Saturday, June 27, 2009

Adam's Tongue (pt 3)

(classic depiction of Saussure's arbitrariness of the sign claim)

This is the third in a series of posts detailing my notes and thoughts about the book Adam's Tongue as I prepare to lead a book discussion meeting July 6, 2009 in the DC metro area (see my first post here and second here).

Ch 3 - Thinking Like Engineers

I've spent the last 5 years working in natural language processing and with engineers and I agree that there is something very valuable for a linguist to "think like an engineer" so I was curious from the start about this chapter, but I was also weary because the Chomskyan syntacticians also "think like engineers" and I believe they have led linguistics down a garden path of false starts and flawed theories for 40 years. So I read on cautiously.
  • DB notes that he came into linguistics via pidgins and creoles and they bear on his thinking about language evolution. But does this bias him too, like the man who has a hammer and sees everything as a nail? We shall see.
  • DB says there's no syntax when we try to speak with people who don't share our language (p 39) because we don't know enough of the language, the foreign words just pop out as we grope for them. Now, I certainly defer to DB's far greater expertise of pidgin & creole formation, but this thought experiment of his does not jive with my own experiences. Like many travelers, I've had this exact experience in places like Guangzhou China and Prague but I don't think the foreign words "just popped out" quite as randomly as he suggests. I'm tending to side with Slobin here.
  • He claims that protowords must not have had any internal morphological structure (41) because early language users would have had no rules defining that structure. On it's face, this makes sense, nonetheless this begs the question: which came first, the word or the morphology? Is it not plausible that some neurologically based process for seeking internal structure to sounds developed prior to the advent or words? I just don't know.
  • The boom vocalization of the Campbell's monkey occurs 30 seconds before the alarm (42). My first reaction: wow! this is stretching the limits of transitional probabilities, isn't it? Can we plausibly claim that an association between sounds 30 seconds apart is neurologically feasible?
  • DB claims these booms are not modifiers (p42) because the boom "cancels out" the alarm. I'd have to review the literature on these boom carefully, but my first reaction is: does it really cancel the alarm? If I understand the context, it simply means "not immediate threat (but still a threat)". That's not a cancellation. It's more like epistemic modality: "there MIGHT be danger."
  • Page 44 -- The gavagai problem restated.
  • Confused: I'm confused by DB's claim on page 45 that "words combine as separate units -- they never blend. They're atoms, not mudballs." I'm not sure what he means. Blending and combining are different, in that blending suggests some elements of both previous words/calls are preserved in the new word/call. This happens all the time in contemporary linguistic change (classic example: motel blends motor + hotel, persevering bits of each's morphology as well as semantic blending). But I suspect DB is not referencing that. So what is he referencing?
  • He makes a nice distinction between ACSs and Language: ACSs are primarily for manipulation of behavior while language is primarily for information sharing. I have no clue if this is really true, but if yes, it's a good point (p 47).
  • He writes "language units are symbolic because they're designed to convey information." A nice follow-up point on the difference point above, but it begs the question: what is "information"? Any answer which supports DB would have to couch a definition in abstraction, right? E.g., Information is a conceptualization that is independent from direct reference.
  • DB makes a bold claim on page 52 that strikes at the heart of post-Saussurean linguistics: displacement is a more important factor to language evolution than arbitrariness. But it's worth noting that both are functions of abstraction, so perhaps this is just another version of his previous point that the jump to abstract thought is the key.
On to chapter 3 -- Singing Apes....

Adam's Tongue (pt 2)

This is the second in a series of posts detailing my notes and thoughts about the book Adam's Tongue as I prepare to lead a book discussion meeting July 6, 2009 in the DC metro area (see my first post here. UPDATE: My third post is here).

Ch 1 - The size of the problem
  • This chapter is designed to walk through what's wrong with other theories of language evolution.
  • The basic point of the chapter seems to be this: no animal communication system (ACS) allows itself to refer to things distant in time and space, therefore they are not likely the precursors of language. Everyone who has taken or taught a Language Files course knows these two criteria as Hockett's two communication features unique to human language (Bickerton get's to Hockett in due course).
  • On the very first page of this chapter, I noted, "Is there a gavagai problem here?" By which I meant, how can we know what one of these ACS references really refers to? Bickerton's index lists nothing for either "Quine" or "gavagai," though he skirts this issue repeated for the next few chapters (and possibly the whole book). This dilemma become particularly critical in chapter 4 Chatting Apes, but I'll come to that later.
As a background, here's a passage from Wikipedia's Indeterminacy of translation page describing Quine's famous example:

Consider Quine's example of the word "gavagai" uttered by a native upon seeing a rabbit[1]. The linguist could do what seems natural and translate this as "Lo, a rabbit." But other translations would be compatible with all the evidence he has: "Lo, food"; "Let's go hunting"; "There will be a storm tonight" (these natives may be superstitious); "Lo, a momentary rabbit-stage"; "Lo, an undetached rabbit-part." Some of these might become less likely – that is, become more unwieldy hypotheses – in the light of subsequent observation. Others can only be ruled out by querying the natives: An affirmative answer to "Is this the same gavagai as that earlier one?" will rule out "momentary rabbit stage," and so forth. But these questions can only be asked once the linguist has mastered much of the natives' grammar and abstract vocabulary; that in turn can only be done on the basis of hypotheses derived from simpler, observation-connected bits of language; and those sentences, on their own, admit of multiple interpretations, as we have seen.
  • No gradual move from ACS to human language (17): Since evolution is gradual and slow, there would have to be a "missing link" (my term, not DB's); an ACS that made the jump from referring to the here and now to referring to the distant and far. No such link exists
  • Therefore, ACSs grew out of non-communication behaviors
  • Uniqueness of language also not relevant because many species have unique features (Pinker's elephant trunk, 20).
  • Humans suddenly had something else to "talk" about other than the here and now and THAT'S what spurned language.
  • The new thing humans had was abstract concepts (22). We can talk about dogs as a category (he makes an important distinction between categories and concepts much later in chapter 4 at the bottom of page 87).
  • This new ability to abstract is not associated with evolutionary fitness.
  • Critical Point: other species didn't develop language because they didn't need language (p 24).
  • Bickerton's 4 tests for any theory of language evolution: 1) uniqueness, 2) ecology, 3) credibility, 4) selfishness (p28). Bolles' blog Babel's Dawn discusses these criteria at length here.

Space and Thought

(two of Boroditsky's stimuli, pdf here)

Yet again, Andrew Sullivan treads into the area of linguistics and cognition research. But at least this time he's wise enough to make no comments about the studies he links to (he's typically misguided, or flat out wrong in his linguistic sensibilities, see here, here and here).

This time he reprints a quote here from an article titled How Does Our Language Shape The Way We Think? written by Stanford assistant professor Lera Boroditsky regarding how language influences thought. Of course, Sullivan reprints the least interesting piece of information in the article, a mere behavioral anecdote about how speakers of different languages use different direction terms. This fact has been well known for a long time (I first learned about it in an introductory cog sci course in 1998 and it was old news then). The more interesting fact is the following effect she observed during a test to compare Russian and English speakers' ability to discriminate shades of blue (color terms is a classic topic within cognitive science going back to Berlin & Kay's work in the sixties, see here):

The disappearance of the advantage when performing a verbal task shows that language is normally involved in even surprisingly basic perceptual judgments — and that it is language per se that creates this difference in perception between Russian and English speakers.

After skimming Boroditsky's article, I felt had it was a very good review of the field of language and thought studies as I remember it, but it didn't add much, if anything, but it's clearly a layperson's article, so I looked at her Stanford page and skimmed her list of publications and more critically, the references she cites.

My first impression was, "she doesn't cite much, does she?" I'm used to experimental psychology articles containing lists of references almost as long as the article itself, but most of her (first author) papers have a handful of citations. But the more surprising thing was the notable absence of two names, Len Talmy and Jürgen Bohnemeyer. I'll grant that I'm a little biased because both of them were professor's at my grad school, but the granting ends there. I can't imagine writing a serious research paper on how language shapes thought without references to one or both of these researchers, especially as Talmy has written an extensive, typologically rich, two volume set on the relationship between language ands thought: Toward a Cognitive Semantics and he has a forthcoming book The Attention System of Language (his work in progress handout on the same topic can be read in this pdf).

Don't get me wrong, I basically like Boroditsky's research methods and approach. I just think it's time for her to review Talmy and Bohnemeyer.

Wednesday, June 17, 2009

Yo! Google This!

(screen shot of Google Spell Check)

Apparently, Gmail spell check does not recognized "googled" as a word (past tense of "to google"). Will Microsoft spell checkers recognize "binged" as a word?

...and that's all I have to say about that.

UPDATE: It looks like Roger Shuy at Language Log has gotten on this bandwagon (a month after us, but he's welcomed aboard). He re-iterates Faldone's point (see my comments) and suggests that Microsoft might be bucking this trend of verbing a trademark with bing. He then makes (ahem) the same joke I did. Welcome aboard Roger.

Monday, June 15, 2009

Adam's Tongue pt 1

(pic of hardback cover of Adam's Tongue)

On July 6th, I will be leading my DC area book club, Books and Banter, in our meeting on the new book Adam's Tongue: How Humans Made Language, How Language Made Humans by Derek Bickerton (Hardcover - Mar 17, 2009).

Amazon’s Product Description:
Language is unique to humans, but it isn’t the only thing that sets us apart from other species—our cognitive powers are qualitatively different. So could there be two separate discontinuities between humans and the rest of nature? No, says Bickerton; he shows how the mere possession of symbolic units—words—automatically opened a new and different cognitive universe, one that yielded novel innovations ranging from barbed arrowheads to the Apollo spacecraft.

(opening page 3 of my copy of Adam's Tongue)

Since this book coheres closely with this blog’s topic of linguistics, I’m going to be posting my notes and thoughts as I read and prep for the discussion. I won’t guarantee that I’ll revise and clean up my notes into entirely coherent prose (see pics above for my typically messy method of “reading” a book on linguistics...the left page was originally blank), but if you’re reading the book too, I hope this encourages your thoughtfulness and stimulates your critical reading.

