diff --git a/pmd-java/etc/grammar/Java.jjt b/pmd-java/etc/grammar/Java.jjt
index ee668ea9c4..cc4afba736 100644
--- a/pmd-java/etc/grammar/Java.jjt
+++ b/pmd-java/etc/grammar/Java.jjt
@@ -232,7 +232,7 @@ options {
   STATIC = false;
   USER_CHAR_STREAM = true;
   JDK_VERSION = "1.5";
-  TOKEN_FACTORY = "JavaTokenUtils";
+  TOKEN_FACTORY = "JavaTokenFactory";
 
   MULTI = true;
   VISITOR = true;
@@ -2999,14 +2999,14 @@ void AssertStatement() :
 void RUNSIGNEDSHIFT() #void:
 {}
 {
-    LOOKAHEAD({ JavaTokenUtils.getRealKind(getToken(1)) == RUNSIGNEDSHIFT})
+    LOOKAHEAD({ JavaTokenFactory.getRealKind(getToken(1)) == RUNSIGNEDSHIFT})
     ">" ">" ">"
 }
 
 void RSIGNEDSHIFT() #void:
 {}
 {
-    LOOKAHEAD({ JavaTokenUtils.getRealKind(getToken(1)) == RSIGNEDSHIFT})
+    LOOKAHEAD({ JavaTokenFactory.getRealKind(getToken(1)) == RSIGNEDSHIFT})
     ">" ">"
 }
 
diff --git a/pmd-java/src/main/ant/alljavacc.xml b/pmd-java/src/main/ant/alljavacc.xml
index 02629e976e..75faa43c60 100644
--- a/pmd-java/src/main/ant/alljavacc.xml
+++ b/pmd-java/src/main/ant/alljavacc.xml
@@ -145,20 +145,20 @@
 
 
         <replace file="${target-package-dir}/JavaParserTokenManager.java">
-            <replacetoken><![CDATA[t = JavaTokenUtils.newToken(jjmatchedKind, curTokenImage);
+            <replacetoken><![CDATA[t = JavaTokenFactory.newToken(jjmatchedKind, curTokenImage);
 
    t.beginLine = beginLine;
    t.endLine = endLine;
    t.beginColumn = beginColumn;
    t.endColumn = endColumn;]]></replacetoken>
             <replacevalue>
-                <![CDATA[t = JavaTokenUtils.newToken(jjmatchedKind, curTokenImage, beginLine, endLine, beginColumn, endColumn);]]></replacevalue>
+                <![CDATA[t = JavaTokenFactory.newToken(jjmatchedKind, curTokenImage, beginLine, endLine, beginColumn, endColumn);]]></replacevalue>
         </replace>
 
         <replaceregexp file="${target-package-dir}/JavaParserTokenManager.java" flags="s">
             <regexp pattern="protected net.sourceforge.pmd.lang.ast.impl.JavaccToken jjFillToken.*?}" />
             <substitution
-                    expression="protected net.sourceforge.pmd.lang.ast.impl.JavaccToken jjFillToken() {return JavaTokenUtils.newToken(jjmatchedKind, input_stream);}"/>
+                    expression="protected net.sourceforge.pmd.lang.ast.impl.JavaccToken jjFillToken() {return JavaTokenFactory.newToken(jjmatchedKind, input_stream);}"/>
         </replaceregexp>
 
 
diff --git a/pmd-java/src/main/java/net/sourceforge/pmd/lang/java/ast/JavaTokenUtils.java b/pmd-java/src/main/java/net/sourceforge/pmd/lang/java/ast/JavaTokenFactory.java
similarity index 84%
rename from pmd-java/src/main/java/net/sourceforge/pmd/lang/java/ast/JavaTokenUtils.java
rename to pmd-java/src/main/java/net/sourceforge/pmd/lang/java/ast/JavaTokenFactory.java
index 6ac09c6d35..05e09d790c 100644
--- a/pmd-java/src/main/java/net/sourceforge/pmd/lang/java/ast/JavaTokenUtils.java
+++ b/pmd-java/src/main/java/net/sourceforge/pmd/lang/java/ast/JavaTokenFactory.java
@@ -9,15 +9,18 @@ import net.sourceforge.pmd.lang.ast.impl.JavaCharStream;
 import net.sourceforge.pmd.lang.ast.impl.JavaccToken;
 import net.sourceforge.pmd.lang.ast.impl.TokenDocument;
 
-final class JavaTokenUtils {
+final class JavaTokenFactory {
 
-    private JavaTokenUtils() {
+    private JavaTokenFactory() {
 
     }
 
     static JavaccToken newToken(int kind, CharStream charStream) {
         JavaCharStream jcs = (JavaCharStream) charStream;
 
+        // Most tokens have an entry in there, it's used to share the
+        // image string for keywords & punctuation. Those represent ~40%
+        // of token instances
         String image = JavaParserTokenManager.jjstrLiteralImages[kind];
 
         switch (kind) {
@@ -32,12 +35,12 @@ final class JavaTokenUtils {
                 jcs.getEndOffset(),
                 jcs.getTokenDocument()
             );
-        case JavaParserConstants.WHITESPACE:
-            // unlikely that anybody cares about that, and since
-            // they're still 30% of all tokens best make this assumption
 
-            // btw 40% of all tokens have a compile-time string constant
-            // as image (jjstrLiteralImages) so they're shared.
+        case JavaParserConstants.WHITESPACE:
+            // We don't create a new string for the image of whitespace tokens eagerly
+
+            // It's unlikely that anybody cares about that, and since
+            // they're still 30% of all tokens this is advantageous
             return new LazyImageToken(
                 kind,
                 jcs.getStartOffset(),