L2 Learners' Processing of Grammatical Gender Varies According to Cognate Status and Proficiency: An ERP Study
Open Access
- Author:
- Schempp, Patricia Lynne
- Graduate Program:
- German
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- May 09, 2017
- Committee Members:
- Carrie Neal Jackson, Dissertation Advisor/Co-Advisor
Carrie Neal Jackson, Committee Chair/Co-Chair
Barry Richard Page Jr., Committee Member
Michael Travis Putnam, Committee Member
Adriana Van Hell, Outside Member - Keywords:
- German
Grammatical Gender
ERP
L2 Learners
Cognate Status
Second Language Acquistion
SLA
L2
Cognate
Gender
Psycholinguistics
Cognitive Psychology
Linguistics
Language Learning
Bilingual
Bilingualism - Abstract:
- This study investigates the developmental trajectory of learning a notoriously difficult grammatical feature in the second language (L2), namely grammatical gender, by investigating the impact of explicit training on the online processing of this feature and by testing L2 learners at different proficiency levels. Evaluation was conducted behaviorally, and by using event-related potentials (ERPs), to provide insights into the real-time processing of grammatical gender information among L2 learners. This study also assesses the impact of cognate status on the online processing of grammatical gender violations. Finally, the study explores the role that individual differences in proficiency may play in the processing of grammatical gender. Previous research has shown that late L2 learners can acquire grammatical gender (Gillon Dowens, Guo, Guo, Barber, & Carreiras, 2011; Morgan-Short, Sanz, Steinhauer, & Ullman, 2010) and use it predictively (Hopp, 2013), once they are highly accurate in grammatical gender assignment, i.e., the ability to assign the correct gender to individual nouns. Other research has shown that training results in improvements to learners’ sensitivity to L2 morphosyntactic violations (Batterink & Neville, 2013), and grammatical gender violations more specifically (Lemhöfer, Schriefers, & Hanique, 2010; Morgan-Short, Finger, Grey, & Ullman, 2012; Morgan-Short et al., 2010). Prior studies using ERP methodology have also shown that L2 proficiency often modulates the type of online processing that learners engage in (e.g., Foucart & Frenck-Mestre, 2011; Tanner, Inoue, & Osterhout, 2014). Novel to the present study is the combination of training a group of less proficient learners (intermediate) and comparing them to a group of more proficient learners (advanced) and a group of native speakers. Previous research has either trained a group of L2 learners and monitored online processing over the course of learning (e.g., Batterink & Neville, 2013; Lemhöfer et al., 2010; Morgan-Short et al., 2012; Morgan-Short et al., 2010; White, Genesee, & Steinhauer, 2012) or tested learners at different L2 proficiency levels using a cross-sectional design (e.g., Ojima, Nakata, & Kakigi, 2005). By employing both a training study and cross-sectional comparisons across proficiency levels, the experimental method in this study allows for a more nuanced understanding of the developmental trajectory of the acquisition of L2 grammatical gender. Furthermore, previous research has demonstrated cognate facilitation effects at both the level of processing individual words in isolation (e.g., Midgley, Holcomb, & Grainger, 2011) and at the sentence level, when such words are embedded in a larger context (see Lauro & Schwartz, 2017, for review). However, it is unknown how cognates influence L2 morphosyntactic processing. In order to assess whether cognates might facilitate the online processing of gender violations, this study also examined how learners at different proficiency levels process grammatical gender for cognate versus noncognate nouns. In order to address the questions of how L2 learners process grammatical gender violations online and how training, cognate status, and proficiency modulate processing, this study looked at two groups of L1 English- L2 German learners (intermediate and advanced) and L1 German speakers. Experiment 1, involving intermediate L2 German learners, consisted of two sessions, seven days apart. In session 1, participants’ sensitivity to grammatical gender violations was assessed via an ERP task in which participants read sentences containing grammatical versus ungrammatical gender-marked determiners (e.g., Wo ist der.MASC /*das.NEUT AnkerMASC in diesem Bild? “Where is the anchor in this picture?”). Additionally, half of the target nouns were cognates (e.g., Anker “Anchor”), meaning they shared form and meaning in English and German, and half were noncognates (e.g., Tasse “cup”). The participants were then trained on the nouns and their grammatical gender in an explicit training task. One week later, they were tested on the nouns, re-trained and then they completed a second ERP task. The behavioral results