On Measuring the Impact of Hyperlinks on Reading

Gemma Fitzsimmons, Mark Weal, & Denis Drieghe

University of Southampton Contact email: G.Fitzsimmons@soton.ac.uk Introduction
We spend a vast amount of time on the Web and much of that time is spent reading. One of the main differences between reading Web and non-Web based text is the presence of hyperlinks There is an on-going debate about hyperlinks and whether they have a negative influence on reading behaviour Displaying hyperlinks in blue has become part of the online culture and most people would recognise a blue word on a Web page as a hyperlink [5] Carr [1] suggested that hyperlinks within the text are a distraction and therefore hinder comprehension of the text Other research has suggested that it does not disrupt reading, but actually assists in the retention of the hyperlinked word [6] Very little research has investigated how the presence of hyperlinks influences reading behaviour on the Web. This area is important to investigate because it has an impact on how hyperlinks are generated on Web pages Recording eye movements is an objective way of collecting data, providing an unobtrusive, realtime behavioural index of visual and cognitive processing [4, 7, 8] SR-Research EyeLink 1000 eye tracker was used (records 1 sample per millisecond) We based our methodology on that used by reading researchers, who use a number of eye movement measures to investigate reading behaviour Early measures which reflect any initial processing difficulties : Skipping probability - probability that the target word was skipped on first-pass reading First fixation duration - duration of the first fixation on a word Single fixation duration - where the reader made exactly one first-pass fixation on the target word. Gaze duration - sum of all first-pass fixations on the target word before moving to another word Late measures which reflect difficulties in integrating a word: Go-past times - the accumulated time from when a reader first fixated on the target word until their first fixation to the right of the target word Total reading time - sum of all fixation durations on the target word regardless of whether this reading happed during first pass or later Web Science

Method
Experiment One
Black Blue Green Red Grey

Experiment Two
High Frequent/Hyperlinked High Frequent/Unlinked Low Frequent/Hyperlinked Low Frequent/Unlinked

Figure 1. Example stimuli from Experiment One

Does a coloured word impair reading behaviour?

30 participants 30 stimuli (6 stimuli per condition)

Does a hyperlinked word impair reading behaviour?

32 participants 80 experimental sentences inserted into 20 edited Wikipedia pages (4 in each, 1 per condition)

Figure 3. Example stimuli from Experiment Two

Figure 2. SR-Research EyeLink 1000 eye tracker

Results
Experiment One Experiment Two

Figure 4. Scan path from a trial in Experiment One

Participants were less likely to skip a target word if it was any colour except black which suggests that the saliency of the colour draws attention to it (see Figure 7) The grey target word had significantly longer fixations across all eye movement measures due to its reduced contrast making it more difficult to process and read [2] (see Figure 8) The other coloured words were not fixated for longer than the black words, suggesting that colouring words does not either hinder or help the reading of those words
0.35 Skipping probability percentage First fixation duration (ms) 0.3 0.25 0.2 0.15 0.1 0.05 0 Black Blue Green Red Colour of target word Grey
260 250 240 230 220 210 200 190

Figure 6. Word Frequency x Hyperlinked/Unlinked interaction for Go-past times and Total Reading Time Figure 5. Scan path from a trial in Experiment Two

Black

Blue

Green Red Colour of target word

Grey

Figure 7. Experiment One skipping probability percentage

Figure 8. Experiment One first fixation durations (in ms)

Participants showed no difference in skipping probability There was a significant effect of word frequency across all eye movement measures, with the low frequency word being fixated for longer due to the increased difficulty [3] In the early eye movement measure (first fixation duration, single fixation duration and gaze duration) there was no difference in the fixation times whether the word was hyperlinked or not However, in the later measures (go-past times and total reading time) there was a significant interaction between whether the target word was hyperlinked or not, qualified by an interaction with frequency (see Figure 6) The low frequency hyperlinked words had significantly longer fixations than the other conditions suggesting that those words caused regressive eye movements due to difficulty processing. This means that participants are reaching the low frequency hyperlinked words and rereading the preceding content to re-evaluate it

Conclusion
Conclusions
Experiment One showed that a coloured word is skipped less often than a black word and that reduced contrast colours make reading more difficult [2] Experiment Two showed different effects of colour to Experiment One indicating that coloured words are processed differently to hyperlinked words Experiment Two also showed a difference between whether the target word was hyperlinked or not, qualified by an interaction with frequency. Low frequency hyperlinked words had significantly longer fixation times in the late measures of reading. Participants had difficulty with these words and would reread the preceding content to re-evaluate it Hyperlinks indicate that the word is important. When the hyperlinked word is a low frequency word the reader may wonder why that word is hyperlinked and want to re-evaluate the preceding content to make sure that they understood it, or try to decide why it is important

What does this mean?

These experiments have shown that coloured text does not hinder reading, but also that hyperlinks can cause us to reread previous content if the word is a low frequency/difficult word in order to re-evaluate the content In terms of Web design and layouts, the present results highlight the importance of carefully considering which words are hyperlinked in Web pages. The key lesson here is that Web designers should only hyperlink important words in pages, taking extra caution with words that are uncommon or that may be difficult to process

Future Research
By basing our future research on the vast amount of research already conducted on eye movements and reading we can build an understanding of how we read hyperlinked text In future research we aim to explore reading behaviour alongside the navigation and decision making elements that hyperlinked text entails
5. Nielsen, J. When bad design elements become the standard. Nielsen Norman Group (1999), http://www.nngroup.com/articles/when-bad-design-elements-become-thestandard/ 6. Nikolova, O. R. Effects of Visible and Invisible Hyperlinks on Vocabulary Acquisition and Reading Comprehension for High- and Average-Foreign. Apprentissage des langues et systmes dinformation et de communication, 07 (2004), 2953. 7. Rayner, K. Eye movements in reading and information processing: 20 years of research. Psychological bulletin, 124, 3 (1998), 372422. 8. Rayner, K. Eye movements and attention in reading, scene perception, and visual search. Quarterly journal of experimental psychology, 62, 8 (2009), 1457506.

REFERENCES 1. Carr, N. G. The Shallows. W. W. Norton, New York, NY, 2010. 2. Drieghe, D. Foveal processing and word skipping during reading, Psychonomic Bulletin & Review, 15 (2008), 856-860. 3. Inhoff, W. A., & Rayner, K. Parafoveal word processing during eye fixations in reading: Effects of word frequency. Perception psychophysics, 40, 6 (1986) 431439. 4. Liversedge, S., & Findlay, J. Saccadic eye movements and cognition. Trends in cognitive sciences, 4, 1 (2000), 614.