The regex given to tokenize determines the separator between tokens, not the
content of the tokens.
I'm not quite sure what result you expect here (or in general). You can
probably tackle it using xsl:analyze-string, which is more powerful than
tokenize().

<xsl:analyze-string select="$string" regex=".{{1,6}}\^">
 <xsl:matching-substring>
   <line count="{ ?? }"><xsl:value-of select="."/></line>
 </xsl:matching-substring>
 <xsl:non-matching-substring>
   <non><xsl:value-of select="."/></non>
 </xsl:non-matching-substring>
</xsl:analyze-string>

<line count="1">....
<line count="2">....

thanks
andrew