Candidates for cross-modal language mixing: Pooling of linguistic resources

L2 learners of a second language who are more advanced in their narrative development than young infants, have been found to concatenate elementary structures to express complex meanings despite remaining gaps at the structural and lexical levels. Several examples in the participants’ data indicate that they pool their resources in several ways, for example, through the use of functional elements, loan translations, or lexical and syntactic borrowing.

Functional elements. The use of functional elements at a time when their associated grammatical properties are not yet attained can be taken as an indication that learners pool their linguistic resources during the early stages of their acquisition of German. Hamida’s use of the complementiser weil (‘because’) in (652), for example, marks a difference to early verbless productions in child L1 acquisition. Indeed, in L1 German acquisition complementisers tend to appear late, often after the production of preconjunctional clauses (Rothweiler 1993). L2 learners, by contrast, have been found to use L2 functional elements such as complementisers at a time when they do not yet master the associated target grammatical properties. Typically this occurs after an initial stage at which functional elements are missing (cf. Klein 2000 for a concise summary of the so-called “basic variety” in natural second language acquisition situations). By assumption, L2 learners borrow these functional elements from their L1 which, applied to the situation of our participants, implies that they use these elements because they know them already in DGS. Following this assumption, learners are confronted with the task of learning the target structural properties associated with these items at a later stage (cf. also Plaza-Pust 2000 for a detailed discussion of the relation of lexical and syntactic learning). Consequently we assume that, at this early stage, functional elements are combined with elementary structural formats via adjunction (for further illustration compare the sketch provided in Table 4.20).

Table 4.20: Adjunction of functional elements at the VP stage.

Word order. Other candidates for language mixing reveal a sophisticated borrowing from DGS. Consider, for example, (653) produced by Fuad in file 1. As we remarked upon before, the example involves the combination of two propositions, i.e. “there is a deer there” and “the deer has antlers on his head”, arranged in a way that is reminiscent of DGS expressions following the figure-ground principle. Certainly, the introduction of the deer as a character through the verbless expression “dann da ein Resch” strikes us in its similarity to constructions used for referential establishment in DGS. Also, the relative order of the elements in the second proposition seems to be organised according to the figure-ground principle. Note that a target-like equivalent would involve the reverse order of ‘head’ and ‘antlers’, as illustrated in (654). Interestingly, the analysis reveals that elements are not only arranged according to the figure-ground principle in sequences with verb drop, as is the case of (653), but also in constructions with inflected main verb forms (for further illustration compare the sketch provided in Table 4.21). Example (655), produced by Hamida in file 1, is a remarkable example in this respect. Notice that this sequence involves a finite verb in sentence-final position, as it would be required in DGS. Further, da is also used to assign a location to a referent; hence da fulfills the function det_lqc would fulfil in DGS. The DGS construction provided in (656) illustrates how close the sequence produced by Hamida is to what we might consider to be the equivalent DGS construction.

Table 4.21: Language contact phenomena at the VP and the IP levels (figure-ground, verb placement).

Candidates for language mixing, involving the calquing of a complex sentential DGS constructions, are easy to spot because they involve a combination of propositions that are built in a way that is not possible in target German and has neither been found to occur in the data of early learners of the language. The situation is different in the case of two or three-word combinations with verb drop, as we learned previously, because such sequences occur also in the productions of other learners of German. So they cannot be unambiguously interpreted as candidates for language mixing. A similar case obtains with SOV sequences in the learner data. SOV sequences, such as the one produced by Christa in example (657), might be taken to reflect the borrowing of the head-final value of the VP headedness in DGS; but this interpretation must be qualified given that German is also an SOV language and sentence-final verb placement is a frequent phenomenon in child language acquisition (though not a target main clause word order). In any case, given the participants’ initial adherence to a rather strict SVX pattern, the production of SOV, if only on an occasional basis, might be taken as an indication of underlying language learning processes: for one, learners do not encounter this type of clause pattern in their input.

Loan translations. The analysis of the data reveals that cross-modal language mixing involves not only relexifications of DGS structural formats (e.g. figure- ground, SOV), but also loan translations of complex DGS meanings that would be simultaneously expressed in space. As we already advanced in Plaza-Pust (2008b) cross-modal translations are illustrative of the lexical and structural adaptations of the expressions borrowed across modalities. Crucially, such adaptations are determined by the properties of the recipient language (in our case a learner variety of that language), as is the case in other types of borrowing (Winford 2003: 42f.). In the case of cross-modal language contact phenomena in the written language, such adaptations are determined by the limitation to use one modality of expression (unlike in spoken language production where signed elements might be combined with spoken ones). Owing to the difference in the predominant type organisation (simultaneous for DGS and sequential for German), cross-modal borrowing involves, at times, a sophisticated translation of simultaneous DGS expressions into sequential German expressions that goes well beyond a 1sign-to-1word translation. By assumption, sequences like (658), produced by Simon in file 3, involve such a subtle type of borrowing: a DGS classifier construction is analysed into meaning units or thematic roles; these elements in turn are mapped onto German lexical items and arranged sequentially. In our discussion of Simon’s data we remarked on the sentence-final placement of the preposition in (‘in’) used to refer to the location of the THEME (= the head) in this verbless construction. The unusual (target-deviant) position of the preposition might be taken as an indication of a lexical gap in German, as it appears in the place of a verb that would express the dog’s sticking his head into the jar (that is, reinstecken, ‘to stick into’). Certainly, the sentence-final position would strike us as odd unless we consider DGS as a potential source of this order (verbs appearing in the final position in that language). Finally, the arrangement of “Glas” and “Kopfen” indicates that Simon adopts the figure-ground principle in this case, too, which can be taken as an additional indication of borrowing.

