Game-based learning, serious games and gamification form a family of approaches that bring games, play or game mechanics into teaching with an educational purpose. Meta-analyses of digital game-based interventions in schools reported a medium effect on overall learning (Hedges g = 0.54),(1) and a meta-analysis of digital games in science, technology, engineering and mathematics reported a medium to large effect over conventional instruction (g = 0.624, 95% CI 0.457 to 0.790).(2) A synthesis of 33 studies in kindergarten to higher education arrived at a similar moderate estimate,(3) and a meta-analysis of eighteen studies in primary schools found positive effects of digital game-based learning, particularly in mathematics, language and science.(4) Evidence on gamification and serious games also points to attitude, enjoyment and usefulness as the main predictors of the intention to use them.(5) These estimates sit next to conceptual uncertainty, because the field has not settled on a consistent definition of educational video games (6) and the psychological accounts of gamification remain fragmented.(7) A bibliometric analysis of gamification-based learning found that most research originated in the United States, followed by Spain and Britain,(8) so the part contributed by Latin American journals is less well described.

Two developments suggest that the emphasis of the regional literature on digital and augmented technologies changed over the last decade. A systematic review reported that publications on augmented reality applications in education doubled between 2016 and 2020,(9) and a scoping review of augmented reality in Latin American engineering education began from the observation that little was known about its use in the region before charting 36 papers.(10) In parallel, university students in Chile, Venezuela and Ecuador went through emergency remote teaching during the pandemic,(11) within a longer regional history of web-based and interactive distance education.(12) The pattern anticipated for regional journals was therefore a rise of digital and gamified formats after 2020, with augmented and virtual reality appearing later and in a minority of articles. This expectation framed the analysis and was not registered as a hypothesis.

SciELO keeps open-access collections of national, peer-reviewed journals and operates in 14 countries.(13,14) Portuguese and Spanish together are the second most frequent publication language in Web of Science,(15) although Colombian researchers describe added costs when they publish in English,(16) and educational research published in Latin American journals concentrates in higher education.(17) Reviews of game-based learning in the region have addressed a single game type, such as role-playing games in Ibero-America,(18) a single country, such as digital games in Brazilian education,(19) or gamification across several databases at once.(20) The composition of the SciELO-indexed literature as a whole, by language, journal, subject and technology, has not been described in those reviews. This study aimed to describe which topics dominate game-based learning articles in Latin American journals indexed in SciELO between 2012 and 2024 and how the emphasis on digital and augmented technologies changed over that period.

The linkage of 3,377 OpenAlex candidates to ArticleMeta records produced 1,856 SciELO records, of which 478 met the core criteria (390 strict and 88 weak); the broad set added 214 records. Of 100 core records audited, 87 were judged relevant (87.0%, 95% CI 79.0 to 92.2). Among the records with a digital term in the title or keywords, 43 of 45 had a digital technology as the object of the study (95.6%). Of the 478 articles, 20 (4.2%) were classified as reviews.

Annual output climbed from 12 articles in 2012 to 90 in 2024 (Figure 1), a compound annual growth of 18.3%. The slope stayed at 13 articles a year or fewer until 2016, steepened from 20 in 2017 to 82 in 2022, and reached its peak in 2024. Annual counts and publication year coincided closely (Spearman ρ = 0.92, p <0.001). The language mix moved with the volume: Portuguese made up 54.3% of the articles in 2012-2016 and 21.1% in 2021-2024, while Spanish climbed from 13.0% to 60.7% (χ² = 38.4, 4 degrees of freedom, p <0.001). English held between 18.1% and 32.6% across periods (Table 1).

The distribution by origin clustered around three collections (Figure 2). Brazil supplied 40.6% of the articles (194), Colombia 18.6% (89) and Mexico 12.6% (60), and 13 collections contributed at least one article. The language profile differed by collection: 58.8% of the Brazilian articles were in Portuguese and 33.0% in English, against 96.7% in Spanish in the Mexican collection. Journals spread widely, with 187 titles and 96 of them (51.3%) contributing a single article. The ten most frequent journals held 22.2% of the articles, and the two leading journals held 14 each (Herfindahl-Hirschman index 107 on a 10,000 scale). The journal names contained an education term in 38.1% of the articles. The median number of authors was 3 (interquartile range 2 to 4), without a trend over years (ρ = −0.05, p = 0.286).

Higher education and health stood at the front of the distribution by level and area (Figure 3). Higher education was the most frequent stated level, with 24.5% of the articles, ahead of secondary (8.4%) and primary education (5.4%), while 55.6% of the articles stated no level. Health led the primary areas with 21.1%, followed by natural sciences (10.7%), mathematics (8.6%), languages and literacy (8.6%) and business and management (8.4%); 23.6% had no specific area. Business and management dropped from 19.6% in 2012-2016 to 5.4% in 2021-2024 (odds ratio per year 0.84, 95% CI 0.77 to 0.92, adjusted p = 0.003), and physical education dropped from 10.9% to 4.0% without a significant trend. Natural sciences rose from 4.3% to 11.4%, and higher education from 13.0% to 26.2%, also without a significant trend after adjustment.

The slope of gamification dominated the technology trend (Figure 4). Gamification appeared in 2.2% of the articles of 2012-2016, in 29.9% of 2017-2020 and in 44.3% of 2021-2024, which corresponded to an odds ratio of 1.32 per year (95% CI 1.21 to 1.44, adjusted p <0.001). The line of digital technology stayed level near 49% in the three periods (50.0%, 49.3% and 48.7%; odds ratio per year 0.98, 95% CI 0.92 to 1.04, adjusted p = 0.714), and non-digital games stayed between 15.8% and 17.4%. Mobile devices peaked in the windows ending in 2018 and 2020, and their share was 5.0% in 2021-2024. Serious games reached 14.9% in 2017-2020 and 9.1% in 2021-2024 (Table 2).

Augmented and virtual reality stayed marginal. No article of 2012-2016 carried the tag, 1.5% of the articles of 2017-2020 did and 3.0% of those of 2021-2024 did, so 11 articles in total carried it (6 for augmented reality and 5 for virtual or immersive reality). The first augmented reality article was published in 2018 and the first virtual reality article in 2021. The odds ratio per year was 1.36 (95% CI 0.98 to 1.89, adjusted p = 0.195). Artificial intelligence or adaptive systems appeared in 2 articles. The sensitivity analyses gave similar estimates. With the strict criterion only, the odds ratios per year were 1.00 for digital technology (95% CI 0.93 to 1.07) and 1.32 for augmented or virtual reality (95% CI 0.91 to 1.92). In the broad set the values were 1.01 (95% CI 0.97 to 1.07) and 1.18 (95% CI 0.95 to 1.47), and with the years 2015 to 2024 only they were 0.98 (95% CI 0.91 to 1.06) and 1.34 (95% CI 0.95 to 1.89).

The topic model with 5 topics reached the highest mean coherence in the range tested (normalized pointwise mutual information 0.202), and the refits with other initializations returned topics with a mean best cosine similarity of 1.000 (477 of 478 articles had an English abstract). Dominant topics clustered in two blocks (Figure 5). Motivation, strategies and academic performance in university courses (topic 1) held 33.3% of the articles and climbed from 19.6% in 2012-2016 to 36.9% in 2021-2024 (Spearman ρ of weight with year 0.16, adjusted p <0.001). Video games, children and social skills (topic 3) held 23.2% and dropped from 28.4% in 2017-2020 to 20.1% in 2021-2024 (ρ = −0.18, adjusted p <0.001). Digital and didactic practice in schools, with teachers, mathematics and physical education (topic 5), held 24.3% without a trend (ρ = −0.03, adjusted p = 0.563). Health education technologies, content validation and board games (topic 2) held 12.3% (ρ = −0.11, adjusted p = 0.020), and foreign-language learning (topic 4) held 6.9% (ρ = −0.01, adjusted p = 0.825). The labels refer to the highest-weight terms: motivation, strategy and university (topic 1); health, nursing, content, validation and board (topic 2); video, videogames, children and design (topic 3); language, English and foreign (topic 4); and digital, school, pedagogical and didactic (topic 5).

