16 weer Language, Culture, and Environment Tupt-Guarani Plant Names Over Time WILLIAM BALEE AND DENNY MOORE We may note, in passing, that the double or compound names are the most doubtful. They may consist of two mistakes; one in the root or princi- pal name, destined almost always to indicate the geographical origin, some visible quality, or some comparison with other species. The shorter a name is, the hetter it merits consideration in question of origin or antiquity; for it is by the succession of yeats, of the migrations of peoples, and of the trans- port of plants, that the addition of often erroneous epithets takes place. Alphonse de Candolle, Origin of Cultivated Plants [orig. Fr. 1886) Both because of its many languages and associated indigenous peoples as well as its extraordinary biological and ecological diversity, Amazonia constitutes an ideal setting for investigating relationships among language, culture, sub- sistence, and habitat. Several recent studies have been made of linguistic and cognitive systems that Amazonian peoples illustrate in naming and classify ing the organisms and habitats of their immediate surroundings (e.g-, Berlin 1992; Grenand 1980; Posey 1983). This essay contributes a diachronic (tem- poral) dimension to such studies. We examine many synchronic (current) lin- guistic and cultural practices in order to help better comprehend diachronic processes, which, through their inferred action over time, have apparently generated many of the interrelationships among language, culture, and envi- ronment in Amazonia that one discerns today. Specifically, this essay assesses patterns of plant naming in the Tupi-Gua- rani family of languages. Since by definition these languages are descended from a common mother language, Proto-Tup{-Guaranf, the measure of simi- larity among them (excluding resemblances due to borrowing, which appears to be relatively insignificant) in terms of plant names should also indicate the364 BALEE AND MOORE degree to which these names have been retained from the mother language through time. In the semantic domain of plants, one can estimate the proportion of names from Proto-Tupi-Guarani, a language spoken in Amazonia at about the time Latin was spoken in Rome, that somehow survived the vicissitudes of time and history. In addition, it is most intriguing that the plants whose names have most tended to endure through time seem to possess certain common, culturally defined features. As for the five languages represented in this study, it is striking that some plant names, such as those for a medium-sized copal tree (Hymenaea parvifolia Huber, Caesalpiniaceae family), which is most often found in old fallows (sites where indigenous peoples had swiddens more than forty years ago), are very similar, whereas others, such as those for a large tree, ‘moela de mutum’ (Lacunaria jenmani [Oliv.] Ducke, Quiinaceae family), which is usually seen only in high, primary forests (forests that lack evidence for past swidden horticulture), are highly divergent: LANGUAGE COPAL TREE ‘MOELA DE MUTUM’ Araweté (Ar) ita’ itwd-pedi Asurini (As) yet iwa-kaw-'iwa Ka’apor (K) yeta-’é kupapa-’iran-’é Tembé (T) utai-"iw iwa-au-’iw-ran Waypi (W) yita-'é mit-’ay If a pair-by-pair comparison among the five languages is carried out for these two plant species, it can be shown that there are ten pairs of similar words for the copal tree yet no pair of similar words for ‘moela de mutum’ among these languages. The present essay seeks to explain why words for cer- tain plant species in these five related languages display stability and genetic similarity, as with these words for the copal tree, whereas words for certain other plant species, such as ‘moela de mutum,’ in the vocabularies of these same languages, seem to be labile and highly unlike. In other words, we seek to identify certain linguistic and cultural processes involved in the observed patterns of similarity and dissimilarity for a common corpus of plant names in these five different yet related languages. The data (which are fully presented in Balée and Moore 1991), consist of 625 words for 167 botanical species native to the neotropics in the Araweté (henceforth Ar), Asurini do Xingu (henceforth As), Ka’apor (henceforth K), Tembé (henceforth T), and Wayapi (henceforth W) languages of eastern Amazonia (see Appendixes 1, 2, and 3). These data are highly appropriateTupi-Guarani Plant Names 365 for isolating the factors related to similarity and dissimilarity in plant names among different languages of the same family for many reasons. First, the five languages are dispersed in four linguistic subgroupings of Tupi-Guarant (A. R. Rodrigues 1984/85; A. R. Rodrigues, personal com- munication, letter, 1988), with only W and K being classified within the same subgrouping. Second, these five languages are spoken in three ecologi- cally diverse regions: the Xingu River Basin of north central Brazil for Ar and As, the Gurupi/Turiagu River Basins of extreme eastern Amazonia for K and T, and the Oiapoque River Basin of northern Amazonia for W. Third, al- though all five groups practice horticulture, they exhibit significant differ- ences in crop staples and utilization of nondomesticated species. For exam- ple, the Ar rely heavily on maize (Viveiros de Castro 1992), in contrast to the other groups who traditionally are more dependent on tubers, especially manioc. Fourth, collections and scientific