Monkeys With Syntax

jamie writes "The Campbell's monkey has a vocabulary with at least six types of basic call, but new research published in the PNAS claims that they combine them and string them together to communicate new meanings. (Login may be required on the NY Times site.) For example, the word for 'leopard' gets an '-oo' suffix to mean 'unseen predator.' But when that word is repeated after 'come over here,' the combination means 'Timber!' — a warning of falling trees. Scientists have known for some time that vervet monkeys have different warning calls for different predators — eagle, leopard, and snake — but unlike the Campbell's monkeys, vervets don't combine those calls to create new meanings, a key component of syntax. The researchers plan to play back recordings to the monkeys to test their theories for syntax errors."

Comments

It was the blurst of times.

[gandhi_2]

FP!

Re:It was the blurst of times.

[Vombatus]

Get enough of them together and it will be like watching a Shakespeare play

Re:It was the blurst of times.

[JustOK]

It's already better than Reality TV.

Re:It was the blurst of times.

[GNUALMAFUERTE]

On another note, the scientists have confirmed that they can pronounce the words "internet" and "nuclear" correctly.

Re:It was the blurst of times.

[causality]

Strangely enough, your sig link sucks any humor out of your triumphant 'FP!'. I'm surprised you didn't try to get the monkeys to start tea-bagging in the Name of Freedom. You could probably get them to go 'oo-oo-oo' if you presented them with an autographed copy of Sarah Palin's 'book'.

Go ahead, mod me as -1 Troll. Just make sure you mod parent as well.

I for one have no problem separating the man's political views from the humor in his post. He's entitled to them, and a link in a sig that I'd have to decide to follow does not constitute a case of him shoving those views down anyone's throat. Sorry but targeting him for that is worse than anything he could write in a blog. I actually view it as a tiny microcosm of how religious wars get started.

For what it's worth, I don't usually visit links in sigs. There are so many of them and I'd rather just read the comments. However, your comment piqued my curiosity and caused me to visit his blog. I think you gave him some free publicity.

Re:It was the blurst of times.

[Anonymous Coward]

that's nice, you would expect a scientist to be able to pronounce "internet" and "nuclear" correctly.

Monkey syntax errors aren't so bad

[istartedi]

But when they throw "exceptions", look out!

Re:Monkey syntax errors aren't so bad

[norpy]

throw new ClumpOfPooException();

Re:Monkey syntax errors aren't so bad

[Cryacin]

I pity the object that catches that.

Re:Monkey syntax errors aren't so bad

[Anonymous Coward]

Nobody will catch it. It'll just lead to a gigantic core dump.

Re:Monkey syntax errors aren't so bad

[von_rick]

Monkeys have an amazing ability to fling core dumps.

ook?

[Suchetha]

Ook!

OOK!

This is what linguists have been waiting for

[snowgirl]

There are so many people out there who have been pushing for "animals can speak!" and "we taught monkeys to use sign language!" And it's like, as a linguist, one has to pull out all sorts of jargon and details about why this isn't actual language.

Those scientists who have been studying animal language as a non-pseudoscience have been waiting for anyone to show SYNTAX in animal language. You have have 1 trillion different words in a language, and it has a finite range of expressions... meanwhile you can have 10 different words, that with the right syntax can generate an infinite range of expressions.

That's why I think this is so cool... a chance to really look at a real proto-syntax, because all human languages have a very strongly developed syntax.

Re:This is what linguists have been waiting for

[Internalist]

Those scientists who have been studying animal language as a non-pseudoscience have been waiting for anyone to show SYNTAX in animal language. You have have 1 trillion different words in a language, and it has a finite range of expressions... meanwhile you can have 10 different words, that with the right syntax can generate an infinite range of expressions.

While this is true, it's not clear to me that what's documented here is, in fact, syntax. The researcher in question (Zuberbühler) has written about this stuff before and has been much more cautious in attributing full-on linguistic properties (a search of LanguageLog will turn something up from 2006).

I'll reserve absolute judgment for when I get a chance to look at the actual paper, but this quote from NYT gives me pause: Two booms can be combined with a series of "krak-oos," with a meaning entirely different to that of either of its components. This is not (typically) how human language works...meaning is compositionally built up from bits of syntax, whereas what's described here looks more like idiom. In fact, it looks more like phonology (*maybe* morphology) to me...meaningless bits that can be put together to make meaningful bits.

What they need to do now is get a linguist in there so slice & dice the recordings, play them back to the monkeys in various reconstructed forms, and see how they react.

Also...

[...] a chance to really look at a real proto-syntax, because all human languages have a very strongly developed syntax

some would argue against the subordinate clause here (pointing at Piraha, for example), but I'm not one of those. However, it might be the case that this "syntax" has developed in parallel to human syntax from some common protolanguage (since these are monkeys and not even apes, we're talking REALLY far back), and so this may be relatively uninformative with respect to human syntax.

Re:This is what linguists have been waiting for

[HeronBlademaster]

Hey, that's my luggage combination!

Re:This is what linguists have been waiting for

[Nutria]

I must assume that you have never heard Dubya speaking.

The war is over, you won, W is gone. Now GIVE IT A FUCKING REST ALREADY!!!!

Re:This is what linguists have been waiting for

[Estanislao Mart�nez]

I don't know - a predator (danger that's out to get you) and danger from falling trees aren't entirely unrelated.

But, to borrow a bit from another comment of mine, neither are disco and psycho , right? Lots of psychos go to discos, after all.

The point is that the "krak-oo" example is at best unclear as evidence of syntax. If you want to argue that there's human language-like syntax in monkey calls, you need to find a clearer example, and preferably one that leads to a combinatorial explosion, where n calls can be combined to yield something in the order of n^2 meanings in a predictable manner.

Re:This is what linguists have been waiting for

[Nutria]

A little nervous, eh?

No, I'm sick and tired of the extended partisan hatred: Dems against Nixon, Reagan and W, Republicans against the Clintons (although it all seems to have shifted towards BHO).

The partisan vituperation against most sitting presidents in the past 40 years is also really frickin' old.

Re:This is what linguists have been waiting for

[mestar]

You have have 1 trillion different words in a language

Yet you keep using the same one.

Re:This is what linguists have been waiting for

[DynaSoar]

Those scientists who have been studying animal language as a non-pseudoscience have been waiting for anyone to show SYNTAX in animal language.

Then linguists should have been paying closer attention, and/or been been more accepting of the definition of syntax that applies to sign language: simultaneous/parallel modifiers to sign displays that alter the meanings; taken together they can be considered the primary means of development of language -- compounding components into single components with specific meanings. The novel constructions that result can be instantly recognized and meaning determined by another user of the language despite not having encountered that specific combination before. If the latter, it would only fit the Skinnerian learning model; if the former, Chomsky's 'generative grammar'.

Either millions of sign users around the world are not using language because they're not using syntax, or Koko has been using language for quite some time because she has been using syntax in constructions to modify the meanings of combinations of signs.

Penny Patterson writes: Koko uses several aspects of ASL syntax in the utterance, "You sip?". She indicates a question by maintaining eye contact, holding the sign for an extended period of time, and raising her eyebrows. She adjusts the subject of the phrase from

"I sip" to "you sip" by moving the sign away from her lips and turning it toward me, thereby altering the direction of the sign. Her pursed lips and forward-leaning posture are additional grammatical inflections.

The sign "sip" is Koko's invention, a combination of the signs "eat" (fingers to mouth) and "drink" (thumb to mouth). "Sip" can be a noun or a verb; the distinction is marked in ASL by repetition of the contact motion if the sign acts as a noun, and by a single contact if it acts as a verb. Koko regularly uses this syntactic feature of sign.

Interested readers can see Koko's sign language in action in the 1999 PBS Nature documentary, "A Conversation with Koko."

(We now return to our /. post)

None of the linguists I've worked with ever had a problem considering the modifier components of sign as syntax, particularly if they were used generatively. And they had no problem recognizing Koko's signs as such. This was at the Nation Institute on Deafness and Communications Disorders at NIH, which means two things to my mind: (1) what would you expect from linguists working at NIDCD?, but then (2) NIDCD doesn't bother with linguists who can't manage to expand their thinking beyond the restrictive serial language syntax constructions. The latter adhere to a limited form of Chomky's theory, taking generative grammar to mean people in different cultures develop different syntax/grammars evidenced by different patterns of construction (especially noun/verb ordering) specific to those cultures. These are easily refuted by (1) presenting sentences with ordering uncommon to the language used, with comprehension in intact (says Yoda understand what I'm saying you can), (2) tonal languages which have a simultaneous modifier that, while is a vocal component, performs exactly like the modifiers that are considered syntax in sign.

