There is a mechanism $a$ used in Charset-X-Coder.java which is preprocessed
to "a" or "an", as appropriate to the following word.

However, a number of places where it was needed were missed.