diff --git a/README.markdown b/README.markdown
index 4a8f26d..e4ea511 100644
--- a/README.markdown
+++ b/README.markdown
@@ -11,13 +11,13 @@ Features of this plugin:
 * Focused on the most common of typos
 * Pure Vimscript using the efficient `iabbrev`
 * Included auto-correct entries limited to 350 (in 2016) to ensure fast
-  loading (see **new** expansion policy below)
+  loading (see expansion policy below)
 * Buffer-scoped behavior (won’t touch global settings)
 
 Note that this plugin is not intended to be a replacement for teh spell
 checker in Vim. It’s best used with spell-check enabled.
 
-## Expansion policy (**NEW**)
+## Expansion policy
 
 Since it was created in 2013, _litecorrect_ has limited the number of
 default auto-correct entries (such as `teh`->`the`) to 300, avoiding the
@@ -62,9 +62,9 @@ in `markdown` and `textile` files, place in your `.vimrc`:
   augroup END
   ```
 
-Alternatively, you can build on the defaults by providing your own
-corrections. Note that the corrections are stored as key-value entries
-where the value is a list of the common misspellings for the key.
+Optionally, you can build on the defaults by providing your own corrections.
+Note that the corrections are stored as key-value entries where the value is
+a list of the common misspellings for the key.
 
   ```
   let user_dict = {
@@ -122,17 +122,16 @@ that's -> that’s
 
 ## Criteria to add (or modify) default entries
 
-Note that the number of default entries will be limited to 300 for fast
-loading.
+Note that the number of default entries will be limited for fast loading. See
+policy above.
 
-Suggestions for improving the defaults are welcome, but good evidence is
-needed that a suggested auto-correct entry is more common than the one it
-will replace.
+Suggestions for improving the defaults are welcome, but good evidence is needed
+that a suggested auto-correct entry adds value to the list.
 
 ## Related projects
 
-If load time performance isn’t an issue, you may seek a more
-comprehensive approach:
+If load time performance isn’t an issue, you can seek a more comprehensive
+approach:
 
 * [wordlist.vim](https://github.com/vim-scripts/wordlist.vim) - nearly 800 entries
 * [vim-autocorrect](https://github.com/panozzaj/vim-autocorrect) - over 12K entries!
diff --git a/autoload/litecorrect.vim b/autoload/litecorrect.vim
index f8223e4..703cd6f 100644
--- a/autoload/litecorrect.vim
+++ b/autoload/litecorrect.vim
@@ -25,7 +25,9 @@ function! litecorrect#init(...)
   ia <buffer> THat That
   ia <buffer> THe The
   ia <buffer> Teh The
+  ia <buffer> Youre You're
   ia <buffer> abotu about
+  ia <buffer> aboutit about it
   ia <buffer> acn can
   ia <buffer> adn and
   ia <buffer> aer are
@@ -186,6 +188,7 @@ function! litecorrect#init(...)
   ia <buffer> mysefl myself
   ia <buffer> myu my
   ia <buffer> nad and
+  ia <buffer> niether neither
   ia <buffer> nkow know
   ia <buffer> nkwo know
   ia <buffer> nmae name
@@ -242,6 +245,7 @@ function! litecorrect#init(...)
   ia <buffer> tehy they
   ia <buffer> tghe the
   ia <buffer> tghis this
+  ia <buffer> thanit than it
   ia <buffer> thansk thanks
   ia <buffer> thast that
   ia <buffer> thats that's
@@ -283,6 +287,7 @@ function! litecorrect#init(...)
   ia <buffer> tyhe the
   ia <buffer> tyhe they
   ia <buffer> uise use
+  ia <buffer> untill until
   ia <buffer> veyr very
   ia <buffer> vrey very
   ia <buffer> waht what
@@ -305,6 +310,7 @@ function! litecorrect#init(...)
   ia <buffer> witha with a
   ia <buffer> withe with
   ia <buffer> withh with
+  ia <buffer> withit with it
   ia <buffer> witht with
   ia <buffer> withthe with the
   ia <buffer> witn with