Havelock Orality Models - a HavelockAI Collection

HavelockAI 's Collections

updated Feb 6

BERT-based models for analyzing text on the oral–literate spectrum, operationalizing Walter Ong's framework from Orality and Literacy (1982).

HavelockAI/bert-orality-regressor

Text Classification • Updated Feb 17 • 57

Note Document-level regression model outputting a continuous 0–1 orality score.
HavelockAI/bert-marker-category

Text Classification • 0.3B • Updated 19 days ago • 42

Note Binary span classifier: oral vs literate. Top level of the classification hierarchy.
HavelockAI/bert-marker-type

Text Classification • 0.3B • Updated 19 days ago • 49

Note 25-class span classifier for functional marker families (e.g., repetition, subordination, direct_address). Mid-level of the classification hierarchy.
HavelockAI/bert-marker-subtype

Text Classification • 0.3B • Updated 19 days ago • 45

Note 70+ class span classifier for fine-grained rhetorical devices (e.g., anaphora, epistemic_hedge, vocative). Finest level of the classification hierarchy.
HavelockAI/bert-token-classifier

Token Classification • Updated Feb 16 • 20

Note BIO token tagger for span detection across 70+ marker types (145 labels). Identifies marker boundaries in running text. Uses focal loss for class imbalance.