Answers are grounded in this paper and cite their sources.
When the Qing ceded Taiwan to Japan in 1895, the island's classical poets faced a second fall of a realm within living memory of the first, the Ming collapse that had driven loyalists across the strait to Koxinga's Taiwan. This paper asks whether that second rupture left a measurable trace in the verse itself. Working only from openly available texts and an auditable lexicon of dynastic–loss and loyalist allusion (黍離, 採薇, 文天祥, 屈原, and the local 延平–東寧 tradition), we score the density of loyalist allusion across a corpus dated by era and find a clear bimodal shape: loud at the Ming–Zheng founding, nearly silent through the mid-Qing, and rising again into the cession years. Against the stable mid-Qing baseline the rupture-era density is roughly three times higher (0.087 → 0.268 per hundred characters, p < 0.0001), while a frequency-matched set of neutral allusions stays flat, so the rise is specific to loyalism. The surge is carried by the classical canon of martyrdom and refusal-to-serve (採薇/文天祥/田橫/蘇武, p = 0.0065), not the local Koxinga vocabulary, and survives every detector variant. Two independent checks support it: an officially compiled gazetteer reproduces the mid-Qing baseline almost exactly (p = 0.71), and a classical-Chinese language model that never saw our lexicon separates loyalist from control poems well above chance (0.70 vs. 0.51, p < 0.0001). The one corpus that would settle editorial selection at the peak is no longer reachable; we mark that as the clear next step rather than claim it.
No takes yet. Share an insight, caveat, or question.
Aryan Maity (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: