Thelast part has mentioned the general procedure. Here I will present the results.
In phase 1, the result is displayed in the chart, we can see clearly, the mean coverage of each set of headwords and headwords plus members. So, for question1 and 2, the answers are shown in the chart.
In phase 2, because the EWL is designed for L2 beginners, the size of EWLshouldn’t be larger than 1,000 items. Then, the lexical coverage of the first 1000 items is mapped into a graph. we can clearly see, the lexical coverage ofheadwords plus members is slightly higher than that of headwords. And, there isa huge drop in coverage between 1stand 2nd100-headword level. However, beyond the 8 hundred word level, the curve flattens out andthe additional coverage of the rest headwords is less than 1%, this result canalso be seen from the table. So there are two possible cut-off points：100 wordsand 800 words. But when we consider the practicability, 800 words will be abetter option, because lexical words are central to L2 beginners, but the first100 words contain very few lexical words, but so many function words. The 800cut-off point is also supported by the amount of lexical coverage shown in thetable. The coverage of the top 800 headwords is 61.15%, and that of the top 800headwords plus members is 75.09%. These figures proved that these 800 words canprovide a decent foundation for L2 beginners. So for question 3, the answer is800.
In phase 3, I have made a graph and table using excel. From the graph we can see,the mean coverage of the EWL is the highest both regarding headwords and level2 word families. The table gives the precise data, which shows the same result.so the answer to both the fourth question and fifth question is yes, EWL headwords and word families have the highest mean coverage. what about theoverlap? To answer the next two questions, I have made some graphs using theresearch data as well. 692 words in the EWL appear in the 800 best spokenheadwords, so the overlap is 86.5%. The number of overlapping words for thewritten corpora are 698, so the overlap is 87.25%. Therefore, the answer to thesixth question is 86.5%, and to the seventh question is 87.25%.
Interestingly,590 words in the EWL appear in both the best spoken and written headwords. Thisfact supports that EWL is a quite balanced word list, which is beneficial for bothL2 speaking and writing.
Inthe discussion, I will talk about the advantages and limitations. There are apparentlymany merits, but I will focus on the most important three of them. Firstly, EWLis more useful to L2 beginners, as all the work is aimed at them. Secondly, ithas greater validity because EWL excludes non-words like affixes and propernouns like cities’names. Thirdly, it is more representative for it is derivedfrom 18 corpora with various types of English and discourses.
Thereare also limitations: The EWL is so far only theoretically superior to other word lists. Ihave searched for relevant literature regarding the application of this wordlist, but surprisingly I find very few. It would be more convincing if thisstudy can go along with a longitudinal study testing the effectiveness of theEWL. Meanwhile, it is just a word list, which has no link to contexts. If itlinked closely to certain textbooks which can provide useful contexts, the EWLwould be more attractive and more likely to come into use in English as a SecondLanguage classes.
Edited: rankingthe headwords in the master list, four steps are needed. First, the frequency of each headword is examined ineach corpus, by using an online program. The second step is to calculate thecoverage of each headword, that is frequency divided by the number of runningwords in each corpus. The third step is to calculate the mean coverage of eachlevel 2 family in all 18 corpora. The fourth step is to rank headwords from themaster list according to their mean coverage.
there are two steps. The first step is to determine the mean coverage of each set of100 headwords and level 2 word families in the master list. In total 2000headwords and level 2 word families are examined in this step. The second stepis to determine the cut-off point of the EWL according to three criteria:practicability, change in the lexical coverage curve, and amount of lexicalcoverage.
there searchers compare the mean coverage of EWL headwords and the best headwordsform each source list, and compare EWL word families and the best word familiesin each source list. Then, they will find out the overlap between EWL headwordand the best headwords in nine spoken corpora and nine written corporarespectively.