Skip to content

Using Our Words

I have a new all-consuming hobby: teaching my pets to talk.

I’m nearly finished writing a Nine Lives that puts this statement in a larger context — the tragic history of research into animal cognition and communication, and how one speech-language pathologist, video-based social media, pandemic lockdowns, and a few thousand pet owners combined to not only give the scientific field new life, but directly led to what’s going on in my household right now — but for the moment let’s just go with this: the unexpected addition of a puppy to our family of cats and humans changed the calculus for me on whether it was worth it to spend my carefully hoarded savings testing out animal communication aids on our pets.

I’m writing this account now not just share with interested people, but also to get the timeline correct and all in one place for my own future reference. (I’ve had to go back to various text messages to Jak and other friends to piece together the number of days between each development, as once we’re a few days out from any given event my recollection of time gets vague and often inaccurate.)

So: at the end of August I came back from our wedding trip to Michigan with a whole extra suitcase mostly full of dog supplies for Tashi, including twelve buttons which play a recorded sound when pressed and three rubber foam ‘hextiles’ with which to create a custom soundboard.

We had been having some pretty intense behavioral difficulties with Tashi, and part of my hope in starting up with buttons is that having better communication and more agency would alleviate and/or redirect some of his emotional tantrums … much like we teach toddlers to ‘use your words’ instead of throwing frustrated fits. So when choosing the early words to introduce, I did so with Tashi in mind; the adult cats I put off for later.

Behavior problems, from this adorable sleeping creature? Surely not.

I wanted to start with two buttons — enough so that he would understand that different buttons produce different results, but not so many that he risked being overwhelmed and confused out of the gate. Having observed that Tashi’s main desire in life was ‘play’, I chose to lead with PLAY and SKRITCHES. Late on Saturday, September 2, I set down the first two buttons. I had carefully kept my expectations low as a bulwark against getting discouraged too soon; calculating a rough average across anecdotal uptake lead times, I had prepared to model for several weeks with a dog and several months for a cat.

Instead I found myself having to recalibrate my timetable rather precipitously. It took Tashi only one day to figure out 1) the overall idea of button-activation and 2) that by hitting PLAY he could (usually) get a human to play with him. The pleasant but less-desirable concept of SKRITCHES followed a day or so later.

So by the end of the first week, I’d already added two more buttons; still going with the theme of ‘motivating requests’, I chose OUTSIDE and KONG. He had already picked up the spoken word ‘outside’, so literally within minutes of my placing the OUTSIDE button, he used his first two-button combination: PLAY OUTSIDE.

KONG involved food, so that didn’t take very long either. I’m feeding Tashi a carefully calculated custom raw diet, a portion of which is chicken gizzards. Rather than including them with the rest of his meal, I stuff the day’s portion of gizzards into rubber kongs and freeze them solid. I figured the cold would feel nice to him while he was teething; also getting the frozen gizzards out entertains him for twenty minutes or so, and it slows down at least that much of his food intake, which otherwise he would gulp down in literally one second (usually resulting in hiccups). We started at the beginning of July with two kongs a day; as he’s grown we’re now up to five (!), so it’s a request that we can often say ‘yes’ to.

Right around the same time that I added OUTSIDE and KONG, we started verbally modeling ‘all done’ — a critical concept for giving us poor humans a break from all the playing and outsiding — along with ‘now’ and ‘later’. At about the two-week mark, I introduced a fifth button for ALL-DONE. It took a few days before Tashi clearly demonstrated that he understood what it meant, and of course he didn’t use that button nearly as often as the request buttons. But he started doing things every once in a while like getting up after he’d emptied a kong and pressing ALL-DONE KONG. (Interestingly, although we had only modeled ALL-DONE in the final position — KONG ALL-DONE, PLAY ALL-DONE, et cetera — Tashi consistently used it in the initial position.)

Tashi happily gnawing on a kong in our living room, near the starter soundboard.

Once ALL-DONE felt solid, about two and a half weeks in, I added buttons for NOW and LATER. At first his use seemed experimental and/or random; two whole weeks went by — so about four weeks since we started verbally modeling ‘now’ and ‘later’ — before I was certain that Tashi understood those concepts. I made this judgment in part due to some contextually appropriate button use, but also because his understanding when I used the words was evident in his behavior. Importantly, it helped reduce the amount of frustrated lashing out when I refused a request to play. He still doesn’t like the answer ‘play later’, but he accepts it gracefully a lot more often than before buttons. (I can’t rule out that this improvement was just a function of puppy development stages, but the timing and speed of improvement suggests understanding ‘later’ — as opposed to taking all refusals as a flat ‘no’ — has helped.)

The other thing that happened during the two weeks I was trying to teach him about time is that Tashi’s use of SKRITCHES, which previously had been almost flawless, stopped making any contextual sense.

