JOURNAL OF SOCIAL ISSUES

VOLUME -, NUMBER 2, 1967

Models of the Interaction of Language

and Social Setting
Dell Hymes
University of Pennsylvania

Diversity of speech, within the community and within the indi-

vidual, presents itself as a problem in many sectors of life-in education,
in national development, in transcultural communication. When those
concerned with such problems seek scientific cooperation, expecting
to find a body of systematic knowledge and theory, they must often be
disappointed. Practical concern outpaces scientifk competence.
The questions which arise from diversity of speech are questions
addressed to an understanding of the functional roles of languages.
They take for granted a world in which communities have a plurality
of languages (or code-varieties) and in which languages have a plural-
ity of roles, the two, codes and roles, often being related in complex
and distinctive ways. In expecting to find a scientific theory of such
interaction of language and social setting, one in effect expects theory
based on successfully asking (at least as a start), what code is used,
where and when, among whom, for what purpose and with what result,
to say what, in what way; subject to what norms of interaction and of
interpretation; as instances of what speech acts and genres of speaking?
How do community and personal beliefs, values and practices impinge
upon the use of language, and upon the acquisition of such use by
children?
No such body of systematic knowledge and theory as yet exists.
There is not even agreement on a mode of descriptive analysis of lan-
p a g e in interaction with social setting, one which, being explicit and
8
 INTERACITON OF LANGUAGE 9

of standard form, could ensure development of knowledge and theory

through studies that are full and comparable.

The Phenomenon of Bilingualism

The phenomenon of bilingualism has been the main focus of such
interest as has been shown, both in its own right and as medium of
linguistic diffusion. (Dialectology in the United States until recently
was mostly an abstraction of language from interaction with its im-
mediate social setting, having primarily a geographical and reconstruc-
tive orientation of little relevance to contemporary American society.)
Thus, the signscant work in American linguistics during the 1950's
which provides some sociolinguistic orientation to language deals with
the description of bilingualism ( Weinreich, 1953; Haugen, 1953,1956).
Yet, bilingualism, it must be said, is not in itself an adequate basis
for a model (or theory) of the interaction of language and social set-
ting. From the standpoint of such a model or theory, bilingualism is
neither unitary as a phenomenon, nor autonomous. The fact that two
distinct languages are present in a community or in a person's com-
municative competence is compatible with, and may depend on, a
great variety of underlying functional relationships. Conversely, it is
not necessary that two distinct languages be present for the underlying
functional relationships to appear.
Cases of bilingualism par excellence (as for example French
and English in Canada, Welsh and English in Wales, Russian and
French among pre-revolutionary Russian nobility) are salient, special
cases of the general phenomena of variety in code repertoire and
switching among codes. No normal person, and no normal community,
is limited in repertoire to a single variety of code, to an unchanging
monotony which would preclude the possibility of indicating respect,
insolence, mock-seriousness, humor, role-distance, etc. by switching
from one code variety to another.
Given the universality of code repertoires and of code-switching,
then it does not appear decisive that the code-varieties be distinct lan-
guages (bilingualism par excellence). Relationships of social intimacy
or of social distance may be signalled by switching between distinct
languages (Spanish : Guarani in Paraguay), between varieties of a
single language (Standard German : dialect), or between a pair of
pronouns within a single variety ( t :~ uous). Segregation of reli ious
P
activity may be marked linguistically by a language not general y in-
telligible because it is of foreign provenance (e.g., Latin, Arabic), or
because it is a lexically marked variety of the common language
(Zuni), Conversely, shift between codes may mark a shift between
wholly distinct spheres of relationship and activity (Standard Norwe-
gian : Hemnes dialect), or it may simply mark the formal status of talk
within a single integral activity (e.g., S h e in New Guinea),
10 DELL HYMES

If the community’s own theory of code repertoire and code-switch-

ing is considered, as it should be in any serious descriptive approach,
matters become even more complex and interesting. Among the Ameri-
can Indian peoples, such as the Wishram Chinook of the Columbia
River in the state of Washington, it was believed that infants knew
first a special language shared with certain guardian spirts and in-
terpretable only by men having those spirits; the native language,
Wishram, was in native theory a second language to everyone. Further-
more, one pair of communities may strain to maintain mutual intelli-
gibility in the face of great differences in dialect, while another pair
of communities may refuse to maintain intellegiblity although the
differences seem minor. In native theory, then, cases indistinguishable
by objective criteria of linguistic differentiation may be now mono-
lingual, now bilingual, depending on local social relationships and
attitudes.
Finally, while it is common to look for specialization in the function,
elaboration and value, of a language in a bilingual situation, such
specialization is merely an aspect of a universal phenomenon that must
be examined in situations dominantly monolingual as well. In doing
so it must be borne in mind that language is not everywhere equivalent
in communicative role and social value; speaking may carry different
functional loads within the communicative economies of different
societies. One type of hunting and gathering society, the Paliyans of
South India, “communicate very little at all times and become almost
silent by the age of 40. Verbal, communicative persons are regarded
as abnormal and often as offensive” (Gardner, 1966, 398). Thus the
role of language in thought and culture (Whorf‘s question) cannot
be assessed for bilinguals until the functional role of each of their
languages is assessed; the same is true for monolinguals, since in dif-
ferent societies language may enter differentially into educational ex-
perience, the transmission of beliefs, knowledge, values, practices, the
conduct of life (Hymes, 1966b).
What is needed, then, is a general theory and body of knowledge
within which code-switching and diversity of code repertoire could
find a natural place, and within which salient bilingualism could be
properly assessed. Little of such a theory and body of knowledge now
exists partly because the social scientists asking the right sort of func-
tional questions have not had the linguistic training and insight to deal
with the linguistic face of the problem, and because linguistics, the
discipline central to the study of speech, has been occupied almost
wholly with analysis of the structure of language as a referential code.
In defining such structure as object of study, linguists have tended to
dismiss or ignore the functional role. Sometimes as a matter of simpli-
fying assumption, sometimes as a matter of principle, linguistic theory
has been almost exclusively concerned with the nature of a single
 INTERACXION OF LANGUAGE 11

homogeneous code, shared by a single homogeneous community of

users, and (by implication) used in a single function, that of referential
statement. There have been notable exceptions to such a view, es-
pecially among linguists of Prague and London, but only very recently
has there emerged something tantamount to a movement to redress the
situation. This movement has come to be called sociolinguistics, es-
pecially when it focuses attention upon language proper in relation to
sociological categories, or ethnography of communication, where there
is focus upon verbal art, native taxonomy of speech types and functions,
and other features more typically studied by anthropologists ( Whiteley,
1966). In point of fact, an adequate study of language in interaction
with social setting will enlist scholars from all the social sciences in a
common enterprise. Throughout this article I shall use sociolinguistics,
intending by it the name of a problem area mediating among disci-
plines. Ultimately such a term may become redundant, if linguistics
comes to accept the sociocultural dimensions of its subject-matter and
its theoretical bases; one might then speak simply of linguistics
(Hymes, 1964a, 1966a, ms. a.).

