Profiling The New General Service List: A Study Based On Spoken English Corpora

Vocabulary knowledge is indispensable in second/foreign language(SL/FL) learning. Blachowicz, Fisher, Ogle & Watts(2006) shows that vocabulary learning is a complex process. To facilitate vocabulary teaching and learning, numerous vocabulary lists have been developed. Of them, West's(1953) General Service List(GSL) is the most influential general-purpose word list for its high coverage and wide distribution across various texts. Nonetheless, a number of limitations of the GSL have been identified, such as its old age, qualitative principles of item selection and format of word family. To address the limitations, Brezina and Gablasova(2015) provides a new word list named the New General Service List(new-GSL). It claims that the new-GSL serves well across a variety of texts in both written and spoken texts.However, most of the source corpora were written texts and its spoken component explained only 0.08% of the corpora. For the unbalanced feature of the written and spoken components in its source corpora, the validity of the new-GSL in spoken contexts is necessary to be examined. Therefore, the present study aimed to investigate the coverage of the New General Service List(the new-GSL) in different spoken contexts. Three research questions are addressed: 1) To what extent do the new-GSL words cover spoken English across various spoken contexts? 2) To what extent do the new-GSL words cover spoken English across different frequency bands? 3) Is there any word frequently used in spoken contexts but not included in the new-GSL list? If so, what are they?Three corpora of spoken English with different domains and formality were used in the study, i.e. the British Academic Spoken English(the BASE corpus), the spoken components in the Open American National Corpus(the OANC Spoken) and the corpus of Cable News Network Transcripts(the CNN Transcripts). These corpora represented various spoken contexts, different corpus sizes and language use through time. The Stanford Core NLP and the language Python were used to process the corpus data.The results showed that the new-GSL coverage was generally lower in the spoken corpora(from 71.35% to 78.01%) than that in the corpora(approximately 80%) largely composed of written data as found in Brezina and Gablasova(2015). In addition, the new-GSL coverage across different corpora and frequency bands was not stable, which went up along with the increasing formality of the spoken data. The findings revealed that the newGSL might not serve as a core vocabulary in spoken contexts as well as it did in the written data. More importantly, the results challenged Brezina and Gablasova's(2015: 17) point that there existed a core general vocabulary “across a variety of language corpora including different written and spoken contexts”.The current study argued that there may not be necessary to propose a single core general vocabulary. It is suggested to develop lists of basic words for different groups of target learners based on the pedagogical purposes.Pedagogical implications are mainly from the perspectives of learners, teachers and material designers, and pedagogical research. First, learners may need a specific spoken vocabulary list as an effective reference for self-study in spoken English. Second, teachers and material designers may demand a learner-targeted and purpose-oriented vocabulary guide for spoken English instruction and material development. Third, a learner-specific spoken vocabulary list may be essential to serve as a baseline for more advanced language study in spoken English, such as to develop more specified spoken vocabulary list.