2.

1 Branches of linguistics 35

2 providing data for general linguistics or in testing conflicting

theories and hypotheses, but because they wish to produce
a
Linguistics reference grammar or dictionary for practical purposes. But this
fact does not affect the interdependence of the two complementary
subfields of general and descriptive linguistics.
Throughout the nineteenth century, linguists were very much
concerned with investigating the details of the historical develop-
ment of particular languages and with formulating general
2.1 Branches of linguistics hypo-
theses about language: anch of the discipline
As we have seen, both language in general and particular languages deals with these matters is now known
can be studied from different points of view. Therefore, the field of
linguistics as a whole can be divided into several subfields according
ies inguisics. T s obvious that in’ historical Tgeiiies oo i,
naturally enough, as histor-

non-historical linguistics, one can be interested in language in

to the point of view that is adopted or the special emphasis that is general or in particular languages. It is convenient to mention at this
given to one set of phenomena, or assumptions, rather than point the more technical terms ‘diachronic’ and ‘synchronic’, These
another. were first used by Saussure (whose distinction of ‘langue’ and
The first distinction to be drawn is between general and descrip- ‘parole’ was referred to in the preceding chapter). A diachronic
tive linguistics. This is in itself straightforward enough. It corres- description of a language traces the historical development of the
ponds to the distinction between studying language in general and language and records the changes that have taken place in it be-
describing particular languages. The question “What is language?” tween successive points in time: ‘diachronic’ is equivalent, there-
which, in the previous chapter, was said to be the central defining fore, to *historical’. A synchronic description of a language is non-
question of the whole discipline is more properly seen as the central historical: it presents an account of the language as it is at some
question in general linguistics. General linguistics and descriptive particular point in time.
linguistics are by no means unrelated. Each depends, explicitly or A third dichotomy is that which holds between theoretical and
implicitly upon the other: general linguistics supplies the concepts applied linguistics. Briefly, theoretical linguistics studies language
and categories in terms of which particular languages are to be and languages with a view to constructing a theory of their structure
analysed; descriptive linguistics, in its turn, provides the data which and functions and without regard to any practical applications that
confirm or refute the propositions and theories put forward in the investigation of language and languages might have, whereas
general linguistics. For example, the general linguist might formu- applied linguistics has as its concerns the application of the concepts
late the hypothesis that all languages have nouns and verbs. The and findings of linguistics to a variety of practical tasks, including
descriptive linguist might refute this with empirical evidence that language-teaching. In principle, the distinction between the theo-
there is at least onc language in the description of which the distinc- retical and the applied is independent of the other two distinctions
tion between nouns and verbs cannot be established. But in order to drawn so far. In practice, there is little difference made between the
refute, or confirm, the hypothesis the descriptive linguist must terms ‘theoretical linguistics’ and ‘general linguistics”: it is taken for
operate with some concepts of ‘noun’ and ‘verb’ which have been granted by most of those who use the term ‘theoretical linguistics”
supplied to him by the general linguist. that the goal of theoretical linguistics is the formulation of a satisfac-
There are of course all sorts of reasons why one might wish to tory theory of the structure of language in general. As far as applied
describe a particular language. Many of those working in descrip- linguistics is concerned, it is clear that it draws on both the general
tive linguistics will be doing so, not because they are interested in and the descriptive branches of the subject.
36 Linguistics 2.2 Is linguistics a science? 37

‘The fourth, and final, dichotomy has to do with a narrower and a to mention several other equally relevant viewpoints). Most lin-
broader view of the scope of the subject. There is no generally guists nowadays would say that it is theoretical synchronic micro-
accepted terminological distinction for this: we will use the terms linguistics that constitutes the core of their discipline and gives it
‘microlinguistics’ and ‘macrolinguistics’, saying that in microlinguis- whatever unity and coherence it has. Almost half of this book will
tics one adopts the narrower view and in macrolinguistics the be devoted to this central core; the rest will be concerned with
broader view. At its narrowest microlinguistics is cong solel. historical linguistics and with selected areas of macrolinguistics.
with the structure of language-systems, without regard to the way in
which languages are acquired, stored in the brain or used in their 2.2 Is linguistics a science?
various functions; without regard to the interdependence of Linguistics is usually defined as the science of language or, alterna-
language and culture; without regard to the physiological and tively, as the scientific study of language (cf. 1.1). The very fact that
psychological mechanisms that are involved in language-behaviour; there should be a section, in this book and in other introductions to
in short, without regard to anything other than the language- linguistics, devoted explicitly to a discussion of the scientific status
system, considered (as Saussure, or rather his editors, put it) in of the discipline should not pass without comment. After all, disci-
itself and for itself. At its broadest, macrolinguisticsis concerned plines whose scientific status is unquestioned — physics, chemistry,
with everything that pertains in any way at all to language and biology, etc. — feel no need to justify their claim to be called
languages. sciences. Why should linguistics be so concerned to defend the
Since many disciplines other than linguistics are concerned with validity of its title? And why is it that, in defending his scientific
language, it is not surprising that several interdisciplinary areas credentials, the linguist so often gives the impression of protesting
should have been identified within macrolinguistics and given a topmuch? The reader has every right to be suspicious.
distinctive name: sociolinguistics, psycholinguistics, ethno- The first point that must be made is that the English word
linguistics, stylistics, ete. ‘science’ is much narrower in its coverage than many of its conven-
One point that must be emphasized is that the distinction tionally accepted translation-equivalents in other languages: such
tween microlinguistics and macrolinguisticsis independent of the as ‘Wissenschaft’ in German, ‘nauka’ in Russian and even ‘science’
istinction between theoreticaland applied linguistics. There is, in in French. Linguistics suffers more than most disciplines do from
pinciple, ore spect o evry ranch of macroingutis. the very specific implications of the English words ‘science’ and
It so happens that in such areas of applied linguistics as language- ‘scientific’, which refer, first and foremost, to the natural sciences and
teaching it is_essential ¢ broader, rather than the the methods of investigation characteristic of them. This is still true,
narrower, view of the structure an ons of languages. This even though such phrases as ‘the social sciences’, ‘the behavioural
is why some incorporated
authors have what is here called sciences’ and even ‘the human sciences’ are increasingly common.
macrolinguistics within applied linguistics. Should we then interpret the word ‘science’ in the heading to this
We shall look at some areas of macrolinguistics in later chapters. section to mean simply “properly constituted academic discipline™?
Itmight be thought that, in view of the acknowledged importance of There is rather more to the question than this interpretation
language to so many disciplines, linguistics ought to take the would suggest. Most linguists who subscribe to the definition of
broadest possible view of its subject-matter. There is a sense in their discipline as the scientific study of language do so because they
which this is true. The problem is that there is not yet, and may have in mind some distinction between a scientific and a non-
never be, a satisfactory theoretical framework within which we can scientific way of doing things. They may disagree about some of the
view language simultaneously from a psychological, a sociological, implications of the term ‘scientific’, as do philosophers and
a cultural, an aesthetic and a neurophysiological point of view (not historians of science. But they are in general agreement about the
38 Linguistics 2.2 Is linguistics a science? 19

