A completeness theorem for unrestricted first- order languages