Raw unicode literal that is valid in Python 2 and Python 3?

Why don't you just use raw string literal (r'....'), you don't need to specify u because in Python 3, strings are unicode strings.

>>> tamil_letter_ma = "\u0bae"
>>> marked_text = r"\a%s\bthe Tamil\cletter\dMa\e" % tamil_letter_ma
>>> marked_text
'\\aம\\bthe Tamil\\cletter\\dMa\\e'

To make it also work in Python 2.x, add the following Future import statement at the very beginning of your source code, so that all the string literals in the source code become unicode.

from __future__ import unicode_literals

The preferred way is to drop u'' prefix and use from __future__ import unicode_literals as @falsetru suggested. But in your specific case, you could abuse the fact that "ascii-only string" % unicode returns Unicode:

>>> tamil_letter_ma = u"\u0bae"
>>> marked_text = r"\a%s\bthe Tamil\cletter\dMa\e" % tamil_letter_ma
>>> marked_text
u'\\a\u0bae\\bthe Tamil\\cletter\\dMa\\e'