Loading...
Please wait, while we are loading the content...
Similar Documents
Towards quantification of the effects of typographic variation on readability
| Content Provider | Semantic Scholar |
|---|---|
| Author | Berry, Karl Hargreaves, Kathryn A. |
| Copyright Year | 1998 |
| Abstract | Typographic wisdom holds that regular spacing and well-formed character shapes enhance readability. Are these components quantifiable in ways consistent with human vision theory? We construct texts using irregular spacings and distorted letterforms to verify and compare their effects. Our preliminary data indicate that shape distortion has more effect than spacing distortion. Beginning with classic work of Tinker [16], there has been an effort to verify and quantify the rather strong traditions of typographers regarding what makes text the most readable. In recent years, reading and vision researchers have attempted to examine the relative roles of visual and cognitive components of reading [13, 12, 4] and to explicitly examine typographic factors rigorously [1, 9]. Typographers hold that fixed-width fonts are more difficult to read than proportional-width ones, but the opposite was argued by Arditi et al. [1]. However, in [9] we suggested that poor letter shape explained their data, rather than visual crowding, as they suggest. In experiments similar to theirs but with letter shape controlled, we concluded that there is little difference in reading rates between fixedand proportional-width fonts at normal reading sizes, but that for well-formed characters at very small sizes proportional-width characters enjoy the advantages claimed by typographers. This finding, and some aspects (described below) of the spatial spectrum of text images has led us to search for quantifiable measures of letter shape and spacing in text readability investigations. This report describes our initial work in that direction. There is important evidence from the laboratory of Gordon Legge [6, 7, 2] that purely visual factors have a quantifiable effect on reading rate. Legge and his colleages have manipulated the spatial frequency content of text images and measured the effect on reading rate. This work represents a more controlled investigation than Tinker's and other studies which manipulate text size without quantified control of the spatial spectrum. At normal reading sizes and for normal readers, Legge et al. [6] found that low-pass filtering of text had no effect on reading rates as long as the cutoff frequency was above about 2 cycles per character. Since this is the frequency at which the strokes of characters will become indistinguishable, this work suggests that overall letter shape, rather than precise location of character edges, has the major effect on readability of letters. That conclusion is certainly in accord with type design principles, but in work described by Beckmann et al. at SID91 [2], that same laboratory reported a spatial frequency masking effect for text which suggests visual relations between the high and low spatial frequency content of text—relations which are not found in the visual system's response to simpler stimuli. In these experiments, they superimposed text filtered by a highfrequency band-pass filter on a different text filtered by a low-frequency band-pass filter. Each filter had about 1.5 octave width and the centers were about 2 octaves apart. They found that the high-band text masked the low-band text under normal viewing conditions. But varying the viewing distance—thereby effectively raising or lowering the center frequencies of the filters—can reverse which text is masked and which is visible. In any case, vision researchers find that for many vision tasks, processing takes place in spatial frequency tuned channels that are sufficiently independent so that stimuli 2 octaves apart cannot mask one another. The conclusion of Beckmann et al. is that reading is not such a process. In addition, classic work of Strohmeyer and Julesz found some masking of cosine stimuli by broadband noise distant in center frequency from the stimulus [15]. More recent work of Olzak and Thomas [10] does suggest that channel models may not be fully appropriate to describe the response to complex stimuli. One strongly held belief among type designers is that the inter-stroke distance, i.e., both between strokes within a character and between characters, in well-set text should be constant. This constraint is violated by fixed-width fonts, which is one reason they are regarded Towards quantification of the effects of typographic variation on readability Robert A. Morris Karl Berry Kathryn A. Hargreaves c 1993, Society for Information Display To appear, 1993 SID Digest of Technical Papers Special Session on Text Quality SID 1993, Seattle, May 16–21, 1993 Technical Report 92-6 December 1, 1992 (Revised March, 1993) Department of Mathematics and Computer Science University of Massachusetts at Boston Harbor Campus Boston, MA 02125-3393 |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://www.cs.umb.edu/~ram/dtrg/typography/sid93.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
