Abstract: ［Purpose/Significance］Clarifying the effects of recommended tags and their representational languages on learning behavior in the cross-language environment is an important foundation for optimizing and improving the qualities of tag suggestion and enhancing user experience in cross-language tagging.［Design/Methodology］With the use of controlled experiment, this study compared the results and learning behaviors of cross-language tagging while users annotated texts in three modes, which are tagging with recommended tags in foreign languages, tagging with recommended tags in Chinese and tagging without recommended tags.［Findings/Conclusion］a.There are three learning patterns based on recommended tags in the context of cross-language tagging, including learning based on the tag forms, the long semantic mapping-learning based on the semantics of tags and the short semantic mapping-learning based on the semantics of tags; b.Recommended tags not only provide guidance for users in their learning of tagging, but also bring them some cost. It is consistent with the “anchoring effect”.［Originality/Value］The findings will be helpful for the design and implementation of cross-language tagging system, and also improve the efficiency of information sharing.

Key words: Cross-language, Recommended tags, Learning behavior, Information tagging