Acknowledgements: This paper summarizes a class research project.In addition to the above authors, John Aston, Molly Daniel, SophiaHussein, Jeffrey Oldenkamp, Ugochi Onyejiaka, Taylore Sloan, and ToddWard also made meaningful contributions to the project.
Return to class web page: http://hubel.sfasu.edu/courseinfo/SL03/sl2003front.html
Abstract
The purpose of this experiment was to determine how email textformat and instructions affect reading times, content retention, andreader preference, where text format is as follows: all capitalletters (AC), normal mixed case (MC), capital letter emphasis (CE),and spacing emphasis (SE). Fifty-seven students from an East Texasuniversity were presented with twenty adapted, non-classified U.S.NAVY emails; approximately half of this group was instructed thatthey would be asked questions about the content of each email, andall participants were asked subjective questions about formatpreference. A Web-based computer program calculated the timeparticipants spent reading each email and collected answers to allquestions. It was hypothesized the AC format would result in theslowest reading speeds, the most incorrect answers, and be rated asthe least preferred format. ANOVAs indicated neither text format norinstructions had significant effects on reading times or contentretention; however, Chi Square analyses indicated readers liked textwritten in all upper case letters significantly less than the otherformats, and that it was harder to find important information whenusing all capital letters. Thus, while the use of all capital lettersdoes not significantly affect accuracy or reading times, it mayaffect readers' likelihood to read an email in the first place.
Electronic messages are an increasingly important form ofcommunication. One major benefit is the immediacy with whichinformation is transmitted. Thus it is easy to understand whytime-sensitive information, such as meeting times and locations ormilitary orders, would be best disseminated via email. However, ifthe message text is not easily readable, and the reader must spendadditional time deciphering the content, the time-saving aspect ofelectronically sent information is lost. Readability is the ease withwhich a text may be comprehended (Samuels, 1983). Factors external toa reader, (e.g. text topic and structure, environmentalcharacteristics, and the goal imposed on the reader by an externalsource), and factors internal to a reader, (e.g. backgroundexperiences, knowledge base, and awareness of text structure), duallyaffect the reading process (Samuels). Internal factors could befurther categorized into physical and cognitive influences. Forexample, the structure of the eye, visual cortex, and other corticalareas involved in reading vary with the individual's learning style,word familiarity, and grammar usage to influence the reading process.The scope of the current research is limited to the effects ofexternal factors, (specifically letter case, text format, and"imposed goals") on reading times, comprehension and reader textpreference; thus no physiological functions will be measured.
The current study stemmed from sample U.S. NAVY emails; all NAVYemails are written in exclusively capital letters, and paragraphspacing is not always preserved upon receipt of a message. Theprimary purpose of this research was to determine if this inherentwriting style affects the time a reader needs to finish reading anemail as well as the ability to identify important informationpresented in an email. Secondly, readers' preference of text stylewas determined. Understanding these factors is important, especiallywhen little time is available to decipher an email message, as mightbe the case in hostile or unexpected military situations.
According to Samuels (1983), there is an interaction amongexternal factors to influence overall reading speed, comprehension,eye movements, and reading strategies. These factors include:illumination, print size, font style and legibility (letterrecognition), format design (page width, size, placement of text,etc.), the use of titles and headers, text style (word frequency,sentence construction, and text construction), and the goal ordirection imposed on the reader. A reader whose text has a lesslegible font style but important information separated spatially, mayhave an equivalent reading speed as another reader whose text has amore legible font but unorganized content. Because this interactionis present, the factors influencing reading speed, comprehension, andreader text preference are not segregated for discussion. Allinformation below refers to sighted readers with normal vision.
