Transcribing Oral History in the Digital Age

Transcribing Oral History in the Digital Age

by Linda Shopes

Background  

Transcribing, that is, representing in print that which has been spoken, has long been established as one of oral history’s best practices. Transcribing is related to oral history’s origins and development as an archival practice with the goal creating documents for future use by scholars and other researchers. A transcript emphasizes the contents of an interview, the new knowledge communicated by the narrator’s words.

At its best, transcription– the process of creating an archival-quality transcript– involves several steps:

1. a word-for-word rendering in print of the words (and non-verbal utterances) spoken in the interview with minimal editorial intervention;
2. review of the resulting document by the interviewer for accuracy and correction as necessary;
3. review by the narrator for accuracy and correction, emendation, amplification, and occasionally, redaction or restriction of certain materials;
4. revision per narrator changes;
5. editing and annotation for sense, context, etc;
6. indexing; and
7. cataloguing.[1]

Transcription can claim several obvious advantages. Among those most often noted is that it facilitates use. Researchers have traditionally been much more familiar– and hence comfortable– with paper than with audio and video documents and, arguably, are more likely to consult the transcript of an interview than listen to or view it in its original medium. Because it is easier and faster to skim a written document visually than to listen to an audio or view a video, it is easier both to locate relevant material in a transcript and, importantly, if an index can pinpoint a specific reference, to browse in and around the context of the reference. (However, as discusssed below sophisticated systems of digital annotation and indexing can facilitate a kind of skimming of the audio/video document and may come to obviate this point.)  Furthermore, insofar as it does resemble a traditional document, a transcript might also be understood as conferring a certain legitimacy on what has sometimes been construed as an ephemeral source. It also provides a fixed form for quotation and citation by multiple users. Finally, archival-quality paper remains the most stable medium for long-term preservation.

Transcripts, however, have their limits. Transcribing is both a highly developed skill and a labor-intensive practice. If done by those unskilled with the process, it is often done poorly; if done by professionals, it is expensive. Many oral history programs and projects have a nearly insurmountable backlog of untranscribed interviews; arguably, many never manage to transcribe at all.[2]   Furthermore, even among those programs and projects that do transcribe most of their interviews, the follow-up tasks of interviewer and narrator review, revision, editing, annotation, and cataloguing are frequently left undone or incomplete.[3]  These circumstances point to a larger issue in oral history.  “Doing interviews” is the fun part, the part that captures people’s imaginations and that often yields a satisfying emotional resonance for both narrator and interviewer. As a result, except for professional and well-funded projects, the less charged, often tedious task of transcribing and its accompanying processing is afforded less value and given less attention.

But most importantly, transcribing strips oral history of the oral:  it translates one mode of communication into another, in the process loosing the nuances of embodied expression, paralinguistic cues to meaning, the interpersonal dynamic that occurs when two people talk face-to-face.  Oral historian Alessandro Portelli has observed that the Italian word for “translator” is etymologically very similar to the word for “traitor” – and indeed, there is a way in which we betray a speaker by turning his words into print. Recognizing this, oral historians have generally come to agree that the recorded interview– not the transcript of it– is the primary document or source. Nonetheless, for the advantages noted above and, I suspect, the draw of traditional practice, the practice of transcription remains central to oral history.

Still, many projects, lacking the resources to transcribe but seeking to make recordings accessible, have developed systems for summarizing interview contents, either by specific intervals of time (e.g. five minute segments) or topical segments that were defined by counter number in the days of cassette recorders and are now defined with time codes embedded in digital recordings.  In addition to providing an overall guide to the interview, this method enables a user to pinpoint a relevant passage and then consult the actual recorded interview. One might think of these “half-way measures” as a move towards a greater recognition of the oral in oral history. But perhaps until most recently, they have been understood as a less-satisfactory alternative to transcribing, a point taken up below.

The Current Moment

It is not unreasonable to suggest that oral history is undergoing a paradigm shift, as new technologies redefine all aspects of oral history from initial recording to final access and use. Technology’s utter ubiquity and capacity to provide direct access to the actual recorded interview is causing an explosion in interviewing whose primary object is immediate use in Web sites, museum exhibitions, audio tours, pod casts, and e-publications. Thus, one might argue that oral history practice is moving away from its traditional archival mode to a more engaged presentational mode. Moreover, even those projects that are primarily archival in intent recognize the need for digital recording and access to meet user expectation and extend the reach of the interviews themselves. On the one hand, these developments raise questions about how best to provide digital access to transcripts; on the other, the capacity of new media to present and access audio/video interviews directly, without the intervention of a transcript, suggests that transcribing no longer need be the assumed best practice in all cases. Combining both approaches is the increasingly common practice of linking full text transcripts to the corresponding audio or video files. Perhaps, then, we might think of a continuum of access modes.

