The Language Informant PDF

THE LANGUAGE INFORMANT The linguistic investigator is an outsider to the language he studies. As long as he remains an outsider, the language will be nothing but noise. It is only as he enters into the experience of the community that he begins to "understand," that...

THE LANGUAGE INFORMANT The linguistic investigator is an outsider to the language he studies. As long as he remains an outsider, the language will be nothing but noise. It is only as he enters into the experience of the community that he begins to "understand," that the noise begins to take shape and "make sense." Little by little he learns the rules of the language and relates them to the rules of his own language. This would be a very long process—as long as it is for children—if the investigator did not have the help of an "insider," someone who interprets for him and helps him to bridge the gap until the equation is established. This person is the informant. More specifically, the informant is one who furnishes the researcher with samples of the language, either as repetitions of what has already been said or as creations of what somebôdy might say. He also explains how the utterances were used or what they meant, using for this explana- tion either his own language or some other. The informant can be a chance speaker of the language from whom the researcher has obtained data, but he is more normally one who more or less regularly meets with the researcher for language study. The term informant is therefore a technical one and should be used judiciously outside of linguistic circles. Educated people have been known 20 Need for Informants 21 to become embarrassed when referred to by the phrase "my informant." To some people the term seems to have the same connotation as "in- former." To avoid causing this embarrassment and misunderstanding one can use less loaded terms, such as assistant, helper, aid, or colleague,,. Need for Informants From what has just been said it should be apparent that informants are indispensable for linguistic research of the kind being treated in this book. Yet nowhere in linguistic literature has there appeared an adequate treatment of the part that informants play in linguistic analysis. It would almost seem as if informants were not the indispensable collaborators they really are. In grammars almost always and in journal articles some- times authors will express their appreciation for the patience, faithful- ness, help, and encouragement of the informant, but it is not often that the appreciation is expressed at such length and with such apparent sin- cerity as it is in the following acknowledgment by Bronislaw Malinowski. That he was speaking as an anthropologist makes no difference, for the linguist depends no less on his informant than does his colleague: T h e A n t h r o p o l o g i s t takes full credit f o r s o m e of his discoveries, b u t f o r the real toil, as well as for t h e degree of intelligence in t h e a p p r o a c h , h e c a n only take p a r t credit. T o a large e x t e n t m y i n f o r m a n t s are responsible for t h e correct i n t e r p r e t a t i o n a n d perspective, f o r the sincerity a n d relevance of w h a t is c o n t a i n e d in these v o l u m e s , as also in m y other e t h n o g r a p h i c writings (1935, I:x). But why precisely do we need informants? First, we need informants in order to get the body of data (the corpus) which is necessary for making generalizations about the structure of a language. 1 This is a particular kind of corpus, because the linguist is work- ing under special conditions with special goals in mind. The character- istics of this corpus are not described here, for they are treated in Chapter 4. All that needs to be said now is that a very convenient method for ob- taining the crucial data is by working with an informant. One does not have to wait for further occurrences of something already heard; he can 1 The linguist also collects information which, although not strictly linguistic in nature, may nevertheless be part of the whole speech act, such as gestures. Here are several different cate-. gories of gestures one should pay attention to: (1) gestures of negation and affirmation; (2) gestures indicating measurement of height, length, depth, width, weight, volume, size, direction, position; (3) gestures used to warn, hush, frighten, beckon; (4) gestures and patterned expres- sions showing emotion (pain, fear, sorrow, surprise, disgust, derision, admiration, smell, and so on); (5) suggestive gestures such as snapping of fingers, winking, staring, whistling; (6) ges- tures used in counting; (7) symbolic gestures and dance postures; (8) gestures and postures used in conversation, whispering, oratory, singing; (9) greetings, handshaking (Conklin 1950).  22 The Language Informant obtain them by asking for them. (One cannot say that he uses an in- formant to get "exactly" what he needs. Linguistic research is explora- tion ; one may know in general what to expect, but he has no assurance that he will get to precisely where he wants to go. There will be many op- portunities for getting lost along the way.) One cannot obtain the corpus from everybody in the community. With- out knowing what diversity there is in the speech under investigation, the researcher may acquire an extremely diversified corpus: women's speech may be different from men's as among the Koasati of Louisiana; dif- ferent dialects may be represented by the population; the speech of the oldest generation may be different from that of young people, as among the Badaga. All of these variations within the speech community must in time be incorporated in the linguist's complete description, but he only complicates his task unnecessarily by starting with them. He will , not, of course, ignore any utterances from any speaker which he can : possibly write down with a reasonable degree of accuracy. They too : become a part of the corpus, carefully labelled. Checking the data for accuracy is also an important function of the informant. There is probably no linguistic researcher alive, nor will there ever be, who can write down at the first hearing everything which is linguistically significant. The more different the language is from our own the more chance there is for error in transcription. The errors can be of several types: not hearing enough phonetic differences, hearing differences in the wrong places, wrongly segmenting the stream of sound into phonological units, and so on. Improvement in transcription is spiral and cumulative: as the researcher makes progress in one area of phonology or grammatical analysis he will see the need for going back to earlier utterances. All of this work requires a patient, trained, and re- liable informant. Further reasons for the informant's importance are seen in comparing analyses based on written texts and those based on informant texts. A grammar based entirely on written texts is dependent on the orthography for its statements about the phonology. In spite of the success that linguists have demonstrated in historical reconstructions, they can only guess at the subphonemic characteristics of a language. On the level of the morphemes, identification is made on the basis of the translation. When one has been confronted with extremely knotty problems in living lan- guages, one can appreciate the difficulty of analyzing only from transla- tions. An analysis of Early Archaic Chinese serves to illustrate this prob- lem. In addition to nouns and verbs there seems to be a class of words which show formal characteristics of both nouns and verbs but are trans- latable into English adjectives, adverbs, or intransitive verbs. Whenever these words turn up in verb position they can be translated either "bring Selection of Informants 23 [the noun] into being" or "treat others or behave oneself as [the noun]" (Stimson 1963). Without an informant to provide more data, we have no way of knowing how to better classify these words. Another difficulty with working exclusively with written texts is that one can never be sure that the relations established between grammatical elements are not simply ad hoc descriptions instead of true patterns. The analyst is at the mercy of his corpus. What this means is that without an informant one cannot test hypotheses (that is, one's tentative analyses) and cannot make statements concerning the productivity of morphemic relations—one cannot predict. 2 Selection of Informants Not every speaker of a language can qualify as a language informant. There are good informants and bad informants, but criteria for consider- ing one good and another bad are difficult to enumerate. One linguist who got on famously with an informant might not understand why a colleague found it impossible to accomplish anything with the same person. How the informant is classified does not depend entirely on the informant him- self but also on the personality of the investigator and on other factors. 