diff --git a/modules/unicodejs/unicodejs.wordbreak.js b/modules/unicodejs/unicodejs.wordbreak.js
index f6a9e3c2e8..910038d475 100644
--- a/modules/unicodejs/unicodejs.wordbreak.js
+++ b/modules/unicodejs/unicodejs.wordbreak.js
@@ -23,6 +23,12 @@
 	}
 
 	function getGroup( chr ) {
+		// chr is always converted to a string by RegExp#test
+		// e.g. null -> 'null' and would match /[a-z]/
+		// so return null for any non-string value
+		if ( typeof chr !== 'string' ) {
+			return null;
+		}
 		var group;
 		for ( group in patterns ) {
 			if ( patterns[group].test( chr ) ) {
diff --git a/modules/unicodejs/unicodejs.wordbreak.test.js b/modules/unicodejs/unicodejs.wordbreak.test.js
index 769e610262..ca33b2622f 100644
--- a/modules/unicodejs/unicodejs.wordbreak.test.js
+++ b/modules/unicodejs/unicodejs.wordbreak.test.js
@@ -22,7 +22,7 @@ QUnit.test( 'isBreak', function ( assert ) {
 			// 40 - 50
 			"汉字/漢字 c\u0300\u0327k" +
 			// 50 - 60
-			" c\u0300\u0327",
+			" c\u0300\u0327 a.",
 			/*jshint quotmark:single */
 		textString = new unicodeJS.TextString( text ),
 		breaks = [
@@ -31,7 +31,7 @@ QUnit.test( 'isBreak', function ( assert ) {
 			21, 25, 30,
 			31, 39, 40,
 			41, 42, 43, 44, 45, 46, 50,
-			51, 54
+			51, 54, 55, 56, 57
 		];
 
 	QUnit.expect( text.length + 1 );