Learning to read a second language (L2) is especially challenging when a target L2 requires learning new graphic forms. Learning Chinese, which consists of thousands of characters composed of hundreds of basic writing units, presents such a challenge of orthographic learning for adult English speakers at the beginning stages of learning. In this study, we use an "in vivo" classroom design to extend previous research on how to support orthographic learning. First, we test the hypothesis that learning characters is enhanced by a grouped sequence of characters that share sub-character graphic components. Second, we examine the effects of four encoding methods that have been investigated in laboratory studies--handwriting, visual chunking, passive reading, and stroke-reporting. The results demonstrate that the grouped approach facilitated character production compared with the distributed approach and that visual-chunking outperformed the other three encoding methods under the grouped sequence. We propose that learning via visual chunking with characters grouped by the same chunks enhances the Chinese orthographic representations of beginning L2 learners.