principal differences between the scientific and non-scientific study many of the popular misconceptions about language that are cur-
of language. Let us begin, then, with these points of agreement. rent in our society can be explained, historically, in terms of the
The first, and most important of these is that linguistics is philosophical and cultural assumptions which determined the de-
pirical, rather than speculative or intuitive: it operates with publicl, velopment of traditional grammar. Some of these misconceptions
verifiable data obtained by means of observation or experiment. To. will be listed and discussed in the following section. It must be
be empirical, in this sense, is for most people the very hallmark of emphasized, however, that linguistics,
like any other discipline,
science. Closely related to the property of being empirically based is builds on the past, not ony by challenging and refuting traditional -
that of objectivity. Language is something that we tend to take for doctrines but also by developing and reformulating them. Many
granted; something with which we are familiar from childhood in a recent works on linguistics, in describing the great advances made
practical, unreflecting manner. This practical familiarity with in the scientific investigation of language in the past hundred years
language tends to stand in the way of its objective examination. or 5o, have neglected to emphasize the continuity of Western lin-
There are all sorts of social, cultural and nationalistic prejudices guistic theory from the earliest times to the present day. They have
associated with the layman’s view of language and of particular often been somewhat anachronistic, too, in their failure to treat
languages. For example, one accent or dialect of a particular lan- traditional grammar in terms of the aims it set itself. It must not be
guage might be thought to be inherently purer than another; or forgotten that the terms ‘science’ and ‘scientific’ (or their precur-
again one language might be held to be more primitive than sors) have been construed differently at different periods.
another. Objectivity demands, at the very least, that beliefs like It should also be pointed out that what is generally referred to by
these should be challenged and terms like ‘pure’ and ‘primitive’ means of the term ‘traditional grammar’ — i.e. Western linguistic
either clearly defined or rejected. theory going back through the Renaissance and the Middle Ages to
Many of the ideas about language which the linguist calls into Roman and, before that, Greek scholarship - is much richer and
question, if he does not abandon them entirely, might appear to be more variegated than is commonly realized. Furthermore, it was
amatter of downright common sense. But, as Bloomfield (1935: 3) very often a misunderstood and distorted version of traditional
remarked of the common-sense way of dealing with linguistic ques- grammar that was taught at school to generations of reluctant and
tions: “like much else that masquerades as common sense, it is in uninterested pupils. In the last few years linguists have begun to
fact highly sophisticated and derives, at no great distance, from the take a more balanced view of the contribution that traditional
speculations of ancient and medieval philosophers”. Not all grammar — we shall continue to use the term ~ has made to the
linguists have as low an opinion of these philosophical speculations development of their discipline. There is still much research to be
about language as Bloomfield had. But his general point is valid. done on such of the original sources as have survived from the
The terms that the layman uses to talk about language and the carlier periods. But several histories of linguistics are now available
attitudes that he has with respect to language have a history to which give a more satisfactory account of the foundations and
them. They would often seem less obviously applicable or self- development of traditional grammar than was readily available to
evident if he knew something of their historical origin. Bloomfield’s generation and that of his immediate successors,
We shall not go into the history of linguistics in this book. Some Let us now return to the present state of linguistics. It is un-
general comments, however, are in order. It is customary for doubtedly more empirical and objective, in its professed attitudes
introductions to linguistics to draw a sharp distinction between and assumptions at any rate, than traditional grammar. We shall
traditional grammar and modern linguistics, contrasting the scien- look atsome of these attitudes and assumptions in more detail in the
tific status of the latter with the non-scientific status of the former. following section. But is it as empirical and objective in practice as it
There is good reason to draw this distinction and to point out that claims to be? Here there is certainly room for doubt. There is also
40 Linguistics 2.2 Is linguistics a science? 41

room for dispute, at a more sophisticated level of discussion, as to cism and positivism both by philosophers and by psychologists and
the nature of scientific objectivity and the applicability to the study other social scientists (cf. 7.4).
of language of what is commonly referred to as the scientific Empiricism implies much more than the adoption of empirical
method. methods of verification or confirmation: there is therefore a crucial
Actually, it is no longer so widely accepted by scientists and distinction to be drawn between ‘empiricist’ and ‘empirical’. The
philosophers of science that there is a single method of enquiry term ‘empiricism’ refers to the view that all knowledge comes from
applicable in all branches of science. The very term itself, ‘the c@eriénce — the Greek word ‘empeiria’ means, roughly, “ex-
scientific method’, has a distinctly old-fashioned, even nineteenth- and, more particularly, from perception and sense-
century, ring to it. It is sometimes suggested that scientific enquiry data, It is opposed, in a long-standing philosophical controversy to
must necessarily proceed by means of inductive generalization on ‘rationalism’ — from the Latin ‘ratio’ meaning, in this context,
the basis of theoretically uncontrolled observation. Indeed, this is “mind”, “intellect” or “reason”. The rationalists emphasize the
what many people hold to be implied by the term ‘the scientific role that the mind plays in the acquisition of knowledge. In particu-
method’, But few scholars have ever worked in this way even in the lar, they hold that there are certain a priori concepts or propositions
natural sciences. Whatever scientific objectivity means it certainly (‘a priori’ means, in its traditional interpretation, “known‘indepen-
does not imply that the scientist should refrain from theorizing and dently of experience”) in terms of which the mind interprets the
from the formulation of general hypotheses until he has amassed a data of experience. We will come back to some more specific
sufficient amount of data. Scientific data, it has often been pointed aspects of the controversy between empiricism and rationalism in
out, are not given in experience, but taken from it. Observation our discussion of generativism (cf. 7.4).
implies selective attention. There is no such thing as theory-neutral No distinction need be drawn, for our purposes, between empiri-
and hypothesis-free observation and data collection. To use a cur- cism and positivism. The former has a longer history and is much
rently fashionable phrase, originating with Popper, observation is, broader in scope as a philosophical attitude. But the two are natural
of necessity and from the outset, theory-laden. allies and are closely associated in all that concerns us here. Positiv-
The phrase is suggestive, but controversial. It was produced in ism rests upon the distinction between the so called positive data of
reaction to the strongly empiricist view of science put forward by experience and transcendental speculation of various kinds. It
the logical positivists in the period preceding the Second World tends to be secular and anti-metaphysical in outlook and rejects any
War. Students of linguistics should know a little about empiricism appeal to non-physical entities. It was the aim of the logical positiv-
and positivism. Without such knowledge ~ though it need not be ists of the Vienna Circle to produce a single system of unified
very detailed or profound ~ they cannot be expected to understand science, in which the whole body of positive knowledge would be
some of the theoretical and methodological issues that divide one represented, ultimately, as a set of precisely formulated proposi-
school of linguistics from another at the present time. What follows tions.
is the necessary minimum of background information, presented, Two more specific principles were central to this proposal. The
as far as this is possible, impartially and without commitment to first was the now famous verification principle: the principle that no
cither side in areas of known controversy. The controversies, it statement was meaningful unless it could be verifiedby observation
should be added, are relevant to the whole of science, not just to or standard scientific methods appiied to the data provided by
linguistics. But they have a special relevance for the linguist, in that Siaceaatins Tho secoud:whe thipriicple BETRMMRMNN s
recent developments in linguistics and the philosophy of language, principle that, of the sciences, some were more basic than others —
associated with the work and ideas of Chomsky, have had a very physics and chemistry being more basic than biology, biology being
considerable impact upon the more general discussion of empiri- mare basic than psychology and sociology, and so on - and that in
42 Linguistics 2.2 Is linguistics a science? 43

