Analysis of Meaning and Dictionary Making

Author(s): Eugene A. Nida

Source: International Journal of American Linguistics , Oct., 1958, Vol. 24, No. 4, Franz
Boas Centennial Volume (Oct., 1958), pp. 279-292
Published by: The University of Chicago Press

Stable URL: https://www.jstor.org/stable/1263976

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend
access to International Journal of American Linguistics

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms
 ANALYSIS OF MEANING AND DICTIONARY MAKING

EUGENE A. NIDA
AMERICAN BIBLE SOCIETY

0. The problem in general By cultural context we mean the description

1. Types of dictionaries of a process or object as a part of the culture.
2. Problems of traditional methodology As long as a dictionary is written in the
3. Basic principles same language as the words it is describing
4. Relationship of language to culture and is prepared for people who participate
5. Totality of language-culture correlation
fully in the culture which is being described,
6. Frequency of semantic units the problems of analysis of meaning are ap-
7. Size of semantic units preciably simpler than in the case of dic-
8. Methods of investigation tionaries in which the language of descrip-
9. Componential plotting of a semantic tion is different from the language being
field described. Furthermore, when there are
10. Diagrammatic plotting of semantic serious lacunae, the reader can supply what
distinctions is lacking by virtue of his knowledge of the
11. Componential analysis of individual cultural context. However, as soon as one
terms undertakes to prepare a dictionary of a
foreign language, the problems seem to in-
0. Meaning, the indispensable helper, butcrease with geometric proportion, depending
often repudiated friend, of science, is at last upon the degree of linguistic and cultural
coming into her own. Information theory,diversity. That is to say, a dictionary of
the advance of anthropology into the realm French words written in English is relatively
of personality and group psychology, the simple (because of the essentially similar
inevitable necessity of structural linguists to languages and cultures) in comparison with
deal with meaning (despite its lack of struc-a dictionary of Zulu in English. Too often,
tural neatness), and political events in ourhowever, compilers of dictionaries are not
present world have all joined to force uponsufficiently aware of the profound structural
us an awareness of the necessity, impor-differences between languages and between
tance, and scientific basis of communication. cultures or they take for granted that the
An essential part of investigations in reader will infer the essential distinctions.
semantics is reflected in dictionary-making, There are, of course, a number of different
a highly specialized process, which, how-kinds of dictionaries, depending upon the
ever, too often reflects only a meager ap- purposes for which they are designed, but if
preciation of some fundamental problems we omit from consideration those diction-
involved in the analysis of meaning. aries which are produced solely for com-
mercial purposes and include only those
1. Dictionaries are essentially descrip- dictionaries which are designed to provide
tions of the distribution of language units scientifically useful data, we are left with
(usually words) in terms of linguistic and three fundamentally different types of
cultural contexts, though in general the dictionaries: (1) a list of words with identi-
cultural contexts predominate. By linguistic fying glosses, (2) a list of words with more or
context we mean the phrases or sentences in less full treatment of types of occurrences
which such words are or have been used (in drawn from text material, and (3) a list of
general this means citing literary sources).words with more or less exhaustive treat-
279

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms
280 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS VOL. XXIV

ment of the kinds of cultural contexts in object-response. However, even here we are
which such words occur. In the first instance not sure of the details in the historical de-
the list of glosses, which serve primarily as velopment, despite a rather extensive
identificational tags, assists one materially in literature, for we cannot be certain of usage
an analysis of a text and in the working out during those periods represented by gaps
of structural relationships. The second type in available data, nor can we ascertain the
of dictionary consists of a compendium of developments in the colloquial use of this
usage in which words are classified and word. Furthermore, we cannot assume that
illustrated on the basis of linguistic contexts. the histories of the literary and colloquial
The third type is essentially an "ethno- usage are strictly parallel. Synchronic func-
linguistic dictionary," in that it relates tioning constantly "remakes" the historical
linguistic units of semantic relevance to the patterning. We know, for instance, that
total context of cultural behavior. There is historically the by in bylaw is not the same
little value, for example, in saying thatby a as the by of bypath and byproduct, but for
word means female puberty rite if there isthe no average speaker of English there is no
available ethnological data on the culturesuch in distinction. The historical method is,
question. Similarly, to say that a word means of course, utterly inadequate for languages
good-bye is also relatively useless, unlesswhich we have no available historical record,
know under what circumstances it is ut- for there are no inviolate laws of semantic
development. For example, one of the most
tered: at what time of day or night, in antic-
generally accepted principles is that in
ipation of how long an absence, to what
instances of related object and process
classes of people, after or before other words
of parting, combined with what variety words,
of the latter are derived from the
gestures, intonation, or voice quality,former,
etc. but that is not always the case. In
In actual practice most dictionaries Tarahumara
are a the words mi6uruku shavings,
blend of the three basic types, withrituku
dif- ice, pa6iki an ear of maize, and
fering proportions of data dependingopacaka
upon garment are derivative formations
from the underlying forms mi6uru to make
the practical needs of the intended users.
shavings, ritu to be icy, paci to grow ears of
2. No doubt much of our difficulty in and opaca to be dressed.
maize,
dealing with bilingual and bicultural dic-
For anyone working with aboriginal
languages the historical method obviously
tionary problems has been the inadequacy
has its strict limitations, except where
of certain traditional methods of semantic
analysis and the tendency to mix classi-
certain reconstructions seem warranted on
ficatory criteria and to shift methodology the basis of comparative data. As a result
depending upon the ease of treating any the tendency has been to substitute a 'logical
particular word. The two most commonly arrangement' of meanings, on the basis that
employed techniques have been based upon certain meanings can be logically derived
(1) historical lineage and (2) central-pe- from each other. This logical framework,
ripheral plotting. however, as useful as it may be for certain
Where there is abundant historical data classificatory purposes, does not necessarily
and where the line of descent seems rela- reflect either the historical development or
tively clear, one encounters few difficulties.
the relationships between different meanings
For example, the Greek word kharis means as they may be understood by native
outward loveliness, kindness (as a quality of
speakers of the language.
The so-called logical method of analysis
personality), favor (an act), gift, delight, and
thankfulness. This can be treated as a linealand arrangement of meanings is, however,
development: appearance-character-action- very difficult to apply, for categories which

