Improving Statistical Machine Translation for a Resource-Poor Language Using Related Resource-Rich Languages