determinations of voucher plant specimens for the vast majority of the 167 plant species in this sample were obtained for the five languages— identifications for the other plants either took place in the field or are based on dictionary references. In particular, as for W, all plant words were obtained from the monograph by Pierre Grenand (1980). Finally, the corpus of data is large enough to test statistically propo- sitions concerning similarity and dissimilarity among plant words in the five languages. The present study differs in methodology from that of Berlin et al. (1973) [also see Berlin 1992, 255-56], which, with respect to two Mayan languages, concluded that names for plants of significant cultural importance were more likely to be retained than names for plants of negligible cultural importance. The present essay compares data from five languages instead of two, which yields a comparison of ten pairs of languages instead of one pair. It is organ- ized, moreover, according to botanical referents (that is, plant species them- selves), rather than indigenous names. We listed these referents and then counted similarities and dissimilarities among indigenous words for them on a plant-by-plant basis among each of the ten pairs of languages, rather than list folk generic names for botanical species held in common between lan- guages and then count the pairs of similar words. The dissimilar as well as the similar terms are also presented in our data, with morpheme-by-morpheme glosses, in order to facilitate insights into the patterns of naming plants and the relationships that these patterns may have to culture. The results of this study indicate that the retention rate of plant names is not directly deter- mined by cultural importance, but rather by the type of name, and that cultural practices that are associated with the degree of domestication of a plant ultimately determine the type of name that the plant receives.366 BALEE AND MOORE Method of Comparison Measuring similarity and variation of vocabulary between related languages is different from the procedures of historical linguistics— reconstruction using the comparative method. This exercise first asks the questions, given a botan- ical species, what are the words for it in various languages of the same family and are these words similar or different? These questions reflect well the common sense notion of what one means by similarity, and their answers can be quantified in a straightforward way. In historical linguistics, by contrast, one searches for cognate terms showing systematic sound correspondences, allowing, if necessary, a considerable degree of semantic “shift” of the refer- ent. There is no concern with determining noncognacy, which is difficult with incomplete collections and without knowledge of the full range of mean- ing for each word. In order to determine whether words in two languages are similar or differ- ent, it is necessary to specify a comparable range of meaning for them (such as the biological species) to prevent problems of overlapping (see Alcorn 1984, 270). Consider, for example, the K and T names for jarana and cacador, which are two species in the Brazil nut family: K NAME NAME Lecythis ef. chartacea Berg iwiri-’é iwiri-"iw (“jarana”) Lecythis idatimon Aubl. yasi-amir iwiri'éw-pita (“cacador”) If the range of meaning were not restricted and were one permitted to count as being similar those pairs of words that occurred in vertically different positions, then from the point of view of T, both words would have one simi- lar K counterpart (the K word for Lecythis cf. chartacea), on the basis of the initial morpheme, iwiri. And both words would have one dissimilar K coun- terpart (the K word for Lecythis idatimon), insofar as neither T word shares any of its constituent morphemes with yasi-amir. That is to say, there would be two paits of similar words and two pairs of dissimilar words between K and T for these two species. Such a result would be meaningless for the purposes of determining lexical similarity for plant names between the two languages. This is because from the point of view of K, only one of the two words has similar counterparts in T. Such nonrestricted comparison would become yet more confusing upon considering five instead of two languages. Further, a skewing could result insofar as a pair of languages having relatively complete plant collections would offer greater opportunity for finding similar words than would a pair of languages having relatively incomplete plant collections.Tupf-Guaran{ Plant Names 367 Therefore, it is necessary to restrict the range of meaning of the referent, and then examine whether the words for it are similar or dissimilar on a pair-by-pair basis. We decided to restrict referents to the taxonomic rank of botanical species, mainly because one may argue that the botanical species is the most objective level of abstraction for distinguishing between individual plants. The species is more objective than higher order units, such as tribes, genera, and families, given that “rank is not inherent in supraspecific groups” (Cronquist 1968, 31). We subscribe to the view that species are natural units, morever, not merely products of mind or fantasy (see Gould 1980, 204-13). Species, for the purposes of this analysis, are also more suitable referents for comparing indigenous plant names than are taxa of infraspecific ranks. For