Published online by Cambridge University Press: 22 April 2026
The Semitic languages of Ethiopia occupy a geographical domain in which Cushitic was and still is employed. When the Semites of South Arabia conquered Ethiopia, they imposed their Semitic language on this Cushitic domain. A stage of bilingualism followed, which still endures. The Cushitic group lost ground, most probably because of the political prestige of the conquering Semites, but not without exerting a great influence on the structure of the Semitic languages. In fact, the Semitic languages of Ethiopia are in some respects quite far removed from the typical Semitic pattern. Some of these differences are no doubt due to the internal development of Semitic in Ethiopia; others may be inherited from the Semitic group that came from South Arabia, concerning which we are not sufficiently informed; but many of them are certainly a result of Cushitic influence.
1 Part of the investigation of the South Ethiopic languages was carried on during my second trip to Ethiopia in 1950. The trip was made possible through a grant of the Social Science Research Council, to whom I wish to express my sincere thanks.
2 Abbreviations of book titles: Brockelmann, Grundriss = C. Brockelmann, Grundriss der vergleichenden Grammatik der semitischen Sprachen (Berlin, 1908–13); Cerulli = E. Cerulli, Studi etiopici, II: La lingua e la storia dei Sidamo (Rome, 1938); id., Harar = Studi etiopici, I: La lingua e la storia di Harar (Rome, 1936); Cohen, Etudes = Marcel Cohen, Études d‘éthiopien méridional (Paris, 1939); id., Nouvelles études = Nouvelles études d‘éthiopien méridional (Paris, 1939); id., Traité = Traité de langue amharique (Paris, 1936); Dillmann-Crichton = A. Dillmann, Ethiopic grammar, transl. by J. A. Crichton (London, 1907); Leslau, Doc. Gurage = W. Leslau, Ethiopic Documents: Gurage (New York, 1950); id., Doc. Tna. = Documents Tigrigna (Paris, 1941); id., Influence = The influence of Cushitic on the Semitic languages of Ethiopia, Word 1.59–82 (1945); Moreno = M. M. Moreno, Manuale di Sidamo (Milano, 1940); id., Azione del Cuscitico = L'Azione del cuscitico sul sistema morfologico delle lingue semitiche dell'Etiopia, Rassegna di studi etiopici 7.121–30 (1948); id., Burgi = Note di lingua burgi, Rivista degli studi orientali (RSO) 17.350–98 (1938).
Names of languages are abbreviated as follows: A, Aymallal; Amh, Amharic; C, Chaha; E, Eža; Ed, Endegeň; En, Ennemor; G, Gogot; M, Muher; Ms, Masqan; S, Selti; W, Wolane; Z, Zway.
3 C. Conti Rossini, Storia d'Etopia 91 ff. (Bergamo, 1928).
4 This division agrees with the one proposed by Cohen, Études 100 ff.; the details differ.
5 Grammatica della lingua galla 19 (Milano, 1939); Oriente Moderno 1938.54.
6 Cohen, Études 45 ff.
7 Conti Rossini, La langue des Kemant 39 (Vienna, 1912), thinks that the high plateau of Eritrea was occupied by the Kunama or Barya, peoples of non-Cushitic languages (see also RSO 1910.850). According to Cohen, Études 44, this explains the fact that North Ethiopic (especially Geez) is the group least exposed to Cushitic influence.
8 To cite only F. Praetorius, Beiträge zur Assyriologie 2.312 (1894); id., Zeitschrift der deutschen morgenländischen Gesellschaft 43.317–26 (1889), 47.385–94 (1893); Cohen, Études 44–6 and elsewhere; Cerulli, Harar 440–1 and elswhere. More recently, the present writer listed in Word 1.59–82 (1945) about thirty features in the domains of phonology, morphology, syntax, and vocabulary which show the influence of Cushitic on Ethiopic. M. M. Moreno in an article in Rassegna di studi etiopici 7.121–30 (1948), treated the effect of Cushitic on the morphological system of Ethiopic.
