The “hello” squabble now seems silly, but it’s the same thing that’s going on in the 2010s with “hey.” “Hey” as a summoner, as in “hey you,” is a word we’ve had since at least the 1200s, and “hi” is simply a variant on the vowel, dated back to at least the 1400s. Etymologically speaking, “hey” and “hi” have similar trajectories as “hello,” minus the phone part: all three started out as ways of getting attention and have evolved into greetings. But as recently as 1960, researchers for the Dictionary of American Regional English found just 60 people who would greet someone they knew well with “hey”—in comparison to 683 who’d say “hi” and 169 who’d say “howdy”—and most of those 60 were in a single region, the South and Lower Mississippi Valley. According to the same survey taken in 2014, “hey” had slightly surpassed “hi” in popularity. Linguist Allan Metcalf reported a recent college graduate explaining to him around the same time: “I almost always say ‘hey’ in speech but I have free variation between ‘hi’ and ‘hey’ in writing. . . . I have a sort of three-way formality distinction for greetings—‘hey’ for friends my own age or younger, ‘hi’ for adults I know well or people my own age I’m just meeting, and ‘hello’ for adult strangers.” Someone born in 2000 could justifiably point out that “hey” has been used as a greeting for their entire life, while someone born in 1950 could, equally justifiably, say, “But I don’t feel greeted by ‘hey,’ I feel summoned!”
Shifts in greetings are especially noticeable in email. When we email, we don’t face the total lack of knowledge about who we’re communicating with like the telephone users had. Instead, the first time we email a new person, we know just enough about them to be dangerous—we have a name, probably, but very little background information on their desired style of greeting. In speech, it’s easier to skip over the details and let our messages be carried by tone of voice. But in writing, you can’t get away with answering, “Good, how’re you?” to “What’s up?”—it’s ruined by the other person’s ability to reread your message and think about it for an extra half second. Moreover, real-time, embodied communication gives us extra cues for selecting a greeting: to swiftly size up someone’s likely attitude towards “hey” based on their age, or adjust our greetings on the fly to mirror the other person’s.
The first emails dealt with this lack of cues by simply not caring. In 1978, said technologists Albert Vezza and J. C. R. Licklider about email, “One could write tersely and type imperfectly, even to an older person in a superior position and even to a person one did not know very well, and the recipient took no offense.” Similarly, in an article from 1998, linguist Naomi Baron noted that “most users exercise only a light editorial hand (if any at all) on email messages before they are sent. Many of us chuckle at the error-strewn emails we receive from colleagues otherwise noted for meticulously crafted memoranda.” Amid anarchy like this, who had time to care about the niceties of specific greetings? But by 2001, email systems had spellcheck, and linguist David Crystal wrote, “I receive innumerable e-mails which are anything but fragmented sentences.”
At the same time, Crystal observed that the most frequent email greeting he received was “Dear David,” followed by “David” and then “Hi David.” I remember rereading this passage around 2010 and being surprised. I hardly ever saw “Dear Gretchen” in my inbox, and I certainly would never have addressed someone else with it. “Dear” felt both excessively formal and oddly intimate, the kind of thing that I had inscribed in my best grade-school cursive in thank-you notes to my grandmother for the lovely birthday sweater—certainly not the kind of relationship I associated with work or school. “Hi” felt businesslike, breezy, impersonal, like a polite social smile. A few years later, however, as I got more adept at business emails, I did find myself occasionally using “dear,” at least enough to switch to it when someone else used it first. According to the research of the linguist Gillian Sankoff, I may not be alone here: although much sociolinguistic research finds that the way you talk is pretty much established by late adolescence, Sankoff finds that some speakers may keep changing well into middle age, especially for formal and prestigious bits of language.
In a historical sense, though, my initial instinct was right on target: we’ve been following a trend towards shorter and less descriptive greetings for several centuries. “Dear” is our last relic of what used to be an elaborate system of greetings that describe people in flattering terms, which was popular for well over a millennium. Here’s a sixteenth-century letter from Edmund Spenser to Walter Raleigh on the publication of The Faerie Queene, which is typical of the genre:
To the Right noble, and Valorous, Sir Walter Raleigh, knight, Lo. Wardein of the Stanneryes, and her Majesties lieftenaunt of the County of Cornewayll.
[text of letter]
So humbly craving the continuance of your honorable favour towards me, and th’ eternal establishment of your happines [sic], I humbly take leave.
23 January. 1589.
Your most humbly affectionate.
Ed. Spenser.
But some of these greetings were more rote than sincere. For example, founding fathers Alexander Hamilton and Aaron Burr exchanged a series of letters in 1804 which all closed with “I have the honor to be Your Obedient Servant,” but culminated with them fighting a duel. At the time, it was simply a polite, phatic phrase, just as the modern writer of “Sincerely” isn’t being especially sincere, simply following a social script. But as “Your Obedient Servant” fell from common usage, its literal meaning recovered. The 2015 musical Hamilton, a dramatization of these historic events, highlights this phrase as an ironic juxtaposition by making it the refrain of a song about the origins of the duel. The irony is entirely modern, however: a Founding Father time-traveling to Broadway for opening night wouldn’t notice anything amiss.
“The Right Noble and Valorous” and “Your Obedient Servant” are no longer stock, phatic phrases to us—no one in the twenty-first century would employ them in a generic business email. But the same is gradually becoming true for “dear.” If you didn’t encounter “dear” enough for its meaning to wash out, and the post-letter-writing generations may not have, it feels oddly like calling your boss or your professor your darling. Even if individual people adopt “dear” for older correspondents, as I did, it’s doomed in the long run if people aren’t using it among their peers, as I would never, never do. Reading through the comments on etiquette posts shows a tendency for younger people to resist advice to use “dear,” not through a desire to be rude or informal, but because they simply cannot parse it as anything but intimate. We can only hope that a musical of the 2200s digs up a hostile email thread featuring archival “dears” and writes a song about how dramatically ironic they sound.
It would be easy to attribute the shift in greetings to a broader shift in society: the adjectival greetings could be a bid for affection; the summoning greetings a bid for attention. But such an association is too easy and I think it’s simply wrong. It’s better to realize that greetings are generally phatic: that we pick a particular greeting because that’s what we’re used to, and to acknowledge that we’ve always wanted both affection and attention. (Besides, if the internet is so unaffectionate, why the popularity of the heart-shaped “like” button?) When we zoom out and take a historical perspective, the shift in greetings is simply a change, one that reminds us that language is a thing that lives in the minds of individual humans at individual points in time, a thing that can’t be fully encompassed in a static list of rules like a game of chess.
As someone who receives a fair bit of email from people I’ve never met, I greatly enjoy the chance to treat their various greetings as a game of a different kind, a never-ending multiplayer guessing game of what generation someone’s in and what they’re trying to signal based on how they address me at the top of their emails. What the random strangers in my inbox have in common is hope. Hope that I’ll read their link, hope that I’ll reply b
ack with the answer they’re looking for—hope, even, unfortunately, that I’ll purchase something from their marketing campaign. The solution is less to try to stamp out variation and more to try to exercise the same kind of generosity with each other as my schoolmates unwittingly extended to me in my “How are you?” “Not much” experiments. There’s enough genuine malice in the world that we don’t need to go hunting for more of it in what is truly a case of harmless difference.
Chat and Interruption
Babies learn the rhythm of having a conversation before they even learn the words to do so. When we talk to them, we tend to ask them questions, leave spaces for them to reply, and react to their cooing and babbling as if they’re participating with us. “Are you sleepy?” *baby gurgles and rubs eyes* “Yeahhh, I think you’re sleepy.”* What we learn, long before we utter any words, is that conversations are made up of turns. A conversation isn’t a cacophony of voices all talking at once: it’s a smoothly synchronized back-and-forth.
How do we know when it’s our turn? It would be easy to assume that we must pause after we’re finished saying something, and that other people notice that pause and interpret it as an invitation to speak. But conversation analysts find that actually we don’t pause much, any more than we normally pause between each word. If I ask you a question and you don’t start answering immediately, I’ll probably treat it as a break in communication. Even if just 0.2 seconds go by, I’m likely to repeat the question again, try a different way of phrasing it, or switch languages (to the eternal bane of would-be polyglots). This finely tuned timing is what our caregivers taught us as babies when they treated our gurgling and babbling as conversational turns. (If you’ve ever found yourself unable to get a word in edgewise, or doing all the talking around someone frustratingly taciturn, it’s probably because your cultural timings are ever so slightly miscalibrated for each other, points out the linguist Deborah Tannen.) If you’re acting in a play, it’s just as important to know when you’re supposed to come in as it is to know your actual lines: if you wait for a full pause before every time you start speaking, you’ll sound incredibly stilted. While another person is talking, we’re not just composing our own reply, we’re predicting when they’re going to finish, so we can smoothly transition from speaker to speaker.
How do we do this seamless coordination? It can’t be length: a turn can be as short as one word (“Yep”) or as long as a whole story. Instead, we listen for signals that someone might be at the end of a turn and, if we’re in a group, who they’re expecting to speak next. Some of the signals are linguistic, like if someone asks you a direct question with your name in it. Some are gestural: a raised hand is often a signal that you want to get a word in, and people tend to look away during their turn and look back towards the group or the next speaker when they’re finishing up. A lot of these signals are about intonation and rhythm: people will speed up if they get to a potential turn end but don’t actually want to stop talking, or they’ll use a rising intonation to prompt a response.
But none of these cues are a hundred percent reliable, so sometimes we simply guess. Conversation analysts find that “interruptions” aren’t randomly distributed in conversation: instead, they’re at points when it seems like the main speaker could be finished talking but it turns out they aren’t. In face-to-face conversation, a syllable or two of overlap is nothing terrible (humans are good at associating words with the locations of specific people), and we sort out most such confusions without thinking much of it. In conversations mediated by technology, overlaps can be a bigger problem. Walkie-talkies don’t allow overlaps at all, so guessing becomes painful and people instead say “over” at the end of a turn. Telegraph operators used to end all their turns with “GA” for “go ahead” for the same reason. Some early chat systems had a similar problem with overlap: the chat feature on a system called TENEX in the early 1970s was a single text file that you and your conversational partner edited together, one keystroke at a time. If you tried to type at the exact same time, you’d end up interleaving your letters between each other, so if the letters started jumbling, one person had to stop typing and cede way for the other, or you’d just end up with a horrible mess.
Hey, how’s it goignogod how are you?
Some systems tried to prevent jumbling with conventions of their own: users of one text-based chat system developed the habit of typing in two linebreaks between each turn. Other chat systems, such as the Unix talk program, split the screen into several areas and assigned each person their own text box to type in: if I’m typing only in the top box and you’re typing only in the bottom one, our letters can’t possibly jumble together. You could add more boxes for more people: a chat program created in 1973 called Talkomatic, on a system called PLATO at the University of Illinois, provided five boxes, stacked on top of each other, so it could support up to five participants. One of these boxes was yours to type in, and then you had to keep scanning the others to see what the other people were saying, keystroke by keystroke, each in their own box. Here’s a demonstration, because this paradigm is really very different from all the chat systems you’re used to.
Talky McFirst
hello?
I am demonstrating chat boxes
Chatter O’Second
hi!
I am a second participant
Speech Thirdova
I am a third
the responses may look like they’re out of order, because everyone types in their own box
Words Fourthescue
I am a fourth
I’m good, how are you number five?
Typo von Fifth
I am the fifth
how’s everyone doing?
Scanning between boxes and letters that get jumbled up with each other? Well, we can maybe guess why they didn’t catch on. But instantly appearing letters? That sounds fancy! Why did chat systems stop using those after the 1970s and 1980s? Indeed, Google tried to bring instant letters back in 2009 with the short-lived Google Wave, but it never caught on. The problem with keystroke-by-keystroke chat is that it treats conversation as a thing of letters rather than turns. Not only is it painfully slow to watch someone type in a message letter by letter, because we can read faster than we can type, and not only does it make us self-conscious for other people to watch our backspacings, but we also run into the turn-taking problem of radio and telegraph operators. Have I stopped typing because I’m finished, at least for the moment, or just because I’m figuring out how to say something? You have no way of knowing. We could try to impose a convention like “over” or “GA” or a linebreak or some lesser-used punctuation character, but if we’ve learned one thing from the history of email, it’s that people are inconsistent about following etiquette as long as it’s a mere suggestion. Far better to hardwire the turn break into the form of the chat interface itself, by letting us send each new message turn by turn, rather than letter by letter.
So where did the upwards-scrolling, dialogic chat interface that we know and love today come from? The oldest example that I’ve found is from 1980, in the form of a chat program called CB Simulator, which was also the first dedicated, online chatroom to be widely available to the public. It wasn’t even called a chatroom yet: instead, it was inspired by shared radio waves. Citizens band, or CB, radio is a kind of radio that doesn’t just broadcast out: anyone in the local area can tune in and talk with each other, and enthusiasts do just that. (It’s similar to ham radio but even more decentralized.) An employee of CompuServe, an early online service provider, thought CB radio would be a neat thing to emulate in typed form, and CB Simulator was born. But radio conversations are based on a wholly different paradigm than the shared text boxes of early chat systems: instead of people each talking in their own box, radio sees conversation as a single stream of turns that an individual could choose to add to. The multi-box style imposed a hard limit on how many p
articipants could be involved (there’s no place for a sixth person in a five-box chat, and it quickly gets difficult to scan back and forth between boxes to follow a lively group conversation). By contrast, the number of participants in stream-style chat was a lot more flexible. CB radio enthusiasts were used to a bit of chaos in their word stream: voices crackling in and out, signals dropping and overlapping. The text format actually made the chaos easier to manage. If several people sent a chat message all at the same time, the system coped gracefully: you simply saw all their messages in a stream rather than hearing their voices talking over each other.
Overlapping letters had been terrible, but a stream of overlapping phrases turned out to be chat’s key feature. The multi-box chat style stuck around for a while during the 1980s, but by 1988 the stream-style chat was dominant and here to stay: that’s the year Internet Relay Chat (IRC) was created. It was the system that powered the classic public internet chatroom as we know it, and IRC used a stream. These public chatrooms were much analyzed through the 1990s, and one of the things that researchers consistently noticed was how chaotic chatroom conversations were: multiple conversations would happen at once, messages interleaved with each other, and users didn’t seem to mind. Here’s a nineties example of such overlap:
*** Signoff: puja (EOF From client)
Because Internet Page 21
