April 20-21, 1979
Papers from the Conference on
April 18, 1979
1050 E. 59TH STREET
In conclusion, it is stressed that the notion "universal
rule" should not be accepted a priori, before assessing the
extent of its contribution, if at all, to the theory of syntax.
Specifically, linguists must try to assess the explanatory value
of a substantive constraint in terms of a "universal inventory
of rules", as opposed to a formal one, and try to devise empirical
tests for validating the notion "universal rule".
Bach, E. (1965) "On Some Recurrent Types of Transformations"
Georgetown University Monograph Series on Languages and
Linguistics, Vol. 18, ed. by C. Kreidler, Washington, D.C.
Bach, E. (1971) "Questions" Linguistic Inquiry. Vol. 2, 153-66.
Bach, E. (1974) "On the VSO Hypothesis" Linguistic Inquiry. Vol. 5,
Chomsky, ~ . (1965) Aspects of the Theory of Syntax. Cambridge,
Mass.: M.I.T. Press.
Chung, S. (1975) "On the Subject of Two Passives in Indonesian".
Subject and Topic, ed. by Charles N. Li. Academic Press.
Johnson, D. (1974) Toward a Theory of Relationally-Based Grammar.
University of Illinois Dissertation.
Keenan, E. (1975) "Some Universals of Passive in Relational
Grammar". Papers from the Eleventh Regional Meeting of the
Chicago Linguistic Society.
Keenan, E. and B. Comrie (1977) "Noun Phrase Accessibility and
Universal Grammar". Linguistic Inquiry. Vol. 8, 53-100.
Peters, S. (1970) "Why There Are Many Universal Bases". Papers
in Linguistics. Vol. 2, 27-43.
Sheintuch, G. (1977) Same Rule in a Transformational Theory of
Syntax. University of Illinois Dissertation.
Sinha, A.K. (1978) "Another Look at the Universal Characterization
of Passive Voice". Papers from the Fourteenth Regional
Meeting of the Chicago Linguistic Society.
Stockwell, R., P. Schachter and B. Partee (1973) The Major
Syntactic Structures of English. New York: Holt, Rineholt,
and Winston, Inc.
* language Structure and Linguistic Ideology
Michael Silverstein
The University of Chicago
Cases are rare in which a people have begun to
speculate about linguistic categories, and these
speculations are alroost always so clearly affected
by the faulty reasoning that has led to secondary
explanations, that they are readily recognized as
---Franz Boas (1911: 71)
Were I to begin by observing that "Webster's dictionary
defines ideology as ... ," you would have an example of a very camon
American linguistic ideology in action.
It would be the rhetorical
appeal to the published dictionary as the codified authority on
what words really mean. Even the whimsical force of such rhetoric
rests on a large set of rationali7..ations about the nature of the
dictionary's authority in such matters. Part of our educational
establishment--and especially the publishers--encourage it as
llRlch as possible. Or again, I might start by pointing out that the
word canes fran the Greek root for 'idea,' illustrating another
camon ideological proposition about the "true" or even central
meanings of words lying in their etynDlogical origins, knowledge
of which sarehow allows us to use words correctly. In the works
of ideologues such as Edwin Newnan (1974, 1976), these confusions
of etynnlogy and sanantics at the phrasal level become the basis
for declarations of pet likes and dislikes about contenporary
usage; "clear," or "literal (and correct)" usage is generally so by
historical priority, as in the usual Malinowskian charter myth.
But I do not address myself only to articulated beliefs that
are incorrect or contanptible. I should clarify that ideologies
about language, or linguistic ideologies, are any sets of beliefs
about language articulated by the users as a rationalization or
justification of perceived language structure and use. If we
~ a r e such ideologies with what goes under the name of "scientific"
about language, we might find that in certain areas
the ideological beliefs do in fact match the scientific ones, though
the two will, in general, be part of divergent larger systans of
discourse and enterprise. We need have no conceit one way or the
other, however, that autanatically privileges so-called "scientific"
description, or autanatically condanns native ideological rational-
In fact, I want to develop here SCIre aspects of the subject
that will, I hope,show the relationship between ideology and structure
in the realm of language to be llRlch the same as in any other realm
of social life, a phenarenon of no little significance for the prac-
tice of linguistics.
'lb develop this thEme, I will first indulge in a sketchy his-
toriography of one of the lines of development of American linguistic
anthropology, tracing the definition of the problem by one of the
IIDst misunderstood writers of the century, Benjamin Lee Whorf. It
was Whorf, I will claim, who clarified one aspect of the problem, as
it was posed by his academic grandfather Franz Boas. For Whorf
proposed that the users' native ideology of reference, of how
language serves as a system for segmenting, classifying, and thence
speaking about the universe of experience "out there," is systemat-
ically related to, and at least in part systematically derives
from, the gramnatical structure of the language.
More particularly, Wharf saw regularities in the distorting
relationship between native awareness of language as a referential
systan and referential structure itself. Wharf's scientific
perspective (not, it should be noted, any anti-ccmparative nihilism)
led him to fonnulate a principle of "linguistic relativity" that
might, he concluded, stand in the way of pure positivistic
science (no little concern for the MIT graduates to whom he
addressed a number of popular, and subsequently misconstrued,
articles in 1940-41). The analytic lesson here for linguistics
I will term a linguistic uncertainty principle in keeping with
Wlx>rf's original (and, in Ietxospect, unfortunate) rretaphor: those
who would think that native linguists can directly penetrate to
the linguistic coding of referential "reality out thp.rp." by examining
their own propositional systelll--no matter heM "deeply"--or by examining
others I with crude approximation-translations of propositional content,
unrecognizably distort the object of investigation in
This point leads us into the second area of questlOns, deal1ng
with ideology vs. structure for other areas of language use, other
"functions" of language than the referential, as we now say. From
the writings of the philosophers Wittgenstein and Austin and their
interpreters fran the work of Dell Hyrres and his students that
has carved a field called the "ethnography of speaking," from
the development of whole areasof research called sociolinguistics
and ethnomethodology, it has becorre clearer that not only
speak about, or refer to,the world "out there"--outslde of lan-
guage--they also presuppose (or reflect) and create (or fashion)
a good deal of social reality by the very activity of using language.
We should ask, in particular, how the seaningly reflective and
creative or "performative" functions of language (or, rather, of
language use) relate to native IDVareness and native ideology. Can
we generalize Whorf' s penetrating insights from the plane of
reference to the whole of language function? I think we can discern
the same disjunction between ideology and structure, one, m::Jreover,
which assimilates function to reference and thereby affects the
strategy of language use. .
Answering these questions in this way, we come, as ln any
social science to the problan of accounting for history. I will
briefly how various generalizations about historical change
of linguistic structure, at both the referential and m::Jre broadly
functional levels of analysis, sean to be the outcome of a structure-
ideology dialectical process. This contrasts with views of change
as autonarous internal evolution of rule structures, from some
tendency to analogy, or fran sorre systemic striving for. psychological
econany of rule-ordering relationships, or from some gomg-to-
ccmpletion of otherwise variable rules. The "dynamic synchrony"
that many have seen as the basic condition of human
following Jakobson and the Prague Circle, is, by our
precisely the tension between linguistic structure and varlOUS
institutionalized and non-institutionalized ideological under-
standings of that structure. Thus the necessary conditions for
the formation of ideologies, and the sufficient conditions for
their institutionalization, ought really to be the heart of
historical explanation, for this illustrates on a large scale
what we are, for better or worse, constantly doing to language
in microcosm whenever we think about it.
But in the beginning was Boas. Whether or not Boas had
intellectual contact with his great sociological contanporaries
in France and Germany, I am not certain. By the 18805, he was
hard at work translating the lessons he learned in psychophysics
about what we must call ethnoclassification into an emerging
discourse about "ethnological" or cultural form and its history
(see Stocking 1968: 133-60, 195-233). And the cardinal problan
here, as I see it, was to differentiate between "primary"
cultural classification (as shown in la)--the segmentation and
ordering of the supposedly shared social universe of experience,
which m::Jved along on its own historical plane independent of the
personal will of individuals--and what Boas called "secondary
explanation" or rationalization (as shown in lb)--the edifice of
ideological beliefs about the system of categorizations implicit
in institutionalized social action.
1. (a) "Primary"ethnological phenomena ("fundamental ethnic ideas")
1Cultural Pattern I organizing> "range of personal experience"
e.g., system of religious ritual activity; (referential) language.
(b) "Secondary" explanation (secondary reinterpretation)
about>ICultural pattern
or avallable to consclousness) .
Language, or rather, the social activity of using language,
plays an eXaJPlary role in Boasian theory, precisely because, it
is claimed, the "primary" cultural categorizations of using
language, described by a gramnar, m::Jve along in history m::Jre
independently of secondary overlays than any other phenanenon
of social life. 'Thus, as Boas wrote in his 11 Introduction" to
the Handbook of American Indian Languages, published in 1911,
if we adopt this point of view, language seans to be one of
the m::Jst instructive fields of inquiry in an investigation I,
of the formation of the fundamental ethnic ideas. The great
advantage that linguistics offer in this respect is the
fact that, on the whole, the categories which are forrred
always remain unconscious, and that for this reason the
processes which lead to their formation can be followed
without the misleading and disturbing factors of secondary
explanations, which are so caIlIDn in ethnology, so much so
that they generally obscure the real history of the
developnent of ideas entirely (1911: 70-71).
Here, our discussion turns to Whorf. For, as I noted, he
developed the sharpest tools for describing the granmatical
categories of propositional language we have yet seen, of which
the anergence of transfonnational-generative granmar is merely a
notational refinement. And Wharf then used these analytic tools
to specify in detail an answer to the Boasian question, which
parts of language emerge in "secondary rationalization" and
associated in European linguistic ideology and the technical gram-
matical tradition that grew out of it with a kind of direct
representational relationship with the things "out there"--
hence, as Boas notes, they are considered to be "the parts
expressing the material contents of sentences," as shown in
figure 2. In saying that granmatical study, study of all
2. European ethnocentric view (i.e., secondary explanation)
[c!. Ib].
categorizations inplicit in the activity of using language, subsumes
and logically precedes the study of lexicon, Boas discovered
the structuralist principle that his contemporary Saussure was
enunciating in his Geneva lectures at approximately the very same
too. And if the lexical elanents of language, the roots and
stans of words, are continuous with the other elanents of language,
previously thought to be in the separate realm of granmar, the
categories inplicit in lexical elanents are the very SaJOO in
kind as those implicit in so-called granmar. If you want to
study the lOOanings of words, that is, the way words categorize
the universe of eXPerience they reflect, study granmar, because
you cannot in principle directly study the roots and stans of
words as vehicles of categorial meaning. 'Ibis is schematized in
figure 3. But how to study granmar?
3. Boasian (structuralist) view [ef. la]
"nodifying relations"
lexicon: unpredictable [including
minimal] surface-segmentable
granmar: all structural relations;
"material contents"[
"range of personal
eXPerience' ,

"range of personal eXPerience"
If, as Boas tells us, "the fundamental concepts illustrated by
human languages are not distinct in kind fran ethnological .
phenarena" (1911; 73), but distinct only in degree along the dimen-
sions of the structure/ideology dichotany, what parts of lan-
guage are relatively lIOre subject to "secondary explanations"
than the rest? Though Boas himself gives sane interesting examples,
characteristically, he never formulates a positive approach to
this problem. Indeed, it was only in the 19308, after an academic
generation had passed during which Sapir and Blocmfield had
matured that Whorf was able to take advantage of the progress
in analysis and re-address the issue.
In order to appreciate the brilliance of Whorf's research,
which as I noted above, has been incorrectly maligned as vague,
circuiar or tautologous by several academic generations,4 you
must one :important problem of Boasian linguistic
analysis, especially as Whorf himself sharpened and the
analytic tools. For the Boasians, language, or the SOClal ac-
tivity of using it, is the medium of the universal human faculty
of rationality, the ability to manipulate proIJ<?Sitional .
When we convey linguistic IOOssages, we engage ln camnIDlcatwg
prOpositions about entities and. their
and relationships "out there," 1.e., outSlde of language, the
cultural organization of which is reflected in the categories of
granmar. This reflectionist point of view, to be sure, is an
ancient inheritance of our own folk ideology, or "secondary
explanations" about language, a point we will be able to make'
lIOre precisely below. For the nanent, I want just to point out
that by the late nineteenth century, this reflectionist point
of view saw language as consisting of two relatively indePendent
orders of phenanena, "granmar" and "lexicon."
Ever true to character, Boas rmstered example after example
fran North Arrerican languages to make the negative point that the
European understanding of the relationship between granmar and
lexicon was entirely misguided, inaccurate on a universal scale,
and anpirically UIl\IDrkable in these languages of profoundly
different structure. "In the discussion of language," Boas
the parts expressing the material contents of sentences
appear to us as the subject-matter of parts
expressing the nodifying relations, as the subJect-matter
of granmar. In lIOdern Indo-European the
of ideas which are expressed by subordwate elanents lS,
on the woole, limited, and for this reason the dividing-
line between granmar and dictionary appears perfectly clear
and well drawn. In a wider sense, however, all etYJOOlogical
processes and word canpositions must be considE;red as parts
of the grallJllaI" and if we include toose, we fwd that, even
in the number of classifying ideas
is quite large (J911:33-34).
By "etYJOOlogical processes and word canpositions," Boas JOOanS the
structure of what were considered the word stans of languages, always
thereby historically affect cultural practice. I will not darnn-
strate this in total detail; perfonning the necessary philology
and historical contextualization of Whorf I s oeuvre would take a
large volume. I will just outline Whorf' s approach to the two
central issues. First, where do ideologies of reference (or
representation) cane fran? Second, how, by providing an objectified
meUlphysics of "nature" against which social practice (including
language) becanes interpretable by the participants, do refer-
ential ideologies relate to historically-particular traditions of
social belief and practice, which we call cultures?
Tb begin with, the Whorfian view of language makes a distinc-
tion between what we now call "segmentable surface foun"--
the recordable, transcribable stuff of speech that we divide up
into words and their explicit parts--and grammatical structure.
'Ibis last he presents as an elaborate set of fonnal-semantic cate-
gories that deteunine words and parts of words. For every explicit
part of a word, we must give the organization of categories
manifested by that part, or of which that part is a manifesta-
tion in combination with some other linguistic element. In order
exhaustively to analyze a language in this fashion, carrying
through the Boasian point of view, Whorf had to postulate dis-
tinctions along four general types of categorial dimension as
schanatized in figure 4.
4. Whorfian categories of grammar
(a) Overt (phenotypic)/covert (cryptotypic)--by ubiquity of
fonnal expression of the category in "segmentable surface
fonrl'; cryptotypes discoverable only by transfonnational
(Harris) relationships.
(b) Selective [semantic] / rrodulus [semantic] --by obligatoriness
of mutually-exclusive classifications of lexical forms;
Primary selective category = "lexemic" category
(c) Selective [isosemantic]/alternative [isosemantic]
Cd) Specific/generic
First, Wharf distinguished between overt (or phenotypic)
categories and (or cryptotypic) categories, where the criterion
of assignment rests on the ubiquity and overt fonnal expression of
the category in "segmentable surface foun." An overt category
is a category having a fonnal mark which is present (with only
infrequent exceptions) in every sentence containing a menber
of the category. 'Ibe mark need not be part of the same word
to which the category may be said to be attached in a para-
digmatic sense; i.e., it need not be a suffix, prefix,
vowel change, or other 'inflection, I but may be a detached
word or a certain patterning of the wooIe sentence. 'Ibus in 5
English the plural of nouns is an overt category... ( [1945)1956: 88)
On the other hand, a covert category
is marked, whether morphemically or by sentence pattern, only
in certain types of sentence and not in every sentence in

which a word or element belonging to the category occurs. 'Ibe
class nenbership of the word is not apparent until there is a
question of using it or referring to it in one of these
special types of sentence, and then we find that this word
belongs to a class requiring same sort of distinctive
treatment, which may even be the negative treatment of
excluding that type of sentence.... ln English intransi-
tive verbs foun a covert category marked by of the
passive participle and the passive and causative voices ...
([194511956: 89)
The presence of covert categorizations in any segrnentable surface
foun, then, as opposed to overt ones, depends on finding same
other particular construction in the language where a special
pattern of fonnal treatment necessitates postulating class mem-
bership for the foun in question. In a technical foun we might
paraphrase Whorf and clarify with current concepts by' saying
that we have to find a (Harris- )transfonnationally related
construction in which the cryptotypic category is marked by
special fonnal treatment, or as Whorf calls it a specific
"reactance. " For example, we can test a verb stem
for intransitiVity by seeing if it appears in an otherwise
propositionally-constant passive construction (where be- -p.p.
constitutes the overt "reactance"). Or we can test aparticular
noun stem for gender Classification by seeing what pronaninal
foun replaces it in anaphoric constructions, e.g., conjoined
sentences of parallel structure (where constitutes
the overt "reactance"). -
. For Whorf, the way in which language was meaningful and
ratlonal was encapsulated in such facts of configuration or
"rapport" between words. And where any distinction of semantic
relevance was to be drawn between overt segmentable surface
foun and covert categorization, it was the latter kind of category
according to Wharf, wherein lay the true, primary ethnological '
phenanena that Boas had been after. In that same Boasian historicist
:rein, Wharf writes that in the emergence of a cryptotype category
1n a language, the set of its menbership
bec<m3s increasingly organized around a rationale it
attracts semantically suitable words and loses menbers
that are semantically inappropriate. Logic is now what
holds it together, and its logic becomes a semantic associate
of that unity of which the configurative aspect is a bundle
of nOOJlX)tor linkages JlX)Qring the whole fleet of words to
their Semantically, it has. bec<m3 a deep
persuas10n of a pnnc1ple behind phenomena like the ideas of
inanimation, of "substance," of abstract of abstract
of force of causation--not the' overt concept
(lexat10n) corresponding to the word causation but the covert
idea, the "sensing," or, as it is often called... the
"feeling" that there must be a principle of causa-bon.
