Softwaerewolf
commented
3 months ago Although this suggestion feels overloaded, I have put additional information about why this functionality should be pursued as a matter of course because of impacts it has in relation to internationalisation for certain languages.
The same functionality that is a "nice to have" in English and Latin and similar languages becomes "essential" in relation to Chinese and possibly Japanese and other languages.
Philippe-Choquette
Currently it seems suggestions strongly prefer offering matching single words or hyphenated words.
My most common typing mistake is missing space and hitting "v" instead. In Swiftkey it's suggestions substitute these incorrect characters for a space or at least offer multiple word corrections.
Although possible, FUTO keyboard seems extremely adverse to multi-word suggestions. It will do it if you select a bad correction and then undo and replace it with your preferred correction.
For example: Helloeorld =(select)=> HelloWorld =(manually delete and correct)=> Hello world
Now it will offer "Hello world" as a suggestion (not guaranteed in all circumstances). But it should do this automatically from the start.
The behavior is actually even more erratic.
For example: "Thelazyfox" offers "Tulztocracy", but it does seem aware that these could be 3 separate words if I simply don't select a correction, then move my cursor to the beginning of the concatinated word. Now it offers "The lazy fox", but if I type out the same concatinated words again it does not.
Additionally, "thevlazybfox" should detect and offer the same three word correction. It does not under any circumstances.
Finally, but relatedly, partial substitutions. Where some part of a word is known but the rest isn't or is ambiguous.
For example "Sixallles" could be "Six allies", " Six apples" or "Six alleles" but there is a high confidence that the first word is "Six".
Far future: There are knock-on benefits to partial corrections which will be necessary for certain languages such as Pinyin input for Chinese where words share pinyin and sometimes characters.
For (probably bad) example: 舊時 jiushi - old times 舊識 jiushi - old friend 舅氏 jiushi - uncle
Knowing which is the first character informs which is the most likely second character.
Typing also commonly involves typing entire sentences with just initials, eg. "Ndmzshenm" becomes "你的名字什么" or "Ni De MingZi Shen Me"
Note, I am not Mandarin speaker, my information is all observational.