example, the classification of many neotropical cultivars (i.e., varieties) of a single botanical species is still far from possessing taxonomic exactitude. In their taxonomic revision of the genus Manihot (which includes cultivated manioc, Manihot esculenta Crantz) Rogers and Appan (1973, 34) poignantly noted that, “It is impossible to apply formal subspecific taxon epithets to fleeting variants which are not related to some precise geography or ecologi- cal region.” Names for species that were introduced to the neotropies within the last two thousand years are excluded from the list of plant names under study since these names would not, by definition, be of Tup{-Guarant origin. There is some doubt, nevertheless, about the origins of a few species included in our analysis, all of which are domesticates. These include papaya (Carica papaya L.) (Storey 1976, 23; Sousa 1974, 99), bananas and plantains (Musa spp.) (Smole 1980; Lery 1960, 157; Lisboa 1967, 122; Sousa 1974, 98; Vascon- cellos 1865, 136), and bottle gourds (Lagenaria siceraria Mol.) (Sousa 1974, 95; Heiser 1979, 114-16). All domesticates included in the analysis, how- ever, are of sufficient antiquity in the neotropics (i.e., probably older than the five languages in our analysis) that they can be considered, for histori- cal linguistic purposes, to be neotropical. In other words, the plant vocabu- lary of ProtoTupi-Guaranf, in principle, could have had names for these species. Only species for which names in three or more of the five languages were obtained are included in the analysis. This criterion guarantees that each species occurs in at least two of the three ecological regions; it also limits any possible effects of linguistic borrowing, which would be most likely to occur among the proximate pairs of languages (the K/T and the Ar/As pairs). As field collections were being made, it began to appear that the pattern of similarity and dissimilarity in plant names between these Tup{-Guaranf lan- guages was at least partly a function of the degree of domestication of plants368 BALEE AND MOORE (Balée 1989). Patterns of plant nomenclature appear to segregate traditionally domesticated and nondomesticated plants. These patterns may be summa- tized as follows: (1) life-form heads (for example, K mira, kaa, sipo) are not incorporated into names for traditional domesticates; (2) animal morphemes are incorporated into names for traditional domesticates only when the ani- mals are not ecologically associated with the plants themselves; (3) “obscure” plant names (i-e., names that do not incorporate plant morphemes, such as K akwi-nami “agouti-ear,” which refers to a small forest shrub in the coffee family) do not denote traditional domesticates; (4) morphemes referring to divinities (such as K kurupir) and to the state of being “false” or “similar” (K -ran, At ri, As -rdna, T -ran, W -/@) are only incorporated into plant names that do not refer to traditional domesticates (Balée 1989). Three basic kinds of plant species can be identified in terms of degree of domestication: nondomesticates, semidomesticates, and domesticates. Non- domesticates typically occur in primary well-drained forest, archaic vine forests, or swamp forests. These are zones where contemporary human inter- ference in species composition and dominance is, or recently has been, negli- gible. Although some nondomesticates may sporadically occur in zones of recent human interference, such as old swiddens and old fallows, they do not appear to gain dominance other than in fairly undisturbed forests. Nondo- mesticates are wild species. Semidomesticates, in contrast, do not generally appear to become eco- logically dominant without human interference, usually by horticultural fires or by the seemingly random tossing away of viable seeds. A few of these spe- cies (such as Annona montana Macf. var. marcgravii ‘araticum’ and Theobroma grandiflorum Schum. ‘cupuacu’'—see Appendix 2) may be deliberately planted and carefully protected from time to time by individuals from one or more of the five societies, but such management lacks cross-cultural regularity. The category of semidomesticates corresponds to that of “protected plants” in Berlin et al. (1973, 146). In addition, most of the semidomesticates in Ap- pendix 2 are disturbance indicators, that is, their presence tends to indicate formersites of human habitation and horticulture. Semidomesticates are gen- erally found in swiddens of all ages and old fallows. Neotropical plant domesticates (Appendix 3) are completely dependent on human management for their long-term propagation; most, if not wholly incapable of setting seed, are producers of minuscule quantities of viable seed. These species are often parthenocarpic as a result of human interfer- ence. The genotypes of domesticates have been altered by human interven- tion over long periods of time.Tup{-Guarant Plant Names 369 To measure the degree of similarity in plant names between two languages, we consider pairs of words, such that any word in language A corresponds to a word in language B for the same species. The number of such pairs that are similar and the number that are dissimilar phonologically are then tabulated, the ratio between them being the degree of similarity. In order to define similarity adequately, it is necessary to distinguish literal plant words from metaphorical or descriptive (henceforth metaphorical) plant