the grand synthesis of unified science the concepts and propositions structs and its own methods of obtaining and interpreting the data.
of the less basic sciences were to be reduced to (i.e. reinterpreted in What was referred to in the previous chapter as a fiction ~ the
terms of ) the concepts and propositions of the more basic sciences. language-system — can be described, in scientifically more respect-
Reductionism, unlike the verification principle, was characteristic able terms, as a theoretical construct. Questions can be asked
of a much wider group of scholars than the members of the Vienna about the reality of such constructs, just as they can be asked about
Circle forty years ago. the reality of the theoretical constructs of physics or biochemistry.
The verification principle has now been abandoned (though it has It is more profitable, however, to enquire of each theoretical con-
played its part in the formation of the truth-conditional theory of struct that is postulated what explanatory purpose it is fulfilling with
meaning: cf. 5.6) and the principle of reductionism s far less gener- respect to the data.
ally accepted by scientists and philosophers of science than it was All that has just been said about empiricism, positivism and the
when Bloomfield wrote his classic textbook of linguistics in 1933. T current status of the so called scientific method is intended to be
mention Bloomfield at this point because, not surprisingly, he was more or less factual and uncontroversial. We now turn to points of
strongly committed to empiricism and positivism. This is made very controversy.
clear in the second chapter of his textbook. He was, in fact, closely The first has to do with the implication of Popper’s notion of
associated with the Unity of Science movement and subscribed fully theory-laden observation. This is controversial in the use of the
to the principle of reductionism. It was Bloomfield more than term ‘theory’. What Popper had in mind, and was attacking, was the
anyone else who set for linguistics, especially in America, the ideal sharp distinction drawn by the logical positivists between observa-
of being truly scientific. There is therefore a historically explicable tion, itself held to be theoretically neutral, and theory-construction,
legacy of empiricism and positivism in linguistics. held to be a matter of inductive generalization. He was undoubtedly
Reductionism, and more generally positivism, is no longer as correct in challenging the sharpness of the distinction and, more
attractive to most scientists as it once was. It is now widely accepted especially, the view that observation and data-collection can, and
that there is no such thing as a single scientific method applicable in must, proceed in advance of the formulation of hypotheses. It is
all fields; that diverse approaches are not only to be tolerated, as a commonly the case that the selection of data is determined by some
matter of short-term necessity, in different disciplines, but may be hypothesis that the scientist wishes to test; and it does not matter
justifiable, in the long term too, by virtue of irreducible differences how this hypothesis has been arrived at. The fact that the positivists’
of subject-matter. Ever since the seventeenth century ~ from the notion of unselective observation and data-collection is invalid does
time of Descartes and Hobbes ~ there have been doubts expressed not mean that there is no distinction at all to be drawn between
by some philosophers of science about the positivists’ programme pretheoretical and theoretical concepts. It is an abuse of the term
of accounting for mental processes in terms of the methods and ‘theory’ to subsume under it all the preconceptions and expecta-
concepts characteristic of the physical sciences. Much of twentieth- tions with which one approaches what is observable and makes
century psychology and sociology, like much of twentieth-century one’s selection. We will draw upon the distinction between
linguistics, has been positivistic in spirit. But in all three disciplines, pretheoretical and theoretical concepts at several places in later
and most obviously in linguistics, positivism has recently come chapters; and we will assume that observation, though necessarily
under attack as being either unworkable or sterile. selective, can be made subject to satisfactory methodological con-
In short, the question whether a discipline is or is not scientific trols, in linguistics as in other empirically based sciences.
can no longer be satisfactorily answered, if it ever could be, by A second point of controversy — and one that is of particular
making reference to the so called scientific method. Every well importance in linguistics at the present time — has to do with the role
established science employs its own characteristic theoretical con- of intuition and the methodological problems that arise in this
“ Linguistics 2.2 Is linguistics a science? 45

connection. The term ‘intuition’ carries with it certain rather unfor- at least as unreliable, though often for other reasons, as the intui-
tunate everyday associations. All that is meant when one refers to tions of the layman. The linguist may be less concerned than the
the native speaker’s intuitions about his language is his spontaneous: layman about conventional standards of correct usage (e.g. admit-
and untutored judgements about the acceptability or unacceptabil= ting quite freely that he normally says Ir's me, rather than It is I).
ity of utterances, the equivalence or non-equivalence of utterances, But his judgements are more likely to be distorted by his awareness
and 50 on. There was a time when some linguists thought that it was of the implications that they have for this or that theoretical issue.
in principle possible to escape from the necessity of asking native The linguist’s introspections about his own language-behaviour
speakers to make such intuitive judgements about their language by and that of others may very well be theory-laden, even if direct
simply collecting a large enough corpus of naturally occurring data. observation of spontaneous conversation is not.
and submitting it to an exhaustive and systematic analysis. Very few There are in fact quite serious methodological problems
linguists take this view nowadays. It has become clear that many attaching to the collection of reliable data for a whole range of
naturally occurring utterances are, for linguistically irrelevant issues in theoretical linguistics. But they are no more serious than
reasons, unacceptable and also that no corpus of material, however the methodological problems that confront those working in
large, will contain examples of every kind of acceptable utterance. psychology, sociology or the social sciences in general, And in
But the linguist’s appeal to intuitive evidence remains controver- certain respects the linguist is better off than most social scientists,
sial. There are two aspects to the controversy. since it is fairly clear how much of what is observable is language-
The first relates to the question whether the intuitions that the behaviour and how much is not. Furthermore, there are very con-
linguist makes reference to are part of the native speaker’s linguistic siderable areas, in the description of any language, for which the
competence as such. If so,on Chomsky’s definition of ‘competence’ reliability of the native speaker’s intuitions, and even of the lin-
and his formulation of the goals of linguistics, the intuitions them- guist’s introspections, is not a serious problem. One must not make
selves become part of what the description of any particular lan- too much, therefore, of the methodological problems that arise in
guage must directly account for. Most linguists would probably not the course of linguistic research.
want to say that the description of a language must treat the native Reference was made in the previous paragraph to psychology,
speaker's intuitions as data. We will come back to this question in sociology and the other social sciences. Many linguists, perhaps the
our discussion of generativism (cf. 7.4). majority, would classify their discipline among the social sciences.
The second part of the controversy has to do with the reliabilityof But linguistics is not readily classifiable within any division of
the native speaker’s judgements, considered as reports or predic- academic research which takes as fundamental either the distinc-
tions of his own and others’ language-behaviour. The general con- tion between science and arts or the tripartite distinction of the
sensus of opinion among linguists would seem to be that such natural sciences, the social sciences and the humanities. The in-
judgements are, in particular respects at least, highly unreliable. creasing use of such phrases as ‘the life sciences’, ‘the behavioural
Not only do native speakers frequently disagree among themselves sciences’, ‘the human sciences’ or ‘the earth sciences’ indicates that
about what is acceptable, when there is po other reason to believe many disciplines feel the need for strategic or tactical regroupings
that they speak different dialects, but their judgements have been which have little regard to the conventional distinctions. Whether
shown to vary over time. Moreover, it often happens that a native linguistics, as a university subject, is housed in one faculty rather
speaker will reject as unacceptable some utterance put to him by the than another is largely a matter of administrative convenience.
descriptive linguist and then be heard, or hear himself, producing Linguistics, as has been emphasized before, has natural links with
that very utterance in some natural context of use. As far as the awide range of academic disciplines. To say that linguistics is a
linguist’s introspections about his language are concerned, they are science is not to deny that, by virtue of its subject-matter, it is
 46 Linguistics 2.4 Linguistics is descriptive, not prescriptive 47
closely related to such eminently humane disciplines as philo: unfortunately no generally accepted set of notational conventions
and literary criticism. by means of which these and other distinctions can be drawn. In the
In the following sections, a number of principles will be me present work, we shall make distinctive use of single quotation-
tioned and discussed which are generally taken for granted marks, double quotation-marks and italics. For example, we shall
adays by linguists. For the most part, they can be seen as derivii
distinguish between “table” and table, the former being the mean-
from the scientific ideal of objectivity. Since modern linguistics, i ing and the latter the form (or one of the forms) of the word ‘table’.
asserting its objectivity, has so often proclaimed its distinctiven
By making use of these conventions, we can keep distinct, as we
in this respect from traditional grammar, they are frequen shall see later, at least two of the senses of the word ‘word’: in the
presented in contrast with the principles that determined thy first sense it refers to something that we should expect to be listed in
characteristic attitudes and assumptions of the traditional gr: the dictionary of the language; in the second sense it refers to what
marian, would be printed between spaces as a sequence of letters in a
written text.
2.3 Terminology and notation
Other notational conventions will be introduced later, enabling
Every discipline has its own technical vocabulary, Linguistics is no- us to distinguish spoken forms from written forms; and spoken
exception. Most of the technical terms used by linguists arise in forms of one kind (phonetic) from spoken forms of another kind
the course of their work and are easily understood by those who (phonological); and so on. The general point being made here is
approach the subject sympathetically and without prejudice.
that various notational conventions are, if not absolutely essential,
The objection is sometimes made that the terminology, or jar- atleast very useful for the purpose of referring to language-data and
gon, of linguistics is unnecessarily complex. Why is the linguist so
making it clear what is being talked about. They have the further
prone to the creation of new terms? Why is he not content to talk
advantage that they force the linguist to think carefully about
about sounds, words and parts of speech, instead of inventing such
certain distinctions that might otherwise pass unnoticed. Very
new technical terms as ‘phoneme’, ‘morpheme’ and “form class’?
often, it proves difficult to be absolutely consistent in the applica-
The answer is that most of the everyday terms that are used with
tion of some particular notational convention; and this difficulty
reference to language — many of which, incidentally, originated as
then leads to a re-assessment of the theoretical distinction for which
technical terms of traditional grammar - are imprecise or ambigu-
the notational convention was first established. This is one of the
ous. This is not to say that the linguist, like all specialists, may not
ways in which progress in any discipline is made.
be guilty at times of misplaced terminological pedantry. In prin-
ciple, however, the specialized vocabulary of linguistics, if it is kept
under control and properly used, serves to clarify, rather than to 2.4 Linguistics is descriptive, not prescriptive
mystify. It eliminates a good deal of ambiguity and possible mis- The term ‘descriptive’ is here being employed in a different sense
understanding.
from the sense in which it opposes either ‘general’, on the one hand,
As with terminology, so with notation. We have to use language or ‘historical’, on the other. The contrast that is relevant here is the
in order to talk both about language in general and about particular
one that holds between describing how things are and prescribing
languages. In doing so, we need to be able to identify exactly what how things ought to be. An alternative to ‘prescriptive’, in the sense
bits, parts or features of a language we are referring to. The use of
in which it contrasts with ‘descriptive’, is ‘normative’. To say that
special notational conventions makes this a lot easier. For example,
linguistics is a descriptive (i.e. non-normative) science is to say that
we might need to distinguish between the meaning of a word and its
the linguist tries to discover and record the rules to which the
form and between each of these and the word itself. There is members of a language-community actually conform and does not
48 Linguistics 2.4 Linguistics is descriptive, not prescriptive 49

