American Anthropologist 68/4 (1966)

Dialect,Language,Nation'
EINAR HAUGEN
HarvardUniversity

Theimpossibilityof statingpreciselyhowmany"languages" or "dialects"arespokenin the

worldis dueto theambiguitiesof meaningpresentin theseterms,whichis showntostemfrom
theoriginaluse of "dialect"to referto theliterarydialectsof ancientGreece.In mostusages
theterm"language"is superordinate to "dialect,"butthenatureof thisrelationshipmaybe
eitherlinguisticor social,thelatterproblem It is
falling in theprovinceof sociolinguistics.
shownhowthedevelopment of a vernacular,popularlycalleda dialect,into a languageis in-
timatelyrelatedto thedevelopment of writingand thegrowthof nationalism.Thisprocessis
shownto involvetheselection,codification, acceptance, andelaborationof a linguisticnorm.

THE taxonomy of linguistic description-that is, the identification and

enumeration of languages-is greatly hampered by the ambiguities and
obscurities attaching to the terms "language" and "dialect." Laymen naturally
assume that these terms, which are both popular and scientific in their use, re-
fer to actual entities that are clearly distinguishable and therefore enumerable.
A typical question asked of the linguist is: "How many languages are there in
the world?" Or: "How many dialects are there in this country?"
The simple truth is that there is no answer to these questions, or at least
none that will stand up to closer scrutiny. Aside from the fact that a great
many, perhaps most, languages and dialects have not yet been adequately
studied and described, it is inherent in the very terms themselves that no
answer can be given. They represent a simple dichotomy in a situation that is
almost infinitely complex. Hence they have come to be used to distinguish
phenomena in several different dimensions, with resultant confusion and over-
lapping. The use of these terms has imposed a division in what is often a con-
tinuum, giving what appears to be a neat opposition when in fact the edges are
extremely ragged and uncertain. Do Americans and Englishmen speak dialects
of English, or do only Americans speak dialect, or is American perhaps a sepa-
rate language? Linguists do not hesitate to refer to the French language as a
dialect of Romance. This kind of overlapping is uncomfortable, but most lin-
guists have accepted it as a practical device, while recognizing, with Bloom-
field, "the purely relative nature of the distinction" (1933:54).
The two terms are best understood against the perspective of their history.
In English both words are borrowed from French. Language is the older, hav-
ing partially displaced such native words as "tongue" and "speech" already in
Middle English. The oldest attestation in the OED is from 1290: "With men
pat onder-stoden hire langage." The French word is itself late, being a popular
derivative of Latin lingua with the probable form *lingujticum, first attested
in the 12th century. Dialect, on the other hand, first appears in the Renais-
sance, as a learned loan from Greek. The oldest OED citation is from 1579 in
reference to "certain Hebrue dialectes," while the earliest French I have found
(in Hatzfeld and Darmesteter's dictionary) is only 16 years earlier and speaks
922

This content downloaded from 195.78.108.174 on Sun, 22 Jun 2014 20:41:14 PM

All use subject to JSTOR Terms and Conditions
[HAUGEN] Dialect, Language, Nation 923
of Greek as being "abondante en dialectes." A 1614 citation from Sir Walter
Raleigh's The History of the World refers to the "Aeolic Dialect" and confirms
the impression that the linguistic situation in ancient Greece was both the
model and the stimulus for the use of the term in modern writing.
There was need for some such term in Greece, since there was in the
classical period no unified Greek norm, only a group of closely related norms.
While these "dialects" bore the names of various Greek regions, they were not
spoken but written varieties of Greek, each one specialized for certain literary
uses, e.g., Ionic for history, Doric for the choral lyric, and Attic for tragedy.
In this period the language called "Greek" was therefore a group of distinct,
but related written norms known as "dialects." It is usually assumed that the
written dialects were ultimately based on spoken dialects of the regions whose
names they bore. These spoken dialects were in turn descended by normal lin-
guistic divergence from a Common Greek language of an older period, which
can be reconstructed by comparison of the dialects with each other and with
their Indo-European kinsmen. In the postclassical period, however, the Greek
dialects disappeared and were replaced by a rather well-unified Greek norm,
the koine, essentially the dialect of Athens. So, in the Hellenistic period
"Greek" became the name of a norm that resulted from a linguistic converg-
ence. The differences among the dialects were eliminated in favor of a single,
triumphant language, based on the dialect of the cultural and administrative
center of the Greeks.
The Greek situation has provided the model for all later usage of the two
terms "language" and "dialect." Much of the unclarity in their application
stems from the ambiguities present in that situation. This has become evident
with their extension to other countries and with their adoption into the tech-
nical terminology of linguistics. In a descriptive, synchronic sense "language
can refer either to a single linguistic norm, or to a group of related norms. In a
historical, diachronic sense "language" can either be a common language on its
way to dissolution, or a common language resulting from unification. A "dia-
lect" is then any one of the related norms comprised under the general name
"language," historically the result of either divergence or convergence.
Since this historical process can be indefinitely repeated, the two terms are
cyclically applicable, with "language" always the superordinate and "dialect"
the subordinate term. This is also clear from the kind of formal structures into
which they can be placed: "X is a dialect of language Y," or "Y has the dialects
X and Z" (never, for example, "Y is a language of dialect X"). "Language" as
the superordinate term can be used without reference to dialects, but "dialect"
is meaningless unless it is implied that there are other dialects and a language
to which they can be said to "belong." Hence every dialect is a language, but
not every language is a dialect.
In addition to the ambiguities provided by the synchronic and diachronic
points of view distinguished above, increasing knowledge concerning linguistic
behavior has made the simple application of these two contrasting terms ever
more difficult.

This content downloaded from 195.78.108.174 on Sun, 22 Jun 2014 20:41:14 PM

All use subject to JSTOR Terms and Conditions
924 American Anthropologist [68, 1966
In French usage a third term developed, patois, which applied primarily to
the spoken language. The term dialecte is defined in the dictionary of the
Academie frangaise and other French dictionaries as "variet6 regionale d'une
langue." Littre (1956) explicitly requires that a dialect "include a complete
literary culture" (comportant une complete culture litteraire). As pointed out
by Andre Martinet (1964), this usage reflects the special French situation, in
which there were a number of regional written standards, which were then
superseded by the written standard of Paris. The French dialects were regional,
like the Greek, and literary, but not functionally distinguished like the Greek.
When the dialects ceased to be written, they became "patois": "Apres le
XIVe siecle, il se forma une langue litteraire et &crite,et les dialectes devinrent
des patois" (Littre). Even more succinctly, Brun (1946) writes: "Un patois est
un dialecte qui s'est degrade." A patois, then, is a language norm not used for
literary (and hence official) purposes, chiefly limited to informal situations.
Thus Provengal might be considered a French dialect, but its local, spoken
varieties are all patois. This distinction introduces a new dimension in our dis-
cussion: the social functions of a language. In terms of the language-dialect
distinction, we may say that a patois is a dialect that serves a population in its
least prestigious functions. The distinction of patois-dialect is therefore not
one between two kinds of language, but between two functions of language.
The definition in Littr6 (and others like it) clearly suggests a pejorative atti-
tude toward the patois, since it no longer carries with it "a complete literary
culture."
In English the term "patois" has never been seriously adopted in the de-
scription of language, and "dialect" has carried the full burden of both scien-
tific and popular usage. Older writers, cited in the OED, often used it for any
specialized variety of the language, e.g., "the lawyer's dialect." Samuel Butler
(Hudibras, 1663) railed against "a Babylonish dialect, which learned pedants
much affect." General usage has limited the word largely to the regional or
locally based varieties, such as "Lancashire dialect" or "Irish dialect" in
reference to varieties of English. It is less customary to speak of "London
dialect" or "Boston dialect," except in reference to the lower-class speech of
those cities. Nor is it common to speak of "British dialect" in reference to
cultivated English speech, and Americans are generally resentful of being told
they speak "American dialect" when reference is had to the speech of edu-
cated people. Martinet is therefore beside the mark when he writes that in
America "the term denotes every local form of English but without any sug-
gestion that a more acceptable form of the language exists distinct from the
dialects" (1964:146). It is quite different with the word "accent": an Amer-
ican may inoffensively be described as having a "New England accent" or a
"Southern accent," and, of course, all Americans speak of the English as hav-
ing an "English accent." "Dialect" is here as elsewhere a term that suggests
informal or lower-class or rural speech. In general usage it therefore remains
quite undefined whether such dialects are part of the "language" or not. In
fact, the dialect is often thought of as standing outside the language: "That

This content downloaded from 195.78.108.174 on Sun, 22 Jun 2014 20:41:14 PM

All use subject to JSTOR Terms and Conditions
HAUGEN] Dialect, Language, Nation 925
isn't English." This results from the de facto development of a standard lan-
guage, with all the segregation of an elite and the pyramidal power structure
that it has usually implied.
As a social norm, then, a dialect is a language that is excluded from polite
society. It is, as Auguste Brun (1946) has pointed out, a language that "did
not succeed." In Italy, Piedmontese is from every linguistic point of view a
language, distinct from Italian on the one hand and French on the other, with
a long tradition of writing and grammatical study. But because it is not Tus-
can, and Tuscan became the standard language of all Italy, Piedmontese is
only a "dialect," yielding ground to Italian with every generation and kept
alive only by local pride and linguistic inertia (Clivio 1964). Only if a "dia-
lect" is watered down to an "accent"-that is, an intonation and a set of artic-
ulations, with an occasional lexical item thrown in for color-does it (say in
Germany or Italy or England) become "salonfiihig." As a complete structure
it is out in the cold limbo of modern society. In America the stigma is placed
not so much on local dialects, since these are few and rarely heard, as on
"bad" English, which is quite simply lower-class dialect. The language of the
upper classes is automatically established as the correct form of expression.
They cannot say only, "L'Vtat,c'est moi," but also "Le langage, c'est le mien."