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms
NO. 4 ANALYSIS OF MEANING AND DICTIONARY MAKING 281

are familiar to us as foreigners often lected so as to make possible the accounting

seem
incapable of application to the array of
for alternative forms by certain generally
meanings of words in foreign languages. applicable morphophonemic statements, we
Furthermore, the more one becomes in- should set up such a central form. However,
timately familiar with a language, especially when the data do not justify any such
one with widely different linguistic and selection, we must be content with the facts
semantic structures, the more it becomes of the language as they are and not impose
evident that logical criteria drawn primarily any arbitrary categories upon the material.
from one language-culture complex cannot Something similar exists in the case of the
be easily or validly applied to another. classification of what we may call the
Accordingly, rather than attempt a allosemes of a sememe.
logical ordering of meanings, those preparing We do not wish to imply that the tech-
dictionaries tend to describe meanings in niques of historical, logical, and central-
terms of central and peripheral meanings, peripheral analysis and description are not
though usually these distinctions are not useful. They are. But their usefulness is
explicitly stated. What often happens isstrictly limited and, as I hope to point out
that the first meaning given is assumed to bein the rest of this paper, there are certain
central and that the meanings which follow other techniques which seem to provide
are arranged more or less in terms of suc- more fruitful results, both in terms of
cessive distances from the center. Despiteanalytical methodology as well as descrip-
certain serious problems encountered in the tive arrangement.
application of this method, it must be
recognized that it is distinctly superior to 3. Though in general those who compile
the practice of trying to state the funda-dictionaries are supposedly aware of the
mental "core" or the "common denomi- basic principles of semantic correspondence
nator" of meaning, sometimes described (or lack
as of it), some dictionaries are for-
the "central idea" inherent within each mulated with apparent disregard for the three
fundamental presuppositions which must
meaning and one from which the related
underlie all adequate semantic analysis:
meanings are presumably derived. Never-
(1) noin
theless, even the arrangement of meanings word (or semantic unit) ever has
terms of central and peripheral involves exactly the same meaning in two different
two fundamental limitations: (1) theutterances; multi- (2) there are no complete
dimensional relationships of the datasynonyms (which within a language; (3) there are
cannot easily be squeezed into what is no exact correspondences between related
essentially a two-dimensional analysis) and words in different languages. In other words,
(2) the false assumption that words should perfect communication is impossible, and all
or always do have "central" meanings. In communication is one of degree. The state-
many instances it simply is not possible nor ment of equivalences, whether in diction-
useful to try to describe a series of meanings aries or in translations, cannot be absolute.
of a word in terms of any central meaning We are faced, therefore, not with a problem
and peripheral uses. But this should not of 'right or wrong' but with 'how right' or
unduly alarm us. We have had to abandon a 'how wrong.' Perhaps, because of the es-
similar practice on a morphological level. sentially negative character of these prin-
For example, we no longer feel compelled ciples and the difficulties of application, the
always to choose one allomorph as the basic compilers of dictionaries react in favor of
allomorph from which all others are de- describing what is known rather than what
scriptively derived. Of course, if in a series is not known. Nevertheless, the essentially
of related forms one allomorph can be se- negative elements in the basic principles of

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms
282 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS VOL. XXIV