9 Conti Rossini, Etiopia e genti d'Etiopia 132–3 (Firenze, 1937).
10 Conti Rossini, Studi su popolazioni dell'Etiopia 153–4 (offprint); RSO 11.115 (1926–8).
11 The sources for Sidamo and Gurage are the following. For Sidamo, Cerulli and Moreno; for Kambatta, Alaba, Tembaro, Gudella, Qabena, personal investigation in 1950; for Gurage, C. Mondon-Vidailhet, La langue harari et les dialectes éthiopiens du Gourâghê (Paris, 1902); Cohen, Études; H. J. Polotsky, Études de grammaire gouragué, BSL 39.137–77 (1938); and Leslau, Doc. Gurage. Most of my information on the Gurage dialects as well as on Argobba and Gafat comes from personal investigation during my stay in Ethiopia in 1946–7 and in 1950.
12 It might be that ṃ (in the class of emphatics), cited by Cerulli 43, represents the sound described here.
13 As in the case of the preceding sound, tn and tl were not noted by Cerulli and Moreno. Possibly the sound ṇ, cited by Cerulli 43, represents tn.
14 Cerulli gives several examples of this alternation; add Alaba qäqiččo ‘few’: Qabena kəkiččo.
15 For
, see §4.
16 For ṭ: *, see §4.
17 Note also the alternation ṭ: t (see §3).
18 For q:
, see Cohen, Études 396.
19 k is weakened to h.
20 Studi etiopici 3.3.
21 For ṭ,
:', see §4.
22 Nouvelles études 40–1.
23 In Ǧanǧero, too, q
ḍ become a glottal stop (Cerulli, Studi etiopici 3.8).
24 The same phenomenon occurs in some Arabic dialects, but it is independent of the Ethiopic facts.
25 Conti Rossini, Oriente Moderno 28.113–4 (1948), reports on the basis of E. Rossi's documents that Tigre of the islands of Dahlak shows the loss of ṣ and
, and cites the examples 'aḥay 'sun' for ṣaḥay, ewo ‘salt’ for ṣewa, ifir ‘nail’ for
əfər, etc. What we probably have here is the change of ṣ,
to a glottal stop, rather than actual loss. The Tigre phenomenon would be independent of the one in Ennemor-Endegeň-Gyeto.
26 For d of Ennemor representing t of the other Ethiopic languages, see Leslau, Word 5.273 (1949).
27 Amharic has also ba'ela (Cohen, Nouvelles études 41).
28 For some examples of the change of ṭ,
to the glottal stop in Harari, see Cerulli, Harar §6. Cerulli also considers a Sidamo influence.
29 For q: k, see §3.
30 On the alteration of b in Amharic, see Cohen, Études 384 ff.
31 In the comparisons, not all the Gurage dialects are mentioned, nor are the etymologies given.
32 Polotsky, BSL 39.140 ff. (1938); Leslau, Doc. Gurage 13. For some examples with l in Chaha, see ibid. §1.4.
33 Cerulli, Studi etiopici 3, pp. 3, 4. Cerulli, ibid., 94, classes Ǧanǧero in Septentrional Sidamo, whereas Moreno, RSO 17.397 (1938), classes it with the language of Omo.
34 Archaic features in South Ethiopic, JAOS 71.219 (1951).
35 This is not the case in all the Sidamo languages. Kambatta expresses the feminine by the suffixed morpheme -t.
36 Dillmann-Crichton 279 ff.; for the geminated 2d radical in Geez, see E. Mittwoch, Die traditionelle Aussprache des Äthiopischen 27, 28 (Berlin, 1926).
37 Leslau, JAOS 65.182 (1945).
38 Note that in Tigre the form is räqiq (with simple q).
39 Leslau, Doc. Tna §38. My documents give only the form qättal for the masculine qättil, with a geminated 2d radical. This is not in agreement with P. Mauro da Leonessa, Grammatica analitica della lingua tigray 57 (Rome, 1928), who gives a masculine form qätil and feminine qätal, both with a simple 2d radical. In view of Tigre and the traditional pronunciation of Geez, the form is probably qättal. As for the masculine adjectives given by Mauro da Leonessa, märir ‘bitter’ agrees with Franceso da Bassano, Vocabolario tigray-italiano 82 (Rome, 1918), whereas the adjective for ‘red’ qäyəḥ is given as qäyyəḥ (with geminated y) in Bassano, ibid. 275.