Later this covert idea may be more or less duplicated in a
word and a lexical concept invented by a philosopher;
or Qbjectification. This lies at the basis of what he variously
calls the "world view," the "habitual thought patterns," or
the "natural lQgic" of speakers. It generally constitutes the
natives' ideology of the way their language serves as a propo-
sitional systEm representing and talking about what is "out
there. " In his rrost precise essay on the subject "The relation
of habitual thought and behavior to language" 134-59),
Whorf analyzes the "Standard Average European" projection of
abstract 'quantity,' 'substance,' 'form' (including dimensionality
in 'space'), 'time' as "natural" or metaphysical categories Qf
reality. He traces these projectiQns to the primary selective
categories Qf the maximally-expanded noun phrase. Thus, fran such
an Engli:m noun phrase as scoops of sugar, which is
the partlcular structure that exhibits the cryptotypic selective
categQries Number, Size, Substanceless FQrm, Formless Substance, in
that order, as shown in schena Sa, speakers easily develop a
secondary ratiQnalizatiQn, or "objectification," in which all
these cryptotypic categories are projected as attributes or
features of every object of naninal reference.
e.g., causation. ([ms. HI36/37] 1956:81)
So, in this PDasian way of seeing things, the philosophical
invention of the abstract noun-stan causation, which is itself
a canplex form and which itself has manbership in several
categorizatiQns of nouns, is the secondary ration-
allzatlOn upon the cryptotypic category of English (and other
languages') verb phrase structure manifested in the proportiQnal
sets (be) red : redden :: die : kill :: fall : fell :: ... ,
a category called 'causative'in our granmatical
The second Whorfian division of categories relevant to our
discussion is selective sanantic categories vs. roodulus sanantic
categories. Independent of the Qvert or covert nature of the
formal markings, a selective category
is a granmatical class with narbership fixed, and limited
as canpared with sane larger class. A selective
category, or lexanic category, is one canpaJ'P to which
the next larger class is the total lexicon of the language.
certain sanantic and granmatical properties are assured in
the word by selecting it fran a certain class of fixed
nanbership not coterminous with the whole vocabulary.
([1945] 1956; 93)
(a) Np [tree large sC1pS of sugr]
Number Size Substanceless Formless
form substance
So, as an English noun,. cannQt substitute for in the
example 1D the same sense as lt can substitute for scoop- of
thus, the phrase three trees, as in 5b. And the
objectification of reference Whorf characterizes as the projection
onto such entities as trees the very abstract properties we would
expect by analogy: a certain bounded form (say rrultiple branches
each with circular cross section) and a certain'otherwise formless'
substance (say, wood). We can test the validity of this cbaracterizatiol
of intuition by just thinking about the referent of the phrase
sugars, as in 5c, which might be paraphrased by the
native as three (of)
whether in the context of the coffee-klatsch or of the -
But Wharf does nQt stQP there. He goes on to point out that
this "objectification" has extended the analogy to create another
fonnless substance, 'time, I which we measure into units with the
(c) Np [three large
NtIDlr Si*.:
It is important to see that selective categories, regardless of
whether they are overt (phenotypic) or covert (cryptotypic), are
basic granmatical subdivisiQns, that classify the overt segmen-
table fQrms into IllUtually-exclusive classes, such as noun-stans
vs. verb-stans in Latin (along with the other primary selective
categories). Thus, a selection of one or the other implies both
the form and part of the sanantics of the very word it defines,
noun vs. verb. Modulus categories, on the Qther hand,
are generally applicable and ranovable at will ... to any
word caning within a certain prerequisite larger category,
which may be either selective or another roodulus category.
The cases, tenses, aspects, roodes, and voices of Indo-
European... languages are roodulus categories--cases being
IOOduli of the larger category Qf nouns; aspects, tenses,
etc. rooduli of the larger category of verbs .... in widely
different types Qf speech, these familiar types of meaning and
function cease to be associated with selectivity and roodulatiQn
in the same way: ... ([1945]1956: 95)
It should be clear that the distinction here is bound to Whorf' s
inheritance of the word as the relevant danain of granmatical
analysis, though what he wishes to express lies actually in the
realm of what we navv call the phrase- and clause-level syntax
of the sentence, at which parts of speech and their paradigms are
definable. 7
In any case, by using these Whorfian analytic tools, we can
navv formulate with Whorf the principle of referential projection
(b) Np[three

large tryes]
J, .-!.-
Si:<:e Substance
rretaphorical unitary dimensionality of a line, and which we refer to
with noun phrases of parallel structure to the rreasure phrases we have
already seen. And in tenns of this objectification, we understand our
utterances to be predicating states of affairs at points and in intervals
along a seaningly speech-independent time line. 'The pragmatic (indexical)
category of t tense,' then, is a predicational category interpreted by
the users as really being tanporal in reference, and this sense of
tenporality anchors the category as 'tense' rather than as sanething
else, e.g. ,aspect or status. .
I cannot deal here with Wharf's presentation of the contrastmg
Hopi data, where the categories of the verb phrase a rather
different ideology of processual, rather than substant1ve prImacy.
'This should be read carefully in the original. I emphasize, however,
that Whorf is not talking about sensation and perception at the level
of the individual's physiological-psychological processes, as can-
rnentators have asserted even though he sanetimes uses the tenn
"seeing" (to be sure, a new, technical sense) in his. discussion.
Wharf is talking about the way people who speak a certam language
fonn an ideology of reference, an understanding at the conneputal
level of how their language represents "nature." And he goes on to
say that in practical situations of task-orientation, where people
have to reason out a course of action vis-a-vis the v.urld, "people
act about situations in ways which are like the ways they talk about
then" ([1941]1956: 148). Using a series of channing fire-insurance
examples, arrong other kinds of evidence, Whorf tries to reconstruct
the rnechaniSllS of practical rationality. . .
In his IIDdel, we recognize the disjunction between Imgu1st' s
elaborate categorial analysis of language and the rnechan1SllS of secon-
dary rationalization put to the service of practical rationality..
Just as in our earlier it as. though people"quas1-
consciously rationalize about practlcal sltuatlons based on all
the analogical and suggestive value of the patterns" ([1941]1956: .147)
of their language; taat is, they objectify on the basis ot analog1es to
certain pervasive patterns and .
act accordingly. 'This secondary of the IlJll?Uistlc
system is, however, understood by the nat1ve as. a d1rect "
denotative relationship between surface fonns ::eallty out
It can renain implicit, as in IIDst practical Sltuat1ons, or, as m
technical reasoning, it can becane explicit through the
of new tenns or "lexations" as Whorf calls then. 'These tenns, 1t
should be recalled have their own underlying categorial structure,
distinct fran that' of the linguistic foms fran which they arise by
objectification. . .
It is at this level that Wharf's analysis constitutes a cr1t1que
and undennining of the constructivist scientific project of
because his principle of projection leads him to assert the poss1b1llty
that "objectiiied" scientific tenninology--which we use to reason about
certain practical situations called 'data'-is itself a phenanenon of
secondary rationalization that does not really get near non-cultural
(non-symboliC) "reality" (as sane would say, the "really real"). As
hilosophers might observe the conceit of "essentialist" scientists .
, " "thi 1 e " m
is that they are fashioning an "object language, or ng '. .
which every truly distinct individual entity in Wlll
tically differentiated (specifically with reternng lex1cal expresslOns)Wlthin
theoretical discourse, and all truly distinct relations will be
describable as such. As Boas, himself sanething of a positivist, had
put it earlier,
when we try to think at all clearly, we think, on the whole,
in v.urds· and it is well known that, even in the advancement
of inaccuracy of vocabulary has often been a sturrbling-
block which has made it difficult to reach accurate conclu-
sions. (1911: 71-72)
Wharf points out that the abstractions of scientific uses of
ordinary language may well emerge fran rationalizing cryptotypic
selective categories through pervasive analogical patterns. It
thus is indeterminate to what extent those scientific concepts
based on anyone language structure fonn a true "thing language"
or are, rather, just a "quasi-syntactic language" as
ca.rnap and Morris ([1938] 1971:30) would have it, a set of
"sentences which are apparently thing-sentences, and so about
objects which are not signs," but which "turn out under analysis to
be pseudo-thing sentences which. must be interpreted as syntactical
statements about language." Viewed fran the perspective of the
different possible systems of objectification that play a role in
ideologies of "the real," this is Wharf's principle of "linguistic
relativity." Viewed fran the perspective of the positivist project,
at the time a dcminant philosophical and scientific concern, this
is rather a principle of linguistic uncertainty. If we have cause
to doubt that "our own concepts of 'time, t 'space,' and 'matter' are
"given in substantially the same fonn by experience to all lOOn," but
rather are "in part conditioned by the structure of particular
languages" (Wharf [l94111956: 138), then how can we neatly distinguish
the tv.u classes of rreaningful statements (see footnote 8) so as to
be able to carry through the reduction (Carnap's Aufbau) of all
empirical knowledge to space-time coordinates?9
We can sunmarize the line of reasoning leading to Whorf' s
projection and relativity/uncertainty principles in a IIDre IIDdern
tenninology, unavailable to Wharf himself. In using language as a
device of propositional reference in practical situations (the
apparent IIDde of reasoning about such situations), speakers pre-
suppose a reality "out there" that language codes and categorizes.
And the presupposed categorizations emerge as though speakers
analogically project underlying semantic categories fran certain
maximally-expanded surface lexicalizations (segmentable v.urd-stens,
phrases, etc.) into a rretalanguage of conceptual labels taken to
be "object language." Awareness of so-called reality is at least
partially precipitated out of this process, Wharf v.uuld claim, so
that for speakers of a given language, contextualized social action
is intimately bound up with this projection.
'Ibis leads into the second area for discussion, when language
use is itself seen as contextualized social action, of the sarre
order of phenanena as any other cultural behavior. Can we distin-
guish what we might call an ideology of language use in a given society,
distinct fran what we might call the social system of language use
as we v.uuld "scientifically" (see footnote 2) describe it? If
we think of linguistic behavior as having significance and consequence
for those who engage in it, we can rephrase the first part of this
distinction in the following way. Is there a set of beliefs in
terms of which people rationalize the use of particular linguistic
forms to achieve certain socially-tUlderstood purposes, to be used in
certain socially-tUlderstood contexts, etc., because of such factors
as inherent power or force of language itself, suitability of language
fonns for the context in which they should be (or are characteris-
tically) used, etc.? The second tenn of the distinction is
equivalent to asking if a systematic cross-cultural study of language
use is possible as a genuinely social (or genuinely anthropological)
linguistics. Is there a way of analyzing the uses of language
fonns so as to relate them systematically to particular kinds of
systems of role and status, institutional structures, corporate
group structures, etc.; in short, is it possible systematically
to analyze a social organization of language usage? Observe that in
this second area, we are not dealing with whether or not a given
language contains nanes for (has phrases or words referring to)
these aspects of the society relating to its use. 'The question is
whether or not there is a cuIturally-detenninate distribution of
linguistic forms in socially-constituted contexts of use, and how
norms of such usage--and departures fran thEm--can be understood by
the users.
These two distinct "functionaliBllS" have not, tUlfortunately, been
kept analytically distinct in the literature. (See also Merton 1968:
73-138 and references there.) In particular, as we shall see below,
linguistics has characteristically taken at face value a native
ideology that objectifies "force" in language itself, concretizing
this force in tenns of propositional-structure-as-usual. Social
anthropological studies hare characteristically analyzed native ideology
as though it were an accurate "scientific" picture of the relation
of language fonn to social context. Both of these, I will argue, are
expectable in tenns of a broadened understanding of the bases of
ideology that Wharf proposed, especially as such broadened understanding
explains the tendency to assimilate our own "scientific" views to the
source frem which they have anerged, our own European folk ideology
of language.
Consider in this connection the traditional distinction between
granmar as the description of language structure versus rhetoric as
the description (or prescription?) of strategic language use. As
it has developed into the science of linguistics, the study of
granmar has attenpted to encoopass first and foraoost the systEm-
aticity of how linguistic fonn relates to propositional reference-
and-predication. The methods and theoretical apparatus of this
study of granmar necessitated ultimately the scientific .
of recurrent sameness (Blocmfield (1926) 1970:129-30; cf. S11verstem
1978: 237-8), a constant referring-and-predicating potential of
linguistic fonns, specifiable by a lIDdel, independent of their
context of use. 'Thus, in scientific study of granmar, by this inher-
itance, we have worked at specifying "langue" or "caJlletence," the
context-independence of which is an assumption necessary to turn
prescriptive granmatical doctrine into descriptive linguistic science,
while at the same time preserving its characteristic methods of
dealing with linguistic data.lO
To the extent hCM'ever, that certain fonns of language code
indexical-referential categories, their meaningfulness I;JroIJ<?S-
itional tenns cannot be defined independent of sane spec1f1catlOn
of the context in which the fonns are uttered. As it turns out,
a great ntlllber of the categories of reference-and-predication are
indexical and an even greater number of categories sean to be
indexical'independent of the referential-and-predicational value of
utterances. To the extent that we can give rules that tell us .
the regularities of indexical reference-and-predication, w111
involve sane theory of kinds of recurrent contextual cond1t1ons.
For exarIJlle, the social role of speaker,. of what
individual speaks an utterance, is the Ill11UIllUIll recurrent contextual
feature necessary to define the propositional contribution of the
English class of indexical forms of lime. Such a theory of
recurrent contextual conditions in which tokens of forms appear,.
necessary even to canplete the theory of language as a
and-predicationalsystem, is already an implicitly social descnptlOn
of what we migbt tenn the structure of contexts of utterance.
How much of the theory of social context must of necessity
enter even an attenpted account of the folk inheritance of language
structure as propositional system? It is not as have
noted elsewhere (Silverstein 1973; 1976b; 1977) th1S 1S but
a patching-up approach, in as much descnptlOn
fran the folk realm of rhetor1c (language use) as 1S needed to
preserve the analytic fiction that there is a well-defined systen
corresponding to the scientifically-winnowed folk ::ealm ?f
A much more useful analytic distinction is, folloWlng .
aIlDng the types of sem1osis, or meaningfulness, called 1C?n1C, mdex1cal,
and symbolic, a distinction which cross-cuts any folk-denved
notion of propositionality as a well-defined realm. Not all of the
linguistic features that effect referential-and-predicational
tion" can be accanodated in the analytic plane of Peircean symbol1BT1;
but such as can be are readily analyzed in terms of generative
grammatical I always use the tenn senantic (or senantico-
referential) for this type of meaningfulness. It is that con: .
tribution to the meaningfulness of utterances made by categones wh1Ch
are specifiable through context-independent gramnatical analysis
based on (logical) synonymy, antonymy, hyponymy fLyons 1968: 453-55) ,
partonymy, etc. of sentences and their elaoonts. 2 .
By contrast I maintain that the indexical plane of meanmgfulness
properly the folk realm of rhetoric (the systen of lan-
guage use), how language signals derive
effects in various socially-constituted sltuahons of discourse.
'!bat is to understand how speaking (or other similar uses of language)
is social action, accanplishing such various social ends
as warning insulting, marrying, condenming, christening, growing
yams sores heal, creating ligbt in the world, etc:, we
must'systenatize the description of relationships of coex1Stence
(tUlderstood copresence) that hold between elanents of speech and
elenents canprising the context in which. speech elanents are uttered.
In the context of our discussion, this can perhaps be best :mderstood
by developing the contrast of two functionaliBTIS (Silverstem 1976b;
44-45i ffi3. 1978), the second encanpassing the first as the appropriate
means of cross--linguistic (and cross-cultural) canparison and
description. These tiro functionalisns have not been carefully
enough distinguished in the literature, and I hope that the dis-
cussion here helps to explain why this should be so.
'lb begin with, there is one sense in which language is
"functianal," inaswch as its use seans to the natives to be poten-
tiallypurposive, or actually effective, or the like, in their own
individual experience. Let us call this goal-directed and sanetimes
goal-achieving categorization of occasions of use the function
language. From the discussion above, it should be clear that
is closely related to native understanding of the individual's
ability to use fonns of language strategically, and in a manner
subject to in accordance with an ideology of approp-
Insofar as function, is externalized in verbalizations
about language-what Bloanfield l1ked to ridicule as "secondary and
tertiary responses to language" ( [1944] 1970; 413-25)--it implies
a metalinguistic functionl for language with ei .
special vocabulary and/or syntax, or a partJ.cular meta1J.ngu:LstlC
use of otherwise fonnally undifferentiated linguistic material. As
we shall see in detail below, the ideology of "perfonnativity" deron-
strates nicely the interdependence of metalinguistic function
formation of linguistic ideology.
But there is a contrasting sense in which language is "func-
tional," inasnuch as by characteristic distribution of particular
fOImS in certain contexts of use, these foImS (or, rather, tokens
of than) serve as specifically linguistic indicators (or indices)
differentially pointing to (indexing) configurations of contextual
features. Let us call this indexical quality of speech fOImS, or
indexical rrode of their signification, functionZ' It is very
important to realize that indexical fonns are not to be restricted
to surface.-segm:mtable lIDrphanes, or caIDinations of morphanes. Any
linguistic configuration is potentially indexical. There is, to be
sure, a trivial sense of this, in which any occurrence of speech
minimally indexes the individual in the role of speaker; but there
is also a nontrivial sense in which any particular abstractable
feature(s) of speech might be discovered to be indexical of par-
ticular features of context, fran "phonane- or lIDrphane-sized" chunks
of language all the way "up" to choice of particular "language" itself.
Furthe:mx>re, any particular "surface" stretch of language will
probably figure in multiple indexical functions2. Perhaps in one
such function2 it will be isolable as the total indexical fonn, while
in another such functioll
it will be isolable as a canponent of an
indexical fonn.
In actual speech situations--Le., at the level of tokens or
instances of usage-we can identify tiro contrasting kinds of
indexical relation of a speech elanent in particular to its context
of· utterance. Where the participwts understand the copresence of
sane indexed aspect of the context independently of the occurrence
of the indexical feature of language-even though there is such an
indexical relationship-we might say that the participants' indexical
understanding of speech-fonn to context presupposes the-existence-
in-context-of the indexed feature. Contrastively, where the participants
understand the copresence of sane indexed aspect of the context only
by the occurrence of of the indexical feature of language, we
might say that the participants' indexical understanding of speech-
fonn in context creates the existence-in-context-of the indexed
feature. ClJviously, presupposition/creativity in this sense
depends upon the particular configuration of sinultaneous carr-
rnunicative codes in use at a particular "nx:ment" defined by the
indexical feature of speech at issue, upon the relevant prior
cannunicative behavior of the participants, etc. However, in
discussing regularities of speech usage in this way, we must
fonnulate the range of characteristic indexical presupposition/
creativity at the level of indexical types (i.e., the elanents of
linguistic structure) for the various indexical features of language.
Thus, the true locative deictics (locative shifters) in English
are relatively presupposing, where they can be isolated as lexical-
fonn indexes independent of syntax. since they fail to refer
indexically unless the existence of their referent can be
guaranteed on other grounds (e. g., visually, by syntactically-
connected reference or predication, etc.). With a finer analytic
delicacy, the isolable English deictic that is more highly
presupposing than this; which fact becanes all the more important
in the discourse-internal deictic (corefe.rence) functions2 of these
words. Inversely, many linguistic indexes of speaker-hearer status
relationships are relatively creative, as for example the American
English alternant fOImS of indexical reference to addressee
(Brown & Ford 1961), e.g., names, since they are characteristically
the very means of establishing the social dimensions they index in
particular interactions, potentially quite independent of other
camnmicative codes.
A great deal of the give-and-take of actual interaction
depends on the constantly-shifting camnmicative negotiation and
ratification of indexical presupposition vs. indexical creativity,
of language vs. other cannunicative media. The functionall game
is played through the functioning2 of indexicals. In this larger
sense, it is through indexicals that any signalling systan such
as language makes contact with other systans that fonn the universe
of conventional social praxis, and hence, this analytic danain of
language is rightly called pragmatics. Formal surface features of
language are thus not either sanantic or pragmatic; they function
pragmatically in many indexical systans at the same time as they
contribute to the artificially-abstracted reference-and-predication
danain of context-independent propositionality. This latter danain,
sanantics, is never identifiable with any actual utterance-type
or locatable in any actual function
language; indeed, sanantics in this sense is an ideall.zation
abstracted fran the pragmatic systans of language, a point to
which we can return below.
Here, I have merely sketched two notions of "function" which
have been current in the l:lterature, and which must be distinguished
so as to pennit a lIDre precise statanent of the role of linguistic
ideology in language use. If pragmatics is the descriptive danain
of language use, then we will now concern ourselves with what we might
the native Pragmatic ideology, expressed in metapragmatic
theories, or ethno-metapragmatics. SUch theories, as rationalizations
about the use of language, are to be distinguished in principle fran
metalinguistic (in particular, metapragmatic) features of language,
those fonns which allow reference-to
and predication-aboutI language
itself. In what follows, in fact, I will be endeavoring to daJDnstrate
a relationship between metapragmatic features of language and
ethrJo.-metapragnntics, as a further example of the kind of
phenarenon that Whorf dealt with at the level of an ideology of
reference. To be sure, metapragmatic features of language constitute
a functional I subsystan of referring-and-predicating use of
language, where language use itself happens to be the topic of
discourse. We can see here an expectable objectification parallel
to toose discussed above. Such objectification can be illustrated
with one particularly influential theory of "speech acts" that
has anerged in Anglo-American learned circles.
But, llDre generally, I think we can subsume this kind of
projection-by-objectification as a particularly obvious factor
contributing to a wider phenarenon that underlies native pragmatic
ideologies. '!his is the tendency to rationalize the pragmatic
systan of a language, in native understanding, with an ideology
of language that centers on reference-and-predication. '!hat is,
native ideology explains or rationalizes about function
(presupposing/creative indexical effect) by analogically projecting
basic structures of reference-and-predication (propositionality)
as units of functional
effectiveness. '!here would appear to be
specific, identifiable caJilOnents of this tendency. First, there
is the tendency of pragmatic ideology to focus upon identifiable
surface lexical itEm>. '!he effectiveness of language in context is
explained by locating the power in words, phrases, and similar
surface-segmentable itEm> of propositional analysis. Second, there
is the tendency to focus upon these units' contributions to pro-
positionality as the starting point for explanation of other effects.
Particular functional
effects are explained as "metaphors" of
"literal" referential-and-predicational meanings, or "extensions"
of basic reference, used to effect certain functions, thanselves
perhaps constituted fran metalinguistic propositiona! material.
'!hird, there is the tendency to understand the functionl of language
in tenns of presupposing functional2 relationships (rather than
creative ones). Appropriate language use is, as it were, con-
stituted as a "metonym" of its context, a making explicit of what
one can purportedly know independent of the occurrence of speech.
failures are recognized to the extent they are so
reducible. In a sense, this is parallel to the way referential
ideology values a speech fonn as a "metonym" of the referent
projected fran it, seemingly a speech-independent entity existing
"out there" which language merely reflects. '!he result of this
constellation of tendencies is that natives' understanding of
their own syst6lS of linguistic usage frequently conflicts with
the caIParative-functional perspective.
Qle of the clearest examples of this, it seans to me, is the
pragmatic ideology that centers all language use, Le., function
(and thence function
) on the concept of "perfonnativity," fran
which has anerged a philosophical doctrine of sane persuasiveness
in linguistic circles. To discuss this, I will briefly review
sane of the salient characteristics of the so-called perfonnative
constructions as such, and then discuss the Whorfian foundations
of the ideology of perfonnativity. '!bis example should point up
the mutual relationship between ideology and functionl' at the
same time as it clarifies what we should be looking for as
canparatists of functional systEm> of language.
So-called "explicit (Austin 21975:68-69) or
explicitly perfo:nnative constructions in a language such as English
have the overt fonn of first person Subject, second person (Indirect)
Object propositional statanents, with simplex Verb in fonnally
Present non-Durative inflection, and sanetimes a Conplanent of
clausal or equivalent structure. '!bough they are overtly
statanental in fonn, they sean to acccrnplich, to "do," sane
specific predicated transfonnations of the social relations and
other contextual understandings in the situation of speech, for
example transforming the non-speech social roles of the participants.
'!bus, uttering 1. pranise that ... puts the speaker in sane new
understood relationship of obligation to the addressee. '!be
uttering of this fOImlla under certain "presupposed" conditions
conventionally "creates" new conditions in the context of utterance.
Similarly, many s:imple granmatical transfonns of otherwise seaningly
explicit perfonnative constructions conventionally effect similar
contextual transfo:nnations, e.g., Passengers are hereby warned
that. ... Observe that here it is possible directly to reconstitute
a kind of full, statanental-like syntactic fonn, [Speaker-as-
Authority) hereby passengers [-as-Addressees) that ...,of which
the earlier fonn is an agentless passive.
So long as our native
understanding of such transfonned constructions depends on
reducing than to "explicit" fonn, the latter serve as the starting
point for all ideology of perfonnativity.
I should stress here that the explicit perfo:nnative formula
describes the conventionally-understood activity that speaker
(and addressee, etc.) are engaged in at the m::xnent of utterance
by virtue of speech. It is, as I tenned it metapragmatic.
In a llDre technical sense, we could say--Austin ( 1975: 6; 70-71)
notwithstanding-that it predicates of the speaker the doing of
a certain kind of activity with respect to the addressee, etc.
And the conventionally-based understanding of the ccrnpulsive
transforming effect of engaging in this particular linguistic
activity means s:imply this. '!he zero tense-aspect fonn (hence,
prmm,tically, the residual present-nondurative) of the particular
predicated proposition has all the nonnal effects of statanental
speech. Natives have the expectation that the predicated trans-
fonnative activity will be effected--that the performative will be
"felicitous"--given that certain prerequisites are satisfied, in
exactly the same way that predications are taken as UIlll3.J'kedly
'true' propositions in the absence of contraindications. We can
then understand why, for exanple, it is a legal concern to establish
that the presupposed contextual conditions obtained in sane situation,
that certain perfonnative formulae--or their equivalents--were in
fact uttered under these conditions, and then that the speaker did
or did not carry through on the expected transfonned social relations
vis-a-vis sane addressee (typically now a plaintiff). Qle might
mention breach of pranise of narriage, or the Marvin vs. Marvin
case. Legal reasoning, we can see, makes explicit the ideology of
perfonnativityor its equivalent. What is of concern is whether
or not the equivalent of what we can name as a social interaction
of a particular type did or did not in effect take place, and
effectively take place.
Of course not all perfonnative formulae becane the topics of
legal procedures, nor are legal procedures always couched in tenns
of trying to discover whether or not the equivalent of sane per-
fonnative fonrnla was uttered, under specific conditions. But
this kind of phenanenon is the paradigm case of native ideological
reasoning about the effective use of language. And further, I am
not cla:ilning that the ideology of explicit perfonnativity here
obscures what is going on; on the contrary, explicit perfonnativity
here is what I \\Duld term transparent to accurate ideological
reflection, at least to that degree of delicacy of giving part of
a functional
typology of speech events in which people engage.
Ideology and praxis reinforce each other in this respect; a point
which will, I hope, becane analytically precise below.
The question for the linguist is, rather, if all effective
use of language can be "scientifically" treated in exactly the
same way. Can all the conventional understandings about how
language is a transforming social mediwn be expressed in tenns of
this functionall ideology of explicit perfonnativity? Do these
speech-event names that occur as predicates of fonrnlae used in
certain circumstances, and seemingly accomplishing certain social
ends, delimit a principled area for "scientific" study? Can we
language use as a functional
realm in this way comparatively,
Wlthout sane framework for functiona1
explanation? I would maintain
that this approach dffiOnstrates the pnenanenon of "secondary
explanation" with which Boas and Whorf were concerned, the Emergence
of a mode of discourse about language use that shows all the
properties we have discussed above.
The rrost extensive exemplification of this ideology is found
in the \\Drks of the philosopher J.L. Austin (especially 21975, based
on 1955 lectures) and his followers. To Austin we owe the
trichotany "locution---illocution--perlocution" as three "acts"
that are engaged in every time saneone uses language. These are
essentially three kinds of abstraction to be analyzed out of any
given use, any given social event of speaking (or its equivalent).
In effect, any given social event of using language carbines or
laminates the "acts" into one spatia-temporally manifested behavioral
interaction. That is, to use language is to engage in locutionary
acts, illocutionary acts, and perlocutionary acts all at once,
each of these being a way of understanding-and a way of analyzing--
the use of language in a single social event. Where do Austin 's
ideas cane fran about the nature of these "acts" and the "forces"
and "effects" they anbody? I would claim that we can see their
origin in the objectification that Whorf talked about I the projec-
tion of cryptotypic selective categorizations of lexical fonns in
the typical metapragmatic discourse of a language such as English.
Fran this objectification we can see projected various "forces"
that constitute "acts" of various types. Consider the set of kinds
of "acts" that Austin distinguished: "phonetic," "phatic,"
"rhetic, II these together making up the "locutionary act,"
"illocutionary act," and "perlocutionary act." I would claim that
each of these levels in effect results fran an objectification of
sane particular alternative way of reporting "what happened" in
an event of using language as a social activity. Each of these
"acts" corresponds to a particular metapragmatic form that reports
the event fran a particular point of view. Let us take these in
SUppose that, in our presence, a particular man utters to
a particular v.ana.n-here I try to duplicate the signal used in as
precise transcription as possible-' I [glqf±: l .' I This is certainly
not what would call an utterance of toe English language. But, if
asked, "What happened?" one acceptable way in which we could report
is describe the event as follCNlS: "He said to her, I [glqf;l,:] , . "
that we have described the event in terms of a particular
Agent ("he") engaging in a particular kind of activity with respect
some recipient/goal of the action ("her"). Notice, more
JJll.POrtantly, that we have described the activity with a construc-
"say , :" that quotes or duplicates the articulated
framed by, or mtroduced by, a specifically linguistic
framing verb indicating that what is framed was a noise camunicated
through articulate pronunciation.
The specific meaning of the verb (in sane systemtic
presentation of semnticity) does not concern us here' what concerns
us the fact that this verb occurs in discourse as framing
for a token representing actual linguistic material
quoted fran a particular speech event. If we may put this' into
the fraID8\\Ork of Whorf, the verb is a lexical form one of
cryptotypes of which is 'engage in speech
the resulting utterance-signal '. Austin
this ,meaning-category by declaring any so-describable
a phonetlC act." Its resulting signal as we have seen
the material that fonns the granmatical object' of the verb in the
metapragmatic construction, Austin calls a "phone." Thus a
"phonetic act" results in a "phone."16 '
, Now, let us imagine another evert of interaction. SUppose that,
our a particular man utters to a particular waran--
here agam I try to duplicate the signal used in a conventionalized
transcription-"I will buy a loaf of bread." If again asked, "What
happened?" an acceptable way in which we could report is to describe
event "He said to her, 'I will buy a loaf of bread' ."
ThlS descnptlon differs fran the first only in the particular
that, is framed by the verb say. Here, the particular
rephca consists, as Austin-r-1975:92) says, of "certain
vocables or words, i.e. noises of certain types belonging to and
belonging to a certain vocabulary in a certain construction
i. e. to and as conforming to a certain granmar, with
certun mtonatlon, & c." The framed signal, in other \\DreIs, is
a token of sentence of a particular language, formed according to
the gI'llllImtlcal rules of selection and canbination of meaningful
lexical units and suprasegmimtal features.
So _ a way in which the verb co-occurs with
other ImgulStlc m metapragmatic discourse. It frames a
signal that is an exact repetition of a well-formed linguistic
expression of sane language. We might say with Whorf that the
verb is a lexical form, one of the selective cryptotypes of
which is 'engage in language-specific speech activity with the
resulting utterance-signal '. Austin objectifies this meaning-
category by declaring any so-describable action a "phatic act."
Its resulting abstraction he calls a "pheme," that type which is
reproduced framed in the utterance-token describing what went on.
Let us continue with the consideration of this second inter-
action, and note that there is another way to report "what happened."
We might report, ''He said to her that he \\QuId buy a loaf of
bread. " We should focus our attention here on the forms that
differ frcxn the :imnediately preceding type of report. Instead of "I"
in the framed canplement we have used "he;" instead of "will," "would";
, d "
introducing this wOOle stretch of the sentence is the wor "that.
So the report starts out the same as the previous one (' 'He said to
her ... "), but then continues with "that" followed by a sentence-form
with change of pronoun and change of tense frcxn what was actually
uttered in the original social interaction. The changes are, of
course, systanatic, and distinguish what has cane to
"indirect quotation" in our inherited rhetorical traditlon,
opposed to "direct quotation." But note that we face the task of
having to account precisely for what "ranains the same" in these
t\\Q nodes of reporting a speech interaction, and what changes.
Briefly, and without treating all the details, we can
that what ranains the same in the "direct" and "indirect" quotatlon
forms is the propositional content, the value of speech as a
statement-about sanething or someone. In the fully "direct"
quotation, we frame the exact linguistically-describable
matical) form that was uttered, presumably thereby presernng
whatever propositional value it had. In the fully "indirect"
quotation, we frame a signal-form that is distinct fran the
original utterance, but conventionally to it in prop-
ositional value in this particular construCtlOn. Thus, there
are characteristic transformations in the overt forms of various
indexical categories which, in this construction-frame, IlR.ISt be
understood (indexically valuated) relative to those of the metaprag-
matic frame in order to appreciate the equivalence of original
message (in its contexhof utterance) and the reported OIE (in its
context of utterance). This use of framing constructions
would lead us to observe, with Whorf, that is a lexical form,
one of the selective cryptotypes of which is'engage in referring-
and-predicating linguistic activity with the resulting propositional
content '. Austin objectifies this meaning category by
declaring any so-describable action a "rhetic act." Its resulting
abstraction he calls a "rheme," that which is the specific propositional
value characterized in the framed report construction.
All three of these acts, according to Austin, are involved in
the canposite he terms the "locutionary act," the use of physical
signals organized into conventional words--and-sentences of sane .
language, to make propositions. '!be very same metapragmatlC
framing verb, occurs in three basically distinct granmatical
constructions, or potentially distinct granmatical constructions,
in the metapragmatic discourse describing the social action of
using a language such as English. As we would now expect, following
Wharf, the native speaker, here Austin, captures both the sameness
of the explicit metapragmatic lexical verb sten (hence uniting
all these into a level called "locution") and the cryptotypic
(or senantic co-occurrence) distinctions of the underlying gram-
matical categorizaions of syntax (hence distinguishing "phonetic,"
"phatic," and "rhetic" acts as subcanponents) by objectifying the
metapragmatic capability of his own language into a "native
theory" of the true nature of speech. Actual behavior "consists of"
these acts, according to the native view, and the technical theory
of "locution" describing then is the projection of this objectifying
process. Again let me emphasize that the correctness or incor-
rectness of the theory--or its utility, if one wishes--is not at
issue. '!he claim I make here is that it is not by chance, not
unexpected as the "unmarked" outcome, that the native theory
matches precisely the syntactico-senantic and lexical properties
of the metapragmatic di§course of the language under investigation
by the native speaker.
If we return to the interaction we have just described in
t\\Q different ways I we might add a third possible description,
"He pranised her that he would buy a loaf of bread," or even,
"He pranised her to buy a loaf of bread." Note in the first of
these that the framed construction is precisely the same as the
"indirect quotation," frcxn Vthich it would be easy to conclude that
pranising is really saying in the three senses noted above, plus
sanething else. That is, it is easy to conclude that pranise is
a hyponym of especially since it is both a metapragmatic
framing verb, as used here, and also a "performative" iten,
as in the usage, "I pranise to buy a loaf of bread." In the
latter, as we noted above, it seans to be central to the description
of the very action that is ongoing-which seans to consist of
talk. Without sane in-depth and systanatic study of the syntactic
properties of the two verbs, 20 a native speaker might rationalize
that the seaning linguistic hyponym relationship reflects a real
inclusion relationship at the level of Vthat has happened. One
might conclude, namely, that to pranise is to say sane formula
"rhetically" equivalent to .! pranise that ... in its performative
'!his is, ultimately, precisely the understanding that emerges
in Austin's notion of an "illocutionary act." Such an act is
projected fran an objectification of the metapragmatic usage that
masquerades as syntactic indirect quotation. The fact that this
metapragmatic iten figures in a fonnula of direct discourse, the
"explicit performative" formula, Vthere it is a creative indexical,
leads to the postulation of a special "force," "illocutionary force,"
that is seen1ngly located in the utterance of a token of the explicit
performative. Given Austin's concentration on lexical itens , in
particular on the predicates of the explicit performatives, it is
alnDst as if these lexical forms aJiJodied the "force" (a further
devel0Jlllent that, to be sureI characterizes the writings of many of
his followers wOO, understandably enough, search for loci of forces).
Thus, illocutionary forces are distinguished only insofar as distinct
explicit performative formulae can be recognized; in effect, this
reduces the study of function
to classifying all uses of particular
linguistic tonns by many-to-one relations with explicit pertonnatives.
Obviously, this yields no unique partition of fonns, nor has it resulted
in a coherent description, beyond a limitless profusion of "indirect
illocutionary acts" and such. I will return to this problan below.
First, however, let us return to the interaction about which we
have been answering the question, "What happened?" '!here v.ould sean
to be another response we might give, namely, a description akin to
"He made her happy." There is a \\bole class of such utterance-types,
to be sure, tor exalq)le, ''He pleased her," etc. The following ranarks
apply only insotar as these have sanantic-syntactic parallelism to
this tonn, that is, insofar as the description codes the speech event
as one in \\bich an individual A brings about (' causes t) sane state in
individual B (or any similar 'effect') as a result consequent upon
reaching a certain point of the event (here, the endpoint). There
is a cryptotypic systen of inherent ("lexicalized") aspect in English
(as in any language), which interacts with factors of understood cross-
clause co-reference (e.g., object-raising) and sequencing (e.g.,
relative tense-aspect). We might Sl.BIJlW'ize this granmatical
analysis by the schena; [A's engaging in sane activity up to a
certain point) consequently theretore [B's being in sane particular
state). One kind of 'cause' and 'eftect' is objectified tran this
kind of grarnnatical equivalence set, one in which purposivityI inten-
tionality, etc. are not invariably grammatically coded.
But it would seen that if it is possible to describe such an
event as A speaking to B by this kind of construction, which can also
be used for many other kinds of activities that A and B might engage
in, we can objectify the event of speaking as "causing sane particular
effect." Thus, the social actiVity of using language can be said to
bring about "perlocutionary" results; that is "the perlocutionary
... is the achieving of certain eftects by saying sanething" (Austin
1975; 12I);"" Whether thIS be "the acfiievanent of a perlocutionary
object (convince persuade) or the production of a perlocutionary
sequel" (Austin 21975; 118). Unfortunately, as even Austin himself
pointed out, none of the proposed "tests" to distinguish "illocution"
fran "perlocution" v.orksj and it seans obvious \\by not, given the
sufficiency of the schena above in suggesting the objectification into
"perlocution;" for example, whatever can be so predicated ot B in the
schena might be considered a "perlocutionary effect."
To be sure, ntllrerous internal contradictions have been and
might be found in this native pragmatic ideology, and numerous
"perlocutionary" sequels. There is no reason why an ideology that
grows piecaneal fIull various metapragmatic tormulations ot a language
should show internal consistency, nor indeed give adequate analytic
insight in areas of social practice relatively unsusceptible
to the kinds of processes we have postulated. My discussion here
has rather been directed to showing how these processes can explain
the vivid reality of native metapragmatic projections to the users
of a language. Through these "forces" and so forth the user can
rationalize functionl, the sense of conventional, goal-directed
purposivity in the social activity of using language. Where
functionl in fact merges with function2' as in the explicit perfor-
mative formulae, each felicitous use of \\bich is a creative
inde::,ical, of goal-situation, the ideological
reahty f1nds 1ts conf1nnation in social actuality. Fran these
paradi@ll situations, and their transparent functional trans-
into ideological tenns, canes the frequently or
1ncoherent account of typical native theory.
We might generalize this analysis to frame an hypothesis about
the precursors to the fonnation of an Austin-like native ideology of
as a theory of language use
found ill the configuration of linguistic structure and rneta- '
pragmatic discourse. We have seen that partiCUlar features of
the English (and similar) configurations sean to serve as under-
pinnings to particular objectifications. These have been;
1) a specifically linguistic, agentive metapragmatic tenninology
verbs), that is understood as a taxonany of hyponyms of (that
1S, of a specifically linguistic, or at least "camunicative"
verb), that allows at least partially parallel canplanent
. Observe that the various rnetapra@llatic construc-
t1<:>ns 1tself are in one-one relation with certain
obJect1f1cat10ns of "locution"; this might also be a generalizable
property. 2) a structural equivalent to "explicit pertonnative"
usage of at least a subset of the fonns inO}. This is
the zero fonn of an inherently-
perfechve .metapra@llat1c verb, together with appropriate first-
inflections/syntax, the \\bole schena
servrng as a creative indeXical. 'Ibis usage will render
as function
, facilitating identification of
purpoS1V1ty 1n these tenns, and grasping function as "illocution"
or such. 3) possibil:ityof equivalent descriptiorts of carmunication
with 'A c::use/make/do'-pIUS-'B effect/result' rnetapragmatic
constructlOns that do not specifically encode agency/intention of
t1;Je Agent of the verb 'cause/make/do'. 'Ibis pennits the distinc-
bet\reen conventional (intentionally puxposive) action-type in
actual (effected) action-token, thereby establishing
:; level It should be possible to find numerous
var1ations of , this distinction among various societies,
of direct agency (or conventionallY-codable
::-ntent:l.On) :unpl1ed in the higher clause, types of linkage of
cause' and 'effect' clauses, etc.
" . c;>f view of extending Whorf I s understanding of
proJect10n, lt 18 1nteresting that, just as trees or any objects
are said to be characterized by all the language-independent
catego::ies of attributes of scoops of sugar, all uses of speech
to be characterized by the categories of locutionary--
aspects that grow fran apprehending
exphc1t pertonnatives and similar metapragmatiC fonns' thus
locutionary "meaning" of grammatical construction
"force" of the construction used in a type of context perlocutionary
"effect" of the utterance by its occurrence. As noted the fact
of explicit performative facilitates the identification of
functionall distinctions, and contributes to the native sense of
multiple Nevertheless, explicit perfonnativity
the exJ.stence of certain metapragmatic configurations
an tile quaB1-propositional indexical (functional
) usage of than '
in certain contexts of discourse. Explicit perfonnativity is the
result of the surface-structural intersection of the metapragmatic
propositional function and of various contextualized functions
with specific functionlU
properties. This
surface-fonnation renders max1mally transparent creatJ.ve mdex1cal
(functional ) relationships otherwise difficult to appreheI.td: This
allows the strategic fo:nnulation of perfonnat1v1ty
as presupposed indexical (functional
) relationships of the surface-
structural fonn, plus the functional
\\bich, at an ideological level, properly applied, .
has its creative indexical effect .. It 1S no wonder
that in societies which hyper-ratlOnahze effect1ve means to
practical ends, such as our own, there is a constant historical
creation of new explicit perfonnative fomulae fran older metaprag-
matic means cast into appropriate verb or verb-phrase fonns
(cf. n. 16 and refs. there).
As a second exanple we might carpare and contrast the so-called
"linguistic etiquette" of Javanese, as an ideologic:a
rationalizing function] and as a funct10na1
systan of
The fonner aspect has neen very elegantly presented and g7ven an
analysis in Clifford Geertz' s Religion of Java (1960) \\b11e the .
latter aspect has been stressed by such writers as Elmor Horne, m
her pedagogical and reference works (1961; 1963; 1974), .
E.M. Uhlenbeck (1970) \\bo has discussed Geertz's presentatJ.on fran
the point of view of lingustic structure and actual li
and S. Poedjosoedarm:> (1968; 1969), \\bo brings hngu1StJ.C
description to his (otherwise native?) canmnd. A Class1C systan of
linguistic expression of social status such as those of
age, wealth, occupation, and of social distance: th1s <Xl?P
alternation of lexical fonns is only partly of inherent 1ndex1cal
significance, indicating status and/or familiarity. of
the user (speaker) and addressee, as will be descnbed. Yet 1I.t
tenns of the first fomulation, which. seans to be based essentJ.ally
on the interpretations of fairly highly-placed participants, in
using the systan of linguistic etiquette "one surrounds the other
with a wall of behavioral (lair) fonnality which protects the
stability of his inner life (OOtin)" (Geertz 1960:255), . according as
the addressee's position vis-a-vis the speaker danands 1t. In .
other words, the ideology of use of these fonns focuses on
addressee-oriented functional value of fonns, the use of Wh1ch. 1S
dictated by certain presuppostng indexical relationships between
speech and its social context. Ideologically, etiquette vocabulary
is ranked in levels that match the linear conception of social
stratification' functionallY2' there are many interacting systans.
all of \\bich. c:m be concentrated on speaker-addressee relationships,
but they haveIi'""ciiiPlex and sanetimes contradictory
quality Let us take the ideologically-based fomulatlOn of Geertz
first, then \\hat we can make of the rore analytic
data to hand. .
As we would expect, the first view locates the effectJ.veness
of language use in lexical fonns, particularly in. of ra.nke<!-
alte:mant \\Ords for the same referential-and-predwatlOnal meanmgs.
These llsets of linked conjugates" of lexical fonns, as Geertz (1960;253)
calls than, range in size fran one, where the referential meaning is
indifferent to status-ranking, up to three, or in sane cases five,
each alternant ranked with respect to the others. As Uhlenbeck
(1970: 442) observes, the Javanese metapragnntic tenr.inology for
this phenanenon is ambiguous:
The main tenns, namely KRAMA polite behaviour,
MADYA, middle, in between, and NGCKO, canparable as to meaning
and fonnal structure to French tutoyer, are handled in two
different ways. CXIe may say that one speaks in or uses KRAMA,
MADYA, or NGOKO, or that a sentence [Le., sentence-token or
utterance--MS] is in KRAMA, MADYA or NGOKO. In that case the
tenns refer to the speech style chosen by the speaker. But
one may also say of a word or a grlllJJmtical elanent that it is
Given the variation in nuni:>er of lexical alternants in the
paradignBof these etiquette itans, there is obviously no way of
assigning similar levels directly both to words and to the syntagmatic
canbinations in which. they occur. Thus, the native theory has
typically derived a fine partition of levels of (or speech
styIe) fran canbinations of certain kinds of ranked alternants,
subdividing the three basic levels into as many as nine varieties
of refinanent (Poedjosoedanno 1968; 59-62; Horne 1974: xxxii-xxxiii;
Uhlenbeck[ 1970: 443] reports eight), each appropriate as a means of
paying the right alIDunt of deference to the person of the inter-
locutor, based on speaker's social distance and relative status.
But I as Geertz points out and illustrates, not everyone in
traditional Javanese society controls the use of these lexical
conjugates to the same degree. Using the tenn "dialect" to refer
to a speaker's repertoire of alte:mant styles, he observes (1960:249)
A further crnplication is 1hat status meanings are cemnunicated
in speech, not only intent:!i.onally in tenns of word selection
within the speaker's dialect but unintentionally in tenns
ortiie dialect he uses as a whole. Not only are there "levels"
of speech within the dialect \\bich are ranked in tenns of their
status (or alus (refined] /kasar [coarse] ) connotations;
the various dialects in the-ooiiiimity as a \\bole are also
ranked in tanns of the alus to kasar spectrum, this latter sort
of ranking being characteristic-;-o!course, of any stratified
'!hus, the greater the numerical delicacy of alternant levels that
a speaker uses, as well as the higher peak on the spectrum of levels
that he achieves, the rore refined (and statusful) a persona he
creates for himself, associated with higher and higher segments
of Javanese society.
The analysis that Geertz offers, then, as sunmarized in the charts
in 6. reproduced fran his discussion (1960: 250-52) I distinguishes
the "dialects" of each of three typical groups (relevant to the larger
social analysis fran which the account is drawn) I each such dialect
being a repertoire of ranked divisions, made up of basic "stylanes"
(keyed by numbers) as IIPdified or 1.lIIlOCldified by "honorifics" (keyed
alphabetically). The basic stylanes are expressed fonnally by
fran one to three lexical conjugates for a particular understood
English translation-equivalent. "In sane cases the madya conjugate
is the san:e as the ngoka (e. g., Ian [' and' in charts I, II, levels 1,2]);
sanet:i.Ioos it is the same as the krama (e.g., samp€jan, selrul
['you', 'eat', 'rice' in charts I, II, levels 2, 3]); and of course,
sanetilres the conjugate is the san:e in all three cases (e.g.,
['cassava' in all charts, levels 1, 2, 3])" (Geertz 1960:253; bracketed
material added). There are additional lexical conjugates, however,
"\\hich occur independently of the first kinds of conjugates and \\hich
act to raise the level of speech indicated by the first, inevitable
[stylene] selection, one 'notch' higher--or, better, one-half notch."
Thus the level 3a in charts I and II I, and the level lb in chart I II ,
the former showing this "honorific" usage in addition to Krama,
the latter showing this "honorific" usage in addition to Ngoko.
However, there is one IIPre "honorific" usage, associated with
basic Ngoka stylene, and hence called by Geertz "low honorifics,"
"the use of krarna \\Ords •.. lifting ngoko biasa (level 1), to
ngoka madya (level la)," as shown in charts I, II, and III, for
the glosses 'you' [kowe I : saJl¥?€3an la] and 'eat' [ mangan I : la] .
This is an analysis based IOOre on ideologically-interpreted
functional, statements about examples such as the one illustrated than
on form-fU1'lction
covariation, since a number of puzzles energe, which
other authors (particularly Uhlenbeck 1970) have called attention to.
The illustrated example, as a question with a second person subject,
seans to concentrate all of the functional power of the lexical
alternants on the speaker-hearer relationship; this is a fonn
both addressed to and questioning a proposition predicable of
the addressee, the referent of the explicit subject. Further, how
can we understand the facts as shown in chart III, that for prijaji
speech the half-Ievel-raising honorifics occur with both basic
Ngoka (level I) and basic Krama (level 3), but that Ngoko also
has certain Krarna \\Drds used (in level la) without
raising the level fran I to It should be observed that these
are precisely the \\Drds which, in charts I and II, are seen to
span levels la, 2, 3 as their range of usage. We might consider
why these Krarna items do not raise level la above level lb in
chart III, that is, take it out of the realm of the Ngoko stylene.
And, though there appear to be certain specifically level 2
(Jdadya) terns shown in charts I, II, there appears to be, inex-
plicably, no "honorific" lIXXiification. To be sure, the "low
honorifics" (as Geertz points out, really just Krarna conjugates),
sean already to be used in expressing level 2; but what of the
"high honorifics" that constitute both levels 3a and lb? How can
this and the other asynmetries of the chart be accounted for?
Finally, as Geertz observes in a footnote (1960: 255n), "sanetimes
the status of a third person referred to, especially if he [sic]
be quite high., may detennine the fonn used: ...[ e.g., ] the high,
krama fonn of 'houseI \\hen speaking of the one the district officer
Iives in." The speaker-addressee focus of the system as presented
makes this seem sanehow extraneous. Presumably it is so in the


•• 0

o c
..0 •
c E
o 0
.. a.
c "
c c
o 0

c ,

" .!!.

" .!!.

" .!l



E ·0
o "
" .!l
Chart II
Level are you going to eat rice and cassava now Complete sentence
2 napa odjeng sekul saniki
Napa sampejan adjeng nedo seltul
Ian kaspe soniki? •
sampe-jan neqo
1a Ian kas"e
Apo sampejon arep n e ~ a sego Ian
kaspe saiki?
I- apa arep sega saiki
1 kowe mangon
Apa Icowe arep mangan sega Ian
kaspe saiki?
Cl1art III
Level are you going to eat rice and cassava now Complete sentence
30 pandjenengan dahar
Menapa pandjenengan ba"'e "'ahar
sekul kalijan kaspe samenika?
bade sekul kalijan kasptl sameniko
3 sampejan ne40
Menapa sampejan b a ~ ! e neqa seloul
kalijan kospe samenika?
Ib pandjenengon qahar
Apo pandjenengan arep 40har sega
Ian kaspe $oiki?
