by Rich Horton
Whether we said what was needed, whether it needed to be said: that is some other kind of question. Did anyone suppose that the confluence of us all would be a diplomat?
The imago pressed us close, but that was an enquiry. There was pattern in the pressure: we could see it, we could read it almost, those of us with finger-talk or bubble-talk or both. What lives, what choices? Swim or fly, drown or burn? Swallow or be swallowed?
We knew, we thought, how to press back, how to pattern a reply. Mr. Holland gave us what we lacked: content, poetry, response. Meaning more than words. Sometimes the map declares the territory.
For he who lives more lives than one
More deaths than one must die.
He would have turned the bitterness all against himself, but our collective consciousness couldn’t sustain that. We all wanted our share, we all deserved it: all but Barley, who had no hidden other self, who’d had no time to grow one.
Suddenly he couldn’t hold us together any longer. Fraying, we fled back to Durand, back to our waiting bodies—and the imago pursued, flying by sheer will in the dreadful night, wreaking havoc in its own frozen body. It followed us to the Dolphin and hurtled against the conservatory where we were anything but sheltered, battering at the windows like a moth at the chimney of a lamp, until the only abiding question was whether the glass would shatter first or the machine, or the creature, or ourselves.
Hello, Hello
Seanan McGuire
Tasha’s avatar smiled from the screen, a little too perfect to be true. That was a choice, just like everything else about it: When we’d installed my sister’s new home system, we had instructed it to generate avatars that looked like they had escaped the uncanny valley by the skins of their teeth. It was creepy, but the alternative was even creepier. Tasha didn’t talk. Her avatar did. Having them match each other perfectly would have been . . . wrong.
“So I’ll see you next week?” she asked. Her voice was perfectly neutral, with a newscaster’s smooth, practiced inflections. Angie had picked it from the database of publicly available voices; like the avatar, it had been generated in a lab. Unlike the avatar, it was flawless. No one who heard Tasha “talk” would realize that they were really hearing a collection of sounds programmed by a computer, translated from the silent motion of her hands.
That was the point. Setting up the system for her had removed all barriers to conversation, and when she was talking to clients who didn’t know she was deaf, she didn’t want them to realize anything was happening behind the scenes. Hence the avatar, rather than the slight delay that came with the face-time translation programs. It felt wrong to me, like we were trying to hide something essential about my sister, but it was her choice and her system; I was just the one who upgraded her software and made sure that nothing broke down. If anyone was equipped for the job, it was me, the professional computational linguist. It’s a living.
“We’ll be there right on time,” I said, knowing that on her end, my avatar would be smiling and silent, moving her hands to form the appropriate words. I could speak ASL to the screen, but with the way her software was set up, speaking ASL while the translator settings were active could result in some vicious glitches. After the time the computer had decided my hand gestures were a form of complicated profanity, and translated the chugging of the air conditioner into words while spewing invective at my sister, I had learned to keep my hands still while the translator was on. “I’m bringing Angie and the kids, so be ready.”
Tasha laughed. “I’ll tell the birds to be on their best behavior.” A light flashed behind her avatar and her expression changed, becoming faintly regretful. “Speaking of the birds, that’s my cue. Talk tomorrow?”
“Talk tomorrow,” I said. “Love you lots.”
“I love you, too,” she said and ended the call, leaving me staring at my own reflection on the suddenly black screen. My face, so much like her computer-generated one, but slightly rougher, slightly less perfect. Humanity will do that to a girl.
Finally, I stood and went to tell my wife we had plans for the next weekend. She liked my sister, and Greg and Billie liked the birds. It would be good for us.
“Hello,” said the woman on the screen. She was black-haired and brown-eyed, with skin that fell somewhere between “tan” and “tawny.” She was staring directly at the camera, almost unnervingly still. “Hello, hello.”
“Hello!” said Billie happily, waving at the woman. Billie’s nails were painted bright blue, like beetle shells. She’d been on an entomology kick again lately, studying every insect she found as raptly as if she had just discovered the secrets of the universe. “How are you?”
“Hello,” said the woman. “Hello, hello, hello.”
“Billie, who are you talking to?” I stopped on my way to the laundry room, bundling the basket I’d been carrying against my hip. The woman didn’t look familiar, but she had the smooth, CGI skin of a translation avatar. There was no telling what her root language was. The natural user interface of the software would be trying to mine its neural networks for the places where she and Billie overlapped, looking for the points of commonality and generating a vocabulary that accounted for their hand gestures and body language, as well as for their vocalizations.
It was a highly advanced version of the old translation software that had been rolled out in the late 2010s; that had been verbal-only, and only capable of translating sign language into straight text, not into vocalizations that followed spoken sentence structures and could be played through speakers. ASL to speech had followed, and then speech to ASL, with increasingly realistic avatars learning to move their hands in the complex patterns necessary for communication. Now, the systems could be taught to become ad hoc translators, pulling on the full weight of their neural networks and deep learning capabilities as they built bridges across the world.
