LINGUIST List 2.671

Thu 17 Oct 1991

Disc: Whorf Part 2

Editor for this issue: <>


  1. "Bruce E. Nevin", Whorf-Sapir Hypothesis

Message 1: Whorf-Sapir Hypothesis

Date: Wed, 16 Oct 91 08:28:26 EDT
From: "Bruce E. Nevin" <>
Subject: Whorf-Sapir Hypothesis
I want to outline the views of Sapir and of Whorf on linguistic and
cultural relativism as I understand them and survey some of what has
been done with these ideas, both as deriving explicitly from their
writings and as arising from less clearly articulated cultural and
intellectual antecedents that it is difficult for any of us not in some
measure to share as we grapple with universals and idiosyncrasies of
language and culture.
These ideas arose for Sapir in the context of his work on language typology
on the one hand and psychology on the other. In the background lay social
Darwinism, or at least the pervasive evolutionist perspective of 19th-
century anthropology, and in this respect Sapir's interest here was a
continuation of Boas' restitution of "primitive" languages as on an equal
footing with the languages of familiar literate cultures, and an all-
important entre'e into "the network of cultural patterns of a civilization,"
which "In a sense . . . is indexed in the language which expresses that
civilization." (1929:162)
In his conception of the relation of language, personality, culture, and
"the world," Sapir distinguished between social reality
 "Language is a guide to `social reality.' . . . it powerfully
 conditions all our thinking about social problems and processes. . .
 the world of social activity as ordinarily understood"<1>
1. Hoijer, in the 1953 conference proceedings, adduces passages of a
similar sort in the writings of Boas.
and objective reality, as had Durckheim and others, and affirmed of the
former that
 "No two languages are ever sufficiently similar to be considered as
 representing the same social reality. The worlds in which different
 sodcieties live are distinct worlds, not merely the same world with
 different labels attached."
It was in this sense that he made his famous assertion "The fact of the
matter is that the 'real world' is to a large extent unconsciously built up
on the language habits of the group." (Preceeding quotations all loc.
The core of the matter for Sapir, however, was an identification of
language, specifically grammatical categories, with thought:
 I quite frankly commit myself to the idea that thought is impossible
 without language, that thought is language. (In a letter of 8 April
 1921 keeping Lowie abreast of progress on the manuscript of Language;
 quoted in Darnell 1990:99.)
In other places, Sapir severely divorces language from culture, but in this
he appears to mean material culture, the "inventory" of cultural artefacts.
The correlation of these things with associated vocabulary he regarded as
2. Darnell (1990:434 n) says
 Sapir's strongest relativity statement was a brief note titled
 "Conceptual Categories of Prinitive Languages," an abstract of a paper
 read to the National Academy of Sciences in 1931. This was published
 only after his death.
Her bibliography lists it as appearing in Science 74:578. I have not seen
it and cannot comment.
Whorf may have been a Theosophist. His philosophical interests
attracted him to Sapir and to linguistics, and his fascination with the
"hidden metaphysics" of languages remained always the central thing for
him, for which the tools of linguistics were subordinate means. From
the point of view of an emerging profession, then, he was quite
literally eccentric, in that specific sense. His ideas began to
crystallize with preparation to teach a course at Yale during Sapir's
leave in 1937-38. His intention was to "excite [students'] interest in
the linguistic approach as a way of developing understanding of the
ideology of other peoples" (letter to Spier). He would focus on "a
psychological direction, and the problems of
 meaning, thought and idea in so-called primitive cultures," aiming to
 "reveal psychic factors or constants" and the "organization of raw
 experience into a consistent and readily communicable universe of ideas
 through the medium of linguistic patterns" (to Carroll; both quoted in
 Darnell 1990:381).
 Whorf developed his ideas about linguistic relativity during Sapir's
 illness and elaborated it after his death, so Sapir never had a chance
 to comment. Whorf died in 1941 at the age of forty-four, leaving less
 sympathetic colleages to pursue the implications of his work.
 (Darnell 1990:375)
Sapir had confined the constitutive role of language to social reality.
Whorf went farther, and developed the claim that
 It is the grammatical background of our mother tongue, which includes
 not only our way of constructing propositions but the way we dissect
 nature and break up the flux of experience into objects and entities
 to construct propositions about. (1956:239)
The identification of language and thought takes an adversive twist:
 [T]hinking . . . follows a network of tracks laid down in the given
 language, an organization which may concentrate systematically upon
 certain phases of reality, certain aspects of intelligence, and may
 systematically discard others featured by other languages. The
