token.[] -> edgeType='tok' tokenStr=content.surface
analysis.[] -> edgeType='morph' stem=content.transliteratedLexiconItem