The Case for Sociolinguistic Description

For some of the most brilliant workers on the interaction of lan-
guage and social setting, a general theory and body of knowledge is to
be achieved by selecting problems that contribute directly to present
linguistic theory and social theory. The mode of progress is direct
action: use of multiple working hypotheses and strong constraints on
relevance and verification in quite particular problems, intended to
satisfy adherents of traditional linguistic theory and social theory.
Studies in exotic societies are not particularly valued, since strong
control over data and hypothesis-testing cannot be maintained. Infor-
mation of the sort given (most often incidentally) in reports from other
societies is not found to be convincing. A concern to secure reports
focused on sociolinguistic information from such societies is thought
pointless, since it suggests a prospect of endless descriptions that, what-
ever their quality and quantity, would not as such contribute to present
theory.
My own view is different. I accept that intellectual tradition which
since the eighteenth century has sought to understand the unity of
mankind through both its ethnographic diversity and its general evolu-
tion, In that tradition a theory, whatever its logic and insight, is ulti-
mately unsatisfying if divorced from the natural and existential world
of mankind as a whole. The concern is consonant with that of Kroeber,
reflecting upon Darwin:
, . . anthropologists . . . do not yet clearly recognize the fun&-
mental value of the humble but indispensable task of classifying-that
12 DELL HYMES

is, structuring-our body of knowledge, as biologists did begin to recognize

it two hundred years ago ( 1960,14).
One recognizes that communities differ significantly in patterns of
code-repertoire, code-switching, and, generally, in the roles assigned to
language. Ethnographic reports indicate differences with regard to
beliefs, values, reference groups, and the like, as these impinge on the
on-going system of language use and its acquisition by children, Since
there is at present no systematic understanding of the ways in which
communities differ in these respects, we need one. We need, in effect,
a taxonomy of sociolinguistic systems.
From this standpoint, each of a variety of diverse cases may be
felt to be of value in its own right, as an expression of mankind. In
any case, such instances are valued as enlarging and deepening insight.
We require a widely ranging series of sociolinguistic descriptions be-
cause a particular model, let alone an integrating theory, is not con-
vincing unless it has met the test of many diverse situations, of a mass
of systematic data. (Recall that Darwin’s exposition of evolution was
convincing for such a reason.) A taxonomy and a descriptive model
are joint conditions of success.
Information from exotic societies, analyzed with the goals of taxon-
omy and descriptive models in mind, is in fact interdependent with
detailed work in one’s own society. Each may provide insight and a
test of significance for the other. Thus it has been suggested that there
is only a class-linked British relevance to Bernstein’s sociological model
of elaborated vs. restricted forms of code, governed by personal vs.
positional types of social control. [Elaborated codes are largely now-
coding, adaptive in lexicon and syntax to the ad hoc elaboration of
subjective intent, whereas restricted codes are largely then-coding,
adaptive to the reinforcement of group solidarity through conventional
expression. Personal social control appeals to individual characteristics,
role discretion, and motivation; positional social control bases itself
on membership in categories of age, sex, status, etc.]

From the Standpoint of Taxonomy

From the standpoint of taxonomy, the model takes on a new di-
mension. It is found to be a valuable set of polar ideal types, applicable
to the comparison of whole societies as communicative systems (see
the description of Arapesh and Manus by Mead, 1937), and suggestive
of new hypotheses linking socialization and adult religious experience.
Among the Hopi and Zuni of the American Southwest, for instance,
Severe socialization pressure is initiated at about two years of age,
before the child can have reasons verbally explained, and thus is
necessarily experienced as positional control. Among the Wishram
Chinook socialization pressure is withheId until the child can talk and
have reasons verbally exphined; the native view of socialization is
 INTERACTION OF LANGUAGE 13

explicitly one of personal control through verbal means. Adult religious

activity among the Hopi and Zuni is dominated by positional relation-
ship ( clan membership, etc. ) ,and its verbal aspect is highly prescribed.
Among the Wishram it is dominantly unique to each individual; the
verbal aspect is private between person and guardian spirit, and in-
terpreted by the person according to his own life experience.
Furthermore, I believe that failure to postulate a model and taxon-
omy of sociolinguistic systems as a goal will perpetuate a long-stand-
ing, unsatisfactory state of affairs, namely, the failure of scientifk
study to address itself to the unity of language and social life. This
unity is rooted in the use of language in social life, in the integrity of
the message as an act. Because of the common divorce of the study of
language, as grammar, from the study of society, the unity does not
come into view, Each of the separate specialisms abstracts from the
speech act its own aspect for its own purposes. A theory of langua e
in society, when envisaged, is usually thought of as uniting the resu ts
of such separate enterprises, institution-free grammar and grammar-
f
free institutions. But these enterprises, having made their abstractions
in quite disparate frames of reference, and never having been respon-
sible for the study of speech acts as such, are quite incapable of sup-
porting the act of reintegration. What from the standpoint of the
actors and the community is an integral act, motivated and subject to
shared rules of interpretation, remains invisible. It is both less and
more than it was: less, because it lacks its own form and motivation;
more, because having been dismembered according to conflicting
claims of jurisdiction by specialisms, each concerned to gerrymander
speech to its own taste, the parts to be fitted now overlap. The act,
still lifeless, has grown grotesque. All approaches in which the relation
between language and social life is regarded wholly as a matter of
correlation, or of variation, are vitiated by the implicit assumption
that integration is a matter of post-hoc putting together of separate
results, none obtained with the integral object in view.
In short, there must be a study of speaking that seeks to determine
the native system and theory of speaking; whose aim is to describe the
communicative competence that enables a member of the community
to know when to speak and when to remain silent, which code to use,
when, where and to whom, etc. (This view is an application to speak-
ing of the general view of ethnography as the construction of descrip-
tive theory that has been elaborated by Goodenough, Conklin, Frake,
Sturtevant, myself and others; for its application to speaking, cf. Hymes
1962, 1964b, 1964c, and ms. b.)
In considering what form sociolinguistic description might take,
and what form an integrated theory of such description might take,
one needs to show sociologists, linguists, ethnographers and others a
way to see data as the interaction of language and social setting, The
need for this is clear from the frequency with which researchers have
14 DELL HYMES