words, a distinction that proves to be of crucial analytical importance. In our usage, literal plant words are those that contain a literal plant morpheme; they may contain other morphemes as well. Literal plant morphemes are here defined as those that have as their sole referent a specific class of plants, excluding general life form morphemes, such as “tree” or “herb.” The word pine in English, for example, refers only to specific trees—the association between the word and its referents is purely arbitrary. The terms jack pine, white pine, and red pine are also literal since they contain the literal morpheme “pine.” Likewise, in the Tupf-Guarant languages under study, the words for an ingé tree found in old fallows (Inga capitata Desv. —from example 6, Ap- pendix 2) are literal in the five languages: Ar, ifia-paka-', ‘L-long-tree’s As, ueruru-inja, ‘toad-L’; K, inja-hu-’s, ‘L-big-tree’; T, tapi’i-virja-"iw, ‘tapir-L-tree’s and W, inja-mulua-ya, ‘L-pregnant-owner,’ since they all incorporate the lit- eral morpheme ifialirja, which we gloss as ‘L.’ Two literal plant words are considered to be similar if their literal plant morphemes are similar, regardless of the other morphemes occurring in the word. Thus, the three words for ‘inga” in this example, upon being systematically compared, constitute three pairs (Ar/As, Ar/K, arid As/K) of similar, literal plant words. In our usage, metaphorical names are those that (a) do not contain a literal plant morpheme or (a) do contain a literal plant morpheme that is being used only metaphor- ically (i.e. the class of plants designated by the whole metaphorical name is not included in the class of plants designated by the literal morpheme). For example, in folk English, dogwood is an example of a metaphorical plant term, since neither dog nor wood refers to a specific plant, as does pine. The term poison oak is also metaphorical, since it is not botanically a kind of oak (which, in folk English, when applied to species of the genus Quercus, is otherwise a literal plant term). Likewise, in the Tupf-Guaranf data, the K word for Tapirira guianensis Aubl., tayahu-mira, ‘white-lipped peccary-tree,’ is metaphorical, since its constituent morphemes do not denote a specific plant and the W word for Conceveiba guianensis, a’i-miniyu, ‘sloth-cotton,’ is metaphorical because this tree of the spurge family is not a kind of cotton. The relation between a metaphorical plant term and its referent is, in a sense,370 BALEE AND MOORE less arbitrary than that of a literal term to its referent, because some culturally given interpretation of the plant intervenes between the metaphorical term and its referent. It is more complicated to judge the similarity of metaphorical plant terms since these may contain several morphemes, and some degree of arbitrariness is unavoidable. Two metaphorical terms are deemed similar (1) if they share two morphemes that are similar in sound and in meaning (e.g., K yasi-sépo-pe, ‘tortoise-vine-flat,’ and T iwipo-pew, ‘vine-flat’) or (2) if one of the principal nominal components is similar in sound and in meaning, excluding life-form morphemes or common plant part morphemes (e.g., T zani-ro-’#u, ‘oil-bitter- tree,’ W yani, ‘oil,’ referring to the crabwood tree of the mahogany family, Carapa guianensis Aubl.). Phonetic resemblance between forms must be apparent for them to be considered similar (such as ifia and ina, ‘ing4,’ above). Given the number of languages involved in this study, the lack of descriptive work on two of the languages (Ar and As), the uncertainty of some phonetic details, and the limited size of the corpus, we cannot, in general, reconstruct the Proto-Tupi- Guarant forms with certainty and then identify borrowings by the fact that they do not show the same systematic sound correspondences as do the recon- structable words. Borrowing, however, appears to have been very minimal. Berlin et al. (1973, 152) also observed that borrowing of plant words between Tzeltal- Tzotzil, which are very closely related Mayan languages, was a “relatively rare occurrence.” In addition, as a general principle, names for domesticates would be the most likely of the three categories of plant names to be bor- rowed, yet these names strongly tend to reconstruct in ProtoTupf-Guarani (Aryon Rodrigues, personal communication, letter, 1988). It is very doubtful that much borrowing occurred, since a society would have had to lose the domesticate as well as its name and then regain the domesticate together with a new name. It seems unlikely this would have happened often. Second, if borrowings were extensive from language A to language B, one would expect these two languages to be conspicuously more similar to each other than is either to languages C, D, and E, But among these Tupi-Guarant languages, no such significant pairings were found. There has been, however, minor Portuguese influence (K kuyer-’¢ and T wira-kuzer,‘spoon-tree,’ from colher, K kanei-’é-tuwér, ‘tesin-tree-white,’ from candeia, and the words for “wild cacao” and “cacao”). It should be noted that the referents of these words are of neotropical origin. The similarities between the languages for these words probably reflect independent borrowing from Portuguese in the remote past.Tupf-Guarant Plant Names 371 Similarity between literal plant terms must, in general, be due to their retention in the languages since splits in the protolanguage. This is probably