American Structuralism Source: https://fr.scribd.

com/document/420028921/LinguisticsLecturesfor2nd-
yearLMDStudents

Introduction
It is agreed upon that the American linguistic studies emerged from the institutes of anthropology
rather than from the institutes of languages. The American scholars were anthropologists who developed
structural ideas far away from European work. They worked on existing languages, the Amerindian
languages. Field work techniques of anthropologists characterized their approach. These languages did
not have written records or previous descriptions_as opposed to the European languages. Therefore,
their historical aspects were discarded. The Amerindian languages were very different from the European
ones. Thus, American structuralists, avoiding the prescriptive attitude, were in need to develop fresh
descriptive frameworks fitting these languages’ actual features. American work emphasised the uniqueness
of each language’s structure, similar to the European tradition. The leading figures of the American
structural studies were Franz Boas, Edward Sapir, and Leonard Bloomfield.

I. Features of American Structuralism

In order to avoid the dangers implicit in traditional grammar, American linguists had the following
aims:
▪ To describe current spoken language, not dead languages;
▪ To focus on language form as a sole objective, thus neglecting meaning to a subordinate place;
▪ To perform the description of language using an organized, unprejudiced and meticulous method which
allows the analyst to extract the grammar of a language from a corpus of recorded data in a quasi-
mechanical way following four steps:
a) Field recordings of a corpus of data;
b) Segmentation of the utterances of the corpus at different levels: phoneme, morpheme, word, group,
clause and sentence;
c) Listing an inventory of forms thus obtained from each level and stating the distribution (possible
environment) of the forms;
d) Classifying the forms (by giving them names) and utterances of the language being studied.
Only such an essentially classificatory method could enable them, it was thought, to concentrate
systematically without any predetermined framework, on the unique structure of the language under
examination.

II. American Structuralists

A. Franz Boas (1859–1942)

Boas was the leading figure in anthropological work in early 20th century. Interested in describing the
Amerindian cultures, particularly American Northwest ones, Boas focused on languages because they
represented the best channel for classifying the aboriginal cultures. He objected to the use of grammatical
categories of the Indo European languages in describing Native American languages. For him, such a
tradition would distort the features of these languages. The most important publication of Franz Boas was
the Handbook of American Indian Languages (1911–1941).

B. Edward Sapir (1884–1939)

Edward Sapir was one of the students of Franz Boas. He was himself an anthropologist and a linguist at
the same time. His important publication was his book Language (1921). Adopting a descriptive approach,
he studied, together with Boas, a number of Amerindian disappearing languages. By and large, Sapir’s
approach to language was based on the exploration of the relations with literature, music, anthropology and
psychology. His outlooks on language insist on its impact on every part of human life.

Sapir is well-known for a theory called the Sapir-Whorf Hypothesis (also relativity, determinism,
Humboldtism, or Whorfian Hypothesis). Developed after his death in the 1950s, it was the product of the
beliefs of both Sapir and Benjamin Lee Whorf (1897- 1941) on the relationship of language to thought.
According to the strong version of the theory, our vision of the world is heavily determined by our language:
the grammatical structures of a language shape its speakers’ perception of the world. Much criticism
was levelled at the Sapir and Whorf hypothesis; for example, translating between languages is possible, and
this process does not impose a change in world view. On that basis, a weaker version of the hypothesis
appeared, stating that language influences thought.

C. Leonard Bloomfield (1887–1949)

Leonard Bloomfield is the father of modern American linguistics. His masterpiece in linguistic studies
Language (1933), established the track of the scientific study of language in the United States till the early
1950s. Crucial in Bloomfield’s work was his influence by behaviouristic psychology, which rejects all that
is non-physical or non observable in search of being empiricist in approach. He conceived of language
basically as couples of stimuli and responses. Bloomfield maintained that language should be studied like
a natural science. Most importantly, he made influential contributions to the development of vigorous tools
for the analysis of language.

1. Immediate Constituent Analysis (ICA)

In addition to his remarkable contribution to the fields of phonology and morphology, Bloomfield’s name
is usually attached to a pioneering syntactic theory called immediate constituent analysis (ICA).
Basically, ICA is an explicit method of analysing sentences grammatically by dividing them into their
component parts. It is structural in nature because it no longer considers a sentence as a sequence or string
of isolated elements, but it is made up of layers of groups or constituents. A constituent is a group
of words or morphemes with closer relationships between one another than between the elements of the
other groups or constituents within the same sentence. The constituent is part of a larger unit.

