Thank you Dave! Sounds challenging, we'll see... :)
To clarify: To pull something like this off (if at all possible, which will depend on a number of factors), you need to effectively work with multiple daisy-chained trials. See e.g. http://www.millisecond.com/forums/Topic6989.aspx#6989 for something remotely related.
No, a <trial> collects a *single* response.
Is it possible to record more than one response given with the same key within a single trial?
I am trying to develop audio-motor synchronized tapping task, where participants are instructed to tap together with rhythmically presented tones (one one per trial). Thus, within a single trial, I can get at least two responses, one "delayed" response for the previous tone and another "premature" response for the tone to come.