1

Cushitic
Maarten Mous
1. Geographical distribution and speakers
The Cushitic family consists of more than thirty languages spoken in
Northeastern and Eastern Africa. Ethnologue (Gordon 2005) mentions forty-seven
languages because it distinguishes six Oromo languages and six Somali languages.
The Cushitic languages fall into a number of groups: Beja; the Agaw or Central
Cushitic languages: Awngi, Bilin, Khamtanga, Kemant, Xamir and a number of
smaller endangered languages including those of the Falashas, the Ethiopian Jews; the
Highland East Cushitic languages: Kambaata-Timbaaro-Alaaba-Kabeena, Hadiyya-
Libido, Sidamo, Gedeo and Burji; the Lowland East Cushitic languages: Afar, Saho;
Oromo, Konso, Dirayta; Dhaasanac, Arbore, Elmolo; Yaaku; Bayso; Dullay (Harso-
Dopase, Gawwada-Gulango, Tsamakko); Rendille, Boni, Somali; and the South
Cushitic languages: Dahalo; Aasx, Kwadza, Burunge, Alagwa, Gorwa, and Iraqw. I
do not include the mixed language Maa/Mbugu in this overview. This is a language
classified as (South) Cushitic on the basis of its lexicon and its historical origin. The
people once spoke a Cushitic language, shifted to a Bantu language and tried to shift
back by creating a parallel lexicon with root forms from their original language, from
Maasai, from Gorwa and from their dominant Bantu language but manipulated in
form; see Mous (2003). Regardless of ones views on classification of this mixed
language, typologically the language is Bantu.
There is a separate chapter on Omotic languages. Omotic is not taken to be
part of Cushitic, as it is in some proposals for classification. This does not reflect a
point of view on the part of this author, who actually has no opinion on this issue. The
separation of Omotic and Cushitic makes the writing of the respective chapters
definitely more manageable. The debate on the classification of Omotic is dealt with
in the chapter on Omotic.
Many variants of language names exist in the literature. Only the most
common ones that may lead to confusion are given here with the variant that is used
in this paper first: Gedeo = Darasa, Oromo = Galla, Boni = Aweer, Dirayta = Gidole,
Yaaku = Mokogodo, Alagwa = Wasi, Aasx = Aramanik. The following are simply
spelling variants: Afar = Qafar = Afar, Khamtanga = Xamtanga, Bilin = Blin.
The northernmost and easternmost language is Beja, spoken in Sudan and
Eritrea. See map 1Other Cushitic languages spoken in Eritrea are Saho and Bilin; and
in Djibouti, Afar. The bulk of the Cushitic languages are spoken in Ethiopia. Kenya
has several Cushitic languages: Dahalo, Elmolo, Yaaku, and several dialects of
Oromo (Boraana, Orma, Waata). The southernmost Cushitic languages are spoken in
Tanzania: Aasx, (Kwadza),Alagwa, Burunge, Gorwa, Iraqw.
According to calculations by the Moscow school of glottochronology and
long-range comparison, the time-depth of Cushitic is deeper than that of any other
branch of Afroasiatic (Militariev 2005). Despite this time-depth, the Cushitic
languages are typologically relatively homogeneous. In the main Cushitic-speaking
area, the languages are in contact mainly with other related languages, Cushitic or
Ethio-Semitic. In the more southern areas, the languages are, and have been, in
contact with unrelated languages, mainly Bantu and Nilotic. There must have been
intensive contacts between Cushitic languages and Nilotic languages at several points
in time and space. In fact, Heine et al. (1979) postulated a now-extinct Cushitic
language, proto-Baz, on the basis of loan words in Nilotic languages, and similarly
now-extinct Cushitic languages have been hypothesized in the Taita hills of Kenya
2
(Ehret and Nurse 1981). Nurse (1988) and Ehret (1998) argue for a number of now-
extinct South Cushitic languages in Tanzania on the basis of loan word evidence in
Bantu languages. The intensive contact between Cushitic languages and Ethio-Semitic
languages in the highlands of Ethiopia has led to the proposal of an Ethiopic
Sprachbund (among others, Ferguson 1970). The validity of the Sprachbund and the
nature of the observed similarities has been critically discussed by Tosco (2000b).
The Cushitic languages vary greatly in number of speakers. The largest is
Oromo, with more than 20 million speakers according to Ali and Zaborski (1990), and
the second-largest is Somali, with more than 7 million speakers. Estimates of the
numbers of speakers for these largest Cushitic languages diverge immensely due to
political factors that are involved. The Oromo language has spread enormously in the
past and is spoken over a vast area in Ethiopia and Kenya. Oromo was for a long time
the lingua franca of Southern Ethiopia, but this role is now taken over by Amharic.
Somali is the official language of Somalia, but also spoken in all neighbouring
countries. Somali became a fully functioning official language used in education at all
levels in a remarkably short period of time (Laitin 1977). Somali and Oromo are both
taught in several places in the world and teaching material is available. Afar, Beja,
Hadiyya, Kambaata and Sidaamo have each about a million speakers or more. Awngi,
Khamtanga, Konso, Iraqw have 100,000 speakers or more. The other languages have
fewer speakers, and some of them are endangered: Yaaku and Elmolo in Kenya have
only a handful of very old (semi-) speakers left. In Tanzania, Kwadza is extinct and
the situation of Aasx is unclear (Winter 1979). The Agaw languages Kunfl and
Kailina (Appleyard 1996, 1999), and Kemant (Zelealem 2003) are highly endangered.
The Cushitic peoples traditionally depend on animal husbandry and
agriculture. Several groups such as the Afar, Arbore, Beja, Bayso, Dhaasanac, various
Oromo groups, Rendille, and Somali have specialised in transhumant animal
husbandry; others such as the Konso, Iraqw, Dullay, and Highland East Cushitic
groups have a highly developed agricultural economy. Some were specialised in
hunting: Dahalo, Yaaku, Aasx, Kwadza, Boni, and Elmolo (fishermen).
In Ethiopia the Ethiopic script is used for a number of Cushitic languages.
Oromo has opted for a Latin-based script. In Somali the choice was ultimately for a
Latin base too. The pharyngeal sounds pose problems for those languages that have
them. Somali uses c for the voiced pharyngeal fricative, as in the name Cali; Afar
uses q as in the language name Qafar; Iraqw uses /. In this chapter the official
orthographies of Somali and Oromo are used when the sources do so, but otherwise I
use for this sound. The voiceless pharyngeal is rendered x in Somali, c in Afar, hh in
Iraqw and in this chapter; except for where the source uses the official orthography.
I use double symbols for length and the following symbols for consonants: C for
ejectives and implossives, for the retroflex d, sh for the palatal fricative and for the
voiceless fricative; see the section 4.4 on consonants.

2. Internal classification
The present state of insight into the internal subclassification of Cushitic is given in
Tosco (2000a); see Figure 1. A historical overview of Cushitic classifications can be
found in Lamberti (1991). The main issues regarding Cushitic internal classification
are the position of South Cushitic and the relationship between Highland East
Cushitic and Central Cushitic (Agaw). Hetzron (1980) proposed the inclusion of
South Cushitic in East Cushitic on morphological grounds, and most experts would
agree. Ehret, for example, has changed his position on this; compare Ehret (1980)
3
with Ehret (1987). The actual position of South within East is difficult to establish due
to the lack of lexical correspondences (Kieling and Mous 2003) and also because of
the anomaly of the presence in South Cushitic of a lateral fricative and affricate which
is claimed to be cognate with Afroasiatic laterals (Dolgopolsky (1987), Takacs
(2000)). Kieling (2001) reviews Hetzrons arguments for the inclusion of South in
East on the face of new data and insights, specifically in South Cushitic. Hence in the
tree in Figure 1 South appears tentatively at the highest level or as a branch within
East Cushitic. The positions of Bayso and Yaaku within Lowland East Cushitic are
not clear, but the data for these languages are limited. Blaek and Tosco (1994) have
proposed a closer link between Dahalo and Yaaku. The classification of Burji as
Highland East Cushitic had been obscured by the various ways in which Burji
participates in the Dullay-Konsoid-Burji contact area (Sasse 1986).
Most of the debate on classification is about external matters: the genetic
relation of Omotic as separate or part of Cushitic (see chapter on Omotic) and the
position of Beja as a Cushitic language or a separate branch. In the Tsamay area there
is a language, Ongota (or Birale), that has so far escaped classification; see Sav and
Tosco (2000a).

Figure 1 here.

Figure 1: Cushitic family tree (Tosco 2000a)

Cushitic phonological and lexical reconstruction is very much in progress.
Lowland East Cushitic lexicon and phonology was reconstructed by Paul Black in his
unpublished PhD thesis, and the phonology of Proto East Cushitic was reconstructed
by Sasse (1979). Heine (1979) has reconstructed proto-Sam (Somali-Boni-Rendille).
South Cushitic has been reconstructed by Ehret (1980) and recently in more detail by
Kieling (2002) and Kieling and Mous (2003). The Highland East Cushitic lexicon
has been reconstructed by Hudson (1986). A reconstruction of the Agaw languages
has recently become available (Appleyard 2006). Blaek is working on an
etymological dictionary of Beja (Blaek 2003). Ehret has published a lexical and
phonological reconstruction of Cushitic (Ehret 1987), and Arvanites (1991), of the
glottalic consonants of Cushitic.

3. Scholarship on Cushitic
There is no comprehensive up-to-date bibliography of Cushitic languages. For Somali
one can use Lamberti (1986a) and M Diriye Abdulahhis website (Diriye 2002); for
Highland East Cushitic there is the Bibliography of Highland East Cushitic (Hudson
2006). Publications on Cushitic languages can be found in general African languages
journals such as the ones mentioned in the chapter on Chadic, and in the Journal of
Ethiopian Studies, and in the now-discontinued journals and series:
AfroAsiaticLanguages, Journal of Afro-Asiatic Languages, Rassegni di Studi
Etiopici. There are regular conferences on Cushitic and Omotic studies: Bonn
(Bechhaus-Gerst and Serzisko 1988), Torino, Berlin (Griefenow-Mewis and Voigt
1996), Leiden (Amha, Mous, Sav 2007) and Paris in 2008. Introductions and
overviews on Cushitic can be found in Palmer (1970), Hodge (1971), Flemming
(1976), Zaborski (1987), Sasse (1981, 1987a, 1987b), Cohen (1988), Tosco (2000a),
and Gragg (2001). Palmer (1970) gives an overview of the historical development of
Cushitic studies, and Zyhlarz (1956) explains the origin of the term Cushitic.
4
In the following list I give the most important descriptive works (grammar,
dictionary, texts) for each language. The list is selective and criteria vary by language.
More references do not mean better described. For most languages there is some data
available or will soon be available. There remain a number of languages that are in
need of description: Bayso, Bilin, Boni, Burji, Harso-Dopase in Dullay, and Bussa
(Mosiye); more detailed studies are definitely necessary for Arbore, Dahalo, Rendille
and Saho; for the highly endangered languages Aasx, Elmolo and Yaaku description
is very urgent or too late.

Beja: Roper 1928, Hudson 1964, Morin 1995, Wedekind, Wedekind and Musa 2003,
various articles by Vanhove.
Agaw as a whole: Hetzron 1976, various articles by Appleyard.
Awngi: Hetzron 1969; Kunfal
Bilin: various articles by Palmer, Kiflemariam Hamde 1993
Kemant: Zelealem 2003, Appleyard 1975
Khamtanga: various articles by Appleyard
Highland East Cushitic: Gebre-Tsadik et al. 1985
Hadiyya: Sim 1981, 1989 various articles by Plazikowsky-Brauner
Kambaata-Kabeena-Alaba: Crass 2005, Schneider-Blum 2006, Treis
(unpublished)
Sidamo: Moreno 1940, Gasparini 1983, Wedekind 1980, Anbessa Teferra
2000, Kawachi 2007.
Gedeo (=Darasa): Gasparini 1994, Wedekind 1990
Burji: Sasse 1982
Lowland East Cushitic:
Afar-Saho: Bliese 1981, Hayward 1998, Parker and Hayward 1985, various
articles by Hayward
Yaaku: Heine 1975
Dullay: Amborn et al. 1980, Sav 2005, Tosco (unpublished)
Oromoid:
Oromo: Ali and Zaborski 1990, Argaw and Philippson. 1988 [1991],
Baye 1987, 1988, Clamons 1992, Gragg 1982, Lloret 1988, Owens
1985a, Stroomer 1987, 1995.
Konso-Dirayta: Black and Shako Otto 1973, Black 1973, Bliese and
Sokka Gignarta 1986, Wondwosen 2006.
Omo-Tana:
Dhaasanac: Tosco 2001
Arbore: Hayward 1984a
Elmolo: Heine 1980
Bayso: Hayward 1978a, Lamberti and Haberland 1988
Rendille: Oomen 1978, 1981, Pillinger and Galboran 1999, Schlee
1978, Sim 1981
Boni: Heine 1982, Sasse 1980
Somali: Andrzejewski with Sheila Andrzejewski 1993, Andrzejewski
and Lewis 1964, Johnson 1974, Laitin 1977, Lamberti 1986a, 1986b,
1986c, Luling 1987, Puglielli et al. (eds.) 1985, 1998, Puglielli 1981,
1984, Saeed 1984, 1987, 1999, Serzisko 1984, Tosco 1997
South Cushitic:
Dahalo: Tosco 1991
5
Alagwa: Mous (unpubished).
Burunge: Kieling 1994
Iraqw: Mous 1993, Mous, Qorro and Kieling 2002, Berger and Kieling (ed.)
1998.

4. Phonology
4.1. Syllable structure and word structure constraints
Cushitic languages have open and closed syllables. Most languages do not allow an
empty onset; the onset is minimally filled with a glottal stop. Onsets are usually
simple and consist of one consonant. For example, in Arbore onsets and codas contain
maximally one consonant if we disregard glides and laryngeals (Hayward 1984a: 58).
The coda allows the same set of consonants as the onset and is also simple: either
empty or consisting of one consonant.
Word-internal consonant sequences are limited. They are either geminate
ambisyllabic consonants or sequences of two consonants that obey the sonority
hierarchy as in Arbore {glide/vibrant} < lateral < nasal < spirant < stop (Hayward
1984a: 60), and in Dhaasanac liquid < nasal < fricative < coronal stop < non-coronal
stop (Tosco 2001: 51-53); in Kabeena the first consonant of a word-initial series is
either the glottal stop or a sonorant (Crass 2005: 38).
A number of Highland East Cushitic languages and Bayso have sequences of
glottal stop and sonorant that are sometimes analysed as complex phonemes; see
Hudson (1989: 11ff), Cerulli (1938: 41), and sometimes as sequences of two
phonemes. Hayward (1978a: 543) discusses both analyses for Bayso and opts for the
complex phoneme solution on the basis of syllable structure and the variation in
phonetic realisation of the glottalized labial obstruent [b] ~ [p]. Kabeena has l, m,
n, r, y, which are analysed as consonant sequences by Crass (2005: 39). These
complexes generally arise through metathesis, but in Hadiyya l sequences arise
through dissimilation of l-t in which -t is the 2sg/3f agreement suffix; see Sim (1989:
14-15); however, non-derived l sequences exist as well.
Due to rich morphological systems, words are often long even if the roots are
relatively short. Vowel epenthesis and vowel deletion are common processes when
longer words are formed. Oromo (Lloret 1988), Somali (Saeed 1999: 26-27) and
Iraqw have a rule that deletes the vowel of a short syllable between two (short)
syllables if a morpheme boundary is involved, e.g. Iraqw gawid-en /difficult-PL/
gawden, lawala-u lawlu spears (Mous 1993: 30). Various languages have root
structure restrictions that involve the quality of the vowel (e or o versus i, a, u), length
of the vowels and accent. For example, in Oromo there are strong tendencies to
several co-occurrence restrictions on vowels in roots with the vowels being identical
or alternatively either V1 or V2 is a (Owens 1985a: 16-17). These restrictions are
violated at the word level and do not hold for loan words; thus they are more relevant
for language history than for synchronic analysis.
The minimal word is a heavy syllable in Iraqw, which does not have CV
words with a short vowel (Mous 1993: 26); Kabeena, however, allows CV words
with a short vowel (Crass 2005: 35).