The methodology of ICA consists in splitting a sentence up into two immediate constituents, which are
analysable into further constituents. This process of segmentation continues until the smallest indivisible
units, the morphemes, are reached. The latter are called the ultimate constituents, and each is given an
identifying label. As a principle, the partition in ICA is binary. Let us take Bloomfield’s classical example
“Poor John ran away”. To show divisions in this sentence, it is possible to use two ways:

- Bracketing

According to ICA, a sentence is not seen a string of elements but it is made up of layers of constituents (or
nodes). Thus, constituent structure is hierarchical.
2. Weaknesses of ICA
In spite of its popularity and scientific rigour, ICA was shown to involve inherent limitations because as a
model of language description, its descriptive framework did not cover all the aspects of language that
constitute the knowledge of a native speaker, and it contained some analytical inconsistencies. The main
weaknesses for which this analysis is reprimanded are the following:
a) In some sentences, it is not always clear where the division should be;
b) ICA does not indicate the role or function of constituents as they are not labelled. When parsing is done,
some implied grammatical information is included (circularity of argument);
c) In ICA division is arbitrarily binary, while some sentences may have alternative analyses;
d) The analysis in ICA does not go beyond the morpheme;
e) Because it focuses only on the surface of the sentence (formal properties), ICA cannot show the
syntactic relationship between sentences which are superficially different (active/passive,
positive/negative) and fails to show the differences between sentences which are superficially similar;
f) ICA cannot handle lexical and syntactic ambiguity in the sentence;
g) ICA does not demonstrate how to form new sentences;
h) ICA cannot handle sentences with discontinuous elements;
i) ICA cannot handle complex sentences.
Post-Bloomfieldianism (Distributionalism)

Distributionalism is a branch of American structuralism developed in the 1940s and 1950s. Its goal is
an experimentally verifiable, objective description of the relations inherent in the systems of individual
languages, excluding all subjective and semantic factors. The notion of distribution is a central feature of
American work. It was introduced by Zellig S. Harris in his book Methods in Structural Linguistics
(1951). In fact, Harris developed Bloomfieldianism to its extremes, and paved the way for the
appearance of transformational generative grammar, being Chomsky’s teacher.
The structure of each individual language can be described by means of experimental methods, the so-
called discovery procedures, in which essentially two analytical steps are applied: (a) segmentation of the
material through substitution. i.e. through paradigmatic interchangeability of elements having the same
function (paradigm); and (b) classification of elements as phonemes, morphemes, among others, on the
basis of their distribution and environment in the sentence.
Distribution is the number of possible environments (contexts or positions) in which a particular linguistic
item (phoneme, morpheme, phrase, and clause) can occur in a given language. Every item has specific
distributional properties. The following types of distribution can be distinguished:
∙ Equivalent distribution: when elements occur in identical environment either in (i) free variation
without distinguishing meaning, (e.g. in the alternation of [i:] and [ay] in either) or in (ii) contrastive
distribution distinguishing meaning (e.g. initial /g/, /k/, /t/ in game, came, tame)
∙ Partially equivalent distribution: two elements occur largely, but not exclusively, in the same
environment, in which either (i) the distribution of one element includes that of the other, e.g. the
distribution of the velar plosives /k/ and /g/ includes that of the velar nasal /ŋ/ since the first two occur
word-medially and word-finally, while the last one does not occur word-initially; or (ii)the distribution of
two elements overlaps (also: partially complementary), /h/ and /ŋ/ both of which occur word-medially
(inherent, angle), while only /h/ occurs word-initially (heart) and only /ŋ/ occurs word-finally (song).
∙ Complementary distribution: two elements never occur in the same environment, e.g. [t] and [th] are
said to be in a relation of complementary distribution since the latter does not occur after word-initial /s/.
 Distribution is used to determine and define different basic linguistic elements: equivalent
distribution uncovers phonemes functioning as distinguishers of meaning, while complementary
distribution uncovers allophones and allomorphs, among others.

Transformational Generative Grammar

Introduction
From the late 1950s onwards, structural linguistics has sometimes been used with less popularity because
supporters of generative linguistics initiated by Noam Chomsky have regarded the work of American
structuralists as too limited in conception. They have argued that it is essential to go further than the
position of items to produce a grammar which reflects a native speaker’s knowledge of language.

