Signposts for the implementation of the IP: auxiliary and modal verbs
A diagnostic criterion commonly used to establish the nature of the structure available to L2 German learners is the production of periphrastic verb constructions with auxiliary and modal verbs. Recall that auxiliary and modal verbs are assumed to be base-generated in INFL. Hence, constructions with these verbs can be taken as an indicator of the availability of the functional category INFL. From a developmental perspective we might assume that complex verb forms act like signposts of an expanded structure for the language learner: these constructions involve two verbs distributed in two different positions of the clause.
The application of this criterion in the analysis of our data reveals two important findings. First, the timing of the emergence of complex verb forms with modal or auxiliary verbs is subject to individual variation. And, second, the use of these verb forms is subject to intra-individual variation as the forms produced are not always target-like; errors occur in the choice of the appropriate auxiliary or modal verb form, as well as in the choice of the appropriate main verb form. These are important findings, in particular, if we consider that all participants are attending the same bilingual programme. For one, variation at the inter-individual level revealing that learners progress at a different pace suggests that the impact of the input available in the context of a formal teaching/learning environment needs to
be qualified. In a similar vein, variation at the intra-individual level can be taken as an indication that there is more into language acquisition than the repetition “by rote” of explicitly taught grammatical structures as only those properties that are acquired will be productive.
Turning to the structural change (expansion by an additional structural layer, the IP) we assume to be associated with the emergence of auxiliary and modal verb constructions we noted in the analysis of the data that we need to be careful in our interpretation. Indeed, for some early constructions with modal verbs it is questionable whether they reflect the availability of a new structural layer. Instead we might assume that these verbs are adjoined to the available VP structure. Consider, in this respect Muhammed’s examples repeated in (668) and (669). If we look first at the second example in (669), we can see that elements are arranged in a target-deviant manner in this sequence, in which the modal verb kann (‘can’) is followed by the negator and the lexical verb precedes the object. Because the target-deviant verb-complement order is reminiscent of the SVX schema we might conclude that “kann nicht”, used as an unanalysed formula or idiomatic expression, is adjoined to the elementary sentential pattern. In line with this assumption, the modal verb in (668), too, appears to be combined with the main verb via adjunction.
Incidentally, the target-deviant head complement order that becomes apparent in these examples is strikingly similar to the word order observed in the early productions of L2 German learners in other acquisition situations. Indeed, as we can see in (670) learners with a Romance L1 produce constructions that parallel (668) above.
Notice that sequences like (670) are commonly assumed to result from a temporary borrowing of the L1 VP headedness parameter value, which in the case of the L1 Romance learner is head-initial. The critical question to ask at this stage is why bilingual learners whose L1 DGS is an SOV language would produce a type of error that is not compatible either with the German or the DGS structure. Taking up our earlier considerations concerning the learning tasks faced by those learners who start out with the SVX schema as their “base structure” of German, the data can be taken as an indication that these learners set the VP headedness parameter to the target-deviant initial value. The apparent coexistence of alternative structural formats as it occurs in (671), a complex clause in Maria’s file 1, not only provides additional evidence for this assumption. It illustrates also the variation characteristic of reorganisation phases preceding the eventual implementation of the target option.
Evidence for the eventual implementation of a structural layer above the VP can be found in the written productions in the form of sequences with complex verbs in which objects, adverbials or negators appear inside the verb bracket (compare the example in Table 4.24). These structures clearly represent evidence for the availability of an additional IP layer: auxiliary or modal verbs appear in the left peripheral verb position, whereas lexical verbs appear in the right-peripheral verb position. The relative order of complement and main verb indicates further that the VP headedness is correctly set to the target head-final value.
Table 4.24: Example of a target-like distribution of finite and non-finite verb forms.