Following the trend of computer assisted language learning (CALL), in Taiwan, most language classes now have equivalent media support for language teachers and learners. Implementing videos into classroom activities is one of the choices. The current study explores the process of implementing American TV drama in L2 vocabulary learning from learners' perspectives. Twenty-eight Taiwanese EFL adult learners participate in the study. Authentic video clips from three different American dramas-"How I met your mother", "The King of Queens", and "Reba" are adopted. After three sessions of class activities including clip watching, class discussion, word listing, etc., students complete a 4-likert scale questionnaire and are individually interviewed by the researcher to give their opinions. The results reveal positive comments on the facilitative role of TV drama in learning new vocabulary as previous studies suggested, but learners' comments point out some crucial factors while learning L2 vocabulary with TV drama. First of all, the interest level and the familiarity of the content is an important factor. In addition, the images, subtitles and repetition help participants to "remember" the target words. Other factors such as the authenticity of the language, the contextual meaning of the words, and the dramatic performances all contribute to the learning of the L2 vocabulary. In the end of the study, working memory system, the context, acquisition-learning hypothesis (Krashen, 1981), and noticing hypothesis (Schmidt, 1990) are further discussed.

Adopting a corpus-driven approach, the study aimed to explore the vocabulary knowledge in English short talks including word patterns, features, and usages that are most likely to be encountered by language users in the real context. A specific corpus TED was conducted through a collection of English talks that are less than 20 minutes from the website TED Talks. In addition, the existed corpus BASE (British Academic Spoken English) was included in the study as a sample of talks longer than 20 minutes. Applying three corpus tools, AntConc (Anthony, 2003), RANGE (Nation & Heartkey, 2002), and KfNgram (Fletcher, 2007), the researcher was able to compile frequency-ordered word lists, concordance lines, vocabulary coverage, and lists of lexical bundles. The results showed that although the most frequently-used words in TED corpus and BASE corpus were similar grammatical items, the order was quite different. Moreover, the chi-square test showed a significant difference among four pronouns I, You, We, They between the two corpora and also in different parts of the TED corpus. Finally, the results of concordance lines and lexical bundles presented the "typical" and "frequent" word usages in the beginning, middle, and ending part of English short talks. It is suggested that teachers can build their own corpus to meet specific teaching purposes or learner's needs, and to generate the corpus results into classroom materials while teaching English short talks.