diff --git a/CHANGELOG.md b/CHANGELOG.md
index 6500f7a..7250461 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -11,6 +11,12 @@ instead of version numbers.
 
 * Only cluster by leading tags in the sort order, so if you click to sort by
   e.g. Updated then there's no longer awkward clustering by repeated tags.
+* Search no longer supports quoting phrases with single quotes, so that it's
+  easier to search for possessives (e.g. `Erik's`).
+  [[#638](https://github.com/edemaine/coauthor/issues/638)]
+* Fix escaping search queries with backslash: `\:`, `\"`, `\|`, `\(`, `\)`, `\\`
+* Newly allow escaping of `*` with `\*`
+  [[#29](https://github.com/edemaine/coauthor/issues/29)]
 
 ## 2023-11-17
 
diff --git a/README.md b/README.md
index a3504d0..9e63d47 100644
--- a/README.md
+++ b/README.md
@@ -200,11 +200,13 @@ To see what's changed in Coauthor recently, check out the
   * Negative match with minus sign
     (e.g., `-word` excludes documents with whole `word`).
   * Search for a regular expression via `regex:...`.
-  * Use quotes (`'...'` or `"..."`) to search for phrases or `regex:"..."`
+  * Use double quotes (`"..."`) to search for phrases or `regex:"..."`
     to search for regular expressions with spaces in them; normally,
     spaces act as an AND query.
   * Connect words/phrases with `|` to get an OR query instead.
   * Use parentheses to mix AND and OR arbitrarily, e.g. `always (this | that)`.
+  * Escape special characters with backslash:
+    `\:`, `\"`, `\|`, `\(`, `\)`, `\\`, `\*`.
   * `by:username` searches for messages coauthored by a specified username
     (which can include `*`s or use regular expressions via `regex:`);
     `by:me` is shorthand for searching for your own username.
diff --git a/lib/search.coffee b/lib/search.coffee
index 8a6a9b3..faf43ef 100644
--- a/lib/search.coffee
+++ b/lib/search.coffee
@@ -35,7 +35,11 @@ unescapeRegExp = (regex) ->
 export parseSearch = (search, group) ->
   ## Quoted strings turn off separation by spaces.
   ## Last quoted strings doesn't have to be terminated.
-  tokenRe = /(\s+)|((?:"[^"]*"|'[^']*'|[^'"\s\|\(\)])+)('[^']*$|"[^"]*$)?|'([^']*)$|"([^"]*)$|([\|\(\)])/g
+  tokenRe = ///
+    (\s+) |
+    ([\|\(\)]) |
+    ( (?: " (?:[^"\\]|\\.)* (?:"|$) | [^"\s\|\(\)\\] | \\. )+ )
+  ///g
   wants = []         # array of Mongo queries that will be $and'd together
   options = [wants]  # array of wants that will be $or'd together
   stack = []         # stack of strict-ancestor options objects
@@ -63,8 +67,8 @@ export parseSearch = (search, group) ->
   while (token = tokenRe.exec search)?
     continue if token[1]  ## ignore whitespace tokens
 
-    if token[6]  # top-level grouping operators
-      switch token[6]
+    if token[2]  # top-level grouping operators
+      switch token[2]
         when '|'  # OR
           options.push wants = []
         when '('  # start group
@@ -84,19 +88,12 @@ export parseSearch = (search, group) ->
     colon = /^-?(?:(?:regex|title|body|tag|emoji|by|root|is|isnt|not):)*/.exec token[0]
     colon = colon[0]
     ## Remove quotes (which are just used for avoiding space parsing).
-    if token[4]
-      token = token[4]  ## unterminated initial '
-    else if token[5]
-      token = token[5]  ## unterminated initial "
-    else
-      token = (token[2].replace /"([^"]|\\")*"|'([^']|\\')*'/g, (match) ->
-        match[1...match.length-1]
-      ) + (token[3] ? '')[1..]
+    token = token[3].replace /(^|[^\\](?:\\\\)*)"((?:[^"\\]|\\[^])*)(?:"|$)/g, "$1$2"
     ## Remove leading colon part if we found one.
     ## (Can't have had quotes or escapes.)
     token = token[colon.length..]
     ## Remove escapes.
-    token = token.replace /\\([:'"\\])/g, '$1'
+    token = token.replace /\\([:"\|\(\)\\])/g, '$1'
     ## Construct regex for token
     regexMode = 0 <= colon.indexOf 'regex:'
     colon = colon.replace /regex:/g, '' if regexMode
@@ -109,14 +106,16 @@ export parseSearch = (search, group) ->
         if starStart
           word = word[1..]
           return unless word
-        starEnd = word[word.length-1] == '*'
+        starEnd = /[^\\]\*$/.test word
         if starEnd
           word = word[0...word.length-1]
           return unless word
         regex = escapeRegExp word
         ## Outside regex mode, lower-case letters are case-insensitive
         regex = caseInsensitiveRegExp regex
-        regex = regex.replace /\\\*/g, '\\S*'  ## * was already escaped
+        regex = regex
+        .replace /(^|[^\\])\\\*/g, '$1\\S*' # * in input becomes singly escaped
+        .replace /\\\\\\\*/g, '\\*'  # \* in input becomes doubly escaped
         if not starStart and regex.match /^[\[\w]/  ## a or [aA]
           regex = "\\b#{regex}"
         if not starEnd and regex.match /[\w\]]$/  ## a or [aA]