Tuesday, December 20, 2016

talking brains

In a recent paper (here), Tecumseh Fitch (TF) and colleagues argue that monkey vocal tracts are structurally adequate for the production of human speech sounds. Why is this important? Because, as the paper puts it:

Our findings imply that the evolution of human speech capabilities required neural changes rather than modifications of vocal anatomy. Macaques have a speech-ready vocal tract but lack a speech-ready brain to control it.

This, in other words, puts another nail in the coffin of those that look to provide a continuity thesis style of explanation of human linguistic facility based on a quantitative extension of what appears in our nearest cousins (i.e. If this is right it then Phil Lieberman was wrong). If TF is right, then all that effort expended in trying to teach primates to speak was a waste of time (which it was) and the failure was not one that could be resolved by teaching them sign (which in fact didn’t help) because the problem was neural not vocal. IMO, the futility of this line of inquiry has been pretty obvious for a very long time, but it is always nice to have another nail in a zombie’s coffin.

The results are interesting for one other reason. It suggests that Chomsky’s assumption that externalization is a late add-on to linguistic competence is on the right track. FT provides evidence that vocalization of the kind that humans have is already in place engineering wise in macaques. Their vocal tracts have the wherewithal to produce a range of vowels and consonants similar to those found in natural language. If they don’t use this to produce words and sentences (or movie reviews or poems) it is not because they lack the vocal tract structure to do so. What they lack is something else, something akin to FL. And this is precisely Chomsky’s suggestion. Whatever changed later coupled with an available system of externalization. This coupling of the new biologically unique system with the old biologically speaking more generally available system was bound to be messy given they were not made for each other. Getting the two to fit together required gerrymandering and thus was born (that messy mongrel) morpho-phonology. FT supports this picture in broad outlines.

One more point: if externalization follows the emergence of FL, then communication cannot be the causal root of FL. Clearly, whatever happened to allow FL to emerge came to take advantage of an in-place system capable of exploitation for verbal communication. But it seems that these capacities stayed fallow language wise until the “miracle” that allowed FL to emerge obtained. On the assumption that coupling FL with an externalization mechanism took time, then the selective pressure that kept the “miracle” from being swept away cannot have been communicative enhancement (or at least not verbal communicative enhancement). This means that Chomsky-Jacob suggestion (here) that the emergence of FL allowed for the enhancement of thought and that is what endowed it with evolutionary advantage is also on the right track.


All in all, not a bad set of results for MP types.

18 comments:

  1. That link is wrong. Here's the correct one: http://advances.sciencemag.org/content/2/12/e1600723

    ReplyDelete
  2. "It suggests that Chomsky’s assumption that externalization is a late add-on to linguistic competence is on the right track. "

    I don't quite see how this follows; presumably this result implies that early hominids had a language ready tract before language emerged in its current form. Why does that mean that externalisation would appear after Merge? (if that's the right interpretation of this claim)

    ReplyDelete
    Replies
    1. I think there is at least one relevant point along this line. A speech-ready vocal tract is a necessary part of Chomsky's claim, because otherwise both Merge and then a speech-ready vocal tract would have to evolve. Without the truth of this finding, Chomsky's claim would be harder to defend. It perhaps does not provide any other interesting evidence in support of the claim as far as I can see.

      I do think Norbert's logic is not quite right here, though. Fitch's claims about neural hard-wiring are only relevant to speech, not FL. FL is perfectly happy with either sign or speech (and indeed, the exact same brain areas are involved, modulo the auditory/visual-spatial difference). Apes can learn to make signs, but not produce speech sounds. Fitch's claims are therefore relevant to externalization capacities, and whether these capacity differences between humans and apes and/or monkeys resides in the vocal tract itself or in the nervous system. So I don't think Minimalists should get too excited about this research.

      Delete
    2. Externalization of merge created objects would appear after merge on the assumption that hooking the merge system, i.e. G, to external mechanism was not automatic.

      I agree with William that MPers should not get TOO excited about this. Just suggestive in the right direction.

      Delete
  3. This comment has been removed by the author.

    ReplyDelete
  4. Being familiar with the relevant literature, we are quite surprised by this interpretation/reading of the recent study by Fitch et al. As Fitch discussed in, e.g., his Evo of Lg book in 2010, currently the best hypothesis we have is this: the critical component that would make our relatives' "speech-ready" would be a direct connection between the motor cortex and the larynx. Such a connection, whose molecular properties we are beginning to understand (see work by Jarvis et al on the SLIT-ROBO pathway, and how this ties in with the literature on FOXP2), is present in vocal-learning birds. These birds have a 'speech-ready' brain. But no one from the minimalist/chomskyan camp would claim that the evolution of the vocal learning property of the birds' brains support the externalization-late-add-on 'thesis' (in fact, we know of quite a few passages from the minimalist/chomskyan literature, including passages of this blog) that typically dismiss work on birds as 'oh, that's just externalization, not the core language'. So, we find it funny that both the non-vocal-learning-ready brain of primates and the vocal-learning-ready brain of birds are used by some to make the point that "language is special, different, unique, etc"

    Cedric Boeckx
    Bridget Samuels
    Pedro Tiago Martins

    ReplyDelete
    Replies
    1. Not sure i follow the point. I took the point to be that the wherewithall for externalization existed before whatever ling capapcity is existed. I take this to mean that the structures needed for speech existd prior to the mergence f language. This suggests that whatever happend to allow for speech was somthing more abstract than this and the more abstract thing had to accomodate itself to the prevailing articulatroy structures to be usable to talk. This in turn suggested that the Chomsky idealization, first merge then externalization, made sense. It was compatible and suggestive. Are you suggesting it isnt?

