Referential establishment and maintenance

Referential establishment and maintenance, expressed in sign languages through the linguistic use of sign space, is a complex phenomenon. As remarked upon previously (cf. section 3.1.4.2), knowledge from different levels of linguistic analysis needs to be integrated in a skilful manner. In the course of their narrative productions, competent signers use the sign space like a “referential map”: they pick out loci to associate them with referents, they might become a part of the map when they adopt the perspective of one of the referents - associating their body with the locus selected for that referent. As they use this map in their narrations they are confronted with the challenge of being consistent. Their narrative account will only be comprehensible if they control for the multiple intersections that make up what can be considered to represent a filigree linguistic network serving the purpose of providing a cohesive narration. So, what do the data reveal about the participants’ mastery of these complex tasks? What linguistic means do they use to establish and maintain reference? Do they use these devices in a consistent manner?

Linguistic means used. In general terms, the analysis of the data reveals that participants make use of various linguistic means to establish and maintain reference, including determiners, pronouns, agreement verbs, and referential shifts. Individual variation becomes apparent with respect to range and

frequency of the linguistic devices used. Muhammed’s file 1 examples in (113), repeated here in (362), illustrate a consistent use of referential loci to express referential identity, whereby the locus established via det_loc in (362a) is picked up by the agreement verb see and det_exist in (362b), and also by the agreement verb take in (362c).

As for the means used to establish referents, the analysis reveals that article determiners and pronouns are used fairly infrequently by all participants. Only Muhammed makes a more frequent use of these linguistic devices, not only in file 1, as documented in the previous examples, but also in file 3, that is, in the context of a narration that is remarkably lively in style, including several rhetorical passages, in which he addresses the audience by using a pronominal determiner directed toward the centre of the sign space. The choice of a locus on the sagittal axis corresponds with the canonical location of referents with whom signers interact when they shift reference (Perniss 2007: 1319). In Muhammed’s data we also find pronouns in reported dialogue contexts involving POVs, as is illustrated in (363), a passage in which the boy’s parents give their consent to his intention of going to bed. Notice that in the sequence preceding (363) Muhammed establishes the locus for the parents to his right, when he recounts that the boy tells his parents that he is tired. In this narrative passage, pronouns, determiners, and referential shifts are skilfully used to create cohesion. The loci are established contrastively, and they are used consistently.

In contrast to article determiners and pronouns, det_exist is used more frequently already in file 1, in particular by Muhammed, Maria, Fuad and Hamida. Maria’s systematic use of det_exist and agreement verbs for reference establishment and maintenance in file 3 is sketched in Table 3.48 on the basis of examples (202)- (203). (notice that the same pattern is applicable also to Muhammed’s examples (362)-(362) above).

Table 3.48: Expression of referential identity via det_exist and agreement verbs in Maria’s file 3.

Evidence of a systematic use of referential loci in the narratives of the third sample are indicative of the progress made by those learners whose file 1 narratives documented remaining gaps in this respect (compare Simon’s example (166) repeated in (364), in which the use of referential loci also patterns with the sketch provided in Table 3.48).

Choice of loci. As already remarked upon previously, participants vary regarding a consistent and contrastive use of loci picked up in the sign space. By way of illustration of this variation, we may consider the distribution of referential loci in the narratives of two participants, Muhammed and Hamida. Muhammed’s consistent and contrastive use of the sign space for referential establishment and maintenance in file 3 is illustrated in Figure 3.14, which represents a sketch of the distribution of loci along the horizontal, sagittal, and vertical axes. This participant also reassigns loci consistently to avoid referential ambiguity.

Figure 3.14: Contrastive use of referential loci in Muhammed, file 3.

The filigree network of referential loci used by Muhammed contrasts with Hami- da’s use of the sign space in her file 1. This participant, as we remarked upon previously u ses SRFs fairly frequently. POVs are marked non-manually, via a change in body orientation and eye gaze direction. While these linguistic means clearly distinguish FRFs from SRFS loci are not picked out contrastively, with the effect that different protagonists are associated with the same locus (for further illustration see Figure 3.15).

Figure 3.15: Referential loci in Hamida’s file 1.

Narrative passages remain difficult to understand, in particular, where protagonists are not introduced or reintroduced via other (lexical) means. It becomes apparent then that Muhammed’s and Hamida’s use of sign space reflect different strategies regarding the use of the sign space to mark reference in the narration of the frog story, which consists of multiple sub-events involving several characters. Variation in this respect has also been observed in the productions of adult signers. In his study on narrative development in BSL, Morgan (2000), points out that representational space is “cleared” and “reused” several times. Signers are reported to “divide sign space up at several levels, both by assigning different areas of sign space to different events and by overlaying different events in the same sign space” (Morgan 1999: 52). According to Morgan (2000), the reuse of sign space occurs through the signalling of a new perspective by an overt reference form. However, some authors have also remarked that signers do not always use overt linguistic means to mark a change perspective. In a discussion of person deixis in ASL, Meier (1990: 182) (pace Bahan & Petitto 1980; Loew 1984), reports that role-playing is not always marked by body shifts, with the effect that the same locus is used to refer to distinct individuals, whereby the distinction of referential identity occurs on the basis of the discourse context.

