If to measure is to “assign numerals to events,” queried audiologist Ira Hirsh in 1952, “what are the observable events in hearing?” In the newly professionalized field of audiometry, the test material for a typical audiogram—a set of individual tones—lent itself to precise electronic control, but it did not reflect everyday hearing situations and capacities. This chapter examines the use of speech to test hearing, from preliminary clinical applications of phonograph recordings in the late nineteenth century, to mass “screenings” with electrical recording and playback machines in the 1930s, to postwar diagnostic tests in which speech itself—from nonsense syllables to spondees to sentences—was forged into an objective measuring tool. The chapter argues that the quantification of “hearing loss for speech” derives from articulation testing in the field of telephone engineering. More specifically, the molding of speech sounds into yardsticks of “useful hearing” arose in the historical context of quality control, as did the notion that human hearing should be “screened” and inspected in industrial fashion.