Japanese preschoolers’ cooperative engagement in lunch monitoring activities

Abstract In most Japanese preschools, children are involved in setting up lunches as monitors’ activities. The current study conducted six-month long naturalistic observations of lunch monitors’ activities in a nursery school in Tokyo. Ten five-year-old children were included in the analysis. By engaging in the activities with their peers, the lunch monitors learned serving skills and how to serve food and related items accurately. There were individual differences in the type of cue children relied on while placing items, based on which they were divided into two groups: self-helping and social groups. Between these two groups, role-sharing and cooperative commitment were observed. The obtained results were discussed in terms of development of peer cooperation, the effect of small group activities with same members, and teachers’ careful arrangements of small group activities.


Cooperation in young children
Cooperation is a quality of human beings that forms the foundation of our cultures (Tomasello, 1999). Recently, there has been growing interest in the early development of cooperation and ABOUT THE AUTHOR Eating is not just a biological necessity but also a highly structured socio-cultural activity for individuals. From the very beginning of children's social development, children acquire several socio-cultural skills by eating together with their family and peers. My research interest lies in studying how children acquire various skills in the socio-cultural setting of mealtimes. For this purpose, I have conducted naturalistic observations of children during mealtimes at preschools and at homes. There are cultural variations in the kind of skills adults expect children to acquire during mealtimes. In Japan, emphasis is laid not only on nutrition, but also on good manners and pleasant conversations. Consistent with this cultural belief, adults eat and interact with their children to guide their cultural participation during mealtimes. The current paper focused on preschoolers' lunch monitor activities and examined the social skills children acquired through this activity. (142)

PUBLIC INTEREST STATEMENT
The ability to cooperate is a human quality that forms the foundation of different cultures. For most young children, preschool is their first group living experience where they learn a variety of social skills. Japanese preschools maintain a duty system where children are responsible for tasks like serving school lunches and cleaning. The current study conducted longitudinal naturalistic observations of lunch monitors' activities at a nursery school in Tokyo where, over a period of six months, all five-year-old preschoolers became skilled lunch monitors. Individual differences based on the cues children depended on for placing items were observed. While some children preferred self-helping cues, others relied on social cues. Among these different sets of children, role sharing was observed. The obtained results were discussed in terms of developing peer-cooperation, the effect of small group activities containing the same members, and teachers' careful arrangements of small group activities. (146) social coordination with peers (Brownell & Carriger, 1990;Eckerman et al., 1989;Melis et al., 2016;Smiley, 2001). With adults, children exhibit cooperative behavior, such as turn-alternation and reciprocity, quite early during infancy (Ross & Kay, 1980), whereas with peers, cooperation develops a little later, that is, in the second and third years of life. At around 20 months of age, cooperative behavior becomes evident in peer interactions (Brownell & Carriger, 1990;Brownell et al., 2006;Eckerman et al., 1989). Children's social coordination with peers is associated with the development of social skills, e.g., attention sharing with adults and in language comprehension and production about self and others (Brownell et al., 2006), and understanding of intention (Smiley, 2001). After toddlerhood, peer interaction becomes more elaborate and responsive to one's partner (Ashley & Tomasello, 1998;Gräfenhain et al., 2009), and also more selectively (Hermes et al., 2016). Between the ages of three and five years, young children also became able to adopt a turn-taking strategy with familiar peer partners for their mutual benefit (Melis et al., 2016).
As these studies suggest, cooperative behaviors with peers seem to become more sophisticated and partner-sensitive during preschool years. In these previous studies, children's cooperation, while engaged in some experimental playful tasks, was observed in experimental situations (Brownell & Carriger, 1990;Brownell et al., 2006;Eckerman et al., 1989;Gräfenhain et al., 2009;Hermes et al., 2016;Melis et al., 2016;Warneken et al., 2012) or within their homes (Smiley, 2001). These studies examined how children built cooperative interactions while engaging in short-term tasks. Due to such research methods, an important question of how children get to develop cooperative relationships with each other on a certain long-time basis appears to be unclear. The aim of this study was to examine this question by conducting longitudinal naturalistic observations in a Japanese preschool.

