I would like to introduce a project I have spent the last few weeks working on. Put simply, leŋoba is an IAL with the goal of using AI to rapidly produce a vocabulary the most representative of the most people and language families. I will briefly explain in this post, but see the documentation booklet for a more complete breakdown of the languages features and my reasoning for designing them such.

As a quick preface for the rest of this post: I am not a linguistics expert, just a random computer nerd with an idea and the stamina to google until my fingers bleed. If anything in this post or the documentation booklet is inaccurate or could be improved, please let me know.

-----------------------------------------------------------------------------------------------------------------------------------------------------

Phonetic Inventory

​

Orthography

Leŋoba is written using the Latin alphabet, with the only tricky phoneme being /ŋ/. I spent a lot of time on this, but in the end decided that <ŋ> should be used when possible, and the unused Latin <g> should be used in all other cases. Despite this, I acknowledge that many speakers will still use <ng>, and thus most speakers will also end up recognizing this.

Vocabulary

This is the true distinguishing feature of leŋoba, as it uses AI to produce a lexicon hopefully more neutral and universal than that of other IALs. The way this algorithm functions is explained in more depth in the documentation booklet, but I will explain it briefly here.

When synthesizing a word, the algorithm first searches for the corresponding word in each major language, and groups them by similarity. It then selects which group has the greatest total number of speakers, and finds a word that is most similar to each word in the group. The result is the synthesized word, a list of words it was drawn from, an estimate of how many speakers will recognize the word, and whether or not it used its bias correction methods for this word.

Notably, this lexicon produces words that sound similar, not words that look similar. I expect every learner of the language to have to learn most if not all words in this vocabulary, but for this method to generate words incredibly easy to learn.

Want to Help?

If you would like to help contribute to this project, feel free to leave comments or suggestions on the documentation booklet on ways it could be improved, including sample texts you have translated, or anything else. Suggestions and comments are also available on the English-Leŋoba dictionary for those who would like to comment on words it think the AI may have made a mistake in producing.

Lastly, if you speak English well enough to talk to me and at least one other language fluently, I would love to enlist your help in translating both the the documentation booklet and the dictionary into your language!

-----------------------------------------------------------------------------------------------------------------------------------------------------

I hope that the information I have provided is complete and correct enough for you all to understand it, and I hope you have enjoyed the concept. So what do you guys think, can AI fix the problems with IAL vocabularies?

​

-- Jester