3 The nature of the study, the circumstances under which it is done, the informant's health, and innumerable other factors will influence one's judgment about a candidate. Difficulties in agreeing on the qualities of a good informant do not, however, justify ignoring the consideration of who best serves the investigator. Proper deliberation in the selection of an informant is required by the fact that it is so often difficult to dismiss one. Where the informant in some way represents the community to the field worker or where he is in some special relationship to an important personage in the community, the dismissal of an informant will engender ill will. So great might the ani- mosity be that the linguist, not being able to find a replacement, might have to give up his project. For this reason and others one should make a thorough reconnaissance before selecting the informant. This involves soliciting, if possible, the help of the language community in determining the availability and aptitudes 2 This statement is not meant to be categorical. One can, for example, always divide his texts: one part for the analysis and the remaining part for testing the analysis. If the nonanalyzed part is describable in terms of the analysis, then one has succeeded. Obviously, the problem of judging adequacy is a statistical one. 3 Kurath (1939:41) documents how field workers for the Linguistic Atlas of New England tended to show preferences for certain types of informants (one for "the interesting old-fash- ioned local type," another for "the quicker middle-aged type") in spite of the precise instruc- tions they had received concerning the three types of informants they were to seek out.  24 The Language Informant of several candidates. The informant must first of all have enough time to more or less regularly meet with the field worker. He must also be a good speaker of the language. People show a surprising ability the world over in evaluating the verbal gifts of their fellow men. Comments people make about individuals in the community will often contain information which is useful in identifying potential candidates. One ethnographer was alerted to the existence of a truly remarkable informant by the statement that he was always eager to talk at great length about the work that he was doing (Osgood 1940:51). One can also use questions such as the following in assessing the usefulness of a particular candidate: Does he speak your language well? Does he use words from other areas? Is there anything wrong with his pronunciation? Is he a good conversationalist, storyteller, orator, quoter of proverbs? Can he explain things well? Answers must never be taken at face value, for a particular question may have been in- terpreted in a way different from the intended one. For example, an af- firmative answer to the question "Does he know your language well?" might only mean that the man was a good storyteller and would not indicate that having lived away from the community for several years, he used forms which no one else did. Recommendations for an informant should always be carefully scru- tinized. Inquiry might reveal that a recommendation issued from a desire to sponsor a nephew or cousin. If the sponsor had learned the qualities that the investigator was looking for, he could make a very convincing case for his candidate. Bad recommendations can also issue from a misunder- standing of the nature of the project. If the people somehow should imagine that the investigator was interested in flora (perhaps because he simply had asked for the names of ten or fifteen trees and plants in the neighborhood), they might suggest a person whose special craft made him well acquainted with this part of the environment. The foregoing discussion may have given the impression that potential informants were in such abundance that the linguist had only to take his choice of the most suited individual for his work, but getting informants can be a most frustrating experience. Where the communities are small and few in number, one will first have to locate the people. This is not always an easy task. Finding Bushmen, for example, can require weeks of exhausting and patient search over the desolate plains of South Africa. The Indians of the Amazon will disappear with hardly any warning what- soever to be gone for days in search of monkeys, pigs, or bees. Such is the case among the Brazilian Aripaktsâ who frequently depart on three-day hunting or fishing expeditions, leaving in the village the few children under six and the lame adults. Very often one must take what informants he can get. In Brazil this may mean working with captured Indians. When Kenneth L. Pike was trying to analyze the tonal system of Mixtec (a lan- Selection of Informants 25 guage of Mexico), he went several months without a regular and depend- able informant with whom he could do the tedious checking. The informant who did finally agree to work with him was available for only two hours before sunrise, after which the informant would go to his corn- field (E. V. Pike 1956:43). Often the only people who are more or less willing to spend time at the apparently meaningless activity of saying things for the investigator are irresponsible and derelict members of the society. One may have trouble with such people because of their undependability, but one may also have trouble with the rest of the community. Such people, being marginal to the society, are often under suspicion because of the danger they pose to the established order by exposing the full workings of a society (Berreman 1962). Overcoming all these difficulties will hardly ever be easy. One may leave the field without ever having had an informant of whom he could be proud. Tact, patience, ingenuity—and some luck—will all figure in one's complete or partial success. It might be thought that remuneration in the form of money, goods, or services would be a compelling reason for a person to make himself avail- able as informant. Sometimes pay is the only means of getting informants, and when work is regular and intensive, the informant will usually expect it; but often the winning of an informant is achieved by ingenious ways of gaining his confidence or interest. Because Lowie had had the foresight to learn to make a number of cat's cradle designs as part of his preparation for field work (in 1907), he was able to enlist the help of a Blackfoot in- terpreter. It happened in this way: "[I strolled] through the camp, develop- ing different figures with a piece of string while apparently looking neither right nor left. The Blackfoot came out of their tipis, staring at me in rapt attention, and finally themselves summoned Rex [the only Blackfoot bilingual in the camp] to discover the meaning of my strange antics" (1959:17). The accomplishment is all the more remarkable since the In- dians were at that time in a state of passive rebellion against the white authorities. In some places the promise of pay is enough to entice almost anyone to work as informant, but in others hardly any amount of remuneration would strengthen an individual to go against the will of the community. If the researcher has established friendly relations with the community, a person might even be embarrassed in receiving pay publicly. He might want to consider the service an act of friendship. The rate of pay depends most of all on the local conditions—and not on the resources of the investigator. Where a money economy is well established one will probably have to rate his informant as some kind of semi-skilled employee: a schoolteacher, clerk, professional letter-writer.  26 The Language Informant He certainly would not be paid less than a house-servant or grass-cutter. Wherever possible it would be advisable to begin with a lower rate than what one intended on finally giving. There are two reasons for this proce- dure. First, it is easier to dismiss a poorly paid employee than a highly paid one. Second, the increase in salary can serve as an encouragement for good service. Different pay scales can also be attached to different kinds of work. In fact, the field worker should not lead people to think that everything done for him is of equal importance. Survey work would be well-nigh impossible if interviewees demanded the same pay that a regular informant received. More important than the actual salary is the kind of relationship it establishes. The informant might think of his part as an act of friendship or as a disagreeable task which is justified only by the financial return. For his part, the investigator may feel that the arrangement is strictly business and that the pay must be returned in equal value of work. It would therefore be wise for the investigator to give his views on this matter careful scrutiny. One of the undesirable aspects of our money economy in field work is that it allows us to think that the informant is obligated to us for the salary he receives. This puts us in the position of power from which we can make all kinds of demands. The Tightness of our position may appear to us incontestable since the work-pay pattern is so easily taken for granted in our Western society. The informant would see little sense in our view if it were explained to him, and when he does not understand the reasons for the investigator's demanding attitudes and words, he will be hurt, embarrassed, or angered. A good example of the kind of under- standing which the field worker must exhibit is found in the following words by Robert Lowie about his Crow Indian informant: Our relations were on a plane of noblesse oblige. Theoretically, I was paying him four dollars a day [in 1910] for interpreting and transportation, but to insist on the eight hours' daily stint would have proven fatal. If Jim had been on a spree the night before, there was n o use expecting him at nine in the morning; he might turn up at noon or he might not turn u p at all. I gained stature in his eyes by never reproving him for such irregularities and actually lost nothing at all. For when [Jim] Carpenter had once overcome his initial suspicions, he worked for me whether I was about or not [that is, in getting information for Lowie] (Casagrande 1960:430). Financial remuneration may actually be impossible in some situations, not because the informants do not want the money or cannot use it, but because they prefer another way to seal the bargain. In some societies the giving of gifts is the only acceptable behavior, sometimes because a loosely structured relationship permits more freedom in making demands upon the donor: the chicken may not have been big enough, the last gift was given three weeks ago, the service that was rendered was especially diffi- cult, because the informant should have been working in the garden, and Number of Informants 27 so on. Again, the refusal to take pay may be motivated by the desire to put the investigator under the obligation to "act like one of us." Thus, when a sociologist was making a study of attitudes among the Japanese interned in American concentration camps, her informants—and even her teacher of Japanese—refused any pay. The resultant relationship is described in the following way: Informants had indicated that they were willing to accept me as a friend or friendly acquaintance but not as an employer. The value of the role hinged upon the fact that it put me under a n unverbalized obligation. My informants and I knew that I was getting information on the strength of a personal rela- tionship. I had no means of recompensing them except by returning their friendship and accepting the obligations it implied, the most important of which was observing the complicated taboos of the in-group (Wax 1960:169). The principle here is one that field workers in sociology and anthropology have long recognized, that the relationship established in field work can be conceived of as involving a reciprocal exchange between the participat- ing parties: the investigator wants data, but the interviewee has his own set of motives. He is not cooperating "for nothing." If for no other reason, there is some psychological satisfaction such as ego-building or curiosity (Gusfield 1955). By bringing the informant into the project, we satisfy his curiosity and repay him in some small measure. My own Gbeya in- formant was repaid in part by the prestige that he acquired by constant association with me which led eventually to his being able in spite of his blindness to marry a respectable widow with children. The investigator should not overestimate the "rewards" that apparently accrue to the informant. His cooperation must be seen from the viewpoint of his peers. They may consider him a shirker of culturally-approved work (such as going to the corn gardens as every man must do to support his family), a betrayer of village secrets and traditions, a gossip, a toady of the stranger. The suspicion and criticism he must endure must be over- balanced by the rewards he finds in cooperating with the investigator. Tricking informants into rendering free service is an extremely danger- ous undertaking; an investigator who practices duplicity usually deceives nobody but himself. An investigator who is suspected of committing this evil may find himself tricked and obstructed by an irate community. A better motive for reciprocity in field work is the respect the investigator maintains for himself and his endeavors, a commodity whose value greatly exceeds the price which must be paid for it. Number of Informants In a critical appraisal of American field linguistics the statement was made that "the linguist must not restrict himself to only one in-  28 The Language Informant ' formant" (Uhlenbeck 1960:433), because a grammar which ostensibly : comprehends the whole language must necessarily be based on a fully I representative corpus. But the question of how many informants one 1 should use cannot be answered with a single categorical statement. There is indeed a relation between the kind of linguistic investigation being undertaken and the number of informants who are used. The most obvious kinds of study which require many informants are dialectological and sociolinguistic ones. Where language features are to be correlated with age, class, occupation, or any other sociological factor, a scientific study demands careful sampling. 4 But where one is concerned with de- termining the structural outline of a language in its broadest form, there is usually no need for more than one good informant. (The word "good" is crucial in that statement. What this term implies is explained in the next section.) A representative speaker has had built into him all the linguistic rules needed for interacting efficiently with the other members of the speech community. We can say that he has within him a microcosm of the linguistic structure. I The more one expects diversity in the language at some point or another, the more he needs to have a plurality of informants. It would have been foolish for me, for example, to have based my description of Sango on one or even half a dozen speakers of this Central African language. Since it is a lingua franca used by about a million speakers, mostly as a second language (although many children learn it along with the language of their parents), it was necessary to determine to what extent there were varia- tions correlated with geography, language background, amount of ac- culturation to European culture, age, sex, and religion. The final corpus of about 37,000 words represents the speech of 56 informants (see table). There are dangers surely in working with just one informant; one is wisely cautioned against doing it. But the reason is not that an informant caq,so adulterate elicited data as to conceal the structure of the language. Native speakers of languages have, of course, reported instances of such sabotaging. One well-educated Mexican Indian reported that he knew of one informant who had been encouraged by his fellow villagers to dis- tort the language for a period of seven years. After a change of heart, he confessed his action and advised the linguist to destroy everything he had written down. There is evidence that something like this did in fact hap- pen, but we demonstrate extreme naiveté in believing with the informant that he alone could be responsible for a poor analysis of the language. What he had in mind was very likely certain words which had great 4 The relevance of sampling to linguistic field work has generally been ignored in linguistic literature, but the naiveté of linguists is probably not as bad as Pickford (1956) would make it out to be. For a sophisticated use of data obtained in a sociological survey see the work of 4 Labov on the social stratification of English in New York City (1964a, b). Number of Informants 29 SANGO INFORMANTS Sex Age Degree of Religion sophistication (1)-14 p-10, c-0, o - 4 Adult—32 (2)- 5 p - 2, c-0, o - 3 (3)-13 p - 0, c - 2 , o - l l Male—37 (1)- 3 p - 2, c - 1 , o - 0 Youth—5 (2)- 1 p - 0, c-0, o - 1 (3)- 1 p - 1, c-0, o - 0 (1)~ 7 p - 7, c - 0 , o - 0 Adult—15 (2)- 5 p - 2, c - 0 , o - 3 (3)- 3 p - 0, c-0, o - 3 Female—19 (1)- 2 p - 0, c - 0 , o - 2 Youth—4 (2)- 0 p - 0, c-0, o - 0 (3)- 2 p - 2, c-0, o - 0 Degree of sophistication: (1) refers to the lowest degree, (2) to people who have had a minimum of education and/or travel and broadening employment, and (3) to those with a fairly high degree of education and/or travel and employment. Religion: p = Protestant, c = Catholic, o = Other or unknown religion. cultural value to him and his fellow villagers. If a person did try to delib- erately corrupt the data, we should, in any case, find it