I'm posting this merely for the sake of offering another opinion on
legibility; I don't think that any syntax changes in this area are
good, because they can break existing and compliant R5RS code.  For
instance, I've seen this naming convention somewhere: FOO has to do
with an exclusive interval, FOO: with an interval exclusive on the
lower bound and exclusive on the upper bound, :FOO with an interval
inclusive on the lower bound and exclusive on the upper bound, and
:FOO: with an inclusive interval.  For another example, SRFI 42 would
break if the meaning of a colon prefix were changed in Scheme's
lexical identifier syntax, although if any extension were to be made
I'd prefer the colon prefix.

   Date: Tue, 11 Apr 2006 17:18:17 -0400
   From: Marc Feeley <xxxxxx@iro.umontreal.ca>

   I find colon at the end to be more natural since it closely
   corresponds to the use of colon in English (and many other natural
   languages).

English has a few differences, though, most prominent among them that
there is other punctuation separating the pairs.  Compare:

  foo: bar, baz: quux, zot: mumble    or
  foo: bar; baz: quux; zot: mumble    with
  foo: bar baz: quux zot: mumble

In the last one I find it hard to tell where one pair ends and the
next one begins.  On the other hand, with a prefix, the delimitation
of the pairs is very clear:

  :foo bar :baz quux :zot mumble

   1) I prefer prefix syntax, e.g.   (button :text "OK" :action quit)
   2) I prefer suffix syntax, e.g.   (button text: "OK" action: quit)
   3) I don't care

I opt for (1).