Interview by Joseph Wilson

https://www.dukeupress.edu/how-machines-came-to-speak

Joseph Wilson: The combination of legal discourse analysis and technology studies in this book is a fascinating mix. How Machines Came to Speak examines the legal definition of speech (as part of the concept of free speech) and how it morphed over the years as new technologies changed the way Americans communicated. How did you first come up with this topic? Did you come at it from a legal angle or from an STS angle?

Jennifer Petersen: I came to the book from the history of media technologies (informed by STS) and communication history, and a set of interests—or perhaps a puzzlement—about the status and authority of law. In particular, I was interested in free speech law as a site where debates about rationality and emotion, ideas and action (for example, when does speech become a threat or harm; how is money considered speech) take place, with social and political consequences.

In graduate school, I had written a paper looking at how judges were determining whether and how the First Amendment applied to new media – at the time, software, URLs and hyperlinks. I was fascinated by the way the judges were theorizing communication in these cases. The paper left me with a lot of unanswered questions, so when I finished my first book, these nagging questions returned. When I thought about the scholarship I would need to read to work through these questions, I was all in.

Joseph Wilson: You write that in the 20^th century written speech (newspapers, books, pamphlets) was often privileged over other modes of ‘speech’ (film, radio, flag-burning) because of its association with rationality, civilization, masculinity and the ‘world of the mind’. With the rise of new visual/aural media like podcasts, TikTok videos, or even video games, do you think these Enlightement-era binaries are starting to break down? Is the written word still considered the archetypical medium of speech?

Jennifer Petersen: I do think the binaries that were in place at the beginning of the century have eroded in some ways. This is perhaps most evident in examples of expressive conduct. In the early 20^th century, physical activities and non-verbal expression were clearly physical conduct. By the mid- to late-20^th century, it was possible to consider many forms of physical conduct – from flag burning to sit-ins and other forms of silent protest, like Colin Kaepernick’s silent protest –as expressive, even when they do not clearly translate into words, or a verbal message. And I do think that pictorial and visual media will continue to erode this distinction.

However, in many ways the written word is very much the archetype of speech in legal discussions of the First Amendment. In determining whether new or controversial cases are speech, judges and justices often turn to words and writing. For example, when arguing that naked dancing was a form of speech (albeit a putatively lesser one), one of the justices anchored his reasoning in the assertion that dancing is like writing with the body. Given this, it is no coincidence that the current challenge to anti-discrimination laws in Colorado is being brought for a business that is proposing to build websites and wants to exclude same-sex customers. If you remember, there was a similar case a few years ago involving a baker who refused to make wedding cakes for same-sex customers. Cakes were less evidently expressive (though there was a lot of discussion of the fact that bakers use icing to write messages on cakes!) than websites are. I think the current case has probably been engineered to appeal explicitly to the Court’s bias toward writing.

Joseph Wilson: You point out that during the radio era it was legally understood that the speech one heard on the radio did not necessarily represent the views of the radio announcer, but was instead the work of a team of people: station owners, distributors, advertisers, announcers, and so on. This reminds me of Goffman’s distinction between animator, author, and principal. Do you think there is a danger that this kind of distribution of responsibility can be used as a defense against the consequences of truly offensive or dangerous speech?

Jennifer Petersen: The question of distributed speech is one that really interests me. Your last question asked about erosion of older binaries. One of the questions that came out of this book for me was about how the experience of algorithmic media – from Tik Tok to bots—might erode some of the dualisms around speech and subjectivity or personhood. I wondered whether our current experiences might re-shape our understandings of speech, so that in the future the idea of an autonomous speaking subject is less central. One of the ways in which we currently experience algorithmic mediation is as a change in the structure of speech or authorship. In many such contexts, either 1.) what we say is understood not to be entirely of our own making; or 2.) complex systems designed and prompted by people produce outcomes that are not direct expressions of their thoughts, beliefs, knowledge or intention (for example, ChatGPT). Put differently, in many algorithmically mediated contexts, we understand ourselves to be speaking with or through complex sociotechnical systems; the chains of intentionality and agency behind any expression are distributed rather than direct. I am very curious about how this experience will shape the next generations of judges’ and justices’ assumptions about what speech is and who (or what) can speak.

This has some attractive political and ethical implications for how we might understand ourselves and constitute community. In some ways, when we decenter ourselves and our intentions, our sphere of responsibility may expand. For example, if what makes my words racist or damaging is not my thoughts or intentions but their effects in the world, my responsibility for those words is expanded. I think that many of my students are coming to express this sense of ethics and accountability.

Also, it’s the conception of speech as the direct expression of an individual’s ideas and beliefs does not bring with it a lot of accountability. We are held legally accountable only for expression that harms reputation (defamation), that incites or is very likely to incite criminal actions, and that poses a “true threat” (such as, burning a cross on someone’s lawn as a treat of racial or religious violence).

Joseph Wilson: There has also been a lot of criticism recently (a so-called techlash) about data collection, privacy policies, and the ability for social media companies to manipulate what people understand to be the truth, much of which is currently protected under freedom of speech laws. This is a version of what you call a ‘posthuman conception of speech’. Are we due for a (legal) correction to this broad understanding of speech, one that rehumanizes speech as the realm of only humans? 

Jennifer Petersen: When I talk about the posthuman conception of speech, I’m looking at instances where lawyers and judges reason about speech without reference or regard to speakers. When the courts say that the flow of information is speech, or that particular artifacts are speech, and attempt to analyze speech without reference to social dynamics or agents involved, they are engaging in what I call a posthuman approach to speech. It is one that does not require persons.

Interestingly, in the examples you mention in this question, tech companies have been drawing on classic liberal humanist arguments. For instance, Google’s search results have been classified as speech because they represent the ideas and opinions of Google employees and/or the company. In practice, the company becomes endowed with protections crafted for persons – in particular, political persons, in effect granting these companies more political and legal capacities. These companies, in sum, use arguments about civil liberties in order to gain market and legal leverage.

Yet, I don’t think saying that speech rights are only ever for natural persons, or individuals—or rehumanizing speech –will answer these problems. There are reasons to think that organizations, even corporations, might play some role in our public sphere, and expressive and political landscape. In fact, part of the problem I see is that speech is understood too narrowly in terms of a classic, liberal vision of persons. It is this close articulation of speech to a form of agency associated with natural persons that allows corporations to game the system.

I think, rather, that a less dichotomous approach to speech and agency may be in order. The examples of distributed, algorithmically-mediated speech noted above may be able to help us think more productively about expression. They may help us dis-articulate speech from the minds of discrete subjects, and help us think through expression as having a variable relation to things like the opinions, intentionality, conscience, and will of persons.

We need, I think, a broader and variegated conception of what it means to speak today – a broader ecology of means and forms of speaking as well as of different types of speakers. And we need ways of adapting the law to these different forms rather than an all or nothing logic. I think this would be hard. It would bring its own problems. And it, too, would be open to opportunism. But the opportunities would be fewer without the current winner takes all stakes.

Joseph Wilson: In recent years, the concept of free speech has drifted from being a supposedly typical left-wing issue favored by social democrats, towards the right, where it is often used as a defence against accusations of political incorrectness. How does this political reframing fit into your argument that speech is a ‘historically contingent’ concept?

Jennifer Petersen: I think that when we think we know what speech is – and that this is always what it has been – it is easy to think that what is going on right now is about “the left” abandoning speech or “the right” taking up the rhetoric if not the actual substance of free speech (the right, after all, is busy banning books and dictating curricula, both textbook examples of censorship).

To some extent, I think that this framing can act as something of a red herring, especially for the types of legal opportunism outlined above, in which companies use the First Amendment to avoid regulation as well as to enhance their standing as persons, or subjects before the law. In many of the key legal battles today, the expression being fought over does not look like a dissenter on a soap box or sound truck. Rather, these cases often center on categorizing economic transactions, services, and products as expression in order to gain favorable regulatory outcomes. The political and social stakes of cases like these are substantively different from the debates over whether sit ins are a form of expression—yet, when courts support these claims, they often do so in the names of political freedom and civil liberties.

I think that a large part of what enables these opportunistic uses of the First Amendment is the way speech as a legal category has transformed over time without serious attention and reflection. When we engage in discussions of whether free speech is a left or right issue, I think we lose sight of the pressing issues around technology and opportunism outlined above. To me, this is an essential political and theoretical misstep.  Looking instead at the underlying definitions of speech can help to highlight what exactly is at the heart of free speech disputes. It can also provide new arguments and insights for charting a future as technologies again change what it means to speak.