This article first appeared online in the Atlantic on January 16, 2018 and quickly became one of the most read articles in the magazine for a week, surpassing even an argument for why Donald Trump should be impeached. Eventually it would gather over a million online views. Since it hides now behind a paywall on the Atlantic's website, I'm putting it up here. I have not corrected it against the current online version.
It forms the basis for my book, Bye Bye I Love You, which is the product of this initial curiosity. You can find the MIT Press description of the book here.
Mort Felix believed that his name, read as two Latin words, meant “happy death.” When he was sick with the flu, he used to jokingly remind his wife, Susan, that he wanted Beethoven’s “Ode to Joy” played at his deathbed. But when his life’s end arrived at the age of 77, he lay in his study in his Berkeley, California home, his body besieged by cancer and consciousness cradled in morphine, uninterested in music and refusing food as he dwindled away over three weeks in 2012. “Enough,” he told Susan. “Thank you, and I love you, and enough.” When she came downstairs the next morning, she found Felix dead.
During those three weeks, Felix had talked. He was a clinical psychologist who had also spent a lifetime writing poetry, and though his speech often didn’t make sense, it seemed to draw from his attention to language. “There’s so much so in sorrow,” he said at one point. “Let me down from here,” he said at another. “I’ve lost my modality.” To the surprise of his family members, the lifelong atheist also began hallucinating angels and complaining about the crowded room—even though no one was there.
Felix’s 53-year-old daughter, Lisa Smartt kept track of his utterances, writing them down as she sat at his bedside in those final days. Smartt majored in linguistics at UC Berkeley in the 1980s and built a career teaching adults to read and write. Transcribing Felix’s ramblings was a sort of coping mechanism for her, she says. Something of a poet herself (as a child, she sold poems, three for a penny, like other children sold lemonade) she appreciated his unmoored syntax and surreal imagery. She also wondered if her notes had any scientific value, and eventually she wrote a book, Words on the Threshold, published in early 2017, that analyzed the linguistic patterns in her father’s final sentences.
Despite the book’s faults, it’s unique—it’s the only published work I could find when I tried to satisfy my curiosity about how people really talk when they die. Since the 17th century, anthologies of final utterances by famous personages (and condemned criminals) have been published in numerous languages, but such “last words,” eloquent and perfectly enunciated, can’t literally be how people’s final utterances sound or what their linguistic abilities are like. “Authentic or not? Apocryphal or not?” inquired Harvard German scholar Karl Guthke, in his long, pleasantly droll essayistic tour, Last Words, through Western culture’s fascination with them. One point that Guthke makes repeatedly is that they’re not “historical facts of documentary status but artifacts,” which makes them unsuitable as descriptions of how the dying person’s ability to interact, verbally or nonverbally, changes.
Despite this unsuitability, very few have collected the actual linguistic patterns present in the speech of the dying. To find any sort of rigor, one has to go back to 1921, to the work of anthropologist Arthur MacDonald. “This seems to be the only attempt to evaluate last words by quantifying them,” Karl Guthke reflected, “and the results are curious.” In order to assess people’s “mental condition just before death,” MacDonald mined last word anthologies, the only linguistic corpus then available, dividing people into ten occupational categories (statesman, philosophers, poets, etc.), coding their last words as sarcastic, jocose, contented, and so forth, then counting the words for each emotion expression. MacDonald found that military men had the “relatively highest number of requests, directions or admonitions,” while philosophers (which included mathematicians and educators) had the most “questions, answers and exclamations.” The religious and royalty used the most words to express contentment or discontentment, while the artists and scientists used the fewest. Curious, indeed.
§§§
As the way we die has changed, ideas about language at the end of life have too, and so contemporary approaches take a more emotional, relational approach. Books like Final Gifts, published in 1992 by hospice nurses Maggie Callanan and Patricia Kelley, and Final Conversations, by Texas State University communications scholar Maureen Keeley, aim to sharpen the skills of the living for having important, meaningful conversations with the dying. (A prolific scholar, Keeley has published academic papers on the same topic, such as this.) Previous centuries’ focus on last words has ceded space to the contemporary focus on last conversations, even interactions that don’t involve words. “As the person gets weaker and sleeper, communication with others often becomes more subtle,” Callanan and Kelley write. “Even when people are too weak to speak, or have lost consciousness, they can hear; hearing is the last sense to fade.”
I spoke to Maureen Keeley shortly after the death of George H. W. Bush, whose last words (“I love you too,” he reportedly told his son, George W. Bush) were widely reported in the media, but she said they should properly be seen in the context of a conversation (“I love you,” the son had said first) as well as all the prior conversations with family members leading up to that point.
At the end of life, the topics may not change, Keeley says, but the majority of interactions will be nonverbal as a person’s body shuts down and they lack the physical strength, even the lung capacity, for long utterances. “People will whisper, and they’ll be brief, single words—that’s all they have energy for,” Keeley said. She also noted that family members often take advantage of a patient’s comatose state to speak their piece, taking advantage of the dying person’s inability to interrupt or object and also assuming they’re still able to hear.
Many people die in such silence, particularly if they have advanced dementia or Alzheimer’s that robbed them of language years before. No one appears to have quantified even this simple characteristic — Final Gifts and Final Conversations both give the impression that dying is unavoidably chatty. (People with Alzheimer’s were excluded from Keeley’s study, along with suicides and accidental deaths, which required that both people had to know a death would soon occur.) Other anecdotes center on the banality of the vernacular. From a doctor I heard that people often say, “oh fuck oh fuck.” Often it’s names of wives, husbands, sons. “A nurse from the hospice told me that the last words of dying men often resembled each other,” wrote German writer Hajo Schumacher in a September essay in Spiegel. “Almost everyone is calling for ‘Mommy’ or ‘Mama’ with the last breath.”
It’s still the interactions that fascinate me, partly because their subtle interpersonal textures are lost when they’re written down. A linguist friend of mine, sitting with his dying grandmother, spoke her name. Her eyes opened, she looked at him, and died. What that plain description omits is how he pauses when he describes the sequence, and how he makes his eyes mimic hers.
Whether last words or last interactions, there’s no systematic description of the basics of either in the scientific literature. The most detail exists about language in delirium, which involves a loss of consciousness, the inability to find words or understand what people say, restlessness, and a withdrawal from social interaction. Delirium strikes people of all ages after surgery and is also common at the end of life, a frequent sign of dehydration and oversedation. Delirium is so frequent, wrote a New Zealand psychiatrist, Sandy McLeod, “it may even be regarded as exceptional for patients to remain mentally clear throughout the final stages of malignant illness.” In the late 1990s, McLeod devised a test of delirium among terminal patients: ask them to write down their names and addresses. In delirium, McLeod wrote, “writing is the most significantly impaired.” In fact, one study in an Alberta hospital found that 88% of patients with advanced cancer experienced delirium in the last 48 hours of their lives. In that study, patients’ delirium significantly impaired communication at a critical moment: when loved ones were trying to communicate with them. [NB: there’s some literature on speech during delirium that I could discuss, but becuase it wasn’t strictly among dying people but among post-operative elderly people, I’m leaving it out. But it’s interesting because some people recover from delirium and can talk about the dislocating powerlessness of the experience, from which we might infer something about the experience of dying].
I’ve puzzled over the absence of more detailed information along these lines, given that we have such a rich picture of the beginnings of language, thanks to decades of scientific research with children, infants, and even babies in the womb. But if you wanted to know how and when the ability to construct sentences and understand language shuts down in the dying, there’s next to nothing to look up, only firsthand knowledge gained painfully. I recognize the cultural reasons for this asymmetry — babies are appealing, it’s a badge of modern parenting to offer up offspring to science — but it still strikes me as out of balance with the reality that every first word will someday be paired with final conversations, last words, squeezes of the hand, and fluttering breath. It’s as if the culture of science must still, despite its advertised relentless curiosity, avert its eyes at the lifespan’s other end.
Lisa Smartt, to her credit, didn’t avert her attention. After her father died, she was left with endless questions about what she heard him say in his last days of life. She approached graduate schools, proposing to study last words, but was rebuffed and then began interviewing family members and medical staff on her own. That led to a collaboration with Raymond Moody, Jr., the Virginia-born psychiatrist best known for introducing the world to “near death experiences” in a 1975 best-selling book, Life after Life. He has also been interested in what he calls “peri-mortal nonsense.” When she couldn’t get funding for a mainstream academic study, she published Words on the Threshold, based on her father’s utterances as well as ones she’d collected via a website, www.finalwordsproject.org.
One pattern she calls “prepositional nonsense.” When Felix used pronouns like “it” and “this,” they didn’t clearly refer to anything. One time he said, “I want to pull these down to earth somehow…I really don’t know…no more earth binding.” What did “these” refer to? Also, his prepositions hinted at how his sense of his body in space seemed to be shifting. “I got to go down there. I have to go down,” he said, even though there was nothing below him.
He also repeated words and phrases, often ones that made no sense. “The green dimension! The green dimension!” (Repetition is common in the speech of people with dementia and also those who are delirious.) Smartt found that repetitions often expressed themes such as gratitude and resistance to death. But there were also unexpected motifs, like circles, numbers, and motion. “I’ve got to get off, get off! Off of this life. I’m dying. I’m dying,” Felix had said.
Smartt was most surprised by narratives in her father’s speech that seemed to unfold, piecemeal, over days. Early on, he talked about a train stuck at a station, then days later referred to the repaired train, and then weeks later, how the train was moving northward.
“If you just walk through the room and you heard your loved one talk about, ‘Oh, there's a boxing champion standing by my bed,’ that just sounds like some kind of hallucination,” Smartt said. “But if you see over time that that person has been talking about the boxing champion and having him wearing that, or doing this, you think, ‘Wow, there's this narrative going on.’” She imagined that if medical staff tracked these storylines, they could be clinically useful, particularly as the storylines moved toward some sort of resolution, which might reflect a person’s sense of an impending end.
