Langston, Psychology of Language, Notes 1

Langston, Psychology of Language, Notes 1 -- Introduction/History

I. Goals.
A. What is psycholinguistics?
B. Why are psychologists interested in language?
C. Brief history.
D. What is the plan for this course?
E. The big problem.

II. What is psycholinguistics?
A. Psycholinguistics as a conversation killer; telling your mother what you do.
B. Psycho linguistics
1. Psycho is from psychology. We aren't going to study language, but how it's used.
2. linguistics is the study of language.
Linguists study the language itself. They generally ask themselves if something sounds correct, and if it does count it as part of a language. Psychologists want to know about the people using the language. When is it used?; How is it used?; How is it understood?
Example of markedness.
C. What sorts of topics are psycholinguists interested in?
1. Language basics: How do you understand spoken language?; How are sentences represented?; What happens during reading?
2. Language use: How is language used to communicate?; How is language used to control/manipulate others?
3. Language acquisition: Where does your knowledge of your language come from?

Top

III. Why are psychologists interested in language?
A. Language is fundamental. If anyone has been in a place where their language is useless (me in Bucharest), you can appreciate how it underlies everything that we do.
B. Intellectual concerns. Language is an interesting problem. As we'll see, no matter what aspect of language production, comprehension, or use you look at, the best explanation for what's taking place is "it's a miracle." Coming up with more scientific answers is a challenge.
C. Practical concerns.
1. Machine translation: During the cold war, it was vitally important to the US government to be able to read every word published in the USSR. The task of translating all of this stuff was enormous. The government funded a lot of research by psychologists and computer scientists to automate this task. Now, business needs for machine translation are still driving research in the area.
2. Language deficits: One way to treat problems with language is to understand how language is ordinarily used. If we work it out we can help people who have strokes or particular language deficits to recover.
3. Learning to read and making that easier.

Top

IV. Brief history. This is superficial. The goal is to outline the origins of some of the debates that are still active today.
A. The middle period. A bunch of influences emerged in psychology, and these impacted the study of language. The big players:
1. Behaviorists: As part of their rejection of mentalism, the behaviorists pretty much killed the investigation of language for a few decades. Gradually, the study of language reemerged as the study of verbal behavior. Some of the things they said about language will help clarify their position:
a. Miller: "In order that we learn to respond correctly to the word 'chair' it is necessary for another organism to intervene and reward us each time we respond correctly."
It's similar to this situation (Staats & Staats, 1963): The kid reaches for the TV and the mother hits its hand (UCS) and says "No!" (CS), and the baby snatches back its hand (UCR). After a few repetitions, the CS ("No!") will get the CR (pulling hand back). Now the kid knows what "no" means.
b. Skinner: "In teaching the young child to talk, the formal specifications upon which reinforcement is contingent are at first greatly relaxed. Any response that vaguely resembles the standard behavior of the community is reinforced. When these begin to appear more frequently, a closer approximation is insisted upon. In this manner very complex verbal forms may be reached."
In other words, the kid says "wahwah" and gets a glass of water. But, as it gets older, parents hold out for "wahda," "watah," "and water."
This is not accurate. There are two concerns here, but this is only the tip of the iceberg.
a. Brown, Cazden, and Bellugi (1969): With respect to grammar, parents reinforce truthfulness of utterances, not how grammatically correct they are:
Child: "He a girl."
Mother: "That's right."
and
Child: "There's the animal farmhouse."
Mother: "Wrong." (Because it's a lighthouse.)
b. Some psychologists were also concerned with the behaviorists' use of nonsense syllables (Ebbinghaus' nonwords). The idea was if you want to investigate acquisition of language, you had to start with something a person didn't already know. You know a meaning for "book," but not for "plak." This didn't really seem like studying language.
2. Information theory: The phone company wanted to get more information to go on the wires. Information theory developed to solve that problem. The basic idea is that information comes in bits (the unit of measurement). Each bit has two values: on or off. If you're Paul Revere, there are two ways the British could invade, so you need one bit (lantern) to transmit it. Land = on, Sea = off. If the British could come 4 ways, you'd need two bits (lanterns): Land = on on, Sea = on off, Air = off on, Submarine = off off. This can extend indefinitely. If you have N bits, you can transmit 2^N unique messages. Or, N messages can be sent with log₂N bits. If you can say 64 things, you only need 6 bits...
Psychologists were very excited by this. Talking is transmitting information and listening is receiving information. We can use the math from information theory to characterize that. Some stuff:
a. Language is full of redundancy. How much information can a letter transmit? There are 26 letters (messages), so each one transmits log₂26 bits. That's 4.5 bits. But, each letter isn't equally likely, so if you take that into account, each letter actually transmits about one bit of information. In other words, 3.5 bits of potential is wasted. You can see it in this example:

(1) MST PPL CN RD THS SNTNC.

