<<

THE HYPOTHESIS

by

Michele Nathan

A Thesis Submitted to the Faculty of the College of Social Science in Partial Fulfillment of the Requirements for the Degree

of Master of Arts

Florida Atlantic University

Boca Raton, Florida

December 1973 THE LINGUISTIC RELATIVITY HYPOTHESIS

by

Michele Nathan

This thesis was prepared under the direction of the candidate's thesis advisor, Dr. John D. Early, Department of , and has been approved by the members of his supervisory committee. It was submitted to the faculty of the College of Social Science and was accepted in partial fulfillment of the requirements for the degree of Master of Arts.

SUPERVISORY COMMITTEE:

&~ rl7 IC?13 (date) 1

ii ABSTRACT

Author: Michele Nathan

Title: The Linguistic Relativity Hypothesis

Institution: Florida Atlantic University

Degree: Master of Arts

Year: 1973

Although interest in the linguistic relativity hypothesis seems to have waned in recent years, this thesis attempts to assess the available evidence supporting it in order to show that further investigation of the hypothesis might be most profitable. Special attention is paid to the fact that anthropology has largely failed

to substantiate any claims that correlations between culture and the of do exist. This has been due to the impressionistic nature of the studies in this area. The use of statistics and hypothesis testing to provide mor.e rigorous methodology is discussed in the hope that employing such paradigms would enable anthropology to contribute some sound evidence regarding t~~ hypothesis.

iii TABLE OF CONTENTS Page

Introduction • 1

CHAPTER

I THE.HISTORY OF THE FORMULATION OF THE HYPOTHESIS. 5

Notes to Chapter I 20

II SOME PERIPHERAL ASPECTS OF LINGUISTIC RELATIVITY. 21

III THREE OF THE FOUR LEVELS OF EVIDENCE. 25

Notes to Chapter III • 40

IV THE ANTHROPOLOGICAL CONTRIBUTION • 41

v FURTHER RESEARCH POSSIBILITIES. 84

VI CONCLUSION . 90

BIBLIOGRAPHY. 94

iv Introduction

In my opinion, the language-and-culture problem is the greatest challenge to since the phoneme • • • it is largely through the language-and-culture field that linguistics can become most clearly relevant to the other social sciences, and thereby become a meaningful social science itself. I think this is what Whorf had in mind • • • • And this is where most of American linguistics has marked time since Whorf and Sapir. --Garvin 1956:568-569

The provocative of had great influence on the fields of linguistics, anthropology, and in the 1940's (Heidbreder 1958:87), when they were first published and then, later, when they were collected by John B. Carroll into one volume in 1956, along with some unpublished articles. The excite- ment that attends such a revolutionary idea as linguistic relativity, however, eventually dies down, and so we find that the Whorf Hypoth- esis, although frequently mentioned in academia, is no longer inspiring the conferences and research that it once did. This waning of interest is the result not only of loss of novelty, but also of the difficulty in researching the problem and of the disappointing results of much of the research that has thus far been performed.

However, the question of linguistic relativity--that language differences can be related to differences in perception, , and culture--is ~= oo small importance and certainly deserves a more

1 2

thorough exploration than has as yet been accorded it. As a special

type of cultural relativity (Fearing 1954:49), linguistic relativity

could be most valuable in aiding cross-cultural understanding,

especially when the nature of a culture and its language are inspected

together (Hoijer 1954:102). As Wharf himself suggested (1971:78-80),

insights of ethnology and psychological linguistics can cross­

fertilize, with benefit to both sciences. If the hypothesis is true,

understanding different will give us a fuller understanding

of human nature (Wharf 1971:263-264).

There are also more practical benefits that exploration of the hypothesis may bring. If it is true, study of different languages

will give us exposure to a variety of useful ways of thinking

GAlexander 1936:263-269; Carroll 1958:34-35). Thus, a contribution

of linguistic relativity, according to Wharf (1971:84-85, 218), will

be the greater appreciation of patterns of thought other than those

engendered by the Indo-European language family; we may come to fully appreciate that the languages of modern civilization are not superior

to other languages and, in fact, are inferior in dealing with certain

ideas. Further, Wharf (1971:81-83) suggests that the hypothesis is relevant to problems of human relations, understanding, education, and language reform. Thus, Chase (1954:78) mentions that even a brief visit to the United Nations would show the problems of inter-language

communication and, indeed, a study by Glenn (1954:173-178, 166-167) of discussions in the Security Council shows how complementary linguistic and cultural analysis could be fruitful in avoiding international dis­ agreements; for example, applying the term "expanding economy" to the 3

United States is heresy to th~ Russian since the term is reflexive in the Russian language and implies that capitalism is inherently expansive. As a matter of fact, the School of Languages and Lin­ guistics of the Foreign Service Institute at one time considered

Wharf's writings to be essential reading for foreign service agents

(Leary 1952:196). Also, it has been claimed that poor thinking due to poor use of language can be linked to America's political and economic difficulties, as well as the problems of war and slums (Chase

1938:18-29). And if language and culture are integrated, accultura­ tional situations have the potential for disturbing this integration with disastrous results, and such problems must be cons~dered by foreign service programs (Thompson 1950:562-563). A final practical consequence of the hypothesis that might be mentioned is its implications for teaching foreign languages; if thought varies with language, a new language should perhaps be taught as a new world structure at variance with that inherent in the student's native language (Carroll 1963:17-19; Cassirer 1965:133).

Thus, although Penn (1972:55-56) maintains that the only contribution of the Whorfian Hypothesis has been to challenge linguistics and force it to be empirical in searching for universals, it would be more correct to say that the hypothesis has pointed to a relation between language, thought, and behavior which, if it has any at all, has relevance not only to the theoretical aspects of linguistics, anthropology, and psychology, but is also extremely pertinent to practical problems such as international communication and formulation of scientific languages. While Hockett (1954:107-111) 4 cautions that we should not let investigation of this hypothesis cause us to neglect other relations between language and culture, there has never been any real danger of that, even at the time when Wharf's ideas were most popular (see, e.g., Casagrande 1955; Goodenough 1964:

39; Greenberg 1964; Kroeber 1941; Nida 1945; Reichard 1950:201-204;

Voegelin and Harris 1945:456-461). Rather, the danger is that what will be neglected is linguistic relativity, and this would be an unfortunate mistake in view of its potential significance. What is needed ac this point is an assessment of the work that has been done in this area so far in order that research on the problem may find new direction and take steps toward ascertaining specifically in what ways language is a key to thought and behavior. In this paper I shall attempt to provide such an assessment of the results and problems of the research that has so far been directed toward exploring linguistic relativity, with special emphasis on the anthropological literature investigating this problem. I

THE HISTORY OF THE FORMULATION OF THE HYPOTHESIS

Although writers have disagreed in their beliefs regarding the strength of language's influence and the areas in which it is exerted

(Carroll 1963:1-2), one form or another of the linguistic relativity hypothesis has been discussed by many scholars in many ages. In a sense, in fact, the history of the hypothesis is as old as mankind itself for, as Fishman (1960:325) points out, the association of a people's individuality with their language is an ethnographical commonplace. In Western history, the question of the relation between language, thought, and behavior can be traced from through

Francis Bacon to John Locke (Penn 1972:41-48). In more recent times, linguistic relativity has been proposed by numerous scholars, thus earning it such titles as the "Humboldt-Boas-Cassirer-Sapir-Whorf-

Lee" Hypothesis (French 1963:391-392; see also Lounsbury 1959:195).

The ideas of each of these scholars will be discussed briefly below.

Begun as a reaction against Kant's innate ideas (Penn 1972: 1 48-51), the German lineage of linguistic relativists starts with , proceeds to Johann G. Herder, and then to (Penn 1972:19-22, 48-55), and the tradition has lived on among the recent Neo-Humboldtists, Jost Trier, Walter Porzig, and

Johann . Thus, Hamann maintained that language is a clue

5 6 to how reality is experienced by a people, that it provides "access to the spirit of a people" (O'Flaherty 1952:23-25). Although Herder was not consistent in his views (Sapir 1907:137-139), at one stage he did agree with his mentor, Hamann, that are one (Penn 1972:51-53).2 Of this German tri~p1rate, Humboldt is perhaps best known for his linguistic . In Ohman's" words (1953:

123-124), he saw language as a 11 central force influencing and forming the mind and behavior of human beings." Even stronger than this, he essentially saw language as a determinant of how the world is exper- ienced (Penn 1972:19-22).

The Neo-Humboldtians have developed the linguistic relativity idea even further than their forerunners. They have concentrated on the of the "lexical field" as a clue to the way in which language is a screen between the observer and his experience. The concept of the field, although not a very new one, was only used infrequently before 1924, when Ipsen first used the term "field" to

II denote words that form a unit of (Ohman 1953:124-125).

Although they actually discuss different kinds of lexical fields, the various Neo-Humboldtists hold much in common. Thus, Trier has worked with the manner in which conceptual fields are divided by the words of a language, using his method to study the evolution of the vocabu- lary of the intellect in German (Bhman 1953:126-128). For Trier, a word and terms related to it conceptually are the constituents of a field; the meaning of the grade "good," fo~ example, depends on this term's position in a configuration of grades (Waterman 1957:202-203).

There are those who feel that he has devised a good technique for 7

II investigating the tnfluence of language on thought (Ohman 1953:132-

134). On the other hand, Porzig's "elementary semantic field" con-

sists of a nucleus v.~rb or adjective and the nouns and other words

that it presupposes (as "throw" presupposes "hand"); his method

II remains to be testud on a large scale (Ohman 1953:129, 134). The

t~ird most im~~rtant Neo-Humboldtist is Weisgerber, who adopted

II Trier's "field," though modifying it somewhat (Ohman 1953:131-132).

He has applied his concept to of nature, the senses,

and material culture, among others (Basilus 1952:101-102). To s~

up, lexical fields are organized conceptual spheres that analyze

experience, providing a certain world view (Ullmann 1966:250-251).

It might also be mentioned that there are several differences between

these scholars and the Whorfian school--they concentrate on European languages, which Whorf avoided, and they focus on , while

Whorf was more concerned with grammar (Ullmann 1966:253), Whorf seems

to have gone beyond the N~o-Humboldtians in his formulations

(Waterman 1957:205-208, 211).

Another name frequently associated with the hypothesis is that of Ernst Cassirer. Lenneberg (1955:512) suggests that Cassirer holds a controversial position in this area due to a change in the views expressed in his writings. As will become apparent as this discussion progresses, consistency has rarely been a virtue among linguistic relativists, and this has surely added to the confusion surrounding work in this area. Thus, in Die Philosophie der

Symbolischen Formen, Cassirer maintained that language is a reflection of a people's spirit, obligatorily imposing its Weltanschauung upon 8

the native speaker, while in An Essay on Man, he is more concerned

with language as a universal human possession (Lenneberg 1955:520-

522). Language and seems to reflect Cassirer's earlier stage.

In it he suggests that language influences myth, just as myth in­

fluences language (1953:43), and he goes so far as to say that an

individual "lives with his objects only as language presents them

to him" (1953: 28).

It has been suggested that Humboldt's ideas were taken over the Atlantic Ocean by and from him were transmitted to

Edward Sapir (Fishman 1960:325; Penn 1972:53-55). Thus, Boas mentions

the "intimate ties between language and ethnic psychology" (1904: 518), and states (1930:77-78) that grammar and vocabulary serve to

direct "the attention of the speaker in definite lines." There are

those who deny that Sapir's ideas are related to this hypothesis

(see, e.g., Landar 1965:25-26, 216, 224), although it is generally agreed that Sapir was indeed a forerunner of Whorf. Perhaps his

inconsistency in this area may be held responsible for the debate concerning his actual position (cf., e.g., 1949:218-219 and 1964). There is, however, no mistaking a that different languages are only "loosely equivalent" and are "incommensurable" (1964), or that "We may see and hear and otherwise experience very largely as we do because the language habits of our community predispose certain choices of interpretation" (1963b:l62).

The idea of linguistic relativity is, of course, most associ­ ated in America with Whorf, Sapir's pupil. Before discussing Whorf's formulation of the theory, however, I should like to mention the work 9 of his contemporary, Alfred Korzybski, and the linguistic movement he

initiated. Whorf was well aware of the importance of his work for

scientific investigation (see, e.g., 1971:51-56, 78, 83-84, 217,

221-222, 233-.245, 269), often suggesting ways in which different

languages could profitably serve to emphasize different aspects of reality important to science. This idea has also been discussed by

Stuart Chase (1954:79, 81). Further, as mentioned above, Whorf was concerned with other practical implications of linguistic relativity as well. However, Korzybski devoted himself to such practical prob­ lems as Whorf never did.

General Semantics may be said to date from the publication of

Korzybski 1 s The Manhood of Humanity in 1921 (Reiser 1943:4), although perhaps his most famous book is Science and Sanity. In the former, he expresses the idea that up-to-date knowledge requires a language structured as like the real world (as known by science) as possible

OReiser 1943:5-7), and in much of his later work he sets out to show how language may be improved to be more exact. He sought to devise a linguistic system that is "non-Aristotelian, non-Euelidean, and non­

Newtonian" (Chase 1938:94), a system that, through the use of

"extensional devices" would promote exact thinking, in which the individual's map of the world in his head would correspond maximally with the territory of the actual world (Rapoport 1958:86-87). In other words,

Korzybski maintained that verbal thinking projects a language's struc­ ture onto thought, such that speakers of Indo-European tend to adopt an Aristotelian system of thought. In order to bring thought up to date with modern science, the individual should be careful not to con- 10

fuse language with reality and may use language in closer congruence with reality if he adopts the use of certain language conventions to

be added to his native language. For example, instead of saying "Harry

has gray hair," which implies an unchanging attribute, one would say

"Harry has gray hair," commenting on Harry's hair as it appears at 1973 a certain point in time, 1973. (For a full discussion see Korzybski

1951).

The principles of may also be seen in

Language in Thought and Action, by S. I. Hayakawa. Concerned that

people often let language mask reality, Hayakawa also presents the reader with a of rules to be used to cut through language so that he may face reality (1949:252-253, 281-285, 291-293). Hayakawa (1949:

307) says that the purpose of his book is to show how to use language as a mechanism of survival rather than as a creator of conflicts.

Perhaps this statement, more than any other, sums up the essential driving force behind General Semantics.

The work of Dorothy Lee will be discussed in a section below.

This scholar does deserve mention here, though, because she has been one of the most forceful proponents of linguistic relativity in America.

Her orientation toward the problem is somewhat different from Whorf's, but she too is concerned with the manner in which language may be a clue to a people's culture. Believing that "grammar contains in crystallized form the accumulated and accumulatin~ experience, the

Weltanschauung of a people" (1938:89), in her work she extensively employed linguistic analysis as the first step in con~tructing a culture's world view. 11

It was, however, Wharf who inspired the major concern with

linguistic relativity in the United States, so his statement of the hypothesis merits a more extensive discussion. A 1918 M.I.T. graduate

in chemical engineering, Wharf was trained in fire prevention by an insurance company, but managed to combine notable careers in both business and linguistics (Carroll 1971:1-5). He developed the terms phenotype and cryptotype to aid in language analysis. A phenotype, or overt category, is a linguistic that is always used in con- nection with a formal mark, as are nearly all plural nouns in English

(Wharf 1971:72, 88-89, 93). While grammar has dealt mainly with phenotypes, Wharf (1971:72-73) argued that there is much more to meaning than that. A covert class, or cryptotype, has no overt mark, but is known intuitively as part of a rapport~system and is given a distinctive treatment in its use. An example is gender in

English, which is marked only when pronouns are used, such that "ship" is a member of the cryptotype "feminine" (Wharf 1971:69-71, 89-92).

Thus, a full understanding of meaning in a ~norfian analysis (Wharf

1971:110) requires analysis of both phenotypic and cryptotypic 3 factors. Thinking, Wharf (1971:66-69) says, is largely linguistic, manipulating linkages, or whole class paradigms. Since different languages have different grammatical categories and different vocabu- lary classifications, it follows that speakers of different languages think differently. "We are thus introduced to a new principle of relativity, which holds that all observers are not led by the same physical evidence to the same picture of the universe, unless their linguistic backgrounds are similar, or can in some way be calibrated" 12

(Whorf 1971:214). Language channels nonlinguistic experience such

that even of time, matter, and space may vary (Whorf 1971:

158-159). A language, then "conceals a metaphysics" and is a clue

t~ culture and behavior (Whorf 1971:58-59). Although we think we

understand language and thought since we talk effortlessly, Whorf

(1971:250-252) declares, in reality we know little about them. It is

the job of the linguist to study the background phenomena of language

that the layman misses (Whorf 1971:211), and hence to show the work­

ings of linguistic relativity.

Before explaining the evidence Wharf used to support his

theory, one point in his formulation should be clarified if possible.

Much has been made of Whorf's concern with the "metaphysics" embodied

in a language. Some scholars (e.g., de Madariaga 1969:175) are quite

agreeable to the idea that a philosophy is contained in a grammar, but others, while admitting the possibility of interplay between

language and world view, deny that a language can contain a world view (e.g., Hymes 196lb:36). The problem becomes that of deciding what features of language are to be considered, and what is meant by

a "metaphysics." For example, Carroll (1963:7) argues that while

in a language does not comprise a world view, the

system structuring classification, that is, the principles of classi­

fication, may, and indeed it seems most likely that world view is

related to characteristic modes of distinguishing categories, whether

or not it may be said to be related to the presence of certain mor­

phemes in a language. Likewise, if "metaphysics" is taken to mean a

screening of perc-~tion or cognition (Lounsbury in Hoijer et al. 1954: 13

136-137, 145-146) or any analysis of experience (Greenberg 1954:17-

18), it is entirely possible that it is intimately related to

language.

Although Olmsted (1950:9) is of the opinion that Whorf "has

made a pretty good case" for the linguistic relativity hypothesis,

he is, beyond a doubt, expressing a minority point of view. Rather

the consensus seems to be, as Bedau (1957:291) expresses it, that

Whorf gave "no good reason" to support the relativity hypothesis.

This is not to say, however, that he did not purport to present

evidence supporting his assertions. Indeed, he presents evidence of

three kinds. One of these, correlations observed between language

and culture, will be discussed in a later section. Another is the

supposed in "thought worlds" pointed out by inter-language

. Despite a minority opinion (Carroll 1971:29-30) that

linguistic differences can be significant even though not shown to

be correlated with nonlinguistic behavior, most scholars agree that

Whorf's literal translations, although suggestive, do not establish linguistic relativity (see, e.g., R. Brown 1958:233, 243; Lenneberg

1953:463; Luh 1948:107; Penn 1972:30). Indeed, Black (1962:254-255) and Osgood (1966:317-318) note that Whorf has actually identified language with an individual's perception, cognition, and world view; hence the tautologous "discovery" that it determines them. This is not the only problem with such evidence, though. As

Lenneberg (1953:465) points out, Whorf's translations are quite literal, but there is no necessity for being so strict, and more liberal translations might not show such great language differences. 14

Further, he says (1953:466), such strict translation assumes that meaning is dependent on morphemes, ignoring possible connotative associations that may be important, and translations can never really be exact even in their denotata. What Lenneberg is saying, in essence, is that translation is a complex problem, and it is not scientific of Wharf to present a passage translated according to his own personal tastes, and then to say that the translation is evidence . - "' of a type of thought different from his own. Another point Lenneberg

(1953:465) brings out is that,. if language does influence psychology, a translator substitutes his own psychology for the original, so the translation cannot be used as evidence for the psychology of the original. Another important problem with the translation evidence is that of dead , that is, metaphors whose literal meanings are no longer conscious, such as "good-bye" in English (R. Brown 1958:

140-142; Carroll 1964:107; Lenneberg 1953:465; Walker, Jenkins, and

Sebeok 1965:194-195). Likewise, although etymology may be indicative of cultural attitudes or perception (Casagrande 1955:24), this does not mean that in all cases this is so, and one must be careful not to rely on etymologies as revealing current thought (R. Brown 1958:

241-242; Luh 1948:108). Although a or etymology may tell us something about how its originators thought (Hymes 196la:331), in other words, it may be irrelevant to present thinking. Thus, trans- lation, as used by Wharf, can provide no evidence for linguistic relativity, and we must search for other methods of supporting the hypothesis.

It is interesting that Gastil (1959) resurrects the transla- 15

tion method of proving linguistic relativity. He proposes that in­

specting translations of scholarly works may show characteristic

distortion of ideas. If a certain distortion teniency could be demon­

strated between two languages, there would be a case for linguistic

relativity, in the form of a predisposition for each of the two

languages to think along somewhat different lines. Gastil does point

out the limitations in his method--that it is subjective and relative

and keeps language differences at a minimum--but it is significant

that he has improved on Wharf's method by using translations devised

independently of the investigator. A similar idea is suggested by

Kaplan (in Hoijer et al. 1954:374), who proposes an experiment in

which speakers of two languages translate sentences back and forth to

see if there are characteristic distortions that appear after a round

of translation.

Whorf also used his own personal experience in the insurance

business in support of his theory. He notes (1971:135-137) cases in which poor labeling of the machinery or material being used in a

factor~led to inappropriate behavior that started fires. While many

scholars (see, e.g., Lenneberg 1953:464; Longacre 1956:307; Penn 1972:

30-31; Whatmough 1956:186-187) maintain that these examples merely

show wrong use of language and are irrelevant to the question at hand,

they do at least show that one behaves in accordance with the labels one uses, whether or not these labels are the most appropriate possible

in the language. Names do seem to influence behavior.

From the above it is apparent that Whorf's ideas were not well

supported by his own evidence. However, they were provocative and 16

inspired more appropriate research. Actually, though, Whorf never

provided a working hypothesis capable of empirical verification

(Lenneberg and Roberts 1961:494), because he never specifically out­

lined the nature and degree of linguistic influence on behavior and

thought (Heidbreder 1958:96).· Perhaps the most appropriate conclusion

concerning Wharf's contributions is that he is a source of important

ideas--an inspiration rather than an explanation (Heidbreder 1958:

108; Rapoport and Horowitz 1960:362-363).

It is not surprising, then, that such ambiguous writings have

been subjected to numerous different interpretations (see, e.g.,

R. Brown 1958:260-262; Brown and Lenneberg 1961:481; Carnes 1970:58;

Carroll 1958:32-35; Carroll 1963:2; Carroll 1966:46; Chase 197l:vi;

~earing 1954:52; Penn 1972:10, 29-30). Much of the discussion centers on the problem of strong versus weak interpretations of linguistic relativity. Perhaps the best discussion is presented by Swanson

(1962). First, he mentions the strong, controversial interpretation-­

that language determines the way the world is seen. Next, he mentions a weaker interpretation--that translation always involves a loss of nuance and emotive value--which is surely true. Third, there is another weak interpretation, maintaining that linguistic categories favor certain organizations, such that the speaker is prone to make certain distinctions. Then, there are also some more unusual interpre­ tations. Wells (1966:704-707) states that Wharf only meant to say that learning different languages is one way of learning other world views, these world view variations being due to differences in learning.

While this seems in line with the position of Lee, it seems to this 17

writer that Wharf was saying something more than this, giving language

the role of a primary determinant of thought. Hymes (196lb:26), on

the other hand, sees Wharf's main goal as having been to develop a

language taxonomy emphasizing cognitive styles. While it is true

that Wharf did indulge in a taxonomic decision, lumping European

languages together as being nearly identical in their world views,

this seems to be beside his main point of stressing the relation of

language to other behavior.

Given the fact that Whorf's writings are so ambiguous, it might be expected that this would have led to much confusion among

scholars investigating "the" Wharf Hypothesis, and Penn (1972 :17)

suggests that experiments with "conflicting" results have often been

testing different hypotheses. Thus, one thing that must always be kept clear is precisely what hypothesis is being considered; labels such as "the Wharf Hypothesis" are too ambiguous for clear communica­

tion.

Let us return now to Swanson's three interpretations of the hypothesis. The second one requires no proof, and will be excluded from the discussion here. The strong interpretation will require a somewhat lengthier consideration. As Penn (1972:33) explains, the strong hypothesis is not capable of being empirically tested and is actually a matter for philosophy. It is most probably false. First, it assumes that thought and language are identical, resulting in the

~hilosophical problem of how language could have been created, since prior to its creation there would be no thought present to develop it (Penn 1972:18, 40). Second, it leads to what Penn (1927:33) calls 18

the "egocentric quandary"; if thought depends on language, it is im­

possible to make any assertions about reality, including language and

thought (see also Black 1962:255-256; Carnes 1970:60-62; Longacre 1956:

303-304; Whatmough 1956:235). Third, inter-language translatability also makes it unlikely that language determines thought (Black 1962:

249; Penn 1972:40), though Swanson (1962:189-192) argues that a single case of inability to translate completely would support this strong interpretation and proceeds to demonstrate the direction in which one might look to find such a case of non-translatability. Fourth, language does change as its speakers learn more about the world

(Whatmough 1956:226), and it changes to meet new needs (Feuer 1953:

95); this indicates that there is some influence proceeding from thought to language, as well as in the reverse direction. Finally, it has been argued that man's thought must be adequate for survival or he would be extinct, so distortions of thought caused by language must be at a minimum (Feuer 1953:97; Penn 1972:40; von Bertalanffy 1955:

257). This objection is not necessarily valid, however. To use Wharf's own analogy, "Just as it is possible to have any number of geometries other than the Euclidean which give an equally perfect account of space configurations, so it is possible to have descriptions of the universe, all equally valid, that do not contain our familiar contrasts of time and space" (1971:58). In other words, there is not necessarily only one or even a handful of appropriate ways of cate­ gorizing the universe; each codification has its own aspect of truth

(Lee 1963a:l03), so it would seem that survival does not require any one categorization of reality. 19

An extreme hypothesis of linguistic relativity, then, appears

to be untenable. We have still, however, to consider Swanson's third

hypothesis, that language influences thought and behavior. This is

a hypothesis that can be tested empirically, and there is favorable

evidence regarding it, but the nature of such linguistic influence

has only begun to be explored.

Indeed, it seems quite reasonable to suppose that the specific

features of a language have some sort of effect on thought. As Brown

and Lenneberg (1961:484) suggest, it is certainly possible that a

child hearing different words is signaled that there are significant

differences to attend to in the nonling~istic environment, and in

this way language may serve to teach him what features in the environ­

ment he ought to pay attention to. Similarly, Carroll (1963:11-13) mentions that the child learning a language must learn to match

language with his preverbal percepts, and this may require some alter­

ation in those percepts; as a result, verbal children come to distin­

guish according to the categories in their own languages, rather than

solely according to a universal human perceptual structure.

Kilpatrick (1955:263) cites an experiment that points out the

high degree to which people take language for reality, rather than

trying to perceive reality in its own terms. Subjects who were told

that the room they were in was distorted from the normal rectangular

prism shape were slower in learning about the nature of the room and

how to deal with it than were those given no such explanation. Thus, when verbal categories are available they are accepted and clung to, with a resulting lessening of attention to the environment. 20

Notes to Chapter I

1. This being an anti-Kant reaction is ironic because Wharf has been charged with displaying aspects of the "Victorian popularity of the decadent part of Kantian philosophy" (Landar 1965:29). The paradox may be resolved if one considers the fact that linguistic relativists, like Kant, conceived of the observer as imposing structure on the world; however, unlike Kant, they held that language, and not uni­ versal categories of human understanding, is the source of this structure. (Swanson 1962:185-186). 2. Ironically, too, since they were rebelling against Kant, Hamann and Herder were as unempirical in positing the source of language as Kant was in positing innate ideas (Penn 1971:53).

3. See Wharf (1971:102-111) for an example of the way in which meaning of a cryptotype may be ascertained. II

SOME PERIPHERAL ASPECTS OF LINGUISTIC RELATIVITY

Granted this potential for language to stand between the individual and reality, what specific evidence do we have to support the hypothesis of linguistic relativity? The first area that may be considered is the effect of language on language perception. For example, Miller (1951:200-204) notes that one's reading speed varies with one's familiarity with the language, pseudo-words are more easily remembered the more they approximate actual letter frequen.:y in the language of the subject, and a subject can more quickly per­ ceive words that are more frequently used in his language. Then there is the work by Murray Glanzer (Miller 1962:751) showing that combina­ tions of nonsense words and actual words are learned better if.they are recognizably analogous to actual phrases in the language. (Thus,

"snat and snoo" would be more easily remembered ·than "snat red snoo.")

Finally, an experiment by Miller and Selfridge (1950) shows that recall of nonsensical strings of words is improved the more they approximate actual strings in the language.

There is much more to linguistic relativity than varying per­ ception of language, however. Another area to be looked at is that of phonological structure and audial discrimination. Although there is no inter-language difference in overall audial acuity, it is evident

21 22

that speakers of different languages attend to different phonetic

features.in.listening and speaking. Differential training in dis­

tinguishing phonemes is the source of the "foreign accent" and of

the fact that it can be understood (Bloomfield 1933:81-85). Thus,

an English speaker can understand the Frenchman's uvular /r/ as the

English phoneme /r/, because he has not been trained to attend to

those differences that distinguish the French and English /r/'s. On

the other hand, the Frenchman may not even realize how deviant his

phonetic pronunciation of English is, since he has not been trained

to attach significance to the difference between a uvular and

English /r/ either.

Liberman and Harris (1957) conducted an experiment to see if

audial discrimination is better across phoneme boundaries than within

them, and their results indicated that, for English speakers at least,

this is indeed the case. Sounds classified in different phonemes were more often judged as sounding different, such that two /b/ allophones would not be distinguished, although either of them would be dis­

tinguished from a highly similar /d/ allophone. Similar results were

obtained in an experiment by Brown and Horowitz (R. Brown 1956:

293-294; Brown and Lenneberg 1958:11-12). Navaho- and English-speaking

subjects grouped eight color chips differently due to a different way of grouping their nonsense names; some of the color names differed only by length of the vowel, which is phonemic in Navaho but not in

English, so most Navaho, but few English-speakers, paid attention to vowel length in grouping the colors. After several trials, however,

English-speakers did learn to attend to vowel.length, indicating the 23 ability to transcend one's language's patterns.

In any language there are definite probabilities regarding what two phonemes will appear next to each other in a word. Roger

Brown (1956:261-262) found that nonsense syllables, when transmitted from subject to subject in a series, tended to be altered into more probable sound combinations. Here is further evidence that one "hears" in accordance with the probabilities of one's language.

Finally, a relevant "thought experiment" concerns onomatopoetic words. Is it not likely that an American hears a rooster's crow more like "cock-a-doodle-doo" than a Chinese? This author, for example, although not quite sure of the vowel sound, hears a distinct "m" in the mooing of a cow. Although such words do imitate nonlinguistic sounds, they are somewhat conventionalized (Langacker 1968:25-26) and therefore might contribute to linguistic relativity.

Linguistic relativity regarding language and sound perception, though, is somewhat peripheral to the main import of the hypothesis.

Evidence regarding meaning is more central to the issue. In line with

Fishman's analysis (1960), the evidence shall be presented in terms of four levels of relativity--codifiability and cultural reflections, codifiability and specific experimental behavior, language structure and cultural reflections, and language structure and specific experi­ mental behavior. A word of explanation is required by these last two levels. It is true, as Dona Brown (1960:339-345), Wells (1962:698-

699), and Marshall (1965:80) argue, that grammatical form and meaning are arbitrarily associated, and these writers maintain that, there­ fore, not the structure itself but what is taught about its meaning 24 is the factor that might influence thought. If this is the case, of course, it greatly increases the difficulty of the linguistic rela­ tivist working with language structure, since he cannot merely study the structure and ascertain its semantic. Rapopbrt and Horowitz

(1960:348-349) suggest that the problem be broadened to state that the relationship to be considered is that between the semantics of language, whether it be taught or inherent, and the correlated behavior. This is surely the safest way to look at the problem, and perhaps our third and fourth levels of analysis should be termed language structure semantics and cultural reflections, and language structure semantics and specific experimental behavior. III

THREE OF THE FOUR LEVELS OF EVIDENCE

The first level of evidence, linking codifiability and culture, is the weakest, most of its data being anecdotal and not rigorously dealt with (Fishman 1960:326-329). Evidence on this level points out that language vocabularies are not mirrors of reality, but rather are part of a process creating things and relations uniquely for each language (Lee 1963c:78-82). The vocabularies of different languages classify reality differently, emphasizing different areas in making fine discriminations (Kluckhohn 1954:938); the conclusion is that these differences are related to differences in interest, environment, and cultural concepts (R. Brown 1956:279-281; R. Brown 1965:317, 336,

338-339; Brown and Lcnneberg 1958:16-18; Cassirer 1965:134; Henle

1958:6; Hoijer 1953:557-558).

Thus, Chinese has more discriminating kinship and ethics vocabularies than does English, due to the greater social significance of these areas in China than in England or America (Chang 1952:22?.). Also, the etymology of Navaho words for anger, evil and warrior is in accord with the cultural attitudes toward these things, the Navaho have no word for "passion" since intimacy for them is more relevant to harmony than emotion, and their word most closely corresponding to "reverence" more nearly means "doing things in the prescribed

25 26 order," the very essence of Navaho religion (Reichard 1950:195-201).

Although Olmsted (1950:11-12) sees music, dance, painting, and sculpture as being outside linguistic relativity, Wescott has not found this to be true for the Bini tribe of West Africa (1960).

Although he finds some cases of non-correspondence between vocabulary richness and cultural importance, he notes that the abundant vocab­ ulary for shape, sound, and motion corresponds to rich music, sculpture, and dance, as well as to pride in having a good body and carriage and a well modulated voice. Similarly, he finds that the dearth of color words is reflected in a lack of pictorial art and in the drab clothes and buildings of the Bini.

It is conceivable that the structures of semantic domains also have other kinds of cultural correspondences. For example,

Lenneberg and Roberts (1956:32) wonder whether esthetics may be in­ fluenced by boundaries in a domain, such that, for example, colors on category boundaries may be less esthetically pleasing than those within boundaries. Further, if Douglas (1966:55) is correct that the Israelites' dietary laws considered unclean those animals not conforming to the norm of their class or to "the general scheme of the world," it seems entirely possible that classifications of reality leave an imprint on religious tabus.

One may also consider the problem of a language that has a given word with multiple meanings. This might lead to certain confu­ sions and identifications in the thought of its speakers. Gastil

(1959:27-29), for example, suggests that the fact that French has only one word denoting both "conscience" and "consciousness" is 27 responsible for the fact that the French have more of a tendency to identify these concepts than do English-speakers.

Now the fact that the Eskimos have several words1 for "snow" does not mean that they have more visual discrimination than we do; we can name these types of snow, too, only not so conveniently (Brown and Lenneberg 1961:481-482). The work of George Zip£ is relevant in this regard. From studies of Chinese, Latin, and English, he arrived at the conclusion (1965:24-38) that terms of high frequency tend to be shorter than those rarely used; probably due to a pressure for economy, when a term comes to be used often it is either shortened or a shorter term is substituted for it. If we generalize from Zipf that the length of any name tends to decrease as it is used more often

(R. Brown 1965:338), we may consider name length as an indication of the frequency of the practice or idea referred to (Doob 1952:98), although this relation does not necessarily hold in every case.

Voegelin and Harris (1945:461-463), while agreeing that the frequency of the use of a morpheme may be indicative of the use of the referent, also suggest that it may be indicative of the referent's prestige; further, they propose that complex words are most likely to have technologically complex referents and be used for more recent innovations.

We may now proceed to discuss the experimental results that comprise the second level of evidence for the linguistic relativity hypothesis, relating codifiability to experimental behavior. This is a level on which hypotheses are easily tested by control experi­ ments and, as a result, there is an abundance of literature on this 28 area. The discussion of level two, therefore, rather than trying to be comprehensive, will be aimed more at presenting a representative

sample of ex~erimental results obtained thus far.

The.earliest experiment in this area, reported in 1889 by A.

Lehman, showed that numbering shades of gray helped subjects discrim- inate between them (Fishman 1960:329). Kurtz and Hovland (1953) have shown that verbalization while seeing an object helps a subject to retain a memory of it. In an experiment conducted by Pyles (1932), children were asked to choose which one of five objects covered a toy.

