This is in response, and speaking directly, to the creator of this video.
Edit: On 9/23/22, Soup Emporium actually contacted me about this article, and his classy AF response is below. Folx, this is how scientists should behave. Thank you, Soup, for your humility and openness to learning. Everyone else, subscribe to his channel!
Hello there!
Sorry if this is a bit of an intrusion. I’m the guy who made the Koko video (I guess you already knew that).
Sorry its taken me so long to find your blog, I was only sent it today! I wish I’d found it earlier. I loved reading your thoughtful piece on my video. I found myself nodding along to a lot of it. A lot of the points you raised are things I’ve thought about and come around to with the benefit of time and perspective, as well as a lot of other points I didn’t consider.
Your first point – yeah, I definitely had a lot of gaps in my knowledge around behviourism – I’ve had a few non-Chompsky linguists reach out to me about it, and it was definitely a shortfall on my part – one you explain very nicely in your blog. I’m not trained as a linguist, I’m a geologist by trade. I did the arrogant “Oh I can figure this out thing” and, like you said, its an area I needed to tread carefully. The chats have been very eye opening. I still find Chompsky’s ideas convincing, and I still find Terrace’s 1979 paper a more persuasive case than not, but all the same, I didn’t tread as carefully as I could have, or handle the topic with the degree of nuance it deserved. It’s definitely something I missed the mark on.
I’m also glad that you brought up the implicit ableism angle up as well. I didn’t catch the views Terrace was echoing about autism on first read – going back to look at it, it should have been something I noticed.
I should have been more vigilent to not reinforce the kinds of ways we priviledge typical communication when making the case I was – especially when considering neurodivergence. It didn’t come on my RADAR, but, like you said, it should have. In a bit of a twist, a lot of comments on the Koko video were people accusing Keller of being a fraud — which of course she wasn’t — and investigating that/eventually turning it into a video (which is being released today, funnily enough) has been quite an eye opener on a lot of my own internalised ableism that I’ve not interrogated as thoroughly as I should have.
Anyway, this was a long winded way of saying thank you for taking the time to write the piece you did. I wanted to ask if you were happy with me linking your blog piece on my Channel’s community page? I’d like to be able to point people to resources like yours.
I look forward to more thoughtful pieces from you in the future!
Warm Regards,
— Soup
And now, the article as it was first published:
Having gone through your sources and watched your video several times, I understand where you are coming from. My assumption is that creating this video comes from a place of promoting good science, picking apart bad science, and generally combating misinformation. I couldn’t agree with you more there. The way the media handles science and spreads “information” for the sake of ratings and attention has snowballed into the anti-science nightmare world in which we both live today, on both sides of the Atlantic.
However, you’re inadvertently perpetuating a problem in two parts:
- Application of a colloquial understanding of behaviorism.
- Implicit ableism to which you seem completely oblivious.
Take a second with that, because I’m sure you’re now feeling very defensive. No one likes being called ableist. No one likes being called anything-ist. No one likes being told that something they spent a huge chunk of their time researching and creating is immensely problematic and flawed.
Here’s the thing, though: I work with and stand by the disabled community, and your video is a problem. My hope is, as a person who appreciates science and facts, you will take this article as constructive feedback and maybe even make a follow-up to your video with new information. After all, good science changes its conclusions based on new information.
Honestly, if you still don’t agree that Koko could “talk” even after reading this, that’s fine, just find a better way to defend it.
In case you’ve made it all the way here without going “who the hell is this bitch?” and heading over to the About page: I am an actual expert in behavior and learning, and I am going to show you how your conclusions were ill-informed and promote problematic ideas that make the world less safe for disabled humans.
Flawed definitions
Let’s start with definitions: What does “talking” even mean? What is language? Throughout your video, you present multiple supposed non-examples of language, but never actually establish in any concrete way what language is. There are definitely implied aspects, but you never actually said anything along the lines of “this is what we are measuring as language.”
Surely you understand that’s a bad start to any study. How can you measure and compare if you’ve never established what it is that you’re measuring and how?
The definition you use vs. the actual definition
Based on the various arguments and points you make in the video, the general assumption I was able to glean was that your operating definition of language is that it’s complex, containing words strung together to create phrases of (apparently) no less than 4 words, and containing syntax and grammar. There also seems to be an assumption that the audience will just understand this, that it is a generally accepted understanding of what language is.
As someone who helps others build language skills every day, this was an odd assumption to me. However, you’re probably right– given the amount of stories from disabled people with communication difficulties and differences regarding how they are treated by the general public, the majority of people likely does accept your implied definition of language.
That doesn’t make it OK, though, and it doesn’t make it an accurate judgement on Koko’s communication skills. Language encompasses a wide array of skills and variables, but your argument wasn’t that she didn’t have language, your argument was that she couldn’t “talk.” “Talking” is a vague, colloquial term that can mean many things, including communicating.
From a simplified developmental and learning perspective, communication requires 2 things: a speaker and a listener. Expressive language is made up of 3 components called verbal operants: Manding (requesting), Tacting (labeling), and Intraverbals (basically conversation). The tools for building these skills are imitation and receptive language; receptive language of course being your ability to understand what is said to you.
Therefore, communication is any effective means of delivering an idea from a speaker to a listener.
