BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL. 35(4)


 DATA AND METHODS

 Data Sources


 A total of 625 names for the 167 botanical species were obtained from the
 five languages. These data are presented in Tables 1, 2, and 3. Indigenous
 plant names are divided among the five languages as follows: 114 (Ar), 90 (As),
 160 (K), 125 (T), and 136 (W). All 136 names in W are derived from
 Grenand's published study (1980), and these are supported by his voucher
 numbers and determinations. Of the remaining 489 names in the other four
 languages, 399 are represented by voucher numbers on the series Bal6e
 (voucher specimens are deposited at the New York Botanical Garden with
 duplicates at the Museu Paraense Emflio Goeldi). In other words, 535 (87%)
 of the plant names in our data are supported by voucher numbers, either by
 Bal6e or Grenand (1980). Many names that were not documented by voucher
 numbers refer to species whose identities were unmistakable in the field,
 especially domesticates. Of the 45 Temb6 names not supported by a voucher
 number, 42 were obtained from Boudin's published dictionary (1978). Other
 names for some of the 19 species in Arawet6 and 17 in Asurini which were not
 documented by voucher numbers were obtained from Eduardo Viveiros de
 Castro (pers. comm. 1988), Aryon D. Rodrigues (pers. comm. 1988), and
 Velda Nicholson (1982). These names tend to refer to extremely well-known
 domesticated species. The nine Ka'apor names not documented by voucher
 numbers were supplied by Bal6e based on reliable field determinations of
 species.
 Each plant name associated with a voucher number in the series Bal6e
(for Ar, As, K, and T only) was elicited from several informants by Bale at the
moment of its collection. Each name was later checked for accuracy in the
village. Data were recorded in phonetic transcription. Bal6e is a native
speaker of English, fluent in Portuguese, with reasonable speaking fluency in
Ka'apor; he has some linguistic training.


 Data Format


 Tupi-Guarani is not phonologically difficult. The accuracy of the
transcription of plant words in these five languages is reasonably high. For
example, the c/1 allophony and the schwa phoneme in Temb6 appeared in
Bal6e's transcription as predicted by Bendor-Samuel (1966). There are
probably some minor errors in transcription, for example, in i vs. a vs. -,