40 Praetorius, Amharische Sprache 148 (with references); E. König, Neue Studien zur Schrift, Aussprache, und allgemeine Formenlehre des Äthiopischen 87–8 (Leipzig, 1877), considers the formation archaic; Dillmann-Crichton 279 explains it by ‘interpolation’ of the feminine ending a into the root; Brockelmann, Grundriss 1.345, 349.
41 Brockelmann, ibid. 345.
42 Agau is the ‘substratum’ language in the domain of Geez, Tigre, and Tigrinya.
43 In the Occidental group, Gogot alone seems to use the repetition of the last radical; cf. bora ‘ox’: pl. borarä; ämar ‘donkey’: ämararä. Tigrinya and Amharic also occasionally use the repetition of the last radical for the expression of the plural; cf. Tigrinya komorot: komoraru ‘onorario del prete o del fabbro ferraio’ (Mauro da Leonessa, Grammatica analitica 47); Amharic wåndəm: wåndəmamoč ‘brother’. On the plural, see also Moreno, Azione del Cuscitico 124.
44 Harari expresses appurtenance by the order determinant–determined without any particle: ṭay basar ‘the meat of the mutton’, lit. ‘mutton meat’. For the position of determinant before determined, see Leslau, Influence §26.3.
45 Sidamo expresses appurtenance by the order determinant–determined without any particle: handó ánni ‘the owner of the oxen’, lit. ‘oxen owner’ (Moreno 27).
46 Brockelmann, Grundriss 2.257.
47 Brockelmann, ibid. 2.228, 327.
48 See also Moreno, Appunti sulla lingua darasa, Rendiconti della Reale Accademia dei Lincei 1937. 215.
49 Leslau, JAOS 63.7 (1943).
50 Cohen, Le système verbal sémitique 75–138 (Paris, 1924).
51 There is also an impersonal ‘there is’: Geez bo, Tigre bu, both containing the preposition b ‘in’.
52 Geez alone has hallo as full verb with all the tenses and a causative 'a-halläwä. The other languages have only the form of the perfect with the meaning of the present 'he is'.
53 In Amharic the copula näw can also occasionally express existence: ənυe bäbetυe näñ ‘I am in my house’.
54 Cohen, Études 150, connects Muher nänä with Chaha närä and derives närä from nbr. The derivation of Chaha närä from Ethiopic nbr might be possible (in fact, the form yänbar is used for the jussive; but this may be a suppletive form). Since there is b-anä ‘he was’, y-anä ‘who is’, t-anä ‘while he is’ (all connected with allä ‘to be‘), and since Ennemor has anä and Gogot ino, one is tempted to connect Chaha närä, Muher näno, and Masqan nänä with allä plus a prefixed element n. Polotsky, JAOS 69.41 (1949), also doubts the connection of Chaha närä with nbr.
54a For the verb of existence in the negative or in tenses other than the present, other elements are used.
55 Dillmann-Crichton 497 ff.
56 Leslau, JAOS 65.193 (1945).
57 The copula of Gogot will illustrate the form. Sg. 3d masc. (ə)n, fem. na; 2d masc. nəhä, fem. näx; 1st com. näu̯h. Pl. 3d masc. nämw, fem. näma; 2d masc. nähəmw, fem. nähma; 1st com. nänä.
58 Geez occasionally uses the subject personal pronoun of the 3d person for all persons, but only in the singular; the pronoun of the plural is used for the plural.
59 Mondon-Vidailhet, La langue harari 101, connects it with Tigrinya 'əyyu.
60 Tigrinya also uses the negation 'ay- preceding the predicate complex: 'ay-ḥaddis 'əyyu 'he is not new'; but the copula 'əyyu changes in accordance with the person to be expressed: 'əyya 'she is', 'ək̠a 'thou art', and so on. Occasionally, the negation 'i- is also used in Tigre: 'i-wålye '(he is) not my son'; in this case the sentence is a nominal one.
61 The element yeš ‘there is’ is optional; thus, li or yeš li ‘I have’.
62 Dillmann-Crichton 407.
63 Leslau, JAOS 68.131 (1948).
64 For the influence of Cushitic on this formation, see Leslau, Influence §14. See also Conti Rossini, Proverbi, tradizioni e canzoni tigrine 33 (Verbania, 1942).
65 -u, -u̯ is probably weakened from -m, which is still used with the 3d pl. negative al-säbbäru-m.
66 For an-käfätäčə-da.
67 For an-käfätšə-da.
68 Probably for an-käfätku-da.
69 For Oriental Gurage and Harari, see §18. In Aymallal, the imperfect in the main clause has a final -u (thus yiqärsu), whereas in subordinate and relative clauses the form is without -u (thus yiqärs).
70 Relative clauses without a relative element occur also in Gafat, and occasionally in Amharic (Polotsky, JAOS 69.37 n. 8 [1949]).
71 That is, the simple imperfect with the verb allä ‘to be’, becoming al.
72 Note that the indicative has the compound -äl (in yədäbəläl) whereas the relative has al (in yədäbəlal). It is quite possible that -al of the relative imperfect results from *yä-äl (> *yal > al)—that is, the relative morpheme yä placed between the verb and the compound -äl. For yä- as relative element in Zway, see fn. 76; for the relative morpheme placed between the imperfect and the compound, see Harari above.
73 I did not record the relative imperfect in Wolane. The subordinate clause with the conjunction ‘until’ uses the compound imperfect (yəwådqan läǧigät ‘until he falls‘), whereas the other Ethiopic languages would use the simple imperfect here.
74 Note the place of the relative element zä- between the imperfect and the compound al; see fn. 72.
75 Sidamo has a simple and a compound imperfect (Moreno 44, 47).
76 The other Ethiopic languages use the simple imperfect in the negative. Zway also uses the simple imperfect here: yaydäbəl säb ‘the man who does not add’. Note the relative morpheme *yä-; for this morpheme in the positive relative imperfect, see fn. 72. The relative element yä- is also used in Zway in a subordinate negative clause, with the jussive: yaydäble hum ‘that he may not add’ (yä + negative jussive + conjunction e + hum). For the similar expression in Harari, see §19.
77 The origin of the negation a in a-ynägr-əm in Amharic and in the other Ethiopic languages is no doubt the morpheme *al- used in the perfect (see §16). The l of al is noticeable only with the 1st person sg. al-ənägər; before the prefix y of the 3d pérson, l disappears (a-ynägr-əm); it is assimilated to t of the 2d person (at-tənägr-əm).
78 It must be added, however, that in the subordinate negative the suffixed morpheme is most often omitted.
79 The Gafat negative imperfect: sg. 3d masc. tifärkam, fem. təttəfärkam, 2d masc. təttəfärkam, fem. təttəfärkiyam, 1st təlfärkam; pl. 3d com. tifärkimwam, 2d com. təttefärkimwam, 1st com. tənnəfärkam. It can be seen from these forms that the negation is təl-. For əl- as negation of the imperfect, see Selti-Wolane (§19); for the origin of t-, see Praetorius, Amharische Sprache 521, and Leslau, Gafat Documents 68.
80 The relative element is not expressed with the imperfect; see §17 and fn. 70.
81 The negative element in the main clause is the same in Gafat, Aymallal, and Zway.
82 The conjunction t is different from the t of the negative təl of the main clause, although etymologically it may be the same element (see fn. 79).
83 The sentence is Selti. For the expression of the relative imperfect in Selti and other Ethiopic languages, see §18.
84 Note that in the relative negative the compound imperfect is used, whereas in the subordinate negative the simple imperfect is used. Harari, too, makes a distinction between the subordinate clauses used with and without the relative element zä-.
85 For similar use in Zway, see §18.
86 See fn. 84.
87 The forms of Sidamo are those of the imperfect indicative with the negation. It seems to me that Cerulli 79 wrongly speaks about the negative subjunctive.