 individual is utterly unaware of this organization and is constrained
 complete within its unbreakable bonds. (256)
 if a rule has absolutely no exceptions, it is not recognized as a rule
 or as anything else; it is then part of the background of experience
 of which we tend to remain unconscious.
In the background always is Theosophy, as in The Voice of the Silence:
 The mind is the great slayer of the real. (Quoted on p. 253)
His views were recast in terms more acceptable to prevalent conceptions of
operational test and verification, as by Eric Lenneberg in 1953, summarized
by Roger Brown (Reference: In Memorial Tribute to Eric Lenneberg, Cognition
 I. Structural differences between language systems will, in general, be
 paralleled by nonlinguistic cognitive differences, of an unspecified
 sort, in the native speakers of the two languages.
 II. The structure of anyone's native language strongly influences or fully
 determines the world-view he will acquire as he learns the language.
 (p. 128)
Behind this was the assumption (presumably "part of the unconscious
background" of every student in the Boas-Sapir tradition, and indeed of
virtually everyone as has been argued on the LINGUIST list) that
III. Languages, and hence cognitive systems, can vary without constraint.
Proposition II has generally been presumed to be untestable because of
the identification of language and any means of communicating one's
world-view. Attempts to verify or falsify the hypothesis have concerned
themselves either with I or III (with indirect evidence for II sought
from III). It would be interesting to see a resumption of attention to
III e.g. employing techniques developed for study of non-human
A conference organized by Robert Redfield in 1953 drew together a
relatively small number of linguists and anthropologists with the aim of
defining problems related to the hypothesis, reviewing work undertaken and
plans for future work relating to it, and attempting to establish a minimal
framework of institutional support for these research interests. Their
proposals concerned mostly methods for getting at I. Their conclusions
were cautious, as noted above, in keeping with the temper of the times.
Kay and Kempton (AA 86:66), perhaps somewhat parochially but truthfully as
regards empirical research, claim that most of this research has been in
the domain of color. They give citations of work bearing on III beginning
about the time of the Redfield conference (Ray 1952, Conklin 1955,
Lenneberg and Roberts 1956, Gleason 1961, Bohannan 1963), and probably the
best known study, their own (Berlin and Kay 1969). They remark that
"studies before 1969 tended to support III; those since 1969 have tended to
discredit III" (loc.cit.) They accept the finding of Kay and McDaniel
(1978) explaining universal constraints in color classification in terms of
the neurophysiology of human color vision, and discrediting III with
respect to color. They affirm of course that research into II and III is
an open matter for domains other than color perception, in particular
domains (they mention religion) where characteristics of peripheral neural
mechanisms like those of color perception have no bearing.
A parallel tradition of research into aspect I of the hypothesis has been
carried out primarily by psychologists, and Kay and Kempton (1984) is a
continuation of this. They cite Brown and Lennebert 1954, Burnham and
Clark 1955, Lenneberg 1961, Lantz and Stefflre 1964, and Stefflre,
Castillo, and Morely 1966. This line of research seeks a correlation
between a linguistic variable (codability and communication accuracy) and a
nonlinguistic cognitive variable (memorability) within a single language,
and is thus a weak form of I.
After initial claims of success in finding a positive correlation between
the memorability of a color and its value on a linguistic variable, Rosch
showed that both memorability and the combined variable of of codability
and accuracy of communication is determined universally by focality or
perceptual salience. The assumption that the linguistic variables of
codability and communication accuracy differ across languages (III again)
was falsified by this research, and therefore any correlation between
memorability and a linguistic variable was not relevant to the hypothesis.
Lucy and Shweder determined that the problem of focality or salience was an
artefact of how the color chips were presented, and devised an array by
repeatedly re-randomizing chips from the initial array so that there is no
relation between focality and findability. By this means they have
reinstated the earlier correlation in favor of I with respect to color
categories. There remain problems of interpretation and relevance to the
broader aims of the enterprise, as unfortunately often happens in narrowly
empirical work.
Research of a broader sort has gone on in many fields. In social and
cultural anthropology it is difficult to find anything that is
absolutely irrelevant to the Whorf-Sapir hypothesis, though the latter
can be made irrelevant to some forms of anthropological work essentially
by legislating a rather narrowly realist, anti-constructivist
perspective for science. Among clearly relevant issues I name questions
