187 56 24MB
English Pages 244 [245] Year 2017
TZELTAL N U M E R A L CLASSIFIERS A STUDY I N E T H N O G R A P H I C S E M A N T I C S
JANUA LINGUARUM STUDIA MEMORIAE NICOLAI VAN WIJK DEDICATA
edenda curai
C. H. VAN SCHOONEVELD INDIANA
UNIVERSITY
SERIES P R A C T I C A 70
1968
MOUTON THE H A G U E
• PARIS
TZELTAL NUMERAL CLASSIFIERS A S T U D Y IN E T H N O G R A P H I C
SEMANTICS
by
BRENT B E R L I N UNIVERSITY OF CALIFORNIA, BERKELEY
1968
MOUTON THE HAGUE • PARIS
© Copyright 1968 in The Netherlands. Mouton & Co. N.V., Publishers, The Hague. No part of this book may be translated or reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publishers.
LIBRARY OF CONGRESS CATALOG CARD NUMBER: 68-13345
Printed in The Netherlands by Mouton & Co., Printers, The Hague.
To William E. Bittie
"To the ethnologist, the semantics of the language of the people in whom he is interested is a subject of considerable interest since it presents him with a practically exhaustive classification of the objects in the cultural universe of the speakers." Joseph H. Greenberg (1948:142)
" . . . t o ask what it is that enables a man to tell the difference between A and B involves a different procedure from asking what goes to make up A and B. The difference in procedure is due to the fact that the subject cannot give a precise account of how he goes about making distinctions. But he can tell whether A and B are the same or different. The scientist's job is to analyze the difference and thereby uncover the hidden system of his subjects." Edward Hall (1959:108)
"... and even Stigand, the patriotic archbishop of Canterbury, found it advisable —" "Found what?" said the Duck. "Found it", the Mouse replied rather crossly: "of course you know what 'it' means." "I know what 'it' means well enough, when / find a thing", said the Duck: "it's generally a frog or a worm. The question is, what did the archbishop find?" Lewis Carroll (1960:47)
PREFACE
In the Fall of 1960,1 began descriptive linguistic work on a dialect of Tzeltal spoken in the community of Tenejapa, state of Chiapas, Mexico. This research continued until the following summer under the auspices of the University of Chicago and with the supervison of Norman A. McQuown. 1 The descriptive linguistic background in Tzeltal led to an interest in the semantic implications of a series of linguistic forms termed classifiers with which the present research deals. The months of June to December, 1962 and of January through June, 1964 were spent in the field collecting data related to this problem and, in part, analyzing it under the auspices of the Department of Anthropology at Stanford University. A large portion of the work was generously supported by funds made available from the National Science Foundation. 2 The form of the study as it now stands is a revised version of my Stanford University dissertation, some aspects of the research having appeared elsewhere.3 Funds from the Laboratory of Social Relations at Harvard University have contributed to the preparation of the final manuscript and the comparative appendix of TzeltalTzotzil numeral classifiers.4 In attempting this study in ethnographic semantics I have striven to employ as often as possible linguistic techniques in the analysis of what may be considered an essentially ethnographic problem. As such, I would hope that it might contribute to the growing body of substantive studies in the general area of linguistic anthropology. I have spent a good deal of space in outlining in detail the actual procedures employed at each stage of the analysis. It may appear at times that I have been overly explicit. However, should the possibility of replication be an important consideration in ethnographic semantics, some effort must be made to discuss the pure mechanics of ethnographic elicitation. In conducting and in bringing this work to its present stage I am deeply grateful for the generous support of numerous persons. I would like to thank Floyd Louns1 Funds granted by the National Science Foundation, and National Institute of Mental Health, Norman A. McQuown, Principal Investigator. 2 Grant G22152, A. Kimball Romney, Principal Investigator. 3 See Berlin and Romney (1964). 4 I wish to thank Robert F. Bales, Director of the Laboratory and Evon Z. Vogt for help in obtaining this financial assistance.
10
PREFACE
bury, whose theoretical orientation is evident throughout, and Norman A. McQuown, both of whom have read the manuscript in its final form; A. Kimball Romney, Joseph H. Greenberg and Cornelius van Schooneveld for their helpful comments on an earlier draft as members of my dissertation committee; Roy D'Andrade, Charles O. Frake, Sidney Lamb, and William Sturtevant, who, as participants in the 1963 Merida Conference on transcultural studies in cognition, provided helpful criticisms of the work reported in Berlin and Romney (1964); Duane Metzger and Gerald Williams to whom I owe an obvious methodological debt; Terrence Kaufman whose earlier unpublished treatment of Tzeltal numerals (1961) has been influential in the formal linguistic treatment of numeral classifiers here and whose Aguacatenango materials on classifiers have been included in the comparative appendix; Robert Laughlin and Harvey B. Sarles for permission to utilize their unpublished Tzotzil linguistic data in appendix 4; Christopher Day and Nicholas A. Hopkins for comments and criticism relating to the development of the comparative sketch given in appendix 4; Dell Hymes to whom I owe the use of the term "ethnographic semantics"; and A. Kimball Romney for contributing freely of his time and effort since the beginning of this study in 1962 and for consenting to write the Foreword. For direct help in solving specific problems relating to analysis, implications of the research, form of presentation and general intellectual stimulation, I am very grateful to Paul Kay. My wife, Elois Ann, has read the manuscript in its final form and has contributed greatly to the compilation of appendix 4. Sr. Vicente Kramsky, of San Cristóbal las Casas, has been helpful in producing the photographs used throughout the volume. I cannot, of course, express my debt to /h?alus mentes ton/, informant, teacher and friend, for his patience and genuine concern with the production of this work. To /hpetul peres konte/, my second full-time informant, I am also indebted. Finally, I offer my thanks and dedicate this work to my mentor in anthropology at the University of Oklahoma, William E. Bittle, who, with paradoxical infatuation and skepticism, led me first to the mazeways of language and culture. B.B. San Cristóbal las Casas, Chiapas, México, September, 1965
TABLE OF
CONTENTS
PREFACE
9
FOREWORD by A. Kimball Romney
13
1.
17
INTRODUCTION
1.1. 1.2. 1.3. 1.4. 1.5. 2.
3.
4.
Previous studies of numeral classifiers in the Mayan family Limitations of the description Informants The Tenejapa Tzeltal A general overview of Tenejapa Tzeltal numeral classifiers
. . . .
18 18 19 19 20
METHODS OF DATA COMPILATION
24
2.1. 2.2. 2.3. 2.4.
25 26 28 30
THE
The compilation of the inventory of numeral classifiers Eliciting the denotata of numeral classifiers The elicitation of sub-sets of numeral classifiers The elicitation of sub-sets of "object" classifiers DERIVATION
OF
CRITERIAL
ATTRIBUTES
OF
TZELTAL
NUMERAL
CLASSIFIERS
33
3.1. 3.2. 3.3. 3.4.
33 34 35 40
A note on terminology Three procedural suggestions for the study of criterial attributes . . The production and classification of objects Summary
THE SPECIFICATION OF THE DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
4.1. 4.2. 4.3. 4.4. 4.5. 4.6. 4.7. 4.8. 4.9.
The form of the presentation of the analysis Semantic implications of distributional features The semantic domains of numeral classifiers Semantic domains exhibiting contrastive distribution Semantic domains exhibiting complementary distribution Semantic domains exhibiting mixed distributions Classifiers unassigned to semantic domains Semantic domains encompassed by "action" classifiers Summary
42
42 43 61 62 132 134 162 162 172
12 5.
TABLE OF CONTENTS
DISTINCTIVE FEATURES AND HIERARCHIES: THEIR RELEVANCE TO TZELTAL NUMERAL CLASSIFIERS
173
5.1. 5.2. 5.3. 5.4.
173 173 174 180
"Distinctive" features of classifiers The "hierarchal" ordering of numeral classifiers Objects and their "inherent" and "temporary" states Conclusions
Phonetic key for Tenejapa Tzeltal segmental phonemes . . . . 183 2. An alphabetical listing of Tzeltal numeral classifiers 191 APPENDIX 3. Special problems 215 APPENDIX 4 . A comparative sketch of Tzeltal-Tzotzil numeral classifiers (with notes on Chontal and Yucatec Maya) 218 APPENDIX 1.
APPENDIX
REFERENCES CITED
239
FOREWORD
The mutual relevance of linguistics and anthropology has long been recognized. The convergence of interest may be seen in the continually challenging, and frequently puzzling, area of meaning or semantics. In the past decade psychologists also have joined in the attempt to unravel the various facets of semantics. The philosophers, of course, have long been at the task. As our knowledge and information has accumulated, it has become increasingly apparent that any solutions are going to depend upon a base of careful, detailed enquiries into strategic empirical cases of potential theoretical import. The present investigation is an example of such a study. In broad terms a consideration of numeral classifiers may be viewed as contributing to an understanding of the relation between a noun and its modifier. In Tzeltal, as in many other languages, counting something is associated with classifying or modifying indicators about the thing being counted. The occurrence of numeral classifiers is obligatory in specific numeral phrases in Tzeltal and invariably indicates some semantic features of the noun in the phrase. Numeral classifiers are not universal. Their function as "nominal qualifiers" is, however, universal. Thus many of the specific problems encountered in the present investigation are likely to be general to the area of nominal qualification in every language. Some of the suggested solutions, by the same token, should be generalizeable. Numeral classifiers in Tzeltal are a particularly strategic point of entry into the problem area of nominal qualification for a number of reasons. In the first place it is a simpler, more delimited problem than is the general case. It is simpler especially in respect to context. They occur only between a numeral and a noun (implied when not explicitly stated). Furthermore, they take a specified canonical shape and their derivation is fairly straightforward to specify. For these reasons it has been possible to place precise boundaries on the universe of phenomena under study. Even so, the magnitude of the study is considerable. For example, by presenting informants with the 4,410 phonemically possible cononical forms, some 528 numeral classifiers were isolated. Since most nouns in Tzeltal may occur in a numeral phrase the combinatorial possibilities are still enormous.
14
FOREWORD
Let me now turn to a few of the issues discussed by Berlin that may have rather wide implications for the study of semantics. The first of these is the use of the concept of "semantic domain". Anyone interested in formulating a more precise definition of the concept of semantic domain would do well to analyse Berlin's treatment of the problem. The challenge of discovering the various semantic domains into which the 528 numeral classifiers were categorized presented some novel problems. In the first place, there were no explicit "cover" terms for the various domains. Lack of cover terms is not unique in work of this kind; it is rather the fact of the number of domains that were to be isolated. The major methodological innovation was to use explicit judgments of native informants as a criteria for distinguishing one domain from another. Over a hundred semantic categories were isolated on the basis of judged similarity. One suspects that further advances in the study of semantics will incorporate the use of informant responses beyond the traditional linguistic limitations implied in "allor-none" and "same-or-different" judgments. Berlin has also clarified the well-known notion that utterances occur contrastively in context. Thus, for example, when an individual selects an adjective to modify a noun, the choice involves a selection of context, i.e., he decides to talk about "A" rather than " B " . Similarly, as we have said in another context, "When a Tzeltal speaker is selecting a numeral classifier to characterize the particular shape or state of an object in his physical environment, he first must select a context which allows for some limited range of meanings, e.g., he decides to talk in the context of 'shape' rather than 'color'. This context includes presumably the range of meanings of all other semantically appropriate classifiers which could conceivably be alternatives. For example, a classifier occurring in a context dealing with the enumeration of objects in terms of some components of shape would presumably not be an appropriate alternative for the context of 'kinds of actions of eating'. One may view classifiers as linguistic units that indicate certain semantic components of objects in some specified contexts." A second important issue raised by Berlin has to do with the internal structure of a semantic domain. It is overwhelmingly clear that the majority of the semantic domains with which he deals are neither perfect paradigms nor perfect taxonomies in the technical sense that these words have acquired. The implication is that we need new methods that will handle "loosely structured" domains. Another contribution to the understanding of the internal structure of the semantic domain relates to the distinction between two basic types of internal structure. In the first type, the numeral classifiers are in contrast, both in distribution and in denotative meaning. In the second type, the numeral classifiers are in complementary distribution (or free variation) and are denotatively synonymous. This distinction is a very general one, and will probably be applicable to the relation between a noun and its modifier in all languages. The work as a whole rests on a solid base of empirical data, and anthropologists
FOREWORD
15
and linguists will be eagerly watching for further developments in the area represented by this pioneering effort. Harvard University Cambridge, Massachusetts 1965
A. Kimball Romney
1. I N T R O D U C T I O N
1.1. PREVIOUS STUDIES OF NUMERAL CLASSIFIERS IN THE MAYAN FAMILY
The occurrence of a set of linguistic forms called numeral classifiers in various of the some twenty-five Mayan languages has been reported in grammatical works of the eighteenth century. Some of the first materials available include partial inventories for Yucatec Maya (see Beltran de Santa Maria [1746]). It was primarily from these sources that Tozzer compiled his list of some eighty numeral classifiers in his classic A Maya Grammar (1921). Moisés Romero Castillo has recently discussed modern Yucatec Maya numeral classifiers — which he calls "classificatory morphemes" — and presents a list of thirty-three forms (1961). Romero Castillo notes that many of the classifiers included in Tozzer's report are now in "disuse" in modern Yucatec Maya, although it would be premature to state whether the terms have been lost. Two specific studies of numeral classifiers have also been reported for certain dialects of the Chontal of Tabasco (see Harris [1947], Keller [1955]). The Harris paper lists only thirteen forms but Keller, in an admirable analysis, includes seventyeight forms. The earliest record of numeral classifiers in Tzeltal appears in a dictionary and grammar by Vicente Pineda (1888).1 There have been several more recent descriptive linguistic studies of Tzeltal and of Tzotzil, Tzeltal's closest relative, which include partial inventories of numeral classifiers (see Castro [1955], Kaufman [1961], Sarles [1961] for the Tzeltal materials; Berlin [1964b], Colby [1963], Laughlin [n.d.], Sarles [n.d.] for the Tzotzil materials). All of the published data on classifiers cited above have noted the apparent classificatory features of these linguistic forms. Only the articles by Romero Castillo and Keller offer essentially semantic interpretations. Both of these contributions, however, are not detailed in the specification of the semantic features of individual classifiers 1
There is comparative evidence to indicate that the Tzeltal described by Pineda was closely related, if not the same dialect, to that spoken in Tenejapa at present (Terrence Kaufman, personal communication). The three forms he lists are /tul/, used in the enumeration of humans, /koht/ [Pineda transcribes cot], used in the enumeration of animals and /cehp/ [Pineda chep], used in the enumeration of inanimate objects. The first and second forms may be presumed to have identical semantic ranges in modern Tenejapa Tzeltal, the latter, however, is quite distinct — my informants use /cehp/ for small loads in net bags and for units of one-hundred.
18
INTRODUCTION
nor do they specify the relationships of classifiers to one another. There is also reason to believe that neither the Chontal nor the Yucatec materials approaches the full inventory of numeral classifiers in their respective languages. The present study is, to my knowledge, the first attempt at specifying in varying degrees of completeness the semantic implications of the approximately complete inventory of numeral classifiers for any Mayan language.
1.2. LIMITATIONS O F THE DESCRIPTION
Tzeltal Numeral Classifiers is admittedly descriptive and exploratory. While the general methodology is one borrowed directly from an eclectic appreciation of pretransformational, American descriptive linguistics,2 a good many of the problems encountered required the development, many times by simple trial and error, of special eliciting techniques and procedures. At the time this study was being carried out, some of the procedures had already been discussed in reference to ethnographic method and some of them are employed here (see Conklin [1954], [1962a]; Frake [1961, 1962]; Metzger and Williams [1962a, 1962b, 1963a, 1963b]). I have not, however, systematically conducted the research in terms of any existent or proposed methodological perspective, and this is a weakness of the study. Furthermore, any generalizations from the research, in reference to the content of the materials at least, must take into consideration the fact that only one dialect of Tzeltal is considered. There may be as many as seventeen distinct dialects. I have some indication that numeral classifiers are not used as frequently in some of these Tzeltal dialects, especially those that have come under strong influence from Spanish. A comparative survey of classifiers in several Tzeltal dialects is obviously a next step in the research. (Appendix 4 indicates the importance of future comparative work.) An additional limitation of the study concerns the small number of informants consulted. I have depended almost exclusively on the aid of two native Tzeltal speakers with informal and non-systematic checks of reliability from three additional informants near the end of the study period. While I feel confident of the generality of the findings for these few informants (the production of objects which take particular numeral classifiers were strikingly similar for all informants in many independent trials), a larger sample, including women and children, would have been highly desirable. An additional, but certainly not final, weakness concerns the fact that only a selected number of classifiers have been considered in great detail, those numeral classifiers which refer to certain semantic features of physical objects. A large number 2 In addition to Norman A. McQuown, I have been most strongly influenced in problems of descriptive linguistic methodology by Terrence Kaufman, Nicholas A. Hopkins and Harvey B. Sarles. I am grateful to them all for their continuing interest.
INTRODUCTION
19
of classifiers refer to actions or events but, due to the difficulty of recording the referents of these classifiers for visual examination (possible only with cinema film or a series of still shots), I have included them with only minimal English glosses. While it is nevertheless true that I examined with my informants the actual actions or events referred to by the total inventory of what I have called "action classifiers", no permanent record for examination by the critical reader is included here. Fortunately, this is not the case with the classifiers which refer to objects and which can be more or less easily photographed. It is hoped that these considerations do not detract from the study as an initial step in delineating at least some of the pertinent semantic features of numeral classifiers in this dialect of Tzeltal. 1.3. INFORMANTS
The two principal informants consulted in this study were native speakers of the dialect of Tzeltal spoken in Tenejapa, Chiapas. Both were male and bilingual, Spanish being their second language. My main informant /h?alus mentes ton/, age twenty-eight, was born in the paraje (small, hamlet-like, political unit) of /kul?ak' tik/. My second informant, /hpetul peres konte/, age twenty-five, resides in the paraje of /pokolum/. Both have had little formal education. At the beginning of the research, /hPalus mentes ton/ had developed excellent control in the phonemic transcription of his dialect, having been employed earlier in the descriptive linguistic studies of Tenejapa Tzeltal carried out in 1960-61 (see Berlin [1962a, 1962b, 1963a, 1963b]; Kaufman and Berlin, [1963]). /hpetul peres konte/ developed this skill shortly after he began working late in 1962.
1.4. THE TENEJAPA TZELTAL
In the central highland region of Chiapas some 158,000 Tzeltal and Tzotzil speakers reside in relative isolation from the Mexican mestizo (or Ladino) population. Tzeltal and Tzotzil are bordered on the north and east by two related Mayan languages, Choi and Toholabal; on the west and south by Zoque, a non-Mayan language, and Spanish. About 50,000-60,000 Tzeltal speakers inhabit the eastern slopes of the highlands while some 90,000-100,000 Tzotzil speakers reside in the western area. In fact, Tzeltal and Tzotzil cut this mountainous section of Chiapas almost in half. The Tenejapa Tzeltal number nearly 10,000 and inhabit an oblong shaped municipio (county or township) of the same name located some twenty miles northeast of San Cristobal las Casas. It is a highly conservative area, most of the population speaking no Spanish. (Liberal estimates of bilingualism in the Tzeltal-Tzotzil region indicate that some 30 percent of the population have some "knowledge" of Spanish.) The majority of Tenejapa Tzeltal Indians practice milpa, or swidden, agriculture
20
INTRODUCTION
over some one-hundred square kilometers that cut through several ecological zones. The municipio of Tenejapa is divided into twenty-one political sub-divisions of different sizes called parajes. While some differences in choice of certain lexical items appear to correlate with paraje boundaries, phonologically and grammatically, Tenejapa Tzeltal is fairly homogeneous as a dialect of Tzeltal.
1.5. A GENERAL OVERVIEW OF TENEJAPA TZELTAL NUMERAL CLASSIFIERS
The numeral classifiers of Tenejapa Tzeltal are a series of predominantly monosyllabic, bi-morphemic stems of the canonical shape CV(h)C. They have a fairly restricted distribution, occurring only in quantifying expressions where they appear as second members of endocentric compounds, i.e., specific numeral + numeral classifier ± noun. Classifiers are of particular interest to the linguistic anthropologist in that they indicate, with a high degree of specificity, certain qualitative features of the referents of the nouns with which they occur. The specific numeral preceding the numeral classifier indicates the number of items in the particular state or condition specified by the classifier and/or the number of specified actions performed on that particular item. Indeed, if it were not the long established precedent in linguistic circles to refer to such forms as numeral classifiers, a more functionally appropriate designation would appear to be "nominal qualifiers". Readers familiar with the work of Hoijer (1951), Carroll and Casagrande (1958) and others on Navaho verbal categories will find interesting structural and semantic parallels with Tzeltal classifiers. Like certain Navaho classificatory verb stems, Tzeltal numeral classifiers are obligatory in specific (as opposed to general) numeral phrases. Thus, in those instances of enumeration of objects, events or actions, a Tenejapa Tzeltal speaker must make a decision as to which of a number of alternative classifiers to employ in terms of some qualitative feature of the object, action or event. Or stated another way, numeral classifiers are linguistic markers which indicate certain physical attributes which a set of objects, actions or events may manifest. A major aim of this study is to outline what these attributes are for each classifier in the inventory derived thus far. Morphologically and syntactically, numeral classifiers form a large but finite sub-class of Tzeltal nouns in that they may be inflected by nominal personal affixes. They are derived by an infixed {-h-} morpheme (which occurs between and C 2 ) primarily from transitive verb roots, positional verb roots, bi-valent verb roots and noun roots. A small number of classifiers are derived from intransitive verb roots or are of unknown derivation. In terms of the form classes from which numeral classifiers are derived one may recognize at least the following sub-classes: (1) Numeral classifiers derived from transitive verb roots. Classifiers of this class refer, broadly speaking, to (a) actions of the transitive verbs from which they are
INTRODUCTION
21
derived, (b) sounds emitted as a result of the action of the transitive verbs from which they are derived, (c) certain objects resulting from the action of the transitive verb. Examples are: /haht'/ 3 'action of breaking branches, limbs, etc. off at base' < tv /-hat'/ 'to break off at base' /lohf£/ 'individual blows with the hand' < tv /-lo£/ 'to strike with the hand' /wohj?! 'action of mixing corn dough in water to make corn gruel' < tv j-wof/ 'to make corn gruel' Note that the transitive verb roots of this class may be inflected as simple transitive verb stems. They do not take positional verb root derivational affixes, cf. (2) below. (2) Numeral classifiers derived from positional verb roots.4. Classifiers of this set refer broadly to objects which have taken, or have been placed, in a certain specified position. Examples are: /koht/ /hoht'/ /t'uhd/
'enumeration of animals' < pv /kot*/, e.g., /-kohtan/ 'to place on all fours' 'enumeration of animate objects in squatted position' < pv /hot'*/, e.g., /-hoht'an/ 'to squat (it down)' 'enumeration of objects in vertical position due to inherent properties of the object' < pv /t'uc*/, e.g., /-t'uhSan/ 'to place in vertical position'
Note that the positional verb roots of this class may not be inflected as simple CVC verb stems. (3) Numeral classifiers derived from bi-valent verb roots. Classifiers of this set refer broadly to objects which are the result of the action of the verb and/or the result of placing the object in some position. Examples are: /b'ah^S/
/k'aht/
/k'ol/
3
'very large loads normally carried in net bag' < bv /b'a£/, e.g., /-b'a0/ 'to make up a large load in net bag'; /-b'ah^an/ 'to place large load in net bag (someplace)' 'oblong, non-erect, solid objects, roughly rectangular' < bv/ k'at/, e.g., /-k'at/ 'to form into semi-block shape'; /-k'ahtan/ 'to place semi-block shaped object (someplace)' 'spherical solid objects' < bv k'ol, e.g., /-k'ol/ 'to form into spherical solid object'; /-k'olan/ 'to place spherical solid object (someplace)'
The derivational morpheme -f-h-} which occurs to form the majority of classifiers has two allomorphs, /-h-/ and 0 . /-h-/ is present when C 2 is voicless but not /?/ or /h/. The allomorph 0 occurs when C 2 is voiced or is /?/ or /h/. There are a small number of roots, primarily nouns and transitive verbs which refer to sounds, which function as classifiers with no change in phonemic shape. 4 Positional verb roots, as opposed to transitive verb roots, are recognized formally by two principal derivational features: (a) they do not occur as simple CVC inflected stems and (b) upon derivation with the transitive stem-formative {-an} or the intransitive stem-formative {-ah} they exhibit an infixed {-h-} morpheme. Transitive verb roots, on the other hand, occur most frequently as simple inflected stems and, upon derivation with the above mentioned suffix, do not exhibit infixed {-h-}. The term 'positional verb root' has been used by Slocum (1948), Kaufman (1963) and Berlin (1962a, 1963b) in agreement with certain obvious semantic similarities of glosses referring broadly to position.
22
INTRODUCTION
Note that the CVC roots of this class may be inflected as simple CVC transitive verb stems or may take the positional verb derivational affix {-h} in conjunction with the transitive stem-formative suffix {-an}. (4) Numeral classifiers derived from noun roots. Classifiers of this set, of which there are few, refer to containers and simple objects. Examples are: /b'ohc/ /c'is/ /ji'ib'/
'gourds-full' < n /b'ohc/ 'gourd' 'long, pointed, non-flexible objects' < n/ c'is/ 'spine' 'marks, written letters' < n/ (¿'ib'./ 'written marks, letters'
(5) Numeral classifiers of unknown derivation. Classifiers of this set comprise what Kaufman (1961) has referred to as "quantifying classifiers" or what Silverman (1962) has called "numericals". The class also includes forms which appear to refer to sounds, actions, and in some cases, objects. The numericals of this class form a small sub-set of eight classifiers. Five of the forms are restricted to quantifying expressions with the numeral {hun} 'one'. 5 /h?ahk'/ /hyalel/ /hteb'/ /htab'/ /hpik/
'once' 'all at once' 'a little' 'twenties' '8000s'
The remaining three numericals may occur with other numerals as well. /b'ahk'/ '400s', e.g., /hb'ahk'/ '(one group of) 400'; /caPb'ahk'/ '800 (i.e., two groups of 400)'; /Posb'ahk'/ '1200 (i.e., three groups of 400)' /mil/ 'thousands' ( < Spanish numeral /mil/ 'one thousand'. This is the only classifier which is clearly Spanish in origin), /hmil/ '(one group of) 1000'; /caPmil/ '2000 (i.e., two groups of 1000)'; /Posmil/ '3000 (i.e., three groups of 1000). /cehp/ 'hundreds', e.g., /h5ehp/ '100 (one group of 100); /ca?5ehp/ '200 (i.e., two groups of 100)'; /Poscehp/ '300 (i.e., three groups of 100)' Classifiers of this set which refer to sounds may be illustrated by the following examples: /^'ihp/ /sin/
'duck cries' < ? 'cries of field mice' < ?
As stated above, numeral classifiers occur obligatorily in specific numeral expressions. However, it should be noted that there are two series of quantifying expressions in Tenejapa Tzeltal, a general series and a specific series. Numeral classifiers do not 6
{hun} 'one' has three allomorphs, /h-/, /hu-/ and /hun/. /h-/ is the allomorph which occurs with classifiers indicating one instance or occurrence.
INTRODUCTION
23
occur in the general series. In general numeral expressions, numeral roots are derived by the general numeral derivational suffix {-eb'}. To illustrate: General numeral series
Specific numeral series
/Poseb' fi?/ 'three dogs'
/Poskoht fi?/ 'three (objects of the animal class) dogs' /cankoht f i ? / 'four (objects of the animal class) dogs' /hoPkoht fi?/ 'five (objects of the animal class) dogs'
/caneb' f i ? / 'four dogs' /hoPeb' f i ? / 'five dogs'
My feelings are that the general numeral series in Tenejapa Tzeltal is coming into greater use by bilingual speakers with a knowledge of Spanish, the phrase structure being quite analogous. The exact conditions, however, which predispose a speaker to use specific numeral expressions where a numeral classifier is obligatory and specifies some semantic features of the thing being counted are not as yet determined. It was also noted above that the noun of the specific numeral expression may or may not be present. To illustrate: /caPtul winik/ 'two (persons of the human class) men' /hoPsoht c'ahan/ 'five (pieces of rope-like material in circular coils) cords' The noun may be absent, but the referent of the expression is always clearly understood, either by contextual features or by the highly specific nature of the numeral classifier of the phrase. For example: /caPtul/ 'two (persons of the human class)' /hoPsoht/ 'five (pieces of rope-like material in circular coils)' The actual occurrence of numeral classifiers in recorded texts is relatively low, due primarily to the restriction of their occurrence in numeral phrases. The published inventories of classifiers from other Mayan languages should not be taken, therefore, to represent the total inventories within those languages. In fact, only systematic elicitation for classifiers per se, as seen in Chapter II, can hope to claim partial exhaustiveness.
2. M E T H O D S O F DATA C O M P I L A T I O N
While the methods of ethnographic semantics, at least as undertaken by linguistic anthropologists, vary enormously from investigator to investigator, there are several comparable, if not identical, general procedural steps employed by all of them. The earliest explicit statement of these procedural suggestions, to my knowledge, appeared in Goodenough's (1951) treatment of Trukese kinship terms. These suggestions were amplified and revised in Goodenough (1956) and by Lounsbury (1956) in what by now are the classic articles in the area of "componential analysis" of kinship terminology. Wallace and Atkins (1960) summarize the methodological considerations for kinship analysis in terms of five steps: "(1) the recording of a complete set (or defined sub-set) of the terms of reference or address, using various boundary-setting criteria, such as constant syntactic context, a type of pragmatic situation, or common inclusion within the extension of a cover term for kinsmen'; (2) the definition of these terms in the traditional kin-type notation (i.e., as Fa, FaBr, DaHuBr); (3) the identification, in the principles of grouping of kin-types, of two or more conceptual dimensions each of whose values ('components') is signified (not connoted) by one or more of the terms; (4) the definition of each term, by means of a symbolic notation, as a specific combination, or set of combinations, of the components; (5) a statement of the semantic relationship among the terms and of the structural principles of this terminological system." (1960:60) Lounsbury, in speaking of the total field of descriptive semantic analysis of lexicon, states the general procedures as follows: One moves to undertake "... (a) the compilation of raw lexicographic data on particular denotations, to (b) the assembling of the denotata of each single linguistic form as a semantic class of objects, or designatum, to (c) the discovery — when possible — of the classificatory dimensions imposed upon the field by native linguistic usage, and (d) the specification of the distinctive features defining each of the constituent semantic classes as a kind to (e) an ordering of the semantic units of the various hierarchical levels within the total structure of the system" (1963:574). In general, I have followed these suggestions in the present research, although not always within the identical sequence.
METHODS OF DATA COMPILATION
25
2.1. THE COMPILATION OF THE INVENTORY OF NUMERAL CLASSIFIERS
While there may be general agreement concerning the initial step of compiling the "... raw lexicographic data ..." (Lounsbury 1963:574) or "... the recording of a complete set ... of the terms ..." (Wallace and Atkins 1960:60) or "... determining the universe ..." (Goodenough 1956:198) in any descriptive semantic study of lexicon, there is hardly any agreed upon automatic procedure for doing it. Sturtevant points out the necessity "... for determining in a nonarbitrary manner the boundaries of the major category or classification system being analyzed ..." (1964:104) in order to avoid an incomplete if not erroneous "... description of the internal structuring of the d o m a i n . . . " {Ibid.) However, he notes that the delimitation of selected "domains" remains "... one of the more difficult problem areas of ethnoscience [ethnographic semantics] ..." {Ibid. 105). A good many of the general problems presented in determining domain boundaries, e.g., semantic extension of the ranges of certain terms, were fortunately by-passed in the definition of the larger lexical domain of classifiers. Problems involved in the determination of sub-domains of numeral classifiers within the larger domain are discussed in 2.3. I have been able to determine the set of all potential numeral classifiers by resort to strictly formal distributional criteria. That this was possible depended on several factors: (1) prior knowledge gained through morphological analysis of the highly regular canonical shape of numeral classifiers (which, it will be recalled from section 1.5, is predominantly CV(h)C). (2) knowledge of the limited distributional characteristics of classifiers (e.g., in specific numeral phrases of the construction numeral + numeral classifier ± noun) (3) knowledge of the inventory and distributional restrictions of the segmental phonemes of Tenejapa Tzeltal (which allowed for the automatic generation of all phonemically permissable CV(h)C sequences). 1 This list was large (4410 items) but finite. (4) the discovery of a constant Tenejapa Tzeltal "syntactic context" (Wallace and Atkins 1960:60) or what have been termed eliciting frames in which generated forms may be substituted and evaluated (by native informants) as to their occurrence or non-occurrence as numeral classifiers in this dialect of Tzeltal. Given a generated list of all phonemically permisable CV(h)C canonical forms in Tenejapa Tzeltal, it was necessary to present the list to Tzeltal informants in order to determine which of the 4410 possible forms actually occured as numeral classifiers. This was accomplished by presenting independently to two informants, both of whom were literate, the generated list and the following elicitation frame: 1
Harvey B. Sarles was the first to employ this automatic generated list of forms for the Oxchuc dialect of Tzeltal in constructing a monosyllable dictionary. The use of the device was first suggested, to my knowledge, by McQuown. Cf. Sarles (1961), Mc Quown (1952).
26
METHODS OF DATA COMPILATION
Iya b'al stak' kahtatik te b'itik
-e/
Lit. gloss. 'Can one we count what things ?' Free gloss. 'Is it possible for us to count things (with the form)?' All generated CV(h)C forms were systematically substituted in the above frame (in reduplicated form, to indicate continuous enumeration). To illustrate, one generated CV(h)C form was /koht/. This expression, reduplicated and substituted in the interrogation frame, appeared as: /ya b'al stak' kahtatik te b'itik skohtokohte/ 'Is it possible to count continuously something with the form /koht/?' /koht/, as we have seen in the Introduction, occurs as the numeral classifier for animals and informants' responses to the question were positive. Expressions for which a positive response was received were considered tentatively to function as numeral classifiers in this dialect of Tzeltal. The number of forms isolated by this method was 528. 2.2. ELICITING THE DENOTATA OF NUMERAL CLASSIFIERS
The second step in data compilation in descriptive semantic studies is Lounsbury's (b), "... assembling the denotata of each single linguistic form as a semantic class of objects, or designatum ..." (1963:574). Lounsbury defined the terms denotata and designatium in his first work on Pawnee Kinship (1956). There he referred to "... particular referents of a term as denotata; to the range of meaning, or class of denotata, as the designatum of the term ..." (1956:167-8). To exemplify his usage of this terminology drawn primarily from Charles Morris, (1938, 1946), he stated: "For the Pawnee terms, then, any particular kin type is a denotatum of some term (e.g., MB is a denotatum of tiwatsiriks), while the class of all denotata of that term [which included, for example, in addition to MB, FFF, FFFB, MMMB, etc.] constitutes its designatum" (Ibid.: 168). It is useful to consider the descriptive task involved in the numeral classifier research as comparable. Thus, I have considered the referents of a particular numeral classifier, i.e., the nouns with which they occur, as the denotata of any particular classifier. The class of all denotata (i.e., nouns) I have considered as the designatum of any particular classifier. It is, practically speaking, impossible to feel confident that the inventory of nouns elicited in conjunction with a numeral classifier is "complete". However, as will be seen below in reference to step 4 (the discovery of the criterial attributes of numeral classifiers), deriving the "true" inventory of nouns which potentially occur with any particular classifier is not absolutely crucial. The elicitation of the denotata of classifiers was accomplished in an analogous fashion to the actual elicitalion of the numeral classifiers themselves, i.e., by the use
27
METHODS OF DATA COMPILATION
of Tzeltal interrogation frames. Each of the CV(h)C forms isolated in the first step of the elicitation (see 2.1) were substituted in the following frame: /b'itik hehcuk ya stak' Pahtael te
-e/
Lit. gloss: 'What things perhaps can one counting
?'
Free gloss: 'What things can one count (with the form)?' An example of a completed frame would be as follows: /b'itik hehcuk ya stak' Pahtael te stulutule/ 'What things can one count one by one with the form /tul/?' If the substituted CV(h)C form were a classifier (the frame thus serving as a check on the forms isolated by frame 1), informants were able to list immediately a series of exemplary nominals. As mentioned above, I have no reason to believe that the lists of nouns elicited in this fashion are complete. I do have confidence that those listed occur fairly often with the classifiers under consideration. The actual filing of classifiers and their accompanying nouns was accomplished by entering them on four by six inch paper slips commonly used in linguistic analysis. After running both informants independently for several classifiers and noting the high degree of agreement in nouns listed, I divided the inventory of classifiers, giving each informant half, and requested that they write down all the nouns possible which were appropriate to the classifier on each individual slip. The completed slips were then exchanged, and each was instructed to read the list of the other, adding nouns where he thought one or more to be lacking. Two examples of such file slips are given below. The isolated numeral classifier is given in the upper left hand corner, the nouns which informants list as occurring with it are given in the central columns. Glosses of nouns are included for the convenience of the reader. numeral classifier
nouns
cahp
c'ahan 'sinew' cilc'ahan 'small, fine cord' laso 'rope' c'ahan tak'in 'wire' ?ak' 'vine'
t'uhc
p'in 'pitcher' p'is 'shot glass' limete 'bottle' k"ib 'jar' cumanteP 'tree trunk'
28
METHODS OF DATA COMPILATION 2.3. THE ELICITATION OF SUB-SETS OF NUMERAL CLASSIFIERS
To review to this point I have now succeeded, to some extent, in satisfying the first two steps in the descriptive semantic study of a portion of lexicon. That is to say I have (a) elicited the theoretically complete set of lexical items in the domain and (b) I have determined the partial inventories of nouns which occur with those lexical items, (i.e., I have a tentative set of the denotata for each of the 528 odd numeral classifiers). This brings us to step three of Lounsbury's outline, i.e., "... the discovery — when possible — of the classificatory dimensions imposed upon the field by native linguistic usage ..." (1963:574). The simple, automatic formal linguistic procedures which allowed for the delimitation of the total set of numeral classifiers (cf. 2.1) was unfortunately inapplicable when it became obvious that the larger set was comprised of a series of smaller lexical sets. The observation that there existed smaller sets of classifiers was made, quite by chance, during a regular eliciting session with an informant early in the research. During the elicitation, my informant clearly indicated that some numeral classifiers were more closely "related" semantically than others. Thus, when discussing informally the characteristics of a set of objects in a particular type of "pile" or "heap", the particular "state" of the heap being indicated by the specific classifier under consideration, my informant voluntarily submitted a series of additional numeral classifiers which specified distinctive types of "heaps" and "piles". In discussing a classifier which appeared to specify some features of "shape", additional classifiers could be easily elicted which referred to varying types of "shapes". Before the session was over, I became sufficiently convinced of the existence of semantically structured sub-sets of classifiers that the goal of delimiting them from the total set clearly presented itself as the next step in the research. At least two methods of procedure could have been followed at this point. One obvious step would be to group all classifiers which occurred with one or more identical nouns into tentative lexical sets. The results would have been something similar to a concordance indicating the freedom of occurrence of each classifier and noun in the corpus. [The rationale against this procedure is given below in 2.4], A second method, the one followed and one recently suggested by Goodenough (1964) is to search for "... at least one criterion that will divide the universe into at least two subuniverses" (1964:225). Clues concerning what that criterion might be were given during informal discussion with informants concerning individual numeral classifiers. In the discussion of some particular classifier, informants would often employ the actual physical referent of the noun with which that classifier occurred, manipulating the object into the state appropriate for that classifier. The "meaning" of other classifiers, however, could be illustrated only as some physical action was in progress. That is, enumeration occurred only during the process of the action. (There is some evidence that there are classifiers which refer to both objects and actions, but this is tentative and awaits further corroboration from informants.)
METHODS OF DATA COMPILATION
29
Two large groupings, however, are clearly delimited by the following Tzeltal interrogations: Question 1
/b'itik hehcuk te k'alal ya kiltik ya stak' Pahtael ta b'a^'il sb'il ta Palel/ Lit. gloss: 'What things perhaps when we see it one can counting in truth its name in saying?' Free gloss: 'What things when we see them can be counted by their actual names (i.e., classifiers)?' Question 2
/b'itik hehcuk te ya stak' Pahtael te k'alal ya stak' pasel, haPte ta patil ma sciknah sb'il Pahtael/ Lit. gloss:
'What things perhaps one can counting when one can making, but afterwards no certainly its name counting?' Free gloss: 'Which things can be counted when they are being made, but afterwards their names cannot be counted?'
Of the total inventory of classifiers, approximately one-third of the forms refer solely or primarily to physical objects. It is this portion of classifiers in Tenejapa Tzeltal which I have selected for detailed discussion, and a lesser amount of attention has been devoted to those forms which might be referred to as "action classifiers". Two considerations are foremost in this decision. The first is that the actual observation of physical objects and the consequent inductive procedures utilized in arriving at criterial attributes is considerably more rigorous. That is, once an object or group of objects, has been acted upon or arranged such that it takes a particular classifier, one has a permanent, easily observable record on which to begin analysis. Numerous different objects, which manifest the attributes of the same classifier, can be examined for self-consistency of the supposed attributes and numerous examples prepared by different informants can be compared. Such objects can be photographed or drawn and presented along with the description for the scrutiny of the reader in checking the analysis. On the other hand, "action classifiers" are much more difficult to record in a permanent fashion without the aid of cinema film or the like. Without it, the reader is left entirely with the verbal glosses of the investigator on which to base his judgements of attributes. Likewise, comparisons are difficult to accomplish over numerous informants, since readily checkable, visible records are not available. A second consideration for which I can offer only intuitive support is the feeling that "action classifiers" actually occur with less frequency and have somewhat less semantic import that do classifiers referring solely to objects. The world is filled
30
METHODS OF DATA COMPILATION
with objects in varying states and conditions which are often the focus of enumeration. Actions, it is presumed, are less often counted. While these intuitions could obviously be checked against the data, the inference concerning frequency is supported somewhat by observing the published lists of classifiers that have been abstracted from textual materials. The vast majority of classifiers reported by Tozzer and Romero Castillo for Yucatec refer to physical objects, as do those given by Keller for Chontal of Tabasco. And for Tzeltal proper, from Kaufman's list of forty-eight classifiers in Tenejapa, Aguacatenango and Oxchuc forty, or eighty-five percent, refer to an object or objects.
2.4. THE ELICITATION OF SUB-SETS OF "OBJECT" CLASSIFIERS
If Goodenough's ideas concerning dichotomous segmentation of the universe appear appropriate for the first major cut, continual spliting in terms of some additional criterion failed to be productive. What did appear to be the case was that each major class (both those classifiers which referred to "action" and those which referred to "objects") were comprized of a series of few membered sub-classes, these sub-classes formed in terms of criteria many of which were relevant only to the specific sub-class of classifiers. The delimitation of such sub-classes was accomplished primarily in terms of native speaker reactions, i.e., in terms of sub-groupings made by informants in response to Tenejapa Tzeltal eliciting frames. Initially, my two principal informants were requested to read through the lists of isolated "action" and "object" classifiers and apply the interrogation frame below: /Pay b'al yic'oh shoytak te
-e/
Lit. gloss: 'Does it exist it has its companions the ?' Free gloss: 'Is it the case that (the form under question) is similar to some other forms?' A later equally productive eliciting frame was as follows. /Pay b'al b'i kol pahaluk sok ta Palel te
-e/
Lit. gloss: 'Does it exist what is almost equal with in saying Free gloss: 'Are there things almost equal (to the form in question) when it is said?'
?'
The informant would substitute the particular classifier under consideration in the substitution slot. The systematic application of these eliciting frames with two informants resulted in the isolation of some eighty lexical sets of numeral classifiers, thirty-eight sets of which are "object" classifier sets. As will be seen in Chapter III, the number of clas-
METHODS OF DATA COMPILATION
31
sifiers in any set may range from two to twenty-six. It should be pointed out that not all classifiers were judged by informants to participate as a member of some set. These "single" or "isolated" classifiers have not been examined in detail, but are included, with tentative glosses, in Appendix 2, the inventory of all classifiers 1 have elicited to this point. Before proceeding, I should discuss an alternative method for identifying sub-sets of classifiers which I mentioned briefly in 2.3. This would have been to simply compare the distributional features of all classifiers in reference to all nouns in the corpus. This procedure was not followed in terms of two considerations. The first, although not the more important, concerned the large amounts of labor involved in making a concordance of all nouns and classifiers without resort to mechanical aids (such as computers). Such facilities were unfortunately not available to me at the time the research was carried out. The second and more crucial consideration concerned native speaker reactions to classifiers which, while sharing several nouns in common, were nevertheless deemed to be members of distinct semantic lexical sets. For example, the noun /laso/ [ < Sp. laso] 'rope' occurs with the numeral classifiers /b'ehc'/, /hiht'/, /lihk/, /hil/, /t'im/, /tim/, /k'am/, /soht/, /cahp/, /pihs/, /pehc'/, and /lak'/ as well as numerous others. At the lowest level of taxonomic differentiation, informants consistently treat the first two classifiers, /b'ehc'/ and /hiht'/, as a set separating them from the remaining classifiers which also form a lexical set regardless of all the classifiers'joint occurrence with the same noun, /laso/. As will be seen below, such grouping is made in terms of a set of semantic criteria not apparent from the strict distributional data and made available to the investigator only by native speaker participation in the analysis. It seemed apparent that grouping made in terms of some simple distributional criteria alone might well result in arbitrary and culturally inappropriate lexical sets of numeral classifiers. Interestingly, other investigators in linguistic anthropology have recently voiced a similar concern with native speaker reaction in semantic analysis. Thus Hymes writes: "An unfortunate distrust of the native speaker as anything but a source of sounds has sometimes led to the ignoring of information of an [ethnolinguistic] sort as 'unscientific'. As it happens, reliable data of this sort could be of immense importance to students of psycholinguistics as well as ethnolinguistics" (1960:82-3). An excellent statement has been made by Werner as regards his descriptive semantic work in Navaho. "It is almost a tradition of anthropological linguistics that the informant is never involved in the linguist's work directly. He is expected to respond to all questions exactly, rapidly and with tenacious patience regardless of the linguist's insensitivities toward his native language. ... The antithesis is the complete involvement of the native speaker in the problems of analysis. In this role he ceases to function as informant and contributes more like a consultant of his specific expert knowledge (cf. Voegelin's terminology — personal communication)" (1964:8).
32
METHODS OF DATA COMPILATION
Reliability of sub-groupings of numeral classifiers by native speakers is, of course, of utmost importance. Unfortunately no large scale reliability check as to the stability of these lexical sets across large numbers of informants has been attempted, (as was mentioned in the Introduction). However, some initial work has been done which is highly suggestive and encouraging. Two classes of information are relevant here: (1) When verbally presented a list of classifiers from two presumed sets of classifiers in a random order, informants do not recall classifiers in a random fashion but tend to cluster presumably related classifiers in groupings quite similar to the sets arrived at via elicitation. (2) When presented objects with purposely erroneous classifiers, informants tended to correct such errors by selecting the appropriate classifiers not at random but from predictable semantic groupings arrived at in elicitation. This procedure has been suggested by Frake (1961:118-119; 1962:81-82).
3. THE DERIVATION O F CRITERIAL ATTRIBUTES O F TZELTAL NUMERAL CLASSIFIERS
In 2.2., I discussed the compilation of the denotata of the numeral classifiers, which, it will be remembered, were defined as the nouns with which numeral classifiers occur. The class of denotata for any particular numeral classifier, or the total inventory of nouns with which such a classifier occurs, was defined as the designatum of the term. Lounsbury, again following the usage of Morris, defines yet another term that is critical to descriptive semantic analysis, i.e., significatum. The significatum of any term is considered to be"... the distinctive features of [the] class [i.e., the designatum]" (Lounsbury 1956:168). In another place he defines the significatum of a term as "... the necessary and sufficient conditions for membership in it ..." (1964:1074). It is the discovery of the significata of numeral classifiers that the third step of the semantic analysis is aimed.
3.1. A NOTE ON TERMINOLOGY
Before discussing the actual procedures employed in the derivation of the criterial attributes of classifiers, a point must be made concerning terminology. In 2.3, I referred to the classes of numeral classifiers as lexical sets, after Conklin (1962:124). Recently, there has been a tendency to refer to such sets of linguistic forms as contrast sets. I have found it somewhat confusing to employ this terminology due to certain distributional characteristics of classifiers in a limited (to be sure) number of such sets. The notion of contrast set has not been rigorously defined. However, in an explicit discussion of the concept (contrasting it with that of paradigm, a notion only recently formalized by Kay [1966]), Sturtevant states that a contrast set is a complete set of terms which "... occur in the same culturally relevant environment (setting, context, substitution frame, surroundings, situation, etc.)" (Sturtevant 1964:107). I have reason to believe that the lexical sets of numeral classifiers which I have isolated fulfill the criteria of being termed contrast sets, (i.e., cultural relevance and exhaustiveness). The linguistic forms in each lexical set do co-occur as "... culturally appropriate responses which are distinctive alternatives in the same kind of situations..." (Frake 1962:62). However, as will be seen below, it may be the case that a sets of linguistic forms function as alternatives (i.e., occur in some identical cultural context)
34
THE DERIVATION OF CRITERIAL ATTRIBUTES
and do not occur in some identical linguistic context, such as a common substitution frame. Examples of such cases are sets of classifiers complementarily distributed in reference to the nouns with which they occur, (or, as English herd, covey, and school). To allow for this distributional feature among semantically related, but linguistically n o n - c o n t r a s t i n g f o r m s a n d t o r e t a i n t h e strict
linguistic
usage of the term contrast,
I
have chosen to refer to these lexical sets of classifiers as semantic domains.
3.2. THREE PROCEDURAL SUGGESTIONS FOR THE STUDY OF CRITERIAL ATTRIBUTES
Charles O. Frake has pointed out three distinct procedures for the derivation of the criterial attributes of a conceptual category, or, in his terminology, for "... determining the rules which govern the assigning of one name rather than another in a particular diagnostic situation" (Frake 1961:123). These are (a) analytic (b) perceptual and (c) explicit. The analytic method is the one analogous to that utilized in formal linguistic and kinship analysis whereby some independent etic grid is employed as a meta-language by which the investigator may derive and display the analytic meaning of a concept. The familiar phone types of linguistics and the kin types of kinship analysis are well known and do not require discussion here. The second procedure, the perceptual, is one whereby the investigator attempts to discover the actual "... stimulus discriminations made by informants when categorizing" (Ibid.: 123). The problems involved here are great and have in recent years been the concern of many social psychologists in refuting much of the earlier linguistic work done on "world view" (see Whorf 1956, Hoijer 1951 and the criticisms by Lenneberg 1953, Brown and Lenneberg 1954, Carrol 1963). Recently, more sophisticated ethnographic methods have allowed for a more formal account of the data dealing with exotic classifications which by-pass somewhat the problems of translation parallels. The degree of fit between these formal accounts and so-called psychological reality, (i.e., the stimulus perceptions of one's informants) is discussed below. The third procedure mentioned by Frake is the explicit verbalization by informants of criterial features. Such procedures yield "... the culture's explicit definitions or criteria of categories ..." (Frake 1961:123, cf. also Bruner et al., 1956:30). It is obvious that the application of one, or a combination, of these three above procedures depends almost entirely on the nature of the categories under investigation. In the present research it appeared impossible to utilize an analytic approach analogous to the study of phonology or kinship. It is even now doubtful that kinship can be studied with the present etic-grid of descriptive kin-types alone. As Hymes has recently noted: "It may sometimes be assumed that although languages segment experience differently, what they segment is the same, as if it were a matter of different jigsaw puzzles fashioned
35
THE DERIVATION OF CRITERIAL ATTRIBUTES
from the same painting. But recent work shows that structural analysis of meaning must first demonstrate that a domain is a domain for speakers of the language in question. What the domain includes, what it excludes, what features define it and its elements, cannot be prescribed in advance, even for kinship (cf. Connant 1961) or color terms (Conklin 1955)" (Hymes 1962:19).
Even a cursory examination of numeral classifiers clearly indicates that no universal etic grid exists such that one may discuss the large variety of semantic domains encompassed by them in terms of a few, limited dimensions. Weinreich notes that "In ordinary language only some areas are marked by a degree of 'terminologization' ... in the highly patterned or 'terminologized' domains of vocabulary, such as kinship or color, distinguishing components recur in numerous sets of signs, [but] the bulk of the vocabulary is of course more loosely structured a n d is full of components
unique to single pairs, or small numbers ofpairs
of
synonyms."
(1963:149) [emphasis mine]. Likewise, the explicit verbal formulation of criterial attributes by informants proved to be equally inappropriate. The "objects" or "events" in the world of Tenejapa Tzeltal informants which manifest the criterial attributes of the numeral classifier with which they occur appear to be such that "... category discriminations and decisions do not require verbal description" (Frake 1961:123). In fact, when I pushed well-trained informants to give verbal descrip:ions as to why such a referrent of a noun was classifier "A" and not classifier "B", I received highly variable responses with low information content. On the contrary, informants, apparently frustrated by failure at verbal explanation, would readily bsgin manipulation of exemplary objects such that they could be classified by the appropriate classifiers (e.g., taking pieces of clay-like 'corn dough' /ma^'/, and forming it into shapes with differences in relative size and shape, to take an example from one semantic domain of classifiers). With the exception of those few classifiers whose occurrence with nouns makes their meaning fairly clear by simple inspection (e.g., /koht/, which occurs only with nouns indicating individual animals), every semantically meaningful explanation I have received has been by way of demonstration on the part of my informants. These demonstrations of the distinctions that hold between classifiers in a set were accomplished primarily, therefore, by examining the actual physical objects or events which were characterized by specific classifiers. It is from the observation of objects in states appropriate to their classification by particular numeral classifiers that I have made my tentative, initial statement of those criterial features of discrimination which define the structure of semantic domains. 3.3. THE PRODUCTION AND CLASSIFICATION OF OBJECTS
As stated above, the major focus of this study is a description of classifiers which refer primarily to specific features of physical objects, as opposed to those classifiers which
36
THE DERIVATION OF CRITERIAL ATTRIBUTES
refer principally to actions and events. The actual production of objects which are characterized by various classifiers which refer to these objects was accomplished by the application of Tenejapa Tzeltal interrogation frames systematically applied to each classifier within each semantic domain. I have found, with the aid of two informants, two highly productive frames which allow one to produce the physical denotata of nouns which occur with classifiers. They are as follows : /b'i yilel {specific numeral phrase) (noun)/ Lit. gloss:
'How does it look
?'
Free gloss : 'How do (some number of objects characterized by a particular numeral classifier) look?' The second frame, is produced by (a) inflecting a transitive root or derived positional verb root by thè transitive imperative mode suffix {-a}, followed by (b) a particular specific numeral phrase whose numeral classifier is a derivative of the above verb form, followed by (c) a noun selected from the list of nouns cited as occurring with that classifier. This is best illustrated by an actual example. Selecting a classifier /b'uhs/ 'small piles of individuated objects with maximal horizontal extention' from one semantic domain, one sets out to determine the form-class from which it is derived, i.e., whether the root is a transitive verb, positional verb, bi-valent verb, noun, or none of the preceeding. For transitive verb roots, this is easily established by inflecting the root with tense and person morphemes, e.g., /ya sb'us/ (ya, present tense particle + s-, '3rd person subject) Should the informant reject such a construction (which actually occurred with this example) the root is then derived as a positional verb root by the positional verb root stem formative {-h-} and by the transitive verb derivational suffix {-an}., e.g., I ya sb'uhsan/ 'he makes it into the form, state, etc., indicated by the transitive verb stem /-b'uhsan/' As was pointed out in Chapter I, rarely does one find that the root cannot be inflected in either of the above forms, as the vast majority of all classifiers are derived from either transitive verb, positional verb, or bi-valent verb roots. To continue with the example, the derived positional verb root /b'us/ is then inflected by the transitive imperative mode suffix {-a} as was indicated in step (a) above, e.g., /b'uhsana/
37
THE DERIVATION OF CRITERIAL ATTRIBUTES
This form is followed by the classifier /b'uhs/ in a specific numeral phrase as indicated in step (b), e.g., /b'uhsana £a?b'uhs/ Finally, one selects a noun from the list of nouns occurring with /b'uhs/ to complete the frame. An example of the completed interrogation is as follows: /b'uhsana 5a?b'uhs ton/ "perform the action indicated by the transitive verb stem /-b'uhsan/ on two (groups of) objects in the state specified by the classifier /b'uhs/, the object(s) being stone(s)" One or both of the above interrogations were employed for each classifier and for a representative number of nouns occurring with each classifier. The nouns chosen to substitute in these frames had denotata that were fairly easily obtainable (e.g., corn, beans, stones, rope, mud, corn dough, pieces of wood, etc.). I attempted to obtain the denotata of the nouns in the set and to have them before the informant during the application of the questions. In several cases, i.e., those cases involving denotata of nouns which referred to animate objects, it was possible, with a little training, to employ simulated animate objects, such as flashlight batteries, bottles, drawings, etc., during elicitation. The objects could then be manipulated or arranged into appropriate states or positions for classification. Such simulation was made clear to informants by the frame: /ya stahinotik sokin li? ini/ Lit. gloss: 'We are pretending with right here' Free gloss: 'Let us pretend that (substituted noun) is right here' The particular noun under consideration is substituted in the substitution slot giving the completed frame as exemplified: /ya stahinotik sok in winiketik li ?ini/ 'Let us pretend that these are men here' nouns
/laso/ /5'ilé'ahan/ /?ak'/ /?ak/ /c'ahan/
numeral classifiers
'rope' 'cord' 'vine' 'grass' 'belt'
/b'ehö'/
/hiht'/
X
X
X
X
X
X
X
X
X
X
etc. Figure 1 — Distribution matrix: semantic domain 1
38
THE DERIVATION OF CRITERIA!, ATTRIBUTES
Before the total number of numeral classifier semantic domains are presented, it might be helpful to present one simple semantic domain as an illustration of the application and results of these procedures. For one set of classifiers, the distribution of nouns vis a vis classifiers appeared as indicated in Figure 1. Glosses are provided for the nouns of the set. I procured the denotata of as many of the nouns on the list as possible, (i.e. pieces of rope, cord, vine, grass, etc.). Those that I could not have immediately available were supplied by my informants at the time of elicitation. For each classifier and noun (i.e., those which occurred together as read directly from the distribution matrix), I applied the frame:
b'i yilel {specific numeral phrase) {noun) such that the following series of questions resulted: /b'i yilel canb'ehc' laso/ 'How does (a piece of) rope in /b'i yilel canb'ehc' cilc'ahan/
the state of four /b'ehc'/ look?' 'How does (a piece of) cord in
/b'i yilel Canb'ehc' ?ak'/
the state of four /b'ehc'/ look' 'How does (a piece of) vine in
/b'i yilel 5anb'ehc' ?ak/
the state of four /b'ehS'/ look?' 'How does (a piece of) grass in the state of four /b'ehc'/ look?' etc.
The same procedure was followed for the classifier /hiht'/.
/ b'i yilel canhiht' laso/
'How does (a piece of) rope in the state of four /hiht'/ look?' 'How does (a piece of) cord in the state of four /hiht'/ look?' 'How does (a piece of) vine in the state of four /hiht'/ look?' 'How does (a piece of) grass in the state of four /hiht'/ look?'
/b'i yilel canhiht' óilc'ahan / /b'i yilel canhiht' ?ak'/ /b'i yilel canhiht' ?ak/ etc.
Experience with the production of objects indicated that it was convenient to employ numerals greater than one in enumeration (e.g., /5a?/ 'two', /?os/ 'three', /can/ 'four', etc.). As each object was produced, it was placed aside for further reference. In the example just given, one notes a series of contrasting forms all of which share a
THE DERIVATION OF CRITERIAL ATTRIBUTES
Domain meaning: semantic domain 1 'slender-flexible objects in various wrapped conditions around non-flexible objects'
Criterial attributes of numeral classifiers : Semantic domain 1
Plate I - /b'ehc'/ 'individual wraps of slender-flexible objects in sequential spiral around some long-non-flexible objects, as a piece of wood'. Included in photo: /lahunb'ehc' laso/ 'laso in the state of ten sequential wraps around long non-flexible object'
Plate II - /hiht'/ 'individual wraps of slender-flexible objects in sequential lash-loops around two pieces of long-non-flexible objects at 90° angles to one another, as in fence making. Included in photo: /hoPhiht' laso/ 'laso in five lash loops around two pieces of long non-flexible objects' [noted to the left of photo, the rope in state of /Posb'ehc'/ 'three continuous wraps']
40
THE DERIVATION OF CRITERIAL ATTRIBUTES
common feature of meaning, uniting the set as semantic domain. I infer this common feature of meaning, from observation of objects, to be 'slender-flexible objects in various wrapped conditions around non-flexible objects.' (A similar gloss was used by Maclay 1958:221). "The features of difference among the forms" are indicated by the contrasting classifiers, each denoting a distinctive "wrapped state" of the slenderflexible object. The domain meaning (cf. Lounsbury's root meaning of a paradigm 1964:1073), and the distinctive features of each classifier within this domain are seen in Plates I and II. I have included one object only in the photos in a state characterized by the two classifiers, i.e., /laso/ 'rope'. In numerous of the remaining sets, a number of representative objects are included for visual comparison.
3.4. SUMMARY
In Chapter III, I have attempted to point out the procedures utilized in the discovery of criterial attributes of Tenejapa Tzeltal numeral classifiers. I first reviewed three procedures suggested by Frake (1961): the analytic (i.e., one where-by some independent etic grid is utilized as a meta-language for discussing, analyzing, and displaying the data), the explicit (i.e., one whereby one attempts to elicit native linguistic criteria for categories), and the perceptual (i.e., one whereby one attempts to determine the actual physical stimuli focused upon as criterial by one's informants). It was shown that neither analytic nor explicit procedures were satisfactory in the analysis of numeral classifiers, due to (a) the lack of an independent etic grid that would cover the total semantic space encompassed by classifiers and (b) the lack of explicit verbalization in Tenejapa Tzeltal of the criterial attributes of classifiers. The method followed here is primarily a perceptual one, to use Frake's terminology, whereby the actual physical objects, i.e., the denotata of nouns, are produced such that they manifest the criterial attributes of each classifier considered. Finally, I have shown the actual procedures utilized in the production of physical objects for observation by the use of eliciting frames. By taking an actual semantic domain, I demonstrated how one generates these objects, as it were, by employing interrogation frames while reviewing the distribution matrix of nouns and classifiers in any particular set. The presumed criterial attributes of the classifiers in this set were then presented, depending greatly on the use of photographs as visual aids to the reader. The conjunction of verbal glosses and visual stimuli should enable one to make inductive observations as to what features are criterial. One tests the adequacy of one's hypotheses as to the criteriality of supposed attributes by the degree to which one is able to make culturally appropriate verbal and non-verbal responses to newly encountered materials. Or, stated in other terms, one's analysis is prescriptive (Teeter 1963:342) to the extent that it allows a naive observer to perform correctly (culturally speaking) in all contexts circumscribed by the description. While no
THE DERIVATION OF CRITERIAL ATTRIBUTES
41
systematic validation of the description by naive observers has yet taken place, the materials to be presented in Chapter IV are now amenable to just such validation procedures.
4. T H E S P E C I F I C A T I O N O F T H E D I S T I N C T I V E F E A T U R E S OF NUMERAL CLASSIFIERS
The aim of this chapter is to specify the distinctive features of a limited number of numeral classifiers in Tenejapa Tzeltal which refer to observable objects and which have been grouped by my informants into a limited number of semantic domains. I will also discuss briefly those semantic domains comprised of'action' classifiers. The specification of the distinctive features of any one classifier within a semantic domain requires, as well, a specification of the relationship of that classifier to all other numeral classifiers within that domain (cf. Frake 1962). My discussion of these relationships is likewise based on distributional considerations. I hope to show in the following sections that differences in the distributional characteristics of classifiers within sets have, as well, important descriptive semantic implications.
4.1. THE FORM OF THE PRESENTATION OF THE ANALYSIS
Much has been said recently concerning the claims made in ethnographic semantic studies as to the "psychological reality" of the resulting analyses (cf. Wallace and Atkins 1960; Wallace 1962; Burling 1964; Hymes [in reply to Burling] 1964; Frake 1962; 1964 [the latter in reply to Burling]; Romney and D'Andrade 1964). I have no intention nor desire to enter into this discussion which, in some respects, has taken on the aura of what a close colleague of mine calls "knit-picking". It is important, however, to state at the beginning of this section what the following presentation claims and does not claim to represent. Firstly, given the fact that the study is essentially descriptive, I make no claims whatsoever as to the "cognitive saliency" of my analyses for my Tenejapa Tzeltal informants. Such a claim could be made only if I were to resort to systematic psychological testing of the sort begun by Brown (1956, 1957), Brown and Lenneberg (1954), Lenneberg and Roberts (1956), Carroll and Casagrande (1958), and recently conducted for kinship by Romney and D'Andrade (1964). The claim that I do make is similar, although much more modest, to that of Lounsbury's concern with a "formal account". Sturtevant characterizes Lounsbury's thoughts as presenting analyses which attempt to discover "... the culturally significant sets and their included units ... [and] to determine the most economical componential
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
43
[descriptive semantic] analysis which will define (or 'generate') their paradigmatic [-like] relationship — we are concerned only with predictability, economy and inclusiveness, not ... with cognitive saliency" (1964:109). It is questionable whether the analyses of the delimited semantic domains of numeral classifiers fulfill these criteria, although I have made every attempt that they should. When future attempts are made at discussing the important question of psychological saliency I am convinced numerous revisions will be entailed. However, the analyses are now a first approximation, offered strictly as a first approximation, and nothing more. A non-arbitrary presentation of the data in the present analysis that aims at predictability, comprehensibility and simplicity was difficult to arrive at. The form that it has taken, however, presents as much of the data as possible alongside the analysis for the benefit of the reader, sacrificing elegance for ease of presentation and predictability. Where possible, I have accompanied the English statement of criterial attributes either by photos or line drawings as were seen in Plate I. These provide the reader with as close a replication as possible of the actual physical objects that were available to me and from which I made the deductions of which attributes were criterial. I have aimed to present enough data so that "... fellow anthropologists [will] have sufficient evidence to reject ... [one's] plausible, esthetically pleasing interpretations ..." (Metzger 1964:12). Following each semantic domain will be a tentative componential definition of each classifier (where applicable) "... as a specific combination, or set of combinations, of the components ..." (Wallace and Atkins 1960:60) that I have isolated as criterial. The advantages gained by this presentation, if any, will be discussed in Chapter V. I have, after Wallace and Atkins, {Ibid.) adopted the convention that where a term does not discriminate on a dimension, no numeral subscript accompanies the letter representing that dimension. The order of presentation is arbitrary but consistent. It is determined primarily by (1) the gross semantic distinctions among classifiers and (2) the distributional pattern of classifiers vis a vis nouns within each set, i.e., contrastive, complementary or mixed. Classifiers which have been grouped into mutually exclusive semantic domains will be discussed first. Single or isolated classifiers, (i.e., those which by informant judgements were not considered to participate in any isolated semantic domain) will be discussed last.
4.2. SEMANTIC IMPLICATIONS OF DISTRIBUTIONAL FEATURES
In Chapter II it was seen that two basically distinct distributional patterns of classifiers in reference to nouns in each set occur in the data — contrastive and complemen-
44
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
tary. A large number of domains exhibit distributions incorporating both features, such sets being called, for the time being, mixed. The semantic implications of each such distribution may be made clearer by a general discussion of each distributional pattern and its variants in reference to actual semantic domains. In Figure 1 [p. 38] the distribution matrix accompanying semantic domain 1, one observes a set of two classifiers which exhibit distributional features such that each classifier occurs within the same linguistic environment, i.e., each occurs with each noun in the set. This distributional feature is further observed to be contrastive in that the occurrence of each classifier with each noun is also accompanied by a change in the referential meaning of the noun (cf. photographs of the contrasting classifiers, Plates I and II). To demonstrate contrast between any two classifiers, one is concerned primarily with establishing, by informant response, same or different. Just what the differences may consist of is the problem of distinctive features, a related but quite different question. As will be seen below, some classifiers exhibit the same distributional features as do the classifiers in semantic domain 1, but for which I can show no difference in referential meaning. If such distributions are stable, one is forced to consider them as examples of referential synonyms. However, I have for the time being treated them as distinct members of identical semantic domains. Contrast between any two classifiers is possible in terms of at least two logical possibilities: (1) the denotata of the nouns with which such classifiers occur are such that they occur in at least two distinct natural states which are characterized by the classifiers in this context and/or (2) said denotata can undergo certain transformations such that they acquire the requisite criterial attributes indicated by the two classifiers. In reference to the semantic domain under consideration here, one might state the semantic relationship between /hiht'/ and /b'ehc'/ as follows: /hiht'/ and b'ehc'/ are labels which indicate separate or distinct "bundles" of criterial attributes. The denotata of the nouns of this set are such that when they occur with /hiht'/ they manifest the criterial attributes indicated by that classifier, as these contrast with the criterial attributes indicated by the classifier /b'eh5'/. The relationship is, of course, symmetric in that when the denotata of the set occur with /b'ehc'/, they manifest the attributes indicated by that classifier as they contrast with those indicated by /hiht'/. To paraphrase Frake again, one observes in this instance a situation (i.e., context) indicated by the two classifiers /hiht'/ and /b'eh5'/ where a Tenejapa Tzeltal speaker must make a decision about the category membership of some objects by selecting from a set of alternatives (in this case, presumably two) a verbal label (i.e., one of the two classifiers) with classificatory import. A variant of the contrastive distribution given above is one where the distribution of nouns with classifiers is defective. This distribution is one where some but not all classifiers occur with lists of non-overlapping nouns and where some but not all
45
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
nouns /hehp/ /si?/ X 'firewood' /wale?/ X 'sugarcane' /te?el lo?b'al/ X 'stem of banana' /?ahan/ 'young corn' /lo?b'al/ 'banana' / /h/. I have indicated morpheme boundaries by a hyphen for the convenience of the reader.
Plate XIX - /puh?!'/ 'bends in thin, oblong objects such that breakage is observed, but not through surface of object, such as bends in a paper match' Included in photo: /hpuh^' te?el nicim/ 'one bent flower stem', /'hpuh0' wale?/ 'one bent piece of sugarcane', /?ospuh0' te?/ 'three bends in a piece of tree (stem)'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers: Semantic domain 2
Plate XX - /p'ih/ 'partial breakage in thin, oblong objects such that surface area is severely broken as cracked pencil' Included in photo: /ca?p'ih si?/ 'a piece of firewood broken in two places', /ca?p'ih te?/ 'a piece of tree (stem) broken in two places', /hp'ih wale?/ 'a piece of sugarcane broken in one place'
Plate XXI - /k'ahs/ 'complete breakage of thin, oblong objects' Included in photo: /caPk'ahs wale?/ or /hk'ahs wale?/ 'two pieces of broken sugarcane' or 'one piece of sugarcane broken once;' /ca?k'ahs lapis/ or /hk'ahs lapis/ 'two pieces of broken pencil' or 'one pencil broken once'; /CaPk'ahs si?/ 'two pieces of broken firewood' or /hk'ahs si?/ 'a piece of firewood broken one time' (for a discussion of the ambiguity in this classifier, see Chapter V).
65
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Table 5 — Componential definitions of numeral classifiers: semantic domain 2 Number of Dimensions: 2 Definition of dimensions
Values
'breakage' (A)
:
indication of breakage (ai) ; non-indication of breakage (a2)
'degree of breakage' (B)
:
total (bj) ; surface (b2) ; internal (b3)
Definition of terms by
/ huy/ /puh*'/ /p'ih/ /k'ahs/
: : : :
a2 ai at a!
components
b b b
Semantic domain 3 nouns
numeral classifiers
/p'in/ 'earthen jar' Isefl 'plate' /moc/ 'basket' /p'is/ 'drinking glass' /ti?limete/ 'mouth of jar' /litro/ 'liter bottle'
/nuh/ X
X
/ mahk/ X X X X X X
/muhk/ X X X x
/suhk/ X
X X
Figure 7 - Distribution matrix: semantic domain 3 Domain meaning: semantic domain 3 - 'closing or covering of openings of objects'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers: Semantic domain 3
Plate XXII - /nuh/ 'covering opening of jar, container, etc. with non-flexible objects, the covering object being placed with convex side up' Included in photo: /hnuh p'is/ 'one covered glass'; /hnuh moc/ 'one covered basket'; /hnuh p'in/ 'one covered earthen jar'
Plate XXIII - /mahk/ 'covering openings of jar, containers, etc. with non-flexible objects, covering object being placed with convex side down' Included in photo: /hmahk p'is/ 'one covered glass'; /hmahk moc/ 'one covered basket'; /hmahk p'in/ 'one covered earthen jar'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers : Semantic domain 3
/ifjEiligl
Plate XXIV - /muhk/ 'covering openings of jars, containers, etc. with broad, flexible objects such as cloth fabric' Included in photo: /hmuhk p'in/ 'one covered earthen jar'; /hmuhk moc/ 'one covered basket'
Plate XXV - /suhk/ 'closing of opening of jars, containers, with "stopper-like" object' Included in photo: /caPsuhk limete/ 'two closed bottles'; /hsuhk p'in/ 'one closed earthen jug'
68
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Table 6 — Componential definitions of numeral classifiers: Semantic domain 3 Number of Dimensions: 3 Definitions of dimensions
Values
'type of cover' (A) 'flexibility'(B) 'position of cover' (C)
: stopper (a x ); non-stopper (a2) : flexible (bj); non-flexible (b2) : inverted ( c j ; non-inverted (c2) Definitions of terms by components /nuh/ /mahk/ /muhk/ /suhk/
: : : :
a2 a2 a2 ax
b2 b2
c2 Ci
bi
Semantic domain 4 nouns
numeral classifiers
/cenek'/ 'bean' ¡mafl 'corn dough' /?i6/ 'chili pepper' /tolha?/ 'water' /cilha?/ 'cane beer' /ti?b'al/ 'meat'
/pahc/
/b'ohc/
X
X
X
X
X
X
X
X
X
X
X
X
Figure 8 - Distribution matrix: semantic domain 4 Domain meaning: semantic domain 4 - 'open containers of edible objects'
Discussion: semantic domain 4 The total amount of edibles is considered during enumeration, as 'plateful', 'bowlful', 'basketful'. However, the 'container' is understood with the use of /pah5/ and not obligatorily stated. The classifier /bohc/ is unique in that its form is identical to the noun /bohc/ 'gourd' from which it is derived. Thus the expression /5a?b'ohS £enek' ta b'ohc/* 'two helpings of beans in gourds' is not semantically appropriate while /5a?pahS cenek' ta pulatu/ 'two helpings of beans in plates' is correct.
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of rumerai classifiers: Semantic domain 4
Plate XXVI - /pahc/ 'enumeration of open containers of edible objects' Included in photo: /Pospahc cenek'/ 'three containers of beans'
Plate XXVII - /b'ohc/ 'enumeration of open container of edible objects in gourd' Included in photo: /hb'ohc cenek'/ 'one gourd-full of beans'
69
70
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Table 7 — Componential definitions of numeral classifiers: semantic domain 4 Number of dimensions: 1 Definition of dimensions
Values
'container type' (A)
:
gourd ( a j ; other (a2)
Definition of terms by components / pahc/ /b'ohc/
a
2
a!
Semantic domain 5 nouns
numeral classifiers /t'uy/
/cohak'/ /wo sail/ etc. etc.
/hihp/
x
X
X
X
Figure 9 - Distribution matrix: semantic domain 5 Domain meaning: semantic domain 5 — 'hanging, flexible containers'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
C r i t e r i a l a t t r i b u t e s of n u m e r a l classifiers : Semantic domain 5
Plate XXVIII - /t'uy/ 'hanging flexible container with contents' Included in photo: /ht'uy cohak'/ 'one hanging net bag'
71
72
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers: Semantic domain 5
Plate XXIX - /hihp/ 'hanging flexible container without contents' Included in photo: /h-hihp cohak'/ 'one hanging net bag'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
73
Table 8 — Componential definitions of numeral classifiers: semantic domain 5 Number of Dimensions: 1 Definition of dimension
Values
'contents of hanging object' (A)
:
present (a x ); absent (a2)
Definition of terms by components /t'uy/ /hihp/
: :
ax a2
Semantic domain 6 nouns
numeral classifiers
/wah/ 'tortilla' /kaslanwah/ 'bread' /hun/ 'paper'
/seht/
/seht'/
X
X
X
X
X
X
Figure 10 - Distribution matrix semantic domain 6 Domain meaning: semantic domain 6 — 'torn fragments of thin, pliable objects'
DISTINCTIVE FEATURES OF N U M E R A L CLASSIFIERS
Criterial attributes of numeral classifiers : Semantic domain 6
Plate XXX /seht/ 'torn fragments of thin pliable objects relatively larger than /seht'/' Included in photo (left): /?os-seht/ 'three torn pieces of tortilla' /seht'/ 'torn fragments of thin pliable objects, relatively smaller than /seht/ Included in photo (right): /caPseht'/ 'two torn pieces of tortilla'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
75
Table 9 — Componential definitions of numeral classifiers: semantic domain 6 Number of Dimensions: 1 Definition of Dimension
Values
'size of torn, flexible fragments' (A)
:
small fragments (a^ ; large fragments (a2)
Definition of terms by components /seht/ /seht'/
: :
a2 a1
Semantic domain 7 nouns
numeral classifiers
/k'apal/ 'debris' /tunim/ 'cotton' /jzSojziil holol/ 'hair' /jioiiil cih/ 'hair of sheep' o0il t'ul/ 'hair of rabbit' /?ak/ 'grass'
/pahji/ x X x X x
/6'eht/ X
x
Figure 11 - Distribution matrix: semantic domain 7 Domain meaning: semantic domain 7 'balls of soft, pliable, flexible, strand-like objects'
/woh# X X X X X
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers : semantic domain 7
Plate XXXI - /c'eht/ - /pah**/ - /wohfi/ /c'eht/ - 'balls of soft, pliable flexible materials relatively larger than /pahfi/ or /woh0/' Included in photo: /hc'eht ?ak/ 'one ball of grass' /hc'eht tunim/ 'one ball of cotton' /pahji/ - 'balls of soft, pliable, flexible materials relatively larger than /wohfi/, relatively smaller than /c'eht/' Included in photo: /hpah^ Fak/ 'one ball of grass'; /hpah£ )io0il t'ul/ 'one ball of rabbit hair' /wohji/ - 'balls of soft, pliable, flexible materials relatively smaller than /c'eht/ and /pah^/' Included in photo: /hwoh^ ?ak/ 'one ball of grass' /hwoh^ ^ofiil t'ul/ 'one ball of rabbit hair'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
77
Table 10 — Componential definitions of numeral classifiers: semantic domain 7 Number of Dimensions: 1 Definition of dimension
Values
'size of soft, wool-like spheres' (A):
large (a t ); medium (a 2 ); small (a3)
Definition of terms by components : : :
/pah 0/ /6'eht/ /woh 0/
a a a
Semantic domain 8 nouns
numeral classifiers /say/ /yom/ /keh^/ /peht/ /suht'/ /torn/ /tehfi/ /tuh0/
/te?/ 'tree' /Ci/ 'istle' /wamal/ 'leafy plant' /¡¿ojiSil holol/ 'hair' /?ak/ 'grass' /si?/ 'firewood'
X
X
X X
X
X
X X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X X
X
X
X
X
X
X
X
Figure 12 - Distribution matrix: semantic domain 8 Domain meaning: semantic domain 8 'potentially handable bunches of slender, lengthy objects (flexible/non-flexible), in various bound and un-bound states, severed and not severed'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of n u m e r a l classifiers : Semantic d o m a i n 8
Plate XXXII - /say/ 'small, handful size bunches, not bound, severed' Included in photo: /hsay te?/ 'one bunch of tree stems'; /hsay ?ak/ 'one bunch of grass'; Ihsay ci/ 'one bunch of agave fibers'
Plate XXXIII- /yom/ 'larger than /say/, handful size bunches, not bound, severed' Included in photo: /hyom ?ak/ 'one handful of grass'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers: Semantic domain 8
Plate XXXIV - /kehtf/ 'larger than /yom/, bunches difficult to hold with two hands, not bound, severed' Included in photo: /hkehji ?ak/ 'one double-handful sized bunch of grass'
Plate XXXV - /peht/ 'larger than /kehji/, armful-size bunches, not bound, severed or not severed (/peht/ crosscuts dimension of severedness. Stands of certain plants, especially grass, are also /peht/ cf. semantic domain 29) Included in photo: /hpeht ?ak/ 'one armful of grass'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers : Semantic domain 8
Plate XXXVI - /suht*/ 'bunches of varying sizes, bound by slender-flexible object in center, severed' Included in photo: /hsuht' te?/ 'one bunch of tree stems'; /hsuht' tah/ 'one bunch of pitch-pine'; /hsuht' ?ak/ 'one bunch of grass'
Plate XXXVII - /torn/ 'bunches of varying sizes, bound by slender-flexible object at one end only, severed' Included in photo: /ca?tom ?ak/ 'two bunches of grass'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers : Semantic domain 8