So we’ve established what your operating definition of language is (I hope) and what the definition of language is from a developmental perspective. Now, let’s consider your sources.
Flawed sources
If you’re not an expert in the field, it’s hard to see a problem with your sources. They were “experts” and contemporaries in the field, some even participating in ape research themselves and criticizing their own work. From an outside perspective, that’s great research.
I’m sure you can hear the “but” coming…
BUT, Herbert Terrace was a practicing psychologist in the ’70s, around the same time as Ivar Lovaas. For the sake of length, I won’t go into the full significance of that comparison, but I invite you to read about him here and here and draw your own conclusions. The short version is that this was a period of time in which people who couldn’t communicate in the same way as the majority of the population were literally considered less than human.
Terrace’s paper, Why Koko Can’t Talk, echoes Lovaas’ thought processes on autistic children from his published works. Yes, he’s talking about apes primarily, but he’s comparing them to typical child development and almost aggressively dismissing “atypical” development. Terrace’s paper also somehow asserts that grammar is the most important part of human language, an idea you seem to have taken very much to heart as it’s the basis for most of your arguments.
Again, this is a time in which psychology and behaviorism viewed anyone who wasn’t “normal” as something less than human. Lovaas even compares treating the autistic children in his care as “human” rather than “patients.” He stated that they were not “whole people,” but people he needed to “build” by forcing them to perform in ways that appeared neurotypical. This is the atmosphere in which Terrace was practicing.
Updated sources matter, even when looking at old studies
Page 2 of Terrace’s paper also highlights a flaw in his own methodology. He notes that Nim didn’t communicate spontaneously, but rather describes a classic example of what is called a chain. He would grab for what he wanted, then be prompted to sign, and would then emit the sign, which was reinforced.
He was not properly trained to communicate. What he was trained to do was engage in a behavior, expecting the trainer to prompt him, then engage in the prompted behavior. It’s a common error in teaching, and often, trainers (especially if they were not adequately trained to do their job) don’t even realize they’re doing it.
There’s a saying in behaviorism: The rat is always right. It means if the behavior you’re seeing isn’t the one you wanted, you did it. The learner is doing only what you taught.
That being said, Nim does demonstrate some communication skills, but I’ll get to that later.
The point is that there are better sources out there now and better research on the validity of different patterns of language development. Furthermore, we have a much better understanding of not only how to measure and track language development, but also how varying patterns in development can still result in effective communication. Looking at research in mental health and development from previous decades is useful for exploring the foundations of the sciences, but that’s pretty much it. Move on quickly, and certainly don’t use them to draw conclusions. Otherwise, you are carrying their problematic and dated ideas into the modern day, allowing them to continue doing harm and preventing them from dying.
Evidence and measurement are important
Another excellent source (on the surface) is the testimony from the deaf employee working with Nim, beginning around 31:07. The former trainer perfectly describes his own lack of training or appreciation for the researchers’ goals. None of that is his fault, they should have a) trained him better with a clearer definition of what responses they were looking for, and b) had enough respect for him as a deaf person to explain the difference between the sign he speaks and the sign they were trying to teach.
At no point did Nim emit the sign for “eat” as he understood it, and eventually the chimp just started screaming.
Some actual science
Let’s pause to explain some behavior science. A discriminative stimulus is something that appears in the environment to indicate that reinforcement is available for a certain response. The simplest example is when pigeons are trained to peck at a button when it lights up green but not blue. If they peck at the button while it’s lighting up green, food comes out. If they peck at the button while it’s lighting up blue, nothing comes out. So, the green light is a discriminative stimulus indicating reinforcement is available for the behavior of pecking a button.
Now, once that behavior has been reinforced for a while and becomes consistent, let’s say you stop reinforcing any pecking. Green light comes on, pigeon pecks, nothing comes out.
What do you think the pigeon will do?
Well first, it will peck the fuck out of that green-lit button. It will peck a lot more, it will peck harder, it may even then attack the button, try other ways of pressing it, try other buttons, etc., before eventually ceasing the pecking behavior altogether. This is called an extinction burst or a behavior burst. It happens when reinforcement for a previously reinforced behavior is removed. We all do it. Think of your own behavior when a vending machine eats your money. Chances are, you push the button more, you push other buttons, you push them harder, you may kick the machine, swear, etc., but eventually you will move on when none of these behaviors are reinforced by either getting your money back or getting your snack from the machine.
Let’s go back to Nim. There’s no video of this incident, so we have to go by the provided testimony. The description is great, though, and quite frankly, I think we can both agree that he has nothing to gain from making the whole thing up.
So the ex-trainer says that as soon as he appears in front of Nim, the chimp begins moving his hands constantly. He then starts screaming and moving his hands even more. At no point did the trainer see anything that indicated “eat” or “drink” (as he understood it), so he didn’t reinforce any of the hand movements.
The discriminative stimulus was present (trainer with milk), the ape emitted responses which were not reinforced, so the responses increased, including other, more intense responses such as screaming.
This should sound familiar; Nim is demonstrating a behavior burst, which implies that the responses he was emitting had been reinforced in the past. He was in the presence of something that indicated that reinforcement was available for these responses, but then the responses were never reinforced. The trainer even states that non-deaf employees would be jotting down tons of notes while the deaf employees saw nothing worth recording.