6
4.2. Accent and tone
The Cushitic languages are accentual or restricted tone languages. The distinction
between tone languages and accent languages is not straightforward, because the term
accent stresses the organisation of prominence in the word and the term tone
stresses the physical realisation by pitch. Pillinger (1989) described Rendille as an
accent language underlyingly and a tone language with downdrift on the surface.
Somali has been analysed as a tonal-accent language by Hyman (1981); for Oromo
see Andrzejewski (1970), Owens (1980) and Banti (1988); for Afar, Hayward (1991).
There are a number of Cushitic languages in which tone does not play a role at all;
they are purely accentual. Crass (2005) describes Kabeena as a language with stress
on the ultimate or penultimate syllable depending on the status of the final vowel
(whispered or fully voiced). Other Highland East Cushitic languages are also stress-
accent languages. Whether stress, tone, or tonal-accent, for all Cushitic languages the
role of prominence at the lexical level is minimal, but the role in morphology is
considerable, also in the Highland East Cushitic languages; see Sasse (1981: 205) for
such a characterisation of Cushitic tone. Lexical minimal pairs in tone exist for Iraqw
and other South Cushitic languages but only marginally so, and partly because of a
tone rule that derives names. For example, in Iraqw there are minimal pairs such as
konkomo rooster and konkom insect sp., hlooro foam and hloor locust sp.;
these words for insects show qualities of names, Mous (1993: 21). The tonal systems
of the Agaw languages are fairly uncommon; Hetzron (1997: 483-489) describes
Awngi as having four tones in which High and Mid are the most important ones
compared to Low and High-Mid fall. Joswig (2006a: 17) reanalysed the Mid as a Low
tone and Hetzrons Low tone as a contextually conditioned variant of his Low tone
and Hetzrons Mid.
The role of tone in morphology and syntax is important. In Oromo, hin is a
negative morpheme but hn is a focus morpheme. Tone is essential in case marking, in
gender distinction, in verb conjugations, and so on. In Somali l with and la one,
someone and ku y ou but k in differ in tone only (Saeed 1999: 42-43); these
differences in tone are due to the fact that adpositional particles such as l with and
k in have a high tone and tone is linked to this grammatical category. In Iraqw all
definite nominal suffixes (possessives, demonstratives) are high toned (Mous 1993
21).
The Cushitic tone-accent languages are atypical accent languages in the sense
that not all words contain an accent (high tone). For example, in Somali, the preverbal
adverbial clitics wada together and kala apart are toneless. For some of the
monosyllabic toneless grammatical morphemes, one could argue that they are clitics.
However, there are also syntactic positions or functions that are marked by the fact
that they are toneless. In the Southern Cushitic languages the majority of the nouns
are toneless in all positions and receive tone only through suffixation of high-toned
morphemes; Hayward (1984a: 98) notes the same for Arbore.
For none of the Cushitic languages is tone distinctive on every syllable of the
word. Tone is distinctive on the final syllable(s). For example, for Somali, Saeed
(1999: 42) distinguishes three accentual patterns or melodies: High on the last mora
and Low elsewhere, High on the penultimate and Low elsewhere, and Low on all
moras. Once the tone falls in a word, it is rare that it can rise again in the same word.
But it does occur, for example, Arbore lkkutsut his hens (Hayward 1984a 99). It
also occurs in Somali when a high-toned suffix is added to a word with a high tone on
the penultimate, e.g. grigi the house (remote) (Saeed 1999: 43). High tones are at
7
the end of the lexeme (ultimate or penultimate) but due to suffixation with suffixes
that impose a High tone, a word may end in a series of High tones, for example, Iraqw
gajr--d-r isa /work-my-that-of yesterday/ that work of mine of yesterday.
There are grammatical morphemes that consist of a (change in) tone only. The
Beja first singular possessive is only a low tone and the third-person possessive only a
high tone in the underlying form (Appleyard 1991: 7, based on Hudson 1976). In
Somali, the subject case in many nouns is marked by removing the high tone and
genitive case by shifting the high tone to the final mora, for example dbi bull
absolutive dibi bull masculine nominative, dib bull genitive, bisd cat
absolutive, bisadi cat feminine nominative, bisd cat genitive (Saeed 1999: 44).
Banti (1988) analyses the tone distinctions as accent feeding tone: the underlying
accent is realized in the absolutive on the ultimate or penultimate mora, while the
genitive is characterised by an accentual pattern of ultimate accent and the nominative
has a number of allomorphs involving adding an (empty) mora to the end. In several
Cushitic languages gender is distinguished by tone differences only. This is valid for
Somali, e.g. nan boy, inn girl, nyl male lamb, nyl female lamb (Saeed
1999: 19), and similarly in Rendille nam boy, inm girl, mr bullock, mr
heiffer (Pillinger 1989); in Afar; see Appleyard 1991: 21-24) who suggests a role for
tone in gender marking in proto (Lowland) East Cushitic.
Certain grammatical suffixes require tonal changes on the preceding moras or
the high tone on the preceding syllable is part of the suffix, e.g. in Oromo the nominal
plural marker -le, magalaa--le magalale markets and adjectival plural
markers -a and -o (Owens 1985a: 93-94). Inherent tone on suffixes can have a
different effect on preceding tones and lower preceding high tones. This happens in
Beja (Hudson 1976: 101-102); for example, the high tone of 1sg n disappears in tam-
a-n-ee-k if I ate, compare tam-a-n I ate. The reverse also happens in Beja in ti-
df-a you went where the high tone of the root suppresses the accent of the past tense
suffix -`a. In many Cushitic languages the addition of high-toned interrogative
suffixes removes all preceding high tones, e.g. Somali gri-ke gurige /house-
which?:M/ which house (Saeed 1999: 43). This is also the case for Iraqw and
Alagwa (Mous unpublished) and it could well be a phonologized intonational pattern.
The possessive suffixes in Arbore take away any immediately preceding high tone but
not high tones that are separated by a toneless syllable/vowel, e.g. buur-h-sut
/porridge-M-his/ buuruhsut his porrridge, but lkku-t-sut /hens-F-his/
lkkutsut his hens (Hayward 1984a: 98-99).
The tone-bearing unit is the mora; this is, among others, shown for Somali
(Banti 1988a: 13; Saeed 1999: 41), and Dhaasanac (Tosco 2001 36).
In terms of historical development of tone/accent systems, Appleyard (1991)
has proposed that Highland East Cushitic lost its tone; Kieling (2002) has shown
how high tone (partly) developed in South Cushitic.
Hetzron (1997: 483) describes Awngi as having stress independent of tone.
Stress falls on the penultimate and is accompanied by a slight rise in pitch.

4.3. Vowel system
Cushitic languages typically have ten vowels, five long, five short, i, e, a, o, u.
A number of Cushitic languages have whispered vowels word- or clause-
finally. This is true for Oromo (Voigt 1984, Stroomer 1988, Lloret 1989, 1997), the
8
South Cushitic languages Burunge (Kieling 1994) and Alagwa (Mous unpublished),
and for Kabeena (Crass 2005).
The Agaw languages have no length opposition in their vowels, and they tend
to have a sixth vowel: Bilin and Kemant have , Kemant has an additional (Hetzron
1976: 12). Joswig (2006a) shows that the occurrence of Awngis sixth vowel is,
except for a limited number of exceptional words, predictable.
Somali has full tongue root advancing (ATR) vowel harmony. Vowels within
one word are either pronounced with or without ATR (see Kraska and Kim 992, Pia
1965, 1984).

4.4. Consonantal systems
The consonant systems are presented here in three tables: first, the system that Sasse
(1979) reconstructed for East Cushitic (see Table 1); second, the system of
Tsamakko with glottalised obstruents and pharyngeals (see Table 2); and third, that
of Afar with no glottalized obstruents but a retroflex d and pharyngeals (see Table 3).
9

Table 1: Proto East Cushitic consonants

t k
b d g
d d
1

k
f s sh h
m n
z
l
r
w y


Source: Sasse 1979.
10
Table 2: Tsamakko consonants

p t c k
b d g
b d g q
ts c
s x
z h
m n ()
l
r
w y

The voiced glottalized stops are implossive and the voiceless q' is ejective.
Source: Sav 2005.
Table 3: Afar consonants

t k
b d g

f s (sh) h
m n
l
r
w y


Source: Hayward 1974.
11
I use the following symbols for consonants: , , h, , and C for ejectives and
implossives, for the retroflex d, sh for the palatal fricative and for the voiceless
fricative.
The gap of the absence of a voiceless plossive p is common in Cushitic and
can be observed, among others, in Beja, Agaw, Sidamo, Gedeo, Somali, and Rendille.
A number of other languages do have p. Black (1974a) and Sasse (1979) did not
reconstruct p for Proto (Lowland) East Cushitic (see Svolacchia 1987 for an overview
of this phenomenon).
Most languages have glottal consonants and h. There is often complete
assimilation of vowels through the glottal stop, as in Oromo (Lloret 1995a: 62), Iraqw
(Mous 1993: 36-37), and historically in Cushitic (Sasse 1979: 53). In some languages
there is no opposition between two identical vowels separated by a glottal stop and a
long vowel, for example, Oromo (Stroomer 1988) and Gedeo (Wedekind 1980: 140;
1990: 129), where VV alternates with glottalised [V:] in fast speech.
Several languages also have the pharyngeal fricatives and . These are Bilin
as only Agaw language, Afar, Somali, Rendille, Dullay (Dobase, Gollango,
Tsamakko), and the South Cushitic languages. The pharyngeal fricatives are absent
in the Agaw languages (except for Bilin), in Highland East Cushitic, in Dhaasanac-
Arbore-Elmolo, Yaaku, Oromo-Konso-Gidole, and Boni. The voiced pharyngeal is
sometimes realised with glottal closure. This is the case in Dullay (Hayward and
Hayward 1989: 183 n.10) including Tsamakko. In Iraqw there is no closure but
creaky voice (Mous 1993: 18).
Pharyngeal consonants have a centralizing effect on neighbouring vowels; in
Rendille for example, all vowels are centralised in the environment of a
pharyngealised consonant (, g, k, x, ) evidenced by an increase in F1 and a slight
increase of F2 (Esser 1991: 147-148). Dullay a is fronted in the environment of ,
and to a lesser extent q (Amborn et al. 1980: 67); according to Hayward and Hayward
(1989: 183) this extents to the glottal consonants as well.
Cushitic languages provide arguments for a feature grouping guttural
consonants together (see Hayward and Hayward 1989). The pharyngeal and glottal
consonants and to some extent also the uvular stop and fricative behave similarly in a
number of ways. In Afar consonants in one root are either identical or non-
homorganic; , , and h count as homorganic. In Iraqw there is morphophonological
vowel assimilation that applies to the vowel of the (final) verbal derivation which
assimilates completely to a primary vowel i, a or u in the syllable preceding it
provided that the intervening consonant is a guttural, a uvular stop or in some cases
even a velar fricative (Hulst and Mous 1992).
Cushitic languages typically have glottalic consonants: implossives, ejectives
or both. In Konso the opposition in stops is primarily along the lines of glottalic
versus pulmonic; it has four implossive consonants b, d, j and g which are
devoiced when geminated and four pulmonic stops p, t, c, k which are voiced
whenever a vowel follows. Neighbouring Burji has a voiceless ejective series p, t,
c, k plus an implossive d. Dhaasanac has the four implosives which are devoiced
word-finally and realised with egressive air stream mechanism (Tosco 2001: 19).
Oromo has p, t, k and d. Lloret (1995a) shows that Oromo d differs
phonologically from the ejectives and behaves like a glottal stop or a plain voiceless
stop in many respects: geminated dd alternates with plain t in verbal conjugation;
Lloret proposes t as underlying form and mutual assimilation. Lloret (1995b)
proposes a feature analysis (using feature geometry) for Oromo in which glottalic
12
consonants are specified for constricted glottis but not for voice, and implossives
differ from plain stops only in terms of [constricted voice] but not in voice; ejectives
have pharyngeal as second place of articulation in addition to their oral place of
articulation. The Cushitic languages that have ejectives in the four places of
articulation tend not to have pharyngeals; this is true for the Highland East Cushitic
languages; the Agaw languages except for Bilin, Arbore, Oromo. Konso, Dhaasanac,
and Boni have implossives where the others have ejectives and no pharyngeals. The
languages that have pharyngeals in their inventory, e.g. Afar, Rendille and Somali, do
not have ejectives/implossives in their inventory. Separate from this is the presence of
an implossive (Oromo, Arbore) or pulmonic retroflex d (Afar, Rendille, Somali). Note
that the parameter glottalic consonants versus pharyngeals does not align with genetic
units. South Cushitic has pharyngeals and ejective affricates ts and t but no other
implossive/ejective stops. The ejective lateral affricate is cognate with d
1
in Sasses
(1979) reconstruction (see Kieling and Mous 2003). The bilabial glottalic stop in
Arbore is a devoiced or voiceless implossive (Hayward 1984a: 53). The Cushitic
languages do not seem to have constraints on co-occurrence of glottalic consonants in
one root (Wedekind 1990).
Gemination of consonants is common. Many languages have gemination as a
morphological process. For example, Konso geminates the final root consonant of a
verb to form a singulative verb stem, and Gedeo uses final gemination for imperative
plurals (Wedekind 1990: 51). In those languages that have gemination as a
morphological process, there is usually no restriction and any consonant occurs long
and short; in Konso all consonants occur long; in Kabeena all but the glottal
consonants (Crass 2005: 37), in Oromo all except h (Lloret 1997: 499). Many
geminate consonants arise through assimilation of consonants that come together in
morphological concatenation, e.g. Oromo laal-ne laalle we watched, moor-nii
moorri fat:nominative, did-te didde you refused (Owens 1985a: 22). This is
also the case in the CVC- reduplication in Dhaasanac (Tosco 2001: 46-48).
Dhaasanac has also prosodically conditioned gemination: the consonant after a short
open syllable or after a diphthong is geminated, thus
h
agsu hurry! is realised as
h
aggso (the initial
h
is phonetic), and guoro-m-i as guorrome I was tired.
Degemination occurs too. In Arbore geminates undergo degemination when
followed by another consonant or a word boundary, e.g. hw steer is underlyingly
hww, and iy heyt
h
e she became replete from /heyy-t-e/ (Hayward 1984a: 62-63). In
Dhaasanac prosodically defined degemination occurs optionally when the following
syllable also starts in a geminate (Tosco 2001: 49-50).

4.5. Metathesis
Metathesis is relatively common in consonant clusters, but the extend to which it
occurs differs from language to language. In Dirayta metathesis has occurred in its
recent history in order to fulfill the sonority conditions in consonant clusters:
(semivowels <) liquids < nasals, fricatives < stops; dilk elbow against related
Konso dikla; tak honey against Konso takm (Black 1974b). Burji has metathesis
of causative s and preceding stop (Sasse and Straube 1977: 249). In Rendille there is
consonant metathesis of the final consonant and the consonant of the preceding
syllable plus vowel drop before the plural suffix -; the final consonants involved are
r but also b and , e.g. bab-o bab armpits, ta-o ukt goatskins, ugr-
o urg skinbags (Oomen 1981: 50).
13
The language in which metathesis is synchronically most productive is
Sidamo, in particular in the verb conjugation; consider the following verb forms with
the 1pl ending -inemmo: amad-inemmo amandemmo we grasp, gat-inemmo
gantemmo we are left over, got-inemmo gontemmo we sleep, etc. (see
Yri (1990: 35), also Murray and Vennemann (1982)).
Bilin has various consonant alternations, e.g. devoicing and/or spirantization
of root stops in derived singulars, lx-a, singulative of lk fire, drgum-a, singulative
from drkum sycamore. Fallon (2006) unifies these alternations as consonant
mutation: Blin displays a variety of mutation processes, all of which appear to be
morphologically (or lexically) determined. The mutations involve the features [voice],
[continuant], [sonorant] and [lateral], as well as complex mutations involving
combinations of these features. Some of these mutations may have originally been the
result of lenition processes induced by affixation of a vowel-initial suffix (Fallon
2006: 117).

4.6. Reduplication
Reduplication occurs lexically and as a grammatical process. The former is
presumably often the result of the latter. Grammatical reduplication includes plural
formation in nouns, frequentative on verbs and habitual on verbs. Repetition (which I
do not consider a phonological process) of a word is used for distributive meaning, for
example, the repetition of a number in (1). In Kabeena the repetition of a modifier
indicates maximal validity of the characteristic (Crass 2005: 291). Ideophones often
display expressive repetition. But ideophones also show reduplication. In Somali
reduplication and insertion of l is common in ideophones, e.g. malaf ~ maf ideophone
for wipe out, exterminate (Salaad and Tosco 1998: 127).

