“Intergenerational” is an adjective used to describe what actually occurs between different generations (Sánchez and Kaplan, 2014; Dauenhauer et al., 2016). Previous research investigating intergenerational relationships focused mainly on the reciprocity between parents and their children (e.g., Bó et al., 2020; Chai et al., 2020; Wang et al., 2020). Intergenerational learning is defined as the reciprocal exchange of knowledge between individuals from different generations. In doing so, people of all ages, whether relatives or strangers, are able to learn together and from each other (Sánchez and Kaplan, 2014). It should be noted that intergenerational learning emphasizes the centrality of exchange and reciprocity between different generations (Sánchez and Kaplan, 2014). In a multigenerational classroom, members of different generations possess some generational experience of belonging to different generations (Sánchez and Kaplan, 2014). Intergenerational learning not only refers to learning about others but also learning about oneself and one's own generational bearings (Sánchez and Kaplan, 2014).

Macroanalytic and microanalytic coding systems are different approaches to exploring behavior and interaction. The former have a global and gestalt-based focus, and are developed to investigate the major themes and bigger units of interaction. The latter are more appropriate when researchers aim to investigate smaller interaction units (Verstaen, 2017). Previous research on observing interaction among individuals mainly focused on couples and family members (e.g., Gottman and Driver, 2005; Darling et al., 2008; Friedlander et al., 2019). Studies on intergenerational partners and their interaction processes have, however, been accumulating because of the growing need for social structure (McKee and Scheffel, 2019; Sun et al., 2019; Yin et al., 2019).

For example, Rubin (2001) developed a play observation scale, which categorized play behaviors as unoccupied (e.g., inactivity), watching solitary (e.g., activity without social interaction), parallel, and cooperative. Also, Jarrott et al. (2008), modeling on Rubin's (2001) scale, developed an intergenerational observation scale (IOS) to investigate intergenerational interactions and affect. They expanded IOS to distinguish social targets such as age peers (e.g., interactive peers) and intergenerational partners (e.g., interactive intergenerational). In this scale, intergenerational behaviors are accordingly coded as seven categories: interactive intergenerational, parallel intergenerational, interactive peer, parallel peer, staff, watching solitary, and unoccupied.

To explore intergenerational interactions involving adolescents, institutionalized elderly, and older volunteers, Santini et al. (2018) developed two similar coding frameworks (one for older adults and one for adolescents). The categorization of adolescents' behaviors includes representation of older adults (e.g., bad/good, burden, experience exploitation, and useful), intergenerational relationship (e.g., conflictual, friendly, teach ICTs, understanding, listening, psychological support, tell, read, teach and mentor, and take care), lessons learnt from (awareness and changes of attitudes toward volunteering), and suggestions for improving the relationship with the youngest (the role of adults and the role of school authority). On the other hand, the categorization of older adults' behaviors involves the representation of young people (e.g., egoist, altruist, and able to listen), intergenerational relationship (e.g., indifference/distance, friendship, physical help, ICT literacy, company, love, and listening to each other), impact on (able to feel emotions and able to give love), and suggestions for improving the relationship with the youngest (more meeting chances, talking, systematic visits, more time spent together, and trips outside and fun).

According to the aforementioned coding systems, two key elements are extracted: behaviors and targets. Rubin (2001) used categorizations to represent a behavior without a specific target (e.g., solitary, parallel, and cooperative). For example, the categorization of cooperative means an individual doing something with a target; however, it still failed to refer to a specific target when doing something (i.e., age peer or intergenerational peer). Therefore, Jarrott et al. (2008) and Santini et al. (2018) employed categorizations involving one behavior and a specific target to represent with whom the individual does something. Nevertheless, few behavior coding systems have been customized and developed for collaborative communication during intergenerational learning. The current research accordingly aimed to develop a coding system for the assessment of collaborative communication regarding intergenerational learning.

There is more to intergenerational learning models than just the participation of learners of different ages and generations. Diversity and enrichment could result not only from advancing age but also from various experiences, individual life events, and historical events (Whitehouse, 2017). To assess intergenerational group members' collaborative communication during intergenerational learning, the current study reviewed previous research especially focusing on the theme of intergenerational collaborative communication. According to previous research findings, the power process (McHale et al., 2015; De Schutter et al., 2017), communication skills (Suhr et al., 2004; De Choudhury and Kiciman, 2017), and the way of responding to bids (Driver and Gottman, 2004; Gable et al., 2006) are key factors of individual behavior and interaction among dyadic or group partners. Therefore, the current framework included these three core sub-systems, as described below.

The current research addressed the following research aims. First, the present research sought to develop a coding scheme for collaborative communication during intergenerational learning. Second, to establish the reliability of our coding scheme, we aimed to examine the inter-rater reliability of the coding scheme across coders. Third, to verify the validity and the application of the various learning conditions of intergenerational learning, the study aimed to explore how elderly adults and University students learn from each other, and how the balance of power is compromised and negotiated by creating two different collaborative tasks—one focusing more on the process of creation and the collection of the game elements and design, and the other placing more emphasis on the completion of an executable prototype. Finally, the current research aimed to investigate and compare two different collaborative communication patterns shown in two different learning contexts.

