^{page 567 note 1} My thanks are due to Mr. Musa H. I. Galaal, now of Radio Somali, Hargeisa, who, as my research assistant in Somaliland (1950–1) and in London (1951–4) helped me in the collecting and testing of the linguistic data upon which this article is based. I also acknowledge my debt to Professor Malcolm Guthrie for the help and advice which he gave me during my work on the formulation of the statements presented in this article.

^{page 567 note 2} This dialect is spoken by the Isaaq tribal family who inhabit the British Protectorate of Somaliland. For further geographical and ethnographic information see: Bryan, M. A., The Distribution of the Semitic and Cushitic Languages of Africa,Oxford University Press, 1947, International African Institute;Google ScholarHunt, J. A., A General Survey of the Somaliland Protectorate, London, 1951, Crown Agents for the ColoniesGoogle Scholar; Lewis, I. M., Peoples of the Horn of Africa: Somali, Afar and Saho, London, 1955, International African Institute.Google Scholar

^{page 567 note 3} Lilias E., Armstrong, ‘The Phonetic Structure of Somali’, MSOS, XXXVII, III, 1934, pp. 116– 61. On p. 149 of her article Armstrong speaks of ‘retracting’ and ‘fronting’ vowels and thus sets up two series of vowels.Google Scholar

^{page 567 note 4} Schleicher, A. W., Die Somali Sprache, Part I, Berlin, 1892, p. 89— ‘Vokalharmonie’.Google Scholar

^{page 567 note 5} Reinisch, L., Die Somali Sprache, Vol. III Grammatik, Kaiserliehe Akademie der Wissenschaften, Sildarabische Expedition, Vienna, 1903, p. 28— ‘Vocalharmonie’.Google Scholar

^{page 567 note 6} Warsama, S. and Abraham, R. C., The Principles of Somali, London, 1951, 2nd edition (cyclostyled); ‘ Vowel Harmony’ is chiefly dealt with on pp. 396–9, and there are numerous references to it throughout the book.Google Scholar

^{page 568 note 1} Note that Armstrong's i corresponds to my (i) and [i], and her ii corresponds to my (ii) and [ii]. These additional distinctions of vowel quality which I have introduced are necessary, as otherwise pairs like

^{page 569 note 1} At the end of a paragraph, or when words are given in isolation pause marks are unnecessary and are therefore dispensed with.

^{page 569 note 2} In the examples given in this article all the symbols, except the vowel symbols and the symbol y are identical with those of Armstrong. The correspondences between my vowel symbols and those of Armstrong have already been given in § 2. The symbol y represents a sound articulated by a movement of the tip of the tongue towards the upper teeth or the teeth-ridge but without actual contact. The tip of the tongue is raised to the level of or above the lower teeth but there is no substantial raising of the mid part of the tongue towards the hard palate. This sound occurs only intervocalically and is always voiced. Berghold uses the symbol δ for this sound and describes it on pp. 125–7 of his ‘Somali Studien’, WZKM, xm, 2, 1899. Tone and stress are not marked here as they are not relevant to the problem under discussion; to use the tone and stress marks would require a lengthy explanation.

^{page 570 note 1} Note that under certain conditions le when followed by a vowel is pronounced as leh.

^{page 570 note 2} The symbol ∫J corresponds to S as used in most of the works on Somali in German and Italian and to sh as used in most of the works in English, including Bell's, C. R. V.The Somali Language, London, 1953 (Longmans).Google Scholar

^{page 570 note 3} In this context Mahamuud would be either the actual head of a family or an eponymous ancestor of a tribe.

^{page 570 note 4} See the preceding footnote.

^{page 571 note 1} It is very difficult to measure or to classify the speed of pronunciation. It is even more difficult to induce a speaker of this language to keep to the same speed, say, while he narrates a story.

^{page 571 note 2} This appears to happen only, however, when higher speed is accompanied by greater spontaneity of utterance and lower speed by its greater deliberateness.

