POETRY IN PERFORMANCE ENTAILS THE SOUNDING OF A TEXT in a live performance context. Sound events become voice documents through the application of sound engineering values that supplement an authorial voice-practice. This engineering affects the pitch or tempo of a performance and changes the relationship between sonic figure and sonic ground. This essay details the technical, aesthetic, and ideological values that Caedmon Records brought to bear upon sound recording and reproduction and compares these with those of other publishers, such as Columbia and Folkways, of recorded poetry during the same period. It also explains how the relatively slow tempo of postwar public speech, spoken word, and sound recording regimes imprinted the voice documents of the monaural era. As Charles Bernstein has noted, "[A] large archive of audio and video documents, dating back to an early recording of Tennyson's almost inaudible voice, awaits serious study and interpretation" (5). Bernstein argues for a formalist analysis or "close listening" to poetry in performance. The listening I propose, which is materialist rather than formalist, allows for analysis of the heterogeneity of poets' voice-practices as preserved in the voice documents of the recorded poetry archive I propose an aural criticism that attends to the material production of voice documents in a manner that is analogous to the focus of critics such as Jerome McGann and Lawrence Rainey on the material production of textual documents.
Many of the voice documents of the recorded poetry archive were produced during the monaural LP era, a period in media history that extended from 1948 to 1958. That era was part of a postwar "audio revolution" that marked a technical divide between partial and full sound spectrum recording and reproduction. It was enabled by three innovations: the long-playing vinyl record (LP) as the first medium to allow full sound spectrum reproduction of speech and sound events; magnetic audiotape as a medium of full sound spectrum recording and sound engineering; and high-fidelity sound reproduction systems. Several historical factors contributed to the social production of poetry as the spoken word content of the LP medium, but the editorial choices of Barbara Holdridge and Marianne Mantell, the two Columbia graduate students who founded Caedmon Records, were pre-eminent. Founded in 1952 with a recording of Dylan Thomas, the label soon began publishing recordings of the first generation of modern American poets and prose stylists, including T. S. Eliot, Ezra Pound, Gertrude Stein, William Carlos Williams, and Wallace Stevens. The aesthetic and technical choices of the company's first sound engineer, Peter Bartok, shaped the formal characteristics of these voice documents and contributed to the distinctive Caedmon sound.
Bartok was an exemplar of a sound reproduction value I call sonic realism. He was also an articulate spokesperson for the technical and aesthetic values that lay behind "lifelike" high-fidelity sound reproduction. In a 1954 article published in The Nation, Bartok defined the aesthetic and technical high fidelity standard in terms of its effect on the listener: "If the effect on the listener of a loudspeaker-produced sound is the same as it would be from a 'live performed' sound in a suitable auditorium with proper acoustics, then let us call the reproduction good" (485). Conversely, he also noted, "Any deviation in the sound characteristics due to acoustical conditions, in the time-relationship of the various sound components to each other, in their intensity relationship, in the components that are present, makes for bad reproduction" (485). The sound recording value I call "sonic romanticism" involves the manipulation of an electrical sound signal to produce certain forms of affect through, for example, the addition of reverb or the changing of the pitch of a speech or sound event. The sound recording value I call "sonic idealism" depends on sound editing, the manipulation of the figure/ground relationship, and a high signal-to-noise ratio. Caedmon's earliest voice documents reflected Bartoks value of sonic realism, but sonic idealism later prevailed as Caedmon co-founder Marianne Mantell (among others) began to supervise sound recording, editing, and mastering.
