It could highlight differences for you. Current models already seem competent enough to take into account a lot of context and not to alter the meaning.