out soon enough because of the inconsistencies. Even in the best of circumstances one will find the use of several in- formants profitable. It will turn out, for example, that one person is a very good storyteller, another very keen on the use of words, another adept in paraphrasing, and so on. A plurality of informants is especially important in the collection of texts, for one needs to avoid making the error of at- tributing to the whole language characteristics of one informant's style. Even a dullard can be of real assistance, as Sarah Gudschinsky has pointed out. One of her Mazateco informants was ideal for tonal study, because she never tired of repeating words in frames as the complex tone changes were being carried out. The imaginative and gifted person very soon grows weary of such routine, monotonous work. One does not necessarily want several of one's informants at the work session together. The relations between the informants may be such that some will be reluctant to speak in the presence of the others. When William Labov was conducting his study of English on the island of Martha's Vineyard, Massachusetts, he had no difficulty in interviewing individuals. However, when he tried to observe what happened in natural speech among acquaintances, he was less successful; the people did not want to talk casually in a group. Sometimes informants disagree so much among  30 The Language Informant themselves and over inconsequential matters, that the field worker finds it difficult to make progress. On the other hand, the presence of another person at the work session might serve to check the desire of the informant to please the investigator by not correcting wrong or unnatural construc- tions. If the investigator is adept, he can also use groups of people in less structured work sessions. By acting as their host but letting them carry on their own conversations and activities, he can stimulate linguistic dis- cussions among them. In a description of a "typical" day among the Parina of the Philippines the anthropologist Harold C. Conklin describes this technique in the following way (the time is 7:30 in the morning): Pinungu, the old man of Arasa'as and the best archer on Mt. Yagaw, arrives with a gift of 5 fresh eggs and a handful of medicinal jungle plants which he thought we might have missed (we had). Ayakan and two other Parina elders come into my house to join Pinungu in a chew of betel and a round of gossip. It seems " N u n g u " has just come from a nearby settlement where he officiated, as eldest relative, in the trial of his grand-nephew for polygynous marriage sans pangagduway (compensatory payment to first wife). Sitting in a circle on the floor, the others [at least 11 people] listen intently as the old boy relates the whole affair blow by blow. I appear to be attending to some other business at my bamboo desk, but actually I am recording on 4" x 6" slips as much of the sociological information uncovered in their conversation as possible (1960è: 119-120). Qualifications of Informants AGE. Since the investigator needs to select informants who are truly repxesentafive.ofi.thc speech community, lie must find people who are thoroughly experienced in it. A young age generally disqualifies a person from this point of view; a six-year-old obviously has not had as much experience in his language as his senior by 30 years. There are exceptions to this generalization; a very good one is Conklin's remarkable seven- year-old Hanunôo informant.... Maling h a d demonstrated an astonishing maturity of interests and experi- ence, richly illustrating the way in which a Hanunôo child, without formal instruction, acquires an increasingly detailed acquaintance—direct or vicari- ous—with all sectors of the local adult world. Geographically, this is a small universe, limited often to an area within ten kilometers of one's birthplace.... But this small orbit comprehends a comparatively vast realm of knowledge in all provinces of which any member of the society is expected to be at home. In this setting, Maling's parents never thought it particularly precocious that on some occasions she should be as interested in contraceptives as in learning to spin cotton or take care of her younger sister. Nevertheless, I was constantly impressed with her independent thinking and utter frankness which seemed Qualifications of Informants 31 to recognize n o b o u n d a r i e s , except of degree, between child a n d a d u l t k n o w l - edge ( C o n k l i n 1960a: 107-108). In a strongly bilingual community, where one language is more pres- tigious than the language being studied, there will also be the danger of incomplete knowledge on the part of the children. Such is the case among the Coatlân Zapotec where only about 10 percent of the children have any control of Zapotec (Robinson 1963). In this community the men, though knowing Zapotec, carry on their marketing and political and religious activities in Spanish. As for the women, only about one-third of those of child-bearing age can speak the language. Obviously one would have to find his informants here among the adult men. A similar situation obtains among the Chontal (Waterhouse 1949). Children make poor informants also because they very often do not seem to understand what is wanted in elicitation. Their intellectual powers not having fully matured, they have more difficulty in making sense of particular lines of inquiry. They also have a shorter interest span. One other difficulty in working with children is the possibility of insulting their seniors who might consider their own relation to the investigator a pres- tigious one. (If a child is a better informant than some elder, one can get around this problem by using the child to elicit—or pretending to elicit— from the elder, writing down instead what the child repeats.) There are, on the other hand, some advantages in working with children: they very often are patient about repeating things and in giving utterances slowly; in some societies too, children of a certain age do not have many responsi- bilities, so they are available for questioning when the adults are otherwise occupied. Old people present different problems and opportunities. In their favor, obviously, is their experience in the culture. Again, their age often makes them more available as informants than younger people, and they appre- ciate the attention which comes to them in an investigation. On the other hand, advanced age can be responsible for things which make informant work quite difficult: deafness, poor health, tendency to fall asleep, in- ability to concentrate on one problem for a long period of time, lenis articulation, and so on. It is probably also true that elderly informants would not be good for elicitation which did not have immediate cultural relevance. SEX. There is no reason why female informants would not be as good as male informants, 5 but it is probably wise for an investigator to 5 The claim has sometimes been made (Pop 1950,1:725) that women preserve a more archaic usage than men, but this has been disproved. In Atwood's study of Texas speech the women represent a slightly older group, but their use of archaic forms is almost equal to that of men (1962:118).  32 The Language Informant work primarily with informants of his (or her) own sex. In some instances the investigator might be disturbed by the physically determined speech differences. Women, for example, have greater difficulty in adjusting to the pitch levels of men's speech than of women's speech, an important fact when working on tone languages. In those languages which require women to speak to men and women differently, a man might find it dis- concerting, at least in the initial stages of his investigation, to have to make the proper adjustments throughout the work session. There might also be rules of etiquette which would make it difficult to work with a person of the opposite sex. For example, among the Gbeya of the Central African Republic it is improper for a woman to look directly into the face of a man who is not closely related to her. A women investigator working with a Gbeya man would stand the risk of being considered "loose." A man working with a Gbeya woman would have two difficulties: he would be annoyed by not being able to watch her face closely for greater ac- curacy in transcription; she for her part would be embarrassed by the stare. One must also reckon with difficulties accompanying the investigation of a cultural domain which is proper for only one of the sexes. In most preliterate societies there are large portions of the culture which only the men are supposed to know anything about. That the women are not entirely ignorant of the lore might very well be conceded by the men, yet their right to the knowledge would never be. Of course, what constitutes the men's and the women's worlds can hardly be predicted without pre- liminary investigation. C U L T U R A L AND PSYCHOLOGICAL QUALITIES. There are several different attributes of good informants which are not easily discussed under separate headings; they overlap at one point or another. The first group concern the informant's relation to his own culture. A good informant is one who can talk freely and naturally on a wide range of subjects relevant to his own culture. This does not mean that he must be a specialist in every art and skill—in a highly specialized society this in any case would be impossible—but that he should be as informed as one can possibly be in his own society. Ignorance of the basic activities and values of the society will be evidence either of an inferior intelligence or of incomplete enculturation. It is evident that when a person has been away for many of the formative years of his childhood, he will have less knowledge of his culture than his peers who stayed at home. Imperfect knowledge will affect the relations between the investigator and the in- formant by the fact that embarrassment at his own ignorance will pro- gressively undermine the informant's self-confidence; but any discom- fiture on the part of the language assistant is something which the investigator must consciously avoid. Secondly, an imperfect cultural Qualifications of Informants 33 knowledge will lead to an uninteresting content in the corpus. Although information about the people's lives is not the primary goal of a linguistic investigation, no field worker can be excused for ignoring a wealth of ethnographic data which no one might ever again have the opportunity of getting. Besides, grammatical exemplification can just as well be based on interesting texts as it is on uninteresting and uninformative sentences as, "He bought two yellow bananas." One must be careful not to reject people too quickly as candidates for the job of informant. A "bad reputation" might be the clue to a truly independent spirit and creative mind, qualities which eminently qualify a person for being an informant. Such a person Casagrande's Ojibwa informant seemed to have been: Everyone at Court Oreilles seemed to know John Mink. His name was one of the first mentioned to us when we arrived on the reservation in June of 1941. And the more we heard about him the more redoubtable he appeared and the more curious we became. Some called him medicine-man, priest, friend; others called him sorcerer, pagan, scoundrel. But all, Indian and white alike, agreed that his knowledge of the old ways was unsurpassed by any of the 1700 Ojibwa on the reservation (Casagrande 1960:468). A few more words need to be said about the independence of an in- formant. Unless he is free from the pressures of his kin or other members of the community, he will work reluctantly and sporadically. This is a corollary of the statement that the informant most useful to an investiga- tor is one who, like Lowie's Crow interpreter, identifies himself with the investigator and the project. There is again a whole array of qualities which we recognize as dis- tinguishing the superior person. Only a few are mentioned here. Intelli- gence is the hardest to define. It is enough to say that an intelligent informant, by whatever standard we measure this quality, is going to be more helpful than one who is not. Bernard Bloch gives an amusing de- scription of a nearly illiterate butcher in a Massachusetts industrial town whose mental sluggishness impeded the dialect study he was engaged in: "[the informant] persisted through a dozen hours of interviewing in the belief that it was only my pitiful ignorance that led me to ask him such questions as, for instance, 'What do you call the thing you fry eggs in?' Instead of giving me the word he used (frying pan, skillet, spider), he would stare at me in astonishment, then say, as if he could hardly believe his ears: 'Don't you really know the name of that?"' (1935:4fn). In this same place Bloch explicates the need for "sympathetic intelligence" on the part of the informant: "A man with quickness enough to understand our purpose, with alertness enough to co-operate in our work, and with a memory good enough to call up old words and idioms from his child-  34 The Language Informant hood, is naturally preferable to one who hardly knows the words he uses himself and has to be coaxed or dragged along over every step of the way without a notion of the object" (3-4). Locating an intelligent informant may not be an easy undertaking. It is unlikely that all societies will understand intelligence as we do. Any attempt to talk about an "intelligent person" in the Gbeya language would be a failure; I know no way of circumventing the difficulties linguistically. There are practical means, however, of evaluating the in- telligence of informant candidates. If they are literate, one can ask them to submit lists of names of plants, animals, kinship terms, and others, a procedure which was used by Conklin in his work in the Philippines. The importance of a good memory is seen in the work session when the investigator asks the informant to repeat something he has already given, perhaps the day before or just a few minutes ago. If an informant cannot recall exactly what he said, there will be moments of malaise between him and the investigator as the latter, puzzled, says, "No, that wasn't what you gave before," or "Didn't you say it another way before?" Unable to recall having said anything different, the informant may interpret the linguist's statement as a criticism of his lack of competence. Alertness is related to memory. What is needed is a person who pays attention and is not easily distracted either by his environment or his own fleeting thoughts. Nothing can be more exasperating in a work session than repeatedly having to bring the informant back to the task at hand. The alert informant will also be aware of mistakes or contradictions he has made in response to the questions of the investigator. No less important are some social qualities: patience, honesty, depend- ability, cheerfulness. Patience is a virtue which the informant demonstrates in his relations to the investigator when the latter fails to pronounce words correctly, forgets words which have already been obtained, and so on. Impatience on the part of the informant can lead an embarrassed investi- gator to respond in several ways detrimental to the project: by pretending to know something he had really forgotten, thus giving evidence to the informant of his own dishonesty; by going on with the point under inves- tigation in a careless and superficial way; or by becoming angry. Honesty on the part of the informant prevents him from accepting anything from the linguist which is not completely natural in his own language. De- pendability characterizes the informant who fulfills all the obligations agreed upon with the field worker, for example, to work for so many hours a day on such-and-such days, and so on. My own Gbeya informant Gounnté was truly remarkable in this respect. When I was first working on the language, I met with him five days a week after lunch for about one hour or so. In spite of the fact that he was totally blind and had to make Qualifications of Informants 35 his way from his village a couple of miles away, he failed to show up with- out advance notice on only a few occasions, all of which were fully justi- fied. He never arrived at my house late, and if he arrived early (which he often did because no one in the village had a clock), he would patiently wait until my siesta was over. His dependability was not only a credit to him, but it was also a challenge to me, for there were many times when I had other work to do or when I simply did not feel like doing language work. Quite different was the experience of the person trying to analyze the language of the Amuesha Indians of Peru: After hiring someone as an informant with good pay, we have found that he comes one day and misses a week. He may have good intentions when he promises to come each day, but outside circumstances change his mind. H e must get his food supply almost daily. If he sees the river rising, he knows that the catfish will bite. When it drops and the water becomes clear, he takes advantage of spearing fish. If it looks like a fine day, he goes hunting.... If there are several sunny days in succession, he remembers that his new clearing needs to be burned if he is to plant for the next season. If it rains, he stays at home. Some day he runs out of kerosene or matches and comes to buy some from us. Then we take the opportunity and get some language data from him while he sits down to visit a while.... Again we repeat our offer and give him a special gift, hoping this may induce him to come more often" (Fast 1952:80). LANGUAGE. Of all the qualifications for an informant none are so important as the linguistic ones. Anyone