For example, text is typically preprocessed by tokenization.