diff --git a/src/main/java/io/bdrc/lucene/bo/TibPattFilter.java b/src/main/java/io/bdrc/lucene/bo/TibPattFilter.java
index 703a072..26c8960 100644
--- a/src/main/java/io/bdrc/lucene/bo/TibPattFilter.java
+++ b/src/main/java/io/bdrc/lucene/bo/TibPattFilter.java
@@ -80,6 +80,20 @@ public final static String normalizeR(final String in) {
         }
     }
     
+    public static class ReorderFilter extends PatternReplaceCharFilter {
+        public ReorderFilter(Reader in) {
+            super(rReorder, repl, in);
+        }
+        // https://github.com/buda-base/lucene-bo/issues/17
+        // reorder vowel + subscript into subscript + vowel
+        public static final Pattern rReorder = Pattern.compile("([ཱ-྇]+)([ྍ-ྼ]+)");
+        public static final String repl = "$2$1";
+        public final static String normalizeR(final String in) {
+            final Matcher matcher = rReorder.matcher(in);
+            return matcher.replaceAll(repl);
+        }
+    }
+    
     public static Reader plugFilters(Reader in) {
         in = new MergedSylFilter1(in);
         in = new MergedSylFilter2(in);
diff --git a/src/test/java/io/bdrc/lucene/bo/TibetanAnalyzerTest.java b/src/test/java/io/bdrc/lucene/bo/TibetanAnalyzerTest.java
index 1ac30e5..f81ca5b 100644
--- a/src/test/java/io/bdrc/lucene/bo/TibetanAnalyzerTest.java
+++ b/src/test/java/io/bdrc/lucene/bo/TibetanAnalyzerTest.java
@@ -223,7 +223,7 @@ public void syllableLemaTest() throws IOException {
         System.out.println("Testing TibPattFilter() for Old Tibetan");
         String input = "བཀྲ་ཤིས་བདེ་ལེགས་དགར་ཁོར་ཁྲོ་ཁྲོས་འཐུ་གདུ་གདུམ་ཐེབ་ཐེབས";
         Reader reader = new StringReader(input);
-        List<String> expected = Arrays.asList("བཀྲ", "ཤིས", "བདེ", "ལེགས", "དགར", "དགར", "ཁྲོ", "ཁྲོ", "འཐུ", "འཐུ", "གདུམ", "ཐེབས", "ཐེབས");
+        List<String> expected = Arrays.asList("བཀྲ", "ཤིས", "བདེ", "ལེགས", "དགར", "དགར", "ཁྲོ", "ཁྲོ", "འཐུ", "གདུ", "གདུམ", "ཐེབས", "ཐེབས");
         System.out.print(input + " => ");
         TokenStream res = tokenize(reader, new TibSyllableTokenizer());
         final TokenFilter resF = new TibSyllableLemmatizer(res);
@@ -241,6 +241,17 @@ public void pattFilterTest() throws IOException {
         assertTokenStream(res, expected);
     }
     
+    @Test
+    public void reorderTest() throws IOException {
+        System.out.println("Testing TibSyllableLemmatizer()");
+        String input = "ཀྲི ཀིྲ";
+        Reader reader = new StringReader(input);
+        List<String> expected = Arrays.asList("ཀྲི", "ཀྲི");
+        System.out.print(input + " => ");
+        TokenStream res = tokenize(new TibPattFilter.ReorderFilter(reader), new TibSyllableTokenizer());
+        assertTokenStream(res, expected);
+    }
+    
     @Test
     public void ewtsOffsetBug() throws IOException {
         System.out.println("Testing TibEwtsFilter() offsets");