Research
8 peer-reviewed publications, with first-author papers at EMNLP (opens in a new tab) and COLING (opens in a new tab).
Interpretable Topic Modeling of Spontaneous Speech in Depression Using Large Language Models (opens in a new tab)
We identify interpretable topics in spontaneous speech across four cohorts and examine how they relate to depression, to the questions that elicit them, and to demographics.
Formalizing Style in Personal Narratives (opens in a new tab)
We model narrative style as recurring patterns in how people describe experiences, found with language models and pattern mining, and linked to psychological observations.
Textualized and Feature-based Models for Compound Multimodal Emotion Recognition in the Wild (opens in a new tab)
We compare modeling visual and acoustic features directly with translating them into text for a language model. We placed third in the ABAW compound expression challenge.
Improving Language Models for Emotion Analysis: Insights from Cognitive Science (opens in a new tab)
We use theories of emotion and communication to identify what emotion-label prediction overlooks and propose directions for annotation, modeling, and evaluation.
Sequence-to-Sequence Language Models for Character and Emotion Detection in Dream Narratives (opens in a new tab)
Supervised sequence-to-sequence models (opens in a new tab) outperform in-context learning with a larger model at detecting characters and emotions. I released their DreamBank annotations (opens in a new tab).
Emotion Recognition based on Psychological Components in Guided Narratives for Emotion Regulation (opens in a new tab)
We introduce French autobiographical narratives (opens in a new tab) annotated for behavior, feeling, thinking, and context, and examine how these components contribute to emotion recognition.