Castledown Journals
Not a member yet
968 research outputs found
Sort by
Exploring the role of conversational AI in developing communicative skills: A systematic review
No full textConversational AI tools, with their capacity for interactive and dynamic language exchanges, offer promising avenues for enhancing language learning. While the potential of these tools is significant, their application in developing learners’ communicative skills remains relatively under-explored. This study investigates conversational AI tools and their applications in supporting communicative skills development among language learners. To this end, thirty-one studies published between 2010 and December 2024 were systematically reviewed across seven dimensions: (1) tools; (2) languages and skills; (3) theories and pedagogical approaches; (4) research methodologies; (5) mechanisms of employing conversational AI; (6) findings; and (7) challenges. Based on the results, conversation AI tools were categorized into five distinct types: embodied conversational agents (ECAs), speech recognition systems, chatbots, advanced AI-powered chatbots (such as ChatGPT), and general conversational agents. Advanced AI-powered chatbots emerged as the most prevalent type, likely due to their ability for personalized and real-time feedback. English was the most frequently targeted language, with a primary focus on speaking and writing skills. Most studies tended to adopt mixed methodologies, combining quantitative and qualitative techniques. Improvements were observed in learners’ communication skills, alongside increased motivation and autonomy. Methodological limitations, challenges in technological implementation, and ethical considerations emerged as key challenges. This review recognizes the transformational potential of conversational AI in developing students’ communicative skills and recommends strategies for addressing the challenges of scaling up these technologies across diverse language contexts and effectively integrating them into educational practices
Experimenting with corpus-driven approaches: On paper or on computer
No full textLanguage teachers who have received training in corpus linguistics can resort to DDL as a supplement to their conventional teaching in two ways – direct or indirect. This means that learners can use concordances indirectly through corpus-based materials designed by teachers as handouts or they can have direct computer-based experience with corpora (Chambers, 2022). The aim of this study is to provide useful skills and processes in the use of language corpora as a resource for grammar instruction and to experiment with paper-based and computer-based approaches of DDL. The study recruited 32 Armenian adult learners studying English at elementary and pre-intermediate levels outside academic context. The analysis of the evaluation data collected from pre- and post-tests reveals that while the elementary group improved their grammar competency more from the paper-based work, the pre-intermediate learners benefited from both the paper-based and hands-on training. Additionally, the semi-structured interviews sought to enhance insights into the language learning process and brought to light some of the benefits and challenges of using DDL. These findings demonstrate that DDL can be meaningfully adapted for lower proficiency learners through appropriate scaffolding and may inform future curriculum design and teacher development in comparable EFL contexts
Analyzing intercultural communicative competence in an international English textbook designed for young learners in the Indonesian context
No full textIntegrating intercultural activities in English language textbooks that foster intercultural communicative competence (ICC) is crucial for promoting global awareness and tolerance in young learners. However, previous studies have identified persistent gaps in ICC representation, especially in textbooks designed for international use. In the context of Indonesia’s Kurikulum Merdeka, which highlights global diversity as one of the Profil Pelajar Pancasila values, ICC integration becomes essential. This study investigates how ICC is reflected and to what extent it is integrated in an internationally published English textbook for Grade 1 students in Indonesia, Explore English Stage 1. A content analysis design was adopted, combining quantitative and qualitative techniques. A modified ICC checklist, adapted from Byram and Masuhara (2013), consisting of 19 indicators across four components, Knowledge, Awareness, Attitudes, and Skills, was applied. The findings indicate that the intercultural content is limited, with only 8 indicators appearing across all 58 pages. The overall ICC presence reached 60.34%, categorized as Moderate Integration based on Mishan and Kiss’s (2024) adapted scale. Moreover, the analysis of illustrations, character names, situational contexts, and classroom activities shows that the textbook mainly features observable cultural elements without fostering deeper reflection or intercultural awareness. These results align with previous research, suggesting that English textbooks often meet linguistic goals while overlooking intercultural dimensions. Therefore, textbook developers are encouraged to incorporate more explicit exposure to diverse cultures to meet curriculum expectations better and support the development of ICC in young learners
Identity positioning in teaching Intercultural Communication: A case study of Chinese EFL teachers
