JOURNAL OF LANGUAGE AND LINGUISTIC STUDIES Journal of Language and Linguistic Studies: A fifteen-year bibliometric quest for a bigger impact

To assess scientific progress from global to author levels, a large body of bibliometric studies could be found in many fields but relatively scarcer in the realm of language and linguistic studies, especially at a journal level. Motivated by the commitment and competence shown by the Journal of Language and Linguistic Studies, this study presents its bibliometric portrait as captured by two pieces of software, Publish or Perish and VoSviewer. from the Microsoft Academic, one of the global abstract and citation databases. The descriptive and network analyses of the journal’s bibliometric data revealed the patterns of publication, citation, authorship, keyword co-occurrence, and bibliographic coupling from 2005 to 2019. The first 15 publication years have seen 482 articles by 552 authors from 142 organizations of which Turkish ones were in the majority contributing to 1,291 total citations along with regional and global indexations including in Scopus, one of the largest bibliographic databases. Evolving around the unity of psychology, linguistics, and education/ pedagogy viewed from different perspectives, the journal has been a scholarly outlet for not only local but also global issues in the realm of language and linguistics. The next 15-publication years will see the journal’s growth in “internationality” in terms of authorship due to the combination of increased annual publication and recent international indexation. Perish


Introduction
Since 2005 the Journal of Language and Linguistic Studies (JLLS) has professionally published peerreviewed empirical and non-empirical articles covering the big enterprise of language and linguistics including the field of language teaching. Recent years have seen the growing recognition of the journal's scientific contribution to the fields. In 2016 JLLS was included in the European Reference Index for the Humanities and the Social Sciences (NSD -Norwegian Centre for Research Data, 2019) as the regional recognition of its scientific presence in the two subject areas. In 2017, as one of the Diamond Open Regarding the indexation, in 2019 JLLS was also indexed in Scopus, one of the major academic search engines and bibliographic databases (Baker, Kumar, & Pattnaik, 2020;Gusenbauer & Haddaway, 2020). The inclusion in Scopus is, as Erfanmanesh, Tahira, and Abrizah (2017) highlight, often viewed as a very significant achievement of not only the included journal and its publisher but also the country of publisher. For the journal and its publisher, the inclusion represents high editorial and managerial qualities. For the country, it could improve the national scientific quantity and quality. The inclusion could lead to a higher submission rate because, as Kwiek (2020) noted, publishing in the Scopus-indexed journals could help authors' chances of promotion, position, and funding.
Motivated by the JLLS's continued 15-year publication along with the recent regional and global recognition, this study exhibits a bibliometric portrait of JLLS from 2005 to 2019. Using bibliometrics as the bedrock, this study could provide a retrospective of its publication and citation pattern during the time span. The analyses of scientific quantity (publication) and quality (citation) might help reveal how well JLLS has contributed to the fields by involving "scholars not only from Turkey, but also from all international academic and professional community" in presenting "different theoretical and thematic approaches to linguistics and language teaching" (JLLS, 2020a). The next part of the Introduction concerns bibliometrics in general and a little body of literature about it in the subject area of language and linguistics.

