Tuesday, January 21, 2014

Tecumseh Fitch’s The Evolution of Language – a highly recommended read

Guest post from William Matchin:

Tecumseh Fitch’s The Evolution of Language was published in 2010, making a quick review of the book a long time coming, and perhaps not as apropos as it might have been. Still, I found the book to be a quite informative synthesis of many areas of research in the speech and language sciences. A more appropriate title for the book might be The Evolution of Proto-language or perhaps The Evolution of Speech, given the author’s heavy focus on the development of vocal communication in humans, and much less discussion regarding the higher-level components of language, particularly recursive syntax. I was surprised at this, given the author’s contribution to the seminal paper The Faculty of Language: What Is It, Who Has It, and How Did It Evolve? (Hauser, Chomsky & Fitch, 2002). One of the major conjectures in this paper is that the core, human-specific component of language is recursion, leading to major questions about how it evolved and its instantiation in the brain. This was the main reason I picked up the book in the first place, but the topic is mostly avoided aside from some introductory discussion, perhaps for good reason, given the uniqueness of its place in human language (and the accompanying difficulty of applying the comparative method). The book’s discussion of syntax is nicely summed by the following quotation (pg. 185):

In conclusion, animals which actually generate call sequences that appear random seem to be exceptional, and in many species there are rules (or constraints) upon vocal sequences that can reasonably be termed “animal syntax.” However, the types of rules that govern these arrangements in primates are very simple compared to human linguistic syntax: they typically can be captured by trivial finite state grammars, and only the propositionally meaningless “songs” of birds and whales require more complex grammars. Thus, current data support the existence of a large gulf between animal “syntax” and that employed in any human language.

Oh well – looks like I’ll have to wait to figure out the answer to syntax. At any rate, the book is an excellent introduction into interesting nuances of evolutionary theory, the comparative method and theories of proto-language evolution, and has plenty of goodies that I think neurolinguists should pay attention to. Here are some salient points that I was personally unaware of before reading the book:

1. The descended larynx is not uniquely human.

The “uniquely” descended larynx in humans is often touted as indicative of the selective pressure of vocal communication in humans. It turns out, though, that this trait is not uniquely human. In most non-human animals without a permanently descended larynx, the larynx descends while vocalizing. Other mammals, such as deer, koalas, and the big cats (lions, tigers, etc.), do have a permanently descended larynx. What function does a descended larynx serve? The lower the larynx, the lower the pitch, which is a reliable cue to an animal’s body size. This is an important signal both inter- and intra-species for potential conflict between animals. Lowering the larynx may have been adaptive for humans as well, and studies show that human subjects use pitch as a cue to body size, even though in humans pitch is not a reliable cue to body size, indicating that perception of body size through pitch is a cognitive relic from a previous ancestor. Also, the lack of a lowered larynx does not appear to be an impediment for nonhuman primates to produce speech, given that (i) they can lower their larynx while vocalizing, and (ii) they have plenty of control over their vocal apparatus for other functions, like eating and biting.

2. A crucial difference in vocal control between humans and other primates/mammals consists of neuronal control of the larynx.

One crucial difference is the development of novel direct connections between frontal motor cortical areas and brainstem motor neurons involved in laryngeal control. While nonhuman primates lack these connections, their presence in humans suggests that an important stage in the evolution of speech involved developing neuronal control mechanisms of the larynx. This hypothesis is supported by fossil evidence concerning the enlargement of the thoracic canal in modern humans (including Neanderthals) relative to existing primates and earlier fossil hominids. The thoracic spinal cord is critical in the control of muscles for breathing. The thoracic canal is the opening in the base of the skull through which the thoracic spinal cord travels, and its diameter is enlarged in late hominids compared to earlier hominids and non-human primates.

3. Gestural proto-language theories face a very hard problem: sign language.

At first, this argument may seem counterintuitive: doesn’t the existence of sign language support the evolution of language from a gestural system? However, the hard problem in the evolution of language is why humans, uniquely among primates, use speech exclusively as the default signal of communication, and how humans developed such intricate vocal control. If a gestural system served as the proto-linguistic step on the way to complex language, why not stay in the visual-manual modality? The viability of sign language proves that there really is no reason to make the extensive evolutionary jump to speech. Arguments in favor of speech, such as communication at a distance, etc., face equally strong counterarguments in favor of sign, such as silent communication during hunting, etc. Given a community of speakers already using a gestural system, there appears to be no selective pressure on the development of vocal communication. This is not to say that gesture isn’t an important part of modern language use, or didn’t form part of the proto-linguistic system of communication in humans, but that a gestural proto-language doesn’t have much explanatory power as a bridge between the communication of our least common ancestor and modern language. Arguments against a gestural proto-language often come from experts in the study of sign language (e.g., Karen Emmorey, 2005).