"Good sound" has aesthetic, technical, and ideological aspects. It is also historically variable. Electrical gramophone sound and L11 sound had different sound recording and sound reproduction values. The first was a product of modern sound recording and resulted from the use of electronic microphones and sound-absorbent recording studios, which produced a characteristically dead or non-reverberant modern sound. This sound was a product of the mass culture industry during a period that extended from the mid 1920s to the late 1940s. In contrast, postwar sound recording was characterized by the "lifelike" reproduction of an original sound event. Fundamentally, postwar sound was highly reverberant. The main difference between prewar partial spectrum sound recording and postwar full spectrum sound recording was that reverberant fields were recorded and reproduced only after 1947. This allowed for the reproduction of a speech or sound event reverberating in live or primary performance context with particular acoustical attributes. High fidelity recording was first used to record classical music. The symphony hall "without walls" was part of the opening of the field of elite or minority culture to mass audiences. Such recording was not only a technical and aesthetic value but also an ideological one.
Caedmon transferred the high fidelity sound into the field of spoken word recording because Bartok specialized in recording classical music. Bartok achieved a high fidelity sound by paying close attention to every element of the sound recording process, beginning with the acoustical attributes of the environments in which raw recordings were undertaken. Caedmon's earliest recordings were undertaken at Steinway Hall, also known as Carnegie Hall B. The warm acoustics of the small hall allowed Caedmon to simulate a speech event resonating in a small but public performance context as an alternative to the dead or non-reverberant and intimate voice favoured by large-scale industrial recorders. When not recorded at Steinway, recordings were undertaken in writers' homes, which were usually acoustically warm or reverberant because they were constructed from wood. These small room settings were intimate, but midrange microphone placement simulated a public address and preserved the sense of a speech event resonating in a primary performance context with particular acoustical characteristics.
Postwar spoken word recordings also manipulated the figure/ground relationship. Microphones had to be chosen to suit the vocal characteristics of each speaker. Some writers were seasoned public speakers; others not. Regardless of a writer's capacity for vocal projection, Bartok favoured mid-range mike placement because he aimed to simulate a speech event reverberating in a primary or live performance context. This choice produced a voice document with some environmental noise. Honking horns and other traffic sounds add a faintly percussive element to some Steinway Hall recordings. Writers' homes were generally quieter, but recording in these contexts involved other environmental sounds that were not as easily transcribed onto vinyl discs. Sean O'Casey's Caedmon recording was undertaken at a rustic home that lacked electricity and central heating. Bartok avoided reproducing the sound of the generator that was required to make the recording but thought that a background fire would lend a certain ambience to the record. The fire registered as a series of clicks when transcribed onto vinyl discs and each crackle had to be edited from the master tape However, the romantic sound of a distant train whistle was allowed to remain on the final voice document.
Unlike musical records, spoken word records required editing. Perfection was achieved by splicing together two or more taped performances. Bartok and other sound engineers typically removed pauses, errors, repetitions, coughs, and other bodily sounds. As a sonic realist, Bartok preferred to leave in background and breathing sounds, whereas Mantell favoured the removal of all bodily sounds, including breaths and the onsets and offsets of words. Her sonic idealism produced disembodied voice documents and significantly shaped the formal characteristics of each poet's voice-practice as reproduced on LP records. The most salient example of this sonic idealism involved William Carlos Williams's Caedmon record. The effects of a series of strokes hampered Williams's performance As a result, Mantell had to construct the cadence of Williams's spoken line based on her intuition of how this would sound if the poet were capable of reading "normally." The poet's wife later validated her choices, as did Williams's personal physician, who noted that Mantell had intuitively mimicked the rhythms of the poet's heartbeat (personal interview 27 May 2002).
Spoken word LPs began with the recording of a sound event in magnetic audiotape and the editing of that raw performance in order to produce a master tape Production of a record master and the pressing of copies followed. Pressing was a purely mechanical process, but record mastering involved aesthetic sound sense and technical expertise. Record grooves had to be spaced closely together to fit as many minutes of reading time as possible on each record, but when too closely spaced there was a transfer of sounds from one groove to the adjacent groove. This also occurred if records were mastered during the summer months before the advent of air conditioning, when the grooves in record masters softened before electroplating. Ghostly echoes between words are found on many of the voice documents from the gramophone and early LP era, although they often do not survive transcription into another medium.
