diff --git a/src/orca/scripts/apps/Thunderbird/script.py b/src/orca/scripts/apps/Thunderbird/script.py
index 9da4a188e54d2e9f61244cbfe7beb09c7b806fd1..ef4494cf8502725eb97d19828b3dab2cc05fbb27 100644
--- a/src/orca/scripts/apps/Thunderbird/script.py
+++ b/src/orca/scripts/apps/Thunderbird/script.py
@@ -385,19 +385,6 @@ class Script(Gecko.Script):
             debug.println(debug.LEVEL_INFO, msg, True)
             self.sayAll(None)
 
-    def sayWord(self, obj):
-        """Speaks the word at the current caret position."""
-
-        contextObj, offset = self.utilities.getCaretContext(documentFrame=None)
-        if contextObj != obj:
-            super().sayWord(obj)
-            return
-
-        wordContents = self.utilities.getWordContentsAtOffset(obj, offset)
-        textObj, startOffset, endOffset, word = wordContents[0]
-        self.speakMisspelledIndicator(textObj, startOffset)
-        self.speakContents(wordContents)
-
     def toggleFlatReviewMode(self, inputEvent=None):
         """Toggles between flat review mode and focus tracking mode."""
 
diff --git a/src/orca/scripts/web/script.py b/src/orca/scripts/web/script.py
index ba056851f1d0c22153081fc88bb9b8e9621912b6..e84682347d506e7bff3df0cc5917bd5ced55dc6d 100644
--- a/src/orca/scripts/web/script.py
+++ b/src/orca/scripts/web/script.py
@@ -887,8 +887,8 @@ class Script(default.Script):
     def sayWord(self, obj):
         """Speaks the word at the current caret position."""
 
-        if not self._lastCommandWasCaretNav \
-           and not self.utilities.isContentEditableWithEmbeddedObjects(obj):
+        isEditable = self.utilities.isContentEditableWithEmbeddedObjects(obj)
+        if not self._lastCommandWasCaretNav and not isEditable:
             super().sayWord(obj)
             return
 
@@ -897,7 +897,7 @@ class Script(default.Script):
         if keyString == "Right":
             offset -= 1
 
-        wordContents = self.utilities.getWordContentsAtOffset(obj, offset)
+        wordContents = self.utilities.getWordContentsAtOffset(obj, offset, useCache=not isEditable)
         textObj, startOffset, endOffset, word = wordContents[0]
         self.speakMisspelledIndicator(textObj, startOffset)
         self.speakContents(wordContents)