A Study on Human Perception of Document Paragraph Layout

We identify a list of characteristics of a text document that we use as the basis for a study on human perception of what the structure of a document should look like. Our study reveals that the native language dimension has a significant impact on the human perception of some such characteristics, including the average number of paragraphs starting with a verb, as well as other characteristics describing average paragraph proportions. The study results also show what the means values are, for the different document characteristics, and their distribution across the native language dimension, therefore providing some idea of what a normalised structure of a document should look like. The results of the study have a direct application in a new method for embedding secret messages in text documents that has been recently proposed by the authors and which uses manipulations in the paragraph layout of a document.