(1) qaacc-e=ma dookko dookko baddam-inki
bush-P=to/in one.M one.M hide.oneself-3PLCONS.A
One by one they hid themselves in the bush. Tsamakko (Sav 2005: 79)

In Alagwa, lexical roots show the same three types of reduplication that occur
as grammatical processes. Reduplication of the final root consonant, which is frequent
in derivations, is rare in the lexical domain, but in the lexical domain we see
reduplications of medial root consonants, lug<aag>-o tree sp., 'om<om>oro ant
sp., sar<ar>aakwi tree sp.. The reduplication of the initial CV- and initial CVC- of
the root is more common in the lexicon than as a morphological process. As a
morphological process the CVC- reduplication involves an epenthetic vowel a or i,
but lexically Alagwa also has initial CVCV reduplication which does not occur as a
process, e.g. sungu-sungu-moo ant sp., kuti-kuti ~ kut-kuti puppy, and CVC
reduplication in which there is no epenthetic vowel, e.g. ar-ara termite. Some
lexical reduplications involving roots that have an r, l or s as final root consonant add
an epenthetic homorganic nasal, buru-m-bur-moo worm, siili-n-siil-imoo bird sp.,
giri-n-giri-t be round; and with CV- reduplication bo-m-boqoori calabash for
divination stones, kwi-n-kwiisi epilepsy. Iraqw has tsur-uun-tsuur (v) gather at one
place in disorder, kwi-n-kwir harlequin quail (a bird species), din-dirmo small
hill, xwaa--xwaa bridge of the nose, pu'-uum-pu'i circle, mu-u-n-r Icuna
bean. There are also lexical CVCV- reduplications that replace C
2
in the reduplicant
14
by r, or CV-r-epenthetic vowel reduplications, uru-utli arrow with ornaments,
kara-(n)-kaaa palate.
Below I discuss seven reduplication processes that have been repeatedly
reported for Cushitic languages. Variants of these and others occur as well.
Gemination of the final consonant could be considered a special type of reduplication,
but I discuss gemination separately. The epenthetic vowel is given here as a. This is
the most commonly used epenthetic vowel in reduplication; outside reduplication, the
most common vowel for epenthesis in Cushitic is i; this latter vowel also occurs in
reduplication, sometimes with a difference in meaning. The epenthetic vowel will
undergo the regular vowel assimilation processes of the language, e.g. Iraqw eeees
from eees to finish by assimilation through a guttural consonant. Different
languages may use one particular type of reduplication for different functions; for
example, final reduplication is used for nominal pluratives in Somali and Alagwa but
for habitual verb forms in Iraqw. Tendencies for a particular type of reduplication to
be used for a specific function are mentioned in the discussion of that type of
reduplication. The choice between reduplication types is functionally and lexically
rather than phonologically determined. Reduplication may provide hints whether a
segment is complex or not. For example, in Alagwa we see that only the velar stop
part of rounded velars is reduplicated, but the evidence is inconclusive as to whether
Cw is a unit or a sequence because it is followed by a round suffix vowel which may
have absorbed the reduplicated rounding; similarly, in the reduplication of
prenasalized stops, only the oral part is reduplicated.

1. C
1
V
1
(V
1
)-.
The V
1
may be lengthened in the reduplicant. This type of reduplication is
indistinguishable from type 2 when V
1
is a or identical to the epenthetic vowel.

Examples: Boni has frequentative sisii from sii give and duduud from duud
consider (Heine 1977: 280-281). Dhaasanac has fafa from f (Tosco 2001: 142).

2. C
1
a-.
Iraqw has a frequentative reduplication for which most examples are indecisive as to
whether the vowel is reduplicated or epenthetic a; in the case of tatumbiim it has to be
epenthetic and in the case of gogoow it has to be reduplicated (type 1), but in all other
cases in (2) below V1 is either a or the vowel in question could have undergone
assimilation through a guttural consonant or l (Mous 1993: 31-33, 180-183)

(2) Iraqw C
1
a- reduplication for frequentative
tatumbiim tumbiim to splash in water
mamaw maw to leave
aaw aw to get
papaa paa to push aside, pass
aakuut akuut to jump
ooos oos to excrete
eeees eees to finish
gogoow goow to flee

3. C
1
V
1
C
1
-
15
The V
1
is usually shortened. The second radical of the reduplicated form becomes a
geminate provided C
1
is admissible as geminate. This type is used in verb roots for
frequentative (see section 28.4). Note that Tosco (2001: 46-48) analyses this as a
subtype of type 4 with reduction of the second consonant and compensatory
gemination.

4. C
1
V
1
C
2
-.
The V
1
is usually shortened. This is indistinguishable from type 3 for those C
2
C
1

combinations in the second radical of the reduplicated form that show full
assimilation. This type is often used on verb roots for the frequentative. Examples:
Rendille has furfura from fura be open (Pillinger and Galboran 1999: 33). Iraqw has
kum-kumiit from kuumiit to continue, aawaw-aw from aaw to waste time
(Mous 1993: 182).

5. C
1
V
1
C
2
-a.
The V
1
is usually shortened. The extra vowel a is epenthetic and may assimilate
according to the assimilation rules of the language. This type is difficult to distinguish
from a type C
1
V
1
C
2
V
2
-. However, so far I have not found an instance of an
indisputable C
1
V
1
C
2
V
2
- reduplication. Alagwa has tsanq-a-tsanqas from tsaanqas
wet a bit, dah-a-dah from dah enter, ag-a-ag from ag eat, al-a-al from al
decorate, am-a-am from am be lost, disappear, dead, and possibly with
assimilation of the epenthetic vowel un-u-unuus from unuus chase at high speed,
tso-o-tsoomit from tsoomit be distressed, struggle. In fact, the Alagwa examples
do not allow us to decide whether the reduplication is C
1
V
1
C
2
-a or C
1
V
1
C
2
V
2
-. But
Iraqw has huw-a-huw from huw to bring in which the a must be epenthetic, and
Tsamakko has this type of reduplication lexically in nouns, e.g. kutt-a-kutt-o (m)
small braid, dang-a-dangacc-o (m) porcupine, and in the imperfective derivation
in verbs (3). In this productive process, C
2
should be C
2
C
3
, as the complex codas of
monosyllabic roots are reduplicated, as are the geminate consonants in that position,
including those that arose from the productive derivational process of gemination of
the final consonant for singulative action, e.g. bitt-a-bittam to keep on buying one
thing.

(3) Tsamakko imperfective reduplication
el to drop el-a-el to keep on dropping
gere to steal ger-a-gere to steal continuously
bitam to buy bit-a-bitam to keep on buying
azaz to order az-a-azaz to keep on ordering
ziir to extract ziir-a-ziir to keep on extracting
upp to blow upp-a-upp to keep on blowing
aw to be ripe aw-a-aw to be ripe for a long time
baqqal to sprout at once baqq-a-baqqal to sprout continuously
bittam to buy one thing bitt-a-bittam to keep on buying one thing
(Sav 2005: 188-189)

6. -aC
f

This form is used predominantly for number formation in nouns; in which case it is
often followed by a vowel number suffix or terminal vowel. It is also used for the
multiple reference form of adjectives.
16
Somali has -aC
f
as one of its plurative formations, e.g. f-af mouths,
languages, dab-ab fires, san-n noses (Saeed 1991: 61). Iraqw has a rare
singulative suffix -aaC
f
-i (f) for collectives, e.g. kwa-aa-i (f) bead, singular of
kwau (p), bal-aal-i (f) cob of grain, singular of balangw (m), war-aar-i (f) seed,
singular of warangw (m). Alagwa has several plurative suffixes involving
reduplication of the root-final consonant with a preceding epenthetic vowel a with
variable vowel length and followed by an additional plurative vowel suffix.

(4) Alagwa reduplication of final consonant in pluratives
a. The plural suffix -a(a)C
f
-ee (f)
Base (sg) Meaning Plural
deesu (m) snake deesasee (f)
milambu (m) trough mlambabee (f)
qaambu (m) tree sp. qaambabee (f)
saalu (m) leather loincloth for men saalalee (f)
waqaantu (m) entrails waqaantaatee (f)
b The plural suffix -aC
f
-aa (f)
siraa (f) anus siraraa (p)
kwari (m) year kwararaa (p)
qameesu (m) pot shell qameesasaa (f)
gaamu (m) thing gaamamu (p)
filu (m) aardvark filalu (p)
c. The plural suffix -aC
f
-u (p)
gibisa (f) diaphragm gibisasu (p)
kat (f) hump (of cattle) katatu (p)
kebi (f) hearthstone kebabu (p)
qubee (f) tortoise qubabu (p)
d. When C
f
is a rounded velar
tligw (f) caterpillar sp. tligwagu (p)
yakwa (f) calabash, container yakwaku (p)
senko (f) adze, sickle senkoku (p)
e. The plural suffix -aC
f
-a'u (p)
lipampi (f) calabash for milk or grains limpapa'u (p)
tsini (f) point tsinaana'u (p)

5. Morphological typology and processes
Cushitic languages show all types of morphological processes: suffixation,
prefixation, infixation, ablaut, stem alternation, reduplication, and tonal marking. The
languages are usually rich in morphology, and grammatical categories are realised by
(segmental and suprasegmental) morphemes. Suffixation dominates over prefixation
and infixation is rare. Kieling (2003) discusses how infixation arose historically in
South Cushitic through the combination of more common processes such as
suffixation and reduplication in combination with historical changes. Sequences of
suffixes are not uncommon, for example, gadye-r-e-d-r is /work-F-mine-that:of-F
yesterday/ that work of mine of yesterday (Iraqw, Mous (1993: 230)).

17
6. Lexical categories
Nouns and verbs are clear-cut categories. Adjectives are less clear as an independent
class in some Cushitic languages. For Somali, there is a debate whether the category
of adjectives is an independent word category. Afars adjectives are stative verbs,
whereas Tsamakkos adjectives are nouns. Other lexical categories that are
recognized are postpositions, conjunctions, and ideophones. Derivational processes
are mainly from verb to noun; the processes from noun to verb overlap with verbal
derivation.

7. Nominalisation
Cushitic languages may have many different deverbal nominalisation suffixes.
Kabeena has twenty-eight (Crass 2005: 71), Iraqw has twenty (Mous 1993), Sidaamo
has close to twenty (Anbessa 2000: 67-69). A limited number of these are fully
productive and some have clear specific semantics. A functionally common
nominalisation consists of the agentive noun, for example, Sidaamo hatasr-aano
butcher from hatar to butcher (Anbessa 2000: 67). Other nominalisations include
instrument nouns and abstract nouns. Sidaamo uses the same -aano suffix for
instruments, e.g. fey-aano broom from fey sweep (Anbessa 2000: 67), and a
suffix -imma, -ima for abstract nouns, e.g. but-ima poverty from but be poor,
gedd-imma old age from geedd grow old (Anbessa 2000:7); this suffix is also
used to nominalize adjectives, e.g. dan-imma goodness form dana good
(Anbessa 2000: 69). Several nominalisations of the same verb can co-occur, e.g.
Iraqw faara, faaro counting from faar count, where the derivation in -a is
completely productive but refers to a specific act of counting while the in -o is
lexically restricted and refers to counting in general (Mous 1993: 75).

8. Nouns
Nouns have the grammatical property of gender. Gender is defined on the basis of
agreement. The number of agreement classes is either two or three. For those where it
is three, the third one interrelates with the category of number; there are alternative
analyses reducing these systems to a two-gender system. Thus the analysis of gender
is closely linked to the analysis of number. Number is derivational in nature, and most
languages have a variety of number derivation for both plural and singular.
Definiteness is less generally marked and interrelates with case marking. Case
marking is of the marked nominative type, that is, the subject (both subject of
intransitive and agent of transitive) is marked and the object is in the unmarked
citation form but not in the Agaw languages where the accusative is marked. There
are no other nominal categories. Pronouns and numbers can be considered special
nouns but are discussed separately (see sections 23 and 19). The final vowel of nouns
drops under certain circumstances and can be argued to be not part of the noun, see 10
on terminal vowels.

9. Nominal number
Number is not an obligatory category. One can use an underived basic form of the
noun that is neutral for number in situations where the specification of number is
considered irrelevant. There is number agreement in the subject marking on the verb,
but for several languages, agreement on the verb is with gender and not with number.
This is the case in Iraqw, for example (Mous 1993). Within the noun phrase there is
number agreement on the adjective. Sometimes this agreement is semantic rather than
18
morphological. For example, in Iraqw one can say noto r /paper.money (notes) big/
a lot of money or noto ur-n /paper.money (notes) big-PL/ large denomination
notes with a distributive reading for the plural adjective. Most languages have more
than one plural derivation and also singular derivation. The forms of the number
derivation diverge greatly within Cushitic (see Zaborski 1986 for an overview).
Number derivations impose gender on the noun. The choice of the number derivation
is lexically determined but partly correlates with the gender of the base noun. As a
consequence there is a phenomenon called polar gender, meaning that the plural
form has the opposite gender of the singular form. The derivational nature of number
is also evident from the fact that lexemes vary in the number of number forms that
they have and in the nature of their interrelatedness. For example, in Tsamakko (Sav
2005: 61-65) we can have a lexeme like zilanqa (f) rainbow which has only one
form and one like gurl-o (m) cat which has the following number derivations,
singular: gurl-itt-o (m), gurl-itt-e (f); plural: gurl-ad'd'-e (p). In the following I will
discuss these number properties in a bit more detail.
Mass, collective and/or transnumeral nouns can be distinguished on the basis
of their morphosyntactic properties. In Oromo uncountable nouns such as water do
not take numeral modifiers (Owens 1985a: 94), and in Somali transnumeral nouns
need a relative clause to count two that is orange(s) (Saeed 1999: 58). The small set
of transnumeral nouns have only one number form, but that can refer to an individual
(singular or plural), the substance, or a collective (Serzisko 1992). In Somali one has
to distinguish countable nouns, mass nouns, collective nouns, and transnumeral
nouns. Only countable nouns regularly form plurals. In the construction Number +
Noun in which Number is the head, the countable noun is not in the plural, e.g. lab
kob two cups (Saeed 1999: 56-57).
The feature number has two categories based on agreement: singular
reference and multiple reference. In Bayso (Hayward 1978a) there is an extra
category of paucal reference. Because plural will be used as a value for gender, I
follow the terminology that Hayward (1984) has suggested: multiple reference and
singulative reference. Derivationally nouns can be of three sorts: base, derived plurals
or pluratives, and derived singulars or singulatives. The base is often semantically
neutral for number and is used when number is irrelevant (see for example Sav 2005:
47) for Tsamakko, Crass (2005: 63) for Kabeena. In Oromo most nouns do not have
derived plurals.
External agreement with number is shown in subject agreement of the verb.
Some languages have words that are singular in meaning but require plural agreement
on the verb. For those, plural is considered to be an exponent of gender, and
agreement on the verb is with gender only. I will come back to that after the
discussion of gender. For languages with only two exponents of gender, subject
number agreement on the verb can be lexically determined. For example, in Somali,
mass nouns have either singular or plural agreement on the verb depending on the
lexeme; those that require plural agreement end in , which is a plural suffix (Saeed
1999: 57). In many languages there are agreement-reduced verb forms that do not
show number (and gender) agreement. There is no agreement with the object in
number. Internal agreement includes agreement on adjectives. Adjectives show plural
agreement through initial reduplication, for example, in Somali (Saeed 1999: 108), in
South Cushitic (Kieling 2002, Mous 1993, unpublished), in Oromo (Owens 1985a :
87, 93), but number agreement is not strictly obligatory. Plural suffixation occurs in
19
Oromo. Other modifiers such as demonstratives and possessives do not to show
number agreement.
Plural derivations vary greatly in form: reduplication of final consonant (Bilin,
Somali, South Cushitic, Konso, and Rendille), gemination (Bilin, Arbore, Dhaasanac,
Kabeena, Konso, and Rendille), change of stem vowel (i.e. broken plural or ablaut)
(South Cushitic), and suffixation with various shapes of suffixes: -V, -(V)C(C)V, etc.
An example of infixation is the Iraqw plural formation <ee>_i for which the vowel ee
is infixed before the final root consonant and the root is followed by a suffix i, e.g.
digeemi boundaries derived from digma; this combination of infix and suffix is an
allomorph of the plural suffix -eeri used with three consonantal stems, and its shape is
explainable by a preferred light-heavy syllabic pattern for the plural (Mous 1993: 53).
Many East Cushitic languages have four to six different plural formations
(Oromo, Somali, Konso, Dhaasanac, Tsamakko, Bayso, and Kabeena). Arbore has
more than ten, and the South Cushitic languages have even more. In several languages
there is irregular allomorphy (or similarity in plural formatives) involving length of
the vowel or consonant of the plural formative (Dhaasanac, South Cushitic, and
Konso). For example Dhaasanac has plurals in -a(a)m: deger barren, plural: deger-
aam; kur knee, plural: kurr-am; fuoc-u bride-wealth, plural: fuoc-am (Tosco 2001:
86-88). Suppletive plurals typically occur for the following lexemes: women,
cattle, goats, people, sisters, children, uncles in Burunge (Kieling 1994:
60).
The choice of the plural formative is lexically determined, but often
correlations with the base have been observed in terms of the following properties of
the base: gender, quality of the final vowel, presence of a particular singular suffix,
syllabic structure, and accent type of the base noun. Thus, for Somali, number
derivations have been described in terms of declensions where each declension is
defined by inter alia (i) whether there is gender polarity between singular and plural,
(ii) the form of the plural suffix, (iii) the accent pattern in singular and plural and,
sometimes, (iv) the final vowel of the base (Andrzejewski 1960). In Arbore multiple
reference suffixes that end in o are all plural and have a feminine base. In Rendille
plurals in -aC (p) and -Ce (p) are restricted to polysyllabic masculine bases and
plurals in - to feminine bases, with some exceptions. In Khamtanga there are several
different plural formations, the most common being -tan; other formations include
drop of final vowel a, change of consonant, and gemination of the final root
consonant (Appleyard 1987a). The other Agaw languages have similar complex
number formation, e.g. Bilin (Palmer 1958). All languages for which we have dialect
information show regional variation in choice of plural marker for some of the
lexemes.
Singulatives are common, and not only for individual entities of collectives,
masses or sorts but also when there is no apparent semantic motivation. For example,
Dhaasanac has a derived singular bil-ti knife from bilu (Tosco 2001: 79). Singular
human and animal individuals are often derived by distinguishing males and females,
for example, Dhaasanac la (f) lions, sg: luoc (m) lion, looti lioness (Tosco
2001: 79); Arbore zze (f) gazelle: zze-t (m) male gazelle, zze-t (f) female
gazelle. Sometimes the feminine singulative is the second derived form, derived from
the male, for example Arbore hokkl lame (people), hokkol-an a lame male,
hokkol-ant a lame female; geleb (f) Dhaasanac, geleba-n (m) male Dhaasanac,
geleba-n-t (f) female Dhaasanac (Hayward 1984: 162). The singulatives are used
for the singular of pairs, Dhaasanac gunu (m) testicle, sg: gunti (f); for the singular
20
of collectives Dur (m) hair, sg: Duiti single hair (Tosco 2001: 79-80); for the
partitive of mass nouns, e.g. Tsamakko and-e (p) water, sg: and-itto (m), and-itte
(f) drop of water (Sav 2005). Most languages have about five different singulative
formations and often at least one of them contains -t-.
There is a strong interplay between singulative and definiteness in Oromoid,
and in Bayso the singulative -ti ~ -titi indicates individualization or particularizatoin
(Hayward 1978a: 106).
Names for people and their languages have their proper suffixes. The -ac in
Dhaasanac is such a suffix; in Arbore individuals of an ethnic group are derived by
suffixes that are not used for other words, Hayward (1984: 183); in Alagwa the
suffix -aisa derives language names such as imbeek-a'isa (f) Maasai language from
imbeek (f) Maasai (Mous unpublished); in Kabeena there is a suffix -sinata that
derives language names from names for people (Crass 2005: 83).
It is not uncommon for a Cushitic language to have a large number of
nominalizing formations. Crass (2005: 71-80) gives twelve different deverbal
nominalizers and a few less productive ones for Kabeena. Some of these have a
specific meaning such agentive, result of unaccusative verbs, or extent (Mass /
Menge); but some suffixes are used for a variety of meanings and many different ones
denote abstract nouns; ultimately it is a lexical matter which deverbal nominalizers
can be used with a specific verb. In Kabeena there is some overlap between deverbal
and denominal nominalizers; this is a common situation.