Literature review
This study is methodologically founded on bibliometrics. The term "bibliométrie" was coined in 1934 by Paul Otlet in "Traité de Documentation" or Treatise on Documentation (Rousseau, 2014). It was then anglicized by Pritchard (1969, p. 349) who defines bibliometrics as "the application of mathematics and statistical methods to books and other media of communication." However, long before it was coined, some bibliometric studies to describe the scientific progress of a certain field had been conducted by keeping a tally of the number of scholars and their work (Lei & Liu, 2019a). One of the earliest bibliometric attempts was made by Cattell (1906) who analyzed a total of 1,000 American scientists and how they were distributed in terms of disciplines, organization, and regions. In 1964 bibliometrics was revolutionized by Eugene Garfield who introduced Journal Impact Factor in his Science Citation Index (Garfield, 2007). Since then bibliometrics has evolved around the trinity of publication, indexation, and citation (Nylander, Österlund, & Fejes, 2020).
So far bibliometrics has been colored by a wide range of metrics from calculating publication or citation to combining the trinity of publication, citation, and indexation by using different algorithms. Harzing (2011), for example, lists eight simple metrics such as total number of publication and, at least, six complex metrics such as the h-index (Hirsch, 2005). Two major journal-specific metrics introduced by two key players in the business of indexing service and citation database are Web of Science's Journal Impact Factors and Scopus' CiteScore which are principally on the same basis of citation count (Fernandez-Llimos, 2018;Roldan-Valadez, Salazar-Ruiz, Ibarra-Contreras, & Rios, 2019). The assessment of scientific contribution around the publication and citation also specifically applies to institutions and countries as listed in the Scopus-based SCImago Institutions Ranking (SCImago, n.d.a) along with SCImago Journal and Country Rank (SCImago, n.d.b).
Much work on bibliometric analyses in many fields has mined the bibliographic data from different academic search engines and bibliographic databases. Much work on the comparison of different bibliographic databases such as Google Scholar/ GS (https://scholar.google.com/), Microsoft Academic/ MA (https://academic.microsoft.com/home), Web of Science's databases including its Core Collections (http:// webofknowledge.com/), and Scopus (https://www.scopus.com/) has also been carried out, e.g. Harzing (2019), Hug, Ochsner, andBrändle (2017), andThelwall (2017). More recent comparative research was more comprehensively carried out by Gusenbauer and Haddaway (2020) who compared 28 academic search engines and bibliographic databases from open to pay walled access.
Furthermore, bibliometric researchers have been armed with a lot of bibliometric software. A piece of free software, Publish or Perish/ PoP (https://harzing.com/resources/publish-or-perish; Harzing, 2007), for instance, could be used to retrieve bibliographic data from seven academic search systems including the aforementioned ones and to describe the patterns of publication and citation in terms of 27 indicators at the levels of field, topic, journal, institution, and author (Harzing, 2011). The bibliographic networks such as co-citation could also easily be mapped by adopting software such as CiteSpace (http://cluster.cis.drexel.edu/~cchen/citespace/; Chen, 2017), and VOSviewer (http://www.VOSviewer.com/; . The software is a dream come true as it will help to increase efficiency and effectiveness when conducting bibliometric analyses. For at least 12 decades bibliometric indicators such as citation have been used to assess research progress. Because of the growing significance of bibliometrics for the purpose of research assessment twinned with the fast developments of information and computing technologies, bibliometrics have been widely adopted in various disciplines such as finance, advertising, and engineering (Martínez-López, Merigó, Valenzuela-Fernández, & Nicolás, 2018). Despite this, very little has been published on the bibliometric overview of the language and linguistics including the multidisciplinary field of language teaching. As a matter of fact, Eugene Garfield whose Science Citation Index has escalated bibliometrics to a new level is a Ph.D. holder in Structural Linguistics (Masic, 2017). To make it worse, even though the few bibliometric studies in the fields have been done at the levels of topic, region, and country, much fewer could be found at the journal level.
Despite few examples of bibliometric attempts in the fields related to the focus and scope of JLLS, they could offer a wide range of contexts from global to journal levels. At the global level, Zhang (2020) analysed the major trends in the domain of Second Language Acquisition (SLA) from 1997 to 2018. The Web of Science was used to retrieve the data of (co)citation and keywords of almost 8,000 articles in 16 top journals such as Modern Language Journal and Applied Linguistics. The bibliographic data including 7,866 titles and abstracts receiving nearly 160,000 citations, authors from 2,041 affiliations in 87 regions, 791 keywords, and more than 180,000 unique references were then submitted to Bibexcel (https://homepage.univie.ac.at/juan.gorraiz/bibexcel/; Persson, Danell, & Schneider, 2009), a citation counting software program, and VOSviewer (http://www.VOSviewer.com/; Eck & Waltman, 2020), a network visualization software. Performing the analyses of citation-based impact, co-citation, and keyword, he could discover the changes and trends with regard to the hottest topics, themes, and theories, along with journals' focus. This work could successfully exhibit the bibliometric portrait of more and more diversified SLA.
One of the most interesting findings in Zhang's (2020) work is how China could emerge as one of the top three countries in the field during the 2007-2018 period. In a bibliometric study at the level of country or region focusing on Mainland China and its three related territories, i.e. Hong Kong, Macao, and Taiwan, the progress had been well explained by Lei and Liao (2017). One of the explanations was that the "publish or perish" system applied to faculty members of some universities especially in Mainland China which envisaged reaching an international status with the focus on journals indexed in the Web of Science's Social Science Citation Index. The system was also well covered supported by the scheme of research and development expenditure. In doing the bibliometric research, Lei and Liao (2017) explored the Web of Science database to extract bibliographic data of around 1,400 articles and book reviews by the Chinese scholars during the 2002-2012 period. Despite a significant annual increase in scientific productivity by the four regions, no significant difference in research impacts could be observed between them. Hong Kong could top the list in terms of not only quantity (total and annual documents) but also quality (impact factors and citations). Even though their analysis was not assisted by a piece of bibliometric software, both of the authors could still vividly capture the scientific contribution to the arena of SLA by the researchers in Hong Kong, Macao, Mainland China, and Taiwan.
At least 41 internationally reputable journals in the realm of language and linguistics were mentioned in the two bibliometric studies. However, only one out of them were lucky enough to be bibliometrically analysed, i.e. System. It was Lei and Liu (2019b) who identified the top discussed topics along with the top cited authors, articles, and references in System from 1973 to 2017. Mining bibliographic data of 1,589 articles including their references from Scopus database, the two authors run TreeTagger (https://www.ims.uni-stuttgart.de/en/research/resources/tools/treetagger/; Schmid, 1995) and AntCon (https://www.laurenceanthony.net/software; Anthony, 2018). The first piece of software was used to annotate the retrieved abstracts with their lemma information and parts of speech whereas the second one was adopted to extract word groups of a maximum of five words from the abstracts previously engineered with the first software. During the 45 years, some topics such as learning strategy classified under the classic teaching learning practice and cultural context classified as newly developed sociocultural and technology-based practice attracted growing attention in the periods of 1970s and 1980s, 1990s and 2000s, along with 2010s. This study is also filled with revealing changes in the field. One of the them is how the theorists and practitioners in the field mainstreaming Communicative Language Teaching and Monitor Theory had made grammar step down from the focus of attention in the first examined period. For over four decades, System, as Lei and Liu (2019b) conclude, has relatively been successful in achieving its mission to solve some problems of foreign language teaching and learning, especially found in the developing countries, through the unity of proper educational technologies and applied linguistics.
Not only the above three bibliometric studies but also a lot of other studies from global to journal levels explored the two major academic search engines and bibliographic databases, i.e. the Web of Science and Scopus. Both of them are, however, pay walled. The subscription-based platforms bar researchers with no institutional subscription access from mining their databases. The pay walled platforms thus set a limit on the "reproducibility (also "replicability," "reliability," and "repeatability") and transparency" (Gusenbauer & Haddaway, 2020, p. 184, emphasis in original) in conducting bibliometric analyses.
Moreover, the two platforms could not be used to mine the full set of bibliographic data of a certain journal beyond the publication years covered in their databases. For example, it would be hard to take a full length bibliometric portrait of JLLS through the lens of Scopus. The coverage years of JLLS in Scopus began in 2019. In fact, how JLLS had grown before it was indexed in Scopus, one of the hallmarks of high quality academic venue, could reveal some revealing changes and trends both in the field and the journal itself.
Examining the (first) 15-year continued commitment, competence, and contribution of JLLS are therefore of great importance. Framed by the mission of JLLS to be "… a platform for different theoretical and thematic approaches to linguistics and language teaching" by publishing articles that could develop theoretical, conceptual, or methodological approaches to language and linguistics, present results of empirical research that advance the understanding of language and linguistics, explore innovative policies and programs and describe and evaluate strategies for future action, and analyse issues of current interest (JLLS, 2020a) this study performs descriptive and network analyses of JLLS between 2005 and 2019.