Early on, Smartt called on Berkeley linguist George Lakoff, whose classes she’d taken 30 years earlier. Back in the 1980s, Lakoff co-wrote Metaphors We Live By with philosopher Mark Johnson, which outlined the deep conceptual and cognitive basis of metaphors, and he encouraged Smartt to look closely at the metaphors in her collection. In Final Gifts, hospice nurses Callanan and Kelley note that “the dying often use the metaphor of travel to alert those around them that it is time for them to die.” They quote a 17-year-old, dying of cancer, distraught because she can’t find the map. “If I could find the map, I could go home! Where’s the map? I want to go home!” Smartt noted the prevalence of such journey metaphors as well (though she goes further and says that dying people get more metaphorical in general, not just in describing their dying experiences). This attention to travel metaphors makes them seem universal, but anthropologists who study dying in other cultures told me that journey metaphors aren’t actually as prevalent elsewhere. This seems to be an important cultural fact to include in trainings for hospice and palliative care nurses who don’t share the same cultural frame as their patients.
Such training is only one practical application of the basic descriptions of language at the end of life that we don’t have eyt. Experts told me that a more detailed roadmap of changes could help counter people’s fear of death and provide them with some sense of control. It could also offer insight into how to communicate better with the dying. One early academic collaborator of Lisa Smartt’s, Erica Goldblatt-Hyatt, a social work professor at Rutgers University, was drawn to studying the language of the dying in order to create a translation app — when a delusional patient talks about a train, family members could understand what it means. Another collaborator, a nursing researcher from the University of Pennsylvania, Alexandra Hanlon, also saw a clear application in understanding the internal psychological state of the dying person, particularly in regards to clarity and peace. [NB: I have numerous interviews with speech language pathologists about their relevance in palliative care settings, but where they don’t often get used because of the way hospice is funded.]
The broader relevance for understanding the trajectory of changes of language at the end of life lies in how we die today. We are more likely to die in strikingly different ways than our counterparts a century ago, who tended to die quickly and abruptly from accidents, infections, or childbirth. Of course, we can’t know the precise moment of our deaths, but thanks to medical advances and preventative care, we do know that most of us will die either from some sort of cancer, some sort of organ disease (foremost being cardiovascular disease), or simply advanced age. We also know that we’re increasingly likely to die in hospitals or nursing homes, overseen by teams of medical experts who will want us to participate in decisions about our care for as long as possible, as much for us (because it improves our chances of dying the way we want) and our loved ones (because it will ease their bereavement) as for the professionals themselves (because of the emphasis their training puts on communicating well). But this seems possible only if we know more about how language ends.
Understanding language at the end of life also has theoretical implications for the evolutionary origins of language. Linguistic anthropologist Steve Levinson has posited that humans possess an “interaction engine,” a drive to interact that is so fundamental to human language and cognition that, as Levinson wrote in 2006, “when language is lost, interaction doesn’t disappear.” This would seem to describe many anecdotes about deathbed interactions, where dying people are still able to respond non-verbally, as with a hand squeeze. Such turn-taking is such a fundamental part of the architecture of interaction that it has been noted across animal species, who go back and forth when they grunt, chirp, or coo. It’s also the first social thing that the newest of newborn babies learns to do, whether it’s with eye gaze or gurgles. So the beautiful symmetry is that it may work the same way at the other end of life.
But studying language and interaction at the end of life remains a challenge, because of cultural taboos about death and ethical concerns about having scientists’ at a dying person’s bedside. Experts also pointed out to me that each death is unique, which presents a variability that science has difficulty grappling with.
Within health care, the overwhelming interest is in improving the abilities of the living, mainly for doctors, and that’s where the funding has gone. “I think that work that is more squarely focused on describing communication patterns and behaviors are much harder to get funded because we are in the business of supporting research that reduces suffering from cancer,” said Sylvia Chou, a program director in the Behavioral Research Program at the National Cancer Institute of the National Institutes of Health who oversees funding on patient-doctor communication at the end of life.
Despite the faults of Smartt’s book (it’s not scientific, for one thing, and it’s too interested in the afterlife), it takes a brave step toward simply noticing what people are saying, and toward focusing less on what they mean as much as how they do it. It takes on greater significance in an analogy with child language studies: that field didn’t take off until natural historians of the 19th century, most notably Charles Darwin, began writing down things their childen said and did. (In 1877, Darwin published a biographical sketch about his son, William, noting his first word: “mum.”) Such “diary studies,” as they were called, eventually seeded a more systematic approach.
Early child language research has itself moved away from studying first words, and it’s good to question the cultural fascination with last words, no matter who says them. “Last words” are the cornerstone of a romantic vision of death that obstructs the realistic planning that we should do before illness or tragedy strikes. “The process of dying is still very profound, but it’s a very different of profoundness,” said Bob Parker, Chief Clinical Officer of the Hospice and Palliative Nurses Association. “But last words — it doesn’t happen like the movies. That’s not how patients die.”