had informal field experience of great sociolinguistic interest, but,

lacking precedent and format for its presentation, have let the infor-
mation lie fallow as at best a matter for anecdotes.
Only a specific, formal mode of description can guarantee the con-
tinuation of the present interest in sociolinguistics. Such interest is
sustained more by fashion and practical issues, perhaps, than by scien-
tific conviction and accomplishment, It was the development of a
specific mode of description that ensured the success of linguistics as
an independent discipline in the United States in the twentieth century,
and the lack of it that led to the decline and peripheral status of folk-
lore, both having started from a similar base, i.e., the interlocking in-
terest of anthropologists and humanistic scholars in language, in the
one case, verbal tradition, in the other.
Such a goal is of concern to practical and applied work as well as
to scientific theory. When a problem of bilingualism is to be studied,
for example, the componets of speaking that are taken into account will
depend upon a model, implicit if not explicit, of the interaction of
language with social setting. The significance attached to what is
found will vary with the understanding of what is possible, what uni-
versal, what rare, what linked, in a comparative perspective. What
survey-researchers need to know linguistically about a community, in
selecting the code of questioning, and in conducting questioning, is
in effect an application of the community’s sociolinguistic description.
In turn,practical work, if conducted with the needs of taxonomy and
theory in mind, can make a special contribution, for it must deal
directly with the interaction of language and social setting, and so pro-
vide for a test of the relation between theory and practice.
The goal of sociolinguistic description can be put in terms of the
present situation in the disciplines whose interests converge in socio-
linguistics. Whatever his questions about language, a linguist is clear
that there is an enterprise, description of languages, which is central
and prerequisite. Whatever his questions about society and culture,
a sociologist or an anthropologist is clear that there is an enterprise
(whether called ethnography, social sbucture, social organization )
that is concerned with the concepts and methods prerequisite to par-
ticular studies and answers, a system that provides a coherent, general
guide to inquiry. In other words, such workers understand what it
means to describe a language, a social system or the culture of a com-
munity. We need to be able to say the same sort of thing, i.e., what it
means to describe a sociolinguisticsystem.

Toward a Descriptive Theory ...

Sociolinguistic systems may be considered at the level of national
states, and indeed, of the emerging world society. The concern of this
 INTERACTION OF LANGUAGE 15

paper is with sociolinguistic systems at the level of the speech econo-

mies of individual communities. The interaction of language with social
setting is viewed first of all as a matter of human action and of the
knowledge, sometimes conscious, sometimes unconscious, that enables
persons to use language in social life. Larger systems, it is true, may
have properties not reducible to those of the speaking competence of
persons, just as the world economy has properties not reducible to
those of the economies of nations, communities or persons. Such com-
petence, however, underlies communicative conduct, not only within
communities, but in encounters between them. Whenever a larger
system is dependent upon communication among persons, then the
point of departure is persons. The speaking competence of persons
may be seen as entering into a series of systems of encounters at levels
of different scope. The considerations to be advanced here apply in
principle to analysis of any social relationship in which norms of com-
municative conduct entailing speech have arisen. The examples will
typically be from the analysis of relationships characteristic of individ-
ual communities.
An adequate descriptive theory would provide for the analysis of
individual communities by specifying the technical terms required for
such analysis, and by specifying what form the analysis should take.
That form would be formal (i.e., would deal with the actual forms of
speaking in a wholly explicit way) and standard (in the sense of being
subject to general constraints on order, interrelationship and the
character of rules). However, only extended empirical work, and ex-
tended experimentation with alternative modes of statement, can show
what form of descriptive theory is to be preferred. When achieved,
such a theory, by providing for the explicit, standard analysis of indi-
vidual systems will at the same time provide a theory of their universal
features.
Some Notions w i t h which the Theory M u s t Deal
Among the notions with which such a theory must deal are those
of speech community, speech situation, speech act, speech event, fluent
speaker, native speaker, factors (or components) of speech events,
functions of speech, rules of speaking, types of speech event and act.
I have sketched a partial approach to such notions, first in “The ethnog-
raphy of speaking” (1964a, 33-44; 196413). I shall not repeat
the ways in which the approach has already been developed, but show
how it has been modified in the course of recent work. It must be em-
phasized that the discussion at this stage is of a heuristic guide to the
anaIysis of systems, and that an eventual theory will have properties
that can emerge only from the results of such analyses.
The recent work has been aimed at analyzing ethnographic data
so as to provide at least a preliminary taxonomy of the variety of socio-
16 DELL HYMES

linguistic systems that impinge upon education and the child. In

socialization a child acquires not only language(s), but also sets of
attitudes and habits with regard to the value and utilization of lan-
guage (s). A child capable of any and all grammatical utterances, but
not knowing which to use, not knowing even when to talk and when
to stop, would be a cultural monstrosity. Often enough a child is con-
fronted not only with more than one code, but also with more than
one system for the use of the codes. There may be what linguists have
come to call interference (Weinreich, 1953) not only between two sets
of code habits, but also between two sets of habits for the use of codes.
In educational situations made complex by diversity of speech, then,
whether saliently bilingual or not, one needs to understand the general
patterns of communicative competence being acquired by children
as background for understanding the outcome of the small fraction of
communicative experience encountered by children in school, A com-
parative perspective may help one to understand problems of a par-
ticular case.
In the course of this work a guide to the analysis of socialization,
focused on speaking, has been prepared. (Existing guides neglect
speech as a variable.) In one sense, such work focuses on the acquisi-
tion of a pre-existing system of interaction of language with social set-
ting. In a deeper sense, such work studies the whole system as viewed
from the standpoint of the child.
The guide is organized from the more general to the more specific,
and is consciously designed to present the acquisition of linguistic
codes as but a part of the acquisition of communicative competence as
a whole. Its outlines are:

