It is a combination, for sure. There is no denying that Chrysanthemum is extra long, but Minerva is also long to me despite a rather streamlined set of letters and just three syllables. My perception includes the weight and timbre of a name and its composite sounds, I think. Wilhelmina is light and springy as the Willa-mina pronunciation, and while it is a lot of letters and an extra syllable, it is in the same bucket as Minerva for me. Will-HELM-ee-na is in the Chrysanthemum bucket because of the extra weight and clunk.
I also think that bias plays a huge role, and one bias more than all others. That is your own surname (or that of your child or future child if that is how you think about names). I share a rather brief monosyllabic surname that has lengthened the average of my favorite names since being with my now husband and choosing to take his name. I grew up with a richer two syllable surname and I still loved many of the same names but not as much. My high school sweetheart I was with when I discovered Nameberry 10 years ago had a three-syllable surname and I gravitated toward shorter names at that time in my life. I think this has way more play than we think on Nameberry. We might love a wide variety of names of different length however we perceive it… but we spend the most time hanging around with the names we come to see as usable and not cumbersome.
@katinka, with her “Winterfield,” spends more time hanging out with Astrid and Flora than with Rosamund and Endellion, despite loving them all. I’m over here spending more time playing with the opposite set, despite also loving all those names, too. And I think that even though we see all those names and love them, we tend to offer advice on the forums that leans slightly toward our own “style,” which is married to the bias of the own length we perceive as wearable for our own children… And then there is added bias of an Emma and M@x growing up not a Lilian and J0nathan. The bias isn’t some horrible prejudice, but it is just light influence that winds up exuding large differences over time.
Hopefully that make some sense…