Research Methods in Applied Linguistics最新文献

This article provides an introduction to finite mixture models in an applied linguistics context. Mixture models can be used to address questions relating to whether there are unknown subgroups in one's data, and if so, which participants/texts are likely to belong to which subgroup. Put differently, the technique enables us to assess whether our data might come from a heterogeneous population that is made up of latent classes. As such, mixture models offer a model-based framework to answer research questions for which the field previously has either attempted to use nonparametric heuristic techniques (e.g., cluster analysis) or has left entirely unanswered. An example of such research questions would be, ‘Does the treatment work equally well for all the participants, or are there unknown subgroups in the data that respond differently to the treatment?’ The article starts by introducing univariate mixture models and then broadens the scope to cover bivariate and multivariate mixture models. It also discusses some known pitfalls of the technique and how one might ameliorate these in practice.

In response to the call for integrating a praxis orientation to understand processes of language development through active engagement with teachers and learners, this article introduces techno-reflective narrative interview (TRNI) as a tool to understand how learners invest in their English as a second language (ESL) language and literacy practices as social practices in English medium instruction (EMI) courses. It is timely in integrating both technical and critical guidance for methodological innovations in the field of applied linguistics. To illustrate how this aim is achieved, this article first justifies the significance and theoretical basis of TRNI by elaborating on the definitions of and underpinning rationale for it, followed by the relationship among second language (L2) learners, investment, and TRNI. It then discusses the methods adopted by the 69 most cited empirical studies from Google Scholar that used the model of L2 investment as the theoretical model and compares these methods with TRNI. Finally, it discusses how TRNI was developed and adopted in two action research projects conducted by the author and two lecturers among local undergraduate students in the Hong Kong context. Through these steps, I demonstrate how TRNIs can be adopted or adapted to provide an interactive space for ESL learners to reconstruct and perform their English speaker and learner identities, share their learning experiences with the agency to introduce their positions, voices, and stories, and therefore, claim legitimacy as English users in EMI courses.

The use of natural language processing tools such as part-of-speech taggers and syntactic parsers are increasingly being used in studies of second language (L2) proficiency and development. However, relatively little work has focused on reporting on the accuracy of these tools or optimizing their performance in L2 contexts. While some studies reference the published overall accuracy of a particular tool or include a small-scale accuracy analysis, very few (if any) studies provide a comprehensive account of the performance of taggers and parsers across a range of written and spoken registers. In this study, we provide a large-scale accuracy analysis of popular taggers and parsers across L1 and L2 written and spoken texts, both when default and L2-optimized models are used. Accuracy is examined both at the feature level (e.g., identifying adjective-noun relationships) and the text level (e.g., mean mutualinformation scores). The results highlight the strength and weaknesses of these tools.

Taboo topics tend, by their definition, to be censored in language use, and are thus often absent from discourse. When such topics are discussed, they tend to be referred to indirectly, for example through euphemistic language. This presents a challenge for the design and construction of topic-specific corpora, then, when the topic being investigated might be considered taboo within the culture and/or discourse context under study. In this article, we explore the challenges involved in attempting to construct a corpus of news media texts that are ‘about’ a taboo topic. Focusing on the case of incest – an issue deeply entrenched in social and linguistic taboos – we present an iterative, corpus-assisted approach to designing, assessing and (re)constructing a corpus of UK newspaper articles about this topic. As well as contributing to our understanding of the representation of incest in UK news media, this article underscores the importance of transparency and reflexivity in the process of (iterative) corpus design and serves to demonstrate how the reporting of this process might proceed in other studies whose data similarly represent a product of iterative design.

This paper examines how multilingual English language teachers use narrative for learning and teaching writing given their own writing practices and beliefs. The data was collected and analyzed using different narrative approaches to explore the advantages and disadvantages of each of these three models: Labov, Ochs and Capps, and De Fina & Georgakopoulou's Social Interactional Approach (SIA). Each model highlights different aspects of narrative. However, due to the contextual nature of multilingual writing, the SIA model proved to be the most comprehensive for knitting together teacher writing practices, beliefs about language-learning, and their pedagogies. Nevertheless, applying all three models proved useful for highlighting the dynamics of language use in different contexts along with multidimensional aspects of storytelling.

