Solvatochromism sounds pretty interesting (I was unaware of this phenomenon--thank you Honclbrif!), maybe you could try different solvents as well and see if there is a correlation between the solvent polarity and color change.
Have you tried to crystallize the material dissolved in DCM or chloroform? Going along with the possibility of dimerization, maybe you could crystallize out the dimer--mount both your yellow crystals and the new ones on a diffractometer and see if they are the same thing or not.
Another idea (probably a better one) would be to do an NMR experiment. Dissolve your compound in CDCl
3, and monitor the NMR spectrum over time. See if there are new peaks arising. If there are, and considering you said it is a slow process, then you could also extract the kinetics of the process.
Hope the suggestions are helpful!