of symbolism, including especially money and symbols of political and/or
religious stature, magic and cargo cults, studies of kinship systems and
their role in the construction of interpersonal and social relations,
and work in social categories. To this must be added work of more
obviously linguistic nature, such as projection of prehistoric cultures
from reconstructed protolanguages, Studies of the bases of prejudice, of
stereotyping, and of national character in a more genuine sense (as
pioneered by Gregory Bateson) . . . the list is seemingly endless.
The fields of ethnolinguistics and sociolinguistics, themselves
extremely broad and diversified (and themselves polarized rather as the
right and left hemispheres of the brain of the archetypal
anthropological linguist), have obvious bearing on the hypothesis.
Hymes has urged a reinterpetation of the hypothesis, investigating
patterns of language use rather than of language structure per se.
The perhaps contentiously named field of cognitive linguistics has a
strong constructivist bent. Work in psycholinguistics in general often
has clear bearing, though the direction of interest (and funding) to
linguistic universals has tended to obscure investigation of linguistic
idiosyncrasies that might correlate with cognitive differences.
>From Bateson's work on communication and learning and in particular the
discovery of the double bind in relation to these have developed lines of
clinical research that have developed practical techniques of reframing and
use of metaphor, and an understanding of human systems in cybernetic terms,
as therapy (particularly the field of family therapy).
Lastly, I must mention the resurgence of feminism in all its many forms,
especially as a scholarly concern in anthropology.
I will describe in a little more detail a new test of aspect I of the
hypothesis devised by Kay and Kempton (1984) so as not to be so
restricted in interpretive scope as the previous
communicability/codability studies had been. Speakers of Tarahumara (a
Uto-Aztecan language of northern Mexico) lack the basic lexical
distinction between green and blue (as do various other languages,
including Achumawi). Aspect I of the hypothesis predicts that speakers
of English will polarize their perceptions near the border of green and
blue, but speakers of Tarahumara will not.<3> In the first experiment,
English- speaker's judgements reflected the division of green against
blue in 29 trials out of 30; Tarahumara speakers responded even-handedly
with 13 out of 24, extremely close to a 50-50 split, vindicating the
3. This phenomenon of polarization, by the way, is the reason speakers of
English can disagree so strongly about the assignment of marginal colors to
either green or blue. A slight difference in idiosyncratic placement of
the boundary makes a large difference in categorization. This would
provide the basis of an interesting study relating to I.
These experiments involve discriminating among three chips. In the first
experiment, the subject had an opportunity to assign a color name to the
intermediate chip, and this may have prejudiced the later step of the
experiment, when the alternate comparison was made. The second experiment
made the comparisons with the three chips adjacent in a box with a sliding
cover that covered the chip on one end. In the setup stage, the subject
agrees that the middle chip is greener with respect to one chip, and then
that it is bluer than the other. It thus has both names associated with it
when the subject is invited to alternate views as often as desired, and
judge which difference is greater. In this experiment the polarization
effect disappears.
This accords with an interpretation by categorization (experiment 1) versus
an interpretation by discrimination (experiment 2). An exact parallel
could be made with the fact that people can discriminate differences
between sounds with indeterminate fineness (phonetics), but discriminate
relevant differences that make a difference4 in small numbers of categories
(phonemes, contrasts, distinctions) and displaying characteristic
polarization effects at the boundaries. A culturally/linguistically
determined contrast can be repeated, a difference requiring perceptual
discrimination can only be imitated.
Kay and Kempton interpret these findings as disconfirming what they call
radical linguistic determinism, in which "human beings . . . are very
much at the mercy of the particular language" (Sapir, quoted
previously). Because the polarization associated with naming can be
made to disappear simply by not naming, we are not hopelessly at the
mercy of our language. To this I would add that it is difficult to do
many sorts of things cooperatively with other human beings or with
social consequence and recognition without employing the categories
inherent in language. The exceptions, it seems to me, are in the realms
of art, of religion, of play and creativity. These are the domain of
the _pleroma_ in Bateson's terms, the realm of cybernetic explanation,
as opposed to the _creatura_, the realm of forces and impacts dealt with
in the conventional categories of one's shared language and culture.
	c 1991 Bruce Nevin
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue