W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > October to December 2000

RE: proper use of abbr and acronym

From: Al Gilman <asgilman@iamdigex.net>
Date: Sat, 18 Nov 2000 13:06:05 -0500
Message-Id: <200011181734.MAA103593@smtp2.mail.iamworld.net>
To: "WAI Interest Group \(E-mail\)" <w3c-wai-ig@w3.org>
[Kynn gave an excellent summary 
on this point which is all anyone really needs to know.  

Here, for those interested in reading it, is a more prolix and pedantic
in substantial agreement with what Kynn said. -Al]

This is indeed a FAQ.  If you review the bidding with


It says:

 About 680 documents match your query.

Please see item 13 in

 WAI Page Author Guidelines Working Draft Changes

At the present time, the content guidelines working group is unaware, so
far as
I know (unofficial results based on searching issues lists) of an
rationale for distinguishing between acronyms and abbreviations in HTML
type or in any WCAG guideline or checkpoint.  The last record I see of this
issue being touched is the above, where the resolution was that for
accessibility there is no distinction worth making between acronyms and other

The accessibility dependency would appear to be limited to the following:

Lexemes (strings that parse like a word) that are in certain categories are
prima_facie suspect of not being in the text-to-speech technology's
pronunciation dictionary or fit the algorithm used for a given language for
items not in the dictionary.  Here it is important to give supplemental
information.  Abbreviations and acronyms are both in this category.  To apply
the relevant rules for accessible markup, it is not necessary to recognize any
distinction between abbreviations and acronyms.

This is fortunate, because in common usage there are multiple theories as to
what abbreviations are acronymns, without a clear consensus.  In particular,
the usual suspects for definition of the dividing line between these two
[nested] cases in English do not port well to other natural languages.

So, unless someone can come up with a clear articulation of a compelling
accessibility case for this distinction in the markup to be defined and
enforced, the WAI-PF group will probably continue to advise the HTML Working
Group that this distinction is "semantic sugar" to appease those who believe
there is a difference.  The format itself recognizes the difference in name
only.  There are no defined semantic differences between these two HTML
types that are actually of the format specification beyond the tokens used as
element type name.

There is at least the following potentially machinable use for a distinction
between acronyms and other abbreviations:  The definition or explanation of an
acronym is better if it is able to be displayed with the letters in the
emphasized in the presentation of the fully spelled out equivalent, according
to a mnemonic mapping of the letters into the definition.  This mnemonic
pattern of which instances of those letters are emphasized should be
for all times the definition is so presented.  

One could value the same sort of "form of the definition entry" application to
the 'acronym' concept if you tied it to pronunciation.  Then the defintion of
an acronym would require a say-as pronunciation entry where the definition of
an abbreviation would not.  Then again, the say-as property or method is still
defined; it defaults to the fullword expansion.  

But so far as HTML and access rules and their enforcement are concerned, this
distinction would appear to be in the "You just don't want to go there"
category.  It is a distinction that _should_ be "under our radar."

The best way to increase the light:heat ratio in this area, may be by an
explicit information model.  A model that is not keyed off the pronunciation
test is the following.

An abbreviation is a short spelling which has an associated equivalent long
spelling.  The definition of an abbreviation is a relationship or
correspondence between a short form and a long form.  They are equivalent as
references.  Both the short and long form refers to one thing, the same thing
for either form.

An acronym is a special case of an abbreviation, where in the information
pattern of its definition there is in addition to the above structure, a
mapping of the letters in the short form to specific occurrences of the same
letters in the long form and this mapping is considered to be of mnemonic

In addition to "acronyms are initialisms and their close kin" there are some
other statistical trends that should be noted:

The preponderance of letter-inclusions in acronyms come from word-initial
letters, with a significant but minority participation from a) other letters
immediately following letters already included and b) other letters of
convenience to make a pronounceable token.

Abbreviations for long compound terms are often formed as acronyms.  

Abbreviations for common terms are usually not called acronyms.  I can't think
of an abbreviation for a single word that should be called an acronym. 
Although the current usage of i18n etc. initial-count-final symbols comes
close.  Although the French abbreviation Mlle. consists of a filter of the
string 'Madamoiselle,' the usage is so common that I don't think that spelling
the long form MadamoiseLLE would be considered of significant mnemonic

There is a reasonably widely held theory that to be an acronym, an
must have a pronunciation other than its spelling.  Kynn supports this theory
in an earlier post, as does Charles here.  


My personal preference is to lean toward the construction-based definition,
with the mnemonic value of the letter selection pattern as the primary test,
not the pronunciation test.  

The mnemonic value of the etymological derivation acro-nym i.e. "'tall' name"
is that it is a name composed of "the tall letters."  This is essentially
synonymous with the term 'initialism.'

As Sean pointed out, Webster's Collegiate does not impose a pronunciation


[Kynn PS:]  The reason we have both is that we have people literate in Latin,
French, and Greek all influencing the lexicon in English.  Not necessarily
we had a particular distinction we needed to encode.

please note, however, that the phraseology "XUL is an acronym, rather than an
abbreviation" is logically ill-formed, as acronyms are a special case of
abbreviations.  It would, on the other hand, be logically well-formed to say
"XUL should be marked as an acronym, rather than as an abbreviation" because
the markup syntax does reduce this to an either/or choice, and one may feel
that there is added value in using the narrower concept (whatever _that_


At 07:33 PM 2000-11-17 -0800, Charles F. Munat wrote:
>Sean B. Palmer wrote:
>"In specific, it appears to me that XUL is an acronym, rather than an
>Only if it is pronounced as a word ("zuhl"), rather than spelled out. An
>acronym is a word formed from the first letter or first few letters of other
>words. For example, RaDAR from RAdio Detecting And Ranging or CReEP from
>Committee to RE-Elect the President (remember that one?).
>So XML, HTML, W3C, XHTML, etc. are abbreviations, not acronyms.
>SMIL and SQL could be considered acronyms if they are pronounced "smile" and
>"sequel" respectively. If they are spelled out, then they are abbreviations.
>Frankly, so few people seem to understand this that acronym will shortly
>lose its meaning (sadly). My guess is that the originators of the HTML
>elements knew the difference and wanted to indicate when an abbreviation was
>an acronym, perhaps to provide a clue to how it should be pronounced. If it
>were up to me, we'd teach people what makes an acronym special when we
>taught HTML. I'd even put an explanation in the standard. But then, I'm the
>kind of guy who winces when people pronounce forte for-tay (it's "fort"),
>gnashes his teeth when he hears "rather unique" (there are no degrees of
>unique), and becomes positively homicidal when confronted with statements
>such as "all cars are not blue" (oops, wrong: some cars ARE blue, but
>perhaps they meant "not all cars are blue"). So feel free to ignore me...
>Charles F. Munat,
>Seattle, Washington
Received on Saturday, 18 November 2000 12:34:01 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 13 October 2015 16:21:10 UTC