A computational approach to English core vocabulary based on the British National Corpus

Abstract

Core vocabulary has been a long-time concern for lexicographers, language teachers and other linguists. Many researchers have realized the existence of core vocabulary in English studies and defined core vocabulary by a variety of criteria. Ogden's (1930) concept of Basic English, the vocabulary control movement in the 1920s and 1930s, and the proposal of culture-free Nuclear English for international use (Quirk 1982; Stein 1979a) all boosted the research into core vocabulary. Past research tends to focus on word frequency in written texts rather than both of the written and spoken texts. Moreover, the factor of age has been ignored in previous studies, but modern corpora with the tags of age information have made this kind of research possible. In the light of previous theories and studies, core vocabulary is now defined as those words not only with high total frequency (raw frequency) but also with a wide distribution across different age groups and genres in both spoken and written modes of large corpora. The core words are now being selected and evaluated by the following parameters: 1) distributed word frequency in different age groups, 2) distributed word frequency in different text genres, and 3) word familiarity, imageability and age of acquisition. Making use of the age-group and genre information tagged in the British National Corpus XML Edition (BNC XML 2007), our research adopts Carroll's Usage Coefficient (U) to calculate a word's distributed frequency in different age groups and text genres for evaluating the core degree of words. The findings are as follows: First, higher cumulative coverage can be achieved when we select core vocabulary by the combined parameters of a word's dispersion index and distributed frequency across age groups rather than raw frequency alone. Second, young age group under 15 rely more on core vocabulary than adults. For age group over 15 years old, core vocabulary occupies a stable ratio in their mental lexicon despite age increase. Each age group tends to acquire more core words selected on the basis of distributed-frequency-cum-age than those selected on the basis of raw frequency or raw-frequency-cum-range (e.g., Longman Defining Vocabulary, Oxford Defining Vocabulary, i.e. Oxford 3,000). Third, a core vocabulary selected according to distributed frequency across genres achieves higher cumulative coverage than a core vocabulary selected by raw frequency alone. Fourth, a word's core index and its familiarity, imageability and age of acquisition are significantly correlated. There is a significant positive correlation between a word's core index and its familiarity, while the correlation between a word's core index and imageability and age of acquisition is significantly negative. That is, familiarity is a good indicator of core vocabulary and core vocabulary should be familiar to people of all age groups and different backgrounds. However, highly imageable words are not necessarily core words and core words tend to be lower in imageability. Moreover, core words tend to be acquired earlier in language acquisition and language development. Fifth, by applying our research result to the compilation of the electronic English Frequency Dictionary for English Learners, the distributed frequency across different genres is effective for learners to discriminate English synonyms in terms of genre, language formality, cultural indication, register, and spoken and written domains. With distributed frequency across genres displayed in a histogram, EFL learners can gain a better insight into the difference between English synonyms. Our study is thus significant in bringing into play the importance of balanced vocabulary selection based on distributed frequency across age groups and genres. The findings of the present research will not only contribute to English teaching and learning but also lexicography studies. Moreover, our research into correlation between a word's core index, familiarity, imageability and age of acquisition deepens our understanding of the psycholinguistic features of the mental lexicon.

Date of Award	15 Feb 2012
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Chengyu Alex FANG (Supervisor)