Salmoneus wrote:Dolphin language in particular seems extremely expressive, but almost none of it, other than their system of personal names, has been satisfactorily deciphered.
That "system of personal names" has, afaik, also not been "deciphered". See e.g. here
No offence, but that's an unusually stupid/fatuous article from languagelog. Sure, it's vacuously true that names in dolphin language do not, so far as we are aware, have exactly the same syntax as explored in pages 515-523 of the Cambridge Grammar of the English Language. Because, you know, they're dolphins.
But the existence of signature whistles seems uncontroversial - although one guy has lead the research, for 50 years, because there aren't that many dolphin language specialists in general, that doesn't make him a crackpot, and many other scientists have accepted his ideas, and been published in respectable peer-reviewed journals. Pullum, on the other hand, has no idea what he's talking about, so I'll take the experts over him any day. Signature whistles are constant vocal expressions that uniquely identify specific individuals, and are recognised as doing such by other individuals - therefore they are by definition names. To say otherwise is just philosophically ignorant. If they are not used exactly as humans use names, that's interesting, but it doesn't stop them being names. And Pullum's counterargument is that dolphins repeat their name if you say it to them, and humans don't? That's just silly. Dolphins aren't humans.
[it's also not true. If you go up to Geoffrey Pullum and say "hi, I'm Geoffrey Pullum!", it's actually quite likely that he WILL say something like "oh, I'm Geoffrey Pullum too!" (or "no, I'M Geoffrey Pullum!" depending on whether he believes you). Likewise, since dolphin names seem primarily used in introductions, that experiment is essentially going "(I'm) XYZ!" and the dolphin shouts back "(but I'm) XYZ!?", which I think seems like a pretty human response, actually.]
Other species (primates, elephants, birds) may have more complex grammar than we realise, due to lack of understanding of these communication systems.
Even though me may not "understand" their communication systems, it is pretty easy to test whether there is any complexity to a communication system, by checking for repeated elements etc. So far, these elements are absent, so to state that they would have a "more complex grammar than we realise" is imho nonsense.
No, it's not easy to test by checking for repeated elements etc. For one thing, many animals DO communicate in streams that incorporate many repeated elements, including elements that incorporate other repeated elements in various combinations. That's uncontroversial.
The two questions are:
a) are the same patterns of repetition and incorporation used across multiple utterances, much more than would be the case through chance?
b) do these patterns bear any consistent relationship to any hypothetical meaning?
These questions are harder to answer, particularly because it's almost impossible to elicit specific utterances from animals in the wild - so much of the time we've no idea what the animal is "trying to say", so we can't compare the hypothetical meaning to the string to deduce any consistent relationships. Sometimes we can have a pretty good guess - for instance, although songbirds have "grammar" almost (if not in some cases as) complicated as human grammar, with repetition and embedding and discenable generative patterns, we're reasonably confident that they're not actually "saying" much at all, and that their songs are more analogous to music (which also, of course, is grammatical) than to speech, because of the restricted and repetitive circumstances of their utterance - they mostly seem to just be saying "I'm here, this tree is taken!" and "hey pretty girl, look at me!", and the great complexity of some of their songs doesn't seem to represent any specific function beyond that.
Other animals, though, do seem to communicate in more complicated and hence less consistent ways, which makes it really hard to guess what 'meaning' (if any) they are trying to convey, and hence if they do so grammatically or not. Most likely, they don't - but that's a guess, and a guess heavily reliant upon attempts to teach animals human languages, which may be entirely unrepresentative.
A more fundamental problem, however, is that any attempt to record what words an animal says relies upon assumptions about what 'words' are that are increasingly being challenged. It seems that many of the smarter animals - dolphins, elephants, chimps - use multiple channels of information simultaneously. Dolphins use whistles, but probably also clicks (but not all clicks, because some are obviously echolocative), and probably also physical gestures (but not all their gestures, because some are just random movements). Studies of chimps in the wild suggest that their "words" rely as heavily on things like posture and orientation as on sound. But again, of course, a lot of their posturing and orienting and gesturing and even some of their vocalising is clearly NOT linguistic in intent. So it's hard to know what type of 'phonemes' to be looking for, and then hard to work out which are actually phonemes and which are non-linguistic 'interference' in that channel, and then hard to work out what the 'phonemes' specifically are (which postures are significantly "the same" despite having obvious differences, and which are "different" despite being very similar?).
