Você está na página 1de 4



Because we have a word language, we assume that there must be some corresponding entity for the word to denote (see section 32). However, the linguist
Saussure (1969 [1916]: 19) points out to us that language is not an entity.1
Dening something like The English Language turns out to be a dicult task.
Part of the problem is that the language has so many dierent aspects. We
can view it as a social fact, as a psychological state, as a set of structures, or as a
collection of outputs.
A language is a social fact, a kind of social contract. It exists not in an individual, but in a community.
It is a treasure buried by the practice of speech in people belonging to
the same community, a grammatical system which has virtual existence in each brain, or more exactly in the brains of a collection of individuals; because language is not complete in any individual, but exists
only in the collectivity. (Saussure 1969 [1916]: 30, my translation, see
the footnote for the original French2)
A language can also be viewed as a mental reality. It exists in the heads of
people who speak it, and we assume its existence because of peoples ability to
learn languages in general and their practice in dealing with at least one


La langue nest pas une entit.

Cest un trsor dpos par la pratique de la parole dans les sujets appartenant une mme
communaut, un systme grammatical existant virtuellement dans chaque cerveau, ou plus
exactement dans les cervaux dun ensemble dindividus; car la langue nest complte dans
aucun, elle nexiste parfaitement que dans la masse.


particular language. [A] grammar is a mental entity, represented in the

mind/brain of an individual and characterising that individuals linguistic
capacity (Lightfoot 2000: 231). Note that Lightfoot here talks of a grammar
rather than of a language, but one possible denition of a language is precisely
that it is the grammatical system which allows speakers to produce appropriate utterances. Grammar has as many meanings as language (see section 4).
In this sense, we might see a language as a set of choices, a set of contrasts.
We can say Kim kissed the crocodile or The crocodile kissed Kim, but we cannot
choose to say, as a meaningful sentence of English, Kissed crocodile Kim the.
There is a system to what orders the words have to come in if they are to make
sense. We choose, in English, whether to say towel or cowl, but we cannot
choose, in English, to say something with a consonant half-way between the
/t/ of towel and the /k/ of cowl to mean something which is part towel and
part cowl (or, indeed, to mean anything else). There is a system to what sounds
we use in English. So a language can be viewed as a system of systems. This
view is usually attributed to Meillet: Every language forms a system in which
everything is interconnected (Meillet 1903: 407 [my translation]3). But he has
forerunners who make the same point in similar terms, e.g.: Every language is
a system all of whose parts interrelate and interact organically (von der
Gabelentz 1901: 481, as cited and translated by Matthews 2001: 6; see the footnote for the original German4).
Another alternative way of considering language is to ignore the way in
which speakers go about constructing utterances, and consider instead their
output, an actual set of utterances or (in a more idealised form) a set of sentences. A language can be dened as a set of sentences:
the totality of utterances that can be made in a speech community is
the language of that speech community. (Bloomeld 1957 [1926]: 26)
[A] language [is] a set (nite or innite) of sentences, each nite in length
and constructed out of a nite set of elements. (Chomsky 1957: 13)
The question of whether we should be dealing with utterances (things produced, whether in speech of in writing, by speakers) or sentences raises another
potential distinction. Chomsky (1986) introduces the notion of a distinction
between E-language and I-language. Smith (1994) already talks of this distinction as a customary one, which may be an overstatement of the case, but he
draws the distinction very clearly:


chaque langue forme un systme o tout se tient.

Jede Sprache is ein System, dessen smmtliche Theile organisch zusammenhngen und


E-language is the external manifestation of the internally (i.e. mentally) represented grammars (or I-languages) of many individuals.
E-languages are the appropriate domain for social, political, mathematical or logical statements; I-languages are the appropriate domain
for statements about individual knowledge. That this apparently narrower domain is worth considering follows from the fact that, as a
species, humans appear to be essentially identical in their linguistic
abilities. . . . [E]very child brings the same intellectual apparatus
(known as universal grammar) to bear on the task of acquiring his or
her rst language. (Smith 1994: 646)
So the utterances are E-language, while the sentences may well belong to Ilanguage, that hypothesised rather less error-prone system which we have in
our heads. But the intellectual apparatus which allows children to construct a
language like English for themselves is also, it is suggested, language in a rather
dierent sense. The language capacity, the feature which distinguishes humans
from other animals, is sometimes also simply called language.
There are so many complexities here that we might argue that it would be
better for linguists to give up attempting even to describe particular languages,
let alone language in the abstract. What are they to describe? Are they to
describe the social structure which is complete only in the collectivity, or the
mental structure which speakers of that language must be assumed to carry in
their heads, or the set of systems which are presumed to allow speakers to create
new utterances for themselves, or the actually produced utterances? All of these
have been tried. But note that there are logical inconsistencies between these
various potential objects of description. If language as a social fact exists only in
the collectivity, no individual speaks the language; any individual must have
only a partial knowledge of the language. This isnt hard to prove: open any large
dictionary of English at random, and read the rst fty headwords you come to.
You did not know all of these words before you started reading (you probably
dont after youve nished), but somebody (or, more likely, a set of individuals)
knows them and has used them or they wouldnt be in the dictionary. So the
description of what is in any persons head can never provide a full description
of a language in the sense that English is a language. Many linguists prefer to use
the term for the language of an individual. So you dont speak English,
you speak your idiolect. That seems simple enough until we ask what English
consists of. Presumably it consists of the sum of all the idiolects of people who
we agree are speaking English. But some of these people have conicting ideas
about what is part of their language. To take a simple example, there are millions
of people speaking what we would call English, for whom the past tense of the
verb dive is dove. For these speakers dived sounds like baby-talk, as writed would
instead of wrote. There are also millions of speakers for whom dived is the only


possible past tense of dive, and dove sounds like the kind of joke you make when
you say that the past tense of think must be thank or thunk. The example is trivial,
but it means that we must allow for a lot of dierent answers to what is English,
even mutually incompatible ones. So it must be true that there is no clear-cut line
where English stops and something else begins (and it is frequently not clear
what that something else is). The language English is not well-dened (and the
same will be true for any other language which is given a name in this way).
Neither is language in the sense language faculty well-dened. A lot of work
has gone into trying to understand Universal Grammar (or UG as it is usually
termed) within Chomskyan approaches to linguistics (see section 8), and we do
not yet understand what it must look like or how it must function. There is even
dispute as to whether it is a specically linguistic set of functions, or whether it
is a general set of cognitive abilities which together allow human beings to be
language users.
If neither a language nor language (the language faculty) is easily denable,
we have to ask what it is that linguists deal with. Linguists have to dene language for their own purposes. They do not have an external denition of language or of a particular language which is clearly sucient for their needs. This
is not necessarily a bad thing, but it does mean that care is required.
Bloomeld, Leonard (1957 [1926]). A set of postulates for the science of language.
Language 2: 15364. Reprinted in Martin Joos (ed.), Readings in Linguistics. Chicago
and London: University of Chicago Press, 2631.
Chomsky, Noam (1957). Syntactic Structures. The Hague and Paris: Mouton.
Chomsky, Noam (1986). Knowledge of Language. New York: Praeger.
Chomsky, Noam (2000). New Horizons in the Study of Language and Mind. Cambridge:
Cambridge University Press.
Gabelentz, Georg von der (1901 [1891]). Die Sprachwissenschaft. 2nd edn. Leipzig:
Lightfoot, David (2000). The spandrels of the linguistic genotype. In Chris Knight,
Michael Studdert-Kennedy & James R. Hurford (eds), The Evolutionary Emergence
of Language. Cambridge: Cambridge University Press, 23147.
Matthews, Peter (2001). A Short History of Structural Linguistics. Cambridge:
Cambridge University Press.
Meillet, Antoine (1903). Introduction ltude comparative des langues indo-europennes.
Paris: Hachette.
Saussure, Ferdinand de (1969 [1916]). Cours de linguistique gnrale. Paris: Payot.
Smith, N[eil] V. (1994). Competence and performance. In R. E. Asher (ed.),
Encyclopedia of Language and Linguistics. Oxford: Pergamon, Vol. 2, 6458.