also the general cause of resemblances between metaphorical terms, though in some cases resemblance may be due to independently similar cultural in- terpretations of the plant. No effort is made here to exclude these, because there are no clear means of identifying such cases and their contribution to the overall proportion of similarity is certainly extremely limited. Results The results show clearly that the more intensively managed plants have higher rates of similarity in their names from one language to the other. Combining data from all ten pairs of languages: ALL WORD PLANT CATEGORY PAIRS SIMILAR DISSIMILAR Nondomesticates 441 136 (30.8%) 305 (69.2%) Semidomesticates 278 164 (59.0%) 114 (41.0%) Domesticates 198 159 (80.3%) 39 (19.7%) The differences between the three categories of plants—nondomesticates, semidomesticates, and domesticates—in terms of degree of similarity (30.8%, 59%, and 80.3%, respectively) are very significant (chi? = 146.483, 2 df, p <..coor). In other words, lexical similarity between the ten pairs of lan- guages very significantly increases along a scale of increasing human domesti- cation of plants (see fig. 16.1). The results also indicate that the type of plant name, whether literal or metaphorical, is also strongly influenced by the degree of domestication: ALL PLANT CATEGORY worps* LITERAL METAPHORICAL Nondomesticates 323 137 (42.4%) 186 (57.6%) Semidomesticates 175 128 (73.1%) 47 (26.9%) Domesticates 113 110 (97.3%) 3 (2.7%) *Fourteen indeterminate words are excluded. The differences between the proportions of metaphorical words in the three categories of plants (57.6%, 26.9%, and 2.7%) are very significant and show that these words were taken from fundamentally different samples (G hetero- geneity = 143.482, p < -ooor). In other words, the proportion of metaphor-372 BALEE AND MOORE oo} | same SEGRE LEXICAL SIMILARITY (IN PERCENT) Nondomesticates Semidomesticates Domesticates PLANT CATEGORY Figure 16.1. Lexical similarity by plant category. ical words declines considerably as a function of increasing plant domestica- tion (see fig. 16.2) Another important finding is that the literal plant terms are much more similar from language to language than are metaphorical terms. Overall, the similarity of pairs of literal words compared to metaphorical words is: SIMILAR DISSIMILAR TOTAL Literal Word Pairs 393 (77.0%) 118 (23.0%) 511 (100%) Metaphorical Word Pairs 49 (23.2%) 162 (76.8%) 211 (100%) (The fact that the percentage of similar literal word pairs is almost exactly the same as the percentage of dissimilar metaphorical word pairs is a coinci- dence, not a mistake). It is important to note that the overall proportions of similarity of literal plant name pairs for each of the three plant categories are not significantly different (68% for nondomesticates, 78.6% for semidomesti- cates, and 80.8% for domesticates). The ratio of literal to metaphorical plant words, combining words from all plant categories, is not significantly different between the five languages (chi? = 1.7, df = 4, p > .05). Discussion The above results are generally in accord with the hypothesis that cultural importance influences the retention of plant names in sister languages (Berlin etal. 1973):Tupf-Guarant Plant Names 373 Degree of cultural importance of plants ———> Degree of retention of names of plants But our results suggest that this process can be further elucidated by recogniz- ing as analytical variables (1) the degree of plant domestication (domesti- cated, semidomesticated, nondomesticated); (2) a widespread nomenclatural pattern among these languages, in which words for traditional domesticates tend to be literal, words for nondomesticates tend to be metaphorical, and words for semidomesticates tend to lie between these extremes; and (3) the much higher stability of literal, as opposed to metaphorical, plant names. In this model, the types of names that the nomenclatural pattern assigns to domesticates tend strongly to be literal, the types assigned to semidomesti- cates show an increasing proportion of metaphorical terms, and the majority of those assigned to nondomesticates are metaphorical. For some reason, literal terms are more stable over time and hence are more apt to be similar from language to language. That is, synchronic cultural factors of degree of plant domestication and the plant naming system combine with the linguistic properties of names and diachronic linguistic processes to produce similarity and variation in plant vocabulary: (x) Synchronic Factors: Degree of plant domestication + Nomenclatural system ——> Proportions of literal vs. metaphorical names (2) Diachronic Processes: Differential retention of proportions of similarity and literal/metaphorical names ——> Similarity among names 200 180 160 140 120 100 80 60 40 20 NUMBER OF WORDS IN ALL LANGUAGES Nondomesticates ‘Semidomesticates ‘Domesticates PLANT CATEGORY Figure 16.2. Metaphorical vs. literal plant words in the five languages.374 BALEE AND MOORE When the data of the Mayan study by Berlin et al. (1973) are reanalyzed in the above terms, it appears that the retention of plant words in the Mayan languages follows the same patterns found for Tupi-Guaran‘. Using the glosses of Tzeltal plant words given in a later publication (Berlin et al. 1974), it can be determined that, of the Tzeltal plant names cognate with Tzotzil names, seventy-nine are literal and only thirty-two