seek to impose upon them other (i.e. extraneous) rules, or norms, negative” (I didn't do nothing); “Don’t end a sentence with a
of correctness. preposition” ( That's the man I was speaking to); “The verb ‘to be’
It is perhaps confusing to use the term ‘rule’, as I have just done, takes the same case after it as before” (so that, by the application of
in these two very different senses. Rightly or wrongly, linguists tall this rule, If's me should be corrected to It is I); “Ain’t is wrong”;
in these terms. It might be helpful, therefore, to illustrate th “You should not split the infinitive” (as in I want you to clearly
difference between the two kinds of rules - let us call them imman: understand where clearly is inserted between fo and understand).
ent and transcendent, respectively ~ from something other than the Consideration of the above examples quickly shows that they are
use of language. Let us take sexual behaviour in a given society. If quite heterogeneous in character. There are some dialects of
we adopt the purely descriptive (i.e. non-normative) point of view English in which the so called double-negative is never employed
in the investigation of sexual behaviour, we will try to find out how (i.e. in which I didn’t do nothing is never used as the equivalent of
people actually behave: whether they practise premarital sex, and, Standard English I didn’t do anything); there are others in which it
if so, of what kind and from what age; whether husbands and wives is, from a purely descriptive point of view, the correct construction.
are equally faithful or unfaithful to their partners; and so on. In so When reasons are given for the condemnation of the double nega-
far as the behaviour of particular groups within the community is tive as incorrect, in terms of some prescriptive principle with ref-
governed, in practice, by determinable principles — whether the erence to which actual usage may be judged and found wanting,
members of these groups profess, or are even aware of, these logic is made the court of appeal. Logic tells us, it is said, that two
principles or not - we can say that their behaviour is rule-governed: negatives make a positive. This calls for several comments. First, it
the rules are immanent in their actual behaviour. But such rules (if betrays a misunderstanding of what logic is and how it operates: but
they are rightly called rules) are very different, in status if not in we need not go into the nature of logical axioms and the complex
content, from the rules of conduct that the law, the established question of how the so-called natural logic of ordinary language-
religion or simply explicit conventional morality might prescribe. behaviour relates to the systems of logic that are constructed and
People may or may not conform, in practice, to what I am calling investigated by logicians. The point is simply that there is nothing
the transcendent (i.e. extrancous or non-immanent) rules of sexual inherently illogical about the so-called double-negative construc-
behaviour. Furthermore, there may be differences between how. tion. In the dialects in which it is regularly employed, it operates
they behave and how they say, or even think, that they behave. All quite systematically according to the grammatical rules and prin-
hese differences have their correlates in respect of language- ciples of interpretation that are immanent in the behaviour of the
sehaviour. The most important distinction, however, is the one that dialect-communities in question. A second point to be borne in
10lds between transcendent (i.e. prescriptive) and immanent (i.e. mind is that the so-called double-negative construction cannot be
escriptive) mlem‘;?rs are commands ( Do/ properly described, as it operates in certain dialects of English,
Yon't say X!); descriptive dos and don'ts are statements (People without taking into account features of stress and intonation. The
aldon’t say X). rules of Standard English (i.e. the rules immanent in the language-
The reason why present-day linguists are so insistent about the behaviour of the speakers of a particular dialect of English) permit /
stinction between descriptive and prescriptive rules is simply that didn't do nothing (with the meaning, roughly, “It’s not true that I
wditional grammar was very strongly normative in character. The did nothing”) provided that didn't is stressed or, alternatively and
immarian saw it as his task to formulate the standards of correct- with additional implications or presuppositions, do or nothing is
ss and to impose these, if necessary, upon the speakers of the pronounced with particularly heavy stress. In dialects in which /
iguage. Many of the normative precepts of traditional grammar didn’t do nothing (with normal unemphatic stress) can mean “I
I be familiar to the reader: “You should never use a double- didn’t do anything" it also has the meanings that it has in Standard
 Linguistics 2.4 Linguistics is descriptive, not prescriptive 51
50

English, but stress and intonation prevent confusion. Finally, it may minded grammarian would regard as good English will say and
be noted that there are many languages in which the so-called write between you and I, He told you and 1, etc. Such constructions
double-negative construction occurs in the standard literary dialect: violate another traditional prescriptive rule for English: **Verbs and
e.g. French, Italian, Spanish, Russian - to mention but a few of the prepositions govern their object in the accusative.” They result,
more familiar modern European languages. Even the most presti= presumably, from what is often called hypercorrection: the exten-
gious dialect of Ancient Greek — Classical Greek as used in the sion of some rule or principle, on the basis of a misunderstanding of
writings of Plato, Sophocles, Thucydides or, the very founding its domain of application, to a range of phenomena to which,
father of logic, Aristotle himself - had a double-negative construc- originally, it did not apply. The nature of the prescriptive rule is
tion. And traditional grammar, after all, had its origins in the misunderstood - the more so as many speakers who might naturally
description of the literary dialects of Ancient Greece! say You and me will go would never say either Me will go or He told
Other normative precepts of traditional grammar — such as the 1. It is interpreted instead as an instruction (under pain of being
condemnation of the split infinitive (. . . to clearly understand) or of considered a speaker of bad English) to substitute you and I for you
It’s me - derive from the application to English of principles and and me (or me and you) in all positions of occurrence. This resultsin
categories established in the first place for the description of Greek the production, not only of what the traditional grammarjan would
and Latin. It so happens that the forms to which the term ‘infinitive’ accept as correct, You and I will go together, etc., but also of what
is applied are one-word forms in Greek and Latin, as in French, he would condemn, between you and I, He told you and I, etc. It is
German, Russian, etc. Traditionally, the two-word forms to under- not being implied, of course, that every speaker of English who says
stand, to go, etc., are also called infinitives, though their function is between you and I, He told you and 1, etc., has himself performed
only partly comparable with the function of, say, the Latin infini- the operation of applying and misapplying the traditional rule.
tives. As we shall see later, whether a form can be split (in the sense These constructions are now so common in the speech of middle-
in which we talk about the split infinitive) is one of the principal class and upper-class speakers of Standard English in England that
criteria which the linguist applies to decide whether the form in they must have been learned naturally in the normal process of
question is a one-word form or a two-word form. Given that, by language-acquisition by perhaps the majority of those who use
other criteria and by the writing conventions of the written lan- them. There is little doubt, however, that their origin lies in the
guage, the so-called infinitives of English are two-word forms, there. process of hypercorrection.
can be no objection, in principle, tosplitting them. As to the castigation Neither logic nor the grammar of Latin can properly serve as the
of It's me, etc., the fact of the matter is that what are referred to in court of appeal when it comes to deciding whether something is or is
traditional grammar as differences of case (1 vs. me, she vs. her, he not correct in English. Nor can the unquestioned authority of
vs. him, etc.) are not found in all languages; nor is anything that can tradition for tradition’s sake (“That's what I was taught, and my
be identified in terms of its function and grammatical characteristics parents, and my parents’ parents”) or the usage of what are thought
as a verb meaning ‘“‘to be”. Furthermore, in languages that have to be the best writers in the language. Itis a widely held view in our
both case and a verb identifiable as the equivalent of the Latin ‘esse’ society, or has been until recently, that linguistic change necessarily
or English ‘to be’, the diversity of the constructions that are found is involves a debasement or corruption of the language. This view is
such that the traditional rule, “The verb ‘to be’ takes the same case indefensible. All languages are subject to change. This is a matter of
after it as before™, stands out immediately for what it is - a Latin- empirical fact; and it is the task of historical linguistics to investigate
based normative rule which cannot be supported on more general the details of language-change, when they are accessible to inves-
grounds, tigation, and, by constructing an explanatory theory of language-
Interestingly enough, many speakers of what the traditionally change, to contribute to our understanding of the nature of
 Linguistics 2.4 Linguistics is descriptive, not prescriptive 53

