Updated phrasing for the manual.

doc_src/en/Dialogs_ProjectProperties.xml

@@ -125,12 +125,12 @@
 		  <para>Switching from <emphasis role="bold">sentence</emphasis> to
 		  <emphasis role="bold">paragraph</emphasis> segmentation
 		  mid-translation may force OmegaT to upgrade old translation memories
-		  that did not use sentence segmentation, but not vice versa. However,
-		  OmegaT will attempt to create fuzzy matches for paragraphs by
-		  combining existing sentence translations. See <link
+		  that did not use sentence segmentation, but not vice versa.</para>
+		  <para>OmegaT can also optionally attempt to create fuzzy matches for paragraphs
+		  by assembling existing sentence translations. See <link
 		  linkend="dialog.preferences.tm.matches.paragraph.from.segmented.tmx"
-		  endterm="dialog.preferences.tm.matches.paragraph.from.segmented.tmx.title"/> to enable/disable
-		  the feature.
+		  endterm="dialog.preferences.tm.matches.paragraph.from.segmented.tmx.title"/> to enable
+		  that feature.
 		  </para>
 
 		  <para>If you change the segmentation while translating, you will have

doc_src/en/HowTo_UseTM.xml

@@ -167,20 +167,21 @@
 	  </listitem>
 
           <listitem>
-            <para>When you ever run projects which use <link
+            <para>If you switch to paragraph segmentation in a project that originally used <link
 			    linkend="dialogs.project.properties.options.segmentation"
 			    endterm="dialogs.project.properties.options.segmentation.title"/>,
-		    and the current project is switched to the paragraph segmentation, you can enable a preference
-		    option <link
+		    you can have OmegaT present matches assembled from existing segments by enabling the
+		    <link
 	    linkend="dialog.preferences.tm.matches.paragraph.from.segmented.tmx"
-	    endterm="dialog.preferences.tm.matches.paragraph.from.segmented.tmx.title"/> that allow OmegaT matches
-		    in special way. In addition to normal fuzzy matches, OmegaT try to segment source text into
-		    sentences, and conduct fuzzy match on the segmented TMX entries, then construct paragraph from
-		    the result. A suggested paragraph can be one that is not in the translation memories. It is why
-		    an origin of TM is shown as empty.
-		    Please refer <link linkend="dialog.preferences.segmentation.setup.type"
-				    endterm="dialog.preferences.segmentation.setup.type.title"/> to know pros and
-		    cons of segmentation configuration.
+	    endterm="dialog.preferences.tm.matches.paragraph.from.segmented.tmx.title"/> option in addition to
+			showing regular fuzzy matches.</para>
+			<para> OmegaT will try to segment the source text into sentences,
+			apply fuzzy matching on the segmented memory entries, and then assemble a paragraph from those matches.
+		    No origin is shown for the match because the assembled paragraph may not exist in the
+			translation memories.</para>
+		    <para>Refer to the <link linkend="dialog.preferences.segmentation.setup.type"
+			endterm="dialog.preferences.segmentation.setup.type.title"/> section for more information on
+		    the benefits and drawbacks of the different types of segmentation.
             </para>
           </listitem>
 	</itemizedlist>

doc_src/en/OmegaT_Preferences.xml

@@ -1484,12 +1484,10 @@ ${filePath}></programlisting></para>
 	  </varlistentry>
           <varlistentry id="dialog.preferences.tm.matches.paragraph.from.segmented.tmx">
 		  <term id="dialog.preferences.tm.matches.paragraph.from.segmented.tmx.title">
-			  <option>Paragraph from segment matches of segmented TMX on a non-segment project</option></term>
-            <listitem><para>Enable matches for paragraph, which is constructed from segment matches search on segmented
-		    TMX entries searched with sentences from a segmented source text, even when running on a
-		    non-segment(paragraph segmentation) project.</para>
+			  <option>Show matches assembled from multiple segments</option></term>
+            <listitem><para>Enable matches for paragraphs assembled from matches found in segmented
+		    memory entries from a segmented source text in a project that uses paragraph segmentation.</para>
             </listitem>
-
           </varlistentry>
 	</variablelist>
   </section>