Bartok and Mantell brought considerable aestheticism to the mastering process. That of Mantell, who lacked formal training, sometimes caused problems with the engineers who mastered Caedmon records. The most salient example of this occurred in the production of one of the last Dylan Thomas LPs during the early 1970s. Caedmon had obtained the recording from the BBC and sent it to RCA for mastering. However, when RCA sent back the test pressing Mantell thought that the pitch of the poet's voice was too high. She surmised that the turntable had been revolving too quickly when the master had been made. Mantell, who had known Thomas, could hear the difference However, she could not convince RCA'S engineers there was a problem. She feigned a headache in order to persuade them to leave her alone in the mastering studio, adjusted the tape speed manually, made a copy, and had the record mastered elsewhere.
Sound engineering clearly modifies authorial voice-practice in ways that complicate a close listening to such practice The speaking subject is not necessarily the person speaking. Similarly, the listening subject is not necessarily the person listening. The ears in the machine belong to engineers. It is their listening processes, rather than our own, that are reproduced in a sound recording. One way of understanding this is to approach sound recording as a representation of the psychological act of listening rather than reproduction of a speech and sound event. We are cultured to attend to information-bearing sounds and to ignore non-information-bearing sounds. A recording apparatus records all sounds according to their amplitude and proximity. Without different forms of sound figuration, sound signals would become lost in noise. (Amateur sound recordings often produce this effect.) In the case of voice documents, the recording process is manipulated in order to amplify the speaker's voice as a signifying sound. The engineer also constructs a social and spatial relationship between speaker and listener as the space of intersubjectivity. In the process, he or she constructs both a speaking and a listening subject.
In this sense, sound recording involves the construction of what Jean-Luc paucity calls a transcendental listening subject (quoted in Williams 56). Alan Williams argues that similar to the movements and placement of the camera, the movement and placement of the microphone function to synthesize "the world for some entity" (56). According to Williams, voice recording does not reproduce a sound or speech event so much as it represents signs of a physical situation in which these events occur (58). This simulation involves the construction of a "surrogate listening facility" that is at once ideological and spatio-psychological (61). Fundamentally, sound recording constructs "the subject that listens" (53). It depends on the relinquishing of individual aural perception and cognition in order to accept an "implied subject that actively perceives, whose demands we accept as our own" (61). This occurs when noise is eliminated from sound reproduction. "Good" sound engineering privileges the linguistic code and eliminates noise
When comparing the sound production values of the major producers of voice documents during the monaural L r, era, it becomes clear that Caedmon negotiated a position between "good" and "bad" sound. The counterculture label Folkways exhibited "bad" sound reproduction values because it eliminated noise. In The Dream Keeper and Other Poems of Langston Hughes read by the author, the audible turning on and off of the recording apparatus can be heard, for example Folkways' recordings also featured a maximum number of grooves per inch, which resulted in considerable echo and pre-echo. Furthermore, the label did not simulate a fixed point of audition. Hughes's voice shifted in intensity and amplitude as he shifted his position in relation to the fixed position of the microphone. As a result, the simulated spatio-psychological subjectivity or fixed sonic point that Williams describes was not achieved. The audient cannot surrender to a surrogate listening facility but remains aware that he or she is listening to a reproduction of Hughes's voice. In contrast, Columbia Records exhibited "good" sound recording values. As produced by Columbia, the voice of the poet was that of a physically proximate aural intimate Columbia favoured extremely close microphone placement. As a result, poets modulated their voices using intimate or conversational speech. Columbia's voice documents were also remarkably embodied because the label's engineers did not remove bodily sounds. The Columbia sound eliminated all other forms of noise, however. All of these elements created conditions in which the audient completely surrendered to a surrogate listening facility.
