From the looks of it, parser_read_dquote_word returns b for every \b it finds. It works well for \", but \t is translated to t and \n is translated to n.
parser_read_dquote_word
b
\b
\"
\t
t
\n
n