Rather than conclude that Nim had no effective language, here are some more likely conclusions to draw from this testimony:
- The researchers had no respect for the deaf employees or ASL.
- The researchers failed to consider the difference between what they were teaching and what the deaf community uses in order to properly train their staff.
- The employees had no clear definition of what behaviors were to be reinforced.
- The researchers weren’t monitoring and/or weren’t properly interpreting the data to realize there was a vast difference between what deaf and non-deaf employees were observing, and thus:
- The researchers never provided adequate training or retraining to ensure Nim’s program was being run correctly. They failed to identify an extinction burst when they saw one.
A conclusion we can’t draw from this testimony? Literally any statement on Nim’s actual ability to communicate. The trainer didn’t have the educational background or proper training to observe that.
Putting bad ideas in the spotlight
Laura-Ann Pettito demonstrates simiar problematic views on language at 32:27: “Nim didn’t do anything with the signs. He only used them for requesting things– and even that’s too anthropomorphic of a description– he never used them in the deeper, human sense of making a request.”
What the fuck is the “deeper, human sense of making a request?” Maybe I need to see video of what she’s referring to– and I would really like to– but goddamn if that statement isn’t ableist as fuck. How can you make an assumption that someone who’s expressive communication is limited to requesting a few things isn’t fully human?
Also, apes aren’t humans and never will be, so again, I’m confused about her point here.
We have no way of knowing what anyone, ape or human, is really thinking. No one can read minds, and we currently don’t have any technology that can allow it.
I realize she (and you) are simply criticizing a poorly-done study on teaching language to chimps, but you have to realize the behaviors and skill levels being belittled here also apply to actual human beings. When you discuss these skills this way, you are making comparisons and statements, whether you mean to or not, about those humans that are incredibly offensive.
Flawed understanding
With flawed sources and limited experience in the field, you are set up for failure. It also makes it even harder for you to interpret the information. This is not meant to be an insult to your intelligence at all; in fact, I applaud the effort you put in to educate yourself on this topic. As I stated before, though, these are fields that require advanced degrees, and without that knowledge, you are going to misinterpret things.
Let’s begin with your paraphrase of Chomsky’s point of view against the behavioral view of language development as “language is too complex to develop via a history of reinforcement.” That is an accurate depiction of his view and also a strong demonstration of its absurdity, as well as his complete lack of understanding for both what he was arguing for and what he was arguing against.
If you consider children as essentially apes at a typewriter, you and Chomsky are correct. It’s incredibly unlikely that randomly landing on a word that is then reinforced is going to develop into the complex languages we have. It also leaves no room for grammar development, etc.
But that’s not how language develops, and that’s not what behaviorism claims.
Side note: Chomsky’s argument is essentially the same one Creationists sometimes make against evolution: “Life is so complex, there had to be an intelligent design!” (read: I don’t understand science, so it must be magic!).
Actual explanations
As I mentioned above, one of the main tools for developing expressive language is imitation. Yes, the babbling infants do can be considered “nonsense,” but arriving at words is not random. Within a few months of being born, a baby’s hearing develops such that they are able to start picking up more sounds. At that point, they also begin to imitate them.
It can be hard to tell, especially if you don’t know what you’re looking for, but here’s an easy way to observe it: Listen closely to infants in different cultural and language environments. You will hear differences in their babbling, even down to variations in dialects, and it starts early.
Vowel sounds are usually the most obvious example. Babbling from an infant living in a home where everyone’s only language is American English is going to sound different from that of an infant growing up in a home where everyone’s only language is British English. Now take a family in which the caregivers speak Spanish and English but only speak English in front of their children, except they speak with a Spanish accent: The babbling from their infant is going to sound different from their monolingual neighbor’s.
It’s not random. They are imitating and developing their vocal imitation skills, until they are finally able to emit a word approximation. That word approximation is then usually heavily reinforced.
You also failed to consider child-directed speech (previously known as Motherese), a natural, instinctual way in which adults talk to children that simplifies language, over enunciates sounds, slows the rate of speaking, and raises the tone to something more pleasing to the infant to hear, thus making them more motivated to attend to the speaker’s sounds. It’s an evolutionary response with the specific purpose of bonding parent and child and helping the child develop communication.
Again, not random.
Flawed Kid vs. Ape Comparisons
You make a lot of statements comparing what the various apes did to what “kids do” that made me wonder when was the last time you spent a significant amount of time around young children.
Let’s start with the assumption you make relatively early on at 15:35 that “kids” will initiate conversations, talk to themselves, make related comments, etc. That’s not a completely true statement. Most typical kids will do these things, but 9.4% of children have ADHD, 1 in 42 children are autistic, nearly 1 in 12 have speech disorders, and around 10% have speech delays. That is a massive portion of the population that you are simply ignoring when you make such a blanket statement about language development in kids.
The relevance of nonsense
Then there’s this utterance at 16:09: “Give orange me eat orange me eat orange give me eat orange give me you.” You state that “a kid would never say something this devoid of meaning or repetitive.”
OK, then why is this funny?
Here’s another experiment to try in the wild. Spend some time alone with a 2 year-old. I say alone because you need to be the only one available to them to get them something they want, like a snack or a super cool toy that’s out of their reach or something. When they ask you for it, try ignoring them for a minute or two. Spoiler alert: You will also find that a child will try to “brute force” a response. Everyone will. Remember the extinction burst? The fact that you have more responses in your arsenal than a 2 year-old (typical or otherwise), or Nim for that matter, doesn’t make what you do any better, and it certainly doesn’t make it more “language.”