      Delete
    2. How can evidence that mechanisms for externalisation were available before Merge be taken to be evidence that externalisation occurred after Merge?

      This is the bit I don't follow. It seems backwards to me. It seems rather to support the externalisation first idea.

      Delete
    3. I am sort of with you, Alex, but I think there is a deeper problem with the logic of the argument put forth by Norbert: as far as I can see, it conflates 2 things: "externalization" (often used by minimalists in the sense of vocal learning or speech; see their (dismissive) treatment of the FOXP2 literature) and "articulation" (both vocal learners and vocal non-learners articulate). What I don't get is how evidence that mechanisms for articulation (but not externalization) are available before Merge could be taken as evidence for anything about externalization.
      If the argument is that articulation preceded language, it can't be used to support anything Chomsky said (assuming, as I think we must do, that Chomsky's argument about externalization was more subtle than 'articulation=externalization')

      Delete
    4. @Norbert (I think I can safely say I'm speaking for Bridget and Pedro here) We are not sure we grasp what you mean. If by "I took the point to be that the wherewithall for externalization existed before whatever ling capacity is existed", you mean 'the point of Fitch et al's paper', we think this is mistaken, because Fitch (and others, too) has been arguing for years that the wherewithall of externalization (for vocal learners) must include a specific brain connection ("speech-ready brain"), and their paper lends support to the idea that this is missing (or not developed enough) in our closest living relatives. In other words, they don't show "the structures needed for speech" existed prior to language, since such structures include special neural connections. That's why we cannot understand the point of the post, and why we think this cannot be used in any way to defend the specific take ("late add-on") on externalization favored by some.
      Our confusion/surprise stems from the fact that when people show something non-human primates lack, the conclusion (for some) is "see, (it's just) externalization". And when people show birds have that thing which non-human primates lack, sort of the same conclusion is reached "yes, but that's just externalization" (aka "speech, which is not language"). So, both the speech-ready brains of birds and the non-speech-ready brains of non-human primates are used to support a distinction between externalization and (core of language/merge). We find this very strange.

      Delete
    5. This comment has been removed by the author.

      Delete
    6. I think Norbert's (and presumably Chomsky's) point is that if there is nothing to externalize then it is not going to be externalized, whether you can articulate it or not. Hominins constitute at least a special case where what is externalized are merge-based structures. Only after these structures emerge can one externalize them (by whatever means). So it is perhaps irrelevant for the argument when and in what species the externalization mechanism itself evolved. Vocal-learning birds clearly do have this mechanism but it is used to externalize something very different from human language.

      Delete
    7. Presumably one can externalise individual lexical items without merging them. Does saying "Hi" involve Merge? Or one could externalise some non hierarchically structured expression. The various proto-language theories may be wrong but they aren't wrong a priori.

      Delete
  5. This comment has been removed by the author.

    ReplyDelete
  6. This comment has been removed by the author.

    ReplyDelete
  7. This is most interesting research. The study of contortions of Emiliano the macaque’s vocal tracts not only highlights aspects of language, it may be fruitfully extended to study other domains as well. For example, the form of research opens the question, “why doesn’t Emiliano sing like Luciano Pavarotti?” It is well-known that, in order to sing like Luciano Pavarotti, you need intricately structured lungs, and it may well turn out that Emiliano has those lungs. So, theoretically, Emiliano could sing like Luciano Pavarotti, but he doesn’t. Why? Because Emiliano doesn’t have the right neural circuits. If Emiliano had Pavarotti’s brain, he would sing ‘Nessun Dorma'. However, the research won’t rule out that, after 50,000 trials and truckloads of banana, Emiliano may be encouraged to sing like Bob Dylan.

    ReplyDelete
  8. "What they lack is something else, something akin to FL. And this is precisely Chomsky’s suggestion."

    Well, maybe; it depends on which version of Chomsky you're considering. It's obviously in accord with what Geoff Pullum calls the "rocks and kittens" version to which Chomsky has retreated in recent years, but that was never really controversial, and it doesn't tell us anything about what FL might be. More to the point, the "rocks and kittens" view is consistent with the idea of FL, but *also* consistent with the idea that humans have a different domain-general learning mechanism, which is responsible for language and all sorts of other things. (I grant that some of Chomsky's more fervent defenders are now inclined to claim that (1) Chomsky always already thought that FL might be domain-general, or that (2) "all sorts of other things" are all language, but I don't find these defenses serious enough to merit further comment).

    "One more point: if externalization follows the emergence of FL, then communication cannot be the causal root of FL."

    I'm not sure I follow. Fitch et al. seem to be saying that macaques have the *capacity* for externalization, but not the neural architecture. If suddenly some switch flips and a bunch of primates gain the neural architecture, then it can immediately be used for communication, because the apparatus for externalization is already in place.

    ReplyDelete