UPDATE: My second post on this is here and my third is here.

This first post will cover only the Introduction, pages 3-15. On general note: as I am no longer affiliated with a university, it is remarkably difficult for me to follow leads involving academic papers; therefore, many of the references Bicketon makes to published works (such as Derek Penn’s intriguing list of things humans can do that non-humans cannot, p8) are, for the time being, locked behind an impenetrable vault for the lowly lay Lousy Linguist and as such must go un-reviewed. Apologies. I shall review all that time and Google together permit.

Shall we begin?

My first reaction is that that the intro is written as a teaser (like most pop writing intros) and as such it leaves lots of questions to be answered. This begs the question: will the rest of the book live up to the tease? I’m a skeptic by nature, so I’m guarded in my expectations. We shall see.

MAJOR POINTS

1. Thought experiment (p 3) – “imagine for a moment that you don’t have language and nobody else has either.” Okay…hmmm…uh…wait, what? First, as a linguist, I HAVE to ask: what is your definition of language? This is a non-trivial question. If you want me to understand how X originated, then you should help me understand exactly what X is. Note: the book index contains no entry for “language” per se.

UPDATE (June 17): the excellent blog Babel's Dawn (on the origins of speech), responds to Bickerton by asking a similar question: how is language to be defined, and then offering definitions here (HT The Outer Horde):

2. Language makes thought meaningful by putting thoughts together into meaningful wholes (pp 3-4). Okay, so language is combinatorial syntax? Can’t we say the same thing about logic? Language is logic?

3. Darwin: having the tool of language caused us to develop greater cognitive capacity (p 5).
Is this similar to Jared Diamond’s Guns, Germs and Steel argument that the coincidental cooccurrence of geography, people, ecosystems was the “ultimate cause” of Western dominance? I fear Bickerton may have an “ultimate cause” nightmare on his hand.

4. FOXP2 – Only one indexed reference to FOXP2 (p110). He disparages the “brouhaha” around FOXP2 and I agree with his point here (there’s no such thing as a “gene for language”); nonetheless, FOXP2 is an interesting gene worth discussing at length, I think. Yet, I wonder if this brevity isn’t an editorial function. I recall the physicist Stephen Hawking retelling a caveat his publisher gave him when writing A Brief History of Time that every mathematical equation he chose to include in the book would cut his readership in half. Perhaps the same could be said for each gene referenced.

5. Magic Moment (p 6) – Apparently he’s looking to explain the “magic moment” when our distant ancestors broke from other communication methods and started using language (uh...cough...hmm...please see 1).

6. Discontinuity (p 9) – evolutionary leaps = differences between species not attributable to gradual change.

7. Niche construction (p 11) – we “guide” our own evolution. I don’t like the use of the word “guide” here. Sounds too intentional. Better if it’s just “affect”.

8. Learning vs. instinct (p11) – he writes “we adapt our environment to suit ourselves, in the same way ants and termites adapt the environment to suit them. We do it by learning, they do it by instinct; big deal." Whoa! Whoa! Yes, this IS a big deal. Let us not trivialize the distinction between learning and instinct. I’ve had just enough exposure to computational neuroscience to recognize that this is no small distinction.

MINOR POINTS

P 4 – “without language there wouldn’t be scientific questions” – here’s my interpretation of what he means: 1) the things we ask questions about exist apart from us but 2) the fact that we ask questions about them (and not others) is a function of our cognitive apparatus (this is a variation on Lakoff’s embodied consciousness, right?). The fact that our embodied consciousness leads us to ask certain questions (and not others) does NOT mean that those questions are a priori more important than other questions; it only means that we consider them more important. We could be wrong.

P 5 – Quoting Darwin does not impress me any more than quoting Aristotle or Buddha or Chomsky: it’s all argument from authority and I have little patience for it.

P 9 – “in this book, for the first time ever, I’m going to show...” This reminds me of a point Foucault made in, I believe, History of Sexuality vol 1, that there is a tempting addiction to being the one who sees and reports the “truth” that others do not. As I recall, his point was that this temptation leads people to report “truths” that are, in fact, not true. Rather narcissistic, really, don’t you think? Is Bickerton a wise man or a narcissist? We shall see.

P 10 – I like this idea of niche construction and “constant feedback loop”. Sounds entirely commonsensical. Of course we affect our environment (despite the claims of global warming skeptics).
P 13 – I like this point that any given communication system is suited only to take care of that species needs (not some lego block building up of features and functions).

P 15 – the big question: what did our ancestors do (that other species did not do) that caused language to explode?

I am a skeptic by nature but I am intrigued, yet doubtful. The difficult part lay before me. 12 chapters of challenging linguistic exploration. Okay, Professor Bickeron. I accept the challenge. Lay on, Macduff, And damn'd be him that first cries, 'Hold, enough!'

Nuts and Bolts of Applying Deep Learning (Andrew Ng)

I recently watched Andrew Ng's excellent lecture from 2016 Nuts and Bolts of Applying Deep Learning and took notes. I post them as a he...