The main finding is that the growth of game-based learning in Latin American SciELO journals came through gamification and not through digital or immersive technology. Output multiplied several times over the period, gamification went from a rarity to a feature of more than a third of the articles, and the share of articles with a digital technology as object stayed level. The expectation that digital formats would rise after 2020 and that augmented and virtual reality would follow was therefore not met: the rise was in a vocabulary and a pedagogical approach, and immersive technologies stayed a small minority. The discussion follows the order of the figures.

The pattern of volume and language places these journals within a wider regional movement. Portuguese and Spanish together are the second publication language in Web of Science,(15) and the 2012-2024 corpus is dominated by the same two languages, with English near a fifth of the articles. The English share of the Brazilian collection, about a third, is consistent with a stated SciELO policy of adopting English alone or together with Portuguese to improve the contribution of multilingualism.(14) The change in the language mix between periods needs care, because the number of collections that contributed articles grew from four to thirteen, and the rise of Spanish partly reflects the entry of Spanish-speaking collections. Researchers in Spanish-speaking countries who publish in English face costs of writing and translation,(16) which gives a reason for national journals to remain an outlet for pedagogical work close to local classrooms. SciELO describes itself as a database of national journals with open access to full text,(13) a design that fits the high share of local-language articles found here.

The distribution by origin and journal repeats a known concentration with a dispersed tail. Brazil, Colombia and Mexico supplied most articles, a trio close to the three countries (Brazil, Mexico and Chile) that concentrated three quarters of the empirical educational research articles of 2014 to 2016 in Latin American journals.(17) Half of the journals published a single article, and the two leading journals held no more than fourteen articles each, so no single venue defined the field. A bibliometric analysis of gamification-based learning found that two journals focused primarily on the field,(8) and the contrast suggests that regional game-based learning is a topic published across general education, nursing and physical education journals. A bibliometric study of Portuguese-language educational research in the SciELO Citation Index analysed 7,448 documents published between 2002 and 2021,(29) which gives the scale of the educational literature in which these articles sit.

The pattern by level and area agrees with the weight of higher education in regional educational research, where it accounted for 43.1% of the empirical papers of 2014 to 2016.(17) A systematic review of gamification in higher education also described it as an increasingly used alternative to traditional classroom strategies.(30) The leading position of health in our primary areas departs from that review, in which health sciences obtained the lowest score of the fields compared, whereas engineering and architecture obtained the highest.(30) Two differences probably explain the discrepancy: their indicator compared bibliographic production across nine databases, and the SciELO corpus carries several nursing journals among its most frequent venues, whose articles in the audited sample often build and validate educational games. Primary and secondary education together reached about one article in seven, whereas reviews restricted to school settings centre on them: primary students were the most common participants in game-based language learning for K-12,(31) and primary lessons were the object of a whole review.(32) Because more than half of the articles did not state a level, the school share is probably underestimated. Digital games in mathematics, programming and early childhood have been the object of Brazilian reviews,(33-37) and the modest and stable share of mathematics here suggests that this interest has not been matched by a rising share of articles in SciELO journals. The decline of business and management is a relative change; a shrinking share does not mean fewer articles, and the growth of other areas can account for it.

The slope of gamification matches the literature that links gamification, motivation and performance. A systematic review of 37 studies selected from 1,706 Web of Science records found that gamification had a positive effect on students' motivation and performance,(38) and a review of gamification strategies in the Ibero-American context reported teacher interest and a positive influence on academic level.(20) The rise of the university motivation and performance topic in the topic model coincides with that body of work. Gamification is also a technology-neutral label: an article can gamify a course with paper cards, a learning platform or an application. This feature explains why gamification could climb while the share of articles with a digital object stayed level, and it warns against reading the gamification trend as a digital trend. The pandemic left no visible step: the share of digital technology did not increase in 2021-2024 even though university students in Chile, Venezuela and Ecuador went through emergency remote teaching(11) on top of a long regional history of web-based distance education.(12) A possible reading is that digital delivery became the background of teaching and ceased to be a distinguishing feature of articles, but the data cannot separate that reading from a true absence of change.