No full textThis qualitative study examines how Chinese EFL teachers position their identities when teaching Intercultural Communication (IC), an interdisciplinary subject. With data collected through interviews, observations, and document analysis of three experienced EFL teachers at Chinese universities, thematic analysis revealed: (1) Teachers adopt strategic roles as “organizers,” “guides,” “transmitters,” and “midwives” in IC pedagogy; (2) Their identity evolution progresses from casual taster to IC promoter, apprentice to independent practitioner, and student to creative professional; (3) Sustained IC teaching expands their identities beyond EFL teachers to encompass IC instructors, learners, and researchers. Findings highlight a reciprocal relationship between IC expertise development and identity formation, with pedagogical growth shaping identity and vice versa. The study offers insights for EFL teacher development in intercultural education and specialized content teaching contexts
Re-imagining knowledge canons in South African education: An inquest into mother tongue-based bilingual education
No full textThis paper examines Mother Tongue-based Bilingual Education (MTbBE) in South Africa as a site for epistemic transformation rather than mere linguistic reform. While policy efforts have widened access by including African languages, classroom and institutional practices often remain anchored in Eurocentric epistemologies. Drawing on decolonial theory and existing empirical studies, the paper argues that MTbBE can only fulfil its transformative promise when language reform is coupled with pedagogical shifts, curriculum design, and institutional reward systems. It shows that translanguaging and bilingual pedagogy enhance comprehension and participation, yet genuine transformation requires teachers to use African languages as reasoning and knowledge creation resources. Re-authoring curricula and restructuring institutional incentives emerge as essential for sustaining such change. The paper contributes to debates on decolonisation by reframing MTbBE as a strategy for epistemic justice that links linguistic inclusion with cognitive renewal. It concludes that when African languages are recognised as instruments of inquiry, education can move from translation to transformation, fostering an intellectual order that reflects South Africa’s multilingual and plural realities
Investigating the role of mindful, meaningful, and joyful learning in promoting deep learning in AI-based language learning
No full textThis study aimed to investigate whether Mindful Learning (MiL), Meaningful Learning (MeL), and Joyful Learning (JL) are predictors of Deep Learning (DL) in AI-based language learning. The study employed a convergent mixed-methods design by combining phenomenology and ex-post facto approaches. This study recruited eight EFL students for the interviews and 276 EFL students for the quantitative part. The data were collected through semi-structured interviews, researcher field notes, and questionnaires. The data were analyzed using inductive thematic analysis, descriptive statistics, Pearson Product Moment, and SEM. The findings revealed that qualitatively, MiL, MeL, and JL are related to DL. However, the SEM outputs indicated that JL is the only non-predictor. The study offers four implications directed for the use of AI-based applications in facilitating DL in AI-based learning in EFL contexts
Assessing L2 Japanese speaking in role-play tasks on an online meeting platform
No full textThis study examines the assessment of L2 Japanese speaking in online role-play tasks from a sociolinguistic-interactional perspective. Twenty L2 learners from China engaged in online role-plays with a professor while performing various speech acts. Four raters evaluated the participants’ video-recorded performances using five criteria adapted from Youn (2015; 2020). Multifaceted Rasch Measurement (MFRM) confirms the raters were able to assess the participants’ oral performances effectively using the rubric. In addition to the quantitative analysis, qualitative transcript analysis highlights differences between higher- and lower-rated learners’ oral performances. Compared with the high-scoring performances, the low-scoring performances showed difficulties such as inaccurate language use and long pauses both within and between turns. When communication breaks down, managing intersubjectivity becomes essential. Speakers with high interactional proficiency handle it subtly, while for low-proficiency learners, the negotiation of meaning, use of repairs, and visible struggles to maintain intersubjectivity are much more apparent. Findings provide insights into L2 learners’ interactive performance in online settings compared to face-to-face contexts, contributing to a deeper understanding of assessing L2 speakers’ sociolinguistic-interactional skills in online role-play. 
The marginalisation of heritage languages, heritage language education and heritage language teachers in Italy: Results from a case study
No full textThis paper investigates the marginalisation of heritage languages (HLs), heritage language education (HLE), and heritage language teachers (HLTs) within the Italian educational system. The inclusive orientation of Italian schools has expanded since the 1970s, with respective efforts mainly addressing students with specific educational needs due to learning difficulties and disabilities, while largely neglecting the specific ressources and needs of students speaking heritage languages, that is, other languages than the language(s) of instruction. Drawing from a migration pedagogical framework, the study analyses the qualifications, training opportunities, and working conditions of HLTs within a multiple case study design. 15 in-depth semi-structured individual interviews were conducted with HLTs and other educational stakeholders. The data were analyzed using a structuring qualitative content analysis.
