Subscribe to:
Post Comments (Atom)
-
Harnad, S. (2003b) Categorical Perception . Encyclopedia of Cognitive Science . Nature Publishing Group. Macmillan. http://www.youtube.com...
-
(10a. Comment Overflow) (50+)
-
Fodor, J. (1999) " Why, why, does everyone go on so about thebrain? " London Review of Books 21(19) 68-69. I once gave a (perfec...
Overall I agree with the logical flow of this paper, that language did not start with vocal language, but initially started based on gestures, and that initially most learning started by induction and then later on instruction based learning conferred an advantage, as it reduced the risk adopted by trial and error/corrective feedback learning. Furthermore, it makes sense that Baldwinian evolution began to favor those with a disposition to learn. I believe what was meant by disposition was both the cognitive capacity to construe propositions but equally the motivation to acquire and share categories through intentional rather than incidental instruction.
ReplyDeleteA point of contention in the paper is, "Baldwinian evolution began to favor this disposition to learn...the tendency to acquire and convey categories by instruction thus grew stronger and stronger in the genomes and brains of the offspring of those who were more motivated and disposed to learn to do so. And that became our species’ 'language-biased' brain..." My question is where does UG fit into all of this? If we suppose that Chomsky is right when he says UG does not have evolutionary origin, yet without it we would not have the well-formed language capacity that we have now, than I believe it is quite impossible to have a language-biased brain without having the UG rules part of our cognition to begin with, it must precede the bias, it allows our species to even have a bias. Until it can explain how UG fits into this instruction vs induction perspective on language development this to me is an 'almost-there explanation' (I disagree with Chomsky that the question of the origin of UG does not need to be answered). I think to answer the origin of language one should focus on answering the origin of UG, much like brain imaging techniques are often correlational, and don't explain the how, but having a brain model, although more complex, and time consuming it provides much more insight on the causal mechanism which advances the field much faster than correlational studies. Essentially, that is how I feel about ignoring the UG question in answering the origin of language.
In the section about "the power to say anything that can be said" I was deeply intrigued by the translatability thesis. That’s to say, anything able to be said in one language, can be said in all other natural languages. In particular, I thought this was interesting because the Saphir Worf hypothesis seems to support that language shapes our perception, thusly languages with more words for things can perceive differences that other languages may not (the Inuit language with snow, etc). I've learnt about this construct in many other psychology classes, and it has always been presented as an open debate. However, this paper, published back in 2013, seems to very easily wrap up any controversy. Why in academia are they not presenting this argument in opposition to the linguistic relativity stance?
ReplyDeleteIf proposition was born from the motivation to teach and learn new categories, wouldn’t this support an evolutionary explanation of language? If elders in a family were trying to teach their kin important tools or lessons to better their survival, thus giving way to propositions in language, perhaps this affinity to teach was passed down preferentially through generations. The distal role would be the enjoyment of learning new categories and the proximal role would be learning more helps you to know more, communicate better and survive longer. If this "language biased brain" evolved over the years, wouldn’t that favor evolution as an explanation?
The idea that we’ve developed language because we are more “motivated”, or rather more ‘socially motivated’ as a species is interesting to me. The idea being that we are more socially oriented, thus end up wanting to help our kin survive, then the most efficient way to do that is to use instruction learning over induction learning, and so through Darwinian evolution this became part of our genes. What does this mean for creating T2 or T3? Does this mean that this kind of ‘social motivation’ needs to be somehow programmed? Would researchers need to pinpoint a specific gene and reverse engineer that? I think what I am mulling over is just how incomplete our social tendencies and evolutionary explanations are for understanding how we learned to “take the ball and run with it”.
ReplyDeleteFirst off, I can't quite accept the reason given for why chimps do not utilize language. If language was so advantageous for category learning and instruction, and could potentially have survival and reproduction benefits, wouldn't any species capable of it use it? A simple lack of motivation seems to be too simple doesn't capture why, if their brains have the capacity for it, they do not produce the propositional quality of human language. If chimps can categorize, why not extend that into the verbal modality so as to instruct their kin, and help them in adapting to their environment, so that not everything had to be learned by induction - as was the explanation for why humans developed propositional language.
ReplyDeleteJust a thought - Harnad mentions that the offspring of those who were more motivated to acquire and teach categories had more of a tendency to do so, and this is how we eventually acquired our "language biased" brain. Could this language-biased brain have something to do with UG? Perhaps UG principles and parameters depend on and are based on category instruction, and this is why they are innate and not attainable through positive evidence? Or maybe UG principles facilitate the process of acquiring and conveying categories, by not allowing us to utter sentences that would confuse the person who is acquiring the category through our instruction.
“We can now see the sense in which the “shapes” of the symbols themselves are arbitrary: 0 is not shaped like nothingness, 3 is not shaped like Threeness, the equals sign is not shaped like equality, and so on.”
ReplyDeleteI understand that symbols are arbitrary and we could choose any shapes to serve as symbols for the formal symbol system to operate, but this passage makes me very curious about the nature and definition of meanings. Without the arbitrary shapes that we created, what are “nothingness”, “Threeness” and “equality”? Will we ever be able to illustrate the true form of these concepts? With that being said, are meanings also arbitrary? Additionally, if something is meaningful, does that imply it has truth value?
One of the main purposes of this paper is to examine the origin of language, and I think it does a great job on giving the readers a clear picture of how language potentially started, by combining the topic of categorization, natural language, and evolutionary theories. In section 5, the authors state that “the power of language was, in the first instance, the power of acquiring a new composite category from other people who already knew the old categories out of which the new one was composed.” This shows that categorization, as an important behavioral and cognitive skill, has a huge impact on the emergence and development of language for our species. The availability of various cognitive components allowed our ancestors to be able to communicate in the first place, and later adaptation and motivation enable the language to develop step by step.
1. “The Kernel is younger than the words in the rest of the dictionary; and within the Kernel, the Core is youngest, with words getting younger and younger the shorter their definitional distance from the Core.” (page 28)
ReplyDeleteI don’t understand this; since the Kernel is the composed of the most concrete words in the dictionary, surely these are the oldest words in the language. Furthermore, the kernel is made up of words that are used to define other words in the dictionary, so surely the words in the kernel must have been invented prior to the words that they are used to define? I don’t see how it’s possible for the words to be younger the closer they are to the core.
2. I really liked Figure 2 from the Masse et al paper, it helped clarify that the kernel acts as the grounding set for the dictionary. I also found it very interesting to think about the following exerpt:
ReplyDelete“This has some of the flavor of our “test” for whether something is a natural language: that you can express any and every proposition in it. Whatever that turns out to be, it seems as if fewer than 1500 category names plus a few functors and syntactic rules is as much as you need in order to get language’s full expressive power.” (page 13)
I wonder whether this is true for all languages, as is claimed here. I often hear that Chinese is a particularly different language to learn due to its large array of characters (of which there are thousands). However, potentially this paper could oppose this view and show that the relative complexity of various languages is, in fact equal.
It didn’t surprise me that learning categories by instruction is not equivalent to learning categories by induction. They’re two different processes with their own unique benefits. Instruction definitely requires less trials, or even any for that matter. On the other hand induction lets the individual experience it for themselves and get a real sensorimotor feel for the category at hand.
ReplyDeleteWithout the core categories learned through induction it would be impossible to learn through instruction. Thus, even if tell has superseded show, I think it’s extremely important to note that tell relies on show as its foundation.
I now understand why "Steven says” T3 capabilities are probably necessary to pass T2. If we want to create a robot that can learn new categories, it must be able to have sensorimotor capabilities to ground the core categories through induction.