diff --git a/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java b/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java
index b92283c..740b86d 100644
--- a/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java
+++ b/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java
@@ -24,6 +24,7 @@
 import com.android.inputmethod.keyboard.ProximityInfo;
 import com.android.inputmethod.latin.SuggestedWords.SuggestedWordInfo;
 import com.android.inputmethod.latin.personalization.DynamicPersonalizationDictionaryWriter;
+import com.android.inputmethod.latin.personalization.DynamicPredictionDictionaryBase;
 import com.android.inputmethod.latin.utils.CollectionUtils;
 
 import java.io.File;
@@ -72,7 +73,7 @@
     private BinaryDictionary mBinaryDictionary;
 
     /** The in-memory dictionary used to generate the binary dictionary. */
-    private AbstractDictionaryWriter mDictionaryWriter;
+    protected AbstractDictionaryWriter mDictionaryWriter;
 
     /**
      * The name of this dictionary, used as the filename for storing the binary dictionary. Multiple
@@ -84,9 +85,11 @@
     /** Whether to support dynamically updating the dictionary */
     private final boolean mIsUpdatable;
 
+    // TODO: remove, once dynamic operations is serialized
     /** Controls access to the shared binary dictionary file across multiple instances. */
     private final DictionaryController mSharedDictionaryController;
 
+    // TODO: remove, once dynamic operations is serialized
     /** Controls access to the local binary dictionary for this instance. */
     private final DictionaryController mLocalDictionaryController = new DictionaryController();
 
@@ -446,9 +449,9 @@
     private final void syncReloadDictionaryInternal() {
         // Ensure that only one thread attempts to read or write to the shared binary dictionary
         // file at the same time.
-        mLocalDictionaryController.writeLock().lock();
+        mSharedDictionaryController.writeLock().lock();
         try {
-            mSharedDictionaryController.writeLock().lock();
+            mLocalDictionaryController.writeLock().lock();
             try {
                 final long time = SystemClock.uptimeMillis();
                 final boolean dictionaryFileExists = dictionaryFileExists();
@@ -483,10 +486,10 @@
                 }
                 mLocalDictionaryController.mLastUpdateTime = time;
             } finally {
-                mSharedDictionaryController.writeLock().unlock();
+                mLocalDictionaryController.writeLock().unlock();
             }
         } finally {
-            mLocalDictionaryController.writeLock().unlock();
+            mSharedDictionaryController.writeLock().unlock();
         }
     }
 
@@ -519,16 +522,16 @@
     private class AsyncLoadDictionaryToMemoryTask extends Thread {
         @Override
         public void run() {
-            mLocalDictionaryController.writeLock().lock();
+            mSharedDictionaryController.readLock().lock();
             try {
-                mSharedDictionaryController.readLock().lock();
+                mLocalDictionaryController.writeLock().lock();
                 try {
                     loadDictionaryAsync();
                 } finally {
-                    mSharedDictionaryController.readLock().unlock();
+                    mLocalDictionaryController.writeLock().unlock();
                 }
             } finally {
-                mLocalDictionaryController.writeLock().unlock();
+                mSharedDictionaryController.readLock().unlock();
             }
         }
     }
@@ -622,4 +625,35 @@
             mLocalDictionaryController.writeLock().unlock();
         }
     }
+
+    // TODO: Implement native binary methods once the dynamic dictionary implementation is done.
+    @UsedForTesting
+    public boolean isInDictionaryForTests(final String word) {
+        mLocalDictionaryController.writeLock().lock();
+        try {
+            if (mDictType == Dictionary.TYPE_USER_HISTORY) {
+                return ((DynamicPersonalizationDictionaryWriter) mDictionaryWriter)
+                        .isInDictionaryForTests(word);
+            }
+        } finally {
+            mLocalDictionaryController.writeLock().unlock();
+        }
+        return false;
+    }
+
+    // TODO: Remove and use addToPersonalizationPredictionDictionary instead!!!!!!!!!!!!!!!!
+    @UsedForTesting
+    public void forceAddWordForTest(
+            final String word0, final String word1, final boolean isValid) {
+        mLocalDictionaryController.writeLock().lock();
+        try {
+            mDictionaryWriter.addUnigramWord(word1, null /* the "shortcut" parameter is null */,
+                    DynamicPredictionDictionaryBase.FREQUENCY_FOR_TYPED, false /* isNotAWord */);
+            mDictionaryWriter.addBigramWords(word0, word1,
+                    DynamicPredictionDictionaryBase.FREQUENCY_FOR_TYPED, isValid,
+                    0 /* lastTouchedTime */);
+        } finally {
+            mLocalDictionaryController.writeLock().unlock();
+        }
+    }
 }
diff --git a/java/src/com/android/inputmethod/latin/ExpandableDictionary.java b/java/src/com/android/inputmethod/latin/ExpandableDictionary.java
index 342dcfc..ba7d1a2 100644
--- a/java/src/com/android/inputmethod/latin/ExpandableDictionary.java
+++ b/java/src/com/android/inputmethod/latin/ExpandableDictionary.java
@@ -265,10 +265,10 @@
         return (node == null) ? false : !node.mShortcutOnly;
     }
 
-    public boolean removeBigram(final String word1, final String word2) {
+    public boolean removeBigram(final String word0, final String word1) {
         // Refer to addOrSetBigram() about word1.toLowerCase()
-        final Node firstWord = searchWord(mRoots, word1.toLowerCase(), 0, null);
-        final Node secondWord = searchWord(mRoots, word2, 0, null);
+        final Node firstWord = searchWord(mRoots, word0.toLowerCase(), 0, null);
+        final Node secondWord = searchWord(mRoots, word1, 0, null);
         LinkedList<NextWord> bigrams = firstWord.mNGrams;
         NextWord bigramNode = null;
         if (bigrams == null || bigrams.size() == 0) {
@@ -297,10 +297,10 @@
         return (node == null) ? -1 : node.mFrequency;
     }
 
-    public NextWord getBigramWord(final String word1, final String word2) {
-        // Refer to addOrSetBigram() about word1.toLowerCase()
-        final Node firstWord = searchWord(mRoots, word1.toLowerCase(), 0, null);
-        final Node secondWord = searchWord(mRoots, word2, 0, null);
+    public NextWord getBigramWord(final String word0, final String word1) {
+        // Refer to addOrSetBigram() about word0.toLowerCase()
+        final Node firstWord = searchWord(mRoots, word0.toLowerCase(), 0, null);
+        final Node secondWord = searchWord(mRoots, word1, 0, null);
         LinkedList<NextWord> bigrams = firstWord.mNGrams;
         if (bigrams == null || bigrams.size() == 0) {
             return null;
@@ -473,37 +473,41 @@
         }
     }
 
-    public int setBigramAndGetFrequency(final String word1, final String word2,
+    public int setBigramAndGetFrequency(final String word0, final String word1,
             final int frequency) {
-        return setBigramAndGetFrequency(word1, word2, frequency, null /* unused */);
+        return setBigramAndGetFrequency(word0, word1, frequency, null /* unused */);
     }
 
-    public int setBigramAndGetFrequency(final String word1, final String word2,
+    public int setBigramAndGetFrequency(final String word0, final String word1,
             final ForgettingCurveParams fcp) {
-        return setBigramAndGetFrequency(word1, word2, 0 /* unused */, fcp);
+        return setBigramAndGetFrequency(word0, word1, 0 /* unused */, fcp);
     }
 
     /**
      * Adds bigrams to the in-memory trie structure that is being used to retrieve any word
-     * @param word1 the first word of this bigram
-     * @param word2 the second word of this bigram
+     * @param word0 the first word of this bigram
+     * @param word1 the second word of this bigram
      * @param frequency frequency for this bigram
      * @param fcp an instance of ForgettingCurveParams to use for decay policy
      * @return returns the final bigram frequency
      */
-    private int setBigramAndGetFrequency(final String word1, final String word2,
+    private int setBigramAndGetFrequency(final String word0, final String word1,
             final int frequency, final ForgettingCurveParams fcp) {
+        if (TextUtils.isEmpty(word0)) {
+            Log.e(TAG, "Invalid bigram previous word: " + word0);
+            return frequency;
+        }
         // We don't want results to be different according to case of the looked up left hand side
         // word. We do want however to return the correct case for the right hand side.
         // So we want to squash the case of the left hand side, and preserve that of the right
         // hand side word.
-        final String word1Lower = word1.toLowerCase();
-        if (TextUtils.isEmpty(word1Lower) || TextUtils.isEmpty(word2)) {
-            Log.e(TAG, "Invalid bigram pair: " + word1 + ", " + word1Lower + ", " + word2);
+        final String word0Lower = word0.toLowerCase();
+        if (TextUtils.isEmpty(word0Lower) || TextUtils.isEmpty(word1)) {
+            Log.e(TAG, "Invalid bigram pair: " + word0 + ", " + word0Lower + ", " + word1);
             return frequency;
         }
-        final Node firstWord = searchWord(mRoots, word1Lower, 0, null);
-        final Node secondWord = searchWord(mRoots, word2, 0, null);
+        final Node firstWord = searchWord(mRoots, word0Lower, 0, null);
+        final Node secondWord = searchWord(mRoots, word1, 0, null);
         LinkedList<NextWord> bigrams = firstWord.mNGrams;
         if (bigrams == null || bigrams.size() == 0) {
             firstWord.mNGrams = CollectionUtils.newLinkedList();
diff --git a/java/src/com/android/inputmethod/latin/LatinIME.java b/java/src/com/android/inputmethod/latin/LatinIME.java
index a8a29a1..3d29c5a 100644
--- a/java/src/com/android/inputmethod/latin/LatinIME.java
+++ b/java/src/com/android/inputmethod/latin/LatinIME.java
@@ -2532,11 +2532,8 @@
                 false /* isPrediction */);
     }
 
-    private void showSuggestionStrip(final SuggestedWords suggestedWords) {
-        if (suggestedWords.isEmpty()) {
-            clearSuggestionStrip();
-            return;
-        }
+    private void setAutoCorrection(final SuggestedWords suggestedWords) {
+        if (suggestedWords.isEmpty()) return;
         final String autoCorrection;
         if (suggestedWords.mWillAutoCorrect) {
             autoCorrection = suggestedWords.getWord(SuggestedWords.INDEX_OF_AUTO_CORRECTION);
@@ -2544,17 +2541,21 @@
             autoCorrection = suggestedWords.getWord(SuggestedWords.INDEX_OF_TYPED_WORD);
         }
         mWordComposer.setAutoCorrection(autoCorrection);
+    }
+
+    private void showSuggestionStrip(final SuggestedWords suggestedWords) {
+        if (suggestedWords.isEmpty()) {
+            clearSuggestionStrip();
+            return;
+        }
+        setAutoCorrection(suggestedWords);
         final boolean isAutoCorrection = suggestedWords.willAutoCorrect();
         setSuggestedWords(suggestedWords, isAutoCorrection);
         setAutoCorrectionIndicator(isAutoCorrection);
         setSuggestionStripShown(isSuggestionsStripVisible());
     }
 
-    private void commitCurrentAutoCorrection(final String separator, final Runnable callback) {
-        // Complete any pending suggestions query first
-        if (mHandler.hasPendingUpdateSuggestions()) {
-            updateSuggestionStrip();
-        }
+    private void completeCommitCurrentAutoCorrection(final String separator) {
         final String typedAutoCorrection = mWordComposer.getAutoCorrectionOrNull();
         final String typedWord = mWordComposer.getTypedWord();
         final String autoCorrection = (typedAutoCorrection != null)
@@ -2588,9 +2589,22 @@
                         typedWord, autoCorrection));
             }
         }
-        if (callback != null) {
-            callback.run();
-        }
+    }
+
+    private void commitCurrentAutoCorrection(final String separator, final Runnable callback) {
+        getSuggestedWordsOrOlderSuggestionsAsync(Suggest.SESSION_TYPING,
+                new OnGetSuggestedWordsCallback() {
+                    @Override
+                    public void onGetSuggestedWords(final SuggestedWords suggestedWords) {
+                        if (suggestedWords != null) {
+                            setAutoCorrection(suggestedWords);
+                        }
+                        completeCommitCurrentAutoCorrection(separator);
+                        if (callback != null) {
+                            callback.run();
+                        }
+                    }
+                });
     }
 
     // Called from {@link SuggestionStripView} through the {@link SuggestionStripView#Listener}
diff --git a/java/src/com/android/inputmethod/latin/makedict/BinaryDictEncoderUtils.java b/java/src/com/android/inputmethod/latin/makedict/BinaryDictEncoderUtils.java
index 1727382..5a21341 100644
--- a/java/src/com/android/inputmethod/latin/makedict/BinaryDictEncoderUtils.java
+++ b/java/src/com/android/inputmethod/latin/makedict/BinaryDictEncoderUtils.java
@@ -27,7 +27,6 @@
 import java.io.IOException;
 import java.io.OutputStream;
 import java.util.ArrayList;
-import java.util.Iterator;
 
 /**
  * Encodes binary files for a FusionDictionary.
@@ -432,7 +431,7 @@
      * @param formatOptions file format options.
      * @return the same array it was passed. The nodes have been updated for address and size.
      */
-    private static ArrayList<PtNodeArray> computeAddresses(final FusionDictionary dict,
+    /* package */ static ArrayList<PtNodeArray> computeAddresses(final FusionDictionary dict,
             final ArrayList<PtNodeArray> flatNodes, final FormatOptions formatOptions) {
         // First get the worst possible sizes and offsets
         for (final PtNodeArray n : flatNodes) calculatePtNodeArrayMaximumSize(n, formatOptions);
@@ -484,7 +483,7 @@
      *
      * @param arrays the list of node arrays to check
      */
-    private static void checkFlatPtNodeArrayList(final ArrayList<PtNodeArray> arrays) {
+    /* package */ static void checkFlatPtNodeArrayList(final ArrayList<PtNodeArray> arrays) {
         int offset = 0;
         int index = 0;
         for (final PtNodeArray ptNodeArray : arrays) {
@@ -508,7 +507,8 @@
      * @param position the position to write.
      * @return the size in bytes the address actually took.
      */
-    private static int writeChildrenPosition(final byte[] buffer, int index, final int position) {
+    /* package */ static int writeChildrenPosition(final byte[] buffer, int index,
+            final int position) {
         switch (getByteSize(position)) {
         case 1:
             buffer[index++] = (byte)position;
@@ -537,7 +537,7 @@
      * @param position the position to write.
      * @return the size in bytes the address actually took.
      */
-    private static int writeSignedChildrenPosition(final byte[] buffer, int index,
+    /* package */ static int writeSignedChildrenPosition(final byte[] buffer, int index,
             final int position) {
         if (!BinaryDictIOUtils.hasChildrenAddress(position)) {
             buffer[index] = buffer[index + 1] = buffer[index + 2] = 0;
@@ -598,7 +598,7 @@
         return flags;
     }
 
-    private static byte makePtNodeFlags(final PtNode node, final int ptNodeAddress,
+    /* package */ static byte makePtNodeFlags(final PtNode node, final int ptNodeAddress,
             final int childrenOffset, final FormatOptions formatOptions) {
         return (byte) makePtNodeFlags(node.mChars.length > 1, node.mFrequency >= 0,
                 getByteSize(childrenOffset),
@@ -616,7 +616,7 @@
      * @param word the second bigram, for debugging purposes
      * @return the flags
      */
-    private static final int makeBigramFlags(final boolean more, final int offset,
+    /* package */ static final int makeBigramFlags(final boolean more, final int offset,
             int bigramFrequency, final int unigramFrequency, final String word) {
         int bigramFlags = (more ? FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_HAS_NEXT : 0)
                 + (offset < 0 ? FormatSpec.FLAG_BIGRAM_ATTR_OFFSET_NEGATIVE : 0);
@@ -702,7 +702,7 @@
                 + (frequency & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY);
     }
 
-    private static final int writeParentAddress(final byte[] buffer, final int index,
+    /* package */ static final int writeParentAddress(final byte[] buffer, final int index,
             final int address, final FormatOptions formatOptions) {
         if (BinaryDictIOUtils.supportsDynamicUpdate(formatOptions)) {
             if (address == FormatSpec.NO_PARENT_ADDRESS) {
@@ -721,7 +721,7 @@
         }
     }
 
-    private static final int getChildrenPosition(final PtNode ptNode,
+    /* package */ static final int getChildrenPosition(final PtNode ptNode,
             final FormatOptions formatOptions) {
         int positionOfChildrenPosField = ptNode.mCachedAddressAfterUpdate
                 + getNodeHeaderSize(ptNode, formatOptions);
@@ -736,35 +736,29 @@
      * Write a PtNodeArray to memory. The PtNodeArray is expected to have its final position cached.
      *
      * @param dict the dictionary the node array is a part of (for relative offsets).
-     * @param buffer the memory buffer to write to.
+     * @param dictEncoder the dictionary encoder.
      * @param ptNodeArray the node array to write.
      * @param formatOptions file format options.
-     * @return the address of the END of the node.
      */
     @SuppressWarnings("unused")
-    private static int writePlacedNode(final FusionDictionary dict, byte[] buffer,
-            final PtNodeArray ptNodeArray, final FormatOptions formatOptions) {
+    /* package */ static void writePlacedNode(final FusionDictionary dict,
+            final DictEncoder dictEncoder, final PtNodeArray ptNodeArray,
+            final FormatOptions formatOptions) {
         // TODO: Make the code in common with BinaryDictIOUtils#writePtNode
-        int index = ptNodeArray.mCachedAddressAfterUpdate;
+        dictEncoder.setPosition(ptNodeArray.mCachedAddressAfterUpdate);
 
         final int ptNodeCount = ptNodeArray.mData.size();
-        final int countSize = getPtNodeCountSize(ptNodeArray);
-        final int parentAddress = ptNodeArray.mCachedParentAddress;
-        if (1 == countSize) {
-            buffer[index++] = (byte)ptNodeCount;
-        } else if (2 == countSize) {
-            // We need to signal 2-byte size by setting the top bit of the MSB to 1, so
-            // we | 0x80 to do this.
-            buffer[index++] = (byte)((ptNodeCount >> 8) | 0x80);
-            buffer[index++] = (byte)(ptNodeCount & 0xFF);
-        } else {
-            throw new RuntimeException("Strange size from getGroupCountSize : " + countSize);
-        }
+        dictEncoder.writePtNodeCount(ptNodeCount);
+        final int parentPosition =
+                (ptNodeArray.mCachedParentAddress == FormatSpec.NO_PARENT_ADDRESS)
+                ? FormatSpec.NO_PARENT_ADDRESS
+                : ptNodeArray.mCachedParentAddress + ptNodeArray.mCachedAddressAfterUpdate;
         for (int i = 0; i < ptNodeCount; ++i) {
             final PtNode ptNode = ptNodeArray.mData.get(i);
-            if (index != ptNode.mCachedAddressAfterUpdate) {
+            if (dictEncoder.getPosition() != ptNode.mCachedAddressAfterUpdate) {
                 throw new RuntimeException("Bug: write index is not the same as the cached address "
-                        + "of the node : " + index + " <> " + ptNode.mCachedAddressAfterUpdate);
+                        + "of the node : " + dictEncoder.getPosition() + " <> "
+                        + ptNode.mCachedAddressAfterUpdate);
             }
             // Sanity checks.
             if (DBG && ptNode.mFrequency > FormatSpec.MAX_TERMINAL_FREQUENCY) {
@@ -773,85 +767,23 @@
                         + " : " + ptNode.mFrequency);
             }
 
-            final int childrenPosition = getChildrenPosition(ptNode, formatOptions);
-            buffer[index++] = makePtNodeFlags(ptNode, index, childrenPosition,
-                    formatOptions);
-
-            if (parentAddress == FormatSpec.NO_PARENT_ADDRESS) {
-                index = writeParentAddress(buffer, index, parentAddress, formatOptions);
-            } else {
-                index = writeParentAddress(buffer, index, parentAddress
-                        + (ptNodeArray.mCachedAddressAfterUpdate
-                                - ptNode.mCachedAddressAfterUpdate),
-                        formatOptions);
-            }
-
-            index = CharEncoding.writeCharArray(ptNode.mChars, buffer, index);
-            if (ptNode.hasSeveralChars()) {
-                buffer[index++] = FormatSpec.PTNODE_CHARACTERS_TERMINATOR;
-            }
-            if (ptNode.mFrequency >= 0) {
-                buffer[index++] = (byte) ptNode.mFrequency;
-            }
-
-            if (formatOptions.mSupportsDynamicUpdate) {
-                index += writeSignedChildrenPosition(buffer, index, childrenPosition);
-            } else {
-                index += writeChildrenPosition(buffer, index, childrenPosition);
-            }
-
-            // Write shortcuts
-            if (null != ptNode.mShortcutTargets && !ptNode.mShortcutTargets.isEmpty()) {
-                final int indexOfShortcutByteSize = index;
-                index += FormatSpec.PTNODE_SHORTCUT_LIST_SIZE_SIZE;
-                final Iterator<WeightedString> shortcutIterator =
-                        ptNode.mShortcutTargets.iterator();
-                while (shortcutIterator.hasNext()) {
-                    final WeightedString target = shortcutIterator.next();
-                    int shortcutFlags = makeShortcutFlags(shortcutIterator.hasNext(),
-                            target.mFrequency);
-                    buffer[index++] = (byte)shortcutFlags;
-                    final int shortcutShift = CharEncoding.writeString(buffer, index, target.mWord);
-                    index += shortcutShift;
-                }
-                final int shortcutByteSize = index - indexOfShortcutByteSize;
-                if (shortcutByteSize > 0xFFFF) {
-                    throw new RuntimeException("Shortcut list too large");
-                }
-                buffer[indexOfShortcutByteSize] = (byte)(shortcutByteSize >> 8);
-                buffer[indexOfShortcutByteSize + 1] = (byte)(shortcutByteSize & 0xFF);
-            }
-            // Write bigrams
-            if (null != ptNode.mBigrams) {
-                final Iterator<WeightedString> bigramIterator = ptNode.mBigrams.iterator();
-                while (bigramIterator.hasNext()) {
-                    final WeightedString bigram = bigramIterator.next();
-                    final PtNode target =
-                            FusionDictionary.findWordInTree(dict.mRootNodeArray, bigram.mWord);
-                    final int addressOfBigram = target.mCachedAddressAfterUpdate;
-                    final int unigramFrequencyForThisWord = target.mFrequency;
-                    final int offset = addressOfBigram
-                            - (index + FormatSpec.PTNODE_ATTRIBUTE_FLAGS_SIZE);
-                    int bigramFlags = makeBigramFlags(bigramIterator.hasNext(), offset,
-                            bigram.mFrequency, unigramFrequencyForThisWord, bigram.mWord);
-                    buffer[index++] = (byte)bigramFlags;
-                    final int bigramShift = writeChildrenPosition(buffer, index, Math.abs(offset));
-                    index += bigramShift;
-                }
-            }
-
+            dictEncoder.writePtNodeFlags(ptNode, parentPosition, formatOptions);
+            dictEncoder.writeParentPosition(parentPosition, ptNode, formatOptions);
+            dictEncoder.writeCharacters(ptNode.mChars, ptNode.hasSeveralChars());
+            dictEncoder.writeFrequency(ptNode.mFrequency);
+            dictEncoder.writeChildrenPosition(ptNode, formatOptions);
+            dictEncoder.writeShortcuts(ptNode.mShortcutTargets);
+            dictEncoder.writeBigrams(ptNode.mBigrams, dict);
         }
         if (formatOptions.mSupportsDynamicUpdate) {
-            buffer[index] = buffer[index + 1] = buffer[index + 2]
-                    = FormatSpec.NO_FORWARD_LINK_ADDRESS;
-            index += FormatSpec.FORWARD_LINK_ADDRESS_SIZE;
+            dictEncoder.writeForwardLinkAddress(FormatSpec.NO_FORWARD_LINK_ADDRESS);
         }
-        if (index != ptNodeArray.mCachedAddressAfterUpdate + ptNodeArray.mCachedSize) {
-            throw new RuntimeException(
-                    "Not the same size : written " + (index - ptNodeArray.mCachedAddressAfterUpdate)
+        if (dictEncoder.getPosition() != ptNodeArray.mCachedAddressAfterUpdate
+                + ptNodeArray.mCachedSize) {
+            throw new RuntimeException("Not the same size : written "
+                     + (dictEncoder.getPosition() - ptNodeArray.mCachedAddressAfterUpdate)
                      + " bytes from a node that should have " + ptNodeArray.mCachedSize + " bytes");
         }
-        return index;
     }
 
     /**
@@ -862,7 +794,7 @@
      *
      * @param ptNodeArrays the list of PtNode arrays.
      */
-    private static void showStatistics(ArrayList<PtNodeArray> ptNodeArrays) {
+    /* package */ static void showStatistics(ArrayList<PtNodeArray> ptNodeArrays) {
         int firstTerminalAddress = Integer.MAX_VALUE;
         int lastTerminalAddress = Integer.MIN_VALUE;
         int size = 0;
@@ -968,49 +900,4 @@
 
         headerBuffer.close();
     }
-
-    /**
-     * Dumps a FusionDictionary to a file.
-     *
-     * @param destination the stream to write the dictionary body to.
-     * @param dict the dictionary to write.
-     * @param formatOptions file format options.
-     */
-    /* package */ static void writeDictionaryBody(final OutputStream destination,
-            final FusionDictionary dict, final FormatOptions formatOptions) throws IOException {
-
-        // Addresses are limited to 3 bytes, but since addresses can be relative to each node
-        // array, the structure itself is not limited to 16MB. However, if it is over 16MB deciding
-        // the order of the PtNode arrays becomes a quite complicated problem, because though the
-        // dictionary itself does not have a size limit, each node array must still be within 16MB
-        // of all its children and parents. As long as this is ensured, the dictionary file may
-        // grow to any size.
-
-        // Leave the choice of the optimal node order to the flattenTree function.
-        MakedictLog.i("Flattening the tree...");
-        ArrayList<PtNodeArray> flatNodes = flattenTree(dict.mRootNodeArray);
-
-        MakedictLog.i("Computing addresses...");
-        computeAddresses(dict, flatNodes, formatOptions);
-        MakedictLog.i("Checking PtNode array...");
-        if (DBG) checkFlatPtNodeArrayList(flatNodes);
-
-        // Create a buffer that matches the final dictionary size.
-        final PtNodeArray lastNodeArray = flatNodes.get(flatNodes.size() - 1);
-        final int bufferSize = lastNodeArray.mCachedAddressAfterUpdate + lastNodeArray.mCachedSize;
-        final byte[] buffer = new byte[bufferSize];
-
-        MakedictLog.i("Writing file...");
-        int dataEndOffset = 0;
-        for (PtNodeArray nodeArray : flatNodes) {
-            dataEndOffset = writePlacedNode(dict, buffer, nodeArray, formatOptions);
-        }
-
-        if (DBG) showStatistics(flatNodes);
-
-        destination.write(buffer, 0, dataEndOffset);
-
-        destination.close();
-        MakedictLog.i("Done");
-    }
 }
diff --git a/java/src/com/android/inputmethod/latin/makedict/DictEncoder.java b/java/src/com/android/inputmethod/latin/makedict/DictEncoder.java
index 89c982e..d1589a3 100644
--- a/java/src/com/android/inputmethod/latin/makedict/DictEncoder.java
+++ b/java/src/com/android/inputmethod/latin/makedict/DictEncoder.java
@@ -17,8 +17,11 @@
 package com.android.inputmethod.latin.makedict;
 
 import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions;
+import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode;
+import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString;
 
 import java.io.IOException;
+import java.util.ArrayList;
 
 /**
  * An interface of binary dictionary encoder.
@@ -26,4 +29,32 @@
 public interface DictEncoder {
     public void writeDictionary(final FusionDictionary dict, final FormatOptions formatOptions)
             throws IOException, UnsupportedFormatException;
+
+    public void setPosition(final int position);
+    public int getPosition();
+    public void writePtNodeCount(final int ptNodeCount);
+    public void writePtNodeFlags(final PtNode ptNode, final int parentAddress,
+           final FormatOptions formatOptions);
+    public void writeParentPosition(final int parentPosition, final PtNode ptNode,
+            final FormatOptions formatOptions);
+    public void writeCharacters(final int[] characters, final boolean hasSeveralChars);
+    public void writeFrequency(final int frequency);
+    public void writeChildrenPosition(final PtNode ptNode, final FormatOptions formatOptions);
+
+    /**
+     * Write a shortcut attributes list to memory.
+     *
+     * @param shortcuts the shortcut attributes list.
+     */
+    public void writeShortcuts(final ArrayList<WeightedString> shortcuts);
+
+    /**
+     * Write a bigram attributes list to memory.
+     *
+     * @param bigrams the bigram attributes list.
+     * @param dict the dictionary the node array is a part of (for relative offsets).
+     */
+    public void writeBigrams(final ArrayList<WeightedString> bigrams, final FusionDictionary dict);
+
+    public void writeForwardLinkAddress(final int forwardLinkAddress);
 }
diff --git a/java/src/com/android/inputmethod/latin/makedict/Ver3DictEncoder.java b/java/src/com/android/inputmethod/latin/makedict/Ver3DictEncoder.java
index 4e28fea..9385ba3 100644
--- a/java/src/com/android/inputmethod/latin/makedict/Ver3DictEncoder.java
+++ b/java/src/com/android/inputmethod/latin/makedict/Ver3DictEncoder.java
@@ -16,13 +16,19 @@
 
 package com.android.inputmethod.latin.makedict;
 
+import com.android.inputmethod.latin.makedict.BinaryDictDecoderUtils.CharEncoding;
 import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions;
+import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode;
+import com.android.inputmethod.latin.makedict.FusionDictionary.PtNodeArray;
+import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString;
 
 import java.io.File;
 import java.io.FileNotFoundException;
 import java.io.FileOutputStream;
 import java.io.IOException;
 import java.io.OutputStream;
+import java.util.ArrayList;
+import java.util.Iterator;
 
 /**
  * An implementation of DictEncoder for version 3 binary dictionary.
@@ -31,10 +37,13 @@
 
     private final File mDictFile;
     private OutputStream mOutStream;
+    private byte[] mBuffer;
+    private int mPosition;
 
     public Ver3DictEncoder(final File dictFile) {
         mDictFile = dictFile;
         mOutStream = null;
+        mBuffer = null;
     }
 
     // This constructor is used only by BinaryDictOffdeviceUtilsTests.
@@ -57,13 +66,166 @@
     }
 
     @Override
-    public void writeDictionary(FusionDictionary dict, FormatOptions formatOptions)
+    public void writeDictionary(final FusionDictionary dict, final FormatOptions formatOptions)
             throws IOException, UnsupportedFormatException {
         if (mOutStream == null) {
             openStream();
         }
         BinaryDictEncoderUtils.writeDictionaryHeader(mOutStream, dict, formatOptions);
-        BinaryDictEncoderUtils.writeDictionaryBody(mOutStream, dict, formatOptions);
+
+        // Addresses are limited to 3 bytes, but since addresses can be relative to each node
+        // array, the structure itself is not limited to 16MB. However, if it is over 16MB deciding
+        // the order of the PtNode arrays becomes a quite complicated problem, because though the
+        // dictionary itself does not have a size limit, each node array must still be within 16MB
+        // of all its children and parents. As long as this is ensured, the dictionary file may
+        // grow to any size.
+
+        // Leave the choice of the optimal node order to the flattenTree function.
+        MakedictLog.i("Flattening the tree...");
+        ArrayList<PtNodeArray> flatNodes = BinaryDictEncoderUtils.flattenTree(dict.mRootNodeArray);
+
+        MakedictLog.i("Computing addresses...");
+        BinaryDictEncoderUtils.computeAddresses(dict, flatNodes, formatOptions);
+        MakedictLog.i("Checking PtNode array...");
+        if (MakedictLog.DBG) BinaryDictEncoderUtils.checkFlatPtNodeArrayList(flatNodes);
+
+        // Create a buffer that matches the final dictionary size.
+        final PtNodeArray lastNodeArray = flatNodes.get(flatNodes.size() - 1);
+        final int bufferSize = lastNodeArray.mCachedAddressAfterUpdate + lastNodeArray.mCachedSize;
+        mBuffer = new byte[bufferSize];
+
+        MakedictLog.i("Writing file...");
+
+        for (PtNodeArray nodeArray : flatNodes) {
+            BinaryDictEncoderUtils.writePlacedNode(dict, this, nodeArray, formatOptions);
+        }
+        if (MakedictLog.DBG) BinaryDictEncoderUtils.showStatistics(flatNodes);
+        mOutStream.write(mBuffer, 0, mPosition);
+
+        MakedictLog.i("Done");
         close();
     }
+
+    @Override
+    public void setPosition(final int position) {
+        if (mBuffer == null || position < 0 || position >= mBuffer.length) return;
+        mPosition = position;
+    }
+
+    @Override
+    public int getPosition() {
+        return mPosition;
+    }
+
+    @Override
+    public void writePtNodeCount(final int ptNodeCount) {
+        final int countSize = BinaryDictIOUtils.getPtNodeCountSize(ptNodeCount);
+        if (1 == countSize) {
+            mBuffer[mPosition++] = (byte) ptNodeCount;
+        } else if (2 == countSize) {
+            mBuffer[mPosition++] = (byte) ((ptNodeCount >> 8) & 0xFF);
+            mBuffer[mPosition++] = (byte) (ptNodeCount & 0xFF);
+        } else {
+            throw new RuntimeException("Strange size from getGroupCountSize : " + countSize);
+        }
+    }
+
+    @Override
+    public void writePtNodeFlags(final PtNode ptNode, final int parentAddress,
+            final FormatOptions formatOptions) {
+        final int childrenPos = BinaryDictEncoderUtils.getChildrenPosition(ptNode, formatOptions);
+        mBuffer[mPosition++] = BinaryDictEncoderUtils.makePtNodeFlags(ptNode, mPosition,
+                childrenPos, formatOptions);
+    }
+
+    @Override
+    public void writeParentPosition(final int parentPosition, final PtNode ptNode,
+            final FormatOptions formatOptions) {
+        if (parentPosition == FormatSpec.NO_PARENT_ADDRESS) {
+            mPosition = BinaryDictEncoderUtils.writeParentAddress(mBuffer, mPosition,
+                    parentPosition, formatOptions);
+        } else {
+            mPosition = BinaryDictEncoderUtils.writeParentAddress(mBuffer, mPosition,
+                    parentPosition - ptNode.mCachedAddressAfterUpdate, formatOptions);
+        }
+    }
+
+    @Override
+    public void writeCharacters(final int[] codePoints, final boolean hasSeveralChars) {
+        mPosition = CharEncoding.writeCharArray(codePoints, mBuffer, mPosition);
+        if (hasSeveralChars) {
+            mBuffer[mPosition++] = FormatSpec.PTNODE_CHARACTERS_TERMINATOR;
+        }
+    }
+
+    @Override
+    public void writeFrequency(final int frequency) {
+        if (frequency >= 0) {
+            mBuffer[mPosition++] = (byte) frequency;
+        }
+    }
+
+    @Override
+    public void writeChildrenPosition(final PtNode ptNode, final FormatOptions formatOptions) {
+        final int childrenPos = BinaryDictEncoderUtils.getChildrenPosition(ptNode, formatOptions);
+        if (formatOptions.mSupportsDynamicUpdate) {
+            mPosition += BinaryDictEncoderUtils.writeSignedChildrenPosition(mBuffer, mPosition,
+                    childrenPos);
+        } else {
+            mPosition += BinaryDictEncoderUtils.writeChildrenPosition(mBuffer, mPosition,
+                    childrenPos);
+        }
+    }
+
+    @Override
+    public void writeShortcuts(final ArrayList<WeightedString> shortcuts) {
+        if (null == shortcuts || shortcuts.isEmpty()) return;
+
+        final int indexOfShortcutByteSize = mPosition;
+        mPosition += FormatSpec.PTNODE_SHORTCUT_LIST_SIZE_SIZE;
+        final Iterator<WeightedString> shortcutIterator = shortcuts.iterator();
+        while (shortcutIterator.hasNext()) {
+            final WeightedString target = shortcutIterator.next();
+            final int shortcutFlags = BinaryDictEncoderUtils.makeShortcutFlags(
+                    shortcutIterator.hasNext(),
+                    target.mFrequency);
+            mBuffer[mPosition++] = (byte)shortcutFlags;
+            final int shortcutShift = CharEncoding.writeString(mBuffer, mPosition, target.mWord);
+            mPosition += shortcutShift;
+        }
+        final int shortcutByteSize = mPosition - indexOfShortcutByteSize;
+        if (shortcutByteSize > 0xFFFF) {
+            throw new RuntimeException("Shortcut list too large");
+        }
+        mBuffer[indexOfShortcutByteSize] = (byte)((shortcutByteSize >> 8) & 0xFF);
+        mBuffer[indexOfShortcutByteSize + 1] = (byte)(shortcutByteSize & 0xFF);
+    }
+
+    @Override
+    public void writeBigrams(final ArrayList<WeightedString> bigrams, final FusionDictionary dict) {
+        if (bigrams == null) return;
+
+        final Iterator<WeightedString> bigramIterator = bigrams.iterator();
+        while (bigramIterator.hasNext()) {
+            final WeightedString bigram = bigramIterator.next();
+            final PtNode target =
+                    FusionDictionary.findWordInTree(dict.mRootNodeArray, bigram.mWord);
+            final int addressOfBigram = target.mCachedAddressAfterUpdate;
+            final int unigramFrequencyForThisWord = target.mFrequency;
+            final int offset = addressOfBigram
+                    - (mPosition + FormatSpec.PTNODE_ATTRIBUTE_FLAGS_SIZE);
+            int bigramFlags = BinaryDictEncoderUtils.makeBigramFlags(bigramIterator.hasNext(),
+                    offset, bigram.mFrequency, unigramFrequencyForThisWord, bigram.mWord);
+            mBuffer[mPosition++] = (byte) bigramFlags;
+            mPosition += BinaryDictEncoderUtils.writeChildrenPosition(mBuffer, mPosition,
+                    Math.abs(offset));
+        }
+    }
+
+    @Override
+    public void writeForwardLinkAddress(final int forwardLinkAddress) {
+        mBuffer[mPosition++] = (byte) ((forwardLinkAddress >> 16) & 0xFF);
+        mBuffer[mPosition++] = (byte) ((forwardLinkAddress >> 8) & 0xFF);
+        mBuffer[mPosition++] = (byte) (forwardLinkAddress & 0xFF);
+    }
 }
diff --git a/java/src/com/android/inputmethod/latin/personalization/DynamicPersonalizationDictionaryWriter.java b/java/src/com/android/inputmethod/latin/personalization/DynamicPersonalizationDictionaryWriter.java
index d446606..e43e74d 100644
--- a/java/src/com/android/inputmethod/latin/personalization/DynamicPersonalizationDictionaryWriter.java
+++ b/java/src/com/android/inputmethod/latin/personalization/DynamicPersonalizationDictionaryWriter.java
@@ -18,6 +18,7 @@
 
 import android.content.Context;
 
+import com.android.inputmethod.annotations.UsedForTesting;
 import com.android.inputmethod.keyboard.ProximityInfo;
 import com.android.inputmethod.latin.AbstractDictionaryWriter;
 import com.android.inputmethod.latin.ExpandableDictionary;
@@ -156,4 +157,10 @@
     public boolean isValidWord(final String word) {
         return mExpandableDictionary.isValidWord(word);
     }
+
+    @UsedForTesting
+    public boolean isInDictionaryForTests(final String word) {
+        // TODO: Use native method to determine whether the word is in dictionary or not
+        return mBigramList.containsKey(word);
+    }
 }
diff --git a/java/src/com/android/inputmethod/latin/personalization/DynamicPredictionDictionaryBase.java b/java/src/com/android/inputmethod/latin/personalization/DynamicPredictionDictionaryBase.java
index a08145b..be4c7c4 100644
--- a/java/src/com/android/inputmethod/latin/personalization/DynamicPredictionDictionaryBase.java
+++ b/java/src/com/android/inputmethod/latin/personalization/DynamicPredictionDictionaryBase.java
@@ -196,12 +196,6 @@
         return mLocale;
     }
 
-    @UsedForTesting
-    /* package for test */ void forceAddWordForTest(
-            final String word0, final String word1, final boolean isValid) {
-        addToPersonalizationPredictionDictionary(word0, word1, isValid);
-    }
-
     public void registerUpdateSession(PersonalizationDictionaryUpdateSession session) {
         session.setPredictionDictionary(this);
         mSessions.add(session);
diff --git a/java/src/com/android/inputmethod/latin/personalization/UserHistoryDictionaryBigramList.java b/java/src/com/android/inputmethod/latin/personalization/UserHistoryDictionaryBigramList.java
index 6c2c9e2..4c1803b 100644
--- a/java/src/com/android/inputmethod/latin/personalization/UserHistoryDictionaryBigramList.java
+++ b/java/src/com/android/inputmethod/latin/personalization/UserHistoryDictionaryBigramList.java
@@ -97,6 +97,10 @@
         return mBigramMap.isEmpty();
     }
 
+    public boolean containsKey(String word) {
+        return mBigramMap.containsKey(word);
+    }
+
     public Set<String> keySet() {
         return mBigramMap.keySet();
     }
diff --git a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.cpp b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.cpp
index 6bb90fc..405628b 100644
--- a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.cpp
+++ b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.cpp
@@ -43,10 +43,19 @@
                 dictBuf, mFlags, MAX_WORD_LENGTH, &pos);
     }
     if (isTerminal()) {
+        mProbabilityFieldPos = pos;
+        if (usesAdditionalBuffer) {
+            mProbabilityFieldPos += mBuffer->getOriginalBufferSize();
+        }
         mProbability = PatriciaTrieReadingUtils::readProbabilityAndAdvancePosition(dictBuf, &pos);
     } else {
+        mProbabilityFieldPos = NOT_A_DICT_POS;
         mProbability = NOT_A_PROBABILITY;
     }
+    mChildrenPosFieldPos = pos;
+    if (usesAdditionalBuffer) {
+        mChildrenPosFieldPos += mBuffer->getOriginalBufferSize();
+    }
     mChildrenPos = DynamicPatriciaTrieReadingUtils::readChildrenPositionAndAdvancePosition(
             dictBuf, mFlags, &pos);
     if (usesAdditionalBuffer && mChildrenPos != NOT_A_DICT_POS) {
diff --git a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.h b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.h
index acc68b3..2ee7c24 100644
--- a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.h
+++ b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.h
@@ -40,9 +40,11 @@
             const DictionaryShortcutsStructurePolicy *const shortcutsPolicy)
             : mBuffer(buffer), mBigramsPolicy(bigramsPolicy),
               mShortcutsPolicy(shortcutsPolicy), mNodePos(NOT_A_VALID_WORD_POS), mFlags(0),
-              mParentPos(NOT_A_DICT_POS),  mCodePointCount(0), mProbability(NOT_A_PROBABILITY),
-              mChildrenPos(NOT_A_DICT_POS), mShortcutPos(NOT_A_DICT_POS),
-              mBigramPos(NOT_A_DICT_POS), mSiblingPos(NOT_A_VALID_WORD_POS) {}
+              mParentPos(NOT_A_DICT_POS),  mCodePointCount(0),
+              mProbabilityFieldPos(NOT_A_DICT_POS), mProbability(NOT_A_PROBABILITY),
+              mChildrenPosFieldPos(NOT_A_DICT_POS), mChildrenPos(NOT_A_DICT_POS),
+              mShortcutPos(NOT_A_DICT_POS), mBigramPos(NOT_A_DICT_POS),
+              mSiblingPos(NOT_A_VALID_WORD_POS) {}
 
     ~DynamicPatriciaTrieNodeReader() {}
 
@@ -95,11 +97,19 @@
     }
 
     // Probability
+    AK_FORCE_INLINE int getProbabilityFieldPos() const {
+        return mProbabilityFieldPos;
+    }
+
     AK_FORCE_INLINE int getProbability() const {
         return mProbability;
     }
 
-    // Children node group position
+    // Children PtNode array position
+    AK_FORCE_INLINE int getChildrenPosFieldPos() const {
+        return mChildrenPosFieldPos;
+    }
+
     AK_FORCE_INLINE int getChildrenPos() const {
         return mChildrenPos;
     }
@@ -129,7 +139,9 @@
     DynamicPatriciaTrieReadingUtils::NodeFlags mFlags;
     int mParentPos;
     uint8_t mCodePointCount;
+    int mProbabilityFieldPos;
     int mProbability;
+    int mChildrenPosFieldPos;
     int mChildrenPos;
     int mShortcutPos;
     int mBigramPos;
diff --git a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.cpp b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.cpp
index cf70f92..6d1bd02 100644
--- a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.cpp
+++ b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.cpp
@@ -26,6 +26,9 @@
 
 namespace latinime {
 
+// TODO: Enable dynamic update and remove this flag.
+const bool DynamicPatriciaTrieWritingHelper::ENABLE_DYNAMIC_UPDATE = false;
+
 bool DynamicPatriciaTrieWritingHelper::addUnigramWord(
         DynamicPatriciaTrieReadingHelper *const readingHelper,
         const int *const wordCodePoints, const int codePointCount, const int probability) {
@@ -56,16 +59,21 @@
         }
         // All characters are matched.
         if (codePointCount == readingHelper->getTotalCodePointCount()) {
-            if (nodeReader->isTerminal()) {
-                // TODO: Update probability.
+            if (ENABLE_DYNAMIC_UPDATE) {
+                return setPtNodeProbability(nodeReader, probability,
+                        readingHelper->getMergedNodeCodePoints());
             } else {
-                // TODO: Make it terminal and update probability.
+                return false;
             }
-            return false;
         }
         if (!nodeReader->hasChildren()) {
-            // TODO: Create children node array and add new node as a child.
-            return false;
+            if (ENABLE_DYNAMIC_UPDATE) {
+                return createChildrenPtNodeArrayAndAChildPtNode(nodeReader, probability,
+                        wordCodePoints + readingHelper->getTotalCodePointCount(),
+                        codePointCount - readingHelper->getTotalCodePointCount());
+            } else {
+                return false;
+            }
         }
         // Advance to the children nodes.
         parentPos = nodeReader->getNodePos();
@@ -76,12 +84,14 @@
         return false;
     }
     int pos = readingHelper->getPosOfLastForwardLinkField();
-    // TODO: Remove.
-    return false;
-    return createAndInsertNodeIntoPtNodeArray(parentPos,
-            wordCodePoints + readingHelper->getPrevTotalCodePointCount(),
-            codePointCount - readingHelper->getPrevTotalCodePointCount(),
-            probability, &pos);
+    if (ENABLE_DYNAMIC_UPDATE) {
+        return createAndInsertNodeIntoPtNodeArray(parentPos,
+                wordCodePoints + readingHelper->getPrevTotalCodePointCount(),
+                codePointCount - readingHelper->getPrevTotalCodePointCount(),
+                probability, &pos);
+    } else {
+        return false;
+    }
 }
 
 bool DynamicPatriciaTrieWritingHelper::addBigramWords(const int word0Pos, const int word1Pos,
@@ -198,12 +208,58 @@
             newPtNodeArrayPos, forwardLinkFieldPos)) {
         return false;
     }
-    int writingPos = newPtNodeArrayPos;
+    return createNewPtNodeArrayWithAChildPtNode(parentPos, nodeCodePoints, nodeCodePointCount,
+            probability);
+}
+
+bool DynamicPatriciaTrieWritingHelper::setPtNodeProbability(
+        const DynamicPatriciaTrieNodeReader *const originalPtNode, const int probability,
+        const int *const codePoints) {
+    if (originalPtNode->isTerminal()) {
+        // Overwrites the probability.
+        int probabilityFieldPos = originalPtNode->getProbabilityFieldPos();
+        if (!DynamicPatriciaTrieWritingUtils::writeProbabilityAndAdvancePosition(mBuffer,
+                probability, &probabilityFieldPos)) {
+            return false;
+        }
+    } else {
+        // Make the node terminal and write the probability.
+        int movedPos = mBuffer->getTailPosition();
+        if (!markNodeAsMovedAndSetPosition(originalPtNode, movedPos)) {
+            return false;
+        }
+        if (!writeNodeToBuffer(originalPtNode->isBlacklisted(), originalPtNode->isNotAWord(),
+                originalPtNode->getParentPos(), codePoints, originalPtNode->getCodePointCount(),
+                probability, originalPtNode->getChildrenPos(), originalPtNode->getBigramsPos(),
+                originalPtNode->getShortcutPos(), &movedPos)) {
+            return false;
+        }
+    }
+    return true;
+}
+
+bool DynamicPatriciaTrieWritingHelper::createChildrenPtNodeArrayAndAChildPtNode(
+        const DynamicPatriciaTrieNodeReader *const parentNode, const int probability,
+        const int *const codePoints, const int codePointCount) {
+    const int newPtNodeArrayPos = mBuffer->getTailPosition();
+    int childrenPosFieldPos = parentNode->getChildrenPosFieldPos();
+    if (!DynamicPatriciaTrieWritingUtils::writeChildrenPositionAndAdvancePosition(mBuffer,
+            newPtNodeArrayPos, &childrenPosFieldPos)) {
+        return false;
+    }
+    return createNewPtNodeArrayWithAChildPtNode(parentNode->getNodePos(), codePoints,
+            codePointCount, probability);
+}
+
+bool DynamicPatriciaTrieWritingHelper::createNewPtNodeArrayWithAChildPtNode(
+        const int parentPtNodePos, const int *const nodeCodePoints, const int nodeCodePointCount,
+        const int probability) {
+    int writingPos = mBuffer->getTailPosition();
     if (!DynamicPatriciaTrieWritingUtils::writePtNodeArraySizeAndAdvancePosition(mBuffer,
             1 /* arraySize */, &writingPos)) {
         return false;
     }
-    if (!writeNodeToBuffer(false /* isBlacklisted */, false /* isNotAWord */, parentPos,
+    if (!writeNodeToBuffer(false /* isBlacklisted */, false /* isNotAWord */, parentPtNodePos,
             nodeCodePoints, nodeCodePointCount, probability, NOT_A_DICT_POS /* childrenPos */,
             NOT_A_DICT_POS /* originalBigramsPos */, NOT_A_DICT_POS /* originalShortcutPos */,
             &writingPos)) {
diff --git a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.h b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.h
index 524f361..16b84ba 100644
--- a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.h
+++ b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.h
@@ -49,6 +49,7 @@
  private:
     DISALLOW_IMPLICIT_CONSTRUCTORS(DynamicPatriciaTrieWritingHelper);
 
+    static const bool ENABLE_DYNAMIC_UPDATE;
     BufferWithExtendableBuffer *const mBuffer;
     DynamicBigramListPolicy *const mBigramPolicy;
     DynamicShortcutListPolicy *const mShortcutPolicy;
@@ -63,6 +64,16 @@
 
     bool createAndInsertNodeIntoPtNodeArray(const int parentPos, const int *const nodeCodePoints,
             const int nodeCodePointCount, const int probability, int *const forwardLinkFieldPos);
+
+    bool setPtNodeProbability(const DynamicPatriciaTrieNodeReader *const originalNode,
+            const int probability, const int *const codePoints);
+
+    bool createChildrenPtNodeArrayAndAChildPtNode(
+            const DynamicPatriciaTrieNodeReader *const parentNode, const int probability,
+            const int *const codePoints, const int codePointCount);
+
+    bool createNewPtNodeArrayWithAChildPtNode(const int parentPos, const int *const nodeCodePoints,
+            const int nodeCodePointCount, const int probability);
 };
 } // namespace latinime
 #endif /* LATINIME_DYNAMIC_PATRICIA_TRIE_WRITING_HELPER_H */
diff --git a/tests/src/com/android/inputmethod/latin/InputTestsBase.java b/tests/src/com/android/inputmethod/latin/InputTestsBase.java
index 0a1c4e9..da1fb6f 100644
--- a/tests/src/com/android/inputmethod/latin/InputTestsBase.java
+++ b/tests/src/com/android/inputmethod/latin/InputTestsBase.java
@@ -44,10 +44,12 @@
 
     private static final String PREF_DEBUG_MODE = "debug_mode";
 
-    // The message that sets the underline is posted with a 100 ms delay
+    // The message that sets the underline is posted with a 200 ms delay
     protected static final int DELAY_TO_WAIT_FOR_UNDERLINE = 200;
-    // The message that sets predictions is posted with a 100 ms delay
+    // The message that sets predictions is posted with a 200 ms delay
     protected static final int DELAY_TO_WAIT_FOR_PREDICTIONS = 200;
+    // The message that sets auto-corrections is posted within a 100 ms delay.
+    protected static final int DELAY_TO_WAIT_FOR_AUTOCORRECTION = 100;
 
     protected LatinIME mLatinIME;
     protected Keyboard mKeyboard;
@@ -221,6 +223,7 @@
     protected void type(final String stringToType) {
         for (int i = 0; i < stringToType.length(); i = stringToType.offsetByCodePoints(i, 1)) {
             type(stringToType.codePointAt(i));
+            sleep(DELAY_TO_WAIT_FOR_AUTOCORRECTION);
         }
     }
 
diff --git a/tests/src/com/android/inputmethod/latin/personalization/UserHistoryDictionaryTests.java b/tests/src/com/android/inputmethod/latin/personalization/UserHistoryDictionaryTests.java
index 1fd1b8a..1a20ec5 100644
--- a/tests/src/com/android/inputmethod/latin/personalization/UserHistoryDictionaryTests.java
+++ b/tests/src/com/android/inputmethod/latin/personalization/UserHistoryDictionaryTests.java
@@ -82,14 +82,29 @@
         }
     }
 
+    /**
+     * @param checksContents if true, checks whether written words are actually in the dictionary
+     * or not.
+     */
     private void addAndWriteRandomWords(final String testFilenameSuffix, final int numberOfWords,
-            final Random random) {
+            final Random random, final boolean checksContents) {
         final List<String> words = generateWords(numberOfWords, random);
         final UserHistoryPredictionDictionary dict =
                 PersonalizationHelper.getUserHistoryPredictionDictionary(getContext(),
                         testFilenameSuffix /* locale */, mPrefs);
         // Add random words to the user history dictionary.
         addToDict(dict, words);
+        if (checksContents) {
+            try {
+                Thread.sleep(TimeUnit.MILLISECONDS.convert(5L, TimeUnit.SECONDS));
+            } catch (InterruptedException e) {
+            }
+            for (int i = 0; i < 10 && i < numberOfWords; ++i) {
+                final String word = words.get(i);
+                // This may fail as long as we use tryLock on inserting the bigram words
+                assertTrue(dict.isInDictionaryForTests(word));
+            }
+        }
         // write to file.
         dict.close();
     }
@@ -103,7 +118,8 @@
         final Random random = new Random(123456);
 
         try {
-            addAndWriteRandomWords(testFilenameSuffix, numberOfWords, random);
+            addAndWriteRandomWords(testFilenameSuffix, numberOfWords, random,
+                    true /* checksContents */);
         } finally {
             try {
                 Log.d(TAG, "waiting for writing ...");
@@ -148,7 +164,8 @@
                 final int index = i % numberOfLanguages;
                 // Switch languages to testFilenameSuffixes[index].
                 addAndWriteRandomWords(testFilenameSuffixes[index],
-                        numberOfWordsInsertedForEachLanguageSwitch, random);
+                        numberOfWordsInsertedForEachLanguageSwitch, random,
+                        false /* checksContents */);
             }
 
             final long end = System.currentTimeMillis();