indicated that cognates facilitated noun naming during the training and behavioral recall tasks, but not gender naming accuracy. ERP responses in session 1 revealed no overall sensitivity to grammatical gender violations. However, correlational analyses revealed that responses to gender violations with cognate nouns varied according to L2 proficiency, with less-proficient L2 learners exhibiting a more negative N400-like response to violation sentences and more-proficient L2 learners exhibiting a more positive P600-like response. In session 2, after training, the behavioral results showed the same pattern as session 1. The ERP responses revealed an overall sensitivity to grammatical gender violations, but these results were not modulated by L2 proficiency. Qualitative and quantitative differences in the ERP signatures for cognates and noncognates were present, with participants exhibiting an N400 and a frontal positivity in response to gender violations with cognate nouns but only a late frontal positivity in response to gender violations with noncognate nouns. That the overall ERP responses to gender violations increased from session 1 to session 2, and that these ERP response patterns were no longer modulated by L2 proficiency, provide evidence that training reduced initial proficiency-based differences and increased participants’ sensitivity to grammatical gender information during real-time language comprehension. In Experiment 2, advanced L2 learners’ and L1 German speakers’ sensitivity to grammatical gender violations was assessed via the same ERP task used in Experiment 1. Following the ERP task, participants completed the same behavioral noun and gender naming recall task that the intermediate L2 learners completed in session 2 of Experiment 1. The behavioral results for the advanced L2 learners showed no cognate facilitation effects in either noun or gender naming accuracy. The ERP responses revealed that the advanced L2 learners were sensitive to gender violations with noncognates, but less so with cognates, in that they exhibited a centralized positivity in response to gender violations with noncognate nouns and only a descriptive frontal positivity in response to gender violations with cognate nouns. The native speakers showed no cognate facilitation effects in the behavioral task. In the ERP task they showed sensitivity to gender violations for cognates and noncognates, exhibiting a P600 effect in response to gender violations with both cognates and noncognates. These results provide evidence that advanced L2 learners are sensitive grammatical gender violations in the absence of training, yet unlike the native speakers, their pattern of ERP responses remains different for cognates and noncognates, suggesting that the online processing of grammatical gender continues to be modulated by cognate status, even among more advanced L2 learners. The present study demonstrates that proficiency has a significant impact on whether L2 learners are sensitive to grammatical gender violations and how they process grammatical gender in real time. With respect to proficiency and training, this study presents data from a continuum of proficiency levels, demonstrating that learners’ sensitivity to grammatical gender violations increases with increased L2 proficiency. It also reveals the large degree of variability in L2 learners’ processing of grammatical gender violations, especially at lower proficiency levels. With respect to cognates, although the facilitation in noun learning and recall did not extend to the acquisition of grammatical gender, the different ERP responses to grammatical gender violations for cognates versus noncognates in both L2 learners groups indicate that facilitated cognate recall does affect the online processing of gender violations. This is a likely result of facilitated lexical retrieval freeing up cognitive resources to retrieve and use morphosyntactic properties of the noun during online processing. In conclusion, the results of this study highlight ways in which L2 learners’ difficulties with grammatical gender stem, at least in part, from how they allocate finite processing resources during online processing, rather than an inability to process grammatical gender per se. Strengthening L2 gender representations through training facilitates access to grammatical gender information, which in turn leads L2 learners to successfully process grammatical gender information—exhibiting sensitivities to grammatical gender violations—during real-time language processing. This dissertation highlights the importance of identifying ways in which explicit training can facilitate the acquisition of grammatical structures known to be difficult for L2 learners, and the need for researchers to more carefully consider the ways in which L2 proficiency impacts the real-time processing of such structures during L2 comprehension.