It’s the exact same behavior; it just looks different.
Thoughts are hidden but still important
Another example of things the apes did that kids supposedly don’t occurs in your discussion of the AOL chat at 20:51. Now, I will definitely agree with you that the chat itself is a bit dodgy, especially Penny’s interpretation of Koko’s responses, but I will get more into that later. The point you bring up is that Koko’s responses are not relevant to the questions being given and you state that this is not a thing that “kids” do.
Once again, that is factually inaccurate.
Allow me to elucidate with a real-life scenario that I had the privilege of witnessing. It even involves a child without a disability, so it fits into your pool of examples thus far. Let’s set the scene:
My supervisor was visiting one day while I was consulting in a preschool that consisted of a mix of disabled and non-disabled children. Among those non-disabled children, was a rather friendly and talkative young lad around age 4. I knew him from my work in the school and he would frequently come up and interact with pretty much anyone. During one such approach, the following exchange took place:
Boss: I’m cold.
Kid: I like cherry.
(kid walks off)
By the logic you’ve established, that child was not using language and may, in fact, have no actual language to speak of, but I’m sure you can see that there are a lot more possible variables here.
Does the kid have no functional language? Of course not.
Did he string together a nonsensical word salad? Well, no. He strung together 3 words in the correct order to make a sentence that conveys meaning in English.
Does he like cherries? I don’t know. He said he did, but we can’t measure anything beyond that because I’ve never seen that child eat anything cherry or cherry-flavored.
Also, he’s talking about cherries when the subject matter was my boss labeling his own internal feelings of temperature. It’s not a related comment, so it can’t be considered a conversation. Apparently he was thinking about cherries in some form, though, because that’s all he had to say.
Back to the AOL chat: What was Koko looking at, thinking about, seeing, hearing, etc.? We don’t know. The only conclusions we can draw are what Penny is tacting (labeling) as interpretations of Koko’s emitted signs. That’s literally it. The chat may highlight some of the flaws in Penny’s practice as an interpreter, but it’s impossible to make statements about Koko’s language abilities based on it.
One of the most important clips you include starts at 19:26. It’s one of the few uncut clips provided, and you completely missed the significance of it. When interacting with Mister Rogers (RIP), Koko displays a level of communication that is actually quite advanced and impressive, especially considering that she’s a gorilla.
Mister Rogers had asks “how do you say love?” and from there, you hyper-focus on the fact that she didn’t answer the question without Penny’s prompt.
But watch that clip again and look at Koko. She’s not looking at Penny or even Mister Rogers. She’s touching and looking at his cufflink. From these behaviors, it’s safe to assume that her attention is on the cufflink, not on Mister Rogers’ question of “how do you say ‘love?’”
Following his question, Koko says at 19:37 “what’s that? Flower,” and points to his sun-shaped cufflink. That is, she uses the vocabulary she has to demonstrate a very high level of mand: Requesting information. This is a skill you don’t see until around age 2 in humans. Before that, mands are usually for concrete things: mama/dada (caregiver/attention), eat/baba/crackers (food/drink), etc. Following these stages of concrete requests is the infamous “why?” phase that anyone who has spent extended periods of time with young children knows well.
You use this clip of Koko engaging in advanced and spontaneous communication as evidence for her lack of communication because she didn’t answer the question.
I am left asking again: When was the last time you spent a significant amount of time interacting with children? Asking a kid to show/tell me something and having them go off on a completely different topic is just Tuesday.
Also, many people– kids and adults– with communication differences and difficulties including autism will often not answer questions and there are a variety of reasons for it. It’s vital that you keep this in mind, especially since disabled people have been killed because of it.
We all play with language
The same goes for the supposed strings of “nonsense” words Koko, Washoe, and Nim all demonstrate. That’s not evidence against their use of language, it’s evidence for it. Playing with language is a natural part of language development and use. Sometimes it’ll be a sound, sometimes it’ll be echolalia (repeating things without an intent to communicate anything). Even if you remove disabled children from your “normative” data entirely, kids still do this, with or without a listener involved.
Older humans do this as well, but with more complex language. Ever get stuck with a particular word and find any excuse possible to work it into a sentence? A friend of mine got really into the word “heinous” for a while. Anything bad was “heinous.” It’s the exact same thing, just typically-developing adults do it with a wider vocabulary, longer phrases, more complex syntax, etc, whereas younger and non-typically developing humans do this with a smaller vocabulary, fewer words, and less complex syntax.
It’s the same thing, it just looks different, but it’s hard to see when one’s understanding of the field is only rudimentary.
The apes’ communication
You’re right at 30:51. The apes are not communicating using ASL. They are communicating using the sign language they were taught. We established from former deaf employee’s testimony at 31:14 that the methods were far from respectful to the deaf community. Of course we’d also established at that point that psychology and behaviorism had no respect for any “non-normative” population during the ’70s, so it’s really not surprising.
Was it still language though? Yes, but it’s hard to see it when you are brute-forcing the rules of ASL as we know it onto gorilla sign, as evidenced at 31:51.