semantic equivalence must be constantly

rapid breakdown of the old cultural patterns
recognized if one is to deal adequately (including
with the complex series of religious
the broader ethnolinguistic relationships.beliefs), this grammatical dichotomy is
Furthermore, though dictionary compilers losing its synchronic validity and rapidly
cannot attain an absolute definition of a becoming another linguistic 'fossil', with
term in another language or culture which
(or all languages are strewn. That is to
even in the same language or culture-for say, this tie between formal language
that matter), nevertheless, they can structure
give and cultural behavior is breaking,
very useful approximate descriptions. but, as in all languages, there are other newly
emerging formations which reflect current
4. Whatever we may personally think cultural
of developments.
When one proceeds from the level of
structural analysis as divorced from mean-
grammatical categories (which are largely
ing or of the influence of grammatical cate-
implicit) to the level of words, which are
gories on thought processes, we certainly
symbols for dynamic and explicit features of
must admit the close relationship between
language and culture. Language cannotthe culture,
be one is obliged to interpret the
properly treated except in terms of meaning
its of such linguistic units in light of
the
status and function as a part, a process, cultural context. That is to say, the
and,
meaning
to some degree, a itodel of culture, with a of a unit must be described in
high degree of reciprocal reenforcement.terms of the sum total of what it signals in
all the contexts in which it is used. Note that
Though one may not wish to go all the way
with Whorf, nevertheless, one cannot we specifically reject meaning as 'a common
escape
denominator' or 'what is common to all
the fact that language seems to provide the
'grooves for thought' in the same waysituations
that in which a term is employed.'
cultural patterns constitute the molds If,for
for example, we analyze the use of charge
more general modes of behavior. in the following contexts, we will find that a
An illustration of a close tie between common denominator would be precious
little
language and culture is provided by the two indeed. It would be only a small part
'possessive' systems in New Caledonian.' total meaning signalled by charge in
of the
These may be roughly distinguished the as
various contexts: charge into the line of
'intimate' and 'non-intimate' possession. charge the gun, charge the battery,
players,
The first class includes such nouns as those charge the pencil, charge the man ten dollars,
charge the culprit with the crime, he gets a
meaning mother, liver, and descendants,
charge out of it, a charge of electricity, he is in
while the second class includes father, heart,
and personal life. The apparently arbitrary charge, he is a public charge. The only way to
character of the distinction can only be 'define' the meaning of charge is to describe
understood if one realizes that New Cale- (usually by illustrative phrases or sentences)
the distribution of the word.
donian society has been traditionally mat-
When, however, we speak of 'the distri-
rilineal, that the liver has been regarded as
bution' of a word we must distinguish
symbolic of the entire person (the liver is
between (1) the specific linguistic context,
used in sacrifices as symbolizing the victim),
which gives a form a linguistic meaning, and
and that one's descendants have a more
(2) the practical-world (non-linguistic) con-
intimate, continuing relationship to a person
text, which provides what is more generally
than even his own life. However, with the
understood by the meaning of a word.
1 Maurice Leenhardt, Do Kamo: La Personne
Obviously, the so-called 'function words,'
et le Mythe sans le Monde M6lan6sien (Paris: (following Fries' usage) have predominantly
Gallimard, 1947, pp. 21-24). linguistic meaning.

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms
NO. 4 ANALYSIS OF MEANING AND DICTIONARY MAKING 283

The cultural distribution must, however, vocabulary. These correlations may be

reckon not only with objective events summarized
butas follows:
with evaluations of the events and the cor- (1). The vocabulary relating to the focus
responding symbols. The emotion which aof the culture is proportionately more ex-
patriot feels when he uses the phrase 'Oldhaustive than that which refers to nonfocal
features. That is to say, the extent of
Glory' is utterly unintelligible to a Nuer of
the Sudan, who neither has nor understandsvocabulary relating to any phase of culture
banners. But when he dances before a is directly proportionate to its cultural
favorite bull and calls outs its name, he
relevance.
experiences a similar 'thrill' in having(2).
ut-Subcultures have proportionately
more extensive vocabularies in the area of
tered an emotionally charged expression.
We can say that the cultural event their distinctiveness.
symbolized by a word provides the denota- The proportionately greater vocabulary in
tive meaning, while the emotional responsethe area of the focus of a culture is almost a
experienced by the speakers in the culturetruism, but nevertheless a fact which is not
(and modeled by the culture) is the basis ofinfrequently overlooked. For the Nilotic
the connotative meaning. Since there is noNuers cattle are the central fact toward
speech without speakers and no speakerswhich almost all the rest of the culture is
without subjective evaluations (absolute oriented and in the light of which most be-
objectivity is an illusion, for we are parts ofhavior has meaning. Accordingly, one finds
as well as students of culture), there are nomany hundreds of words which describe dif-
words without some measure of connotation. ferent colors (including distribution of
Even apparent neutrality of meaning may be color), sizes, shapes, breeds, behavior, and
regarded as connotatively significant, byvalues of cattle. The English language has
virtue of its apparent lack of emotional nothing even remotely approaching such
coloring. specialized 'cattle' vocabulary. On the other
hand, the Nuer language may be regarded as
5. The correlation between language and very 'deficient' in words for mechanical
culture is perfectly obvious when we are artifacts, of which the Nuers have relatively
dealing with isolated words which reflect few, while English abounds in names for
unusual cultural objects, activities, or gadgets, a reflection of the fact that mechan-
attitudes. For example, the Shilluks of the ical technology is the focus of our culture.
Sudan speak of forgiveness as spitting on the The Ponapeans have an extensive vocab-
ground in front of a person, a description of ulary to describe different forms and
the manner in which forgiveness is formally varieties of sweet potatoes, for the growing
indicated. The Uduks, also of the Sudan, of these tubers is one important focus of
their culture. For us sweet potatoes are a
employ a phrase to meet snapping fingers
very minor feature and accordingly possess
again (a concise description of the cultural
no specialized vocabulary. Similarly, the
event) in contexts where we would use the
abundance of terminology relating to maize:
term reconciliation. The Cuzco Quechuas
its kind, stages of growth, parts, cultivation,
call the year tying up the sun, an obvious
harvesting, and preparation as food, is a
reference to the ancient use of quepus. readily understandable feature in the
However, some of the significant correlations various Mayan Indian languages of southern
between language and culture are not these Mexico and Guatemala. It is easy to under-
more obvious correspondences of individual stand why so many dictionaries prepared by
semantic units (whether single words or outsiders to a culture tend to omit a high
whole phrases), but involve whole sets of percentage of the foci vocabulary, for such

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms
284 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS VOL. XXIV

words are difficult to elicit andlationship extremely between extent of vocabulary and
hard to describe, due to the lack cultural specialization seems even more
of corre-
sponding words and traits in thestriking. For any specialized subgroup with-
compiler's
language and culture. In going through in a culture
suchthe extent of vocabulary is
dictionaries, one is likely to receive proportionately
the im- greater in the area of dis-
pression that the languages in question tinctive specialization,
are since such an area of
quite deficient in words, while actually activity is the
generally the focus of the sub-
apparent scarcity of words results groupinand hence has much greater cultural
large
measure from a failure to give adequate relevance for those people. For example,
treatment to foci vocabulary. people in the fishing villages of Newfound-
Since the extent of vocabulary is roughly land have an abundance of words relating to
reflected in the degree of cultural relevance the sea and their work, while the farming
of the referents of the semantic units, it is people of the interior areas are conspicuously
obvious that such vocabulary is not neces- lacking in such terms. The same principle
sarily a permanent feature. For example, in holds for all occupational subgroups within
many of the Mayan languages of southern a culture-including linguists, who find it
Mexico there is a significant lack of indig- difficult to speak without -eme words.
enous terms for juridical processes and The significance of these correlations be-
government, which must certainly have tween culture and the extent of vocabulary
existed before the conquest and for which should make the dictionary compiler not
there is adequate evidence in the Popol only more aware of the probable volume of
Wuj. However, with the destruction of words in the different areas of the culture,
patterns of indigenous government and the but also more alert to reflect in any abridged
superimposition of foreign authorities, this dictionary (and most dictionaries are
vocabulary largely disappeared. In its place drastically abridged) a truer picture of the
there came into usage a rather meager vocab- proportion of words in the different phases of
ulary drawn from Spanish and reflecting the the culture.
nature of the contacts with the Spanish- However, though there are broad cor-
speaking rulers. A similar shift in the extent relations of a statistical nature between
of vocabulary is taking place in Anuak, culture and vocabulary, the principle of
selectivity operates so extensively in se-
a Nilotic language of the Sudan, in which
there are, for example, eight different terms mantic structure that we cannot anticipate
to describe various methods and stages in the manner in which a particular language
the grinding of corn, but up to within the will treat any given phenomenon. Cultures
last few years only one word for anything may possess the very same traits, but
made of metal, whether a screwdriver or an identify and describe them in utterly diverse
airplane. However, with the rapid increase ways. This is particularly true of psycho-
of contacts with people using metal tools and logical characteristics. For example, the
machines, there is a sizable increase in the Habbes of the French Sudan speak of sorrow
number of borrowed words and coined as having a sick liver. The Bambaras, some-
expressions which are being rapidly intro- what to the west of the Habbes, say that
duced into the language, in order to desig- sorrow is having a black eye. The Mossi
nate these new, culturally valuable objects. people, just north of the Gold Coast, insist
These principles relating to the size that
ofsorrow is having a rotten heart, while
the Uduks
vocabulary in proportion to the cultural in the Sudan describe sorrow as
relevance of certain objects or modes of having a heavy stomach. Psychological phe-
behavior are true not only for cultures as a nomena are not, however, the only features
whole, but for subcultures, where the re- which illustrate the unpredictability of

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms
NO. 4 ANALYSIS OF MEANING AND DICTIONARY MAKING 285

means of symbolization. We speak of the

descriptions of connectives such as de, kai,
eye of a needle, but the Kekchi and Indians
oun. From
of the nature of the lengthy
Guatemala call it the face of thestatementsneedle, of their use one tends to infer
the
Lahu of Southeast Asia and the Piros of that such words are quite important, and
Peru speak of the nostril of the needle, thatthe
they are at least equivalent in
Haka Chins of Burma call it the mouth of
semantic value to the corresponding Eng-
the needle, the Tiddims of Burma referlish conjunctions but, and, and therefore,
to it
by which
as the ear of the needle, the Mitla Zapotecs in they are usually translated. The
Mexico say the face of the needle, and truththeof the matter is that these Greek
Amuzgos, likewise of Mexico, talk conjunctions
about occur so frequently that they
the hole of the needle. mean much less than their English 'corre-
spondants.' A brief comparison of frequencies
6. Information theory has provided
will us
make this evident. Of the first forty sen-
with some very important concepts for tences in Plato's Republic all but six have
making quantitative (and to some extent, some type of connective, of which de, oun,
qualitative) judgments concerning semantic and kai are the most frequent. Of these six
phenomena. The fact that information is in sentences, two begin sections (including one
inverse proportion to redundancy gives us which begins the book) and four are direct
important clues to the relationship between discourses, which are more or less interrup-
certain aspects of frequency and meaning, tive in nature. In the first twenty sections of
though we must not make the mistake of Isocrates' On the Peace every sentence but
assuming that information as used in com- the first has some type of conjunction either
munication theory is the same as information as the first word or as postpositive to the
when used in the popular sense of the extent introductory expression. Of the 55 para-
of meaning which some particular word graphs which make up the treatise On the
might have in a particular context. For Peace all but the first begin with some type
example, if a translator is unable to employof connective, of which de occurs in 22, oun
some indigenous term to describe a foreignin 12, and other connectives in the remaining
object, e.g. phylacteries as spoken of in the 20 paragraphs. No writer of English in any
Scriptures, and so chooses to employ a way approximates this type of frequency.
borrowed word, the unpredictability of In some selected writings of John Ruskin
such a term within the context means that it (reproduced in Twelve Centuries of English
carries a heavy informational load, but it Poetry and Prose, published by Scott,
actually does not mean much to the reader.Foresman and Company) out of the first 35
On the other hand, words such as thing,sentences only four begin with connectives,
matter, object, datum, and item have such and a each of these begins with and. Out of a
high frequency of usage in some types of total of 32 paragraphs, 26 begin without
writing (e.g. in scientific papers, where there connectives, two begin with and, two with
seems to be a premium on generalizations now (in a conjunctive, not temporal, use),
or dullness), that they really contribute very and one each with for and however. James
little to the meaning of a passage. They mean Truslow Adams in his book The Adams
so much that they end up meaning veryFamily (published by Little, Brown, and
little in many contexts. Company, 1930) employs conjunctions to
This problem of meaning and frequencybegin only two sentences out of the first 32.
should be more fully recognized by compilers Of the first 66 paragraphs only 8 have
of dictionaries. For example, in all dic- connectives.
tionaries of Classical Greek with which theOur Greek dictionaries, however, go on
writer is acquainted there are rather full
repeating the traditional statements about