The paper highlights contrasting perspectives among educational stakeholders and national and regional guidelines. Despite national ministerial guidelines advocating for the inclusion and promotion of HLs, HLE is not systematically embedded in the curriculum and is typically initiated at the local level by municipalities or community organisations. Italy’s approach to HLE is fragmented and the results from interviews with educational stakeholders reveal the marginalisation of heritage languages, heritage language education and its teaching staff. The findings point to a need for systematic, research-informed reforms to improve the status, conditions, and professional development of HLTs and to integrate HLE more robustly into the Italian education system
Compassion and emptiness: The Four Immeasurable Minds and the Three Types of Great Compassion in Candrakīrti’s Madhyamakāvatārabhāṣya
No full textCompassion has long been central to Buddhist thought and practice, shaping both per sonal development and the broader vision of human flourishing. This article compares two key frameworks: The Four Immeasurable Minds (brahmavihāra) of early Buddhism and the Three Types of Great Compassion (mahākarun .ā) articulated by Candrakīrti in his Madhyamakāvatārabhāṣya. The Four Immeasurables of loving-kindness (maitrī), compassion (karun .ā), sympathetic joy (muditā), and equanimity (upekṣā) are presented as meditative practices that cultivate boundless attitudes of care and prepare the mind for wisdom. Candrakīrti’s threefold model deepens this foundation by showing how compassion evolves from empathy for beings to recognition of the ignorance underlying suffering, to a non-objectified compassion inseparable from the realization of emptiness (śūnyatā). By placing these frameworks side by side, the study highlights both their continuity and their distinctive emphases: the Four Immeasurables ground compassion in ethical and psychological training, whereas Candrakīrti reframes it as an ontological principle at the heart of the bodhisattva path. The comparison shows that Buddhist compassion is not merely an emotional response but a transformative discipline that integrates ethical practice with philosophical insight. Beyond textual analysis, this study suggests that Buddhist approaches to compassion offer conceptual resources for contemporary reflection on empathy, human relationships, and the cultivation of a more humane society. Rather than providing prescriptive solutions, these teachings inspire ways of linking inner transformation with outward ethical responsibility. In this way, ancient Buddhist traditions continue to inform modern conversations about spirituality, ethics, and the conditions for human wellbeing
Linguistic control in AI text generation: An accessible prompt-based approach targeting L2 Spanish absolute beginners
No full textGenerative AI offers strong potential for developing customized second language learning materials and tools. However, generating texts for absolute beginners—requiring strict lexical and grammatical control—remains a challenge. Although controlled text generation (CTG) techniques exist, they often require technical expertise and infrastructure, limiting accessibility for educators.
This study evaluates, in the context of Spanish, a prompt-based approach that leverages large language models (LLMs) without fine-tuning or specialized tools. Prompts enforce linguistic constraints defined in two attachments: a categorized Spanish vocabulary list, and a set of example sentences illustrating approved Spanish grammatical structures organized by communicative function.
Three variables were manipulated: AI model (ChatGPT-4o vs. Claude 3.5 Sonnet), prompt type (standard vs. extended, with constraint-enhancing techniques), and attachment format (rich-heavyweight vs. lightweight JSON). A secondary variable, text type (city descriptions, personal introductions, and dialogues), was also examined. A total of 720 texts were generated, 30 per condition. Measures included proportions of non-compliant lexical and grammatical items, user-perceived latency, and errors in vocabulary, grammar, and coherence.
Model choice was the primary driver of constraint adherence, with Claude 3.5 Sonnet outperforming ChatGPT-4o. Extended prompts improved adherence across models. Attachment format showed no systematic effect on adherence, but JSON significantly reduced latency and response-time variability. Text type also influenced adherence, and error rates remained low.
Findings offer educators a scalable, low-barrier solution for generating tailored beginner-level Spanish materials and AI-powered tools using LLMs, along with insights into how different design choices affect performance. This approach, transferable to other languages, provides a practical alternative to resource-intensive CTG techniques, addressing a critical gap in AI-assisted language education