A Multi-system Translation Word-order Data Set

System combination is emerging as a powerful tool in statistical machine translation. In this document, we describe a data set designed to enable the study of a sub-problem of system combination: that of combining the word-order decisions made by several translation systems. We outline a data set derived from the MSR-NRC-SRI joint entry to the 2008 NIST machine translation evaluation, and briefly describe each translation system that contributes to the set.