Historically, documents printed in English have been in mixed case(some upper and lower case letters), thus learning and continuing toread effectively requires some ability to recall differences betweenletters of different cases. MacKeben (2000) stated that without goodletter recognition, reading is severely impeded. With frequentexposure, the perceptual system becomes tuned to font regularities inorder to effectively process visual information (Sanocki, 1987 &1988 as cited in MacKeben). These regularities act as cues in wordrecognition, and can include letter ascenders (vertical extensionsabove the letter body), descenders (vertical extension below theletter body), curves, dots, and overall word shape. The shape of alower case word provides significant clues to the reader as to itsidentity, whereas a word in all capital letters has a destroyed shape(no ascenders or descenders), so the word appears to be rectangular(Williams, 2000). When unable to use previous experience andknowledge of word shapes, reading time slows as a result of slowedprocessing. Upper case type is read about 13 percent slower thanlowercase type (Tinker, 1955 as cited in Williams).
Additional problems arise when reading text presented inexclusively capital letters due to the amount of similarity betweenthese letters. Briggs and Hocevar (1975) conducted a study in whichthe letters of the capital letter alphabet were defined based oncommon and distinctive features such as horizontal angularity(horizontal line at the top, middle, or bottom of letter) andcurvature (small curve-convex right, closed curve, etc). Each letterwas tested for confusability with all other letters. Their studyshowed more effort was required to decipher the differences betweentwo capital letters, because they are more similar to each other(have fewer distinguishing features). Staats (1968) too suggests thereading process is affected by the difficulty in learning todistinguish between similar letters in the English alphabet.
Finally, upper case letters are simply larger than thecorresponding lowercase letters. With each eye fixation, a limitedamount of material can be resolved, recognized, and used to guide thereading process. Letter distinctions can be made for four to sixletters from the fixation point, but seven to fourteen letters awaycould be useful for planning the next eye movement, or saccade(McConkie & Rayner, 1975 as cited in Bailey et al., 2003). Thus,the size of the text can influence the efficiency with which thevisual system combines previous information with current informationand plans to incorporate new information by controlling eyemovements. Interruption of this process with the introduction ofdifficult-to-read text may influence the frequency of eye movementsand could ultimately affect reading speed.
In addition to text case affecting reading speed, textorganization can affect reading speed and content retention. Forexample, aesthetically pleasing visual organization and design canimpact cognitive learning (Martin, 1986 as cited in Haag &Snetsigner, 1994). In fact, Samuels (1973) believes textconstruction/coherence is the most important aspect of readability.In describing information presentation on web pages, Williams (2000,pp.384) stated "Good design reveals structure when it visually mimicsthe logical relationships that exist among elements…the humanvisual system attempts to find the structure of information… itdoes so by looking for visual patterns." Logical relationships areassigned based on a text's visual characteristics, and readers assumeplacement of any element on a screen as intentional and as a result,try to assign meaning to it (Williams). Williams identifies spacingas an important tool for organizing information, suggesting thevisual system tries to group elements that are close together orseparated by white space. This spacing simulates a "blocking" visualeffect, helping the reader break down the information into smallermore manageable pieces of information, and in turn can facilitatebetter content comprehension. In addition to manipulating text case,the current study will also include conditions where some text isemphasized using either all capital letters or spacing. This allowedus to examine whether the use of selective text emphasis would affectboth reading speed and retention of emphasized and non-emphasizedinformation.
As discussed above, text readability can be affected by aninteraction of many format and organization factors. However, fewstudies have focused on how more subjective influences (i.e. readerpreferences) impact text readability. Gump (2001) conducted a studyin which participants rated various fonts as either easy or hard toread and indicated the "mood" created by each of the fonts. Whileninety-eight percent of participants rated Arial as easy to read,approximately seventy percent of participants rated Arial as plain.The findings of this survey suggest differences in assessments offont readability and basic font aesthetics. As noted above, textorganization and layout influence aesthetics and retention. Thus, itis possible that basic font styles and letter formats (all capitalletters versus normal mixed case) may also influence retention andreading speed simply because they influence a reader's subjectiveresponses. To determine if text format affects readers' subjectiveresponses as well as objective performance (reading times andretention), subjective questions will also be included.
Finally, reader motivation will be manipulated in order todetermine its influence on reading times. Previous research onreading comprehension indicates the purpose of the reading task maydictate how quickly a participant reads text; thus affecting overallreading performance. Samuels and Dahl (1975, as cited in Samuels,1983) found that when participants were issued reading tests andasked to read for a general as opposed to detailed information, largedifferences in reading speed and amount of information learned wereproduced. Because participants did not know how detailed thepost-reading questions would be, they did not know whether to readquickly or carefully (Samuels & Dahl). This research suggests ifa reader is looking for specific information in preparation of testquestions, more time is required in gathering more information forretention. Therefore, one half of the current participants willanswer retention questions following each email, while the remainingparticipants will read the emails without answering any retentionquestions.
Due to word shape unfamiliarity and a resulting difficulty readingall upper case text, we hypothesized participants would yieldsignificantly slower reading times and more incorrect answers whenreading all upper case text than when reading the emails presented inmixed case. We anticipated that participants who knew they would bequestioned about the email content would have significantly slowerreading times than participants who would not be questioned, becausethey would spend more time reading and trying to remember detailsabout the text. Additionally, we hypothesized that participants wouldhave faster reading speeds and more correct answers when answeringquestions about emphasized text, than when answering questions abouttext with no distinguishing features. Finally, it was hypothesizedthat text written in exclusively upper case letters would be theleast preferred style as a result of this format being the mostdifficult to read.
Participants
Sixty participants from a mid-sized, East Texas university wererecruited from the Psychology department student subject pool. Amajority of the participants in this study were between 18 and 20years old. Participants were compensated with a one-hour experimentparticipation credit and were required to have 20/20 vision normallyor corrected.
Design and Materials
The current research employed a 4x2 mixed design. Because thesample texts were adapted from NAVY emails written in Arial 10 pointtype using black text on a white background, the current testconditions used the same font, size, and colors. The independentvariables were Text Format (all capitals (AC), mixed case (MC),capital letter emphasis (CE), and spacing emphasis (SE)), andInstructions (participants questioned about the material they read(Q), participants not questioned (NQ)).
The following describes each level of the Format variable: AC-email text written exclusively in upper case letters with singleblank lines between paragraphs (ex: THIS IS JOHN'S BOOK); MC-emailtext written in standard format (some lower and some upper caseletters) with single blank lines between paragraphs (ex: This isJohn's book); CE-email text written in standard format with foursentences or phrases in upper case letters and single blank linesbetween paragraphs; SE-email text written in standard format with anadditional blank line separating the same information capitalized asin the CE conditions.
Twenty NAVY emails were each adapted to the four formats describedabove (AC, MC, CE, SE). Each participant was presented with twentyemails; five in each of the four formats. The specific emails thateach participant received in each format was counterbalanced acrossparticipants. Thus, eight versions of the program were created toeliminate any effects of email length and content difficulty.Versions one through four were identical to versions five througheight, however they did not contain questions. To minimize ordereffects within the experimental versions, the experimental computerprogram randomized the order in which the emails were presented toeach participant.
All participants received all treatment combinations for theFormat variable and only one level of the Instructions variable. Thethree dependent variables were reading speed, content retention, andthe reader preference of text format. A participant's reading speedwas measured in milliseconds and calculated as the time between thecomplete loading of a web-page that displayed a single email and theparticipant's request to go to the next page. The timer began as soonas a page was completely loaded (based on the browser and not theserver from which the page was hosted) and ended when the participantclicked a button at the end of the email. Content retention wasmeasured by the number of correct answers to questions pertaining tothe text content. In the Q condition, each email was followed by fourmultiple-choice questions presented on the computer screen andanswered electronically. To better estimate the effects of text caseand emphasis on comprehension in the CE and SE levels, two of thefour answers came from capitalized/spatially separated information,while the remaining two came from information not emphasized bycapitalization or spatial separation. At the end of the twenty emailswere four subjective questions about the readers' preference of textformat. Participants indicated their most (SubQ1) and least (SubQ2)preferred formats and the format they thought was the easiest (SubQ3)and most difficult (SubQ4) to read.
The twenty non-classified NAVY emails were incorporated into acomputer program executed via the World-Wide-Web. The emails variedin their length, with a minimum of 84 words and a maximum of 844words (median of 273). A laboratory with 20 single-user computerworkstations was used to collect data. Each computer was equippedwith Internet access to the program developed for this study. Theprogram collected information about each participant's reading time,content retention, and format preference.
Procedure
Before beginning each experimental session, the computers wereturned on and one of eight versions of the data collection programwas started on each machine so that the experiment instructions weredisplayed. After participants were seated, they were given consentforms and provided with answers to any questions they had.Participants were then given instructions emphasizing therelationship between the experiment and the NAVY, the importance ofusing three initials for data coding purposes, and that the on-screeninstructions would indicate whether or not there would be questionsafter each email presentation. Participant then entered theirinitials and completed the twenty trials. At the end of all twentytrials, each participant was presented with the four subjectivequestions about their preferences among the four formats. Finally,participants were debriefed and provided with documentation ofexperiment completion. Participants took between 10 and 45 minutes tocomplete the experiment.
The scores for three participants were not included in the dataanalysis because they did not follow instructions. Of thoseremaining, 28 participants received experiment versions includingquestions and the remaining 29 received versions without questions.One ANOVA was performed for the dependent variable of reading speedand another was performed using question accuracy (contentretention). For each participant for each condition, the medianreading speed, the total number of correct answers from emphasizedtext, and the total number of correct answers from non-emphasizedtext were calculated; these values were used in the ANOVAs.Chi-square analyses were used to analyze the subjective data.
A 4 (AC, MC, CE, SE ) x 2 (Questions, No Questions) ANOVA revealedno significant effects of text format on reading times. See Figure 1for a graph of mean reading times (seconds) for all conditions. Although it was not significant, there was a tendency for those whoreceived questions to read emails using all capital letters orcapital letter emphasis more slowly than those not receivingquestions. Overall, emails using normal mixed case format were readmost quickly.
Figure 1 - Reading Times (seconds) as a function of textformat and instructions.
A 4 (AC, MC, CE, SE ) x 2 (Emphasis, No Emphasis) ANOVA alsorevealed no significant effects on accuracy. See Figure 2 for a graphof accuracy scores for all conditions. Although it was notsignificant, there was a tendency for the spacing emphasis to improveaccuracy for the emphasized questions, but decrease accuracy fornon-emphasized questions.
Figure 2 - Accuracy as a function of text format andinstructions.
Before conducting Chi Square analyses on each of the foursubjective questions, the answers were tallied for each question (seeTable 1). Only 56 participants were included because one participantneglected to answer the fourth subjective question. All fourquestions showed significance at the p<.005 level. Participantsmost strongly preferred (easiest to read) the two formats usingemphasis. The all-capital-letters format was the single leastpreferred format of the four format types. A similar pattern wasfound for the formats for which it was easiest to find importantinformation (capitals emphasis and spacing emphasis), and for whichit was the most difficult (all capital letters).
Table 2 - Tally of Answers to Subjective Questions.
| Question 1 | Question 2 | Question 3 | Question 4 | |
| | Format Most Preferred (Easiest to read) | Format Least Preferred (Hardest to read) | Format Easiest to find Important Information | Format Hardest to find Important Information |
| AC | 4 | 46 | 5 | 46 |
| MC | 7 | 6 | 4 | 5 |
| CE | 22 | 2 | 27 | 1 |
| SE | 23 | 2 | 20 | 4 |
Although no significant difference was found between groups withrespect to reading times, there were in fact trends that werenoticeable and worth discussing. E-mails that were formatted usingnormal mixed case tended to be read more quickly than the other threeformat options. Perhaps this trend existed because readers are moreaccustomed to reading mixed case text, and because the use of lowercase letters allows word shapes to be used to help recognize words.In contrast, readers tended to take the longest to read text that waspresented using all capital letters, where word shape cues weremissing and familiarity was not as strong. It is also important tonote the interaction trend between format and the presence or absenceof questions. Readers took their time while reading e-mails in allcaps or e-mails with case emphasis if they knew they would be askedquestions. However the presence of questions had no effect on readingtime of e-mails in mixed case or spacing emphases. These resultssuggest that text presented in capital letters may require moreattention and focus in order to accurately retain information.
With respect to accuracy, the trend regarding the use of spacingemphasis is also noteworthy. It seems that while spacing emphasissomewhat helped readers accurately recall information of theemphasized material, it seemed to hinder their recall ofnon-emphasized material. It is possible that the emphasized materialcaused readers to pay close attention to that information and in theprocess of doing so may have caused participants to inadvertentlydisregard information that was not emphasized. Therefore, it iscrucial when emphasizing material to be sure and emphasize all of theimportant material. A similar trend did not exist for the emphasisconditions using all capital letters. This lack of effect for the allcapital letter emphasis may have occurred because these particularemails often contained Navy acronyms in all capital letters. Thus,readers may not have been able to efficiently scan for and use theall capital letters emphasis and, in turn, they effectively ignoredit.
In contrast to the objective measures, participants showedstrongly significant preferences for the different formats.Participants liked and found it easiest to find important informationusing mixed case with capitals emphasis and mixed case with spacingemphasis. Further, they specifically disliked and found it difficultto find important information using the all capital letters format.While they did not strongly like the normal mixed case, they also didnot strongly dislike it. The fact that they least preferred the allcapital letters format may be due to the fact that this is the mostunfamiliar way to read text and that reading this format was morefrustrating due to the lack of word shape cues.
Although readers did not perform significantly differently oncomprehension tasks as a result of emphasized text, they still feltthat it was easiest to find important information in a paragraph thatcontained emphasized sections. In the future, experimenters may wantto require participants to read material with more difficult content.Emphasis may have more of an impact on the reading comprehension ofmore complex messages. Additionally, the use of the all capitalletters emphasis should be tested using emails that do not containacronyms.
The results of this project have implications regarding the Navy'suse of the all capital letter format. Because subjective responsescan influence the likelihood of reading and attending the textcontent, it might be beneficial for the Navy to consider changing itscase format. The trend toward an interaction between format andanswering questions about the email supports this conclusion.Participants only took extra time to process the all capital lettersformat when they knew they would be questioned over the content.Without questions, they spent the least amount of time reading emailsusing that format, most likely because they did not like reading thatformat. Because readers generally are not quizzed over e-mailcontent, the Navy might inadvertently be encouraging its personnel tosuperficially skim any administrative e-mails not prejudged to beimportant to the reader.
Because emails are so widely used, the results of this experimentalso extend beyond the realm of the Navy. General users sometimeswrite emails in all capital letters, either to suggest emphasis (suchemails are often interpreted as "shouting"), or because it is easierto type without using the shift key as often. Such emails may not beread as efficiently, or readers may choose not to read them at all.In general, it is important to know how to best present e-mails tooptimize the transfer of intended information.
Bailey, I.L, Lueck, A.H., Greer, R.B, Tuan, K.M., Bailey, V.M.,& Dornbusch, H.G. (2003). Understanding the relationships betweenprint size and reading in low vision. Journal of Visual Impairments& Blindness, 97(6), 325-334.
Briggs, R. & Hocevar, D.J. (1975). A new distinctive featuretheory for upper case letters. The Journal of General Psychology, 93,87-93.
Gump, J. (2001). The readability of typefaces and the subsequentmood or emotion created in the reader. Journal of Education forBusiness, 76(5), 270-273.
Haag, B. B. & Snetsigner W. (1994). Aesthetics of screendesign: an integration of principles. In D. G. Beauchamp, R. A.Braden, & J. C. Baca (Eds.), Visual Literacy in the Digital Age:Selected Readings from the 25th Annual Conference of theInternational Visual Literacy Association (pp. 92-97). Blacksburg,VA: International Visual Learning Association.
MacKeben, M. (2000). Enhancement of peripheral letter recognitionby typographic features. Visual Impairment Research, 2(2) 95-103.
Samuels, S. J. (1983). A cognitive approach to factors influencingreading comprehension. Journal of Educational Research, 76(5),261-266.
Staats, A. W. (1968). Learning, Language, and Cognition. New York:Holt, Rinehart, and Winston, Inc.
Williams, T.R. (2000). Guidelines for designing and evaluating thedisplay of information on the web. Technical Communication, 7(3),383-396.