You probably have the coding system set to a fixed value. E.g. by using
(setq coding-system-for-read 'utf-8)
in your .emacs file. Setting this variable overrides the settings for individual files and lines like
-*- coding: iso-2022-jp-unix; -*- get ignored. In my installation this skrewed up the generation of the autoloads for auctex during installation and resulted in the same scan error.
Test if this problem applies to you: locate were the elpa archives are extracted and open then file elpa/auctex-11.89/tex-jp.el. If
buffer-buffer-file-coding-system is not set to
iso-2022-jp-unix, then this is the problem.
The solution is to not set
coding-system-for-read globally. See http://ergoemacs.org/emacs/emacs_encoding_decoding_faq.html for how to set the default coding system properly.