Abstract
Keywords
Introduction
Vocabulary acquisition is fundamental to language learning, demanding both systematic instruction and extensive exposure for successful mastery (Nation, 2017; Schmitt, 2008). Building comprehensive lexical knowledge requires substantial cognitive effort, particularly in English as a Foreign Language (EFL) context, where learners must acquire not only word meanings but also their phonological, morphological, syntactic, and sociolinguistic properties (Laufer, 2009; Zhang & Koda, 2018). This multifaceted nature of vocabulary learning makes it one of the most persistent challenges faced by foreign language learners. The process is further complicated by the need for learners to notice and process new lexical items in meaningful contexts (Schmidt, 1990), while also avoiding cognitive overload when presented with dense or decontextualized input (Sweller, 1988).
In recent years, technological advancements have profoundly reshaped language education, establishing Technology-Enhanced Vocabulary Learning (TEVL) as a vibrant and critical field of inquiry. The research landscape has rapidly expanded to encompass a diverse array of approaches, ranging from mobile applications designed for ubiquitous learning (Harger, 2020) and digital games that enhance motivation (Zou et al., 2021), to collaborative social media platforms (Reinhardt, 2018) and, most recently, adaptive systems powered by artificial intelligence (Jia et al., 2022). While empirical studies consistently demonstrate the efficacy of these individual technologies (Hao et al., 2021), this rapid proliferation has created a rich but increasingly fragmented knowledge base. Such fragmentation makes it challenging for researchers and practitioners to discern the overarching intellectual structure, collaborative patterns, and evolutionary trends of the field as a whole.
Positioning the Current Study Within the Landscape of TEVL Research
To justify the present study, it is crucial to first critically map these existing reviews and identify the specific gap this bibliometric analysis aims to fill. Broadly, existing reviews can be categorized into two main streams.
The first stream consists of quantitative meta-analyses focused on effectiveness, which seek to answer the question: “What works?” These studies synthesize empirical results to calculate the effectiveness (i.e., effect size) of specific technological tools. For instance, meta-analyses by Lin and Lin (2019) and Hao et al. (2021) have established the moderate effectiveness of mobile-assisted and technology-assisted vocabulary learning, respectively. Similarly, Yu and Trainin (2022) confirmed the benefits of technology for long-term vocabulary retention. While these focused reviews provide invaluable, deep evidence on the efficacy of individual technologies (e.g., digital games, mobile apps), they inherently operate within “technological silos.” They are designed to assess the effectiveness of a pre-defined path, not to map all available paths or to illustrate how these different technological streams intersect, compete, or evolve together over time.
A second stream of systematic reviews addresses the “why and how” of TEVL, collating theoretical frameworks and critiquing research methodologies. These reviews have identified significant gaps, noting that many studies fail to articulate a clear theoretical foundation (Yang et al., 2021; Zhou et al., 2024). Furthermore, they consistently highlight prevalent methodological limitations, such as small sample sizes and brief intervention periods (Hwang & Fu, 2019; Vnucko & Klimova, 2023). While these reviews provide profound qualitative insights, their conclusions are typically drawn from a curated selection of studies. They can diagnose problems within the field but are not designed to provide a data-driven, macro-level map of its intellectual architecture, foundational pillars, or evolutionary trajectory based on thousands of publications.
While both streams of research are essential, a significant gap remains. Neither the siloed meta-analyses nor the qualitative systematic reviews can offer a holistic, structural view of the TEVL field as a whole. A science map that visualizes the entire territory of inquiry is currently lacking. Such a map would reveal the field’s key contributors, collaborative networks, thematic evolution, and emerging frontiers. By moving beyond evaluating a single intervention or qualitatively summarizing trends, it provides a data-driven, comprehensive knowledge map of the TEVL field over the last decade, complementing existing reviews by offering a perspective that is both panoramic and structurally deep.
Research Questions
This study aims to answer the following research questions:
Materials and Methods
The study conducted a scientific and comprehensive analysis of TEVL via CiteSpace, VOSviewer and the Biblioshiny package in
Data Sources and Extraction
This study employed the Web of Science (WoS) Core Collection database (2014–2024). The WoS Core Collection was chosen as the sole data source for this study due to its reputation for high-quality, peer-reviewed literature and its comprehensive citation data, which is essential for conducting robust co-citation and network analyses. While other databases like Scopus offer broader coverage, the data consistency and quality within WoS are considered highly reliable for mapping the intellectual structure of a scientific field (Aria & Cuccurullo, 2017). This choice prioritizes data quality and analytical rigor, though its limitations are acknowledged.
