2026 Volume 38 Issue 4 Pages 720-729
Estimating transmission lineages of Japanese Lion Dance (shishimai) has become increasingly difficult because of the lack of historical records and the aging of tradition bearers. This study compares multiple distance functions for estimating transmission lineages using only extant motion data. Motion sequences obtained from video analysis and motion capture were analyzed using cosine dissimilarity based on n-gram frequency vectors, Levenshtein distance, and Dynamic Time Warping (DTW). Lineage structures were inferred using the Neighbor-Joining (NJ) method, and topology stability was evaluated through bootstrap analysis by adding normally distributed noise to distance matrices. The results showed a strong correlation between video-based and motion-capture-based analyses when using the Levenshtein distance. In contrast, the n-gram-based approach achieved the highest stability and attained more than 99% confidence in the automatically classified data. These results indicate that local sequential movement patterns are effective for estimating transmission lineages.