Research questions
The first analyses consider some bibliometric indicators from unidimensional metrics such as the cites per paper to multidimensional metrics such as the h-index. The descriptive analyses therefore concern one general research question, i.e. how have the publication and citation structures of JLS evolved through time?
The research questions dealing with the network analyses of co-authorship, citation, cooccurrence, and bibliographic coupling are as follows: 1. Who were the most prolific authors in JLLS in the first, second, and third five publication years along with during the 15-year time span? 2. Which institutions have been the most productive ones in JLLS in the first, second, and third five publication years along with during the 15-year time span? 3. How was the pattern of authorship in in JLLS? 4. Which articles and the article types have been the top-cited articles during the 15-year time span? 5. What key topics have been used most frequently in JLLS in the first, second, and third five publication years along with during the 15-year time span? 6. How have the articles published in JLLS been related in terms of the frequency they cited at each other? 7. How have the articles published in JLLS been related in terms of the number of shared references?
The answers to the two clusters of research question could help take better the bibliometric portrait of JLLS. The portrait could provide not only retrospective but also prospective insights into the contribution of JLLS to the realm of language and linguistic studies.

Method
This section is aimed at supplying details of this bibliometric attempt. It describes the bibliographic data, the free database from which the data were extracted along with the two pieces of user-friendly free software adopted and how they were run. Further details about how the selected database and software could ensure the coverage, consistency, replicability, and transparency of this study (Gusenbauer & Haddaway, 2020) is also set out.

Bibliometric Data
A total of 482 bibliometric items of the articles published in JLLS during the 15 publication years was included in this study. Five of the 482 article titles had no abstracts. For the descriptive analysis, the bibliographic data at the journal level were 27 indicators from uni-to multi-dimensional criteria as listed by Harzing (2011). At the article level, as can be seen in in the online supplementary materials, there were at least 18 bibliographic elements such as the number of citation per article, author(s)' name and their affiliations, publication year, page numbers, and cited references.
When the retrieval of data was carried out on July 6, 2020, 489 bibliographic items could be collected. A closer inspection, however, revealed that one of them was the journal cover, Table of Content, and Editorial. The initial data were also cleaned by merging five couples of articles whose titles were in both English and Turkish and retrieved as two different articles. The data cleaning began with paying attention to two articles with English and Turkish titles by the same author(s) published in the same year. The titles were manually checked by the author going onto the journal's website.

Academic Search Engine and Bibliographic Database
The bibliometric data in this study were retrieved from MA, one of the free to access academic search engines and bibliographic databases. MA was chosen as the database for this study after the author had compared the results of data retrieval by MA with those by other five databases, i.e. Crossreff (https://www.crossref.org/), GS (https://scholar.google.com/), Pubmed (https://pubmed.ncbi.nlm.nih.gov/), Scopus (https://www.scopus.com/), and Web of Science (http:// webofknowledge.com/). MA and Scopus searches could be run by providing a free Application Programming Interface key obtained from the Microsoft Research APIs Portal (https://msrapis.portal.azure-api.net/) and the Elsevier Developer Portal (https://dev.elsevier.com/), respectively.
As can be seen in the online supplementary materials of the comparative results by the six search systems, MA suited this study best in terms of the four principles of conducting bibliometric research, i.e. best coverage, highest consistency, maximum replicability, and greatest transparency. For example, Pubmed understandably retrieved no bibliographic data of JLLS because the database was developed specifically for biomedical and life sciences (Sayers et al., 2020). Scopus database could relatively be free to access but the search results could not satisfy this study. For instance, only the first author' name could be extracted from an article by multi authors. In spite of its broader coverage, GS retrieved too many stray data to clean. Zhang (2020) also observed that the GS search resulted in a citation bias because of not differentiating academic and nonacademic citations and the results of GS search did not supply cocitation data needed in this study. Finally, the only access to the Web of Science was through subscription. In the comparison phase, no data could be extracted from the Web of Science. The bibliographic database in this study has thus carefully been selected.

Bibliometric Tools
Two pieces of free software were adopted in this study, i.e. PoP (https://harzing.com/resources/publish-or-perish; Harzing, 2007) and VOSviewer (http://www.VOSviewer.com/; . Both of them were used to extract the bibliographic data as explained in 2.1. The first tool was mainly used to conduct the descriptive analyses of publication and citation structures of JLL during its 15-year lifetime. As its name implies, the second tool was used for visualizing similarities of the bibliographic items. The similarities represent the connectedness of articles published by JLLS in terms of co-authorship, citation, co-occurrence, and bibliographic coupling. In this study, the two tools complemented each other. Combining the two application in one bibliometric study was also carried out by many researchers, e.g. Kwanya (2020).