The search strategy utilized three interconnected keyword clusters: (a) vocabulary-related terms (“vocabulary” OR “word”), (b) technology-related terms (“technology” OR “AI” OR “artificial intelligence” OR “computer” OR “mobile” OR “digital” OR “online” OR “deep learning”), and (c) educational context terms (“learning” OR “education”). These clusters were combined using Boolean operators to ensure precise capture of relevant publications. The selection of keywords was based on considerations of the technological evolution trends in the research topic.
The search was confined to English-language publications indexed in four WoS indices (SCI-EXPANDED, SSCI, A&HCI, and ESCI) and restricted to articles, review articles, and early access publications. Further refinement focused on four WoS categories: Education Educational Research, Linguistics, Language Linguistics, and Education Special. The downloaded dataset comprised 3,316 records, retrieved in seven sequential batches due to platform export limitations (Figure 1). After removing of duplicates, the final dataset included for analysis contains 3,307 records. The bibliographic data, including authors, titles, abstracts, keywords, references, and citation metrics, were exported in plain text format and analyzed using Biblioshiny package in
Flowchart for selection of documents for bibliometric analysis.
Bibliometric Analysis
The three complementary software tools employed in this study are selected for their own unique strengths. Biblioshiny is an
Results
Table 1 provides an overview of the data collection. The final dataset includes 3,307 documents published between 2014 and 2024, distributed across 651 sources. The corpus demonstrates a robust annual growth rate of 15.01%, indicating increasing scholarly attention to technology-enhanced vocabulary learning. Of these publications, the majority (2,910) are research articles, supplemented by 229 early access articles and 159 reviews, reflecting the field’s emphasis on empirical research.
Main Information About Data.
The collection involves 8,005 authors, with a notable tendency toward collaborative scholarship as evidenced by the average of 2.88 co-authors per document. While 711 publications are single-authored, the majority are collaborative efforts, with 18.54% involving international co-authorships. This collaborative pattern suggests a growing global research network in the field. The research impact is substantial, with documents averaging 11.25 citations each and maintaining a relatively current knowledge base with an average document age of 3.69 years. The intellectual breadth of the field is reflected in 8,353 author-designated keywords and 3,154 system-generated keywords, indicating diverse research focus within the research domain. The substantial reference count of 123,721 demonstrates the field’s rich theoretical and empirical foundation.
Annual Scientific Production
Figure 2 depicts the annual scientific production in TEVL research from 2014 to 2024. The field exhibits a consistent upward trajectory, with publications increasing from 139 in 2014 to 563 in 2024. Three distinct growth phases are observable: (a) a steady growth period (2014–2019) with annual publications rising from 139 to 255; (b) an acceleration phase (2019–2021) showing a marked increase to 373 publications; and (c) a rapid expansion phase (2021–2024) reaching 563 publications. Notably, the post-2020 period demonstrates the steepest growth, reflecting not only the global catalyst of the COVID-19 pandemic but also, arguably, a maturation of the field itself, where research tools and methodologies became more accessible and established, enabling a wider range of scholars to contribute.
Annual scientific productions.
Analysis of Cited Documents
Average Article Citations Per Year
Figure 3 presents the average annual citations of TEVL articles generated by Biblioshiny. Older publications generally display higher total citation counts due to their longer exposure, but their mean annual citations are more modest, as seen in 2014 (2.71) and 2015 (2.2). A notable increase occurs between 2018 and 2020, with mean annual citations peaking at 3.07 in 2020, likely driven by the pandemic’s acceleration of research on adaptive and remote learning technologies. Recent years, particularly 2023 (2.0) and 2024 (1.17), show declining averages primarily due to citation lag and the increasing number of publications diluting citations. However, exceptions exist where recent articles achieve high citation rates, driven by their alignment with emerging research priorities.
Average article citations per year.
Most Highly Cited Documents
The most highly cited papers in TEVL research are examined in Biblioshiny using two important indexes: Global Citation Score (GCS), which shows the total number of citations in WoS at the time the data were retrieved, and Local Citation Score (LCS), which shows the number of citations in the current dataset. Beginning with a historiographical review to identify key literature, I will then analyze globally cited papers, as LCS cannot exceed GCS.
According to Table 2, the top three cited ones are all review articles (Burston, 2015; Golonka et al., 2014; Takacs et al., 2015). Other notable empirical works include Wu’s
Historiography.
Figure 4 presents the 15 most globally cited documents in the field in the past decade. The most cited work by Golonka et al. (2014; GCS: 479) provided a comprehensive effectiveness framework for language learning technologies, catalyzing subsequent empirical investigations. Building on this foundation, Takacs et al. (2015; GCS: 219) and Burston (2015; GCS: 200) contributed crucial meta-analytic evidence, with the former establishing the efficacy of multimedia features in vocabulary acquisition and the latter critically examining methodological rigor in mobile-assisted language learning research.
Most global cited documents.
The field’s theoretical maturation is evidenced by the emergence of more specialized investigations. Hung et al. (2018; GCS: 152) analysis of digital game-based learning and Hsu (2017; GCS: 154) examination of augmented reality applications exemplify this trend, offering empirical validation for specific technological approaches while refining pedagogical frameworks. Recent high-impact works by Lin and Lin (2019; GCS: 134) and Zou et al. (2021; GCS: 121) represent the field’s current trajectory toward integrated approaches. Their analyses synthesize previous findings while identifying emerging trends, particularly in adaptive learning systems and virtual reality applications. This evolution reflects the field’s progression from basic effectiveness studies to sophisticated implementations that combine multiple technological affordances for optimal vocabulary acquisition outcomes. The citation patterns show a linear theoretical progression, with each stage upholding methodological rigor while expanding and building upon earlier discoveries.
Co-citation Analysis of Papers
Co-citation analysis, introduced by Henry Small (1973), measures how frequently two documents are cited together in subsequent publications. This method reveals the intellectual structure of research fields by identifying pairs of works that researchers frequently reference together, suggesting these works share related concepts, methods, or theoretical foundations (McCain, 1990). Through network visualization, co-citation analysis can effectively map the core literature of a field, identify influential research clusters, and demonstrate how different research streams are interconnected, making it a valuable tool for understanding the evolution and structure of academic knowledge domains (White & McCain, 1998).
A co-citation network analysis was conducted using VOSviewer to examine the intellectual structure of the field (Figure 5). The visualization reveals five distinct clusters, representing key research domains and their interconnections through cited references. The prominence of Vygotsky (1978), a foundational theorist of social learning, alongside Schmitt (2008), a key figure in applied linguistics pedagogy, is particularly telling. This duality suggests that the intellectual heart of TEVL research is not purely technological, but rather a dynamic synthesis of sociocultural learning theory and concrete vocabulary teaching principles. This finding provides a data-driven confirmation for the theoretical underpinnings that will be explored in the Discussion section.
Co-citation network of cited references.
Regarding closeness centrality, Kukulska-Hulme and Shield (2008) shows high measures in the mobile learning cluster (blue), with Liu and Chu (2010) and Sylvén and Sundqvist (2012) demonstrating similar prominence in technology integration research (yellow). The five clusters represent: theoretical foundations (red, centered on Vygotsky), mobile learning applications (blue), language teaching methodology (green), educational technology integration (yellow), and language assessment (purple).
Sources
Figure 6 presents the most influential sources in TEVL research over the past decade. Analysis of publication patterns reveals that the top 10 sources contributed 686 documents, accounting for 28.5% of the sampled literature, indicating concentrated scholarly attention in specific journals.
Top 10 relevant journals.
The three most productive journals in this domain are
Authors, Affiliations and Countries
Prolific Authors
Figure 7 presents the publication patterns of the top 20 most relevant authors in the TEVL field from 2014 to 2024. From the productivity visualization, several prominent research groups emerge.
Top-authors’ production over the time.
The first group is led by Zou D. (
According to the visualization, the majority of productive writers continued to pursue active research agendas from 2018 to 2022, and the overlapping publication years show growing patterns of collaboration. The timeline’s bubble sizes highlight times of high productivity; multiple authors produced a lot of work between 2020 and 2022, indicating a rise in interest in TEVL strategies both during and after the global pandemic.