^{page 572 note 1} I have introduced here the term ‘minimum size utterance’ in order to deal with those forms which cannot occur in isolation. By ‘minimum size utterance’ I understand the shortest possible utterance in which a form of this kind can occur. There may be, in the case of some forms, several minimum size utterances for a particular form, but they are never very numerous. When the term ‘minimum size utterance’ is used in this article the following restrictions are attached to i t: all the forms in a minimum size utterance, except the form under examination, must be Back, i.e. their vowels must belong to Series A when they themselves occur in isolation or in a minimum size utterance. These restrictions are necessary in order to remove the possibilities of influence of the vowels of the neighbouring forms described in § 8.

^{page 572 note 2} I use the term ‘Form’ instead of Word’ as very often some forms of the same word are Back and some are Front. Some words, however, are indeclinable and have only one form and in such cases ‘Form’ and ‘Word’ are co-extensive. When some forms of a particular word are Back while some are Front the differences in their distribution can usually be connected with the presence or absence of certain affixes, e.g. Wuu eegay ‘he looked’, Wuu eegayay ‘he was looking’.

^{page 573 note 1} Note that monosyllabic forms are also divided into Back and Front according to the same criteria and are marked in the same way as the polysyllabic ones, e.g. dar ‘clothes’, beer ‘a garden’, god ‘a hole’, geed ‘a tree’.

^{page 573 note 2} Forms which belong to the minimum size utterance but are not under discussion themselves are written in capital letters.

^{page 573 note 3} A compound word; cf. dab ‘fire’ and qaad ‘take’.

^{page 573 note 4} A compound word; cf. reer ‘people, tribe, family’, and magaalo ‘a town’.

^{page 574 note 1} A compound word; cf. daan ‘a jaw’ and deer ‘long’. Compounds of this kind should, in my opinion, always be hyphenated.

^{page 574 note 2} A compound word; cf. jeeni ‘a front leg or paw’ and wareen ‘stab’ (Imper. Sing.).

^{page 574 note 3} Due to such difficulties as, for example, the varying speed of pronunciation and individual variations.

^{page 574 note 4} i.e. probabilities of frequent occurrence.

^{page 574 note 5} i.e. an utterance within which there are no pauses.

^{page 574 note 6} It is very difficult to assess how far back, within an uninterrupted utterance, this tendency reaches. Here, particularly, individual variations and variations according to the speed of pronunciation can be observed. I have often found, moreover, that the more spontaneous the utterance the further this tendency extends.

^{page 574 note 7} For grammatical information concerning these forms consult C. R. V. Bell, The Somali Language, already mentioned in n. 2, p. 570 of this article

^{page 574 note 8} See n. 1, p. 570.

^{page 575 note 1} In this dialect, owing to certain junction features it is often convenient and sometimes necessary to write two or more forms together. In such combinations of forms the tendencies described above operate in the same way as elsewhere. It may be difficult for a reader not familiar with the morphology of this dialect to recognize the component forms, and that is why in a text where the marks for Back and Front Forms are given and the limit marks of harmonic groups are omitted it may be useful to show, in textual notes, the internal composition of mixed combinations, i.e. the combinations which include both Back and Front Forms at the same time. In the case of some combinations of forms it is possible to show their internal composition by giving an alternative, optional arrangement of the component forms, e.g. Ma ‘Alaa? ‘Is it ‘Ali?’ = Ma ‘Ali baa? (with the same meaning). In some combinations, however, there are no alternative optional arrangements of forms and the component forms can only be hypothesized on the grounds of structural analogy. We may say, for example, that Ma tukaa? ‘Is it a crow?’ = Ma tuke baa? (with the same meaning) on the ground of comparison of such analogous structures as:– Ma geed baa? ‘Is it a tree?’: Waa ge.ed ‘It is a tree’ = Ma tukgaa? ‘Is it a crow?’: Waa hike ‘It is a crow’ Such hypothetical breaking up of combinations of forms is indicated by placing the sign + between the hypothetical components.

