English Spell Checker Dictionaries
SCOWL is the basis for the official English dictionary in Aspell and the en_US, en_CA and en_AU dictionaries in Hunspell. Dictionaries for British English are also available for Hunspell.
Hunspell Dictionaries
The latest release is: 2020.12.07 (readme)
- American: en_US, en_US-large
- Canadian: en_CA, en_CA-large
- Australian: en_AU, en_AU-large
The default dictionaries correspond to SCOWL size 60 and, to encourage consistent spelling, generally only include one spelling variant for a word. The large dictionaries correspond to SCOWL size 70 and include common spelling variants. The larger dictionaries, however, (1) have not been as carefully checked for errors as the normal dictionaries and thus may contain misspelled or invalid words; and (2) contain uncommon, yet valid, words that might cause problems as they are likely to be misspellings of more common words (for example, “ort” and “calender”).
British English dictionaries are also available, however they are not considered the official dictionaries for Hunspell. The official ones are maintained by Marco A.G.Pinto at proofingtoolgui.org. Marco’s dictionaries are based on David Bartlett’s now abandoned version. They likely have better coverage of British words and fewer Americanisms. Marco’s dictionaries are also considerably larger than the ones based on SCOWL, and include variant spellings.
- British, -ise (alternative version, traditional spelling): en_GB-ise
- British, -ize (alternative version, Oxford spelling): en_GB-ize
- British large (alternative version, both spellings): en_GB-large
Aspell Dictionaries
The official English speller dictionary for Aspell includes all dialects and is available on the GNU FTP server: ftp.gnu.org/gnu/aspell/dict/en/.
Custom Dictionaries
If none of these dictionaries are suitable to your needs you can create your own from the information in SCOWL using this simple tool.
Contributing
Suggestions for new words are always welcome. I now use ChatGPT (and other LLMs) to help screen words (see #429). To get an idea if a word will be accepted you can use the SCOWL Word Evaluation GPT. LLMs’ analysis is a significant data point, but only one of many factors considered when adding words.
To suggest new words simply create an issue at https://github.com/en-wl/wordlist/issues. If you have multiple related words to suggest please include them in a single issue as it is less work for me to evaluate batches of words than one word at a time.
The old app that checked if a word is in SCOWL is retired. If you want to check if a word is in the latest version use the wordlist-diff repo. This repo contains generated wordlists and the combined scowl.txt file for each commit to the main repo.
SCOWL is in no way perfect, if you find words that don’t belong or other inconsistencies please feel free to report them as an issue.
General feedback is also welcome, you can either email me directly at kevina@gnu.org, or preferably start a public discussion at https://github.com/en-wl/wordlist/discussions.