Data collection and analysis
The bibliometric data of publication and citation structures were retrieved from MA by running PoP. Generally, PoP was operated as described by Harzing (2011). The search queries were filled with "Journal of Language and Linguistic Studies" for "Full journal name" and "1305-578X" for "ISSN". The "Years" query was filled with "0"-"0" (unspecific years), "2005"-"2019" (publishing years), "2005"-"2009" (the first five years), "2010"-"2014" (the second five years), "2015"-"2019" (the third five years). All of the queries were consecutively combined in order to increase the precision, for instance "Years": "2005-2019", "Full journal name": "Journal of Language and Linguistic Studies", and "ISSN": "1305-578X" (Figure 1). The gathered bibliometric data were then cleaned for duplication and error. The second software, VOSviewer, was used to retrieve the bibliometric data from MA and to visualize the bibliographic network of the 482 articles published by JLL between 2005 and 2019. In general, the software was adopted according to  as follows.
Firstly, VOSviewer provided three types of data to choose for creating a bibliometric map, i.e. "network data", "bibliographic data", and "text data". As described in 2.1., the type of data to analyse was "bibliographic data". Secondly, VOSviewer asked which one of data sources to choose. As was stated in 2.2., the bibliometric data were mined from MA through the provided Application Programming Interface key. As shown in Figure 2, two out of seven search queries, i.e. "Journal" and "Year" were filled with "Journal of Language and Linguistic Studies" along with successively " 2005-2009", "2010-2014", "2015-2019", and "2005-2019" to increase the accuracy. The box of "title and abstract" was clicked. Neither the box of "Restrict to primary documents" nor the box of "Restrict to documents with DOI" were checked because, as shown in the online supplementary material, nor some of the bibliographic data were retrieved from the journal's website and a few Digital Object Identifiers of the articles could be found. The data retrieval in VOSviewer was set up as shown in Table 1. The four types of analysis and the objectives was briefly explained by Martínez-López et al. (2018). The full set of text data extracted by VOSviewer can also be found in the online supplementary materials. As with the bibliometric data in 2.1., after being extracted, the data were checked for duplication and error. Because VOSviewer retrieved the data from the same database as PoP did, the search results of PoP could be used for data cleaning.

Results and Discussion
From the database of MA, PoP and VOSviewer retrieved bibliometric data of 482 documents published by JLLS over the 15 publication years. They included 26 metrics as shown in Table 2 and the articles' metadata such as titles and citations provided as part of the online supplementary materials. This combined section is therefore divided into two, i.e. the descriptive and network analyses.

Descriptive Analysis
How JLLS has spread out its two legs of the bibliometric tripod, i.e. publication and citation through time are shown in Figure 3. In terms of publication, a noticeable increase in the number of publications could be observed in 2016. Since then the publication in JLLS has increased more than threefold or by 168% compared to the number of publication in the first and second five publication years combined. Nearly 63% of the total articles was published in the last five publication years. From 18 articles in 2005 to 482 ones in 2019, JLLS demonstrated an over 2578% growth with a compound annual growth rate of 26.47%. The noticeable boost growth could partially be attributed to, like in SLA as noticed by Zhang (2020), the bigger scope, wider diversity, and faster growth of the fields of language and linguistics. The significant growth of publication could also be identified in some leading journals in the relevant fields such as System (Lei & Liu, 2019b). The number of publication in System increased from 80 articles in 2017 to 105 and 109 articles in 2018 and 2019, respectively (System, 2020). In addition, the regional and international recognition given to JLLS during the third period could explain the increasing manuscript submissions which could meet the publication standards of JLLS.
The change of publication number in 2017 and frequency in 2018 (JLLS, 2020b) after over ten publication years could help JLLS win the complex game of scholarly publishing. After being included in reputed indexing services, JLLS is now being engaged in a quest for higher scores of, for instance, CiteScore in Scopus, sometimes considered as an indicator of more significant impact in the fields. Increasing the number of annual publication might lead to a higher impact especially because some of the most widely used bibliometric indicators such as CiteScore, h-index, and Journal Impact Factor are calculated by considering both publication as a quantity/ productivity indicator and citations as a quality/ performance one (Fernandez-Llimos, 2018;Roldan-Valadez et al., 2019). The importance of larger number of publication in achieving higher scores of the Web of Science's Journal Impact Factors and h-index did not escape Harzing's (2011) notice when she was comparing at least eight journals having varied annual publication numbers in the fields of management information systems. Within the next few years, the effect of editorial change in terms of publication number and frequency on the bibliometric indicators could be identified.
As the second leg of the bibliometric tripod, nevertheless, all of citation-related single metrics such as Cites per Year in JLLS decreased over time (Table 2). Nearly 70% of the total citations were received by the documents published between 2005 and 2009. All of the 8 articles receiving over 50 citations were also published in the period. The trend towards a higher citation rate for an article published in previous issues is understandable because the earlier an article is published the more chances it receives citations (Aksnes, Langfeldt, & Wouters, 2019;Tahamtan, Safipour Afshar, & Ahamdzadeh, 2016). To increase the citation rate, JLLS allows authors to publish their manuscripts as preprints (JLLS, 2020a). Besides, its inclusion in one of largest bibliographic databases such as Scopus which may mean better visibility and discoverability could help JLLS to achieve a higher citation rate in the next few years.
Furthermore, the levels of citation concentration and uncitedness in JLLS are worth noting. While the former is the percentage of articles that accounts for a certain percentage of citations in a journal such as 50%, the latter defines the total articles receiving no citation after a certain period such as 10 years (Harzing, 2011). As can be seen in the online supplementary materials, only 17 articles (approximately 3.5%) account for around 50% of the total 1,291 citations presented in Table 2. Moreover, from the first to the fifteenth volumes, JLL has experienced the uncitedness level of 56%, i.e. only 270 of 482 articles were cited at least once. The high concentration and uncitendeness, according to Harzing (2011), is not uncommon in the fields covered by JLLS because the articles in those fields had a longer waiting time for citations than those did in such fields as Biology and Neuroscience. Table 2 compares and summarizes the bibliometric data on the wide range of publication and citation metrics of JLLS. Overall, the h-index of JLLS is 14, i.e. of the 482 articles, 14 articles were cited at least 14 times or more. According to Harzing (2011), another important metric for assessing a journal is the gindex introduced by Egghe (2006)