^{page 575 note 2} dameerkiisii = dameer + kiis + kii (Noun + a Possessive Definitive + the Definite Article).

^{page 575 note 3} See n. 1, p. 570.

^{page 576 note 1} ugaaskii = ugaas kii (Noun and the Definite Article).

^{page 576 note 2} Berberuu = Berbera + buu = Berbera + baa + uu (Proper Name + the Syntactic Particle baa + a Subject Pronoun).

^{page 576 note 3} hilibkaa = hilib kaa (Noun and a Demonstrative).

^{page 576 note 4} Minimal pairs of this kind are not very frequent

^{page 576 note 5} Note that hel ig further differentiated from hel by the fact that the former has final glottalization and the latter final aspiration.

^{page 577 note 1} Seen. 2, p. 567.

^{page 577 note 2} Usually spelt Issa or Esa.

^{page 577 note 3} Usually spelt Gadabursi.

^{page 577 note 4} Usually spelt Darod. I have no adequate information about the Southern (Transjuban) Daarood.

^{page 577 note 5} Usually spelt Ha wiya.

^{page 577 note 6} These devices have been described in §§ 3–5.

^{page 577 note 7} cf. Daniel Jones' recommendation in his The Phoneme (Cambridge, 1950): ‘It is probable that a certain narrowing of transcription would be useful in teaching English learners to make proper use of vowel harmony in Somali’ (§ 675, p. 222).

^{page 577 note 8} Where a high degree of generality of spelling would be required.

^{page 577 note 9} Where some freedom of interpretation must be left to the readers.

^{page 577 note 10} This system of vowel representation would not be general enough for a practical orthography. Moreover, like Armstrong' system it would be too difficult for the general public (both Somali and non-Somali) to handle.

^{page 578 note 1} For the use of these graphic devices see § 7.

^{page 578 note 2} This is one of the basic problems of transcription in general. The broader the transcription the greater its generality and vice versa.

^{page 578 note 3} This system represents all the distinctions of meaning which depend on the differences in vowel series.

^{page 578 note 4} In the same way as the vowel points are sometimes used in Arabic in an otherwise unvocalized text.

^{page 578 note 5} For information about these two systems of spelling evolved by the Somalis the following works should be consulted: Cerulli, Enrico, ‘Tentativo Indigeno di Formare un Alfabeto Somalo’, Oriente Moderno, XII, Gennaio-Dieembre, 1932, pp. 212–3Google Scholar; Maino, M., ‘L'Alfabeto “Osmania” in Somalia’, Sassegna di Studi Etiopici, x, Gennaio-Dieembre, 1951, pp. 108–21;Google ScholarMaino, M., La Lingua Somala, Strumento d′Insegnamente Professionale, Alessandria (Italy), 1953 (Tipografia Ferrari, Occella e Co.);Google ScholarGalaal, M. H. I., ‘Arabic Script for Somali’, The Islamic Quarterly, i, 2, July 1954, pp. 114–8.Google Scholar

^{page 578 note 6} In the section entitled ‘ Grafemi “‘Ismaniya”’ in La Lingua Somala Strumento d'lnsegnamento Professional*, (pp. 33–4), M. Maino gives an account of the fact that the system of spelling which he describes recognizes ten vowel qualities in both short and long vowels by means of a diacritic, whenever this is necessary for eliminating an ambiguity. In practice this diacritic is very seldom used.

^{page 579 note 1} The symbol X corresponds to h as used by Reinisch (see n. 5, p. 567) and to kh as used by Bell (see n. 2, p. 570).

^{page 579 note 2} geela = geel + ka (Noun + the Definite Article).

^{page 580 note 1} The two men had one milking camel each and milked them each for himself, each drinking the milk of his own camel.

^{page 580 note 2} The word ‘cousin’ in Somali is often used as a term of familiar address among men even if they are not cousins.

^{page 580 note 3} i.e. the prospective pupil in the art of trickery.