Analyzing sequential categorical data on dyadic interaction: A latent structure approach