Translation laboratory
Compare 3 public-domain English editions side by side. Enter a verse or range, or find which verses render an English word.
Try 2:255, 112:1-4, or the word “mercy”.
How this was computed
- Corpus:
- v0.8.0
- Editions compared:
- 3 (public-domain English)
- Alignment:
- verse-level only (no word-level translation alignment exists)
Divergence method. For each verse, every edition’s text is lowercased, stripped of punctuation and stopwords, and light-stemmed to a set of significant word-stems. We take the minimum pairwise Jaccard similarity across editions; a verse is flagged materially divergent when that minimum falls below 0.40 (editions share under ~40% of significant vocabulary). This is a lexical measure — it captures both substantive and stylistic difference — so the per-pair score is always shown. With these early editions most verses diverge lexically; the measure sharpens as more, more-modern editions are added. The editions, their licences and the full method are recorded in the dataset’s translations report.