Perceived usefulness/ease.

An essential stream of research on DGBL is learners’ acceptance of these systems [26]. Important factors impacting user acceptance of DGBL systems are commonly found to be relevant to perceived ease of use, perceived usefulness, self-efficacy, and satisfaction, which are usually examined using TAM (see Fig 7 for the close relationship between acceptance and TAM), a reliable tool for predicting end-users acceptance or adoption of new technology. The use of TAM facilitates the understanding of learners’ behaviors and intrinsic and extrinsic factors in DGBL system use. For example, in a TAM-based study [27], learners learning with virtual reality games show higher game satisfaction and self-efficacy than those using traditional high-fidelity simulations. This is because learners are more immersed in and keen to keep playing with games, and meanwhile capable of transferring theories into practice with an integrated feedback mechanism helping them evaluate their practice competency. Focusing on children’s “continued use of interactive computer game-based visual perception learning systems (p. 76)” in special education, Lin et al.’s TAM analysis [26] indicated a positive effect of learner-instructor and learner-system interactions on children’s perceived ease of use and perceived usefulness.

Socio-emotions.

Emerging technologies with socialization features demonstrate motivational effects to support the basic psychological need for relatedness through promoting social belonging, connectedness, and support (see Fig 7 for the close relationship between social network and mobile game and attitude), which is consistent with the claim that “our relationships are increasingly mediated by technology (p. 6) [28].” Web technologies and social networking sites integrated with DGBL strategies provide enormous opportunities for learners to learn through interacting, communicating, and collaborating socially and virtually [29]. The study of whether and how these digital technologies impact learners’ affect and particularly those related to socio-emotions like social support and social anxiety during DGBL is important to understand how to make the best use of socialization functions brought by the technologies during learners’ DGBL. In [30], an interactive online game, by translating “evidence-based social-emotional learning strategies into personalized social problem-solving scenes (p. 959)” in virtual worlds, improves elementary-age children’s social skills, behavior, satisfaction, and confidence in their social relationships.

Motion-sensing games.

In motion-sensing games, players are required to control and complete gameplay through body movements [31]. Players’ posture and body movements also indicate their affective states [32] during gameplay for the understanding of their emotional experiences. In [33], motion-controlled video games promote learners’ information processing and increase their cognitive resources assigned to task performance, which is mainly due to the games’ advantages in keeping learners motivated and engaged during task completion. The increasing interest in studying emotions in motion-sensing games can be a result of the advances in three-dimensional technologies’ improved efficiency in capturing body position and movements, which adds value to detecting and understanding learners’ affective status during DGBL. However, the coding of body movements remains challenging with a lack of specific frameworks defining basic units of gesture and body movements for emotion detection [34].

Competitive gameplay anxiety.

Competitive anxiety (see Fig 6 for the close relationship between competitive and anxiety), as an essential inhibitor of task achievement [35], is increasingly studied during gameplay, particularly in competitive DGBL due to its commonness and potential negative impacts on learners’ performance (e.g., higher competitive anxiety usually results in lower gameplay enjoyment and engagement). However, competitive anxiety’s negative effect on gameplay interest is not conclusive in the literature. Hong et al. [36] found competitive anxiety’s positive relationship with interest in gameplay, and they attribute it to the stimulated players’ interest and the gameplay control triggered by the higher state anxiety. Hong et al. [37] also suggest competitive anxiety’s positivity in eliciting a change from goal-oriented to stimulus-driven attention, thus stimulating players’ focus on competitive situations and their intention to play again. Whether competitive anxiety plays a positive or negative role in gameplay also relates to individual personality. Hong et al. indicate a positive correlation between gameplay anxiety and deprivation-type (D-type) epistemic curiosity (EC), which is because D-type EC involves desires for knowledge and worries for increasing working efficiency, thus triggering anxiety to play.

Location-based games.

With the integration of mobile technologies with location awareness features, location-based learning has been significantly extended from mere content identification to interactive discovery to promote learners’ engagement and immersion in contextual activities and enhance their experiences of learning from surroundings and with each other [38]. During such immersive gameplay, learners’ immersion is measured in the forms of different levels of cognitive and emotional involvement [39]. It is commonly reported that location-based games improve learners’ learning attitudes and achievements by guiding them in finding the target and showing them the learning tasks and materials accordingly.

Subject-specific emotions.

Due to the increasing interest in developing educational games specific for particular subject learning (e.g., game-based writing/vocabulary learning, word game, and code-learning game) and the growing awareness of the importance of learners’ affective status in impacting gameplay performance, diverse emotions related to specific subject learning (e.g., English anxiety, writing attitude, and computer anxiety) (see Figs 6 and 7 for the close relationships between anxiety and language learning; anxiety and English learning; anxiety and Chinese idiom) start to emerge in this period. In [40], learners show significant improvements in Spanish learning enjoyment, engagement, and attitude towards computer science brought by game-like activities integrated into grammar-specific mobile learning applications. Hosseini et al. [41] also suggest that game-based lectures are more engaging and enjoyable, contributing to learners’ improvements in learning perception of computer science concepts, engagement, and teamwork. There are also studies showing variations in learning achievement associated with anxiety levels. For example, Cheng and Chen’s study [42] indicates that learners with lower English anxiety show higher achievement compared to those showing higher anxiety while using an English mobile learning system integrated with interactive games.

Affective regulation (strategy).

Self-regulated learning involves the regulation and management of individual learning to comprehend personal performance and achieve learning goals [43]. Self-emotion regulation, as a crucial element in self-regulated learning individuals adopt to enhance, decrease, or sustain specific affective states for outcome achievement [44], is increasingly researched and recognized as important in determining improvements in DGBL. During gameplay, learners showing higher self-regulation abilities tend to experience positive flow affect [45]. Attention has also been paid to how and under what situations self-regulation elicits the most positive effect. An exploratory study [46] indicates the benefits of engaging in effective regulation of cognitive reappraisal and acceptance to reduce frustration and confusion. Chen and Sun [47] suggest enhancing learners’ immersion in DGBL by balancing between learners’ skill levels and game challenges since learners are more likely to self-regulate their flow state when a difficult choice is given to them.

Intrinsic need/motivation.

As a socio-emotional factor supporting effective interactions, intrinsic motivation, an “innate psychological need for competence and self-determination (p. 3) [48],” includes learners’ interest in their learning subjects and learning processes. Accumulative evidence indicates gamified interventions’ varied impacts on learners’ learning participation cross intrinsic or extrinsic motivation. Compared to extrinsic motivation, according to self-determination theory, intrinsic motivation increases game enjoyment more. Such a principle is increasingly considered during the implementation of meaningful gamification mechanics with a key goal to construct meaningful and intrinsically learner-centered DGBL experiences. Thus, how to trigger more intrinsic motivation through gamification element use becomes a prevalent research topic. In [49], the use of gamified interactive response systems (IRSs) contributes to greater intrinsic motivation to learn English as compared to those using whiteboards. This is because gamified IRSs are attractive and can create easy delectation of learning atmosphere where instructors plan richer and more interesting courses to promote learners’ learning interest and classroom interactions, thus triggering their intrinsic motivation to learn. In addition, when integrated with a feedback mechanism and clear goals [50], educational games more effectively contribute to learners’ intrinsic motivation and thus their perceived learning and learning outcomes.

Emotion dynamics and changes.

According to [51], a broad range of emotions can be triggered during DGBL simultaneously and dynamically. Hence, it is critical to avoid a mere focus on retrospective assessment to understand learners’ perception of a current situation; however, this is contradictory to most extant literature retrospectively measuring learners’ affective states in DGBL. On this account, scholars start to pay attention to emotional dynamics by measuring changes in emotions during DGBL (see Fig 6 for the close relationship between dynamics and engagement). An autoregressive model [52] indicates a significant relationship between learners’ control perception and emotions in every round of gameplay designed for human liver functionality instruction. Therefore, it is essential to determine whether and when an adaptive change in specific design characteristics should be introduced for subsequent game tuning to obtain ideal gameplay experiences and performance.

Mood/affective disorders.

