Cool to see this on HN -- I wrote a program for a psych lab in college that was exploring this effect. The idea was to see if homogenous sensory stimuli had a faster reaction time & accuracy compared to disparate ones.
eg. Was the response to [ Bouba-like image + Bouba-like audio ] faster/more accurate than [ Bouba-like image + Kiki-like audio ]
Seemed like it may provide some insight into speech perception and synesthesia.