If someone were to advance the thesis that dogs were incapable of understanding words I would require some pretty strong evidence before I would believe them.
The soundboard stuff in the article -- not so much. Training a dog to press a button is simple conditioning, but implying any sort of understanding of the language is a bit of a stretch. It sounds like the referenced research (I haven't read the paper) only tests the other way around (that dogs can understand some speech).
If you can show me a dog trained to understand "stand on the yellow bench" and similar commands, but never actually hearing the specific command "sit on the red chair" and being able to execute that, then maybe I'm interested, but otherwise a dog hitting a button that says "I love you" is great for instagram but is indistinguishable from a button that says "look at me and pet me please".