Pattern Recognition, Data Reduction, Catchwords and Semantic Problems

Transcription

1 Pattern Recognition, Data Reduction, Catchwords and Semantic Problems ANNA BIRGITTA ROOTH University of Uppsala IN a paper entitled On the Difficulty of Transcribing Synchronic Perception into Chronological Verbalization I commented on problems connected with the verbalisation of materials and situations that are not easily verbalised. Here I would like to draw attention to some problems concerning the verbalisation of ideas or notions in narratives. In this case the material is suited for narration - in fact its whole raison d etre is to be told as narratives. It is not a problem that concerns good or bad, or active and passive, tradition-bearers or storytellers only. It also involves human thinking and cognition, and for the scholar the problem of arranging data in a taxonomic way. While we are waiting for physiologists, especially brain physiologists, to give us the ultimate solutions, I would like to draw attention to some problems. The first is the pattern recognition of the human mind, and the second is the data reduction that this mind can undertake. The terms are from computer language. As there are some similarities between a computer and the human brain, I will point out some similar ways of operating. It was after my experience of field-work in Alaska and my way of presenting a couple of story-types that I had to ask myself why am I doing it in this way? After considering it, I reached the conclusion that my mind must have worked according to the processes or principles that are so obvious in computering. It seems that the two processes of pattern recognition and data reduction are what scholars work with consciously or subconsciously. It was no solution to me that I had worked with some sort o f historical-geographical method. We always have to consider time and space - the problem was how scholars operate in analysing the narrative. What ideas, thoughts, and notions in the story do we scholars extract and express verbally as catchwords for later references, mapping etc? Is it the concept of the whole story the type in itself or is it just parts of the story, smaller or larger parts or entities in the story? Whatever it is, we then face a semantic problem. How much of the story and what parts of it are we thinking of? How do we verbally formulate the catchwords for the parts that include ideas and notions in the story? The answers to these questions are important, because the recognition of likeness or similarity depends on these formulations or catchwords, for example the basis for the pattern recognition which in turn is fundamental in the discussion of genetic relationship or polygenetic origin. The latter problem can also be phrased in another way: pattern recognition may differ with different people according to their knowledge, and this in turn may explain why some scholars see a pattern where others do not, and vice versa. Verbal formulation of the catchwords, to denote traits or motifs or whatever we want to call the small entities in the story, involves a semantic problem, namely that of describing the idea or part of development of events in the narrative. How big is that part? Is it a catchword for the whole story, like a title that can be given to a special narrative like Cinderella, The Dragonslayer, Snow White or, for example the two stories that I called Wolverine s Trap and Giants Wrestling, where two important traits were used in the title for my own orientation. Or do the catchwords cover smaller parts within the story, like main motifs or detail motifs, motif complexes, episodes, or parts? How big or how small are the entities that the catchwords are supposed to cover? This is very important for the

2 further discussion of similarities or dissimilarities in the stories, so that we do not compare entities of a different order of magnitude. As I will show, we can play verbally with ideas and notions in the story and describe them in many different ways. Hence it is important to formulate catchwords that are not too wide nor too narrow. The possibility of verbally diminishing or magnifying the ideas or parts of the development of events in the stories probably helps to confuse the discussion. You may not know from the beginning whether you are studying a motif an episode, ora type of story, or whatever that special entity should be called. It is probably more convenient to regard the different traits and small entities that stand out through our pattern recognition as motifs and adjust the terms later, if necessary. To me these terms are operational, and they are not absolute in themselves. I use the same terminology as in the Cinderella Cycle, and the terms in themselves are uninteresting since they are just instruments for my research and analysis of the tale. If, however, we want to analyse the scholar s way of working, how his or her mind operates, we should not study the tale itself but the scholar s relation to his material. Then we work with another dimension, namely the scholar s ability to recognise patterns and his way of formulating catchwords for different striking entities in the story. Whether we call them motifs or motif complexes, episodes, parts, or scenes, is then of secondary importance. Then we are interested in why the scholar s brain emphasised some of the traits, while reducing others that could also have been used. The interaction between pattern recognition and data reduction is of paramount interest, but this is a problem that will probably be solved only with the help of brain physiologists. We will have to use inadequate terms that are perhaps hermeneutic to this way of operating, and we may be using them both consciously and subconsciously. This means that when we do research, we get familiar with material that the brain stores and sorts. Which entities does the brain choose as outstanding, characteristic or important to the story? How big are they, and how clear are these entities or segments in the scholar s mind? As I worked with the Cinderella tales once, I will show how I used the term main m otif in that research. I finally singled out 9 main motifs, but mapped only 6, since I found them sufficient for my purpose. This data reduction was made to save time and work. What I called the main m otif (for which I formulated the catchwords or key-words below as expressions for the skeleton or structure of the story) are necessary for the whole narrative. However, they are not stressed in the story itself. It is the scholar who extracts the underlying idea in the story and gives it verbal expression. Two examples of main motifs will be presented here: 1. in the Cinderella story, 2. in the Wolverine s Trap. Example 1. Main motifs in the Cinderella story. If the Cinderella story was told with the catchwords I used for the 9 main motifs,1 it would run: 9 main motifs 1 = orphan (stepdaughter) 2 = helpful animal 3 = helpful animal is slain and buried 4 = fine dresses 5 = sorting grain etc. 6 = visiting a feast (or a church) 7 = dropped shoe 8 = shoe test 9 = wedding Of these, I only mapped 6 main motifs.2 I could leave out the feast or church visit, since that was the reason for her being dressed in a fine dress and shoes, i.e. motif 4. Furthermore the shoe test (no. 8) and the wedding (no.9) could be left out since they were dependent on motif no.7, i.e. the dropped shoe. 6 main motifs 1 = orphan (stepdaughter) 2 = has a helpful animal 3 = helpful animal is slain 4 = fine dresses 5 = sorting grain 6 = dropped shoe Obviously I used data reduction here for practical and financial reasons. Both lists give a synopsis of the Cinderella story, but it certainly would not be worth while telling in this way, and it is not the way we think of the Cinderella story. 349

3 Now to turn to the Wolverine s trap. For my analysis I had made a synopsis of the story in parts or main motifs. I preferred the neutral term parts, as 1 wanted to avoid a debate on terms like motif, motif complex, episode, type. In the synopsis I structured one story in five parts. Synopsis 1 Part I Hero lets himself get caught in trap or grave Part II Hero is carried to home of ogre (or wolverine) Part III Hero is to be butchered Part IV Hero slays ogre and family - except one member or Part V Hero escapes, ogre pursues him (= Tale Type 313), drinks the river or the lake and bursts, causing origin of fog. Later, pondering upon why I had done it in this way and if I could have done it differently, I realised that I could have structured the story in three, and even two parts, resulting in the following two synopses: Synopsis 2 1 Hero is caught and carried to home of wolverine (or ogre) 2 When hero is to be butchered, he slays wolverine 3a One wolverine cub escapes = Action: that is why there are still wolverines in Alaska or 3b Magic flight (Tale Type 313) + Origin of fog. Synopsis 3 1 Hero fights ogre 2 Aetion results The third type of synopsis is too simplified for pattern recognition of type or episode, or motif complex or motif. It is also too simplified as a structure; it is of no significance for a scholarly approach. Later I asked myself if I could have structured the story in more parts than three and five. It turned out that I could have structured it in e.g. 8, 12, or perhaps even 17 parts. Thus, it is possible to magnify and diminish different entities or segments in the narrative to suit our mental decision.3 It is standard scholarly procedure to find a comprehensive term (main motif = catchword) which covers all the detail motifs, pertaining to or clustering around the main motif. The process of finding entities and catchwords on them is related to that of making systems for catalogues for libraries or archives: to arrange details within groups that can in turn be found in a still larger and more extensive group - like a dendrogram with branches emanating from the trunk. It is not relevant to compare millimeters with miles since they are entities of different magnitudes. The same thing applies to the study of stories. There are traits or entities of different order of magnitude and hence importance, and for a valid analysis only traits or motifs of similar order or importance or magnitude may be compared. That is why a detail or small entity may not be compared or identified with a superior entity by means of a verbal technique, which is very easily done. We must not compare incommensurable entities as Freund did. This is important for the discussion of structure, archetypes, genetic or polygenetic origin, because, just as in mathematics, it is necessary that a similar order of magnitude and the same quantitative entity be used and compared. Consequently we cannot compress a whole story into a few striking catchwords, compare them with a detail motif picked from another type of stoi'y from another country and then compare them with a custom or a rite in a third country, and to prove the point, take a value from say, a proverb from a fourth country. Such a mixture of material from different genres is not based on pattern recognition but rather on flimsy associations. We must know the distribution of a story type and its motifs before we use it for any comparison. Since man is a symbol-using animal, he has the ability to conceptualise and memorise verbal art, such as narratives, and retell them at his pleasure. Have you heard this one? That phrase is a simple reference to the pattern recognition of the audience. The story-teller will remember the plot or the pattern of the story as he understood it or conceptualised it, and so will his audience. The scholar will be subject to the same procedure, but when he hears the same story in different villages he will most probably recognise it as a type. He 350

4 Mental concept of story (W.T.) by story-tellers and audience Mental concept of story (W.T.) by scholar -----collecting and recording data for analysis Pattern recognition of: 1 entities in story 2 entities of different order of magnitude verbal expression; semantic problems; for mapping; for comparison; similarities; genetic or polygenetic origin Figure 1. Pattern recognition o f the story Wolverine s Trap, here called W. T. has then experienced pattern recognition of a type. We recognise patterns daily, and that is one basis for a sort of symbol system. I have tried to exemplify pattern recognition in a simple drawing (see fig. 1). Folklorists usually as I did in the investigation of the Cinderella tales rely upon the pattern recognition of scholars like A. Aarne, S. Thompson, J. Bolte, and K. Polivka. They have obviously done a good job, because we have been able to use their terms and the semantic form in which they expressed the types and the motifs they worked with. We scholars mentally divide folk poetry like stories and tales into entities or characteristics, or group them into major or minor segments. We conceptualise and recognise the tale and the story as one entity containing all the details and characteristics. However, when we want to analyse the story and its contents, we divide it into segments or entities that we make the objects of our investigation. This is usually done chronologically, part by part or segment by segment, since the verbalised narrative always has to be presented chronologically. The synchronic, holistic mental notion of the story which is stored in the brain, once heard in a chronological order as well, cannot be synchronically expressed as it is bound to be chronologically verbalised when retold.4 I have tried to illustrate this in the picture representing our pattern recognition of the whole story and also the way we have to handle it in entities of different size in our analysis. The picture also shows how we use the entities, giving them a verbal expression (catchword) which involves a semantic problem, and how these catchwords are used both as traits ( motifs ) for mapping and as objects for the morphological-analytical method in the process of comparison, which in turn leads to solutions concerning a genetic or polygenetic origin. After doing field work in Alaska, I encountered the problem of presenting and investigating a story-type new to me but based upon my own pattern recognition. I had recorded several versions of the same story (which I called the Wolverine s Trap) in different Indian villages. I will present the actual story in order to demonstrate the problems that are involved in the process of data recording or pattern recognition in the verbal formulation of the conceptualised structure and the catchwords used in the presentation. These problems are of consequence for scholars and their work, as well as for their often fierce debates. To me it is of interest why we can sometimes agree upon or recognise the same pattern and sometimes disagree as to pattern, likeness, relationship and genetic relationship. What one scholar cannot see as a pattern is to another at least 351

5 an echo of the same fable. Philip Freund writes about the Cree story of the virgin Kwaptahaw, who survived the Flood.5 She was carried in the claws of a bird over the water and was dropped on a cliff where she gives birth to twins, a son and a daughter. As Freund puts it: An echo of the Greek fable of Leda and the Swan is heard here, but how can we account for the Greeks and the Crees sharing a story, or even a theme? I cannot accept his statement that the Greeks and the Crees share a story. The Skywoman is quite different from Leda and appears in a different set of motifs and development of events. The woman who has fallen from the sky, or the skywoman, is usually a motif that follows a long introduction about how the hole in the sky was created and how she fell through it. In the east Woodland area, it is usually two swans or geese or other birds that rescue her, carry her on their wings to the sea level and ask the sea animals to help create a world for her in this primeval water. The animals then dive in turn to the bottom of the sea and eventually succeed in bringing up some sand or soil and magically make it grow. The Skywoman is placed on this earth. She is pregnant by the sun or by some other means. In several myths she is pregnant when falling and then gives birth to for example a daughter or a pair of twins. The version that Freund refers to without giving the source, is of course related to this Earth-diver myth, which is often intermingled or mixed with the Flood-story. This latter type can be used as an introduction to explain why and how the earth was flooded. When single motifs are taken out of their context as Freund has done here as well as out of their geographically traditional form, a similarity with the Greek notion of Leda and the Swan is achieved, and this is exactly my point. We must not make this kind of analogy with a small unit by verbally diminishing a whole long epic story until it suits our purpose for comparison. If the Greek tradition had a long narrative story of the same structure and detail motifs, then we may discuss the possible genetic relationship. If I say that 1 have found the same story, that of Wolverine s Trap, in Alaska, Northern Canada and Greenland, I am using my pattern recognition of a long narrative story to solve the question of a genetic relationship versus a polygenetic development. The story is not exactly the same, as we shall see, and instead there are some interesting differences. Part IV, for example, is geographically distributed in western North America and part V in eastern North America and Greenland (see maps 7 and 8 on pages 362-3). It was not until I did fieldwork in Alaska that I had to analyse, use my own ability to recognise patterns and give a special verbal form to the catchwords I decided to use. When you have recorded a couple of stories containing more or less the same traits, as a folklorist you recognise it as a kind of type. In Alaska I recorded eight versions or data of the same type of story, which for my pattern recognition I identified as one type and called it The Giant s Wrestling, using the most important trait as a title for the story. The other type of story, of which I have recorded several versions, I called The Wolverine s Trap. These two types were, to my knowledge, not described in Thompson s types of the North-American Indian Tales, so I had to describe them by the entities that struck me, in order to show what occurred over and over again in the stories, and in an attempt to find proper catchwords for these different motifs and traits or entities. Not until the scholarly analysis did I have to use terms like motif, motif complex, part or episode. I had to try to describe the pattern as I recognised it in the stories. In order to show the different problems connected with data recording, pattern recognition and the verbal expression of the pattern of different entities, I will give my examples here. I will use them in order to demonstrate how it is possible to use words like motifs and types in different ways according to your aim and scope. I will also point out how motifs and types might be used for the mental synopsis of the type of story or of the motifs of which the story consists. These operating terms can vary according to what the author wants to stress and is interested in, in his different investigations. How we verbally formulate the catchwords for the motifs is also of importance for the mapping of those motifs. The reading and analysis of the maps is important for the following problem: what is the distribution of the different motifs? And this, in turn, is of importance for the problem of genetic relationship or polygenetic origin. The verbal formulation of the catchwords for

6 the motifs turns out to be very important. What are we mapping? Is it a synopsis, a verbally expressed form of the head motifs or something else, such as the significant or discriminative motifs? Are the significant motifs based only on their numerical appearances, or are there other qualities that make entities discriminative? The problems of discrimination as well as numerical appearances are dealt with in another context. I now want to draw attention to some difficulties that I encountered when I tried to structure or describe this story, which I recorded in Alaska several times. It was told by different story-tellers from different villages. What happened was the following: I recognised the story to me it was the same story or it was a type of story of which I had heard slightly different presentations or tellings, that is to say what I call data, versions or variants. How do I present a story-type that is not presented by Stith Thompson or Antti Aarne? Like a computer, our brain can store and recognise patterns, and like a computer, it can reduce data considered of minor or no interest. But what is it that makes certain parts of a story interesting while others are not? To me it was evident for the first time that some of my students found it difficult to recognise what I recognised as a pattern, and what I considered unimportant could be important to them and vice versa. We were apparently differently programmed depending upon our different experience, and that made me ponder how different scholars work consciously or subconsciously. The problem of using the right catchwords is the one pointed out above: they must be comprehensive enough for the purpose they must not be too wide or too narrow. If we look at motif G, i.e. the catchword Wolverine stumbles and breaks wind, and motif H, i.e. causing the hero to laugh, motif G was not recognised by my assistant. He corrected me and said no, in this version, Wolverine did not stumble he jumped. It is, however, of no importance which physical movement caused him to break wind. In another situation, for example when a patient describes to the doctor how he happened to break his leg, it might be very important for the doctor to know if the patient jumped or stumbled. If he stumbled, this could be an indication of a disturbance in the patient s health or condition. Because of these difficulties, I tried different verbal possibilities to formulate catchwords in order to show their importance for analyses and theoretical discussions concerning terms like motif and type. After having demonstrated the possibilities of multiplying/diminishing and simplifying/magnifying the parts or segments in a story, I will present the structure of the story in motifs and parts. These terms are operational and relative and mean nothing but entities or segments of different magnitude. The catchwords used are presented here in a list, together with the letters A-X to denote the motifs or segments that I have used in my analysis. The list of motifs is based not only upon my own recorded material but also on related or similar data known in ethnological or anthropological literature. Motifs and catchwords in the 5 parts o f Wolverine s Trap 1. A Hero lets himself get caught in a trap B or in a grave Bl or in a cache C He smears the trap with blood D or puts his coat in the trap 2. E Hero is carried to the home of the ogre or wolverine F Hero is tickled (or licked he pretends to be dead) Fl and keeps his breath G Wolverine stumbles and breaks wind Gl because hero holds on to branches H which makes the hero laugh 3. I Hero is to be butchered K Wolverine must look for a knife L because hero wishes that it will disappear N Children say the victim looked up P Father complains the victim was too heavy 4. Q The hero kills all wolverines except one Ql often the youngest R He tries to burn down the tree S The youngest, or one of them, escapes Si The youngest, holds out her hand against the arrows 353

7 T U The young wolverine defecates/urinates on the hero Aetion: Still wolverine in Alaska etc. 5. V Magic flight (Aa 313) Vl mountain V2 river V3 cheated to drink bursts X Origin of fog6 The motifs presented above, within the five parts into which I structured the story of Wolverine s Trap, are also of interest as to their spatial distribution as shown in the maps. We have recognised the pattern of the story of Wolverine s Trap, and wherever we find two data of this type, we have to conclude that they are genetically related to each other. They imply a spatial tradition, that is to say there has been a transmission in space from one place to another. How can this tradition within the distribution area be explained? Was one form prior to or older than the other? This is not always possible to tell. We can only establish where a subtype has developed, and as such it must be secondary to the type. It is only from the internal evidence based on a comparative morphological-analytical method that we can deduce which forms of the types, and possibly which forms of the motif or rather the motif complex, are the original ones and so prior to the others. By means of this internal evidence we may explain the direction in which the story type has spread, for example from east to west or from west to east within the distribution area. The spatial distribution of the type and its motif is also of great importance when we want to compare it with a study of choice, in other words the acceptance and refusal mechanisms at work when people choose or the mechanism behind studies of acceptance and innovation. Unfortunately most of these studies are based on mercantile products or social group patterns in the mobile culture of today. We see only part of this mechanism of human choice if we exclude all the immaterial, mental or verbal elements like rites, supertitions, beliefs and narratives, proverbs, formulas etc. I have commented on these problems in another study. I have concluded from the motifs on the maps that the type as such belongs to the arctic and subarctic area in Alaska, Canada and Greenland. I have not encountered it south of the Canadian border (sec maps 1-8). Furthermore we can conclude that the Western tradition area, in this case Alaska and western Canada, must present an earlier form of the story than we find in eastern Canada and Greenland. These conclusions may be drawn from the existence of the many detail motifs in the western area (see maps 2, 4, 7).7 The analysis and method used in the research on the two arctic/subarctic types are thus based on three components: 1. Pattern recognition 2. Morphological-analytical method of the recognised type and its motifs and parts 3. The spatial distribution related to the material used. When I started to collate other data of the Giants Wrestling, a rather long and elaborate story for the Eskimos, I noticed one entity or segment in the material returning now and then. At that stage of my work I had recognised it as a discriminative feature or entity, but I had not yet given it a designation in mental nor verbal form. When I started analysing this story-type, I assigned to it letter D, thus denoting that it belonged to the exaggeration motifs where bear and fox were related to the giant s size. The literary function of the entities or motifs that were to be recognised as exaggeration motifs or priamelen was to point out the giant s size in a playful way. He was so big that: A he could put the man in his boot (kamik) or cover him with his boot-lace B he calls whales cod or catfish etc. C he calls moose hare/rabbit D he calls bear fox/lemming Di Giant s lice as big as lemmings or muskrats etc. E Giant s finger huge (almost crushes the boy s head) El Giant s helper will wake him up with a stone when bear comes F Giant takes whales in his hand Fl Giant carries kayaks into his house G Moose hang from his belt like rabbits When I analysed the exaggeration motifs or priamelen in this story, I thought of the feature 354

8 here marked El as a motif or entity among others, explicitly describing the size of the giant in relation to different animals and things as well as to the hero in the story. Later in the research process, in the analysis of the forms, meaning and functions of the exaggeration motifs in the story, it turned out that some of them seemed to belong to the Eskimo tradition of the story. What I had worked with vaguely as a motif or entity among other priamelen could, as I then found out, be structured as a narrative form of its own, often with motif A as an introduction. In the first stage of this analytical process I could not assign a name to this entity, and at that stage it was of no interest to my work. Later on, after sorting, checking and mapping, it turned out that what had been designated by the letter El was a product of data reduction that my mind had worked with at this stage of the analytical process. I asked myself if I could structure the notion behind El in a different way, and it then turned out that I could even structure it into 13 traits or entities from a-m (see scheme of the mental data reduction, stage III). Data reduction o f notion in stages I-III I II Pattern recognition when collating material for comparison Subconsciously compressing several traits (13: a-m) into one motif designated by the letter El in the first stages of the analysis III Later analysis of E l: a. the giant tells the hero b. to rouse him c. by hitting him with a stone on the head d. when a bear approaches e. which is so big that it darkens f. a cleft, passage etc. g. the hero rouses the giant h. who laughs: i. the bear is so small that he calls it a fox j. the giant falls asleep again k. the passage is filled with darkness l. the giant is roused with a stone m. and now kills the huge bear. What at earlier stages I, II I had worked with as a trait or entity among others, that at those stages could be called an exaggeration motif among others, could later properly be called a motif complex or a type of its own. With this presentation I wanted to stress that it is futile to try and show that the terms are fixed and that you can give them a proper definition. The terms might well vary with the analytical process, as I have shown here. And because of the way the brain works in solving problems, we should not over emphasise the words, terms and their definitions. They are just temporary analytical tools that vary as the analysis progresses towards the final conclusions. In this process the terms are of minor importance, as there are, to me, other much more important questions to be solved from an ethnological/folkloristic point of view. Consequently I have tried to avoid words like motifs and types, so as not to be entangled in a fierce debate. To me so many problems are involved that the terms motif, motif complex, episode, and type are just the tip of the iceberg. I have written this in order to elucidate the many problems that arise from taking field-notes8 all the way to the last analytical processes and syntheses and the place of the scientific terms in the process. Many of my learned colleagues have not had the opportunity to do field-work in a primitif society and have not encountered the practical obstacles and modus operandi that are an introductory and highly important part for the verbalising of texts. This then is stage III, at which most scholars begin. I suggest that it is very important for us all to know this pre-text stage if we want to know how we work as scholars, depending on our different experiences, from the field (and the library) to the final product. 355

9 PART 1:1 = A = Hero (Beaver) lets himself get caught in a trap = B = or in the grave o = Bl = or in a cache

10 PART 1:2 A = C = Hero smears the trap with blood = D = or puts his coat in the trap

11 PART 2:1 = E = Hero is carried to the home of the ogre or wolverine = Gi = Hero holds on to branches

12 MAP 4 PART 2:2 = G = Wolverine stumbles and breaks wind A = H = and makes the hero laugh

13 54 I I PART 3 a = I = Hero is to be butchered = K = Wolverine must look for knife (axe) 360

14 PART 3 b = N = Children say the victim looked up = P = Father (wolverine) complains that the hero (victim) is too heavy

15 PART 4 = Hero kills all wolverines (except one). Aetion: Still Wolverines in Alaska etc. 362

16 MAPS PART 5 = V3 = Cheated of drink = X = Origin of fog

17 Notes 1. A. B. Rooth, The Cinderella Cycle (Lund, 1951). 2. ibid., Tradition Areas in Europe, Lokalt och glob alt, II (1969); ibid., ARV, XII (1956). 3. A definition of terms like motif, motif complex, type, is premature at this stage. 4. A. B. Rooth, On the difficulty of transcribing synchronic perception into chronological verbalization (1977). 5. Philip Freund, Myths o f Creation (London, 1964), The motifs presented here have been transferred to maps 1-8 to show their geographical distribution. 7. Seven detail motifs are covered by the symbol for part IV. 8. A. B. Rooth, The Alaska Expedition Myths, Customs and Beliefs among the Athabascan Indians and the Eskimos o f Northern Alaska (Lund, 1971),