¹ Ocaina is a member of the Witoto language family of northeastern interior Peru, in the area of the Ampiyacu river.

The language data on which this paper is based were gathered by Ilo Leach, of the Summer Institute of Linguistics. The morphological analysis was prepared by her in consultation with Evelyn G. Pike. The theoretical matrix and presentation is by the author.

² Some of the macromorphemes of the type treated by Robert E. Longacre (Trique Tone Morphemics, Anthropological linguistics 1.4.9, 1959) may be composed of ‘a sequence of two morphemes which, in terms of the systematic contrasts, function as one unit while they are nevertheless morphemically separated by types of random contrasts’. Such a sequence is called a ‘morpheme cluster’, a term borrowed from Garvin's treatment of fused polymorphemic units (IJAL 17.85, quoted by Longacre 38). Their morphemic solutions are related to the one proposed here, without the larger theoretical framework within which our solution is fitted.

³ The Ocaina phonemes are represented as follows, with phonetic symbols in brackets to indicate the norms of certain ones: p, t, t ^y, k, ?; 6, r, d ^y, g; f, s, š, x, h; v [b], y [ž]; ¢, č, z, j̆; m, n, ñ; M, N, Ñ (lenis); i, e, a, o, u [ï]; į, ę, ą, ǫ, ų,; /'/ (high tone). A hyphen marks the presence of a morpheme division.

⁴ A few basic low-toned roots are perturbed to high tone by stem-formative suffixes which are not discussed here. The high-toned allomorph of the object morpheme occurs regularly in these circumstances: umá-ko-d^yoñít ^yo (d ^yoñi ‘looking at’ -t ^yo ‘?‘). Allomorphic changes of roots, other than those involving tone or affecting allomorphs of affixes, are not treated here.

⁵ The stem allomorph huma has single u before the enclitic sequences ñi-xo and ñi-o. The allomorph huumą occurs, without enclitics, above.

⁶ For the theoretical background to such a suggestion, note Kenneth L. Pike, Language as Particle, Wave, and Field, The Texas quarterly 2.37–54 (1959), and Language in relation to a unified theory of the structure of human behavior, Part 3, Chapter 14, §3 (Glendale [now Santa Ana], Calif., 1960).

For related problems in the segmentation of certain verbal affixes of Spanish, see the latter work, 1.129b (1954); for their relation to fused (portmanteau) tagmemes in Spanish (with one morpheme filling two tagmemic slots, like the {o} of ablo), see 3.586. For suggested limitations on the use of the portmanteau concept see Sol Saporta, Spanish Person Markers, Lg. 35.612–15 (1959) : ‘a morpheme may not have just one member [allomorph] if that member is a portmanteau; a morpheme may not have just two members if those members are a portmanteau and zero.‘

⁷ A suggestion of Lorrie Anderson of the Summer Institute of Linguistics.

^7a [The expression make short shrift of is not familiar to me; I know the word shrift only in the expression give short shrift to. The former collocation probably represents a crossing with the semantically related make short work of, to which the author refers immediately below.—Editor.]

⁸ See Pike, Language etc. 1.85b–86, 87b, 89a, 100a, 105b, 126b–27a (1954). For the structural handling of words like shrift, see 1.123a and 3.87.