Asterisks are for presentational purposes only, and parsers must not save asterisks as part of the linguistic data for the utterance.
This should be changed to:
Parsers should pass through asterisks unaltered as part of the data by default. However, it is recommended that parsers provide users with the option to retain or strip emphasis, or access both data formats in different places.
Asterisks placed elsewhere should be stripped from the data and ignored by parsers.
This should be changed to:
Asterisks placed elsewhere should be ignored by parsers.
If an odd number of asterisks is found, they should be stripped from the data and ignored.
This should be changed to:
The behavior of parsers when an unpaired asterisk is found (an odd number of asterisks) is unspecified.
This should be changed to:
This should be changed to:
This should be changed to: