Humanitext Reader is a multilingual parallel reader for the Greek and Latin classics. It presents the original text alongside translations in four languages (Japanese, English, Chinese, and Korean), sentence by sentence, with dictionary look-ups and notes on grammar and syntax. More than a reading tool, it is a participatory library where readers improve the texts together through discussion and error reports.
https://humanitext.ai/apps/reader
What is Humanitext Reader?
Humanitext Reader is an online library for reading the classics of ancient Greek and Latin. It is built for readers who cannot read the originals directly: the source text and translations in four languages — Japanese, English, Chinese, and Korean — are displayed side by side, one sentence at a time, with notes on syntax and usage. Beyond reading, it is a participatory library that improves its texts through dialogue among readers and through error reports.
One idea runs through everything: to open the classics to everyone.
- Bringing you closer to the original words. By setting the source text sentence by sentence beside its translations and notes, with a dictionary and morphological analysis a click away, even readers who have never studied Greek or Latin can engage with the originals themselves. AI is a bridge to the source text, not a replacement for it.
- Every AI-generated part clearly marked. The translations and notes are a developing text — they begin as an AI-generated draft and are revised over time by curators. They are a starting point, not an authority, and reading them alongside the originals and established translations is encouraged.
- A new entry point to under-translated classics. Especially in Japanese, Chinese, and Korean, parallel translations cover only a small fraction of the canon. Through translations and notes in four languages, Reader opens a door to works long closed off by the language barrier.
- A participatory community that grows with its readers. Through error reports and cross-language comments, readers refine the texts together in dialogue — tending the very place where the classics are read as a shared good.
Current Scale
The library currently holds 921 works by 93 authors — roughly 1.8 million sentences across all languages, in four display languages. The originals are Greek (grc) and Latin (lat); the display languages are Japanese, English, Chinese, and Korean.
Reading
- The original and its translation are shown in sentence-by-sentence correspondence. Switch the display language between Japanese, English, Chinese, and Korean from the header.
- Dictionary pop-ups: click a word for its definition and morphological analysis (word form).
- Grammar and usage notes: gathered beneath the parallel text via footnote numbers.
- Alignment highlighting: hover over the translation and the corresponding part of the original lights up. The alignment is generated by AI across the entire text and is not guaranteed to be perfectly accurate — differences in word order and syntax can cause it to drift.
Dialogue (Comments)
- Comment on any sentence; commenting highlights the corresponding sentence in the original.
- Your post is translated by AI into all four languages and gathered into a single cross-language thread. Replies and “likes” are supported. Comments are anchored to the source by span overlap, not sentence id.
- Type tags: each new top-level comment carries one or more content tags (translation / interpretation / context / intertext / grammar / annotation / other). Replies inherit the thread’s tags automatically.
- Citable permalinks: every comment has its own ID with ”🔗 link” and ”❝ cite” actions, so it can be referenced individually from papers (
/read/<urn>?c=<comment_set_id>). - Posting requires a Google sign-in and agreement to the data-handling terms.
Error Reports (objective errors in the original only)
- Click a sentence → ”⚑ Report an error.” Sign-in and consent are required (anonymous reporting has been removed); the report is credited to the reporter’s display name.
- Reports are routed to a moderator and never auto-applied. Their status (open / in progress / resolved) is shown, and others can lend support with ”♥ same issue.” Adopted fixes are recorded in a public changelog.
- Reporters are credited for their contribution, with a +10 score when a report is adopted.
Contribution and Community
- Your profile shows your record (adopted fixes, reports, comments, likes received), plus badges and a score. Score weights: adopted reports/fixes ×10, reports ×1, comments ×2, likes ×1.
- A cooperative leaderboard (“Community”) and notifications (a bell, toggleable by type) round out the participatory features.
Original-Text Sources and Licensing
The originals are based on canonical TEI corpora such as the Perseus Digital Library and Open Greek and Latin, with the source and original licence (CC BY-SA 3.0 or 4.0) shown on each work page. In keeping with ShareAlike, the original texts on this site are also provided under CC BY-SA 4.0. The translations and notes in each language are AI-generated artifacts, separate from the original text.
Humanitext Reader is a new way into the world of the classics — read together, one sentence at a time, and made better by the community that reads it.