Children with autism spectrum disorders commonly have difficulties in recognizing facial expressions and emotions (see Fig 7 for the close relationship between child and autism spectrum disorder and Fig 6 for the close relationship between child, autism, and cognitive). A promising solution to resolve such difficulties is to exploit robotic toys’ potentials to facilitate intervention processes of autistic children by enabling them to experience diverse social behaviors through interacting with robots. An examination of socio-emotional development [53] indicates humanoid robots’ effectiveness to promote high functioning autistic children’s acquisition of facial expression recognition skills during gameplay and their engagement in interactions. The increasing interest in researching affect and emotions in children with mood or affective disorders is partially contributed by technological advances with eye-tracking, EEG, ECG, and facial emotion detection technologies being popularly introduced to more conveniently detect emotions compared to traditional data collection methods like questionnaires and interviews requiring self-reporting.

Impact of human factors on attitude towards personalized DGBL.

There is a trend towards customizing avatars, scenarios, and gaming content tailored to individual learners to promote DGBL immersion (see Fig 3). In [54], learners were highly engaged with customizable avatars that allow them to customize their own unique avatars in terms of gender, appearance, profession, and animal companion in line with their preferences. However, such personalized gamification experiences vary across individuals with different characteristics (e.g., cognitive styles, learning styles, game experiences, and personality traits). Thus, the understanding of how individual characteristics affect gamification experiences is important to provide insights into effective gamified learning intervention design. Chen et al.’s study [54] on the impact of cognitive styles (i.e., Pask’s Holism/Serialism) on learners’ reactions towards customizable avatars suggests that Serialists and Holists experience higher levels of engagement and sense of presence, respectively.

There are also studies (e.g., [55–57]) examining the impacts of learning styles and personality traits on gaming experiences and perception. Sepehr and Head’s study [57] suggests that learners with higher levels of dispositional competition are more intrinsically motivated during gameplay and considered the games more competitive. Focusing on personality traits’ impact on learners’ perceptions of and engagement with gamified learning interventions, Buckley and Doyle [55] suggest that extraverted individuals, compared to conscientious ones, are motivated more by gamification. Buckley and Doyle also found that emotionally stable individuals, compared to neurotic ones, show better performance, and individuals tending towards global learning favor gamification more, whereas highly conscientious individuals commonly show a negative gamification perception. Buckley and Doyle and Ghaban and Hendley indicate that enhancing learners’ motivation cannot guarantee their improvement in learning since varied personalities respond differently to gamification elements. Therefore, for tutoring system designers thinking of integrating gamification, it is crucial to provide learners with gamification elements matching their individual characteristics, and that can be updated dynamically based on their behaviors during gameplay.

Contextual DGBL.

Contextual learning involves providing simulated contexts to apply knowledge and opportunities to mastery learning through repeated practice and personalized guidance and feedback [58]. Digital games provoke active learner involvement during contextual learning through exploration, experimentation, competition, and collaboration. Learners’ immersion in the contextual DGBL activities enhances their learning motivation and interest [59], and has become an increasingly popular practice in healthcare education. In a contextual role-playing game [60], learners learn to save others’ lives in varied cases by collecting data concerning normal and abnormal ECG waves and completing relevant testing tasks. Contextual games’ effectiveness in enhancing learners’ learning motivation, attitudes, and satisfaction is mainly because integrating learning materials into gaming scenarios allows learners to learn with pleasure and interactively [61], and situating them in real-world simulations enables them to link real-life problems to their learned knowledge [62].

Collaborative games and team-based gaming.

Compared to individual learning and conventional instructions, collaborative learning is increasingly considered effective to elicit higher achievement concerning individuals’ cognitive development [63]. The effectiveness of collaboration strategies can be enhanced by integrating gamification mechanisms that are advantageous in promoting motivation and engagement to complete educational tasks [64] (see Fig 6 for the close relationship between group and attitude, engagement, and motivation). Particularly, in an era where technological advances proliferating in every aspect of social life, designers increasingly focus on developing and designing technological support for collaborations in DGBL activities [65] where learners practice self-regulation and social skills to achieve team goals. Hanghøj et al. [66] indicate the positive effect of collaborative role-playing games in promoting the sense of social belonging among learners with social difficulties by increasing their social participation and engagement through team collaborations. Doumanis et al. [64] also claim that learning in virtual multimodal contexts with integrated collaboration and gamification elements promotes learners’ motivation and engagement in accomplishing collaborative tasks by involving learners in intuitive and rich interactions with others and the virtual learning contents.

