• Keine Ergebnisse gefunden

3 Regular Stem Allomorphy

3.1 Overview

3.1.4 Regular stem allomorphy in L2

3.1.4.2 Discussion: Regular stem allomorphy in L2

The present pilot study was designed to explore the processing and representation of regular stem allomorphy in a second language. The stem allomorphy employed in this study was not only regular but also highly predictable. The regularity and predictability of the stem allomorphy were governed by the application of morphosyntactic rules that first converted an adjective into a verb with a simultaneous umlauting of the stem vowel, and then attached a nominalizing suffix to the verbal stem. In line with the declarative/procedural model (Ullman, 2001, 2004), we reasoned that the initial

accommodation of these morphosyntactic rules by an L2 speaker should proceed within the declarative system, while growing experience should increase the reliance on the procedural memory.

We expected the lexical decision task to trigger a search for a lexeme within the declarative system. Taking into account the overreliance of the low-proficiency L2 speakers on the declarative system and the non-existence of the RD items in the German vocabulary, we expected the RD nonwords to be behaviorally classified as such. The factors contributing to the behavioral recognition of the RD items as nonwords would be (i) the failure to find these items in the long term memory and (ii) the application of a memorized umlauting rule offline. Both factors require the involvement of the declarative memory for the task performance. The reliance on the declarative memory was expected to be reflected in the brain response pattern: we expected the participants to demonstrate a semantic integration effect for the RD items similar to that elicited by the real words. The involvement of the procedural memory should be manifested in the sensitivity of the L2 learners’ perceptual system to the violation of the morphosyntactic rules, i.e. in the elicitation of the N400 by the RD items.

The UD and NC items were expected to elicit similar large N400 effects due to the non-existence of their stems.

The results of the pilot study provided empirical evidence for the reliance of the low-proficiency L2 learners on the declarative system: a significant N400 effect was revealed for the NC violation condition. The RD items failed to significantly differ from the W condition. Although the violation-related brain response pattern was reminiscent of that elicited in the native-speaker group, it failed to provide significant difference between the critical conditions (W vs. RD and RD vs. UD). We argued that the acquisition of a morphosyntactic rule required a definite amount of time to get accommodated into the procedural system.

According to the first language acquisition research (Anderson & Nagy, 1992; Nagy & Anderson, 1984; Nagy et al., 1985), the establishment of the morphological relations between words is observed by the end of the first decade of life and, for the most part, due to reading. The L1 children have to run numerous analyses on the linguistic material they perceived/learnt during the first ten years of life and establish connections between the words with a certain phonological or orthographic overlap. Some children might deduce the morphological structure/relatedness on their own, while the others might need instruction to do so. Therefore, the first language acquisition combines, on the

one hand, an analytic work done by a child on his/her own, and on the other hand, the supportive work performed by the parents and teachers. Following this logic, the most important difference between the L1 and the late L2 acquisition is the form of the initial instruction. A late L2 learner comes into the classroom with an already established concept of the linguistic structure of the L1. The acquisition of the new language proceeds via the accommodation of the novel rules into the already existing system as either an equivalent or a superstructure.

In both cases the initial rule learning and application proceeds within the declarative system (Ullman, 2001a, 2004). The growing proficiency of the L2 learners should be reflected in the transition from the declarative system to the procedural one (Pliatsikas, Johnstone, & Marinis, 2014). This transition would mean the change of hardwiring/involvement of certain brain structures indexed by the error-detection brain responses elicited by morphosyntactic violations.

The present ERP pattern demonstrated that the low-proficiency L2 speakers over-relied on the declarative system. A follow-up study with a high-proficiency L2 group could shed light onto the development of the L2 acquisition with respect to the change of hardwiring. According to Hahne (2001), even a high-proficiency L2 group would fail to demonstrate a native-like response pattern in terms of the automatic error-detection (e.g. ELAN). However, she did observe a delayed P600 effect in the L2 group in her study. Taking into account the design used in the present experiment and the semi-automatic characteristic of the N400 component, this violation effect might be observed in the high-proficiency L2 group.

The N400 effect elicited by the NC condition in the present study had the same latency as the N400 observed in the German. This finding was in line with Rossi et al.’s (2006) study, which showed that the timing of the ERP responses in the high-proficiency L2 group corresponded to that of the native speakers. Our findings, having revealed a similar timing of the N400 in both experimental groups, provided the evidence for a slow change of hardwiring. Considering a very small sample size of the present study, we could expect the critical comparisons to reach significance if the sample size should increase. The number of significant comparisons would be the index of a morphosyntactic rule acquisition, i.e. W vs. RD and/or RD vs. UD. With the increasing proficiency the reliance on the procedural system will grow (Ullman, 2004) revealing native-like brain responses to morphosyntactic errors (Friederici, Steinhauer, et al., 2002;

Lück et al., 2006; Osterhout & Holcomb, 1992; Penke et al., 1997; Rodriguez-Fornells et al., 2001; Weyerts et al., 1997). In order to finalize the present pilot

study, the sample size should be increased and a high-proficiency L2 group should be tested. Only in this case, we can demonstrate the transition from the declarative to procedural system with reliable results.

3.2 Discussion: Representation of regular stem