RE: 4.1 expert, general and non-reader

At 06:29 AM 2002-06-04, Lisa Seeman wrote:

>So if the subject matter takes intelligence (an programming for example)
>then assume intelligence, but not literacy or the ability to understand
>non-visual or literal information. A lot of Engineers are LD and have some
>autistic related impair mats (Asperses syndrome for example may be
>especially prevalent among engineers)

This discussion by Lisa is quite good.  I would like to introduce some related ideas:

Distinguish "this utterance" and "the [nominal] topic of this utterance."  Warning -- the latter is a colloquial approximation.  The actual meaning is not "a topic proper to this utterance" at all, but a topic that the utterance falls in, and which distinguishes this utterance from many, many others.  Let's use "nominal topic" for this supertopic that is appropriate to capture in metadata.

How one describes the nominal topic in a way that people can relate to.

How one describes the nominal topic in a way that machines can relate to.

The notion of the intended audience vs. the potential audience.  Multi-criterion profiles of comprehensibility.

Characterizing the "intended audience" by what they will most likely do with this utterance.

We should have a scale for general literacy dealing with the user as well as topicality and targeting that is utterance-centric.  The right policy involves a blending of both factors, both the utterance itself and the whole population in the web audience.

Topic description:
best current practice is keywords in html:meta in html:head of a web page.

These are natural language terms freely chosen without regard for controlled vocabulary, but putting them in the metadata as well as their appearance in the content at large marks them as having special significance for this page.

A possible success criterion for topic documentation would be something like:

There is a domain of discourse that can be recognized from a keyword set which is a subset of the keywords in the page metadata that:

- is not too broad:  The topic description should capture the dependencies of the exposition on reader qualifications, and relate strongly to the actual contribution of the full content of the resource.  People who have a reading knowledge of the topic as defined by the subset keyword tuple will more than 80% of them find this resource comprehensible without undue effort, and more than 5% of those who have expressed an interest in gaining information in the topic so stated would choose to dip into the resource to see if it is indeed something the wish to spend some time with.  Something like 1% or more of those who will answer 'yes' to "do you want information about keywordTuple?" should find the page worth browsing in depth.

- is not too narrow:  The terms used to characterize the topic are understood by essentially all who have a reading knowledge of the current topic, and preferable as broad as possible, either allied general fields or the general speaker population of the natural language at hand.

Then there is the matter of the intended audience.  I believe that the best scale here is a triage into scholar, practitioner, general.  The difference has to do with what the reader will do with the information.

The audience of scholars (of the topic) will themselves create new knowledge that extends what is published in the current utterance.  They need compactness and definition in the information representation, for the purposes of citation; to capture things that they don't themselves have to say in later work, but can cite.  In IEEE circles, this is the nominal standard of the 'Transactions' journals -- archival journals.

The audience of practitioners are those who have a need to apply the information in the current utterance in their daily life or work.

The general readership are those with a passing interest in knowing things.  This is anyone without an articulable reason for being in one of the above two classes.

This is how we qualify readers relative to the topic, which in the objective criteria of the authoring process is relative to the topic as declared and exported for example in a kewords metadata item.

We still need Jonathan's literacy scale which is again a triage:

literacy: extreme, general, limited

Finer gradations would be to sub-divide general literacy as to native speakers of the document language vs. non native speakers, subdivide limited literacy into partial and total reading difficulty etc.  But for the purposes of building profiles of how user-safe some web content is, it is probably safe to say that if we do testing on criteria that presume a man-in-the-street general knowledge of the language at hand, and other testing for how ready the content is to deal with someone with a literacy-reducing condition, we have enough diversity in the checks to get on with life.

The checks themselves, in success criterion terms, have to do with what lengths the user has to go to in order to meet with usability success:

- can process the resource in standalone fashion

- can process the resource with occasional side trips to un-linked but readily available forms of assistance: dictionary, web search, etc.

- can process the resource with systematic side trips to forms of assistance which are linked from the resource.

- can process the resource with the aid of assistance from products and services that are  available in two competing flavors on the open market.

- extreme accomodation:  The user can process the resource with industrial-strength helps that key off the web of components and associations documented in the resource, but pull in re-enforcements and re-flow the dialog structure intensively.  This may be the best we can do for the SLD user group on the Web at large that is targeted to a general audience.  This "try harder delivery context" could include telepresence of a mentor: see the teleCoaching application of the TeleRehabilitation RERC and the recent whitepaper on broadband and disability from Frank Bowe.

Note that "can be conveyed in markup" is pretty much only usable in the last two of these cases.  We should be getting content whose comprehensiblity is maximized in each of the following test modes:

- "just play it to me" animated tour through the content with minimal user intervention in the play sequence.

- nominal web-browsing mode:  user is facile with follow hyperlink action and back button, but if it's more than three links away, it might as well not be there.  Information that is within a radius of three hyperlinks may be depended on, and only that.

- expert infoSpace walker:
Users of 'advanced' query capability to formulate view-specifying report-specifications with selection and presentation aspects.  In some of our "extreme accomodation" scenarios, these queries may be supplied to a PWD user by a rehabilitation professional, reference librarian, etc.

>This is something I was working out with Jason.
>
>What is required by which site.
>
>In ideal terms I would say that the only thing you are guarantied is the
>ability to understand the subject matter.
>
>Thus, figure out what cognitive skills are required for understanding of the
>meaning of the page, assume nothing else.
>
>So if the subject matter takes intelligence (an programming for example)
>then assume intelligence, but not literacy or the ability to understand
>non-visual or literal information. A lot of Engineers are LD and have some
>autistic related impair mats (Asperses syndrome for example may be
>especially prevalent among engineers)
>
>A shopping site should assume that their audience understands what it means
>to shop, and pay, for items.
>
>If you are making a philosophy paper then you can assume the audience has
>intelligence and the ability to understand abstract non visual concepts.
>However one can still have hyperactivity or dyslexia and be in the audience.
>Ironically I (an heavy dyslexic) have a love of linguistics, and the roots
>and formation of language, So I guess dyslexia is everywhere. Hmm I guess
>word game sites can forget about us. (ever plaid scrabble with me - no
>rules)
>
>
>In practical terms of course there is something else going on. We live in a
>system of academic legitimization that is, to a large extent, dependent on
>literacy skills. So people in academia, with communication and learning
>disorders that effect literacy, must have gotten around it somehow. I chose
>engineering, for example, in part because less literacy is required to
>complete a degree. Does that mean i can not understand philosophy?
>Absolutely not. It does mean that I would not have gotten into a philosophy
>degree course without English or history A levels, and I could not get them.
>
>However the internet is and can do a lot to change that. I know, for
>example, a lot about disabilities and accessibility without having read a
>book on the matter. Adult education sites "should" (my opinion) be working
>to change this exclusion.
>
>So I think you need to work out what cognition skills are required to be
>interested (not qualified) in a subject and then you can assume that your
>audience has thoughs skills. If we go with this approach, then we need to
>construct a complete list.
>
>
>Please note that in asking people to use simpler words:
>-  this is only when meaning is not affected
>- and this can be achieved through rendering markup (such as ruby or the
>markup I am working on)
>
>In engineering were a high frequency is not exactly the same thing as
>"often", high frequency is the right term.
>
>
>
>-----Original Message-----
>From: w3c-wai-gl-request@w3.org [mailto:w3c-wai-gl-request@w3.org]On
>Behalf Of jonathan chetwynd
>Sent: Saturday, June 01, 2002 10:08 PM
>To: jasonw@ariel.ucs.unimelb.edu.au
>Cc: Web Content Guidelines
>Subject: Re: 4.1 expert, general and non-reader
>
>
>
>I think it is important that user testing is the criteria, who is to say
>for whom the material is intended. This is too easy a let out, and
>demeans us all.
>
>Every topic worth discussing has a nugget that is of general interest,
>and that is what we should encourage to be made available.
>or  obscure papers that have nothing to communicate to everyone, will to
>my mind never attain this. and why should they.
>
>2 examples (backed by user testing) that clearly might have a level 3
>conformance.
>
>many people have heard of einstein,  know of atomic weapons, and thus of
>the relation between mass and energy,  a few could possibly quote his
>formula.
>
>a more concrete example is http://www.respond.org.uk/help/index.html
>
>this is a brief document for 'discussion', it tries to reduce what is a
>very complex issue, that of personal relationships, to a few words.
>Obviously there are far more weighty tomes on the issues discussed, and
>yet they would be the more valuable if they contributed something to
>this version. Responsible authorities seeking level 3 conformance, are
>well able to create such a digest, and should be required to.
>
>Evidently one needs some means of identifying who the users might be,
>and that remains to be discussed
>expert, general, and non-reader(less than 30 words per page), may be a
>reasonable reflection of our 3 levels of conformance, for the present.
>
>
>thanks
>
>jonathan
>
>
>---
>Incoming mail is certified Virus Free.
>Checked by AVG anti-virus system (http://www.grisoft.com).
>Version: 6.0.368 / Virus Database: 204 - Release Date: 5/29/02
>
>---
>Outgoing mail is certified Virus Free.
>Checked by AVG anti-virus system (http://www.grisoft.com).
>Version: 6.0.368 / Virus Database: 204 - Release Date: 5/29/02

Received on Tuesday, 4 June 2002 08:50:15 UTC