| Summary: | The article presents a study which discussed sampling problems in studies of writing style. According to the author, as the techniques of studying language become more refined, there is a need to take a closer look at the sampling methods. The samples usually drawn for language studies are composed of clusters of words, in sentences or paragraphs. The random sampling of words is suggested as a way to avoid difficulties in word studies. A comparison of the clustered and simple random sampling methods shows that a clustered sample significantly overestimated the percentage of short words, structural words and easy words.
|
|---|