Network analysis 3.2.1. The most prolific authors and institutions
VOSviewer retrieved 73, 121, and 404 authors for the first, second, and third five publication years, respectively. Overall, 552 authors had their articles published in JLLS during the 15 publication years. Table 3 shows the top ten productive authors sorted by the total document (D) and total link strength (T), a co-authorship link of one author with other authors in JLLS. Based on the order, the only author that could stay in the top ten list for the three periods is Arif Saricoban. The second most productive author, Mehmet Demirezen, also authored the same number of articles but with a less total link strength. He is the only author in the overall list of top ten authors who had no link of co-authorship. In the complete list of authors supplied as part of the online supplementary materials, it could be observed that in the third period, he also authored three document but was ranked only 16 th . All of the top ten authors in the overall list except Bengu Aksu Atac could be traced back to the five-year lists. The co-authorship analysis with the analysis unit of authors is visualized in Figure 4. The top ten authors whose articles make a total of about 10% of publication are clearly given the spots. Interestingly, while nine of the top ten prolific authors have co-authorship links, sole authorship is generally in the majority. Figure 3 shows most of the 552 authors are not connected to each other in JLLS. As regards the co-authorship analysis with the unit analysis of organization where the authors work, 144 institutions, ranked according to the number of publications and co-authorship, took part in publishing 482 articles in JLLS. Except for Islamic Azad University (Iran), all of the top ten prolific institutions in the overall list are Turkish. The number of documents by the authors from the top ten prolific institutions constitutes over 38% of the total publication during the 2005-2019 period ( Table 4). All of the top ten institutions in the overall list could be seen in each of the three five-year lists.  Unlike the author-based visualization, the organization-based one shows the connectedness of institutions are higher than that of authors in JLLS. Figure 5 presents the largest set of connected organizations consists of 41 Turkish universities. For instance, when their articles were submitted to JLLS, the top two productive authors were being affiliated with the Hacettepe University, Turkey.

Authorship pattern
The co-authorship data extracted by PoP and VOSviewer were combined in order to examine the authorship pattern of publication in JLLS. As presented in Figure 6, more than half of the 482 articles were authored by sole authors while 45% of them were authored by more than one authors. Sole authorship is thus in the majority in JLLS. The second and third metrics indicate that sole authorship dominated JLLS throughout the 15-year lifetime. The top ten articles authored by more than one authors ranked according to the number of citations are presented in Table 5. Only four of them received at least one citation. With a total of 28 citations, the h-index of the top ten articles in Table 5 is 3, i.e. of the 10 documents, three articles received three citations or more. Multi authorship in JLLS grew through the study period as shown in the "Author per paper" in Table 2.
The dominance of sole authorship as presented in the Authors per Paper of Table 2 is in accordance with not only Figure 4 and 6 but also Table 5. The findings confirm the dominance of sole authorship in the subject areas of Social Sciences including education along with Art and Humanities including language and linguistics (Harzing, 2011). Additionally, the similar authorship pattern has recently been found by two authors (Lei & Liu, 2019b, 2019a who conducted two bibliometric analyses at the global and journal levels, i.e. the scientific trends in the area of applied linguistics and System as one of the top journal in the same area. That sole authorship might be the norm in the areas of language and linguistics could be attributed in part to the solitary and competitive nature of the two areas complicated with a relative lack of collaborated research agendas (Barrot, 2017). Regarding the aim of JLLS, the authorship could be framed by how "scholars not only from Turkey, but also from all international academic and professional community" presented "different theoretical and thematic approaches to linguistics and language teaching" and three other types of manuscripts (JLLS, 2020a) through the 15 publication years. The most prolific authors and institutions presented in Table 3 and 4 are from and located in Turkey, the publisher country of JLLS. Similarly, the complete list of authors and organization provided as the online supplementary materials proves the local authorship, both individually and institutionally. The question is then on whether JLLS failed in its attempt to attract international authorship it has envisaged since its first issue despite its 17 out of 21 editorial board members consisting of international scholars.
In reality, international authorship in most of scholarly journals is not easy to grow. Internationality in terms of the authors' diverse geographic distribution often takes too long. Even after being indexed in the Web of Science as one indicators of internationality, over 50% of 929 journals took four to six years to become "more international" (Gazni & Ghaseminik, 2016, p. 104). Therefore, it is understandable that recent inclusions in the regional and international bibliographic databases have not pushed up the international authorship in JLLS yet. Within the next few years, the combination of increased annual publication and international indexation will make JLLS "more international" in terms of authorship.

