About CorpusForge
CorpusForge is a corpus-linguistics word game. A corpus is a big pile of real text, and corpus linguistics is the study of what that pile actually contains — which letters, sounds, and words appear, and how often. That single question, how often?, quietly decides how every tile game is scored.
What you’re really learning
- Frequency → fairness. Rare units earn more points. Setting a fair bag is reading a frequency distribution.
- Zipf’s law. In any language a few units carry most of the load — the commonest English word, “the”, is about one word in fourteen.
- A tile isn’t a byte. Vietnamese NG, CH, and TH are single graphemes; the single Cyrillic letter щ romanizes to four Latin letters. A bag must be built from the language’s own writing system.
- Rarity has two faces. How often a word appears, and how late we learn it (age-of-acquisition) — a fair bag weighs both.
The synergy
The bag you build isn’t make-believe. A tile bag is exactly the data our other word games (and their kid dictionary) need to score fairly across seven languages. Doing corpus linguistics here produces the same kind of artifact a real language-technologist ships — student work that could become real infrastructure.
What it’s good for (and what we don’t claim)
CorpusForge is good practice for frequency and distribution, Zipf’s law, and thinking about writing systems across languages. It is not a brain-training program, and we make no claim it raises IQ or test scores. The frequencies shown are rounded, illustrative figures — directionally true, chosen to make the lesson clear.
Safe by design
- On your device. Your progress stays in your browser — no account, no sign-in, nothing sent anywhere.
- No dark patterns. No timer, no streak to lose, no scarcity — just the puzzle.
- Units we chose carefully. Every tile is a plain letter, grapheme, or common safe word.