such as topic or focus, or for wh-questions. We suggest that in the absence of A-movement and A-checking, these languages rely instead on Ā-positions for agreement triggers and landing sites for movement.

2 Cross-Clausal Agreement Is Not Case Related

The presence of abstract Case is discernable in morphological case and in Case-motivated movement, but Algonquian languages appear to have neither.³ In this section, we discuss one type of movement that is clearly not motivated by Case considerations. In section 3, we argue that Algonquian has no passive, though it may have a passivelike movement, which again is not Case related.

In the Algonquian languages, under certain circumstances, the verb can agree with an argument in a subordinate clause. This phenomenon is called cross-clausal agreement (CCA),⁴ and one might be tempted to assume that it is analogous to exceptional Case marking (ECM; also known as copy/raising to object). In his original description of CCA in Blackfoot (Plains Algonquian), Frantz (1978) noted that the construction is reminiscent of English ECM, illustrated in (1).

(1)John wants [_IP me to visit him].

We follow Branigan and MacKenzie, however, in analyzing CCA as an instance of Ā-agreement, and not raising. As described by Branigan and MacKenzie (2002), CCA occurs in Innu-aimûn when a verb that selects a clausal complement optionally agrees with an animate argument of that clausal complement. This is reflected in the form of the matrix verb but does not affect the morphology of the embedded verb. In particular, verbs that select clausal complements normally appear in the transitive inanimate (TI) form, but in CCA they are realized in their transitive animate (TA) form. The argument that triggers object agreement on the matrix verb is either the subject or the object of the embedded clause. This is illustrated in (2)and (3). In (2a) and (3a), there is no CCA, and the matrix verb appears in its TI form. In (2b), the matrix verb agrees with the embedded subject 'John and Marie'; in (3b), it agrees with the null embedded object 'you'. The CCA morphemes on the matrix verb are in bold; the overt trigger of CCA is underlined. Note that the form of the embedded verb is invariant in each pair of examples.

1. 2. Innu-aimûn

   1. a. Ni-tshissenitamu-ânân mûpishtuât Shûshepa Tshân
      1PL-know-TI-1PL     visit     Joseph   John
      mâk Mânî.
      and Marie
      'We know that John and Marie visited Joseph.'

   2. b. Ni-tshissenim-ânân-at mûpishtuât Shûshepa Tshân
      1-know-1PL-3PL     visit     Joseph   John
      mâk Mânî.
      and Marie
      'We know that John and Marie visited Joseph.'
      (Branigan and MacKenzie 2002:388, (3))

2. 3. Innu-aimûn

   1. a. Ni-tshissît-en   kâ-uîtshi-shk     Pûn utâuia.
      1-remember-TI PRT-helped-3/2PL Paul father
      'I remember...