dd.say[] and dd.ask[] accept rich-text, so one alternative might be to represent your input options as hidden rich-text fields. You could also use rtext.cat[] to convert an image stored in a canvas into a rich-text table, or use any other method of constructing it on the fly.
Suppose you have a pair of fields named "op1" and "op2". Their "value" attribute is their rich-text, which is represented in Lil as a table of text runs and attributes. It is important to wrap each table in a list with the "list" primitive like so; otherwise the comma operator will join the rows of those tables and make all the options in the prompt "stick together":
dd.ask[ "Some Question" (list op1.value),(list op2.value) ]
