I used to be a normal psycholinguistics graduate student. I wanted to study how the mind parses improbable metaphors, unintelligible accents, and quirky syntax. Sexy things. Things that would play out well at parties. I imagined myself magnanimously explaining how sentences like “The bartender served the bourbon fell down the stairs” were truly grammatical. I imagined myself dropping newspaper headlines like “Iraqi Head Seeks Arms” into conversations with beautiful people. I would defend Internet chatroom slang on local radio. I would exchange holiday cards with Steven Pinker.
But something has happened. I am in my third year of graduate school, and I have fallen in love. I have fallen for pronouns. It’s hard to shut me up about them.
Before we can understand a word, we first have to retrieve its meaning from memory. Most of the time, this happens quickly—so quickly we call it automatic—but sometimes it doesn’t, sometimes the word is good and dusty. Say my lab manager says, “Sally saw Rosemary Clooney on the bus today!” I’ll quickly retrieve some meaningful representation for Sally and saw and bus and today, but Rosemary Clooney might throw me. I retrieve her a bit at a time, one piece leading to another. She sounds familiar. She’s a singer, right? Isn’t she George Clooney’s aunt? Then I remember Rosemary Clooney, George’s aunt, died several years ago. “No way,” I reply, that bit of idiomatic speech rolling off my tongue effortlessly. “She’s been dead for years.”
And here we’ve come to a pronoun. My lab manager delves into memory for representations of dead and years, and finds them, no problem. Delving for she, my lab manager comes back up with a representation of Rosemary Clooney. But she doesn’t always mean Rosemary Clooney. Sometimes she means Sally or Hillary Clinton or the girl I ate lunch with on the first day of seventh grade. She could be anything that can be referenced as a single female—even a ship or a country. But my lab manager knows, straight away, that she is Rosemary Clooney. Pronouns involve that extra step, that discourse mining, that sensitivity to intent and likelihood: that matchmaking. Right here, right now, who is she?
Perhaps you are beginning to see why I am obsessed.
By definition, pronouns only contain vague information, like first-person or plural. In order for something this vague to effectively retrieve a word’s meaning, there has to be a whole lot of context. Imagine all the words contained in your mind as a vast pool of fish. Look carefully and you’ll see that each fish is different from all the others. If you had a hook selective enough, you’d be able to control which fish you catch. But pronouns are not selective hooks. Pronouns are sweeping nets. You have to cast your net shallowly in the hopes that you catch the one noun the pronoun refers to. That’s what context does: it pushes what’s relevant to the surface of the mind.
There are plenty of psycholinguists out there trying to figure out what counts as context. The real answer would be everything. A more helpful answer would be a bit narrower: syntactic things like part of speech and parallel structure matter, as do pragmatic things, like how the world works and what message a speaker intends to convey. Say my lab manager says, “Sally invited Rosemary Clooney to a dinner party! She said, ‘Yes!’” She references Rosemary Clooney and not Sally, since invitations lead to responses. This is context at its most straightforward.
Most psycholinguistic work on pronouns has used third-person pronouns like he and she. This is because psycholinguists prefer straightforward problems that make it easy to design experiments; the first- and second-person pronouns require context of a different sort. A person must know the speaker, as well as the listener, for the referents of I or you to be retrievable. Fortunately, language-acquisition researchers (who often speak with psycholinguists and occasionally even conference with them) have looked at these problematic pronouns, and to interesting effect. I and you are challenging for children to learn: some children go through a stage where, not understanding you’s speaker-specific meaning, they hear their mothers say, “Do you want a cookie?” and respond “You want cookie! You want cookie!” How charming! Interestingly, some deaf children learning American Sign Language make the same mistake. Fluent speakers of ASL point to themselves to indicate I. It’s as iconic as a sign can get. But that young ASL learner will point at his mother until she hands him a cookie. It’s hard not to envy the language-acquisition researchers the adorableness of their subjects.
