Logo Lanfrica

MedikaLtd/Shona-language

Domaine:

natural language processing

Type de record:

softwaretools
Créateur:
Med
Hôte:
Free Shona spellchecker, autocorrect and speech tools - bringing ChiShona into the digital era # Shona Language — Digital Infrastructure for ChiShona 🇿🇼 **Bringing ChiShona into the digital era.** Spellcheck, autocorrect, and speech recognition for 10+ million Shona speakers — free, open source, MIT-licensed. > Type Shona without red squiggles. Autocorrect that suggests *Shona*, not English. > No more "language detected: Indonesian". And a path to live Shona captions on TV. ## 🚀 Install the spellchecker today **LibreOffice / OpenOffice (free):** Download `installers/shona-spellcheck-0.2.oxt` and double-click it. Then set your text language: *Tools → Language → For All Text → More → Shona*. **Microsoft Word (Windows):** 1. Download `installers/ChiShona-Word.dic` (214,677 Shona word forms) 2. Copy it to `C:\Users\ \AppData\Roaming\Microsoft\UProof\` 3. In Word: *File → Options → Proofing → Custom Dictionaries → Add*, select it, set *Dictionary language: All languages* 4. To stop autocorrect mangling Shona: *File → Options → Proofing → AutoCorrect Options* → untick *Replace text as you type* while writing Shona **Firefox / Chrome / Linux / macOS:** copy `dictionaries/sn_ZW.aff` and `sn_ZW.dic` into your system Hunspell folder (`/usr/share/hunspell/` on Linux, `~/Library/Spelling/` on macOS). ## 🧠 How it works Shona is agglutinative — one verb stem produces hundreds of valid words (*ndinotenda, vachatenda, handitendi, ndinokutenda, kutendwa...*). A flat wordlist can never keep up, so this project encodes the **morphology itself** as Hunspell rules: | Layer | Example | Status | |---|---|---| | Subject concords + tense (78 complexes) | **ndicha**enda, **vaka**uya | ✅ v0.1 | | Object concords (two-level prefixes) | ndino**ku**da, vaka**ndi**udza | ✅ v0.2 | | Verbal extensions (-ir/-is/-w/-an/-ik/-isw) | kubatsir**an**a, akashand**isw**a | ✅ v0.2 | | Negative paradigm | **handi**tend**i** | ✅ v0.2 | | 603 curated seed stems → | **214,677 recognised forms** | growing | ## 🛠 Tools | Script | Purpose | |---|---| | `tools/build_wordlist.py` | Feed it any Shona …