Subjects given names for experimental objects learned faster than those given no names that the toy could always be found under the same article when the objects were rearranged. Spiker, Gerjuoy, and

Shepard (Carroll 1964:94) found that children who have a given verbal

~ concept do better than those that do not in a problem hingeing on that concept. Similar results were obtained in Bruner's experiment

(1964:4-5), in which children who could correctly verbalize the prin- ciples of ordering of three rows of glasses did twice as well as those who did not in rearranging the glasses using the same principles. And

A. J. Judson (Cofer 1963:208-209) has al~o had similar results, find- ing that male subjects tended to solve the Maier two-string problem in different ways, depending on whether or not they had previously learned a suggestive relevant word list, although no such effect was found in female subjects. All of these experiments show the potency of word presence in facilitating problem solutions.

To return to the conclusions drawn from Zipf's work, expres- sion length is an indicator of cultural concerns and also can be 29 expected to indicate the category's availability and the frequency with which the relevant perceptual judgments are made (R. Brown 1958:235- 236; Brown and Lenneberg 1961:482). A number of experiments have been conducted relating word length to behavior, an area of special interest to Roger Brown and . In one experiment (R. Brown 1958:

239-241; Brown and Lenneberg 1961:485-588), they found that highly

"codable" colors, as measured by expression length, took less time to be named and tended to be given the same name by different subjects and by the same subject on different occasions. Further, they found that codability is highest for central stimuli in a color category, such as "ideal" colors. Another experiment (Brown and Lenneberg 1961: 488-491) showed that codability correlates with category availability as measured by the abilfty to recognize a color seen previously.

When several colors are to be recognized and when there is a longer interval between seeing the color chips and the recognition task, codability is especially important as an indication of the accuracy of recognition. This correlation between codability and recognition also holds true for Zuni, although codability peaks in that language do not correspond with those in English (Carroll 1964:108; Lenneberg and Roberts 1956:31). Thus, English-speakers recognize yellow and orange most easily since they are the most defined English color categories, but Zuni does not distinguish these colors, and Zuni­ speakers tended not to recognize them so well. Another interesting result of this experiment is that bilingual Zunis showed a transition toward the behavior of English-speakers (Lenneberg and Roberts 1956:

30). It appears, then, that this experimental behavior is a matter of 30

degree and is influenced by the individual's unique language experi­

ence. The results of an experiment by DeLee Lantz shed further light

on the problem of codability. She found (Brown 1965:332-334) that

names given highly codable colors were most effective in communicating

that color to another subject, so that he could identify it. She

also found, corroborating Brown, Lenneberg, and Roberts, that codabil­

ity correlates with a subject's ability to recognize a color seen previously.

Brown and Lenneberg interpret their results in terms of

information theory; they conceive of language as providing a means to encode and store information, thereby increasing the availability of certain cognitive modes (Heidbreder 1958:95-96). The codability inherent in a language, which varies with the array of stimuli and

the linguistic community (R. Brown 1965:335), has been shown experi­ mentally to be related to recognition and communication. Brown suggests (1965:336-337) that experiments on the codability of higher level categories, for example, a set of color ~hips, would produce similar results.

Other experiments have made possible some refinements on the importance of codability for behavior. It seems to be irrelevant if the stimuli to be discriminated are presented simultaneously (Lenne­ berg 1961:381-382). Also, it has been found that in some instances it is low codability that facilitates memory (Landar 1965:227-228).

Lenneberg (1961:376-379) suggests that what is crucial is not codability per se, but semantic structuring. In a particular array the best land­ mark for memory may be either a point of transition or a point of easy 31

labeling. Thus, for example, in an array of poorly codable color

chips, the point of easiest recognition may be the chips of lowest

codability (Lenneberg 1961:382). Semantic habits provide an anchorage

point of a different sort depending on the array and in any case, this anchorage becomes more important as memory is taxed (Lenneberg

1961:382).

In a highly complex experiment on codability (Lenneberg 1957),

Lenneberg obtained information that allowed him to ascertain the semantic structuring of the color spectrum for ·speakers of American

English. That is, for each color chip he used, he could plot the • frequency with which it would be called by different names. He then tried to teach English-speakers new names for the color chips. By varying the probabilities with which different chips were given different names, he was actually trying to teach his subjects a new semantic structuring of the domain. A shift from English in position of highly codable colors, and especially in probability gradients for labeling transition colors, made learning difficult. Thus, learning is easier when there is little distortion from the structure of a domain learned by a subject as part of his native language.

Questioning the results of these codability experiments,

French (1963:410-411) wonders whether the problem involved may not be somewhat more complex. He suggests that there may be such a thing as cultural codability that supplements linguistic codability. As he sees it, there may be easily available unnamed categories, and there may also be long, rare names that communicate well, such as "Swiss­ cheese colored." While there may b·e something to what he says, it 32

would seem that only an exceptional highly available category would

remain unnamed, given man's linguistic inventiveness and his dependence on language. And while certain long names may communicate well, they

are, as he says, conceived of and used only rarely and thus have only

a limited effect on behavior. In short, there may be such a thing as

cultural codability that is not isomorphic with the linguistic

structuring of domains, but it is likely that its effects are greatly

restricted.

To conclude this discussion of codability, codability is related to frequency of expression and frequency of perception and

thought of a category (R. Brown 1956:307-308; Brown and Lenneberg

1958~16). It should be remembered that codability can vary not only between languages and cultures, but also between subgroups within

one language and even between individuals (R. Brown 1958:255-258;

Brown and Lenneberg 1958:15); Western skiers can talk about as many different types of snow as Eskimos (R. Brown 1956:311-312). Another set of experiments pertaining to linguistic relativity

has explored the effects of the labeling of ambiguous figures. Per­ haps the first such experiment is reported by Bartlett (1961:16~21), who found that the names subjects assigned to detailed designs affected their reports of what they had seen, though no such effect was noted with simpler designs. However, it is the experiment of Carmichael, Hogan, and Walter (1932) which seems to have been the

inspiration of a long series of studies. This experiment determined

that the reproductions drawn by subjects of figures seen previously

tended to be assimilated toward the labels the experimenter had given 33 them originally. It has also been found (Hanawalt and Demarest 1939) that labeling during the recall period results in this assimilation effect. This effect increases the longer the delay between exposure and reproduction. Labeling affects immediate reproduction, and not only longer term memory (Bruner, Busiek, and Minturn 1952). The influence of labeling increases when subjects are not told prior to the experiment that they will be asked to reproduce the figures, and there is a tendency for the labeling effect to be greater the shorter the exposure of the original figure (Herman, Lawless, and Marshall 1961).

In view of these positive results, it is somewhat surprising that Prentice (1954) found no significant labeling effect when subjects were merely asked to recognize the figure seen originally. Either his results are anomalous, or we must conclude that labeling does not affect perception or memory, and only affects reproduction. Bartlett's findings mentioned above make the former explanation a definite possi­ bility. Finally, Lenneberg (1961:380-382) found that delayed recogni­ tion of a color was not influenced by its label when first seen. He concludes that thi~ could be because the stimuli were not ambiguous enough. In view of Prentice's results it could also be that labeling does not affect recognition.

Some other experiments test the expressed by Greenberg (1954:11-13) that meanings that are grouped together will correlate with similar behavior. This is a special case of the hypoth­ esis of acquired equivalence of cues--if a subject is taught to make the same response to dissimilar stimuli, he will have an increased 34

tendency to generalize other behavior from one to another (Spiker

1956:254). This hypothesis has been confirmed by many experiments,

many of them using identical names as the similar response made to

dissimilar stimuli (Spiker 1956:254-256, 258-259). In such cases

the subjects are found to generalize a response to all stimuli given

the same name even though the response is taught only for a given subset. Thus, in Shepard's experiment (1956), children taught to

give the same name to red, orange, and yellow lights generalized a

response to be made to red lights to orange and yellow significantly

more than subjects who were not taught the nonsense names for the lights. Similarly, Jeffrey (1953) found that children taught to make

differential responses to white and black patches generalized one or

the other of these responses to gray patches, depending on whether

they were taught, either through naming or a motor response, to include

gray with black or white. Further, it was found that naming served

as a better mediator than motor responses. A counterpart to the hypothesis of the acquired equivalence

of clues is the hypothesis of acquired distinctiveness of cues, which

states that if a subject learns to respond differently to similar stimuli, he will learn more rapidly to make other discriminal respon­ ses (Spiker 1956:254). Experiments testing this proposition with

adults have had conflicting results, but experiments with children have confirmed the hypothesis, and it has been discovered that names are more effective in teaching discrimination than merely learning

to determine whether two stimuli are the "same" or "different"

(Spiker 1956:256-259). Thus, Rossman and Goss (1951) conducted an 35

experiment in which subjects who had mastered nonsense names for similar figures did better in learning to make.differing motor re­

s~onses to these figures than did subjects who had only minimal

exposure to the different nonsense names.

All of these level two experiments except the one concerning

Zuni color codability are intra-cultural, yet they are pertinent to

linguistic relativity since they show how differences in the semantic

structure of a domain or the presence of a word may be related to

behavior. One of the few inter-cultural experiments along these lines

was conducted by Doob (Kluckhohn 1961:904), in which he found that,

in correspondence with the poor color vocabulary in the language of

the Baganda tribe, when these people were given cardboard pieces to

sort they used size or shape rather than color to perform the task,

unless they were educated in the West, in which case they used color.

In another inter-cultural experiment conducted by Carroll (Carroll

and Casagrande 1958:22-26), - and English-speaking subjects were asked to choose which two out of three pictures belonged together.

Due to differences in the denotata of the words descriptive of these pictures in the two languages, it was predicted that the two groups of subjects should sort the pictures differently. The results showed the expected tendency, but were not significant. Carroll suggests that the results might have been more favorable had monolingual been used, and in view of Lenneberg and Roberts' findings that bi­ linguals were transitional in codability experiments, this suggestion seems quite reasonable.

Before turning to the third level of evidence, the fourth 36

level, which can be covered in a short space, will be considered.2

Very little work has been done on this level, that of correlations

between language and experimental behavior, perhaps because of its

great difficulty, since no longer in question is a language's vocabu­

lary, but its grammatical structure. Greenberg (in Hoijer et al.

1954: 208-209) and Dona Brown (1960:341-343) both raise the question

of whether or not one may say that a linguistic class actually has a common meaning that is extended to all of its members, since a minority within the class may seem to differ from the majority meaning.

Rapoport and Horowitz (1960:360) point out, however, that when a

child, by analogy, utters the word "knowed," this constitutes proof

that the suffix "-ed" has a meaning for him. Happily, there is even more proof than this logical piece of reasoning. Roger Brown found himself asking this very question of whether a language's structural features are associated with a semantic (Brown and Lenneberg 1961:

483), so he set about to devise an experiment that would shed some light on this problem. In this experiment (R. Brown 1958:244, 246-

253; R. Brown 1961:504-508), he discovered that English-speaking children will point to the appropriate picture in response to a nonsense word used in a frame as a verb, mass noun, or particular noun. For example, when asked to point out a picture of

"a sib," the children usually chose a picture of an object, rather than a picture of movement or an extended substance. It appears, then that children do discover semantic implications of grammatical classes. The problem remains to ascertain how this semantic structure affects thought in both the child and the adult. 37

Two experiments have attempted to determine just this. One of these, conducted by Joseph Casagrande, was carried out as part of the

Southwest Project in Comparative . This project, which began in 1955, was an attempt to test hypotheses of linguistic relativity and universals among the Navaho, Hopi, Zuni, Hopi-Tewa,

Santa Clara Tewa, Spanish-Americans, and native English-speakers

(Casagrande 1960:777-778). On the whole, this project did not provide a great deal of support for the linguistic relativity hypothesis

(Carroll 1963:13-14; Greenberg 1959:1169; Greenberg 1963:138; Hoijer

1966:123), but not all of its results were completely negative, including this next experiment.

Navaho verbs of "handling" take different stems depending often on the form of the object. Because of this, Casagrande (1958: 26-31) predicted that Navaho children would use form as a criterion of sorting earlier than English-speaking children, who it was known attend to size and color before they learn to attend to form. In the experiment subjects were asked to say which of a pair of objects a third object went best with. The objects were selected such that the criterion the child used would be immediately apparent from his choice of what objects belonged together. The results were that Navaho- and predominantly Navaho-speaking children did tend to use form in sorting when possible, while English- and predominantly

English-speaking Navaho used form less frequently. Bostonian children, however, showed the Navaho response pattern. Casagrande suggests that this unexpected result could be because playing with toys empha­ sizing form problems leads to form-sorting just as the linguistic 38 structure of Navaho does, and Bostonian children are likely to have had much contact with such toys. Casagrande concludes that use of form as a criterion to match objects increases with age, and such behavior can be enhanced either by learning to speak a language that requires the use of such distinctions or by playing with toys in­ volving form problems.

Maclay (1958), testing the same prediction about Navaho form classification, also found it to be incorrect. His explanation is that other possible linguistic classifications in Navaho may have interfered with the expected sorting behavior, and that frequency of use of different linguistic classes is needed to predict behavior.

There is also the problem that he did not distinguish English-speaking

Navahos from monolinguals in his experiment, a fact that, in view of

Casagrande's results, is highly significant.

Luh (1948:118) has argued that it is absurd to maintain that thought in a bilingual changes as he uses dii t'~rent languages alter­ nately. Actually, this is not what this weak interpretation of linguistic relativity implies. Rather, as several of the experiments supporting levels two and four show, bilinguals' behavior differs only in degree from monolinguals'. It is not so much that they are parties to two distinct kinds of thought, but rather that they have incorpo­ rated two different systems of significant features and blended them together, such that they manifest transitional behavior.

To conclude the discussion of levels two and four, it is apparent that Longacre (1956:304) is incorrect in saying that what is most important is whether languages can transcend their own categories; 39 the mere fact of categorization has an effect on behavior, and this must be understood if an individual is even to see a need for trans­ cending the categories that seem natural to him. 40

Notes to Chapter III

1. The exact number of terms cited varies (cf., e.g., Boas 1930:78; Brown and Lenneberg (1961:481).

2. Before proceeding to level four, it should be mentioned that work on componential analysis also pertains to the second level of evidence for linguistic relativity. A good deal of work has been done to determine the features that people attend to as being defining of a particular category, especially in the area of kinship terms. Marshall (1965:80) considers that such studies show "limited instances" supporting linguistic relativity, but discussion of them will be deferred to allow a comprehensive discussion of the special topic of this thesis, level three, or the relation between grammatical structure semantics and culture. IV

THE ANTHROPOLOGICAL CONTRIBUTION

The third level of evidence for the hypothesis concerns

correlations between language structure and culture. It is similar to level one, having the advantage that it deals with more significant language patterns than vocabulary (Hoijer 1953:557-558), and the disadvantage that there are more complexities to be considered. This level includes the third kind of evidence Wharf used to support his thesis. He maintained (1971:156-159) that there are "connections" but not "diagnostic correspondences" between language and culture; the two grow up together, but since language is a system its influence is more limiting. This latter notion should be disposed of immedi­ ately. Culture is systemic as well as language and, as· a system, it too is resistant to change; if, in fact, language is more limiting it would therefore have to be for some other reason, and Hoijer (1953:

566-567) suggests that it would be due to language's close inter­ penetration with all other cultural systems. Wharf's statement is rather problematical in that it is hard to determine exactly what he meant by "connections" and "diagnostic correspondences," but it appears that he may have been indicating that he felt there is no way to predict culture from language. This is a problem which will be re­ turned to later. The potential relationship between language and

41 42

culture is, of course, diffi~ult to discover. There has been much discussion, for example, on the nature of this relationship, be it causal, correlational, or congruent (Lounsbury in Hoijer et al.

1954:270; Levi-Strauss 1953:4; Wescott 1960:21), and whether or not there might be reciprocal influence between language and culture

(Chang 1952:208-209; Greenberg 1954:28-29; Voegelin and Sebeok 1953:

48; von Bertalanffy 1955:255). As Gastil (1959:24) and Glenn (1954:

165) maintain, culture may influence language development and, in turn, be influenced by language as it leaves its impress on current users. Besides such theoretical questions, there are more practical problems for investigators who broach this topic. Thus, even if there is a correlation between language and culture in some instances, it is entirely likely that language may lag behind cultural change (Cohen

1956:45-46; Voegelin and Sebeok 1953:30-31), so that it may not be a reliable indicator of culture, especially when a culture is experi­ encing or has recently experienced a great deal of change. And there are other problems to consider, such as the presence of multiple dialects, use of different linguistic codes in different situations, differences in behavior of subcultural groups, and differences in the use of language in different communities (Cohen 1956:39, 43-47; Hymes

196lb:36-37). Hymes (196lb:42) considers the difficulties attendant on such an investigation to be so great that he despairs of the possi­ bility of determining the language-culture relationship for many peoples.

Considered on this third level of evidence are the special areas of a culture's philosophy and science. Whether or not language 43

is reflected in philosophy and science has been a matter of debate

for some time. Thus, maintained that sound philosophy

demanded linguistic analysis (Feuer 1953:85), and he was not alone in

his view that there is an intimate connection between language and

philosophy and logic (see, e.g., Alexander 1936:261, 263; Carroll

1958:35; Cassirer 1946; Radcliffe-Brown 1957:120). Others take a

more moderate stance, advising caution against exaggeration concerning

correlations (Heidbreder 1958:101-102; Hockett 1948:569-570; Voegelin and Sebeok 1953:66). Then there are those who argue that an axiom

system constructed by specialists is independent of natural language

(Radnitzky 1961:156), and that philosophical systems have, in fact,

not varied between different language families (Feuer 1953:86-91). A

survey by Basson and O'Connor (1947:60-63), although showing some

possibly important language variations, provided no evidence, in the

authors' opinion, that unresolvable philosophical problems are due to

the inadequacy of a language. A statement in accord with the weak

interpretation of linguistic relativity being discussed here is pre­ sented by Luh (1948:119) when he says that, although language need not change as philosophy changes and the same idea can be expressed

in different languages, a language may influence a first formulation of a theory by an individual. In other words, language is not an in­

surmountable barrier to philosophical thinking, but it may have a very real effect before a metaphysical system is subjected to rigorous examination. Similarly, Whatmough (1956:232,240-241) and Sapir

(1963a:l0) remark that language must not be allowed to contravert accumulated scientific knowledge, that its implications should be 44

fought if necessary, or else the language should be changed. Such a view implies that science is not completely constrained by language, though it may be hindered by it.

Before proceeding to a summary of the literature concerning

specific proposed language-culture correlations, one possible argument against the possibility of such correspondences should be dealt with.

It might be argued that language and culture areas do not coincide, and this is most certainly true, but close attention must be paid to the criteria used to define such areas. It will become apparent that the defining features used are inappropriate for the problem at hand.

Boas (1904:518-519), while a believer in linguistic relativity, noted that the traditional language and culture classifications clash and, indeed, could only be expected to. Thus, Roger Brown (1956:

306; 1958:258-259) and Hoijer (1953:567; 1954:104-105; 1964b:455) argue that culture areas are defined by discrete and easily diffused traits, rather than more "genetic" traits relevant to cultural premises, and criteria of language areas are often phonemic and func­ tional rather than semantic traits relevant to cognition. Wissler's

North American culture areas, for example, are defined by such traits as the presence of the dog, basketry, and developed wood work, what the primary food resources are, the level of tool development, and what resources are used for tools (1967). It does not seem surprising that areas defined in such a manner do not correspond to areas defined in terms of semantic structures, let alone language families. We are faced with the problem of how to define culture and language areas so that any relationships of influence might become apparent (Kluckhohn 45

1961:897-898). Bittle (1953) presents a most complete discussion of

this problem. He notes that not only are culture areas based on the

principle of diffusion, but they do not distinguish between diffusion

and genetic relationships, resulting in a confusion of traits.

Language areas, on the other hand, are consciously based on genetic

relationships, phonetic correspondences between language cognates. As

Bittle explains, while areas defined in this manner are not cotermi­

nous, if culture areas were based on genetic factors, or if language

areas were defined by large-scale linguistic patterns and culture

areas by large-scale behavior patterns, they might be found to

coincide. At any rate, redefinition is mandatory if such areas are

to be compared to test language-culture connections.

Kluckhohn (1954:939) has said that validation of the linguistic

relativity hypothesis requires comparison of linguistic and non­

linguistic aspects of culture as well as psychological experiments.

Experiments that have been performed thus far have benefited from

the presence in psychology of a strict methodology by which experi­ menters could guide their work in order to obtain significant results.

Anthropologists, however, have not had such a methodology to rely on, and the consequences of this will become apparent upon examination of the literature concerning cultural reflections of language struc­

ture semantics.

Radnitzky (1961:156-157) suggests that languages may have built-in value systems due to what they can easily explain, as, for example, church language cannot speak about criminals as approvingly as can gang language. This approach, however, does not seem to have 46 been a popular one, and there are no specific studies investigating this type of Whorfian effect.

Some papers have concerned themselves with correlations between languages and cultures, without presenting any specifics of such correlations. Thus, Kluckhohn and Leighton (1962:253-256, 273-279,

293) stress the importance of language study if one is to understand the Navaho, but they never directly state specific cases in which this is true. Thompson (1946:61, 63-66), using the term "psycho-socio­ cultural whole," claims that language and culture have an "inner coherence" and that the Hopi have an especially good integration due to their sedentary, isolated existence. Unfortunately, the specifics of this integration are not disclosed. Hymes (196lb:38-41) analyzes

Wishram-Wasco Chinook, demonstrating that within the past few hundred years changes have occurred in the language that stress the relation between two terminals. He states that there are cultural analogues of this process, but as these are not his prime concern in this article, he fails to enumerate them.

Other scholars have proposed some very general correlations.

Levi-Strauss (1951:159-163) suggests comparing language and kinship at their deepest levels of rules and combining elements in the various geographical areas of the world; he wonders, for example, whether the typical Indo-European kinship system comprised of simple rules joining complex groups correlates with the linguistic feature of simple structure used to put together numerous elements. Altshuler

(1956) has posited that societies centered on the female (matrilineal, matrilocal, etc.) have languages centering on verbs, while societies 47 centered on the male (patrilineal, patrilocal, etc.)_ have languages centering on nouns, and he backs this up with data from North America and Africa. Glenn (1954:172, 167-168) correlates different types of word formation with different logics and , with the proviso that these correlations will hold in specific areas within a language and culture, rather than throughout an entire language and its culture.

Most scholars who have dealt with this third level of lin­ guistic relativity have proposed specific correlations in specific cultures and languages. These studies will be presented in the order of the following geographical areas: Asia, the Pacific, North Amer­ ica, and Europe.

Chinese has been a favorite language for linguistic rela­ tivists to consider. Hockett (1954:111-122) notes several mainly lexical differences between English and Chinese without speculation on any connections with culture, but he also wonders whether the

Chinese manner of attributing qualities located relatively on a scale, as opposed to qualities dichotomously opposed, is related to the

Chinese doctrine of moderation. Hockett has been commended by Good­ enough (1955:242-243) for confining himself to lower level, "tangible if unspectacular" examples, but as a matter of fact the article presents relatively little of significance, perhaps because of the low level of the examples used. Chase (1954:80) is in agreement with

Hockett that the Chinese language is "multi-valued," that is, it allows expression of qualities on a scale rather than pigeon-holing them, and he links this language trait with Chinese tolerance and lack of fanaticism. A similar idea is expressed by Leary (1952:201).

Chang (1952) is the author of a somewhat more extensive

analysis of Chinese language and culture, frequently pointing up their

differences from English. He argues that Chinese ideographs are re-

sponsible for an emphasis on relations between rather than on

underlying substance; Chinese does not inflect word stems but creates

a new for each new word. He indicates that the Chinese lack

of subject-predicate sentence structure makes the language unfavorable

to Aristotelian logic, which is based on the subject-predicate

distinction. The Chinese language is, in his point of view, rela-

tional, not relying on dichotomous exclusion as a language such as

English does. Thus, Chinese has a "correlation logic," emphasizing

the relation of opposition, rather than mutual exclusiveness. The

resulting lack of the law of identity in Chinese is, he feels, tied

in with the emphasis on analogical thinking in China, which in turn

results in a socio-political emphasis in the culture, since analogical thinking is more appropriate to such an area than to science.

In the West, on the other hand, the presence of the law of identity may be correlated with the use of the syllogism and the resultant emphasis on science. Emphasis on socio-political affairs is respon- . . sible for a concern in China with finding out the "how" of things;

Western countries are more concerned, however, with the 11what," since

they emphasize religion and science. The Western concern with science and religion may also be linked to their concern with substance, which in turn leads to the idea of causality, these ideas being less familiar to the Chinese. Chang conceives of the Supreme Being as 49

.being related to the idea of substance, so the absence of this idea in China correlates with the Chinese religion which is not based on any divine entities. Chase (1954:80) quotes Chang as saying that qualities in Chinese are complements rather than substantive oppo­ sites because Chinese lacks the sentence structure, "X is adjective."

Luh (1948) also considers the relation of the Chinese language to the logic and science of that country. He first asks the question whether the fact that Chinese uses classifiers when dealing with space but rarely when dealing with time is due to an early conception that time and space are essentially different. Henthen explains that early Chinese philosophy shows little discussion of these concepts, so the thesis cannot be supported. Further, Indo-European treats time and space as if they were similar, yet the idea that time is actually a fourth dimension came as a surprising discovery in its science. Luh also notes that philosophies similar to that of Plato's and empiricism developed in China despite the dHferences in language.

Further, he finds that the Chinese language may, with a minimum of change, be used for modern science and, while a foreign language must be learned to deal with Western logic, there is no great difficulty in learning the language and logic. In fact, the only case of lin­ guistic relativity that Chang appears to admit is that one has a shorter memory span for digits in a language that is not one's native language. He closes, however, by suggesting that correspondences between language and thought might be greater in ideas concerning the relation of man to nature.

Win (1960) presents an analysis of the Burmese language, such so analysis being, in his opinion, necessary if one is to understand the

Burmese mentality. The lack of tenses in the language can be related to subjective time, a stress on immediately experienced events, and a dislike for planning ahead by making appointments. The fact that subjects are often omitted from sentences implies that action and change are more important than actors. This is further emphasized since each action must take a new sentence. He also argues that the language presents space as subjective. For example, an individual's name changes depending on the person who is speaking of him. Win's main thesis is that these characteristics may be related to the

Buddhist religion. In Buddhism space and time are subjective, and objects are relative to the perceiver, just as the perceiver is rela­ tive to his consciousness. Thus, in Buddhism, as in the Burmese language, existence is found in immediate experience. Since objects are relative, the language concentrates on verbs.

We turn now to work that has been done on Pacific languages and cultures. Following in the tradition of Durkheim and Levy-Bruhl,

Sommerfelt (1938) resurrects the concept of primitive mentality in the name of linguistic relativity. Adhering to the dictum that social facts must be explained by other social facts, he reasons that society is the source of language, which is the source of thought and logic.

He analyzes the language of the Arunta, arriving at the conclusion that it is a language solely of verbs. That is, just as the word for "man" consists of two morphemes, "to throw" and "to fight," other objects are also designated by words divisible into morphemes denoting actions and states. In such a language, actions and qualities can 51 manifest themselves in diverse ways, and these are the basic concepts, since there are no true objects. This would lead to thought based on the Law of Participation, as Sommerfelt sees it. On the other hand, he argues, since Indo-European languages do have objects, the Law of Participation seems absurd to speakers of them, and this is the source of the repudiation Levy-Bruhl has received.

Another tribe of Australia, the Ngarinyin, has been discussed in terms of linguistic relativity by Capell (1960). In the Ngarinyin language nouns are divided into five or six classes for the purpose of alliterative concord throughout sentences. In this grammatical structure Capell sees the semantic of "undifferentiated 'nounhood' subdivided into classes as the people and objects are individualized"

(1960:7). He feels that this semantic is related to the Ngarinyin creation of the "origin of the many from the One." To support this he notes that, in Arnhem Land, there is no notion of primal substance and no noun classification. But in fairness, he also remarks that noun classification similar to the Ngarinyin is found among the Bantu and in Southern Bougainville, where the Ngarinyin world view does not prevail. Capell concludes that the Ngarinyin correlation is due to a pressure of the world view on the language, rather than the reverse process, which he feels Whorf advocated.

Although Dorothy Lee is perhaps best known for her work on the Wintu Indians of California, she also studied the Trobriand

Islanders through the accounts of Malinowski and, daring to differ with her source, came up with some unusual conclusions about the language and behavior of these people. These conclusions may be 52

found in three of her articles (1940; 1959; 1963a). In the Trobriand

language a noun stem often changes completely when its referent's

qualities are changed. In other words, change in an object makes it

something else, such that a ripe yam is referred to by a different

name than a rotten yam. This implies that being is valued, and not

change. Further~ being is evaluated discretely, rather than through

comparison; the language does not have many morphemes signifying

comparison, and words in a sentence are discrete and not relational.

Tense and temporality are not developed in the language, events are often not presented chronologically in speech, and there is no word

for causation or teleology; this is mirrored by a lack of concern with causality and purpose. Much of Trobriand life is centered on gift exchange, yet the value of such exchange is not in the gift

the giver will in turn receive, but in conforming to a cultural fat tern. In fact, Lee says, purposive behavior is despised by these people. Lee also detects a lack of linearity in Trobriand speech and says that this too correlates with a lack of purposive behavior, with valuing a sameness that follows the mythical pattern. This absence of linearity also correlates with the fact that being is valued discretely rather than in relation to other elements. One participates in a pattern because of the value of the pattern and the sameness in following it, and not for the purpose of change or progress. Thus, in their language as well as in their ethics, for the Trobriand Islanders causes and consequences are not evaluated, because the act or object is an end and reason in itself, having its own evaluation. 53

Graves, in a critique of Lee's work (1957a), has argued that

Lee is too quick to say that there are differences between the Tro­ briand language and culture and the West. Many of her examples, he claims, have parallels in English. Thus, English too has nouns that change when the qualities of the referent change; the terms for a salmon, for example, vary depending on the fish's life stage. And while Westerners may generally feel that gift-giving in hope of return is acceptable, there are at least some occasions when it is not-­ hence such terms as "gold digger." Although the line and purpose may be important in the West, its etymology shows its recent development, since its Latin source did not mean "lineal." Also, non-purposive behavior is to be expected in a place such as the Pacific where the necessities of life are easily obtained, but even in the West there are ~arallels, such as children's games. While the Trobrianders do not think of sex as a cause of pregnancy, contraception in the

West has also separated erotic love and parenthood. Finally, Graves notes that, in his eyes, the Trobriand Islanders do at least show pur~ose in their erotic behavior, and whether it motivates them or not, there are often rewards, such as celebrations, following valued activities. Next we proceed to literature on the North American Indian.

Hoijer (1964b:465) menti~ns that the Navaho use of special enclitics when speaking of the dead corresponds with their avoidance of the dead and that the presence of dual number in Apache and Tewa correlates with the fact that many of their deities are twins, or siblings, or come in pairs. Hallowell (1965:64-66) considers the effect of 54

animate and inanimate noun categories in Ojibwa on the perception of

these people, concluding that objects that are in truth inanimate

but are grammatically animate are conceived of as having a potenti­

ality for animation. Mathiot (1964), in an analysis of Papago noun

classification, discovered that criteria used to distinguish the

different classes were often based on gradual perceptual differences.

The language 1 s number system ''as seen to reinforce gradual perceptual

opposition. From this she predicted that the Papago would place

more emphasis on perception than on conception, and that they would

tend to think in terms of relative qualities rather than binary

oppositions. Her predictions were confirmed by a psychologist

familiar with the Papago.

Seeking to discover the Wintu cultural premises through their

language, Lee (1939:181-182; 1944:362) discovered that for these

people reality is unbounded--the generic is primary, with particu­

larity being imposed temporarily. Thus, nouns are primarily

generic--the singular and plural use different roots, specifying number is optional. Garvin (1957b:69) cautions, however, that we must remember that in some instances number is optional in English

too, as when one says "fish for trout." Categories are imposed diffidently by the Wintu (Lee 1939:182-183)--mandatory verb suffixes tell the source of the knowledge being stated, and special verb stems are used if what is referred to is "beyond personal cognition." A preoccupation with form is seen in word etymology; things similar in form are referred to by similar words, and this corresponds to the emphasis on shifting form in Wintu mythology (Lee 1939:184). The primacy of the whole and of society is seen in the fact that one 55

says, "I am ill in respect to my child," rather than "My child is

ill" (Lee 1939:185). Garvin (1957b:73) notes, relevant to this,

that, like the Wintu, we at times regard the other as part of the , as when a nurse uses "we" to refer to her patient. There is no Wintu word for the self or for the body as•. opposed to the controlling mind (1963b:l34). The low status of the self is seen in the fact that judgments must be referred to the source of knowledge, as mentioned above, the first person is derived from the third, and words and their extensions are based on observation, rather than kinesthetics or ~ntrospection (Lee 1963b:l38). The verbal suffix indicating the source of knowledge is considered by Lee (1939:187) to indicate a non-aggressive attitude toward reality, and she feels that this is re- lated to the Wintu's courteous and intimate relationship with nature.

An analysis of the Wintu verb system (Lee 1938) shows two categories of verbs--one in which the speaker participates through indicating the source of his knowledge and in which the grammatical subject

~articipates as a free agent and has an attitude toward the action or idea expressed, and another in which the individual is not a free agent, there is no reference to the speaker, and the stem implies helplessness, ignorance, or unrealized future events. From this Lee concludes that the Wintu think in a small field surrounded by a world of unknowable and ineffable necessity, the second category of verbs indicating this unknowable world. Lee (1938:102) asserts that this systemic grammar, expressing the basic Wintu categories of subjec- tivity versus objectivity, knowledge versus belief, and freedom versus natural necessity was created unconsciously, since the Wintu are not 56

aware of their own Weltanschauung. Note that, like Capell, Lee

believes that world view influences language, rather than that language

influences world view.

Of all the American Indians, perhaps the Navaho have been

mentioned most often in discussions of linguistic relativity. Kluck­

hohn and Leighton (1962:285) noted a great difference between Navaho

and English, in that the former stresses verbs, while the latter

stresses nouns and adjectives. Although they say that this implies a

"different order of thought habits," they do not specify what this

might be. For this we should look to work by Hoijer (1964a:l43-147)

and Astrov (1950). Hoijer argues that the Navaho verbs stress

movement--derivational prefixes are used to specify the nature of movements, verbs that do not seem like movement to a speaker of

English are built up on themes for movement, some substantives are

denoted by the same words as movement verbs, and even neuter verbs

suggest state as being the result of the withdrawal of action. This

emphasis on movement is paralleled by the nomadic existence of the

Navaho and their mythology, which often deals with traveling. This mythology is further elaborated on by Astrov; the creation myth is a

story of travel and creation through motion, myths of chant origins

are often travel accounts of heros, .. in a curing ceremony the patient

becomes a traveler through song in order to make himself holy, the most revered god is the god of motion, and so forth. Astrov suggests

that because of this cultural theme of movement, the Navaho are

adaptable, not persistent, and have no tendency for group coopera­

tion or planning. 57

Hoijer (1954:100-102; 1964a:l47-148) has made another asser­

tion about the Navaho concerning their adaptability, their non­ aggressive attitude toward nature, and their religion emphasizing balance and harmony between man and nature. He correlates these cultural traits with the verb trait mentioned earlier that verb stems in some cases vary with the object or, if the verb is intransitive, with the subject. This has the effect that the stem ties a class of entities with an action, the object suffix linking an individual with that class of entities in motion. Thus, the subject of an intransitive verb does not perform action, but is associated with it, and the agent of a transitive verb does not perform action, but merely links an object with the movement of its class. In Hoijer's eyes, in both the religion and the language of the Navaho, there is an adjusting of the individual to a given universe.

Hoijer has been criticized for this statement of his.

Carroll (1963:9) calls this a "tenuous" correlation, arguing that the language trait in question could be interpreted in an opposite manner to indicate a "physicalist" world view. Here we are running into the problem of what is the semantic that is associated with a given grammatical structure by the people that actually speak the language, rather than by the linguist. Also, while Hoijer is correct that the Navaho religion does emphasize respect toward nature and a balance between man and nature (see, e.g., Downs 1972:82,

95-96), the fact remains that, although the Navaho religion doubtless is unique in its particular flavor, all other Indians have a reverence 58 toward nature that is in stark contrast with Western man's attitude

(Wissler 1966:214). Not only this, but a reverence for nature is to be expected among any people with an undeveloped technology, who therefore live close to nature and are conscious of their intimate dependence on her. What, then, does Hoijer mean by positing a language-culture correlation in which the culture trait is virtually a universal among hunting and gathering people? The argument is, actually, meaningless, for there is no way in which it can be proven.

The same criticism may be leveled against Lee's analysis of the Wintu discussed above, in which she ties their non-aggressive attitude toward nature in with their language structure. To conclude this discussion of the Navaho, the following quote from Carroll (1966:48) seems appropriate: "One may rightly remain skeptical as to whether this evidence truly demonstrates any fundamental relation between linguistic structure and world-view."

We come now to the work of Wharf himself on the Hopi Indians, the third kind of evidence he used to support the linguistic rela­ tivity hypothesis. Wharf stressed the differences in thought, language, and behavior between the Hopi and Standard Average Euro­ pean (hereafter SAE), most, if not all, European languages being considered by him to be essentially the same for the purposes of his investigation. In "The Punctual and Segmentative Aspects of Verbs in Hopi" (1971:51-56), Wharf says that the Hopi, due to their segmen­ tative aspect, have a language that can deal with vibratile phenomena, and therefore modern physics, better than our own. Carroll (1963:8-9) argues that, if this is the case, it is surprising that it is quite 59 easy to gloss the segmentative in English, and that Whorf goes too far in associating this verb aspect with the ability to think in terms of the physicists' particles and fields. He urges more caution in making such inferences.

In another article on the Hopi (1971:59-64), Whorf explains that each language has certain terms that contain in themselves the basic philosophical premises of a people and their culture. For the

Hopi these are tunatya, subjective hoping or striving to manifest, and tunatyava, the objective end of hope through coming true. These terms correspond to the Hopi two "grand cosmic forms," the manifest­ ing and subjective, and the manifested or objective, which take the place of tense in this language.

Whorf's most significant work on the Hopi, however, is in his contrast between it and SAE. He notes (1971:139-146) the follow­ ing aspects ~f the Hopi,language. There are plurals only for entities that can form an objective group, such that the word for "day" has no plural. There are no mass nouns, every noun implying its container or ;form, so that there are no such expressions as "stick of wood," which imply a duality of the formless and form. Seasons and phases are not objectified as nouns, but remain subjective, and there is therefore no implication of a formless "time" as an objective entity.

There is no tense to objectify time. And finally, space terms are not used to express duration, intensity, and tendency, as when one says in English, "a sharp voice." These linguistic traits are said to correlate with the Hopi culture. The meaning of subjective time, according to Whorf (1971:148-152), is that the present carries its 60

impress into the future so that all effort is accumulated and, cor­

respondingly, the Hopi have an emphasis on preparation and repetition

in their culture. Since the language involves no imaginary space, when the Hopi thinks of something existing in space, he has no choice but to believe that his thoughts are traveling in real space and actually influencing nature (Whorf 1971:149-150).

Carroll (1963:8) says that there has been no "adequate critique" of this supposed Hopi world view; the task would be extremely difficult, requiring a thorough examination of the language and then a demonstration that the world view is correct. One of the language traits Whorf used to make his point, however, has been subject to some question. Greenberg (1954:14) questions whether Whorf was correct in saying that the Hopi must say "the xth day" rather than

"x days," and at any rate is skeptical of any world view being attached to such a trait. Whatmough (1956:225-226) adds to this point by noting that the word "day" is not pluralized in Latin either, a language that presumably falls into the category contrasted with

Hopi--SAE.

Let us now look at how SAE contrasts with Hopi in Whorf's view. The following traits in the language are listed in Whorf's comparison (1971:139-146). Subjective experience is objectified by being pluralized--"ten days" takes the same form as "ten men."

Phrases individualizing mass nouns, such as "a stick of wood" imply a duality of the formless and form. Seasons and phases are objectified as nouns, and on analogy with other nouns we create a formless "time" that is put into forms such as an hour or a year ("a year of time"). 61

The tense system is another aspect of objectifying time, putting it as units in a row (past-present-future), Duration, intensity, and tendency are expressed by metaphors of spatial extension, leading to the concept of imaginary space. The resultant world view is then pictured (Whorf 1971:152-156). The form-substance dichotomy is found in much of philosophy and Newtonian Physics. Objectified time leads to an excessive emphasis on time--diaries, interest in history, clocks, expressions such as "saving time" are all indicative of this.

The notion of formless time divided into equal units leads to time­ prorata values used for paying wages, rent, and interest, and is instrumental in helping to "routinize us." Imaginary space leads to the use of gestures to clarify speech, and the importance of space is related to an emphasis on muscle in art and sports. Finally, the confusion of what is and is not spatial leads to synesthesia in art.

This analysis of SAE has not been immune from attack either.

Greenberg (In Hoijer et al. 1954:131) states that the so-called three tenses in English are actually a grammatical doctrine, rather than simple units, and wonders whether English really does contribute to the idea of time as consisting of units on a line. Likewise, Carroll

(1963:10) questions whether the tens~ system really refers to time.

Carroll (1963:11) also remarks that SAE did manage to develop modern physics after all, despite the supposed antagonistic implications of the language.

Roger Brown (1958:259-260) suggests that perhaps Whorf was incorrect in lumping together all European languages and cultures.

This same idea occurred to Garvin who, through an analysis of Czech 62 as a representative of Slavic languages, sought to establish the differences between Slavic and Western Germanic (1949). First

Garvin presents some differences between the Slavic and Germanic languages, seeking only to prove their dissimilarity in certain respects, and not offering any cultural parallels (1949:66-74). He then contrasts the processes of word formation in the different lan­ guages (1949:74-83). English uses many loan words and forms words through compounds. Czech, on the other hand, creates new vocabulary through its own resources for the most part. As a result of this, vocabulary in Czech, even in unfamiliar areas, is created through eyeryday devices, and strange words are easier to learn and under­ stand than in English. Thus, education in Czech is easier and quicker than in English, literature, art, and philosophy are more easily popularized, and there is little prestige among Czech-speakers accorded to knowledge of vocabulary, since there is no reason for confusing this with knowledge of the vocabulary's subject. Garvin concludes (1949:84-85) by saying that the Czech language, although showing the objectification and rationalization of Western SAE, also shows the typically Slavic emotional nature. It constitutes a bridge between the West and Slavic.

Many scholars have speculated on the influence of the on cultures whose people speak it. Rapoport (1958:96) maintains that the English transitive verb causes people to concentrate on the action of the subject and ignore the effect on the subject, with the result that foreign policy concerns what happens to the weak, and what the strong do, but ignores what happens to the strong, as i~ 63

indicated in such sentences as "X conquers Y." Henry Lee Smith, Jr.

and Charles A. Ferguson, in a syllabus for the Foreign Service In­

stitute, suggest that the actor-action-goal sentence structure of

English may be linked with the excessive tendency to use causal

analysis, and they also propos.~. that a type of thinking in terms of

dichotomies with no middle ground may be linked to the language

(Leary 1952:201). Thompson (1946:53-54) argues that the subject­

predicate sentence structure implies that actors and things are

separate from actions and attributes, an idea which led to the notion

that matter is inert and acted upon externally. Our language,

according to Thompson, led science astray from reality, and the use

of mathematics and logic has been needed to correct this. Chang

(1952:211-213) links the presence of the copula, the requirement of

a subject for every sentence, and the distinction between the subject

and the verb, with the law of identity and the idea of substance in

Western logic and philosophy. An interesting idea is presented by

Carroll (1958:36) who says that language may influence psychopathology

and that, for example, a paranoid who speaks a language in which every

sentence must have a subject, as in English, must find the source of

all the threats he feels. Lotz (1964) suggests that athletic goals,

prices, the important wedding anniversaries (25, 50), and so forth,

are influenced by the basic "round numbers" of English, as well as

other European languages. Emeneau (1950:202) notes that number morphemes in English parallel the decimal system, with the exception

of "eleven" and "twelve," which are not compounds expressing the

number of groups of ten and the number of units. 64

De Madariaga (1969:176-182) compares the English "man of action" with his language, and finds the following correlations.

Words are monosyllabic, that is, utilitarian. There is a good verb vocabulary. An empiricism is seen in that word shape is fairly con­ stant and words are interchangeable. Words are easily turned into verbs, indicating that thought and action proceed together. Tenses generally use the present form of the verb with an auxiliary, and such thinking in the present is tied with empiricism. Vowels are fluid, as would be expected to be associated with empirical quick­ acting minds. The importance of intonation correlates with the importance of social organization in England. The tone of the sentence is more important than that of the word, just as there is a tendency for individuals to form harmonious groups. Social conven­ tion, rather than rules, keeps the language uniform, again emphasizing the importance of social organization. Thus, the English language is fluid, changing with social whim, words are formed easily as is needed in a man of action, and words are as free and individual as are the

Englishmen, though they do combine harmoniously (as do the English­ men).

On the other hand, de Madariaga's Frenchman is a man of thought (1969:181-187). French has taken Latin words and chopped them to their essentials, in a process of abstraction. Pronunciation is accurate and ruled, vowels are precise, vowels and consonants are balanced in words, and vowels lose their color, tending to become

"e"; this geometrical, abstract, and clear nature of phonology corre­ lates with a geometrical, abstract, and clear mentality. The grammar 65 is rigid, order fixed, and each constituent has a definite shape and function; that is, style, too, is abstract and precise, and therefore corresponds to thought. Logic is important in the language, so the future has its own form and is not expressed with the present form of the verb, as in English. Stress is even, in line with the French emphasis on the equality of all differences. And as already men­ tioned, vowels tend toward "e," the middle vowel, indicating the

French sense of moderation.

This may be contrasted with de Madariaga's Spanish "man of passion" (1969:187-192). Pronunciation in Spanish is more forceful than in the other languages, words are full and colorful, and pro­ nunciation is straightforward. Words may be full, but there is no superfluosity, as in Italian. Diphthongs and triphthongs are used to build up energy. Energy is thrown off in pronunciation, rather than kept in the word, as in English. Word accent is stronger than sentence accent, indicating the importance of the individual. renultimate word stress often results in stressing the middle syllable of the word, just as the Spanish man is self-centered. Finally, the rules of the language are not rigid; there is spontaneity and freedom.

As a final example of correlations proposed for Europe,

Meillet suggested that Latin-speaking aristocrats were given the opportunity to use words that were as autonomous as they wished to be, for nominals were not fixed in expression (Cohen 1956:40-41).

If language is correlated with culture, a change in one ought to be correlated with a change in the other. Hoijer (1964b:

458) considers some evidence for this--Modern English arose during 66

England's industrialization, Navaho is divergent in both language and culture from other Western Apache groups. One of the most obvious linguistic changes during culture change is alterations in vocabulary to accommodate innovations, and Hoijer (1964b:459-461) explores the possibility that such semantic change could lead to alterations in language structure. Lee (1943:438-440) asserts that contact with the whites has resulted in a new Wintu emphasis on kinesthetics in both behavior and language. Muller (1945) presents a case involving

France. From the fifth through the ninth centuries, the introduction of Christianity gave the individual a new dignity--he became a free agent. Correspondingly, the use of the demonstrative in the language increased and the use of declensions decreased. Words were thereby detached from their functional markers and instead linked to the speaker; they lost their syntactic independence, gaining instead a semantic independence. Christianity, according to Muller, freed man from the terror of fate, making him immortal and outside of nature; likewise the use of demonstratives gave him "possession" of his substantive referents, allowing him to show his new dominance.

Cultural parallels to this linguistic change were that small family farms gained in importance, and the small people eclipsed the prestige of large land-owners, as the individual was given a new importance.

The purpose of this rather lengthy summary of the literature seeking to establish language-culture correlations is not so much to show the reader isolated language and culture traits, but rather to provide the basis for an analysis of the nature of the work that has been attempted thus far. Some of the examples, such as de Madariaga's 67 self-centered Spanishmen, and their words with stress on the middle syllable, may seem amusing and far-fetched, but what is of greater importance is the methodology--or lack of it--upon which the above correlations have been based.

The chief objection that has been raised against the above analyses is that the authors have, for the most part, followed a circular procedure in which cultural data and linguistic data are not arrived at independently, and this objection has been raised by many scholars (see, e.g., Bedau 1957:291-292; Black 1962:252-253;

Kroeber in Hoijer et al. 1954:236; Doob 1952:105; Lenneberg, quoted by Carroll 1971:28; Lounsbury, quoted in Voegelin and Sebeok 1953:

48; Singer in Hoijer et al. 1954:233). The problem of bias that may be introduced in such a procedure is an important one. For example, as mentioned above, Carroll argues that there are at least two possible semantic interpretations of Navaho verb stems, and it is certainly a possibility that the one that occurred to Hoijer was in­ fluenced by his prior knowledge of the Navaho culture. Now even if experimental evidence proved that the Navaho agree with Hoijer's interpretation, the correlation posited with their culture would still be insignificant. This is because Hoijer's methodology allowed him to choose any linguistic or cultural trait that might be useful in proving his hypothesis that language and culture mirror each other.

Thus, if it turned out that the Navaho do not share his interpreta­ tion of their verb stems, Hoijer could co~ce~vably turn to any other linguistic trait that might prove his point.

How, then, might a more strict methodological paradigm pro- 68 vide some control and reduce the effects of the investigator's bias?

As this author sees it, there are two possible paradigms that might

be followed in such analysis. The first, the closest that ethnog­

raphy can come to a controlled experiment, is that of testing

predictions from language to culture, or from culture to language.

The second model is that of using statistics to establish signifi­ correlations from a sample of languages and cultures.

Let us examine first the paradigm of testing predictions.

Both Doob (1952:105) and Redfield (in Hoijer et al. 1954:222-224) maintain that the true test of this third level of linguistic rela­ tivity would be to see if knowledge of a language alone allows correct prediction of behavior and values. Certainly, if this is not the only test, it is at least a very good one, and yet, of all the literature mentioned above only Mathiot followed this method of prediction. And it is significant that even she was not predicting blindly, having had an unspecified amount of contact with the Papago

Indians prior to her analysis. This leads us to the question of whether, indeed, it is possible to predict culture from language with greater than chance success, even if correlations do exist.

Not only did Mathiot have some knowledge of the Papago, but it may have been a happy coincidence that led her to inspect a portion of the language that participated in such a correlation. Thus, Doob

(1952:109), Greenberg (1954:8-9), and Kaplan (in Hoijer et al. 1954:

236-237) would question the feasibility of a predictive method, due to the complicated nature of the problem and the uniqueness of each case. To quote Henle (1958:23-24), "every culture may, perhaps, be 69 correlated with some aspect of the language accompanying it, but there

is not yet enough evidence to suggest what this aspect may be without actual examination of the case"; in his view, we must study both language and culture together to find parallels until a further de­ velopment of linguistics and anthropology allows investigators to make predictions.

In terms of this predictive method, the investigator should ideally perform an analysis of a language whose culture he knows virtually nothing about and, based on the semantics of this language, should make some predictions about the culture. Meanwhile, an anthro­ pological colleague should conduct an ethnographic study of the people in question, perform a cultural analysis, and finally determine if the prediction is confirmed by the facts. Performing the linguis­ tic and cultural analyses independently appears to be necessary so long as there remain some subjective elements in them. Goodenough

(1955:243-244) reasonably maintains that anyone working in this field ought to have both linguistic and ethnographic experience, and certainly this would have to be the case for the individual making cultural predictions from linguistic facts. Such an individual, however, must not have too broad an anthropological background, lest he find that he knows too much world to make blind pre­ dictions.

Now it is true that in some of the research cited earlier the crucial linguistic and/or cultural traits could be determined objectively. For example, in Muller's case, increased use of demon­ stratives and decreased use of declensions are quantitative phenomena, 70

although the problem remains that their semantic interpretation is more

subjective. Or consider Win's article, in which the major cultural

trait he is concerned with is the objective presence of Buddhism.

Note, however, that both the linguistic and cultural traits must be

objectively determined in order for it to be methodologically sound

for the same investigator to study the language, make the predictions,

and study the culture (or, conceivably, to proceed in the reverse

direction); only in such a case would investigator bias be assured of not coming into play. Such an experiment--involving only one

investigator and easily determined traits--would surely be a rela-

tively simple one and therefore might seem desirable. However, due

to the complexities of both language and culture, where different

traits may be clashing with each other, for example, the possibilities

for a successful experiment without conducting a thorough analysis of both language and culture are minimal. What, then, are some of

the procedures that might help in such an analysis?

Although linguistics can certainly be scientifically objective in certain areas, the linguistic analysis required in an investigation of linguistic relativity is, at present, no less subjective or diffi- cult than the cultural analysis. The major reason for this is that in the structuralist tradition meaning was excluded from linguistics, and as a result semantics has been studied very little. Many scholars blame the lack of a developed science of semantics for the lack of results regarding the linguistic relativity hypothesis (see, e.g., Garvin 1956:569; Hymes 196la:337; Ullmann 1959:318-320). Thus, from the very start, we run into the question of the feasibility of 71

the predictive method, at least until the linguist is armed with

techniques needed for the study of meaning.

A few clues for such a study might be mentioned here.

McQuown (1954:27-28) suggests a method of using culture patterns to

help show the ordering, or distribution, and thereby the meaning,

of linguistic forms, a method Weinreich (1955:427-428) terms "tedious

and unproductive." Newman (1954:89-90) points out that study of a

lexeme's contexts allows the determination of its range of meaning.

Roger Brown (1956:311) cautions, however, that a term may not be

expanded into its total cultural place lest any correlations be

tautologous. Problems may arise due to inconsistencies in lexical

divisions (Hockett and Kroeber in Hoijer et al. 1954:140-141).

Goodenough (1955:241) suggests that a study of the processes of

learning meaning might help point to a method for determining

meanings.

Obviously meaning can be an elusive topic of study, and for

this reason Hoijer (1954:95-96) proposes that, rather than attempting

to determine precise meanings, the investigator should concentrate more on how the semantic system organizes experience, how it is

structured. As Weinrich (1955:429) explains this, Hoijer's question

is what is the difference between meanings, instead of what are the meanings.

Another way around this problem of meaning can be to use a universal standard of reference, as Whorf (1971:160-172) attempted with Gestalt technique. Using the universal laws of visual percep­

tion, he determined the order in which different perceptual elements 72 are combined in Shawnee compound verbs and nouns. Hockett (in Hoijer et al. 1954:171), however, questions the need for such a frame of reference.

A further problem is that, following a ,

the linguist must weight different features as to their relative importance. He must know how to determine which may profitably be ignored, which of two contradictory features is the stronger, and so forth, if he is to put together some sort of world view from the linguistic data. As Newman (in Hoijer et al. 1954:199) says, he cannot simply pick out a central meaning from a language, but must account for varied relationships in the grammar. Several scholars have speculated on rules that might be followed to do this. Features that are purely grammatical for their speakers may be ignored (Hoijer

1954:97-98). Newman (1954:82-86) says that obligatory forms have greater weight than optional ones, forms that have extensive ranges are more important than those of limited scope, and categories fre­ quently used are more important than those rarely used. Greenberg

(1954:1D-ll) agrees that compulsory forms are likely to be important;

Weinreich (1955:429), on the other hand, objects to Newman's first rule, since optional forms carry the most information, and therefore have a special importance. Like Newman, Hoijer (1954:97-98) says that the most significant speech patterns are those used by all adults frequently and with ease, and Hymes (196la:328-329) is also in agreement with this statement. While Voegelin and Sebeok (1953:

24) suggest that morphemic absence is not a particularly fruitful area in which to look for evidence of linguistic relativity, Basson 73 and O'Connor (1947:58) feel that absence of abstract terms may be more important than structure, at least in problems of translation. Roger

Brown (1958:253-255) explains that items such as inflexion force observation of certain environmental features, and Henle (1958:8-10) adds that inflexion has a pervasive influence, more so than that of vocabulary. Finally, Wells (1962:698) suggests that grammar is more significant than vocabulary since it is less changeable. Perhaps the most significant conclusion that may be drawn from these sugges- tions is that the problem is a difficult one at best.

Essentially, this problem is one of putting together what Wharf called a "fashion of speaking." Not everyone agrees as to what

Wharf meant by this term (see, e.g., McQuown in Hoijer et al. 1954:

149), but along with Hoijer, (1953:460-461) this author would say that Wharf was referring to a pattern of language put together through study of the diverse systems in a language, from the lexical to the syntactic. More significant effects are to be expected where more than one system leads toward a certain concept (Greenberg 1954:

10-11; Hoijer 1954:98).

Finally, we may turn to the cultural analysis demanded by the predictive paradigm. Trager (1959) believes that a satisfactory structural analysis may be the key to the problem of language-culture correlations. Some (e.g. Kluckhohn 1956:572-573; Hockett 1948:

566-567) suggest extending linguistic methods into cultural areas, feeling that these two spheres are similar enough to warrant such an approach. Some more specific suggestions may be found in Goodenough

(1964), McQuown (1954), and Voegelin, Yegerlehner, and Robinett

(1954). Levi-Strauss/ (1953:6) states that the problem with Wharf's 74 work is that he sought to correlate his linguistic analysis with too crude an analysis of culture--he failed to attempt to correlate things that were on the same level.

One possible type of cultural analysis that may be used is to arrive at a configurational description of a culture. ior example,

Opler (1946) shows how a culture may be characterized by a few major themes. Thompson (1946:65) suggests that the basic integration of a community may be more easily determined if a number of communities with a common heritage be compared.

Finally, a few last problems with the predictive method might be mentioned. As brought up earlier in this paper, even if linguistic relativity is true, languages may at times lag behind cultures; if such a case were chosen to test a prediction the results might be negative, although if the language had been given time to change, the prediction formed from it might have been supported. And conversely, if a culture lags behind a language, a prediction would not hold true until the culture is given time to catch up. Also, linguistic analysis is further complicated by the fact that usage of the language, that is, the ethnolinguistic situation, may vary within a culture, and also, if different codes are used in different situa­ tions, they m~st all be analyzed and somehow integrated.

In view of these difficulties involved in using a method of prediction, perhaps a more reasonable approach to take would be to find evidence in the form of statistical correlations. The study that comes closest to this method is that of Altshuler. Although he does not perform a test of significance, he provides statistics 75 on social organization that show that the importance of the female principle is relatively high in American Indian cultures, as he would predic~-from the prevalence of verb-centered languages among the

Indians. As he realizes, however, this is only a "gross check" of his hypothesis, since he does not discriminate within the Western

Hemisphere to see which specific societies have verb-centered languages. To remedy this, he provides evidence from some Indian and African cultures that supports his hypothesis, but he does not m~~tion if he came across any contrary evidence during his investiga­ tion. Further, in some cases the linguistic evidence used is minimal, and he is subjective in some of his linguistic interpretations.

Altshuler is certainly to be commended for his use of multiple cases to prove a point, but he has not gone quite far enough, neglecting a rigorous use of statistics and injecting too much of his own bias into the study.

Thus, there is no good use of statistics in the examples in the literature, and there is only Mathiot's one dubious study in which the method of prediction was followed. The conclusion to be drawn from this is that all of the work that has been done concerning specific language-culture correlations is nothing more than anecdotal; as a result of careless methodology anthropology has contributed neither favorable nor unfavorable evidence to the linguistic rela­ tivity hypothesis. It is another question, however, whether or not it has contributed some likely hypotheses that might be tested in the future. Before considering this question, let us further examine this second, statistical paradigm.

This paradigm has the disadvantage that it requires results 76

from a large sample of languages and cultures in order to determine

statistical significance of any posited correlations. One must

consider its feasibility, for if linguistic and cultural analyses

are complex and time-consuming even for one case, the problems might only be multiplied by the number of societies in the sample, making the cost in time and money of such a study prohibitive. It

seems to this author that the only way in which this method can

possibly be used to advantage is if it be used to test correlations between relatively discrete traits. As stated before, discrete

traits are likely to be superseded by other traits or blended with other features, and it might seem unlikely that significant correla­

tions could ever be found, if this be the case. However, if care be taken to choose a few frequent or pervasive traits, the possibilities for a correlation might be very favorable indeed. Complete analysis might be needed if the method of prediction is to be followed, but if the aim is merely a statistical correlation, discrete traits ought to be satisfactory, if they are chosen wisely. Not only, then, is the statistical approach at least as feasible as the method of pre­ diction, but it has the advantage that it is not forced to draw a conclusion from any one case, and this adds a certain compelling nature to any of its conclusions. Indeed, given the complexities of analysis described above, it might be that the linguistic relativity problem is too laden with difficulties and sources of error for us to expect anything more than significant correlations in anthropological studies, at least until our techniques of analysis are more developed.

According to this paradigm, randomly selected languages with 77 a common trait or traits would be chosen to determine in how many cases an expected trait or traits were found in the associated culture. Or, conversely, cultures with certain selected features could be tested to see if their associated languages have the expected similarity. Per­ haps the most profitable starting point for selecting the traits to be correlated might be an intensive study of one or two languages and cultures in which the investigator sensed a correlation intuitively through simultaneous study of both language and culture. If the traits thereby selected are objective, one investigator might conduct the entire statistical test. If, however, an element of subjectivity would be involved in ascertaining whether the elements are present in .the test cases, it would be mandatory that people unaware of the hypothesis independently judge languages and cultures, in order to prevent personal bias from entering into the results.

The advantage of this approach, then, seems to be that no complete analyses of language and culture are necessary, except perhaps for the first case from which the hypothesis to be tested is developed. Once the features to be tested are selected, by whatever process, determination of their presence in test cases would require analysis of only certain selected areas. The limitation of the approach is that certain traits may be too unique to find a test sample. Although some of the correlations posited by anthropologists, philosophers, and linguists are far-fetched, and others, such as those by Hoijer and Lee cannot be tested since they involve universals for a certain level of development, some present reasonable hypotheses 78 that might be tested by one of the two methods described above. In presenting a few examples, there will be an opportunity to point out that the nature of a specific problem may lend itself more easily to one approach than to the other.

Consider, for example, Hoijer's notion that the dual number in Apache and Tewa is related to the prevalence of paired deities.

Here we have a relatively straightforward suggestion, requiring no extensive analyses, thereby lending itself very nicely to the statis- tical approach. Now, it might be advisable to expand the cultural trait to be a more variable emphasis on two's, such as the use of important paired ritual implements, of rituals in which individuals perform in pairs, and so forth. The danger here is that this ex~ansion provides a leak in the system that might allow some unwanted subjectivity to creep in. It becomes open to question what kinds of pairs are to be looked for, and how frequently in the culture they have to appear. Perhaps the most satisfactory approach to the problem, therefore, might be to determine a very large sample of those world languages having the dual number, and to then inquire of all ethnog- ra~hers of the cultures in question if, in their opinions, their cultures contain any conspicuous emphasis on paired items. If the ethnographers are not aware of the hypothesis to be tested, their subjective answers would be most acceptable, given the large sample.

Confirmation of the hypothesis would require a control group of languages without dual number to see if their cultures lack the empha- sis on pairs. , Similarly, Levi-Strauss' suggestion on the relationship 79

between the rules and elements of kinship systems and those of lan-

guages could be tested through use of a survey asking linguists and

ethnographers (again, preferably ignorant of the hypothesis) in which , of Levi-Strauss' categories, if any, they consider the languages

and societies they know to belong. A large scale survey is feasible

in these cases, since there are doubtless numerous scholars in the world who already have the knowledge to answer the questigns_ to be asked. A hypothesis asking more unusual questions might require

special investigation, in which case time and money would probably limit the size of the sample.

Similarly, the Human Relations Area Files would provide Altshuler with the information to select a large number of cultures unequivocably centering on the male or the female. A questionnaire could then be sent to the relevant linguists, asking them to charac- terize their languages in terms of the relative importance of nouns and verbs. This would improve on his study, making it more objective, and using statistics to their best advantage.

Capell's case for a relation between noun classes used for alliterative sentence concord and the notion of a primal substance is also amenable to a statistical treatment. As a matter of fact,

Capell himself mentions a few cultures that might be included in the sample.

Correlations involving philosophy, logic, and science, on the other hand, suffer from the fact that the potential sample is limited to those cultures which have reached a relatively high level of devel- opment. The most fruitful approach in such cases would probably be

-- 80 to carry out an extensive linguistic analysis and then make predictions to be tested. In this manner, a maximum amount of information may be obtained from each of the societies available for testing correlations in these areas. For example, although Chang's discussion of Chinese and English in relation to philosophy, science, and religion is most suggestive, a statistical approach might be futile. Of the cultures that have evolved to an appropriate level of civilization to be in­ cluded in a sample, how many languages with the Chinese features will be found, and of these how many of their cultures can be shown to have been relatively uninfluenced by the Chinese culture? This is an important point. A statistical correlation, in terms of today's methods, is valid only if it involves cultures that have not had an extensive interchange of ideas. Because of this, the sample of truly significant cases in any test is reduced, and when it comes to a question of philosophy and an isolating, ideographic language, the sample may become close to nonexistent. This is why questions of science and philosophy are best handled through the method of predic­ tion.

Note, also, that the very nature of Win's Burmese example makes a statistical test an impossibility. Any Buddhist cultures by necessity have been in contact significant enough for the spread of this religion. Any linguistic similarities could conceivably be due to the same contact. On the other hand, if, for example, Indian

Buddhist te~ts failed to show the Burmese linguistic traits, this negative example does say something about the posited correlation, though it constitutes no disproof in terms of the model being used. 81

Thus, while a statistical test cannot prove or disprove a correlation when the sample consists of cultures that have had a great deal of

contact, the absence of a correlation does at least cast some doubt

on it, for, if the traits really are especially compatible, accepting

one would naturally pave the way for accepting the other, unless

something in the recipient's language or culture was exerting a contrary pressure.

Lee's work on the Trobriand Islanders could, conceivably, be approached using either methodology. She cites a number of language traits in her analysis, and they appear to be suitable to a question­ naire format, but given the comprehensive nature of her study, it seems to warrant a more comprehensive treatment of test cases. For example, one of the features she uses is that of changes in noun stems corresponding to changes in a referent's qualities. Graves rightly points out that this trait is even found in certain cases in English, so obviously care must be taken that the language to be used in a test case has this trait occurring with high frequency. For this

Trobriand example, a very careful analysis of noun stems, tense, lineality, causation, chronology, and purpose in the language must be conducted. Granted that so many features are involved to begin with, it might be just as well to analyze thoroughly a test language to get as much information as pessible from it. On the cultural side, again, questions as to purposive behavior and cultural value could be answered through questionnaires, special ethnographic studies if necessary, or through complete cultural analyses. Lee's example could, of course, be divided into a number of more discrete correla- 82 tions, and if this were done, a statistical approach would be favored.

However, her example shows such a nice integration that there would be value in testing it as a whole, in which case the predictive approach would be satisfactory as well, if not favored.

Mathiot's article presents a very interesting approach, that of predicting cultural features from the criteria used in classifica­ tion. Although she did receive favorable results through the analysis of one selected system in the language, a safer course might be an analysis of a number of systems to determine a predominant mode of classification in all of them, if it exists. Construction of folk taxonomies is a time-consuming task, and a statistical approach would require too much time. For this reason a more appropriate methodology would be to make predictions from specially constructed taxonomies, and since the potential for predictive ability seems to be high (a child learning the taxonomies presumably is thereby taught to make certain distinctions, and these would be most likely incorporated into other cultural behavior), the approach should work well.

Perhaps enough has been said now for some conclusions to be drawn. First of all, many of the correlations between language and culture that have been posited thus far, although anecdotal, are definitely suitable for test in the future. Which seem reasonable and which seem too far-fetched to bother with further may in many cases be a matter of personal judgment. These, and future hypotheses, arrived at by whatever means, may be tested in either of two ways.

Predicting a cultural feature from a language, or vice versa, and then testing this prediction is a desirable procedure when the pas- 83

sible test sample is very small or when the specific question· being

asked requires rather extensive analysis that has not yet been done.

Using statistics in search of significant correlations is an advisable method when the potential sample is large, when the cultures in­ volved have not been in extensive contact, and when the information needed either can be obtained from published data or through the use of questionnaires or would require special investigations of a limited nature. Even when an element of subjectivity is involved in the analyses required of investigators, the results of either method can be significant if information sources are kept independent of each other and if, in the case of the statistical technique, test cases are selected randomly. Also, in the predictive method, to allow blind predictions, this task should be accomplished by people with limited ethnographic knowledge; ironically, the less experienced may be more valuable for this step of prediction in the test.

As more predictive tests are performed, some of them may be similar enough so that they might be integrated into an ex post facto statistical paradigm. And once a statistical test has been performed, it can be added to bit by bit as relevant information is obtained through further investigations. Typologies constructed might be useful in facilitating comparisons between languages and cultures, such that diffe~ent scholars may integrate their work (Hockett 1954:

108; Hymes 196lb:37-38). Work will also be cumulative in the sense

that any successful studies that are performed will instruct future investigators as to fruitful areas of language and culture in which to search for correlations. As of the moment, so little is known

that the best places to investigate remain to be discovered. v

FURTHER RESEARCH POSSIBILITIES

Although Carnes (1970:63-65) and Black (1962:250) assert that it is not possible to test "the Wharf Hypothesis," the above sections show that such tests are possible, even on the difficult level of grammatical structure. The task, however, is no easy one.

We are deficient in the necessary typology of language and other culture traits (Hockett 1954:108), in our knowledge of semantics, and in our ability to analyze culture. These are tools that must be developed if rigorous, significant tests are to be performed.

Nevertheless, there is much that can be done in our present state of knowledge, as is indicated by the experimental results summarized earlier and the suggestions for study of language-culture correla­ tions in the preceding section. A discussion of a few other potentially fruitful avenues of approach is in order.

It is helpful to have a framework for such investigation.

Fishman's four levels are useful, although they need some modifica­ tion since they ignore the morphemic level of linguistic analysis, dealing only with lexation and structure, mirroring a gap in the experiments that have been performed thus far. In the discussion in this paper the morphemic level was merged with that of gramma­ tical structure, but it probably deserves special consideration of

84 85

its own lest the focus on words and structure cause its potential

significance to be ignored. Walker, Jenkins, and Sebeok (1965)

present another useful framework; according to this approach four

hypotheses are in question--the influence of cognition on behavior

and of behavior on cognition, and the influence of codification on

cognition and of cognition on codification. These two systems

provide different perspectives for looking at the problem and framing

working hypotheses.

A problem that is encountered in cross-cultural experiments

is how to transcend cultural differences in order to ascertain those behavioral differences due solely to language (Carroll 1966:46;

Carroll and Casagrande 1958:22; Fearing 1954:54). Heidbreder (1958:

105-107) cautions that early motor activities may influence percep­

tion and cognition, and this means that inter-cultural differences

in these processes need not be due to language, or even differences

in cultural interests. One way out of this difficulty is to design

an intra-cultural experiment (Lenneberg 1953:468, 471; Walker, Jenkins,

and Sebeok 1965:195), and, indeed, this has been the most frequent

course taken. Lenneberg (1953:470) goes so far as to say that cross­ cultural comparison may add weight to a hypothesis, but is not needed

to validate it. Some questions, however, may require a cross­ cultural approach, such as a hypothesis involving auditory discrimina­ tion in speakers of different languages (Lenneberg and Roberts 1961:

496).

In a cross-cultural experiment the controlled variable must be carefully chosen. It must be universal in order to test how 86

different people deal with it, it must be variable, and it should be

as simple to measure as possible (Lenneberg and Roberts 1961:497-499).

These criteria are satisfied by what Lenneberg and Roberts (1961:499)

call the "language of experience," words or morphemes referring to

elementary forms of sensation. In view of these considerations,

Lenneberg's frequent use of colors in his experiments is most under­

standable.

Von Bertalanffy (1955:246-247) rightly explains that the

Wharf Hypothesis is part of a larger question--the importance of both

culture and biology on the categories of thought. As he says (1955:

253-254), language cannot alter an individual's experience potential,

but it can change which features are emphasized or underplayed. In his scheme, biology, including heredity as well as temporary physi­

ology and developmental stages, forms the substratum which language and other aspects of culture may influence. Similarly, Casagrande

(1966:293) suggests that there are two approaches to language's influence on thought; the first is its generic function, while the second involves comparative differences. This so-called "generic function," or the biology underlying cultural differences, is most relevant to investigation of the linguistic relativity hypothesis, for as language and psychological universals are uncovered, the areas in which to look for differences come into focus.

Wharf (1971:239) realized that there is "mind" behind language and, as Fearing (1954:72) says, we do need to discover the in;errelations between linguistic and nonlinguistic cognition.