I. Noam Chomsky and TGG

Transformational Generative Grammar (also TG grammar, TGG) is a theory of grammar which

attempted to provide a model for the description of all languages. It was launched and dominated by
Avram Noam Chomsky (b. 1928), and it may be said to have officially begun with the publication of
his book Syntactic Structures in 1957 (the classical theory) though some of the theory had been
prefigured a few years before in introductory papers by Chomsky as well as in articles by Zellig
Harris. Chomsky’s early work falls into two related points:
1. Criticism of structuralism
2. New formulation of linguistic theory

Chomsky has considerably modified his ideas since 1957. Undeniably, the best known theoretical position
is that of Aspects of the theory of Syntax (written in 1965) or the Aspects Model, a position that Chomsky
himself has called the Standard Theory. It is, in fact, the most recognized version of the theory since it
added important considerations to the study of language. TGG was revolutionary and it is, undoubtedly,
the most forceful and prominent in the century. No linguist who wishes to keep track of contemporary
developments in the field can afford to overlook Chomsky’s theoretical contributions.

II. Criticism of Structuralism

For Chomsky, structural linguistics involved some weaknesses in conceptions and in methodology.

A. Corpus Analysis

For American structuralists, an empirical science studies only observable phenomena. For descriptive
purposes, a language was defined in terms of a corpus. A linguistic corpus has a level of phonological
structure, a level of morphological structure and a level of syntactic structure. They believed that when all
elements of the corpus were grouped and labelled at each level, the grammar of the language was complete.
Structural grammars offer an inventory of forms and constructions which appear in a limited corpus; they do
not provide the rules needed to construct an endless range of possible grammatical sentences. For Chomsky,
a corpus can never represent the whole language, but will only cover an incomplete and a selective sample of
it because language is infinite and creative in nature. TGG supporters suggest that instead of describing
a corpus, a linguist can arrive at an inclusive grammar of language by describing its underlying system of
rules, which is not contained within the corpus, but lies beyond it, in the minds of the speakers. The study of
this system is more important than the study of the actual sentences.

B. Surface Analysis (taxonomic analysis)

Structural grammars only describe the surface structure of sentences. They cannot effectively
handle important grammatical facts (which are part of a native speaker’s knowledge of language),
like the relationship between active and passive sentences, positive, negative and interrogative
sentences, and the deep dissimilarities that exist between superficially identical sentences. The
following sentences are seen to be structurally similar if their analysis considers only their surface
layer, but if another layer is considered, they would be revealed to be dissimilar.
 Examples

- John is eager to please.

- John is easy to please.

Chomsky and others criticized structuralist and post-Bloomfieldian theories as a whole as being
based on a representation of a sentence in terms of surface structure alone. Such approaches are
unsuccessful in distinguishing the surface from the underlying structures of a sentence.

C. The Behaviourist Attitude

Bloomfieldians were influenced by behaviourism. Behaviourism is a psychological theory of learning
which takes into account only visible facts, excluding concepts like “mind”, “ideas” and so on. For
behaviourists, learning a language is similar to learning any other behaviour (to walk, to eat, to write . . .).
It is a mechanical process based on habit formation. Learning is controlled by an external factor (a
stimulus) which produces a response. This response is learnt when it is repeated and positively
reinforced. This process is called conditioning. Language is learnt just by imitation of previously
heard language, and the learner is passive when doing this. Chomsky had been the opponent
of behaviourism. He tried to show the unproductiveness of this view and the inappropriateness of its
terminology to the acquisition and use of human language.

D. Languages’ Diversity

Bloomfield and his followers emphasised the structural diversity of languages following Boas .
They tended to overstate the divergences between languages and have placed excessive accent on the
principle that every language is a unique law. To arrive at a complete understanding of each
language’s structure, a linguist adopts a descriptive approach to the data.

III- Key concepts:

I. Competence and Performance

Chomsky's objection to analyses of corpora is based on the distinction that he draws between two
concepts: competence and performance.
Competence refers to a person's internalised grammar (knowledge) of his language. This means a native
speaker's ability to produce and understand sentences, including sentences they have never heard before. It
also includes a person's knowledge of what are and what are not sentences of a particular language. So, it is
the code which underlies all utterances in a given language. A speaker's linguistic competence enables him
to produce only grammatical and well-pronounced sentences, and to avoid the generation of ungrammatical
and mispronounced sentences, and to recognise whether sentences are synonymous, ambiguous, simple,
complex, etc. For Chomsky, linguistics should be concerned with competence. The latter is purely linguistic.

