SpokenWeb Podcast

Welcome to the SpokenWeb Podcast, stories about how literature sounds. [Music fades]

My name is Hannah McGregor–

In past episodes, we’ve talked about the work of transcription, the affordances of sound design, and the messy business of wading through archival collections.
In this new episode, producers Natasha D’Amours, Michael MacKenzie, Sarah Freeman, and Xuege Wu take us inside one of the most common kinds of work that research assistants, working on the SpokenWeb project, participate in: timestamping.

For SpokenWeb, timestamping goes beyond marking moments in an audio recording—it transforms sound into something legible, curates literary events, and preserves ephemeral voices. Each timestamp isn’t just a data point; it bridges raw audio with structured metadata. By logging sonic events alongside their timecodes, we create a detailed map of each recording. These timestamps are then transferred to a public-facing platform where users can engage with the archive, clicking on a timestamped event to jump directly to that moment.

This work sits at the intersection of archival practice and digital humanities. But why does it matter? How does timestamping shape the preservation and curation of literary sound? As Phase One of SpokenWeb nears its conclusion, we turn to three scholars who’ve shaped this discussion: Jason Camlot—

Timestamps within digital systems, in that sense, represent one manifestation of a long history of temporal measurement—timekeeping and its mobilization for meaning-making. There’s a lot of scholarship on ancient timekeeping, both in relation to the seasons and in the development of mathematics as a way of producing a calculus of time for various purposes.

So, the conception of time in terms of measurable units and ideas of temporal precision goes back very far in history. Then I started thinking about Hayden White’s work from the 1980s on history, particularly the distinction between annals and historical narrative. I think that’ll be interesting to think about. [Click-whirr]

White introduced medieval annals—lists of notable events organized by year—as alternatives to homogenizing historical narratives. Like timestamps, annals link moments in time to events that occurred within them, albeit on a much larger scale. A set of timestamps may be similar to annals, attributing equal importance to disparate threads of a sonic event.

Timestamps may also emphasize similarities across a sonic event, creating a unified narrative of that event—but more often, timestamps do a little bit of both. [Click-whirr]

These systems evolved over time. And with the emergence of durational media in the late 19th and 20th centuries, we also began to require new ways of counting and organizing time within those media.

And I would just add to Mike’s point that so much of 19th-century print culture is periodical, right? It essentially divides itself into periods—whether it’s dailies, weeklies, or monthlies—each of which is a different measure of things happening within that span of time. So, the explosion of periodical literature in the 19th century is another strong manifestation of what Mike was talking about.

I still think I’m being a bit of a devil’s advocate here, because I’m not giving that much importance or significance to the timestamp in and of itself without the attached—what I would call—the annotation.

One thing I really like about your pushback, though—because when this question about what is a timestamp? was asked to me and I sort of went back in time—is how much I realized that, as we move from ancient history and conceptions of temporality through analog media and its modes of marking time to digital media, a lot of that shift seems to be about the scale at which one is doing annotations, and the increasing degrees of precision—what Wolfgang Ernst calls micro-temporalities—as we move into digital media.

So, I think a table of contents versus a timestamp, or annals that are annual versus a timestamp of a 3-minute recording with 50 annotations—those are questions of temporal scale that I think are really interesting to think about in relation to timestamps. But personally, I see them as working on a continuum.

In creating these timestamps, we’re not just listening to and analyzing the recordings—we’re transforming them into written data, effectively turning sound into another form of media. [Instrumental music plays and fades]

I have a list of examples here—everything from YouTube’s heat map algorithm, which shows how popular a video is at any given point along its time bar, to EKG (electrocardiogram) monitors that mark the rhythm of a heartbeat and any interruptions in it with beeps and squiggles. There are also the SETI Institute’s protocols for parsing sound from outer space to identify potential alien messages, and seismic monitors or lie detectors that give off sudden bursts of timestamping activity.

These are all examples with very specific goals and motives. Given the wide variety of things a timestamp can do, what brought you to start thinking critically about it?

And I think that does matter. I’m not sure if this is an exact response to your question, but it is something that feels closely related.

And that’s predicated on those great lines toward the beginning of Burnt Norton:
“If all time is eternally present, all time is unredeemable.
What might have been is an abstraction,
Remaining a perpetual possibility only in a world of speculation.”

This is not only a deep philosophical problem—it’s also a very complex practical one.

Then, analog tape recorders began to include time counters. Not right away—in the 1930s, they didn’t have them—but eventually, they did, as it became more important to navigate and manipulate the recordings. So, there’s a strong sense of timestamping as a form of power over time, embedded in whatever media format one is engaging with.

And I guess the point I was going to make about digital technologies, as opposed to analog ones, is that digitized and digital media represent a fuller realization of the timestamp concept in media form.

But where I want to go is actually to my last thesis, in relation to what Tanya was saying, which is that timestamps also have the potential to express. That’s the thesis—it’s an ambiguous or cryptic one, and I’m going to unpack it. I think it has something to do with what Tanya was pointing out about the subjectivity of timestamps. What I mean by that is the possibility of an aesthetic or a poetics of the timestamp in itself. And I love the idea that timestamps, when disconnected from their annotations or from their historical or temporal events, can go awry in a variety of ways.

They are primarily mechanisms of precise control over time—for a variety of purposes, to make arguments, to serve different ends. But the example I have in mind is actually a chapter in Jordan Abel’s book Injun. There are a couple of sections where he offers a timestamped transcript from a lecture he gave at the TransCanada conference. And if you read that text as a poem, I think what he’s performing there is exactly what Tanya was talking about—he’s really attempting, in this piece of writing (which consists of a lot of timestamps), to explore that poetics.

I’ve actually read it out loud, and when I quote from it, a long section might be something like: 15:38:53–15:38:54, right? That’s just the passage of a second. But it carries so much weight in how it’s presented.

And the timestamps, as they appear on that page, I would argue, represent a kind of poetics of the timestamp—one that aims to show how time is always subjectively relative. It depends a great deal on how one is experiencing a moment, a second, or whatever unit of time it may be.

And that is where the expressivity of the timestamp might lie: in the fact that we carry certain values and meanings into the production of timestamps and annotations. Those values and meanings shape the circulation and reception of the objects we subject to those practices.

What I’ve realized as we go through this project is that so much of our work focuses on making sure that the comments—the annotations—are timestamped, that they’re actually linked up to the AV, right?

I’d love to hear, or for all of us maybe to reflect on: what’s the difference in value between a timestamped annotation (which we almost use as one word sometimes) and just an annotation? Like, what if we didn’t timestamp any of our annotations for that event, and we just had this running text? The reader—or the listener—would then have to figure out what’s referring to what. You know?

For example, I was timestamping an interview the other day—mostly just doing transcripts—but I had a range for a particular moment.

But I kept having to remind myself: Oh, it’s not that exact number. The number kept making me think it had to be very precise, when in fact, it could be anywhere within that space. I’m always reminded of Gertrude Stein, right? As human beings, when we’re reading things or trying to understand them, we tend to put them into structures—and we rely on those structures to make sense of what’s happening. [Click-whirr]

So I think one’s intention, if you’re being inexact with timestamping, can be generative, but it can also be confusing. Because when people see timestamps, they tend to assume: this is what’s happening at that exact moment. Or if they see something in sequential order, they assume: this is what’s being talked about first, then this, then the next thing. So you’re still imposing a time-based order, I think.

My question is about what timestamping does in this context: on one hand, it provides accessibility—it allows you to directly find what you need. But at the same time, some artists or performers might want to preserve a certain sense of enigma or opacity in their work. So timestamping can also risk disrupting that intentional ambiguity.

What I’d like to do now is briefly check out a clip from Jackson Mac Low. I’ll share my screen so we can listen to it together—it’s one of the pieces we’ve timestamped. I’ll share my question with you right after we take a look. [Click-whirr]

Warning: you might want to turn your volume down a little bit. [Click-whirr]

“…About—honestly, that… Alison Knowles, Carol… Dickens’ farm… Rose Jackson, Scott… take… Gotti? Haha… there’s something… dark—God? Song? Sing? Maybe that. Slide… please… I—I gotta eat… thank God now… guaranteed… body spa? Please…”

Note: This segment contains heavily distorted and overlapping voices. The transcript reflects best approximations.

What matters from a practical and theoretical perspective when you encounter a difficult piece like this? And specifically, what I mean by that is: when a piece appears to want to remain enigmatic or appear opaque—not necessarily easy to interpret—what happens when we include, inside our timestamps, descriptive words like inaudible or unintelligible? What does that do? How might that affect a reader or listener when the thing they’re listening to might have the express purpose of being noise, or of being inaudible on purpose?

And so, have you thought about the intersection of those things? And also, more generally, have you run into difficult pieces like this that have really made you think about your practice?

First of all, how is this timestamped and annotated on the site where it appears? Essentially, there’s a timestamp before the performance of the Fifth Gatha, where Mac Low explains what they’re about to do. The transcript for that explanation is actually connected to a timestamp—so that’s at 01:06:11, where he says something like, I’ll do a piece called Fifth Gatha, and then he goes on to explain what’s about to happen.

Then you have an annotation at 01:09:35, and it just says performs Fifth Gatha, right? And there’s nothing else there. But I’m looking at this, and I’m actually quite grateful that there are no annotations for the piece itself.

So, following that logic: at 01:09:35 it says performs Fifth Gatha, and then at 01:26:20 he starts talking again. So all we have is the time span during which that piece unfolded, and that’s the only annotation that exists in relation to that timestamp.

And that, probably, would be more interesting to actually include—because we learned quite a bit from conducting an oral history with two of the technicians who were there assisting Mac Low. They told us about the setup in the room: there were five reel-to-reel tape recorders that had been pieced together. I even saw the tech list that Mac Low had submitted, and that made all the technicians laugh when they were together in the room.

So, essentially, to timestamp this would be extremely complex, because it’s an event unfolding in multiple places at the same time—in New York, and wherever else he was—as well as in the present moment. And then that tape would be brought to a future performance, and so on.

So, in this case, I think some explanation of the context of what we’re listening to would probably be even more useful than attempting a transcriptive annotation.

But the other thing they do—and you could say this about a lot of Jackson’s work—is tied to its recursive nature. He’s working with computer technologies and other algorithmic systems to produce his work in the first place, and he’s drawing from discursive corpora of other authors. Gertrude Stein came up earlier in our conversation, and the Stein poems would be a really good example of that—but there are many others as well.

One of the things he’s doing is also challenging our notions of time. The linearity of the performance piece, its monologic subjectivity, and its stability—these are all things that he’s disrupting in a piece like The Gathas. And these, in turn, are the very presuppositions we often bring to the practice of timestamping.

In other words, in a very fundamental way, Jackson’s art forms challenge the basic suppositions we hold when we think about duration, performance, event, and subject. All of those things come under scrutiny—and all of them are disrupted in one way or another by his work. That’s why the limit case you shared with us is such a wonderful way to get us to rethink the fundamental ideas we bring to the activity of marking durational media.

What goes into deciding on a standardized vocabulary or a reduced set of words that we’ll use when annotating or timestamping? What’s the thought process behind determining that perform is going to be the word we use, and not read?

So, let’s say your annotations are transcripts, and you want to add speaker names as tags so you can see everywhere that a particular person has spoken across a project—because all of their annotations have been tagged with their name. The same could be true if, say, you were working within a particular theoretical perspective, and you had terms that might be adopted in the context of that theoretical positioning—you might want to tag a transcript based on those.

I guess the short answer is: it depends on your theoretical perspective. It depends on what it is you’re trying to pull out or identify as significant in a particular recording. You don’t have to do transcripts at all, really. A lot of people tend to do them because they create better access for listeners or users. But I think the larger point is: it depends. It depends on what you’re trying to mark as significant.

So, the very first thing we did with SpokenWeb in the first year—the first task force we struck—was our Metadata Task Force. That was in anticipation of having hundreds of hours of audio that we were planning to describe. We were imagining doing things with that material which, without some kind of controlled vocabulary and grammar for our schema, would have been much more difficult—especially if we wanted the recordings to teach us things about each other when placed in relation to one another.

When I say it allows us to do things we couldn’t do otherwise, I mean we were looking at other existing schemas—their vocabularies and grammars—to ensure that, if we came across other collections that already had some metadata, we’d be able to bring it in, or ingest their metadata, more easily than if we didn’t have a shared structure.

So we wanted to make sure our terms were interoperable with other standardized languages or grammars. But we also wanted, as much as possible, to ensure that we were tagging, annotating, and describing individual recordings in a way that categorized them meaningfully.

For example, if we’re interested in the distinction between performing versus reading, perhaps that becomes a useful search term for locating more experimental kinds of performance within the contents fields themselves—which one could do, if one wanted to, because that vocabulary was adhered to.

So the vocabularies can be useful, but they are all forms of abstraction—necessary forms of abstraction if you want to do this kind of more distant searching or reading of the contents. They’re useful in their own ways, but abstractions also do violence to the particulars of the events themselves. They’re not always accurate, or fully capable of capturing what’s actually unfolding—but they’re useful up to a point, depending on the goals one might have when working with these materials as data.

And that’s the other thing that’s really happening in a lot of this work: we’re converting qualitative humanities content—speech, performance, sound of various kinds—into different forms of data, by using these controlled vocabularies.

So I would say, for example, that the difference between performance and reading—if we designate the activity of a particular author on stage who’s been recorded as reading a work—for me, it connotes the idea that the literary audio performance is secondary to a print version of the text. That you are reading—that it’s derivative, right? That it follows from something else.

Whereas the notion of performance, for me, carries the weight of the performative. In other words, it’s a constitutive medium of its own. A performance produces a text that is not secondary or derivative to the print version—it’s its own beast.

And so, the language that we use to describe the work we’re doing in this regard can carry some pretty heavy connotations. But that only comes under scrutiny—only becomes a matter of discussion—when we sit down and deliberate: how do we describe these things? And what values does it carry to do it one way or another?

That’s the part that’s most exciting for me. Getting to the practicality of it—being able to produce constrained vocabularies to increase searchability, interoperability between different systems, and all of the things that go along with that—is really crucial.
[Upbeat instrumental music starts playing] But the fun part is the conversation that gets you there.

We also explored what it means to timestamp a recording that resists singular meaning—such as Jackson Mac Low’s Fifth Gatha. Most of all, we learned that timestamping is far from a settled field. There is no objective or universal timestamp. Each one results from a series of subjective, contingent, and political decisions made by the time stamper.

Timestamping is an invitation to intervene, to interpret, and—most of all—to create.

Further thanks to Michael O’Driscoll, Sean Lowe, and the SpokenWeb Podcast production team for their support in creating this episode.

Technical support was provided by the Digital Scholarship Centre at the University of Alberta.

This podcast was produced by Natasha D’Amours, Michael MacKenzie, Xuege Wu, and me, Sarah Freeman.

This month’s episode was produced by Natasha D’Amours, Michael McKenzie, Sarah Freeman, and Xuege Wu, and features the voices and smart ideas of Jason Camlot, Tanya Clement, and Mike O’Driscoll.

The SpokenWeb Podcast Team includes supervising producer Maia Harris, sound designer TJ McPherson, transcriber Yara Ajeeb, and co-hosts Katherine McLeod and me, Hannah McGregor.

To find out more about SpokenWeb, visit spokenweb.ca and subscribe to the SpokenWeb Podcast on Apple Podcasts, Spotify, or wherever you like to listen. If you love us, let us know — rate us and leave a comment on Apple Podcasts, or say hi on our social media.

Plus, check out our socials for info on upcoming listening parties and more. For now, thanks for listening.