Re: language and tool results

Hi Shawn,

I don't agree that all of the highlighted sentences are OK as you mention near the end of your email.  Before I get into why, I don't think anyone is going to get hung up on a single sentence.  If the sentence is the best it can be, they will override what Hemingway is saying.  I do think it's very important for editors to seriously consider what Hemingway is saying as a whole, and not discount it because it seems harsh in a situation pulled out of context.  The app is objective and we should at least have thought it through before deciding to leave sentences red.

If that sentence you experimented with was the only sentence in the document with the issue of a long comma separated list, it would be easy to say ok, Hemingway, we have few red sentences and the reading level is pretty low anyway, so we're gonna call it done.  But, the issue is that the document contains many, many of those lists in sentences (sometimes with more than one per sentence).  I'd say it's "scoring poorly" for a good reason and that it's not OK in this case.

There is so much crammed into those sentences, they really are very hard to read.  I pulled out what lists jumped out a first glance below.  It may not be what we should do, but I think the reason Hemingway likes the bullets is that pulling all those bulleted items out of sentences into lists allows users to absorb the items voluntarily, skim, or skip them as needed.  Leaving them in the sentences below forces the users to traverse the "wall of text", or give up reading…

Hemingway lists almost half of the sentences as very hard to read, with almost a third of the sentences in the document containing these lists:

  *   hardware, software, language, culture, location, or physical or mental ability.
  *
  *   range of hearing, movement, sight, and cognitive ability.
  *   when websites, web technologies, or web tools
  *
  *   older people, people in rural areas, and people in developing countries.
  *   mobile web design, device independence, multi-modal interaction, usability, design for older users, and search engine optimization (SEO).
  *   better search results, reduced maintenance costs, and increased audience reach, among other benefits
  *   social, technical, financial, and legal benefits
  *   people who are blind, as well as to people who turned off images on their mobile phone to lower bandwidth charges, people in a rural area with low bandwidth who turned off images to speed download, and others.
  *   web tools, education, and development processes.
  *
  *   designing, writing, and developing for web accessibility.
  *   together people from industry, disability organizations, government, and research labs from around the world
  *   auditory, cognitive, neurological, physical, speech, and visual disabilities
  *   'web content' (websites and web applications), authoring tools (such as content management systems (CMS) and blog software), browsers and other 'user agents', and W3C technical specifications, including WAI-ARIA
  *   by implementing, promoting, and reviewing guidelines; contributing to the WAI Interest Group; and participating in Working Groups
  *   accessibility standards, education, testing/evaluation, project management, and policy.

Rather than discount the app because it favors lists, I think we should be looking at the bigger picture it is communicating.  The document is quite complex.

Regards,
James

On 7/28/17, 12:12 PM, "Shawn Henry" <shawn@w3.org<mailto:shawn@w3.org>> wrote:

Hi James, Sharron, and all,

We might want to caution folks not to get too hung up on the HemingwayApp results.

I was a surprised at the HemingwayApp results on some of the WAI content. I've been experimenting with it and the content from https://w3c.github.io/accessibility-intro/ I tried several types of edits on the first section to get better results -- which didn't do much -- and then tried the lists -- bingo.

An example issue with HemingwayApp results is inline lists.

Starting text: "The Web is fundamentally designed to work for all people, whatever their hardware, software, language, culture, location, or physical or mental ability. When the Web meets this goal, it is accessible to people with a diverse range of hearing, movement, sight, and cognitive ability."
- Grade 10
- 1 of 3 sentences is hard to read
- 1 of 3 sentence is very hard to read
The whole thing is highlighted. :(

Deleting lists (which we wouldn't do here -- this is to show how much impact just a list has):
"The Web is fundamentally designed to work for all people, whatever their hardware. When the Web meets this goal, it is accessible to people with a diverse range of cognitive ability."
- Grade 6
- 0 of 3 sentences is hard to read
- 0 of 3 sentence is very hard to read
For this paragraph, the lists alone were adding 4 grade levels and giving the highlighted "hard to read"s.

Changing in-line lists to bullets:
"The Web is fundamentally designed to work for all people, whatever their:
* hardware
* software
* language
* culture
* location
* physical or mental ability
When the Web meets this goal, it is accessible to people with a diverse range of:
* hearing
* movement
* sight
* cognitive ability
"
- Grade 5 (taking out "fundamentally" dropped it to 4)
- 0 of 13 sentences is hard to read
- 0 of 13 sentence is very hard to read

For this specific paragraph, my perspective is that we do *not* want to bullet out at least the first list -- because it is a very minor point of the overall page -- just setting the stage, and we don't want them emphasized.

---

If someone just looks at HemingwayApp results on this paragraph, they could critique it. However, the inline lists are the main cause of high reading level and highlighted sentences, and those are OK in this case.

But maybe others won't get hung up about it so no need to say anything. :-)

~Shawn