Negative-dimensional integration revisited