Emacs 24.4 ships with a new minor mode called prettify-symbols-mode. Its purpose is to replace the standard text representation of various identifiers/symbols with a (arguably) more aesthetically pleasing representation (often a single unicode character would replace several ascii characters).

A classic example would be lambda from various Lisp dialects that many people prefer to replace with the greek letter λ (small lambda). prettify-symbols-mode allows you to achieve this by relying on a simple mapping expressed in the form of an alist that each major mode must initialize (prettify-symbols-alist). Simply put - major modes have to provide the configuration for prettify-symbols-mode.

Lisp modes do this via lisp--prettify-symbols-alist:

(defconst lisp--prettify-symbols-alist
  '(("lambda"  . ?λ)))

This means that out of the box only lambda will get replaced. You can, of course, add more mappings for different major modes:

(add-hook 'emacs-lisp-mode-hook
            (lambda ()
              (push '(">=" . ?≥) prettify-symbols-alist)))

Let’s see the mode in action. Consider this bit of Emacs Lisp code:

(lambda (x y)
  (if (>= x y)
      (something)
      (something-else)))

After you do M-x prettify-symbols-mode you’ll end up with:

(λ (x y)
  (if (≥ x y)
      (something)
    (something-else)))

To enable this for a particular mode use (add-hook 'some-mode-hook 'prettify-symbols-mode). If you’d like to enable it globally just add the following to your config:

(global-prettify-symbols-mode +1)

By the way, sky is the limit for symbol prettification. One fairly extreme example would be vim’s plugin haskell-conceal+ that goes to great lengths to bring proper mathematical notation to Haskell code. We can achieve more or less the same effect with prettify-symbols-mode, but one have to ask themselves where should we draw the border between tasteful and distasteful prettifications.