The Chomskyan Revolution - Prof. Fredreck J. Newmeyer

1

FREDERICK J. NEWMEYER

UNIVERSITY OF WASHINGTON, UNIVERSITY OF BRITISH COLUMBIA,

AND SIMON FRASER UNIVERSITY

Class 1:The Chomskyan Revolution

2

MY GOAL IN THESE CLASSES

To present the progression of ideas in linguistic theory as they have developed.

To take one topic in each class and chart its development over the decades.

3

MY GOAL IN THESE CLASSES

LECTURE 1: THE CHOMSKYAN REVOLUTION

LECTURE 2: CONSTRAINING THE THEORY

LECTURE 3: THE BOUNDARY BETWEEN SYNTAX AND SEMANTICS

LECTURE 4: THE BOUNDARY BETWEEN COMPETENCE AND PERFORMANCE

LECTURE 5: THE TREATMENT OF GRAMMATICAL COMPLEXITY (at Peking University)

4

SAUSSURE IS A GOOD STARTING POINT

FERDINAND DE SAUSSURE1857-1913

Cours de linguistique générale,

published in 1916

5


All of the important strands of modern linguistics can be found in Saussure’s Cours.

Saussure himself was primarily an Indo-Europeanist.

He accepted the Neogrammarian idea of the regularity of sound change.

6


But Saussure’s greatest contribution went beyond the assumption of mere regularity.

Saussure posited a series of sounds for PIE that had no reflexes in any known IE language.

He was led to this hypothesis by thinking of PIE as a STRUCTURAL SYSTEM.

7


A few years later a linguist suggested that these ‘mystery sounds’ might be laryngeals.

Decades later the first extensive texts in Hittite — an ancient IE language — were discovered and decoded.

Hittite turned out to have sounds (probably laryngeals) in precisely the positions posited by Saussure for PIE!

Saussure’s ‘structuralist thinking’ had been confirmed.

8


The basic idea of the Cours is a ‘structuralist’ one, though Saussure did not use the term.

The idea is that language as a whole can be divided into langue (language/competence/I-language) and parole (speech/performance) and that langue can be studied as a formal system.

Between the two world wars, this idea was being worked out in different ways in different countries.

9

STRUCTURAL LINGUISTICS

The Prague School, with its focus on language universals.

NIKOLAÏ TRUBETZKOÏ ROMAN JAKOBSON

1890-1938 1896-1982

10


The Prague School in many ways is the intellectual ancestor of generative grammar:

UniversalsMarkednessRelating facts about grammar to

language acquisition, language pathology, etc.

But this was all in phonology. Their view of syntax was a very ‘functionalist’ one.

11


The Copenhagen School (Glossematics), with its focus on formal relations among elements.

LOUIS HJELMSLEV 1899-1965

12


The London School, now best known for its work on suprasegmental phenomena.

J. R. FIRTH, 1890-1960

13


The London School linguists can be considered the founders of the field of applied linguistics.

They believed that if a theory is correct it should be applicable to practical concerns, such as language teaching.

M. A. K. HALLIDAY

14


LUO CHANGPEI1899-1958

WANG LI1899-1958

LÜ SHUXIANG1904-1998

15


CHAO YUEN REN (ZHAO YUANREN)1892-1982

The greatest Chinese structuralist

President of the Linguistic Society of America in 1945

16

AMERICAN STRUCTURALISM

American structuralism, famous for its attention to methodology and numerous studies of Amerindian languages.

EDWARD SAPIR, 1884-1939 LEONARD BLOOMFIELD, 1887-1949

17


American structuralism was the most direct ancestor of generative grammar. Zellig Harris, one of the most prominent American structuralists, was the teacher of Noam Chomsky.

ZELLIG HARRIS, 1909-1992

NOAM CHOMSKY

18


American structuralism was famous for its commitment to empiricism.

Only observables and generalizations about observables were admissible.

This followed from the positivist philosophy of science that was dominant in English-speaking countries from the 1920s to the 1960s.

19


The idea was to start with the lowest, most observable, levels, and to build a grammar from that point:

1. phonetics2. phonemics3. morphology4. syntactic categories5. relations governing the distribution of

syntactic categories.

20


Each level was built up by procedures of segmentation and classification from the next lower level:

[pʰɑrkʰɨŋ] phonetic representation/parkiŋ/ phonemic representation{park+iŋ} morphemic representation[V [Npark] [PARTiŋ]] syntactic representation

21


Because (in theory) you couldn’t discover the syntactic categories of a language until after you had done the phonemics, no phonemic statement could refer to a syntactic category.

