Campbell's monkeys: incipient syntax?

« previous post | next post »

Lest anyone should think that the Animal Communication desk at Language Log Plaza is asleep, let me just note that we have indeed taken note of recent reports about how Campbell's monkeys have complex syntax. Among these Ivory Coast primates, according to one report: "males have a repertory of six types of alert calls (Boom, Krak, Hok, Hok-oo, Krak-oo, Wak-oo) but only rarely use them in isolation, preferring to produce long vocal sequences of an average of 25 successive calls (each sequence being made up of 1 to 4 types of different calls). Furthermore, Campbell's monkeys combine calls in order to convey different messages. By modifying a call sequence or the order of calls within a sequence, the messages are changed, and can relay precise information about the nature of the danger…" You read, you decide. The original research paper has been published in Proceedings of the National Academy of Sciences, but I have not yet seen it. At present Language Log has nothing to say, except to utter a long series of about 25 successive alert calls warning you that anything concretely tied to present dangers apparent in the immediate spatiotemporal environment cannot bear a very strong relation to natural use of a human language. The things we say are not just long sequences of "Watch out!", "Fore!", "Timber!", "Stop thief!", "Hey!", "Ouch!", and so on. (At least, not for me; not on a good day.)


  1. Daniel Midgley said,

    December 27, 2009 @ 9:38 pm

    Not really syntax. Probably not even morphology, since the 'oo' that's being added doesn't seem to mean a specific thing.

    About the most we can say is that the monkeys seem to be using signs in an arbitrary way, which is kind of languagey, because our signs are arbitrary too. (So is the waggle dance of the European honeybee.) But not syntax.

  2. John Roth said,

    December 27, 2009 @ 9:53 pm

    Well. If you're looking at it from a viewpoint of language origins, rather than similarity to modern human languages, I think the picture is very different.

    As I understand it, the usual animal calls are one to one – one call to one specific environmental cue that's worth notifying others about. What I get from the public reports so far is that they've got four specific calls and a couple of dozen situations they want to communicate.

    Other than that, since Campbell's Monkeys aren't on the direct line of descent, and since other animals that are on the line of descent don't do it, it looks more like a curiosity that may have nothing to do with the origins of human communication. It would still be interesting to see a tensor diffusion MRI study to see if there's an unusual linkage somewhere in their brains.

    I wouldn't think that the honeybee's waggle dance is all that good a comparison: that has a highly specific semantics that describes a location, while this seems to be describing a wide variety of threat situations.

  3. Nathan Myers said,

    December 28, 2009 @ 2:00 am

    Would it kill you guys to suggest that something involving animal communication might yet become linguistically interesting once more information comes to light? Or does posting about it without outright derision mean just that? (Does kremlinology count as linguistics?)


    [Would it kill you, Nathan, to say something that is not couched in the form of a rhetorical question? —GKP]

    It might be hard to distinguish a transcript of these monkeys' vocalizations from that of a typical sportscaster's, e.g. at a horse race or boxing match, in a language you don't know. I don't know if that would say more about the sportscasters or the monkeys.

    [Thank you. —GKP]

  4. Daniel Midgley said,

    December 28, 2009 @ 2:31 am

    So, Nathan, you're suggesting that the monkeys are using a language, but we just don't understand it? Or what?

    Maybe you haven't yet read story after story of overblown outrageous claims involving animals and language. I find that the OP uses about the right level of skepticism.

  5. Nathan Myers said,

    December 28, 2009 @ 3:14 am

    It's easy to find overblown outrageous claims about most subjects. No, I don't think these monkeys have language, as such, and I don't find the question very interesting. I do think it possible that there is something to be learned, from their vocal behavior, about the origins of human language — even if what they're doing doesn't, in fact, match how our own language facility originated. We won't know if nobody looks, and shouldn't linguists be involved in looking? Even if this behavior turns out not to seem much like the beginnings of language as humans perform it, who is better equipped to study rudimentary examples of meaning conveyed by sequential, constructive vocalizations than linguists?

    Until human language is precisely placed among a broader class of animal behaviors (that for practical, historical reasons won't ever be classed as language), we won't really understand what it is.

  6. Philip Spaelti said,

    December 28, 2009 @ 6:09 am

    The following article from the NYT provides a bit more specific detail:

  7. John said,

    December 28, 2009 @ 1:09 pm

    Seriously, how hard would it have been for Science Daily or the NYT to provide a link to the actual abstract? I'll excuse GKP, since he's clearly admitting he hasn't looked at the paper. (And what do I want– a refund??) (And watch, my link will probably be broken…)

    Do science reporters even read the papers they're reporting on, or do they just regurgitate the press release?

  8. SG said,

    December 28, 2009 @ 4:45 pm

    Campbell's monkeys concatenate vocalizations into context-specific call sequences. 2009. Karim Ouattaraa, Alban Lemassona adn Klaus Zuberbühlerd/

    Published online before print December 9, 2009, doi: 10.1073/pnas.0908118106

    Primate vocal behavior is often considered irrelevant in modeling human language evolution, mainly because of the caller's limited vocal control and apparent lack of intentional signaling. Here, we present the results of a long-term study on Campbell's monkeys, which has revealed an unrivaled degree of vocal complexity. Adult males produced six different loud call types, which they combined into various sequences in highly context-specific ways. We found stereotyped sequences that were strongly associated with cohesion and travel, falling trees, neighboring groups, nonpredatory animals, unspecific predatory threat, and specific predator classes. Within the responses to predators, we found that crowned eagles triggered four and leopards three different sequences, depending on how the caller learned about their presence. Callers followed a number of principles when concatenating sequences, such as nonrandom transition probabilities of call types, addition of specific calls into an existing sequence to form a different one, or recombination of two sequences to form a third one. We conclude that these primates have overcome some of the constraints of limited vocal control by combinatorial organization. As the different sequences were so tightly linked to specific external events, the Campbell's monkey call system may be the most complex example of ‘proto-syntax' in animal communication known to date.

  9. SG said,

    December 28, 2009 @ 4:50 pm

    'As the different sequences were so tightly linked to specific external events, the Campbell's monkey call system may be the most complex example of ‘proto-syntax' in animal communication known to date.'

    — Of course, they meant 'non-human animal'.

  10. Diane said,

    December 29, 2009 @ 2:57 am


    Well, that's a very confusing correction…it seems to me that if you say "non-human animal" here it suggests that there are some examples from human language of 'proto-syntax'. Is that true? I thought all known languages had full-blown syntax. Perhaps some pidgins don't? Or maybe you are counting baby talk?

  11. Darryl McAdams said,

    December 29, 2009 @ 7:28 pm

    See, what I don't get is why anyone would even thing this is remotely syntax like. It's completely non-compositional in terms of its "meaning". John Roth above says that animal communication is usually one-to-one pairing one meaning with one form, but that's certainly not at all true, there's plenty of phonetic overlap in, say, wolf howls, such as voicing, or what would seem like shortened pre-reduplication, etc. Bird chirps are surely even more redundant, using a single note or sequence of notes in a variety of places. Would we be really all that surprised to find two bird calls from one species, one of which goes chirp chirp chirp cheep and another that goes chirp cheep chirp cheep? No, and we wouldn't be tempted to say that because both end in "chirp cheep" that this is some sort of suffix demonstrating proto-syntax. No, what we would say is that this is the inevitably duplication of acoustics when the production inventory is relatively limited. So why are we even entertaining the idea that taking "krak" (which means "oh shit a leopard!") and turning it into "krakoo" (which means "oh no i hear something!") qualifies at all as syntax? "krakoo" isn't about leopards, it's about unseen potential predators. But hey, at least it overlaps in being about predators right? Except ALL of the calls these monkeys make are about predators.

    Here's a quote directly from the NYTimes article:

    — 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 they can combine a series of sounds into a larger series of sounds to get a meaning thats completely different. Big deal. That's not syntax, that's phonology.

  12. Darryl McAdams said,

    December 29, 2009 @ 7:46 pm

    Just to add a minor comment to this, bee dances are considerably more compositional than this crap. We know very precisely how the different aspects of bee-dance are related to different aspects of the message. Not only are the researchers here denigrating actual linguistic research by trivializing the subject matter, they're denigrating other animal communication research as well, research which is far more relevant than theirs is to understanding how compositional systems develop.

  13. Nathan Myers said,

    December 29, 2009 @ 11:47 pm

    Darryl: You seem to be saying these monkeys don't have language. I think we knew that. Nobody is trying to decide whether it's Niger-Congo or Chadic.

    A better question here is whether there's anything in this behavior a linguist could be better equipped than your typical primate biologist to shed light on. The only answer the evidence seems to support is that we don't know yet. What's certain is that we won't know before one looks.

  14. Jesús Sanchis said,

    December 30, 2009 @ 5:25 am

    Can Campbell's monkeys differentiate between a) imperative and b) narrative/description?

    a) Run from the approaching leopard!
    b) A leopard is approaching!

    Maybe there's no answer to the question, but it would interesting to research into this, as it could tell us something about the origin of human language and grammar.

  15. Darryl McAdams said,

    December 30, 2009 @ 7:30 am

    Nathan: And you seem to be trolling.

  16. Mark Liberman said,

    December 30, 2009 @ 11:59 am

    I'd like to suggest that those who want to comment on this topic should read the original article, "Campbell's monkeys concatenate vocalizations into context-specific call sequences", PNAS 2009. If you don't have access to an institutional subscription, write to me and I'll send you a copy. It's a serious and interesting piece of work, and it deserves better than an uninformed back-and-forth that is supported only (if at all) by references to mass-media descriptions.

    In fact, I plan to enforce this suggestion from here on out. Comments that don't contain evidence of knowledge of the article's contents will probably be deleted, in the absence of redeeming qualities of some other kind.

    You might also want to take a look at Karim Outtara et al., "The alarm call system of female Campbell's monkeys", Animal Behaviour 78(1) 2009; Wolters & Zuberbühler, "Mixed-species associations of Dian and Cambell's Monkeys", Behaviour 140(3), 2003; Klaus Zuberbühler, "A syntactic rule in forest monkey communication", Animal Behaviour 63(2) 2002; Alban Lemasson, Martine Hausberger, Klaus Zuberbühler, "Socially Meaningful Vocal Plasticity in Adult Campbell's Monkeys", Journal of Comparative Psychology 119(2) 2005; etc.