Caedmon's highly reverberant sound contrasted with the intimate, dead sound of the postwar spoken word industry. Caedmon featured mid-range mike placement instead of aural close-ups. As a consequence, poets did not use codes of private or conversational speech but instead adopted more public forms of address. This placement resulted in a less embodied sound than Columbia's, especially when combined with sound editing that imposed textuality on the speech stream and contrasted with Columbia's emphasis on speech. Such editing was a "good" sound production value However, Caedmon's early voice documents generally did not allow the auditor to fully surrender to a surrogate listening facility because Bartok's sonic realism, and the background noise he typically left on the finished recording, functioned to prevent it.
Sound recording values were objectively polarized within the postwar spoken word industry. In general, each company's sound appears to have imitated different media cultures. Folkways privileged loose speech as recorded in acoustical era gramophone recording, which did not allow for sound editing; its anachronistic sound also imitated the ethnological disposition of the phonograph and graphophone acoustical era sound recording. Columbia's sound mimicked postwar radio voice as a rehabilitated modern sound regime characterized by the ideological imperative to generate affect. (To this end, postwar radio voice typically simulated spatial and psychological intimacy.) In contrast, Caedmon simulated a live reading reverberating in a public performance context and imitated the high-fidelity sound of the postwar classical music industry. However, its high signal-to-noise ratio also mimicked the black and white of the printed page, effectively functioning as an analogue to the written word.
Voice recording was also unified by a stately tempo that characterized postwar public speech, spoken word, and reading regimes. This tempo reflected the political morality of public secular reading during the postwar period and a liberal democratic imperative to hail the so-called "common man" via technologies of secondary orality. Walter Ong coined the term "secondary orality" to refer to the voicing of texts, and literary texts in particular, by electronic media (136). However, the production of texts as the content of postwar technologies of voice transmission and voice reproduction has a pre-history that Ong did not explicitly acknowledge Public secular reading became discursive during the postwar era because of the role of such reading during wartime as a response to the Nazi attack on literature and the so-called "origins" of print media. As a form of political morality, this practice supplemented the public reading of religious texts. Public reading also became the foundation of the postmodern public sphere during the postwar era and supplemented the role of literacy as the foundation of the modern public sphere
The slow tempo of postwar poetry reading was also shaped by the live performance culture of the period. Poets read largely before university audiences during this era, and, as a result, their readings were educational and monological in disposition (hear CD track 36 of e.e. cummings's "sweet spring is your"). That tempo was also determined by disposition of radio as a medium for mass education, including WGBH and NAEB programs directed to the so-called "common man" outside the university system and VOA programs directed to international audiences who spoke English as a second language. Whether the result of a poet's address to a real or imagined audience or the result of supplemental sound engineering, this tranquilized tempo reflects public speech regimes that are historical.
When undertaking a close listening, one must be sensitive to how historical voice regimes determined a poet's voice-practice and how a poet's address to a particular audience, whether real or imagined, shaped that practice One must also attend to other elements that are completely extraneous to such practice. These include the disposition of the original medium of the voice document, whether this was an electrical era gramophone master or copy, an instantaneous acetate disc recording as the medium used to record radio performances from the mid-1930s to the mid-1940s, or magnetic audiotape as a medium that permitted sound editing. One must also attend to sound engineering practices that supplement authorial voice-practice, including the sound values of institutional and commercial recorders and those of particular engineers. All of this contributes to the "schizophonia" of the recorded poetry archive (Schafer 1977).