10. Terminal vowels
In the Omotic languages final vowels of nouns are often considered not to be part of
the stem; see Hayward (1987) and the chapter on Omotic in this book. In Cushitic a
similar analysis can be argued for; however, in many languages such an analysis is
just one of the possible options. Arguments for a special status of the final vowel
include the following: (i) the number derivations usually erase the final vowel of the
noun, (ii) for several languages not all vowels occur word-finally; for example, in
Konso nouns end in a with the exception of names which may end in i, o or e, (iii) for
some languages there is a correlation between the quality of the final vowel and its
gender. For example, in Kabeena nouns that have a short final vowel -e are feminine
and those that have -a, -aa, -o, -oo, -i, -u or -ee are masculine, unless they contain an
addition formative -t
a
(Crass 2005: 61-62); in Tsamakko nouns that end -o are
masculine, those that end in -a are feminine and those that end in -e are feminine or
plural in gender; no nouns end in u or i (Sav 2005: 51-52). Hayward (1983)
distinguishes between terminal and non-terminal ultimate vowels in Saho-Afar on the
basis of phonological properties.

11. Gender
Gender is very interesting in Cushitic because of its interrelatedness with number.
Here I adhere to the Cushitic practice of recognising plural as a category of gender
for those languages that have this third category. Note, however, that the typological
specialist of gender and number, Grev Corbett, has a different view on Cushitic
plural as exponent of gender (see Corbett 1991, 2000 and Corbett and Hayward
1987).
Gender is a property of nouns in terms of agreement, internal noun-modifier
agreement and external subject-verb agreement. Morphological (automatic) subject
agreement on the verb is either with number and within singular with gender (the
21
typological common situation) or with gender only (the typologically special
situation). As an example of the latter I present the situation in Iraqw (see Example 5).
All nouns fall in one of three groups depending on agreement with the verb. The three
agreement classes are termed feminine, masculine and plural because the first group
of nouns has the same agreement as a third-person female subject (she); the second
one, as a third-person male subject (he); and the third, as a third-person human plural
subject (they). I use multiple reference for the denotation of plural as an exponent
of number.

(5) Iraqw subject gender agreement on the verb:
a. daaqay i giiln. i giiln
boys 3 fight:3SG.M 3 fight:3SG.M
The boys are fighting. He is fighting.
b. ayse i harweeriirin. i harweeriirin
tails 3 make:circles:3SG.F 3 make:circles:3SG.F
The tails make circles. She is making circles.
c. ayso i harweeriirin'. i harweeriirin'
tail 3 make:circles:3PL 3 make:circles:3PL
The tail is making circles. They are making circles.

(6) External agreement in Arbore:.
a. nek y yeecce A lion came.
komayt y teecce A tortoise came.
mmo so yeecce The children came.
b. daac ay gra There is a rat.
ingir ay grta There is a louse.
bce as gira There is water.

Internal noun-modifier agreement requires the same division of the nominals
into three genders. In Iraqw, the combination of the various agreement markers also
require the same nouns to be derived into the same gender classes. This is shown in
Table 4 where the masculine nouns require the linker u the demonstratives; the
feminine nouns the linker r (deleted before an alveolar consonant) and the (p) gender
nouns have no gender linker. In Arbore there are several agreement markers for noun
phrase internal agreement (see Table 5) when modifying a noun (N-), on adjective
(Adj), on possessives (poss), on demonstratives (D) and on the the modifying question
word which?. They require the same three genders as the subject agreement on the
pre-verbal selector (preV) and on the copula (be) do.

Table 4: Iraqw internal agreement patterns: Demonstratives
hiima (m) rope asam (f) dilema gii (p) ghost
DEM1 hiimuw asamar giik
DEM2 hiimusng asamasng giisng
DEM3 hiimuq asamarq giiq
DEM4 hiimud asamad giid

22
Table 5: Internal agreement in Arbore
preV be came N- Adj poss D D which?
masc y gra yeecce -ha - ha- -h- 0 b-
fem y grta teecce -tah - ta- -t- t bto-
plur s gira yeecce -ha -o toha h- 0 to-

Internal gender agreement markers often involve ku for masculine and ta for
feminine, or forms developed out of those (see also Bryan 1959).
The values for gender on the basis of internal Noun Phrase agreement of
possessives and demonstratives in Cushitic languages are summarised in Table 6.
Table 6: Possessive and demonstrative agreement
m f p m/p f m f none
Alagwa, Burunge,
Iraqw, Arbore,
Boni, Dullay,
Kabeena definites
Alagwa pronouns,
Burunge pronouns,
Iraqw pronouns,
Arbore genitive
Elmolo, Oromo,
Somali, Kabeena
demonstratives
Konso, Dhaasanac,
Tsamay, Kabeena
possessives

Personal pronouns do not always show the same gender distinctions as nouns do. For
example, in Iraqw there are only two third-person pronouns: ins s/he and inon
they. In Arbore the sex of the possessor is differentiated in third-person singular
possessives. Also deictic pronouns referring to human antecedents display sex
difference using the words for man, woman, people. Personal pronouns in
Cushitic in general tend to refer to humans only and are primarily used for contrast.
They are more like a subset of nouns than grammatical markers.
Another important and typologically interesting feature of Cushitic gender is
that it is a property of the word and not of the lexeme. Singular and plural forms of
the same lexeme often differ in gender and partly in systematic ways. This has given
rise to the concept of gender polarity (Hetzron 1967, 1972). In Somali a large number
of nouns have the opposite gender in singular-plurals pairs. Serzisko (1982) has
analysed this phenomenon of opposite gender in terms of markedness. The more
general Cushitic picture is not one of polarity of gender but of plural and feminine
as common genders for multiple reference words in combination with correlations
between choice of multiple reference formation and the gender of its base. Thus the
polarity of gender is only part of the picture and not a property of the gender system
as such; the more general picture is rather one where different words, singular and
multiple reference within one lexeme are often different in gender.
Gender is not predictable on the basis of meaning of the word. Words with
male connotations can be feminine and the other way around; for example, in Alagwa:
sereea (f) buffalo, karama (f) castrated bull, isaamu (m) breast, teat, aama (f)
person who is made ill (male or female). For most words the choice of gender has
no semantic base at all; compare the words for gourds in Konso dahaan-aa (p)
gourd(s), hulp-a (m) large gourd for water, murraa-ta (f) gourd for drinking,
xott-aa (p) large water gourd, shaww-aa (p) gourd with handle.
There is some evidence for semantic associations with gender in terms of size
and endearment/pejoration, as is so common in the Omotic and Semitic languages of
Ethiopia. This is the case in the Western Oromo dialects in which the gender system
has developed into one with masculine as basic gender; use of feminine gender is
restricted to females and to express diminutives and pejoratives (Clammonds 1999:
392), as is the case in Agaw (Hetzron 1976: 14). In more general terms, gender
23
denotes the semantic notion of social significance (masculine) vs. social
insignificance (feminine) (Tucker and Bryan 1966: 511, Castellino 1975: 352ff, Sasse
1984: 117). There are parts of the lexicon where gender clearly has a semantic base in
all languages: agentives distinguish male and female sex which correlates with the
gender of the derivational suffix; derived singulars for animates are often sex
specified in the gender.
Gender is partly overt in the formal properties of the noun. Number suffixes
are gender specified, and once a number suffix is recognized, the gender of the noun
is known. Here too there is no full predicative value. Words ending in what seems to
be one of the number suffixes may have a different gender and some homophonous
number suffixes differ only in gender. There are also homonyms that differ in gender
only, e.g. Arbore ell (m) cowrie shell vs. ell (f) fear. We have already seen that
terminal vowels can be indicators of gender. Typical correlations between word form
and gender are those in Afar and Somali: Afar stressed vowel-final nouns are (f);
consonant-final and nonstressed vowel-final nouns are (m); other nouns with final o
and e are (f) (Hayward 1998). In Somali, polysyllabic masculine nouns ending in a
consonant have the accent/high tone on the penultimate vowel; those that are
feminine, on the ultimate (Saeed 1999). Final high tone for feminine is also reported
for Rendille (Oomen 1981: 40-43) except for those feminine nouns that end in a
vowel, while masculine nouns have penultimate accent. She proposes that the
contrastive pitch is caused by the loss of a feminine suffix in feminine nouns (Oomen
1981: 39). The difference in tone or accent placement is related to word-final
reduction processes: In Borana Oromo, feminine nouns mostly have long final vowels
and masculine short final vowels (Stroomer 1987: 70). The assignment of gender to
loan words can show a clearer picture of form-gender correlations; in Iraqw nouns
ending in u tend to be masculine; nouns ending in other vowels tend to be feminine
and loan words from Swahili follow this pattern (Mous 1993: 41). But loan words can
also be assigned to one of the classes, as is the case in Tsamakko where all loans are
feminine (Sav 2005).
Several proposals have been made to recognize default gender. Hayward
(1992) proposes that (f) is the default gender for Afar. Mous (1993) has suggested that
(f) is the default gender for Iraqw on the basis of (f) subject agreement of subject
complement clauses. It issometimes difficult to determine default values for gender
that are independent of other factors such as the gender of the general word for thing
or the phonological properties of the vowel that happens to be the marker of the
default exponent of gender. There are, however, languages with a default gender in
which all non-sex-distinguishable nouns are of the same gender.
The interplay between gender and number is in the (p) exponent of gender.
This class has to be set up because of words that require third plural agreement.
Underived (p) words constitute a relatively small set of words, ranging from 130 in
Konso to 4 in Afar. Many but not all of these words have some connation with
multiple reference, for example, people, children, women in Afar (Hayward and
Corbett 1988: 265). Examples in Alagwa are daaqaay (p) children, tikay (p)
women, wives, yawa (p) cattle, aaraa (p) goats, baaluu (p) days, fayee (p)
marriage settlement, bride price, kwau (p) house of many poles and liquids,
collectives, words for time and geographical concepts (Mous unpublished). Other
kinds of words that often appear in this group are words for part of the day. But also
clearly singular words appear in this class, e.g. tail in Iraqw, healed wound in
Alagwa. For many languages a large number of the derived multiple reference words
24
are (p). In Bayso all paucal words are (p). However, all relevant languages have
derived multiple reference words that are feminine (Alagwa) or masculine (Bayso),
seldom both. For example Iraqw has (p), (f) and (m) derived multiple reference words
but the (m) derived nouns are ambivalent in terms of number. Derivation for singular
reference is never (p) and always restricted to (m) and (f).
Oomen (1981: 56) proposes that (m) is the unmarked gender and used for
[-count] (transnumeral) nouns, and feminine gender is the marked gender to indicate
[+count], either plurality or singularity. Despite the attractiveness of this proposal to
account for the relative rarity of derived (m) multiple reference nouns, the proposal
does not hold for the other Cushitic languages that do have derived singulars in (m).
In Rendille no feminine plurals are derived from a feminine base, and this is also true
as a strong tendency for the South Cushitic languages. The historical explanation that
Oomen offers for Rendille is that these words already had the feminine suffix.
There are additional connotations of (p) and multiple reference in the external
agreement phenomena. Many languages show an alternative of semantic multiple
reference agreement to morphological gender agreement for the subject of the verb. In
Alagwa, multiple reference words that are (f) can be combined with either a third
singular feminine ending verb or a third plural ending of the verb. In the latter case
the agreement is on a semantic base. A second connotation of multiple reference and
plural gender agreement is that the same semantic agreement of a third plural verb is
observed as a resolution of gender conflict for a coordinated structure with mixed
gender; this is one of the strategies to resolve a gender conflict in Afar (Corbett and
Hayward 1988). A further complication is that several Cushitic languages have
reduced verb conjugation paradigms in which the underspecified verb form is
homophonous with that of a third-person singular masculine subject; see the section
24.
Feminine is associated with multiple reference. In Afar, numbers higher than
one trigger feminine agreement. In Arbore, Alagwa, and Iraqw multiple reference
formations are mainly feminine or plural in gender, rarely masculine. In Borana
Oromo (f) adjectives are used for mass meaning (Owens 1982: 47). In Rendille
change to feminine gender marks either sex specification or multiple reference
(Oomen 1981: 43).

12. Case
Case in Cushitic is typologically interesting, since the most common Cushitic case
system is that of marked nominative which is relatively rare among languages of the
world. Marked nominative refers to the fact that the noun is case marked for the
subject function (subject of an intransitive verb and agent of a transitive verb) and the
base form of the noun is used when the noun is not a subject, that is, in isolation, in
object position, as well as when it is the predicative noun in a nominal sentence
(Hayward 1988: n. 8); see Gensler (2000) and Knig (2006) for an overview of this
phenomenon and its areal spread. In both the marked nominative case system and the
nominative-accusative system the subject of a transitive and the subject of an
intransitive clause are treated the same way, in both the marked nominative case
system and the ergative-absolutive system the agent of an transitive clause is marked
and the object of a transitive clause is unmarked. A third general case form is the
genitive, and some languages mark the head of a modified noun (anti-genitive or
construct case). There are additional case forms in some languages, adverbial case
clitics and postpositions. In this section I deal only with core case: marked
25
nominative. The nominative is often termed subject case in Cushitic studies. The
unmarked case is called absolutive, in line with the unmarked case value in the
ergative-absolutive system; some authors prefer the term accusative for the
unmarked case.
Sasse (1984) presents an overview of marked nominative case in Cushitic. He
summarises its characteristics as follows: (i) absolutive rather than nominative
(subject) case is the citation form of the noun, (ii) the absolutive rather than the
nominative case is the predicative form of the noun in a verbless sentence; (iii) the
absolutive is also used for the vocative, measure constructions, and with adverbial
case markers; it also occurs when case is neutralized as a consequence of group
inflection or focus marking. Gensler (2000) adds that the absolutive is used for the
fronted topic and emphatic noun in situ position. Languages differ in the choice of
case they use for the subject in nominal clauses: Oromo uses the absolutive for the
subject of a verbless equative clause (Owens 1985a : 98), while Somali uses the
nominative (Saeed 1999: 187). The focalized subject is in the absolutive; in that
situation there is no subject agreement on the verb, e.g. in Somali and Arbore
(Hayward 1984a: 113). The explanation for this is that these constructions go back to
a fossilized cleft sentence (Hetzron 1972, Hayward 1984a: 113-126). In Borana
Oromo, relative clauses are not marked with nominative case marker (Owens 1982:
53). Dirayta is exceptional within Cushitic in that synchronically the nominative case
form is the unmarked form and the absolutive form is best treated as derived (Tosco
1996). Kabeena marks both accusative and nominative case, but the nominative can
be analysed as more marked and based on the accusative (Crass 2005: 86).
An exemplary overview of use nominative and absolutive case is presented by
Owens (1985a: 98-102). I copy his overview with some examples here (see 6). The
absolutive is unmarked in the sense that (i) it lacks morphological marking, (ii) it is
used as the citation form, (iii) it is the basis of morphological processes such as
genitive marking and coordination marking, and (iv) it is used in a large variety of
other contexts. The nominative is marked in the sense that (i) it needs morphological
marking and (ii) its function is restricted and (iii) can be specifically formulated as
marking the (focussed and non-focussed) subject of a tensed clause (both the subject
of intransitive clause and the agent of a transitive clause).

(6) Absolutive case in Oromo
a. equative predicate
xun bishaan kurshashaa
this water dirty
This is dirty water.
b. direct object
hrre-n ark dolki-t-i
fog-NOM sight prevent-F-IMPFV
Fog reduces visibility.
c. causative object
nama sn intalaa-f xenna xann-isiis-e
man that girl-DAT present give-CAUS-PAST
He made that man give the girl a present.
26
d. goal, location object
magala deema
market go
He will go to the market.
e. time complements
inni saaa afur si bod man tur-e
he hours four you after house stay-PAST
He stayed behind four more hours than you at home.
f. predicative
man adi akka gaari-tti dima dib-e
house white as nice red paint-PAST
He painted the white house red very well.
g. unit of measure
xara-n s rra kilomtri diddm fagata
road-NOM here from kilometers twenty far
The road is twenty kilometers from here.
h. object of postposition
inni xeesmma sun br jira
he guest that near exist
He is near to that guest.

(7) Nominative case in Oromo
a. subject of adjectival clause
hdii-n dim-tuu
lip-NOM red-F
A lip is red.
b. focussed patient subject (S) verbal clause
nyaan-ni n nyaatama
food-NOM focus eat:PAS
The food is being eaten.
c. focussed agent subject (S) verbal clause
sre-n adi-n n iyyi-t-i
dog-NOM white-NOM focus bark-F-IMPFV
The white dog is barking.
d. non-focussed agent subject (A) verbal clause
haat-ti okkte goot-t-i
mother-NOM pot make-F-IMPFV
Mother is cooking.
(Owens 1985a: 98-102)
The nominative (subject) case is often limited to certain nouns. It is restricted
to masculine nouns in Saho, Afar, Dirayta, Sidamo, and Kemant. In Saho only
masculine nouns ending in vowels are involved. In Dirayta the nominative case
involves masculine nouns only; in addition some singulative masculine nouns are
excluded from case marking (Tosco 1996: 28). In Rendille the nominative occurs only
on feminine nouns ending in a consonant provided that the noun is noun phrase-final
27
(Oomen 1981: 45). An overview of such restrictions is offered in Tosco (1994a: 226-
228).
Recurrent formal characteristics of the nominative case are low tone and a
final vowel i. In Somali, the nominative case is primarily marked tonally on the last
element of the subject phrase by lowering of the tone of the noun (but penultimate
high in one of the declensions) and by a suffix i for some feminine nouns (Hyman
1981, Banti 1984, Lecarme 1988). Lowering of tone and final vowel i are also
characteristics of the nominative in Saho and Afar. In Kabeena the subject case is
marked by retraction of the accent by one syllable (Crass 2005: 87). In Sidamo
masculine nouns change the final vowel to i or u for the subject case. In Oromo the
nominative is not always marked; when it is, by -i (m), -ti (f), or -n
i
plus voicing and
lengthening the final vowel.
Nominative-case marking is on the head noun and on its modifiers in Oromo;
the nominative and absolutive demonstrative pronouns are used interchangeably
(Owens 1985a: 87). In Arbore (Hayward 1984a: 150) the nominative is marked on the
head only.
Sasse (1984) reconstructs a nominative case system for Cushitic and case
marking on nouns by changing short final vowel -a for absolutive to -u or -i for
nominative on masculine nouns only; there was a different situation for nouns ending
in long vowels. Consequently word-final reduction processes resulted in the
restrictions of case marking that we find in the present-day languages. New
nominative-case marking also developed, such as Oromo nominative -n. And in Burji
a new feminine nominative case was formed by suffixing the feminine subject
demonstrative pronoun; these pronouns were distinct in case for masculine and
feminine in Proto East Cushitic: *ka MASC:ABS, *ku MASC.NOM, *ta FEM:ABS, *ti
FEM.NOM for the proximal demonstrative pronoun (Sasse 1984: 117).
Other case systems occur too. The Agaw languages have a nominative-
accusative system, and several languages have no distinction of nominative and
accusative case (South Cushitic, Konso, and Beja). In Agaw the accusative is marked
and the nominative (subject) is unmarked and identical to the word form in isolation
(Bilin and Awngi); in Kemant there is marking for the masculine subject as well. In
Bilin and Kemant the accusative is used with definite objects only (Hetzron 1976: 17).
The Agaw languages have non-core case marking as well.
There is a tendency for a correlation between definiteness and case marking.
Hayward (1988) shows that in Burji indefinite base subject nouns are marked by the
nominative case including gender agreement (definite expanded subject nouns are
case marked differently by reduction of their final vowel and for masculine words
addition of i). Indefiniteness is a matter of degree in regard to marking of the
nominative case in Burji. Such a system is a natural development, as is explained by
Comrie (1981: 123), since if case marking is partial only, it might be expected in
objects that are high in animacy and definiteness or in subjects that are low in
animacy and definiteness by the odd-man-out principle, i.e., only the less common
situation would be case marked. In the case of Burji the latter gave rise to an
indefinite subject marker which for masculine nouns is superimposed on definite
marking.

13. Genitive
The genitive links the nominal modifier to the head noun, e.g. Somali gri-ga Cal
/house-the:ABS Ali:GEN/ Alis house (Saeed 1991: 175). The genitive marks the
28
possessor in a predicative possessive construction, e.g. (8). The genitive is also used
with certain modifying suffixes. For example, in Kabeena the ordinal number suffix,
the privative and the similative suffixes require genitive marking on the noun, e.g.
kaakummee teeni-gga /september:GEN rain:GEN-similative/ like rain in September
(Crass 2005: 100).

(8) ti tkut
i
aye-rr
a

DEM1:F:marked animal:NOM who?:GEN-COP:P
Whose animals are these? (Crass 2005: 100)

In Oromo, as in many other Lowland East Cushitic languages, the noun plus a
modifier noun form a tonal phrase and gender-sensitive tone rules apply. In Oromo a
segmental gender-sensitive genitive-case marking in the form of an associative
morpheme is optional, obligatory when self-standing (Owens 1985a: 103-104).

(9) oww-i (xan) ibidd nam gubaha
fire-NOM ASSOC.M fire person burns
Heat from a fire burns a person.

(10) tan intal sun arke
ASSOC.F girl that.GEN saw
He saw that girls. (Owens 1985a : 104).

In Afar there is a distinction between definite (marked by -ih) and indefinite
genitives, with the latter used when no other modifier occurs. Burji also has such a
definite/indefinite distinction and has the additional interesting property that it agrees
in gender with the possessee rather than with the possessor on which it appears, e.g.
goti-nta daga /hyena.M-GEN.F ear.F/ a hyenas ear (Hayward 2002: 63).
In Agaw the genitive precedes the head noun and agrees in gender and number
with the head noun, next to a non-agreeing genitive construction (Hetzron 1976: 19).
In Arbore both the head and the genitive are potentially marked; the form of
the head marking consists of tone patterns and suffixes and depends on the gender of
the head noun and the phonological shape of it, while the genitive noun is marked
with a suffix containing t and has certain tone patterns depending on the tone pattern
of the head noun (Hayward 1984a: 150-157).
The phrasal properties of (genitive)-case marking (e.g. in Awngi (genitive)
case is marked on every element of the noun phrase) are discussed in the section 21.

14. Construct case
In the South Cushitic languages the nominal modifier is not marked and follows the
head noun. Any modified head noun is marked as such by the construct case which is
gender sensitive. In Iraqw the construct case is marked by a high tone on the final
syllable of the head noun: af-r mar'i /mouths:CON-F houses/ doors, muru ayma
/things:M:CON eating/ food. Nouns modified by an adjective or a relative clause are
also in the construct case; if the head noun is understood, a gender-sensitive construct
case pronoun is used, e.g. ar mari /INDEP.CON.F houses/ those (i.e. mouths) of the
houses.
29

15. Non-core cases and clitics
The Agaw and Highland East Cushitic (HEC) languages have additional cases:
Kabeena (HEC) has dative case, instrumental-comitative + locative case, and ablative
case. Formally these are built on the accusative. The instrumental-comitative is also
used for noun coordination; it has partial semantic and formal overlap with the
locative (Crass 2005: 105-106).
Several of the Lowland East Cushitic languages and South Cushitic languages
have case clitics that have a fixed position in the clause: Arbore, Dhaasanac, Elmolo,
Dullay, Boni, Rendille, Somali, Dahalo, Alagwa, Burunge, and Iraqw have case clitics
that are linked to the verbal complex rather than to the noun phrase. The case clitics
have also been termed adpositional clitics, or applicative. They typically have a
fixed position preceding the verb and indicate the role of one of the noun phrases in
the clause. The case clitic is not necessarily attached to that noun phrase (11e,f). The
usual cases are dative (11b), instrumental/comitative (11a,c), locative/allative
(11d,e), and ablative (11d). The clitics get cliticized either to the following verb
(11d,f) or to the preceding (pro)nominal (11c,e), which may often be an object
pronoun but again not necessarily referring to the object of the case relation (11d).
(11) a. um ye k un (Dhaasanac, Tosco 2001)
children me with gather.IMPF.A
Bring me the children.
b. ku lo-s-o ab-it Juma (Alagwa, Mous unpublished)
2SG.M OPT-DAT-O.M tell-2SG Juma
You should tell Juma.
c. anafuumay-ank-
i
ha-gi-ni-ri faa agim
a

1SG meat-N-DEM1 S1/2-O3PL-O.FOC-COM porridge eat.1SG.IPF
ilibaa-goo-ba (Burunge, Kieling 1994: 163)
milk-PRED-NEG
I eat the porridge with this meat, not with the milk.
d. --k-so-weyne (Rendille, Pillinger and Galboran 1999: 30)
FOC-IO-from-to-drove.animals:we
We drove the animals from [there] to [here] for him.
e. buura a-n suum-i qaas-an
beer O.F-EXPEC poison-DIR put-1PL
Well put poison into the beer. (Iraqw, Mous 1993: 246)
f. moor dek k-habta
home children at-remain
It is the children who remain at home. (Boni, Sasse 1981: 256)
Somali and Rendille allow stacking of case clitics, the other languages do not.
The actual forms of the case clitics and a discussion of their history can be found in
Appleyard (1990: 27) and Biber (1984: 51-52); see also Mous (2006) for the addition
of South Cushitic.

16. Adpositions
The word category of pre/post-position is not straightforward. Most languages have
special sets of nouns that translate like pre/post-positions and are at various stages of
grammaticalization towards postpositions (or preposition in the case of the South
Cushitic languages). Hayward (2002) addresses the analytical problem of whether
30
these relational clitics are case markers or postpositions, such as the locative marker l
in (12), and whether some others are postposition or relational nouns, such as the
direction marker ula in (12). Using as guiding principle that case markers do not
attach to adpositions, he argues that relational nouns are not postpositions, because
they often have the genitive case. And as a consequence, the relational clitics to which
no core case can be attached are analysed as postpositions because if they were case
markers, it would be unexplainable that these (relational) nouns cannot receive case.

(12) iyawt-i eel-i ula-l adiik yane
man-NOM well-GEN direction-LOC going he:is
A man is going towards the well. (Hayward 2002: 58)

Konso postpositions can be defined as a separate word category on the basis of
the fact that they do not occur as head in a Noun Phrase in a genitive construction
even if most can be shown to be derived from nouns. Postpositions can be combined
(13) and are often combined with a case clitic such as locative -opa in (14) and
directional adverbs.

(13) iifsa tarapeesa qara-dela ca
lamp table on-upwards be:IPF
The lamp is above the table. (Daudey and Hellenthal 2004: 87)
(14) aree-pp-opa-xata xooye
here-at-DEST-downwards come:IMP
Come down to this place. (Daudey and Hellenthal 2004: 88)

Dullay has cliticized postpositions genitive-locative, benefactive, instrumental,
directive, and ablative (Amborn et al. 1980: 89). Tsamakko has four adpositional
clitics that are attached to the noun phrase. Some, like the clitic nu, have a wide
variety of functions (15) (Sav 2005: 103-107).

(15a) ano beze=nu gaz-o oo-i
1SG.SUBJ Beze=from hair-M shave-3SG.UNM
I have shaved Beze.
(15b) bogol-k-o=nu qol-e cox-inda
king-SG-M=from cattle-P milk-PLUR.IMP.B
Milk the cattle on behalf of the king!
(15c) laabl-e gaan-t-e=nu eeg-i
cloth-F woman-SG-F=from bring-1SG.UNM
I brought the cloth to the woman.

The semantics of the locational nouns/postpositions may reflect a cattle-
focussed culture if the body part back is used for up reflecting the model of a
quadruped animal (cow) (Heine and Reh 1984). This is the case in Iraqw; Carlin and
Mous (1995) argue that the model is not necessarily that of a cow but more that of a
container, e.g. baati i daand do The iron sheets are on top of the house (Carlin and
Mous 1995: 124), but see Reh (1999) for counter-arguments.

31
17. Adjectives
The category of adjective is not clear-cut in all Cushitic languages. In several
languages there is a (static) verb conjugation type that plays an important role in the
function of descriptive modification. This is the case in Somali and Afar, (Banti
1988b: 208-213). These verbs form a conjugation class of their own and follow the
head noun in a subject relative clause, for example, (16).

(16) shalay baa ri-dii caddayd la qashay
yesterday FOC goat-DEF:NOM be.white:PAST:3F IMPERS killed
Yesterday the white goat was killed. (Somali, Banti 1988b: 209).

