Agreement can be defined as a relationship of covariance between two or more sentential elements, as, for example, between subject and verb. In such a relationship, one element serves as the controller of the agreement relation, and one or more of the other sentential elements can be identified as the targets by virtue of some formal exponence that would not appear without this relationship. Handbooks of English usually agree that English has agreement between subject and verb at least in terms of person and number: I leave, he leave-s, the house stand-s here, the houses stand here. This agreement relation is only marked in the third person singular (non-past) by means of the suffix -s. We encountered another agreement relation in Chapter 3 in our discussion of systems of pronominal gender. Demonstrative pronouns show number agreement with their nominal heads (Chapter 5). Negative concord, as introduced in Chapter 9, may also be viewed as an agreement relation, though not a typical one. In the present chapter, we will be exclusively concerned with subject-verb agreement.
In his monograph on agreement, Corbett (2006: 1) begins his exposition by introducing a clearly false hypothesis – for didactic reasons, of course. According to this hypothesis, ‘grammatical information will be found only together with the lexical item to which it is relevant’. Corbett continues by stating that ‘[t]his hypothesis suggests a situation which is iconic, functional, sensible and understandable’. By way of illustration, he introduces English plural marking (dog/dogs) and the marking of the past tense (compute/computed), which fulfil the above criteria. Subject-verb agreement, and agreement in general, blatantly violates such a commonsensical perception of language as that which lies behind the above hypothesis. The main ‘problem’ of agreement is that it expresses information originating in another constituent. The verbal third person -s suffix of standard English mirrors information on the subject constituent.