A Guide to Analysis of Speech Socialization

( A . ) General Aspects of Socialization
I. Life cycle (the ways in which speed and language enter into the
distinguishing and accomplishing of reference points in the life
cycle )
11. Learning and teaching (the place of language and speaking in
native conceptions of acquisition of culture and of modes of teach-
ing )
111. Social control (the relative place of verbal means and verbal
explanations )
(B. ) Competence in Speaking
IV. Speaking competence (general attitudes toward speaking in rela-
tion to valued types of person, satisfactions, normal demeanor; the
system of speaking as something in which competence is ac-
quired and evaluated; conceptions of such competence, and its
place among other modes of communicative competence)
 INTERACTION OF LANGUAGE 17
V. Linguistic code competence (general attitudes toward knowledge of
linguistic codes in relation to valued types of person, satisfactions,
conduct; the repertoire of codes in the community, their uses, their
order and mode of acquisition; conceptions of competence in lin-
guistic codes, and their place among other communicative codes)
( C .) Processes of Acquisition
VI. Communicative environment of the infant (what communicative
behavior is directed at the infant, in what ways its behavior is in-
terpreted as communicative, differential response to its use of
communicative modalities )
VII. Acquisition of speaking competence (conceptions of children's first
speech acts, what speaking is directed at children, how their speech
is responded to, conceptions of sequence of acquisition of speaking
competence, of how competence comes about, what is done)
VIII. Acquisition of linguistic code competence (conceptions of first
words, of sequence of acquisition of code(s), of how it comes
about, what is done)
( D. ) Generalizations, Typological Contrasts
The detailed contents of the guide, and the practical procedures
for their application to data, must be passed over here. What is directly
pertinent is the effect of applying the guide on that part of the guide
which sketches the analysis of sociolinguistic systems. As work con-
tinues, feedback from data continues too; but certain general formu-
lations have been shown to be necessary and have remained stable. To
these I now turn after a brief comment on the background of such
efforts.
Most general treatments of language, speech, rhetoric, literature
and some treatments of other topics, make assumptions, explicit or im-
plicit, as to notions with which a descriptive theory of speaking must
deal. With particular regard to the components of speech events and
the functions served in them, there have been important classifications
by Karl Biihler, Kenneth Burke, Roman Jakobson, Charles Morris, C.
K. Ogden and I. A. Richards and others. With particular regard to
code repertoire in relation to social setting there are important classi-
fications and findings in the work of Basil Bernstein (see above),
Joshua A. Fishman, H. A. Gleason, Jr., John J. Gumperz, Michael
Halliday, William Labov, and others. These studies provide terms and
notions that may prove quite useful, and the same is true of studies of
other aspects of the field of sociolinguistic description: there is much
to be gained heuristically from them. It would not be to the point,
however, to review such studies here with the object of picking among
them, amalgamating them, or the like. While each suggests a schema
that may prove useful in part or in whole, the fundamental problem-
to discover the underlying communicative competence that enables
members of a community to use and interpret the use of language, and
18 DELL HYMES

to provide a formal description that is a theory of that competence-

cuts deeper than any schema so far proposed.
Concern must now be with terms as heuristic input to descriptive
analyses, but just as a theory of language structure must have its uni-
versal terms (e.g., sentence, distinctive feature), SO must a theory of
language use. At least some of the terms now to be discussed will no
doubt survive empirical revisions and permanently remain as part of
the theory.

Toward a Descriptive Theory . . . Social Units

One must first consider the social unit of analysis. For this I adopt
the common expression:
Speech Community. Speech is here taken as a surrogate for all
forms of language, including writing. The necessity and primacy of
the term, speech community, is that it postulates the basis of descrip-
tion as a social, rather than a linguistic, entity. That is, one does not
start with a code and look afterward to its context. One starts with a
social group and looks within it at the codes present.
Bloomfield (1933) and others have in the past reduced the notion
of speech community to the notion of language. Those speaking the
Same language (or fist language, or standard language) were defined
as belonging to the same speech community, and this confusion still
persists in much social science literature, a quantitative measure of
frequency of interaction sometimes being appended. The present
approach requires a definition that is qualitative and expressed in
terms of the use of language. Tentatively, a speech community is de-
fined as a community sharing both rules for the conduct and interpre-
tation of acts of speech, and rules for the interpretation of at least one
common linguistic code. The sharing of code rules is not su5cient:
there are persons whose English I can interpret, but whose message
escapes me. Nor is the sharing of speech rules su5cient: such sharing
may characterize a speech area (Sprechbund-I owe the term to J.
Neustupny ) , comprising several distinct speech communities, not
necessarily sharing a common language, but agreeing in patterns of
speaking.
The speech field (akin to the notion of social field) can be defined
as the total range of communities within which a person’s knowledge
of code and speaking rules enables him to move. Within the speech
field must be distinguished the speech network, the specific linkages
of persons through code and speech rules across communities. To il-
lustrate: one’s speech community may be, effectively, a single city or
portion of it; one’s speech field will be delimited by one’s repertoire of
codes, sometimes by a single language (say to England, Canada,
Australia, and the United States, given a command of English), some-
 INTERACTION OF LANGUAGE 19

times not; one’s speech network, based for example on the practice of
a common profession (say, sociology), may extend across communities,
the common profession providing sufficiently common rules of speech.
Obviously part of the work of definition is borne by the commu-
nity, and the difficulties of defining it are here by-passed. Criteria of
frequency of interaction, regularity of interaction, focus of interaction,
contiguity, degree of commonalty of pattern, etc., may be invoked,
each perhaps representing a continuum. The essential thing is that the
object of description is an integral social unit. Probably it will be use-
ful to reserve the notion of speech community for the social unit most
specifically characterized for a person by common locality and primary
interaction (Gumperz, 1962, esp. 30-32). I have essentially drawn dis-
tinctions of scale and kind of linkage within Gumperz’ wholly general
concept of linguistic community which amounts to any distinguishable
intercommunicating group.
Speech Situation. Within a community one readily detects many
situations associated with (or masked by the absence of) speech. Such
contexts of situation will often be naturally described as ceremonies,
fights, hunts, meals, love-making and the like. It would not be profitable
to convert such situations en masse into parts of a sociolinguistic de-
scription, by the simple expedient of relabelling them in terms of
speech. (Notice that the distinctions made with regard to speech com-
munity are not identical with the concepts of a general communicative
approach, which must note the differential range of communication by
speech, film, art object, music.) Such situations may enter as contexts
into the statement of rules of speaking as aspects of setting (or of
genre). In contrast to speech events, they are not in themselves gov-
erned by such rules, or one set of such rules throughout. A hunt, for
example, may comprise both verbal and nonverbal events, and the
verbal events may be of more than one type.
In a sociolinguistic description, then, it is necessary to deal with
activities which are in some recognizable way bounded or integral.
From the standpoint of general social description they may be regis-
tered as ceremonies, fishing trips, and the like; from particular stand-
points they may be regarded as political, aesthetic, etc., situations,
which serve as contexts for the manifestation of political, aesthetic,
etc., activity. From the sociolinguistic standpoint they may be regarded
as speech situations.
Speech Event. The term speech event will be restricted to activi-
ties, or aspects of activities, that are directly governed by rules for the
use of speech. An event may consist of a single speech act, but will
often comprise several. Just as an occurrence of a noun may at the
same time be the whole of a noun phrase and the whole of a sentence
(e.g., Fire!), so a speech act may be the whole of a speech event, and of
a speech situation (say, a rite consisting of a single prayer, itself a
20 DELL HYMES

single invocation). More often, however, one will find a difference in

magnitude: a party (speech situation), a conversation during the party
(speech event), a joke within the conversation (speech act). It is of
speech events and speech acts that one writes formal rules for their
occurrence and characteristics. Notice that the same type of speech act
may recur in different types of speech event, and the same type of
speech event in different contexts of situation. Thus, a joke (speech
act) may be embedded in a private conversation, a lecture, a formal
introduction. A private conversation may occur in the context of a
party, a memorial service, a pause in changing sides in a tennis match.
Speech Act. The speech act is the minimal term of the set being dis-
cussed, as the remarks on speech events have shown. The work on
speech acts inspired by British philosophers such as the late J. L.
Austin provides many helpful indications of the types of speech acts
and the relationships among them. It also contributes to the task of
determining for English the membership and meaning of native sets
of terms for speech acts.

Toward a Descriptive Theory . . . Components

of Speech
In discovering the native system of speaking, certain familiar guide-
lines may be mentioned. One must determine the native taxonomy of
terms as an essential although never perfect guide. A shift in any of
the components of speaking may mark the operation of a rule (e.g.,
from normal to another tone of voice, from one code to another). Cor-
rection, embarassment, withdrawal and other negative responses may
indicate the violation of a rule. There may also be positive evaluation
(more in some groups than others) of effective use of speech and its
rules.
One must have in addition some schema for the components of
speech events. Traditional in our culture is the three-fold one of
speaker, hearer and thing spoken about. That has been elaborated
upon in various ways, e.g., in information theory. Work with ethno-
graphic data, however, has shown the necessity of a somewhat more
detailed schema. The constraints on such a heuristic guide are that it
should be ample enough to handle data without arbitrariness, yet com-
pact enough to be kept in mind for use. Being heuristic, rather than at
present a theory, there is no harm and a definite advantage in organiz-
ing the schema as a mnemonic device. It so happens that in English
the letters of the term SPEAKING itself can be used rather naturally
for this purpose. (That the analysis of components is not language-
bound, determined by the accidents of spelling, can be shown by the
possibility of alternative keywords, were the spelling different. Thus
what below is I[nstrumentalities] could as well be M[eans], A[gen-
 INTERACTION OF LANGUAGE 21

cies]; what below is K[ey] could as well be T[one], , M[anner], W[ay],

H[ow]; etc. Changes of terminology, rather than exact translation,
may permit an analogous mnemonic device in other languages. (In
French, for example, PARLANT could be adapted to the purpose, as
will be shown below. ) )
The criterion for registering a component is that it should be part
of the definition of a rule of speaking. Rules of speaking, in other words,
entail structured relationships among two or more components.
Organized in terms of the English code-word, components are:
( S ) Setting, or Scene. By setting is intended of course time, and
place, of a speech event. In addition, psychological setting, and cul-
tural definition of the setting as a certain type of scene, may be im-
plicated here, as when, within a play on the same stage, during the
same performance, the dramatic time or place shifts: “ten years later”,
“a battlefield in France”. The types of scene defined by a society may
be basic to an analysis of speech events and the role of speaking. For
example, among the Subanun of the Philippines, described by Charles
0. Frake (ms.), there is a basic division into festive and nonfestive
scenes. Thus, the character of Subanun litigation derives from its oc-
currence in festive scenes and the verbal art appropriate to them. A fre-
quent type of rule is one in which a form of speech act is dependent
on an appropriate scene; of equal importance is the use of speech acts
or the choice of code to define scenes as appropriate.
(P) Participants or Personnel. Schemes of components usually dis-
tinguish Speaker and Hearer (Sender and Receiver, Addressor and Ad-
dressee). From the standpoint of explicit rules for speaking, such a
categorization is at once too specific and too imprecise. It is too specific
in that some rules hold for a participant independent of his role as
speaker or hearer. Thus, in conversations among the Abipon of Argen-
tina, if a participant (speaker or listener) is a member of the Hocheri
(warrior class), then -in is added to the end of every word. It is too
imprecise, because societies commonly differentiate a variety of roles
for participants in speech events, and these must be specified. The im-
portance of the category of auditor or audience, as a constraint on rules
of speaking has recently been emphasized by the sociologist, Allen D.
Grimshaw. Among the Wishram Chinook, formal speech events are
defined by the relationship between sender, or source (e.g., a chief),
a repeater of the sender’s words, and an audience constituted as a pub-
lic; in the major Wishram speech event the addressees at crucial points
are not the audience, but the spirits of the surrounding environment.
It is typically in their definitions of the presence or absence of
participants in speech events (more generally, communicative events)
that societies most differ. Much of religious behavior can be viewed as
application of a native theory of communication, often associated with
elaboration of a specific code and code-switching.
22 DELL H Y M E S

