Re: Proposal: "text-transform" property revision

Brad Kemper wrote:
> I don't think it is any further out of line with CSS purpose than the 
> other text transformation values that exist today ("capitalize", 
> "uppercase", "lowercase").

I disagree. Those are clearly intended to be styling commands, not 
data-cleansing commands. The underlying HTML can be separated from the 
CSS, for example for exposure to assistive technologies, without losing 
content information. See also:

http://www.w3.org/TR/WCAG20-CSS-TECHS/#F2

In the HTML layer, text should be in "natural" case. Using anything else 
creates problems with consuming software that needs to transform the 
text to braille or speech, or match it to speech (e.g. voice browsers, 
screen readers, voice recognition software):

1. They may mistake uppercase text for acronyms and initialisms. For 
example, with the string "US CONCILIATORY OVER MISSILE PLAN" it's very 
difficult for assistive technology to tell whether that's "us" the 
pronoun or "US" the initialism. Conversely, they may fail to recognise 
acronyms and initialisms in lower case text.

2. Less seriously, they may emphasize uppercase text where it's 
inappropriate.

> Speaking for myself, I have had on several occasions been involved in 
> co-branded sites where I could provide a style sheet for the other site 
> to link to, but could not change any of the data itself. If those other 
> sites, which I wanted to conform stylistically with my own, decided on 
> all caps, there was really nothing I could do about it to make them 
> Title Case. On the other hand, if my site had used all caps for the 
> headlines, lets say, then I could have used a style sheet rule to make 
> their headlines all caps as well. Magic, but only in one direction, and 
> this new keyword would provide more parity.

I sympathize with your plight! However, I don't think it's the purpose 
of CSS to fix this problem any more than it's the purpose of CSS to 
correct spelling or provide missing alternative text for images. Data 
problems must be fixed in the data layer, or it's not a real fix.

--
Benjamin Hawkes-Lewis

Received on Wednesday, 10 October 2007 08:36:01 UTC