In the context of the original article, the ability to "speak" is completely irrelevant since it's the privacy risks that are being highlighted. I think "microphone" is a perfectly fine way to capture the essence of the entire Alexa product line in this regard.
Actually all microphones can "speak". All speakers can be used as microphones as well. It's not like the conversion process only works one way.
That's kind of a technicality, the vast majority of non-mic'd speakers don't have all the required hardware or software to save or transmit audio to anything.