In trying to clarify these relationships, linguistic science has been only

moderately successful. Even in the Renaissance it was perfectly clear to seri-
ous students of the subject that the term "language" was associated with the
rise of a nation to conscious unity and identity. George Puttenham wrote in
his book The Arte of English Poesie (1589): "After a speach is fully fashioned
to the common understanding, and accepted by consent of a whole country and
nation, it is called a language." This kind of historical development, by which
convergence was achieved at the expense of deviating varieties, was familiar
to the men of that age. But the arbitrary tower-of-Babel approach to linguistic
divergence was dispelled by the discovery, in the early 19th century, of histori-
cal regularity. The realization that languages have resulted from dialect-split-
ting gave a new content to the terms and made it possible to begin calling lan-
guages like English and German "dialects" of a Germanic "language."
But in the mid-19th century, when scientific study of the rural and socially
disadvantaged dialects began, a generation of research was sufficient to revolu-
tionize the whole idea of how a dialect arises. The very notion of an area
divided into a given number of dialects, one neatly distinct from the next, had
to be abandoned. The idea that languages split like branches on a tree gave
way to an entirely different and even incompatible idea, namely, that in-
dividual linguistic traits diffused through social space and formed isoglosses
that rarely coincided. Instead of a dialect, one had a "Kernlandschaft" with
ragged edges, where bundles of isoglosses testified that some slight barrier had
been interposed to free communication. Linguistics is still saddled with these
irreconcilable "particle" and "wave" theories; this in effect involves the differ-
ing points of view from which any linguistic structure can be seen: as a unitary

This content downloaded from 195.78.108.174 on Sun, 22 Jun 2014 20:41:14 PM

All use subject to JSTOR Terms and Conditions
926 American Anthropologist [68, 1966
structure (a "language"), or as one of several partially overlapping structures
(the "dialects").
Without going into the problems raised by this conflict, we may simply
state that the "particle" theory of language as a unified structure is a fruitful
hypothesis, making it possible to produce an exhaustive and self-consistent de-
scription. But it excludes as "free variation" a great many inconsistencies
within the speech of any informant, and it fails to account for the fact that
communication is possible between users of identifiably different codes. Com-
parative grammar succeeded in reconstructing the common structure from
which "dialects" could be derived. Contrastive grammar has tried to program
the differences between languages in order to ease the learner's task or, on a
higher theoretical plane, to arrive at a linguistic typology. But there is still no
calculus that permits us to describe the differences between languages in a co-
herent and theoretically valid way.

Our discussion has shown that there are two clearly distinct dimensions in-
volved in the various usages of "language" and "dialect." One of these is
structural, that is, descriptive of the language itself; the other is functional,
that is, descriptive of its social uses in communication. Since the study of lin-
guistic structure is regarded by linguists as their central task, it remains for
sociologists, or more specifically, sociolinguists, to devote themselves to the
study of the functional problem.
In the structuraluse of "language" and "dialect," the overriding consider-
ation is genetic relationship. If a linguist says that Ntongo has five dialects, he
means that there are five identifiably different speech-forms that have enough
demonstrable cognates to make it certain that they have all developed from
one earlier speech-form. He may also be referring to the fact that these are
mutually understandable, or at least that each dialect is understandable to its
immediate neighbors. If not, he may call them different languages, and say
that there is a language Ntongo with three dialects and another, Mbongo, with
two. Ntongo and Mbongo may then be dialects of Ngkongo, a common an-
cestor. This introduces the synchronic dimension of comprehension, which is at
best an extremely uncertain criterion. The linguist may attempt to predict, on
the basis of his study of their grammars, that they should or should not be
comprehensible. But only by testing the reactions of the speakers themselves
and their interactions can he confirm his prediction (Voegelin and Harris
1951; Hickerson et al., 1952). Between total incomprehension and total com-
prehension there is a large twilight zone of partial comprehension in which
something occurs that we may call "semicommunication."
In the functional use of "language" and "dialect," the overriding consider-
ation is the uses the speakers make of the codes they master. If a sociolinguist
says that there is no Ntongo language, only dialects, he may mean that there is
no present-day form of these dialects that has validity beyond its local speech
community, either as a trade language or as a common denominator in inter-
action among the various dialect speakers. A "language" is thus functionally

