There's an amusing way to quantify just how systemic the language orthography is: train an ML model to converts words to phonemes and back, and then see how many mistakes it makes. With similar sized training corpus in different languages, you get numbers that are directly comparable. Of course, it's a very rough estimate, but still interesting:
Finnish scored 98% accuracy on "writing" (model converting phonemes to written words) and 92% accuracy on "reading" (written words to phonemes) in this study. The only other languages with both scores above 90% in their comparison are Esperanto - which is explicitly designed for that, of course - Serbo-Croatian, and Turkish. I'd say that's a very good result for a natural language.
For comparison, Spanish is 70% writing / 85% reading, French is 28% writing / 80% reading, and English is at dismal 36% writing / 30% reading.
https://aclanthology.org/2021.sigtyp-1.1/
Finnish scored 98% accuracy on "writing" (model converting phonemes to written words) and 92% accuracy on "reading" (written words to phonemes) in this study. The only other languages with both scores above 90% in their comparison are Esperanto - which is explicitly designed for that, of course - Serbo-Croatian, and Turkish. I'd say that's a very good result for a natural language.
For comparison, Spanish is 70% writing / 85% reading, French is 28% writing / 80% reading, and English is at dismal 36% writing / 30% reading.