> If you write on paper, you need to write out each character by hand.
Exactly like English or French. You also need to remember the correct sequence of letters (see my remark below).
> In contrast, if you write on a tablet or other device, the input system will sort of "autocomplete" for you.
Exactly like English or French. I didn't write half of the letters used to type this comment.
I think you are searching for Japanese exceptionalism where it probably doesn't apply. Remembering the correct kanji, and graphy of said kanji is pretty much the same as remembering the (wild) orthography of French or English words. I'm searching quite often if a letter is a 'a' or 'e' (doesn't help that cognates between French and English often swap those), the conjugation of a verb or if a letter is doubled or not (again often not the same for related words, for example ennemi vs enemy).
Your perception of the difficulty in-between writing systems is twisted by the decades of education you got in the said system vs the few years of learning another system with a second language. When I see some soup of letters in Japan, it's clear that the question is not just about one system being more simple than the other.
Character amnesia is a unique phenomenon to hanzi/kanji and is absolutely incomparable to English and French orthography. Native Chinese and Japanese speakers learn characters largely through muscle memory, which deteriorates rapidly if you don’t regularly write by hand. Many university-educated adults are unable to write a supermaket shopping list without hesitation.[1] Conversely, in English I learn difficult spellings (Gloucestershire, syzygy, rhythm) by aural-mnemonic tricks which stick around in my memory essentially forever.
Whether this had any effect in this study, I don’t know, but it certainly seems relevant that the experience of writing by hand in Japanese is utterly different to any alphabetic script.
Don't worry, I have some knowledge of Japanese and Chinese (traditional) and first hand experience of the phenomenon. If you can visualize the character you can write it, and I maintain that this is not as different as people makes it than visualizing the writing form of a word for a language written with a morphophonemic script (English, French, Korean without hanja, Thai, etc.)
I've been studying Japanese for years and while SRS reading methods help to some degree, writing is the only thing that's really helped me lock in Kanji. Especially at the intermediate level where kanji not only look similar but are pronounced the same.
SRS : Spaced repetition system. Anki is a good example of that.
I've been slowly learning Japanese, mostly with Duolingo, and that's a conclusion I've noted as well. People get hung up on the "complexity" of even hiragana and katakana, complaining about there being two many or those two scripts being duplicates.
There are fewer total hiragana and katakana than there are English digrams and trigrams in lower-case and capital letters. (Because Japanese uses fewer overall phonemes...and the characters are consistent in terms of pronunciation, unlike English.)
I'm not very far into kanji (just some basic ones like 水 or 人 that stick) but the principles are solid: they have shared stroke components (radicals) that related kanji share, much like similar collections of letters, and studying them as you study vocabulary isn't really any different than knowing what an English word looks like on sight, without "sounding it out."
> Participants in the Tablet group used a stylus pen, thereby controlling for the effects of longhand writing with a pen in the Note group.
In this specific study this is not a notable effect between the Note and Tablet groups. For the phone group you're right that it's a completely different input, while also being much more painful all around due to the sheer device size.
It autocompletes kanji? That's impressive. At first I thought that might be a very difficult thing to do but I just talked to some friends who are Japanese natives and they said it's easier to solve the issue of similar kanji by using context.
The participants wrote their notes in Japanese.
If you write on paper, you need to write out each character by hand.
In contrast, if you write on a tablet or other device, the input system will sort of "autocomplete" for you.
Therefore, more cognitive effort is required to write on paper than on device.
It is fairly accepted that greater cognitive effort (either to encode or to retrieve) is associated with greater recall.
So I would doubt the transferability of this result to western context.