Morphological processing in Mandarin Chinese learners of English

A key aspect of learning English as a second (L2) or additional language is establishing the relationships between morphologically related forms, such as the stem create, the inflected form created and the derived form creation. Are these stored as separate but related forms, or do learners establish interlanguage rules allowing such forms to be generated (in production) and accessed via morphological decomposition (in comprehension)? Does the mental representation of stems and their complex forms further depend on the nature of the morphological relationship (e.g., inflection vs. derivation)? The answers to these questions have implications for how morphologically complex words (MCWs) are stored in learners’ mental lexicons.

Previous research into L2 morphological processing and the learning of MCWs has highlighted the relevance of lexical factors, language differences and L2 proficiency. The research reported in this thesis provides further evidence relating to each of these factors, addressing areas that have hitherto been insufficiently investigated. First, with regard to lexical factors, while there is considerable evidence that orthographic and semantic relationships are important, relatively little attention has been directed to phonological relationships. Second, and relatedly, while Mandarin Chinese and English have been studied as typologically distinct languages, little attention has been given to the relative weights given to phonological and orthographic information in first language (L1) processing in these two languages, and to how these weightings might affect L2 processing by Mandarin Chinese learners of English. Last, L2 proficiency as part of a learner’s profile has mainly been measured by self-report and/or vocabulary or placement tests. It is unknown to what extent other aspects of learners’ profiles may affect word processing in L2, such as their reliance on different vocabulary learning strategies, e.g., paying deliberate attention during learning to affixes or to orthographic or phonological information. This study attempts to fill these gaps by investigating 1) how Mandarin Chinese learners of English may process English MCWs; and 2) how their processing is affected by their L2 proficiency and use of vocabulary learning strategies.

While most previous relevant investigations of morphological processing used lexical decision tasks with either masked priming or cross-modal priming, this study used visual lexical decision with mediated priming. This use of mediated priming was designed to separate semantic, orthographic and phonological relatedness between stems and their morphologically related forms, which in prior studies are usually presented as primes and targets. Take the set door-(pour)-poured as an example. door is the prime, poured is the target and the stem pour is the mediator. This mediator rhymes with the prime door. In the main experiment, participants only see and make word decisions for door and poured, and the mediator pour is not presented to participants. If door activates pour via rhyme priming (which is tested in a preliminary experiment), and if there is morphological decomposition then pour will give access to poured and thus door will effectively prime poured, even though the prime and the target are not morphologically or semantically related. This study consists of two stages. The first stage served to select participants and materials for the second experimental stage. Stage 1 used a lexical decision task (LexTALE), a C-test (a modified version of a cloze test) and a background questionnaire, which together provided information about participants’ English learning background, their use of vocabulary learning strategies and their English proficiency. The participants for Stage 2 were selected from this pool of participants, meaning that their background data could be included in the analysis of the experimental studies. Stage 1 also included a familiarity rating task to ensure that all the real word stimuli in Stage 2 were likely to be known to participants. Stage 2 consisted of two experiments. Since mediated priming requires the mediator to actually be primed by the prime, Experiment 1 used a visual lexical decision task involving just the primes and the mediators (e.g., door and pour) to select the materials that are most likely to demonstrate that the primes to be used in Experiment 2 would in fact prime the mediators. Based on the results of Experiment 1, primes and targets were chosen for Experiment 2, the main experiment, where the target was a morphological relative of the mediator tested in Experiment 1 (e.g., the prime-target pair door-poured in Experiment 2 was selected based on rhyme priming being found in Experiment 1 between door and pour). Experiment 2 then used a visual lexical decision task with mediated priming to test how participants process MCWs, particularly inflected words and derived words.

The response time data in Experiment 2 showed no clear evidence that morphological structure was influencing priming. There were, however, some subtle effects in the accuracy rates of responses to the targets that suggest that morphological structure is important under certain circumstances. First, accuracy rates for responses to inflected forms (poured) were higher than those to rhyme controls (board) following the prime (door), but only when the targets were of lower word frequency. I conjecture that the weak form-based activation of these low frequency words by the prime is supplemented by morphological priming through the mediator (pour), facilitating recognition and thus higher accuracy if not quicker response times. Second, accuracy rates for responses to inflected forms were also higher than those to rhyme controls for participants whose language learning profiles showed relatively little reliance on oral repetition strategies. That is, the weaker phonological links for these participants between primes like door and targets like poured are supplemented by the morphological link between the mediator pour and the target, a link which is not available for the rhyme target board. Third, while accuracy rates for irregularly inflected targets remained unaffected by participants’ English proficiency, accuracy was overall higher and increased with increasing proficiency for regularly inflected targets. In other words, the participants examined in this thesis fall across a proficiency range where the influence of regular inflectional processes starts to be evident, but the same cannot be said for irregular processes. Finally, the accuracy results for materials designed to test the influence of derivational morphology showed higher accuracy both for words with real affixes (e.g., heater after the prime street, where the assumed mediator heat rhymes with street) and for words with pseudo-affixes (message, with the pseudo-affix -age, after the prime dress where the assumed mediator was mess), compared with a control condition. This suggests that the learners in this study have had sufficient exposure to derivational affixes for these to have an impact on their responses to target words, but insufficient exposure for them to be able to separate real from pseudo-affixation.

This thesis contributes to our understanding of morphological processing by Mandarin Chinese-speaking learners of English, especially from the perspective of individual differences, with a new approach, mediated priming. The findings also shed some light on how L2 proficiency and vocabulary learning strategies may affect morphological representations in L2 learners. Moreover, this thesis provides some implications for methodology development in L2 research, such as adopting mediated priming in L2 studies, as well as investigating the role of not only L2 proficiency but also vocabulary learning strategies.