And there are plenty of other types of pronouns, all with their own contextual obligations (read: experimenter headaches). Sometimes people use pronouns that are unheralded, meaning they don’t refer to anything that has been mentioned previously in conversation. These pronouns generally need the most context to be understood. Often, as is the case with knowing the relationship between I and you, this context is nonlinguistic. If I say “He’s on fire!” when my cat is on fire, it is pretty obvious that he refers to the cat. I don’t even have to point. Anyone listening to my speech and witnessing the event will have enough context to know who I’m saying is on fire.
The next day, the postdoc I share an office with listens to my story about buying a catnip candle so that I could watch my cat roll endearingly on the carpet while the scent wafted across the room, and how it didn’t quite turn out that way, how instead my cat pounced on top of the candle and the fur on his belly instantly broke into flame. When I say all this and four hours later the postdoc sees me and shakes his head and says, “They should be illegal,” I know that by they he means catnip candles. Unheralded, see? The source of that blazing belly has seared itself so prominently on both our minds that it doesn’t even need to be mentioned to be there.
There are also pronouns that don’t mean anything at all. Dummy pronouns, they’re called, and we come across them all the time (you read one in the previous sentence). They’re those pronouns that exist only because the English language demands that each sentence contain a subject: the it in “It’s raining” or the there in “There is a shed in my back yard.” (Note: the there only works as an example of a dummy pronoun if I am not pointing to a shed, and am nowhere near my back yard.) (Note: most linguistic examples have caveats like this, making the linguist’s life frustrating: “You’re right,” linguists will inevitably find themselves saying at some point. “If the bartender is also drinking bourbon while she mixes drinks for a patron wearing a yellow shirt and standing directly to her left, and if all of this is happening on a Tuesday, well yeah I can get that meaning.”)
One thing we psycholinguists have learned is that using pronouns correctly is a lot of work. A person has to strike a balance between referencing something and not beating other folks over the head with it. Striking this balance requires quite a bit of attention. And when attention goes, so does proper pronoun use.
I ask a professor how Corey’s comprehensive exams went. The professor says something like, “Well, Corey’s a bright guy. I don’t think he studied as much as he could have, but overall he did fine.” But if I ask a professor how Corey’s comprehensive exams went while also asking the professor to rehearse a five-digit number in her head, she will say something like, “Well, Corey’s a bright guy. . . . don’t think Corey studied as much as Corey could have. . . . but overall Corey did fine.” It takes work to keep track of when to use a pronoun; work that we can do easily—automatically—when we’re not keeping track of anything else, but work that becomes damn near impossible if we have to keep rehearsing 9-1-0-7-4 over and over again in our heads. Try it on your friends. Try it on your spouse. Try it on your parents, but only if they’re not too old, because once their memories are shot they’ll be using a lot more pronouns. Not because they’ve become exceptional at keeping track of when the referent has last been named, but because they’ve lost track of the referent altogether.
I‘m lucky. At a large research university, I have plenty of participants (never “subjects,” at least in print) at my disposal. For at least a hundred hours each term I have other people’s lives to exploit as shamelessly (but ethically) as I would like.
I try to run participants during the first few weeks of a term when they’re plucky and ambitious, the sort of undergraduates who like to stay on top of things, who won’t try to fit all seven research hours in on the last possible day, who aren’t likely to rearrange the keyboard to spell [F][U][C][K] [ ] [T][H][I][S] or to accidentally “lock” themselves in the testing rooms by pulling at the doors instead of pushing. I run group after group, sometimes several groups at once, then go back to my desk to make sure that the data are going where they ought and that no software bug needs last-minute eradication. Then, once the numbers start looking funny, once the ceiling is judged by participants to be as likely to smile as the child and response times get painfully long—or impossibly short—I breathe a huge sigh and try to see what it is, exactly, that I’ve collected. If an experiment works, if the numbers look and smell and feel so gorgeously simple that I actually trust them, I’ll sit at my desk long after my adviser has left for home, after the postdocs and other graduate students have called it a day and the lab manager has shut down her computer. I’ll sit staring at those numbers as if they’re a message from somewhere so close I can feel its heat.
You see, psycholinguists have devised—and continue to devise—a number of different tasks to get at how pronouns work. The one I use most often in my own research is the recognition-probe task. Essentially, the task presupposes that what is most prominent in a person’s memory while experiencing language is what the language is “about” at that time. Participants enter a little testing room and sit down at a computer. Short texts appear on the screen, and the participants are instructed to read them, presumably in preparation for a comprehension test. Participants are given a secondary task: to decide as quickly as possible whether they recognize test words that appear unexpectedly on the screen as words they’ve read in the text. If a pronoun’s referent is retrieved, participants should be better at responding “yes” to the referent if it appears as a test word in the recognition task. Say you’re a participant reading along and you come to this: Rosemary threw the jellyfish that hit Anne in the chest. While she laughed and laughed, Anne doubled over in pain. If Rosemary is correctly recovered as the referent of she, then responses to Rosemary as a test word should be faster and more accurate immediately after the pronoun than they would be otherwise. Other words, like jellyfish, should receive no such boost.
Other tasks involve reading-time measures like eye tracking, where an expensive machine actually records participants’ pupils while they read. From this video, researchers are able to reconstruct the eye’s saccades (eye movements) from word to word, and to calculate the amount of time a reader spends on a precise word (or even part of a word). Reading-time tasks presuppose, of course, that where the eyes linger is what the language is “about.”
Yet another task just asks participants straight-out how acceptable they find a sentence, a task that assumes that people’s judgments about the acceptability of a sentence presented in isolation will mirror how much difficulty they would have upon encountering the sentence in a more realistic setting. Researchers using this task tend to be interested in pronouns as freaks of nature, as sneaky critters that don’t play by the rules. For instance, it seems as though, while pronouns can easily take the place of nouns in a sentence, the opposite is not always true. Let’s say a fellow grad student points out, “The gardener who the agency which the neighbors recommended sent did a wonderful job around the pool.” I’d have no idea what he was talking about, and not just because graduate students do not have pools. Let’s say, instead, he proclaims, “The gardener who the agency which I recommended sent did a wonderful job around the pool.” With just the right inflections, I can almost hear him saying this: It’s bad, but it’s not as bad. Psycholinguists talk in terms like these. “Bad, but not as bad” is a notable thing for a psycholinguist.
Recently, psycholinguists have started to act on the pictures-of-the-brain-are-sexy phenomenon that has hit all areas of cognitive science like cats on candles. And at least one big pronoun study has come out of the madness. A team of researchers at the University of South Carolina scanned participants’ brains in an fMRI machine while reading them sentences that contained either proper nouns or their pronominal equivalents. Patterns of activity across the brain (as indicated by changes in blood flow) showed a number of differences in the two cases. Notably, when pronouns—but not proper names—were read, there was activity in areas of the brain associated with spatial processing.
The lead researcher has an interesting interpretation: listening to proper names, over and over again, can be disruptive. Each proper name will bring with it a host of associations—not all of them particularly relevant. Think about it. When you hear the name Emily Dickinson, you may not just retrieve aspects of Emily Dickinson’s life and work; you may also think of every Emily you’ve ever known, as well as the time you considered naming the family dog Emily. Or you may access a beloved high school English teacher, or the embarrassment you felt when you read a Dickinson poem aloud to your seventh-grade class and pronounced all of the dashes.
Words, the researcher suggests, are loaded, and pronouns may be a way of accessing a representation of the proper name that avoids the pitfalls of the proper name itself. It’s a way to maneuver (literally, it appears) around some of the association-juggling that language processing involves, to head straight for a placeholder instead. If his theory sounds crazy, consider this: speakers of ASL will sign a proper name the first time a person is brought up, and then point to a specific location in space, essentially assigning the proper name to that space. The person can then be subsequently referenced with a simple point to that space. In our brains, of course, the space is virtual, but its importance may be very real.
This brain stuff is relatively new to psycholinguists and hard to interpret. The fMRI procedure presupposes that blood flow to the brain means what we hope it means and isn’t a byproduct of something we just haven’t thought of yet. Still, results like this may eventually force psycholinguists to approach pronouns with a different metaphor. Change doesn’t come easily to psycholinguists. After all, we’re still weaning ourselves off the idea that sentences are represented in the mind as special trees that only branch to the right. If pronouns themselves should be treated as occupying a virtual space, can I really talk about them as if they delve into memory to “retrieve” a referent? Will I have to start talking about them as if they are hooks of the coat-rack variety?
One doesn’t have to be a psycholinguist to obsess over pronouns. Pronouns are talking trilobites, the last remnants of an Old English with a very different—and much richer—morphology. English used to always mark case: words were pronounced differently depending on their part of speech. Now the only words that still mark case are pronouns: I, we, he, she, and they when the pronoun appears as a subject; me, us, him, her, and them when it appears as an object. Second-person pronouns used to mark case, too, but ye, the pronoun marked for subject, is already obsolete. After all, we don’t really need case anymore. Word order takes care of everything. We know the relationship of a subject to a verb because of their respective places in a sentence. Still, pronouns are a way of looking back, or, perhaps more interestingly, a way of experiencing languages that rely more on morphological markings and less on word order.
Which brings me to the obvious yet disarming truth that English pronouns could just as easily be something else. In some languages, pronouns are optional; context permitting, they can be dropped out of the sentence entirely. In other languages, third-person pronouns—the words like he, she, and they that are psycholinguists’ bread and butter—don’t even exist. Third persons are instead referenced with demonstratives like this or these. And in yet other languages, pronouns can carry information about the formality of the conversation, and even the social positions of its speakers.
To think about how odd this is, consider how much one type of information carried by some English pronouns—gender—can affect the way we communicate. Last year, arriving late to a departmental Christmas party, I was immediately greeted by a waifish 10-year-old with pale skin, delicate features, neatly braided long brown hair, and a stuffed clown fish. The girl solemnly informed me that her stuffed animal was dying of diphtheria. “Oh no!” I cried in mock horror. “Is your fish contagious?” Perhaps fearing I would launch into a speech about how young ladies should be careful around contagious fish, a fellow graduate student quickly interjected, “He’s sure the fish isn’t contagious. I asked him that same question.” And that is how I learned that the strange girl with the delicate features and the long braid was in fact a boy. How deftly pronominal information is delivered, and gleaned, by fluent speakers! How different the entire situation would have been were I a speaker of Hawaiian or Persian, where gender isn’t marked at all!
There is even evidence that linguistic markers of gender can shape the way we think. Not in big ways. Not in the ways that had so much of social science drooling in the ’60s. But in little ways, more like a nail file than a chain saw. A Stanford researcher presented bilingual English-Spanish and English-German speakers with a picture of a bridge. In Spanish and German, unlike English, even some inanimate objects are referenced with gendered pronouns.
In Spanish the word for bridge is marked masculine (and thus referenced with a masculine pronoun); in German the word is marked feminine. The researcher instructed the participants to describe the photograph. Then an independent group of participants rated all of the adjectives the bilinguals had written as either masculine or feminine.
German bilinguals consistently described the bridge with more feminine adjectives (elegant, beautiful), and Spanish bilinguals described it with more masculine adjectives (sturdy, dangerous). Here’s the kicker: instructions were given in English, descriptions were written in English, and the photograph of the bridge was just that—a photograph. This suggests that pronouns might be important, not just to how we use language, but to how we experience the objects in our world (although, as dear Steven Pinker points out, “Just because a German thinks a bridge is feminine, doesn’t mean he’s going to ask one out on a date”).
Lucky for me, there are plenty of pronouns in need of more study—the diectics (here, there), the reflexives (himself, themselves), the interrogatives (who, what), the possessives (his, mine), the indefinites (somebody, anything)—each with its own relatively unexamined life. Or, for the freshest pronoun around, I could always coin one myself.
In Baltimore, some teenagers already have: their candidate, yo, is a new gender-neutral third-person personal pronoun. As in Yo was tuckin’ in his shirt or Yo sucks at magic tricks. If yo sticks around—and if it spreads—maybe we can put the ever-awkward he or she to rest forever. And what would that mean? What consequences could that have for how we think about our world? Empirical question. Send in the psycholinguists.