Search for: All records

There has been growing interest in developing ubiquitous technologies to analyze adult-child speech in naturalistic settings such as free play in order to support children's social and academic development, language acquisition, and parent-child interactions. However, these technologies often rely on off-the-shelf speech processing tools that have not been evaluated on child speech or child-directed adult speech, whose unique characteristics might result in significant performance gaps when using models trained on adult speech. This work introduces the Playlogue dataset containing over 33 hours of long-form, naturalistic, play-based adult-child conversations from three different corpora of preschool-aged children. Playlogue enables researchers to train and evaluate speaker diarization and automatic speech recognition models on child-centered speech. We demonstrate the lack of generalizability of existing state-of-the-art models when evaluated on Playlogue, and show how fine-tuning models on adult-child speech mitigates the performance gap to some extent but still leaves considerable room for improvement. We further annotate over 5 hours of the Playlogue dataset with 8668 validated adult and child speech act labels, which can be used to train and evaluate models to provide clinically relevant feedback on parent-child interactions. We investigate the performance of state-of-the-art language models at automatically predicting these speech act labels, achieving significant accuracy with simple chain-of-thought prompting or minimal fine-tuning. We use inhome pilot data to validate the generalizability of models trained on Playlogue, demonstrating its utility in improving speech and language technologies for child-centered conversations. The Playlogue dataset is available for download at https://huggingface.co/datasets/playlogue/playlogue-v1. 

Emotion dysregulation in early childhood is known to be associated with a higher risk of several psychopathological conditions, such as ADHD and mood and anxiety disorders. In developmental neuroscience research, emotion dysregulation is characterized by low neural activation in the prefrontal cortex during frustration. In this work, we report on an exploratory study with 94 participants aged 3.5 to 5 years, investigating whether behavioral measures automatically extracted from facial videos can predict frustration-related neural activation and differentiate between low- and high-risk individuals. We propose a novel multi-scale instance fusion framework to develop EarlyScreen - a set of classifiers trained on behavioral markers during emotion regulation. Our model successfully predicts activation levels in the prefrontal cortex with an area under the receiver operating characteristic (ROC) curve of 0.85, which is on par with widely-used clinical assessment tools. Further, we classify clinical and non-clinical subjects based on their psychopathological risk with an area under the ROC curve of 0.80. Our model's predictions are consistent with standardized psychometric assessment scales, supporting its applicability as a screening procedure for emotion regulation-related psychopathological disorders. To the best of our knowledge, EarlyScreen is the first work to use automatically extracted behavioral features to characterize both neural activity and the diagnostic status of emotion regulation-related disorders in young children. We present insights from mental health professionals supporting the utility of EarlyScreen and discuss considerations for its subsequent deployment.