Now, there had always been the occasional fit of ‘press literally all the request buttons’, which I mentally translate as “Mom/Dad, I’m boooored, pay attention to meeeee.” But aside from that, when Tashi requested SKRITCHES he seemed happy enough to get skritches — rolling over to expose his belly, for example. And then, over the course of a couple of days, I realized that was no longer the case: I was responding to SKRITCHES just the same as always, but his body language kept saying that he didn’t want physical affection. (Usually it was clear that what he actually wanted was to play; sometimes I thought he wanted outside or I wasn’t sure.)

I tried for a few days to sort this out, and then just decided something had gone completely haywire with that one word. In hopes of resetting, I took SKRITCHES off the board. He didn’t show any noticeable signs of missing it, and we continued to verbally reinforce ‘skritches’ in moments of active petting and belly rubbing.

Meanwhile I added POTTY — another word he already knew very well — and (once I was very sure he understood NOW and LATER) FOOD. We also started verbally adding ‘soon’ as a third time word, so ‘now’ means more or less immediately, ‘soon’ means sometime in the next half-hour or so, and ‘later’ means sometime today. If it’s a thing we won’t be doing again that day — like OUTSIDE once it’s after dark, or FOOD after his dinner is over — then we said ALL-DONE instead.

At this point, I really started running into the limitations of my equipment, not in number of buttons but in number of hextiles. Tashi was sometimes choosing to push buttons with his nose and sometimes with his feet; the nose is pretty accurate but the feet … not so much. If I were only working with cats it would not have been an issue, but with a sixteen-kilo (and counting) puppy, it turns out that buttons in adjacent slots are often mistakenly caught by overeager clumsy paws. We started getting a lot of PLA-UTSI-OTTY, all jumbled on top of each other.

By this time I’d already decided not only that AIC (Alternative Interspecies Communication) was worth pursuing, but that waiting the four to five months until our next trip to the US would be too frustrating for both Tashi and me. So even though the cost of shipping to Mexico caused my frugal little heart an almost physical pain, I wrote out a preliminary word list and ordered what I hoped was enough equipment (nine more hextiles, and forty-eight new buttons) to last us through January. Unfortunately, even though I placed the order on September 20, various delays meant it would take at least a month to get here, so as of this writing I still don’t have it.

This coming Saturday will mark seven weeks since beginning with the buttons. I’ve been verbally modeling upcoming words like crazy over the last four weeks, but — knowing that with the new equipment I’ll need to completely rearrange the soundboard — I’ve held off on adding any new button words, with one exception: about a week and a half ago (so about five weeks from our start date) I replaced SKRITCHES.

Within fifteen minutes Tashi went and pressed SKRITCHES, and happily accepted them. Okay, I thought, maybe we fixed that confusion.

And then less than an hour later he pressed SKRITCHES OUTSIDE. Which was … odd. I didn’t think he actually wanted skritches-while-outside — there was no precedent for that — so I decided to treat them as two separate requests. I went to the buttons and told him OUTSIDE LATER, SKRITCHES NOW … but this time he did not seem to want skritches at all.

However, I had more information now than I’d had three weeks earlier. I’d continued reading and listening to advice from people who’ve gone through this ahead of me, and I’d learned that — well, first of all that taking buttons away is contraindicated if they’re being used (oops) — but also that pets with limited button options will often repurpose buttons. This sort of ‘off-label’ usage is intentional, if not always immediately comprehensible.

For example, there’s a cat who began saying OUTSIDE BED to mean ‘window’. (Nope, I do not know why ‘bed’.) When his person gave him a WINDOW button, he completely stopped using OUTSIDE BED in favor of WINDOW. (And then one day some weeks later, he used OUTSIDE BED again … which is how his person found out that the WINDOW button had stopped working.) There’s also a dog who requested DINGBERT PUZZLE, a combination which made no sense to her person (DINGBERT being a particular dragon toy) until it clicked that the dog was using toys to signify color — Dingbert was the same teal color as the puzzle she wanted. That same dog also started using FRISBEE as a verb, meaning ‘throw’.

Sure enough, this time I caught a pattern that I’d missed the first time (although in hindsight, I realize it was present then too): when Tashi presses SKRITCHES by itself, with no other buttons, he actually wants skritches. On the more frequent occasions when he presses SKRITCHES in combination with one or more other requests, he does not mean skritches.

Exactly what he does mean by it, I don’t yet know for certain, although it is definitely deliberate and consistent. I originally came up with three possible theories: a) it’s a general request for human attention, b) it’s an intensifier of some kind, showing he wants the object or activity very much, and c) it’s a positive indicator, like saying ‘good’. While I could more easily imagine a conceptual path from the modeled meaning of SKRITCHES to either a) or c), I thought at the time b) had the most evidence going for it.