This content downloaded from 195.78.108.174 on Sun, 22 Jun 2014 20:41:14 PM

All use subject to JSTOR Terms and Conditions
HAUGEN] Dialect, Language, Nation 927
defined as a superposed norm used by speakers whose first and ordinary lan-
guage may be different. A "language" is the medium of communication be-
tween speakers of different dialects. This holds only within the limits estab-
lished by their linguistic cognacy: one could not speak of Ntongo as a dialect
of English just because its speakers use English as a medium of intercommuni-
cation. The sociolinguist may also be referring to the fact that the "language"
is more prestigious than the "dialect." Because of its wider functions it is likely
to be embraced with a reverence, a language loyalty, that the dialects do not
enjoy. Hence the possibility of saying that "Mbongo is only a dialect, while
Ngkongo is a language." This means that Ngkongo is being spoken by people
whose social prestige is notoriously higher than that of people who speak
Mbongo. When used in this sense, a dialect may be defined as an undeveloped
(or underdeveloped) language. It is a language that no one has taken the
trouble to develop into what is often referred to as a "standard language."
This dimension of functional superiority and inferiority is usually disregarded
by linguists, but it is an essential part of the sociolinguist's concern. It becomes
his special and complex task to define the social functions of each language or
dialect and the prestige that attaches to each of these.
What is meant by an "undeveloped" language? Only that it has not been
employed in all the functions that a language can perform in a society larger
than that of the local tribe or peasant village. The history of languages dem-
onstrates convincingly that there is no such thing as an inherently handicapped
language. All the great languages of today were once undeveloped. Rather
than speak of undeveloped languages as "dialects," after the popular fashion,
it would be better to call them "vernaculars," or some such term, and limit
"dialect" to the linguist's meaning of a "cognate variety." We are then ready
to ask how a vernacular, an "undeveloped language," develops into a stan-
dard, a "developed language." To understand this we will have to consider the
relation of language to the nation.
The ancient Greeks and Romans spread their languages as far as their
domains extended, and modern imperialists have sought to do the same. But
within the modern world, technological and political revolutions have brought
Everyman the opportunity to participate in political decisions to his own ad-
vantage. The invention of printing, the rise of industry, and the spread of
popular education have brought into being the modern nation-state, which ex-
tends some of the loyalties of the family and the neighborhood or the clan to
the whole state. Nation and language have become inextricably intertwined.
Every self-respecting nation has to have a language. Not just a medium of
communication, a "vernacular" or a "dialect," but a fully developed language.
Anything less marks it as underdeveloped.
The definition of a nation is a problem for historians and other social
scientists; we may accept the idea that it is the effective unit of international
political action, as reflected in the organization of the United Nations General
Assembly. As a political unit it will presumably be more effective if it is also a
social unit. Like any unit, it minimizes internal differences and maximizes

This content downloaded from 195.78.108.174 on Sun, 22 Jun 2014 20:41:14 PM

All use subject to JSTOR Terms and Conditions
928 American Anthropologist [68, 1966
external ones. On the individual's personal and local identity it superimposes a
national one by identifying his ego with that of all others within the nation and
separating it from that of all others outside the nation. In a society that is
essentially familial or tribal or regional it stimulates a loyalty beyond the pri-
mary groups, but discourages any conflicting loyalty to other nations. The
ideal is: internal cohesion-external distinction.
Since the encouragement of such loyalty requires free and rather intense
communication within the nation, the national ideal demands that there be a
single linguistic code by means of which this communication can take place.
It is characteristic that the French revolutionaries passed a resolution con-
demning the dialects as a remnant of feudal society. The dialects, at least if
they threaten to become languages, are potentially disruptive forces in a uni-
fied nation: they appeal to local loyalties, which could conceivably come into
conflict with national loyalty. This is presumably the reason that France even
now refuses to count the number of Breton speakers in her census, let alone
face the much greater problem of counting the speakers of Provengal. On the
other hand, a nation feels handicapped if it is required to make use of more
than one language for official purposes, as is the case in Switzerland, Belgium,
Yugoslavia, Canada, and many other countries. Internal conflict is inevitable
unless the country is loosely federated and the language borders are stable, as
is the case in Switzerland.
Nationalism has also tended to encourage external distinction, as noted
above. In language this has meant the urge not only to have one language, but
to have one's own language. This automatically secludes the population from
other populations, who might otherwise undermine its loyalty. Here the urge
for separatism has come into sharp conflict with the urge for international con-
tact and for the advantages accruing both to individual and nation from such
contact. Switzerland is extreme in having three languages, no one of which is its
own; Belgium has two, both of which belong to its neighbors. The Irish move-
ment has faltered largely under the impact of the overwhelming strength of
English as a language of international contact. The weakness of the New
Norwegian language movement is due to the thorough embedding of Danish in
the national life during four centuries of union; what strength the movement
has had is derived from the fact that Danish was not one of the great inter-
national languages.
Whenever any important segment of the population, an elite, is familiar
with the language of another nation, it is tempting to make use of this as the
medium of government, simply as a matter of convenience. If this is also the
language of most of the people, as was the case when the United States broke
away from England, the problem is easily solved; at most it involves the ques-
tion of whether provincialisms are to be recognized as acceptable. But where it
is not, there is the necessity of linguistically re-educating a population, with
all the effort and disruption of cultural unity that this entails. This is the prob-
lem faced by many of the emerging African and Asian nations today (Le Page
1964). French and English have overwhelming advantages, but they sym-