Plate XXXVIII - /tehtf/ 'small bunches of long, slender, flexible objects, not severed, bound near base by slender flexible object (in growth of corn stalk, grass around base may prohibit maturation. Rather than cut the grass, which will serve later as roof thatching, the clump of grass is bound in form of /teh0/ at base) Included in photo (right): /hteh0 ?ak/ 'one bound clump of grass' /tuhff/ 'larger than /teh0/ bunches of long, slender flexible objects, not cut, bound near base by slender flexible object (cf. above /tehtf/)' Included in photo (left): /htuhfi ?ak/ 'one bound clump of grass'
82
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Table 11 — Componential definitions of numeral classifiers: semantic domain 8 Number of Dimensions: 4 Definition of dimensions
Values
'form' (A) 'arrangement' (B) 'type-binding' (C)
severed (a t ); non-severed (a2) bound (bj); un-bound (b2) bound at end or base (cj); bound at center (c2) less than handful (d x ); handful (d 2 ); (potential) double handful (d 3 ); (potential) armful (d4)
'quantity' (D)
Definition of terms by components /say/ /yom/ /keh(i/ /peht/ /suht '/ /torn/ /tehfi/ /tuhfi/
ai a1 a1 at ax a i a2 a2
b2 b2 b2 b2 bx bj bx bx
dj d2 d3 d4 c2 ct cx cx
d3 d4
Semantic domain 9 nouns
numeral classifiers /woh /
/?alcas/ 'orange' /loPb'al/ 'banana' /ji'usub'/ 'grape' /suhs/ 'beehive'
/pal/
X X X
X X X
Figure 13 - Distribution matrix: semantic domain 9 Domain meaning: semantic domain 9 - 'types of bunches' Criterial attributes of numeral classifiers: semantic domain 9 (no photographs available) /woh/
'individual bunches, hanging clusters, as bunches of oranges or individual cluster of bananas or grapes on the larger stem'
/pal/
'individual total hanging bunch, as beehive, grape cluster, banana stalk'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
83
Table 12 — Componential definitions of numeral classifiers: semantic domain 9 Number of Dimensions: 1 Definition
of dimension
Value
'type of bunch' (A)
: Definition I Pal/ /woh/
total bunch (a x ); part of bunch (a2) of terms by
: :
components
ai a2
Semantic domain 10 nouns
/hunI 'paper' /?us/ 'fly' /Pahc'al/ 'mud' /5ab'ek'/ 'wax'
numeral classifiers /nahp'/ x
/pahk'/ X X
X X X
Figure 14 - Distribution matrix: semantic domain 10 Domain meaning: semantic domain 10 'adhesion of certain objects to flat surface'
/noh^/ x X
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers: Semantic domain 10
Plate XXXIX - /pahk'/ - /noh0/ - /nahp'/ /pahk'/: 'total adhesion of objects to flat surface due to adhesive agent on exterior surface of object, (as tape, etc.)' Included in photo (left): /hpahk' hun/ 'one adhered piece of paper' /noh0/: 'partial adhesion of object to flat surface due to adhesive agent on interior (back) surface of object' Included in photo (center): /hnoh0 hun/ 'one partially adhered piece of paper' /nahp'/: 'total adhesion of object to flat surface due to adhesive agent on interior (back) surface of object' Included in photo (right): /hnahp' hun/ 'one adhered piece of paper'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Table 13 — Componential definitions of numeral classifiers: semantic domain 10 Number of Dimensions: 2 Definition of dimensions
Values
'type of adhesion' (A) 'degree of adhesion' (B)
: :
exterior ( a j ; interior (a2) total (b x ); partial (b2)
Definition of terms by
/nahp'/ /noh^/ /pahk'/
: : :
a2 a2 ax
b b b
components
85
86
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
a o J3 c0 >0 Oh -y.
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers : Semantic domain 18
'slender, flexible objects in n a t u r a l extended state, n o m a n i p u l a t i o n ' Included in p h o t o : / h l i h k ~ h i l cilc'ahan/ ' o n e extended c o r d ' ; /hlihk — hil laso/ ' o n e extended rope'
Plate LXI1I /t'im/ 'slender, flexible objects in taut, stretched position, as clothesline' Included in p h o t o ( t o p ) : / h t ' i m laso/ ' o n e stretched r o p e ' /tim/ 'slender, flexible objects stretched between two points, but n o t taut, as clothesline, etc.' Included in p h o t o ( b o t t o m ) : / h t i m laso/ ' o n e stretched rope'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers : Semantic domain 18
Plate LXIY - /k'am/ 'slender, flexible objects in short coils, twisted to hold shape (or tied) in center; not circular' Included in photo: /hk'am laso/ 'one coiled rope'; /caPk'am cilc'ahan/ 'two coiled cords'
Plate LXV - /soht/ 'slender, flexible objects in nearly perfect circular coils' Included in photo: /hsoht laso/ 'one coiled rope'; /hsoht c'ahan/ 'one coiled piece of hide'; /hsoht cilc'ahan/ 'one coiled cord'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers : Semantic domain 18
Plate LXVI - /cahp/ 'slender, flexible objects, long coils, the upper end of coil tied off by one end of the object' Included in photo: /caPcahp laso/ 'two ropes in coiled position'
Plate LXVII - /pihs/ 'slender, flexible objects rolled into ball-like form' Included in photo: Ica?pihs cilc'ahan/ 'two balls of cord'; /hpihs laso/ 'one balled rope'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers : Semantic domain 18
Plate LXVIII - /pehc'/ 'slender, flexible objects in braided form (three-strand braids)' Included in photo: /hpehc' ?ak/ 'one braided piece of grass'; /hpehc' laso/ 'one braided rope'
Plate LXIX - /lak'/ 'slender, flexible objects, individual strands of which comprise larger object, e.g., strands of a rope' Included in photo: /canlak' laso/ 'four strands of a rope'; /caPlak' cilc'ahan/ 'two strands of a cord'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
113
Table 21 — Componential definitions of numeral classifiers: semantic domain 18 Number of Dimensions: 4 Definition of dimensions
Values
'degree of manipulation' (A) 'class of manipulation' (B)
natural ( a ^ ; manipulated (a2) stretched (bj); enrolled (b 2 ); balled (b 3 ); braided (b 4 ); stranded (b 5 ) loose (cx); taut (c2) coiled (di); looped (d 2 ); loop-twist (d3)
'type of stretching' (C) 'type of enrolling' (D)
Definition of terms by components /lihk~hil/ /pihs/ /pehc '/ /Iak '/ /tim/ /firn/ /cahp/ /soht/ /k'am/
a
i a2 a2 a 2 a2 a2 a2 a2 a2
b3 b4 bs bi bx b2 b2 b2
cx c2
d d! d2 d3
Semantic domain 19 nouns
numeral classifiers /tehfl
/ton/ 'stone' /te?/ 'tree' /b'akal ?isim/ 'corncob' /hun/ 'paper' /b'ak/ 'seed' etc. etc.
~
/loht'/
~
/hiht/
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Figure 23 - Distribution matrix: semantic domain 19 Domain meaning: semantic domain 19 'objects wegded in forked object'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers : Semantic domain 19
Plates LXX and LXXI - /lahj£'/ ~ /loht'/ - /luht/ 'objects wedged in forked objects' Included in photos: /hlahji'~loht'— luht ton/ 'one wedged stone'; /hlahji'~loht'~luht bakal ?isim/ 'one wedged corncob' (I have been unable to show semantic contrast between any of the three forms in this domain but include it as a tentative semantic domain)
115
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Semantic domain 20 nouns
numeral classifiers
/?al5as/ 'orange' /?isim/ 'corn' /lo?b'al/ 'banana' /ton/ 'stone' /cenek'/ 'beans' /ji'ub'ilum/ 'pebbles' etc. etc.
/b'uhs/
/t'ol/
/kun/
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Figure 24 - Distribution matrix: semantic domain 20 Domain meaning: semantic domain 20 'aggregation or piles of individuated objects'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers : Semantic domain 20
Plate LXXII - /b'uhs/ - /t'ol/ - /kun/ /b'uhs/ 'small piles (relative to /kun/) of individuated objects with maximal horizontal extension' Included in photo (left): /hb'uhs ton/ 'small piles of stones' /t'ol/ 'small piles (relative to /kun/) of individuated objects with maximal vertical extension' Included in photo (center): /ht'ol ton/ 'small pile of stones' /kun/ 'large piles (relative to /t'ol/ and /b'uhs/) of individuated objects with maximal horizontal extension' Included in photo (right): /hkun ton/ 'large pile of stones'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
117
Table 22 — Componential definitions of numeral classifiers: semantic domain 20 Number of Dimensions: 2 Definitions of dimensions
Values : :
'size of piles' (A) 'type of pile' (B)
large (a x ); small (a2) vertical extension (bj); horizontal extension (b2)
Definition of terms by components /b'uhs/ /t'ol/ /kun /
: : :
a2 a2 aj
b2 bx b2
Semantic domain 21 numeral classifiers
nouns
/te?/ 'tree' /sulub'/ 'cattle horn' /patil/ 'shoulder' etc. etc.
/b'ahc/ X X
/b'ehò/ X X X X
/mehc /
/p'uhs/ X X
Figure 25 - Distribution matrix: semantic domain 21 D o m a i n meaning: semantic domain 21 'curves, warped states in slender, non-flexible objects'
/luhk/ X X
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers : Semantic domain 21
Plate LXXIII - /b'ehc/ ~ /b'ahc/ 'warped, slender, non-flexible objects with ends in raised position' Included in photo: /hb'ehc~b'ahc te?/ 'one warped piece of tree stem' Discussion: I have not been able to demonstrate semantic contrast between /b'eh5/ and /b'ahc/.
Plate LXXIV - /mehc/ 'warped, slender non-flexible objects, ends in no specified position' Included in photo: /hmehc te?/ 'one warped piece of tree stem'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers : Semantic domain 21
Plate LXXV - /p'uhs/ 'warped, slender non-flexible objects, ends in inverted position' Included in photo: /hp'uhs te?/ 'one warped inverted piece of tree stem'
Plate LXXVI - /luhk/ 'permanently warped, slender, non-flexible objects, one end in hooklike fashion' Included in photo: /hluhk te?/ 'one hooked tree stem'
120
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Table 23 — Componential definitions of numeral classifiers: semantic domain 21 Number of Dimensions: 2 Definition of dimensions
Values
'position' (A) 'type of warping' (B)
raised ends ( a ^ ; inverted ends (a2) hooked (b^; non-hooked (b2) Definition of terms by components
/b'ahc~b'ehc/ /mehc/ /p'uhs/ /luhk/
a
i
t>2 b2 b2 bi
Semantic domain 22 nouns
numeral classifiers
/ma^'/ 'corn dough' /pom/ 'incense' /kantela/ 'candle' etc. etc.
/pihs/
/poh?!/
x x
x x
X
X
Figure 26 - Distribution matrix: semantic domain 22 Domain meaning: semantic domain 22 'objects enveloped in broad, flexible materials'
DISTINCTIVE FEATURES OF NUMERAL CLASSIFIERS
Criterial attributes of numeral classifiers: Semantic domain 22