Our DSL is token delimited, so for example a word in curlies is one type of variable and one in angle brackets is another. This is working fine matching pairs of delimiters like {}, <>, [] and (), but we also have some cases where the start and end delimiter are the same (eg. start ~variable~ or ^variable^). Using the Language Designer, I have a different explicit pattern for the Start Pattern Group and End Pattern Group in the former case. In the latter case, I have the same pattern, but that doesn't seem to do the trick. When I look at those with the same start/end delimiter in the Live Test, they don't transition out of their lexical state like those with different delimiters do. What is the right way to get them to behave similarly? Do they need to have a state transition?