Monitor activity
Children acquire various cognitive and social skills by participating in daily activities that are culturally structured within each community (Bronfenbrenner, 1979;Rogoff, 2003). Settings of children's everyday lives and adults' involvement in children's activities are diverse among cultures. Children in some cultures regularly participate in specially designed child-focused activities as preparation for adult life or schooling, while children in other cultures spend much time performing the work of adults (Morelli et al., 2003;Paradise & Rogoff, 2009). Japan, where the current study was conducted, would be classified into the former. Formal schooling prevails and children grow up by being separated from the adult workplace throughout early childhood to adolescence.
Previously, preschool practices in Japan have been discussed as having several features. According to Lewis (1995), Japanese pre-elementary schooling is diverse and nearly universal. One caretaker takes care of a large number of children. In East Asian countries, including Japan, the class sizes are generally large. Hayashi and Tobin (2014) discussed that in cases where the emphasis was on the class as a community, low child-caretaker ratios would be viewed as contributing to the development of sociality.
Here, I would like to point out two features that appear to be related to the development of children's peer cooperation. First, in Japanese pre-elementary and elementary schools, children in a classroom are divided into several fixed small groups. With group members, children participate in various daily activities. Typically, members remain constant for at least several months (Lewis, 1984). By working frequently with the same peers, children are expected to learn about one another and respond to the influence of others (Lewis, 1995, p. 83). Second, based on the group system, children regularly engage in monitoring activities, called "tōban-katsudō." In Japan, children perform various activities other than studying, such as cleaning their classrooms by themselves or taking care of animals that are raised in schools. It is common for all children to engage in these activities in rotation (Lewis, 1995). Children in the same group work together and take responsibility for monitor's activities. Repeated long-term participation in monitoring activities with the same members helps children understand their peers' intentions, affect, and needs (Lewis, 1984), and would accelerate collaboration with each other. The current study focused on lunch monitors' activities in a preschool in Japan and aimed to examine these points.

School lunch in Japan
In Japan, school lunches started in 1889 and have a history of more than 100 years. In 2018, 99.1% of elementary schools in Japan provided school lunches (Ministry of Education, Culture, Sports, Science and Technology, 2018). For preschools, whether they receive lunch depends on the type of school. In nursery schools ("hoikuen"), which take care of children who have working parents, usually have food facilities that provide lunches, while in pre-elementary schools ("youchien"), it depends on each school whether they provide lunches or children bring their own lunch boxes.
Children are involved in setting up school lunches as monitors' activities. Lunch monitors (kyūsyoku-tōban) are selected and rotated on a weekly or daily basis to serve lunches to other children. The main task of lunch monitors is to carry items of meal sets to class members. Meal sets usually include staple food items (e.g., rice, bread, and noodles), soup, a main dish (e.g., meat, fish, and soybean products), and a side dish (vegetables). Arrangements of these items are customarily fixed. Thus, lunch monitors are supposed to place each item in their correct position. To accomplish this, lunch monitors help each other out and share their responsibilities.
The current study examined the questions of how young children learn skills to serve meals by engaging in their duties as lunch monitors and how they share tasks with each other to complete the activity. For this purpose, six-month long observations of two classes in a nursery school in Tokyo, Japan, were conducted. Observational data were analyzed by two analyses: Analyses A and B. Before explaining these, the methods of observations and data collection will be explained.

Observation
We visited a nursery school in Tokyo, Japan, nearly once a week, for six months, from February to July 2019. This school had children between the ages of zero and five, and children of ages three to five were mixed in the same classes. We observed two multiage classes of children above three years. The total number of days over which observation was done were 22 and 17 days for Class A and B, respectively. The classes comprised 22 and 23 children for class A and B respectively, with an almost equal number of three-, four-, and five-year-old children. For both classes, two full-time and one part-time caretaker took care of the children. Lunch monitors' interactions during serving time were videotaped by two cameras facing different directions. Sketches of views from these cameras are shown in Figure 1.

Lunch monitors' activities
In Japan, the academic year begins in April and ends in March. In the selected nursery school, the oldest children were supposed to take the role of lunch monitors. The oldest children (five-yearold) and the next oldest children (four-year-old) engaged in lunch monitor activities together for the takeover during the last two months of the academic year, that is, in February and March. By engaging in lunch monitors' activities, the next oldest children, called "junior monitors" were expected to learn lunch monitors' skills as apprentices from the oldest children, called "senior monitors." Lunch monitors first washed their hands, put on aprons and clothes to cover their heads ( Figure 2), wiped tables, and placed luncheon mats on the tables. As shown in (Figure 2), the apron had a slightly large button on the right side at chest height. Caretakers told children to grasp the button by their hands if they could not discriminate between left and right, "This button tells you your right side." First, caretakers and lunch monitors carried food by a cart from the cooking room to the nursing room where children ate lunch. To provide food while it was warm, the food was carried in pots, and in the nursing rooms, the caretakers served food from the pots. There were about six to eight tables in each class, and the time to begin eating varied  for each table. The caretakers did not serve food for all tables at one time to prevent food from cooling. Typically, they served food for two tables. After serving food on a high cupboard, the caretakers placed the food on a low table for distribution ( Figure 1). Next, lunch monitors carried the food and tableware from the serving table to the tables for eating. There were always five items in the meal set, that is, tableware (fork and spoon for 3-and 4-year-olds and chopsticks for 5-year-olds), tea, rice, the main dish, and soup ( Figure 3). Caretakers placed these items in a fixed order, that is, tea, tableware, main dish, rice, and then soup, so that lunch monitors followed this order. The positions of each item were fixed ( Figure 3). The lunch monitors had to place each item in the correct position.
The entire observation period was divided into three intervals for every two months. The 1 st period corresponded to the last two months of the academic year. During this period, the next oldest children (four-year-olds) learned lunch monitors' work as apprentices with the oldest children (five-year-olds) who already had one year of experience as lunch monitors. The 2 nd period began in April, that is, the beginning of the academic year. Without senior lunch monitors, the children who had just become the oldest in the preschool started to engage in lunch monitors' activities. After two months, the 3 rd period began.
During the observations, 19 children were observed at least once. Among them, 13 children were junior monitors and six were seniors from the 1 st period. Since the current study aimed to examine the development of children's skills and peer cooperation, six seniors and three junior monitors who were not observed throughout the three periods were excluded from the analyses. Eventually, 10 children were included in the study. Table 1 shows the frequencies of observations for each child. As (Table 1) shows, total days of observations were five for the 1 st period, eight for the 2 nd period, and nine for the 3 rd period. There were four teams: teams (a) to (d). When a lunch monitor was absent from school, it was usual for the other monitor(s) in their team to take part in the job. However, there was only one exception, in which a child who was not a member of a given team joined as a substitute.

Analysis A: development of individual skill
Analysis A examined the development of each child's serving skills. Right-left recognition is still in progress in preschool years (Rigal, 1994), thus accurate placement would not be easy for four-and five-year-old children. As indices of development of lunch monitors' skills, the rates of correct placement, direction of body, and cues of placement were used. With experiences as lunch monitors, they  would start to place items more accurately. This is the first assumption examined in this study. Second, placing items in correct positions would be difficult because it would require children to rotate mental images. But with experience, they would be able to place items from unusual directions. Third, as children become proficient in lunch monitors' tasks, they would need to rely less frequently on cues.
That is, their serving behavior would become more independent. Analysis A examined these assumptions.

Data coding
Behavioral data were scored through videotaped recordings using ELAN_5.2.0.
(1) Extraction of parts to be coded: From the video, the part in which the lunch monitors carried items for the first and second tables was extracted. After they finished setting the first two tables, other children who were seated on these tables began to eat. Since there were many children in the video frames it was difficult to code the lunch monitors' behaviors after the children started eating. Therefore, the current study analyzed the placement behaviors of lunch monitors for the first two tables.
(2) Carrying episode: For coding, lunch monitors' behavior was divided into carrying episodes. The onset of a carrying episode was determined by the lunch monitor leaving from a serving table with the items to be carried. The end of a carrying episode was marked when they placed the items on a table for eating.
(3) Accuracy: For each carrying episode, when a lunch monitor placed an item in the correct position, a correct placement was coded. Correct positions for each item are shown in Figure 3. When a lunch monitor placed an item incorrectly, incorrect placement was coded.
(4) Bodily direction: Lunch monitors' bodily directions when placing items were coded for each carrying episode: (a) from the front, (b) from the opposite, and (c) from lateral directions. The lateral category included placing items from the left and right.
(5) Cues: When placing items, children sometimes explicitly relied on cues. These cues were divided into three categories: (a) the button of an apron, (b) already placed items, and (c) other persons. At the beginning of the takeover period, caretakers told new lunch monitors that their aprons had a large button on the right side at chest height and that the children could always discriminate left and right by grasping the button. Possibly owing to these suggestions, lunch monitors sometimes touched the buttons with their hands. It was also observed before placing items, lunch monitors explicitly watched items already placed in the next or nearby seats, and then proceeded to place the items. These behaviors were categorized as "already served items." The final category of "other persons" was more simple and direct. When lunch monitors asked other persons where to put items, "where should I put this cup of tea?" the category of "other persons" was coded.