1a apa sompejan arep n e ~ a sego Ian kaspe saiki
Apa sampejan arep netJa sego Ian
kaspe saiki?
1 kowe mangon
Ape kowe arep mangan sega Ian
kaspe saiki?
ideological concentration on the function
of marking speaker
deference to addressee by use of higher and higher "honorific"
levels .diacritically within the basic dimension of speaker-hearer
familiarity ("intimacy" in Brown & Gi1man's [1960:257] sense)
signalled with the "styleme" distinction of Ngoko (for intimates),
Krama (for non-intimates). By this reckoning, it is clear \\hy
prija,ji would claim that within their systan, "the middle stylane--
considered to be vulgar--drops out" (Geertz 1960:254); for in
their view, it is the "honorifics" only that count, and the middle
styleme, level 2, uses none according to this model.
Persuasive as this might sean, since, for example, in many
respects Poec!josoedanno (1968) in part organizes a sanewbat fuller
linguistic treatment around these same unidimensional levels, it
seems that the native ideology of function
does not distinguish
annng several analytically distinct kinds of alternations. Fran
the second point of view, indeed, the example used in the charts
of 6. is particularly poorly chosen (though it is powerful and
revealing in the first perspective), since it conflates various
systens involving the relationship between actual speaker
and the adaressee, speaker and referent of various,
syntactically-definable noun phrases, etc. Additionally,
and perhaps for this reason, it does not display clearly the \\hole
range of fonnally-alternative types of linguistic units that can be
recognized in the broad functional area of 'status marking'. When
we speak of paradigmatic alternations of lexical fonns, there sean
to be no fewer than four special, marked subsystans that cont
with the basic non-polite, infonnal Ngoko lexical conjugates.
Two of these, Krama !Z§d Madya, function
to mark speaker-addressee
relations primarily, \\hile two, Krama Inggil and Krama func-
tion to indicate the speaker's evaluation of the (sanetimes
deference-entitlements of specific referent(s). (bserve
that these last two systens, \\hen applied to addressee as referent,
have the not unexpected effect of merging in function
Wfth the
first two systans. Fran a formal point of view, moreover, Madya
is a snaller set of itens than Krama, and in many cases is
phonologically derived fran it. Similarly, Krama An<;iaP is a
snaller set of itens than Krama Inggil, and the fonner are uniquely
limited to certain verbs that take Dative-like person-referent
objects, only with reference to \\hich do they to
express status. In many ways, the functional a.synm3trles follow
the same directionality: Madya and Krama Anc;lap itens take precedence
when they are to be anploYed' and the other two sets appear
residually. When all of these sets are considered in contrast
with Ngoko, the latter also then appears to be a residual set of
basic fonns, thus negatively characterizable.
Let us briefly characterize each of the functiona1
systens I
starting with those focusing on the referent. The most specialized
Of these is the set of ca. twenty Krama An<;iap fonns, all but three
of \\hich sean to be verbs of social interaction, typically and
in particular the social interactions that involve speech, and
that are directed to sane Dative-role person, such as invite,
obey, [against], and, most significantly for the sanantics of
this set, give [to]. They are, then, typically metapragmatic
verbs \\hich can be used in first-person (speaker) Agent con-
structions, and in second-person (addressee) Dative-object
constructions, in which carbined usage (with special (Krama Anc;laP)
pronaninal fonns, note) they would appear to be akin to t
perfonnative constructions. There is a many-to-one relationshlp
of Ngoko to Krama Ant;iap lexical foms, so that theE is, fran one
point of view, a kind of delicacy by euphanistic allusion, rather
than specificity of metapragmatic (and "illocutionary") import.
These fonns index the speaker's estimation that the referent of
the Agent of the verb engages "humbly" in sane social activity
directed toward a socially elevated person, the referent of the
Dative-object. This "huni>le" relationship is coded in the verb
stan (and sanetimas in associated pronouns) as a true metapragmatic
lexicalization of the manner of the pragmatic speech activity.
That it is so coded makes its objectified reality all the more
salient in the functiona1
anployment of speakers; this level has
all the transparency properties of the explicit perfonnatives of
our own language, indexing as well estimation of deferential
Agent-to-Dative relations.
The larger and residual set of referent-focused fonns, called
Krama Inggil, number about 250-260 (Horne 1961: 56-57; Poedjosoedarm:>
1968: 67; Uh1enbeck 1970: 449),
which morphologically as well as sanantically are rather
haoogeneous. Morphologically they are either nouns or verbs,
with the nouns in the majority. Sanantically they... are all
related to the human person. There are KRAMA INGGIL-words
for the human body and for virtually all its parts, for nearly
all its vital functions and for its products of excretion, for
birth, death and uneamnn illnesses. . .. for nearly all itens
of Javanese male and female apparel and for the things a
person nonna.lly carries with him or uses daily. . .. for
nearly all kinsfolk in the ascending and descending generations
and... for the most eamnn and general human activities such
as bathing, sleeping, sitting, standing, eating, drinking,
walking, thinking, speaking, asking, giVing, receiving, etc.
As Poedjosoedanoo notes, the Krama Inggil words are phonologically
unrelated to their conjugates at other levels, many being historically
borrowings fran Sanskrit or literary Old Javanese. Again here
we sanetimes find a many-to-one relationship of Ngoko to Krama Inggil
conjugates, though apparently not so extensive as in Krama Anc;laP.
FunctionallY2' the Krama Inggil vocabulary indexes residual estimation
of deference entitlement of referents, \\here for example Agent of
higher rank 'gives' CIa) sanething to Dative-object person, or \\here
a phrase referring to the body part or possession of such a (high-
rank) referent is enployed, or any other such appropriate syntactic
construction other than the specific case detennining Krama
usage. By a kind of residual logic, then, Krama Inggil canes
to express in verb lexanes the high-status position of Agents and
(Intransitive) Actors, and in naninal lexemes this evaluation of
Possessors and similar relative semantic roles (e.g., poles
of kin relationships). To be sure, this can apply also to cases where
the addressee is referred to, Le., in cases of functional overlap
where indexical reference and deference are to
the addressee (vis-a-vis speaker).
Both of these cases, that detennining KranR Andap and that
detennining Krama Inggil, are independent of the chOice of other
"levels" of lexical conjugates, such as Ngoko, Krama, Madya. And
indeed they cooccur with all three. What is especially interesting
is that in the first case, we have a set of metapragmatic itans
where the social indexical function presupposes the sanantico-
referential (and hence surface structure) identification
of the particular fonns which are subject to alternation. This
seans to be the case as well in the KranR Inggil fonns, where the
particular sanantico-referential danain, and IOOrphosyntactic
constructions involving the (residual) Agent/Actor or Possessor/
Relatum are presupposed by the choice of social index. Outside
of this presupposed danain, the indexical systan focusing on
particular referents of particular surface-structure categories
does not operate as such, and other strategies III.lSt be found, if
available, to implarent this same I stress this presup-
position of reference-and-predication, and the structures this
implies, to draw attention to the canparability of these cases with
the case of the theory of "illocution," etc. as it is an example of
tI'?.nsparency. In all of these cases, we may apprehend
the functiona1
systan by the way in which particular functions
be associated with surface lexicalizations identifiable through
their referential-and-predicational contributions. Characteristically,
as we saw in the first presentation of Krama Inggil (not specifically
separating KranR Andap), it is given as both "honorific" vocabulary
and "level" labeled'3a in charts I,III of 6., and called Krama Inggil
in the description (Geertz 1960: 253ff.); and, it is apprehended
through the specific function1 2 of marking speaker deference to
addressee-as-referent, in keeping with the overall ideology of the
use (function
) of speech for politeness in speaker-addressee
Turning now to the other sets of lexical conjugates, which
always focus on speaker-addressee relations, we might begin with
the set of specifically Madya itEm3, about 35, many of
which are a kind of attenuated Krama form, seaningly shortened or
abbreviated, and also special second person pronouns. '!be Madya
speech styles, however, "are not only characterized by the presence
of certain special elanents, but also negatively by the exclusion
of certain rn:KO-words," totalling about 50 lexical itEm3 (including
S(IOO prepositions and conjunctions). "In what is called MADYA, they
are replaced by specific MADYA-elarents if such elanents exist, if
these are absent, by their KRAMA-counterparts" (Uhlenbeck 1970: 451).
Additionally, as noted by all writers, certain Krama affixes are
avoided in so-called Madya-style speech, and Ngoko equivalents are
It is thus seen that the Madya style of speaking is basically
negatively defined, and IOOreover, of relatively uncodified charac-
ter. On the one hand, as Uhlenbeck notes, "the process of abbreviation
[of Krama elanents seen in Madya ones] has the important function
of making respect fonns less formal and of putting the relation... on
a IOOre intimate and confidential footing... ," if we take the
process as an iconic metaphor vis-a-vis full Krama function
. On
the other band, Madya as a style, or, as Uhlenbeck (1970: 452-3)
notes, set of styles, leaves the speaker "relatively free to
regulate the mmoor of KRAMA-elarents in the sentence [-token] ,
thereby varying the measure of respect he wants to convey towards
his speech partner." It is no wonder the prijaj i, for whan precision
is all in the strategy, deny using this imprecise,
negatively-defined, ccrnpranise level of 30
'!be Krama set of lexical conjugates, numbering about 850
(Poedjosoeclanro 1968: 64), are enployed not only in so-called Krama
speech stylas, but, to varying degrees, in Madya styles as \I.e11 .
Being functionallY2 specific to speaker-addressee relationships, use
of all the Krama possibilities of lexical conjugates(including
certain affixes) constitutes the various Krama styles, while, as
we have seen, the Madya styles have varying usage of Krama words
both residually (when, avoiding Ngoko, no special Madya conjugate
is available), and sporadically. This latter, sporadic usage of
Krama conjugates, is said to index degree of status difference by
density of occurrence of Krama i tEm3. While sane Krama i tEm3 do
not resanble the Ngoko conjugate in form, nnst sean to fall into
numerous derivational subgroups (Krama derivable fran Ngoko) , and
apparently for the prijaji it is a source of S(IOO lII1I.\sarent that
I!l€!li>ers of other groups hypercorrect, and analogize non-standard
Krama derivations, etc. (Geertz 1960: 258-59; Poedjosoedarmo 1968:
66; Uhlenbeck 1970; 457),31 in their efforts to speak in the alus
As should be clear at this point, the native perception is
that there are linearly-ranked levels or styles of speech, each
of which has a fixed position in a systan that functionsl to
maintain the correct aroount of deference and distance of speaker
to addressee. '!bese levels are characterized by various canbinations
of the vocabulary and affixes just reviewed, when fixed in
focus upon the speaker-addressee relationshiP':" As we
have seen, this functiona1
focus is accessible either directly
through function
, for sucfi lexical sets as Madya and Krama, or
for such a quasi-performative set as Krama Ant;iap, or the func-
tional focus is accessible indirectly through the presupposition of
referefltial identification of the units of alternation, as for
both Krama Anqap and Krama Inggil.
'!bus, in the schena of nine speech levels, the highest is
called Muda Krama ("young Krama"), and "consists of KrOOO affixes,
KrOOx) voc:ibulary, and KrCm3 Inggel words [including syntactically
appropriate Krama Andap] to denote the person, possessions, and actions
of the addressee" (POedjosoedar'mo 1968: 59). 'Ibis constitutes
GeertzI s level 3a on the charts given in 6. '!be second and third
levels, called Kramantara ("equal Krama") and Werda Krama ("old Krama.")
respectively, use the Krama conjugates without KrWm Inggil
indexically focused on the speaker-addressee relationship; they
appear to differ only in a pair of suffixes. 'lbese correspond to
Geertz's level 3. '!be fourth level, called Madya Krama (''middle
Krarna") is like the fifth just below, but arploys addressee-directed
Krama Inggil. '!be fifth level of nine is called Madyantara ("equal
Madya") , and consists "of Ngoko affixes and Mady6 words (Kr&o
words in the absence of Mady6). Kr&ID Inggel oords are not used
to refer to the addressee" (PoedjosoedanID 1968: 60) . This would
seem to be Geertz's level 2. The sixth level, called Madya
Ngoko ("sani-Ngoko"), is similar to the fifth, except that "a few
Ngoko oords may be substituted for either Mady6 or Kr<m5 ones, but
which oords will occur in Ngoko cannot be predicted. The lower the
status of the addressee, the rrore frequent the Ngoko oords will be."
The seventh level, called Basa Antya,is basically Ngoko with Krama
Inggil fonns focused on the addressee, plus "occasional Kraro
oords" where the higher the status of the addressee, the rrore
frequently these are said to occur. '!his is not quite the same
as what Geertz labels la or lb, because both Krama Inggil and Krama
forms will occur, whereas Geertz seems to consider these forms as
alternatives in the basic Ngoko "styleme." PrestmJably, we should
take this as akin to Geertz' s la, since the eighth level, called
Antya Basa, is precisely his level lb. '!he ninth level, finally,
called Ngoko I..ugu ("plain Ngoko"), consists of straight Ngoko
fonns, "except, of course, for KrCInO Inggel oords in
to a respected third person" (Poedjosoedanno 1968: 61). Except
for the last provision, this corresponds to Geertz' s level 1.
'!hat is, even if Krama Inggil occurs with respect to a non-addressee
referent, this does not raise the "level" of the utterance, since
"levels, " as should be very clear, are based on the particular
rationalization afthe function
of speech that makes speaker-addressee
indexical functions
the basis of an ideology of how and why the
systEm oorks.
Just as in our first extended example, then, the ideology
of linguistic etiquette presents a pragmatic systEm organized in
tenns of presupposed dimensions of the context of use, including
rrost importantly the central role of referential-and-predicational
(propositional) structure. Either directly, through the functiona1
alternation of individual "isosernantic" Ngoko--Krama--Madya
lexical conjugates, or indirectly, through the rootapragmatic
functional1 lexicalization of one fran arrong the number of
possibilities of Krama Anqap and Krama Inggil, the
ideology is an understanding fashioned fran a particular point
of View, one that constitutes the functional
systEm as a reinforce-
ment of socially real relationships of speaker and addressee, that
is, as reflecting the structure of a situation, for which speech
of a certain "level" is appropriate. The functional appropriate-
ness of a particular level is thus not to transfonn, but to
"regularize stimuli so that they will not puzzle, shock, or sur-
prise. . .. one should provide an ordered picture for others so as
not to upset than" (Geertz 1960; 248). This means being sensitive
to the givens of the situation, in which one pays attention to what
would be expected by interlocutors of a certain type. '!he dynamic
character of the use of etiquette, as noted in n. 21 and as anphasized
especially by Poedjosoedarmo (1968: 77-78) and Uhlenbeck (1.970:448;
455-56), its taJPOral course as a creative, strategic systEm in use,
is hardly anphasized, nor indeed the indexing
(which Geertz called "unintentional") with respect to the speaker,
again a potentially creative aspect of the use of these fonns. Even
in the case of the Krama An<;lap lexical conjugates, seaningly rrost
like our own explicit perfonnatives in origins, it
is not any transfonnative "illocutionary force" that is objectified
in the native ideology, so much as the situationally-appropriate
"humble" perfonnance of an action with respect to a higher-rank
person, elevated into a function. Indeed, it is no wonder that
in Krama Andap, there is little of distinct kinds
of action; these lexical conjugates focus upon the manner in which
the action is performed, large numbers of distinct Ngoko action-
verbs sumnarized with a single fonn, which serves for all.
Finally, as an aspect of this phenanenon that will lead into
the last section of my discussion, one should note the trends in
historical renewal of the vocabulary levels. Fran Poedjosoedanno's
(1968:73) point of view, this ")dnd of change is always in a downward
direction--that is, KrCmS Inggel becares KrOnJ, or 'Kr&D becares Ngoko,
or sorretimes KrCmS Inggel even becorres Ngoko." It is clear that
this is "downward" change only in tenns of the ideological construct
of speech levels, not in tenns of the sets of lexical conjugates.
For Krama Inggil: [its absence] is functionallY2 different fran
Krama:Ngoko as a lexical opposition; the former is in the plane
of referent-focused function
, and the latter, addressee-focused.
Thus, the changes are emerging fran the functional
restriction of
older Krama Inggil to this latter systEm. Only within this rrore
restricted ideological equation of vocabulary conjugate sets and
speech levels does the directionality of change make sense: those
levels with Krama Inggil itEmS functionallY2 focused on addressee
are, ceteris paribus, higher than levels with only Krama (always
perforce functionally focused on addressee), and those with Krama
higher than those witft Ngoko. So the historical change reflects in
an interesting way the ideological inflection of function of the
etiquette systEm, rather than the functiona1
structure. lparticularly
the renewal of both Krama and Ngoko fran Kraiiia Inggil is in effect
functional narrowing of the latter, parallel to narrowing of
"rooaning" ti.e., sense) in the restricted functional plane of
sanantico-referentiality. I will return to the paralleliBII after
I briefly present my final example for discussion.
