This is part 2 of 2. Part 1 is Why use Semantic HTML (HyperText Markup Language)?
This guide only deals with HTML4/XHTML elements which have a specific, human-readable meaning. The semantics of elements such as link
, which are not seen in normal browsing, have been left out, as have replacement elements like img
or object
. In some cases, I’ve also addressed specific attributes which are critical to providing semantic value to an element.
This is not a guide which demonstrates the opinion of the W3C (World Wide Web Consortium) as represented in the HTML, XHTML (eXtensible HyperText Markup Language - HTML reformulated as XML (eXtensible Markup Language)), or HTML 5 specifications. This is a practical-use guide which indicates my reasoned opinion concerning the best use of each element.
Core Block Elements
div
- The
div
element represents a discrete section of a page which can be meaningfully divided from the content around it. Commonly used to indicate a header region, footer, sidebar, or navigation region; it’s use can extend equally to indicate columns on a page or sections of an article. The element is also commonly used in multiple layers to group lower-level sections together, such as a “content” section which groups a main article, comments on that article, and meta data about the article or author. h1-h6
- The six levels of headings are all used to introduce sections of content (containing
p
(paragraphs),div
(page divisions) or other content) which they describe. They’re perhaps most accurately compared to the structure of an outline:h1
is the top level heading element. The only heading element which can follow anh1
ish2
.h2
, on the other hand, can be followed by either an additionalh2
, if the sections are equivalent and both fall under the precedingh1
topic; anh3
if the following section is logically a child of theh2
, or anotherh1
if the following section is a new topic of the same level of specificity as the first heading. A common preference (although certainly not mandatory) is to use only a single first-level heading on any page and to require all subsequent headings to descend from it. p
- The paragraph element is the fundamental building block of prose text. It is also the most appropriate element for marking up a stanza of poetry or other similar discrete block of text. Different from a
div
principally in that it is specifically intended to indicate text regions, whereas thediv
element is more broadly specified. blockquote
- This is a very specific use element which should be used to indicate a significant block of text which is being quoted from outside the current source. It should always be paired with a
cite
element to indicate the quoted source. It may also, optionally, use thecite
attribute to contain a URI (Uniform Resource Identifier) for the quoted text.
Supporting Inline Semantic Elements
a
- When accompanied by an
href
attribute, the anchor element indicates either an external resource (a resource other than the current document) accessible via hyperlink or an anchored location within the same document. Using scripting, it can be used to perform more complex functions within the current page, but should always maintain a fall-back functionality to retain it’s semantic value. abbr
- The abbreviation element generically indicates a shortened form of a more extensive term or phrase. It is inclusive of an acronym, although the lack of support for
abbr
in Internet Explorer frequently forces developers to ignore that relationship. acronym
- “Acronym” refers to a subset of abbreviations characterized by their formation from parts (letters or syllables) of the words they are used to abbreviate. The definition isn’t strictly agreed on, but it’s generally agreed that abbreviations formed by the removal of letters from a word are not acronyms.
em
- Indicates emphasis. “Emphasis” is a general indication that the emphasized text is in some way more significant than the text surrounding it. Whether a piece of text should be emphasized or not is usually dictated by authorial preference.
strong
- “Strong” is described officially as “Stronger Emphasis.” So, practically speaking, it’s an element you use in much the same scenario as you would use
em
: an authorially determined preference for emphasis. address
- According to the W3C,
address
indicates contact information relevant to a specific document or part of a document. In practical usage, it’s more commonly used to indicate any block of contact information. As a block-level element, it’s generally reserved for significant blocks of information, rather than being used to mark-up a single e-mail address or telephone number. cite
- A citation is fairly broad, and does not necessarily have to be associated with specific quoted information (although the reverse is not equally true.)
cite
is associated with bibliographical information, personal quotations, or references to an external resource used in the research towards preparing a document. code
- Indicates a sample of programming code as a general rule. The W3C specifications are clear that this is intended to refer to computer code; and I haven’t yet come across a situation where I needed to post encryption information which was not computer code. 😉
dfn
- This is one of the more difficult to define elements — which is ironic, given that it’s intended to represent the “defining instance” of a term. It is not intended to contain a definition, it is merely intended to enclose a term at the point in a document where it is used in a definitive state. Sounds very legalese, to me.
del
- Represents information which has been deleted from a document. This should generally be used with date and time information indicating when the change was made, which can be included in the
datetime
attribute in the following format:datetime="YYYY-MM-DDTHH:MM:SS"
. See alsoins
samp
- Sample output from programs or scripts. Differentiated from
code
in that the output of a program may not itself be code, but should still be indicated as an example of output. span
- A generic inline-level HTML element. It should not be concluded that
span
does not contain any semantic value, rather, that it is available to be used when no other element provides suitable meaning. It is preferable to use a generic element and define a meaning for it rather than use an element which has a pre-defined and inappropriate semantic meaning. ins
- The opposite of
del
, above. Represents inserted text following revisions. q
- Indicates a shorter, inline quotation. Unfortunately, support for the
q
element is minimal, and it cannot be readily recommended for any use. kbd
- Indicates text to be entered by the user. Rarely used, but useful in circumstances where you are demonstrating the use of a program, along with
code
andsamp
. sub
/sup
- Superscripting and subscripting of text can be used to indicate footnote references, valence numbers in chemical formulas (such as Fe+3), etc.
var
- Along with
code
,samp
,kbd
, the “variable” element indicates a variable (or program argument.) It should be reasonably obvious at this point that this language was designed by programmers and not by librarians.
List Elements
ul, ol, li
- This is pretty straightforward: lists are used to represent grouped information best represented as a list.
ul
is unordered, and is generally visually represented as a bulleted list.ol
is ordered, and is generally visually represented as a numbered list. It’s common to attempt to apply lists at a significant macro level in organizing the elements in a form or, occasionally, within an entire page, but it’s my opinion that this kind of usage is taking the semantic construct a bit too far. dl, dd, dt
- A definition list literally indicates a list of terms (
dt
) with their accompanying definitions (dd
). Practically speaking, it’s reasonable to use the definition list format for any collection of data characterized by paired relationships with one signifying and at least one descriptive. It’s perfectly reasonable to provide multiple definitions to a single term. Frequently asked questions pages are commonly assembled this way.
Table Elements
table
- Oft abused, the table is the best way of organizing and displaying a data matrix. Any kind of two-dimensionally represented data should be organized within a table.
thead
- Defines a header region for a data table, which would normally contain the headers (
th
) for each column. tfoot
- Defines a footer region for a data table, which should include information referential to the columns of data.
tbody
- The content bearing region of a table, but also includes row headers.
caption
- Briefly describes the table. This is essentially a heading for the table.
th
- A heading for either a row or a column, to indicate the type of information within that row or column.
td
- A data cell, in which content is placed which corresponds to both the headers for the row and column.
- Attribute:
scope
- Scope: applied to
th
, it indicates whether the heading information applies to a row or a column. It can also be applied to a row group, for tables which have been divided into multiple sections. - Attribute:
headers
- A much, much, more complicated way of indicating relationships between data cells and their respective headers. Necessary in complex tables where a given data cell may apply to multiple row or column headers. If possible, just avoid creating tables which are that complex…they’re a headache.
- Attribute:
summary
- Applied to the
table
element, the summary is a more extensive description of the table, intended to provide non-visual users with the equivalent of a “quick scan” of the table to best understand the purpose it serves.
Separator and “Other” Elements
br
- Generates a line break. The semantics of a line break are a commonly debated point – you can read my views in my article “Is a
br
tag semantic?“ hr
- Separates two sections with a visible horizontal line. Although this element conveys no specific semantic meaning which is not conveyed by other elements, it provides the advantage of a visual separator between sections when styles are disabled which is otherwise unavailable. I’m not aware of any advantages for other scenarios.
Discouraged (Presentational) Elements
These elements have not been deprecated; but should generally only be used after careful consideration.
big
small
b
i
tt
pre
Is it semantic, or is it presentational? This can be a more difficult question than it initially appears. Take b
. Presentationally, it renders text as bold. Semantically, it provides no specific emphasis or other specific meaning. Does this mean that it should never be used? Not clearly. Although it’s difficult to describe scenarios in which these elements are useful, if you assume a scenario in which you want bold text but do not want that text to receive additional emphasis, it makes more sense to use b
than it does to use span
and style it to be bold.
Regardless, these are not elements that should generally be used without careful consideration that they are, in fact, the best choice for the job. But it’s your call.
Deprecated Elements
applet
center
font
dir
isindex
menu
s
strike
u
Not all deprecated elements are created equal. I find it ironic that strike
and u
are set right alongside font
and isindex
. Thinking logically, strike
and u
are very much in the same vein as b
and i
. Presentational, but perhaps appropriate in some contexts.
thanks for the article. I’m not new at HTML (HyperText Markup Language), but am at manual coding so this is very helpful!
Finally, a site that gives an understandable definition of semantic in relation to HTML (HyperText Markup Language). Thank you for clearing up the question in my mind.
Oh and I also like the semantic graph tool. Would be nice if it could output in a JPG or something like that.
Nice quick overview of the elements. I am aiming at providing a little more detailed insight into best practices and will be covering each tag and what I have seen people do and not do with it.
Semantic User Interfaces – The Best Practices
That pragmatism, sometimes simple is not easier. Thanks, Joe.
In normal language, that means that the page contains no content which is a child of only that
div
. On this page it’s thediv
with the id “outer.”This div is there purely for visual purposes, and as such is non-semantic. However, within the limitations of HTML (HyperText Markup Language) and CSS (Cascading Style Sheets), sometimes that’s a valid choice. The challenge is over how many extra nested elements continue to be reasonable and valid. One is really just fine, in my view — a simple page structure is pretty effectively retained using the odd extra element for formatting.
The resolution is simply a matter of accepting a pragmatic perspective on semantic use of elements – sometimes there are design choices which are not available to you without introducing some degree of extra markup. The balance between pragmatism versus semantics is personal choice.
Joe,
based on W3C (World Wide Web Consortium) semantic data extractor, this page/post has:
which
div
is it? do you need to fix it? how to resolve it?