Top cited articles
From the database of MA, PoP and VOSviewer retrieved bibliographic data of 482 articles of which 44% (122 articles) received at least one citation. Table 6 lists the top ten cited articles of which four articles are empirical and two articles were authored by multi authors. Only one of the top ten cited articles was not published in the first publication period. Interestingly, the number of citations by the top ten cited articles, i.e. 566 citations, comprises 44% of the entire citations. Finally, the h -index of the top ten cited articles is 10, i.e. the ten documents received ten citations or more. All of 482 articles published in JLLS during the 15 publication years could be categorized into two groups, i.e. empirical and non-empirical ones. The second top cited paper in Table 6 was authored by Kırkgöz (2005) who used a questionnaire to identify what students' motivation was for joining English medium departments in a Turkish public university. She got the answers to her research questions from observations or experiences, in this case through a questionnaire. Such an article belongs to the former group. On the other hand, in authoring the first top cited article, based on his extensive reading a large body of literature, Hişmanoğlu (2005) underlines the importance of literature in teaching English as a foreign language. Such scholarly "papers without data" (Gilson & Goldberg, 2015) belong to the latter group including theoretical, conceptual, and review writing.
The results that six out of the top ten cited articles in Table 6 belong to the non-empirical articles fit fairly well with Lei andLiu (2019a, 2019b). One of the reasonable explanations for the more citations received by non-empirical articles may be that, drawing on a greater body of previous studies, such articles as review papers provide researchers with richer information and knowledge about a defined topic than the empirical ones do (Tahamtan et al., 2016). Another possible reason is that the views are so everlasting that they could attract more widespread interest than the time-bound empirical ones. Test anxiety as satisfactorily reviewed by Aydin (2009) is one of serious problems still found and not well understood in foreign language classrooms.

Key topics in JLLS
With fields of study as the analysis unit, the co-occurrence analysis of the 482 titles and 477 titles retrieved nearly 1,100 keywords. With a minimum number of five occurrences, 147 extracted keywords were grouped by eight clusters. The first cluster, for instance, consists of 34 items such as academic writing and computational linguistics. Figure   The four leading keywords also appear in all of the three divided periods in Table 6. In the first period, only pronunciation and foreign language teaching have not connection with the top ten keywords in the second and third periods. In the second period, they are problem statement and language assessment whereas in the third one they are teaching method, content analysis, and language proficiency. The keyword "english language" can be observed in all but the third period. Interestingly, "turkish" also appear as one of the top ten keywords in the three periods because most of the authors in JLLS were from Turkey or affiliated with Turkish institutions. Overall, through time JLLS has evolved around the trinity of psychology, linguistics, and education/ pedagogy. Even though the journal's name implies to the two areas of studies, i.e. language and linguistics, the trinity of psychology, linguistics, and education/ pedagogy has provided evidence that during its 15 publication years JLLS could be a scholarly proxy with more emphasis given to the language teaching. Whether the psychology, linguistics, and education/ pedagogy was "trinity or unity" in the realm of language teaching has been excellently exposed by Wardhaugh (1968, p. 80). Most of the top keywords in System from 1973 to 2017 also revolved around the trinity (Lei & Liu, 2019b). The relatively similar keyword co-occurrences suggest that JLLS has voiced not only the "Turkish" but also global issues of language teaching.