[of course, multi-channel communication could just be multiple simple communications at once. OR, multi-channel communication could be a single complex communication in which the data in each channel interacts. It's very hard to tell which is which without a vast amount of exhaustive data to analyse]
[non-deaf humans, of course, use essentially single-channel communication, in the form of a stream of vocalisations - gestures and expressions and motions are at most expressive augmentations. But there is no logical reason why all languages must be single-channel.]
Wild animal studies of these things are clearly in the very primitive stage, and nowhere near extensive or sophisticated enough to rule out syntactic behaviour in non-humans. In many cases, we've only just begun to even start looking at what may be relevent channels (in the case of elephants, for instance, their use of subsonic vocalisations has been badly understudied, in part due to lack of availability of the relevent tech in the field). [now, 'syntax' needn't mean syntax as complex as human syntax, of course!]
However, as they're being taught the language of a different species, we don't know how much of that is an inability to use grammar, and how much is an inability to understand alien (i.e. human) grammar.
Time and time again, behaviour that seemed exclusively human has been found in animals, and especially our closest relatives turned out to be capable of such behaviour. Given their close relationship, common descent, rather than parallel evolution is the logical hypothesis here. The fact that they can understand and use abstract concepts, but cannot use any kind of grammar, seems to indicate that they do not posses the ability to understand or use that
. To state that just maybe, they have some kind of "alien" grammar of their own, that we have never discovered even though they are one of the most studied group of animals, is imho very, very wishful thinking.
And to take such primitive and sparse data and completely rule out the possibility of any form of syntactic behaviour ever being discovered is in my opinion even more wishful thinking.
The key thing here is that we don't really know whether 'language' is instinctual (i.e. a behaviour inherent to creatures with our genes) or whether 'language' is a cultural phenomenon. It's almost impossible to test, because almost no human groups have ever developed in complete, non-linguistic isolation
I would assume you are aware of the case of Nigaraguan sign language
? You do not consider this language genesis? Or do you consider them to not be "isolated" enough? Personally I think this makes a very strong case to build a model upon. Also, it kinda mirrors the way creoles originate after a stage of pidgin.
I don't consider it isolated enough. The children in that case had exposure to home sign systems, fingerspelling, and study of spoken linguistic behaviours through lipreading of Spanish - they were being taught intensively. I accept that they did not understand the specific grammar of Spanish, and that ISN is more or less sui generis ab origine. But it is possible that although they did not learn any other language
before creating ISN, they may have learned the concept of language
The analogy here would be to something like the Cherokee Syllabary. The story goes that the guy who invented that was illiterate - he was not able to read any other script. But clearly the idea
of a script was inspired by his observation of the reading-behaviour of Europeans (and many symbols are borrowed from European scripts, in random ways). So when we talk about the origin of writing, Cherokee writing is generally not considered an independent invention of writing. Indeed, it's probable that there have been only two or three independent inventions of writing, though there are many more scripts that have been invented without specific descent from a predecessor script.
Actually, that's an interesting subject. The fact that we know that writing has
been invented independently, repeatedly, shows that writing is, as it were, something hardwired into the brain - something that repeatedly arises in the right circumstances. There is, as it were, a writing faculty. But the fact that it has been invented so rarely, and yet has spread so quickly by imitation once invented, shows that it is primarily a cultural practice: there is a latent capacity that in almost all cases has had to be triggered by external cultural influence. So it's entirely possible that a similar 'third way' option may be valid for spoken language: that humans may have an innate tendency to invent language, but also that most language has spread as a cultural practice from a small number of original innovators.
Anyway, I don't know the answer. Personally, I suspect it's closer to instinct than to culture, and examples like ISN certainly bolster that interpretation. But alternative interpretations have not yet been ruled out, and probably never will be.
Except it's probably the case that nouns develop before syntax does.
Well, yes, that's a bit of a no-brainer, because syntax without nouns doesn't have much use
. Also, the calls of certain monkeys for specific kinds of dangerous animals could be called "nouns", though there's no syntax at all, just single "noun" utterances.
It's not a no-brainer at all - syntax could develop at the same time as nouns.
I would propose the order "single nouns" -> "combining single nouns" -> "nouns + verbs" -> "more complex grammar".
Possible, but far from certain. It's possible to have quite complex grammar without categoriality, particularly in languages that are highly restricted in function.