MEGs and Generativists

Article up in the Sydney Morning Herald about a new magnetoencephalographer (MEG) in the Southern Hemisphere. Now, I'm as excited as the next bloke about MEGs in particular because I'm confident that they are measuring cell firing (as opposed to MRIs which measure deoxygenated blood which is a number of steps removed from ion channels opening and ions passing through.)

What I can't get over is what some of the linguists are hoping to find with this new technology.

Click on to read quotes.

Professor Crain believes most children easily learn whatever tongue they are exposed to because they are born with an innate knowledge about language. Concepts such as "or" and "and", for example, were common to many languages, and might have a biological origin.

"If we are right in thinking that there are special brain structures that correspond to certain properties of language, then we may be able to reveal these using the magnetoencephalography technology."

Commonality may have less to do with biological origin than mere reality. Most languages have "up" and "down", do we think this too is biologically inherited. There may be some idiosyncratic encoding of these concepts, but this may be related to perception. The visual system can distinguish orientation of objects and through experience relative location can be distinguished. That one might want to label relative position does not point to a biological module for linguistic "up-down-ness". The module, insofar as it exists, would be far more general.

Read more!

The symbolic world

[I've been gone for a few weeks getting married, going on my honeymoon, and getting settled in. It's nice to be back.]

One thing I love about what NRLG does is that we are trying to understand what is underneath our symbolic capacity. So it's always exciting to reflect on what our symbolic capacity is capable of.

From Science Blog

The researchers, Constance Steinkuehler and Dmitri Williams, claim that MMOs [Massively Multiplayer Online video games] function not like solitary dungeon cells, but more like virtual coffee shops or pubs where something called "social bridging" takes place.

Their overall conclusion in this newest study: "Virtual worlds appear to function best as bridging mechanisms, rather than as bonding ones, although they do not entirely preclude social ties of the latter type."

The pervasiveness of consciousness, whether non-humans are conscious, is controversial even among this research group. However, I personally think that it should not be controversial to say that the symbolic capacities of humans as far as we can tell outstrips any other living creature. People engaged in social bonding on MMOs are applying their biological systems to a world that does not materially exist. Yet they can experience the bodily responses they might encounter in a face to face interaction.

I guess to return to language, a couple questions are raised
1) What is the neurobiological nature of symbolic capacity?
2) How did it evolve?
3) Is language built on symbolic capacity or vice versa?
4) As always, is language uniquely, modularly in the brain or coopting previously/otherwise deployed systems.

Read more!

Surprise for Linguists: Nouns and Verbs Sound Different.

I think this new finding is very significant and puts many ideas of Semiotics into questions. Here is my argumenet. We thought that symbolic representation is the hallmark of language. This study suggest that language is acutally indexical b/c it suggest that people distiguish different wave lenght in seperating nouns from verbs and not just arbirary signs associated with nouns and verbs. Therefore, the "symblic speice" is not so much symolic in understanding and using language. Here is the the whole news and I will send you the whole article in pdf.

Surprise for Linguists: Nouns and Verbs Sound Different
By Sara Goudarzi
LiveScience Staff Writer
posted: 09 August 2006
09:12 am ET


Linguists have long believed that the sound of a word reveals nothing about its meaning, with a few exceptions of words like “buzz” or “beep” that are known as onomatopoeia.

But a new study analyzing the sounds of nouns and verbs challenges that view.

“What we have shown is that the sound of a word can tell us something about how it is used,” said Morten Christiansen, associate professor of psychology at Cornell University. “Specifically, it tells us whether the word is used as a noun or as a verb, and this relationship affects how we process such words.”

However, if you are mouthing a whole bunch of nouns or verbs and listening for a similar sound in each group, you're out of luck.

"It's not a particular sound," Christiansen said. "It's much more subtle than that."

Christiansen, Thomas Farmer, a Cornell psychology graduate student, and Padraic Monaghan, lecturer at the University of York in the U.K., detailed their findings in the Aug. 8 print issue of the Proceedings of the National Academy of Sciences.