There are two other kind of words in Somali that translate as adjectives: nouns
that are used as predicates in a relative clause, e.g. nin marti ah /man guest is/ a man
who is a guest (Banti 1988b: 214), and attributives, that is, invariable nouns that
modify the head noun, e.g. meel sare /place high/ a high place (Banti 1988b: 217).
These nouns cannot be used independently, and they need a dummy head noun when
used predicatively, e.g. waa kan sare /FOC this high/ the higher one. Many
languages have a group of modifying nouns, e.g. Tsamakko, Arbore, and Kabeena.
Nevertheless, there is evidence for an independent category of adjective in many of
the Cushitic languages. For example, Treis (2005b) argues for an independent
category of adjectives in Kambaata on the basis of derivational category-changing
morphology that is specific for adjectives.
Adjectives can be defined morphologically by number agreement. Adjectives
are often the only word category that shows agreement with number, especially for
languages that have gender agreement in the verb. In the South Cushitic languages,
Rendille, Bilin, Oromo, Arbore, and Dhaasanac adjectives show number agreement,
singular being unmarked. Tsamakko is one of the few East Cushitic languages with
no number agreement in adjectives.
The languages that show number agreement in adjectives tend to have various
ways in which number is marked on adjectives, but these are different from nominal
number marking. The most widespread number agreement marking on adjectives is
by partial reduplication of the initial CV, e.g. in Arbore (plus vowel lengthening in
Dhaasanac, plus gemination of the root-initial consonant in Oromo), or complete
reduplication, as in Rendille. South Cushitic and Bilin have -an for plural number
agreement, Kabeena has -aanut
a
and other markers, Rendille has a prefixed a-
(Oomen 1981: 61), and Oromo has -o or -ot (-o marks (p) gender in adjectives in
Arbore).
Number agreement is claimed to be semantic for some languages. In Borana
Oromo, Stroomer (1995) points to the distributive reading imposed by a reduplicated
adjective expressing a difference between a group of good bulls with a singular
adjective and scattered good bulls with a plural reduplicated adjective. The same is
valid for Iraqw (see the example on paper money in section 9 on nominal number).
In addition to number agreement, adjectives show gender agreement.
Agreement involving adjectives can be complex, as in Arbore or in Burunge and
Iraqw where the modified head noun has a gender marker to allow modification, the
adjective is number marked and has additional gender marking. In Arbore we have
three options for agreement on an adjective modifying a multiple reference noun, see
(17) in which -a on the adjectives marks (masculine/feminine) gender, -o (plural)
gender and reduplication on the adjectives marks number.
32

(17) Arbore adjective gender agreement (Hayward 1984a: 201f)
a. ed-ha fa-fayyaan- /sheep.goats-M RDP-good-M/F/ good sheep and goats
b. enug-m-ta gu-guud- /kids-MR-F RDP-many-M/F many lambs/kids
c. kacc--ha -ils-o /stones-P RDP-heavy-P/ heavy stones

In some languages there is no gender agreement on the adjective itself, only on
the head noun, e.g. Alagwa and Rendille. In others, such as Oromo and Tsamakko,
gender agreement adjectives follow the noun without additional (gender) marking on
the noun.
In Arbore the adjective is invariable and shows no agreement when used
predicatively, in which case it needs a predicative suffix -da.
Adjectival derivation: Adjectives are derived from verbs by inchoative
derivation in Tsamakko, but in this language the inchoative suffix -ay is also present
in the regular gender markers for adjectives -akko (m), -atte (f) and -ayke (p). South
Cushitic has an unproductive -ar deverbal adjectival suffix. Tsamakko has -al (~ -ol)
to derive adjectives from nouns.

18. Adverbs
Adverbs are not a clearly defined major word class in Cushitic languages. Most
grammars include a section on adverbial expression in which expressions for time and
place are presented. These expressions may be nominal in nature, lexicalised phrases,
or difficult to categorise in a word class.
Some languages have a restricted set of adverbs that can be defined
syntactically as admissible in the verbal piece. For Iraqw these are ad quickly, soon,
immediately, adawa all together, ak more, further, al together, bal one day,
ever, never (in combination with negation), geer ahead, firstly, lak almost, mak
somewhat, mal again, first, qar already, nearly, sang now, just, already, tawo
in vain, uselessly, tsibi truly, tsuw for sure, algee slightly, a bit, tibe again.
In Alagwa a number of adverbs seem to be derived by -nkoo, banko first, hinko
now, lanko some time, some day, slanko before; others end in -ee, baree if,
tsige fast, early, tsobolee truly sigee far. Other languages too have adverbs that
end in o, for example, Oromo fgo far, dho near, dkkoo little gddo
very. In Kabeena adjectives in the locative case are used as adverbs. Kabeena has a
few de-adjectival adverbs derived by the similative suffix -gg
a
(Crass 2005: 239).
Oromo has some adverbs that modify the verb or adjective that they precede, such as
much, very, little, very.
In the Oromoid languages Oromo and Konso, postpositions can be used
adverbially, e.g. the Oromo postposition waj together in ni wj d'ufe he came
together (Owens 1985a: 121).

19. Numerals
Most Cushitic languages have basic numbers for 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 100, and
1,000. The highest numbers are often borrowed, e.g., in Alagwa 100 and 1,000 are
borrowed from Swahili; in Tsamakko from Amharic.
In the Lowland Eastern and South Cushitic languages numerals are a subset of
nouns but follow the noun without any genitive marking. The head noun does not
need to be in plural when followed by a number higher than 1; for example, it is not in
Dullay and Somali.
33
Being nouns, numerals have inherent gender. In Iraqw 1 to 9 are feminine, but
the units 10, 100, and 1,000 are masculine; in Arbore only the units1 and 100 are
masculine, the other numerals are feminine or plural. In several languages the nouns
for the units such as 10, 100 and 1,000 tend to have multiple reference forms (though
not the unit 10 in Arbore).
The higher numbers 11 to 99 are formed by the formula {term for 10(s)} n
{term for and} m in which the multiplicator is usually expressed by simple
juxtaposition and the addition by the common coordinator in the language.
Multiplication by 1 is optional or not expressed. This system operates among others in
Arbore, South Cushitic, and Dullay. The word for 10 is in its multiple reference
form if there is one (South Cushitic and Dullay).
A slightly different system is one in which the numbers 20, 30, et cetera, are
lexicalised but etymologically related to the basic numbers 2, 3, et cetera. This is the
case in Borana, Kabeena, Dhaasanac, and Afar.
Tsamakko has a different basic unit, 20, which is a word meaning body; the
numbers between 10 and 19 are formed by 10 followed by the number between 1
and 9, without a coordinator (Sav 2005).
Numerals do not agree in gender with the head noun. In Dhaasanac head
may intervene between the head noun and the numeral as a kind of generalized
numeral classifier. kulic m tikiddi /day head one/ one day, once upon a time. A
common exception, a numeral that does show gender agreement, is the numeral for
1 (Arbore, Tsamakko, and Alagwa). In Dhaasanac the number 1 has a different
form when used independently as in counting and when used as a modifier. The
numeral 1 can take up other functions such as a marked indefinite, e.g. Alagwa
wokoo a certain from wak one.
In the Noun-Numeral construction with no gender agreement the noun is
usually in the singular, e.g. in Oromo, Tsamakko. An example is gulm-a sala
/beer.calabash-F four/ four beer calabashes (Sav 2005: 87-88).
In Somali, numerals are the head of the noun phrase when they modify a noun,
e.g. far naagod /four.ABS women:GEN/ four women (lit. four of women) (Saeed
1999: 70).
Numerals occur in subject, object, predicate and adverbial positions. Arbore
numerals have predicative forms containing the predicative suffix -da which is also
used for predicatively used adjectives.
The syntactic behaviour of numerals is different in Highland East Cushitic,
Agaw, and Beja, where the order is numeral-noun, and the numeral is more a modifier
and less a noun, e.g. Kabeena lamalu annicci-ne /seven:M:marked father:ABL-
1PL.POSS/ among our seven forefathers (Crass 2005: 211).
Not all languages have ordinal numerals: Arbore lacks them; in Dhaasanac
one adds the element ki after the head noun and a determiner after the numeral used as
ordinal, for example ma ki afuur-a /man ki four-DET/ the fourth man (Tosco
2001:106). In some languages ordinals are derived from cardinals; for example,
Borana has a suffix -eesoo for ordinals, e.g. afr-eesoo fourth; this suffix contains the
adjectival suffix -oo (Stroomer 1995:61). The ordinal for first is usually lexically
different, e.g. rkob first in Dhaasanac; dur in Oromo from dr in front.
The distribution of number gestures is according to geographical area. In the
South Cushitic languages, showing numbers by gestures is in line with the common
East African system: 2 = forefinger and middle finger brushing against each other, 3 =
forefinger, middle finger and ring finger (no movement), 4 = forefinger and middle
34
finger as pair separate from ring and little finger in a V, 5 = raised fist, often moving,
6 = 3 + 3, 7 = 3 + 4, 8 = 4 + 4, 9 = 4 + 5, 10 = two fists hitting each other. Dhaasanac
has the Eastern Nilotic system of Maasai and Turkana in which 2 is presented by the
middle and the forefinger extended and brushed against each other and brushed
together; 3 is presented by forming a circle with the top of forefinger and thumb
together and the other fingers extended; 4 as described above and 5 by a fist in which
the thumb appears between pointing and middle finger (see Tosco 2001: 107;
compare with Zaslavsky 1973: 250 and Gulliver 1958). The counting gestures are
different from the gestures showing numbers and they display a great deal of variation
and areal influence. Counting on finger digits is very common in Ethiopia, counting
by closing fingers of one hand starting with the little finger is common in Kenya and
Tanzania.
In Dullay numerals can be used as verbs with inchoative meaning, e.g. sal
become four, from sala four (Amborn et al. 1980: 97).
Other quantifiers may behave like numerals. This is for example the case for
Arbore bl(h) all, every and Alagwa oomee all tk
u
or tk
u
all, whole and
modifying question words such as mi which?, miili which?, mag(a) how many?.
In Kabeena all is similar to the numerals. In Oromo heddu many and dkko
few are numerals and do not co-occur with numbers.

20. Ideophones
Cushitic languages are no exception to African languages in that they have an
extensive word category of ideophones. It has, however, rarely been described. Tosco
(1998) is a thorough study of Somali ideophones with an impressive set of examples;
Crass (2005) pays due attention to them; they are described for the South Cushitic
languages; see also Mous (2000) on ideophones in riddles.
Defining the category of ideophones is generally not easy. In Somali,
ideophones are noun-like; they are feminine nouns; in Dhaasanac they are a subclass
of adverbials or nominals (Tosco 2001: 249-250).
Some language have grammatical morphemes that are limited to sound
symbolic words. This is the case for Kabeena which has a nominalizing suffix -iti
that is restricted to ideophones (Crass 2005: 84, 233). In Iraqw there is a de-
ideophonic verbalizing suffix -eel.
As in many other languages, one of the ways to introduce an ideophone is by
the verb to say or another direct speech introducer. This is the case for Alagwa,
Iraqw, Somali, and Kabeena. In Kabeena the verb to make is used in transitive or
causative constructions (Crass 2005: 229); see also the section 25 on to say.
There are special phonological characteristics of ideophones, uncommon
sounds, and uncommon sequence of sounds. In Kabeena, ideophones often end in a
geminate consonant. And there is often expressive lengthening, tone and
reduplication.

21. Structure of the noun phrase
The Cushitic languages are divided into languages with head-final and head-initial
noun phrases. This does not correlate with differences in basic word order or other
characteristics that are universally associated with the position of the head such as
position of the adverb, pre- or post-positions, or prefixing or suffixing strategy. The
distribution of head-final versus head-initial nominal phrases is areal rather than
genetic: Noun phrases in Lowland East Cushitic and South Cushitic are head initial;
35
in Afar-Saho, Highland East Cushitic, and Agaw, head final. (The areal nature of the
order is evident from map 3 in Banti 1988b: 254).
Tosco (1994b) provides a typological overview of noun phrase syntax in
Cushitic (see Table 7) and proposes a historical scenario in which he proposes that
pre-head modifier order is an innovation in Highland East Cushitic and Saho/Afar
through the grammaticalization of a left cleft construction which explains the
recurrent determiner-gender markers on the preposed modifiers originating from
copulas in Highland East Cushitic.
36
Table 7: Word order patterns in Noun Phrases in selected Cushitic languages.
Language Adj, N Gen, N N, Poss Dem, N Num, N Rel, N No. of
Mod N
NP.s
No. of
N Mod
NP.s
HEC
Hadiya Adj N Gen N Poss+N Dem N Num N Rel N 6
Kambata Adj N Gen N Poss N,
N+Poss
Dem N Num N Rel N 6 (1)
Sidamo Adj N Gen N N+Poss Dem N Num N Rel N 5
Gedeo Adj N Gen N Poss N Dem N Num N Rel N 6
Burji Adj N Gen N Poss N Dem N Num N Rel N 6

Afar Adj N Gen N Poss N Dem N Num N Rel N 6

LEC
Dullay N Adj N Gen N+Poss N Dem N Num N Rel 6

Oromoid
Konso N Adj N Gen N Poss N Dem N Num N Rel 6
Oromo N Adj N Gen N Poss N Dem N Num N Rel 6

Omo-Tana
Somali N Adj N Gen,
Gen
N+Poss
N+Poss N Dem Num N N Rel 1 (2) 5
Arbore N Adj N Gen N Poss N Dem N Num N Rel 6
Bayso Adj N N Gen N Poss Dem N N Num N Rel 2 4
Dhaasanac N Adj N Gen N+Poss N+Dem N Num N Rel 6
Source: Tosco 1994b.
Bilin (Agaw) has both orders: head final and head initial. As in Beja the order Head-
Modifier is used to emphasize the modifier (Morin 1995: 46). In Dhaasanac,
possessive phrases can be either head final or head initial; the head-final phrases
require a final possessor or emphasis marker, e.g. cr bl=l /snake house-EMPH/ the
house of the snake, versus bl caret /house snake:GEN/ snake house/ (the emphasis
marker is absent in compounds with dependent-head order such as bl afu /house
door/ door of the house (Tosco 2001: 254-55).
Within the noun phrase, modifiers require the head noun to be repeated by a
gender-sensitive pronoun. This pronoun cannot function as a subject or object
pronoun. The shape of this pronoun goes back to ku (m), ta (f) and ki (p); but more
often than not these are reduced to kV (m/p) and tV (f) or even further reduced to
suffixes. The amalgamation of such a pronoun with the head noun results in construct
state of the head noun, i.e. the shape of the noun when modified. The South Cushitic
languages require a gender-sensitive genitive pronoun before a modifier: for the
independently used modifiers this has become part of the possessive or demonstrative
pronoun; when following the noun, the pronoun is fused with it resulting in construct
case on the head noun. Take for example the following noun phrase (18) from Iraqw
in which the gender linker is realized as r before the possessive suffix, as ta in the
independent demonstrative and as ar before the final modifying noun phrase.

37
(18) dooa-r-k ta-q' ar bar qaymo
hoeing-F-2.SG.POSS INDEP:F-DEM3 INDEP.CON.F in:CON field
that hoeing of yours in the field...

A head-final language such as Kabeena also requires a gender-sensitive
determiner for nominal modifiers. The determiner ta (agreeing with the head noun in
gender) precedes the nominal modifier hand in (19a) and the genitive pronoun
mine in (19b); it is also needed before relative clauses.

(19) a. ta anga forkott
u

DMD1:F:UNM hand:GEN rawness:NOM
the rawness of this hand ... (Crass 2005: 325)
b. ta ii saat
a
DMD1:F:UNM 1SG.GEN advice:ACC
my advice (Crass 2005: 326)

Some languages require a phrase-final determiner ka. Dhaasanac has this
determiner after every member of the noun phrase, for example, adda k=a ti=a
/aunt your-DET that-DET/ that aunt of yours in which the determiner ka appears as a
(Tosco 2001: 253).
In Oromo a possessor, relative clause, numeral, which, demonstratives, all,
and other can all occur without a head noun. The associative marker must be used in
that case for relatives and third-person possessives, including nominal modifiers.
Modifiers that follow the noun are either suffixes/clitics or independent words.
Determiners such as demonstratives and possessives may be suffixes to the head noun
as in the Iraqw example (18) above. These can also occur as independent words.
The order of modifiers in the noun phrase varies from language to language. In
Iraqw the order is noun-possessive/demonstrative/indefinite suffix -
demonstrative/indefinite pronoun-numeral/adjective/adverb/relative clause. In Oromo
it is noun adjective possessor-relative/numeral/which/other-demonstrative-only/all.
Numerals often follow the adjective, e.g. Tsamakko org-ayn-e busk-e xobin
/male.goat-PL.P castrated-P five/ five castrated male goats.
In addition to linkers with gender agreement, other agreement phenomena may
occur within the noun phrase. Adjectives may agree in gender, in addition to the
gender linker on the head noun and, independently, in number, as discussed in section
9 above. Several languages show case agreement within the noun phrase. In Oromo
both the noun and the following adjective agree in case marking. There is case
agreement in all qualifiers in Awngi. In example (20) the locative which refers to the
whole noun phrase is repeated after each word in the final position, the plural genitive
referring to the doorways of is found after each word in the higher genitival phrase
before the locative, and the masculine genitive referring to the nice house of is
found before that in the lower genitive phrase. There is much variation in this area of
noun phrase syntax among the Agaw languages. In Xamir, case is expressed only at
the end of the NP, not necessarily on the head noun. In Kemant, case is either on the
head noun or on the modifier.