After developing the coding scheme for intergenerational learning, an intergenerational workshop was designed and held to investigate the reliability and application of this scheme, and to explore the communication patterns during intergenerational learning. A total of 16 individuals (seven older adults and nine University students) participated in this workshop. The average age of older adults was 62.33 years (SD=3.14) and the average age of University students was 21.22 years (SD=2.73). During the workshop, all the participants were invited to learn and design digital games with their intergenerational group members. They were required to design a game with two main purposes: (1) play and have fun with intergenerational partners; and (2) the game elements were to involve local culture elements (e.g., understanding festivals or community). Participants of two generations (i.e., older adults and University students) were, respectively, and randomly assigned to four groups; therefore, there was at least one older adult and one University student in each group. All participants provided written informed consent before participating in the workshop, and they were fully debriefed at the end. The participants in each group performing the two collaborative tasks are shown in Table 1.

The current workshop was modeled after the Miami Six-O project (De Schutter et al., 2017), and was tailored according to the collaborative tasks Previous research has investigated the benefits of games for elderly adults (e.g., McLaughlin et al., 2012). Also, digital game design was conducted in previous studies and was found to be a good way to enhance intergenerational relationships because technology has been proven to be a medium which changes the traditional dominant role of older adults into a more balanced and equal power in intergenerational relationships (e.g., Khoo et al., 2009; Al Mahmud et al., 2010; De Schutter et al., 2017; Cucinelli et al., 2018). Therefore, a workshop was accordingly created to explore and design digital games as a medium to facilitate intergenerational learning and intergenerational collaborative communication. Therefore, the aims of the workshop, as advertised to the University students and senior participants, was to learn together with intergenerational classmates to design a paper prototype of digital games for intergenerational players.

The workshop comprised a four-step creative process that was spread across 6 days over a 2-week period. Participants joined this workshop every 2 days during the 2-week workshop. The process was customized for intergenerational learning, and followed previous models (e.g., Howard et al., 2008; De Schutter et al., 2017). Based on the models of Howard et al. (2008) and De Schutter et al. (2017), the four stages were similar to the analysis, generation, evaluation, and implementation phases.

The instructors of this workshops are experts with game design and local culture backgrounds. They designed the courses of the workshop and led the participants based on the four stages (analysis, generation, evaluation, and implementation). Throughout the 2-week workshop, the intergenerational participants were instructed and guided through a structured process to analyze the current resources and conditions. In addition, individuals in each group were videotaped and recorded throughout the whole workshop.

The current coding system involving three sub-coding schemes (see below) was adopted for investigating intergenerational learning, accordingly, examining the first and second aims of the current research. The power process coding scheme included three codes and was adopted to investigate the third research aim. The communication skills coding scheme involved eight codes, and the response to bids coding scheme included seven codes. Three coding schemes were adopted to explore the fourth research aim.

For the quantitative content analysis, since Phases 1 and 2 mainly focused on cultivating team camaraderie and exploring familiarity among team members, the intergenerational participants only started to get to know their group members and had not yet developed enough familiarity and confidence during the discussions. Accordingly, the current study aimed to analyze the patterns of discussion in Phases 3 and 4. In view of this, the researchers only watched and analyzed the videos of “Be a chef!” and “Hey! Show me your game!”

In addition, to investigate the inter-rater reliability of the current coding scheme for intergenerational learning and the exploration of patterns of collaborative communication during different tasks, we chose two 20-min collaborative interactions (one from Phase 3 and one from Phase 4) to code and analyze the intergenerational learning. The video data of one sampled intergenerational interaction were first coded by two well-trained coders with education or psychology backgrounds.

After determining the inter-rater reliability, all video data of collaborative interaction in Task 1 (Be a chef!) were divided into a total of 350 interactional collaborative behavior units across the four groups. Additionally, data of the collaborative interaction in Task 2 (Hey! Show me your game!) were divided into a total of 239 interactional collaborative behavior units across the four groups. Furthermore, to better understand the sequential process of intergenerational discussion, the current research also divided 20-minute collaborative interactions into five intervals (refer to Figures 1–3) to review the change in patterns during Tasks 1 and 2.

Results of Kappa coefficient k revealed the high inter-rater reliability of the codes (Landis and Koch, 1977) for the power process (k = 0.87; p < 0.001), communication skills (k = 0.91; p < 0.001), and responses to bids (k = 0.85; p < 0.001). Table 3 shows the frequencies and percentages of each interaction code during the discussions in Task 1 and Task 2 for each group, individually and across the four groups. The overall findings revealed that, in Task 1, almost one third of the interaction codes show that the older adults were in charge of the discussion; slightly more than one third show that the younger participants played a leading role during the discussion; and one third show the balance of power structure during the discussion. Our patterns tended to be consistent with the findings of De Schutter et al. (2017), showing that the discussions were dominated by either older or younger adults in half of the teams, while the other half of the teams revealed collaborative interactions. The patterns of Task 2 are, however, completely different from those of Task 1, showing that the younger participants in the discussion regarding an executable game prototype guided and led more than their senior counterparts.

