Posteditions from ContentTranslation

You can get dumps of ContentTranslation translations here:

• https://dumps.wikimedia.org/other/contenttranslation/

The TMX files (e.g. 20160923/cx-corpora.es2ca.text.tmx.gz) contain:

    <tu srclang="es">
      <tuv xml:lang="es">
        <prop type="origin">source</prop>
        <seg>André Lotterer durante una tanda de entrenamientos de la Fórmula Nippon 2010 en Motegi.</seg>
      </tuv>
      <tuv xml:lang="ca">
        <prop type="origin">mt</prop>
        <seg>André Lotterer durant una tanda d'entrenaments de la Fórmula Nippon 2010 en Motegi.</seg>
      </tuv>
      <tuv xml:lang="ca">
        <prop type="origin">user</prop>
        <seg>André Lotterer durant una tanda d'entrenaments de la Fórmula Nippon 2010 a Motegi.</seg>
      </tuv>
    </tu>

The original, the MT and the postedition.

If you use the .json output then you can also see which engine was used.