IP structures

Several diagnostic criteria can be used to determine whether an additional structural layer above the VP, that is, the inflection phrase (IP), is available in learner grammars, such as the production of constructions with auxiliary and modal verbs, the marking of subject-verb agreement, and the raising of finite verbs to a position at the left periphery in main clauses. Crucially, not all of these phenomena need to be available at the same time.

Research into the acquisition of German has shown that learners may take different avenues or strategies in structure-building (D’Avis & Gretsch 1994; Gawlitzek-Maiwald 2003). The variation encountered points to the relevance of paying attention to changes in learner grammars that might conspire in the structural expansion of the VP listed previously.

Discovering the relationship between different verb positions. A fundamental step in the acquisition of German word order concerns the establishment of a relationship between the different positions verbs may appear in. Recall that finite and non-finite elements of the verb complex appear in sentence-second vs. -final position respectively in declarative main clauses (section 4.1.1, Table 4.1), whereby finite elements appear in INFL and non-finite elements in the sentence final V position. The availability of an expanded structure in learner grammars is reflected in the production of sentences containing modal, auxiliary verbs or separable verbs with a target-like distribution of finite and non-finite elements of the verbal complex in sentence-second and final position respectively (see (399) for an example of an L1 learner with a modal verb and (400) for an example of an L2 learner with an auxiliary verb).

Variation in L2 German acquisition. Variation concerning the relative order of the verb and its complement in constructions with periphrastic verb forms has been found to occur in productions of L2 learners whose L1 differs from the L2 regarding the VP headedness parameter. If we look at the following L2 German utterances of the L1 Italian learner Bruno^[1] (examples (401)-(378) from Plaza-Pust 2008a: 257), we can see that these constructions follow the verb-object pattern, which is characteristic of VO languages like Italian.

At this stage, it seems, L2 lexical elements are arranged in an order that reflects the L1 parametric option, which amounts to the traditional notion of L1 transfer or influence. Following a dynamic approach to language development (cf. section 2.2.3), the adoption of the L1 parametric value can be rephrased in terms of a coupling of the L2 learner system with the available (L1) language knowledge. Further progress in the attainment of the L2 grammar involves system-internal conflicts and ensuing processes of uncoupling or differentiation. In the L2 German grammar, such non-linear processes can be observed upon the inclusion of the target OV (object-verb) option (examples (403) and example (424) above), four to six weeks after the production of the above examples (examples from Plaza-Pust 2000: 182f.).

What is interesting for present purposes is that the transition from a VO- to an OV-grammar does not occur instantaneously, i.e. in terms of an immediate exclusion of VO orders. In fact, the introduction of the new L2 OV-option is subject to fluctuations. Indeed, the Italian learner continues to produce VO constructions for some time and there seems to be no apparent reason for why and when one word order is chosen over the other as both occur with the same lexical items. As we can see in Figure 4.1, which depicts the relative frequency of the respective orders, both alternate in the data of the Italian learner until file 9, where they are equally frequent. From then on, the proportion of target-deviant structures decreases, disappearing completely as of file 12.

Figure 4.1: Relative frequency of OV and VO sequences in ‘Bruno’s’ L2 German until file 13 (Plaza-Pust 2000: 184, 2008a: 258)

Verb raising, agreement, V2. Another crucial process that is bound to the expansion of the initial VP format by the IP layer is verb raising, whereby finite main verbs are raised to I (in main clauses) to have their features checked (because this process applies only to finite verbs, non-finite verbs remaining in the VP, learners are also assumed to master what is commonly dubbed the finiteness distinction) (compare examples (404) and (405)). As learners of German are also acquiring a language that is a V2 language, they also have to acquire the V2 constraint which requires that some other XP be moved into the preverbal position in SpecI (compare examples (406) and (407)).

In research on L1 learners of German the acquisition of the finiteness distinction, the target agreement paradigm and the V2 constraint have been found to commonly coincide. In some learners, however, the grammatical properties associated with the IP do not become productive at the same time. This holds equally of monolingual L1 and bilingual or L2 learners of German, as is explained next.

(a) Variation in L1 German acquisition. Variation regarding main-clause word order has been found in some L1 German learners. The child ‘Max’, for example, shows that bare VPs and “mobile” IPs (initial/final) may coexist prior to the eventual convergence toward a unified structural format. Note that the diversity of main clause patterns in examples (408)-(411) includes V1, V2, and Vend structures (examples from Fritzenschaft et al. 1991: 89).

(b) Variation in L2 German acquisition. Studies on the L2 acquisition of German by Romance L1 learners have shown that their early L2 German productions do not adhere to the V2 constraint. Instead, word order appears to be determined by grammatical processes that relate to the L1 grammar^[2], which indicates that the IP is initially set to the L1 value.

As L2 learner grammars progress toward the L2, the grammatical processes associated with the V2 parameter are implemented. Examples (412)-(413) and

(415)-(419) illustrate the structural variety observed once non-subject V2 appears in the L2 learner grammar of the Italian learner mentioned previously (Plaza- Pust 2000). As examples (412) and (413) show, the productivity of structures that adhere to the V2 constraint (cf. (412)), does not go along with the immediate exclusion of constructions that do not. Instead, target-like non-subject V2 constructions alternate with target-deviant V3 sequences (cf. (413) from Plaza-Pust 2000) that appear to be based on the L1 Italian structure (for further illustration an Italian utterance is provided in (414).

Examples (415)-(416) illustrate the type of Italian-like constructions with subject- verb inversion (involving the so-called free inversion) with or without a constituent in the preverbal position (compare with the Italian sentence compra un libro Gianni, ‘buys a book Gianni’). In addition, the range of sentential formats produced includes verb initial sequences with subject drop (cf. (417)) or with the subject in postverbal position (cf. (418)). Note that the SVX sequence in example (419) is amenable to both a V2 or non V2 analysis (Plaza-Pust 2000: 229).

• [1] Bruno was one of the subjects studied in the framework of the Hamburg ZISA (= Zweitspra-chenerwerb italienischer und spanischer Arbeiter, ‘second language acquisition by Italian andSpanish labourers’) project (cf. Clahsen, Meisel and Pienemann 1983 and Plaza-Pust 2000 forfurther details).
• [2] The processes include, in particular, free adjunction to IP (which derives V3 structures such as(414) in which the adverbial appears before the subject in preverbal position), and nominativecase-checking under spec-head agreement (as nominative case is not checked under governmentin that language, subjects cannot appear in post-verbal position).