Mirror of the source code for the Plains Cree morphological analyzer/generator.
tânisi! If you’re here, you want to use the FSTs (available here) to generate Plains Cree word forms from a linguistic analysis.
In English, aren’t “ate” and “eat” the same word? Yes, of course they’re the same word – they have the same dictionary definition! – but they differ in that one is past tense and one is present tense. These are two different word forms of the same word. Likewise in Cree, nôhkom and kôhkom are the same “word” but they differ in just whether it’s “my grandma” or “your grandma”. nôhkom and kôhkom are two separate wordforms.
It’s impractical to refer to a word by all its possible word forms. Imagine an English dictionary, that prior to defining a word, enumerated all its possible word forms. It would read “eat/eats/ate/eaten”, “break/breaks/broke/broken”, “drink/drinks/drank/drunk”, and similary for other words. These becomes even more impractical in Plains Cree, with some words having hundreds of forms, and even the ability to be creative with word forms! Instead we choose one wordform to represent them all. This is one word form to rule them all is called the lemma.
First install hfst:
# macOS
brew install ualbertaaltlab/hfst/hfst
# Ubuntu
sudo apt install hfst
Next, download a copy of “crk-normative-generator.hfstol”.
Cozy up to your command line and cd to where you downloaded the
*.hfstol file. We’re about to generate Cree words on the command line!
🧑🏽💻