Figure 8 illuminates the intellectual structure of TEVL research through the interconnections among cited references (CR), authors (AU), and author keywords (DE). The bibliometric network reveals three distinct but interrelated components.
Three-fields co-occurrence plot.
First, the theoretical foundations are anchored by seminal works, including Schmitt (2008), Nation (2017) and Vygotsky (1978), which established fundamental principles in vocabulary acquisition and cognitive development. Second, the author distribution identifies key contributors, with Hwang GJ, Shadiev R, and Zou D emerging as central figures in the field.
The network analysis shows that top researchers have different research paths. Particularly in studies on mobile-assisted vocabulary learning, Hwang GJ and Shadiev R show close ties to the areas of vocabulary acquisition and mobile learning. Zou D’s work emphasizes the integration of digital platforms and demonstrates significant connections to online learning and language instruction. Notably, research of Golonka et al. (2014) on computer-assisted language learning provides an essential link between vocabulary acquisition techniques and mobile learning applications. The keyword analysis also shows that higher education is a noteworthy study environment, with Zhang RF and Kennedy MJ concentrating mostly on tertiary-level TEVL.
Most Relevant Countries
The global cooperation tendencies in research visualized by VOSviewer are shown in Figure 9. Given its greatest node size and central location, the United States stands out as the primary hub, highlighting its crucial role in global research cooperation. The second most important node is China, which has close bilateral ties with the United States and acts as a link to other Asian nations, including South Korea and Japan. The network shows three different geographical clustering patterns: a Middle Eastern cluster that includes Iran, Iraq, and Indonesia; an Asian cluster that is led by China and includes Japan and South Korea; and a Western cluster that includes the United States, Canada, Germany, Spain, and France. The USA-China and USA-Germany partnerships have the strongest links, with the thickness of connecting lines signifying the level of cooperation. Although research collaboration is disseminated globally, this network structure indicates that it is dominated by collaborations between large research-producing nations, especially the USA and China, and is marked by significant regional links. This bipolar concentration of research output in the USA and China, while demonstrating the field’s dynamism in these regions, simultaneously raises critical questions about the global diversity of TEVL research. It suggests that the dominant narratives, research questions, and technological solutions in the field may be heavily shaped by the educational ecosystems of these two superpowers, a point that warrants deeper critical examination.
International co-authorship network.
The USA (1,683 documents), China (1,179 documents), UK (379 documents), Australia (285 documents), and Spain (243 documents) held the top five spots among the top 10 nations that produced TEVL research output throughout the study period (Table 3). The significant contributions from Asian-Pacific nations, such as Japan (144 documents), Australia (285 documents), and other Asian countries, point to the increasing significance of TEVL research in areas where English is the primary language of instruction. While emerging economies like Turkey (221 documents) and Iran (185 documents) show growing research capacity in TEVL, European representation is noteworthy through nations like the UK (379 documents), Spain (243 documents), and Germany (168 documents). This reflects the global nature of this field and its relevance to a variety of educational contexts.
Top 30 Countries/Regions in TEVL Research Production.
Keywords and Research Trends
Keyword Co-occurrence Network
Three separate thematic clusters generated by VOSviewer in research on TEVL are shown in Figure 10. With phrases like “education,”“technology,”“students,” and “teachers,” the red cluster focuses on pedagogical features and emphasizes stakeholder viewpoints and educational implementation. Terms like “vocabulary,”“acquisition,”“English,” and “second-language” predominate in the green cluster, which focuses on language acquisition processes and highlights the theoretical underpinnings of vocabulary development. “Meta-analysis,”“comprehension,”“writing,” and “skills,” among other methodological and outcome-related elements, are represented by the blue cluster, which also denotes research methodologies and learning objectives. The word “vocabulary” connects all three clusters and serves as a bridging idea, showing up as a major node with strong betweenness centrality. The density and connectivity of the network point to a highly integrated field of study where linguistic, educational, and technical elements are intimately related. The size of the nodes, especially for phrases like “vocabulary,”“education,” and “students,” reflects how frequently they appear in the literature and how important they are.
Keyword co-occurrence network.