To meet the third aim of the current research, the study created two tasks with different purposes to explore how elderly adults and University students learn from each other, and how the balance of power is compromised and negotiated during intergenerational learning. As shown in the upper part of Figures 1, 2, both senior and younger participants displayed equality and contributed mutually to the coding scheme of the power process and communication skills.

The patterns indicated egalitarian intergenerational collaboration consistent with previous research findings. For example, according to the research findings of DeMichelis et al. (2015), both older and younger participants actively engaged in the activities. Additionally, intergenerational learning serves as a reciprocal interaction whereby both older and younger generations show unique contributions (Borrero, 2015). However, each generation demonstrated different behavioral patterns in various ways. For instance, older adults tended to offer younger partners real-world understandings of concepts, and were good at mentally reflecting themselves backwards and forwards across different life experiences; conversely, younger counterparts were more adept at providing educational knowledge and learning from others' life courses (Borrero, 2015; DeMichelis et al., 2015). Additionally, according to a study on the healing effects of intergenerational dialogue, the results revealed that the elders were willing to share personal and positively helpful stories from their history and experiences, whereas the youth were eager to learn from older counterparts' past experiences (Wallace et al., 2014).

As shown in the lower part of Figures 1–3, the younger participants played a dominating role in the coding schemes of power process, communication skills, and responses to bids. The patterns indicated that younger partners are more capable of providing their educational experience and their knowledge of technology to their older partners, consistent with previous research findings.

Derboven et al. (2012) conducted a study regarding intergenerational collaboration. Participants were invited to join an intergenerational user interface discussion and game evaluation. The findings revealed that younger participants put more effort and time into guiding and helping their older partners on the user interface. It was also suggested that younger players enjoyed teaching and explaining how to play games to their older partners. Younger partners are more capable of sharing their experience with and knowledge of technology with their older partners (Derboven et al., 2012).

Additionally, Alfrey et al. (2020) suggested that the traditional power structure within intergenerational relationships may be disturbed and overcome if youth are given the opportunity to play the role of teacher to their older counterparts. For example, it was suggested that opportunities be created for younger participants to teach older participants how to use mobile phones.

Overall, the current research makes a unique contribution to intergenerational learning. It also provides implications for future studies and educators as follows. First, our findings developed a coding scheme for intergenerational learning, and illustrated two patterns of collaboration when tasks with different purposes were delivered. Second, the validation of the coding scheme for intergenerational learning showed a reciprocal and mutual contribution during the task regarding the process of creation and collection of the game elements and design. The findings were consistent with previous research results. For instance, Alfrey et al. (2020) conducted intergenerational arts-led pedagogies, and their findings suggested that the intergenerational arts-led pedagogies worked in connecting different generations. Additionally, through pedagogies, stereotypes of the other generation and the power structure between the generations were disturbed and shifted.

Nevertheless, a previous study indicated that egalitarian intergenerational collaboration requires great energy and effort due to the age-based inequality (Taft, 2015). Intergenerational dialogue, despite many efforts to foster youth's authority and agency, continues to exhibit deeply structured patterns of interaction that give elders greater power. However, the current findings successfully showed a technique (i.e., teaching how to design a digital game) to disturb the traditional power structure of intergenerational collaboration. The findings paralleled the results of Guha et al. (2013), indicating that the power issues in intergenerational relationships exist, but that these issues can be resolved. To change the power structure in intergenerational relationships, numerous techniques were used in their study to overcome the power differentials. For example, they arranged intergenerational participants to have fun together by playing outside, sitting together on the floor, and visiting a campus.

Dauenhauer et al. (2016) employed an intergenerational approach to investigate the promotion of balance and strength for fall prevention. The findings revealed that intergenerational intervention is effective in terms of the prevention of fall risk in seniors and younger participants. It was also suggested that future research should pay more attention to the issues of whether intergenerational programs are equal to every generation or serve as age-specific programs (Granacher et al., 2011; Alfrey et al., 2020).

Last but not least, the current findings should be interpreted with caution. We list the limitations with corresponding future research directions as follows. First, the current research investigated participants using the observation method, which is a research technique where researchers observe individuals' ongoing behavior in a natural situation. Accordingly, its conclusions should be limited to this particular collaborative task and a limited number of participants. Future research may collect data using other methods such as experimental tasks, questionnaire surveys, or employing a longitudinal design. Future research may sequentially investigate intergenerational learning and intergenerational collaborative interaction from different waves of observation or investigations, thereby establishing the causal relationships. Future studies could incorporate more investigations in various situations to explore intergenerational learning in greater depth. Second, the collaborative patterns shown in Figures 1–3 were compared based on the five intervals of 20-min interactions, which failed to present significantly sequential behavioral patterns. To better understand the behavioral sequential patterns, future research could consider choosing sequential analysis (Bakeman and Gottman, 1997; Hou, 2010, 2015; Cheng and Hou, 2015; Wu et al., 2016), which is another type of analysis for behavior and interaction. During sequential analysis, the coded data of behavior codes are used to ascertain sequential behavior patterns. Nevertheless, this idea is speculative and should be investigated in future research.