| 总结: | <p>Upon finishing FL learning at primary school, many students lack productive knowledge, for example regarding structures like verb-argument-constructions (e.g., [Verb] about [Noun]), indispensable for increasing communicative agency, as expected by curricula. These curricula adopt a usage-based constructionist approach to language learning and consider 'large linguistic units' (e.g., routines (<em>How are you?</em>) or patterns (<em>My favourite _ is _</em>)) to be catalysts for students' development of a productive linguistic repertoire. However, despite the ubiquity of such units in primary curricula, evidence on the impact of teaching input consisting of such structures on primary FL students’ linguistic development is scarce, as indicated by Schulz et al. (2023), reported in this thesis.</p>
<p>Across cognitive domains, increased initial input variability (the variation in our experience with different exemplars, e.g., [talk/think/rant/wonder] about [god/bicycles/dogs]) can improve generalization (i.e., [Verb] about [Noun]), and enhance learning. In controlled experiments, increased input variability proved beneficial for children’s inductive generalization and extension of linguistic information from input structures to novel contexts (e.g., Wonnacott et al., 2012). Such findings drove the investigation into extending the benefits of input variability to real classrooms.</p>
<p>Following a usage-based constructionist approach to language learning, I report on two quasi-experimental teaching intervention studies (each lasting two weeks) with two British Year 2 classes learning German (age 6; 20 students/class). Experiment 1, comprising of a high (HV) and low (LV) input variability condition, focused on 16 German 'approach' event verb- argument-constructions (<em>Zum X [robbt/schleicht/rutscht/etc.] der/die/das Y; To the X [approach verb] the Y</em>), featuring one (LV) or four different verbs (HV) in the construction’s verb slot. Post- tests indicated that children exposed to increased input variability demonstrated better generalization to novel verbs compared to controls. Experiment 2 focused on three sets of non- adjacent dependencies (cf. Gómez, 2002). The HV and LV conditions included 30 and five ‘intervener positions’, respectively. The learning of non-adjacent dependencies and the ability to generalise structural information and extend it to novel contexts (i.e., unknown interveners) was investigated in post-tests (grammaticality judgments), yielding ambiguous results.</p>
<p>The experiment 1 findings largely aligned with more ‘controlled’ experiments, suggesting that even in 'noisy' classroom environments, increased input variability can positively impact students’ construction generalization and extension to novel verbs. However, the experiment 2 data did not support this variability effect. Instead, they underscored the potential benefits of LV input under specific pedagogical circumstances.</p>
|
|---|