Classifier agreement

In sign language linguistics, handshape units in verb forms that express a meaning related to the subject or the object of a given sentence have received much attention because of the complex information that is expressed simultaneously through constructions that contain these units. The varying terminology used to refer to these units and the constructions they appear in reflect the ongoing debate about their linguistic status (cf. Schembri 2003: 4). As a detailed discussion of the appropriateness of the use of the notion of classifier in sign languages vis-a-vis spoken languages (cf., for example, Slobin et al. 2003: 272) is beyond the scope of this work, we will continue to refer to these handshape components as classifiers and assume that the set of classifier morphemes in a given sign language represents a set of bound morphemes that cannot constitute a word on their own (cf. Sandler 2006: 193; cf. also Benedicto & Brentari 2004). Although some scholars treat classifiers as a separate system because they are reminiscent of gestures, classifier constructions “are not pantomimic analogues” (Sandler 2006: 193). Rather, classifier constructions are rule-governed and they involve a finite set of handshapes and movements (Sandler 2006: 193). Further, as pointed out by Sandler (2006: 194), “[t]he individual morphemes in the classifier subsystem, each a minimal pairing of form and meaning that recombines productively with other morphemes in the system, must be assumed to be independently listed in the lexicon, like other morphemes.” Slobin et al. (2003: 272) highlight the referential function of these sign components which would consist in identifying or designating discourse elements. These authors emphasise the central role of the classifier system as “a flexible discourse tool” that serves to maintain reference and construct coherent and cohesive discourse (Slobin et al. 2003: 272), an aspect we will take up below when we discuss complex classifier constructions (section 3.1.3.4).

Verbal classifiers can be distinguished into two types depending on the argument they agree with, namely subject classifiers and object classifiers (cf. Gluck & Pfau 1997: 42). For ease of reference we provide a summary of the classifiers referred to in this work in Table 3.2 (reference to the corresponding semantic type of classifier is provided in brackets).^[1]

Table 3.2: Verbal classifiers (based on Happ and Vorkoper 2006: 175).

Subject (also: class or entity) classifiers and spatial verbs. Spatial verbs (verbs of motion and location) take class classifiers (also referred to in the literature as entity classifiers, cf. Gluck 2005: 185; Perniss 2007). These verbs agree with specific subject features, reflected in the choice of the handshape (note that the phonological feature of handshape is not determined lexically, which is similar to the lack of specification of beginning and end locations in agreement verbs, see section 3.1.3.5). There is a limited set of handshapes used to express this type of agreement. In DGS, for example, the B-hand form is used to refer to the class of vehicles with four wheels (hence, this would be the handshape used in example (41), from Happ & Vorkoper 2006: 156, our transl.). Constructions with body part classifiers, a sub-group of class classifiers, involve a shift of perspective or referential shift (cf. (42)) (we will discuss referential shift in section 3.1.4.3).

Object (also: handle) classifiers. In some verbs, the choice of the handshape relates to the physical properties of the object or the way the object is manipulated (cf. (43)-(44), from Happ & Vorkoper 2006: 160) (cf. Schembri 2003: 22

for a detailed discussion of the properties involved).^[2] Verbs that belong to this group (cf. the DGS verbs in (45)) are so-called verbs of object transfer (in Happ & Vorkoper’s terms kausale Bewegungsverben, ‘causal verbs of motion’, plus the verb give). Notice that the verbs agree with all their arguments (with the direct object via classifiers, and with subject and indirect object via movement).

In constructions with objects that have been previously specified with a SASS (size and shape-specifier) classifier the verb takes up features of this object through the bound handle-classifier morpheme (notice that SASS classifiers are free classifier morphemes that may fulfil the function of an adjective, cf. Happ & Vorkoper: 2006: 155. SASS classifiers are not used as a morphological part of the verb, but are only used for the introduction of the referent later referred to by the verb, cf. Gluck & Pfau 1997b: 4).

• [1] The terminology used in the literature to distinguish between different types of classifiersvaries. Although we will also use notions based on semantic distinctions, we believe that a distinction based on grammatical roles is particularly useful for a study focused on grammar development.
• [2] In Happ & Vorkoper’s (2006: 161) analysis these are objects with the theta-role THEME. Slobinet al. (2003: 279), by contrast, state that the handshape generally "incorporates the patient ofthe verb”. Schembri (2003: 22) remarks on both the role of ‘patient’ or ‘theme’ assumed by thereferent (and, at times, the instrument role).