Statistical analyses
This was a longitudinal study. Since longitudinal data are correlated, generalized estimating equations (GEEs) were used to analyze the development of individual skills (Analysis A). Two coders coded 100% of the data. Inter-rater reliability ranged from 86.1% to 93.6% (for all kappa values, p < .05). Disagreements between the coders were resolved through discussion.

Results
Averages of the total frequencies of carrying episodes for each child were 69.80 (SD = 35.94, range = -35-154) in the 1 st period, 149.10 (SD = 63.49, range = 37-261) in the 2 nd period, and 69.80 (SD = 106.40, range = 40-215) in the 3 rd period. Since the frequencies of observations were different among children (see , Table 1), the total frequencies of carrying episodes also differed for each child.
(1) Correct rates: Rates of the frequencies of carrying episodes in which children placed items in the correct positions without relying on explicit cues per total frequency of carrying episodes in each period were calculated. This calculation was performed for each item. Figure 4 shows the results for (a) tea, (b) tableware, (c) main dish, (d) rice, and (e) soup. The group means are also shown in this figure.
These data were examined through a 3 (period: 1st, 2nd, and 3rd periods) × 5 (items: tea, tableware, main dish, rice, and soup) GEE analyses. The GEE analyses yielded significant effects for the period, Wald χ 2 (2) = 39.68, p < .001. The correct rates in the 3 rd period (M = .91, SE = .02) were significantly greater than those in the 1 st (M = .47, SE = .07) and 2 nd periods (M = .73, SE = .07), with significant differences between those in the 1 st and the 2 nd periods (all p < .05). The mean correct rates for the senior lunch monitors in the 1 st period were .97 (SE = .01), which were equivalent to the results for junior lunch monitors' performance in the 3 rd period.
The GEE analyses also yielded significant effects of the items, Wald χ 2 (4) = 22.17, p < .001. The correct rates for soup (M = .82, SE = .05), main dish (M = .77, SE = .05), and tableware (M = .72, SE = .06) were significantly greater than those for tea (M = .55, SE = .06), without significant differences among the former three items (p > .05). The correct rates for rice (M = .65, SE = .04) were significantly lower than those for soup (p < .05). The difficulty levels of placing items in correct positions seemed to differ among items. Placing tea and rice appeared to be relatively more difficult than placing tableware, main dish, and soup. There was also a period × item interaction, Wald χ 2 (8) = 1011.94, p < .001. Significant differences among the three periods are shown in Figure 4 (a)-(e). For all items, the correct rates increased with experience, but the pace of improvement differed among items.
In total, frequencies of wrong placements were 57 in the 1 st period, 64 in the 2 nd period, and 22 in the 3 rd period. Among them, left and right errors reached 94% (135/143). In the other eight cases, children made mistakes with regard to near and far placements. These results suggest that it was difficult for young children to discriminate left and right while placing items on meal sets.
(2) Bodily direction: Next, the children's bodily directions while placing items were analyzed. In this analysis, there were no significant differences between the items. Therefore, the results of the items have been summed up. Figure 5 shows the total rates of carrying episodes in which children placed items from unusual directions per total frequency of carrying episodes in each period. A GEE was conducted to examine differences over time. It revealed that placing items from unusual directions in the 1 st (M = .22, SE = .04) and the 3 rd periods (M = .23, SE = .03) were significantly more frequent than those in the 2 nd period (M = .13, SE = .03), without significant differences between those in the 1 st and the 3 rd periods, Wald χ 2 (2) = 10.48, p < .01. The tendency to place items from unusual directions dropped in the 2 nd period but recovered in the 3 rd period. The senior lunch monitors' mean rate in the 1 st period was .21 (SE = .03), which was comparable for the junior lunch monitors' results in the 1 st and 3 rd periods.  (3) Cues: Figure 6 shows the rates of the frequencies of carrying episodes in which each child relied on cues per total frequency of carrying episodes in each period. A GEE was conducted to examine differences over time. It revealed that the rate of relying on cues in the 1 st period (M = .53, SE = .06) was significantly greater than those in the 2 nd (M = .19, SE = .05) and the 3 rd periods (M = .26, SE = .06), without significant differences between those in the 2 nd and the 3 rd periods, Wald χ 2 (2) = 29.28, p < .001. For senior lunch monitors, the mean rate of relying on cues in the 1 st period was .04 (SE = .02). This means that senior lunch monitors seldom relied on explicit cues.
Regarding the cues that children relied on, and the differences over time, the total rates of the frequencies of relying on each cue per total frequency of relying on cues were calculated. Figure 7 shows the results for each cue: (a) button of an apron, (b) already placed items, and (c) other persons. The GEEs were conducted to examine the differences over time for each cue separately. These revealed that the rates of using "button of an apron" in the 1 st period (M = .03, SE = .02) were significantly lower than those in the 2 nd (M = .26, SE = .08) and the 3 rd periods (M = .31, SE = .11), without significant differences between the 2 nd and the 3 rd periods, Wald χ 2 (2) = 7.65, p < .05. For "already placed items," there were no  Notes. ** p < .01, * p < .05.
, Cogent Education (2022), 9: 2070052 https://doi.org/10.1080/2331186X.2022.2070052 significant differences by periods, while for "other persons," the GEE yielded a significance, Wald χ 2 (2) = 19.07, p < .001. The rates of relying on "other persons" in the 1 st period (M = .51, SE = .08) and the 2 nd (M = .36, SE = .06) periods were significantly greater than those in the 3 rd (M = .21, SE = .05) period, without significant differences between those in the 1 st and the 2 nd periods. With experiences, lunch monitors came to rely on "button of an apron" more frequently but still less frequently than asking other persons.

Analysis B: collaboration process
Analysis A showed that some children liked relying on self-helping cues, while others tended to use social cues. The next analysis examined collaboration among different types of children. For this purpose, Analysis B first characterized each child based on their reliance on cue types and divided children into two groups, that is, self-helping and social, and examined how different types of children collaborated to work as a team. More specifically, when children relied on "already placed items" and "other persons," the questions "who had placed the items" and "to whom had the children asked" were examined. If they relied on "already placed items," the order of who placed items would be essential; if they liked to use "already placed items," he/she should not become the first person who placed the item. For this, the first server for each item was coded. Details of follow-up coding were as follows:

Data coding
(1) Social cues: When children relied on "already placed items," the person who had carried the items was coded. If a child explicitly watched an already placed item and the item had been placed by him/herself, self-placed was coded, while if the item had been placed by peer lunch monitors, peer-placed was coded. During the 1 st period, there were junior and senior lunch monitors. Therefore, for the 1 st period, senior-placed or junior-placed was coded. Concerning the cues of "other persons," the person to whom children had asked was checked.
(2) First server: The person who placed each item first for each table was coded. This coding was conducted separately for tea, tableware, main dish, rice, and soup. The cases where the caretakers placed the items were excluded from the analyses.

Statistical analyses
Analysis B intended to examine group differences, and therefore analysis of variance (ANOVA) and chi-square tests were used. Two coders coded 100% of the data. Inter-rater reliability ranged from 94.5% to 98.2% (for all kappa values, p < .05). Disagreements between the coders were resolved through discussion.

