Shifted reference: grammatical aspects
Independently of the functions it may fulfil, referential shift always has the same formal characteristics (Herrmann & Steinbach 2007: 160). Some of the formal constraints are described in the following.
Syntactic contexts of RS. Referential shift has been found to be obligatory in some contexts, as for example, in constructions with embedded questions (Herrmann & Steinbach 2007:160, our transl.):
Happ and Vorkoper (2006: 465f.) point out that the shift of perspective is obligatory in constructions with verbs that select a constituent clause as a complement (cf. (62) and (63))[1]. Note that the verbs listed in (62) denote the attitude of the signer to the statement of the constituent clause, whereas verbs listed in (63) select an imperative or an interrogative constituent clause.
Further, referential shift in DGS has been found to be obligatory in imperative constructions, but optional in constructions with modal verbs (e.g. have-to). Finally, several scholars have remarked on the use of referential shift in constructions that involve a reference to body locations not associated with the signer. In ASL, for example, the expression of the proposition “John shaved Bill’s nose” (cf. (64), from Bellugi et al. 1990: 18) involves a serial verb construction, in which the verb expressing the location follows a referential shift after the expression of the activity through the verb without a specification of the body location.[2]
Pronoun reference and agreement in RS. As we mentioned previously, referential shift affects the reference of first person pronouns and verb agreement (Lillo-Martin 1995: 158; Meier 1990). In shifted referential frameworks new referential loci are automatically established (Bellugi et al. 1990: 18). In addition, one or more loci may be reassigned overtly. In example (65), the referent associated with the first person index changes.
Note that 1pronoun in ASL can serve as a logophoric pronoun in addition to its use as a first-person pronoun (Lillo-Martin 1995: 161). In the former case, the pronoun is interpreted as coreferential with the subject of the matrix clause (rather than with its regular referent, the speaker), a phenomenon that is observed in reported dialogue contexts.
Few authors have specifically addressed the issue of agreement in the context of reported action, which contrasts with the analysis of reported dialogue. In the latter, the shifted predicate (also point of view or POV predicate) is realised as a non-manual 2-place agreement marking, with the first argument being the signer and the second the addressee. As for reported action, we will assume here that the dual perspective characteristic of constructions with reported action predicates is reflected in different agreement phenomena (Happ & Vorkoper 20 06: 567):
- - body part classifiers agree with the subject (the shifted referent adopted by the signer)
- - class (or entity) classifiers are not affected by the perspective shift and therefore agree with the object (or the thematic role of THEME, in Happ and Vorkoper’s terms).
Non-manual markers in RS. In referential shift, the signer often assumes the facial expression, eye gaze and head movements of the character described, onset and offset being linguistically constrained (Emmorey & Reilly 1998: 81). In addition, fixed and shifted referential frameworks differ concerning the referent to which affective behaviours expressed are attributed to: while facial expression, body posture and non-linguistic gesture are attributed to the narrator in the former (“plain” narration in Emmorey and Reilly’s terms), non-manual affective behaviour is attributed to the character portrayed in the latter. The contrast is illustrated in examples (66) and (67) (Emmorey & Reilly 1998: 83).
For referential shifts in quotation environments involving a matrix clause, non-manual marking often begins on the verb of the matrix clause (Herrmann & Steinbach 2007: 160). It should be noted, however, that because referential shifts involve a shifted reference of the locus associated with the signer’s body (that is, the locus of the referent whose words or actions are reported) the introducing matrix verbs need not be expressed, as the non-manual marking unambiguously refers to the person speaking or carrying out an action (cf. Happ & Vorkoper 2006: 465). Hence, this is an instance of non-manual agreement marking.
- [1] Verbs are listed in German in the original in the context of a discussion of the syntactic characteristics of German vs. DGS.
- [2] Morgan and Woll (2003: 303) describe a similar phenomenon for BSL in terms of what they describe as referential shift in paired constructions, involving (a) Verb A, which moves away from thesigner’s body, describing the action (e.g. girl paints), and (b) Verb В followed immediately by thesame verb moving toward the signer’s body with B-Cl(body classifier) (boy painted on the face),representing the signer’s affected body part. Between the two verbs there is a referential shift.