88 M. Cohen, Système verbal sémitique 243 ff. In post-Biblical Hebrew the imperfect has the value of the future, but occasionally it also expresses the present. Normally the present is expressed by the subject personal pronouns with the active participle ('ani koṭeb̠ 'I write', lit. ‘I [am] writing‘).
89 Leslau, Doc. Tna. 89–90.
90 JAOS 65.8 (1945).
91 šä represents the root for ‘wish’; see Mondon-Vidailhet, La langue harari 104.
92 The jussive with šä expresses certainty; the imperfect with te- expresses possibility, probability, doubt, or future intention. The opinion of Cohen, Système verbal sémitique 276 and Études 161, that the imperfect with te expresses an imminent action, is not correct.
93 L. Reinisch, Die Bilin-Sprache 33 ff. (Vienna, 1882).
94 Reinisch, Die Chamir-Sprache 55 (Vienna, 1884).
95 Sidamo, as described by Moreno and Cerulli, has no special form for the future.
96 For the lack of the relative element, see §17.
97 Leslau, Doc. Tna. §104b.
98 The other Ethiopic languages that have a special expression for the future (§20) express the relative and negative future in a different way. Ennemor uses a suffixed morpheme -ka for both the present and the future: yəkäftə-ka məs ‘the man who opens, will open’,
‘the man who does not open, will not open’ (the negation is
). In Tigre the negative future is expressed either by the simple imperfect or by the conjunction 'əgəl + jussive + 'ikon ('he is not'): 'i-fägər or 'dgəl ləfgär 'ikon 'he will not go out' (Leslau, JAOS 68.132 [1948]).
99 The statement of Cohen, Études 33, that qätali is used in all the Ethiopic languages, is not correct.
100 Tigre also uses wäda ‘make’ beside bela, but not for the expression of the causative (Leslau, JAOS 65.25 [1945]).
101 The verb adärrägä ‘make’ also serves to express a transitive verb, e.g. tuss alä and tuss adärrägä ‘pour’ (Cohen, Traité 265). The causative and the factitive of an active verb are expressed by the factitive assäññä. Cohen connects it with the Amharic verb meaning ‘name, have the name of, seem good’. It is to be connected rather with Zway sano ‘make’ (see fn. 106).
102 As in Amharic, aša is also used with an active verb: zäg aša ‘stretch’. The Harari verb aša is thought by Cerulli, Harar 400, to be taken from Sidamo.
103 Chaha also uses mänäm ‘be made’ for the intransitive verb (nəq mänäm 'act slowly') and amänäm for the transitive (quṭ amänäm 'gather').
104 Muher also uses mäññäm ‘be made’ for the intransitive verb: bərq bərq mäññäm ‘scintillate’.
105 For amäññäm.
106 It is tempting to connect it with Arabic ṣana'a ‘make’, but the order of the consonants is not the same: Arabic has ṣn', Zway has s'n (ṣ has become s). For Amharic, see fn. 101.
107 The conjunction and preposition seen in Selti ǧango ‘until’, Wolane lä-ǧigät, Zway lə-ǧəg, is no doubt the origin of the ancient Amharic preposition žig ‘until’ (ž alternating with ǧ is an Amharic feature); for other etymologies, see Praetorius, Amharische Sprache 274, and Cohen, Nouvelles études 309. These prepositions are to be derived from the root ‘arrive’: Selti-Wolane ǧeǧe, Zway ǧiǧi, Harari diǧa ‘come’. For the development of meaning, cp. Amharic dəräs ‘until’ from the root drs ‘arrive’.
108 Moreno 72 ff. For Galla, see Moreno, Grammatica della lingua galla 102 ff. (Milano 1939); for Bedja, see E. M. Roper, Tu Beḍawi 88 ff. (1928). For the influence of Cushitic on the postposition of relational elements, see Leslau, Influence §20.
109 The use is the same in Tigrinya (Leslau, Doc. Tna. 143).
110 For a recent study of calques and linguistic borrowing in general, see E. Haugen, Lg. 26.210–31 (1950).
111 A calque from Sidamo: Alaba läm foläta ‘conceive’, lit. ‘two souls’.