Of course, it also meant that we had moments like this one, two people shouting greetings across an undefined void of linguistic separation. “Billie?” I repeated.
“It’s Aunt Tasha’s system, Mom,” said my nine-year-old, turning to look at me over her shoulder. She rolled her eyes, making sure I understood just how foolish my concern really was. “I wouldn’t have answered if I didn’t recognize the caller.”
“But that’s not Aunt Tasha,” I said.
Billie gave me the sort of withering look that only people under eighteen can manage. She was going to be a terror in a few years. “I know that,” she said. “I think she’s visiting to see the birds. Lots of people visit to see the birds.”
“True,” I said, giving the woman on the screen another look. Tasha’s system was set up to generate a generic avatar for anyone who wasn’t a registered user. It would draw on elements of their appearance—hair color, eye color, skin tone—but it would otherwise assemble the face from public-source elements. “Hello,” I said. “Is my sister there?”
“Hello,” said the woman. “Hello, hello.”
“I don’t think the computer knows her language very well,” said Billie. “That’s all she’s said.”
Which could mean a glitch. Sometimes, when the software got confused enough, it would translate everything as “hello.” An attempt at connection, even when the tools weren’t there. “I think you may be right,” I said, moving to get closer to the computer. Billie, recognizing the shift from protective mother to computer scientist with a mystery to solve, shifted obligingly to the side. She would never have tolerated being smothered, but she was more than smart enough not to sit between me and a puzzle.
“Is Tasha there?” I asked again, as clearly as I could.
The woman looked at me and said nothing.
“I need to know what language you’re speaking. I’m sorry the translator program isn’t working for you, but if I know what family to teach it, I can probably get it up and running in pretty short order.” Everything I said probably sounded like “hello, hello” to her, but at least I was trying. That was the whole point, wasn’t
it? Trying. “Can you say the name of your language? I am speaking casual conversational English.” No matter how confused the program was, it would say “English” clearly. Hopefully that would be enough to get us started.
“Hello, hello,” said the woman. She looked to her right, eyes widening slightly, as if she’d been startled. Then she leaned out of the frame and was gone. The image of Tasha’s dining room continued for several seconds before the computer turned itself off, leaving Billie and I to look, bemused, at an empty screen.
Finally, hesitantly, Billie asked, “Was that one of Aunt Tasha’s friends?”
“I don’t know,” I said. “I’ll call her later and ask.”
I forgot to call.
In my defense, there were other things to do, and none of them were the sort that could easily be put off until tomorrow. Greg, our two-year-old, discovered a secret snail breeding ground in the garden and transported them all inside, sticking them to the fridge like slime-generating magnets. Greg thought this was wonderful. The snails didn’t seem to have an opinion. Angie thought this was her cue to disinfect the entire house, starting with the kitchen, and left me to watch both kids while I was trying to finish a project for work. It was really no wonder I lost track of them. It was more of a wonder that it took me over an hour to realize they were gone.
Angie wasn’t shouting, so the kids hadn’t wandered back into the kitchen to get in the way of her frenzied housework. I stood, moving carefully as I began my search. As any parent can tell you, it’s better to keep your mouth shut and your eyes open when you go looking for kids who are being unreasonably quiet. They’re probably doing something they don’t want you to see, and if they hear you coming, they’ll hide the evidence.
I heard them laughing before I reached the living room. I stopped making such an effort to mask my footsteps, and came around the corner of the doorway to find them with their eyes glued to the computer, laughing at the black-haired woman from before.
“Hello, hello,” she was saying. “I’m hungry, hello, can you hear me?”
Greg laughed. Billie leaned forward and said, “We can hear you. Hello, hello, we can hear you!” This set Greg laughing harder.
The woman on the screen looked from one child to the other, opened her mouth, and said, “Ha-ha. Ha-ha. Ha-ha. Hello, hello, can you hear me?”
“What’s this?” I asked.
Billie turned and beamed at me. “Auntie Tasha’s friend is back, and the program is learning more of her language! I’m doing like you told me to do if I ever need to talk to somebody the neural net doesn’t know, and using lots of repeating to try and teach it more.”
“The word you want is ‘echolalia,’ ” I said distractedly, leaning past her to focus on the screen. “You’re back. Hello. Is my sister there?”
“Hello, hello,” said the woman. “Can you hear me? I’m hungry.”
“Yes, I got that,” I said, trying to keep the frustration out of my voice. It wasn’t her fault that her language—whatever it was—was causing issues with the translation software. Tasha’s neural net hadn’t encountered as many spoken languages as ours had. It could manage some startlingly accurate gesture translations, some of which we had incorporated into the base software after they cropped up, but it couldn’t always pick up on spoken languages with the speed of a neural net belonging to a hearing person. Tasha also had a tendency to invite visiting academics and wildlife conservationists to stay in her spare room, since they were presumably used to the screeching of wild birds.
“If not for them,” she had said more than once, “you’re the only company I’d ever have.”