This content downloaded from 195.78.108.174 on Sun, 22 Jun 2014 20:41:14 PM

All use subject to JSTOR Terms and Conditions
HAUGEN] Dialect, Language, Nation 929
bolize past oppression and convey an alien culture. The cost of re-education is
not just the expense in terms of dollars and cents, but the malaise of training
one's children in a medium that is not their own, and of alienation from one's
own past.
The alternative is to develop one's own language, as Finland did in the
19th century, or Israel did in the 20th. Different languages start at different
points: Finland's was an unwritten vernacular, Israel's an unspoken standard.
Today both are standards capable of conveying every concept of modern
learning and every subtlety of modern literature. Whatever they may lack is
being supplied by deliberate planning, which in modern states is often an im-
portant part of the development process.
It is a significant and probably crucial requirement for a standard language
that it be written. This is not to say that languages need to be written in order
to spread widely or be the medium of great empires. Indo-European is an ex-
ample of the first, Quechua of the Inca Empire an example of the second
(Buck 1916). But they could not, like written languages, establish models
across time and space, and they were subject to regular and inexorable linguis-
tic change. It is often held that written language impedes the "natural" de-
velopment of spoken language, but this is still a matter of discussion (Zengel
1962; Bright and Ramanujan 1964). In any case the two varieties must not be
confused.
Speech is basic in learning language. The spoken language is acquired by
nearly all its users before they can possibly read or write. Its form is to a great
extent transmitted from one generation of children to the next. While basic
habits can be modified, they are not easily overturned after childhood and are
virtually immovable after puberty. The spoken language is conveyed by
mouth and ear and mobilizes the entire personality in immediate interaction
with one's environment. Writing is conveyed by hand and eye, mobilizes the
personality less completely, and provides for only a delayed response. Oral
confrontation is of basic importance in all societies, but in a complex, literate
society it is overlaid and supplemented by the role of writing.
The permanence and power of writing is such that in some societies the
written standard has been influential in shaping new standards of speech. This
is not to say that writing has always brought them into being, but rather to
say that new norms have arisen that are an amalgamation of speech and writ-
ing. This can of course take place only when the writing is read aloud, so that
it acquires an oral component (Wess6n 1937). There is some analogy between
the rise of such spoken standards and that of pidgin or creole languages
(Meillet 1925:76; Sommerfelt 1938:44). The latter comprise elements of the
structure and vocabulary of two or more languages, all oral. They have usually
a low social value, compared to the oral standards, but the process of origin is
comparable. The reawakening of Hebrew from its century-long dormant state
is comprehensible only in terms of the existence of rabbinical traditions of
reading scripture aloud (Morag 1959). Modern Hebrew has shown a rapid
adaptation to the underlying norms of its new native speakers, so that it has

This content downloaded from 195.78.108.174 on Sun, 22 Jun 2014 20:41:14 PM

All use subject to JSTOR Terms and Conditions
930 American Anthropologist [68, 1966
become something different from traditional Hebrew. Similarly with the
standard forms of European languages: one is often hard put to say whether a
given form has been handed down from its ancestor by word of mouth or via
the printed page. "Spelling pronunciations" are a well-known part of most oral
standards, even though purists tend to decry them.
While we have so far spoken of standard languages as if they were a clear
and unambiguous category, there are differences of degree even among the
well-established languages. French is probably the most highly standardized
of European languages, more so than, for example, English or German. French,
as the most immediate heir of Latin, took over many of its concepts of correct-
ness and its intellectual elaboration. French in turn became a model for other
standard languages, and its users were for centuries nothing loth to have it so
considered. When English writers of the 18th century debated whether an
English academy should be established to regulate the language, the idea of
such an institution came from France. The proposal was rejected largely be-
cause the English did not wish to duplicate what they regarded as French
"tyranny."
In France, as in other countries, the process of standardization was inti-
mately tied to the history of the nation itself. As the people developed a sense
of cohesion around a common government, their language became a vehicle
and a symbol of their unity. The process is reasonably well documented in the
histories written for the older European languages. But the period since the
French Revolution has seen a veritable language explosion, which has been far
less adequately studied. In many countries a process that elsewhere took cen-
turies of effort on the part of a people and its writers has been compressed into
a few short years or decades. In a study of the new standards developed since
1800 for Germanic languages, Heinz Kloss has suggested that there may be a
typical profile for what he has called the "Ausbau" of a new language (Kloss
1952:28). First comes its use for purely humorous or folkloristic purposes.
Then lyric writers may adopt it, followed by prose narrators. But it has not
reached a crucial stage of development until success is achieved in writing
serious expository prose, or what he calls "Zweckschrifttum." Beyond this
comes the elaboration of the language for purposes of technical and scientific
writing and government use. Each of these "domains" (as Fishman [1964] has
called them) constitutes a challenge for the language in its attempt to achieve
full development.
While making a survey of the world's standard languages, Ferguson pro-
posed (1962) to classify them along two dimensions: their degree of standardi-
zation (St. 0, 1, 2) and their utilization in writing (W 0, 1, 2, 3). Zero meant in
each case no appreciable standardization or writing. St. 1 meant that a language
was standardized in more than one mode, as is the case, for example, with
Armenian, Greek, Serbo-Croatian, and Hindi-Urdu. He also included Nor-
wegian, but it is at least arguable that we are here dealing with two languages.
St. 2 he defined as a language having a "single, widely accepted norm which is

This content downloaded from 195.78.108.174 on Sun, 22 Jun 2014 20:41:14 PM

All use subject to JSTOR Terms and Conditions
HAUGEN] Dialect, Language, Nation 931
felt to be appropriate with only minor modifications or variations for all
purposes for which the language is used." W 1 he applied to a language used
for "normal written purposes," W 2 to one used for "original research in physi-
cal sciences," and W 3 to one used for "translations and r6sumes of scientific
work in other languages."
These categories suggest the path that "underdeveloped" languages must
take to become adequate instruments for a modern nation. The "standar-
dization" to which Ferguson refers applies primarily to developing the form of a
language, i.e., its linguistic structure, including phonology, grammar, and
lexicon. We shall call this the problem of codification. Ferguson's scale of
"utilization in writing" applies rafher to the functions of a language. We shall
call this the problem of elaboration, a term suggested by a similar usage of
Bernstein's (1962) and corresponding to Kloss's Ausbau. As the ideal goals of
a standard language, codification may be defined as minimal variation in form,
elaboration as maximal variation in function.
The ideal case of minimal variation in form would be a hypothetical,
"pure" variety of a language having only one spelling and one pronunciation
for every word, one word for every meaning, and one grammatical framework
for all utterances. For purposes of efficient communication this is obviously the
ideal code. If speakers and listeners have identical codes, no problems of mis-
understanding can arise due to differences in language. There can be none of
what communication engineers call "code noise" in the channel (Hockett
1958:331-332). This condition is best attained if the language has a high de-
gree of stability, a quality emphasized by many writers on the subject (e.g.,
Havranek 1938). Stability means the slowing down or complete stoppage of
linguistic change. It means the fixation forever (or for as long as possible) of a
uniform norm. In practice such fixation has proved to be chimerical, since even
the most stable of norms inevitably changes as generations come and go. At all
times the standard is threatened by the existence of rival norms, the so-called
"dialects," among its users. It is liable to interference from them and even-
tually to complete fragmentation by them.
Apparently opposed to the strict codification of form stands the maximal
variation or elaboration of function one expects from a fully developed lan-
guage. Since it is by definition the common language of a social group more
complex and inclusive than those using vernaculars, its functional domains
must also be complex. It must answer to the needs of a variety of communities,
classes, occupations, and interest groups. It must meet the basic test of ade-
quacy. Any vernacular is presumably adequate at a given moment for the
needs of the group that uses it. But for the needs of the much larger society of
the nation it is not adequate, and it becomes necessary to supplement its re-
sources to make it into a language. Every vernacular can at the very least add
words borrowed from other languages, but usually possesses devices for making
new words from its own resources as well. Writing, which provides for the
virtually unlimited storage and distribution of vocabulary, is the technological