You ask a fair question at 22:44: “How much is the ape communicating and how much is the human interpreting it?” That same question can be applied across species and variation in that species (i.e. humans and neurodiversity). It’s a question that’s difficult to answer, especially with hard numbers, because we can’t measure intent. However, there are observations and data you can take that can provide about as much evidence as is possible without mind-reading.
It’s the spontaneity that counts
Prompt-dependent language is problematic. It makes it unclear if it is actual communication– such as manding– or simply imitating the modeled word/sign, etc. However, this is how novel communication is acquired. As I already mentioned, very young infants begin with babbling and imitating sounds they are hearing in their environment. Only through reinforcement and pairing these words with their “meaning” do they become functional. Once we’ve seen a baby do this with a few words, what do we do if the baby is showing signs of wanting something?
Let’s say a very young child is pointing to or reaching for a stuffed animal and saying “that,” or “give” or even just sounds. Eventually, you figure out what they want, and say something like “oh! You want the doggy?” etc. The child says “doggy,” everyone cheers and gives lots of attention because it was just so cute. Maybe we also say “yeah! Doggy! This is a doggy!” etc.
In that example, the child wasn’t necessarily using language. They were imitating the word “doggy” because it was prompted by their caregiver. Later though, the child sees the same stuffed animal and says “doggy” to their caregiver in order to get them to give it to them. Now, it’s novel, effective communication, but it had to be prompted and reinforced first.
If you model the word “doggy” for the child, have them repeat it, but never give them the stuffed animal or do any cheering or extra attention, you may hear them say “doggy,” again later, but it will likely be in an unclear context and without function because that combination of sounds was never paired with reinforcement or any other stimulus with which to establish meaning.
That’s the difference.
What is accuracy?
The examples you provide of Koko’s signs are also far more valid approximations for their intended meanings than you are giving them credit for.
You understand that mimicry leads to language, but at 23:52, you make the claim that apes never stop mimicking and, thus, never generalize it to functional language. Yet, you’ve just spent over 23 minutes providing examples to the contrary. There’s video of Koko, Nim, and Washoe emitting signs without a demonstrative prompt from any of the humans around them, (e.g. 10:11, 14:56, 14:59, 19:25, 23:05). The example at 23:50 is especially clear because Koko demonstrates the accuracy of her mand by showing motivation.
She signs “hug,” and then gestures for a hug before Penny moves in for one. None of Koko’s behavior– signing “hug” and then opening up her arms for one– can possibly have been prompted by Penny because it all occurs before Penny does anything.
A prompt is something that elicits the desired response; thus, it has to happen before a response in order to be a prompt, or mimicry.
Just like the dogs in the clips following 23:50, the lack of complexity does not exclude it from being actual language. If you say “sit” to a dog and the dog sits. Then yes, that dog “understands” the word “sit.” There’s literally no other way to measure this.
About that sign language…
Now, all appreciation for ASL as a massively cool language aside, your tone when you say that Koko used “modified” sign at 28:07 rubbed me all kinds of the wrong way.
Full disclosure: I am not fluent in sign by any means. I can say a few sentences such as “nice to meet you,” “I want (insert variety of things that mostly appeal to children),” “do you want more ___?” etc. The rest of my vocabulary is primarily these “floating” signs you speak of. Because I don’t currently support any learners who require me to have a more complex understanding of sign, my level suffices for their needs.
You see, modified sign and gorilla sign are just as valid forms communication as full sign, and frankly, I am not sure why you would treat it any other way. There are many non-speaking people who have fine motor difficulties that make some signs impossible to form and therefore need to modify them– such as infants. Really, that’s all Baby Sign is. It’s sign language modified to accommodate fine motor deficits but still allow someone who cannot speak to communicate.
And yes, single words are communication.
So, you’re saying the researchers learned single signs and “assumed they could slap English grammar on top of it and call it a day (28:46).” Did they though? Was there ever an expectation or even an intention of gorillas developing the ability to have full-blown complex conversations with ASL as the full, functional language it is?
It’s never stated in your video or any of the research I saw on ape communication, so there’s no way to know for sure. If it were me, though, with my experience in the field, I would absolutely start with simple mands, tacts, and maybe fill-ins and simple questions but probably not aim much beyond that, especially for relatively preliminary research.
If the apes showed signs of being able to go farther than that with language, I’d start over with a new ape and then explore their ability to communicate with more complete, fluent ASL.
With that being said, why not “impose” aural aspects onto sign when the organism you’re teaching is hearing? I realize this can be a controversial statement, especially to the D/deaf community and its supporters, but stay with me here. The goal of learning these signs to begin with had nothing to do with ASL or the communities that use it. The morality or amorality of that fact notwithstanding, it had everything to do with the goal of teaching apes to communicate with humans. That’s it.
Why not use it in the way in which you, as a teacher, are most familiar with engaging with it, especially if that way is perfectly functional for the ape? The use of spoken word order is similar to that of vocal speech in young children (29:08) which is what they were using as a guide to determine each goal for their ape as they progressed. If that’s the goal, it’s irrelevant if the word order is correct in standard ASL. They were trying to mimic spoken language in an organism incapable of speaking.
Most importantly, the apes were using the language they were taught and they were using it functionally.
It’s the exact same mechanism and relevance of raising a child to speak any language. You could completely make up your own language and then teach your child to communicate with that language. It doesn’t make their use of that language to communicate “not language,” regardless of how useful that language is to them outside of their family.
