# The Landscape for the second semester

 [[Classnotes MOC]] [[Psycholing Classnotes]]

Date:: 2022-01-23

Class:: Psycholinguistics II

Tags:: #umd #classnotes #umd/psylx

## Central Constraint

Language is complicated and varied. BUT people are really good at it. How to build people or their cognition so that they are good at this complicated thing. They mostly succeed and when they fail they fail in a systematic way.

why aspects of the theory of the syntax?

• it underlies mentalistic approach and explain it very well. it shaped a lot of discussions. It also stakes out the relations what is going under the hood of linguistics.
• in order to make practical progress now, you have to make practical choices. What can be done in 67 is not the same what we can do right now.

why how cognition came into being?

• you have to suspend your belief, kind of.
• it is useful to think about how different people was thinking back in the day about issues in linguistics and cogsci.

## what do people do

### processing people

• it is usually very fast.
• they give us process models, focusing on what people do quickly.

### theory people

• process neutral models
• what people can actually do, freed of time/memory
• talk about mental computation a lot, when you poke them, they show little interest in real-time operations.

It feels like there is a linguistics vs. psychology split. it used be that psycholx people stay in psychology department. that is not true anymore.

there is not much of a divide in language development ppl.

this does not matter tho. Psycholinguistics tries to understand that is not much given in proper category. Levels of representation, level of analysis, and change.

levels of representation: every subfield are interested in different type of representations, phonological, semantic, syntactic, feature-wise.

levels of analysis: we can take the same cognitive system, we can describe that at different levels of details. either high level characterization, or very detailed model.

• for example, standard grammatical analysis is very complicated and fine-grained. however, there is no dimension. we abstract away from time/building it up/time steps.
• or you can use the same sentence/information, and talk about/think about time dimension. or memory. or discreet symbolic representations, thus different levels of activations.
• All of these will give you different information and modelling.

change: how these information changes over time. these systems do not stay the same overtime.

• we are mostly interested in individual psychology. I believe though there is nothing stopping us. what about discourse?
• independent of interaction, we can internalize the discourse and the update that goes in an individual psychology.

### colin’s obsession

levels of analysis $$\neq$$ mechanisms $$\neq$$ tasks

when your goal/task is different, it will affect the rest of the pyramid as well. but it will also depend on how specialized the cognitive system.

## Question Formation

• Wallace suspects that everybody doubts that Gromit makes cracking toast.
• What does Wallace suspect that everybody doubts that Gromit makes?

We do not have any trouble tracking anything in the examples above. When we have “anybody who”i nstead of “everybody doubts”, it is very bad.

• Wallace suspects that anybody who doubts that Gromit makes cracking toast.
• *** What does Wallace suspect that anybody who doubts that Gromit makes?

Sentences are like these are called islands, specifically relative clause islands: a wh-word cannot escape from it.

• Where do these type of high-level linguistic knowledge come in?
• Why English and other languages are like these and what do learners need to do to figure this out.

It seems that it is very hard to learn due to poverty of stimulus or negative evidence. Also, almost all languages do the very same. THUS, we want to say it is already there and not learned. So, we go on and test children and try to show that it exists in children.

One thing that come about is maybe this is actually not very hard to learn. Syntactic Islands and Learning Biases by Pearl and Sprouse. They showed that it is somewhat possible to use readily available information and learn islands. Also, Learning Island-insensitivity from the input by Kush Sant and Strætkvern. You thought island was universal? Nope. okay in Norwegian. More, Lidz and Omaki thought that kids are really bad at processing. Maybe they thought they heard stuff, but it was not actually there at all.

There are also a lot of work that people are interested within adult language processing. Can we use evidence from time course to address disputes among linguistics about the right way to encode the relationship between the word and the relative clause. Some ppl say there is an empty position in the original object position and wh-word and the verb have a relation. Others says the empty space is actually occupied and the relationship is between the placeholder and the wh-word.

Maybe time-course and the wierdness affect how you interpret the question?

How many students did the school enlarge the clasroom for _ ?

People may be eager to put students right after the enlarge. What happens if we put it inside an island? Do you still get that weird interpretation? Guess not.

• #task read the Branigan Pickering paper, focus on it. try to unpack it. 📅 2023-01-30 ✅ 2023-01-30