You Are What You Say: Language Reveals Personality?

By: Reece Akhtar

Organizations have many tools at their disposal to identify and develop talented employees, such as psychometric assessments, structured interviews, and 360 performance appraisals. Although these tools are supported by a plethora of scientific research that demonstrates their predictive power and practical utility, they can be criticized for being time consuming, unengaging, and expensive. Fortunately, recent innovations in computing and psychological science have led to the creation of new, emerging technologies that have the potential to overcome these limitations and help organizations win the war for talent. One such innovation, Natural Language Processing (NLP)—using artificial intelligence to recognize, understand, and generate language—is fueled by a never-ending supply of data: every second we send more than two million emails, search Google 60,000 times, and post 8,000 tweets. Similarly, with the explosion of voice assistants such as Alexa, language has never been more readily created, shared, and logged. With all this chat, can we do a better job at understanding people?

The relationship between language and individual differences is as old as modern personality psychology. Originating from Freud’s psychoanalytic theory, the content, frequency, and style of language has served as the foundation of individual difference theory and measurement. This is because language is the most effective way for individuals to describe themselves, others, and the world—it is the only way we can attempt to know what is happening inside someone’s head. With the rise of digital communication and computational methods, NLP offers a new alternative to assessing an individual’s personality and values. For instance, a team of researchers from University of Pennsylvania analyzed the social media posts of over 60,000 people and were able to accurately predict their Big Five personality scores. These findings, bolstered by the historic tradition of language analysis, do in fact suggest you are what you say.

On Friday, April 20, at the annual conference of the Society for Industrial and Organizational Psychology, I will be presenting research on how we can use NLP to assess psychological constructs within a personnel selection context. I believe NLP offers an unobtrusive method for assessing applicants with the potential to reduce the biases that are often associated with qualitative assessments of unstructured interviews, resumes, and social media profiles. To support this position, we explored the extent to which language data can predict two personality inventories that are widely used by HR and I/O practitioners, specifically the Hogan Personality Inventory (HPI) and the Hogan Development Survey (HDS). Answering this research question not only extends our knowledge pertaining to the relationship between language and personality models beyond the Big Five framework (such as the dark side of personality), but may also have practical utility given the inventories’ prevalence in selection and development contexts.

To learn more about our research and its implications, please contact me at


Our insights and thinking, direct to your inbox.