BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL. 35(4)


 vowel), (high central vowel), a (mid central vowel) a, u, and o
 (mid or low-mid back vowel). In all of these languages, the
 nasals have homorganic prenasalized voiced stops as
 subphonemic variants, e.g. m is sometimes [mb].
 The Wayapi phonemicization (Grenand 1980) is retained,
 although the symbols are standardized. The (surface) phonemes
 are: p, t, k, ', s, m, n, fi, g, w, I, y, i, e, a, u, o,4, and nasalization.
 The Ka'apor data are presented in the phonemicization by
 Kakumasu (1986:399-401): p, t, k, kw, ', s, s, h, m, n, g, gw, w, r, y,
 i, e, i, a, u, and o. The phonemicization for Temb6 is taken from
 Bendor-Samuel (1966) recognizing as phonemes the following: p,
 t, k, kw, ', c (alveolar or alveopalatal affricate), z (alveolar or
 alveopalatal affricate or palatal continuant), h, m, n, g, gw, w, r, i,
 e, 4, a ,a, u, and o.
 For the Asurini language, we have only a very tentative
 phonemicization (Irmazinha Edith 1987:5-7) and some
 unpublished transcriptions by Sidney Facundes of the Museu
 Goeldi. Although the sounds ii, and y are possibly allophones
 of the same phoneme, as are g and g, we employ a broad
 phonetic transcription rather than risk an undifferentiated
 preliminary analysis. The symbols, then, are: p (bilabial stop,
 fricative, or affricate), t, c, k, ', j, g, m, n, f, g, w (bilabial
 semivowel or fricative), r, y, i,-i-, e, a, u, and nasalization.
 There is no phonemicization available for Arawet6,
 although some information appears in Viveiros de Castro
 (1986:145). The broad phonetic transcription for Arawet6 uses
 the following symbols: p, t, k, ', b, d, c, c, h, d; m, n, i, w, r, y, T,
 a, e, i, i, a, u, o, and nasalization.

2. Cronquist (1988:18) later reaffirmed this, stating that:

 "It is perfectly clear that natural, recognizable
 groups of species, and groups of such groups, exist.
 The ranks at which these groups should be received
 are not inherent in the nature of the group, but depend
 on subjective individual judgment . any evaluation
 of the importance of the characters marking a
 [supraspecific] group is likely to be difficult and subject
 to unresolvable differences of opinion."

3. It is interesting to compare this with Berlin et al. (1973), whose
 methods differ somewhat from ours. They explicitly intended to
 exclude as possible cognates "compound names which appear to


260