are metaphorical, indicating a high rate of cognacy for literal terms. ‘Combining the categories of cultivated and protected plants, on the one hand, and those of wild-useful and wild-useless plants on the other (tables 1 4in Berlin et al. 1973) to reach a valid statistical sample, we arrive at the fol- lowing proportions for the 111 cognate sample of the two Mayan languages: cultivated/protected: 43 (37 literal and 6 metaphorical) wild: 68 (42 literal and 26 metaphorical) The ratio of literal to metaphorical cognates (6.2 to 1) in the first category is about 4 times higher than the corresponding ratio (1.6 to 1) for the combined “wild” category. This difference is very significant at p < .o1, chi? = 7.0, df = 1, Clearly, the more managed plants are receiving literal names at a much higher rate than the less managed plants, as in the case of the Tupt- Guarani languages. In addition, as with these Tupi-Guarané languages (Balée 1989), a nomenclatural pattern that lexically distinguishes between culti- vated and noncultivated plants has been explicitly noted for Mayan speakers of the Yucatan peninsula (Marin et al. 1976, 472). Once again combining cultivated and protected plants on the one hand, and useful/useless wild plants on the other (see table 5 in Berlin et al. 1973, 161) to reach a statistically meaningful sample, we find that of the 52 word pairs in the first group, 43 are cognates, whereas of the 205 word pairs in the “wild” group, only 68 are cognates. That is, the Mayan managed plants have a cognacy rate about two and one-half times higher than that of the wild plants, the same pattern as that of Tup{-Guarant. The question remains whether plant utility, aside from degree of plant domestication, as we have defined it, would more economically explain the proportion of similar plant words among the five Tupi-Guaran{ languages in our sample. We quantified the uses of nondomesticated species (see Prance et al. 1987) for the K. The uses were (r) food, (2) construction material, (3) tool, weapon, utensil or container, (4) medicine, and (5) adornment. Fuel and game food were excluded as uses, since these are extremely wide- spread among forest species. Each use is of two types: major or minor. A major use has a value of 1.0, a minor use, 0.5, and no use, o (cf. Turner 1989). The value of plant utility as a predictor of literal as opposed to metaphori-Tup{-Guarani Plant Names 375 cal plant names, which are predictors of cognacy and retention, was tested within the nondomesticated plant category, where the factor of use can be separated from that of degree of plant domestication. Of the nondomesti- cated plants with a high use value, thirteen are metaphorical and twelve are literal. Of the nondomesticates with a low use value, thirty-seven are meta- phorical and thirty are literal. The differences are not great and are not statistically significant. The usefulness of a plant, therefore, at least within the nondomesticated category, is not shown to be a predictor of whether its name will be literal or metaphorical, and, by inference, whether its name will be retained. Why are literal plant terms more stable? One hypothesis is that of Al- phonse de Candolle (stated later, in somewhat different form, by Zipf's law [Berlin 1992, 259]): they are shorter. Although there may be some truth in this, it is an incomplete explanation, since the shorter metaphorical words in our sample (having only one morpheme excluding any life-form or com- mon plant part morpheme) do not seem to have a higher similarity rate than the longer words (two or three morphemes, excluding any life-form or com- mon plant part morpheme). Another possibility is that the literal terms en- dure because of their arbitrariness—the metaphorical terms involve a cultural interpretation of the plant that is susceptible to change. In spite of the strong correlations observed in the section on results, some degree of unpredictability and possibly still unidentified factors, at work in determining naming patterns, remain. For example, some nondomesticated species (in particular, several palms) show stable literal names. It is not yet clear what causes such exceptions. The general patterns explained above, however, appear also to be present in the Mayan Tzeltal and Tzotzil as well as in the Tupt-Guarani languages studied. Perhaps this is the general case in egalitarian, horticultural societies. It would be instructive to compare these observed patterns with those of state societies and foraging societies. Further research might also examine whether similarity and variation in other semantic fields, such as birds, fish, or mam- mals, can be analyzed along the same general principles. APPENDIX 1 Examples of Names for Nondomesticated Plant Species in Five Tupi-Guarant Languages! 1. Anacardium giganteum Hancock ex Engl. (Eng. “wild cashew”; Port. “cajut”): akayu, ‘L-tree’ (K); akazu-'iw'ete, ‘L-tree-true’ (T); akay-u, ‘L-big’ (W). Obs.: Three of three possible pairs are similar. All names are literal.376 BALEE AND MOORE 2. Tapirira guianensis Aubl. (Port. “tatapiririca”): takaré-me'e-’a-'i, ‘?-some-fruit-tree’ (Ar); tayahu-mira, ‘white-lipped peccary-tree’ (K): tata-piririk-’iw, ‘fire-crack- ling-tree’ (T); tata-pilil, ‘fire-cracking’ (T). Obs.: One (T/W) of six possible pairs is similar. All names are metaphorical. 