Previous Entry Share Next Entry
Conversations with toddlers^H^H^H^H^H^H^H^Hprimates
chap
nmg

ias recounted a conversation that marklesuk had with Thea, on the subject of apes and monkeys, which got me browsing aimlessly through Wikipedia. I came across an article on Yerkish, an artificial language designed in the 1970s for use by non-human primates, principally chimpanzees. Yerkish has a vocabulary of symbols (lexigrams) which are used to label buttons on a keyboard, and has a relatively complex grammar. The paper referenced in the Wikipedia article (a scan of a microfiche version of a typed paper original) has an appendix of conversations with Lana (the first chimpanzee to have learned Yerkish) that were so close to conversations that I've had with the garklet that I think they merit reposting.

One of the first of Lana's spontaneous 'generalisations' concerned the lexigram NO. She had learned the use of this lexigram in one specific context, i.e. in answers to questions such as: ? BANANA NAME-OF THIS, when the object ostensively indicated to her was, for instance, her blanket. One morning, Tim Gill, to whose ingenious devising of training situations the project owes a great deal of its success, entered the room with a bowl of banana slices. As he moved around the corner of Lana's cubicle in order to fill the dispenser that responds to the keyboard message PLEASE MACHINE GIVE PIECE OF BANANA, he popped a banana slice into his mouth. Seeing this, Lana adopted a threatening posture and hooted angrily. Then, suddenly, she ran to the keyboard and, three times in succession, vigorously pressed the key bearing the lexigram NO.

Very close to things that the garklet has done. Nice to know that this sort of behaviour is not limited to the juveniles of H. sapiens sapiens.

Chimpanzee 1, Toddler 1

On the preceding, days Lana had learned the lexigrams for a bowl and a metal can, BOWL and CAN. This had been accomplished by first using objects whose names were already known to her, putting an M&M candy inside them, and asking her: ? WHAT NAME-OF THIS. On May 5th she reliably replied with the correct lexigram when the reward was placed in the bowl or in the can. The next morning Tim came in with the bowl, the can, and a cardboard box. While Lana was watching, he put an M&M candy in the box, and the following exchange took place:

Lana:
? TIM GIVE LANA THIS CAN.
Tim:
YES. (Tim gives her the empty can, which she at once discards)
Lana:
? TIM GIVE LANA THIS CAN.
Tim:
NO CAN.
Lana:
? TIM GIVE LANA THIS BOWL.
Tim:
YES. (Tim gives her the empty bowl)
Lana:
? SHELLEY (Sentence unfinished)
Tim:
NO SHELLEY. (Shelley, another technician who worked with Lana, is not present)
Lana:
? TIM GIVE LANA THIS BOWL. (Before Tim can answer, Lana goes on)
Lana:
? TIM GIVE LANA NAME-OF THIS. (A spontaneous generalisation of GIVE, not foreseen by the grammar, since NAME-OF had not been classified as a possible object of GIVE!)
Tim:
BOX NAME-OF THIS .
Lana:
YES. (Short pause, and then)
Lana:
? TIM GIVE LANA THIS BOX.
Tim:
YES. (Tim gives it to her, she rips it open and eats the M&M)

Lana has the garklet beaten here. He's never explicitly asked for the name of an object, but tends to use the catch-all word 'baya' instead; the meaning of 'baya' has changed over the last year, from water (which is now something like 'glok', or sometimes 'drink') to nappy (which he can also say, infuriatingly). 'Baya' also tends to crop up when he doesn't know the name of a thing, so it possibly also means thing.

Chimpanzee 2, Toddler 1

Tim:
? LANA WANT APPLE. (no apple is in sight)
Lana:
YES. (Tim leaves the room and, after a moment, returns with an apple)
Lana:
YOU GIVE THIS TO LANA.
Tim:
? GIVE WHAT TO LANA.
Lana:
YOU GIVE THIS WHICH-IS RED. (since there is no sentence marker, this is an indicative statement and neither a request nor a question)
Tim:
? THIS. (Tim holds up a red piece of plastic)
Lana:
? YOU GIVE THIS APPLE TO LANA.
Tim:
YES. (whereupon she received the apple)

Again, the chimp wins out. Trying this on the garklet would lead to a minor meltdown and some shouting of 'my bapple! my bapple!'. He's not so good at understanding linguistic game-playing, nor does he restate his questions in different terms when we fail to understand. On the other hand, he almost always phrases requests as indicative statements ('more cake').

Chimpanzee 3, Toddler 1


  • 1
Do keep us up to date on when the toddler exceeds the chimpanzee in these things!

William Cooper Aged Two used the word "ibbuk" for "give me" or "want", with no word for what it was he actually wanted. We use it now for "could you pass me my book please which I have inadvertently left somewhere where you can reach it and I can't?". It is accompanied by pointing with a grasping hand, and a screwed up face.

Could it merely be that the garklet is just better at training humans to respond to overt emotional disapproval? :-D

Possibly, but that would suggest that he's got a fairly sophisticated theory of mind. He understands that when he does X, we do Y, but he doesn't always understand *why* we do Y.

Knowing that X does Y is sufficient I'd say. I've seen dogs train humans on the basis that X does Y and they are not nearly as bright as chimps or gerkels. I think you need to experiment more.

Er...that'd be 'garklets' obviously. Gerkels are something else. Sorry, been reading another subject this evening and got confused.

True, there's Pavlovian conditioning, which doesn't require us to formulate a reason for changing our behaviour (and therefore doesn't require him to have a theory of mind).

God, that's depressing. I'm on a par with a dog that dribbles when they ring a bell.

Yeah, both human parents and chimp researchers have tended to over-estimate the extent to which there's something complex and exciting going on in the subject's head. See also dog owners.

Its very tricky in primate research because so much of what you're doing depends on bribing them with food treats. Lana does something, Lana gets food. I've talked with some people (including Stephen Harnad of course, but also linguists) who dismiss all of the results as wishful thinking - one "conversation" a week extrapolated from thousands of random utterances, or minor variations of taught patterns (filling a new word into an existing pattern) over-analysed by experimenters.

Still - you should get to see some cool experimental results reproduced in front of your eyes with the Garklet though. Apparently they usually learn some common irregular verbs and then after a short while unlearn them in favour of regular productions. So "I ran" and "We went" become "I runned" and "We goed". Parents are often dismayed and blame other children (and thus implicitly, their parents) -- as if somewhere there's a parent teaching their child to say "We goed" -- rather than being impressed at the fact that their child has just learned a rule of grammatical tense which they can use to create new words that other English speakers find comprehensible.

Parents are often dismayed and blame other children (and thus implicitly, their parents) -- as if somewhere there's a parent teaching their child to say "We goed" -- rather than being impressed at the fact that their child has just learned a rule of grammatical tense which they can use to create new words that other English speakers find comprehensible.

Ah, logical errors ftw! Language rocks.

He does some of this already; he's generalised 'baby' as an adjective meaning a small instance of some class of objects, for example 'baby marmite'.

  • 1
?

Log in

No account? Create an account