OTEANN: Estimating the Transparency of Orthographies with an Artificial Neural Network

English is singled out because its orthography really is atrociously bad even when compared with other languages. https://arxiv.org/pdf/1912.13321

According to this paper [0] and my own experience, it's way easier to pronounce a word in French given the spelled word than in English. It's slightly harder to spell French than English for the model of the study, but it's really close. Now, in my personal experience, I feel like French has a lot of rules while English has a lot of outliers which do not follow any rules. But my native language is French, so I am obviously biased. [0]: https://arxiv.org/pdf/1912.13321

This is more generally the concept of Orthographic depth [1] A recent study attempted to categorize languages and found English and French to be among the most opaque, Finnish and Korean to be the most transparent, and German in the middle. [2] [1] https://en.wikipedia.org/wiki/Orthographic_depth [2] https://arxiv.org/abs/1912.13321

The complexity of these rules, and the number of exceptions that you need to learn notwithstanding the rules, can be roughly estimated for any given language by training a language model on word <-> IPA correspondence for that language (using a subset of the vocabulary as a training set), and then seeing how well it can predict the remaining words. You can run it in either direction, too, to separately measure the difficulty of reading (word -> IPA) and writing (IPA -> word) that language. This was actually done for a number of languages including English: https://arxiv.org/abs/1912.13321 You can see how languages with true phonemic spellings tend to be in the >90% range on both reading and writing, with Esperanto at 99%. Spanish and German are in 60-80% range. English is dismal at ~30% for both, though, with only French and Chinese being harder to write, and all other languages tested being easier to read.