diff --git a/src/compiler/GF/Text/Lexing.hs b/src/compiler/GF/Text/Lexing.hs
index 4e07c2126..bfe38ca83 100644
--- a/src/compiler/GF/Text/Lexing.hs
+++ b/src/compiler/GF/Text/Lexing.hs
@@ -14,7 +14,7 @@ stringOp name = case name of
   "lexcode"    -> Just $ appLexer lexCode
   "lexmixed"   -> Just $ appLexer lexMixed
   "words"      -> Just $ appLexer words
-  "bind"       -> Just $ appUnlexer bindTok
+  "bind"       -> Just $ appUnlexer (unwords . bindTok)
   "unchars"    -> Just $ appUnlexer concat
   "unlextext"  -> Just $ appUnlexer (unlexText . unquote)
   "unlexcode"  -> Just $ appUnlexer unlexCode
diff --git a/src/runtime/haskell/PGF/Lexing.hs b/src/runtime/haskell/PGF/Lexing.hs
index a49328826..782757b03 100644
--- a/src/runtime/haskell/PGF/Lexing.hs
+++ b/src/runtime/haskell/PGF/Lexing.hs
@@ -30,12 +30,12 @@ unlexText = capitInit . unlext where
     _ -> []
 
 -- | Bind tokens separated by Prelude.BIND, i.e. &+
-bindTok :: [String] -> String
+bindTok :: [String] -> [String]
 bindTok ws = case ws of
-    w:"&+":ws2 -> w ++ bindTok ws2
-    w:[]       -> w
-    w:ws2      -> w ++ " " ++ bindTok ws2
-    []         -> ""
+               w1:"&+":w2:ws -> bindTok ((w1++w2):ws)
+               "&+":ws       -> bindTok ws
+               w:ws          -> w:bindTok ws
+               []            -> []
 
 -- * Code lexing