The marginal place of augmented and virtual reality departs from the international trend, where publications on augmented reality applications in education doubled between 2016 and 2020.(9) A scoping review of augmented reality in Latin American engineering education noted that the regional literature leans on conference papers, which journal indexes miss.(10) The same logic probably applies here: immersive work may circulate in engineering and computing conferences, and the SciELO journals of education and health capture little of it. Studies of digital games in Brazilian physical education journals showed a preference for active games such as exergames and augmented reality games,(39) which agrees with the presence of exergames in several articles of the corpus. The few augmented reality articles appeared in 2018 and the first virtual reality article in 2021, and with so few cases the odds ratios were imprecise; the data are compatible with a slow rise as well as with no change. Non-digital games, in contrast, kept a stable share near a sixth of the articles. The stability agrees with reviews that describe role-playing games as of educational interest in Ibero-America, even if of little impact,(18) with Latin American video game studies that combine physical and digital environments in pedagogy,(40) and with the description of game-production events in Latin America as collaborative learning experiences.(41)

The topics point to a field that moved from games as objects to strategies as practice. The decline of the video games, children and social skills topic and the rise of the university motivation topic echo the two lexical worlds that a review of 49 articles on digital games in Brazilian education distinguished, one theoretical and social and one practical, which treats games as a teaching tool.(19) Foreign-language learning kept a small and steady share, in line with a review in which English as a foreign language was the most common language in game-based language learning in K-12.(31) The fit of the topic model should be read with the limits of matrix factorization in mind: it produces coherent topics for niche corpora(25) but depends on the initialization,(28) which the repeated refits addressed.

The strengths of the study are the use of a fixed list of journals and of the full ArticleMeta record, the reproducible coding, the audit of relevance and the sensitivity analyses that gave the same direction of effects. Limitations are specific. First, ArticleMeta cannot be searched by keyword, so discovery relied on OpenAlex and on a linkage that left 1,346 candidates without a record, more often in the first years; a harvest of all records of the identified journals, with local text search, would remove this dependence. Second, the dictionary coding was checked on a sample with a relevance precision of 87% and left more than half of the articles without a level; a double human coding of a stratified sample would give agreement statistics. Third, the first period contained 46 articles, so period shares are imprecise and a replication should report intervals and use additional years. Fourth, the audit was done by the language model that drafted the manuscript, and an independent human audit is required. Fifth, the corpus covers national collections of Latin America and excludes journals indexed elsewhere, so the findings do not describe the whole regional literature; a comparison with Redalyc or Scopus would delimit that gap. Sixth, topics depended on English text and on the number of topics, and a complementary method with human review would test their stability.

The findings have practical meaning for editors and authors. A journal that wants to receive immersive technology studies could state that scope in its calls, given the thin presence found here. Authors who work with games benefit from the definition of educational video games proposed in recent scholarship,(6) which would help indexing and comparison, and from the explicit statement of the educational level, which was missing in most articles. For indexing bodies, the weight of Portuguese and Spanish and the persistence of English support continuing multilingual policies.(14,15)

The articles on game-based learning in Latin American SciELO journals grew steadily in number, were written mostly in Spanish and Portuguese with English in a minority, and were spread across many journals of general education, nursing and physical education.

Higher education and health were the dominant level and area, and the topic model separated motivation and performance strategies in university courses, children and video games, school didactic practice, health education technologies and foreign-language learning.

Gamification was the technology feature that rose across the period, whereas the share of articles with a digital technology as object stayed level and augmented and virtual reality remained marginal.