Art song, by nature, is a combination of media: text and music. Composers map these two domains onto each other using established correspondences known as conceptual metaphors. The outcome, known as a conceptual blend, manifests a different cognitive perception than if the domains are experienced separately. The purpose of this dissertation is to keep these blends intact when performing instrumental transcriptions of vocal music. Without the ability to utter words, the performer must consider how they can bring life to the narratives through their instrument. A comprehensive set of performance practice guidelines are discussed, largely for practical application to the trumpet. Accompanying this dissertation are seventeen video recorded transcriptions of 20th century art song from German, French, and American composers, which implement a solution to be used along with these guidelines: subtitle translations. The addition of subtitles not only allows for the audience to experience the narratives in real-time with the music, but also creates an opportunity for instrumentalists to perform vocal repertoire that has yet to be explored in great depth. The expressive atonal and twelve-tone works of the Second Viennese School, experimental and Transcendental works of Charles Ives, deep Symbolist poetry set by Lili Boulanger, and evocative, text-driven works of Libby Larsen can be brought to life in compelling ways with their narratives intact.