Words function as gateways to our minds. The terminology we select — alongside the manner in which we articulate them — reveals significant insights about our character and potentially our emotional wellbeing, stated WashU psychologist Josh Oltmanns. “Our thoughts, emotions, and actions are mirrored in our language,” he noted.
Instead of putting individuals through an unending series of assessments, psychologists could derive crucial understanding from samples of language. However, they might require advanced technological support to identify the essential signals hidden among the noise.
AI technologies designed to recognize distinct markers in conversation could transform psychological evaluations, he remarked.
“Psychologists are human, and humans make mistakes, so even a skilled clinician may overlook significant verbal indicators,” stated Oltmanns, an assistant professor of psychological and brain sciences in Arts & Sciences at Washington University in St. Louis. “Conversely, a well-trained computer model will detect those signals.”
In theory, a psychologist could prompt a client to detail their life and issues, a routine component of a preliminary evaluation. By employing their own professional judgment, the psychologist could also input that dialogue into a system aimed at identifying personality characteristics and signs of psychological distress.
“The software could assist in affirming their observations or alerting them to aspects they might have overlooked,” Oltmanns mentioned.
Oltmanns is collaborating with fellow researchers, also PhD candidates at WashU, Tu Do, Tong Li, and Tongyao Ran, to create AI instruments that assist psychologists in revealing these concealed indicators within language. He recently articulated the promise of such technologies in the journal Advances in Methods and Practices in Psychological Science. Mehak Gupta from Southern Methodist University and Jocelyn Brickman from Xavier University joined him as co-authors.
Language possesses the ability to express psychological nuances in various ways. The selection of vocabulary is significant, whether in a profound dialogue or a casual update on X or Facebook. Early in his career, Oltmanns explored how language choices in social media posts generally reflected an individual’s Big Five personality dimensions: openness to experience, neuroticism, agreeableness, conscientiousness, and extraversion.
Additionally, the delivery of words is crucial. “You can glean quite a bit by observing how someone communicates,” Oltmanns stated. “Slow speech can indicate depression, while excessively fast speech correlates with anxiety.”
Rate is merely one metric. Spoken language can also differ in volume, tone, and pitch. “Speech samples encompass hundreds of distinct acoustic properties that may hold significance,” he added.
With so much potential data embedded in each discussion, psychologists have long sought computer assistance for speech analysis. Over two decades ago, researchers created Linguistic Inquiry and Word Count, a software capable of scoring individuals based on various psychological parameters derived from written language. Although those tools have evolved, the emergence of AI introduces a new realm of opportunities, according to Oltmanns.
“AI systems could surpass previous computer models in speed, comprehensiveness, and accuracy,” he mentioned.
Nevertheless, Oltmanns warned that AI poses its own challenges. “It is frequently trained on internet content, which can lead to biases,” he cautioned. If these biases remain unaddressed, there’s a possibility that certain cultural variances in speech could be misinterpreted as indicators of psychological issues.
To mitigate potential bias, AI systems should be developed using diverse patient cohorts. Toward that goal, Oltmanns is examining the extensive hours of interviews gathered over time through the SPAN Study, an ongoing examination of over 1,600 adults from St. Louis that reflects the city’s diversity.
“We aim to analyze speech patterns among white and Black participants to ensure the AI systems treat both groups equitably,” he expressed.
Oltmanns anticipates several further critical inquiries ahead. It remains uncertain how written language choices differ from those in verbal communication, or the quantity of words necessary to truly glean insights into an individual’s psychological state. “We possess numerous hypotheses and considerable work ahead,” he indicated.
Considering the rapid pace of innovation in the AI sector, he is hopeful of obtaining answers sooner rather than later.
“Companies are already marketing AI psychological evaluation tools to medical facilities and professionals, yet I doubt their efficacy or thoroughness in evaluation,” Oltmanns noted. “This type of technology could mark a significant leap in psychological practice, but must be approached cautiously. We need to be prudent.”
Originally published on the Ampersand website.
The post What do our words say about our minds? appeared first on The Source.