The Computational Paralinguistics Challenge [Social Sciences]

摘要

As Friedrich Nietzsche stated more than a century ago, “The most intelligible factor in language is not the word itself, but the tone, strength, modulation, tempo with which a sequence of words is spoken—in brief, the music behind the words, the passions behind the music, the person behind these passions: everything, in other words, that cannot be written” [1]. However, nonlinguistic aspects were broadly conceived as fringe phenomena until the attitude changed at slow pace half a century ago [2]. Paralanguage—literally “alongside” language—is researched more widely only since the term was arguably first mentioned by the linguist Archibald Hill in 1958. Paralinguistics, first named roughly at the same time by George Leonard Trager [3]—can be limited to “vocal factors” according to David Abercrombie and David Crystal roughly a decade later—also all linguists. With the advent of modern computing devices, a new branch of paralinguistics allowed for their automatic processing.