Identifying and Manipulating the Personality Traits of Language Models