Sign language acquisition: diagnostic criteria

In this section, we elaborate a working proposal on what we assume constitute the main milestones in the development of the target sign language structure based on the evidence discussed in the previous sections. Unfortunately, as our review of the available literature makes apparent, there is a lack of longitudinal studies that would provide further insights into how learners develop the target

structure, for example, by identifying the main developmental milestones as it is commonly the case in spoken language acquisition research. Such developmental studies, as those undertaken on the acquisition of German, provide further insights into (a) the progressive development of the target structure and associated grammatical processes, and (b) the spectrum of variation observed at the different developmental stages.

With a few exceptions, notably the studies undertaken by Chen Pichler, Lil- lo-Martin, and Hanel, there is a persistent lack of theoretically founded studies that would address current issues in the broader field of language acquisition. The authors mentioned coincide in their assumption of a continuity view of development, which basically assumes that the target structure is available early on, and that language-specific properties are specified equally early. The alternative hypothesis of a gradual development of syntax (cf. section 2.2.2) is not taken into consideration. Instead, structural characteristics and grammatical processes in language development are largely regarded in isolation (for example, the relation of word order and verb inflection remains unaccounted for).

In our view, however, the Structuring-building hypothesis elaborated previously (cf. section 2.2.2), accounts better for the changes documented in the development of sign language learner grammars. Following this hypothesis, we assume that the evidence obtained in sign language acquisition studies reflects a progressive expansion of the structure of the learner systems, in accordance with the evidence obtained from the input. Further, we argue that intra-individual variation is bound to transitions between stages, in line with the UG based dynamic model of language development we proposed in section 2.2.3.

In our working proposal about the development of DGS we distinguish different milestones (cf. also Table 3.11). Note that the characteristics at each stage serve also as diagnostic criteria in the evaluation of children’s learner grammars which is why the descriptor is followed by an indication of the respective developmental phase in brackets in the following summary.

VP structures (Phase I): No evidence of grammatical processes. VP structures observed at the beginning of sign language development (cf. Hanel 2005: 208) are categorial-thematic in nature (cf. Radford 1990). Grammatical processes related to the functional projection IP run vacuously. Hence, word order at this stage may vary. There is no checking of subject-verb agreement. Neither can null arguments be licensed, as the necessary structural relations are not available (notice that the empty elements, that is, pro vs. topic drop are not distinguished, which will only be possible once INFL is in place, cf. Hanel 2005: 222). The few agreement verbs that are produced at this stage appear in their default or citation form. That referential loci are not established overtly at this stage also reflects

the lack of the relevant abstract features; the same holds of errors concerning the first/non-first person distinction (Hanel 2005: 266).

IP structures (Phase II): verb inflection, pam, complex classifier constructions. IP structures and their (language-specific) associated features are reflected at the level of word order. Structural requirements are in place for grammatical processes like verb raising and feature checking to become operative. Verb agreement and classifier inflection morphology is productively used at this stage. Following Hanel (2005: 259) the establishment of non-present referents goes along with productive verb agreement marking. The temporal coincidence comes as no surprise given the inter-relation between both mechanisms and relevance of abstract (grammatical features) for their mastery. An additional piece of evidence for the availability of the IP concerns the use of pam with plain verbs and adjectival predicates.^{[1] [2]} Although the use of pam was not an issue investigated by Hanel, this author (2005: 244, our transl.) provides one example with the agreement marker (cf. (99)). The sequence illustrates nicely how the sign is used to mark agreement in a construction with an adjectival predicate.

CP structures (part of phase II or a milestone of phase III?): complex sentential constructions, interrogation, referential shift (POV). CP structures and their associated features are reflected in the production of wh-questions, embedded clauses, including those that involve referential shift. Thus far, the development of complex structures and referential shift have only received little attention (there are no available studies on DGS learners). Further, there is no consensus on whether the structural expansion by a CP layer coincides with the projection of the IP. Hanel (2005) assumes that the acquisition of the inflection system with its feature specifications goes along with the activation of an additional syntactic position, namely, a topic position that allows for the licensing of empty elements in constructions with plain verbs (by assumption, a specification of the C-system)22.

Syntax-discourse interface. Learners of a sign language like DGS are confronted with the task of integrating information from different levels of linguistic analysis. In our work, we have focused on the following phenomena:

• (a) Referential establishment and maintenance. Several linguistic devices are used in sign languages to establish and maintain reference (cf. Table 3.6 above for an overview). Verb inflection, for example, involves the picking out of referential loci to mark agreement; yet the consistent use of these loci throughout a narrative to create cohesion, as well as the contrastive choice of loci for different referents represent phenomena that are modelled by discourse requirements. Hence, learners need not only have a command of the processes associated with the IP, that is, the domain of grammatical relations that hold between syntactic constituents (e.g. subject-verb, verb-object). Their mastery of DGS verb inflection is related also to (a) lexical competence (involving the distinction of plain, agreement and spatial verbs), (b) morphological competence (involving the inflection of verbs to encode their arguments [first / non-first distinction, classifier selection, spatial relations], and (c) discourse (involving the overt marking of coindexation).
• (b) Referential shift. The skilful use of fixed and shifted referential frameworks involves (a) lexical knowledge (in particular, where verbs select for a POV complement), (b) the IP level (that is, the level at which grammatical relations, including agreement, need to be marked) (c) the CP level (that is, the level at which referential shift is signalled and marked), and (d) the discourse domain (modelling the choice of loci for the purpose of creating cohesion).
• (c) Reference forms and their functions. We have seen also that signers deal with the functional dimension of the linguistic devices they choose to use to make reference. The challenge here is not only to make an appropriate selection among various lexically overt reference forms. Because DGS is a discourse oriented language and a pro-drop language, learners have to acquire the grammatical constraints and learn the discourse requirements that need to be met in the use of null elements. As they make a choice among the different linguistic forms available, signers will also have to consider the referential function these forms might serve depending on the respective discourse context they appear in.
• (d) Spatial relations. Finally, the intricate interaction of information from distinct linguistic levels also becomes apparent in the expression of spatial relations in the narration of the story characters’ locations or movements. Notice that variation regarding the degree of detail provided is determined, on the one hand, by the overall organisation of a narrative, and, on the other hand, by grammatical requirements. Beyond the issue of narrative style, the question that arises from a developmental perspective concerns the availability and use of the relevant linguistic devices necessary to express spatial relations. The use of complex classifier constructions, for example, involves the competence to integrate informa

tion from different levels of linguistic analysis, including (a) morphosyntax (verb inflection), (b) syntax (word order, in particular where lexical antecedents are involved), and (c) discourse (where h2-classifiers background information provided previously).

Table 3.11: Working proposal about structure-building in DGS.*

*To illustrate the structure-building process, structures are provided bottom-up (from Hanel 2005)

• [1] Notice, that Van den Bogaerde (2000: 218) too, though not in the context of a structural analysis, remarks on the production of the auxiliary “op” at the time subject verb agreement beginsto be marked in the acquisition of NGT.
• [2] The assumption that the activation of the INFL features goes along with specification ofanother functional category raises the question about a potential modality effect. Accordingto Hanel (2005: 268, our transl.) this is so because the R-locus carries referential features as aphi-feature: “It is understood that the location features must be made available discourse grammatically”. For this purpose, referential relations are made visible as overt indices: “It is the overtpresentation that could have an influence on the effect observed in acquisition.”