38
(20) [gud-a-w-sk
w
-da una-w-sk
w
-da [cnkt-k
w
-da
good-FEM-MASC.GEN-PL.GEN-LOC woman-MASC.GEN-PL.GEN-LOC nice-PL.GEN-LOC
n-k
w
-da [wodel-k-da bjl]]]-ka]-da
house-PL.GEN-LOC large-PL-LOC doorway-PL-LOC
In the large doorways of the nice house of the good woman (Hetzron 1976:
37)

22. Demonstratives, definite markers, which?
Demonstratives, definite markers, and certain other markers are discussed together
because they partly overlap in form and function within and across languages. There
are five kind of markers that I discuss here: definite markers, demonstratives,
discourse deictics, particulars, and the question word which?. Boni has six
categories of deixis in Sasses (1980) analysis: proximal, distal, particular, anaphoric
referential, cataphoric, and habitual. Definiteness is never an obligatorily marked
category in Cushitic languages: sometimes the use of definite markers has been
grammaticalized, and the markers no longer necessarily express definiteness.
Demonstratives sometimes function as definite markers. A number of languages have
separate deictic markers for anaphoric and cataphoric use in discourse; others use
demonstratives for those functions. Various languages have a marker for a particular
X. Most languages have a questioning particular marker, which?. Many of the
deictic markers discussed here show gender agreement with the head noun, but the
patterns are mixed; within one language some but not all of the demonstratives may
agree in gender with the head noun; this is the case, for example, in Boni in which
only the proximal has the three-way gender agreement. The deictic markers occur
either as suffixed to the noun or as separate words; most can appear independently in
the separate word form. When noun phrases are case marked, this case marking may
end up on one of the deictic markers. Demonstrative and definite markers may have
case-specific forms.

22.1. Definite markers
Since definiteness is not an obligatory category, the use of the definite marker is in
opposition to that of an indefinite marker. There are various other means to indicate
that a referent is supposed to be present and prominent in the mind of the hearer:
leaving it unmentioned or using a pronoun, the position in the sentence of the noun
phrase referring, and so on. It is very common for a definite referent not to be marked
for definiteness. In section 21 on the noun phrase we have seen that definite markers
may occur several times or at certain positions in the noun phrase for grammatical
purposes only. In Rendille the relative marker is used on nouns that are modified.
When there is a noun phrase with several nouns, it is commonly impossible to
use the position of a definite marker to indicate differences in definiteness between
the nouns. For example, in Somali a noun is marked as definite if modifying another
definite noun and no difference can be made between a and the in car of the
company, baabuurka shirkada.
Several languages have definite markers that are separate from
demonstratives. Kabeena has a definite marker -n-gender linker (always followed by
gender/proximal demonstrative markers) which is added to most numerals. It can be
combined with other definite markers, e.g. gmi-n-ti-s
e
mancot
a
/all:F:NOM-DEF-
F:NOM-DEF.F woman:ACC:COP.F/ All are women (Crass 2005: 120-121).
39
Definite markers are often combined with other definite modifiers. For
example, Somali possessives require an additional definite marker. Definite markers
can also be used with inherently definite words such as personal names. In Somali
personal names that are modified by a relative clause will be marked as definite.
Geographical names contain a definite marker, e.g. ingriiska United Kingdom.
Iraqw allows demonstratives after names and personal pronouns (Mous 1993: 90,
282).
The interplay of definiteness and number is still to be researched. In Somali
the noun form that is unmarked for number will be used in indefinite contexts, e.g.,
dad baa yimin people have come; but in combination with the modifier all, the
noun must be definite, e.g., dadka oo dhan way siman yihiin all people are equal,
and similarly, in general statements such as dadka ma noolaan karocunto la'aan a
person cannot be without food.

22.2. Demonstratives
The languages vary greatly in the number of distinctions in degree of distance marked
by demonstratives, from only one in Konso to four in the South Cushitic languages.
Oromo, Konso, Kabeena, Dhaasanac, and Somali have two, proximal and distal; Afar
and Rendille have three degrees of distance. Within the same branch the number of
distinctions may be different: Within the group of Sam languages Somali has two
degrees, Rendille three and Boni three of which one is used referentially. Whereas
deictic adverbs in some languages refer to same, higher or lower altitude,
demonstratives refer to the distance from the deictic centre, the speaker.
It is common for demonstratives to require gender agreement with the head
noun, as is the case in Somali, Rendille, South Cushitic, Kabeena, and Khamtanga. In
Oromo this is the case for the proximal demonstrative only and not for the distal
demonstrative. There is no gender agreement in Dhaasanac.

22.3. Demonstrative pronouns
The Somali and Rendille demonstratives may be used independently. In the South
Cushitic languages independent demonstratives need a gender-sensitive base which is
different from the gender linker when used as a suffix. Dhaasanac uses the word ee
thing as head when demonstratives are used independently. Kabeena uses
independent forms that have a prefix stem hi plus gemination of the second consonant
(Crass 2005: 128-129); while the Afar demonstratives need a suffix h when used
independently (Bliese 1981: 15-16).
Several languages have separate referential markers, but in others
demonstratives are used to refer back or forward in discourse and in time. Kieling
(1994:80-81) discusses the discourse functions of Burunge third and fourth degree
distance demonstratives which are proximal non-spatial and distal non-spatial, non-
spatial referring to not visible. The proximal non-spatial form is used for a referent
that has been mentioned earlier in the narrative and that should be readily available in
the hearers memory; this demonstrative will be used for the protagonist of the story.
The distal one is used for referents that have been introduced much earlier and might
not be prominent in the hearers memory; it is also used for contrasting the opponent
to the protagonist. In Iraqw the third and fourth degrees of distance demonstratives are
the only ones that can occur reduplicated and this only in referential use.
40
Demonstratives are used as definite markers even in languages with separate
definite markers. Arbore uses the proximal demonstrative -l for definiteness which
can be added to phrases containing the distal demonstrative -tto, (Hayward 1984a:
191); in Konso there are two markers -se and -ose (plural sene) for definiteness and
demonstrative, and the difference between the two is far from obvious. A
demonstrative form se and variants thereof is widespread in South Ethiopia across
language groups: Dirayta has a proximal demonstrative se as an intruder in the
demonstrative system (see Tosco 1996); Dullay has a proximal demonstrative se, and
distal -ssa; Dime (South Omotic) has sini; Koorete (Ometo) has se-, Maale (Ometo)
has se and soo for elevation deixis; Zayse (Ometo) has distal so; Gamo (Ometo) has
sekki; Burji has -shi and outside the area, se is the invariant demonstrative of Yaaku.

22.4. Deictic adverbs
Deictic adverbs here and there are sometimes derived from demonstratives, as in
Alagwa ta, tay-s, ha-qa, and ha-d
a
. But these are used alongside diit
i
and diis, based
on the word dii place and a demonstrative of the first and the second degree
respectively.
Dirayta has a system of elevation deictic adverbs distinguishing (i) higher
elevation ele, (ii) lower elevation hte and (iii) level elevation a-se; the same
distinctions are made in Konso. The distinctions of elevation are only made for
remote distance, i.e. when facing away from the mountain slope (Hayward 1980:
285). Some Omotic languages of the area such as Maale, Dime and Zayse have this
feature too.
Separate referential markers exist for example in Oromo where the
aforementioned one is xani/tani, the proximal demonstrative is xana/tana and the
distal one is sana invariant for gender (Owens 1985a: 87-88). In Dhaasanac gir is
used for anaphoric deixis (Tosco 2001: 226-230). Awngi has a referential article -k
used strictly in the sense of the aforementioned (Hetzron 1976: 39). Boni
distinguishes between back referring in discourse and forward referring (Sasse 1980:
81).

22.5. Particular marker
There are two types of particular markers. One is used as a marked indefinite specific
similar in meaning to the indefinite article in English but only used when it is crucial
enough to mention that the entity from a relevant set is to be understood as new in the
discourse. Examples are Iraqw -koo/-kaa/-kaariya a certain, with double gender
marking, i.e. in the usual gender linker that precedes and in the form of the marker
itself (21). The other kind of particular marker has the meaning one of a set. This is
used in Boni -o to indicate a singular, specific referent (Sasse 1980: 81); Dhaasanac
uses na for particular deixis; it is often followed by the proximal demonstrative and
the general determiner (Tosco 2001: 227-228).

(21) looa-r-ka wak-ee garma-ko i hootat-n
day-F-FINDEF.F one-BGND boy:M-INDEF.M 3 live:HAB-DUR:3M
One day a certain boy was living, ... (Mous 1993: 93)

41
22.6. Which?
Several languages have modifying question words that show similarities with the
markers discussed above. Rendille has a suffix -koh which? (Pillinger and Galboran
1999: 18-19); Oromo has xm/tm (Owens 1985a: 88); Dirayta has hekmm / heknt
/ hekammaddu (Hayward 1980: 286); Arbore has bko / btoko / toko for m/f/p as
selective interrogative definitives, bteh whose? and kaak how many/much?
(Hayward 1984a: 199-200). The question word which? is the interrogative
counterpart of the particular marker. The descriptions are not detailed enough to
determine whether the question word which? refers to a predefined set (which of
those?) or to a set to be construed in discourse.

23. Pronouns
There are various types of personal pronouns. One set of personal pronouns is often
analysed as a special set of nouns; they can be modified by the usual nominal suffixes
such as definite and demonstrative suffixes. Personal (pro)nouns typically distinguish
person, number and gender. Zaborski (1989) provides an overview of Cushitic
independent pronouns. The person distinctions made are first, second and third. In
Somali, Rendille and Dhaasanac there is a distinction between inclusive and exclusive
first-person plural. Gender distinction is usually restricted to the third person but
sometimes extends to the second person, as in Beja and Dahalo, for bound pronouns
only (Tosco 1991: 37). Beja has gender distinction in the second-person plural; and
Dahalo as well, but for the bound pronoun only (Tosco 1991: 37).
1
South Cushitic has
gender distinction in the second person (singular) only and not in the third person.
There is no gender distinction in pronouns in Dhaasanac (Tosco 2001: 210).
Possessive markers (suffixes and pronouns) usually show the same
distinctions as the personal nouns. However, in the South Cushitic languages the
distinction in gender for the second person is not present in the possessives, which do
not distinguish gender of the possessor at all. On the other hand, in Harso-Dullay the
possessive distinguishes gender in the second person but not in independent personal
pronoun (Amborn et al. 1980: 91, 97). In Kabeena the third-person possessive s also
acts as a definite/ demonstrative marker (Crass 2005: 114-115). Final elements of the
noun may elide when a possessive is used, for example, the feminine linker t
a
is
deleted before possessives in Kabeena (Crass 2005: 114). Possessive suffixes and
pronouns often require additional definite markers. In Somali, where this is the case,
such definite marking is left out when the possessed noun is a kinship term; the same
applies to Khamtanga (Appleyard 1988b: 18).
In addition to the independent pronouns, most languages have other sets of
personal pronouns that are more pronominal and less like a subset of nouns. In Somali
there are subject pronouns and two series of object clitic pronouns; the subject
pronouns cliticize to the left of the focus marker, and the object clitic pronouns are
positioned within the verbal complex; the second series is used only when there are
two non-third-person objects; the third-person object pronouns being zero. Third-
person object pronouns are zero in Boni, Rendille, Konso, Elmolo and Dhaasanac.
Dahalo, South Cushitic, Dullay and Oromo have object pronouns too; see Appleyard
(1990) and Biber (1984: 53-54).

1
Iraqw does not have sex distinction in the second person plural pronoun, contrary to Whiteley (1958) and
consequently Zaborski (1989: 650).
42
A non-specific subject pronoun that is different from any of the other
pronouns, like French on or German man, is found in Somali, Rendille la, Boni li;
Arbore na, Elmolo (a)na and South Cushitic ta or da. In some of the languages this
option is in addition to the possibility of a passive extension on the verb. The
semantics of the South Cushitic ta is not just non-specific subject but includes senses
of collectivity and of human agent. For Iraqw, the impersonal subject marker can also
be used to refer to a specific collective group, yet there is no plural marking on the
verb. In Burunge the impersonal subject pronoun is so un-specific that no independent
personal pronoun can be use in connection with it. In Arbore the impersonal subject is
identical in form to the first-person plural marker. In Iraqw the impersonal subject
marker can refer only to human agents, while in the related language Burunge it can
be used even with weather verbs, i.e., verbs in which reference to anything remotely
related to a controlling agent is absent. In Iraqw the same impersonal subject is used
to indicate collective agents. The Elmolo equivalent of the impersonal is termed an
intransitive prefix by Heine, since it suppresses the possibility of expressing two
complements; it too can be used with agentless transitive verbs, such as to have
diarrhoea (Heine 1980). The Arbore impersonal subject construction is also used for
middle situations,
2
as is evidenced by example (22). The object pronoun in the Elmolo
example shows that the undergoer is an object.

(22) na kare (Arbore, Hayward 1984a: 308)
IPS shave
He shaved himself / he was shaved.

(23) kes en-ke-(e)ld-e (Elmolo, Heine 1980: 198)
2SG IPS-O.2SG-have.diarrhea-PF
You have diarrhoea.

All the languages that have object pronouns have a reflexive/reciprocal
pronoun. Oromo has a distinct reciprocal pronoun. A language like Somali has a
combined reflexive/reciprocal object pronoun si; others, like Konso, have a separate
reflexive, isi, and a reciprocal, oll together, each other. The reflexive/reciprocal
pronouns have a root either similar to the Oromo distinct reciprocal root wali, as is the
case in Dhaasanac, Elmolo, and Arbore verb, or to the Oromo reflexive fi, as is the
case in Dullay, Konso, Boni, Rendille, and Somali, or one that is related to the first
plural pronoun, as is the case in Alagwa, Burunge, and Iraqw. The Arbore reflexive is
not a separate object pronoun but consists of the object pronoun followed by tta and
wal- prefixed to the verb (Hayward 1984a).
(24) y-tta (l[e ]) wal-sibe Arbore (Hayward 1984a: 227)
I-SFX lhe
3
REFL-anoint
It was I that anointed myself.

2
A middle situation is not agentless but does not have the clear distinction between agent and patient, as is the
case in a standard active transitive situation.
3
The optional element lhe does not seem to have a semantic contribution and is possibly etymologically related
to a verb to have (Hayward 1984a: 226).

43

24. Verbs
Verbal inflection typically includes the expression of aspect (and tense, mood, and
evidentiality), dependent/independent clause, person (subject) marking, and negation.
Verbal inflectional morphology tends to be complex. Various conjugational classes
have to be distinguished; sometimes this reflects the application of
morphophonological rules, and sometimes the classes are related to derivation but it
often goes beyond that. The most basic aspectual distinction which most languages
display is that of Perfective versus Imperfective, often distinguished by a vowel
difference. In addition many language have a dependent or subjunctive third type with
yet a different aspect encoding vowel. Zaborski (1975) offers a comprehensive
overview of the Cushitic verbal system in a historical approach; see also Zaborski
(1997, 2005), Voigt (1985, 1996), and Banti (1987, 1994, 2001) for historical
scenarios for the developments of the main types of conjugations in Cushitic.
There are many structurally different conjugations. First of all many languages
distinguish between a prefix and a suffix conjugation. The suffix conjugations are
dominant. Then there are paradigms with reduced person agreement. New formations
typically arise in the form of compound tenses or auxiliary constructions.
24.1. Prefix conjugation
The prefix conjugation is the remnant of the pre Cushitic conjugation type and
survives in a number of languages, but typically in a restricted set of frequent verbs.
For example, in Somali it is only the verbs be, come, know, lie, and say
(Saeed 1999); in Awngi only bring, come, know, remain and be (Hetzron
1969: 44ff). Arbore has at least twelve prefix conjugation verbs which follow two
different conjugation patterns (Hayward 1984a: 261-265). Beja and Afar have larger
numbers of prefix conjugation verbs. In Afar the prefix conjugation is growing
because of borrowings from Semitic languages (see Hayward 1978b and Hayward
and Orwin 1991). In the prefix conjugation of Afar the person components and the
first plural number component are prefixed; first singular, third masculine and third
plural are differentiated (see Table 8); aspect and mood are indicated by stem vowel
mutation (identical stem vowels but not a for the perfect and a as first stem vowel in
the non-perfect) (Hayward 1978b: 355-359).

Table 8: Prefix conjugation in Afar
to eat
1Sg okme
2Sg t-okme
3SgM y-okme
3SgF t-okme
1Pl n-okme
2Pl t-okmee-ni
3Pl y-okmee-ni
Source: Bliese (1981: 110).