Research Trends
Eleven primary research streams are depicted in the thematic evolution map by CiteSpace (Figure 11), where diverse research clusters are represented by different colors and the width of lines indicates the strength of thematic links. Several noteworthy tendencies are revealed by the investigation. Research first concentrated on basic elements such as vocabulary acquisition (#2) and literacy (#0; 2015–2018). Following that, the area shifted toward more technology-specific applications, with game-based learning (#3) and mobile learning (#1) being major themes in 2019 to 2021. More focus has been placed on advanced technical applications in the last several years (2022–2024), especially augmented reality (#6), machine learning (#7), and artificial intelligence (#8). This development is consistent with the publication trends of eminent writers like Hwang GJ and Zou D., who have made substantial contributions to the study of gamified and mobile-assisted vocabulary learning.
Thematic evolution map.
Additionally, the visualization shows a move away from conventional computer-assisted vocabulary training and toward more advanced methods. Social media (#9) and gamification (#5) have emerged as separate clusters, indicating an increasing focus on social and interactive learning settings. According to the research interests of leading experts in the field, this tendency is consistent with the growing use of digital tools in language training. Furthermore, the fact that reading comprehension (#4) has remained relevant over time suggests that, despite advancements in technology, it remains a fundamental component of vocabulary acquisition. A growing understanding of the necessity of acquiring comprehensive digital literacy abilities in addition to vocabulary acquisition is indicated by the recent rise of media literacies (#10).
Discussion
This bibliometric analysis provides a structural map of the TEVL field, revealing not only its thematic evolution but also the underlying tensions that shape its trajectory. This section interprets two critical findings that represent the core contribution of this study: the theory-practice paradox and the innovation-inclusion gap.
The Theory-Practice Paradox: A Field Guided by an Invisible Hand
The second research question focused on the field’s intellectual structure and evolution. The analysis reveals a clear trajectory from early computer-assisted tools to a frontier of interactive and intelligent technologies like AI, AR, and social media (Figure 11). This evolution is not merely a technological trend; it reflects a deep, though often implicit, alignment with the principles of Vygotsky’s sociocultural theory.
Vygotsky’s (1978) theory posits that cognitive development originates from social interaction within the Zone of Proximal Development (ZPD), facilitated by a More Knowledgeable Other (MKO). The thematic map provides large-scale, data-driven evidence for this theoretical shift. The emergence of “social media” (#9) and “gamification” (#5) highlights a move toward digital ZPDs where learners collaborate with peers (acting as MKOs). The recent surge in “artificial intelligence” (#8) can be interpreted as the development of scalable, personalized MKOs that provide tailored scaffolding. The co-citation analysis (Figure 5), which places Vygotsky (1978) at the theoretical heart of the network, further confirms this connection.
However, this alignment reveals a profound paradox. While the field’s evolution is implicitly guided by a sociocultural “invisible hand,” the analysis, confirming findings from previous qualitative reviews (Yang et al., 2021), shows that explicit theoretical engagement in individual studies is weak. I argue that this theory-practice paradox is sustained by the field’s dominant methodological habits. The prevalent use of short-term, small-sample intervention designs is pragmatically convenient but inherently ill-suited to capture the longitudinal, social, and cultural dynamics central to Vygotskian theory. Consequently, the field’s methodological preferences may be inadvertently stunting its theoretical deepening, causing many studies to remain at the surface level of technological application without exploring the underlying social learning mechanisms.
The Innovation-Inclusion Gap: Is TEVL Reinforcing Global Inequities?
The first research question examined the key characteristics of TEVL research, including its contributors. The results uncover a second critical tension: a widening innovation-inclusion gap. The research frontier is clearly pushing toward sophisticated, resource-intensive technologies like AI and AR. However, the geographic and co-authorship analysis (Figure 9, Table 3) shows this innovation is overwhelmingly concentrated in two well-resourced hubs: the United States and China, and focused predominantly on English as the target language.
This bipolar concentration, while demonstrating dynamism, raises critical questions about the global diversity and equity of TEVL research. The dominant narratives, research questions, and technological solutions are likely shaped by the educational ecosystems of these two superpowers. This creates an innovation-inclusion gap, where the most advanced pedagogical solutions may be designed by and for a privileged subset of the world’s learners.
