On the (questionable) use of a basic pattern: early SVX

In our discussion of the main developmental milestones in the acquisition of German (cf. section 4.3) we remarked upon an initial stage (VP stage) at which learners produce elementary structures, whereby the relative order of the elements might vary because grammatical processes run vacuously at this stage. At the same time we also acknowledged that, regarding verb placement, studies into L1 acquisition of German agree in the observation of a higher proportion of verb final structures (compare example (644), repeated here for convenience).

Turning to the evidence obtained in our study, it becomes apparent that some of the early productions we assume to be based on VP grammars differ in two respects from the early utterances of L1 German learners, that is, (a) in the use of a rather rigid word order, and (b) in the low frequency of sentence-final verb placement. Indeed, although some early narratives reflect a rather creative use of different word order patterns, the potential of free word order at the VP stage is not exploited by all learners. Word order in Simon’s narratives, for example, follows the SVX schema across the board (disregarding sequences with verb drop). Interestingly, this word order pattern is reminiscent of the early productions of L2 German learners with an L1 Romance. Because the VP-headedness parameter in Romance languages is fixed to the head-initial value, it is commonly assumed that the learners’ Romance L1 and their L2 German input (containing surface V2 clauses) conspire during the initial stage, reinforcing the initial preference of SVX formats. By contrast, learners with an SOV L1 language have been found to initially produce verb-final structures (compare example (645), also repeated here for convenience).

Given that DGS is an OV-language (cf. section 4.1.1.) the question arises as to why there is no reinforcing effect from the L1 that would be reflected in the participants’ preference of SOV orders. Put differently, is there no interaction between these two languages at this level? To answer this question, we have to acknowledge that a substantial amount of the German input these learners are exposed to is provided in the context of a formal teaching/learning situation. So the question to ask is rather: what are the characteristics of this input that would affect the early productions in such a way that they are more similar to the utterances of L2 German Italian learners than to L1 monolingual learners?

In this respect, it is important to note that the teaching of German, at least at the beginning, is oriented towards the inhibition of the learners’ creativity by focusing on the learning of the canonical surface SVO order. Indeed, in the domain of deaf education in Germany, including bilingual education programmes, there is a general consensus that the mastery of this basic sentential format represents an essential step in that it allows learners (a) to produce elementary structures that conform to the surface canonical order of the target language and (b) to develop an awareness about the necessary differentiation of German and DGS (cf., for example, Schafke 2005: 292, and Plaza-Pust 2016 for a critical appraisal of Schafke’s assumptions).

From a psycholinguistic point of view, however, the advantages attributed to this didactic approach might be called into question: learners are encouraged to use a syntactic format without the necessary grammatical processes that would generate it yet in place. For those learners that start out with this sentential format we do not only acknowledge that their early patterns differ from those of L1 learners; from a developmental perspective, we also advance that learners who adhere to the SVX pattern are prone to erroneously set the VP headedness parameter to the head-initial value. Learners who do so are then confronted with the task of restructuring their learner grammar in a way that is more similar to L1 Romance learners attaining L2 German than that of L1 monolingual German learners. Several learner errors observed in this study, particularly concerning verb placement, corroborate this assumption. We will come back to this issue below.

Against this backdrop, the participants’ production of target-deviant structures they do not encounter in their German input, marked by the formal teach- ing/learning situation, deserves special attention: these constructions might provide further insights into what is actually attained or “within reach” in structural terms.⁷ Following this line of reasoning, variation in learner data can be taken as an indication of underlying language learning processes. We turn next to a discussion of what the data reveal in this respect.