^{page 548 note 1} My research assistant was Mr. Mesgenna Almedom of Shuma Negus in the Hamasien district of Eritrea, who has spent most of his adult life in Asmara. Research was carried out in 1951–3, both in London and in Eritrea. Other speakers were used, but the linguistic statements are based on his speech.

^{page 548 note 2} cf. Leslau, W., Documents Tigrigna, Paris, 1941, pp. 30 if. The remarks that follow do not, however, apply to Leslau's analysis, which is neither directly based on the script, nor phonemic.Google Scholar

^{page 548 note 3} cf. Firth, J. E., ‘Sounds and Prosodies’, TPS, 1948, p. 151.Google Scholar

^{page 548 note 4} cf. Firth, J. R., ‘Modes of Meaning’, Essays and Studies (The English Association), 1951, pp. 118–49, esp. pp. 119–21.Google Scholar

^{page 548 note 5} Both the transcription and the translation are provided solely for the identification of the form, for the convenience of the reader, and are no part of the analysis presented here. The transcription does not belong in any way to the phonological analysis. It is based largely on the Ethiopie script used by my assistant. Most of the symbols have roughly the phonetic values indicated in the I.P.A. alphabet, except that † is used for [t'], C for [tf'], s for [ts'], q for [k'], j for [d3], y for [j], h for [h], ? for [?], 3 for [?], and k^w, g^w, and q^w for the labiovelars. Separate symbols are not used, as in the script, for the fricative (post-vocalic and non-geminated) k and q.

^{page 550 note 1} cf. Allen, W. S., ‘Retroflexion in Sanskrit’, BSOAS xvi, 3, p. 556, n. 2.Google Scholar

^{page 550 note 2} Syllabic division is made in terms of syllable types CV and CVC only, where C stands for C and c and V for V and v.

^{page 551 note 1} Or half open central in the alternative form—see below.

^{page 551 note 2} In a full phonological analysis labiovelar and laryngal consonant systems are required. Labiovelarity and laryngality are then prosodic in that they jointly characterize the consonant systems and the initial or final systems—initial and C₁ or C₂, final and C₃ or (structure 1 only) C₄.

^{page 553 note 1} Except where the vowel quality in the singular form is open front, but on this see below.

^{page 553 note 2} It is necessary to set up here unitarysystems (of one term—not including zero, which would mean two terms) which are not usually justified, but are required here in view of their exact parallelism with the two- and three-term systems.

^{page 555 note 1} cf. pp. 553–4. The problem cannot be solved by calling the system ‘front’, since later in this analysis a front system with [i] and [e] as the exponents of the terms is required (p. 560).

^{page 556 note 1} If the singular structure CVCVC were considered alone a seven-term system could be set up for the V of the second syllable. Where, however, central vowels [e] and [a] are the exponents of the terms the related plural form is cvCCVC; this is dealt with in part II. Where [e] is the exponent, there is no corresponding broken plural.

^{page 556 note 2} There are several plural forms of structure 1 with the suffix -ti which is there to be treated as a lexical feature. For its grammaticalrelevance here, however, see p. 55

^{page 557 note 1} There is a single example of such a relation for structure 2: Singular mag^wdi Plural mag^wadu ‘lid’, where the relevant vowel quality of the singular is the exponent of a term in a front system, and the quality of the plural is the exponent of a term in a back system.

^{page 558 note 1} cf.p. 565.

^{page 558 note 2} The fact that no front system is set up should not be regarded as important. Such a system would be required for singular forms not paired with broken plurals. System 1 (i) below might, on phonetic grounds, be labelled‘back’, but since it is a three-term system with [a] as the exponent of the third term, it patterns with the central rather than the back system of A. In fact integration of the systems of A and B can be achieved by considering that the central and back systems of A are paralleled by a single system in B.

^{page 558 note 3} The round brackets indicate that the term is to be set up only by reference to the plural form.

^{page 559 note 1} Phonetically describable as ‘back’ this system is labelled ‘central’ since it patterns with the central system of A. cf. n. 2, p. 558.

^{page 560 note 1} With front-back relation, cf. p. 557.

^{page 560 note 2} cf. pp. 553–4 and 555, and n. 1.

^{page 560 note 3} For different plural forms of the first two examples see above.

^{page 560 note 4} p. 557.

^{page 561 note 1} See p. 556, n. 1.

^{page 562 note 1} The same symbols will be used even for the very different qualities in syllables with labiovelar or laryngal articulation. For the latter even the transcription uses a instead of a, but this will be ignored in the analysis and one set of symbols used.

^{page 564 note 1} cf. p. 563.

^{page 565 note 1} p. 558.

^{page 565 note 2} I do not regard these as examples of ‘gemination’, but of a cluster of two consonant elements.

^{page 566 note 1} cf. Firth, J. R., “Sounds and Prosodies”, TTS, 1948, p. 151: ‘The suggested approach will not make phonological problems appear easier or oversimplify them. It will make the highly complex patterns of language clearer both in descriptive and historical linguistics.’Google Scholar