I took out all the vowels, no problem.
b. The more information something transmits, the harder it is to identify. Words transmit tons of information. If you have just 256 of them, each one transmits log₂256 bits (8 bits). You actually know about 30,000 words (if you're really smart). That's around 15 bits per word. You can see the importance if you try to fill in the word in this sentence:

(2) The witness was examined by the ________. (window)

It's pretty hard to know what's coming.
How is this applied to psychology? Imagine that you're at a loud party and you're trying to listen. If you're talking about your friend's weekend, you can probably follow along pretty well because you can predict the sorts of things to be discussed. If you're trying to catch this lecture, the words carry more information (they're harder to predict), and it should be harder to understand.
Miller, Heise, and Lichten (1951): Have people identify words from a set of two words (one bit) up to a set of 256 words (8 bits) against a noisy background. The more choices there are (the more information each word carries), the less noise people can tolerate.
3. Linguistics: Chomsky (1950s): Language can't be acquired using the behaviorists' learning principles. A sample of the things he said:
a. Poverty of the stimulus argument: Children don't get reinforced for every utterance, and the sorts of reinforcement they do receive won't explain the grammatical rules they eventually develop. Consider the experiment described above.
To put it another way, you know a lot of rules that you, as members in a language community, have a hard time expressing. As an example of this, look at these two sentences:

(3) John wants him to win.
(4) John wants Bill to see him.

In (3), "him" can't be John. In (4) it can. What's the rule that distinguishes the "him" in (3) from the "him" in (4)? Could this be a result of reinforcement for these types of sentences?
b. Associations can't account for all aspects of language. For example:

(5) Colorless green ideas sleep furiously.

is a sentence for most people, even though there are few (or no) associations between its words. (Read some Jabberwocky)
c. Competence/performance: Chomsky puts the mentalism back into play. Competence is what you know about language, performance is what you make. Competence is entirely mental. It can be used to introspect about language (what we did on the last example). It's pretty much unrelated to what you learned in school. In fact, what you learned in school makes up a tiny percentage of what you know about grammar. Example: Pluralize "blit", "blib", and "blish" (out loud). Did you learn the pronunciation rule in school?
This distinction is going to make it hard to study children's acquisition of language. They can usually understand grammatical forms before they can make them. How do we know when they learned them?
C. Present: The cognitive revolution reawakened interest in a lot of language phenomena. A short list: discourse, conversation, sociolinguistics, gender and language, neuropsychology. The basic idea is it's OK to study mental stuff, so we go nuts. A couple of holdovers from behaviorism:
1. We rely almost exclusively on data. Note that a lot of that data is based on introspection of individuals, but it seems more sciency than sitting around on your own thinking about language.
2. We still don't know much about what's going on in the black box, so we focus on the inputs and outputs.

Top

V. What is the plan for this course?
A. We're going to read Pinker's book. This is an introduction to the terms and issues. I'll lecture over material appropriate to each chapter, we'll have a few exercises and discussions. You need to read the book. But, it's light reading, and you should enjoy it.
B. We'll take what we know from the book and apply it to questions that interest us.

Top

VI. The big problem. You have noticed by now that the title of the book is The Language Instinct. Pinker has come down solidly on the nature side of the nature/nurture argument. Obviously, I agree with him (on some things) or I would have chosen a different book. But, this is a controversial position. The other side is that language is learned. Pinker characterizes this as the "meatloaf" model. The brain is a bunch of the same kind of stuff. It uses general learning principles to learn everything, including language. Elizabeth Bates is a major player on that side of the fence. I will try to present balanced coverage of both sides as we go along, but I want you to be aware at the beginning that there are other viewpoints.
As it stands now, choosing sides in this debate is as much a matter of faith as it is science. The data can be interpreted as supporting either side, depending on how you look at it.

Top

Psychology of Language Notes 1 Will Langston

Back to Langston's Psychology of Language Page