Response to Ofcom literature review
Ofcom, the U.K. broadcast regulator, published an Access Services Literature Review (PDF) in March 2006. The review was carried out by i2 Media Research.
The literature review is edifying and is itself a useful addition to the accessibility literature. However, its coverage of typefaces for captioning (errantly called “subtitling,” a mistake other U.K. and Irish publications also make) was nothing short of infuriating. A claimed literature review rubber-stamped the flawed “scientific” research concerning Tiresias Screenfont and completely ignored a published critique of Tiresias’s research and design. There is no credible research basis for Tiresias’s claims of legibility, readability, and suitability for captioning.
The Ofcom literature survey was published as part of a review of television accessibility services. One goal of that review was to identify future research. Verifiable scientific research on captioning typography is urgently needed and is an important and necessary addition to Ofcom’s plans. There is no such research originating in the U.K.
What the literature review says
i2’s literature review states, on p. 45 (footnotes deleted):
The Tiresias screenfont was carefully evaluated using people with a variety of visual and hearing impairments by RNIB and found to be useful. In its development, design considerations included the character shape, character weight (line thickness), intercharacter spacing, compatibility (in terms of maximum type size) with aspect-ratio switching, and character shapes that are difficult to distinguish. It has been adopted by the U.K. Digital Television Group as the standard font for interactive television and subtitling. In March 2000, the set of characters was expanded to cover 40 different languages.
“The Tiresias screenfont” presumably refers to Tiresias Screenfont (and not any of the other variants, like Signfont). The typeface absolutely was not “carefully evaluated.”
What’s wrong with the original Tiresias research?
This activity has extensively documented the shoddy “research” that purports to show Tiresias’s superiority over other fonts. The full documentation is as follows:
- Debunking the research on Tiresias Screenfont
- Design critique of Tiresias Screenfont
- Interview with John Gill (2001)
As detailed in the above documents, here is what really happened in the “research” on Tiresias Screenfont:
The typeface was tested:
- almost exclusively with old people (average age over 60 years);
- with comparatively huge numbers of people with visual impairments but no hearing impairments (35 out of 97 people);
- and with almost no nondisabled people, who were recruited as “escorts” or “signers” of the other subjects.
- To test extended reading of captions – titles that appear and disappear or scroll and disappear – researchers typed up a single sentence on paper and displayed it fixedly on a monitor. (Researchers did not even disclose what the sentence said.)
- A reading test in which type size is important allowed participants to zoom the type size if they wished.
- The comparison typefaces were, first, a dot-matrix font, whose appearance would be judged inferior to nearly any typeface shown at higher resolution, and, second, a print font. (The researchers admitted that print fonts are different from screenfonts but used a print font anyway.)
- The experimenters changed the experiment partway through: “A later version of the Tiresias Screenfont typeface was produced with improvements to the kerning after the testing commenced.” In other words, the first version of the font was so bad they had to fix it, yet the original experimental trials were not rerun.
- A completely different experiment was administered to hearing-impaired subjects.
- The experimenters admitted their research had no scientific validity: “It is freely conceded that this ‘testing’ is far from ideal and could even be described as anecdotal.”
- Viewer preferences were dismissed as “aesthetic except in rare pathological cases.”
- Italics were not tested. In fact, to this day there is no such thing as Tiresias Screenfont Italic or Oblique.
The lead researcher admitted ignorance of typography
In an interview, the lead author of the “research” on Tiresias, John Gill, stated “We did not know enough about the finer points of typography, to say the least.” This near-complete ignorance of typography in general, and onscreen reading and captioning typography in particular, did not prevent Gill from forging ahead with a project to create a new typeface. With so little expertise at hand in the research team, it is not surprising that the resulting typeface and the research ostensibly backing it up are so shoddy. What is surprising is that everyone in the U.K. and Ireland has bought what the researchers were selling.
And on that topic, Tiresias is a commercial exercise for Bitstream, the font’s codeveloper and vendor. A worldwide licence for Tiresias Screenfont can cost up to US$17,500, according to published rates. The Royal National Institute for the Blind cannot be viewed as an impartial observer on the topic of Tiresias’s suitability and effectiveness, as RNIB receives 40% of those licence fees.
The evidence for Tiresias Screenfont is junk science
Tiresias Screenfont was not comprehensively or credibly evaluated. The published reports of its efficacy are junk science.
If the task is to design a font for digital television that looks and performs better than the font used on analogue television, then nearly any font that isn’t blackletter or cursive will fare better in even the worst testing regimen. Most test subjects, if not all, will prefer any font other than the analogue one. The question becomes: Is the font under study, like Tiresias, actually better for captioning? That question has not been asked, let alone answered.
What is urgently required is a body of research, which Ofcom has an active duty to sponsor, to answer these questions:
- What is Tiresias Screenfont’s actual suitability for captioning? How does it compare to existing typefaces that expert typographers would nominate for caption usage, including typefaces custom-engineered for legibility and readability? (In other words, testing Times or Arial for captioning will not produce relevant results, since informed observers would never nominate those fonts for onscreen usage.)
- What characteristics are needed in caption typography for digital television?
- What are the repercussions of reusing analogue captions in digital form? In particular, what are the repercussions of impoverished character sets (missing quotation marks and dashes, missing musical staffnote ♪) and reduced typeface variations (absence of italics, or, more importantly, the use of exactly one font)?
- How do reactions, both objective and subjective, differ among deaf/hard-of-hearing, low-vision, and nondisabled subjects? For the last category, how do reactions differ for native English-speakers and ESL learners?
- Are new fonts needed, and if so, what characteristics should they have? What protocols should be used to test them?
Conclusion: You have been deceived
The U.K. broadcasting industry and Ofcom have been deceived by the junk science purportedly justifying Tiresias Screenfont. You have been taken in. Tiresias Screenfont is not suited to digital captioning; arguably it is not even “good enough” for such usage.
Even if Tiresias’s research basis were solid, there is no evidence that genuinely more readable and actually improved captions result from the adoption of a single typeface, without an italic, that is is used to re-typeset analogue captions.
Ofcom is petitioned to commission and pay for research into the true needs and requirements of captioning typography in the United Kingdom. Some research questions are listed above, but others may be relevant. It would be imprudent to muddy the waters by mislabelling captioning as subtitling, as they are two different things.
If the research concludes that new fonts are necessary, they should be commissioned and tested, again at Ofcom’s expense. To restore desperately-needed credibility, all results for any Ofcom-sponsored research need to be published in peer-reviewed journals. To avoid contamination by commercial interests, rights to the fonts should be held by Ofcom and licensed for free; as an alternative, the fonts could be placed in the public domain or given an open-source licence, again with no cost for licensing or usage.
The United Kingdom is home to several researchers with the ability to carry out such research (among others in different countries).