I studied linguistics so that I could do my neuroscience magic tricks and figure out what the brain was doing during different phases of communication, both ordered and disordered. I also happened to have been an ASL interpreter with experience in sign languages from other countries (ie. not derived from Gallaudet's French version). Through these I came, by necessity, to recognize how much of human communication is non-verbal, that and includes most of ASL 'syntax' in that it's based in kinesics, proxemics and chronemics. Having been so equipped, I found that by simply taking the non-verbal as the primary rather than the semantic "word" unit, I could deconstruct much of animal behavior as display behavior with intentional meaning. So it was of no surprise that in reading Penny Patterson's di

Re:This is what linguists have been waiting for

[theolein]

I find it hilarious that a slashdot piece on monkey syntax gets derailed into a flamefest on Dubya and Obama.

Monkey version of Timber

[Kebis]

The monkey version of "Timber!" is “Boom boom krak-oo krak-oo krak-oo". So, in monkey it's 8 sylables, and in English it's 2. No wonder humans became the dominant species, we had more time to get out of the way after the falling tree warning.

Re:Monkey version of Timber

[koxkoxkox]

"If English was good enough for Jesus, it's good enough for Texas"

PNAS

[Anonymous Coward]

Haha, they said PNAS.

You are hereby notified

[paiute]

The monkeys' lawyers just served papers on the researchers for copyright violations and the making of unauthorized reproductions of the primates' intellectual property. Spokesape Lance Link said "The researchers have submitted my clients' calls to several funding agencies. This is clearly intent to distribute my clients' intellectual properties, and we will therefore be seeking compensatory and punitive damages of one billion bananas for each call infringed upon."

Re:Backstage evolution pass?

[TheKidWho]

They already have?

Careful

[PPH]

Violate Strunk and White just once and they'll fling shit at you.

Ok, and then we can...

[FatdogHaiku]

The researchers plan to play back recordings to the monkeys to test their theories for syntax errors.

Create a very long string of recordings of unrelated calls and play them back to check for buffer overflow errors...

Here's the paper

[damn_registrars]

Straight from PNAS instead of the NYT summary:
Chimpanzees modify recruitment screams as a function of audience composition
The full text should be available to anyone in the US for free, AFAIK (and possibly to those outside the US as well). One thing you will notice on that page is that the NYT is around 2 months late summarizing that article, it was published online in PNAS back in October.

How can you test this well?

[srothroc]

I can't help but feel that you'd have to continuously use new groups of monkeys from the same community, otherwise you'd risk teaching them what you THINK certain calls mean, and they'd begin responding in that fashion...

It depends what one means by syntax...

[Estanislao Mart�nez]

I'm sure there'll be a lot of enlightening commentary about this pretty soon, but my first reaction to it is that the example cited by TFA is not clearly syntactic, in the strictest linguistic sense. Look, for example, at this quote:

"Krak" is a call that warns of leopards in the vicinity. The monkeys gave it in response to real leopards and to model leopards or leopard growls broadcast by the researchers. The monkeys can vary the call by adding the suffix "-oo": "krak-oo" seems to be a general word for predator, but one given in a special context -- when monkeys hear but do not see a predator, or when they hear the alarm calls of another species known as the Diana monkey.

The "boom-boom" call invites other monkeys to come toward the male making the sound. Two booms can be combined with a series of "krak-oos," with a meaning entirely different to that of either of its components. "Boom boom krak-oo krak-oo krak-oo" is the monkey's version of "Timber!" -- it warns of falling trees.

So, the meaning we are told for "krak-oo" is not a clear function of the meanings of "krak" and "-oo." The second paragraph makes an even more problematic claim: "boom" and "krak-oo," combined together, means something completely different than the parts.

What's the problem with this? That one of the paradigmatic properties of syntactic constructions in human language is compositionality, the principle that the meaning of an expression made of parts A and B is a function of the meanings of A and B themselves, and of the manner in which they are combined in the expression. So the meaning of Dog bites man is a function of the meanings of the words, and the way in which they are combined (so that it doesn't mean the same thing as Man bites dog).

This doesn't mean that there isn't no non-compositionality in human language, or even in syntax, but rather that compositionality is typical of syntax, and noncompositionality is typical of morphology. There's in fact tons of noncompositionality in human language, but it's hard to argue that monkeys have a semblance of human language unless you can clearly argue that the meanings of the subparts of the complex calls combine compositionally.

Re:It depends what one means by syntax...

[mr_matticus]

my first reaction to it is that the example cited by TFA is not clearly syntactic, in the strictest linguistic sense.

And in no small part, that's because you're analyzing it as a human language. You go on to suggest that the examples cited tend to indicate morphology. And if this were an elementary study of a phenomenon in a more sophisticated language, I would agree. However, two points:

1. Morphology is fundamentally syntax (underlying mathematics of structure), it's just the syntax with the word, rather than the assembly of words.
2. While morphology is unquestionably more basic than syntax, as a lexicon of words is (we assume) a precursor to the emergence of a language, and though morphology eventually becomes a distinct field in highly developed languages, the initial emergence of syntax (and accordingly, sentences) from morphology is not a black and white line.

Words grow longer and more complicated, and thus carry more and more meaning, until eventually a different structure, a grammar, has to replace a word-based method of communication. The question that this research seeks to answer is whether there is, in fact, a grammar within this language.

The second paragraph makes an even more problematic claim: "boom" and "krak-oo," combined together, means something completely different than the parts.

What's the problem with this? That one of the paradigmatic properties of syntactic constructions in human language is compositionality, the principle that the meaning of an expression made of parts A and B is a function of the meanings of A and B themselves, and of the manner in which they are combined in the expression.

The claim is not problematic and does not necessarily indicate non-compositionality. Again, I believe your perspective is influenced by a study of highly evolved human languages. Consider it more like a machine language and you begin to see things slightly differently.

If you only have a limited range of sounds (as monkeys do, compared to humans) and if you only have a limited storage capacity (again, as monkeys do, compared to humans), then basic syntax enables a great deal of added complexity for relatively no cost. You can recycle the sounds without creating untenably long morphemes.

It is not necessarily that "boom" and "krak-oo" when combined mean something different than the parts, but rather that these primates have multiple working definitions for each of their words, and rather than a contextual association, which is rather advanced cognition and language, the different definition is triggered by the syntactical position of the word.

There's in fact tons of noncompositionality in human language, but it's hard to argue that monkeys have a semblance of human language unless you can clearly argue that the meanings of the subparts of the complex calls combine compositionally

Agreed, but the issue here is a question of whether we fully understand the meanings attached to their sounds. If you assume that one of their morphemes has exactly one fixed definition regardless of combination, your point is valid.

But if the meaning shifts based on sequence, allowing each morpheme to be associated with multiple lexical entries depending on its grammatical position within a basic "sentence", then that is indeed evidence of a much more sophisticated language than is commonly assumed.

Because we have no experience with the development of any human languages at this level, it's hard to say which comes first. I'm of the belief that phonology blurs into morphology, which then blurs into syntax. Is a diphthong a phoneme trying to be a morpheme? Is "boom boom krak-oo [...]" an overextended morpheme, or has it spilled over into a proto-sentence? What is the line between word and sentence, morphology and syntax?

You're assuming the answer to the question they're asking, and thus begging the question. If the "words" always have one meaning, then it's not much of a syntax--but the research aims to show whether those sounds always have the same meaning or if it does vary with composition.

Re:Backstage evolution pass?

[Nutria]

will become more complex. given a few thousand years

I think you're off by a few orders of magnitude.

Vicious circle

[A nonymous Coward]

That's how it began!

Angry monkeys

[nephridium]

They want to play back intentionally malformed phrases? I guess they'll need to prepare to the Campbell monkeys' equivalent of "What did you just call my mother?" ;)

Which monkeys?

[gmuslera]

In humans language is something cultural, even syntax is something you learn from others, is not builtin. If is the same on monkeys maybe the ones from a region have a different syntax or semantics than others from far away.