Results
(1) Dividing children by their reliance on cues: Figure 7(a) in Analysis A showed that there were four children whose rates of relying on "button of an apron" were above average in the 3rd period, and another six children's rates were below average. This index was used to divide children based on their reliance on cues. The former group was called the self-helping group, while the latter was called the Notes. ** p < .01, * p < .05.
, Cogent Education (2022), 9: 2070052 social group. To confirm group differences, the group mean rates of the frequencies of using each cue per total frequency of cues were calculated in each period. Figure 8 shows the results. Two (group: selfhelping and social) × 3 (period: 1 st , 2 nd , and 3 rd period) ANOVAs for each cue showed that relying on "button of an apron" was more frequent in the self-helping group, F(1,7) = 10.51, p < .05, η2 = .60 while relying on "other persons" was more frequent in the social group, F(1,7) = 7.52, p < .05, η2 = .52. For "already placed items," there were no group differences.
In this study, 10 children belonged to four lunch monitor teams (Table 1). In each team, there was one child in the self-helping group, and other members were in the social group. Regarding differences between the two groups in terms of their performance in placing items, Figure 9 shows the group means of the correct rates for each period. A 2 (group: self-helping and social) × 3 (period: 1 st , 2 nd , and 3 rd period) ANOVA did not yield a significant main effect for the groups. Concerning the accuracy of placing items, children in the self-helping and social groups were comparable.
(2) Social cues: Table 2 shows the total frequencies regarding from whom children took hints when they relied on "already placed items" and "other persons" in each group. Since children did not often rely on these cues, each child's results were summed up in each group. For "already placed items," Chi-square tests comparing the frequencies of referring to self-placed items and the frequencies of referring to peer-or senior-placed items by these two groups were conducted separately in each period. Significances were obtained for the 2 nd period, χ 2 (1, N = 44) = 20.46, p < .01, and the 3 rd period, χ 2 (1, N = 15) = 7.81, p < .01. Residual-analyses showed that children in the self-helping group relied more frequently on self-placed items and children in the social group relied more frequently on peer-placed items in both periods. For "other persons," there were no significant differences. As shown in Table 2, when children asked other persons, children in both groups often asked their caregivers, but seldom asked their peers.
(2) First server: Since there were senior-and junior-lunch monitors for the 1 st period, the first servers were divided into these two groups. Table 3 presents the results. To examine differences, binominal tests comparing the frequencies of senior and junior monitors were conducted separately for each item. Significant differences were observed for tea and rice (p < .05). For both, the senior lunch monitors placed these items first on each table more frequently than the junior lunch monitors. Next, for the 2 nd and 3 rd periods similar to the 1 st period, in order to examine group differences, binominal tests comparing the frequencies of children in the self-helping and social groups were conducted separately for each item in each period. Significant differences were obtained for tea in the 2 nd period (p < .01) and for rice in the 3 rd period (p < .05). For both, children in the self-helping group placed the items first more frequently than children in the social group. No significant differences were obtained for tableware, main dish, and soup in any period.

Individual development as a lunch monitor
Over a period of six months, five-year-old children were able to place items of school lunches with increasing accuracy. It was interesting that the level of accuracy was not the same among the items. Placing tea and rice in correct positions was more challenging for young children than placing tableware, the main dish, and soup. Since tea was the first item, there were no previous cues on the luncheon mat for relative placement. However, when placing other items, previously placed items could serve as cues and children could speculate the correct position relative to another object. To place rice, it was necessary to discriminate between the left and right. Since the placement of the rice was followed by the soup, there were fewer cues for guessing the correct position of rice than those of soup. All of these would relate to the differences in placement accuracy among the items.
As shown in the results, discriminating between left and right directions seemed to be difficult for young children. Therefore, placing items from unusual directions was challenging for them. However, placing items from unusual directions was more frequent in the 1 st period than in the 2 nd period. In the 1 st period, junior lunch monitors had just started their duties and were novices throughout the period. Why could new monitors perform such sophisticated serving behaviors? It is possible that working with senior lunch monitors was the key. In the 1 st period, junior monitors would be able to model the placement according to the correct placement of senior monitors and could have asked questions to their seniors. All these factors would have contributed to the unusual serving of novice lunch monitors. After the senior lunch monitors graduated from the school, placing items from unusual directions became less frequent. This suggests that children tried carrying items more carefully and independently in the 2 nd period.

Working as a team
However, at the same time, there were individual differences in their tendency to use certain cues. Some children tended to use self-helping cues, while others were more likely to rely on social cues. Role sharing was observed between these different types of children. The children in the self-helping group placed items first more frequently than children in the social group. Such division of labor was evident for tea and rice, which the children least succeeded in placing in correct positions. Every day small group activities with the same members would have enabled children to discover one another's strengths and weaknesses, and would have contributed to building complementary relationships. Previously, Goldberg and Maccoby (1965) showed that second graders performed more effectively after they had been trained with the same members than with constantly changing members. Consistent membership might play a key role in children learning a lot through peer interaction.
However, it seems unlikely that any combination of children could build cooperative or complementary relationships in a short time. Concerning peer interaction at school, teachers' strategies for building groups would be a key factor in the success of group performance. In Lewis (1995), which examined Japanese preschool and elementary school education based on observations, one preschool teacher explained how they formed small groups: "I put good speakers with poor speakers, children who like to help others with children who are slow. You need a mix of abilities and a mix of outgoing and more inward children." (p. 81) As shown in these responses, teachers' careful design while building groups would complement the groups' performance.

Conclusion
Over a period of six months, all five-year-old preschooler became skilled lunch monitors. There were individual differences regarding the cues on which the children relied while placing items. Some children liked to use self-helping cues, while others relied on social cues. Role sharing was observed between these different types of children. Based on the results, it was discussed that day-to-day activities with the same members and teachers' careful arrangement of group members would have contributed to the children's performance as a team.