We have conducted second language (L2) listener verbal reports (VRs) for many years, and during that time have also advised or trained colleagues and research assistants to use these VRs. Based on these experiences, this article outlines various common missteps we and our trainees have made in conducting the VRs and how these missteps were, or could be, addressed. Through this approach, we provide advice to novice researchers planning to use this method to study L2 listeners. This advice is given as a series of recommendations in the following areas: the type of texts to use in the VR and how to select and trial them beforehand; how to prepare the participants for their VR; the language to use in the VR (L1 or L2); and procedural advice when conducting the VR. In addition, there's an extended section advising on the use of researcher questions to help elicit information from the participants about their listening behaviors during the VR. This is followed by advice on how to generally help train a novice researcher to conduct listener VRs. The article concludes by identifying some possible future research directions for addressing validity threats to listener VRs.

This introduction provides an overview of the state of research on and knowledge about the use of elicited imitation in applied linguistics research, highlighting gaps in our current understanding of this tool. It then introduces the eight articles that comprise the special issue.

Sentence repetition tasks (SRTs) have been extensively used as measures of bilinguals’ language abilities. Most studies relied on SRTs in which the target sentences were not connected to each other. However, participants’ performance may differ if these sentences are embedded in discourse, since discourse provides participants with additional cues for sentence comprehension and interpretation. For the present study, we designed a discourse-based SRT, whereby the target sentences were connected to each other in a story. We examined the effect of discourse on bilinguals’ performance in the SRT and investigated whether this effect varied based on the language of administration, bilinguals’ dominance score and type of target structure. We tested 32 Italian-German bilingual children (7–12 years) living in Germany with two SRTs in each language, one with discourse and one without discourse. Participants showed a better performance in the SRTs with discourse, especially in the heritage language (Italian). The effect of discourse was visible across the board with all target structures. On the whole, SRTs with discourse seem to reduce the processing costs associated with lexical retrieval and shifts in scenarios, thus tapping more directly into children's processing abilities, compared to more traditional SRTs. The results are discussed in terms of ecological validity of different assessment instruments.

Meaning-recognition and meaning-recall vocabulary tests are commonly used to assess knowledge of the form-meaning link as it relates to the receptive skills. Although meaning-recognition is generally more convenient, think-aloud protocols have revealed that in comparison to meaning-recall, meaning-recognition is more susceptible to blind guessing and the use of construct-irrelevant test strategies. Perhaps because of this, meaning-recall tends to be a stronger predictor of reading ability. Following Budescu (1988), this article reports on three studies that investigated an extended-matching test (EMT) format that was designed to address these limitations of meaning-recognition while retaining its convenience. An EMT with 90 target words was developed. It contained three clusters, each with a 30:130 ratio of target words to Vietnamese L2 response options. In comparison to meaning-recall criterion measures, the EMT did not meaningfully differ in terms of internal reliability, mean scores, and, importantly, the strength of the correlation with reading comprehension scores. The consistency of correct/incorrect response classifications ranged from 83 % to 86 %. These initial findings suggest that the EMT format may be used interchangeably with meaning-recall for many research purposes.

This study examines the criterion validity of curriculum-based measurement of writing (CBM-W) for assessing the writing of young language learners in an EFL context. CBM-W has been found to have criterion validity with a variety of tests and student populations, including learners in ESL settings taking standardized tests. However, evidence of CBM-W's validity for use with young learners in EFL settings is limited. This study seeks to address this gap by investigating the criterion validity argument for the following CBM-W scoring procedures: words written (WW), words spelled correctly (WSC), correct word sequences (CWS), and correct minus incorrect word sequences (CIWS). The data set included classroom-administered narrative prompt writing samples and official test scores from the Cambridge Assessment English Young Learners exam reading and writing test collected from 51 Taiwanese English learners aged seven to eight. A Bayesian correlation data analysis found a greater probability favoring the criterion validity argument for CWS and WSC, with strong evidence for CWS and moderate evidence for WSC. This study sets the foundation for future research examining the validity argument of CBM-W for evaluating young EFL learners’ writing in research and classroom settings.