diff --git a/MODES b/MODES
index 595aec0..a8a55d8 100644
--- a/MODES
+++ b/MODES
@@ -36,3 +36,73 @@ to be taken. They do this via self.bindings, a dictionary mapping action names
 Modes subclass mode2.Fundamental, and they call mode2.Fundamental.__init__ to
 run the standard mode initialization (including building the default bindings
 dictionary); they can later modify or overwrite this dictionary if they choose.
+
+There are at least 3 optional behaviors modes can make use of:
+
+    1. Syntax highlighting
+    2. Indentation level detection
+    3. Tag (parenthesis, brace, bracket, etc.) matching
+
+Not all modes can (or should) make use of these features: they are primarily
+useful in modes having to do with programming languages, or other structured
+documents.
+
+4. Syntax highlighting
+
+Syntax highlighting uses a hybrid lexing/parsing process to break each line of
+the buffer down into one or more lexical tokens; these tokens are primarily
+used to highlight parts of the buffer different colors. The colors to use are
+defined by self.colors, a dictionary mapping token-names to a tuple consisting
+of at least a foreground color and a background color.
+
+Explaining how to write a Grammar is outside the scope of this document; see
+lex2.py and mode/*.py for examples. Some important points to note:
+
+    * Regexes are applied to only one line of the document at a time.
+    * All regexes must match at least one character.
+    * All tokens must consist of at least one character.
+    * A rule that matches must generate one or more tokens.
+    * Any input not matched by a rule will end up in a "null" token.
+    * Tokens can't "look" for other tokens (but they can use 0-width assertions
+      to test for data on the current line).
+
+5. Indentation level detection
+
+Indentation level detection hooks into the basic 'insert-tab' action; rather
+than inserting 4 spaces (which is the default), it will instead determine the
+correct "tab depth" for this line of the buffer, and insert/remove spaces from
+the beginning of the line in order to reach the correct number.
+
+To implement this, you must create a Tabber class and assign it to self.tabber
+in the mode. At a minimum, a tabber must support the following methods:
+
+    * __init__(self, mode)
+    * get_level(self, y)
+    * region_added(self, p, lines)
+    * region_removed(self, p1, p2)
+
+Tabber classes can often be tricky to implement correctly. tab2.Tabber provides
+a lot of base functionality that is probably useful; also, you may want to try
+looking at tab2.StackTabber, which provides even more base support.
+
+6. Tag matching
+
+Tag matching allows a closing tag (such as ")") to "show" the corresponding
+opening tag (such as an earlier "(") to help orient the user. Currently, tags
+are assumed to be single characters, although in the future it's easy to imagine
+multi-character tags being useful. Currently they are not supported.
+
+This support is very easy to add, assuming that the mode has a grammar. In most
+cases, here is how it works:
+
+    a. In the mode, create 4 tuples:
+        * opentokens:  tuple of lexical tokens which can be opentags
+        * opentags:    dictionary mapping opentag strings to closetag strings
+        * closetokens: tuple of lexical tokens which can be closetags
+        * closetags:   dictionary mapping closetag strings to opentag strings
+
+    b. Also in the visecmode, create or instantiate actions who subclass
+       method.CloseTag (e.g. method.CloseParen) and assign them the appropriate
+       binding (e.g. '(').
+
+    c. Enjoy!