At least one add-on is using the deprecated speech.REASON_* constants, and source code documentation advises using controlTypes.REASON_* constants instead. Specifically, Nick Stockton’s app modules uses this (based on looking at add-on source codes I have). As no module in the NVDA Core source code uses these and it is trivial for add-ons using the deprecated (and would-be removed) constants to change their behavior, I’ll go ahead and remove speech.REASON_* constants from the version of the NVDA Core source code I’m working on no later than February 10, 2017. Thus, please speak up if you absolutely must use speech.REASON_* constants in your modules before the removal date from my PR (pull request).
For more information on what’s happening, please see NVDA Core issue 6846 (link is above). Thank you.