When I took the question to a community of experienced AIC people, several of them offered variations on a fourth theory: that Tashi conceptualized SKRITCHES as ‘person using their hands’. One said her dog uses PETS to mean ‘hands’, and another said her dog uses CUDDLES to mean ‘come here, do something’.

After another week and a half of pattern-tracking (waiting, waiting for the new tiles and buttons), I’ve downgraded the likelihood of c); Tashi doesn’t use modifier!SKRITCHES in the declarative, only in the imperative. (Although to be fair, with only nine buttons, his ability to narrate anything is extremely minimal.) I am leaning mostly toward a) or d): ‘pay attention, human!’ or ‘help me with your human hands’. (And I find it absolutely fascinating that multiple dogs appear to be making the same apparent cognitive association and linguistic repurposing independently of each other, especially since it’s in a direction that is so non-intuitive for humans.)

As soon as possible (without overwhelming him) once I have the new tiles, I plan to add everyone’s name, plus WANT and HELP and COME, and model them so he has less-confusing ways of asking for attention and help. If I’m lucky, we can get SKRITCHES back to its original meaning (unless a button breaks, I guess!).

So that’s where we are with Tashi and the buttons so far. Next post I will talk about the cats, and my exciting (well, at least to me) plans for the future.

Published inPets


  1. David Gates David Gates

    This is truly a fascinating story. During the first year of covid I decided to settle down and finish a project I had been dreaming about for a long time. I invented my own language. At least 80 to 90 percent of any such project is wrapped up in vocabulary creation. Actually, it can take several years. In hindsight I’d have been better of t have invented useful words and concepts for animals. Have you thought of creating meaningful concepts for your pets that might not be in the human ideational realm?


    • Karawynn Karawynn

      Oooh, a conlang! That’s so cool. Can you tell me about it?

      I’ve never created a conlang from scratch, though I have done extensive work on the three languages used by the cultures in my novel. One is a future permutation of British English, one is a future admixture of American English and American Sign Language, and one is a future Spanish-Portuguese creole — all of them further shaped by their environments.

      I don’t think we know enough about the inner thoughts of non-human animals to really grasp what would be meaningful to them that isn’t meaningful to humans … although one thing about teaching them an English pidgin is that we can begin to find out. When several dogs with no contact independently exhibit the same ‘off-label’ word use in the same type of way, I think we’re getting a glimpse. The fact that dogs seem to generalize a broader concept from ‘skritches’/’pets’/’cuddles’ is one such clue; if we could definitively ascertain how they’re generalizing (attention? hands?), that would be amazing.

      Another concept that both dogs and cats apparently often come to on their own is insults: fascinatingly, multiple teachers have reported that ‘stranger’ gets applied to a known person as a mild insult and ‘potty’/’poop’/’litterbox’ as a stronger insult. The latter constellation also gets used like a metaphor, to label something they’re displeased with, the way a person might say ‘this tastes like shit’ or ‘this [situation] is some bullshit’. Apparently ‘shit’ is a cross-species concept. 😀

      • David Gates David Gates

        Kinda tough to know what to say about my conlang. I called it zziva – pronounced zhiva. Well, It started back around 2004 or 5 with an attempt to list necessary vocabulary for any language. I used a top down category based model not knowing that such efforts date back to at least 1650. They also are generally debunked, but who cares. I ended up with about 4500 words. Nothing original.

        So, when covid struck I remembered all this and thought all I need now is a nice clean grammatical structure (cool verb tenses, articles and noun cases) and naturally a nifty alphabet. What I ended up with was pretty satisfying to me but very pedestrian by the standards of most conlang enthusiasts. There are a gazillion of them out there and they create some very artistic stuff. Plus, many of them are quite advanced linguists – not me. The talk in the discussion groups can be really sophisticated.

        I haven’t looked at this for a couple of years now, so it has grown cold. But, I am pretty proud of what I came up with in terms of its general usability and, if I do say so, a kind of elegant efficiency. Only 18 letters and a complete set of prefixes and postfixes. A pile of conjunctions and subordinate clauses. You get hundreds of these kinds of structures.

        Then there is the question of learning the monstrosity. This is where you get bogged down and need to keep looking everything up in your own damned language – and you give up, especially since the lockdown ends and you go back outside. Your friends stick their heads out of their own holes, and the conlang is archived.

        But, it’s always there if I ever need it – HA!

        The problem nowadays is that I have other projects that I never seem to find time for. Just lots and lots of unfinished stuff and a clock that ticks way too fast. But, thanks for asking. (I assume your life is the same.)

        Keep well and happy,


Leave a Reply

Your email address will not be published. Required fields are marked *