2.2.5 Verb agreement
Verb agreement in sign languages has been analysed in many different ways. A discussion of the competing analyses (e.g. stem-internal changes versus affixation) – not to mention the question whether agreement even exists in sign languages – is outside the scope of this paper (inter alia Liddell 2000; Neidle et al. 2000; Meir 2002; Rathmann & Mathur 2002; Aronoff, Meir & Sandler 2005). In the present context, it is sufficient to sketch some basic properties of sign language agreement.
Agreement inflection is realised manually in signing space. Signers localise referents in the signing space, usually by means of pointing signs (indexes) which identify locations in signing space. Indexes may occur before, after, before and after, or simultaneously with the referent that is localised. These locations (or the location of referents present in the discourse) are used in pronominalisation and verb agreement. To that end, the signing space is usually divided into ‘sections’ analogous to the grammatical category of person. As for pronominalisation, pointing to the signer thus reflects first person, pointing to the addressee second person, and pointing towards any other location in signing space third person (Figure 3). Similarly, the movement of some verbs can be modulated such that the beginning and end point of the movement coincide with previously established locations, thereby expressing agreement with the subject and object. The NGT verb visit, for instance, when describing a movement trajectory from 3a to 1, would be interpreted as ‘s/he visits me’.
Whereas this form of verb agreement is attested in many sign languages from all parts of the world, there are also several sign languages in which no, or only a few, verbs can be modified to show agreement. In Kata Kolok, a sign language from Bali, for instance, the only verb that is spatially inflected with some regularity is the verb baang (‘give’) (Marsaja 2008). AdaSL also shows an infrequent use of verb agreement; it occurs with verbs such as marry and insult (Nyst 2007). In other sign languages, like NGT or ASL, many verbs inflect.
It is important to note, however, that even in those sign languages that make frequent use of agreement verbs, not all verbs inflect. In analysing ASL, Padden (1988) refers to the class of non-agreeing verbs as plain verbs. These verbs cannot be modified in the way described above to express agreement with their arguments, mainly due to phonological restrictions. Since Padden’s (1988) work, the distinction between agreeing and plain verbs has been found in many other sign languages (see e.g. Bos (1993) for NGT). Interestingly, even for those verbs that can inflect for agreement, the realization of agreement appears to be optional (see e.g. de Beuzeville et al. (2009) for Australian Sign Language). Furthermore, agreement verbs do not automatically agree with all of their arguments. Transitive verbs may agree with only the object, and the same holds for ditransitive verbs (Padden 1988; Meier 1987). Whereas in spoken languages, agreement with the subject is the unmarked case (see Figure 1), in sign languages, object agreement seems to be less marked and more common.
Across spoken languages, the relevant agreement features are spelled out in many different ways (see e.g. the table in (7)). In contrast, across sign languages, the (phonological) realization of verb agreement is strikingly homogenous: it always involves similar spatial modulations. There is typological variation in the use of agreement auxiliaries which are capable of realizing agreement in the context of plain verbs (see Steinbach & Pfau (2007) for an overview). Some languages have such an auxiliary, for example NGT and DGS, while others do not, for example ASL. Still, overall, the typological variation in the realisation of verb agreement among sign languages is limited. We conclude that, as far as agreement is concerned, sign languages fall into two types: those that that make frequent use of manual verb agreement and those that show minimal (or no) use of the agreement mechanism. In §4.1, we will show that IUR does not fall neatly in either of these two groups, as it shows agreement, but only to a limited extent.