Today

Homework
Where we are
Introduction to grammar

Where we are

We've looked at how words work:

· Their meanings and the cognitive processes underlying them are grounded in human embodiment – they depend on the particularities of the individual’s interactions with the world, in a particular sort of body, with the sorts of mental faculties we have, in the context of other humans, with social goals, interactions, etc.

· Moreover, words can have multiple meanings or gain new meanings. We've seen cognitive mechanisms for this like metonymy, metaphor, and construal.

· The words that a language has can influence how speakers of that language think - what aspects of the world they attend to, how they categorize experiences, and what they think is similar or different.

Grammar

The third part of the course addresses how we productively combine words and morphemes to create more complex structures, like complex words and sentences. This is grammar: the study of what people know about how words and morphemes combine to produce meaningful patterns.

Background on Chomskian linguistics

Until the late 1950s, the study of the language of individuals was largely behaviorist. The mind was treated like a black box, where all you could study was what went in and came out of human language users.
Noam Chomsky, an MIT linguist, advocated viewing language instead as a window to the mind, and language has since been embraced as one as the major areas of cognitive science.

Properties of Chomskian linguistics

The goal of Chomskian [aka Generative] linguistics: make the study of language more rigorous, by developing a formal model of language, which accounts for grammaticality.

Chomsky observed that there are more grammatical sentences than can be listed, due to the compositionality of language.

The box is on the table in the corner of the room at the top of the stairs in the back of the house...

In fact, language is recursive, so there's an infinite number of grammatical utterances.

The monkey killed the student.

The monkey that the professor trained killed the student.

The monkey that the professor that the police were searching for trained killed the student.

Language on this view is a Generative system, which produces grammatical utterances and not ungrammatical utterances.

I love my good monkey.

*I monkey my love good.

The Chomskian notion of grammaticality has nothing to do with meaning

Colorless green ideas sleep furiously.

Why we should be grateful to Chomsky

Brought language to prominence as a cognitive function - about 1/4 of talks at the annual international Cognitive Science Society Conference are about language
Brought rigor to the study of grammar
Views language as a mental function, not just as something shared by a community

But there are lots of things that are profoundly wrong with a Chomskian approach to language, based on a large number of unsubstantiated assumptions:

The study of language is the study of grammaticality

What is grammaticality? It's whether a sentence is grammatical or not, that is, whether it is produced by a grammar or not. You see the circularity here?
Many people use grammaticality judgments. But it's not clear what these mean

They conflate grammar, conventionality, meaning, pragmatics, use, context
They ask for an opinion, so they invite meta-linguistic processing
They can't be direct reflections of grammatical knowledge, because of the competence-performance distinction, discussed below

Language is for communication, instruction, other social purposes. Studying language through grammaticality judgments is like studying running by asking people about what they think about how they run. Why not study it directly?

The study of language need not be embedded in the study of language use

You can study sentences in isolation, and build theories to explain why sentences are grammatical or not independent of the individuals who actually use language or mechanisms the use for language.
This is so wrong, it's hard to even articulate.

Syntax is modular - i.e. it does not depend on any other aspects of language

There is no evidence for this, and we'll see some evidence against it.

Language is modular, i.e. it does not depend on any other cognitive mechanisms

We've already seen evidence against this.

Variation across individuals is irrelevant to grammar

But there's lots of variation across individuals, which in principle cannot be accounted for by the Generative program, which studies an imaginary "ideal speaker-listener, in a completely homogenous speech community" [Chomsky, 1965:3]

There is a qualitative distinction between core [universal] & peripheral [language-specific] grammar

Only core grammar is part of universal linguistic knowledge and therefore worthy of linguistic study. If you assume this, you miss out on something like 90% of grammar.
Strangely, core and peripheral grammar are very similar in their patterning, acquisition, and behavior. Must just be coincidence.

Competence is distinct from performance

In many contexts, the distinction between language knowledge, competence, and language behavior, performance, is a reasonable distinction, e.g. speech errors
But in practice, it is used to shield grammatical theory from actual language use. Sentences can be deemed grammatical even though people judge them ungrammatical. And vice versa.

E.g. multiple embedding of RCs.

Meaning is lexically compositional - only words and morphemes provide meaning.

Patterns for combining words and morphemes have no meaning - they are simply different formal structures that are related through syntactic transformations.
We will see that they differ in meaning, information structure, use, and pragmatics

Meaning is logical, objective

We've seen that this is not true

On the basis of these faulty assumptions and some bad reasoning, the Generative approach comes to the incorrect conclusion:

One speaker’s grammaticality intuition is evidence about all speakers’ mental systems.

The fallacious reasoning here is exceptional. Assume that [contrary to observation] all speakers of a language have the same grammar. Then conclude that you can observe one individual and learn all there is to know about the grammar of the language.
Even if the assumption were correct, the conclusion would not be. You cannot fully understand the human genome by studying a single individual.

Construction Grammar

The approach to grammar we will be adopting doesn't make any of these assumptions. Rather:

The study of language is the study of all aspects of language behavior, including meaning, interpretation, understanding, acceptability
We study language in terms of the way it's used, including the cognitive mechanisms it uses
We are not theoretically predisposed against the possibility that syntax and language more broadly aren't strictly modular
We assume no competence-performance distinction, but there is self-monitoring
We make no assumptions about variation.
We make no assumption about a core-periphery distinction
We make only two assumptions:

Language knowledge consists of form-meaning [or form-function] pairings, and these may be more or less specific and more or less complex.
Language is embodied - that is, it is part of the human cognitive system.