The last example, or rather class of examples, is too well
known to need extended descriptive treatrrent, for there is a large
literature on the pronaninal alternations in European (and other)
languages that are functional systEmS of speaker-addressee defer-
ence and intimacy marking. Such. "T' (for French tu, German du,
Russian Ho, etc.) VB. "V" (for GeItl'lllilSie, Russian vi,
etc.) alternations have been described in a nunber of sociolinguistic
studies following upon Brown & Gilman's (1960) pioneering and
now-classic paper, e.g., Friedrich 1966 on Russian, Ervin-Tripp 1971
canparatively, Bates and Benigni 1975 on Italian, Paulston 1976 on
Swedish. 33 (I have discussed the universal functional charac-
teristics of such systEmS in S11verstein 1976b: 31;
Basically, the very same pronaninal surface fonns (or the cor-
related verb inflections) that differentiate singular VB. plural
reference, or second person (addressee included) VB. third person
(other) reference, or both of these, in this systEm to
differentiate various relationships of speaker and addressee, along
FlgUl'e 7. The two-dimensional semantic (0) in equilibrium and (b) under tension.
social dimensions that Brown & Gilman surnnarize as "power" and
"solidarity. "
As is shCNlIl in (a) of their figure reproduced in 7., there is
an older tv.o-diIrensional system of usage at a stage in the history
of each of the European languages, in which the asynmetries of
"power" (essentially deference entitlanent) are indexed by asynrnetries
of using T vs. V pronaninals; one defers to "powerful" alter with
V address, and inversely, alter replies with V address to ego. '!he
other diIrension, "differentiating address aIlPng power equals" (1960;
258), and hence sho.vn only in the central area of the figure,
is a synmetric use of V for non-"solidary" (or nonintimate, nonfamiliar)
linguistic address, and T for "solidary" (or intimate I familiar).
In such a systen, there :Ls a priority to the "power" diIrension for
defining pronani.nal usage in a social intMaction, and "solidarity"
is thus the residually indexed diIrension.
Brown & Gilman seek to explain the evolution of the pronaninal
systens of the various European languages, in which, to different
degrees, there arise new systens with "solidarity" as the prior or
only diIrension of social of the pronaninals (1960;259):
Well into the nineteenth century the power senantic prevailed
and waiters, canoon soldiers and anployees were called 1: while
parents, masters, and elder brothers were called y. How-
ever, all our evidence consistently indicates that in the
past century the solidarity semantic has gained supremacy .
. . . '!he abstract result is a siIrq:Jle one-di1oonsional systen
with the reciprocal T for the solidary and the reciprocal
V for the nonsoli<:1arY.
Noting that "the diIrension of solidarity is potentially applicable
to all persons addressed," they use a disequilibrium 100001, as.
shown in (b) of our figure 7., to explain the fact of change, If
not its directionality. As represented in the figure, the dimen-
sions of "power" and "solidarity" are essentially independent
in this IOOdel rather than hierarchiea.1. '!he disequilibrium follows
fran this, inasmch as an addressee represented by the category
at the upper left denands at the SllIOO t:iJre reference both with
"superior" V and with "solidary" T; inversely for the addressee at
the lower right of the figure. So it is the tendency to independence
of these social diIrensions, and the resulting disequilibrium of
function , that leads to the collapse of one of the diIrensions--in
all cases, "power"-and the preservation and even functiona12
inversion of the other (the "solidary". T being the unmarked usage,
according to Brown & Gilman.)35
In a IOOst interesting section of their paper entitled
"Sanantics, social structureI and ideology," however, Brown &; Gilman
reflect on this IOOdel (1960: 267);
It is possible, of course, that human cognition favors the
binary choice without contingencies and so found its way to
the suppression of one diIrension. However, this theory does
not account for the fact that it was the rule of solidarity
that triUIIPhed. We believe, therefore, that the developnent of
open societies with an equalitarian [sic] ideology acted
against the nonreciprocal power semantic and in favor of
solidarity. It is our suggestion that the larger social
changes created a distaste for the face-to-face expression
of differential power.
And indeed, when we examine the questiQl in this light, it would
appear that a disequilibrium IOOdel is sanewhat unnecessary (aside
fran fact that it :Ls not clear such a nodel as in Th. could
represent actual functiona1
norms, rather than areas of undefined
potential aniJiguity, at any stage of history). For, it seans to
roo that, properly put, the question is, how does ideology engage with
such a systen as that in figure 7a, so as to change it. How do the
users of a functiona1
systen explain the way their language has
effect in context and the way it to
effect? It is my contention that I as the citations
fran partisans of linguistic reform (Brown & Gilman 1960; 264-66)
show, functiona1
structure :Ls understood and represented in roota-
pragmatic ideology as a kind of rootaphorical transfer (analogy) fran
the structure and asynmetric synbolism of senantico-referential
categoriesI particularly as EfIixx:lied in lexical form.
(])serve that in the case of pronaninal forms, there is a
straightforward indexical-referential plane of function
, determining
Superior and
Superior and Not
Equal and Equal and Not
Solidary Solidary

Inferior and Not Inferior and
Tt V
Equal and Equal and Not
Solidary Solidary

categories of and there is a straightforward referential
category of 'mIllDer'. In the third person, the basic or
"unmarked" person of the set, the category of 'number' is frequently
and basically a senantico-referential category (as opposed to the
always-indexical total vs. partial "enumerability" entailed by
first or second person naninal categories ) . For the third person,
imreover, the 'singular' is unmarked and the 'plural' is marked
(referentially detenninate or specific). In the first- and
second-person categories, the totally emnnerable fonns (implying
I singular' ), so-called "first person singular" and "second person
singular," are the marked members of the number opposition. Yet,
it wuld seem that typical metapragmatic ideology grasps the social
of the T ; V opposition as metaphor of the function
the thira person (senantico-referential) opposition 'singular' ;
'plural', to describe one vs. many objects. Functiona1
categories are rationalized in this way, literalizing the perceived
metaphor of referential categories: " ... 1' esprit de fanatisme,
d 1 orgueil et de foodalite, nous a fait contracter I' habitude de
nous servir de la seconde personne du pluriel lorsque nous parlons
un seul," a 1793 speech of Malbec is quoted by Brown & Gilman
(1960:264); similarly they cite early Society of Friends literature
fran the mid-seventeenth century "arguing that T to one and V to
many is the natural and logical fonn of address-in all la.ngu"iges"
(1960; 265), hence in English.
The logic of these ideological views is simply that language
use is a reflection of the context, and, the addressee being a
single object to-be.--referred-to (the function
of the pronaninal
then), the pronaninal used should reflect this equality of
ones. When this becares the daninant accepted ideological fonnation
of a particular group, the of the pronaninal systEm has
changed. For, the "solidarity" of adherence to this literalization
of the metaphor is what detennines IlDltual (i.e., reciprocal) T
usage; and the social of the systEm of pronaninals -
indexes the triumph of this ideological solidarity within the
I have been sketching one outcane-the one erphasized by
Brown & Gilman-of the rationalization of pronaninal function
in tenns of an egalitarian value systEm. This "loss" or
(ct. n. 35) sutmergence of the "power"-base of pronaninal
has characterized the continuous linguistic history of the
daninant leading sectors of the Ellropean nations of the Continent.
Hence, the innovative usage has tended to become standard. But
there is obviously another possible outcane of the very same process
of ideological engaganent with the functional systEm, as shown
by the history of our own language, English. result is the loss
of the referential category itself, as has happened in the second
person of the standard language. To the extent that adherence
to the ideologically-specified innovation of unifonnly using T
indexes "solidarity" within a particular group, avoidance of the
innovation (i.e., avoidance of use of the !. fonn) is the only
to differentiate oneself fran the particular innovating group.
And indeed. the category of number has disappeared fran the English
standard language, not haphazardly (it is the claim here), nor by
sane inexplicable tendency for "polite" fonns to be generalized ("polite"
of speaker? of addressee? of "context"?). I think it is the outcane
of a different, negative valuation of the ideological grasp of
the previously-functioning
pragnatic systEm, changing the standard
language secondarily.
Qlr purpose in this part has not been linguistically to account
for the rise of egalitarian ideology in general, nor to make a
claim about the effects of propositional linguistic structure on
"world view," as should be clear. The purpose has been to examine
a particularly interesting historical linguistic change (or set of
historical changes of parallel and interacting character), explan-
ation of which SeEmS to implicate precisely the same function -ideology
relationship we have been stressing in generalizing fran Whorf' s
insights about reference to the first tw pragmatic examples. Here
again we can discern the centrality of surface lexical fonns (pronouns)
that can be defined in tenns of categories of reference (person, mun-
ber), of which other pragmatic dimensions ("power," "solidarity") are
taken to be metaphorical transfers, with the markedness relations
similarly transposed (singular lU] ; plural [M] ;: solidary; non-
solidary). Here again, the ideological fOrIlDllation deals with the
presupposing functional
relationships of the systen of fonns,
rationalizing the opposUion of surface fonns in tenns of "one
form--one function1--one ; thus, tutoyer
" vs.
vouvo¥erl="vouvoyer ," etc., m tfie natIve apprehension of the
systen metapragnatic constitution of a functional 1
Ideology engages metapragmatically with functional
through the constitution of a referentially-based (or referentially-
cente:ect) functional 1 systen, constitution of which is a function
changIng (hence, structure-changmg) process. Our exanples of Englisn
"speech acts," Javanese "linguistic etiquette," and Continental
European "pronouns of power and solidarity" have all d.aJDnstrated
this phenanenon, different though the degree of analytic accuracy
in the ideological apprehension of the structures involved. In
the first case, it was clear that insofar as metapragmatically-
derived function, is in one-to-one relation with i.e.,
for the true so-called "explicit perfonnatives " the ideo!ogy of
"illocution" and are rmtually This was only
partially true of the ideOlogy of politeness in Javanese and the
anal:(sis of itS.functiona1
structure; here, only part of the
machmery of etlquette shows metapragmatIc function
the Krama Inggil/An<;IaP lexicon. But sigIl1ficantly,
I t SeEmS to be that part rn::>st central to historical change and
renewal of lexical sets, i tens from this conjugate danain renewing
other ones, for example. And in the case of the T/V pronaninal
usage, the only mechanien for rationalizing a is the
creation of a metaphorical basis in the systen of reference--a meta-
phorical basis in 'singular': I plural' "muneration," as Whorf
[1941] 1956 would rEmind us--in tenns of which we can calibrate
our understanding of Only by this process we have changed
what function
is, either by fuming creative indexicality into
presupposing mdexicality (a minimal disturbance), or by increasingly
severe changes up to defining a functiona1
category out of existence.
Perhaps reconsidering the nature of the three examples we have
looked at will show how this is a generalization of the principles
that Whorf proposed for the plane of reference. In all three
exlllllJles, we considered the degree to which a native metapragmatic
discourse in terns of function of linguistic forns matches or
fails to match a "scientific" tunctiona_1
analysis of these forns.
We also considered, particularly with thB last exlllllJle, the
consequences over time of this disparity in grasping the functional
systans, or rather of the particular way in which an ideology of
function is the necessary (and, empirically, Ubiquitous) crnponent
of change of functiona1
functional systan, what I have tenned earher the sananhco-
referentiaf granmatical systan, the categories of which are not
indexical. Whorf pointed to the historical process of folk
etynvlogy ([1942] 1956:261-62) as a deroonstration over time of
his principles of projection and objectification. Such a process
rationalizes fonnerly WlIlX)tivated lexical forns into rore rotivated,
referentially-transparent surface constructions, such as sparrow
grass fran fonner asparagus (or, rather [splil r4g(r )as 1, via the.
understanding is 'grass for sparrows'. or Welsh rareblt
fram Welsh rabbit, via the literalization that this is not made
with 'riiEbit' as an ingredient at all, and hence nust be the
(frequently haJDphonous) sequence of rorphanes shown. Whorf also
pointed to the constant creation of "lexations" for objectified
cryptotypes (for exlllllJle, in the passage [ms. 1936/371 1956:81
quoted above), allowing the philosophically and scientifically
inclined users of a referential systan to discourse about these
objectifications while intending to discourse about "reality" as it
appears fram the phenotypes of the referential systan. (Recall again
Carnap and M:>rris here.) Hence, words like causation, gender,
etc., which can never--because of their own cryptotypic categories--
"mean" exactly the same thing as the granmatico-sanantic cate-
goria> they are based upon. In both these areas, it seans to ne,
we have parallels at the level of linguistic function, of pragmatics.
On the one hand, we have the tendency, whether by functional
extensions of metap:ra.!1Jllll.tic referential lexical itans or construc-
tions, or by functima11literalization of function
constantly to rational1ze functiona1
in tne image of
the referential-and-predicational systan. On the other hand,
we have the difficulty that this rationalization through a
apprehension of language can never be the same as the functlOnal
effect of language for which it fOmJlllates the conditions of
strategic (or typical) use. Creative effe<;:ts
particular are lost to functional
wh:ch 18 charac?-
teristically fOmJlllated in terns of presuPPOSlOg funchona1
of speech to context
This parallelian has implications, it VoOuld sean, in three
ways: first, for the nature of historical change as a J?rocess;
second, for the synchronic investigability of language use; third,
for the thane with which we began, how Boas justified the "ethnologic"
relevance of language. I will nerely adUIlbrate a full discussion
of these points here, since my presentation has been in terms of
what I would call characteristic exemplars, not in tenns of ccmparison
carefully controlled by variables of an hypothesis.
If we consider the kinds of historical regularities that
Kury,lowicz ( [1945/491 1966) fOmJlllated out of an older notion of
"analogy" together with the changes studied here, we find that they
have II1lch in caJJlX)n. At the roost obvious level, his second law
(1966: 164-65), that granmatical restructurings "suivent la
direction: fonnes de fondation fonnes fondees, dent Ie
rapport ctecoule de leurs spheres d I emploi," covers, for propos-
itional linguistic structure, the cases we have seen of II1llti-
formations imposing their forns on roore restrictedly
functional formations. Kury,lowicz' s is a structural "sanantic"
(in the ola, wide usage) basis for formal restructuring, covering
changes fram unmarked to marked manbers of an opposition, fran
general to restricted. 'Ihe parallel direction for the indexical
categories in which we are interested is fran referentially-based,
lexicalized pragmatically presupposing (thus, subject to functional
apprehension with identical or near identical netapragmatic forns)
to functionallY2-independent, nonlexica1ized pragmatically creative
ones. Again, oBviously his third law (1966: 165-69) gives a
surface-structural criterion for the direction of reshaping, because
"une structure consistant en membre constitutif plus llHJilre
subordonne fonne Ie fondElllentdu membre constitutif isole, mais
isofonctionnel. " Essentially this deals with the tendency for
Saussurean "relatively rootivated" constructions to determine the
reshaping of dependent "relatively arbitrary" ones. In the larger
functional sphere, the surface-structure linearization of a func-
fonn that metapragmatic ideology provides in a
discourse about language use, serves as the "fondanent" of historical
change in the function
and shape of the dependent fonn. With a
rore careful representation of the extended exanples, 1 think we
could see the parallelians and generalizations of Kury,lowicz' s
first, fourth, and fifth laws as well, as central tendencies
in the functional -structural history of languages.
If this is case, then I would suggest that granmatica1
change is of a piece with functional
-structural change roore
generallyI and that it is plausible to see the same mechanians at
work, an integrally dialectic involvanent with linguistic ideology
in its specific, metalinguistic expression. To what extent this
is an individual phenanenon, and to what extent social (here, that
is, institutionalized as a way of apprehending the functional
nature of language) presumably detennines saoothing about the
extent of the inevitable "change" resulting fran this dialectic. 40
But the principal point here is that the kind of granmatical
change or restructuring in the (sanantico-) referential systan is,
it would sean, the outcare of the same kind of process as is at
work in the pragmatic systan.
But if to rationalize, t(' "understand" one's own linguistic
usage is potentially to chang", it, precisely because of the
inevitable distortion of functiona1
what does this imply atlout the nature of granmatica1 description?
On a micro-scale, there is a dialectic process in our caning to
grips with any (and every) event of using language. And the processes
of systamtization, regimentation, and seeing the rule-governed
"underlying relationships" of linguistic forns--which we demand
always of our sources of data, especially ourselves--would seem to
be capable of destroying the very data themselves: '!here is obviously
a hierarchy of "elicitability" of kinds of functiona1
through seeking judgments about them, and one would hOpe, with
Wharf, that this caution can be developed further in the rrost
precise (though negative) terns, yielding a scale of confidence in
(or, rather, distrust of) the accuracy of structural data.
Linguistics has gone through a period of intense "methodological"
concern which focused upon the transcriptions of forms as data,
and logically-reconstructed canons for justifiable analysis.
There is, unfortunately, little methodological concern at present,
all the rrore unfortunate, it should be realized, because of the
strivings to becane an actual social science studying "pragmatics"
of language along with semantico-referential structure. Pragmatics
cannot be done in a principled manner until the "Whorfian Paradox,"
or Whorfian doubt--vs. Cartesian certitude-is faced squarely.
This starts fran seeing language as of the same "cultural" order
as the rest of social life.
Thus, it is ironic in a sense how we are led back to seeing
language as a "paradigm" of things cultural. It is not as Boas'
epitcme of systematic social determination of the unconscious
cognitive categories carrron to the manbers of a society, his
"primary ethnologic phenanena." It is as a strikingly clear
and revealing area for derronstrating the formation of what Boas
called "secondary rationalizations," what we would call an ideology
of lingustic fonn and function. With language, that is, we might
jcane rrost clearly to fonnulate the social scientist's dilemma of
"structure" vs. "action" in a never-ending historical rrovement.
If "structure" is a set of (fonnalizable) patterns according to
which "action" (contextually-situated social behavior) is inter-
pretable, a so-called synchronic statement (or rrodel) of "structure"
tells us in what respect "action" remains the same within a social
system, in what sense discernible instances of social behavior
remain "the same" action. What we find, however, when we attempt to
apprehend everything in such structural terns (here we return to
Wharf's theme of "indeterminacy"), is that plus c'est la mane chose,
plus s:a change.