It was hard to argue with that. It was just a little frustrating that one of her guests kept calling my kids. “Can you please tell Tasha to call me? I want to speak with her.”
“Hello, hello,” said the woman.
“Good-bye,” I replied and canceled the call.
Both children looked at me like I had done something terribly wrong. “She just wanted someone to talk to,” said Billie mulishly.
“Let me know if she calls again, all right? I don’t know who she is, and I’m not comfortable with you talking to her until I’ve spoken to Tasha.”
“Okay, Mom,” said Billie.
Greg frowned but didn’t say anything. I leaned down and scooped him onto my shoulder. That got a squeal, followed by a trail of giggles. I straightened.
“Come on, you two. Let’s go see if we can’t help Mumma in the kitchen.”
They went willingly enough. I cast a glance back at the dark computer screen. This time, I would definitely remember to call my sister.
As always, reaching Tasha was easier said than done. She spent much of her time outside feeding and caring for her birds, and when she was in the house, she was almost always doing some task related to her work. There were flashing lights in every room to tell her when she had a call, but just like everyone else in the world, sometimes she ignored her phone in favor of doing something more interesting. I could have set my call as an emergency and turned all the lights red, but that seemed like a mean trick, since “I wanted to ask about one of your houseguests” wasn’t really an emergency. Just a puzzle. There was always a puzzle; had been since we were kids, when her reliance on ASL had provided us with a perfect “secret language” and provided me with a bilingual upbringing—something that had proven invaluable as I grew up and went into neurolinguistic computing.
When we were kids signing at each other, fingers moving almost faster than the human eye could follow, our hands had looked like birds in flight. I had followed the words. My sister had followed the birds. They needed her, and they never judged her for her differences. What humans saw as disability, Tasha’s birds saw as a human who was finally quiet enough not to be startling, one who wouldn’t complain when they started singing outside her window at three in the morning. It was the perfect marriage of flesh and function.
After two days of trying and failing to get her to pick up, I sent an email. Just checking in, it said. Haven’t been able to rouse you. Do you have houseguests right now? Someone’s been calling the house from your terminal.
Her reply came fast enough to tell me that she had already been at her computer. A few grad students came to look @ my king vulture. He is very impressive. One of them could have misdialed? It’s not like I would have heard them. ;) We still on for Sunday?
I sent a call request. Her avatar popped up thirty seconds later, filling the screen with her faintly dubious expression.
“Yes?” she said. “Email works, you know.”
“Email is too slow. I like to see your face.”
She rolled her eyes. “It’s all the same to me,” she said. “I know you’re not really signing. I prefer talking to you when I can see your hands.”
“I’m sorry,” I said. “Greg’s ASL is progressing really well. We should be able to go back to real-time chat in a year or so. Until then, we need to keep the vocals on, so he can get to know you, too. Look how well it worked out with Billie.”
Tasha’s expression softened. She’d been dubious when I’d explained that we’d be teaching Billie ASL but using the voice translation mode on our chat software; we wanted Billie to care about getting to know her aunt, and with a really small child, it had seemed like the best way. It had worked out well. Billie was fluent enough in ASL to carry on conversations with strangers, and she was already writing letters to our local high schools, asking them to offer sign language as an elective. Greg was following in her footsteps. I really was pretty sure we’d be able to turn off the voice translation in another year or so.
To be honest, I was going to be relieved when that happened. I was lazy enough to appreciate the ease of talking to my sister without needing to take my hands off the keyboard, but it was strange to hear her words, rather than watching them.
“I guess,” she said. “So what was up with the grad students? One of them called the house?”
“I think so,” I said. “She s
eemed a little confused. Just kept saying ‘hello’ over and over again. Were any of them visiting from out-of-country schools? Someplace far enough away that the neural net wouldn’t have a solid translation database to access?” Our systems weren’t creating translation databases out of nothing, of course—that would have been programming well above my pay grade, and possibly a Nobel Prize for Humanities—but they would find the common phonemes and use them to direct themselves to which shared databases they should be accessing. Where the complicated work happened was in the contextual cues. The hand gestures that punctuated speech with “I don’t know” and “yes” and “I love you.” The sideways glances that meant “I am uncomfortable with this topic.” Bit by bit, our translators put those into words, and understanding grew.
(And there were people who used their translators like Tasha did, who hid silent tongues or a reluctance to make eye contact behind computer-generated faces and calm, measured voices, who presented a completely default face to the world and took great comfort in knowing that the people who would judge them for their differences would never need to know. I couldn’t fault them for that. I was the one who asked my sister to let me give her a voice, like grafting a tongue onto Hans Christian Anderson’s Little Mermaid, for the duration of my children’s short infancy.)
“I don’t know,” she said, after a long pause. “Only two of them spoke ASL. The other three spoke through their professor, and I’ve known her for years. Why? Did she say something inappropriate to the kids?”
“No, just ‘hello,’ like I said. Still, it was strange, and she called back at least once. Black hair, medium brown skin. I didn’t get a name.”