who serves in this capacity must be chosen for his ability to provide a corpus which is abundant, accurate, and thoroughly representative. The requirement is that the informant be a native speaker of the lan- guage and dialect being studied. In some circumstances one may need to add that he be monolingual or "monodialectal," for where there is contact between speakers of several languages or dialects, one runs the danger of selecting a person whose speech shows the result of much interference. (This topic is more fully discussed in Chapter 4.) The speech of the informant should also be characterized by good diction. Some speech defects will be identified by the field worker, but unless he has some familiarity with the phonological norm of the language, he may be unable to spot others. For example, he may mistakenly accept a particular kind of postdental voiceless fricative as the normal realization of an / s / phoneme instead of identifying it as the aberrant phone that it is. There is one published grammar based on data obtained from an in- formant with a speech defect. If there are not more, this is not to say that other field workers have not had to redo much of their work after belatedly making the discovery of their informants' idiosyncracies. There must also be precise articulation and voice resonance that is  36 The Language Informant both sharp and pleasing.6 Mumbling, slobbering, and strident talk can make informant work disagreeable and difficult. In a contact situation some of these defects can be overcome by careful attention to the inform- ant's face as he speaks. For tape recording, however, it is extremely im- portant that the informant have the closest approximation to a "radio voice." This should be ascertained by test recording before he is fully committed. Contrary to what one might imagine, certain practices of personal orna- mentation do not seem to affect speech seriously. Thus, in New Guinea men wear elaborate discs and rods in their noses, usually on special occa- sions; in some African tribes—like the Yakoma and Kaba of the Central African Republic—there is the filing of teeth or the removal of two upper incisors; among the Gbeya, large plugs are worn by women in each nos- tril; some people also insert circular objects, reeds, or sticks at the center of the upper or lower lip—or both. However, such deformations as the enlarging of the upper and lower lips (as among the "Ubangis"—really Sara-Madjingay of the Chad) affect speech to a greater extent. An informant should also be talkative (see Chapter 5). Work sessions can be extremely tedious when a taciturn informant gives only what is asked of him. If he cooperates only to the extent that a food-vending ma- chine cooperates—milk or coffee or cocoa, depending on what button is pushed—the analysis will be doomed to a snail's pace. But when the lin- guist's questions stimulate a whole array of associated responses, his corpus will be richer and more varied. My first Gbeya informant was brighter and more sophisticated than Gounnté, but he braked my study by being both dour and linguistically unresponsive. Although I could oftentimes improve his humor by small talk at the beginning of our ses- sions, I cannot recall having taught him to make unsolicited contributions to my study. The opposite kind of informant poses different problems. When my Tamachek informant at Timbuctoo responded very congenially to my requests for names of objects and short sentences with a torrent of words, I was unable to capture the part that I had asked for. A good informant must therefore be able to control his talk, providing first of all what is requested and, if necessary, at a speed that will permit easy tran- scription. Some informants have been so naive about language use that any speed slower than the one they were used to seemed unnatural and ridiculous. Such people are almost impossible to use when the linguist is still un- 6 Mcintosh (1952:90) is right, however, in insisting that certain speech defects, like tooth- lessness and adenoidal trouble, not be overemphasized. Some people with these handicaps may be better informants than those without them. Besides, if the characteristic is a common one in a speech community, the investigator would err in expecting a different standard of articula- tion than that which exists. Qualifications of Informants 37 familiar with the language. One investigator amusingly describes his first elicitation of verbal forms from a Waunana informant in the following words: I was so eager to get ahead in the work that I skipped my siesta and began furiously writing out verb forms in the paradigms: I run, you run, she runs, it runs, we run, you run, they ran. Then followed all the other tense forms, as far as the English grammar permitted. The idea was that in the afternoon we would only have to fill in the equivalent Waunana forms beside the English pattern already worked out. Finally the informant arrived and we started our work. " H o w do you say 'I run' in your language?" The Indian was quiet for a while. First he looked down; then he looked out. Suddenly his face lit up as if struck by a sudden flash of inspiration. He spoke very rapidly. If I had been able to transcribe what he said, it would have spread across the page several times. I gulped and bravely started to write; but after a few syllables, I was already hopelessly bogged down. "How did you say that?" With his repetition I added two more syllables, then bogged down again. When I asked for the third repetition, the informant began to waver and finally to change his story, and so I had to give up entirely. To my half self-justifying and half self-accusing, "But that surely doesn't all mean just 'I run,' " he said, " W h y of course not. It means I was sitting here with you; then I looked out of the door and saw a deer, so I quickly grabbed my spear and now I am running after it." Then, almost philosophically, he added to himself, "Only a fool would run for nothing" (Loewen 1964:189). When an informant can add to all his other virtues the ability to be analytical, he will be able to serve as the field worker's colleague. A poor informant is not able to see the components of long words in an inflected language such as Nyanja (spoken in Malawi and Zambia). He can pick out from a stream of speech the nouns and perhaps the verb phrases, hardly more. He finds it difficult to relate parts of words to linguistic en- vironment or meaning. A good informant is one who can make fine dis- criminations quickly and realistically. This ability is of considerable importance to the linguist for whom meaningful differences are the keys to analysis. A Gbeya who tells me that go and sôko are the "same" is telling the truth to the extent that they both are connectives in identical sentences, but he would have rendered a much greater service if he could have pointed out that the second one implies that the action of the first clause was com- pleted before the second was performed. (Informant help almost never comes in statements as sophisticated as that one, however!) Akin to the informant's ability in distinguishing fine shades of meaning and use in his own language is his intolerance of unrealistic and muddled sentences. To a good informant the language is a model which must be accurately copied by the field worker. He remains frank and critical in his appraisal of the linguist's every attempt to approximate the model. These analytical qualities are not generally found among "naive"  38 The Language Informant speakers of a language. Linguistic sophistication implies the ability to reflect on how one uses language. It takes language out of the realm of the unconscious and automatic and puts it in the realm of the conscious and deliberate. Neither control of a set of verbal arts (for example, poetry, storytelling, proverbs, orations, and eulogies) nor bilingualism are in themselves sufficient to release one from naiveté, although a person so gifted is in some respects more qualified as an informant than one who is unskilled. The use of these arts, however, can be applied to linguistic analysis only with the guidance and training of the field worker. Empirical studies have not yet been made of the analytical skills of language informants. It is therefore impossible to suggest how the best candidates might be found in a society. One opinion is that a more exactly verbalized and deeper knowledge of the structure of one's language is one of the fruits of being fluently literate in one's own or another language. This correlation might hold only when the literates have had formal edu- cation of the Western type. My own experience with informants literate in the Sango language leads me to this view. In the Central African Re- public people learned to read Sango either from some religious literature or from a set of primers but always in informal classes, usually led by an African. The acquisition of the skill of reading is rarely accompanied by any process of intellectualization which would better qualify a person for objective discussion of his language. This is not to say, of course, that literacy is a skill to be ignored (see below). Hitherto the terms "naive" and "unsophisticated" have been used in talking of an informant's lack of skill in dealing objectively with his lan- guage. It should be understood that these terms, and their opposite, "sophisticated," are useful only at the extreme ends of a continuum of characteristics. One informant might be more sophisticated in some re- spects than another informant, and the same informant may be less sophisticated at one point in time than he will be a year later. In fact, by virtue of being involved in an investigation of his language the informant becomes progressively sophisticated. This kind of sophistication is what the linguist seeks to develop; more is said on this topic in the next section. "Sophistication" can also be used in connection with a person's ac- quaintance with the world beyond the boundaries of his own speech com- munity and culture. Thus, it can be said that because a person lived for a year or two in his nation's capital, he is more sophisticated than when he was still in the village 200 miles up-country. It should be obvious, however, that cross-cultural awareness and linguistic objectivity are two very dif- ferent kinds of sophistication, neither one of which implies the other. A culturally sophisticated informant is not necessarily a better inform- ant, not for purely linguistic research at any rate. In fact, an awareness of values different from his own might lead to some skewing of the lexical Qualifications of Informants 39 information he provides. For example, when a Philippine Tausug inform- ant was asked for the word for "house" in a field methods session at an American university, he furnished a word which did not refer to the mud houses of his own community but to the more prestigious city houses. At that moment he had rejected his own crude mud house as something in- appropriate in a highly sophisticated university context. An informant's attitude toward his language is determined by personal and cultural factors. The cultural factors are implicit in the role that lan- guage plays in the speaker's culture. Mcintosh states (1952:91) that the Scots "will not react in a linguistically unsophisticated way, no matter how they are questioned," for they are used to thinking about the rela- tionship between Scots and English and between their own and other forms of Scots. In some societies language is little more than a useful but unconscious tool for communication. In other preliterate societies lan- guage has a very important role. In these correct or elegant speech may be highly prized. Those learning it, whether children or outsiders, will be expected to strive for the culturally-determined norm. Two such societies are the Dogon of Mali (in West Africa) and the Tewa of First Mesa, Hopi, Arizona (Dozier 1951). The Dogon cosmology is an extremely complex one approaching the sophistication of a well-articulated philosophy. In the creation myths language is described as partaking of the original sub- stance from which the universe was made; the use of language is viewed as a continuation of the creative process. With such respect for language the Dogon naturally place great emphasis on proper speech and take great pains to teach their children about words and their use. This section on the qualifications of an informant cannot close without the reminder that there are great differences between verbal skills among all human beings. In our own society we know people whose intellectual brilliance is not matched by oral articulateness. One would expect of the more intelligent person a better control of his native language, but the facts clearly contradict this assumption. Speaking is an oral skill different in many respects from writing. It is a myth which is tenaciously held that there is one model language to which all forms of speech must conform. Where language style is a ticket to upward social mobility, the existence of this myth is not difficult to explain, but it is nonetheless false. In short, since the investigator wants samples of the best possible use to which a language can be put, he will do his best to seek native speakers who can produce this speech. This simply means that he wants texts which demon- strate the highest level of verbal planning. Equally false is the myth which states that because preliterates live in face-to-face, communal societies where pluralism and diversity are at a minimum, there will be little diversity in the skill of speakers of the native language. It might be summarized in the statement, "One informant is as  40 The Language Informant good as another." No person did more to debunk this myth than Leonard Bloomfleld. By carefully describing his Menomini informants he not only sets a standard of linguistic and cultural perceptiveness which field work- ers should strive to achieve but also outlines the variety of verbal skills which they must expect: Red-Cloud-Woman, a woman in the sixties, speaks a beautiful and highly idio- matic Menomini. She knows only a few words of English, but speaks Ojibwa and Potawatomi fluently, and, I believe, a little Winnebago. Linguistically, she would correspond to a highly educated American woman who spoke, say, French and Italian in addition to the very best type of cultivated, idiomatic English. Her husband, Storms-At-It, a shaman, is half Potawatomi, and speaks both languages. Of English he knows not even the cuss-words. In Menomini he often uses unapproved—let us say, ungrammatical—forms which are current among bad speakers; on the other hand, slight provocation sets him off into elevated speech, in which he uses what I shall describe as spelling-pronunci- ations, together with long ritualistic compound words and occasional archa- isms. He corresponds, perhaps, to a minister who does not put on much "dog," speaks very colloquially in ordinary life, but is at the same time very intelligent and able to preach or exhort in the most approved semi-biblical language. Stands-Close, a man in the fifties, speaks only Menomini. His speech, though less supple and perfect than Red-Cloud-Woman's, is well up to stand- ard. It is interlarded with words and constructions that are felt to be archaic, and are doubtless in part really so, for his father was known as an oracle of old traditions. Bird-Hawk, a very old man, who has since died, spoke only Menomini, pos- sibly also a little Ojibwa. As soon as he departed from ordinary conversation, he spoke with bad syntax and meagre, often inept vocabulary, yet with occa- sional archaisms. White-Thunder, a man round [s/c] forty, speaks less English than Menomini, and that is a strong indictment, for his Menomini is atrocious. His vocabulary is small; his inflections are often barbarous; he constructs sentences on a few threadbare models. He may be said to speak no language tolerably. His case is not uncommon among younger men, even when they speak but little English. Perhaps it is due, in some indirect way, to the impact of the conquering lan- guage. Little-Doctor, a half-breed, who died recently in his sixties, spoke English with some Menomini faults, but with a huge vocabulary and a passion for piling up synonyms. In Menomini, too, his vocabulary was vast; often he would explain rare words to his fellow-speakers. In both languages his love of words sometimes upset his syntax, and in both languages he was given to over- emphatic diction, of the type of spelling pronunciation. Little-Jerome, a half-breed, now in the fifties, is a true bilingual. H e speaks both English (the dialectal type of the region) and Menomini with racy idiom, which he does not lose even when translating in either direction. He contrasts strikingly with the men (usually somewhat younger) who speak little English and yet bad Menomini (Bloomfleld 1927:437). It would be valuable to know if in preliterate societies there were any sociological correlates of language proficiency such as one finds in our The Training of the Informant 41 own society. It has been demonstrated, for example, that at the economi- cally lower strata of American and English societies language structure is not as "rich" as it is in the higher strata, even at an early age. This is so not only in the use of more unusual adjectives and adverbs but also in the use of sentences with dependent clauses (Lawton 1963, 1964).7 The Training of the Informant For everyone but the person who has already served as a linguistic informant, the first weeks or months will be a time of training. Even the best of candidates will have failings which must be corrected. The extent to which an informant is trained depends on the