¹ Amuzgo is a member of the Amuzgoan family, proposed by Robert E. Longacre in his paper ‘Progress in Otomanguean reconstruction’, Proceedings of the Ninth International Congress of Linguists 1016–25 (The Hague, 1964). Amuzgo is spoken in eleven or twelve villages in southeast Guerrero and southwest Oaxaca in Mexico; there are approximately 15,000 speakers of the language, most of them monolingual. The data were collected during several field trips from 1952 to 1960. A number of informants assisted in the analysis; those who contributed most are Isaura Sebastián, Florentina López, Juan Sabino Apóstol, and his wife Victoria Valtierra de Apóstol. Genaro Santiago worked with me on the tone analysis. All are natives of the large municipio town of Xochistlahuaca, Guerrero; their dialect forms the basis for the present description.

I wrote a preliminary draft as a graduate student at the University of Michigan in 1961. I acknowledge especially the help of Kenneth L. Pike of the University of Michigan and the Summer Institute of Linguistics in checking the vowel and tone hypotheses in 1959, during a linguistic workshop in Mitla, Oaxaca, Mexico. I am also grateful to Robert E. Longacre of the Summer Institute of Linguistics for his suggestions.

² The terms ‘ballistic’ and ‘controlled’ are taken from Pike, Language in relation to a unified theory of the structure of human behavior II (Glendale, California, 1955). They are useful for describing the dynamics of the breath force in Amuzgo syllables. Another analysis which uses these terms is that of William Merrifield, ‘Palantla Chinantec syllable types’, Anthropological linguistics 5:5.1–16 (1963). The similarity of our descriptions and our use of the terms ‘ballistic’ and ‘controlled’ are the more striking as we worked simultaneously and independently.

³ Pitch details are not written in phonetic data except in the section which deals with tone.

⁴ Tone is written with superscript numbers after the syllables to which they apply: ¹ is high, ² is mid, ³ is low. Phonetic variations are indicated by the use of plus and minus. ^1– denotes a pitch lower than the theoretical norm for high but not as low as ²⁺, ³⁺ denotes a pitch slightly higher than the theoretical norm for low, and so on. Tone sequences are indicated by superscript ¹³, ³², ²¹.

⁵ When glottal stop is final in a stressed syllable which is followed by an unstressed syllable within the phonological foot, the glottal stop may be considered to belong phonemically to the preceding syllable as terminus, or to the following syllable as onset. On the other hand, it may be considered to function as both terminus and onset, the syllable boundary being indeterminate. For the present description I have chosen to consider the glottal stop in this position to be a terminus.

⁶ This does not imply that tones are absolute, but merely points out that the end point of a phonemic low-mid sequence may still be perceptibly lower than a following mid tone. So also, the phonemic mid-high sequence may have a perceptibly lower end point than a following high tone.

⁷ Tone sandhi occurs between affixes and stems; there is no sandhi between words. In the most common type of sandhi the tone of each suffix remains stable while the stem tone and syllable dynamics vary in rather complex patterns. The details of tone morphology are not treated in this paper.

⁸ Although the allophone [l] of the phoneme h and the allophone [ll] of the phoneme l appear to be phonetically similar, the fact that these phones occur in different positions and with distinct conditioning factors makes it plausible to assign them to different phonemes.

⁹ While at the University of Michigan in 1961, I undertook a brief study of these clusters and unit phonemes with the use of a spectrograph. The resulting data, though limited, confirm the duration and voicing contrasts here described.

¹⁰ ‘Focal consonant’ as used here refers to that consonant which appears to be more prominent and more basic in relation to the other consonants which occur in the cluster with it. It usually occurs in a position which allows for greater diversity of membership. It has a higher phonetic rank, according to criteria set up by Eunice Pike, ‘Phonetic rank and subordination in consonant patterning and historical changes’, Miscellanea phonetica (Le maître phonétique) 2.25–41 (1954).

¹¹ In the former case the [am] which occurs in syllable types +0 +Ń ±T and +O +N ±T would be in complementary distribution with both [ṃ] which occurs in syllable type +Ń, and [ṃ] which occurs in syllable type +N before labials. The three could be considered allophones of a single phoneme. I have allowed distributional criteria, however, to carry more weight than phonetic similarity in the assignment of these allophones. The [ṃ] of syllable type +N has been analyzed as a phonemic syllabic nasal which precedes only syllables of types +O +N ±T and +O +Ń ±T. On the other hand [әm], and the [ṃ] of the syllable type +Ń, occur in positions filled by vowels.