This involves building statistical models that capture the patterns and relationships that exist in natural language data.