informant himself and on the skill of the investigator in teaching him. The general goal of training is to make of the informant an enlightened, interested, and cooperative coworker. The training involves getting him to understand the routine and mechanics of the work sessions: the time and place, the way he should respond when asked a question, how often he should say an utterance before and after it is written down, when addi- tional information is to be given and what kinds are most desired, not ac- cepting ungrammatical or meaningless utterances from the investigator, and looking toward the investigator (if not directly in his face) when giving an utterance. Comprehension of these details does not follow the first explanation of the nature of the linguistic project and the informant's part in it. Even college students who have acted as linguistic informants have required several explanations from different points of view before they began to understand. The ultimate goal is to get the informant to think about language as the investigator does, that is, in terms of broad generalizations based on what is actually said or could be said. Success in this type of training can be achieved only by deliberately and carefully explaining the purpose and techniques of linguistic analysis, not in a few informal "lectures'^ but over a prolonged period of time in connection with specific problems that the investigator is working on. The investigator should not assume that the informant will appreciate being told about the problems and nature of linguistic analysis. Either because he has little capacity for abstract reasoning or because he does not see the relevance of analysis to the speech that everybody, even chil- dren, has obviously grasped, the informant will waste the investigator's 7 However, R. Robinson (1965) concludes after experimentation that the contrast between restricted and elaborated codes may exist only in formal writing and that there may be other reasons, motivational and attitudinal, for explaining why working-class children use the re- stricted code informally.  42 The Language Informant time with his disinterest; he may even resist the investigator's excursions into seemingly irrelevant discussions. If the informant should become only partially sophisticated, he may take more liberty in talking about his language than he is qualified to. It is for this reason that people who have been used as "language teachers" have often made such bad informants; they take their experience as a certificate of competence in all linguistic matters. Whether or not the observations by a partially trained informant are accurate, he may mo- nopolize too much of the work sessions with discourses about matters of interest to himself. The investigator's task in training the informant therefore involves restraining both his imagination and his discussion. The full responsibility for analysis lies on the shoulders of the investi- gator, not those of the informant. Yet there have been grammars produced by well-trained investigators which contained inaccurate descriptions based on the opinions of the informants. One recent grammar of an African language errs in this respect. For example, the word màkànzï ("chief') is described as being "most often considered as a compound," ' that is, md ("mouth") + ko ("death") + nzi (a segment whose meaning is not identified here although in other places it is "memory"). How "mouth + death + X" results in "chief' is not explained (Thomas 1963:46). Only linguistically naive people who had learned the trick but not the science of morphemic segmentation could have produced such an unrealistic compound. They could not possibly know that màkànzï is a borrowing from Sango which in turn borrowed it from a Bantu language. It should be obvious from the foregoing discussion that there can be great variation in the degree to which an informant can be trained. Al- though no informant can remain "naive" after three or more months of working with a linguist, it is wrong to say that all informants become "sophisticated." Much depends on the intelligence of the informant, the nature of the linguistic project, the length of exposure to the habits of the field worker, the amount of training the field worker attempts to give.8 Up to this point we have been talking about the linguistic training of the informant. There also are skills that he can be taught with great profit to the linguistic project. The most important of these is writing. A literate informant becomes extremely valuable, because he is set free as an inde- pendent collector. He may be asked to collect names of objects (plants, insects, animals), further examples of the use of certain words or mor- 8 It is surprising, therefore, that it should be held that American undergraduate students in beginning rhetoric were less sophisticated than field informants who "become something more than naive native speakers of the language being described" (Maclay and Sleator 1960:276). Although some field informants may be more sophisticated linguistically than some university students, the converse must also be true. For other reasons too the statement just quoted is not useful as a generalization about the two groups of people. Gudschinsky (1958) clearly docu- The Training of the Informant 43 phemes, and connected texts. This ancillary collecting adds to the efficiency of the work sessions, since corpus collecting is always extremely time consuming. The immediate problem is training the informant to write legibly and according to the conventions established by the investigator. Unless the field linguist wants to prepare his own writing lessons, he should try to obtain writing manuals in his own language or in the standard language of the country being visited. Even where the Roman alphabet is used, there may be local styles of writing which the informant, as a member of that society, should be taught to master. With a truly illiterate person, one who neither reads nor writes in any language, the field worker must recognize that he is obliged to teach the concept of the visual representation of speech. This is not a highly com- plex task, but it demands some sophistication nonetheless, that is, an awareness of what letters are most easily confused with each other, what letters have the greatest usefulness because of either frequency or occur- rence in productive morphemes, and so on. In short, one needs some familiarization with the art of primer writing (Gudschinsky 1957). Because the introduction of letters in a primer will follow a sequence determined by criteria different from those which determine the sequence in learning to write, the would-be teacher must carefully distinguish between the two operations and the difficulties involved. The investigator must expect inconsistencies and inaccuracies even though the informant is recording his own language. Consider the case of an Osage Indian who was serving as informant for a field methods course: A reading of Bighorse's word lists reveals nothing, unless the reader happens to know both English and Osage. The Bighorse orthography is no more than an interesting failure because it is not systematic—at least not recognizably systematic. Not only was the recording of every form a problem in itself, but, as the many variant spellings in the corpus show, every recurrence of a form presented an entirely new problem which had to be solved without reference to the fact that the form had already been recorded at least once. This is all the more surprising in view of the fact that the informant apparently had a "good ear"... and was striving for phonetic accuracy (Wolff 1958:35; see also Hockett 1948:121). As for other skills—such as typing, manipulating a tape recorder or duplicating machine—nothing needs to be said except that the time in- volved in teaching must always be weighed against its return. Nothing must be taken for granted. Even a recording session has many aspects ments the way in which a native speaker's training affects awareness of linguistic contrasts in his own language. Her conclusion is that the responses of a linguistically naive speaker are different from those of a linguistically sophisticated one.  44 The Language Informant which demand a certain amount of skill or knowledge on the part of the informant. The more that the linguist makes explicit and prepares his in- formant for, the better will be the returns in quality of work accomplished. By now the reader will have seen the justification for spending all these pages discussing the language informant; he is an important member of a linguistic field project. Unfortunately, circumstances may require a field linguist to work with far less than the optimum. The first informant of K. L. Pike's long and productive linguistic career was a Mixtec Indian who was old, blind, toothless, and bedridden with illness!