The learner corpus of English has been developed and researched throughout the world. However, the learner corpus of Spanish is still not completely developed. Because of the popular trend of corpus linguistics and the emerging demand for researching a third language in applied linguistics, we began constructing the Taiwanese Learners' Written Corpus of Spanish (CEATE, Corpus Escrito de Aprendices Taiwaneses de Español) in 2005. In the past 7 years, we collected 2,425 compositions comprising approximately 440,000 words, written by Taiwanese learners whose L1 is Mandarin Chinese and who learn Spanish as a third language and English as a second. With teamwork and technical support of computational linguistics, our construction of CEATE has mainly been dedicated to sharing academic resources. We are currently releasing data collected during the first 3 years (Phase I) for public use (http://corpora.fl1d.ncku.edu.tw). In contrast to a raw corpus, ours is POS-tagged and correction-annotated. Therefore, by accessing this corpus, researchers, teachers and learners can conduct an efficient, practical, and systematic search directly related to their particular interests. In the near future, we will not only increase the number and types of data, but also improve the techniques of annotation and function of searching to offer a useful resource for study of Spanish language learning.