Sustainable language education necessitates scalable, accessible learning environments that foster long-term learner autonomy and reduce educational inequality. While online courses have democratized access to language learning globally, persistent deficiencies in instructor-student interaction and learner engagement compromise their sustainability. The “face effect,” denoting the influence of instructor facial appearance on learning outcomes, remains underexplored as a resource-efficient mechanism for enhancing engagement in digital environments. Furthermore, effective measures linking psychological engagement to sustained learning experiences are notably absent. This study addresses three research questions within a sustainable education framework: (1) How does instructor identity, particularly facial appearance, affect second language learners’ outcomes and interactivity in scalable online environments? (2) How can digital human technology dynamically personalize instructor appearance to support diverse learner populations in resource-efficient ways? (3) How does instructor identity influence learners’ flow state, a critical indicator of intrinsic motivation and self-directed learning capacity? Two controlled experiments with Japanese language learners examined three instructor identity conditions: real teacher identity, learner self-identity, and idol-inspired identity. Results demonstrated that the self-identity condition significantly enhanced oral performance and flow state dimensions, particularly concentration and weakened self-awareness. These findings indicate that identity-adaptive digital human instructors cultivate intrinsic motivation and learner autonomy, which are essential competencies for lifelong learning. This research advances Sustainable Development Goal 4 (Quality Education) by demonstrating that adaptive educational technology can simultaneously improve learning outcomes and psychological engagement in scalable, cost-effective online environments. The personalization capabilities of digital human instructors provide a sustainable pathway to reduce educational disparities while maintaining high-quality, engaging instruction accessible to diverse global populations.
Depression represents a global mental health challenge requiring efficient and reliable automated detection methods. Current Transformer- or Graph Neural Networks (GNNs)-based multimodal depression detection methods face significant challenges in modeling individual differences and cross-modal temporal dependencies across diverse behavioral contexts. Therefore, we propose P$^3$HF (Personality-guided Public-Private Domain Disentangled Hypergraph-Former Network) with three key innovations: (1) personality-guided representation learning using LLMs to transform discrete individual features into contextual descriptions for personalized encoding; (2) Hypergraph-Former architecture modeling high-order cross-modal temporal relationships; (3) event-level domain disentanglement with contrastive learning for improved generalization across behavioral contexts. Experiments on MPDD-Young dataset show P$^3$HF achieves around 10\% improvement on accuracy and weighted F1 for binary and ternary depression classification task over existing methods. Extensive ablation studies validate the independent contribution of each architectural component, confirming that personality-guided representation learning and high-order hypergraph reasoning are both essential for generating robust, individual-aware depression-related representations. The code is released at https://github.com/hacilab/P3HF.
Speech emotion recognition (SER) infers affective states from speech signals, but positional encoding for acoustic tokens remains underexplored in Transformer-based SER. Existing models often reuse encodings designed for text and do not explicitly account for the different sequential and two-dimensional structures of acoustic representations. We propose Fibonacci Position Embedding (FPE) and Fibonacci Target Shutter (FTS). FTS constructs overlapping candidate-index sets over time–frequency token grids, and FPE samples a Fibonacci index and applies its modulo-wrapped, dimension-dependent phase rotation to query and key vectors. The modules are integrated into STRE-Former, which fuses Wav2Vec, log-mel spectrogram, and MFCC representations through asymmetric cross-representation attention with representation-specific positional encodings. We also introduce an implementation-consistent conditional-entropy formulation that quantifies uncertainty in recovering a token location from its sampled positional representation; this quantity characterizes positional ambiguity rather than downstream modeling capacity. Experiments over 64 positional-encoding combinations on IEMOCAP and MELD identify dataset-dependent highest-observed configurations, reaching 74.21% weighted accuracy on IEMOCAP-4, 74.54% on IEMOCAP-6, and 49.44% on MELD. These empirical observations suggest that the relative behavior of positional-encoding strategies may depend on the acoustic representation and evaluation dataset, rather than supporting a single universally optimal scheme.