3. Capparis sp. (Eng. “caper tree”): yapa-ta’i, ‘crested oropendola-spicy-tree’ (Ar); iwira-tdi, ‘tree-spicy’ (As); sawiya-mira, ‘rat-tree’ (K). Obs.: None of the three possible pairs is similar. All names are metaphorical. 4. Conceveiba guianensis Aubl.: ina-i-sina, ‘Inga-tree-black’ (As); arapuha-mira, ‘red brocket deer-tree’ (K); wira-ki'iw-ran, ‘tree-chili pepper-similar’ (T); a’ -miniyu, ‘sloth-cotton’ (W). Obs.: None of the six possible pairs is similar. All names are metaphorical. . Sapium marmieri Huber (Eng. “tallow tree”): éka-i,‘latex-tree’ (Ar); yuwa-'iwa, ‘bird lime-tree’ (As); wakura-mira-hu, ‘nighthawk-tree-big’ (K). Obs.: None of three possible pairs is similar. All names are metaphorical. Pychotria poeppigiana M. Arg. (Port. “mata cavalo”): akuéi-wird, ‘agouti-tree’ (An); yawéi-inga, ‘tortoise-Inga’ (As); tapi’ika’a, ‘tapit-herb’ (K); ka'a-rahi, ‘herb- poison’ (T). Obs.: None of six possible pairs is similar. All names are meta- ba phorical. . Simaba cedron Planch. (Port. “serve p’ra tudo”): ari-kara-’ wa, ‘?-yam-tree’ (As); pere-pusan-'é, ‘skin eruption-remedy-tree’ (K); akuti-wira, ‘agouti-tree’ (T). Obs.: None of three possible pairs is similar. All names are metaphorical. 8. Ampelocera edentula Kuhlm.: yaci-pdpe-'i, ‘tortoise-claw-tree’ (Ar); iwi-payé, ‘earth-shaman’ (As); tapi'idkwape-’i, ‘white bearded manakin-tree’ (K). Obs: None of three possible pairs is similar. All names are metaphorical. 9. Rinorea passoura (DC.) Kuntze: yana-'i, ‘spider-tree’ (Ar); wayaw-a'iy, ‘guava- seed’ (As); piwa-'s, ‘arrow point-tree’ (K); piwa-'iw ‘L-tree’ (T). Obs.: One (KIT) of six possible pairs is similar. All names are metaphorical. 10. Minquartia guianensis Aubl. (Port. “acariquara”): yiwoy-'é, ‘boa constrictor-tree’ (K); wakari-"iw, ‘kind of fish-tree’ (T); wakali-’s, ‘kind of fish-tree’ (W). One (TW) of three possible pairs is similar. All names are metaphorical. APPENDIX 2 Examples of Names for Semidomesticated Plant Species in Five Tup{-Guarant Languages? 1. Spondias mombin L. (Eng. “hog plum”; Port. “taperebé”): akaiya-’i, ‘L-tree’ (Ar); kayuwa-'iwa, ‘L-tree’ (As); taperiwa-'s, ‘old village-fruit-tree’ (K); tawera-"it, ‘old village-tree’ (TT); akaya, ‘L’ (W). Obs.: Four of the ten possible pairs are simi- lar (Ar/As, Ar/W, As/W, K/T). Two (K and T) of the names are metaphorical. 2. Annona montana Macf. var. macrgravii (Eng. “soursop”; Port. “araticum”): tayaha- "si, ‘white lipped peccary-fruit-tree’ (Ar); aratiku-'s, ‘L-tree’ (K); aragiku-'iu, ‘Letree’ (T). Obs.: One (KIT) of three possible pairs is similar. One (Ar) of the names is metaphorical.Tupi-Guaran{ Plant Names 377 3. Didymopanax morototoni (Aubl.) Decne & Planch. (Eng. “matchwood”; Port “morotot6”): murete-towi-'i, ‘L-?-tree’ (Ar); moroto-'é, ‘L-tree’ (K); molototo, ‘I? (W). Obs.: Three of three possible pairs are similar. All names are literal. 4. Maximiliana maripa (Cort. Serr.) Drude (Port. “inaja"): ndya-'i, ‘L-tree’ (Ar); inayd, ‘! (As); inaya-'i, ‘L-tree’ (K); naza-'itu, ‘L-tree’ (T); inaya, (W) [L]. Obs.: Ten of ten possible pairs are similar. All names are literal. 5. Hymenaea parvifolia Huber (Eng. “copal tree”): yiita-'i, ‘L-tree’ (Ar); yeta-'iwa, ‘Litree’ (As); yeta', ‘Ltree’ (K); zuta-i-'iuy ‘L-small-tree’ (T); yéta-', ‘L-tree’ (W). Obs.: Ten of ten possible pairs are similar. All names are literal. 6. Inga capitata Desv. (Port. “ingé”): ifta-paka-i, ‘Lelong-tree’ (Ar); kururu-ija, ‘road-L? (As); arja-hu-'s, ‘L-big-tree’ (K); tapi'icrina-'iu, ‘tapir-L-tree’ (T); inja- mulua-ya, ‘L-pregnant-owner’ (W). Obs.: Ten of ten possible pairs are similar. All names are literal. +7. Eugenia patrisii Vahl. (Port. “pitanga”): me'e-'a-i, ‘some-fruit-tree’ (Ar); ardkwa- mira, ‘little chachalacha-tree’ (K); iwai-'iw, ‘fruit-black-tree’ (T); éwa-pita, ‘fruit-red’ (W). None of six possible pairs is similar. All names are metaphorical. 8. Theobroma grandiflorum Schum. (Port. “cupuagu”): kapi-', ‘L-tree’ (Ar); kipi- huc’i, ‘L-big-tree’ (K); kupi-'a-'iw, ‘L-fruit-tree’ (T); kapiai, ‘L (W). Obs.: Six of six possible pairs are similar. All names are literal. o. Fleurya aestuans (L.) Gouv. (Eng. “stinging nettles,” Port. “urtiga”): pina, ‘L (Ar); pint, ‘L (As); purake-ka’a, ‘electric eel-herb’ (K). Obs.: One (Av/As) of three possible pairs is similar. One name (K) is metaphorical. 10. Lantana camara L. (Eng. “lantana”): fiéma-'i, ‘nothing-tree’ (Ar); kanami-ran, ‘Clibadium sylvestre (a domesticated fish poison)-similar’ (K); arakwa-wira, ‘little chachalacha-tree’ (T); yakale-pili, ‘caiman-aromatic plant’ (W). Obs.: None of six possible pairs is similar. All names are metaphorical. APPENDIX 3 Examples of Names for Domesticated Plant Species in Five Tup{-Guaran{ Languages’ 1. Anacardium occidentale L. (Eng. “cashew”; Port. “caju”): akayi, ‘L (Ar); akayn, L (K); akazu, ‘L (T); akayu, ‘E (W). Obs.: Six of six possible pairs are similar. All names are literal. 