As a consequence, when Bernstein suggests that "the unanticipably [sic] slow tempo of Wallace Stevens's performance tells us much about his sense of the poem's rhythms and philosophical seriousness" (6), it seems relevant to query if the poet was conforming to codes of academic decorum that govern the performance of poetry in educational settings, including his recorded Harvard readings (on CD track 37, listen to Stevens reading from "The Idea of Order at Key West"). Alternatively, one might question if Stevens's voice-practice reveals the exigencies of educational broadcasting on the WGBH Amherst/Boston educational radio station or even that of the VOA, particularly given the fact that another example Bernstein cites (Carl Sandburg's "melodramatic" style when reading "in the toooooombs, in the coooool toooooombs") seems to exhibit the slow tempo of Basic English as the speech regime favoured by the network. In fact, Stevens recorded for both. In sum, it is necessary to query the provenance of the voice documents of the recorded poetry archive, particularly with reference to what can be termed, after Bernstein, second or third edition recordings or recordings that were not undertaken by commercial recorders but which were nonetheless subsequently commercially published by them.
This kind of close listening is highly relevant when considering recordings of modern poets that were made during the postwar era. The archive contains comparatively few voice documents from the era of partial sound spectrum recording. For all intents and purposes, institutional recording of poets began in the mid-1930s and consolidated on a commercial scale during the late 1940s and early 1950s. However, accounts of earlier performances have been preserved in writers' letters, diaries, and memoirs. For example, Virginia Woolf described T. S. Eliot's private 1922 performance of The Waste Land for herself and her husband Leonard in her diary. The poet "sang it & chanted it rhythmed [sic] it" (Lee 443). Eliot's stylized reading contrasts with his voice-practice when reading the work for institutional recorders (including his 1946 reading for the NBC University of the Air program, excerpted on CD track 38, which was recorded by the Library of Congress and was the most widely disseminated performance of The Waste Land) or his 1953 live reading at a football stadium in Minneapolis where he read for a crowd of thirteen thousand. Different audiences determined the characteristics of Eliot's voice-practice on these occasions. The "eerily depersonalized vocal style" described by Bernstein can be interpreted as a refusal of secondary orality on aesthetic grounds, even as Eliot clearly understood and supported the ideological function of large-scale public reading during this period (12). Eliot's disembodied voice was also a manifestation of late electrical-era gramophone recording and of poor mastering by the Library of Congress. His voice is notably warmer, deeper, and more lifelike on his 1955 Caedmon LP (CD track 39: Eliot reading from "The Love Song of J. Alfred Prufrock").
The heterogeneity of the recorded poetry archive of the LP era emerges most clearly when listening to recordings of Dylan Thomas, a poet at the centre of the postwar popularization of poetry. Thomas's 1952 and 1953 Caedmon albums were very different from one another because of the genre of the works recorded, his voice-practice, his health and state of mind on each occasion, and the nature of his address to those present at each recording session (compare Thomas reading from "Fern Hill" CD track 40, then from "A Refusal to Mourn the Death of a Child by Fire, in London" CD track 41). This heterogeneity was multiplied with the posthumous publication of Dylan Thomas Reading His Complete Recorded Poetry in 1964, which drew from the entirety of the Thomas archive, including his Caedmon LPs, his radio performances, and his live readings before university audiences. The voice that emerged on the album was of a seemingly fragmented personality who performed very differently in different contexts. However, much of this multiplicity arose from the sound recording values of different institutional and commercial recorders, and the disposition of different sound recording media, which seem to speak as much as the poet himself Some of this schizophonia resulted from the reengineering of previously published voice documents, including an archival recording of three poems that had been included on side two of Dylan Thomas Reading, Volume 3, in which Thomas spoke in a slow, thick voice and had difficulty pronouncing the patterns of consonance in his works (listen to CD track 42 of Thomas reading from "Of Light Breaks" version 1). That recording appears to have been technically manipulated, primarily through the addition of reverb, in order to simulate vocal projection and a public acoustic context that was lacking in the earlier version (listen to CD track 43 of Thomas reading from "Of Light Breaks" version 2).
All analogue-era voice and sound documents have been re-engineered in the digital era. When listening to recordings from the recorded poetry archive in digital form, one is listening to eighth, ninth, or tenth generation recordings that bear little resemblance to the first or second edition voice documents. The tempo and pitch of an author's voice are often different on analogue and digital versions of the same recording. In addition to material differences between analogue and digital sound, "good" recording values have typically been imposed in the re-engineering process. CD voice is non-reverberant, disembodied, and highly edited, although mid-mike placement still simulates a live reading. Perhaps most importantly, environmental noise is now removed. This editing includes the background noise that grounded poets' archival readings before live audiences. It also includes the environmental sounds Bartok left on Caedmon's voice documents and the sound of a speech event decaying at low frequencies in empty rooms with acoustically warm characteristics, which was the essence of Caedmon's sound: very literally, this was the sound of a poet's voice reverberating in space and time.
Voice documents are produced from the interaction between an authorial voice-practice and a sound engineering practice, but the dynamic between these during the LP era is not recoverable without the various forms of noise that were so much a part of the voice documents of this period. This noise does not survive digital re-engineering or, indeed, even amateur transcription to another medium. In losing this noise, we lose not only the closest approximations of the authorial voice-practices of poets who are now dead--"and of poetry as an event in sound and time, ramifying through the social, language, technology, history" as the call for papers for this volume has it--but also the sound figuration practices, and sonic history, that grounded the publication of their voices during the LP era.
Audio tracks cited in this article are available on the compact disc accompaning the print version of this special issue Some of the audio tracks cited in this article may also be available at www.arts.ualberta.ca/~esc under the "Extras" tab.
Bartok, Peter. Taped personal interview. 5 February 1998.
Bartok, Peter. "It's All in the Groove" The Nation. 4 December 1954: 484-85.
Bernstein, Charles, ed. Close Listening: Poetry and the Performed Word. Oxford: Oxford up, 1998.
Bourdieu, Pierre. The Field of Cultural Production. Trans. Randall Johnson. New York: Columbia up, 1993
The Caedmon Treasury of Modern Poets Reading. up. Caedmon Records, 1961.
cummings, e.e. e.e. cummings Reading His Poetry. LP. Caedmon Records, 1954
Eliot, T. S. T. S. Eliot Reading Poems and Choruses. Le. Caedmon Records, 1956.
--. "The Waste Land." The Caedmon Treasury of Modern Poets Reading. LP. Caedmon Records, 1961.
Eargle, John. Sound Recording. 2nd ed. New York: Van Nostrand Reinhold, 1980.
Holdridge, Barbara. Taped personal interview. 5 August 2001.
Hughes, Langston. The Dream Keeper and Other Poems of Langston Hughes. Le. Folkways, 1955
Lee, Hermione. Virginia Woolf. London: Vintage, 1997
Levy, William Turner. Affectionately, T S. Eliot. Philadelphia: J. B. Lippincott, 1968.
Mantell, Marianne Taped personal interview. 25 and 27 May 2002.
Ong, Walter. Orality and Literacy: The Technologizing of the Word. London: Methuen, 1982.
Pleasure Dome: An Audible Anthology of Modern Poetry Read by its Creators. LP. Columbia Records, 1949.
Schafer, R. Murray. The Tuning of the World. Toronto: McLelland and Stewart, 1977.
Stevens, Wallace. Wallace Stevens Reading His Poetry. LP. Caedmon Records, 1957
Thomas, Dylan. Dylan Thomas ReadingA Child's Christmas in Wales and Five Poems. LP. Caedmon Records, 1952.
--. Dylan Thomas Reading Volume z. LP. Caedmon Records, 1953
--. Dylan Thomas Reading Volume 3. LP. Caedmon Records, 1956.
--. Dylan Thomas Reading His Complete Recorded Poetry. LP. Caedmon Records, 1964.
Williams, Alan. "Is Sound Recording Like a Language?" Yale French Studies 60 (1980): 51-66.
University of British Columbia
SARAH PARRY teaches at the University of British Columbia. Her expertise includes acoustical poetics and sound and voice publishing history. She has published articles and reviews in Performance Research and New Formations.…