This content downloaded from 195.78.108.174 on Sun, 22 Jun 2014 20:41:14 PM

All use subject to JSTOR Terms and Conditions
932 American Anthropologist [68, 1966
device enabling a modern standard language to meet the needs of every spe-
cialty devised by its users. There are no limits to the elaboration of language
except those set by the ingenuity of man.
While form and function may generally be distinguished as we have just
done, there is one area in which they overlap. Elaboration of function may lead
to complexity of form, and, contrariwise, unity of form may lead to rigidity of
function. This area of interaction between form and function is the domain of
style. A codification may be so rigid as to prevent the use of a language for
other than formal purposes. Sanskrit had to yield to Prakrit, and Latin to the
Romance languages, when the gap between written and spoken language be-
came so large that only a very few people wverewilling to make the effort of
learning them. Instead of being appropriate for "all purposes for which the
language is used," the standard tends to become only one of several styles
within a speech community. This can lead to what Ferguson (1959) has de-
scribed as "diglossia," a sharp cleavage between "high" and "low" style. Or it
may be a continuum, with only a mild degree of what I have called "schizo-
glossia," as in the case of English (Haugen 1962). In English there is a marked
difference between the written and spoken standards of most people. In addi-
tion, there are styles within each, according to the situation. These styles,
which could be called "functional dialects," provide wealth and diversity
within a language and ensure that the stability or rigidity of the norm will
have an element of elasticity as well. A complete language has its formal and
informal styles, its regional accents, and its class or occupational jargons,
which do not destroy its unity so long as they are clearly diversified in function
and show a reasonable degree of solidarity with one another.
Neither codification nor elaboration is likely to proceed very far unless the
community can agree on the selection of some kind of a model from which the
norm can be derived. Where a new norm is to be established, the problem will
be as complex as the sociolinguistic structure of the people involved. There will
be little difficulty where everyone speaks virtually alike, a situation rarely
found. Elsewhere it may be necessary to make some embarrassing decisions.
To choose any one vernacular as a norm means to favor the group of people
speaking that variety. It gives them prestige as norm-bearers and a headstart
in the race for power and position. If a recognized 6lite already exists with a
characteristic vernacular, its norm will almost inevitably prevail. But where
there are socially coordinate groups of people within the community, usually
distributed regionally or tribally, the choice of any one will meet with resist-
ance from the rest. This resistance is likely to be the stronger the greater the
language distance within the group. It may often be a question of solidarity
versus alienation: a group that feels intense solidarity is willing to overcome
great linguistic differences, while one that does not is alienated by relatively
small differences. Where transitions are gradual, it may be possible to find a
central dialect that mediates between extremes, one that will be the easiest to
learn and most conducive to group coherence.
Where this is impossible, it may be necessary to resort to the construction
of a new standard. To some extent this has happened naturally in the rise of

This content downloaded from 195.78.108.174 on Sun, 22 Jun 2014 20:41:14 PM

All use subject to JSTOR Terms and Conditions
HAUGEN] Dialect, Language, Nation 933
the traditional norms; it has been the aim of many language reformers to dupli-
cate the effect in new ones. For related dialects one can apply principles of
linguistic reconstruction to make a hypothetical mother tongue for them all.
Or one can be guided by some actual or supposed mother tongue, which exists
in older, traditional writings. Or one can combine those forms that have the
widest usage, in the hope that they will most easily win general acceptance.
These three procedures-the comparative, the archaizing, and the statistical
-may easily clash, to make decisions difficult. In countries where there are
actually different languages, amounting in some African nations to more than a
hundred, it will be necessary either to recognize multiple norms or to introduce
an alien norm, which will usually be an international language like English or
French.
Finally, a standard language, if it is not to be dismissed as dead, must have
a body of users. Acceptanceof the norm, even by a small but influential group,
is part of the life of the language. Any learning requires the expenditure of time
and effort, and it must somehow contribute to the well-being of the learners if
they are not to shirk their lessons. A standard language that is the instrument
of an authority, such as a government, can offer its users material rewards in
the form of power and position. One that is the instrument of a religious fellow-
ship, such as a church, can also offer its users rewards in the hereafter. National
languages have offered membership in the nation, an identity that gives one
entree into a new kind of group, which is not just kinship, or government, or
religion, but a novel and peculiarly modern brew of all three. The kind of signi-
ficance attributed to language in this context has little to do with its value as
an instrument of thought or persuasion. It is primarily symbolic, a matter of
the prestige (or lack of it) that attaches to specific forms or varieties of lan-
guage by virtue of identifying the social status of their users (Labov 1964). Mas-
tery of the standard language will naturally have a higher value if it admits one
to the councils of the mighty. If it does not, the inducement to learn it, except
perhaps passively, may be very low; if social status is fixed by other criteria,
it is conceivable that centuries could pass without a population's adopting it
(Gumperz 1962, 1964). But in our industrialized and democratic age there are
obvious reasons for the rapid spread of standard languages and for their impor-
tance in the school systems of every nation.
The four aspects of language development that we have now isolated as
crucial features in taking the step from "dialect" to "language," from vernacu-
lar to standard, are as follows: (1) selection of norm, (2) codification of form,
(3) elaboration of function, and (4) acceptance by the community. The first
two refer primarily to the form, the last two to the function of language. The
first and the last are concerned with society, the second and third with lan-
guage. They form a matrix within which it should be possible to discuss all the
major problems of language and dialect in the life of a nation:
Form Function
Society Selection Acceptance
Language Codification Elaboration

This content downloaded from 195.78.108.174 on Sun, 22 Jun 2014 20:41:14 PM

All use subject to JSTOR Terms and Conditions
934 American Anthropologist [68, 1966
NOTES
1 This paperwas writtenas a contributionto the workof the Seminaron Sociolinguistics,
held
at the Indiana UniversityLinguisticInstitute in the summerof 1964, under the directionof
CharlesA. Ferguson.It has profitedfromextensivediscussionwith the membersof the Seminar.