Journal's self-citation
The citation analysis is used to determine how many times the 482 documents (for documents as the analysis unit), 552 authors, or 142 organizations cited each other in JLLS. Such practice is best known as self-citation. The self-citation of the three units of analysis is compared in Figure 8.
Only 10 clusters of at least two connected organizations can be observed at the institutional level. Six universities, i.e. Anadolu University, Ankara University, eastern Mediterranean University, Islamic Azzad University, Kirikkale University, and Shiraz University have made the largest set of connected organization. At the document level, there are only 21 clusters of connected articles. The largest set of connected documents consists of 7 documents authored by Mehmet Demirezen (6 documents) and Nurdan Kavaklı (1 document). At the author level, there are only 16 clusters of connected authors. Derya Sabriler, Erkan Arkın, Gizem Doğan, İsmail Hakkı Mirici, Ozgur Yeldirim, Reza Abbasian, Serkan Çelik, and Yaser Khajavi are the eight authors who have made the largest set. For all units of analysis, most of them are not connected to each other. Viewed in this way, the level of self-citation in JLLS is relatively low.
A relatively low self-citation rate in a journal could be considered positive while a high self-citation, to some extent, might be suspected of being editorially engineered. Even some journal editors of reputable publishing houses got caught making the citation of previously published papers in their journals as one of the requirements for acceptance. Although the practice of self-citation is not uncommon for "majority of journals" to increase their bibliometric performance (Merigó, Pedrycz, Weber, & de la Sotta, 2018, p. 258, emphasis added), inflated self-citation rates might result in the journals' discontinuation of indexations in such leading citation databases as the Web of Science (Oransky, 2020). That such unethical editorially engineered self-citation could not be detected in JLLS along with higher visibility and discoverability of references outside of JLLS could explain the low rate of self-citation in JLLS despite the concentration of Turkish authors and organizations.

Bibliographic coupling in JLLS
The last analysis, bibliographic coupling, deals with the same references cited by each of analysis unit. Put it simply, it analyses how many same references cited in the 482 articles by the 552 authors affiliated with 142 organizations when the articles were submitted to JLLS. The results are compared in Figure 8.
Even though JLLS has evolved around the trinity of psychology, linguistics, and education/ pedagogy, the bibliographies cited in JLLS are relatively diverse. At the document level, only 19 clusters of coupled bibliographies could be identified. The largest set of documents consists of 14 articles including two of the top ten cited articles in Table 6, i.e. "Postmethod condition and its implications for English language teacher education" by Arda Arıkan in 2006 and "EFL teachers' perceptions of the place of culture in ELT: A survey study at four universities in Ankara/ Turkey" by Okan Önalan in 2005. At the institutional level, 16 clusters could be observed with the 7 Turkish institutions such as Afyon Kocatepe University and 4 international universities such as Utah State University (USA) making the largest set. In terms of author, out of 28 clusters, the largest relatedness of shared bibliographies comprises articles by 19 authors including Mehmet Demirezen, one of the top ten prolific authors in Table 3. Taken as a whole, the relatively low connectedness of bibliographies cited in the 482 articles by 522 authors of 144 organizations suggests the rich research profile of JLLS.

Conclusions
As one of the in-depth studies mapping how JLLS contributed and trended bibliometrically upward, this work could exhibit the descriptive and network portrait of JLLS during its first 15 publication years captured by PoP and VoSviewer through the lens of MA. Regardless the observed concentration of Turkish authors and organizations, the results suggest JLLS has been a scholarly outlet for not only local but also global issues in the realm of language and linguistics. As such some regional and global indexing services were impressed by the competence and commitment demonstrated by the gatekeepers of JLLS.
The 15 publication years have seen the bibliometric growth of JLLS around the trinity of publication, citation, and indexation. Voicing the unity of psychology, linguistics, and education/ pedagogy viewed from different perspectives, the 482 articles by 552 authors from 142 organizations of which Turkish ones were in the majority contributed to 1,291 total citations along with regional and global indexations including in Scopus, one of the largest bibliographic databases. The next 15-publication years will see the growth of JLLS in "internationality" in terms of authorship due to the combination of increased annual publication and recent international indexation.
Considering several bibliometric indicators from uni-to multidimensional ones, this study could fill one of the academic lacunae in the ever-changing landscape of language and linguistic studies. Presenting different bibliometric perspectives, this work could help readers of different backgrounds including the gatekeepers of JLLS to take best possible steps from policy to ground levels to achieve a bigger scientific impact. Besides, even though this study is basically retrospective and bound up in the investigated timespan and bibliographic database, the observed patterns such as (co)authorship and citation could inform the current status of language and linguistic studies, at least at the journal level. The revealed developmental patterns could also pave the way for further studies such as the relationship between the journal's contribution to the fields before and after being indexed in Scopus. Lei, L., & Liu, D. (2019a)