  17. Forrest said,

    December 30, 2009 @ 8:28 pm

    Hok Wak-oo Krak Boom Boom Hok-oo Krak Wak-oo Hok Boom Krak-oo Krak Hok-oo Boom Krak Krak-oo Wak-oo Hok-oo Krak Hok Krak-oo Boom Hok-oo Boom Hok-oo!!!

    For whatever it's worth – almost certainly nothing – it feels redundant to past 25 of these 'words' to match the length of an average call.

  18. Darryl McAdams said,

    January 1, 2010 @ 7:24 pm

    Having read the article in question, and their previous article on the suffixing aspect, I would like to provide summaries of their findings:

    From their first article,

    — Predators never trigger "boom" calls, "boom" only draws other monkeys to the caller.
    — General alerts were "krak-oo".
    — Presence-of-Leopard alerts were "krak".
    — Presence-of-Eagle alerts were "wak-oo".
    — Visual eagle-model alerts were "hok" and "hok-oo".

    There's no sign of any overall pattern in this alone, contrary to their claim in the follow-up paper that "Adding the suffix 'oo' to krak or hok calls altered these calls' meanings in predictable ways.".

    The follow-up paper also goes on to elaborate how sequences of calls affect "meaning", as summarized:

    — Repetition of an item doesn't change it's meaning ("krak", "krak krak", "krak krak krak", etc. all signal the presence of leopards, for instance).
    — New calls can be derived from sequencing basic calls with unrelated meanings ("boom krak-oo" signals falling wood, "boom krak-oo hok-oo" signals presence of neighbors).
    — Extra sequences exist for the same meanings ("krak krak-oo" signals leopards, "wak-oo hok-oo" signals eagles).
    — "boom" at the beginning of a sequence are indicative of a non-predatory context
    — "Predatory" calls were repeated more when the threat was greater.

    Unfortunately, the authors make some nonsensical claims about their own data. Quoting, with minor paraphrasing for formatting reasons:

    1) "these call combinations were not random, but the product of a number of principles, which governed how semantic content was obtained"
    2) "adding the suffix 'oo' to krak (krak -> krak-oo) or hok (hok -> hok-oo) calls altered these calls' meanings in predictable ways"
    3) "callers combined two meaningful sequences into a more complex one with a different meaning (e.g. boom + krak-oo = falling wood)"
    4) "callers added meaningless calls to an already meaningful sequence and, in doing so, changed its meaning (e.g. boom + krak-oo = leopard; wak-oo + krak-oo = crowned eagle)"
    5) "boom calls, indicative of a nonpredation context, always preceded any other call types"

    2 was discussed above, but it's worth pointing out their repetition of this false claim.

    3 and 5 are consistent, but 4 and 5 are not (boom is supposed to indicate a nonpredation context, so how can book + krak-oo indicate a leopard?).

    4 claims that boom + krak-oo is composed of atleast one meaningless call, but we know this is not true since boom and krak-oo mean "come here" and "leopard", respectively.

    4 also claims that wak-oo + krak-oo contains a one meaningful and one meaningless call, which is true (wak-oo means nothing, and krak-oo means leopard), and that the sequence changes the meaning of the meaningful call, which is a bit silly since krak-oo by itself means leopard but somehow we're supposed to believe that this is "changed" (but not completely overridden) into "eagle". The only consistent meaning is "predation" which is common to _all_ sequences that done start with boom, and so couldn't even be considered part of the meaning of krak-oo that's preserved by the change. It seems that wak-oo + krak-oo is simply completely unrelated to krak-oo.

    1 is just completely untrue given the "composition principles" they list after it: 2 isn't actually a composition principle, 3 derives unpredictable meanings, 4 derives unpredictable meanings. The only "composition principle" that actually signals anything is sticking boom at the beginning of a sequence, which turns otherwise necessarily predation-related sequences into potentially non-predation-related sequences.

    My conclusion is that these authors are a bunch of nitwits trying to make their findings sound more impressive than they actually are. I expect press people to BS some stuff due to misunderstanding, but this is ridiculous. The authors _themselves_ are BSing in their discussion; the NYTimes articles only repeated their BS. For once, science journalism successfully reported on the content (however BS-filled it was) of a scientific finding.

  19. SG said,

    January 1, 2010 @ 9:09 pm

    @ Diane, good point if 'proto-syntax' is really not used at all by humans — although presumably 'proto-syntax' originated before syntax in pre-modern humans or in hominids (presumably syntax didn't arrive fully formed in our early history), in which case I think that my suggested clarification would be awkward but technically valid.

  20. Paul Kay said,

    January 24, 2010 @ 1:14 pm

    "Do science reporters even read the papers they're reporting on, or do they just regurgitate the press release?"

    Often, sadly the latter. A discouraging experience I've had too many times is that of unsuccessfully urging a reporter to read the paper being reported on.

RSS feed for comments on this post