In this dissertation, I propose a reconsideration of the idea of speech that animates the writing and reading of lyric poetry. The project revisits competing linguistic and literary definitions of speech that emerged from nineteenth-century debates in the European academy over the nature of language, and reads the transformation of the lyric poem and its diction as a reaction to the Romantic ideal of language—a diction taken from “real language of men,” as Wordsworth famously formulates it in the Preface to Lyrical Ballads. I argue that poetry critics can revisit the definition of poetic language by turning to the principles of dialectology (or glottologia), a branch of historical linguistics pursued primarily by Italian scholars in the latter half of the nineteenth century, especially its resistance to the widely held idea that speech is inferior to the written word. Through readings of neglected archival sources in dialect lexicography and grammar alongside both canonical and non-canonical lyric poems by Leopardi, Coleridge, Barnes, Hopkins, and Manzoni, I offer a general theoretical claim about the nature of the lyric: that poetic language ought to be read as the speech of geographically-bounded communities of speakers over the longue durée, not merely, as many critics propose, from an immediate historical context nor from an idealized timeless present tense. The work of these poets confronts us with the idea that the specific figurative or rhetorical uses of words and phrases in individual poems may draw discriminately from deep reservoirs of historical meaning as a consequence of the social, geographical world that the poem establishes. By extension, these poets show through their art that everyday speech contains deep roots in social communities. In each of four chapters, my project reevaluates a different common assumption of literary criticism on how speech influences poetic language. The introductory chapter argues that Locke’s influential rationalist conception of language, which continues to dominate our thinking about the spoken language, saw crucial revisions toward a socio-historical plane (primarily in Condillac, Cesarotti, and Ascoli). Based on new analysis of the internal anachronism of spoken language, these critics redefined language function not in terms of its immediate contextual usage but in terms of its deep history. In the main critical chapters, I then focus on two canonical poets of the early nineteenth-century lyric, Leopardi and Coleridge, each of whom presents a different model of poetic language based on figures found embedded in the linguistic strata of individual words. For Coleridge, the lyric speaker demonstrates the limits of individual speech capacities by using an artificially hybridized diction that illuminates the German and French roots of English; by contrast, Leopardi’s use of archaisms reveals the deep resemblance of spoken Italian to Vulgar Latin—a demonstration of the permanence of speech against the dramatic linguistic changes of written Italian. My final chapter considers the linguistic turn to dialectology through the experimental philology and poetry of William Barnes. By creating an artificially standardized language with no real-world equivalent speech community, Barnes’s experiments with dialect poetry and a universal Standard English make visible the inherent anachronism of speech. This project ultimately aims to be prospective: I affirm the need for literary scholars to reconsider their implicit allegiance to a simple, transparent concept of speech by ignoring the historical and material dimension of poetic language—important precisely for a hybrid genre like the lyric, dependent as it is on the deep history and evolution of speech as its touchstone.