>In particular, an *icon* means nothing to me until I can *name* it
>audibly.  I can't recall it, I can't recognize it ... until I
>sound-out the name of it again, and then I get access to all the
>semantics and learning I've had about that *name*, but never the icon
>directly.

Jef Raskin makes a similar criticism of such use of icons in "The 
Humane Interface."  An interesting read.