{"slug": "the-divergence-hypothesis-unmasking-lexical-interference-and-label-bias-in-nlp", "title": "The Divergence Hypothesis: Unmasking Lexical Interference and Label Bias in Mental Health NLP", "summary": "A new diagnostic framework called TSS (Triple-Stream Stress probe) reveals that computational mental health classifiers suffer from lexical interference and label bias, with adding lexical features to the style channel reducing Macro-F1 on human-labeled data by a mean of 0.072 (p<10^-4) but not on auto-labeled data. The framework, introduced in an arXiv paper (2608.20353v1), proposes a Degree of Divergence (DoD) statistic for label-source auditing, with a headline estimate of DoD(BC-A) = 0.0374 (95% CI [0.0097, 0.0651], p=0.0032). Interventional masking retaining 95-99% of Channel C's performance on human datasets indicates the style channel does not rely primarily on lexical surface form, positioning TSS as a diagnostic audit framework rather than a clinical screening tool.", "body_md": "arXiv:2608.20353v1 Announce Type: new\nAbstract: Computational mental health (CMH) classifiers often degrade under distribution shift because human annotators and distant-supervision pipelines reward different linguistic signals. We introduce TSS (Triple-Stream Stress probe), a multi-channel diagnostic framework that decomposes text into (A) lexical character n-grams, (B) a small, mostly content-free morpho-syntactic channel, and (C) a 154-feature psycholinguistic style channel. Across four English datasets (N=12,906), TSS reveals a lexical interference effect: adding lexical features to the style channel reduces Macro-F1 on human-labeled data (mean drop 0.072, p<10^-4) but not on auto-labeled data. We propose Degree of Divergence (DoD), a difference-in-differences statistic adapted from econometrics for label-source auditing, with instance-level bootstrap inference; the headline estimate is DoD(BC-A) = 0.0374, 95% CI [0.0097, 0.0651], p=0.0032. A platform-stratified Twitter-only DoD (which removes the Reddit vs. Twitter contrast) reproduces the pattern with bootstrap inference: DoD-Tw(BC-A) = +0.096 (p<0.001) and DoD-Tw(AC-A) = -0.089 (p<0.001). Interventional masking (pos_only) retains ~95-99% of Channel C's performance after destroying content words on human datasets, indicating that the style channel does not rely primarily on lexical surface form. TSS is positioned as a diagnostic audit framework, not a clinical screening tool: it flags label-source-specific shortcut learning before generalization claims are made.", "url": "https://wpnews.pro/news/the-divergence-hypothesis-unmasking-lexical-interference-and-label-bias-in-nlp", "canonical_source": "https://arxiv.org/abs/2608.20353", "published_at": "2026-08-24 04:00:00+00:00", "updated_at": "2026-08-24 04:14:24.976825+00:00", "lang": "en", "topics": ["natural-language-processing", "machine-learning", "artificial-intelligence"], "entities": ["arXiv", "TSS", "Triple-Stream Stress probe", "Degree of Divergence"], "alternates": {"html": "https://wpnews.pro/news/the-divergence-hypothesis-unmasking-lexical-interference-and-label-bias-in-nlp", "markdown": "https://wpnews.pro/news/the-divergence-hypothesis-unmasking-lexical-interference-and-label-bias-in-nlp.md", "text": "https://wpnews.pro/news/the-divergence-hypothesis-unmasking-lexical-interference-and-label-bias-in-nlp.txt", "jsonld": "https://wpnews.pro/news/the-divergence-hypothesis-unmasking-lexical-interference-and-label-bias-in-nlp.jsonld"}}