2. Xanthosoma sp. (Eng. “cocoyam”; Port. “tajé”): taya, ‘L (K); taza, ‘L (T); ta-st, ‘Lthomy’ (W). Obs. Three of three possible pairs are similar. All names are literal. . Crescentia cujete L. (Eng. “calabash tree”; Port. “cuieira”): kai, ‘L (Ar); kuyd, ‘E (As); kwi, L (K); kui, E (T); kwi, ‘Z (W). Obs.: Ten of ten possible pairs are similar. All names are literal. Bixa orellana L. (Eng. “annatto”; Port. “urucum”): irka, ‘L (Ar); urukii ‘L (As); uruku, ‘LE (K); seruku, ‘E (T); uluku, ‘LZ (W). Obs.: Ten of ten possible pairs are similar. All names are literal. *378 BALEE AND MOORE 5. Neoglaziovia variegata L. (Port. “caraué”): kirawd, ‘ (Ar); kurawa, ‘L (As); kirawa, ‘L (K); kurawa, ‘L (T); kulawa, ‘L (W). Obs. Ten of ten possible pairs are similar. All names are literal 6. Cucurbita moschata (Lam.) Poir. (Eng. “crookneck squash”; Port. “jerimum”): yurumi, ‘L (At); yerima, ‘L (As); yurumi, L (K); zurumii, ‘L (1); asikala, ‘L (W). Obs.: Six (At/As, AvK, Av/T, ASK, AST, KIT) of ten possible pairs are similar. All names are literal. Dioscorea trifida L. (Eng. “yam”; Port. “inhame”): ard, L (Ar); kard, ‘L (As); hard, ‘L (K); kard, ‘L (T); kala, ‘EZ (W). Obs.: Ten of ten possible pairs are similar. All names are literal. Gossypium barbadense L. (Eng. “cotton”; Port. “algodo”): méniyu, ‘L (Ar); miniyti, ‘L (As); maneyu, ‘L (K); manizu, ‘L (T); miniyu, ‘L (W). Obs.: Ten of ten pos- sible pairs are similar. All names are literal. 9. Musa X paradisiaca L. (Eng. “plantain”; Port. “banana”): patsisi, ‘L’ (Ar); pako, ‘L (As); pako, ‘L (K); pako, ‘L (T); bako, ‘L (W). Obs.: Six (As/K, AsT, As/W, KIT, KAW, TIW) of ten possible pairs are similar. All names are literal. ro. Zea mays L. (Eng. “maize”; Port. “milho”): awatsi, ‘L (As); awati, ‘L (As); awa, > L(K); awaci, L (1); awasi, 'L (W). Obs.: Ten of ten possible pairs are similar. Alll names are literal. 2 NOTES This is a shorter and somewhat revised version of Balée and Moore (1991). The reader wishing to examine all the data on which the arguments of this paper are based should also consult the longer version. We would like to thank Dr. Manuel Ayres of the Universidade Federal do Paré in Belém for very helpful advice on the statistical analysis. 1. Botanical species’ names and English or Portuguese folk names (when known) are followed by the indigenous names, given in boldface. Morpheme-by-morpheme glosses, enclosed within single quotation marks, follow each indigenous name. Hyphens indicate morpheme boundaries. The letter L in a gloss indicates a literal plant morpheme. The language code for each indigenous name follows the gloss. 2. Seen. 1 above. 3. See n. 1 above. REFERENCES Alcorn, J. B. 1984 Huastec Mayan ethnobotany. Austin: University of Texas Press. Balée, W. 1989 Nomenclatural patterns in Ka’apor ethnobotany. Journal of Ethnobiology 9 (1): 1-24.Tup{-Guarani Plant Names 379 Balée, W., and D. Moore. 1991 Similarity and variation in plant names in five Tupi-Guarané languages (eastern Amazonia), Bulletin of the Florida Museum of Natural History (Biological Sciences) 35 (4): 209-62. Bendor-Samuel, D. H. 1966 Hierarchical structures in Guajajara. Ph.D. diss., University of London. Berlin, B. 1992 Ethnobiological classification. Princeton, N.J.: Princeton University Press. Berlin, B., D. E. Breedlove, R. M. Laughlin, and P. H. Raven. 1973 Cultural significance and lexical retention in Tzeltal-Tzotzil ethno- botany. In Meaning in Mayan languages, edited by M. S. Edmonson, 143-64. The Hague: Mouton. Berlin, B., D. E. Breedlove, and P. H. Raven 1974 Principles of Tzeltal plant classification. New York: Academic Press. Boudin, M. H. 1978 Diciondrio de Tupi modemo. Vols. x and 2. Sao Paulo: Conselho Estadual de Artes ¢ Ciéncias Humanas. Cronquist, A. 1968 The evolution and classification of flowering plants. Boston: Houghton Mifflin. Gould, S. J. 1980 The panda’s thumb. New York: W. W. Norton. Grenand, P. 1980 Introduction a l étude de l'univers Waypi: Ethnoécologie des Indiens duu Haut- Oyapock (Guyane Frangaise). Paris: seLar. Heiser, C. B. 1979 The gourd book. Norman: University of Oklahoma Press. Léry, J. 1960 Viagem 2 terra do Brasil. Translated by S. Milliet. Sao Paulo: Livraria Martins Editora Lisboa, C. 1967 Histéria dos animais e érvores do Maranhao. Lisbon: Publicagdes do Arquivo Hist6rico Ultramarino e Centro de Estudos Histéricos Ultramarinos. Marin, A. B., A. B. Vazques, and R.M.L. Franco 1976 Nomenclatura etnobotnica Maya: Una interpretacién taxonémica. Instituto Nacional de Antropologia e Historia, Centro Regional del Sureste: 436. Mexico City: Coleccion Cientifica. Nicholson, V. 1982 Breve estudo da lingua Asurini do Xingu, no. 5. Brasilia: Summer Institute of Linguistics.380 BALEE AND MOORE Posey, D. A. 1983 Indigenous ecological knowledge and development of the Amazon. In The dilemma of Amazon development, edited by E. F. Moran, 225-57. Boulder, Colo.: Westview Press. Prance, G. T., W. Balée, B. M. Boom, and R. L. Cameito. 1987 Quantitative ethnobotany and the case for conservation in Amazonia. Conservation Biology 1 (4): 296-310. Rodrigues, A. R. 1984-85 —_Relages internas na familia lingufstica Tup{-Guarant. Revista de Antropo- logia 27/28:33-53.- Rogers, D. J., and S. G. Appan 1973 Manihot, manihotoides (euphorbiaceae). Flora Neotropica monograph BB. New York: Hafner Press. Smole, W. 1980 Musa cultivation in pre-Columbian South America. Geoscience and Man 21:47-50. Sousa, G. S 1974 Notfcia do Brasil. Sio Paulo: Departamento de Assuntos Culturais do Mec. Storey, W. B. 1976 Papaya, Carica papaya (Caricaceae). In Evolution of crop plants, edited by N. W. Simmonds, 21-24. London: Longman. Tumer, N. J. 1988 The importance of a rose: Evaluating the cultural significance of plants in Thompson and Lilloet interior Salish. American Anthropologist 90 (2): 272-90. Vasconcellos, S. 1865 Chronica da companhia de Jesu no estado do Brasil. Lisbon: A. J. Fernandes Lopes. Viveiros de Castro, E. 1992 From the enemy's point of view: Humanity and divinity in an Amazonian society. Translated by C. V. Howard. Chicago: University of Chicago Press.