Reward mechanisms in DGBL.

Similar to games where rewards play an essential role as motivators and feedbacks, in education, rewards are used to motivate and engage learners and recognize their efforts. The effectiveness of rewards as a form of either identified regulation (learners’ consciously-valued rewards like tokens or badges) or integrated regulation (rewards aligning with learners’ goals in enjoying gameplay) is commonly recognized [67]. However, in practice, rewards have varied effects on gameplay, especially those showing no positive impacts on learners’ game experiences, thereby limiting their values to learners. Effective rewards are expected to facilitate positive game experiences by motivating learners to actively participate in learning tasks to obtain the rewards [68]. Also, prompting rewards from gamified systems are informational feedback on learners’ short-term performance, which can promote their sense of achievement [69] and encourage them to set new goals and concentrate on gaming progresses [70]. The variations in different rewards’ effects on DGBL and the significance of effective rewards stimulate scholars’ interest in comparing different types of reward mechanisms and their effects on learners’ gameplay. A reward strategy comparison during gamified instruction [71] indicates that forfeit-or-prize and prize-only strategies stimulate learners’ learning motivation. This is because learners experiencing the forfeit-or-prize strategy tend to experience higher levels of anxiety. Therefore, it is necessary to balance between rewards and learning anxiety (pressure, stress, or competition) via reward strategy design to boost motivation but suppress pressure. In [69], performance-contingent and completion-contingent rewards were compared. With an example of English vocabulary learning via arrow-shooting gaming, the authors suggest that both performance- and completion-contingent rewards facilitate learning and motivation. The higher perceived competence and autonomy brought by the performance-contingent rewards is due to the rewards’ success in delivering sufficient competence information, which, according to cognitive evaluation theory [48], negates the controlling effect, thereby positively influencing learners’ motivation.

Affective states and socio-emotional factors.

In terms of affective states and socio-emotional factors, four dimensions (i.e., emotions, mood, personality traits, and socio-emotional factors) are used to categorize popular concepts appearing in the corpus. Referring to evolution analysis, overall, affective state emotions and socio-emotional factors begin to be considered in DGBL in an early stage, particularly satisfaction, motivation, engagement, and encouragement. Comparatively, affective state mood (e.g., confusion, frustration, anxiety, and enjoyment/joy) and personality traits (e.g., attitude, aggressiveness, and emotional stability) appear in the latter stage of DGBL research. Though the volume of literature concerning learners’ affective status in DGBL is increasing, there is a paucity of longitudinal studies. As it is common that once the novelty of instructional strategies like DGBL wears off, learners would be fatigued and bored, thereby eroding their engagement and motivation. Hence, as suggested in several studies (e.g., [60, 75]), it is worthwhile to conduct longitudinal studies to evaluate gamification’s long-term effects on learners’ positive learning emotions. For example, since it was the first time for participants in most studies to participate in DGBL, the investigation of the engaging impacts of gaming activities when they are implemented regularly and become less novel [76] appears essential.

As discussed before, learners differ in demographic factors (age, gender, education levels, and cultural contexts) and personality traits (competitiveness orientations, spatial ability, and learning and cognitive styles), which may impact gamification’ effect on affect and emotions (e.g., [60, 77]). Hence, the influence of these variables ought to be assessed (e.g., how competitiveness engages or enables learners’ problematic behaviors) [78] to gain a holistic and contextual understanding of DGBL design and implementations specific for certain learning contexts and different groups of learners. Also, varied educational conditions (e.g., study subjects and instructional strategies) under which gamification more positively affects individual learners (e.g., [79, 80]) should be examined and compared.

Types and design of games.

