Agreement: some points of controversy

... does it matter to the syntax that verb agreement is realized spatially? (Lillo-Martin 2002:252)

The linguistic status of agreement in sign languages has been called into question by some authors who remark, among other things, on the selective nature of agreement (agreement is a phenomenon that is exhibited only by some verbs) and the lack of specification of verb inflection (the actual agreement markings are not listed in the lexicon) (cf. Table 3.5 for an overview of the controversial issues) (cf. Mathur & Rathmann 2012 for a discussion of the relevant literature).

Table 3.5: The status of agreement: controversial issues.

A detailed discussion of the arguments that have been subject to a longstanding debate is beyond the scope of this work. However, we shall briefly outline the position adopted in this work, which roughly maintains that the information encoded through the initial and final locations of agreement verbs is best characterised as verb agreement (for a detailed discussion of the main arguments see Lillo-Martin 2002: 249 f.). The main tenets of this approach are as follows.

Verb agreement and thematic structure. The choice of verbs participating in the class of verbs that agrees with their subject and object in person and number is not random but constrained by their s-selection features (i.e. their thematic structure) (see Rathmann & Mathur 2002 for a detailed discussion).

Grammatical information encoded. Feature sharing between agreement verbs and their arguments concerns particular syntactic roles (subject and object). The grammatical processes are also reflected in various syntactic phenomena such as the licensing of null arguments (Lillo-Martin 2002: 251).

Person distinction at the lexical level. Following Meier (1990) two categories of loci can be distinguished concerning their specification. The locus used to refer to the first person is listable (the location is fixed). The loci used to refer to a non-first person represent a bounded set of loci falling within the signing spaced which is, however, not listable. The infinity issue, as Rathmann and Mathur (2002: 377) put it, is thus one of listability.

Person distinction at the syntactic level. The dual first/non-first person distinction is relevant to syntax. For example, in the context of referential shift the first person pronoun may pick out a referent other than the signer (a characteristic reserved to the first person). By contrast, the difference between various non-first locations is irrelevant at this linguistic level (no syntactic process treats a location on the right differently than a location on the left) (Lillo-Martin 2002: 255).

Referential identity at the discourse level. The expression of co-reference involves choice of the same locus. Hence, pronouns with identical referential indices are interpreted as picking out the same referent at the discourse level (Lil- lo-Martin 2002: 255) (we will expand on the constraints relevant at this level in section 3.1.4.2).