A big problem:Verbs often have final stress in English:

perMIT, transPORT, conCEIVE, reFUNDNouns often have initial stress: PERmit,

TRANSport, CONcept, REfund

22


Meaning (semantics) was also a big problem for American structuralists.

Meanings are essentially unobservable.

There was no way to arrive at them by procedures of segmentation and classification

So many American structuralists did not talk about meaning at all.

23


By the late 1940s, Zellig Harris was working out procedures for relating sentence types to each other.

He called the result of these procedures ‘transformations’.

N1 V N2 <-> N2 is V-ed by N1 (passive transformation)

N1 V N2 <-> it is N2 that N1 V (cleft transformation)

But these transformations were stated directly on the surface.

24


What all of the approaches to structural linguistics had in common was the idea that a grammar is an inventory of elements in a structural relationship to each other.

At the same time, these models posited very little ‘below the surface’.

25

EARLY GENERATIVE GRAMMAR

The field of linguistics underwent a major change of direction as a result of the work of Noam Chomsky in the early and mid 1950s.NOAM CHOMSKY

26


Chomsky’s bookSyntactic Structures, published in 1957

27


Generative grammar was different from all previous approaches in that it was syntax-centred.

Saussure for the most part thought that syntax was part of parole.

Saussure felt that because speakers have a ‘choice’ of what syntactic construction to use, syntax was part of language use.

But Chomsky saw rule-governed creativity as central to competence/langue.

28


What made Syntactic Structures important?

1. Looking at a grammar as a ‘theory of a language’

2. Stressing the irrelevance of methodology3. Providing evidence for the autonomy of syntax4. The abstractness of particular grammatical

analyses

29


1. Looking at a grammar as a ‘theory of a language’

That is, the twin tasks of:Characterizing the notion ‘possible human

language’For any particular human language,

providing the best (that is, most psychologically realistic) analysis of that language.

30


2. Stressing the irrelevance of methodology

Scientists don’t care how somebody arrived at their theory (Kekulé on the structure of benzene)

The only important thing is whether the theory adequately models reality.

So all of the structuralists’ procedures get thrown out the window.

31


• 3. Providing evidence for the autonomy of syntax.

• Chomsky argued that syntax had a patterning of its own, independent of meaning, function, and discourse.

• The American structuralists believed that too, but for them it was a consequence of their methodology.

• Chomsky provided empirical evidence for autonomy (the next class).

32


4. The abstractness of particular grammatical analyses

Mary must have be en be ing se en

perfect progressive passive

The elements of the English auxiliary are both overlapping and discontinuous.

There was no way for structuralist methodology to come up with an adequate analysis.

33


Chomsky’s solution:

Posit 2 classes of elements: verbal (v) and affixal (af)

Posit an abstract phrase structure rule:

AUX Tense (Modal) (have + en) (be + ing) (be + en) af v v af v af v af

•Posit an abstract transformational rule: af + v v + af

34


Mary must have + en be + ing be + en see

Mary must have been being seen

This analysis extends to capture the placement of supportive do in sentences like Did Mary see the play? and Mary did not see the play.

35


Syntactic Structures also emphasized and re-emphasized the need for ‘simplicity’ and ‘economy’ in grammatical description.

This would become a major theme in later minimalist theorizing.

36


The idea of ‘elegant’ grammatical descriptions did not — of course — start with Chomsky:

“Grammarians rejoice more over the saving of half a mora than over the birth of a son.” (the ancient Indian grammatical tradition)

INDIAN STAMP HONOURING PANINI

37


“Saussure’s dearest concern was to cast the theory of language into the mould of a mathematical treatise.” (Robert Godel)

FERDINAND DE SAUSSURE 1857-1913

38


The American structuralists — Chomsky’s teachers — were the most adamant of all that linguistic descriptions need to be formal, simple and elegant.

ZELLIG HARRIS, 1909-1992

39


The striving for elegance is the one factor that unites all mainstream versions of generative grammar throughout its history — including early transformational grammar, Generative Semantics, and the Minimalist Program.

And it is a factor largely rejected by most functionalists and cognitive linguists.

40

GENERATIVE PHONOLOGY

At the same time in the early 1950s, Morris Halle (along with Chomsky and others) was developing a generative approach to phonology.