Re: Which monkeys?

[Black Parrot]

In humans language is something cultural, even syntax is something you learn from others, is not builtin. If is the same on monkeys maybe the ones from a region have a different syntax or semantics than others from far away.

But the capability seems to be at least partly built-in.

The big debate is between the "speech is special" crowd, who think the built-in stuff is only good for language and only present in humans, vs. those who think language is to a big extent based on more general cognitive capabilities.

I'm in the latter group, so I find this utterly unsurprising. The discoveries of the past few decades should have disabused everyone by now of the notion that human cognition is of an utterly different caliber than animal cognition.

Still, there are those who will contest this report vehemently.

Measurable results

[Pike]

I guess they'll know they had a "syntax error" if the monkey fails to understand the warning and gets killed by the falling tree.

Re:I second that...

[dziban303]

Okay, okay, you're both linguists. But are you cunning?

Re:Meaning is not a key component of syntax.

[mr_matticus]

Actually, "meaning" isn't just limited to sense and reference (semantics).

Meaning, that is, syntactic meaning, is a key component of syntax. Without meaning, syntax can't exist.

Knowing that a repeating pattern has a logical definitional rule behind it is a key element of meaning. If I say the word "mine" to you, without syntax, you have no idea of the semantic meaning. Is it a verb? An object? A noun? If it is a noun, does it refer to the kind for digging or the kind for exploding? Syntax plays a huge role in meaning.

Consider that the monkeys have a semantic inventory of distinct sounds A , B, and C. Semantically, they have three concepts and no more--because they lack syntax. With a simple syntactic structure, the sounds get new meanings because sequence suddenly informs meaning.

Without syntax, words can only have one meaning. As the article argues and as the sentence describes, the fact that position changes the meaning of sounds is key evidence of the use of syntax in the language. If semantic meaning were unaffected by sequence, that would be evidence of the absence of syntax.

Semantics cannot be divorced from phonology and syntax in oral language. Phonological meaning plus syntactic meaning is fundamentally semantic meaning. More advanced languages have more complicated systems of context and idiom that add layers onto this. But the basic point remains that meaning is certainly an element of syntax.

Is the word for "leopard" really "tree"?

[jonaskoelker]

I have an alternative hypothesis to the one presented in the summary. (Haven't RTFA, fwiw).

I propose that the word for "leopard" really is the word for "tree". Why?

Well, suppose the suffix "-oo" means "get up into", and the "come[s] over here" part refers to the trees, not the monkeys.

Observe that getting up in the trees is a good way to avoid leopards, and that when you yell "Timber!", it's because trees are coming your way. That way, what the monkeys say should still produce the same behaviour as with the summary's language, but the words seem to have more stable, consistent meanings.

If this were not the case, one might expect the monkeys to say "leopard + comes-over-here" and "tree + comes-over-here", or something similarly systematic.

Also, observe how (human) children apply simple and logical (but sometimes wrong) rules to construct sentence patterns; something like the thought "hey, the expression "you're going down" must mean that relative to you, I'm going up. Yeah! "I'm going up, you [word]!"". Key point being: simple rules, a consistent inverse relationship between up and down. Wouldn't it make sense that monkeys have a similarly simple and consistent language?

Note also that the monkeys signal different behaviours when they observe or suspect eagles and snakes. The word for "eagle" might really mean "duck and cover", and the word for "snake" might really mean "stand really still, on your toes, and look down", since that is how they handle these different kinds of predators.

It might also be more effective to say "get up in the trees" and "get up in the trees" versus "there's a leopard coming" and "there's a [different non-climber] coming"; that way, you can get away with a smaller vocabulary, a more restricted vocal apparatus (since you don't need many different sounds), etc. Just cheaper overall.

My cents tw-oo ;-)

Re:Monkeys with syntax

[ColdWetDog]

Interesting... Except for the syntax thing, you'd have Slashdot editors.

Monkey see, Monkey dupe?

Re:Meaning is not a key component of syntax.

[ArsenneLupin]

No, because you can have meaning without syntax.

... and you can also have syntax without meaning. Just any regular expression defines a grammar or syntax. That doesn't mean that any string matching that regular expression has a meaning.