24.2. Person Marking
Person marking on the verb usually has seven exponents: 1sg, 2sg, 1pl, 2pl, 3masc,
3fem, 3pl. Beja is deviant in that it consistently distinguishes between 2masc and
44
2fem. A common characteristic is what Tucker called the block pattern: the ending of
2sg and 3fem are identical and also 1sg and 3masc are identical. The identity of a 2sg
and a 3fem verb form, t or its reflex, is indeed shared in all of Cushitic; the identity of
1sg and 3masc is common but not always valid and is the result of a merger of two
different endings (Banti 2001). The 2pl and 3pl forms are often plural forms based on
2sg and 3masc respectively. Thus there are usually at least three different person
forms: 1sg/3masc, 2sg/3fem, and 1pl; but in Dhaasanac there are only two forms: the
distinct 1pl form is absent and replaced by two forms: the 1sg+ form for the inclusive
1pl form and the 2sg+ form for the exclusive 1pl form: 1sg/3masc/3pl/1pl.incl and
2sg/3fem/1pl.excl/2pl (Tosco 2001: 112). In addition Cushitic languages often have a
verb form that is not inflected for person (or that has the 3masc ending, which is often
zero). This form is used with an unspecific subject (French on, German man); in
Kabeena it is also used for respect (Crass 2005: 157). In Arbore the third-person
plural verb form is used for impersonal clauses; in addition, a clitic na has to be used
on either the focussed element, or the selector (Hayward 1984a: 304-308).

24.3. Suffix conjugation
The most common conjugation type in Cushitic, the suffix conjugation, goes back to a
construction of a nominalised verb followed by an inflected auxiliary, a proposal
which is attributed to Praetorius (1893) (see Banti 2001 for a full discussion). The
suffix conjugation typically distinguishes the seven person forms mentioned above
and is marked for aspect. As an example I give the conjugation of Tsamakko (Table
9).
45

Table 9: Suffix Conjugation in Tsamakko
to drink to eat
1Sg g- -
2Sg g-d -t
3SgM g- -
3SgF g-d -t
1Pl g-n -n
2Pl g-d -t
3Pl g- -
Source: Sav (2005: 146)
4


Conjugation classes of the suffix conjugation are defined along formal and/or
semantic lines. In Somali three main conjugation types are based on whether the verb
has no derivational suffix, a causative suffix, or a middle suffix. In Iraqw as well,
different conjugations are set up for verbs ending in w, as in the inchoative suffix, and
those ending in m, as in the durative suffix. In Dhaasanac verbs ending in an coronal
form a class apart (Tosco 2001: 123). The Awngi verbal conjugation is complicated,
with tonal differences for groups of verbs (see Hetzron 1969).

24.4. Reduced paradigms
Several languages have inflectional paradigms with reduced distinctions in subject
marking. Banti (2001) calls it the Cushitic second suffix conjugation or the East
Cushitic Stative conjugation; he also discusses the possible correlates of this paradigm
in Semitic and Old Egyptian and suggests that the lack of distinction in gender in the
third person is a result of the fact that the person endings go back to possessives,
except for the third person where no possessive was used (Banti 2001: 21). In the
reduced paradigm there is no difference between masculine and feminine third person.
Examples are the Saho affirmative and negative non-past paradigms, the rare Somali
inflected paradigm for affirmative and negative non-past (usually the non-past is
rendered by an invariable form plus an inflected form of to be), the Burji affirmative
past, the Konso adjectival conjugation. In Saho-Afar and in Somali the reduced
paradigm is characteristic of a lexically defined group of verb roots including to be
or copula, to have and to lack, emotion-cognitive verbs such as hate, love,
know, and adjectival concepts such as white, red, new, long, bad. Hayward
(1978c) shows that so-called adjectives in Afar are in fact a category of stative verbs
and their verbal category is, amongst other criteria, shown by the subject agreement
on them. Additional arguments are presented by Vanhove (2000) for the Djibouti Afar
dialect of Tadjoura. Negative stative forms also exist in Afar but interestingly these
cannot occur in the attributive position (Hayward 1978c: 13). The stative verbs of
Afar are the quasitransitives have, lack, like, love and hate (quasi because
these verbs cannot be passivized); equatives to be (copula) and not to be (negative
copula); attributives such as good, thin, difficult and many more. An example of
a reduced paradigm is the following:


4
The two types of suffix conjugation have a tonal difference and represent two different lexical classes.
46
(25) Saho affirmative non-past of usuba be new
1sg usubiy
2sg usubit
3sg.m usub
3sg.f usub
1pl usubin
2pl usubitn
3pl usubn (Banti 2001: 8)

Language internally and cross-linguistically the reduced paradigms often
alternate with invariable paradigms. For example, in Somali the focussed form of the
inflected affirmative non-past is invariable, and in Rendille the equivalent of the
Somali reduced paradigm is invariable (Banti 2001: 7-8). Such invariant paradigms
are common in negative and focussed tenses. The Konso negative dependent
paradigms are invariable; the perfective paradigm ends in nin; the imperfective
paradigm in ninkinnin. The verb in Table 10 is kat sell, the final stem consonant
assimilates; the initial elements are the subject clitics that I discuss in 27.

47
Table 10: Negative Dependent forms of kat sell in Konso
Perfective Imperfective
1sg an kannin an kanninkinin
2sg ak kannin ak kanninkinin
3sg an kannin a kanninkinin
1pl ainu kannin ainu kanninkinin
2pl aishina kannin aishina kanninkinin
3pl aishoona kannin aishoona kanninkinin
-nin -ninkinin

24.5. Compound and auxiliary constructions
Most languages have various conjugations that have compound and auxiliary
constructions; i.e., an auxiliary construction in which the inflected element is an
auxiliary verb and lexical verb infinitival, or a compound consisting of an invariant
form and an inflected element. The Konso negative present in the Table 11 consists of
an invariant verb form ending in nin plus the negative to be.

48
Table 11: The Konso compound negative present continuous and related paradigms
PresCont NegPresCont Neg "be"
1sg in kanni an kanin-co co
2sg ik kanni ak kanin-kkitto kitto
3sg.f i kanni kanin-kitto kitto
3sg.m i kanni kanin-co co
1pl in kanninna an kanin-kinno kinno
2pl ik kannittan ak kanik-kittan kittan
3pl i kanni kanin-can can

New verb paradigms easily arise through grammaticalization of a infinite
(converb) form followed by a inflected light verb. The renewal of paradigms is clear
in Kambaata from the double person marking, e.g., a 2PL marker in the
imperfective -teennta consists of -t-een--nta -2-PL-aspect-2PL (Treis 2005a; see also
Tosco 1996).

24.6. Converbs
The Agaw and the Highland East Cushitic languages have converbs, that is, non-finite
verb forms that function in linking clauses. Kabeena has five types of converbs. The
converb that consists of the stem inflected for person and followed by an epenthetic
whispered i is used for sequential events. Other converbs are inflected and can be
marked for progressive, negation, or telicity (Crass 2005: 176-188). Converbs are also
reported for Agaw (see Hetzron 1969). Depending on the definition of converb, other
languages also make use of converbs; see section 34 on clause chaining.
24.7. Aspect
Apart from person, the verb is usually conjugated for aspect. Most languages have
different paradigms for perfective and imperfective; a third conjugational paradigm is
the optative. The major difference between these three is often in the aspectual vowel,
e, a, or o respectively. Another distinction that is often made is between so- called
independent and dependent sentences, requiring different paradigms. Many Cushitic
languages have additional semantic categories coded in the system. For example,
Awngi encodes tense and evidentiality (Hetzron: 1978). Kambaata expresses
intimidative (hortative not to do something) (Treis 2005a).
24.8. Negative verb forms
Negative verb forms consist of a different set of paradigms that are not based on the
affirmative set, as is the case, for example, with the reduced paradigms for negative
forms (see above). In Highland East Cushitic and Agaw negative suffixes or infixes
are used with modified forms of the affirmative verb (Appleyard 1984: 202-203).
There is often a fair degree of differentiation among negative verb forms within a
group of related languages. Such is the case in Agaw (Appleyard 1984: 203) and also
in South Cushitic where Iraqw has as suffix -ka originating from a verb kaa be
absent, lack and Alagwa has -ba from the quantifier without; in both cases they are
preceded by nominalisation morphology (Kieling 2002: 381-389).

49
24.9. Imperative
The imperative usually has two forms: a general one consisting of the verb stem and a
second one specifically used with multiple addressees. A number of languages have a
vowel ending for the general imperative or for certain verbs. This is often i (Konso,
Somali, and Dullay) but can be different, e.g. u, for middle derived verbs (Konso,
Dhaasanac, and Somali ). In Dahalo the imperative has an ending i for a singular
addressee in the imperfective, a copy vowel in the perfective, and e and o respectively
for plural addressees; a particle signaling direction towards the speaker (hither) can be
added (Tosco 1991: 59). Negative imperatives (or prohibitives) are often part of the
negative subjunctive/optative paradigm or need an extra predicative negation marker
such as ma ( see example 26 below). In the South Cushitic languages there are more
elaborate imperative paradigms with additional and combined marking of direction
(towards speaker or not) and the presence of an object argument; see the following
example of Iraqw and note that the low tone on -ang and -are indicates the presence
of an object.

(26) Iraqw imperatives
do dig!
doo- dig! (to many)
doo-eek dig it!
doo-aak dig it! (to many)
doo- dig for me!
doo-ar dig for me! (to many)
huw-a bring it to me!
huw-are bring it to me! (to many)
ma doo-aar dont dig
ma doo-ara dont dig (to many)

25. The verb to say
It is a common phenomenon among Cushitic languages to form compounds with a
verb to say. Appleyard (2001) calls these composite verbs, while Cohen, Simeone-
Senelle and Vanhove (2002) use the term descriptive compounds. Many of the
Semitic and some of the Omotic languages of Ethiopia share this feature with the
Cushitic languages; it is one of the features of the suggested Ethiopic Sprachbund,
and the origin is Cushitic (Appleyard 2001). The construction is also proposed to be
the origin of the suffix conjugation in Cushitic (Praetorius 1893); Appleyard (2001)
proposes the verb *iy(y) to say as the basis for the suffix conjugation. The verb to
say typically incorporates ideophones, and it is universally common for ideophones
to be introduced by a verb to say. Some Cushitic languages use to say to
incorporate quotes, e.g. Oromo muu jede express dislike, lit. say m-m (Appleyard
2001: 4). A language such as Afar (Cohen et al. 2002) uses the verb to say or to
put to make intransitive or transitive verbs respectively. The form of the incorporated
root (verb root, adverb, ideophone, noun) may be reduplicated, lengthened, or
sometimes attenuative sense may be added, e.g. daf-iyye /sit.down-3.M.PERF:say/ he
hardly sat down, but dffa-iyye /sit.down-3.M.PERF/ he sat down. The newly derived
verb functions as a lexical unit in the sense that it can undergo further derivation, but
prosodically the unit can be broken by adverbial particles and personal pronouns, e.g.
db ko-t he-yyo /IDEOPHONE you-on put-1S.FUT/ Ill beat you (Cohen et al. 2002:
232). Vanhove (2004) shows that in Beja this auxiliary to say has grammaticalized
50
into an intention or purpose marker, e.g. kwt ne strike:AOR1SG say:ACC1SG/ I
intended to strike (Roper 1928: 84, quoted in Vanhove 2004: 154). Vanhove argues
that the remarkable fact is that the usual intermediate stages of quotative and
complementiser were skipped in this instance.

26. Copula, verbs to be
Equative or identificational nominal clauses of the type this is water, xun bshaan in
Oromo (Owens 1985a: 79-82), often do not need a copula; juxtaposition of the subject
and predicate noun phrase (in that order) is a complete clause. Alagwa allows NP NP
clauses without the copula, nyaraw xulxumbimoo a scorpion is an insect; the Dahalo
copula -su suffixed to the subject (first) NP is optional (Tosco 1991: 89-90). The
subject may even be left out and understood as a third person, e.g. Dullay (so)
tiirakk he is a man (Amborn et al. 1980: 104-106). This is more common when
there is a copula present, as in Iraqw a l it is true, lit. is truth (Mous 1993: 235).
In Somali the sentence-type marker waa is used in such clauses, Cali waa bre Ali is
a teacher, and the subject NP can be left out, waa rn it is true, lit. is truth (Saeed
1999: 186-189). Several other languages use the subject clitic in such sentences, e.g.
Dullay u-tiirakk he is a man with focus on man (Amborn et al. 1980: 104-106);
Dhaasanac ma-ti-a
h
e daasanac /man-that-DET 3SUBJ Dhaasanac/ that man is a
Dhaasanac where the subject marker
h
e is optional (Tosco 2001: 288); and Alagwa
hareror na dooumusooo my wife is indeed a cultivator with a subject focus
marker na. The subject noun phrase in nominal equative clauses is in the nominative
in Somali but in the absolutive in Oromo. Several languages have copula suffixes on
the (second) predicative noun phrase. In Arbore this is the copula , , or d
(Hayward 1984a: 114, 122); Oromo has a copula suffix -da (after long vowels), -a
(after short vowels), -i (after consonants), e.g. lafee-da it is (a) bone (Ishetu 1989:
85); in Kabeena the gender-agreeing copulas are suffixes that derive from
demonstratives (Crass 2005: 263-272), and the subject NP need not be expressed.
Oromo has a second enclitic copula -iti, used in nominal sentences involving
possession, e.g. (kun) godaa annan-iti /this container milk:of-is/ this is (a) container
of milk (Ishetu 1989: 90). Kabeena also has such an alternative, -t
i
with lengthening
of the preceding vowel and invariant for gender; it is used in particular after names,
adverbs and demonstrative (in the genitive).
A suffixed copula on the predicative NP is common with questioning and
negation. Arbore has an interrogative copula -ko (Hayward 1984a: 122) which is
required with interrogation intonation, and in Alagwa and Burunge a copula suffix is
required for the questioning intonation and preceding a negative suffix, e.g. Alagwa
angaangay xulxumbim-ko /chameleon insect:QUES-M.PREDis a chameleon an
insect? and kurunkur tsirari-ko-ba /bat bird-M.PRED-NEG/ a bat is not a bird
Mous (unpublished). Rendille has -mee (Pillinger and Galboran 1999: 35-36); Harar-
Oromo has -mihi as a negative copula suffix which lengthens the preceding vowel and
changes its tone to High, e.g. bishaani-mihi it is not water (Owens 1985a: 79-82),
or a free form miti as a negative copula, sentence-finally (Ishetu 1989: 90); Kabeena
has the negative element -ba following the copula suffix; Arbore has a sentence-final
element and for negation; and Dhaasanac has muuni. Other languages use negative
forms of a verb to be, such as yahay in Somali.
Past tense requires either the use of full verb to be rather than a copula
(Oromo tur- , Dullay, Konso, Dhaasanac) or a past tense inflectional element on the
subject pronoun (Alagwa, Iraqw, Dahalo). Kabeena uses a past marker that follows
51
the copula suffix, comparable to the formation of a negative nominal equative clause.
Future also requires a full verb. The verb to become is used for future nominal
clauses, e.g. Oromo tah be, become; in Konso this is c-aad to become which is
middle derived from to be; similar to Dullay which had middle derived ooll-ad be,
stay, wait which is used in all tenses. Subordinate nominal clauses require verbs
rather than copulas: Kabeena ih to be, Oromo (with the exception of the
affirmative adjective which can be used on its own in a relative clause).
Locative and existential nominal clauses usually require a verb. For example
Oromo jir- locative be, exist (Ishetu 1985: 90), Somali jir-, Dhaasanac iddik be
there, stay, live, Dullay ak be, be somewhere, live, Alagwa and Burunge waar;
Kemant has deictic locative and existential wan- and locative smb. Iraqw has a
locative copula a/i/ta for first-and second-person subject, third-person subject and
collective third-person subject respectively; this is different from the equative copula;
in addition, an existential defective verb deer exists, and there is a full deictic locative
verb diirii be here developed form diir this place (Kieling 2002).
Clauses with adjectives used predicatively often have some verbal qualities.
For a number of languages adjectives are in fact stative verbs (see Afar, above). In
Iraqw and its relatives the adjective is preceded by a sentence element (selector; see
section 27 below) which is comparable to the one used in passive clauses; in
Dhaasanac adjectival clauses are negated with the element ma, used otherwise in
verbal clauses and different from nominal clauses. Somali uses the verb yahay to be
with adjectives.
Bilin has a copula that is used for nouns and adjectives as complements, and a
locative verb to be is used with locative adverbs and nouns in the locative case. Both
have suppletive forms. These verbs and the verb to have have the intriguing property
that there is a complete reversal of the usual aspect/time relations. For example, in the
pairxx
w
who is and s x
w
who was the former has the vowel sequence and
tone pattern that is otherwise used for the past tense, e.g. tamx
w
who tasted, and the
latter has the properties of the present tense, compare with tam x
w
who tastes (see
Palmer 1965). