Temporal Variability and Stability in Infant-Directed Sung Speech: Evidence for Language-specific Patterns