*1 acknowledge gratefully the support of the John S1rron
Guggenheim MeJrorial Foundation during the preparation of this
version of the text. Earlier or excerpted versions were delivered
at Brown University (Charles Colver Lecture Series, 10 NoveIDer 1978),
Johns Hopkins University (History of Ideas Seminar, 5 Decanber
1978) I Washington Linguistics Club (6 December 1978), and
Temple University (Anthropology Department Colloquium, 26 January
1979) . The reactions of those audiences--as also of the audience
at the Chicago Linguistic Society--gave me the stinulation to attempt


yet again to clarify the issues raised. Erving Goffman is not
responsible for my errors.
lone should thus not be startled to see its camPn appear-
ance in the works of linguists. No fewer than two articles in
a single recent issue of Language (vol. 53, no. 2) begin this
way. see Morgan 1977: 277 and Bickerton 1977: 466.
2By "scientific" study of language, I intend to include
all canparative (cross--linguistic) systematization grounded
in formal-·functiona1 analysis, an attribute, it would seem,
of all relatively successful approaches to date, regardless
of practitioners' rhetorical stances as being formal or struc-
tural vs. semantic or functional. Thus, any description of a
single language implies corrparison of systenJ3 for justifica-
tion of the very terns of analysis.
3In a real senseI Whorf united the two important lines
of work. On the one hand, his discussions of the nature of
linguistic categories make analytically precise the concern
of Sapir for "fo:rm-feeling." 'Ibis is not adequately explained
(indeed, it is left alrrost mystical) by the system of linguis-
tic categories or "concepts" of Language (Sapir 1921:ch.5).
On the other hand, Whorf's clarification was possible only by
employing the analytic niceties that derive from Bloanfield's
system of granmatical description in (1933: ch. 16), the
tenns from which appear frequently in Whorf' s technical writ-
4See now Alford 1978 for an exuberant preliminary expose
of SClIll3 of the discrepancies between Whorf' s writings and
the claims of his critics. The expose itself lacks perspective
on certain points, however.
as indicated, the original, posthUllDus publica-
tion date was 1945 (in Language), editorial footnotes in both
the original and reprinted editions give 1937 as the date of
writing, which was done "at the request of Franz Boas," with
a view to publication in IJAL. 'Ibis is of importance primarily
to show the priority of Wharf's position as linguistic analyst
and theorist with fully respected professional credentials,
in the light of which all his late popularizations must be
6nro points should be underlined in this connection. First,
for Whorf the covert idea or cryptotype can only be "rrore or less
duplicated in a oord and a lexical concept," i.e., a lexical item
of sane meta1inguistic sub-part of the language in question. This
is because each such metalinguistic lexical item itself has
both phenotypic and cryptotypic aspects of its meaning, hence of its
referential valueI which can never be precisely the same as the
cryptotype out of which it was precipitated. Second, as to the
proportional sets of words that illustrate a cryptotypic category,
the advance of syntactic analysis to the refinaoont called "gen-
erative semantics" would require that syntactic constructions be
included in the proportion, e.g., ron : cause to ron :: ... (with
all the surface bracketting that this i.rrq>lies)-:- Specifically fran
such putative proportions of meaning involving both lexical items
and gramnatical constructions, has care the rediscovery of the
principle of the sanantic harogeneity of "lexical" vs. "gramnatical"
!reaning, and hence the representation of pre-lexicalized abstract
underlying fonns. It has often been remarked that the "abstract"
and "underlying" elements generally look like and share certain
essential properties with surface lexical items, for exanple, the
"abstract" predicate CAUSE and the surface lexical This
is, of course, no accident, given the logical fonn of proportions
lying behind the analyses in the first place, as illustrated.
However, precisely the same ultimate differentiation of abstract
underlying vs. surface lexical element is insisted upon in modern
theories, as we see in Wharf's differentiation of the scientific
vs. the native's rretalinguistic lexation.
There are t\\O rrore Whorfian types of category, which I
sketch here for the sake of First, there is the
distinction between specific vs. generic categories, that is
"an individual fonnal class existing in an individual language"
vs. "a hierarchy fonned by grouping classes of similar or (and)
canplementary types" ([1945] 1956: 100) . Thus' past tenporal
reference' is a specific category of English, while 'tense' is
a generic category that hierarchically subsumes the pair of
categories 'past I and 'non-past.' Second, there is the distinction
between selective vs. alternative isosenantic categories, that
is, necessary (obligatory) purely fonnal distinctions in words
vs. optional purely fonnal distinctions, neither of which has any
effect on (referential) meanings. Thus, conjugation markers of
French verb classes are necessary fonnal distinctions with no
contrastive effect on !reaning, hence, selective isosenantic classes.
Optional differentiation of special latinate or Greek plurals
(as opposed to English ones) in certain learned words in English,
e.g., vacua; vacUtmlS, schenata : schenas, indices: indexes, etc.,
has no effect on the referential !reaning of 'plural', and hence
learned plurals are an alternative isosenantic category (to be
sure, with pragmatic effect as indices of speaker status, etc.).
These t\\O category types are of lesser i.rrq>ortance to Whorf' s cen-
tral than the t\\O discussed in the text.
Indeed, the parallelisn of the analyses of these philsophers
and that of Wharf is not fortuitous, even though in a sense they
eama· to opposite conclusions in the constructivist or physicalist
debate on fashioning a pure object language. Morris, in par-
ticular I represented in his writings both the Vienna Circle
positivisn of Carnap and the American pragmatisn of the school of
Dewey (and thenceI of Peirce). It is clear that Wharf knew of this
philosophical tradition (one wonders to what extent it was discussed
anong Sapir's circle at Yale), at which he directs barbs in several
passages of his Technol9fQT Review articles in 1940-41; for
The situation is not likely to be aided by the philosophical
and mathenatical analyst who may try to exploit the field
of higher linguistic symbolism with little knowledge of
linguistics itself. Unfortunately the essays of rrost modern
writers in this field suffer fran this lack of apprenticeship
training. 'lb strive at higher mathenatical fOIlllllas for
linguistic meaning while knowing nothing correctly of the
shirt-sleeve rudinents of language is to court disaster.
Physics cbes not begin with atonic structures and cosmic
rays, but with rrotions of ordinary gross physical objects and
symbolic (mathenatical) expressions for these rrovements.
Linguistics likewise does not begin with rreaning nor with
the structure of logical propositions, but with the obliga-
tory patterns made by the gross audible sounds of a given
language and with certain symbolic expressions of its own for
these patterns. Out of these relatively si.rrq>le tenns dealing
with gross sound patterning are evolved the higher analytical
procedures of the science, just as out of the si.rrq>le experi-
ments and matbenatics concerning falling and sliding blocks Of
\rood is evolved all the higher mathenatics of physics up
into quantum theory. ( [1940b]1956: 222-23)
Note Wharf's reliance on the fundamental analytic basis of
phonology (phonenics) as the precedent for all of "scientific"
linguistics, in contrast to the seemingly a priori senantics of
the philosophers he criticizes. In this, he duplicates Leonard
Bloanfield who, all the while endorsing the programs of both
behaviorisn (influenced by Albert Paul Weiss) and physicalisn,
criticizes the Vienna Circle authors and even refers then to
his Language (1933) in a trenchant footnote to his 1935 LSA
Presidential Address, "language or Ideas?" (Bloanfield [1936)
1970; 322-28; see 323n. 4). After presenting sane exanples fran
these writers, Bloanfield observes that
Carnap, so far as I have found,. nowhere mentions the fact
that the discourse of logic presupposes descriptive
linguistics and uses the technical tenus of this enpirical
science. The linguistic background of logical
and mathenatical statement is generally ignored by
philosophers and logicians; ...
Given, however, that Bloanfield's endorsement of the physicalist
doctrine in that paper took the following fonn:
Rudolf Carnap and Otto Neurath have found that all scientifically
meaningful statements are translatable into physical teI'llS--
that is, into statanents about rrovanents which can be ob-
served and descrihed in coordinates of space and tine.
Statanents that are not made in these tenus are either
scientifically meaningless or else make sense only if
they are translated into statements about language.
--It is no wonder that Wharf, already having been trained as an
engineer (B. S., MIT, '18), concentrated his "relativist" attacks
on precisely this issue, fran 1936 on. The chronology of give-and-
take, even within the lingustic profession, cannot, of course,
be fully doClllrented here, though this illuminating task should be
systanatically undertaken. I have merely been pointing out that
Wharf's writings reflect preoccupations that were shared, a canron
discourse about issues, and a pointed concern not only with the
explicit addressees of an article but with the implicit audience
(e.g. Bloanfield himself in an appendix to the Shawnee[Algonquianl
stan-list of C.F. Voegelin that Blocmfield was sure to read;
see [1940a] 1956:162-63).
9rro be sure, Whorf proposes that scientific linguistics (in
the sense of n. 2 above) will be no little aid both in understanding
the implications of the relativity and uncertainty principles,
and in overcaning them. To what extent this is all tongue-in-cheek
is an interesting question, since, like nost positiVists (or those
whose scientific training was in that tradition) I Wharf had a mystic
inside screaming to get out. Yet, just as Blocmfield took up the
proselytizing cause of an independent, professionalized linguistics
as Science rescuing language fran students of culture (sc., literature)
in such journals as Classical Weekly, American Journal of Philology,
Modern Language Journal, Modern Philology, etc., so Whorf took up
the same cause as Cultural SCience reSCUing students of science
(sc., physics, chemistry) fran language in The Technologv Review.
In these rhetorical poses, each was rescuing himself fran a
fonner existence, a canron happening in academic life which any
reader of this will recognize inmediately. Hence the vehemence
and fervor of the rhetorical schema: "How could [addressee(s)] ( =
he [audience, i. e., prior influence] ( = 1. [speaker]) ) have been
so taken in?" -
10Hence, one is led to what Labov ( [1970] 1972: 185-87) has
called the "Saussurian Paradox: the social aspect of language
[1. e., langue--MS] is studied by observing anyone individual,
but the individual aspect [ i. e., parole--MS] only by observing
language in its social context" (186). Unfortunately, however, this
psychologistic interpretation of langue as Chansky's "conpetence"
is a misreading of Saussure who says of langue rather that
C' est un tresor depose par la pratique de la parole dans les
sujets appartenant Ii une m6ne cannaunite, un systane gram-
matical existant virtuellement dans chaque cerveau, ou plus
exactement dans les cerveaux d'un ensemble d'individus;
car la langue n' est canple'te dans aucun, elle n' existe
parfaitEment que dans la masse. (Saussure 51960; 30 [§Introd. ,
III.2 ])
What Saussure describes as "existant virtuellement dans chaque cerveau"
later !IIoorican linguists, themselves misreading Bloanfield' s
curious treatment of the issue, have literally located in every
(hence "anyone") speaker. Saussure intends only to characterize
the Durkheimian understanding of carmunity-wide nonns for the systematic
relationship of linguistic utterance-types (signn:raDts) to
intensional classes of referents (signifies), such nonns assured to
be independent of the context of use, the psycho-physiological
of the individuals, etc.
1 Now-classic statements of at least portions of this point
of view are Jakobson [1957] 1971:130-47; 1960 and Hymes [1964]
1974: 3-27; [1972] 1974: 29-66. 'Ibe first of these is explicitly
semiotic and fonnally analytic; the perspective of the other
three is clearly "functional," but not especially semiotic, in light
of the results of the first. 'Ibis rapprochement is a IOOre recent
developnent in many later studies-including my own-influenced by
the ones cited.
12In this respect, there is one plane of meaningfulness that
is entirely equivalent to, though the obverse of, granrnar. Chansky
clearly means this by the notion of an interpretative semantics
of "logical fonn" entailed by an autonarous norphosyntax. Also,
that part of generative semantics that works--that deals with
traditional 1 sense' rather than actual 'reference' --rests on the
same property, i.e., is sanantic in the same way. The partial
equivalence of these two approaches ought to be given wider
recognition. They differ rather in the treatment of context-
fOmHreaning. for which their methods are simply inappropriate.
A large part of the general linguistic writings of such Prague
School scholars as Havranek, MukaTovskY, et al. was devoted to
a theoretical basis for rationalizing the functionall
differentiation of Czech and similarly nodernizing languages of
post-World War I national autonany. See Garvin [1955]31964:
3-69; Jakobsen [1963] 1971; 522-26. Similarly, BloanfieldI s
celebrated ilLiterate and illiterate speech" ( [1927] ) poses a
functional1 problem about good vs. bad speech, but, failing to
provide a solution in purely fonnal tenns of grammtical structure,
falls back on speculation about certain individuals being "better
nodeIs of conduct and speech than others" ( [1927]1970: 156) .
In the course of his discusssion of Menanini, however, Bloanfield
provides many examples of what I will call functiona12 differentiation
(see Wlow) of linguistic features.
l<.brhus, we must distinguish between the functionl of reference-
or propositionality, and the plane of meaningfulness
of linguistic elements called semanticity. Insofar as the latter is
an empirically-investigable danain, as Jakobson has pointed out
many times (esp. ms 1956: 8-16), following Peirce, it presupposes
the metasanantic functionl of language, the use of a language itself
to state equivalences of sense-relations (see above and n. 12), what
seem to be context-independent equivalences of referential-and-
predicational contributions of linguistic surface fonns. Unfor-
tunately, semantics has been confused at times with propositional
functionl, or worse, with referencel' But only through the mediation of
does semanticity becaJE a kind reference2'
15a:>serve also that the addressees are characterized with a
term that specifies nore precisely than a simple YQl.l. which
of the chance receivers of the message are socially-constituted addres-
sees which, residually, are socially-constituted audience).
1 In the fraIJle\\Ork of the \\Orks of n. 11 above, we might say
that the metapragmatic construction focuses on the signal-form as
produced over a certain channel, or at least as produced fron a
certain source nodality. We might speculate on other, similarly
constructed channel-signal foci of report in a rodern cemnunications
system such as English, e.g., "He wrote,' • ,"and the extent to
which it is conventionally understood signal is a
replica of the reported instance. Would Austin (and his followers)
also see "acts" of these types for every such possibility? Notice
that anbedded replicas of speech--as in "He 'Good Morning! 'ed me three
times within an hour! "--fom the basis of what Benveniste has called
"verbes delocutifs" ([1958 )1966: 277-85), perhaps best to be
translated as Here, the metapragmatic construc-
tion rank-shifts the quoted signal to inflectable status, and thence
is historically derived a fom which frequently is itself an explicit
perfonnative, e.g., Je vous salue. The lexicalized focus here is
purely on the signal, the source and channel being implicit in the
degree of imitation in the replica. An extensive cross-linguistic
treatment, drawing out the implications for such topics as onanatopoeia,
is desideratum.
Thereby hangs a huge logico-linguistic literature on so-called
opaque contexts of reference and related matters, about whi.ch it is
not purpose to write here.
8Note that certain reduced forns in English, lacking that in
indirect reports, and even with converging stress-intonation-juncture,
sanetimes attenuate the opposition.
19rro be sure, Aust in's theory is the historical descendant of
a long line of native analyses in the European tradition, a true
historically-continuous ideology of language that has affected all
our technical traditions of dealing with language. In turn, these
have affected the characteristic metapragmatics of the European
language-camunities thEm:>elves. Thus, I should point out that the
historical question of the rise of these various syntactic constructions
or with other, rore technical lexical itEm:> (e.g., predicate
(v. ), utter) in the varioos European languages as precursors of, or
consequences of, such native theories, is not being considered here,
though an essential problem. Below, in fact, I will hold that there is
a necessary dialectic relationship betl\een the two, because ideology
informs and thereby the native grasp of function2' ultiIIately
As we noted above on the subject of Boas' structuralist discovery,
the only "scientific" study of lexicon (and lexical sanantics) is through
granrnatical systems. Thus, for exanple, there is a profoond difference
between these two verbs in terns of their semantic-role coding, reflec-
ted in the discourse properties that emerge in carqJlex sentences. canpare
"He pranised to buy a loaf of bread" and "He said to buy a loaf of
bread. " In the first, the explicit subject of the framing verb pranise
is coreferential with the (anaphorically deleted) understood Agent of
the f:ramed In the second, the explicit subject of is
pointedly not co-referential with any putatively deleted Agent of
The understood Agent of £!&. can presumably be established by establishing
who was the addressee of the original described interaction. Such
lexicalized (hence cryptotypic) same- vs. switch-reference properties
of language are not, in general, systematically considered by native
speakers without the caJParative perspective gained fran "scientific"
study of cross-clause reference-maintenance in languages.
21Notwithstanding Geertz' s use of the characterization "unin-
tentional" for this rather creative indexical relationship, it is
clear fran his discussion of "The role of etiquette" in general in
the. upper reaches. of Javanese society that this is merely an unideo-
log:L:;:ed, or unrat10nalized functionr, though obviously available as
a SOClal strategy. Note the following (1960: 243-44):
means to humble oneself politely and is the
correct behavior to adopt toward anyone who is either of
equal rank or higher. '" I have seen llWly
prlJap [ehte] conversations that seared to consist alnost
ez.ttirely.of an attenpt of the participants to put
himself In the lower posltlon, a kind of obsessive competition
u;> be bottom dog. (The coopetition is pretense, of course. If
elther were to flatly acknowledge the other's inferiority in
such a situation it 'oroUld be a grave insult.) ... The
lished prijaji can express all sorts of nuances of status
(and insult), many of which escape Western perception alto-
and a true virtuoso can reduce novices to qUivering
:J.llJIDblhty. As a Javanese put it, ' ... when I am going to
reply to [such a friend] ... I can't [be andap--asor too] so I
feel ashamed. I '
As we. shall see, the accaJplishment here is to employ as many
systems as possible to focus on the speaker-addressee
relatlonshlp, and in so doing, to daIDnstrate such esoteric
canpetence as only \\Qrthy of similar treatment \\QuId have.
As Geertz notes, this can spiral out of hand to a western
"1blS 'oroUld appear to be a conflict between the locally
paradigmatic lexical-conjugate perspective on the problem and
perspective of syntagmatic canDinations, since in tenns of the
fust, Geertz states, "the occurrence of one of [the linked
conjugates in a set] for any given meaning...will predict the
occurrence of the other [at the same level of "style"] if the meaning
conce.J;l1ed occurs... " (1960: 253).
'syntactic definition' of a surface fom I mean the
catego::ization resulting fran a gramnatical
analysls of sentences. In this way, such units as Noun Phrase
Noun, etc. receive characterizations in terns of the fonnal '
structure of sentences. I call attention to this because it may
not be apparent to all readers that every functional -functional
sy:;>tem implies its own segmentation and organizationlof surface forns
functionally ambiguous roost of the time. '
Poedjosoedanoo (1968: 64; 1969: 190) characterizes and lists
the membership (then 67 i tans) of a further, truly coarse and
vulgar. Q<asar) set of lexical items, considered to be a subset
of baslc Ngoko, "for the IlX>St part nouns, adjectives, and verbs
expressing such very OOlllDn things as parts of the body conditions
of the body, and bodily functions." These are like a of
slang obscenity, inasmch as there are ordinary Ngoko equiva-
pthout the VUlgar connotations.
. . "Though, as exanples given by Poec!jOsoeda:n!o (1968: 77)
mdicate, the presence of an audience in particular relationships to
the addressee (e.g., in-laws [audience] of an elder sibling [ addressee]
of speaker) is ala:> regularly indexed in this way. Actually, as
the exarrples sean to indicate, this usage might be studied as a kind
of stylized display of the basic speaker-addressee indexes, m::>ved to
a plane during "relations in public."
a:>observe that in many languages, the first-person Agent on
second-pers:>n [+human) Patient (or Dative [praroted to Patient) )
has an inflection that bespeaks politeness or deference, or just
avoidance of specification. This interacts with splits of ergativity-
accusativity in "global" (conbined Agent-an-Patient) inflectional
systEmS (Silverstein 1976a: 118; 124-25), or merges first-on-second
forms with passives, :inpers:>nals, etc. '!he fact of lexicalization
of the phenanenon in a predicational fonn (Le., verb) used in
discourse is what we should highlight in Javanese.
Note in relation to this usage that the speaker never
refers to himself with Krama Inggil vocabulary (save the King), which
usage would be theoretically possible (where referent equals speaker) I
but in direct contravention of the Andap-asor ethic cited in n. 21.
Th don§'O would be arrogant.
"""Even in the case of Krama. Andap pronaninals, the social
indexical function2 presupposes the' indexical-referential functi0n2
'person' in the particular noun phrase.
Indeed, Poedjosoedanno (1968: 67) seans to echo the attitude
of Geertz's Prijaji a:>urces (quoted above) in characterizing the
Madya conjugates as of two sorts: "Sane are the result of sane kind
of corruption of standard Kr2m3 words, while others ... sean to be
old borrowings of Krano words from sane local non-prestigious
dialect." Cbserve the negative bias of this essentially historical--
rather than synchronic-characterization.
30Recalling Geertz's observation quoted above that "in sane
cases the madya conjugate is the same as the ngoko... ; sanetimes
it is the same as the krama" (1960:253), this structural character-
ization of the "level" Madya gives the basis for the seaning dis-
parity. Also, these remarks on attitude perhaps explain why the
(apparently) prijaji view even of the urbanized non-prijaji style
repertoire (as srown in chart I) shows no "honorific," i.e.,
Krama. Inggil m::>dification of the level 2 Madya.
31Horne (1963; 122n. 13) also notes the existence of a Krama Desa,
a "fonn of Krama Madya that is spoken in the country villages of Java;
there are only a fffY{ special KD words." This is presumably on the
same order of variability as the geographically-distinct notions of
what is standard Krama fonn, what hypercorrect, etc., to which
refers in his discussion. Perhaps this imputation
of a special Krama. Desa to villagers, equated with Geertz' s
stylene 2 in charts I and II of 6, explains the perception that
have no actual Krama (level 3) usage.
As was noted above, Uhlenbeck gives eight "levels" of speech,
only two subvarieties of Ngoko, "NGOKO proper and NGOKO-ANpAP." I
would imagine that the latter is equivalent to Poedjosoeclanro' s
levels 7 and 8, the fonner to level 9.
::i3U should also be noted that in the course of presenting
field-based graIIIIRt1cal rmterial, many linguists incidentally
mention ernparable pronaninal usage, e . g., Ne\wJa.n (1944: 101).
who, explaining a gloss notes "the polite fonn of address [in Yokuts)
practiced by a man toward his m::>therc:in-law; he must address her as
rm'ak', 'you (dual)', and refer to her as 'amak', 'they (dual) ',"
orO' Grady (1964: 60), whose glosses for various pronaninal suffixes
simply indicate the equivalence of second person singular addressed
to m::>ther' s brother and second pers:>n dual otherwise, for Nyangurmta
be sure, both Friedrich 1966 and Ervin-Tripp 1971 refine
the of the social dimensions of a situation, specifying
m::>re partlcular statuses and attitudinal factors. '!his m::>re fine-
grained analysis shows the same fonnal property of a hierarchy of
indexical effect, however, one which Ervin-Tripp represents in terms
of a !:J5eision flow-chart the outcomes of which are T vs. V.
'!hus, "power"-laden relationships, such as 'father-of' "are
now reinterpreted for purposes of T and V" as relationships of'
"solidarity," such as 'sarne-family=as' (Bro.m & Gilrmn 1960:259).
This reinterpretation effectively raroves the social relations of
"power" fran the set controlling T : V usage, hence from the
functiona12 danain. Note, however, the "interesting residual
of the power relation in the contaJPOrary notion that the right to
initiate the reciprocal T belongs to the member of the dyad having
the better power-based claim to say T without reciprocation." What
might be a ITllch IIDre realistic, non-Ideological interpretation of
the historical change is that "power" has beccme an indirect social
variable, indexed only by the unremarkable initiation of shift
of usage in the course of an interaction or long-tenn a:>cial relation.
See the canparable use of names in Anerican address over the
course of interactions, as described by Brown & Ford (1961:
fig. 1 and discussion).
36while this is obviously not the place for a treatment of
categorial systans of language, or of theory of markedness, I should
refer the puzzled non-linguist who has not been familiarized with
this stock-in-trade of linguists to the refs. SUJIIJlarized in
Silverstein 1976a:116-22, on person, munber, and rmrkedness, or
to 1968: 79-80; 270-83.
Ccrrpare Brown & Gilman's (1960:266) observation:
Sane Friends use "thee" today; ... Interestingly many Friends ala:>
use "you." "Thee" is likely to be reserved for Friends aIlDng than-
saves and "you" is said to outsiders. '!his seans to be a survi-
val of the solidarity semantic. In English at large of course
"thou" i.s no longer used.... the forces at work sean'to have '
included a popular reaction against the radicalisn of
Quakers and Levelers ...
Cf. French leave, Dutch treat, Dutch oven Oklahana credit
all of which nationality/geographical construc-
tlons generally denote (hunorously) sanething not an exemplar of
the head noun at all. '!he last example, which I ONe to Mr. Hiram
::inith, of Wann Springs, Oregon, is apparently an in-joke referring
b a length of rubber hose to be used as a gasoline-tank siphon fran
the Cfs of other, unsuspecting residents of "Indian Territory!"
9rIhe exanples analyzed at length will show the varieties of
direct vs. indirect rationalization, "explicit perfonnativity"
being the JOOSt direct hence transparent to
predicational JOOtivation)>> and sarething like highly.
discontinuous intonational-contour indexes the JOOst llldlrect
hence very opaque to referential-and-predicational JOOtivation and
thence restructuring).
40It is at this level of process that diachrony and synchrony
merge» in a way related to the ancient, the later Neogranma.rian,
and even Saussurean-Bloomfieldian assertion that "analogy" was
really a fact of (synchronic) granmar.
Alford, Danny K.H. 1978. The demise of the Whorf hypothesis. (A
major revision in the history of linguistics)..
of the fourth annual meeting of the Berkeley LingulstlCS
Society, pp. 485-99. . .
Austin, J.L. 21975. Hoyt to do things with VtQrds, second editlOn.
Edited by J.O. UInSOn & Marina Sbisa. CaniJridge, Mass;
Harvard University Press. edition, 1962). .
Bates Elizabeth & Laura Benigni 1975. Rules of address ln Italy;
sociological survey Language in Society 4-271-88.
Benveniste, EInile [195811966. Les verbes delocutifs. In,
Studia philologica et litteraria in honoran L. Spitzer,
ed. by A.G. Hatcher & K.L. selig, pp. 57-63. Bern; Francke
Verlag. In, Probl€mes de linguistique generale, pp. 277-
85. Paris: Gallimard.
Bickerton, Derek. 1977. Review of Perspectives on Black English,
ed. by J.L. Dillard. Language 53. 466-69. .
Bloomfield, Leonard. [1926l. A set of postUlates for the SClence
of language. [Language 2. 153-64] In Hockett, ed. 1970,
pp. 128-38. .
[1927]. Literate and illiterate speech. [Amencan Speech
--2. 432-39] In Hockett, ed. 1970, pp. 147-56.
[1936l.Language or ideas? [Language 12. 89-95]. In Hockett,
--ed. 1970, pp. 322-28.
[l944l.secondary and tertiary responses to language.
--[Language 20. 45-55]. In Hockett, ed. 1970, pp.
Boas, Franz. 1911. Introduction to Handbook of Amerlcan
Indian Langwges, ed. by F. Boas. Bull. Bureau of Amer. Ethnol.,
no. 4, part 1, pp. 1-83. Washington, D. C.: Government
Printing Office. . .
Brown Roger & Marguerite Ford 1961. Address 1ll Amencan English.
Journal of Abnonnal and Social PsyclDlogy 62. 375-85.
Brown Roger & Albert Gilman 1960. The pronouns of power and
:.olidarity. In, Style in Language, ed. by T.A. sebeok,
pp. 253-76. cambridge, Mass: MIT Press. .
Carroll, John B. ed. 1956. Language, and real1ty: selected
writings of Benj amiD Lee Whorf. CanDndge, Mass: MIT. Press .
Ervin-Tripp, Susan 1971. In, Advances 1ll the
sociology of language, ed. by J. FislInan, vol. 1, pp. 15-91.
The Hague-Paris: Mouton. .'
Friedrich, Paul 1966. Structural inplications c;f Russlan pronoounal
usage. In, Sociolinguistics, ed. by W. Bnght, pp. 214-59.
The Hague: Mouton.
Garvin, Paul L., ed. [1955] 31964. A Prague SChool reader on
esthetics, literary structure, and style. Washington, D.C.:
Georgetown University Press.
Geertz, Clifford 1960. The religion of Java. Glencoe, Ill: The
Free Press.
Hockett, Charles F. ed. 1970. A Leonard Bloomfield anthology.
Bloanington: Indiana University Press.
Horne, Elinor C. 1961. Beginning Javanese. New Haven & London:
Yale University Press.
1963. Intennediate Javanese. New Haven & London: Yale Univer-
--sity Press.
__1974. Javanese-English dictionary. New Haven & London:
Yale University Press.
Hymas, Dell H. [1964] 1974. Introduction: Toward ethnographies of
caJm.II1ication. [In, the ethnography of camunication, ed. by
J.J. Gumperz & D. Hyrres pp. 1-34. American Anthropologist
vol. 66, no. 6, part 2.\ In, Foundations in sociolinguistics,
an ethnographic approach, pp. 3-27. Philadelphia: University
of Pennsylvania Press.
[1972] 1974. Models of the Interaction of language and social
--life. [I,n, Directions in sociolinguistics, ed. by J.J. GUJperz
& D. Hymas, pp. 35-71. New York: Holt, Rinehart, & Winston].
In, Foundations in sociolinguistics, pp. 29-66.
Jakobson, Hanan [1957] 1971. Shifters, verbal categories, and
the Russian verb. [C!mbridge, Mass: Harvard University
Department of Slavic Languages and Literatures (dittoed)].
In, Selected writings, vol. 2., Word and language, pp.
130-47. The Hague-Paris: Mouton
1960. ConclUding statanent: linguistics and poetics. In,
--Style in language, ed. by T.A. Sebeok, pp. 350-73. CaniJridge,
Mass.: MIT Press.
[1963] 1971. Efforts toward a means-ends IOOdel of language
--in interwar Continental linguistics. [In, Trends in IOOdern
linguistics, ed. by C. Mohnnann et.al., pp. 104-8.
Utrecht-Antwerp; Spectrum] In, Selected writings, vol. 2,
pp. 522-26.
__ms. 1956. Metalanguage as a linguistic problan. LSA Presi-
dential Address, 27 Decenber.
Kurykllvicz, Jerz,y [1945/49] 1966. La nature des proces dits
"analogiques." [Acta Linguistica 5. 121-38]. In, Readings
in linguistics, II, ed. by E. Hanp et aI., pp. 158-74.
Chicago; University of Chicago Press.
Labov, William [1970] 1972. The study of language in its social
context [Studium Generale 23. 30-87]. In, Sociolinguistic
Patterns, pp. 183-259. Philadelphia; University of Penn-
sylvania Press.
Lyons, John 1968. Introduction to theoretical linguistics.
Cambridge: at the University Press.
Merton, Robert K. 1968. Social theory and social structure,
enlarged edition. New York; 'Ihe Free Press
Morgan, J. L. 1977. Conversational postulates revisited. Language
Morris, O1arles [ 1938] 1971. FOillldations. of the theory of signs.
Unternational Encyclopedia of illlified science, vol. I,
no. 2. Chicago: University of Chicago Press. ] In, Writings
on the general theory of signs, pp. 13-71. Approaches to
SEmiotics, no. 16. The Hague-Paris: Mouton.
Ne\'4lWl, Edwin 1974. Strictly speaking; will America be the death
of English? Indianapolis: Bobbs-Merrill.
1976. A civil tongue. Indianapolis: Bobbs-Merrill.
Ne\\llRIl, Stanley 1944. Yokuts language of California. Viking fund
pUblications in anthropology, 2. New York: Viking Fund.
oI Grady, Geoffrey N. 1964. Nyangurnata granmar. Oceania linguistic
rn::mographs. No.9. Sydney : University of Sydney.
Paulston , Christina Bratt 1976. l'ronoillls of address in Swedish:
social class sanantics and a changing systen. Language in
society 5. 359-86.
Poedjosoedan1D, Soepcno 1968. Javanese speech levels. Indonesia
6. 54--81.
__ 1969. Wordlist of Javanese non-Ngoko vocabularies. Indonesia
7. 165-90.
Sapir, Edward 1921. Language; An introduction to the study of
speech. New York: Harcourt.
Saussure, Ferdinand de. 51960. Cours de linguistique generale,
cinqui€me edition. Paris; Payot (First edition, 1916).
Silverstein, Michael. 1973. Linguistik illld Anthropologie. In,
Linguistik illld NacllParwissenschaften, ed. by R. Bartsch &
Th. Vennenann, pp. 193-210. Kronberg: Skriptor Verlag.
__1976a. Hierarchy of features and ergativity. In, Grammtical
categories in Australian languages, ed. by R. M. W. Dixon,
pp. 112-71. Canberra: Australian Institute for Aboriginal
1976b. Shifters, linguistic categories, and cultural
--description. In, Meaning in anthropology, ed. by K. Basso
& H. Selby, pp. 11-55. Albuquerque; University of New
M:!xico Press.
1977. CuItura.l prerequisites for granmatical analysis.
--In, Linguistics and Anthropology, ed. by M. Saville-Troike,
pp. 139-51. (Georgetown University Roillld Table, 1977).
Washington, D.C.: Georgetown University Press.
1978. Review of 'The view fran language, selected essays,
-·-1948-1974, by C.F. Hockett. IJAL. 44.235-53.
lIS. 1978. The three faces of I filllction ': prelilllinaries to
~ a psychology of language. Chicago: center for Psychosocial
Stocking, George W., Jr. 1968. Race, culture, and evolution;
Essays in the history of anthropology. New York; Free Press.
Uhlenbeck, E.M. 1970. The use of respect fonns in Javanese.
In, Pacific linguistic studies in honour of Arthur Capell,
ed. by S.A. Wunn & D.C. laycock, pp. 441-66. (Pacific
Linguistics, Ser. C., no. 13) Sydney; A.a. & A.W.Reed, for
Linguistic Circle of Canberra..
Wharf, Benjamin L. [l940a]. Gestalt technique of sten-canposi.tion in
Shawnee. [In C.F. Voegelin, Shawnee stens and the Jacob P.
Dunn Miami dictionary, pp. 393-406. Indianapolis: Indiana
Historical Society]. In Carroll, ed. 1956 pp' 16Q....172.
__[l94Ob]. Linguistics as an exact science. [Technology review 43.
61-63, 8Q....83] In Garroll, ed. 1956, pp. 22Q....232. .
__[1941]. The relation of habitual thought and behavior to
language. [In Languge, culture, and personality; Essays in
/llEm)ry of Edward Sapir, ed. by Leslie Spier et. al. pp. 75-
93. Menasha, Wis. ; Sapir Marorial Publication Fund]: In
Carroll, ed. 1956, pp. 134-59.
__U942]. Language, mind and reality. The Theosophist (Madras).
In Carroll, ed. 1956, pp. 246-70.
__[1945]. Gramnatical categories. [Language 21. 1-11]. In
Carroll, ed. 1956, pp. 87-101.
__ ms. [1936/37]. A linguistic consideration of thinking in
prilllitive camn.mities. In Carroll, ed. 1956, pp. 65-86.