It is interesting to note that striking similarities between signed and spoken narratives become apparent where reference is not marked or expressed contrastively. Notice that failure to pick out referential loci in a contrastive (and consistent) manner in sign language discourse has a similar effect as the one observed in spoken language production, when narrators choose to use the same pronoun in events that involve different characters as is illustrated in example (365) (from Berman & Slobin 1994: 56). Consider, in particular, the referential ambiguity of the pronoun “he” in the latter proposition (“he starts running”) which we also repeatedly observed in our DGS data.

(365) And then he stands up on the rock and hangs onto some branches, then it turns out they are antler - a deer’s antlers, so - and he gets - he lands on his head and he starts running. (E9k - 9:11)

Incidentally, the example also illustrates nicely that we are confronted with a problem at the pragmatic or narrative level, as the utterances are well-formed at the local syntactic level.

Non-manual means. Non-manual means are commonly attributed a secondary role in referential establishment, as they are considered to be optional elements that might appear in addition to linguistic means used to pick out referential loci in space. However, our data analysis reveals that referential loci are established and maintained at times via non-manual means, in particular, in the context of referential shifts. This is illustrated in example (159) repeated here

in (366), a passage, in which Simon recounts that the boy falls on the deer and the deer then runs toward the precipice. Typically, Simon does not reintroduce the boy as a protagonist by using overtly expressed linguistic devices (NPs, pronouns), which reflects his choice of the boy’s perspective as the thematic perspective in his narration (we will take up this issue below, section 3.11.2.2). Where the recount of the boy’s activities involves referential shifts, the signer signals and marks POVs through a change of eye direction slightly to the right. This is the case also during the production of the POV in (366a), in which the audience is informed about the boy’s surprise after falling on the deer. As we remarked upon previously, referential identity in (366a) is problematic because the surprise could represent either the boy’s reaction -after falling on the deer- or the deer’s -realising that something has fallen on his back (the former interpretation being more likely because the deer is introduced only after this sequence, that is, in (366b)). As Simon goes on to narrate the deer’s surprise in (366b) he changes eye gaze direction once again, this time to his left. Notice that in this case, the protagonist (the deer) is introduced via an NP and det_loc. After the FRF in (366d), Simon again takes up the perspective of the deer, and recounts that the deer rears, throwing the boy down from his head. In this case, the POV is marked non-manually by picking out a locus toward the centre space in front of the signer, which coincides with the final locus of the spatial verb go signed previously. Hence, the locus associated with the deer is re-assigned by the directional verb go and correctly picked out in the following POV. Finally, the boy’s falling is expressed through an FRF.

The preceding observations make apparent that non-manual means (eye gaze direction, body orientation) constitute linguistic devices that are used, at times, as the sole markers to establish and maintain reference. While it goes beyond the scope of this work to discuss the status of these devices in detail, it seems that they might serve this function only where they are used consistently, as it is the case in Simon’s file 1.

Choice of story pictures as a substitute. Some participants associate referents with a locus that corresponds roughly with the location of the elicitation material (that is, the story book pictures). In Muhammed’s file 1, for example, the boy as a referent is associated with a locus to his left, that is, toward the location of the pictures. This is the case in example (109) discussed above, and repeated here in (367), in which pron_pers refers to the boy.

The example is illustrative of the relevance of considering the presence of elicitation material as a factor in the analysis of narrative data. Indeed, the presence of this material might not only influence the distribution of referential loci in the sign space, it might also have the effect that narrators choose not to provide some information explicitly. Moreover, we also observed that several participants do not recount the story (or extended narrative episodes) from memory but rather set out to describe the events in a picture-by-picture fashion. Not surprisingly, the constant checking of the elicitation material by the latter type of narrator affects the recounting of the story, the consistency in the establishment and maintenance of reference via referential loci and the use of non-manual means to mark reference.^[1]

Interestingly, a high frequency of deictic forms used to refer to protagonists (such as “this one”) and to pictures and locations (such as “here”) was also observed in studies using the frog story picture book to elicit spoken language

narratives. Berman remarks on the “excessive reliance on deixis and other inappropriate means of referring to characters and situations” in this type of picture-based study (Berman 2004: 269). According to Berman and Slobin (1994: 24), the choice of such deictic forms was found to differ in relation to whether or not the interlocutor of the children could see the picture book during the participants’ narration, as in the latter context such forms reduced to about 1.5%. No difference was observed with respect to the narrative abilities reflected in the data.

All in all, however, the effect of the elicitation material on the narratives collected in this study along the dimensions outlined remains minimal. Instead, what we can glean from the data is that the mastery of referential establishment and maintenance in narrative production involves an array of linguistic devices that are used to build up and control what can be conceived of as a filigree network of referential loci.

• [1] Johnston et al. (2007), too, remark on the impact of stimulus drawings in sign language datacollection, in that participants look at this material during their productions, which “interferedwith natural phrasing, such as pauses, head movements and eye gaze, and made the task ofestablishing clause boundaries difficult.”