Ellipsis as informational compression in Vietnamese proverbs: A pragmatic and corpus-informed analysis

This article examines structured non-expression in Vietnamese proverbs—the patterned absence of otherwise expected yet interpretable elements—as a resource for informational compression. Rather than treating proverbial brevity as stylistic, it asks which forms of non-expression are supported by source-attested data under conservative diagnostics. Documentary extraction yielded 713 source occurrences (616 from Nguyễn Văn Ngọc's 1928 collection and 97 from Huỳnh Tịnh Của's collection, cover dated 1897; internal title dated 1896), reconciled to 705 distinct surface forms. The archive retains an aggregate candidate-stage count of 520 for reconciliation, but item-level lineage to the 705 surface forms and 312 records is unavailable, precluding use as an analytical denominator. The workbook contains 312 proverb-level documentary records: 225 facsimile-verified, proverb-supported Nguyễn forms constitute the primary analytical base, whereas 87 Huỳnh records are comparison-only and excluded from primary frequency analyses. Of the 225 primary forms, 154 contain at least one supported target locus, 35 show none, and 36 remain unresolved. Supported primary assignments comprise subject non-expression (54), argument omission (13), and interclausal relational underspecification (87); none meets the conservative criterion for predicate reduction in parallel structures. Counts describe this bounded sample; interpretability is assessed analytically, not through listener performance. A category-linked compression-density heuristic covers all 154 supported primary assignments (54 LOWER, 13 INTERMEDIATE, 87 HIGHER), but its distribution reflects coding coverage, not an independently validated processing or inferential gradient. The findings refine the framework by locating supported recovery in participant, argument, and discourse-relation domains and rejecting parallelism alone as grounds for a missing-predicate analysis.