Class 13 - Where do we go from here?

Question 1

EEG

Answer 1

captures the electrical brain response to stimulus presented.

Question 2

Unified theory of language research

Answer 2

Approach that combines neurobiology (NB) with computational representational theories (CRT), allowing CRT to be the center of a unified theory of language research, while still leaving space for NB to:

• select CR theories
• explain these CR theories

The need for such a theory comes from the lack of interpretation of abundant data coming from NB research.

Question 3

Computational representational theories (CRT)

Answer 3

the basis of language research, made up from data from linguistics, psycholinguists and computational linguistics.
Not a direct measurement of brain activity though, that comes from neurobiology.

Question 4

Neurobiological theories (NB)

Answer 4

More accepted since they provide direct observations of the brain, but they should work together with CRT according to the authors.

Question 5

Marr’s levels

Answer 5

Marr proposed a model that handles complex information processing. The authors use this model to explain how research has been conducted so far in the field of linguistics.

Made up of three levels:

• computational level
• representation level
• implementation level

The authors suggest that CRT research (computational linguistics, psycholinguistics) has been focusing on the first 2 levels, while NB has been focusing more on the 3rd one.

Question 6

Computational level

Answer 6

What and why? (considered the first most important till now)

Question 7

Representation and algorithm level

Answer 7

How? (considered the second most important till now)

Question 8

Hardware, implementation level

Answer 8

Where? (not considered that much till now, but the authors argue against this): they argue we should combine CRT and neurobiology.

Question 9

Unifying issues

Answer 9

Occur when we try to combine CRT and NB

Question 10

OIP (Ontological Incommensurability Problem)

Answer 10

NB and CRT do not have counterparts for some of their elements in each other. They have their own valid definitions for themselves and they shouldn’t change them. This “mismatch” is only a problem in theory, it doesn’t prevent from aiming at a unified theory of language.

Question 11

GNP (Granularity mismatch problem)

Answer 11

There is a mismatch between the ‘conceptual granularity’ of the elemental concepts of linguistics and the elemental concepts of neurobiology and cognitive neuroscience

source: http://www.ling.upenn.edu/~embick/lingneuro.pdf

Question 12

Correlational (neuro)linguistics approach

Answer 12

Finding correlations between CRT and NB data.

Has been done a lot in the past of research (Obleser)

Question 13

Integrated (neuro)linguistics approach

Answer 13

NP data will help the researchers to select among the appropriate CRT theories.

Question 14

Explanatory (neuro)linguistics approach

Answer 14

Brain structures are specialized in specific processing and representations and these explain brain structure.
Example of barn owls –> their auditory system is shaped specifically for one task: surviving by hunting.

Question 15

Barn owls

Answer 15

Its head is shaped to detect sounds: its feathers channel sounds to its ears.
Example of the Explanatory (neuro)linguistics approach in animals, we don’t have an example for humans.

Question 16

Magnet effect

Answer 16

Prototypical sounds of our native language map non-perfect sounds we hear.

Question 17

According to the experiment led by Obleser presented in the lecture, can phonemes be compared to atomic structures which can’t be further “decomposed”?

Answer 17

No: the experiment showed that phonemes with different articulations/formants also have different brain representations. Therefore, we can think of phonemes as units formed by smaller particles (formants, articulations, etc..) that make them up.