The Austronesian Game Taxonomy: A cross-cultural dataset of historical games

Play, particularly rule-based games, is a significant aspect of human culture and development. While play is widespread across cultures, research on the structure and content of games is limited. Previous studies often focused on few cultures or lacked data accessibility, hindering generalizability and replicability. Moreover, the prevailing definition of games as competitive interactions has overlooked the diversity of cooperative and solitary games. This study addresses these limitations by providing a comprehensive dataset of games from Austronesian-speaking groups, known for their linguistic and cultural diversity. The Austronesian language family's shared linguistic ancestry and extensive ethnographic documentation make it ideal for studying game distribution and function. This dataset facilitates cross-cultural comparisons of game cooperativeness and provides a foundation for broader investigations into human behavior and cultural evolution.

Existing literature on cross-cultural games often suffers from limitations in sample size and data accessibility. Many studies focus solely on competitive games, neglecting cooperative or solitary forms. While some research suggests a link between game cooperativeness and cultural levels of cooperation, the small number of cultures studied restricts generalizability. A commonly used definition of games emphasizing competition has limited understanding of cross-cultural variation in game structure and function. The lack of large, accessible datasets has hampered broader comparative research on the role of games in human development and cultural evolution.

The Austronesian Game Taxonomy dataset was compiled from four primary sources: the eHRAF, the Pulotu website, and the *Journal of the Polynesian Society* and *American Anthropologist*. Additional sources were also included. A total of 1738 sources were examined, with 219 yielding game-related information. Games were defined using Whittaker's criteria, encompassing competitive, cooperative, and solitary forms. A new coding scheme for game cooperativeness was developed, categorizing goal structures into five types: Solitary, Competitive, Competitive vs. Solitary, Competitive vs. Cooperative group, and Cooperative group. Each game description was assessed for information on location, language, time frame, and goal structure. Cultural group identifiers were assigned using ABVD codes, Glottolog codes, and ISO 639-3 codes. Record linkage was employed to identify and merge descriptions of the same game. Optional filtering steps were incorporated using an R package to allow researchers to tailor the dataset based on their research questions. Inter-rater reliability was assessed for goal structure coding (κ = 0.94) and introduced game coding (κ = 0.487 for the original coding and κ = 0.808 for the binary version). Time frame data was gathered from the original sources and matched against the Pulotu database. The ABVD codes were also matched to the Austronesian language phylogeny by Gray et al. (2009) to enable phylogenetically informed analyses.

The dataset includes information on 907 games from various Austronesian-speaking cultural groups. After applying several filters (detailed in Table 5), 452 games from 55 ethnolinguistic groups were used for analyses focused on the Austronesian language phylogeny. Figure 2 visualizes the distribution of goal structures across these cultural groups. Competitive (n=228) and cooperative group vs. cooperative group games (n=121) were the most common. The distribution of game cooperativeness varied considerably across cultural groups. The dataset's structure allows researchers to explore various research questions, for example: examining the relationship between game types and cultural characteristics, investigating the role of games in children's social learning, studying the spread and evolution of games across Oceania and comparing the prevalence of different types of games across cultures with varying degrees of social complexity.

This study provides a large, accessible dataset addressing prior limitations in cross-cultural game research. The diverse range of games, the detailed coding, and the provision of various filtering options facilitate numerous research avenues. The findings highlight the substantial variation in game cooperativeness across different cultural groups. The dataset supports comparative analyses investigating relationships between cultural contexts and game characteristics. Future research could explore the impact of factors such as social structure, environmental pressures, and historical influences on game evolution and distribution. The findings contribute to our understanding of the role of games in shaping cultural norms, social interaction, and cognitive development.

The Austronesian Game Taxonomy provides a unique and valuable resource for cross-cultural research on games. Its size, accessibility, and detailed coding allow researchers to investigate various questions related to game evolution, cultural transmission, and the relationship between games and other aspects of culture. Future studies could expand the dataset, incorporate additional coding schemes, and analyze the data using phylogenetic methods to further enhance our understanding of human behavior and cultural diversity.

The dataset is not exhaustive, representing a sample of games from Austronesian-speaking groups. The reliability of the ‘introduced’ game coding was moderate, possibly due to inconsistencies and ambiguities in original source descriptions. The availability of ethnographic data varies across cultures within the dataset. These limitations may influence the generalizability of findings, but the dataset still provides a significant resource for comparative research.