The rise of research from countries like Turkey and Iran signals a positive trend toward localization. However, the stark absence of research from the Global South (e.g., Africa, South America) in the dataset is a sobering reminder. The field’s technological progress, if unchecked, may inadvertently reinforce rather than dismantle existing educational inequities. This finding, made possible by the large-scale bibliometric approach, presents an ethical and scholarly imperative for the TEVL community to address.
Synthesizing the Field’s Trajectory and Methodological Profile
Viewed through the lenses of these two tensions, other findings gain deeper meaning. The prominence of review articles and meta-analyses among the most cited documents (Figure 4; Golonka et al., 2014; Takacs et al., 2015) suggests a field actively engaged in self-reflection. However, it also reflects a response to the fragmentation caused by the theory-practice paradox, as scholars attempt to synthesize findings from atheoretical studies. Similarly, the concentration of research in high-impact journals like
Future Directions
Based on the trends and gaps identified in this bibliometric analysis, future TEVL research should deepen and expand in the following areas to advance the field toward more personalized, interactive, and theoretically-driven approaches.
First, resolve the theory-practice paradox. To bridge this gap, future research must prioritize longitudinal and ecologically valid designs. Instead of asking if an AI tutor works in a one-hour session, studies should investigate how it mediates learning within a classroom ecosystem over a semester. This requires embracing qualitative and mixed-methods approaches to capture the rich social dynamics that the co-citation analysis shows are theoretically central to the field.
Second, close the innovation-inclusion gap. Addressing this is an ethical and scholarly imperative. Future research must consciously expand its scope to Less Commonly Taught Languages (LCTLs) and low-resource contexts. This is not just about geographic diversity; it’s about asking new questions. How can TEVL be designed for offline use? How can technology support vocabulary learning in multilingual, under-resourced classrooms? Answering these questions is crucial for the field’s global relevance.
Third, integrate technologies with pedagogical purpose. The analysis shows clusters for mobile learning, gamification, and AR. The next step is to move from studying these in isolation to integrating them with clear pedagogical goals derived from theory. For example, future projects could explore how AR (providing authentic context) can be combined with gamified social tasks (motivating peer interaction as MKOs) on mobile devices (enabling ubiquitous learning), explicitly testing a Vygotskian framework.
Conclusion
This study has mapped the knowledge landscape of Technology-Enhanced Vocabulary Learning (TEVL) from 2014 to 2024 through a bibliometric analysis of 3,307 documents. The core contribution of this research is that it moves beyond traditional effectiveness assessments (meta-analyses) and theoretical collations (systematic reviews) to provide a macro-level, structural perspective on how the field is organized. The findings reveal that the TEVL field has undergone significant growth and evolution, progressing from traditional computer-assisted tools to a complex ecosystem characterized by mobile, gamified, social, and intelligent technologies. This thematic evolution aligns conceptually with Vygotsky’s sociocultural theory of learning, even though the application of theoretical frameworks in research practice remains underdeveloped. Furthermore, the study uncovers a landscape of both concentration and expansion in global research patterns, alongside systematic limitations in existing research regarding methodology (e.g., short-term, small-sample studies) and scope (e.g., an overemphasis on English). By interpreting these findings within a theoretical framework and critically reflecting on its own methodological limitations, this study not only provides scholars and practitioners with a comprehensive “topographical map” of the TEVL field but also charts a course for future research: one that calls for stronger theoretical grounding, more diverse methodologies, broader linguistic and cultural contexts, and a deeper commitment to bridging the gap between research and practice.
This study’s findings should be interpreted in light of several methodological limitations that also point toward directions for future research. The exclusive reliance on the Web of Science database, while justified for its data quality, introduces a known coverage bias. This decision inevitably underrepresents literature from other significant databases like Scopus, as well as social sciences, humanities, non-journal literature (e.g., books, conference proceedings), and particularly non-Anglophone scholarship, which is a key constraint for a global field like TEVL. Furthermore, the results are shaped by the specific algorithms of the analytical tools; VOSviewer excels at intuitive network mapping but is weaker in time-series analysis, whereas CiteSpace’s strength in identifying emerging trends may overemphasize new topics over foundational research. Finally, the analysis is affected by a natural citation lag, meaning the impact of recent research from 2023 to 2024 is likely underestimated due to incomplete publication and citation data.
Footnotes
ORCID iD