language. The factors that determine language-change are complex. prescription, the linguist is not saying that there is no place at all for
and, so far, only partly understood. But enough is known now —ai the establishment and prescription of norms of usage. There are
has been known since the middle of the nineteenth century- forit obvious administrative and educational advantages, in the modern
be clear to any unprejudiced observer of change in language thal world, in standardizing the principal dialect that is employed within
what is condemned at any one time as a corruption or debasemen a particular country or region. This process of standardization has
of traditional standards of usage can always be matched with taken place over a long period in many Western countries with or
earlier change of the same kind which brought into being the usa; without government intervention. It is now happening on an
that traditionalists themselves treat as unalterably correct. accelerated scale, and as a matter of official policy, in several of the
As to the principle of conforming to the standards set by the developing nations in Africa and Asia. The problem of selecting,
acknowledged best writers, this principle too is indefensible — inde= standardizing and promoting one particular language or dialect at
fensible, that is to say, in relation to the use that is commonly made the expense of others is fraught with political and social difficulties.
of it. There is no reason to believe that the writer, genius though he. It is part of what has come to be called language-planning — an
be, is vouchsafed, by special dispensation, a sure and certain know= important area in the field of applied sociolinguistics.
ledge of the transcendent rules of correctness that is denied to the. Nor should it be thought that in denying that all change in
rest of us. It so happens that traditional grammar had a very strong language is for the worse, the linguist is implying that it must be for
literary bias. The reason is that at several important periods in the. the better. He is merely calling into question the unthinking appeal
development of European culture - from the period of Alexandrian. to empirically discredited criteria. He concedes that it might be
scholarship in the second century B.c. to that of Renaissance possible, in principle, to evaluate dialects and languages in terms of
humanism — grammatical description, first of Greek, then of Latin, their relative flexibility, range of expression, precision and aesthetic
was subordinated to the practical task of making the literature of an potential; and he certainly accepts that the use that is made of their
earlier age accessible to those who did not, and in the nature of dialect or language by individual speakers and writers may be more
things could not, speak naturally the dialect of Greek or Latin upon or less effective. However, he cannot but report, on the basis of the
which the language of the classical texts was based. The literary bias more scientific work that has been done on language and languages
of traditional grammar is not only historically explicable, but per- in recent years, that most of the judgements that are made about
fectly justifiable, as far as the description of Greek and Latin was such matters are extremely subjective. As an individual member of
concerned. It is quite unjustifiable when it comes to the gram- a language-community, the linguist will have his own prejudices,
matical description of modern spoken languages. either personal to him or deriving from his social, cultural and
There are no absolute standards of correctness in language. We geographical background; and he may be either conservative or
can say that a foreigner has made a mistake, if he says something progressive by temperament. His attitudes towards his own
that violates the rules immanent in the usage of native speakers. We language will be no less subjective in this respect than those of the
can also say, if we wish, that a speaker of a non-standard social or layman. He may find a particular accent or dialect pleasing or
regional dialect of English has spoken ungrammatically if his utter- displeasing. He may even correct his children’s speech, if he finds
ance violates the rules immanent in Standard English. But, in them using a pronunciation, a word or a grammatical construction
saying this, we are of course assuming that he was intending, or that is frowned upon by purists. But, in doing 5o, if he is honest with
perhaps ought to have been intending, to use Standard English. himself, he will know that what he is correcting is not inherently
And that is itself an assumption which requires justification. incorrect, but only incorrect relative to some standard which, for
It must now be emphasized — and this point is frequently mis- reasons of social prestige or educational advantage, he wishes his
understood - that in drawing a distinction between description and children to adopt.
54 Linguistics 2.5 Priority of synchronic description 55
All this may seem highly theoretical and abstract. But it has
As far as his attitude to the literary language is concerned, th some
very practical implications. The first has to do with what
linguist is simply pointing out that language is used for many pur- I will call
the etymological fallacy. Etymology is the study of the
poses and that its use in relation to these purposes should not be origin and
development of words. It had its source, as far as
Jjudged by criteria which are applicable, solely or primarily, l? the the Western
grammatical tradition is concerned, in the speculations of
literary language. This does not mean that he is in any way hostile to certain
Greek philosophers of the fifth century before Christ. The
literature or to the study of literature in our schools and univer: term
‘etymology’ is itself revealing. It contains a Latinized transc
sities. On the contrary, many linguists have a particular interest ription
of a form of the Greek word ‘etumos’ meaning “true” or
the investigation of the literary purposes to which language is put “real”.
According to one school of fifth-century Greek philosophers
and its success in achieving those purposes. This is one part - and a all
words were naturally, rather than conventionally, associa
very important part — of the branch of macrolinguistics known as ted with
stylistics. what they signified. This might not be evident to the layman
, they
would say; but it could be demonstrated by the philosopher
able to
2.5 Priority of synchronic description discern the reality that lay behind the appearance of
things. To
penetrate the often misleading appearances and, by careful
The principle of the priority of synchronic description, which analysis
of changes that had taken place in the development of
characteristic of most twentieth-century linguistic theory, impl the form or
meaning of a word, to discover the origin of a word and
that historical considerations are irrelevant to the investigation of thereby its
real meaning was to reveal one of the truths of nature.
particular temporal states of a language. Saussure’s terms ‘synchro What [ am
referring to as the etymological fallacy is the assumption
nic’ and ‘diachronic’ were introduced earlier in the chapter (cf. 2.1), that the
original form or meaning of a word is, necessarily and
We can use one of Saussure’s own analogies to explain what: by virtue of
that very fact, its correct form or meaning. This assump
meant by the priority of the synchronic over the diachronic. o tion is
Let us compare the historical development of a particular, widely held. How often do we meet the argument that
because such
and such a word comes from Greek, Latin, Arabic,
language with a game of chess that is being played in front of us. or whatever
language it might be in the particular instance, the correct
state of the board is constantly changing, as each player makes his. meaning
move. At any one time, however, the state of the game can be fully of the word must be what it was in the language of
origin! The
described in terms of the positions occupied by the pieces. argument is fallacious, because the tacit assumption of
an originally
true or appropriate correspondence between form and
(Actually, this is not quite true. For example, the state of the game meaning,
upon which the argument rests, cannot be substan
is affected, as far as the possibilities for castling are concerned, b tiated.
moving the king from, and back to, its original position. We may Etymology was put on a sounder footing in the nineteenth
century
than it had been in previous periods. It is no longer
neglect such minor points of detail in which Saussure’s analogy fair to say, as
breaks down.) It does not matter by what route the players have: Voltaire is reported to have done, that etymology is
a science in
arrived at a given state of the game. Regardless of the number, which the vowels count for nothing and the conson
ants for very
little! As it is nowadays practised, it is a respectable
nature or order of the previous moves, the present state of the game branch of
historical, or diachronic, linguistics. As we shall see in
is synchronically describable without reference to them. So it is, Chapter 6, it
has its own methodological principles, whose reliabil
according to Saussure, with the historical development of ity depends
languages. All languages are constantly changing. But each of upon the quality and quantity of the evidence upon which
they are
brought to bear. In favourable instances the reliability
successive states of a language can, and should, be described on if of etymo-
own terms without reference to what it has developed from or wi logical reconstruction is very high indeed.
it is likely to develop into. One point that became clear to nineteenth-century etymol
ogists
56 Linguistics 2.5 Priority of synchronic description 57