MORRIS HALLE

41

morphophonemics

phonological

rules

allophonic rules

PHONETIC LEVEL

PHONEMIC LEVEL

MORPHOPHONEMIC LEVEL

GENERATIVE PHONOLOGY

42

THE EARLY DAYS OF GENERATIVE SYNTAX

There were many changes to the theory between 1957 and 1965, the publication date of Chomsky’s Aspects of the Theory of Syntax.

Most of these are still assumed to be correct.

Given the tiny number of generative grammarians at the time, most were made by Chomsky and his colleagues and students at MIT.

43


1. A separate lexicon.

Syntactic Structures:VP V (NP) (PP)NP (DET) NPP P NPV run, eat, think, persuade, …DET the, a, this, …N book, boy, John, Mary, …P in, of, over, …

44


By the early 1960s, each word was listed in a separate lexicon, with its own idiosyncratic properties:

boy N, +human, +animate, /bɔi/hit V, +physical action, +___NP, /hit/elapse V, +time, +____#, /ilæps/on, P, +space, /ɔn/

The separate lexicon simplified the phrase structure rules and allowed for the statement of purely lexical generalizations.

45


The lexicon also solved the problem of subcategorization.

Here are some generalizations:The noun boy is human and commonThe noun book is nonhuman and commonThe noun Charlie is human and properThe noun Egypt is nonhuman and proper

• Before a separate lexicon the would be two very cumbersome ways of stating this generalization:

46


N -> N human N nonhuman

N human -> N human and common N human and proper

N nonhuman -> N nonhuman and common N nonhuman and proper

N human and common -> boy,... N human and proper -> Charlie,... N nonhuman and common -> book,... N nonhuman and proper -> Egypt,...-------------------------------------------------------------------------------------------------

N -> N common N proper

N common -> N common and human N common and nonhuman N proper -> N proper and human N proper and nonhuman N common and human -> boy,... N common and nonhuman -> book,... N proper and human -> Charlie,... N proper and nonhuman -> Egypt,...

47


Here’s how to do it with a separate lexicon with features:

+HUMAN -HUMAN

+COMMON boy book

-COMMON(PROPER)

Charlie Egypt

48


The early 1960s also saw a solution to the ‘traffic rule’ problem: how rules interact with each other.

It seemed clear that some rules were optional and some were obligatory:

PARTICLE MOVEMENT (optional): I looked the answer up and I looked up the answer (both are grammatical).

AFFIX HOPPING (obligatory): You can’t leave the auxiliary like *Mary must have + en be + ing be + en see

49


It seemed clear that rules had to be ordered with respect to each other:

The subject of a passive gets nominative case, even though it started out as an object:

He saw her.She was seen by him.*Her was seen by he.

So Case Marking has to apply after Passive.

50


The next traffic problem to solve was the order in which transformations apply when there are several clauses in one single sentence.

I think that Mary knows that Bill wants to leave.

The question was ‘What gets embedded first, second, etc.?’

I think S Mary knows S Bill wantsBill leave

51


1964: Grammatical rules apply on the most deeply embedded clause first, then the next most deeply embedded clause, that is starting from the bottom and moving to the top.

CHARLES FILLMORE, 1929-2014

52


Mary knows

Bill leave

Bill wants

I think

53


It was quickly realized that, if that is true, we don’t need embedding transformations at all!

The phrase-structure rules can generate one object, now called DEEP STRUCTURE.

Transformational rules CYCLE from the bottom to the top.

54


Mary knows

Bill leave

Bill wants

I think

55


1964: The Katz-Postal Hypothesis: Transformations do not change meaning (= everything for interpretation is in the Deep Structure)

PAUL POSTALJERROLD KATZ, 1932-2002

56

THE 1965 ASPECTS OF THE THEORY OF SYNTAX MODEL, THE ‘STANDARD

THEORY’

57

MAJOR INFLUENCES ON GENERATIVE GRAMMAR

I. Structural linguistics

II. Developments in mathematics and formal logic A. The axiomatic-deductive method: starting with a small

number of axioms and procedures, one derives (generates) a set of propositions (Harwood)

B. Recursive function theory (Gödel, Church, Turing, Kleene, Post)

C. Constructional system theory: reformulating philosophical propositions in a mathematical language (Carnap)

D. Simplicity measurements of formal systems (Goodman)

58


III. Developments in the philosophy of science

Before the 1950s, empiricism was dominant in Anglo-American philosophy.

The idea of empiricism: All knowledge comes from experience.

The old idea was that scientists observe, measure, and generalize, but not much more.

