The study examines how race moderates the relationship between depression and language use on social media. Using a matched sample of Black and White English speakers in the United States, the researchers found that while depression severity predicts increased use of first-person pronouns ("I-usage") in White individuals, it does not in Black individuals. Additionally, White individuals tend to use more negative emotions related to belongingness and self-deprecation. Machine learning models trained on White individuals' language performed well when tested on White participants but poorly on Black participants, even when trained exclusively on Black individuals' language. The study highlights the need for better understanding of race-based differences in the expression of depression in natural language and the importance of ensuring racial invariance in language-based models for detecting psychological phenomena before integrating them into clinical practice.The study examines how race moderates the relationship between depression and language use on social media. Using a matched sample of Black and White English speakers in the United States, the researchers found that while depression severity predicts increased use of first-person pronouns ("I-usage") in White individuals, it does not in Black individuals. Additionally, White individuals tend to use more negative emotions related to belongingness and self-deprecation. Machine learning models trained on White individuals' language performed well when tested on White participants but poorly on Black participants, even when trained exclusively on Black individuals' language. The study highlights the need for better understanding of race-based differences in the expression of depression in natural language and the importance of ensuring racial invariance in language-based models for detecting psychological phenomena before integrating them into clinical practice.