Ok… I finally had the time to do a bit of literature research. There is a lot of research out there looking at the learning of English words. And it appears that there learning in thematic clusters (think beach, sunny, swim) is helpful whilst learning in semantic clusters (think dish, bowl, plate) is not [1].
Another study found that splitting those similar words up helps a lot. This study loooked at Japanese perople learning english and they splitt the semantic clusters up such that there were seven other words inbetween. Doing so increased the recall after a week by a factor of 1.6 (from 17% to 27%) [2]. Sonds like even a little space between the words might help…
Now Japanese is different and those words that use the same kanji are not only similar in meaning but also in sound and looks.
I found a study looking at Spanish looked at both sets of phonetically similar (similar endings or starts of the word) and semanitically similar words. It found that phonetic similarity by it self is harmful to retention [3].
Sadly I didn’t find anything looking at Japanese and words with the same Kanji (or Chinese and Hanzi for that matter).
For my part I’ve started learning new words with the Lesson Picker which helps a lot (thanks again @mitrac). I feel it helps a lot in preventing this for me. But I feel that this has the potential for the big Crabigator to improve things for everyone.
[1] TINKHAM, Thomas. The effects of semantic and thematic clustering on the learning of second language vocabulary. Second language research, 1997, 13. Jg., Nr. 2, S. 138-163.
[2] NAKATA, Tatsuya; SUZUKI, Yuichi. Effects of massing and spacing on the learning of semantically related and unrelated words. Studies in Second Language Acquisition, 2019, 41. Jg., Nr. 2, S. 287-311.
[3] WILCOX, Amanda Elizabeth. Category Effects of Vocabulary Presentation in L2 Vocabulary Acquisition in Novice Learners. Auburn University, 2011.