59


But empiricism had begun to break down by the 1950s.

Empiricists tried and tried to characterize what makes a statement ‘scientific’.

First idea: a statement is scientific if it can be directly verified by empirical evidence.

NO!! That would rule out general laws (and also rule out historical sciences – evolution, historical linguistics, etc.)

60


Second idea: a statement is scientific if it can be falsified by observational evidence.

NO!! That would rule any statement with an existential quantifier (‘For every compound there exists a solvent’; ‘There exists a galaxy more massive than ours’).

61


Third idea: a statement is scientific if it can be given an operational definition, that is, linked to observation by a set of procedures.

NO!! Many terms in science cannot be defined operationally: ‘ideal gas’, ‘perfect vacuum’.

We could calculate the speed of light before we knew what light was. Atomic theory before atoms. Genetic theory before genes.

62


The dominant current idea: A theory is scientific if it is:

A. FormalB. Explanatory and predictiveC. SimpleD. Empirically testable

These were the criteria that Chomsky used in Syntactic Structures and elsewhere.

63


IV. Developments in psychology

Empiricist philosophy has a counterpart in psychology: behaviourism

Behaviourism says that there are no innate predispositions: all behaviour is a result of stimulus, response and reinforcement.

Behaviourism was also on the retreat by the 195os (the problem of planning, for example)

64

THE CHOMSKYAN REVOLUTION: SOME SOCIOLOGICAL ASPECTS

Syntactic Structures was published in 1957. By 1965 (if not earlier) there was talk of a ‘Chomskyan Revolution’ in linguistics.

How did they theory succeed so quickly (at least in the United States)?

Many linguists found the foundations of the theory plausible and the analyses convincing.

65


The sheer amount of effort that Chomsky and his colleagues went into publicizing the theory.

Between 1957 and 1969 Chomsky published 6 books, 20 articles and reviews in refereed journals, and 25 articles in edited volumes. In addition to English, he published in Hebrew, French, and Japanese.

The unusual situation of MIT.

66


The expansion of the American university system.

The 1960s were a period of confrontation.

Pure good luck: The 9th International Congress of Linguists, held in Cambridge, Massachusetts in 1962.

67


Before the 1970s virtually all generative grammarians were American men.

Things are very different today.

Most would agree that well over 80% of the most prominent generative grammarians are not in the US.

Four out of the last five presidents of the Linguistic Society of America have been women.

68


It has always been the goal of the MIT department to get beyond Americans working on English.

Of the six faculty members in the MIT Linguistics Department in the late 1960s, four were known primarily for their work in languages other than English:

Kenneth Hale for Amerindian and Australian; G. Hubert Matthews for Amerindian; Paul Kiparsky for general Indo-European;Morris Halle for Russian.

Chomsky wrote a partial generative grammar of Hebrew before working on English.

69


Of the 28 doctoral dissertations written in linguistics at MIT in the 1960s, 17 dealt primarily with languages other than English:

Stephen Anderson (West Scandinavian)George Bedell (Japanese)Thomas Bever (Menomini)James Fidelholtz (Micmac)James Foley (Spanish)James Harris (Spanish)Richard Kayne (French)Paul Kiparsky (various languages)Sige-Yuki Kuroda (Japanese)Theodore Lightner (Russian)James McCawley (Japanese)Anthony Naro (Portuguese)David Perlmutter (various languages)Sanford Schane (French)Richard Stanley (Navajo)Nancy Woo (various languages)Arnold Zwicky (Sanskrit)

70


Generative Linguistics in the Old World, founded in 1977.

JAN KOSTER JEAN-ROGER VERGNAUD HENK VAN RIEMSDIJK 1945-2011

71


Generative linguistics was developed in Taiwan before the PRC.

Tang Ting-Chi (Ph. D. University of Texas, 1972)

Mei Kuang (Ph. D. Harvard University, 1972)

Both became important figures in Taiwanese linguistics.

72


Tang and Mei taught or inspired some of the leading Taiwanese generative linguists:

C.-T. JAMES HUANG Y.-H. AUDREY LI W.-T. DYLAN TSAI

73


Generative grammar started making an impact in mainland China in the 1980s.

NING CHUNYAN XU LIEJIONG

74


In 1985 and 1987 Ning organized two Harbin Conferences on Generative Grammar at Heilongjiang University.

They were attended by 400 linguists from inside China and 60 from outside China.

75


Education

The Chomskyan Revolution - Prof. Fredreck J. Newmeyer