Fearing (1954:61-75) presents a framework for inspecting this; 87 although the infant avails himself solely of perception, as he matures

he begins to conceptualize, putting symbolization between himself and

the environment. It is along this perception-conceptualization

spectrum that the interaction of nonlinguistic and linguistic cogni­

tion can be studied, and then applied to problems of linguistic relativity. Goodenough (1955:242) says of this paper by Fearing that its value lies in putting Whorf in perspective, providing a psycho­ logical orientation for problems of linguistic relativity.

A developmental approach to linguistic relativity is also advocated by Werner and Kaplan (1956). They suggest that experience is organized at several levels, language being only one of these, being itself divided into levels ranging from close to the preverbal to the completely verbal. In terms of such an approach, some lan­ guages and modes of thought are "primitive"; that is, they have not developed to as high a level as others. Primitive thought is not limited to certain tribes with supposedly poorly developed languages, and the authors propose a method in which primitive behavior may be elicited even from individuals whose language hides primitive, un­ differentiated thought.

Lenneberg (1962:105-106) cites evidence supporting the state­ ment that concepts arise prior to language, and Furth (1961) presents an example of how deaf children may be used in experiments to help determine the nature of preverbal concepts. There is a problem, how­ ever, in applying such experiments to linguistic relativity, since nonverbal cultural factors may interfere in them, such that the results are not indicative of universal human tendencies. 88

Aside from this developmental approach, another help for re­ search will be the determination of psychological and linguistic universals, an area of special current interest (!Marshall 1965:

78-79). A good deal of work has been done on Osgood's "semantic differential," in which subjects rate concepts on bipolar scales, these ratings then being subjected to factor analysis in order to determine the structuring of semantic space. Work by Osgood (1963:

28-34), Suci (1960:26-29), and others (Ervin 1961) with subjects speaking very different languages have shown that the factors that emerge tend to be highly similar. English-speaking subjects tend to have three main factors--Evaluation, Potency, and Activity--and other subjects tend to have at least the first of these and often also

Potency or a combination of Potency and Activity as a second factor.

The most anomalous results so far have been obtained for Navaho subjects (Suci 1960:26-29). As Osgood (1963:34-36) notes, the struc­ ture of semantic factors varies more across concepts than across people. Despite this universality, however, there may be differences in the system, such as in usage of the scales (Osgood, quoted in

Kluckhohn 1961:905). Osgood (1966:317, 319-322) suggests that possibly the universals are in the structure and factors of the af­ fective and discriminatory systems since these are based on the human nervous system, while linguistic relativity may rule outside these systems, where specific associations depend more on exp~rience.

Finally, Osgood (1963:34) mentions that his work is also relevant to linguistic relativity in that the universal semantic structure uncovered may be used as a cross-cultural standard, allowing objective 89 measuring of a culture's concepts.

Osgood (1966:313-314) has also discovered a universal agreement in synesthesia between vision and words; subjects looking at a pair of drawings tended to agree in their judgments of which one was best described by certain adjectives. Roger Brown (1958:146-149) conducted an experiment in which students asked to describe operatic voices with words they generally did not apply in such a manner, tended to agree with each other in their descriptions. Many experiments have been conducted investigating the phenomenon of universals in phonetic symbolism in which subjects guess the meanings of foreign words.

Positive results have been obtained for sense antonyms in Japanese,

Chinese, Hungarian, Polish, Czech, and Hindi (Brown, Black, and

Horowitz 1955:388-393), and for nonsense words (Miron 1961). Further experiments investigating semantic, synesthetic, and phonetic universals will do much to help delimit the possible limits of ~ linguistic relativity.

This discussion of linguistic and psychological universals leads back to the consideration that must always be kept in mind that in linguistic relativity we are dealing with ease of expression and tendencies for behavior, rather than with definite potential for thought (see, e.g., Fishman 1960:335; H~kett 1954:122•123; Hoijer

1953:560; Lenneberg 1953:467; von Grunebaum in Hoijer et al. 1954:

228-229). · This means, of course, that we need a statistical model, allowing for such contingencies as individual differences (Roberts in

Hoijer et al. 1954:239-240; Weinreich 1955:430).

The admitted importance of universals also leads to the 90 question of how important the differences might be. This question has generated a good deal of discussion, and there are those (see, e.g., Carroll 1964:111; Landar 1965:26; Lenneberg 1962:107•108;

Luh 1948:119; Osgood 1966:322; Penn 1972:39) who appear to consider the significance of the differences to be minimal indeed. Their negative attitude can be attributed not only to the established fact of linguistic universals, but also to the limited linguistic influence that has been confirmed through experiments and to the fact that linguistic bias can be overcome and even imitated, as in

Casagrande's experiment. VI

CONCLUSION

Hymes (196lb:46) has said that the notion of "cognitive styles" in language has suffered from proponents who give it an excessive importance and who present work without using appropriate controls.

To this it might be added that it has suffered from the demanding nature of the proof required. As Carroll (1963:4) has remarked, work on the hypothesis requires an "exacting knowledge" of psychology and linguistics (to which this author would add anthropology).

What, then, has been confirmed thus far? The lexicon does seem to be related to such things as cultural interests, though study of this has only been anecdotal. Language structure may well have cultural correlations but, again, the data thus far is purely anec­ dotal, and the time may be right for attempting to establish such correlations using a rigorous methodology and whatever resources linguistics and anthropology can furnish. Controlled experiments have resulted in some more definite conclusions, though they have been limited. The presence of verbal concepts can influence problem­ solving, codability can influence recognition, communication, and learning new category patterns, labeling distorts delayed reproduction of a figure, if not recognition of it, meanings given the same name may elicit similar behavior, objects given different names are more

91 92 easily differentiated, and language structure may be one way of in­ fluencing specific behavior in certain directions.

Even in the realm of control experiments there remains much more that could be done with present methods. Something other than colors ought to be used in codability experiments. More work is needed to help show whether Prentice is correct in that labeling does not affect recognition memory. Much too little has been done testing the influence of language structure on behavior. Some interesting and feasible tests of the linguistic relativity hypothesis have been discussed in the literature (see, e.g., D. Brown 1960:345; Hoijer and

McQuown in Roijer et al. 1954:275; Kaplan, Hoijer, and Fearing in

Hoijer et al. 1954:371-373; Rapoport 1958:91-94; Rapoport and Horo­ witz 1960:350-352, 361; Voegelin and Sebeok 1953:27; Walker, Jenkins, and Se·beok 1965: 20D-202). There is also much that could be done with present resources even in making language-culture studies less impressionistic. All that is needed for further investigation of the linguistic relativity hypothesis is some imagination in framing tests of it.

Thus, the present state of the research concerning linguistic relativity does not indicate that the problem should be abandoned.

Even if the influence of language is limited, we now know that it exists, and further investigation is needed to specify just what this influence is--its nature and power--and how it interacts with bio­ logical psychological factors. Also, the difficulties attendant in such research (experimental design, proper analysis of language and culture, the complexity of the systems under investigation) present 93 another possible interpretation of the limited confirmation thus far received--that we have here a problem that "cannot be solved in a single lifetime" (Landar 1965:220), and that progress will be slow before we arrive at more significant results.

What is needed, then, is not so much proof or disproof of

"the Whorf Hypothesis," but a delimitation of what in language can tnfluence what behaviors, and what the nature of this influence is

(Fishman 1960:337). We might expect that the differences between languages that are most precisely defined are the most trivial in terms of the hypothesis and that those that are hard to define will turn out to be the most important (Hockett 1954:122). Gastil (1959:

24-25) explains that not only is linguistic relativity slight in concrete areas, but also in the natural sciences, since there has been virtually world-wide development of modern science. Also, language's influence is likely to be more important in such areas / as philosophy than in practical activities (Hockett 1954:123).

In conclusion, not only is there a need to continue investiga- tions of the sort already begun and to improve our tools for linguistic and cultural analysis, but there are a variety of new types of linguistic relativity experiments to be added to our reper- toire. Important results will be obtained, provided we do not succumb to the difficulties of the task. As Fishman (1960:336) says, the effect of linguistic relativity may appear to be innocuous enough from the results we have so far, but without studying the phenomenon we will find ourselves ignorant of something that we could and should learn to provide for. BIBLIOGRAPHY

Alexander, H. G. 1936 Linguistic Morphology in Relation to Thinking. The Journal of Philosophy 33:261-269. Altshuler, Nathan 1956 Linguistic Forms as of People. International Journal of American Linguistics 22:106-112.

Astrov, Margot 1950 The Concept of.Motion as the Psychological Leitmotif of Navaho Life and Literature. Journal of American Folklore 63:45-56. Bartlett, Frederic c. 1961 Remembering; A Study in Experimental and Social Psychology. Cambridge, University Press.

Basilus, Harold 1952 Neo-Humboldtian . Word 8:95-105.

Basson, A. H. and D. J. O'Connor 1947 Language and Philosophy: Some Suggestions for an Empirical Approach. Philosophy 22:49-65. Bedau, Hugo A. 1957 Review of Language, Thought, and Reality: Selected Writings of Benjamin Lee Wharf, ed. John B. Carroll. Philosophy of Science 24:289-293.

Bittle, William E. 1953 Language and Culture Areas: A Note on Method. Philosophy of Science 20:247-256.

Black, Max 1962 Studies in Linguistic Relativity: The Views of Benjamin Lee Wharf. In Models and Metaphors; Studies in Language and Philosophy, by . Ithaca, Press, pp. 244-257.

Bloomfield, Leonard 1933 Language. New York, Henry Holt and Company.

Boas, Franz 1904 The History of Anthropology. Science 20:513-524.

94 95

1930 Anthropology. In Encyclopedia of the Social Sciences, Edwin R. A. Seligman, ed. New York, Macmillan, pp. 73-110.

Brown, Dona Worrall 1960 Does Language Structure Influence Thought? Comments on the Psycho-Linguistics Experiment at Michigan. Etc., a Review of General Semantics 17:339-345. Brown, Roger w. 1956 Language and Categories. In A Study of Thinking, by Jerome s. Bruner,.Jacqueline J. Goodnow, and George A. Austin. New York, John Wiley & Sons, pp. 247-312. l958 Words and Things. New York, The Free Press. 1961 Linguistic and the Part of Speech. In Psycholinguistics; A Book of Readings, Sol Saporta, ed. New York, Holt, Rinehart and Winston, pp. 503-509.

l965 Social Psychology. New York, Free Press. Brown, Roger w., Abraham H. Black, and Arnold E. Horowitz l955 Phonetic Symbolism in Natural Languages. The Journal of Abnormal and Social Psychology 50:388-393.

Brown, Roger W. and Eric H. Lenneberg 1958 Studies in Linguistic Relativity. In Readings in Social Psychology, Eleanor E. Maccoby, Theodore M. Newcomb, Eugene L. Hartley, ed. New York, Holt, Rinehart and Winston, pp. 9-18.

1961 A Study in Language and Cognition. In Psycholinguistics; A Book of Readings, ed. Sol Saporta. New York, Holt, Rinehart and Winston, pp. 480-492. Bruner, Jerome S. 1964 The Course of Cognitive Growth. American Psychologist 19:1-15. Bruner, Jerome S., Robert D. Busiek, A. Leigh Minturn 1952 Assimilation in the Immediate Reproduction of Visually Perceived Figures. Journal of Experimental Psychology 44:151-155. Burnham, R. w. and J. R. Clark 1955 A Test of Hue Memory. Journal of Applied Psychology 39:164-172.

Capell, A. 1960 Language and World View in the Northern Kimberley, Western Australia. Southwestern Journal of Anthropology 16:1-14. 96

Carmichael, L., H. P. Hogan, and A. A. Walter 1932 An Experimental Study of the Effect of Language on the Reproduction of Visually Perceived Form. Journal of Experimental Psychology 15:73-86. Carnes, Ralph L. 1970 A.Perceptual Model of the Whorfian Thesis. Etc. A Review of General Semantics 27:55-65.

Carroll, John B. 1958 Some Psychological Effects of Language Structure. In Psychopathology of Communication, Paul H. Hoch and Joseph Zubin, ed. New York, Grune & Stratton, pp. 28-36.

1963 Linguistic Relativity, Contrastive Linguistics, and Language Learning. International Review of Applied Linguistics in Language Teaching 1:1-20.

1964 Language and Thought. Englewood Cliffs, New Jersey, ~rentice-Hall, Inc. 1966 The Study of Language. Cambridge, Press.

1971 Introduction. In Language, Thought, and Reality: Selected Writings of Benjamin Lee Whorf, John B. Carroll, ed. Cambridge, M.I.T. Press, pp. 1-34.

Carroll, John B. and Joseph B. Casagrande 1958 The Function of Language Classifications in Behavior. In Readings in Social Psychology, Eleanor E. Maccoby, Theodore M. Newcomb, Eugene L. Hartley, ed. New York, Holt, Rinehart and Winston, pp. 18-31.

Casagrande, Joseph B. 1955 Comanche Linguistic Acculturation: III. International Journal of American Linguistics 21:8-25. 1960 The Southwest Project in Comparative Psycholinguistics: A Preliminary Report. In Men and Cultures, A. F. C. Wallace, ed. Philadelphia, University of Philadelphia Press, pp. 777-782.

1966 Language Universals in Anthropological Perspective. In Universals of Language, Joseph H. Greenberg, ed. Cambridge, M.I.T. Press, pp. 279-298.

Cassirer, Ernst 1946 L'influence du langage sur le developpement de la pensee dans les sciences de la nature. Journal de psychologie normale et pathologique 39:129-152. 97

1953 Language and Myth. trans. Susanne K. Langer. New York, Dover Publications, Inc.

1965 An Essay on Man: An Introduction to a Philosophy of Human Culture. New Haven, Yale University Press.

Chang Tung-Sun 1952 A Chinese Philo~opher's Theory of Knowledge. trans. Li An-che. A Review of General Semantics 9:203-226.

Chase, Stuart 1938 The Tyranny of Words. New York, Harcourt, Brace and Company. 1954 How Language Shapes Our Thoughts. Harper's Magazine 208 (l247) :76-82. 1971 Foreword. In Language, Thought, and Reality: Selected Writings of Benjamin Lee Whorf, John B. Carroll, ed. Cambridge, M.I.T. Press, pp. v-x.

Cofer, Charles N. 1960 An Experimental Analysis of the Role of Context in Verbal Behavior. Transactions of the New York Academy of Sciences 22:341-347.

1963 Verbal Behavior in Relation to Reasoning and Values. In Groups, Leadership and Men, Harold Guetzkow, ed. New York, Russell & Russell, pp. 206-217.

Cohen, Marcel 1956 Social and Linguistic Structure. Diogenes 15:38-47.

De Laguna, Grace A. 1929 Perception and Language. Human Biology 1:555-558. de Madariaga, Salvador 1969 Englishmen, Frenchmen, Spaniards. New York, Hill and Wang. Doob, Leonard w. 1952 Social Psychology. New York, Henry Holt and Company.

Douglas, Mary 1966 Purity and Danger. New York, Praeger Publishers.

Downs, James F. 1972 The Navajo. New York, Holt, Rinehart and Winston, Inc.

Emeneau, M. B. 1950 Language and Non-linguistic Patterns. Language 26: 199-209. 98

Ervin, Susan 1961 Abstracts of Articles on Psycholinguistics. International Journal of American Linguistics 27: 259-263.

Fearing, Franklin 1954 An Examination of the Conceptions of Benjamin Whorf in the Light of Theories of Perception and Cognition. In Language in Culture; Conference on the Inter­ relations of Language and Other Aspects of Culture, Harry Hoijer, ed. Chicago, Press, pp. 47-81. Feuer, L. S. 1953 Sociological Aspects of the Relation Between Language and Philosophy. Philosophy of Science 20: 85-100.

Fishman, Joshua A. 1960 A Systematization of the Whorfian Hypothesis. Behavioral Science 5:323-339. French, David 1963 The Relationship of Anthropology to Studies in Perception and Cognition. In Psychology: A Study of a Science, vol. 6, Sigmund Koch, ed. New York, McGraw-Hill, pp. 388-428.

Furth, Hans G. l961 The Influence of Language on the Development of Concept Formation in Deaf Children. Journal of Abnormal and Social Psychology 63:386-389. Garvin, Paul L. 1949 Standard Average European and Czech. Studia Linguistica 3:65-85.

1956 Review of Language in Culture; Conference on the Interrelations of Language and Other Aspects of Culture, Harry Hoijer, ed. American Anthropologist 58:568-569.

Gastil, Raymond D. 1959 Relative . Anthropological Linguistics 1 (9):24-38.

Glenn, Edmund S. 1954 Semantic Difficulties in International Communication. Etc., A Review of General Semantics 11:163-180. Goodenough, Ward H. 1955 Review of Language in Culture; Conference on the Interrelations of Language and Other Aspects of Culture, Harry Hoijer, ed. Language 31:241-245. 99

1964 Cultural Anthropology and Linguistics. In Language in Culture and Society, Dell Hymes, ed. New York, Harper & Row, pp. 36-39. Graves, Robert 1957a Comments on Lineal and Nonlineal Codifications. Explorations 7:46-51.

1957b Comments on Symbolization and Value. Explorations 7:67-73.

Greenberg, Joseph H. 1954 Concerning Inferences from Linguistic to Nonlinguistic Data. In Language in Culture; Conference on the Interrelations of Language and Other Aspects of Culture, Harry Hoijer, ed. Chicago, University of Chicago Press, pp. 3-19.

1959 Current Trends in Linguistics. Science 130:1165-1170.

1963 Language and Linguistics. In The Behavioral Sciences Today, Bernard Berelson, ed. New York, Basic Books, pp. 126-138.

1964 Linguistics and Ethnology. In Language in Culture and Society, Dell Hymes, ed. New York, Harper & Row, PP• 27-35. Hallowell , A. Irving 1965 Ojibwa Metaphysics of Being and the Perception of Persons. In Person Perception and Interpersonal Behavior, Renato Tagiuri and Luigi Petrullo, ed. Stanford, Stanford University Press, pp. 63-85.

1970 Cultural Factors in the Structuralization of Perception. In Social Psychology at the Crossroads, John H. Rohrer and Muzafer Sherif, ed. Freeport, New York, Books for Libraries Press, pp. 164-195.

Hanawalt, Nelson G., and Isabelle H. Demarest 1939 The Effect of Verbal Suggestion in the Recall Period upon the Reproduction of Visually Perceived ~orms. Journal of Experimental Psychology 25:159-174.

