Abstract:Crowdsourced translation has impacted on and is also bringing challenges to the traditional translation industry. This new translation model has redefined the working environment, methods, and processes of translation, as well as translators' behaviors and roles. From a text-linguistic perspective, crowdsourcing, given its task-segmentation nature, subverts the traditional concept of "text". Texts in crowdsourcing are no longer a unified whole that represents coherent meaning. Instead, texts can be divided and subdivided into increasingly smaller segments, sometimes even smaller than morphemes. As such, when crowdsourced translation is being studied, "texts" and "non-texts" as by-products of the translation procedure should be given the same amount of attention as translation products. The by-product-texts can serve as quality language resources for building up corpuses for language services, localization, language research, and theme research to promote the diversification of translation technologies. Meanwhile, these language resources can facilitate the analysis of translators' styles and leverage the translator resources globally to unfold the cultural characteristics of the translated texts of multiple languages.