Turns out, for gorilla sign language, sounds do matter.
At 29:39, you say the signs for “pink” and “stink” aren’t similar, but they are. It’s easier to see if you have an understanding of motor movements and errors that occur when motor movements are limited in some way. Similarly, Koko replacing “need” with “knee” at 30:01 makes sense given how she is being trained to communicate.
If you are teaching visual symbols as pairings with auditory stimuli, it’s not that far-fetched that the two would be interchanged occasionally. “Knee” and “need” sound very similar which can easily lead to Koko erroneously switching these 2 words, especially if she is being inadvertently trained to occasionally interchange them.
Consider the flawed teaching methods used with Nim and Koko. The two words “knee” and “need” could easily have been reinforced with interchangeable meanings. The trainers were potentially interpreting what Koko “meant” and prompting accordingly, such as saying “did you mean ‘need?’” or even then prompting her to “correct” the sign. Thus, they were inadvertently reinforcing these words to have double meanings.
Think about the situation of chaining I discussed with Nim. Instead of learning to say “need,” Koko learns that if she signs either “need” or “knee” she will get what she wants eventually, just sometimes there will be a step in between (i.e. saying “knee,” then being prompted to say “need,” then saying “need,” and receiving the reinforcer).
It’s less efficient, but it’s still leading to reinforcement. The errors are being incorporated into the gorilla’s repertoire and being used communicatively. It’s still language. It’s just not language you’re considering acceptable.
Simple language can still be complex
As I mentioned before, communication is an interaction between a speaker and a listener in order to effectively convey an idea. That idea may be “give me food” or “may I please request the filet mignon delivered thinly sliced to my domicile in a timely fashion?”
The level of complexity is irrelevant because language doesn’t just start being language when it hits a certain level; it’s language as soon as it’s effective communication. The first time an infant says “dada” not just as a babble, but as a mand– to request attention from their father– it’s language.
Koko goes even beyond this. She demonstrates intraverbals: Emitting different verbal responses to another person’s verbal response without a visual, such as filling in partial statements, answering questions, and making related comments. It’s the most complex form of language, and Koko was doing it.
Was the “speech” she gave later really her thoughts? Probably not. But it doesn’t make her language abilities any less real or valid.
Because of your insistence on a certain level of complexity in the phrases as a measurement of effective language, you again missed something really impressive about what Washoe did when she said “water bird.” Previously her training had been in mands. What she did in the park was a tact. Instead of using language in order to gain access to a desired thing (requesting), she was identifying an item purely for the sake of labeling it.
This is (very broadly speaking) a slightly more complex use of language. Moreover, she demonstrated generalization of language by using words currently in her vocabulary, words previously only ever used as mands, to create a new idea. Spreading a response onto novel situations is called generalization, and it’s essential for effective communication. What Washoe did was generalization in two ways: Using vocabulary previously taught through mand training as a tact (label), and putting together 2 words that had not been paired before to create a new idea.
It shows that she is grasping and utilizing language as a communication tool.
“Errors” aren’t mistakes
I am reminded of a story in which a person was talking with their friend for whom English was not his first language. He forgot the word for “carousel” and, in his attempt to communicate the idea, came up with “horse tornado.”
Apparently this was not a unique “error” either.
Not only did they still effectively communicate the idea, but quite frankly, I find this to be a far superior way of referring to the spinning ride with fake horses than “carousel.”
This has also been demonstrated in nonspeaking children who communicate through the Picture Exchange Communication System (PECS), which is exactly what it sounds like. They use a set of specific picture icons to communicate various ideas. An obvious flaw in the system is that their vocabulary is limited to the pictures in their supply.
It’s similar to everyone being limited by the spoken vocabulary in their own mental repertoire, but with the added barrier that the vocabulary is made up of tangible items and thus, needed pictures cannot just appear in their communication books the way new words can appear in our brains.
Some researchers decided to explore this. After being taught a variety of mands using PECS, non-speaking children were then shown a novel candy they had never seen before and thus, had no picture icons with which to request it. So they did essentially the same thing Washoe did. The candy was licorice, and all of them found a different way to effectively convey the idea that they wanted some of it, such as “I want black candy” “I want string candy,” “long food, please,” etc.
At 22:57 you state that Washoe could have just as easily been simply signing “water” and “bird,” but if she did this while oriented at a swan, it seems an unlikely coincidence that she would just happen to put these words together randomly. It’s valid to interpret the signs she made as a label using the vocabulary she had. Like pictures, new signs cannot just appear and be emitted as soon as a new thing appears in our line of sight. The specific sign would need to be taught, otherwise you are left with the tools you have to try to make that communication happen. Even if she was simply stating there is a bird and there is water, the language is still there. What she is specifically labeling is irrelevant.
The studies themselves
There are some very fair criticisms of Penny’s paper, as there are and should be with any scientific paper. However, some of the criticism you mentioned didn’t make sense to me. Penny defined signs as spontaneous and appropriate given the context, so unless the data is falsified (which is absolutely a possibility), the data presented shows acquisition of language.
Those “lists” and “internal videos” you have trouble with are just raw data. It’s not great data, but it’s not totally invalid either. As practitioners, we sometimes have lists of requests and other demonstrations of vocabulary for the people we support. It’s a valid way to track language progress if you’re not going to take video of literally every moment of a person’s life.
