This data probably best reflects a language acquisition curve rather than actual frequency of abuse.