What is the deal about these semantic vs syntactic commands?

Question

I keep hearing people on this site talking about "semantic" vs "syntactic" commands and that the "semantic" ones are somehow better. I don't really get what they are talking about, could you explain and give some examples that show the difference.

Should I care about this at all?

Answer 1

Suppose, purely for the sake of argument, that one were typesetting a book which had frequent occasion to reference publications by name --- Science, the New York Times and so forth. In a WYSYWYG editor, one would highlight the name of the journal and click the I button. With syntactic markup, like plain ol' HTML, one would wrap the journal name in codes indicating that it is to be italicized: <i>Science</i>. With semantic markup, by contrast, one could define a command which says, "This text is the title of a publication, and is to be handled in such-and-such a manner". Then, if one wished to change the style of the document, one could tweak the code defining that command, and the change would be reflected throughout, without having to modify each instance manually (and run the risk of missing something).

This has the added advantage that one can build extra features into the command. Instead of expanding out as "put this text in italics", the "publication name" command could do something more elaborate: "put this text in italics, and add the appropriate entry to the book's index". That way, a reader who wanted to find all the places where the book references a New Yorker article or an item in the IEEE Transactions on Love Poetry could just flip to the back and look up the pages.

The LaTeX implementation of that idea would look something like this, using the makeidx package to implement the indexing:

\newcommand{\journal}[1]{\emph{#1}\index{#1@\emph{#1}}}

Having defined the \journal command in this way, we could type \journal{Science}, and LaTeX would handle both the visual formatting and the auxiliary function of indexing for us.

Edit to add: Yes, this can easily become overkill. But it's fun!

Answer 2

Semantic is defined as "Of or relating to meaning". Syntactic is defined as "described by grammatical structure." When expressions are simple, it's fairly easy to read either structure. When they're complex, it's much easier to read things which are defined by the meaning of their expressions rather than by the order of their symbols.

\set{a, b, c} is semantic, it tells you there is a set of three things: a, b, and c. Instead, \{a, b, c\} tells you there is an opening left brace, followed by some symbol a, followed by some coma, followed by ..., but who knows what those symbols mean? Both expand to the same thing, assuming a properly defined 'set' macro, but the former is easier to read, understand, and debug in source code.

Answer 3

Consider hello!, bon jour! sup!. Natural and computer language taxonomy makes (amongst others) these distinctions:

• lexical : each occupies a different place in any lexical ordering (e.g., a dictionary)
• syntactic : most grammars (or 'production rules') parsing these terms would differ in the sense that two of the three would swallow two tokens (don't forget the !), while the other would consume three. To clarify, syntax (or grammar) is the set of rules governing how and where terms can legitimately stand in relationship to one another in a valid program
• semantic : the three utterances broadly intend or mean the same thing
• pragmatic : while the producers of these lexically, syntactically and semantically valid utterances all intend to convey a greeting, the context, place or pattern of their usage will determine their "efficiency" (not just in the program, but also within the community of producers, consumers and cultural guardians of the patterns).

Of course, this taxonomy applies as equally to TeX/LaTeX as any other language that might or might not exist (swap the word phonemic for lexical in purely spoken lingua if you like).