¹ This article was written during my stay in London, while enjoying the hospitality of the School of Oriental and African Studies. I am indebted to the John Simon Guggenheim Memorial Foundation which, through a generous Fellowship, made it possible for me to spend a year in the stimulating environment of SOAS.

² Rossini, Carlo Conti, ‘Noti sugli agaw. II. Appunti sulla lingua Awiyā del Danghelà’, Giornale della Società Asiatica Italiana, XVIII, 1905, 103–94.Google Scholar

³ Sponsored by the Near Eastern Center and the African Studies Center, University of California, Los Angeles.

⁴ Hetzron, Robert, The verbal system of Southern Agaw (Near Eastern Studies, 12), Berkeley, Los Angeles, University of California Press, 1969.Google Scholar

⁵ Hetzron, Robert, ‘The Agaw languages’, Afroasiatic Linguistics, III, 3, 1976.Google Scholar

⁶ e.g. ǝγoši ‘milk’ ῃf ‘ant’, etc. In several cases, the first consonant, the one following the ǝ, is γ or ῃ, but these may also occur initially. The word-initial occurrence of ǝ before a consonant cluster, on the other hand, is predictable and non-phonemic: ǝnkonu ‘tapeworm’, ǝcci ‘worm’, etc.

⁷ In the dialect studied by me, the low tone occurs only in the perfect definite endings, the pl. 2/3 ending of the imperfect definite, and in the interrogative particle -m.

⁸ It is remarkable that most of these words are borrowings from Amharic and that some of these end in -a in Amharic which may have sounded like a feminine to the Agaw borrowers, e.g. Amh. ṭäräpäza. This must be a recent borrowing because Amharic ṭ is represented by t here, and not by c as in the older borrowings.

⁹ This may be a survival from the time when -i was a singulative ending, a reconstruction suggested by comparative evidence. The exact declensional features of the short plural have not been investigated.

¹⁰ Note that Gafat, a Semitic language geographically close to Southern Agaw, has ǝrrašä for‘field’.

¹¹ It seems that this was originally a postposition. It is probably related to the noun ǝšew ‘heart’ with a figurative use as ‘inside’. -Šo may come from the translative case of this noun: ǝšewa *‘into’.

¹² For the relative, which has the same endings and formal behaviour as the genitive, see Hetzron, , The verbal system of Southern Agaw, §1.2.4.1.Google Scholar

¹³ The accusative should be used as a classificatory criterion, because it is always vocalic, whereas the genitive and the dative consist of only consonants after a vowel.

¹⁴ Curiously, the few nominals contained in this class and pattern, are all names of animals: dúr-i ‘cock’ (dúr-a ‘hen’), fís-i ‘mole’, múr-i ‘snake’, yínc-i ‘rat’. The next pattern is also represented by one noun denoting an animal.

¹⁵ The only example of this pattern attested.

¹⁶ Only gǝmb ‘stick’, kǝb ‘gourd’, and yiw ‘wrist’.

¹⁷ Note the fate of -u: bǝrǝndu+ô = bǝrǝndô. If the consonant preceding the -u is a velar or a uvular, they merge into a labio-velar or a labio-uvular before a suffix: bǝrčǝqu ‘glass’, fem. bǝrčǝq^wa pl. bǝrčǝq^w-ka, but acc. bǝrčǝq-ô (since before a labial vowel there is no distinction between q and q^w).

¹⁸ The only example of this pattern attested.

¹⁹ Seen. 17, above.

²⁰ Only šǝngúrč-i ‘onion’ and ǝnzrt-i ‘spin, loom’.

²¹ Only one more noun in this class: fiyél-í ‘goat’, another domestic animal.

²² Only three more examples: gséῃ ‘dog’, ménč ‘much, many’, and, potentially overlapping with class (E), néw ‘calf’ (pi. néwari, with no plural in -ka).

²³ One more example: sǝndáy ‘wheat’. Another nominal belongs to two classes: ǝmpl, fem, ǝmpl-á, pl. ǝmpǝl-ká, acc. ǝmp-o or ǝpl-ô ‘one’ (an adjective). By virtue of ǝmpl-o it belongs to this class, but ǝmpl-ô aligns it with sǝgl in (G).

²⁴ The only example of this pattern attested.

²⁵ See n. 23 for ǝmpl. What makes this word so special is the configuration -CCǝl which makes the ǝ deletion before a vowel optional. sǝglô most probably comes from *sǝg-o, i.e. an original (F).

²⁶ The only example of this pattern attested.

²⁷ See p. 130, n. 17.

²⁸ One more word in this pattern, the interrogative wóšin-í. This is a morphologically composite element (wó- interrogative, -iní adjectival formative), and címark^w-í may also be an original compound (cf. ǝrk^wí ‘tooth’).

²⁹ Two more examples: čáγ-a ‘bird’, dúr-a ‘hen’.

³⁰ One more: búnk-a ‘struggle’.

³¹ One more: zíban-a ‘female of dǝkkuli antelope’.

³² The only example of this pattern attested.

³³ One more: sén-a ‘sister’. Note that both of these nouns are derived from masculine nouns, bún ‘coffee’ and sén ‘brother’.

³⁴ The only example of this pattern attested.

³⁵ Two more: berdst-a ‘vessel’ and canqún-a ‘thirst’. The first one probably has an ǝ before the plural ending, but I have not recorded its tone.

³⁶ One more: wená ‘pregnant animal close to giving birth’.

³⁷ The only example of this pattern attested.

³⁸ Two more celestial bodies are found in this class and pattern: árf-á ‘moon’ and béW-á ‘star’. Four more nominals: cnc-á ‘fly’, dáb-á ‘wooden ball’, léw-á ‘right side’, séb-á ‘side-pain’, and šér-á ‘pregnant woman’.

³⁹ The only example of this pattern attested.

⁴⁰ Three more: tarénk-á ‘kind of antelope’ (I have not recorded the tone of the anaptyctic ǝ in the plural), tǝγ^wán-á ‘bug’, wizazr-á ‘lady’.

⁴¹ See Hetzron, R., ‘Agaw numerals and incongrnence in Semitic’, JSS, XII, 2, 1967, 169–97.Google Scholar

⁴² It is no wonder that the relatively closest masculine counterpart of cangéyá ‘left side’ is vocalic: cangíyí ‘left-handed’.

⁴³ Note that this compound is considered a plural in agreement rules: ǝmplaní láῃa araγúnášínà ‘once upon a time there were a husband and a wife’ (lit.: once two husband-wife were').

⁴⁴ y- before the accusative -a and the dative -ǝs, and yí- elsewhere.

⁴⁵ The endings -ji were added later, as a part of the ‘renewal of plural marking in pronouns’, see Hetzron, K., Ethiopian Semitic: studies in classification, Manchester, 1972, §T.2, and cf. n. 48.Google Scholar

⁴⁶ With the labial appendix before the accusative (kówa) and the dative (kús), but a mere k- elsewhere.

⁴⁷ See n. 45.

⁴⁸ The polite pronouns are survivals of the original plural pronouns, before the addition of -ji; see n. 45.

⁴⁹ ǝntu in isolation, ǝnt- before suffixes. I did not put these in two different columns, for subject and oblique, because this is a phonetic adjustment, whereas in sg. 1/2 two independent allomorphs are involved.

⁵⁰ See n. 45.

⁵¹ See n. 48.

⁵² Note that unlike in compounds, the final -a is dropped before the plural ending -ka. The -a is maintained in -γ^wa-ka.

⁵³ Probably derived from the root ǝtti- ‘fall’, as ‘fall-out’. This suggests that Amharic duqet for ‘flour’ may also come from the root wdq ‘fall’, i.e. *wǝdqet. The i is not the result of the closure rule (§1.4); the feminine is ǝttíná.