Graphing words

The researchers took the sounds of more than 3,000 words in English and subdivided each by its phonetic features—what a person does with their mouth to produce the sounds of each word.

"We could then represent each word in a multidimensional space," Christiansen told LiveScience. This multidimensional space is similar to a simple coordinate-system graph. It gave researchers a chance to see where each noun and verb falls relative to another.

"Each word is a point in this sound-based, or phonological, space," Christiansen explained. And the distance between the words could be calculated.

The nouns were closer to other nouns, and the verbs were closer to other verbs. About 65 percent of all nouns have another noun as its nearest neighbor and about the same percentage of all verbs have another verb next door, Christiansen said. [See the graph]

Hearing sounds

To demonstrate that people were sensitive to this fact, the researchers timed volunteers while they read words of a sentence, appearing one at a time on a computer screen.

They measured how long it took to read each word. The researchers found that volunteers had an easier time processing verbs that sound more like the typical sounding verbs, such as "amuse." The same went for nouns that were more "nouny," like the word "marble."

The volunteers used the relationship between how words sound and how they are used to guide their comprehension of sentences.

"This affects how you interpret a sentence, something that can help you in reading and practicing faster," Christiansen said. This information can also be used in language acquisition.

The researchers performed the analysis for the English language only, but suspect that there are cues in words of other languages as well.


The 3,158 words in the study are plotted in a complex mathematical calculation that places them in "phonological space." Note how the nouns (yellow) tend to cluster, as do verbs (blue). Credit: Morten Christiansen et al.

> Click to View

Read more!

What would it tell you?

Reflecting on a conversation with a computational linguist:
His question was if you knew exactly which neurons were firing when a person heard the word "tree" what would it tell you?

This was a very difficult question to answer because honestly I would have to say not much. However, I would be interested in knowing what neurons fired together in different contexts, different uses, and over time for one subject. And perhaps I'd be interested in the networks across many subjects as well. I would be interested because I'd like to know how fixed these networks are, I think from a Fusterian perspective they might not be that fixed. Perhaps phoneme perception might be more fixed, but the association/recognition of the sound sign would likely vary from use to use. But really, when I think about his question...

I think that we care about fundamentally different issues.

His work is on algorithms that can generate sentences and how generative an algorithm is. Well, that's as far as I understand it. The more any algorithm can account for grammatical utterances the better it is, as far as I can gather. The proof is in the grammatical pudding perhaps. But from my perspective, were I to accept this algorithm, I'd like to know how it is instantiated in the brain. From his perspective, what I am doing explains nothing with regard to grammar which is language to him (well, most likely). However, for me language is more than its grammatical organization. That would be like saying a car was its engine or even that a car is just its physical instantiation without considering the status it conveys, the access it grants to distances further away, or its capacity to carry (or not carry) loads of people and things and so forth.

With this in mind, I look at the neurobiological underpinnings of social behavior and bonding which involves neurotransmitters such as dopamine and opiods under the influence of gonadal steroids, oxcytocin and vasopressin. Then I draw a connection between social behaviorn and bonding to the language learning process and suggest that these systems are involved in the parts of language learning influenced by social bonding.

Where a theoretical linguist will want to know how does a learner acquire the past tense, I will want to know what systems for "other" behaviors may be coopted for language learning. In this way, a) I don't have to wait for a neuron to neuron resolution for word recognition b) I can use what can be done in neuroscience to advance an understanding of language learning and use. Additionally, where a theoretical linguist may believe in a top down rule governing language use, I believe in a bottom up emergence of and socialization to rules. This difference makes communications across disciplines very difficult because it appears to be a difference of worldview and those are nigh impossible to work around.

Read more!

Arbib Interview

Here's an interview with Arbib and his bits on language and mirror neurons. (From Science and Society)

Interesting in its own right, but also interesting to hear Arbib speak to "everyday" people and the interviewers give questions from the educated-but-not-really-knowledgable-about-language perspective.

Nothing below the fold today.

Read more!