Slightly Higher Coverage And Slightly Lower Coverage For Movies

By Author: Alex Lee
Total Articles: 366
Comment this article

In answer to the second research question, there would be 447 unknown tokens in a typical movie for learners with 95% coverage and 179 unknown tokens for learners with 98% coverage. The average running time of movies analyzed in this study was 113.5 minutes. This means that for learners who have 95% coverage there would be 3.9 unknown tokens per minute and there would be 1.6 unknown tokens per minute for learners with 98% coverage. It is important to note that tokens and word families are different and that if there were 447 tokens, the number of unknown word families would be considerably less. This is because some tokens would occur more than once and word families contain different derivations and inflections of the same token. To determine the approximate number of unknown word families in movies for learners who knew the most frequent 3,000 word families, a sample of six randomly chosen movies from the corpus was analyzed to determine the number of low frequency (4,000-14,000 word level) tokens and word families. Thomas Sabo Bracelets

The number ...
... of low frequency tokens ranged from 194 to 706 and the number of low frequency word families ranged from 120 to 303. The number of low frequency tokens tended to reflect the total number of tokens in each of the six movies, which ranged from 5,145 to 14,324 tokens. On average there were 367 low frequency tokens and 139 low frequency word families in the six movies. Learners who knew the most frequent 3,000 word families would encounter 1.2 low frequency word families per minute or 3.2 low frequency tokens per minute in those six movies. In response to the third research question, the results indicated that there was little difference between the American and British movies if 95% coverage was the goal. A vocabulary of the most frequent 3,000 word families plus proper nouns and marginal words provided 95% coverage for both American (95.76%) and British movies (95.50%). However, to reach 98% coverage, knowledge of the most frequent 6,000 word families plus proper nouns and marginal words provided 98.14% coverage of the American movies while knowledge of the most frequent 7,000 word families plus proper nouns and marginal words was necessary to reach 98% coverage of the British movies (98.29%). It should also be noted that because the BNC consists primarily of British text (Nation 2004), it may be more likely to represent the language in British movies than in American movies. This is clearly apparent when looking at high frequency headwords taken from the 1,000 and 2,000 level BNC word lists. For example, bin, biscuit, bloke, bugger, chap, fortnight, lorry, motorway, muck, nick, nil, nought, parish, parliament, pence, petrol, pint, pound, pub, quid, rubbish, sack, and sod are all found in the first and second 1,000 word lists but are likely to occur at less frequent levels in an American corpus. This suggests that coverage of American movies could be higher if they were measured with lists derived from an American corpus. Thomas Sabo

Coverage may be dependent on the amount of discourse (American or British) in the corpus from which the frequency lists are derived. Unfortunately there is currently no word frequency lists derived from an American corpus using the same criteria Nation (2004) used to create the BNC lists. Research comparing British and American movies with frequency lists derived from American discourse may show slightly higher coverage for American movies and slightly lower coverage for British movies.