Designing for Unschooled Users
From CS260SP09
Designing for Unschooled Users
Readings
Text-Free User Interfaces for Illiterate and Semi-Literate Users Indrani Medhi et al., International Conference on Information and Communication Technologies and Development, 2006. ICTD '06.
Optimal Audio-Visual Representations for Illiterate Users of Computers Indrani Medhi et al., WWW Conference 2007, pp 873-882.
Background
Excerpts from Cognitive Development: Its Cultural and Social Foundations A.R. Luria, Harvard University Press, 1976.
Excerpts from The Psychology of Literacy Sylvia Scribner, Michael Cole, Harvard University Press, 1981.
Discussions
Please post your critiques/commments on the required readings below. To do that, first login by using your user name and password, then click the "edit" tab on the top part of this page (between the "discussion" page and the "history" page), New to wikis? Read the Wiki editing guide. . Hint - Please put a whole line == ~~~~ == (literally) at the beginning of your submitted critique, so the wiki system will index, sign and date your submission automatically.
Simon Tan 05:16, 29 January 2009 (UTC)
The fact that semi-abstracted drawings are better than photorealistic pictures at conveying (at least) healthcare symptoms is something that the Hesperian Foundation, a non-profit publisher of healthcare materials, has known for quite some time. They have invested much effort in doing similar studies to discover the best way to present their information within the context of a book, and are worth looking into if you are interested in healthcare for the developing world.
I was working on a way to convert their books into a multimodal experience with a "Digital Library" of sorts, and this paper is a perfect guide to doing so. It quantitively verifies that semi-abstracted static drawings + voice annotation is the best combination of presentation for illiterate users.
I find it interesting that all the numbers show that voice annotation almost always provides a positive effect on comprehension, but the authors downplay the bimodal presentation of information because it caused confusion with some of their participants (particularly those from economically worse slums). It seems to me that the next step with this research is to try a voice-only presentation and see how it compares.
Seth Horrigan 19:26, 30 January 2009 (UTC)
The reference to preference for semi-abstract cartoons over photographs or fully cartoon illustrations reminded me of similar conclusions found in a completely different domain. In Assessing the Effect of Non-Photorealistic Rendered Images in CAD from CHI 1996, Schumann et al found that sketches and non-photorealistic images had clear advantages over photorealistic or other CAD imaging. This was among highly educated, fully literate populations. Furthermore, in Identification and Validation of Cognitive Design Principles for Automated Generation of Assembly Instructions (Heiser et al, AVI'04), the research team found that hand-drawn step-by-step action diagrams were far more useful than text or photographic structural diagrams in assembly instructions. Certainly, the difficult in abstract symbolic representation is more likely in unschooled users ("roads can never be yellow and they are always black"), but the increased perceptual usefulness of sketches and "semi-abstract cartoons" is not restricted to unschooled users.
Also in the first paper, the researchers repeatedly emphasize that their research subjects did not have access to PCs. Why then is the PC the mode of providing the software? Who will maintain this software? How will it be useful to them? Certainly it is nice to know about how to design software on a PC for illiterate individuals, but wouldn't it be more sensible to look at a medium they might be able to use? For example, designing text-free interfaces on mobile phones as many ICTD projects have done before? Why design a system that will be impossible for the target users to ever access? The researchers described how the user testing was more effective when they allowed a more naturalistic setting (a group of women working together to understand and use the system) - and this matches the logical equivalent in western countries, where the trend is away from the artificial user testing formerly conducted in labs and towards ethnographic methods where the software can be observed in context (either through paired observation or remote usability testing). Why not extend this logic a small step further and develop something that the subjects really would be able to use?
KetrinaYim 23:17, 1 February 2009 (UTC)
What's most fascinating in the use of semi-abstract cartoons to depict objects and concepts is that it only takes a few lines to turn an image depicting an object into an image depicting an action. In the first paper, the authors explain the importance of including action elements in a drawing to depict an activity. An image of a faucet with dishes beneath it doesn't say much, but add a simple representation of running water in the faucet and the image suddenly becomes a symbol of dish-washing. Semi-abstract cartoons also go along with the fact that richer information isn't always needed; if added realism in the image doesn't improve understanding, just leave those details out.
And in response to at least some of Seth's questions, I can imagine the use of text-free interfaces as a gentler introduction to computers. I see value in how the tests presented text-free interfaces as a way to find better jobs. There could even be some organization out there that could use text-free interfaces as a gateway to literacy. It's not necessarily impossible for the mentioned individuals to access computers; illiteracy is the obstacle standing in their way of getting access. That and the occasional wariness of technology. However, I do agree with Seth's suggestion of text-free interfaces for mobile phones, which are generally more accessible than computers in developing nations.
On a side note, I was surprised that some people could have difficulty believing that the computer could help them find a job (to the point where someone would refuse to continue with the study!).