Hanks, L. M., Jr. 1954 A Psychological Exploration in the Blackfoot Language. International Journal of American Linguistics 20:195-205.

Hayakawa, S. I. 1949 Language in Thought and Action. New York, Harcourt, Brace and Company. 100

Heidbreder, Edna 1958 Woodworth and Whorf on the Role of Language in Thinking. In Current Psychological Issues, Georgene S. and John P. Seward, ed, New York, Henry Holt and Company, pp. 86-110. Henle, Paul 1958 Language, Thought, and Culture. In Language, Thought, and Culture, Paul Henle, ed. Ann Arbor, Press, pp. 1-24.

Herman, David T., Richard H. Lawless, and Richard W. Marshall 1961 Variables in the Effect of Language on the Reproduction of Visually Perceived Forms. In Psycholinguistics; A Book of Readings, Sol Saporta, ed. New York, Holt, Rinehart and Winston, pp. 537-551. Hockett, Charles F. 1948 Biophysica, Linguistics, and 'the Unity of Science. American Scientist 36:558-572. 1954 Chinese versus English: An Exploration of the Whorfian Thesis. In Language in Culture; Conference on the Interrelations of Language and Other Aspects o£ Culture, Harry Hoijer, ed. Chicago, University of Chicago Press, pp. 106-123.

Hoijer, Harry 1953 The Relation of Language to Culture. In Anthropology Today, A. L. Kroeber, ed. Chicago, University of Chicago Press, pp. 554-573.

1954 The Sapir-Whorf Hypothesis. In Language in Culture; Conference on the Interrelations of Language and Other Aspects of Culture. Chicago, University of Chicago Press, pp. 92-105.

1964a Cultural Implications of Some Navaho Linguistic Categories. In Language in Culture and Society, Dell Hymes, ed. New York, Harper & Row, pp. 142-153.

1964b Linguistic and Cultural Change. In Language in Culture and Society, Dell Hymes, ed. New York, Harper & Row, pp. 455-466.

1966 Anthropological Linguistics. In Trends in European and American Linguistics, 1930-1960, Christine Mohrmann, Alf Sommerfelt, and Joshua Whatmough, ed. Utrecht, The Netherlands, Spectrum Publishers, pp. 110-127. 101

Hoijer, Harry, et al. 1954 The Discussions. In Language in Culture; Conference on the Interrelations of Language and Other Aspects of Culture. Chicago, University of Chicago Press, pp. 127-280.

Hymes, Dell H. 196la Linguistic Aspects of Cross-Cultural Personality Study. In Studying Personality Cross-Culturally, Bert Kaplan, ed. New York, Harper & Row, pp. 313-359.

l96lb On Typology of Cognitive Styles in Language (With Examples from Chinookan). Anthropological Linguistics 3 (1):22-54.

Jakobsen, Roman 1953 Chapter Two. In Results of the Conference of Anthropologists and Linguists, by Claude Levi-Strauss, Roman Jakobsen, c. F. Voegelin, and Thomas A. Sebeok. Indiana University Publications in Anthropology and Linguistics, Memoir 8. Baltimore, Waverly Press, pp. 11-21.

Jeffrey, Wendell E. 1953 The Effects of Verbal and Nonverbal Responses in Mediating an Instrumental Act. Journal of Experimental Psychology 45:327-333.

Kilpatrick, F. P. 1955 Perception Theory and General Semantics. Etc., A Review of General Semantics 12:257-264.

Kluckhohn, Clyde 1954 Culture and Behavior. In Handbook of Social Psychology, vol. 2, Gardner Lidnzey, ed. Reading, Massachusetts, Addison-Wesley Publishing Company, Inc., pp. 921-976.

1956 Review of Language in Culture; Conference on the Interrelations of Language and Other Aspects of Culture, Harry Hoijer, ed. American Anthropologist 58:569-574.

1961 Notes on Some Anthropological Aspects of Communication. American Anthropologist 63:895-910.

Kluckhohn, Clyde and Dorothea Leighton 1962 The Navaho. Garden City, New York, Doubleday, The Natural History Library. Revisions by Lucy H. Wales and Richard Kluckhohn. Korzybski, Alfred 1951 The Role of Language in the Perceptual Processes. In Perception, Robert R. Blake and Glenn B. Ramsey, ed. New York, Ronald Press Company, pp. 170-205. 102

Kroeber, A. L. 1941 Some Relations of Linguistics and Ethnology. Language 17:87-91.

Kurtz, Kenneth H. and Carl I. Hovland 1953 The Effect of Verbalization During Observation of Stimulus Objects Upon Accuracy of Recognition and Recall. Journal of Experimental Psychology 45: 157-164.

Landar, Herbert 1965 Language and Culture. New York, Oxford University Press.

Langacker, Ronald W. 1968 Language and its Structure. New York, Harcourt, Brace & World, Inc. Leary, William G. 1952 Studies in Language and Culture in the Training of Foreign Service Personnel. Etc., A Review of General Semantics 9:192-202. Lee, Dorothy Demetracopoulou 1938 Conceptual Implications of an Indian Language. Philosophy of Science 5:89-102.

1939 Linguistic Reflection of Wintu•' Thought. International Journal of American Linguistics 10:181-187.

1940 A Primitive System of Values. Philosophy of Science 7:355-378.

1943 The Linguistic Aspect of Wintu•~ Acculturation. American Anthropologist 45:435-440.

1944 Categories of the Generic and the Particular in Wintu•L. American Anthropologist 46:362-369.

1950 Lineal and Nonlineal Codifications of Reality. Psychosomatic Medicine 12 (2):89-97.

1963a Being and Value in a Primitive Culture. In Freedom and Culture, by Dorothy Lee. Englewood Cliffs, New Jersey, Prentice-Hall, Inc., pp. 89-104. 1963b The Conception of the. Self Among the Wintu Indians. In Freedom and Culture, by Dorothy Lee. Englewood Cliffs, New Jersey, Prentice-Hall, Inc., pp. 131-140.

1963c Symbolization and Value. In Freedom and Culture, by Dorothy Lee. Englewood Cliffs, New Jersey, Prentice­ Hall, Inc., pp. 78-88. 103

Lenneberg, Eric H. 1953 Cognition in Ethnolinguistics. Language 29:463-471.

1955 A Note on Cassirer's . Philosophy and Phenomenological Research 15:512-522. 1957 A Probabilistic Approach to Language Learning. Behavioral Science 2:1-12.

1961 Color Naming, Color Recognition, Color Discrimination: A Re-Appraisal. Perceptual and Motor Skills 12: 375-382.

1962 The Relationship of Language to the Formation of Concepts. Synthese 14:103-109.

Lenneberg, Eric H. and John M. Roberts 1956 The Language of Experience. Indiana University Publications in Anthropology and Linguistics, Memoir 13. Baltimore, Waverly Press, Inc.

1961 The Language of Experience. In Psycholinguistics: A Book of Readings, ed. Sol Saporta. New York, Holt, Rinehart and Winston, pp. 493-502. , Levi-Strauss, Claude 1951 Language and the Analysis of Social Laws. American Anthropologist 53:155-163 •. 1953 Chapter One. In Results of the Conference of Anthropologists and Linguistics, by Claude Levi­ Strauss, Roman Jakobsen, C. F. Voegelin, and Thomas A. Sebeok. Indiana University Publications in Anthropology and Linguistics, Memoir 8. Baltimore, Wav~rly Press, pp. 1-10.

Liberman, Alvin M., Katherine Safford Harris, HowardS. Hoffman, and Belver c. Griffith 1957 Th~'u1sctimination of Speech Sounds Within and Across Phoneme Boundaries. Journal of Experimental Psychology 54:358-368.

Longacre, Robert E. 1956 Review of Language and Reality, by Wilbur M. Urban, and Four Articles on Metalinguistics, by Benjamin Lee Wharf. Language 32:298-308.

Lotz, John 1964 On Language and Culture. In Language in Culture and Society, Dell Hymes, ed. New York, Harper & Row, pp. 182-184. 104

Lounsbury, Floyd 1959 Language. In Biennial Review of Anthropology, 1959, Bernard J. Siegel, ed. Stanford, Stanford University Press, pp. 185-209.

Luh Chih-wei 1948 Language Forms and Thought Forms. Yenching Journal of Social Studies 4:107-119.

Maclay, Howard 1958 An Experimental Study of Language and Non-Linguistic Behavior. Southwestern Journal of Anthropology 14: 220-229.

Malkiel, Yakov 1956 Review of Language in Culture; Conference on the Interrelations of Language and Other Aspects of Culture, ed. Harry Hoijer. International Journal of American Linguistics 22:77-84.

Mandell, Sibyl 1931 The Relation cf Language to Thought. The Quarterly Journal of Speech 17:522-531. Marshall, John C. 1965 Review of Language, Thought, and Reality: Selected Writings of Benjamin Lee Whorf, ed. John B. Carroll. Linguistics 15:78-81.

Mathiot, Madeleine 1964 Noun Classes and Folk Taxonomy in Papago. In Language in Culture and Society, Dell Hymes, ed. New York, Harper & Row, pp. 154-163.

McQuown, Norman A. 1954 Analysis of the Cultural Content of Language Materials. In Language in Culture; Conference on the Inter­ relations of Language and Other Aspects of Culture, Harry Hoijer, ed. Chicago, University of Chicago Press, pp. 20-31. Miller, George A. 1951 Language and Communication. New York, McGraw-Hill.

1962 Some Psychological Studies of Grmmar. American Psychologist 17:748-762. Miller, George A. and Jennifer A. Selfridge 1950 Verbal Context and the Recall of Meaningful Material. American Journal of Psychology 63:176-185. 105

Miron, Murray S. 1961 A Cross-Linguistic Investigation of Phonetic Symbolism. Journal of Abnormal and Social Psychology 62:623-630.

Muller, Henri F. 1945 Phenomenes sociaux et linguistiques: un cas demonstrable de concordance entre phenomenes d'ordre social et phenomenes d'ordre linguistique. Word 1: 121-131.

Newman, Stanley 1954 Semantic Problems in Grammatical Systems and Lexemes: A Search for Method. In Language in Culture; Conference on the Interrelations of Language and Other Aspects of Culture, Harry Hoijer, ed. Chicago, University of Chicago Press, pp. 82-91. Nida, Eugene 1945 Linguistics and Ethnology in Translation-Problems Word 1:194-208. 0' ;Flaherty, James c. 1952 Unity and Language: A Study in the Philosophy of Johann Georg Hamann. University of North Carolina Studies in the Germanic Languages and Literatures, No. 6. Chapel Hill.

II Ohman, Suzanne 1953 Theories of the "Linguistic Field." Word 9:123-134. Olmsted, David L. 1950 Ethnolinguistics So Far. Studies in Linguistics,. Occasional Papers 2:1-16.

Opler, Morris Edward 1946 An Application of the Theory of Themes in Culture. Journal of the Washington Academy of Sciences 36: 137-166.

Osgood, Charles E. 1963 An Exploration into Semantic Space. In The Science of Human Communication, Wilbur Schramm, ed. New York, Basic Books, Inc., pp. 28-40.

1966 Language Universals and Psycholinguistics. In Universals in Language, Joseph H. Greenberg, ed. Cambridge, M.I.T. Press, pp. 299-328. Penn, Julia M. 1972 Linguistic Relativity versus Innate Ideas. Janua Linguarum, no. 120. The Hague, Mouton. 106

Prentice, W. C. H. 1954 Visual Recognition of Verbally Labeled Figures. American Journal of Psychology 67:315-320. Pyles, Marjorie K. 1932 Verbalization as a Factor in Learning. Child Development 3:108-113.

Radcliffe-Brown, A. R. 1957 A Natural Science of Society. Glencoe, Free Press.

Radnitzky, Gerard A. 1961 Some Remarks on the Whorfian Hypothesis. Behavioral Science 6:153-157. Rapoport, Anatol 1958 General Semantics: Its Place in Science. Etc., A Review of General Semantics 16:80-97.

Rapoport, Anatol and Arnold Horowitz 1960 Does Language Structure Influence Thought? The Sapir-Whorf-Korzybski Hypothesis: A Report and a Reply, Etc., A Review of General Semantics 17: 346-363.

Reichard, Gladys A. 1950 Language and Cultural Pattern. American Anthropologist 52:194-204. Reiser, Oliver L. 1943 Historical-Cultural Significance of Non-Aristotelian MOvement and the Methodological Contributions of Korzybski. In Papers from the Second American Congress on General Semantics, M. Kendig, ed. Chicago, Institute of General Semantics, pp. 3-10.

Rossman, Irma L. and Albert E. Goss 1951 The Acquired Distinctiveness of Cues: The Role of Discriminative Verbal Responses in Facilitating the Acquisition of Discriminative Motor Responses. Journal of Experimental Psychology 42:173-182.

Sapir, Edward 1907 Herder's "Ursprung der Sprache." Modern Philology 5:109-142.

1949 Language. New York, Harcourt, Brace & World.

1963a Language. In Selected Writings of in Language, Culture and Personality, David G. Mandelbaum, ed. Berkeley, University of California Press, pp. 7-32. 107

1963b The Status of Linguistics as a Science. In Selected Writings of Edward Sapir in Language, Culture and Personality, David G. Mandelbaum, ed. Berkeley, University of California Press, pp. 160-166.

1964 Conceptual Categories in Primitive Languages. In Language in Culture and Society, Dell Hymes, ed. New York, Harper & Row, p. 128. Shepard, Winifred 0. 1956 The Effect of Verbal Training on Initial Generaliza­ tion Tendencies. Child Development 27:311-316.

Smith, Henry Lee, Jr. and George L. Trager 1952 Metalinguistics. Etc., A Review of General Semantics 9:163-164.

Sommerfelt, Alf 1938 Les formes de la pensee et !'evolution des categories de la grammaire. Journal de psychologie normale et pathologique 35:170-184. Spiker, Charles c. 1956 Experiments with Children on the Hypotheses of Acquired Distinctiveness and Equivalence of Cues. Child Development 27:253-263.

Suci, George J. 1960 A Comparison of Semantic Structures in American Southwest Culture Groups. Journal of Abnormal and Social Psychology 61:25-30.

Swanson, J. W. 1962 Linguistic Relativity and Translation. Philosophy and Phenomenological Research 22:185-192.

Thompson, Laura 1946 In Quest of an Heuristic Approach to the Study of Mankind. Philosophy of Science 13:53-66.

1950 Science and the Stu~y of Mankind. Science 111:559-563.

Trager, George L. 1959 The Systematization of the Wharf Hypothesis. Anthropological Linguistics 1 (1):31-35.

Ullmann, Stephen 1959 The Principles of Semantics. Glasgow, Basil, Blackwell & Mott, Ltd.

1966 Semantic Universals. In Universals of Language, Joseph H. Greenberg, ed. Cambridge, M.I.T. Press, pp. 217-262. 108

Urban, Wilbur Marshall 1961 Language and Reality. London, George Allen & Unwin. Vigotsky, L. s. 1961 Thought and Speech. In Psycholingu:f.l3tics: A Book of Readings, Sol Saporta, ed., trans. Helen Kogan, Eugenia Hartmann, and Jacob Kasanin. New York, Holt, Rinehart, and Winston, pp. 509-535. Voegelin, C. F. and z. S. Harris 1945 Linguistics in Ethnology. Southwestern Journal of Anthropology 1:455-465.

Voegelin, C. F. and Thomas A. Sebeok. 1953 Chapter Three. In Results of the Conference of Anthropologists and Linguists, by Claude Levi-Strauss, Roman Jakobsen, C. F. Voegelin, and Thomas A. Sebeok. Indiana University Publications in Anthropology and Linguistics, Memoir 8. Baltimore, Waverly Press, PP• 22-67. Yoegelin, c. F., John F. Yegerlehner, and Florence M. Robinett 1954 Shawnee Laws: Perceptual Statements for the Language and for the Content. In Language in Culture; Conference on the Interrelations o~ Language and Other Aspects of Culture, Harry Hoijer, ed. Chicago, University of Chicago Press, pp. 32-46. von Bertalanffy, L. 1955 An Essay on the Relativity of Categories. Philosophy of Science 22:243-263.

Walker, Donald E., James J. Jenkins, and Thomas A. Sebeok 1965 Language, Cognition, and Culture. Section 7.4 in Psycholinguistics: A Survey of Theory and Research Problems, Charles E. Osgood and Thomas A. Sebeok, ed. Bloomington, Indiana University Press, PP· 192-203. Waterman, John T. 1957 Benjamin Lee Whorf an~ Linguistic Field-Theory. Southwestern Journal of Anthropology 13:201-211.

Weinreich, Uriel 1955 Review of Language in Cultura; Conference on the Interrelations of Language and Other Aspects of Culture, ed. Harry Hoijer. Word 11:426-430.

Weinstein, Edwin A. 1958 Changes in Language Pattern as Adaptive Mechanisms. In Psychopathology of Communication, Paul H. Hoch and Joseph Zubin, ed. New York, Grune & Stratton, pp. 262-271. 109

Wells 1 Rulon 1962 What Has Linguistics Done for Philosophy? The Journal of Philosophy 59:697-708.

Werner, Heinz and Bernard Kaplan 1956 The Developmental Approach to Cognition: Its Relevance to the Psychological Interpretation of Anthropological and Ethnolinguistic Data. American Anthropologist 58:866-880. Wescott, Roger W. 1960 The Metalinguistics of Bini: A West African Language. Anthropological Linguistics 2 (6):19-21.

Whatmough, Joshua 1956 Language: A Modern Synthesis. New York, St. Martin's Press. Whorf, Benjamin Lee 1971 Language, Thought, and Reality: Selected Writings of Benjamin Lee Wharf, John B. Carroll, ed. Cambridge, M.I.T. Press.

Win, Khin Maung 1960 The Psychology of the Burmese Language, Main Currents in MOdern Thought 16:110-114.

Wissler, Clark 1966 Indians of the United States. Garden City, New York, Doubleday & Company, Inc.

1967 North American Culture Areas. In The North American Indians: A Sourcebook, Roger c. Owen, James J. F. Deetz, and Anthony D. Fisher, ed. New York, Macmillan, pp. 28-41.

Yuen Ren Chao 1959 How Chinese Logic Operates. Anthropological Linguistics 1 (1):1-8. Zipf, George K. 1965 The Psycho-Biology of Language. Cambridge, M.I.T. Press.