( E ) Ends. Here an English homonymy is exploited, two types of

ends being meant: ends in view (goals, purposes), and ends as out-
comes. In one sense, intentions and effects; in another, manifest and
latent functions, Previous schemata of speech events have most often
not provided a place for intention and outcome. (Kenneth Burke, 1945)
is the exception, perhaps because of an unconscious behaviorism. Anal-
ysis of speech events from several societies shows the category of
purpose and that of outcome to be crucial to the distinguishing of
varieties of event. Among the Waiwai of Venezuela, for example, the
varieties of the central speech event, the oho-chant, are to be dis-
tinguished as to whether the purpose is a marriage-contract, a trade,
a communal work-task, an invitation to a feast, or a composing of social
peace after death. Rules for participants and settings vary accordingly.
Among the Yakan of the Philippines a taxonomy of four levels of event
focused upon speech is to be differentiated in terms of purpose and
outcome. Interpreted in a linguistic mode of statement, one has:
( a ) [Focus] [talk about a topic]
/ miting
[Outcomel + [no special outcome1
That is, the type of speech event called miting has as its focus simply
talk about some topic; no special outcome is expected.
( b ) [Focusl [talk about an issue1
/ qisun (“conference”)
[Outcomel [decision]
That is, the type of speech event called qisun has as its purpose simply
talk about something regarded as an issue, as when to plant rich, when to
take a trip, and a decision is expected as the outcome.
( c ) [Focus] [talk about a disagreement] mawpakkat
[Outcomel * [settlement] 1 ( negotiation)
That is, the type of speech event called mawpakkut has as its purpose
talk about a disagreement involving conflicting interests, and as its ex-
pected outcome, a legally binding resolution, or settlement.
( d ) [Focus]
[~utcomel * [talk about a dispute1
[ruling]
, hukum (litigation)

That is, the type of speech event called hukum has as its focus a dis-
agreement arising from a charge that an offense has been committed, and
as its expected outcome, a legal ruling, based on precedent and carrying
special sanctions. The Yakan examples are from a paper by Frake, Struck
by Speech, to appear in a volume on the ethnography of communication
edited by John Gumperz and myself.
As the varied wording of the Yakan account has shown, where the
focus of a speech event requires special attention, it seems most natu-
rally to be an aspect of this portion of the heuristic scheme. Contrast
in focus is often important for comparative study. Thus litigation
 INTFJUCI’ION OF LANGUAGE 23

among the Subanun (Philippines) has special focus on message-form,

in the elaboration of verbal art, in keeping with its occurrence among
Subanun festive scenes; whereas Yakan (Philippines) litigation has
focus only on topical content, in keeping with its place among Yakan
informal scenes.
(A) Art Characteristics. Here two closely linked aspects of acts
of speech are grouped together: the form, and the content, of what
is said. The technical terms message-form and topic, respectively, are
adopted for these. One context for the distinction is in the reporting of
speech events: “He prayed, saying ‘. . . .’” (preserving message-form)
vs. “He prayed that he would get well” (preserving topic only).
Perhaps the gravest and most common defect in most reports of
speech events is that it is impossible to recapture the rules for message-
form. Without such rules, however, it is impossible to characterize the
nature of the competence in speaking of members of the society. Com-
monly one reads that a certain use of language is important, even cru-
cial to a society-gossip, for example, among the Makah Indians of
northwestern Washington or among fox-hunting English aristocrats.
If one does not know how to gossip correctly, one cannot be an ade-
quate member of the group. It must, then, be possible to say of an act
of speech that it does or does not fit the rules. Possibly the rules for
gossip are defined entirely in terms of participants, topics and settings,
and not in terms of message-form; but it seems far more likely that
some forms of presenting the gossip-content are acceptable and some
not. In any case it is certain that where there are genres of speech act,
such as gossip, there is differential skill in their accomplishment, and
such skill will include handling of the message-form.
A concern for the details of actual form strikes some as picayune
and removed from humanistic and scientific importance. Such a view
betrays an impatience that is a disservice both to humanistic and
scientific purposes. It is precisely the failure to unite form and content
in the scope of a single focus of study that has held back scientih
understanding of the fundamental human skill, speaking and vitiates
so many quasi-scientific attempts to prove the significance of expres-
sive behavior through content categories alone. One can never pre-
scribe in advance the size of signal that will be crucial to content and
skill in a communicative genre. The more the genre has become a
shared, meaningful expression within a group, the more likely that the
crucial cues will be efficient, that is, slight in scale. If one balks at such
detail, perhaps because it requires technical skills in linguistics, musi-
cology or the like, one should face the fact that one is simply refusing
to take seriously the human meaning of one’s object of study and the
scientific claims of one’s field of inquiry.
A further consideration is that such genres which become shared,
meaningful expressions within a group acquire a partial autonomy, an
24 DELL HYMES

inner logic of their means of expression, that conditions and sometimes

even controls their content. For members of the group, then, “freedom
is the recognition of necessity”; mastery of the detail and formal logic
of the genre is prerequisite to personal expression. Again serious con-
cern for the human meaning of such genres requires analysis that goes
beyond gross content to precise, explicit statement of the rules and
features of the form.
While such a perspective may seem to apply first of all to genres
recognized as conventionally aesthetic, it applies as well to conversa-
tion in daily life, Only the most painstaking analysis of form (similar
to that of literary criticism) can reveal the fantastic depth and ade-
quacy of the elliptical art that is talk.
( K ) Key. This component is introduced to distinguish the tone,
manner or spirit in which an act is done. Acts otherwise the same as re-
gards setting, participants, message-form and content, may differ in
key, as between mock :serious; perfunctory :painstaking; and the like.
The communicative significance of key is underlined by the view
that, where the two are in conflict, the manner of an act overrides the
content in determining its true significance. The signalling of key may
sometimes be a part of the message-form itself, but may be nonverbal
such as a wink, gesture, attire, musical accompaniment.
( I ) Instrumentalities. Here are grouped together two closely
linked components, those of Channel and Code. By choice of Channel
is understood the choice of oral, written, telegraphic, semaphore or
other medium of transmission. By choice of Code is understood a
choice at the level of distinct languages. Where the distinction is
necessary, varieties within a language may be designated subcodes.
For the student of bilingualism, of course, rules linking choice of
Code (or subcode) with the other components are of primary interest.
Each component seems to covary with choice of code in some case or
other: setting, participants, ends and outcomes, message-form and
topic, key, channel and (to be cited below) norms of interaction and
interpretation, and genres.
( N ) Norms of Interaction and of Interpretation. By Norms is
meant not the normative character that may attach to all rules for
choice among components, but specific behaviors and proprieties that
may accompany acts of speech-that one must not interrupt, for ex-
ample; that normal voice must not be used except when scheduled
(e.g., in church service). Here, too, may be considered shared rules for
the understanding of what occurs in speech acts, e.g., as to what can
be ignored or discounted.
In a thoroughgoing analysis of a community, the notion of norms
of interaction would implicate the social structure-the members’ cate-
gories of kinds of person (role, status and the like), and the norms of
interaction obtaining between them. Analysis of these norms would be
 INTERACTION OF LANGUAGE 25
prerequisite to adequate statement of rules governing modes of ad-
dress and the symbolic import of other choices, such as choice of code
( see discussion of formal rules below).
The notion of norms of interpretation implicates the belief system
of a community. In the history of ethnographic analysis of language,
the classic precedent is the treatment of symbolic meanings of elements
of Trobrian magical formulae and ritual by Malinowski (1935),under
the heading of Dogmatic Context. Malinowski’s other rubrics for
analysis are roughly related to those presented here in the following
way: Sociological Context and Ritual Context subsume information as
to setting, participants, ends in view and outcome, norms of interaction,
higher-level aspects of genre; Structure reports salient patterning of
the verbal form of the act or event; Mode of recitation reports salient
characteristics of the vocal aspect of message-form.
( G ) Genres. By Genres are meant categories or types of speech
act and speech event: conversation, curse, blessing, prayer, lecture,
imprecation, sales pitch, etc.
[With reference to French, the heuristic set of components might
be presented in terms of PARLANT: ( P ) Participants; ( A ) Actes
(form, content); ( R ) Raison, Resultat (= ends, outcomes); ( L ) Local
(= setting; the English adaptation of the French word, locale); (A)
Agents (channels, codes); ( N ) Normes; (T) Ton ( = Key); Types ( =
Genres) .]

Toward a Descriptive Theory . . . Formal Rules

Rules of speaking do not usually refer to all components of a
speech event, and often to as few as two or three. Choice of code may
be defined in terms of code and interlocutor alone; or code and topic
alone; or code, interlocutor, and setting; etc. It is necessary to dis-
tinguish the entire range because in a given case any one may be de-
fining, Moreover, a non-defining component may yet condition the
success or other aspect of the outcome of a speech event.
Many generalizations about rules of speaking will take the form of
statements of relationship among components. It is not yet clear that
there is any priority to be assigned to particular components in such
statements. So far as one can tell at present, any component may be
taken as starting point, and the others viewed in relation to it. When
individual societies have been well analyzed, hierarchies of precedence
among components will very likely appear and be found to differ
from case to case. Such differences in hierarchy of components will
then be an important part of the taxonomy of sociolinguistic systems.
For one group, rules of speaking will be heavily bound to setting; for
another primarily to participants; for a third, perhaps to topic.
Experimentation with the form of statement of rules of speaking
26 DELL HYMES

has not proceeded very far. Work of Joel Sherzer and myself with some
American Indian data suggests the possibility of adapting a linguistic
mode of statement. In such a format, generalizations applying through-
out a speech event are stated at the outset in a sort of lexicon. The
sequential form of the act itself is stated in a sort of syntax by means
of context-sensitive rewriting rules (Chomsky, 1965). When prose de-
scriptions of events have been so restated, there has been a consider-
able gain in understanding of structure. The explicit form of statement
makes demands upon description that go beyond what is usually in
prose accounts. The form of the event is disengaged, as it were, from
the verbal foliage obligatory in prose sentences, and can be more
readily seen. Such formal restatement is essential, if comparative work
is to proceed. One must be able to compare events within a society,
and across societies, in concise and standard format. Such comparison
cannot depend upon memorizing or shuffling of prose paragraphs
vastly different in verbal style. And it is through some formal restate-
ment that one can commit oneself to a precise claim as to what it is a
member of a society knows in knowing how to participate in a speech
act.
A grave defect in many studies which examine the interaction of
language and social setting has been the failure to state precisely ( a )
the difference and ( b ) the interrelationship between values pertaining
to the sociolinguistic feature, on the one hand, and the values pertain-
ing to the social context in which it can occur.
A related defect has been failure to state precisely the differince,
and the interrelationship, between the normal, ordinary, or “un-
marked value of a sociolinguistic feature, on the one hand, and the
“marked,” or specially loaded values, on the other. Studies of the use
of a given code in multilingual situations, like studies of modes of
address, may state the range of contexts in which the code can occur,
appending information about its use in each, but without contrasting
the effects of varying code and context. However, as the Uruguayan
linguist, J. P. Rona, has insisted, sociolinguistics deals not only with
linguistic facts in contexts, but with linguistic facts having social value
in contexts.
Just as the linguistic sign is a relation between a linguistic form
and a linguistic value (e.g., the form “dog” and meaning “dog”), so a
sociolinguistic feature, such as a choice of code, is a sign, a relation
between a form (here, the linguistic fact, such as a code) and a socio-
linguistic value (say, respect, or formality). The set of code-varieties
within a community may thus be analyzed as a semantic set. One may
determine for normal contexts the dimensions of meaning along which
choice of code-variety implies contrast. To complete the analysis, one
must then state separately the domains-settings, role relationships or
whatever-across which the normal meanings of code choice are de-
 INTERACTION OF LANGUAGE 27

fined. Having done so, one may now state the ways in which code
choice may be used to insult, to flatter, to boast-the marked, specially
loaded uses-by stating the corresponding, diflerent relations between
code values and domain values that govern these uses.
All these relationships can take the form of linguistic rules, such
as those developed by Chomsky (1965) for handling lexical elements.
In effect, one specifies form, content, and context-an overt element,
semantic values, and rules governing its selection in those values.
Such a form of statement is no more than an elaboration of the form of
statement familiar to us in dictionaries now, e.g., ?each (form), “a
tack sailed with the wind coming more or less from abeam” (semantic
value), with the specification (selection rule), “in nautical usage”. To
pursue such a form of statement in sociolinguistics,however, will have
as consequence the inseparability of sociolinguistic analysis from the
full-scale analysis of social life itself, for it is in the analysis of social
life that the requisite rules of selection for sociolinguistic features are
to be found and stated. (For a detailed example of this mode of
analysis, see my “Quasi-Korean Modes of Address”, submitted to
Anthropological Linguistics, with the Korean data on Richard Howell
( 1965), it can be shown that values of authority and intimacy assigned
to the modes of address (values often assigned to choice of code in some
societies) are in fact properties of the social relationships in terms of
which the use of the modes of address is to be defined. The modes of
address themselves form a set on a single dimension of social distance.
The formal separation of the set of linguistic choices and the set of
social relationships reveals the true nature of the relationship between
them. For major detailed work on Speech Communities from this
standpoint see Gumperz ( 1964) and Labov ( 1966).)
Such a mode of analysis permits formal treatment of many of the
functions served in acts of speech. The conventional means of many
such functions can indeed be analyzed as relations among components,
e.g., message-form, genre and key in the case of the -y form of the ac-
cusative plural of masculine nouns in Polish, which has the value
-solemn” in the genre of poetry, and the value “ironic, pejorative” in
the genres of non-poetic speech. Functions themselves may be statable
in terms of relations among components, such that poetic function, for
example, may require a certain relationship among choice of code,
choice of topic and message-form in a given period or society.
It would be misleading, however, to think that the definition of
functions can be reduced to or derived from other components. Such a
thought would be a disabling residue of behaviorist ideology. Ulti-
mately the functions served in speech must be derived directly from
the purposes and needs of human persons engaged in social action, and
are what they are: talking to seduce, to stay awake, to avoid a war.
The formal analysis of speaking is a means to the understanding of
28 DELL HYMES

human purposes and needs, and their satisfaction; it is an indispensable

means, but only a means, and not that understanding itself.
REFERENCES
BLOOMFIELD, LEONARD. Language. New York: Henry Holt, 1933.
BURKE,KENNETH. A Grammar of motiues. New York: Prentice-Hall, 1945.
CHOMSKY, NOAM.Aspects of the theory of syntar. Cambridge: M. I. T. Press,
1965.
GARDNER, PETERM. Symmetric respect and memorate knowledge: the structure
and ecology of individualistic culture. Southwestern Journal of Anthro-
pology 1966, 22, 389-415.
GUMPERZ,JOHNJ. Types of linguistic communities. Anthropological Linguistics
1962 4,( l ) ,28-40.
GIJMPERZ,JOHNJ. Linguistic and Social Interaction in t w o communities. In John
J. Gumperz and Dell Hymes (Eds.), The Ethnography of Communication.
Washington, D.C.: American Anthropological Association, 1964, 137-153.
HAIJGEN,EINAR.The Norwegian language in America: A study in bilingual be-
havior ( 2 vols). Philadelphia: University of Pennsylvania Press, 1953.
HAUGEN,EINAR.Bilingualism in the Americas: A bibliography and a research
guide. (American Dialect Society, No. 26.) Montgomery: University of
Alabama Press, 1956.
HOWELL,RICHARD.Linguistic Status markers in Korean. Kroeber Anthropological
Society Papers, 1965, 33, 91-97.
HYMES,DELL H. The ethnography of speaking. In Anthropology and human
behauior. Washington, D.C.: The Anthropological Society of Washington,
1962, 15-53.
H m , DELL H. Directions in (ethno)-linguistic theory. In A. K. Romney and
R. G. D’Andrade (Eds.), Transcultural studies of cognition, washington,
D.C.: American Anthropological Association, 1964a, 8 5 6 .
HYMES,DELL H. Introduction: toward ethnographies of communication. In
John Gumperz and Dell Hymes (Eds.), The ethnography of communica-
tion. Washington, D.C.: American Anthropological Association, 1964b.
1-34.
HYMES, DELL H. On “Anthropological linguistics” and Congeners. American
Anthropologist 1966, 68: 143-153.
HYMES,DELL H. Two of linguistic relativity. In William Bright (Ed.),
Sociolinguistics. T e Hague: Mouton and Co., 1966. 114-165.
H m s , DELL H. Why linguistics needs the sociologist. Social Research, (in
press ) .
HYMES,DELLH. On Communicative com etence. (To ap ear in a volume edited

KROBER,
l
by Stanley Diamond on anthropofogical approac es to education. )
A. L. Evolution, history, and culture. In Sol Tax (Ed.), Euolution after
Donuin. Chicago: University of Chicago Press, 1960, 1-16.
LABOV,WILLIAMA. The Social Stratification of English in New York City.
Washington, D.C.: Center for Applied Linguistics, 1966.
MALINOWSKI, BRONISLAW. Coral gardens and their magic, Vol. 11. London: Allen
and Unwin. 1935.
MEAD,MARGARET. Public opinion mechanisms among primitive peoples. Public
Ovinion Ouarterlv,“ - 1937,
. 1., 5 1 6 .
WEINRE& U&. Languages in contact. New York: Linguistic Circle of New
York, 1953.
WHITELY,W. H. Social anthropology, meaning and linguistics. Man, new series,
1966, 1, 139-157.