Himanshu Sharma 23:55, 1 February 2009 (UTC)
Even in the case of first study, application scenario seems to be very practical to me. Kiosks having computers connected to internet can be set up allowing illiterate individuals to perform job search. It seems more feasible than having the software designed for mobiles that should be having internet connectivity, ability to play videos, animations. It should be easier to have access to computers at some common location than having mobile with required capabilities for everyone.
These studies are focused on urban illiterates, but there are many more in villages. I am not sure if it can be a factor.
Kenrick Kin 08:28, 2 February 2009 (UTC)
Like the previous commenters, I thought the observation (even if previously well-known) that semi-abstracted drawings are better than photorealistic pictures was better for conveying information was interesting. Too much information can be confusing, and abstract drawings really distill what is important, while still being understandable. The success of sound seems obvious, because understanding speech is certainly independent of technology.
It's hard to imagine not understanding what a mouse click does, because it is so ingrained in our culture. For unschooled users, would touch interfaces be more "intuitive" than the indirect input of a mouse?
@Seth - I was never quite clear as to what the setup was that the subjects performed on, but I agree that mobile devices such as PDAs/phones might be a useful medium to explore and possible be more difficult if you need to fit a lot of images on a small display. However, it's not obvious if mobile devices would be any more accessible than PCs.
David (Tavi) Nathanson 09:17, 2 February 2009 (UTC)
In the first article, I found it fascinating that the initial graphic for a residence (the standard house symbol) was confusing to the subjects. The subjects *did* realize that it was the symbol for a hut/house; the problem was that the subjects didn't feel that it would make sense for their employers to *live* in a hut, as they would be living in an apartment complex! So not only are there representations that might be interpreted differently by members of another culture, but there are also representations that might be interpreted the *same* way by members of another culture, yet that interpretation might not make sense in light of other factors! This is not something that I've considered before; I typically only think about misinterpretations of the representations themselves.
In the second article, I felt the experiment selection was a bit strange. All 5 types of visual representations were paired with each type of audio representation (audio or no audio), but I wonder why no visual representations were paired together (i.e. video + static drawing)?
Nicholas Kong 10:33, 2 February 2009 (UTC)
@David Perhaps the hut/house symbol was a misinterpretation of the representation, in that the unschooled users took the symbol to mean something more specific than it actually was. That is, they thought the hut was actually a hut, instead of a representation of a dwelling in general. Also, I think no two visual representations were paired together because you can attend to an audio and a visual stimulus simultaneously, but not necessarily two visual stimuli.
In the audio-visual representation paper, I agree with Simon that the next logical step is to test audio-only. It seemed self-evident that audio significantly enhanced the reaction time in all cases, since the participants were being told exactly what the visual representation was meant to be. The data in Figures 1 and 2 also show much less variance in response accuracy/time with audio than without, so I wonder if the visual representation was rendered moot. However, the authors do point out that in some cases the subjects did not know the name of the condition (e.g. Kannada for lockjaw), in which case I would expect the audio-only case to do just as poorly as text.
Priyanka Reddy 11:55, 2 February 2009 (UTC)
I thought it was really interesting that in the second article, they actually tested the difference between designer expectations and the best representations. Rather than just figuring out what the best representations were, they were trying to find out how accurately they could have predicated those findings, which gives a lot of insight into how much we can use our own expectations when designing for unschooled users.
One thing I found weird was that they concluded with the sweeping statement that semi-abstracted static imagery accompanied by voice annotation is the best representation for illiterate users when I really think that their experiment showed that the best representation depends on that particular situation. Even though the quantitative results showed that static imagery had the greatest accuracy, other representations also had high accuracy. And, it could be possible that the questions that weren't correct with the static imagery may have been correct in the other representations, in which case the method really depends on the situation.
Brian Tran 17:05, 2 February 2009 (UTC)
I found it most interesting when the articles mentioned that audio and visual together usually led to the user allowing either audio or visual dominate (but usually audio according to the reading). I think it would be most helpful to have a mouse-over on an icon prompt an audio representation much like children's electronic books that output a sound clip when the child touches a picture ("The cow says moo"). This would allow the user to choose which sense they want to utilize for navigating the UI.
I completely agree with Seth on wondering why the researchers were designing a UI for people that do not use computers in the first article. It seems nonsensical to design a UI for people that will not use it anyways. On the other hand, I guess it could be argued that computer illiteracy would be a big reason of why these people do not use computers in the first place. It may seem that many of these illiterate people cannot afford their own PCs, but they still could possibly use publicly-owned computers for things like checking in at the front desk of hospitals if the UI of the software on that public computer is conducive for illiterate individuals.
Anuj 18:59, 3 February 2009 (UTC)
I believe the rationale behind not talking about mobiles to begin with was that the research team wanted to handle one problem at one time. Creating usable interfaces on a mobile is a huge problem in itself, and is something that has been puzzling researchers for some time now. I think there goal was to come up with design principles in these two studies and then may be later use the same principles or a modification of them on a mobile. However, I believe the researchers had an inherent assumption in saying that such interfaces can be designed at all. The research question should have been, "can there be an interface (text-free or otherwise) that can be usable for a first time PC user?".
