In this case I think it is just a popularity contest. Takete/Maluma were the first version of the study in the 1920s. Kiki/Bouba replicated the study in the 2000s. So recency bias plus "simpler" fake words can explain some of the popularity.