Nonetheless, given the continuing hold of the transcript on oral historians’ practice– and the labor and expense involved– the first question generally asked about transcribing in the digital age is the viability of voice recognition software and automatic transcription. This issue is addressed in detail in Doug Oard’s essay; here let it be said that, at this time, no viable system exists to transcribe automatically the often nonstandard, idiomatic, and idiosyncratic speech of oral history narrators with any accuracy. Some oral historians are finding some software programs useful for producing a rough draft of an interview transcript, or more typically, of a desired interview segment that can then be reviewed against the recorded interview and revised for a more accurate rendering. And some argue that a “dirty transcript” can serve as a summary of sorts, leading users to relevant – if poorly rendered – sections of an interview.  Because automatic transcribing tools work best with a single voice speaking slowly and in “standard” English (or any langauge), some oral historians rerecord an interview in their own voice – that is, repeating everything a narrator says – and then have that transcribed automatically.  Also, in what might be a promising innovation for some projects, George Mason University’s Roy Rosenzweig Center for History and New Media has developed Scripto, defined as “a free, open source tool enabling community transcription of docuent and multimedia files.”  It holds potential for spreading the time and tedium of transcribing to multiple individuals, while also ensuring editorial control.

Articles elsewhere in this document, including those by Michael Frisch, Douglas Lambert, Mark Tebeau, and Erin Bell, address more completely issues of content management and user access in a digital environment. Here it may be useful simply to outline a range of possibilities. Some examples and variations are listed at the conclusion of this document.

The approach most in line with traditional practice is to make full transcripts available via a project web site. This predicates the availability of extant transcripts or resources to transcribe, and so is often the option preferred by longstanding oral history projects, who post PDF versions of transcripts. Most typically, these are word-searchable, a boon for researchers seeking a specific reference but not especially helpful if one is looking for topical or thematic references not easily captured in a word or phrase. For this latter, a more complex indexing/data management system needs to be in place.  The efficiency of work searching can also lead to bits of data, all too easily stripped of the rich expressive context characteristic of oral history interviews.

Related but more cautious approaches adopted by some institutions wishing to retain tighter control over transcripts (including those that are restricted in some way) are either to place descriptive summaries online or to develop a more comprehensive, thematic tagging system that refers researchers to specific interviews or transcript pages, but does not link to the interviews themselves. In either case, researchers need to either consult transcripts in situ or order copies.

An increasing number of projects are providing full access to both transcripts and audio and/or video recordings via a project or program web site, with varying degrees of synchronization of text to sound/visuals and varying degrees of cataloguing, annotating, and indexing– and, hence, capacities to search– both media. While the restoration of the oral to these oral histories is perhaps the greatest strength of this approach, developing these varying types and levels of access is also labor-intensive and can require considerable financial outlay to develop and maintain (though open-source tools are also increasingly available).

A more modest approach– one that is less demanding of human and financial resources– is to present short digital audio or video interview excerpts on a web site, sometimes linked to transcripts, sometimes not. These communicate “the flavor” of the oral (and the visual) and are tantalizing in that way, though it’s not clear that they lead anyone to a lengthier, more in-depth encounter with audio or video recordings.

Finally, some projects provide full access to digital audio or video only, bypassing transcribing and its expense altogether. As with those that link transcripts with audio-visual materials, access is facilitated by a variety of systems for summarizing, cataloguing, and indexing; the more detailed and thoughtful the system, the greater the capacity for identifying and connecting “bits” within and across interviews. Insofar as this approach demands attention to the oral and the visual, for some it might be understood as at the cutting edge of oral history practice. It might also inhibit use by some who are uncomfortable in a nonlexical environment, even as– along with projects that provide both transcripts and audio and/or video recordings– it facilitates multiple, creative approaches to use within the digital environment.