and is now taken for granted by all linguists is that most words in the know) that all languages started from the same state of the board, as
vocabulary of any language cannot be traced back to their origin, it were, and then developed differently, but it is also impossible to
Words that are deliberately created, by borrowing forms from other date the beginning of a language except by arbitrary convention and
languages or using any other principle, are untypical of the vocabu: very approximately. We cannot say, for example, at what point in
lary as a whole, and certainly of what might be thought of as th time spoken Latin became Old French, or Italian, or Spanish. Nor
more basic, non-technical vocabulary of a language. What the can we say at what time a particular language ceased to exist —
present-day etymologist does is to relate words of one synchroni: except in respect of languages which have become extinct, more or
cally describable state of a language to words, attested or recol less suddenly, when their last native speakers died. Languages,
structed, of some earlier state of the same language or of some other considered from the diachronic point of view, do not have deter-
language. But the words of the earlier state of the same language 0 minate beginnings or ends. In the last resort, it is a matter of
of the earlier language have themselves developed from earl; convention and convenience whether we say that Old English
words. Whether the form or meaning of these earlier words i and Modern English are two states of the same language or two
recoverable by the techniques of etymology will depend upon th different languages.
evidence that has survived. For example, we can relate the presents There is yet another way in which Saussure’s analogy breaks
day English word ‘ten’ to the Old English word whose forms were down. The game of chess is governed by explicitly formulated rules
either ren (with a long vowel) or tien. And we can relate the Ol and, within the limits imposed by these rules, the players determine
English word, through successive hypothetical states, to a recon: the course of any particular game that is being played in opposition
structed Proto-Indo-European word with the form *dekm, a to one another and with reference to a recognized goal. As far as we
meaning “ten”’. But we cannot go back with any confidence beyond know, there is no directionality in the diachronic development of
that. And yet the Proto-Indo-European word *dekm— the prefixed languages. There may well be certain general principles which
asterisk indicates that it is reconstructed, not attested (cf. 6.3) — i§ determine the transition from one state of a language to another.
obviously not the origin, in any absolute sense, of all the words thaf But, if there are such principles, they are not comparable with the
have developed from it in the languages that we can identify as rules of a man-made game like chess. We shall look at the so-called
belonging to the Indo-European family. It must itself have d laws of language-change in Chapter 6.
veloped from a word (which may or may not have meant “‘ten The principle of the priority of synchronic description is usually
there is no way of knowing) belonging to the vocabulary of some understood to carry the implication that, whereas synchronic de-
other language; and that word in turn from some earlier word in scription is independent of diachronic description, diachronic de-
another language: and so on. Generally speaking, etymologists are’ scription presupposes the prior synchronic analysis of the successive
not concerned nowadays with origins. Indeed, they would say that states through which languages have passed in the course of their
in many instances (e.g. the word ‘ten’) it does not make sense to historical development. This may not have been Saussure’s view.
enquire about the origin of a word. All that the etymologist can tell But it follows from what are now widely accepted assumptions
us, with greater or less confidence according to the evidence, is that about the nature of language-systems.
such and such is the form or meaning of a particular word’s earliest Linguists sometimes talk, rather misleadingly, as if the passage of
known or hypothetical ancestor. time was of itself sufficient to explain language-change. There are
This brings us to one of the more obvious ways in which Saut& many different factors, both within a language and external to it,
sure's analogy breaks down. Every game of chess, played according which may cause it to change from one synchronic state to another.
10 the rules and completed, has a determinate beginning and end. Some of these factors and possibly the most important of them,
Languages are not like this. Not only is it not the case (as far as we are social. The passage of time merely allows for their complex
58 Linguistics 2.6 Structure and system 59
interaction to bring about what is subsequently recognized as a:
transition from one state of the language to another. 2.6 Structure and system
Moreover the notion of diachronic development between succe One of the definitions of ‘language’ that I quoted in Chapter 1 was
ive states of a language makes sense only if it is applied with res Chomsky’s “a set (finite or infinite) of sentences, each finite in
to language-states that are relatively far removed from one another length and constructed out of a finite set of elements” (cf. 1.2). Let
in time. I have already referred to what I have called the fiction of us adopt this as a partial definition of the term ‘language-system’,
homogeneity (cf. 1.6). Up to a point, this is both useful and neces: which was introduced, it will be recalled, in order to eliminate some
sary. However, if it is assumed that language-change involves th of the ambiguity that attaches to the English word ‘language’.
constant transformation through time of what is at any one time a In so far as they are by definition both stable and uniform,
perfectly homogeneous language-system, the whole process of language-systems cannot be identified with real natural languages:
change in language appears to be much more mysterious than it they are theoretical constructs, postulated by the linguist in order to
really is. What is characteristic of the speech of an apparently account for such regularities as he finds in the language-behaviour
insignificant minority of the members of a language-community of the members of particular language-communities — more
one time may spread throughout most of the community in the precisely in the language-signals that are the products of their
course of a generation or two. It might be quite legitimate for the language-behaviour. As we have seen, real natural languages are
linguist describing the language synchronically at either of thes neither stable nor homogeneous. However, there is sufficient sta-
two points in time to discount the speech of the divergent minori bility and homogeneity in the speech of those who would generally
But, if he does so, and then goes on to talk diachronically of one be considered to speak the same language for the linguist’s postula-
synchronically homogeneous language-system being transformed tion of a common underlying language-system to be useful and
into another equally homogeneous language-system, he will scientifically justifiable, except when he is dealing explicitly with
guilty of distorting the facts. Worse than that, he will run the risk o synchronic and diachronic variation. Throughout the next three
creating for himself insoluble theoretical pseudo-problems. Once chapters, we shall take for granted the validity of the notion of the
we realize that no language is ever stable or uniform, we have made language-system as it is here defined and explained,
the first step towards accounting, theoretically, for the ubiquity and Among the language-signals produced by an English speaker,
continuity of language-change. If we take two diachronically deter- over a given period of time, some would be classified as sentences of
mined states of a language that are not widely separated in time we the language and some would not. We need not enquire at this stage
are likely to find that most of the differences between them are also what are the criteria by virtue of which this division into sentences
present as synchronic variation at both the earlier and the later and non-sentences is made. Obviously, there are principles that
time. From the microscopic point of view ~ as distinct from the determine the construction of larger texts and discourses. Further-
macroscopic point of view that one normally adopts in historical more, some of these principles are such that anyone violating them
linguistics — it is impossible to draw a sharp distinction between might reasonably be accused of breaking the rules of the language.
diachronic change and synchronic variation. Although it has not gone unchallenged in recent years, the tradi-
In summary, the principle of the priority of synchronic description tional assumption that most, if not all, of what is involved in
is valid, But, in so far as it rests upon the fiction of homogeneity,it knowing a language can be accounted for in terms of the construc-
must be applied sensibly and in full recognition of the theoretical tion and interpretation of sentences is still accepted by the majority
status of the concept of the language-system. It is to this point that of linguists.
we now turn. b Sentences, let us say, are what would be conventionally punctu-
ated as such in the written language. As we have seen, natural
60 Linguistics 2.6 Structure and system 61

languages have the property of medium-transferability (cf. 1.4) independence of syntax and phonology is demonstrated often quite
‘This means that, in general, any sentence of the written langu; dramatically in the process of creolization (cf. 9.3).
can be put into correspondence with a sentence of the spoken Natural languages, then, have two levels of structure and the
language and vice versa. Spoken sentences are not, of cous levels are independent, in the sense that the phonological structure
punctuated as such with anything that is strictly equivalent to of a language is not determined by its syntactic structure and its
initial capital letter or the closing full-stop, or period, of writ syntactic structure is not determined by its phonological structure.
sentences. For present purposes, however, we can establish a rou Itis unlikely, to say the least, that any two natural languages should
and ready equivalence between the punctuation-marks of a writtel exist such that every spoken or written sentence of the one can be
language and the intonation-patterns of the corresponding spoken: heard or read as a sentence of the other (with or without the same
language. j meaning). But it frequently happens, as a consequence of the
The term ‘structure’ figures prominently in modern linguistics, as independence of phonological and syntactic structure, that the
it does in many disciplines. If we adopt the point of view that w same combination of elements (sounds in speech and letters in
first clearly expressed by Saussure and is now accepted by all those alphabetic writing) realizes not one, but two or more, sentences.
who subscribe to the principles of structuralism, we will say not ol The sentences may be distinguished one from another by intonation
that a language-system has a structure, but that it is a structure. For or punctuation, as the case may be. Thus
example, in so far as written and spoken English are isomorphie (1) John says Peter has been here all the time
(i.e. have the same structure), they are the same language: there
is nothing but their structure that they have in common. The is distinguished from
language-system itself is, in principle, independent of the medium (2) John, says Peter, has been here all the time
inwhichitis manifest. Itis, in this sense, a purely abstract structure.
Language-systems are two-level structures: they have the in written English by means of punctuation; and they would nor-
property of duality (cf. 1.5). Spoken sentences are not just com- mally be distinguished from one another in spoken English by the
binations of phonological elements; they are also combinations of intonation-pattern superimposed upon them. But even without
syntactic units. Chomsky’s partial definition of a language-system differences of intonation or punctuation the same combination of
as a set of sentences, each of which is finite in length and constructed elements can realize more than one sentence. For example,
out of a finite set of elements, must be extended to take account of (3) We watched her box
this essential property of natural languages. It is logically possible
for two language-systems to be isomorphic on one level without could be either of two different English sentences, in one of which
their being isomorphic on the other. Indeed, as has been pointed heris an adjective-form (cf. his) and box a noun-form (cf. suitcase),
out already, it is because the so-called dialects of Chinese are in the other of which her is a pronoun-form (cf. him) and box is a
sufficiently close to being syntactically isomorphic (though they are verb-form (cf. wrestle). We need not bother about justifying the
far from being phonologically isomorphic) that the same, non- traditional syntactic analysis of (3) to which I have covertly
alphabetic, written language can be put into correspondence more appealed. This issomething that will be taken up later. It suffices for
or less equally well with each of them. It is also possible for lan- the present that we should have established that sentences, as
guages to be phonologically, but not syntactically, isomorphic. This traditionally defined, cannot be identified, and distinguished one
possibility is actualized to a greater or less degree by, let us say, a from another, in terms of the phonological elements of which they
native speaker of English speaking grammatically perfect French are composed. Indeed, as we can see from (3), they cannot even be
with a particularly bad English accent. More interestingly, the identified in terms of the syntactic units of which they are composed
62 Linguistics 2.6 Structure and system 63