This is similar to Saussure's concept of langue, but Saussure stressed the social aspect of langue (the
collective shared knowledge), whereas Chomsky stressed the individual nature of competence: He sees it as
a set of processes possessed by the individual and developed in him as part of his maturation. “Langue” is
extracted from utterances after they were produced, but “competence” is the system which creates sentences
never heard before.

Performance, on the other hand, refers to the realisation of this code in actual situations. It is the
person's concrete use of language in producing and understanding sentences. Performance represents only a
small sample of the utterances of language and is influenced by external non-linguistic factors such as lapses
of memory, lapses of attention, malfunctioning of the mechanisms related to speech, stress, fatigue,
noisy surroundings and so on. As a result, a speaker may produce false starts, changes of plan in mid-course,
restructuring of what the speaker wants to say, etc.
 For Chomsky, “performance” is not the object of study in linguistics (but psychology). For Saussure,
however, “parole” does provide the data from which statements about “langue” can be made.

II. Deep Structure and Surface Structure

Chomsky, rejecting the formal analysis of sentences, distinguishes two levels of syntactic struucture in a
sentence: the surface structure and the deep structure. The surface structure (SS) is the syntactic structure
of the sentence which a person speaks or hears: it is the observable form of the sentence. Chomsky argues
that any analysis based on the surface structure encounters difficulties. Therefore, another level of
sentence structure should be taken into account. The deep structure (DS) is much more abstract and is
considered to be in the speaker's mind. It refers to certain important generalisations about the structure of
the sentence which are different from its surface. The deep structure contains all the syntactic information
needed for the understanding of a given sentence. The deep structure is converted into a surface structure
after the application of a specific kind of rules called transformational rules (TRs).

Examples
_ John is eager to please. (I)
_ John is easy to please. (II)

In the deep structures of these two sentences, it is clear that “John” is either the subject of pleasing or its
object.
This distinction between surface and deep syntax became a major dichotomy in TGG, and, for many
people, it is the main difference between the old and new approaches to syntax. For Chomsky, grammar is
not confined to formal description but it should incorporate the internal processes that take place in the
speaker's mind.

III. The Mentalist Attitude

According to Chomsky, language is creative and behaviourism is totally unable to explain creativity. He
argues that the comparison of the sentences a speaker has heard (the input) with the sentences a speaker
produces (the output) shows differences between them. That is to say, the output contains sentences the
speaker has never heard before. On the basis of this evidence, one can deduce that there is “something”
between the input and the output. Chomsky calls it the language acquisition device (LAD)

The LAD is an inborn capacity (a genetic mechanism or apparatus) which is present in the brain right
from the beginning and which enables children (by the age of 3 to 4) to extract the rules of language from
speech when they are exposed to it and to use them productively. Animals do not possess this capacity.
For this reason, their learning of language-like behaviour stops at a definite stage even if they are
exposed to it.

Language acquisition takes place not as a result of imitation (stimulus + response) but as a result of the
functioning of the LAD. In fact, what happens is that the child, when exposed to adult language, tries
mentally to form hypotheses about its rules, then he tests the validity of these rules continuously and adapts
them until he internally masters the abstract system of rules that adults have as part of their competence.
So, language acquisition is part of the maturational process.
 This view about language learning is called mentalism. It is based on the premise that human beings
possess minds.
IV. Language Universals

Chomsky sees that linguistic theory should be concerned with linguistic universals, i.e with the
common characteristics between human languages. According to him, the deep structure is common, and
languages differ only at the level of transformational rules which produce different surfaces.

V. Chomsky's Definitions of Language and Grammar

For Chomsky, "A language is a set (finite or infinite) of sentences, each finite in length and constructed
out of a finite set of elements."(Syntactic Structures, p.13). This definition concerns all languages
(natural and man-made). It implies the following points: ∙ Language is a collection of the infinite number
of possible sentences.
∙ Every sentence is finite in length.
∙ Every sentence is made up of elements that can be collected in a set, and that can be counted (sounds,
morphemes and words)
∙ Language is defined in terms of “sentences”

Grammar is defined as "a device which generates all and only the grammatical sentences of a language."
This definition implies the following points: ∙ The sentence is the basic unit to be described by grammar.
∙ A grammar generates sentences. That is to say, it produces an infinite number of sentences out of precisely
specified rules. (A word taken from maths = to “generate” is to act as a base of a given set
∙ The rules of generative grammar represent knowledge.
∙ A grammar generates “all and only” the grammatical (intuitively accepted as well formed) sentences of
a language. That is to say, grammar should be able to generate all possible grammatical sentences of the
language, and it excludes the ungrammatical (ill-formed) ones.