NotesFAQContact Us
Search Tips
ERIC Number: ED443131
Record Type: Non-Journal
Publication Date: 2000-Aug
Pages: 9
Abstractor: N/A
Reference Count: N/A
Once Is Not Enough: Statistical Simulation of Textual Data.
Hogenraad, Robert
Reproduction is part of the regulations of daily speech. Everyday experience shows there are different ways to convey the same information: starting sentences that are quickly rephrased using other words, manuscripts passing through draft states before the writer decides upon the final version. In more ways than one, any text is always but the sample of another one and the words read are but the visible part of an iceberg. James Joyce's "Ulysses" was content analyzed using the Regressive Imagery Dictionary. Among other things, this dictionary allows a person to assess the degree of presence, in texts, of primordial thought content. A negative linear profile is observed, i.e., primordial content decreases as the one-day journey goes by. Pondering this negative linear profile generates wonder about the enterprise of statistical analysis, how little it says, how brittle is a quantitative analysis of a literary work in the face of the uncertainty that surrounds such data. New ways to look at literary data consist of resampling thousands of time data that by nature exist in only one exemplar, bringing empirical study of literature down to what literature is made of--inking out, excisions, alterations, disappearances. The rates of primordial thought content were repeated 2000 times, using a bootstrap algorithm. The best contribution a researcher can make to empirical studies of literature is to "hammer" scholars to see how well resampling allows researchers to assess the degree of uncertainty associated with any measurement. (Contains 3 figures, 5 notes, and 16 references.) (Author/NKA)
Publication Type: Reports - Research; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A