Effects of Splitting Text into Multiple Columns

Introduction

The problem of readability of text is a very important one.  It arises as more and more people move to their computers as the primary source of news, information, and literature.  There are multiple factors that affect reading speed and comprehension of text presented on a computer screen.  They include display size, resolution, number of lines per screen, and others.  One of the important factors is the length of the text lines.  This is especially important for texts presented on the web pages, as users often like to maximize their web browser windows.

Most of the computer applications that display text, including popular word processors and web browsers, show identical behavior when their window is resized horizontally: unless the page format specifies a particular width for the contents, they will rearrange the text on the page to fill all the available horizontal space.  For example, if the initial window size allowed about 10 words per line of text, and then the window's width were doubled, then about 20 words per line would be displayed.

As large size, high-resolution displays become more common, the size of the computer users' workspace increases.  The width of a typical application window, such as a web browser, can now be over 1900 pixels, which allows for extremely long lines of text.  If a page utilizes tables or some other techniques to carefully arrange areas of text and images in anticipation of various window sizes, this may have positive or just no effect on the readability of the page.  However, if the page is not well formatted, it may become difficult to read in a very wide window.

One possible solution to the problem is to reformat the text into multiple columns, decreasing the length of each individual line.  If automated, this approach would allow users to freely resize their browser window, without affecting their comprehension.  Of course, users should be given an option to disable this feature in their applications.

In this study, we try to determine how much splitting text into multiple columns affects readers' comprehension and subjective satisfaction.  Our hypothesis is that when the line length exceeds a certain value, comprehension will improve if the text is split into multiple columns, and more so for wider pages.  It's supported by a number of psychological theories that have shown a correlation between how far the eye has to travel in order to go from one line to the next and the reading comprehension.  In general, the greater this distance is (which means longer lines of text), the lower the comprehension rates are.  Very short lines of text, however, do not yield better comprehension rates, and so one of the purposes of this study is to determine whether there is a particular combination of page width and line length that yields the best comprehension rate.

Dyson and Kipping conducted a study similar to this one in 1997, titled "The Legibility of Screen Formats: Are Three Columns Better than One?"  In their experiment, subjects read text displayed in a window with a single column layout using scrolling, single column layout using paging, and three-column layout using paging.  The reading speed and comprehension rates were measured.  It was found that a single column paged was read faster than either three columns paged or one column scrolled, while there was no significant difference between three columns paged and one column scrolled.  At the same time, the researchers did not find a significant difference in comprehension rates among the subjects, although faster readers also showed somewhat better comprehension abilities.  When the speed of reading and comprehension rate results were combined, there was no significant difference between formats.  However, all the subjects stated that the three-column layout was easier to read than the other two layouts.  The conclusions of the experimenters were limited to the facts that paging is faster than scrolling and that a single wide column is read faster than three narrow columns (when paged).  In our experiment, we attempted to control the width of the columns by varying the size of the text window between 600, 800, and 1000 pixels in order to find a relationship between the specific widths of columns and reading speed and comprehension rates, rather than merely varying the number of columns between 1 and 3.

Department of Computer Science: Direct questions and comments to the student editorial team