The absence of transcripts, it must be noted, obviates some of oral history’s traditional protocols, including narrator review, annotation, and correction of narrator errors, which rely on transcripts. Omitting these steps can diminish the completeness and accuracy of the record. The lack of transcripts can also result in misquotation and inconsistencies in quotations of the same material appearing in different publications. The absence of paper transcripts is especially problematic for long-term preservation, given the instability and obsolescence of digital media.

Here’s a word about ethics. John Neuenschwander’s article Major Legal Issues Facing the Digitization of Oral History elsewhere in this document addresses issues of consent, privacy, and copyright as they may pertain to placing both transcripts and audio and video recordings online. Here it is sufficient to say that projects need to weigh carefully the advantages of web-based access against the danger of possible misuse or even harm to narrators. While a project can institute certain gatekeeper functions in a digital environment, they lack the authority– and the oversight– of an actual archivist working in actual archive.

Choices and Questions 

In a digital environment, oral historians need to consider transcription, alternatives to transcription, and when, in the entire lifespan of a project, a given approach to access may or may not be appropriate or necessary. Projects sometimes adopt a triage approach to transcribing, prioritizing interviews or sections of interviews to transcribe. A similar approach can operate in a digital environment, developing priorities for different modes of access for different collections or interviews.

When making decisions, the following broad considerations, clustered into three broad areas, can serve as a guide. The first two, General Planning and Perceived Value/s, loosely articulate with the five broad categories affecting decision making in the curation of oral history in the digital age outlined by Michael Frisch, et al. in the introductory essay to this section. The last area is issue-specific to issues raised in this chapter. Though these considerations are geared towards new projects, they may also be useful in making decisions about extant collections.

            General planning:

• What are the goals and desired outcomes of the project?
• What are the intended/anticipated audiences and uses?
• What is the size of the collection, both in terms of number and length of interviews?
• What resources and capacity, including financial, human, and technical (both infrastructure and expertise), are available?

            Perceived value/s:

• Is interview content considered of prime importance? Or is the entire gestalt of the interview, its oral and visual dimensions and the process of communicating content, accorded equal value?
• Are the interviews of a quality and significance that long-term preservation and access are desired?
• Is there a desire to respond to an increasing digital imperative?  Is there an allure to the digital environment that sparks creative, new methods of thinking about and using oral history?

             Additional issues:

• What transcribing practices can facilitate incorporation of transcripts (as text and hypertext) into an intended electronic publication?
• If interviews are transcribed, will access to the audio/video files– and the orality of oral history– be provided?  If so, how?
• If interviews are not transcribed, how will long-term preservation be managed? How will traditional protocols that rely on transcripts be handled?

• If interviews can be accessed via both transcripts and audio/video files, how will the one articulate with the other?

Careful consideration of these and other issues addressed elsewhere in this report, in what is inevitably a nested set of considerations, will help those developing and managing oral history interviews/collections decide whether to transcribe, to place existing transcripts online, to develop alternative methods of access, or to develop some create combination of approached. No one factor will likely tip the scales in one direction or another but, cumulatively, they will perhaps give weight to a particular set of decisions.

For example, if an oral history collection is of considerable significance and intended for long-term, archival preservation and the content of the interviews likely of most interest to scholars and other serious researchers, managers might appropriately privilege transcription. If the collection also has broad educational aims and a large budget, they might consider both transcription and digital access. If a project believes that the oral and visual dimensions of a collection of interviews convey important meaning and value, if it seeks to manage a large quantity of data about interview content, if it intends a sophisticated web presence that is interpretive as opposed to simply archival, and/or if multi-media presentation is the goal– and if technical expertise is available – managers might appropriately consider digital-only access. If the interviews are of sufficient quality and funding is available, managers might also consider transcription. If a collection of interviews (or clips from them) are intended as an enhancement for a web site or other media presentation, if they are of rather local significance, or if the goal is student learning, than perhaps digital-only access is appropriate. And so on. These are simply examples, designed to provoke thinking about decisions related to transcription. Specifics will necessarily vary, as will decisions. There is no single “best practice’ for oral history curation in the digital age; it depends on a constellation of factors.

Examples.