That being said, it’s totally fair to be skeptical of a list of signs– after all, it’s literally just taking someone’s word on what is going on– but more complex-looking data isn’t inherently better. Other data would have been better, such as how often certain utterances were used, percentage of accuracy, number of errors, etc., but it doesn’t negate that language acquisition took place. Not to mention there is actual video of these apes using sign effectively to communicate in your own video.
There are absolutely flaws in Penny’s conclusions, methodology, and certainly her treatment of Koko as her caregiver. No disagreements with you there. Her data though just isn’t as useless as you state. A lack of evidence of abstract thought (18:10) has nothing to do with Koko’s language acquisition, especially at the point in her progress in which Penny published her paper.
Sapolsky’s interpretation is also confusing at 18:48 in which he says there’s no actual data, but then in the same breath states that there is “heartwarming” video of Koko. I know he’s referring to her “speech” about saving the planet, but there is still other video of Koko using sign to communicate. Are all 2,000 claimed signs recorded? No. Is that a problem? Of course, but you’re throwing the baby out with the bathwater.
Other points
Penny’s statement at 22:05 that a relationship needs to be developed for communication to occur isn’t as woo-woo as it sounds either. She makes a fair point that Nim was being passed from scientist to scientist and mistreated frequently over the years. This will absolutely result in stunted skill development for Nim and a breakdown of previously acquired skills over time.
Just like in humans.
Selective mutism occurs for a variety of reasons in children and adults. If abuse, neglect, unstable environments, or any other traumatizing event occurs, especially within the window in which children’s brains are primed for language acquisition, you can absolutely see patterns very similar to Nim’s in their language. I’m not an expert on apes, but I am an expert on learning and behavior and applying these principles to child and language development.
All I can say is, when you describe Nim’s tragic experiences, the patterns of his language development aren’t far-fetched or providing evidence against language acquisition in the first place. Again, the interpretation you provide is a colloquial understanding of something far more complex.
Anyone with a stronger bond with a child is going to experience more language and more language development from that child than a parade of strangers. In fact we, as a field, fully acknowledge that when we assess skills in a person we have just met, that initial assessment is going to be our least accurate and usually an underrepresentation of the person’s actual skills. When we’ve been interacting with that person extensively over 6 months and developed a relationship with them, that is when we see what they can actually do.
That being said, it’s fair to say that there’s a problem if any organism, human or otherwise, is only communicating with one other person (23:35). For one thing, how can language be effective if it’s not generalizing across environments and listeners? Occasionally, I will come across those who resist fine-tuning their loved ones’ communication skills because “we understand them,” but if the learner is not able to effectively communicate outside of their immediate family or home/school environments, their communication isn’t as useful as it could be.
But they are still communicating effectively in those environments.
There are problems with the methods, but language isn’t one of them.
Validly, at 24:17, you question the authenticity of Koko’s “last words” of “fix earth”, but only because of the simplicity of the phrases. The problem with that whole video isn’t that the language is very simple, it’s that we can’t see if there is someone behind the camera prompting her with models, questions, etc., as you point out at 25:05.
A language’s simplicity does not negate it as language. When you show the description of the video at 24:57 as a set of specific clips, it also suggests that it wasn’t meant to be anything other than exactly how it appears: Use of a famous gorilla to tug on heartstrings and bring about awareness and action to a hot topic at the time. It doesn’t seem like it was ever meant to imply that Koko was giving us a video essay all on her own.
It’s definitely misleading; I’m not justifying the method, but it’s not necessarily evidence that Koko had no functional communication, especially given the evidence you’ve presented up to this point.
The implications of your conclusions
Language depends on the function of your behavior when you use it– that is, why you are doing what you’re doing or saying what you’re saying. When you say “sit,” to a dog, you are communicating your intent for the dog to demonstrate a particular response (sitting). The dog reinforces your behavior of saying “sit” by putting its butt on the floor. Thus, you are an effective speaker, the dog is an effective listener, communication is happening, language is being utilized.
Your point of “privileging communication that is understandable to us” at 46:08 is ironic considering the points you’ve made and how you’ve made them throughout the entire video. You mention how Helen Keller’s speech was portrayed in popular culture as an example. Her communication is portrayed as far more vocal than it ever actually was, which undervalues non-vocal speech and the specific way in which she communicated with the world. And yet, you’ve constantly undervalued language and communication that doesn’t meet your standard of complexity.
Why is Helen Keller’s use of non-vocal language more valid than Koko’s?
If it’s not more complex than a 1-3-word mand, if it’s not perfectly emulating the development of syntax seen in typically developing human toddlers, if it’s not conveying complex ideas that may or may not even be present in an ape’s mind, you’re denying its status as language completely. As someone with an in-depth understanding of learning and language, your conclusion here is confusing. What exactly meets your definition of language and “human” communication?
What meets your definition of human?
Confusing criteria
I’m also confused as to why you celebrate and elevate Helen Keller’s communication when she too often communicated with a single person, who would then interpret for her to the rest of the world. What is the difference? From what you’ve presented, the only difference seems to be the complexity of the language being translated and interpreted for us. That’s a serious problem with more repercussions than you might realize. This idea that complex language = intelligence = value is exactly why disabled people are marginalized and mistreated in the first place.