Another potential candidate for borrowing is example (659), produced by Muhammed in file 3. In our discussion of Muhammed’s data we remarked upon the arrangement of elements following the figure-ground principle (deer= ground, Paul= figure) and the repetition of the full NP referring to the ground. This latter aspect deserves further attention for two reasons. For one, because it reflects Muhammed’s lack of the German pronominal system at the time. So overt reference to the same referent occurs through the repeated use of the full NP. Secondly, “liegen mit Hirsch” might be regarded as a calquing of the DGS spatial verb lie_{on deer} used in the expression of the spatial relation of the boy and the deer. Hence, it is plausible to assume that “liegen mit Hirsch” has the status of a complex verb expression and that as such, and “calquing” DGS, it appears in the right periphery of the sentence. Table 4.22 summarises our previous observations concerning potential candidates for language borrowing involving loan translations of classifier expressions.

Table 4.22: Language contact phenomena (figure-ground, classifier constructions).

Against this backdrop, the sequence in (660), produced by Hamida in file 1, could be interpreted as a blend of a DGS and a German sentential format as the DGS-like setting of the ground (“Bei Wasser”) is combined with an SVX clause in which this setting is repeated.

Modifying expressions. Finally, we might consider sequences like (661) and (662), produced by Christa and Hamida respectively. These sequences are more difficult to interpret. Notice that they involve SVX patterns with additional prepositional phrases, arranged in a target-deviant manner. At closer inspection, the prepositional phrases that seem to be adjoined in a random manner, might be interpreted as modifying expressions that relate to the previous noun (for further illustration of this sophisticated arrangement of elements in a clause compare Table 4.23). Following this assumption, a target-like equivalent of the propositions combined would require a juxtaposition of separate clauses or the subordination of a relative clause (as illustrated in example (663)). At this stage, however, learners do not master either the lexical or the structural means necessary to overtly express relations between propositions. As a consequence, they place prepositional phrases in a DGS-like fashion right to the “ground” PP complement of the main clause they refer to.

From a developmental perspective, modifying structures combined with main clauses via adjunction might be attributed the status of precursor structures, to the extent that they potentially pave the way for more complex sentential structures. It must be noted, however, that the attribution of such a precursor status can only occur a posteriori, that is, on the basis of data that corroborate further progress. What do the data reveal in this respect?

Table 4.23: Potential precursors of relative clause structures.

If we look at Christa’s narratives produced after file 1 we must conclude that we cannot attribute the status of precursors to the early PP structures because Christa’s later narratives do not contain any evidence for the development of subordination. By the end of the study, Christa does not use the option of PP-adjunction anymore. What we observe instead is a tendency to produce a series of full main clauses, in which reference to the same referent occurs via repetition of the full NP (compare example (664), produced by Christa in file 5). This way of stringing together main clauses reflects a development that departs from the initial adjunction of related propositions. Only the use of the personal pronoun in (664c) can be interpreted as a first indication of the use of linguistic devices for the purpose of cohesion.

We are left with the question of why the potential implicit in those early structures that appear to be candidates for language borrowing is not exploited any further. Once again, we can only speculate on the impact of the teaching/learning situation and assume that the sophisticated nature of the structures is not recognised as such. What is more, the question arises as to whether the input these learners are exposed to really is sufficiently rich so as to promote the acquisition of relative clauses early on.

While our conclusions about the nature of the input must remain tentative at best, the data collected allow for the conclusion that the impact of the input must be relativised in the face of the scope of inter-individual variation reflected in the data regarding the way complex meanings are expressed structurally.

If Christa resorts to a paratactic concatenation of propositions in her later narratives, Hamida’s recount of complex narrative episodes reveal a rather creative, though not always target-like, use of available linguistic means, including occasional mergers of alternative sentential formats. This learner continues to use verbless clauses until the end of the recording time, but these do not include the type of adjoined prepositional phrases described previously. Unlike Christa, Hamida uses various linguistic means to create cohesion, including coordination with the conjunction und (‘and’) (cf. (665)), the expression of locative relations via the adverbial da (‘there’) (cf. (666)), and, occasionally subordination (cf. (667)). Relative clauses, however, are not produced by the end of the recording time. So, once again, we are led to conclude that the early PPs cannot be attributed the status of precursors structures, but remain instances of language borrowing.