The syntax table contains information that tells Emacs how to operate on text, words, sentences etc. It will make Emacs know enough about all the symbols in a buffer. Syntax tables are used, for example, for moving forward or backward by word. The Syntax Tables may be different for different buffers, but usually it is controled by the current MajorMode of the buffer. Syntax Table may inherit standard syntax table. The syntax tables of most major modes inherit the standard syntax table (see below) for characters from 0 to 31 and from 128 to 255.

Working with Syntax tables

You can see the current syntax table by hitting C-h s or F1 s. This table lists for each character or character range the corresponding syntax class.

Each character belongs to a certain syntax class. This syntax class is identified by one of the following symbol.

  -   whitespace character        /   character quote character
  w   word constituent            $   paired delimiter         
  _   symbol constituent          '   expression prefix        
  .   punctuation character       <   comment starter          
  (   open delimiter character    >   comment ender            
  )   close delimiter character   !   generic comment delimiter
  "   string quote character      |   generic string delimiter 
  \   escape character            @   inherit from `standard-syntax-table'

For example, the backslash \ is the only character with escape syntax in the text-mode-syntax-table. On the other hand, in that table, the word constituent class contains thousand characters. There are of course letters a..z, A..Z, numbers 0..9, but also $, %, ‘, currency units, accented letters, kanjis, chinese ideograms, indian glyphs, mathematical symbols and more.

You can see the syntax of the + character in the current syntax table by typing

   (char-to-string (char-syntax ?+))

and then evaluating that line with C-x C-e.

The Unicode character with code 222B (the mathematical integral sign) can be entered by typing ?\u222B.

Syntax classes can be matched in regular expressions using \sc, where c is one of the syntax symbol listed above. For example, \s( matches all opening delimiters. On the other hand, \Sw matches all characters which are not in the word constituent class. This is the same as \W.

Basic Syntax Tables

How to define a new syntax table

Syntax tables are char-tables. Hence they can be defined using the make-char-table function. By default, elements of a char-table are initialized to nil. Some specific default constituent class can be specified with the help of the string-to-syntax function.

   (setq my-table (make-char-table 'syntax-table nil))
   (setq my-table (make-char-table 'syntax-table (string-to-syntax "w")))

In the last example, all characters are put in the word constituent class.

Note however that syntax-tables are usually defined by starting with some basic syntax table and altering it. Here is how to make a syntax table inheriting all settings from the standard syntax table, or from the text mode syntax table.

   (setq my-table (make-syntax-table))
   (setq my-table (make-syntax-table text-mode-syntax-table))

Next section explains how to modify a syntax table.

Hacking the syntax table

The syntax of a character can be modified with the help of the modify-syntax-entry command.

Put the quote character in the expression prefix class (the default in Lisp mode)

   (modify-syntax-entry ?' "'")

Put the brackets [] in the word class, so as to remove their interpretation as delimiters

   (modify-syntax-entry ?[ "w")
   (modify-syntax-entry ?] "w")

Make ^ and $ delimiters matching each other

   (modify-syntax-entry ?^ "($")
   (modify-syntax-entry ?$ ")^")

Modify the text-mode syntax table, so as to consider $ as a delimiter matching itself.

   (modify-syntax-entry ?$ "$" text-mode-syntax-table)

Here is a short snippet that defines a new syntax table, by modifying the current syntax table.

 (defvar my-wacky-syntax-table
      (let ((table (make-syntax-table)))
        (modify-syntax-entry ?[ "w" table)
        (modify-syntax-entry ?] "w" table)

Put this in the InitFile, and call it from a buffer with

      M-: (set-syntax-table my-wacky-syntax-table)

This will change the syntax of the brackets [], leaving the syntax of the other characters untouched.

You can specify my-wacky-syntax-table as the default table in text-mode by putting in your InitFile

  (add-hook 'text-mode-hook (lambda ()
              (set-syntax-table my-wacky-syntax-table)))

You can also set which syntax table is used on a per command basis. Here is how to use my-wacky-syntax-table with the C-M-f shortcut, bound to forward-sexp by default.

 (global-set-key "\C-\M-f" '(lambda ()
           (with-syntax-table my-wacky-syntax-table (forward-sexp))))

The syntax table, what for ?

Words in emacs are just a succession of characters which belong to the word syntax class. So, commands like forward-word (M-f) or backward-kill-word (M-DEL) use the syntax table to find what string they should take into consideration.

The command forward-list (C-M-n) moves across one balanced group of parentheses. What is interpreted as a parenthesis is given by the syntax table. For example, () [] are delimiters in the lisp syntax table, whereas [] are just word constituents. In the text syntax table, the characters {} [] () are all valid delimiters.

In regular expressions, the \w symbol matches any character in the word syntax class. So it is possible to hack the syntax table so as to modify what \w actually matches. This also means that a regular expression can match different strings, depending on the buffer current mode. The string I'm counts as a single word in text mode, whereas it is considered as two words in lisp mode: I and 'm. Indeed, the quote ‘ is recorded as an expression prefix in the lisp syntax table, and as a word constituent in the text syntax table.