Someone who’s only expressive communication (that you are able to understand) is requesting wants and needs is not only seen as “not communicating” but in a sense “not quite human.” Even without saying those words, the implication and perception is still there, that this “non-communicative” person is somehow a little less than a complete human. It’s something Terrace might have agreed with, and something his contemporary, Lovaas, definitely agreed with. It’s a dangerous thought process, even when you’re just talking about apes.
Your frustration at Koko being the inspiration for hearing people to learn sign, as opposed to the deaf community itself, at 34:29 is also both understandable and ironic. Again, you’ve spent a considerable amount of time even up to this point in the video indirectly belittling other disabled communities whose use of language is too “simple” for you to find interesting.
Defining humanity
At 36:53, after reviewing Kanzi’s use of language and ability to follow relatively complex instructions (multiple nouns, verbs, prepositions, etc.), you state “is that using language the way humans do? I don’t think it is.”
Why not? Because “lower” life forms such as dogs are able to do it? Where do you draw the line? It pulls into sharp relief, from the perspective of someone who spends a lot of time with humans with highly varied levels of communication, that you may not have much experience with non-neurotypical people. It’s not completely your fault, but it becomes a pretty big concern if you are going to make statements such as what does and doesn’t constitute being human.
You aren’t considering the weight or repercussions of your words.
Side note: Consider the heartbreaking story you share starting at 39:15. While we’re talking about prioritizing some disabled communities over others, I’d like to take this moment to draw your attention to the fact that people with disabilities are still treated in very much the same way. If you really care about the treatment of those who are unable to fully express or defend themselves, consider taking action.
Some real-life, human examples
Your criticism of Nim’s language at 15:25 is incredibly problematic. His lack of syntax and grammar as well as basic etiquette (interrupting at random) is implied as being evidence of his lack of actual language ability.
It may surprise you to learn that there are many, many humans whose communication also “stops” at around Nim’s level, including interrupting at random, and failing to demonstrate what we consider “etiquette.” I work with them every day.
First of all, neurotypical etiquette for communication varies in some ways from non-neurotypical etiquette, and forcing these norms as essential criteria for communication alienates people whose neurologies are such that these norms make communication unnecessarily difficult.
Then there’s the criticism that Nim’s communication is mostly “pragmatic” and primarily requests for desired objects and activities. How many autistic people or people with cognitive delays or disabilities have you interacted with on a daily basis? Are you going to tell me that they don’t actually communicate if their communication is limited to a few requests?
At 15:39, you mention that kids “will talk to themselves” and that Nim was never observed using signs to talk to himself. Many disabled people will not “talk” to themselves, at least not in a way you would recognize or necessarily be able to observe. Also, if your language is limited to requests (i.e. dependent on a listener), why would you use these skills when you’re alone?
We’re talking about communication here, not necessarily the whole of what makes a language. They are two different things, and Nim’s training definitely didn’t seem to progress to the point where he would have been exposed to any language that didn’t require a listener.
Maybe you didn’t intend on it, but you are essentially downplaying the language and communication abilities of disabled people because you can’t have an intricate conversation with them.
That’s where it’s hard for me, as someone with deep relationships with disabled people and who supports their language growth every day, to stay civil when I watch this video. It’s hard to hear someone describe language skills so similar to those of the learners I support with such disdain and invalidation– to deny their very humanity. Even though you’re talking about apes, you’re comparing them to humans, and providing a very narrow definition of what it is to be human.
Whether you intended to convey these ideas. or were even aware of it, your statements are hurtful and harmful to the disabled community.
There are humans in the world right now who are unable to do what Nim did, such as asking for food by arranging the same 3 words in different orders (15:42). There are humans who will say “want eat” but will not say “eat want.” Their communication is still valid.
Getting to the point
My goal in this post was not to scold you, insult you, or even downplay the amount of work you put into creating this video. Most of your points were completely understandable given the resources at your disposal. The issue was the video makes many problematic statements about communication and language and creates a stratification of legitimacy.
There are also many statements about what is and isn’t human, which is a dangerous area to tread in.
If you’re going to draw conclusions about what is and isn’t human, you need to take into consideration the entirety of humanity, or else you risk perpetuating, or even creating, marginalization. When you draw lines across the definition of human, inevitably, someone’s humanity isn’t going to make the cut.
It’s also always important to keep in mind what you don’t know. It’s going to be impossible to have a full understanding of any science without experience in it, and no amount of independent research is going to compare to formal training.
This is how we ended up in the mess we’re in today with people refusing to get vaccines or wear a mask during a deadly pandemic. It seems like an extreme example, but when you consider that the mechanisms are exactly the same, it’s not so far-fetched. People lacking any expertise, experience, or training are exploring a topic, drawing conclusions, and then spreading those conclusions as fact.
It downplays the need for expertise in complex topics, such as scientific fields, and reduces the value of actual experts in these fields. What a rando on the internet says, regardless of how many hours of research they have put in, should not carry the same weight as someone with formal education, training, and experience in an area, but it does.
As one who clearly appreciates good science, I hope you will take this as new information and reconsider your thoughts and certainly your presentation. Track down experts in the fields you’re exploring and make sure you have some modern takes on old research, especially when that research– directly or indirectly– affects other humans.
Ideas have power, wield them with care.