Working memory and input-based training effects on L2 morphological and morphosyntactic processing
Description
TitleWorking memory and input-based training effects on L2 morphological and morphosyntactic processing
Date Created2022
Other Date2022-05 (degree)
Extent295 pages
DescriptionProcessing morphology is essential for human communication, as grammatical information is crucial for language comprehension. Speakers use morphological constituency to process their first language (L1) but exhibit systematic trouble computing morphology of their second language (L2) in real time. I investigate whether cognitive abilities (i.e., working memory, WM) and language experience (i.e., lexical frequency and proficiency) affect L2 morphological processing (Study 1) and L2 morphosyntactic processing (Study 3), and whether training facilitates L2 morphological processing (Study 2) and L2 morphosyntactic processing (Study 4). Study 1 investigates the effects of lexical frequency and WM on the processing of verbs as stem+affix or as whole units in beginning and advanced learners and monolinguals. Previous studies researched lexical frequency effects on morphological processing in different languages using corpora, and some suggest word structure processing is language dependent (Beyersmann, Duñabeitia, Carreiras, Coltheart & Castles, 2013). However, most of these studies are about English and Dutch. It is unclear which are the processing mechanisms in Spanish, and whether individual lexical frequency is a better predictor of decomposition than corpus-based frequency. Beginning and advanced anglophone learners of Spanish and Spanish monolinguals completed an auditory lexical decision task containing verbs with two (e.g., abarc-ar) and three (e.g., a-braz-ar) morphemes and decided whether the word was a real word in Spanish. Participants also completed a self-reported lexical frequency task via a Likert scale and a WM task. Beginner L2 learners needed more time to process infrequent morphologically complex words, but monolinguals and advanced L2 learners were unaffected by morphological complexity or surface frequency. Also, working memory did not modulate morphological processing. Taken together, the findings suggest a dual-route mechanism at initial stages of L2 acquisition and a whole-word route for Spanish monolinguals and advanced L2 learners. The findings also support the extension of dual-route models to L2 populations and suggest that the selection of a morphological processing route results from linguistic rather than cognitive reasons.
Study 2 takes a step forward by exploring whether intensive exposure to low frequent L2 words helps L2 learners compute verbs as wholes like monolinguals. This question is critical to elucidate the role of input on the selection of morphological processing mechanisms. Previous studies show that learners decompose low frequency L2 words (e.g., Durand López, 2021; Gürel & Uygun, 2013), suggesting the existence of an extra step in the word recognition process. Increasing exposure to low frequency words may switch learners’ L1 processing patterns to resemble those of the L2. Beginning English learners of Spanish were randomly assigned to a treatment group and a control group. Both completed a pre-post auditory lexical decision task identical to that of Study 1 and a self-reported lexical frequency task. In addition, those in the treatment group completed a series of lexical training tasks in which they read short texts containing low frequent L2 words for four sessions. The results show that both groups took significantly longer to process morphologically complex words with three morphemes than those with two morphemes at pretest. While this pattern was observed also at posttest for the control group, the treatment group took roughly the same to process both types of morphologically complex words following training. Also, the lexical frequency ratings were significantly higher at posttest for treatment group. Taken together, the findings suggest that (a) lexical frequency can be increased experimentally in the laboratory, and (b) an increased lexical frequency switches beginner learners’ morphological computation behaviors from decomposition to whole word processing. Finally, the results support hybrid word recognition models claiming that morphological computation routes rely on frequency.
Study 3 examines whether WM and locality (i.e., agreement within or across constituents) modulate L2 gender agreement processing in L2 learners and Spanish monolinguals. Some scholars suggest that distance between two agreeing words recruits a high degree of cognitive resources (e.g., Keating, 2010), but these studies have failed to disentangle structural and linear distance. Study 3 addresses this issue by isolating structural distance. Beginning and advanced English learners of Spanish and Spanish monolinguals completed a self-paced reading task containing sentences in which gender agreement was established within the determiner phrase (local agreement) or between the direct object and a secondary predicate (structural distance). In addition, they completed a verbal WM task. Results show that Spanish monolinguals had sensitivity to gender agreement violations in local domains and in syntactic distance conditions, while beginner L2 learners were not sensitive to violations in either condition. Advanced learners, on the other hand, detected violations in local domains, and their verbal WM updating spans were associated with sensitivity to violations across phrases. Taken together, the findings suggest that (a) syntactically complex structures consume cognitive resources in great number, and (b) L2 processing is qualitatively similar albeit quantitatively different from native processing, thus providing evidence that late bilinguals can process the L2 in a native-like manner.
Study 4 explores whether WM training facilitates L2 gender agreement processing in structural long-distance conditions in L2 learners, with the aim of elucidating whether a larger cognitive capacity aids morphological processing across words. Advanced English learners of Spanish completed a battery of WM tasks to assess their WM updating skills and a self-paced reading task identical to that of Study 3. Participants were randomly assigned to an experimental group and a control group. The experimental group trained their WM for seven sessions through the completion of a dual n-back task, while the control group participated in pre-post tasks only. The results show that the treatment group presented near-transfer effects to untrained WM tasks at posttest and at the delayed posttest, while the control group did not. Also, the treatment group yielded far-transfer effects on the morphosyntactic processing task. Specifically, while neither group showed sensitivity to within- or across-phrase violations at pretest, the treatment group was sensitive to both types of violations after WM training, and this improvement in morphosyntactic processing was sustained at the delayed posttest. The control group, on the other hand, remained insensitive at posttest. Taken together, the findings suggest that 1) WM is a malleable system prone to changes due to stimulation and 2) it may be used as a cognitive tool to facilitate L2 morphosyntactic computation. Finally, the results support accessibility accounts claiming that cognitive resources are one of the factors responsible for the achievement of native-like computation of the L2.
The findings of the four studies of my dissertation are relevant to the field of Second Language Acquisition, because (a) they provide evidence that late bilinguals can acquire the same morphological and morphosyntactic processing mechanisms as native speakers of the target language, (b) they identify the key factors involved in the modulation of word decomposition (i.e., word frequency) and syntactic distance processing (i.e., WM), and (c) they suggest that it is possible to accelerate L2 acquisition via training regardless of L2 proficiency. The findings are also important to the field of Cognitive Psychology because they advance our understanding of the malleability of human memory, its general-purpose nature, and its association with language processing in the monolingual and bilingual mind.
NotePh.D.
NoteIncludes bibliographical references
Genretheses
LanguageEnglish
CollectionSchool of Graduate Studies Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.