Trends and Patterns in Distance Education (2014–2019): A Synthesis of Scholarly Publications and a Visualization of the Intellectual Landscape

The field of distance education (DE) is dynamic and constantly evolving; it reflects and adapts according to changes in socio-cultural, demographic, political, and technological domains. Thus, there is a need to understand past and present activities in the field, in order to better inform future research. The main purpose of this study was to examine DE research through data mining and analytics approaches, using social network analysis (SNA) and text mining to conduct a bibliographic analysis. The findings highlighted three main strands of DE research: (a) issues related to open education; (b) the design, support, and quality assurance of online DE; and (c) the implementation and use of educational technology, media, and digital tools. SNA of the bibliometric data identified pivotal theoretical contributions, including that the fields of distance education and educational technology converge. The article concludes with recommendations for future research directions.


Introduction
The field of distance education (DE) has long been dedicated to providing flexible ways for a wide diversity of learners to access knowledge and learning opportunities. In particular, for generations the field has benefited from the use of educational technologies (EdTech) to provide a variety of educational opportunities to learners beyond traditional paper-based study materials (Bozkurt, 2019;Moore & Kearsley, 2012;Zawacki-Richter & Naidu, 2016). Over time, DE has adapted and developed DE-specific curricula, educational strategies, and learning models that meet the needs of learners, by taking advantage of increased capacity and affordances of prevalent technologies. Following the advent of computers, online technologies, and online networks, as well as socio-cultural, demographic, political, and technological shifts all over the world, DE has metamorphosed as online learning and emerged as a mainstream educational model, rather than being peripheral or supplementary. Years of experience in the theory and practice of DE have provided a clear pathway for the transformation of higher education due to changing characteristics of teaching, learning, and educational institutions. DE proved its value during the interruption of education due to the COVID-19 pandemic; further, the field of DE has been a significant change agent in the digital transformation of higher education.
In the digital age, affected by the globalized knowledge-intensive economy, 21st-century learners demand ubiquitous, personalized, flexible, and socially collaborative educational experiences; DE has striven to meet these aims by merging theory and practice, and implementing educational technologies where appropriate and relevant. The dynamic nature of DE requires a broad, multidimensional understanding of the past and present evolution of the field, not only to understand past trends but also to anticipate future needs and opportunities. It is vital to gain deeper insights into the constantly changing characteristics of learners, learning, educational technology, and educational institutions. In order to embrace change, enable smooth transitions during the phases of transformation, and more importantly, to be resilient, agile, and adaptive, we need to know how the field of DE has framed and processed scholarly knowledge. This research study aimed to map and visualize the intellectual landscape of the DE field and synthesize the findings of scholarly publications by analyzing scientific networks.

Literature Review
Scholarship is based on the notion that to better predict the future of a field of study and navigate through it, one must first look at its past. In this way we can gain insights by tracking the growth and progress of a field in terms of scholarly publications and reflect on what makes these insights valuable. The following section describes some earlier works that sought to analyze the field and provide a base for future directions by distilling past and present aspects of DE research. Zawacki-Richter and Naidu (2016) suggested that "online educational technologies will play a critical role in education broadly, and especially in relation to an education for all agenda" (p. 264). Similarly, different authors have claimed that EdTech (Sim, 2017) and instructional design for online learning have triggered the advancement of DE (Çakiroğlu et al., 2019) and moved online distance education from the periphery to the centre of higher education (HE) offerings (Xiao, 2018). Naidu (2019) further emphasized that it was not the mere contribution of technology, but learning methods, approaches, and strategies that were transforming distance education. Providing a base and justification for this area of research, Xiao (2018) Berge and Mrozowski (2001) reported that from 1990 to 1999, common concerns in DE publications were pedagogical topics such as "design issues, learner characteristics, and strategies for active learning and increased interactivity" (p. 17).
The same authors also noted the importance of identifying research gaps and providing a research agenda in the DE field, which provided a rationale for the studies that followed. Lee et al. (2004)  Change is continuous, so it is necessary to examine trends and patterns, and build upon previous research.
Thus, structural gaps and weak points in the area can be identified and buttressed. Such an approach is vital for improving the field on solid ground and navigating through uncertainty. Against this background, the main purpose of this research study was to explore and examine research trends and patterns in the field of DE between 2014 and 2019. This study sought to answer the following research questions:

Methodology Research Method and Design
This study applied data mining and analytic approaches (Fayyad, et al., 2002) such as text mining (Feldman & Sanger, 2007) and social network analysis (SNA; Hansen et al., 2010;Scott, 2017) to explore publication patterns and thematic trends in DE research between 2014 and 2019. We used text mining to analyze the Gephi and NodeXL were used to perform SNA, and Leximancer to conduct text mining. The purpose of using multiple analytic approaches is to triangulate the data to increase the reliability and validity of the research findings (Thurmond, 2001), and to explore different layers of the research corpus to reach a broader view in addressing the research questions.

Research Corpus and Inclusion Criteria
The journals for inclusion in the research corpus were selected based on the following three criteria: they

Analysis Process
After building the research corpus, four separate analyses were conducted. First, titles and abstracts of the selected articles were analyzed using t-distributed stochastic neighbor embedding (t-SNE) to visualize "high-dimensional data by giving each datapoint [words in titles and abstracts] a location in a two [sic] or three-dimensional map" (van der Maaten & Hinton, 2008, p. 2579). Second, social network analysis (Scott, 2017) of the keywords was performed to better identify thematic clusters and significant nodes with strategic positions in the keyword network. In this analysis, each keyword identified by the authors of the article was considered a node and their co-occurrences considered a relationship. Third, lexical analysis of the titles and abstracts was performed to visualize a thematic concept map and to identify major themes emerging from the research corpus. Lexical analysis "employs two stages of co-occurrence information extraction-semantic and relational-using a different algorithm for each stage" (Smith & Humphreys, 2006, p. 262). In the fourth and final stage, social network analysis was used to examine the references and citation patterns in the articles in the research corpus. Accordingly, "citing articles and cited articles are linked to each other through invisible ties, and they collaboratively and collectively build an intellectual community that can be referred to as a living network, structure, or an ecology" (Bozkurt, 2019, p. 498). Chen (2006) further argued that pivotal contributions in the scholarly landscape can be identified through their ability to answer critical and needed questions; their strategic position can be estimated in terms of impact which, in turn, can be measured by means of bibliometric metrics. In this regard, the final stage of analysis examined scholarly progress in the DE field by identifying pivotal contributions and turning points.

Strengths and Limitations
The strength of this study was in its methodology and scope. First, the research built a representative research corpus by including full data from six sampled DE journals. Second, the research used computer- and thus this study provided only a partial view. Second, the findings were based on the data extracted from six double-blind, peer reviewed journals and limited to publications written in English.

The Thematic Structure of DE Research
To address the first research question, three types of complementary data analysis explored the thematic structure and research topics in the 1,362 articles that made up the bibliographic corpus of this study. The three analyses-t-SNE analysis of titles and abstracts to identify major research topics, SNA of keywords, and content and co-word analysis-are described in detail below.

t-SNE Analysis
In order to provide a first rough description of the focal points in DE research, t-SNE analysis was conducted based on titles and abstracts of the articles in the research corpus. t-SNE is an unsupervised "nonlinear dimensionality reduction technique that aims to preserve the local structure of data" (van der Maaten & Hinton, 2008, p. 2580), used for exploring and visualizing high dimensional data. The data set derived from the 1,362 articles produced 246,950 total words in 10,730 unique, singular word forms. From among these, the top 100 words with raw frequencies were included in the t-SNE analysis. The settings for t-SNE were adjusted as follows: cluster value: 3; dimension value: 2; and perplexity value (probability distribution): 75.  The majority of research was undertaken in a higher education context (i.e., university in the green, and higher education, academic in the pink cluster), in contrast to DE and technology-enhanced learning in school settings (i.e., school in the blue cluster). This result is in line with findings by Zawacki-Richter and Naidu (2016), who carried out a content analysis of the journal Distance Education over 35 years. They also identified students and learning as the two major topics in 515 articles that were connected via the thematic region of interaction. "Learning is seen in these articles as a social process that is facilitated by interaction Quality in open and distance learning is another important concern of researchers in the field. Student support systems, high-quality content, and interaction are important factors related to quality assurance and development (i.e., quality, support, content, and interaction in the pink cluster). Prerequisites for highquality online learning and teaching include institutional strategies and policies for digital education, support, and resources to overcome the challenges of innovation and change. Professional training for faculty members and teachers is critical for them to develop the necessary skills to design and facilitate meaningful online learning experiences (i.e., institutions, strategies, challenges, professional, skills, and faculty in the blue cluster).

Social Network Analysis of Keywords
SNA provides insightful solutions to map, summarize, and visualize networks as well as identify key nodes that occupy strategic positions in these networks (Hansen et al., 2010). Keywords are representations of the articles at the granular level, highlighting their thematic scope. Of the 3,383 keywords in the 1,362 articles, those with a minimum occurrence of three were included in the analysis-a total of 282 keywords (see Appendix C for the top 40 keywords). Essentially, SNA examines nodes and their relationships-in this study, these were the keywords and their co-occurrence. The analysis revealed eight major clusters concerned with (a) delivery modes, (b) learning environments and contexts, (c) theories and concepts, (d) technology and media, (e) variables, (f) target groups, (g) research methods, and (h) miscellaneous territorial keywords. Figure 2 illustrates the SNA results; each cluster is discussed in more detail following Cluster two is related to DE settings, contexts and learning environments. The findings confirmed that the publications focussed mainly on the higher education context, with less emphasis on K-12 school settings.
Articles in this area dealt with learning environments that can be formal (e.g., an institutional learning management system) and informal (e.g., social networks like Twitter, Facebook, or other Web 2.0 tools).
Again, the concept of openness played a prominent role, with open access applications and open textbooks seen as important tools and resources for the design of learning environments and networks.

26
The third cluster provides an overview of the theories and concepts applied in DE studies. Perhaps the most influential theoretical framework for online learning has been that of Garrison at al. (2000), the community of inquiry (CoI) model with its core elements of social, teaching, and cognitive presences-all these were among the top keywords in this area. As distance education promotes independent and autonomous learning, self-directed and self-regulated learning theories formed the foundation of many articles. On the other hand, many researchers were concerned with questions about how distance learning can be made more interactive, and how to facilitate communication with students and teachers in order to build a sense of community. These ideas resonated with Lave and Wenger's (1991) community of practice or Moore's (1989) theory of transactional distance. Since distance education is facilitated by educational media and tools, technology acceptance model (TAM) was important concepts to measure their acceptance and usage by students and teachers.
The keywords in cluster four confirmed that DE was closely linked to educational technology, especially two-way media as a constituent element, as evident in Keegan's (1980) definition of DE. Learning and teaching in DE have been supported by various synchronous and asynchronous media and tools; in particular, video-based formats have gained importance in the years that were analyzed.
Cluster five includes a wide range of input or output variables that studies have measured to evaluate and investigate learning designs (e.g., students' attitudes and perceptions, their engagement and satisfaction, drop-out rates, or learning styles and strategies). Many of these keywords were related to issues dealing with student support and service quality, as well as students' skills and individual traits. The need for teacher training was highlighted, particularly in terms of providing academic support and guidance to students.
Cluster six includes keywords related to target groups within DE, mainly adult learners, but also high school students. However, the majority of keywords in this area showed again that research focused on the higher education context (e.g., university student, graduate students, academics) as opposed to the school context.
Whilst research into distance education in its early stages was criticized as being "atheoretical and predominantly descriptive" (Perraton, 2000, p. 1), the keywords in cluster three and cluster seven indicated that current research methods have applied a wide range of theories, as well as qualitative and quantitative research methods. SNA and learning analytics, as well as design-based methods, occurred as prominent research approaches over the range of years included in this study.
The eighth and last cluster is related to miscellaneous territorial keywords. What was prominent here was the mission of the DE field to widen access to educational opportunities, especially in developing countries.
These efforts were supported, for example, by the Commonwealth of Learning in Canada, and the open education initiative of the Khan Academy.
The findings of the SNA of keywords resonated with the t-SNE analysis of titles and abstracts, provided more details of the plethora of research topics under investigation. We now proceed to discuss the third method of analysis-a text-mining approach for content and co-word analysis.

Content and Co-Word Analysis
Words gain their meanings in their context, and the implicit lexical relationship among words within texts can be revealed and visualized explicitly through text mining. For the purpose of the study, the titles and abstracts of 1,362 articles were analyzed and visualized through text mining using the Leximancer software tool. The software drew a concept map that revealed lexical patterns of concepts, grouped through relational co-word analysis in thematic regions (Figure 3). "The map is an indicative visualization that presents concept frequency (brightness), total concept connectedness (hierarchical order of appearance), direct inter-concept relative co-occurrence frequency (ray intensity), and total (direct and indirect) inter-concept co-occurrence (proximity)" (Smith & Humphreys, 2006, p. 264). Depending on the connectedness of concepts, a thematic region was formed (the colored bubbles) and named by the most frequent concept within this region.   Dalsgaard & Thestrup, 2015). However, challenges and quality issues were addressed by researchers in this thematic area-such as discoverability, sustainability, and remixing of learning materials (Luo et al., 2019); it has been further critically argued that we should first "remove the barriers from within" (Mishra, 2017, p. 378). Some authors argued that open educational practice (OEP) paves the way for OER, and openness should not be limited to OER issues only (Koseoglu et al., 2020). Similar to discussions on OER and OEP, MOOCs, for instance, emerged as an extension of openness in education. However, they have been highly criticized because of high dropout rates, motivational issues, and low instructional quality. Lambert (2020) argued that MOOCs and other practices rooted in openness in education should be praised for contributing to student equity, social inclusion, and widening participation. In fact, when we revisited the core values of DE, it was not quantified measures that defined its success, but rather quality and the degree of openness, flexibility, and learning opportunities through DE available for those who would not otherwise have access to education.  Figure 3, see concept paths social-activity-designknowledge, and interaction-social-environment-experience-online). Social interaction is a key factor in supporting learners and learning processes (Fisher & Baird, 2005), and changes in online networks require the application of innovative technological and pedagogical approaches to foster learning communities (Conley et al., 2017;O'Connell, 2016). For instance, Robinson et al. (2020) argued that maintenance of a climate of care was significant for online learning which highlighted a need to revisit emotional and affective domains in online learning environments.

Pivotal Contributions and Turning Points
To investigate the second research question regarding pivotal contributions to the field of DE and theoretical and technological shifts, an SNA of the references in the selected articles was conducted to reveal the network of citing and cited articles as a scholarly communication system (Garfield, 1972). In this sense, a total of 43,318 valid processed distinct references (99.93%; captured by the software) and 27 invalid unprocessed distinct references (0.07%; illegible by the software) were collected from the 1,362 articles and examined through SNA. In this analysis, cited articles were the nodes in the network, while the citations (by one article to another) were considered as the ties between them.

Theoretical and Conceptual Growth of DE
Timeline visualization enabled us to identify chronological meta-narratives by linking references and acknowledging their pivotal contributions (Figure 4). Referring to generic educational theories, it was Bandura (1977) and Vygotsky (1978) who proposed early explanations of social learning processes. Short et al.'s (1976) study on the social psychology of telecommunications was a significant pivotal contribution and provided a solid base from which to investigate DE communication processes. By the 1990s, when computer-mediated communication (CMC) came to the fore, DE-related theoretical or conceptual studies drew more attention. For instance, Lave and Wenger (1991), who introduced the community of practice concept, and Moore (1989) who proposed three types of interaction in DE, appeared to be early pivotal contributions in DE research (see Figure 4 and Figure 5).