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms
286 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS VOL. XXIV

Greek conjunctions, but they do not compilers,

by dictionary in- for they are hard
to 'alphabetize'
dicate the differences in relative frequency and are too frequently
between the correspondingjudged Greek to beand
slang or transitory. However,
English conjunctions. As a result, the they often play a major role in the process of
average user of a Greek dictionary is badly communication. The actual inventory of
misled in his judgments about the relative different words in some of the so-called
importance of such conjunctions in the re- primitive languages of the world may not be
spective languages. conspicuously high, but the number of
highly exocentric idioms is often very great.
7. Semantic analysis begins with the For example, if one were to construct a
morpheme and concludes with the discourse. dictionary of Anuak along traditional lines,
For the dictionary maker, however, the the language would appear to be almost
derivational layer of word formation has devoid of terms to describe psychological
generally been regarded as the minimal unit. states and attitudes, while as a matter of
Between the morpheme and the 'dictionary fact, the language abounds in such ex-
level' there is the great no-man's land which pressions. However, most of these are
the structural linguist usually refuses to exocentric phrases containing the word
touch, and which the dictionary maker cwiny liver: he has a cwiny (he is good),
regards as beneath his level of analysis. It his cwiny is good (he is generous), his
is very true that the meaning of sub-word cwiny is bad (he is unsociable), his cwiny is
units is hard to define and describe, for the shallow (he gets angry quickly), his cwiny is
meaning is primarily determined by the heavy (he is sad), his cwiny is stubborn
linguistic context. For example, bad, good, (he is brave), his cwiny is white (he is kind),
kind, and full can be described in terms of his cwiny is cold (he will not be impolite in
their cultural contexts, but -ness, which may eating ahead of others), his cwiny is burned
be suffixed to bad, good, kind, and full, but (he is irritable), his cwiny is sweet (he is
not to well, cannot be so easily defined. happy)-to mention only a few.
Above the level of the word the dictionary It may be argued by some that for the
compiler is generally not in too much dif- Anuak such phrases are not exocentric, but
ficulty as long as the combinations in the endocentric, that is to say, they constitute
language are largely endocentric in semantic language myths which have 'objective
structure, that is to say, if the meaning of reality' for the people of that culture. No
the whole can be determined by adding up doubt to some extent this is true (even as it
the meanings of the parts (which is true of is somewhat true for comparable idioms in
most utterances). However, so-called idioms all languages), but on the whole the Anuak
are a problem precisely because the meaning people cannot be accused of being ap-
of the whole is not the meaning of the sum preciably more naive about their idioms than
total of the parts. Such expressions are are English-speaking people, who readily
exocentric-and in varying degree. The recognize that one cannot insist on a literal
expression he is in the house has only an word-for-word objectivity in such phrases as
endocentric meaning, but the slightly dead tired, drunk with power, and nothing but
different expression he is in the doghouse horsefeathers.
may be either endocentric, if it refers to an However, having recognized the im-
animal, or exocentric, if it means a man. Of portance of idioms, we must not assume that
all languages have the same types or employ
course, there is the possibility of a man being
inside of a doghouse, but that is an ex- them with the same degree of frequency.
tremely unlikely meaning for the sentence. Cuna, spoken by the San Blas Indians of
Idioms have very often been overlooked Panama, is a language which not only has an

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms
NO. 4 ANALYSIS OF MEANING AND DICTIONARY MAKING 287

abundance of semantically exocentric ex-

various participants in the conference
pressions and uses them with highrevealed what were either (1) obvious
frequency,
but also readily admits borrowedomissions
metaphorsor oversights on the part of the
linguistic
from other languages. On the other investigators or (2) strange lacunae
hand, the
Tarascan language of Mexico notinonly the lingustic
falls usage of the language in
considerably below the average in the question.
number and frequency of metaphors (in Once, however, the problem of exploring
comparison with most languages in Mexico) the semantic field has been accomplished,
but does not readily admit the introductionthere still remains the difficulty of trying to
of idioms from foreign languages. relate the various meanings of individual
Any accurate semantic analysis (and, terms so that the area of meaning covered by
accordingly, adequate dictionary) of a a word or semantically endocentric phrase
language must reckon with the various sizes may be understandably relatable to the
of meaningful units and describe the culture in which it is employed. Moreover,
semantically exocentric combinations. With- one must also be in a position to attempt to
out this one cannot know the extent or determine whether two apparently different
nature of the semantic resources of ameanings given are relatable at all, or whether one
language. must classify the forms as two homoph-
onous expressions.
8. One of the most serious problems These diverse, but closely related, prob-
encountered by any field investigator is thelems require several different investigative
procedures, which we may describe as (1)
exploration of a 'semantic field,' an area of
semantically related terms. The problem is componential plotting of a semantic field,
not too complex in the case of a more or less (2) diagrammatic plotting of semantic
distinctions, and (3) componential analysis
well-delineated series such as kinship terms,
for one can plot the various components, of the meanings of individual terms.
e.g. sex, age, ascending and descending
generations, consanguinity, affinity, etc., 9. As in the case of the terms for shaman,
which may be evident in any particularnoted above, it seemed wise to attempt to
system and then fill in the 'holes'. Similarly, determine the number of words or phrases
in the case of the color spectrum, it is which might be employed in any one
possible to provide informants with thelanguage and how they could be related to
entire range of colors and elicit the cor- one another. This was done by placing in
responding terms, thus providing a complete the vertical listing all the names for shaman
nomenclature for such a limited semantic which had been found by the investigator
field. in any one language. For example, in a
However, the problems become much chart for Aguacateco there were eight
more complex when one is dealing with a different terms in the vertical listing, while
series of terms which seem to have no such in the description of the Kekchi data there
easily definable limitations and in which were only four. In the horizontal listing we
there appears to be a considerable areaplaced of all the functions of shaman, regard-
overlap. Recently, for example, in a con- less of what word might be employed, e.g.
ference of linguistically oriented field in- healing sick, casting spells, foretelling the
vestigators we were attempting to study future, determining the cause of drought,
some of the problems involved in terms burial for of the dead, dedication of new build-
shaman in some of the Mayan languagesings, in officiating at weddings, performance of
Guatemala and southern Mexico. The com- rites before planting and harvesting, and
parison of lists of terms submitted by the
consulting with the spirits of the dead. The

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms
288 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS VOL. XXIV

forms by filling out paradigmatic sets.

next step in procedure was to ask informants
The vertical
which type of person as given in the only difference here is that we are deal-
listing performed which type ofing functions
with the possibilities of the occurrence of
which had been noted. This procedure certain terms within a delimited series of
produced two results: (1) the clarification of semantic functional positions, rather than
functions for the titles employed and (2) the testing the occurrence of morphemes in a
addition of other titles to designate function- specific series of morphological combinations.
aries not already properly included. More- There is, of course, a degree of arti-
over, rather than find a neat division of ficiality about such an approach to a
labor between the various types of such semantic field, for the discovery of such
socio-religious functionaries, it was dis- terms in text material would be preferable,
covered that there was a quite unsuspected but the same problems apply here as in the
degree of overlapping in certain functions. case of morphological analyses: we cannot
At one stage in the investigation, however, always wait for the chance occurrence of a
we became almost hopelessly confused be- form which seems necessary to fill out a
cause of an unsuspected error in the hor- crucial spot in the structural analysis.
izontal listing, for we had introduced not
only functions, but techniques, e.g. burning 10. Once we have explored any semantic
of copal, use of candles, reciting of prayers, field by the processes of componential
going into a trance, speaking in a strange plotting, it becomes immediately evident
language, dancing, use of rock crystals, that a single term may exhibit a wide
making of medicine bundles, counting red diversity of meanings, but within this range
beans, and changing onself into the form of there are certain clearly recognized relation-
an animal. This confusion of function and ships, generally reflecting certain culturally
techniques resulted in a hopeless arrange- significant facts. For example, the Biblical
ment of data, but a separation of the two Hebrew root *kbd occurs in a very wide
soon revealed some interesting correlations range of linguistic-cultural situations, giving
between techniques and functions. More- rise to a number of so-called 'meanings',
over, it soon became quite evident that some listed in an English dictionary as heavy,
of the names used for shaman were primarily much, many, slow, abundant, burdensome,
technique terms, e.g., aj pom master ofdifficult,
the grievous, sluggish, dull, riches,
copal (Aguacateco, Quiche). Others identi- respect, honor, and great. One may regard the
fied primarily a function, e.g. aj cun brujo, meanings as given in English as designations
witch-doctor (Aguacateco, Quiche). Still for classes of contexts in which the root
other terms might be derived primarily*kbd may occur. Such glosses, therefore,
from a technique, but be applicable prin- constitute a kind of grid by which we may
cipally to a function, e.g. the Aguacatecodescribe the range of occurrence, while
term aj wutz mes medium, who always recognizing two essential facts: (1) the
employs a table, as implied in the title, which English glosses stand only for a cluster of
includes a borrowing of Spanish mesa. closely related contexts in Hebrew and (2)
By means of a componential plotting of despite the faithfulness of such a grid to the
the semantic field, even though all the linguistic-cultural contexts in Hebrew, there
various functions and techniques might notis an inevitable degree of skewing by virtue
be known at the time, it is possible, to of the differences between English and
achieve a high degree of satisfactory Hebrew. Nevertheless, the analyst, who
eliciting of data. This is, of course, not sub-must generally make use of a foreign lan-
stantially different from what is done when guage in at least the initial stages of study,
one is attempting to elicit morphological can with profit-and care-employ the

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms
NO. 4 ANALYSIS OF MEANING AND DICTIONARY MAKING 289

foreign language glosses as a useful in- are used in English, but as t

vestigative tool. ployed in Biblical Hebrew, in which the
For example, in the study of the range of nomadic background of th
meaning exhibited by the Hebrew root important factor in giving r
*kbd, one will immediately note four dif- meanings in which a quantit
ferent classes of glosses, as determined by the could be either the basis fo
nature of the referents and the value judg- cultural value or the cause
ments involved: (1) those which designate venience if it were relatively
quantity (but without any value judgment) arrows are intended to show
in terms of mass and number, e.g. heavy, direction of derivation, on t
much, many, abundant, (2) those which de- we know about such pat
scribe certain aspects of inertia, slow (with- language usage, and the two
out evidence value judgment) and sluggish solid and broken, designate t
and dull (with disapproval), (3) those semantic relationship.
which specify certain culturally valued Such a diagram has the advantage of per-
features, e.g. riches, respect, honor, great, mitting a valuable organization of diverse
and (4) those which denote abundance as a data into a relatively small compass, thus
source of features which have only negative allowing for the study of otherwise over-
value: burdensome, difficult, and grevious. looked relationships. On the other hand,
This division into four classes suggests such a diagram includes a number of serious
certain relationships. The first class is limitations: (1) the representation of the
neutral (or central) to the two poles of relationships between the meanings is overly
positive and negative values. The meaning simplified (several planes, plus a dimension
of slow seems to fall somewhere between the of time would be necessary if one were to
neutral series and the culturally disfavored diagram all the factors accurately), (2) the
ones, since the meanings sluggish and dull, arrows imply a kind of etymological descent,
being closely related to slow, are disfavored. which may or may not be true, (3) instead
We may plot certain of these relation- of two degrees of interrelationship, there are
ships diagrammatically in the manner actually several, with a considerable recip-
shown in Diagram A. rocal re-enforcement, which cannot be
shown in such a diagram, and could not eve
CULTURALLY CULTURALLY CULTURALLY be fully ascertained from the av
FAVORED NEUTRAL DISFAVORED dence, and (4) despite all attem
riches < heavy > burdensome such a diagram conform to what we know
of the cultural and linguistic situation re-
flected in the Bible, it is inevitable that the
language employed as a grid will tend to
respect abundant slow difficult skew the data. This last fact should not,
I i/ X \S,B | however, unduly alarm us, for at certain
l o much many\ stages in the semantic analysis of any lan-
glory grievous
i i \'x givu guage the influence of the language of
great +- ' sluggish investigation must be reckoned with. The
% same type of problem confronts the phoneti-
dull cian, who brings to his study of a new lan-
DIAGRAM A guage a background of language experience
which inevitably colors his investigation and
Diagram A is not intended to reflect the introduces judgments which can
relationship between these glosses as they corrected by later phonemic ana

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms
290 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS VOL. XXIV

which the data is analyzed on the samebasis

dangers
of of subjectivity as the deter-
the self-contained system. In the miningcase ofthemes of a culture. The ulti-
of the
semantic analysis we must do substantiallymate validation of either the cultural themes
the same thing, but in the past or there has components lies in the degree
the semantic
been the tendency to use the glosses to whichpro-
they help describe (and hence to
vided by foreign languages without recog- explain) the complex nature of the multi-
nizing either their arbitrary character orrelational data. By applying such a com-
their limitations. ponential analysis to the semantic problems
of a number of languages, we should be able
11. Rather than attempt to diagram all to refine our procedures and reduce the
the relationships which exist between words, degree of subjectivity, but despite the recog-
it is often more meaningful to analyze the nized risks of subjectivity the advantages
various meanings from the standpoint of of such a technique seem to far outweigh its
the glosses and the cultural componentslimitations.
which are to be found in such uses of words.
The above diagram does not, of course,
Diagram B consists of the same meanings indicate all the components which occur
of the Biblical Hebrew *kbd charted with with these meanings (e.g. nonphysical domi-
five culturally relevant components: nance, as in great and respect), but only

heavy much many an great riches t glory slow sug- dull den- dii- grie
spect gish some vous

Physical weight....... + t 4 4- + - + =- -
Inertia ............... + t + + -+ - -
Culturally desirable.. t - d + + + +
Culturally undesir-
able ................ +- - + + + +
Increased degree ..... + + + - +
DIAGRAM B

The meanings listed along the top of those which reveal the broader patterns of
relationships.
diagram B really represent certain classes of
contexts in which the root *kbd may occur. The plus sign means that the feature in
It is not necessary to use such meanings in question occurs; the symbol i means that
English, or any other language, for we couldthe feature may or may not occur. A blank
indicates that the feature does not occur,
simply identify different contexts, but there
would be a great many such columns, e.g.but there is no attempt to state whether the
heavy, as a stone over a grave, much, as of absence of such a feature is relevant in the
water in flood, many, as of men in battlecontext. The occurrence of these symbols
array, etc. By using the English terms depends upon the cultural data of the lan-
heavy, much, many, etc., we group contextsguage in question. In this particular in-
together. Moreover, this is done here on thestance the close relationship between the
basis of distinctions peculiar to English. Ifmeanings heavy, riches, and burdensome can
we should use another language, the "grid"be readily understood if one considers the
would, of course, appear somewhat different. early nomadic culture which gave rise to
The components are selected on the basisthe use of *kbd in these diverse ethnolin-
of (1) their cultural relevance and (2) the guistic contexts. The same heavy objects
number of contexts in which they occur. could be riches as well as burdensome, de-
The choice of components is subject to thepending upon the viewpoint.

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms
NO. 4 ANALYSIS OF MEANING AND DICTIONARY MAKING 291

One of the very evident problems and seem never to be the site of any com-
involved
in any type of componential plotting munity ritual.which
employs a grid dependent upon a foreign
5. "The medicine man is jwok."
language is an inevitable degree 6.of skewing,
"The white man is jwok." Any person
despite all cautions which may who be hastaken.
special abilities (something which is
Hence, in order to appreciate somewhat regarded as true of all white men) is spoken
more fully the value of such an investiga- of as jwok.
tive and analytical procedure, we need to 7. "Radios, cars, airplanes, phonographs,
apply it to the uses of a term in which there and electricity are jwok." Any object whose
are no corresponding terms in the languagefunctioning is inexplicable in terms of the
of the analyst. Furthermore, there are many Anuak frame of reference is jwok.
instances in which one cannot line up a 8. "Anything startling is jwok." The one
series of so-called meanings as based upon exception to this is the appearance of a
distinctions in a foreign language, but oneghost (spirit of a deceased person), which is
must employ a series of contexts as the basis called tipo.
for any horizontal listing. 9. "The sick man has been taken by
A profitable example of the application ofjwok." In this type of context the creator
componential plotting to such a series in jwok may or may not be implied.
which we must speak of cultural contexts, 10. "What can we do now? It all depends
rather than glosses, is provided by the useon jwok." When people give up hope, as in
of jwok in Anuak, a Nilotic language of thethe case of apparent fatal illness, they insist
Sudan. The following ten contexts include that the outcome is up to jwok, but there is
all the principal uses of the term in question, no evidence that they always have in mind
for which in most instances there is obviouslythe creator jwok.
no possible corresponding English gloss: The variety of these ten contexts in which
1. "The one who made the world and jwok may be employed is so great that it
everything in it is jwok." In this type seemsofalmost impossible to 'define' jwok, if
context jwok is always referred to as a by definition we mean the traditional type
person, but any traits of personality areof summary statement which will include
mentioned only in rather vague terms. all the attributes and functions. The word
2. "The juu piny must be placated by jwok not only includes what we generally
offerings and sacrifices." The juu piny (juu regard as God (context 1), but also demons
is the plural form of jwok) are literally gods(context 2), mana (contexts 3, 4, 5, 6, 7, and
of the earth, most of whom seem to have 8), fate (context 10), the sacred (contexts
been borrowed from the neighboring Nuers.3 and 4), and the mysterious (contexts 8 and
For the most part they are malevolent and9). At first glance the use of jwok seems to
they differ in activity and power. The rela-imply confused and inconsistent thinking,
tionships (1) between the juu piny them-or at least a serious incapacity for analytical
selves and (2) between the juu piny and the judgments. But this is not the case. From
creator jwok are not defined. the Anuak viewpoint jwok is employed very
3. "The family shrines are jwok." Small consistently, for by it the Anuaks express
village and family shrines mark places clearly their view of the supernatural. A
where the juu piny are propitiated by offer- componential analysis of the meanings of
ings and sacrifices. jwok reveals some of the essential coherence
4. "That grove of trees is jwok." A few and unity in this term (see Diagram C).
places are regarded as jwok. They are not We have not indicated all the components
numerous, are generally quite isolated fromwhich are present in the various contexts,
any village, are for the most part avoided,but we have listed those which are more sig-

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms
292 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS VOL. XXIV

CONTEXTS
COMPONENTS

1 2 3 4 5 6 7 8 9 10

Extraordinary power .............. + + + + + + + +

Personality
non-human ..................... + - + ? =-
hum an ......................... + +
Fear . ........................... + +- + - 4- + +
Respect......................... + + + + + 4 +
Unfamiliar cause-effect sequences. + + + +
Objects.......................... + + +
Processes ........................ + + + +

DIAGRAM C

nificant and which indicate the unity andspondence with jwok. In contexts 5 through
consistency of the Anuak point of view. We
9 there is no special evidence of taboo, and
have purposely omitted such components aseven in contexts 3 and 4 there is practically
ethical vs. nonethical and secular vs. sacred
no ritual avoidance nor any abrupt lines of
for the very reason that these distinctionsdemarcation.
in their traditional formulations are not
According to the traditional method of
particularly valid or important in Anuak defining meanings there would be endless
religious beliefs. It is true that the creator
questions as to whether jwok meant a per-
jwok is usually benevolent and the juu piny
sonal God or an impersonal power. Com-
are for the most part malevolent; and yet
the benevolent or malevolent characteristics ponential analysis makes it possible to
reject such an 'either-or' proposition in
are not primary nor absolute, and they are
favor of the more culturally relevant 'both-
never related to ethical or nonethical stand-
ards. Similarly, though there is a sense in and' statement. The meaning of jwok can-
which the Anuaks recognize a distinctionnot be stated by means of any simple
between the secular and the sacred (con-formula, but only in terms of the culturally
texts 3 and 4), nevertheless, this distinctionrelevant component features which occur in
is very poorly defined and, in so far as it isdifferent combinations in the diverse con-
employed, shows no one-to-one corre- texts.

This content downloaded from

93.182.108.69 on Sun, 06 Feb 2022 10:36:44 UTC
All use subject to https://about.jstor.org/terms