REFERENCES CITED

BERNSTEIN, BASIL
1962 Linguistic codes, hesitation phenomena and intelligence. Language and Speech
5:31-46.
BLOOMFIELD, LEONARD
1933 Language. New York, Holt.
BRIGHT,WILLIAM, ANDA. K. RAMANUJAN
1964 Sociolinguistic variation and language change. Proceedings, IX International Con-
gress of Linguists. The Hague, Mouton. Pp. 1107-1114.
BRUN,AUGUSTE
1946 Parlers rdgionaux: France dialectale et unite frangaise. Paris, Toulouse: Didier.
BUCK,C. D.
1916 Language and the sentiment of nationality. American Political Science Review
10:44-69.
CLIVIO, GIANRENZO
1964 Piedmontese: a short basic course. Mimeographed ed. Center for Applied Linguistics,
Washington, D. C.
DEL'ACADEMIE
DICTIONNAIRE FRANVAISE
1932 Paris, 8th ed.
FERGUSON,CHARLES A.
1959 Diglossia. Word 15:325-340.
1962 The language factor in national development. Anthropological Linguistics 4, 1:23-
27.
FISHMAN,JOSHUA
1964 Language maintenance-language shift as a field of inquiry. Linguistics 9:32-70.
GUMPERZ,JOHN
1962 Types of linguistic communities. Anthropological Linguistics 4, 1:28-40.
1964 Hindi-Punjabi code switching in Delhi. Proceedings, IX International Congress of
Linguists. The Hague, Mouton. Pp. 1115-1124.
HATZFELD,A., AND A. DARMESTETER
1920 Dictionnaire general de la langue frangaise. Paris, 6. ed.
HAUGEN, EINAR
1962 Schizoglossia and the linguistic norm. Monograph Series on Languages and Lin-
guistics. Georgetown University, Washington, D. C. No. 15, 63-69.
HAVRANEK, BOHUMIL
1938 Zum Problem der Norm in der heutigen Sprachwissenschaft und Sprachkultur.
Repr. in A Prague school reader in linguistics. J. Vachek, ed. Bloomington, Ind. Pp.
413-420.
HICKERSON,H., G. D. TURNER,ANDN. P. HICKERSON
1952 Testing procedures for estimating transfer of information among Iroquois dialects
and languages. IJAL 18:1-8.
HOCKETT,CHARLES F.
1958 A course in modern linguistics. New York, Macmillan.
KLOSS,HEINZ
1952 Die Entwicklung neuer germanischer Kultursprachen von 1800 bis 1950. Munich,
Pohl.

This content downloaded from 195.78.108.174 on Sun, 22 Jun 2014 20:41:14 PM

All use subject to JSTOR Terms and Conditions
HAUGEN] Dialect, Language, Nation 935
LABOV, WILLIAM
1964 Phonological correlates of social stratification. In The ethnography of communica-
tion. John J. Gumperz and Dell Hymes, eds. American Anthropologist 66, no. 6,
pt. 2:164-176.
LE PAGE,R. B.
1964 The national language question: linguistic problems of newly independent states.
London, Oxford University Press.
LITTRE,E.
1956 Dictionnaire de la langue frangaise. Paris.
MARTINET, ANDRP
1964 Elements of general linguistics. Chicago, University of Chicago Press.
MEILLET, ANTOINE
1925 La methode comparative en linguistique historique. Oslo, Institutet for sammen-
lignende kulturforskning.
MORAG, SHELOMO
1959 Planned and unplanned development in modern Hebrew. Lingua 8:247-263.
ENGLISH
OXFORD DICTIONARY
1888 ff. James A. H. Murray et al., eds. Oxford University Press.
SOMMERFELT,ALP
1938 Conditions de la formation d'une language commune. Actes du IV Congr?s interna-
tional de linguistes. Copenhagen. Pp. 42-48.
VOEGELIN, C. F., and Z. S. HARRIS
1951 Methods for determining intelligibility among dialects of natural languages.
Proceedings of the American Philosophical Society 95:322-329.
WESSEN, ELIAS
1937 Virt rikssprdk: NAgra huvudpunkter av dess historiska utveckling. Modersmils-
litrarnas F6renings Arsskrift. Pp. 289-305.
ZENGEL, MARJORIE S.
1962 Literacy as a factor in language change. American Anthropologist 64:132-139.

This content downloaded from 195.78.108.174 on Sun, 22 Jun 2014 20:41:14 PM

All use subject to JSTOR Terms and Conditions