Snapshot of Timeline Visualization From 1970s to 1990s
However, in terms of the overall timeline analysis, the biggest impact has been that of Garrison et al. (2000) who introduced the community of inquiry model and its components, namely social, teaching, and cognitive presences. Another study by the same authors (Garrison et al., 2001) maintained the legacy of the CoI. In terms of theoretical contributions, Siemens' (2005) work on connectivism was another milestone, as it indicated a shift in attention to networked learning. Following these developments and capacity increase in delivery modes, MOOCs emerged as an evolved form of DE and online learning (see Daniel, 2012;Ho et al., 2014;Kizilcec et al., 2013;Liyanagunawardena et al., 2013;Yuan & Powell, 2013 Figure 4 and Figure 6). Snapshots of Timeline Visualization From 2000 to 2020

Theoretical and Conceptual Growth of Educational Technology (EdTech)
In parallel with the development of DE, we observed pivotal contributions related to EdTech. A pioneering theoretical contribution was by Davis (1989) who introduced the TAM, which was followed by pioneering theoretical contribution by Venkatesh et al. (2003). After the first decade of the 2000s, the focal point was online learning and online learners (Allen & Seaman, 2013;Braun, 2008), and e-learning became part of mainstream education (Garrison, 2011; see Figure 4 and Figure 7).

Snapshots of Timeline Visualization From 1990s to 2010s
To summarize, it can be argued that DE evolved by incorporating social learning theories and, starting in the 1990s, the field has focused on CMC (e.g., interaction and presence types, as well as online community formation). This focus was followed by the emergence of ubiquitous learning opportunities offered by

Conclusion and Recommendations
This study provides orientation and new insights into the scholarly structure of DE as an academic discipline. Our findings demonstrate that research in the period from 2014 to 2019 published in leading DE journals focused on three broad areas: • Openness and open education extended the raison d'être of many distance teaching and open universities (see Tait, 2008), namely to provide access to learning opportunities for those who were previously excluded or unable to participate in educational programs and services.
• Learning design, academic and institutional support services, and quality assurance were major concerns of researchers in DE. Student support was the critical link in DE (Dillon et al., 1992, p. 29).
• The application and use of educational technology, media and tools provided a means to develop learning content in various presentation modes, and to facilitate synchronous and asynchronous interaction between students and teachers, as well as among students, in order to support collaborative learning.
Our findings show that published DE research has been clearly dominated by studies in the higher education context. Moving further into the mainstream of education-especially in times such as the global COVID-19 pandemic-we recommend that DE researchers should also turn their attention and transfer knowledge and experience to the K-12 school context. This would support teachers and parents, and prevent them having to re-invent the wheel when adopting remote teaching and homeschooling practices.
Furthermore, the results of this study suggest there is very little research that considers learning with digital media in non-formal settings (see Latchem, 2018).
The field of online DE research has been developing dynamically in the era of digital transformation. With the need for emergency remote education (Bozkurt & Sharma, 2020a;Hodges et al., 2020) in the current COVID-19 pandemic, the roots of online learning and teaching (namely theory, practice, and empirical research in the broader field of DE) have gained added relevance. It is crucial to build on and relate to this theory, practical experiences, and knowledge, especially for researchers and practitioners who venture into the field of online distance education.
DE research is well-grounded in foundational theories such as adult learning theory, self-directed learning theory, as well as emerging theories. The discipline of DE (most prominently the CoI framework and the theory of connectivism) has not only spawned new theories, but has also borrowed theories from related fields, such as media theory, instructional design theory, and motivation theory (see an overview in Jung,

33
For generations, DE has skillfully served to balance inequity, address the knowledge gap, and democratize education. However, we are on the verge of a major transformation due to the COVID-19 pandemic, which has been called the great reset, and there is now a new/next normal and a new order all around the world (Bozkurt & Sharma, 2020b). In addition to the effects of the pandemic, the world is being decentralized, learners are coming from increasingly diverse backgrounds, learning is emerging as a liminal process, learning sources are being distributed across online and offline networks, and-most importantly-the nature of the learning ecology is evolving. We, therefore, need to redefine our roles, take on new responsibilities, and embrace them. As distance educators, we should ask ourselves how we envision the future and how well we are prepared for tomorrow's learning landscape. In this regard, it is more critical than ever to ask what our research agenda will be in the changing world. This study suggests that the role of DE is more significant than ever and there is a need to develop new policies and strategies; it is critical to identify a proactive research agenda to better respond to changes, to be resilient, and to meet ongoing and sometimes sudden challenges in terms of educational needs.