Abstract:
This thesis looks at the effect of vocabulary control on reading materials used with beginning readers. Three types of reading material were examined - a series of books, designed for use with beginning readers with a strictly controlled vocabulary, an uncontrolled series also designed for beginners and a set of library books and traditional fairy tales with no vocabulary control. Other qualitative and quantitative comparisons were made.
The study shows that vocabulary control can increase the percentage of words that are likely to be known. However, when different assumptions are made about what words are likely to be known and they include the 1000 most frequently occurring words in general English, words outside the 1000 most frequent but that are commonly found in children's literature, character names, and words defined by the context, then 95% coverage would seem possible with texts with no vocabulary control. In other words, qualitative features of text design can also enhance coverage.
Assuming that readers had a vocabulary consisting of only the 1000 most frequently occurring words in general English, 19% of the running words in the two uncontrolled series could not be read. With the controlled series 14% of the running words could not be read.
Not surprisingly, it was found that both of the uncontrolled reading series contained much larger vocabularies than the controlled series. The number of different words used varied from 1259 for the controlled series to 3849 for one of the uncontrolled series. However, none of the series could be read with 95% coverage (the coverage necessary for successful independent reading) with only a 1000 word vocabulary.
As it was considered that the presence of homographs could skew any counting of words, they were treated separately. Although a fairly large percentage of the words were potentially homographic, very few were actually used in more than one sense.
A set of words was identified which appear to be used more frequently in children's literature than in other kinds of literature. It is argued that these words should be treated as likely to be known. The proportion of the text in each series made up of these words was also very similar.
If the 1000 most frequently occurring words in general English, character names and the words that occur frequently in children's literature are counted as known then the coverage of known words exceeds 95% for both of the series designed for beginning readers. Even the series consisting of library books and fairy tales can be read with over 90% coverage.
The study showed that qualitative features of text design such as the differences in contextual strength and the impact of illustrations can also increase the proportion of words that are likely to be known as well as increasing the interest the reader has in the text.
It is argued that 95% coverage is even more likely if teachers prepare their own materials, basing stories on their students' experiences and if they use the shared book method as this will ensure adequate support for the unknown words.