We identified different types of digital games, for example, simulation games, serious games, role-playing games, and motion-sensing games, where learners’ affective status is considered. However, as most studies examined gamification’s effects as an overarching concept, future studies are advised to evaluate gamification’s specific elements to parse out different mechanics’ effectiveness [25, 81]. For example, in gamified system design, researchers are suggested to make the most of digital gaming mechanics that are best employed using computers and virtual worlds [25]. Also, as greater numbers of gamification mechanics cannot necessarily contribute to better learning performance [82], the optimal number of badges maintaining active engagement ought to be determined [76]. The isolation of specific game mechanics and assessment of their effects on learners’ affect and emotions enables a better understanding of nuanced and effective gamified learning intervention design to maintain and promote motivation and engagement.

According to Hwang and Wang [83] and Calvo-Ferrer [84], a clear goal with a moderately high cognitive load is more likely to facilitate effective learning. It is thus important to consider challenging learners’ thinking skills and integrating clear learning objectives into game dynamics during game design. Also, to enhance enjoyment and decrease anxiety, cognitive and metacognitive scaffolding are important to foster high levels of subjective control [84]. Wei et al. [85] and Wu and Huang [86] also suggest the contribution of assistance tools and strategies to promote learners’ motivation and engagement by constructing a low-anxiety learning atmosphere and enhancing learners’ immersion and intrinsic interest. It is thus essential to consider integrating personalized assistance and providing scaffolding in educational games to release or moderate anxiety to thus promote game performance. Game developers and designers can employ open-source models and stimulate learners’ to customize games via source code manipulation [87]. This would provide more personalized learning experiences for learners to better engage in game activities, thus promoting learning and game performance. In addition, it is essential to integrate grouping rules and tactics to achieve effective heterogeneous collaborations, thereby contributing to peer interaction in real time.

Applications and platforms.

Our analyses reveal that traditional computers and devices are popularly adopted to support DGBL, which is mainly because game systems are commonly developed based on existing packages or systems to save costs in human resources and time [88]. With the advances in information technologies in wearable and handheld devices, there is an increase in educational games to be supported by mobiles, tablet computers, and virtual/augmented reality. To maintain learner engagement and ensure enjoyment while concentrating on solutions and resolving challenges during gameplay, handheld devices are effective by providing individual learners with equal opportunities to participate in the games and interact with others. Therefore, effective ways to arrange learning content and materials in handheld and mobile devices are expected to provide learners with prompt peer feedback and equal learning opportunities [89] to trigger positive emotions during DGBL, thus promoting gameplay and learning performance and outcome. Another promising stream of research lies in the wearable learning technologies with personalized data to facilitate the understanding of individuals’ affective status during gameplay to allow more personalized learning experiences. This is in line with the definition of “wearable personal learning technologies [90],” involving gathering data from learners wearing the devices and the surroundings to promote instruction differentiation and learner engagement.

Analytical technologies.

Increasingly diverse technologies are used to facilitate the understanding of learners’ affect and emotions during DGBL. Traditional statistical methodologies (e.g., Bayesian network, SEM, regression analysis, survival analysis, and sequential analysis) are mainly employed to analyze structured data collected using traditional questionnaires and survey data after gameplay. However, more attention should be paid to how to directly capture data concerning learners’ activities and emotions during gameplay. A potential methodology is by combining questionnaires with game log and assessment data and by increasing measurement robustness [91]. Additionally, techniques like “think aloud” during gameplay, video analysis, click-data, and psycho-physiological measures like EEG, ECG, and skin conductance, are expected to gauge learners’ state-to-state emotional experiences during playing [59] to infer their learning status (e.g., emotions, cognition, and attention). Alongside technological advances (e.g., Internet of Things, wearable and cloud technologies) is the realization of collecting learning data at high-frequency, fine-grained, and micro-levels in a real-time mode. These multimodal data with great diversity can be analyzed using affective computing and learning analytics, a powerful tool for developing adaptive educational technologies [92], to facilitate the understanding of learners’ trajectories in the games and explore their relations to learners’ self-reported goals and purposes and outcomes like engagement [93]. A promising direction is multimodal learning analytics, which detects, incorporates, and analyses learning traces obtained from varied sources/channels and on a large scale to enable a comprehensive understanding of learning processes. The analysis results on affect and emotions can be combined with individual learners’ preferences, interests, and goals to facilitate the provision of personalized support and interventions via recommender systems like collaborative filtering technologies to promote their engagement in DGBL.