UTF-8-encoded output in Emacs shell

[kj]

I have a Python script that writes UTF-8-encoded Japanese text to
stdout.  This script produces nice-looking Japanese output when I
run it on my regular shell, running on OS X's Terminal utility.
But if I start "emacs -nw" the same Terminal window, then from
within the Emacs session I start a new shell (with M-x shell), and
now run the script within this Emacs-internal shell, the output
looks like garbage.  ("Gobbledygook" is the technical term for
this, I think.)

What must I do to get the right-looking Japanese output when I run
this script within Emacs?

(BTW, I naively tried setting the coding for the *shell* buffer to
utf-8, with M-x set-buffer-file-coding-system, but this had no
effect.  But, from Tomas's reply to an earlier query of mine, this
maneuver may not be enough in any case, or may be meaningless in
a buffer like *shell* that is not associated to a file.)

TIA!

~K