The purpose of the present experiment was to replicate and extend the literature on using selection-based instruction to teach responses to interview questions by (a) evaluating the emergence of recombinative (i.e., combinations of taught) and novel (i.e., untaught) topography-based intraverbal responses, in addition to exact repetitions of taught responses, (b) providing a measure of social validity for the emergent response subtypes, and (c) including a touch-screen video interviewing component. Participants were two young adult males with a learning disability who attended a local vocational development center. Increases in accurate intraverbal responding and decreases in inaccurate responding across most interview questions were observed in both participants at posttest. Increases in the number of accurate recombinative responses were observed for both participants, and increases in accurate exact responses were observed for one participant. One accurate novel response was observed at posttest for each participant. Results suggest that selection-based instruction can generate appropriate topography-based responses that sound unscripted and the addition of a video component was viable.