without taking into account other aspects of syntactic structure, with acceptability, potentiality for use or even meaningfulness.
including the assignment of syntactic units to what are traditio There are indefinitely many sentences of English and other natural
called parts of speech (noun, verb, adjective, etc.). languages that, for various reasons, would not normally occur: they
The syntactic units out of which sentences are constructed, unli might contain an unacceptable juxtaposition of obscene or blas-
the phonological elements, are very numerous. Nevertheless, il phemous words; they might be stylistically awkward or excessively
the phonological elements, they are finite in number. Let us say thaf complex from a psychological point of view; they might be self-
every language-system presupposes the existence of a finite inven! contradictory or describe situations which do not occur in the world
of elements and a finite vocabulary of (simple) units, together witha inhabited by the society using the language in question. Any com-
set of rules (of perhaps several kinds) which interrelate the bination of elements or units of a given language, L, which is not
levels of structure and tell us which combinations of units are well-formed in terms of the rules of L is ill-formed with respect to L.
sentences of the language-system and, by implication, if not expli= Ill-formed combinations of elements or units may be marked as
citly, which are not. It should be noted that, as we shall see later, the such by means of a preceding asterisk.! Thus
vocabulary of a natural language is much more than a set of syn
tic units. None of the modifications or terminological refinements to (4) *He weren't doing nothing
be introduced in subsequent chapters affects the substance of wh is ill-formed, and indeed ungrammatical, with respect to Standard
has been said here. English. It is, however, grammatically well-formed in certain non-
For the time being, what we have been calling syntactic units ma standard dialects of English. This example illustrates the more
be thought of as forms: i.e. as combinations of elements such thai general point that different languages may be constructed out of the
every distinguishable combination is a distinct form. But forms, in same elements and units, what is well-formed in one language being
this sense of the term, have a meaning, and their meaning is far from ill-formed with respect to another language. Although the point has
being independent of their syntactic function. That this is so is clear beenillustrated with reference to two dialects of the same language,
in the case of the forms her and box in (3) above. The traditional it holds, in principle, for what would be thought of as quite different
view would be that there are (at least) two different words in the languages.
vocabulary of English, let us represent them (with single quotation= More could be said here about the structure of language-systems.
marks) as ‘box,’ and ‘box,’ respectively, which differ both in mean= It is best left for the chapters dealing with phonology, grammar and
ing and in syntactic function, but share the same form, box. We semantics, where the general points can be introduced gradually
shall later make more precise this traditional distinction between a and exemplified in greater detail 2
form and the unit of which it is a form; and, in doing so, we shall see We began this section by accepting Chomsky’s definition of a
that the term ‘word’, as used both by linguists and by laymen, i§ language (i.e. a language-system) as a set of sentences. It is
highly ambiguous (cf. 4.1).
1 The use of the asterisk to indicate ill-formedness must not be confused with its
Every sentence is by definition well-formed, both syntactically equally common, and longer established, employment in historical linguistics to
and phonologically, in the language-system of which it is a sentence. mark reconstructed forms (cf. 2.5), The context will make it clear which use is
The term ‘well-formed’ is broader than, but subsumes, the more intended,
2 The terms 'structure’ and ‘system’ are frequently used, especially by British ling-
traditional term ‘grammatical’, as the latter is broader than, but uists, in a specialized sense: “system’ for any set of elements or units that can occur
subsumes, the term ‘syntactically well-formed’. The nature and in the same position; ‘structure’ for any combination of elements and units that
results from the appropriate selection in particular positions. Defined in this way
limits of grammaticality (i.e. grammatical well-formedness) will bs { ‘structure’ and ‘system’ are complementary: each presupposes the other. Systems
discussed in Chapter 4. Here it is sufficient to make the point thnt- are established for particular positions in structures; structures are identified in
terms of the selections made from systems (cf. Berry, 1975). In this book, ‘system’
well-formedness (including grammaticality) must not be confused and ‘structure’ are employed in a more general sense.
64 Linguistics Questions and exercises 65

preferable, however, to think of a language-system as being com- are ‘unscientific’, not amenable to direct observation, variable and
posed of an inventory of elements, a vocabulary of units and the untrustworthy. It seems to us that this is not a valid objection ., . .
rules which determine the well-formedness of sentences on both (Smith & Wilson, 1979: 40). Discuss.
levels. And this is what we shall do from now on. Arguably, under . Devise an appropriate context for the Standard English utterance /

=S
the appropriate definition of ‘sentence’ the two ways of thinking of didn’t do nothing (with the appropriate prosodic structure).
language-systems coincide. What, if anything, is wrong with (a) between you and I and (b) You and
me will go together? Can logic or traditional Latin-based principles help
FURTHER READING us to decide?
What is the difference between the descriptive and the prescriptive (or
Generally, as for Chapter 1. In addition, Crystal (1971), chapters 2-3;
normative) approach to the investigation of language?
Lyons (1974).
Of the textbooks asterisked in the Bibliography, Robins (1979a) is the most Exemplify, from your experience if you can, the phenomenon of hyper-
comprehensive, and also the most neutral in its presentation of controver- correction.
sial issues; Lyons (1968) emphasizes the continuity between traditional . “The word ‘alibi’ is commonly misused these days: it is a legal term
grammar and modern linguistics, is restricted to synchronic microlinguis- which comes from the Latin word meaning “somewhere else” and
tics, and is slanted towards (a now outdated version of ) transformational should not be used as if it was synonymous with the everyday noun
grammar; Martinet (1960) is in the tradition of European structuralism; ‘excuse’.” Discuss.
Gleason (1961), Hill (1958) and Hockett (1958), together with Joos (1966), . Explain what is meant by the priority of the synchronic over the
give a good account of the field from the point of view of so-called post- diachronic point of view in linguistics.
Bloomfieldian linguistics; Southworth & Daswani (1974) is particularly
12, Give a critical account of Saussure’s famous comparison between a
good on linguistics in relation to sociology and anthropology, and also on
language and a game of chess.
applied linguistics; so too though less comprehensive is Falk (1973); Akma-
jian, Demers & Harnish (1979), Fromkin & Rodman (1974) and Smith & 13. A naive view of literal translation might be that it consists in the
Wilson (1979) are all consistently Chomskyan in inspiration and, generally one-for-one substitution of the word-forms of the target language for
speaking, stress the biological, rather than the cultural, in language. For the word-forms of the source-language. Ts this what is normally meant
discussion of the various trends and schools in modern linguistics, and for by the term ‘literal translation’? Can you identify some of the reasons
further references, cf. Chapter 7. why the naive view is unrealistic as far as natural languages are con-
cerned?
Historical (i.e. diachronic) linguistics is dealt with later (Chapter 6). So too
are most branches of macrolinguistics (Chapters 8-10). . “The language-system itself . . . is a purely abstract structure” (p. 60).
Consider this statement with reference to the use of simple codes and
On applied linguistics, see Corder (1973) and for more detailed discussion
Allen & Corder (19758, b, ). ciphers based on the principle of (a) letter-for-letter and (b) word-for-
word substitution in written messages. Do such cryptographic tech-
niques necessarily preserve or destroy isomorphism?
QUESTIONS AND EXERCISES
- Can you devise a simple code or cipher which exploits the indepen-
. In what sense is linguistics a science? Does this imply that it is not one of dence of the two levels of structure in a language-system and changes
the humanities? the one without affecting the other?
. “since every branch of knowledge makes use of language, linguistics
»

may, in some respects, be said to lie at the centre of them all, as being
the study of the tool that they must use” (Robins, 1979a: 7). Discuss.
3. “The only useful generalizations about language are inductive gener-
alizations” (Bloomfield, 1935: 20). Discuss,
. Why do linguists tend to be so critical of traditional grammar?
[TAFS

. “Itis often felt, by both philosophers and linguists, that . . . intuitions