4. Musical/prosodic communication is an interesting, viable theory of human proto-language.

Think birdsong, in humans. This theory was actually first proposed by Darwin, and supported by a variety of pieces of comparative evidence, and derided by linguists at the time. He noted the similarities between birdsong and human speech, critical facts such as (i) learned vocalizations, (ii) babbling in young birds and humans, and (iii) local “dialects” in both birdsong and human language. He appealed to sexual selection in birds as the mechanism that drove the evolution of vocal imitation in humans, and the emergence of meaning through imitation of natural sounds using both sounds and gestures (onomatopoeia and imitation of innate human vocalizations like laughter or crying). This last statement will probably prompt many linguists to point out the fundamental, Saussurean notion of arbitrariness of the linguistic sign against this theory, but the crucial point is to explain how vocalizations began to acquire meaning, not to account for the development of the rest of the lexicon. In fact, sign languages have many iconic gestures, which explains how the signs were first developed, but iconic signs are treated the same as any arbitrary word once the full system of language is in place by signers. So the theory gets you to sound-meaning correspondences, and assumes that arbitrariness took over from there.

The nice thing about musical proto-language theories is that they explain music – as an ancestral relic of our proclivity for structured, vocal utterances that do not depend on analytic meaning.

5. The importance of the comparative method in studying cognition, even in traits that are not homologous with humans.

This is driven home throughout the entire book, and there are many examples that embody the utility of the comparative method. The descended larynx is one example – larynges descended in other mammals, but this trait is not homologous to humans. The repeated evolution of laryngeal descent does provide insight into why such a trait evolved, and we can apply this insight to humans. Likewise for birdsong – birdsong isn’t homologous to humans, but there is much insight to be gleaned from studying it. In particular, animal models of the infamous FOXP2 gene highlight this quite nicely – transgenic mice with the FOXP2 mutation that the KE family has (speech motor control issues) show approximately normal vocalizations, but FOXP2 expression in songbirds increases during song learning in the bird homolog of the basal ganglia. Downregulation of FOXP2 expression in living birds show incomplete and inaccurate song learning. This shows suggests very conspicuous connections to the function of this gene in humans, and about how vocal learning and vocal control in humans operates.

The book reads exceptionally well – the points are laid out very clearly, and the main goal of the discussion is always kept in sight. Fitch also excels at exploring the theoretical foundation of many of our modern ideas – his exposition of Darwin’s thoughts on language evolution (and the observation about how rarely Darwin is cited about this issue) was interesting and fun to read through. Fitch is very fair when discussing different perspectives on the issues, almost too fair – he is quite charitable when discussing Rizzolatti and Arbib’s mirror neuron hypothesis of language evolution, even while pointing out important problems. He is also quite thorough, exploring several important hypotheses while not rambling pedantically through every possibility. If you’re looking for a book crammed with witticisms and ways to get undergraduates interested in language a la Pinker, you won’t find it here; instead, you’ll find a useful, interesting and exhaustive resource regarding the evolution of language (minus recursion).

Fitch, W. T. (2010). The evolution of language. Cambridge University Press.

Hauser, M. D., Chomsky, N., & Fitch, W. T. (2002). The faculty of language: What is it, who has it, and how did it evolve? Science, 298(5598), 1569-1579.

Emmorey, K. (2005). Sign languages are problematic for a gestural origins theory of language evolution. Behavioral and Brain Sciences, 28(02), 130-131.


Carrie said...

A good follow up could be "The Unfolding of Language" by Guy Deutscher (a linguist), which I highly recommend, in turn! Not a lot of recursion there either, but it does very satisfyingly explain how complex linguistic systems might arise from simple proto-languages.

William Matchin said...

Given that there's now a vacancy on my "books being currently read" shelf, I'll take you up on that! Thanks.

"O" the Humanatee! said...

I believe you have misunderstood an important distinction in speech production and perception. The descended larynx does not affect pitch (a.k.a. fundamental frequency) but rather formant frequencies. These are the peaks in the filter function of the supralaryngeal vocal tract, i.e., the frequencies that the tract filters out least. I used to know Tecumseh Fitch (we had the same doctoral adviser, Philip Lieberman), and the association between body size, larynx length, and formant frequencies was cental to his doctoral research. I'd be very surprised if he did not make the, er, fundamental difference between pitch and formants clear in his book.

William Matchin said...

You're completely right - this was my error!