Signs of variation: verb inflection morphology

In our discussion of individual written German profiles, we paid special attention to subject-verb agreement in order to establish whether or not the mastery of verb inflection goes along with an increasing complexity at the syntactic level and the attainment of the finiteness distinction, as is usually the case in infants acquiring German as their mother tongue. Following the assumptions put forward in current linguistic theory (see section 2.1.2), the coincidence of both developments would be expected. Alternatively, if subject-verb agreement markings are not used as a cue in the endeavour of identifying verb positions and their relation, could it be that they are acquired as a result of the grammatical processes involved in structure-building?

One of the main conclusions that can be drawn on the basis of the analysis is that participants vary regarding their overall development in this linguistic area. What is common to all of them is that they do not master the German verb inflection paradigm at the onset of the study. For further illustration of these observations consider Figure 4.23, in which the results of the error measures for each participant are put together. In general terms, we can see that all participants continue to produce errors in the domain of verb inflection by the end of the study. However, we can see that whereas the overall proportion rises in the narratives of Simon, it remains at about the same rate in the final narratives of Muhammed and Christa. For three participants, namely, Maria, Fuad and Hamida we acknowledge a decrease of the overall error rate by the end of the study.

Frequency measures of learner errors, as the ones we have provided for each participant in this study, are valuable in that they allow to discern (a) the proportion of target-deviant forms in a file as well as (b) an overall trend in the learner’s development in this area. However, a closer look at the type of errors produced allows for a more in-depth analysis of the nature of the deficits. In particular, we were interested in determining (a) the status of finite forms at the onset of the study, and (b) the typology of errors produced and the development over time.

Status of finite forms at the onset of the study. Turning first to the status of verb forms produced at the onset of study, we note that some participants already produce some target-like inflected verb forms. Consider, for example, Muhammed’s file 1. In this narrative 8 verb forms are target-like, all of them appearing with the 3^rd person singular -t ending. These forms make up a relatively high proportion of target-like forms, namely, 40%. Can we take this proportion as an indication for a rule-based verbal inflection at this stage? To answer this question we need to look more closely at the verb forms produced by Muhammed at this stage. What we can see is that correct subject-verb agreement markings at this stage appear with a selection of verbs only. As we remarked previously, target-like inflection appears with the verbs gehen (> geht, ‘goes’), sagen (> sagt, ‘says’) and schauen (> schaut, ‘looks’). The other verbs used, that is, fallen (‘fall’), finden (‘find’), klettern (‘climb’), liegen (‘lie’), nehmen (‘take’), reinfallen (‘fall into’), rufen (‘call’), suchen (‘search’), (weg)schenchen [>wegscheuchen/ (‘shoo away’), wunschen (‘wish’)) appear in their infinitive form. In view of the discrepancy observed, it seems plausible to conclude that verb inflection is not a productive process at this stage. So, what is the status of correctly inflected forms? In our view these forms are stored as unanalysed units in the lexicon in addition to non-finite forms, commonly infinitives, that predominate at the time. We may conclude therefore that in the acquisition of German, verbal morphology may serve as a cue for the establishment of a (derivational) relationship between the different positions verbs may appear in (Roeper 1992: 351). However, the apparent dissociation of the acquisition of verb second and the correct morphological realisation of subject-verb agreement in some learners provides evidence against a uni-directional cause-effect relationship, as would be assumed within the lexical learning hypothesis (cf. Plaza-Pust 2000 for an extended discussion and also Hohenberger 2002: 141).^[1]

Typology of errors and development over time. From a developmental perspective, we are interested to determine whether and how learners progress in the area of verb inflection. We noted previously that errors continue to occur by the end of the recording time, although with a similar or even lower frequency rate. Only for one participant, Simon, we observe a dramatic increase of errors toward the end of the study.

Given the theoretical underpinnings of the relation of verb raising and subject- verb agreement and finiteness, the apparent variability concerning inflectional morphology raises the question whether the mechanisms assumed to apply in young children’s first language development are missing in this type of acquisition situation. In the domain of the acquisition of German as a first language, inflectional morphology has been assigned a triggering effect for V2 by some authors (cf. Clahsen 1988, 1992), while others have argued against this connection on theoretical and empirical grounds (cf. Prevost & White 2000; Jordens 1990, 2002). Indeed,

some children exhibit a liberal use of the different positions verbs may appear in and produce finite forms in sentence second and final position. While the latter phenomenon tends to predominate in the data, there is also evidence of non-finite forms appearing in V2 contexts, see examples (683) and (684).

Figure 4.23: Proportion of verb inflection errors and verb drop in participants’ files 1-5.

