diff --git a/osh/word_parse.py b/osh/word_parse.py
index d965ecd388..3b469704ab 100644
--- a/osh/word_parse.py
+++ b/osh/word_parse.py
@@ -667,7 +667,7 @@ def ReadSingleQuoted(self, lex_mode, left_token, tokens, is_ysh_expr):
 
                 # x = $'\z' is disallowed; ditto for echo $'\z' if shopt -u parse_backslash
                 if is_ysh_expr or not self.parse_opts.parse_backslash():
-                    p_die("Invalid char escape in C-style string literal", tok)
+                    p_die("Invalid char escape in C-style $'-string literal (parse_backslash, no verbatims)\n\t* use \\\\ to denote a \\ ?\n\t* 'chain'\"different\"$'literals' ?\n\t* (ysh) use u'- or b'-literal instead ?\n\t* (ysh) concat r'- ++ u'- or b'-literals ?", tok)
 
                 tokens.append(tok)
 
@@ -944,7 +944,7 @@ def _ReadLikeDQ(self, left_token, is_ysh_expr, out_parts):
                         if (is_ysh_expr or
                                 not self.parse_opts.parse_backslash()):
                             p_die(
-                                "Invalid char escape in double quoted string",
+                                "Invalid char escape in \"-string literal (parse_backslash, no verbatims)\n\tAllowed: \\\", \\$, or \\\\ to denote \", $, or \\\n\t* use \\\\ to denote a \\ ?\n\t* 'chain'\"different\"$'literals' ?\n\t* (ysh) concat \"- ++ u'- or b'-literals ?",
                                 self.cur_token)
                     elif self.token_type == Id.Lit_Dollar:
                         if is_ysh_expr or not self.parse_opts.parse_dollar():
@@ -1611,7 +1611,7 @@ def _MaybeReadWordPart(self, is_first, lex_mode, parts):
             ch = lexer.TokenSliceLeft(tok, 1)
             if not self.parse_opts.parse_backslash():
                 if not pyutil.IsValidCharEscape(ch):
-                    p_die('Invalid char escape (parse_backslash)',
+                    p_die("Invalid char escape (parse_backslash, no verbatims, no quotings)\n\t* remove \ ?\n\t* use \\\\ to denote a \\ ?\n\t* use \"raw\" '-string-literal ?\n\t* use \"interpreted\" $'-string-literal ?\n\t* (ysh) use \"raw\" r'-string-literal ?\n\t* (ysh) use \"interpreted\" u'- or b'-string-literal ?",
                           self.cur_token)
 
             part = word_part.EscapedLiteral(self.cur_token,