The grammatical information encoded in verbal morphology of sign languages, as we learned in section 3.1.3, comprises agreement with the object, or the subject and the object, aspect, and location. Studies on the acquisition of verb inflection
have largely focused on the acquisition of person agreement and the expression of spatial relations through complex classifier constructions. In this section, we will focus on the evidence obtained regarding the former. The latter will be taken up in section 18.104.22.168, in the context of the discussion of those properties that involve the syntax-discourse interface.
With respect to acquisition of verb agreement, the developmental pattern that emerges from the evidence discussed in several studies allows for the distinction of two broad developmental stages which, in our view, reflects the gradual development of the target structure (based on the theoretical outline provided in section 3.1.5, phenomena that are commonly described in terms of developmental phases are collapsed in this analysis).13
Stage I. Agreement verbs have been found to appear about half a year after the production of the first multi-word combinations (between age 2 and 2;5 [Baker et al. 2005] [2-2;6 in Schick 2003]). Before age 2 children do not produce agreement verbs, but rather use verbs that do not participate in the agreement system (Lillo-Martin 1999: 538).M Interestingly, early agreement verb forms produced appear in their citation form with a short movement in the neutral sign space (Lillo-Martin 1999: 538; Meier 2002, 2006). Such forms have also been observed at the initial stage of a late learner of DGS (exposure to DGS at age 3;7), one year after her exposure to DGS (cf. (78), (Leuninger & Happ 1997: 92, our transl.).
The few correct forms that appear at this stage in the productions of some children represent unanalysed forms (Morgan 2006: 35; Schick 2006: 109). Examples (79) and (80) illustrate a DGS learner’s use of the same verb form irrespective of whether the subject person is first or non-first (cf. Hanel 2005: 169, our transl.)i5.
- 13 Ages at which the emergence/productivity of phenomena were observed are provided in brackets; where age specifications provided in the literature differ, the respective references are provided.
- 14 Unfortunately, no data are available for DGS. The recordings in Hanel’s study started at age 2;2.
- 15 In the examples quoted, underscored numbers indicate first person (1), and second person (2) respectively (1c indicates contact with the signer’s chest).
Some authors remark on the children’s use of overt (lexical) arguments with citation forms (Lillo-Martin 1999: 554) (cf. example (81)); others observe the use of pointing to present referents (cf. Morgan 2006: 30) or pictures of a story-book to indicate arguments (compare example (82)). Evidence of the use of pointing with a movement between locations, as is the case in example (83), produced by a learner of BSL, indicates that, at a stage where verb marking is not yet available, learners might resort to the sequential expression of the relation between the verb and its arguments. Incidentally, the use of the index for this purpose is reminiscent of DGS constructions involving pam, that is, the sign used in that language to mark the agreement relation in constructions with plain verbs.
The sequential expression of complex meanings prior to the target-like simultaneous expression of meanings, either through the simultaneous combination of manual and non-manual components or through the modulation of elements in the sign space, seems to represent a recurrent developmental pattern in sign language acquisition (recall that this pattern was observed regarding the non-manual marking of sentence-types [section 3.2.1]; further, we will see in section 22.214.171.124 that lexical elements are used to introduce referential shifts prior to the use of non-manual markings).
It is interesting to note in this context that deaf children exposed to sign language might encounter this type of non-target-like use of agreement verbs in their input, as has been reported for children learning BSL and NGT respectively. In a study on the acquisition of BSL, Morgan et al. (2006: 34) provide a preliminary analysis of a mother’s BSL productions in the interaction with her child (referred to as child-directed-signing). Morgan et al. (2006: 34) remark on the alternative use of agreement verbs in their inflected and in their citation form, the latter accompanied with overt pronominals for subject and object (for example, you ask him), with the same verbs in the same session. Thus far, the effect of the inconsistent use of agreement markings in the input is unclear. As Morgan et al. (2006: 39) put it “[w]e are left with the problem of deciding which is the more important factor: does the child omit inflections because of performance
limitations in perception and production or because he observes omissions in the input?” Notice that this holds independently of the mothers’ motivations for their linguistic behaviour.16
A discrepancy between the input provided to children and the adult language use is also remarked upon by Van den Bogaerde (2000) in a study of the interactions between deaf mothers and their deaf and hearing children acquiring NGT and Dutch. As for the input provided to the children in NGT, the author (2000: 213) remarks that “[i]n general, we find only very few morphological markers on verbs in the signed input”. The high incidence of citation forms used by the mothers (more than 72%) was found to remain constant throughout the recording time (1-3 years of age) (Van den Bogaerde 2000: 212). Van den Bogaerde speculates, as does Morgan (2006), on the potential connection between the input to and output from deaf children (incidentally, the main topic of her work) (notice that for the children, a rate of 89% of citation forms is documented). On a critical note, these observations raise the critical question about when mothers change their signing behaviour and whether this would be related to changes in their children’s output (upon their attainment of the target grammatical properties), as would be assumed by proponents of the motherese hypothesis regarding spoken language development.
Stage II. Children begin to mark verbs for agreement with present referents several months after the production of the first agreement verbs in citation form (about 3-3;6 [Baker et al. 2005], [2;3-2;8 in Hanel 2005 for DGS]; [3;0 in Van den Bogaerde 2000:218 for NGT]). It is important to note in this context that there is some disagreement in the literature about the timing of agreement markings associated with present vis-a-vis non-present referents.
According to Hanel, children learning DGS mark agreement verbs productively for present and non-present referents at the same age. Examples (84) and (85) (from Hanel 2005: 224, our transl.) illustrate the establishment of loci for non-present referents, and the production of correct agreement forms where needed. Further, Hanel (2005: 223) also remarks on an increase of pronominal references to non-present objects or persons.
16 The problem is well known in the area of spoken language acquisition where the facilitating effect of motherese has been called into question. One important argument against its alleged use concerns the absence of metalinguistic information that would tell the child at which aspect she should pay attention and why, if the child was to understand this meta-information at all at this age (cf. Tracy 1991 for a detailed discussion).
An additional piece of evidence for the progression in the acquisition of the target grammatical properties of DGS is the productive use of the sign da (‘there’) (also glossed detexist, as we noted previously) to establish non-present referents, and to mark agreement (cf. examples (86)- (88) (Hanel 2005: 224, our transl.).
Hanel’s observations contrast with the findings obtained in studies on the acquisition of ASL and BSL, indicating that there is a temporal lag in the acquisition of verb agreement with non-present referents. The discrepancy is acknowledged by Hanel (2005: 267) who argues that it might be an effect of the methods used in the respective data collections (spontaneous DGS data vis-a-vis elicited ASL data). Unfortunately, the author does not elaborate on what she argues to be a well- known “delay effect” associated with elicited data. Nevertheless, and in line with Morgan et al. (2006), one might speculate on the impact of the different cognitive demands imposed by different tasks. “[M]astery of narrative”, as this author (Morgan et al. 2006: 27) remark, “involves additional cognitive demands that may influence the age at which inflections are used. The use of agreement in narratives with non-present participant roles develops late, with children showing a prolonged period of acquisition that continues past age 5;0 (ASL: Loew 1984; BSL: Morgan 2000), marked by the use of appropriate movements in agreement verbs but without identification of their arguments.” In this context, it is useful to recall the sophisticated use of referential frameworks in signed narratives described in section З.1.4.7. Given that this involves the interface between syntax and discourse, the issue will be taken up in section 126.96.36.199. Suffice it to mention here that late mastery comes as no surprise in view of the complexity of the task.
What is interesting for present purposes is that studies coincide in the observation of variation in the production of verb agreement markings. Hanel (2005: 242), for example, remarks that verb agreement is not applied across the board. The author remarks on a high incidence of errors, whereby both errors of omission (for example, verbs produced in their citation form) and of commission (for example, verbs with erroneous inflection markings) were observed. Example (89) illustrates the erroneous repetition of a verb, whereby the first form agrees with the subject and the second with the object, the intended meaning being “you inform the person filming”, cf. Hanel 2005: 233, our transl.):
Children learning ASL, too, have been found to occasionally mark the wrong arguments (for example, producing verb forms that agree with the direct instead of the indirect argument, compare example (90) (from Bellugi et al. 1988, cited in Hanel 2005: 132). Other types of error documented include occasional overgeneralisations of agreement marking with plain verbs (for example with the ASL verb eat to agree with the subject; Hanel 2005: 132 pace Bellugi et al. 1988). Notice that the latter type of error reflects the rule-based character of agreement marking at this stage, on the one hand, and the remaining challenge of learning the lexical properties of the verbs participating in the class, on the other hand.