The persistent variation concerning the use of finite and non-finite forms in the data of the present study is reminiscent of the variable use of agreement morphology in adult second language acquisition. In this domain of research, the apparent optionality has been subject to a controversial debate (Plaza-Pust 2000). Basically, two assumptions can be distinguished. Following the Missing Inflection hypothesis, the variability results from “difficulties in identifying the appropriate morphological realization of functional categories” (Prevost & White 2000: 108). Learner errors would thus pertain to the surface morphological level in that they reflect a problem regarding “the mapping of abstract features to their surface morphological manifestation” (Prevost & White 2000: 108). Alternatively, variability is related to a lack of (cf. Meisel 1991) or erroneous specification (cf. Eubank 1992) of the relevant functional categories, an assumption dubbed “Impaired Representation Hypothesis” by Prevost and White (2000: 110).

In line with Prevost and White (2000: 125) we assume that those learners who established the IP in their learner grammars use non-finite forms as default forms in finite contexts as these learners provide evidence of a knowledge of finiteness and the relating syntactic processes (verb raising, V2). It is assumed therefore that these forms “behave syntactically like finite verbs” (Prevost & White 2000: 108). The data do not confirm the random use of these forms as predicted by the “Impaired Representation” Hypothesis. Finite forms do not appear in the right- peripheral position, with the exception of those learners, whose IP headedness is mobile, as is the case of Hamida. Additionally, we observe a developmental progression in the target-like use of finite forms.^[2]

Indeed, a detailed analysis of the types of error produced makes apparent that errors that appear idiosyncratic at first sight represent recurrent phenomena. As we can glean from Table 4.28, which provides an overview of the error types identified, the range of errors is limited. It is important to note that not all participants produce all types of error and that the frequency varies inter-individually.

Table 4.28: Error types in the domain of verb inflection.

Table 4.28: continued

Furthermore, we might look at the erroneous verb forms in terms of the information they encode or fail to encode. Based on the assumption that the mastery of target-like inflection involves the interaction of several grammatical modules, errors reflect the remaining deficits in this respect. A differentiation of the errors along these lines derives a typology of target-deviant forms according to the type of information that is encoded (cf. also Table 4.29) (capital letters in brackets relate to the examples listed in Table 4.28, they are also included in Table 4.29):

Lexicon (no interface information) [A, B]:

Non-finite forms (infinitives) used as default forms are retrieved directly from the lexicon. They might appear in combination with expletive forms of the copula. The status of the copula is unclear (does it serve as a tense marker?).

Syntax-morphology interface (partial) [C, D]:

Default and mismatch errors occurring at a time when the finiteness distinction is available (distributional/syntactic criterion) reflect a deficit at the interface between morphology and syntax as person-number encoding is still problematic (subject-verb agreement/morphological distinctions).

Syntax-lexicon interface [E, F, G, H]:

Erroneous verb combinations of auxiliary and modal verbs with finite/ infinitive/ participle forms reflect the availability of the finiteness distinction but indicate that deficits remain at the level of the lexicon (selective properties) and morphology (participle formation). Target-like inflection of the finite verb parts indicates that subject-verb agreement is operative.

Morphology-lexicon interface [J, K]:

Failure to correctly inflect irregular verbs (finite forms or participles) reflects a rule-based verb inflection (morphology) but indicates a deficit at the interface between morphology and the lexicon.

In sum, the mastery of the target-like inflection morphology, on the one hand, and the choice of the correct verb form to mark grammatical relations involves several grammatical modules. Learners are tackling with the interfaces between these modules. In line with the dynamic model of development presented in section 2.2.3 we might assume that feedback processes are involved. Clearly, the development is not linear, but dynamic: orderly states are followed by chaotic, which in turn might precede further orderly states. By the end of the recording time covered in this study, the eventual convergence with the target grammar has not been accomplished by the majority of learners. Only Maria seems to have reached an orderly stage in the sense outlined.

Table 4.29: Typology of verb inflection errors and information encoded from the different modules (u= operative, e = problematic).

• [1] On theoretical grounds, the apparent dissociation is also compatible with the variation encountered across V2 languages. As pointed out by Schaner-Wolles (1994: 216), Afrikaans is a V2language without overt verbal inflection (cf. also Vikner 1995, 1998).
• [2] As pointed out by Prevost and White (2000: 129) child first and adult second language acquisition coincide in the progressive replacement of default forms by target-like inflected forms.However, while children eventually give up the use of non-finite forms in finite contexts, adultL2 learners do not or not always do so, which marks a difference between both acquisition types.For a discussion of the potential reasons see Prevost and White (2000: 129 f.).