            The Virtual Oral/Aural History Archive of California State University Long Beach at www.csulb.edu/voaha; provides full audio access to some 300 interviews related to Southern California’s social history. Interviews are broken into short (usually 2 to 3 minutes) audio segments. Each segment is summarized briefly in print and can be accessed through multiple pathways, including a master topical index, interview series, and individual interviews. The intention was to “enable you, the user, to hear the voice, pitch, and rhythm of the narrations as well as the emotions these convey. You will hear the actual spoken words of oral history narrators, rather than seeing a written version of them in the form of a transcript.”  The interviews have not been transcribed; project directors wanted to direct resources to more interviews, not transcribing.

            USC Shoah Foundation Institute at http://college.usc.edu/vhi/ allows web-based searching of some 52,000 video interviews (comprising about 100,000 hours) via a cataloguing and indexing system of of 1 minute segments of individual interviews based on a controlled vocabulary of 50,000 terms plus 12 million names; searching only identifies segments, it does not link to the actual video segment.  Registration is required.  With the exception of approximately 1100 English langauge interviews that can be viewed at the Visual History Archive On Line, access to the actual interviews is available only at selected locations.  Interviews are not transcribed.  One can only speculate, but the scale of the project and the desire to be at the cutting edge of technology may have figured into the decision not to transcribe and to create a sophisticated system of digital indexing. Perhaps equally important is a moral demand the Shoah Foundation wants to place on users, requiring that they confront an embodied “real” person describing the Holocaust, rather than read words on a page.

            The History Makers Digital On Line Archive at http://www.thehistorymakers.com/digital-archive provides access to both video and transcripts appromimately 300 of its thousands of interviews.  Access incorporates a complex system of catalogues, annotations, indexes by 4 to 6 minute segments, and synchronization of audio and text; access also requires registration at a cost of $30/month.  Some addtional History Maker interviews are transcribed but transcripts are only available in situ; the project admits to a large backlog of untranscribed interviews, the result of both the priority placed on conducting interviews and funding constraints.

            Illinois State Museum Digital Audio-Video Barn at http://avbarn.museum.state.il.us/ provides access to both digital audio/video files and full transcripts of 130 interviews. Audio files have been indexed at the level of segments of several seconds to several minutes duration and can be searched via keywords and controlled vocabulary, which directs users to specific interview clips pertinent to the topic at hand. While one can listen to the interview and read along in the transcript simultaneously, it seems as if there’s no linkage between clips and transcript segments.  In other words, if you clip on an audio segment, you can’t easily link it to the corresponding part of the transcript; however, the page for each lip does include a brief summary of the conent of the clip. The site is intended as an educational resource; I suspect that, and the manageable scale of the project, led to the decision to provide access to both transcripts and audio/video files.

            Densho: The Japanese American Legacy Project at www.densho.org provides full web access to both the original video recording and the transcript of 680 (more than 1300 hours) interviews, (as well as to thousands of pages of print documents and photographs).  Each video recording is divided into segments of several minutes; segments are indexed – and hence searchable – by broad topic, location, and chronology, as well as by keywords. Users can search across interviews (and other archived material) and, unlike the Digital Audio-Video Barn, listen to interview segments and read the transcript of that segment simultaneously. However, like the Digital Audio-Video Barn, Densho has an explicit educational agenda, emphasizing classroom use of its materials. It also promotes a more public agenda of “explor-[ing] principles of democracy and promot[ing] equal justice for all.” These goals have undoubtedly influenced the project’s deep web presence and helped it secure its considerable funding.

            Southern Oral History Program (SOHP) at www.sohp.org contains nearly 5000 interviews, all of which can be keyword searched via comprehensive database by: (1) abstract, (2) transcript, (3) interviewee name, (4) interviewer name, (5) interview number, or (6) subject term. Many, but not all, interviews are transcribed. Digital transcripts and audio are available for the 500 interviews included within the Oral History of the American South program, a pilot project funded by IMLS.  Otherwise, interviews must be accessed on site or via special arrangement.  For the 500 that are available via the web, access to transcript/tape segments for searched terms seems idiosyncratic:  sometimes a search leads to the place within a transcript where the searched term appears and sometimes only to a whole transcript, which then must be searched again to find the appropriate term. Sometimes there is very easy synchronization of time-coded transcripts and audio segments; sometimes this seems unavailable (or the download time is long).  SOHP’s digitization efforts are under development. Their Web site states, “We are working with our partners at the Southern Historical Collection . . . to make our interviews more accessible and available online.”  It seems to be a case of an old and distinguished oral history program playing catch-up in the digital era.