Back in the early eighties my grandmother had a stroke and could only say 4 words, we were amazed when we put music on to hear her sing as if she had no speech problems whatsoever.
Hypothetically, if I met him the next day at the grocery store, what are the chances he would be able to respond with that phrase? I'm interested in the generalization of this method in everyday life.
The chances are quite good. If you complete the three stages for every-day phrases you usually see good results and there is also a generalization effect that has been proven by Schlaug et al. 2009. Similar to the fasciculuts arcuatus, which connects the Broca and Wernicke area, researchers found evidence for the formation of nerve fibres on the non dominant language hemisphere that fulfill a similar task when using M.I.T. However, in reality, there are only two types of patients that I would use this method with. 1) Patients in the acute phase. Here the symptoms can change dramatically and you never know whether therapy has the biggest influence on it. 2) Patients with global aphasia. Here the patient's understanding of speech is heavily impacted. So chances are they won't even understand you when you try to communicate with them in the store. So, yes there are positive effects in every-day communication by learning these phrases but you need to complete all the stages and work very frequently and on more things than just speech prodcution.
@Gimmo Sleep is unfortunately wrong. I have spent the past three years researching the neural mechanisms and neural plasticity of melodic intonation therapy when used in speech language therapy, and unless its provided by a trained musical therapist, and quantifiable changes can be measured via fMRI, it is not appropriate (and not good EBP in any manner in CSD). 1. It does not generalize. 2. as noted by gimmo, theres a similar area in the right hemisphere that contributes to language. not noted by gimmo, MIT can *transfer* the language function to the right homologous region and this only will exacerbate language disorders. 3. there is not one good quality experiment investigating MIT that also measure maintenance and generalization. Also, using it in the acute phase also might just be the worst time to use it. Target the left hemisphere and target the underling neural substrates at level of impairment.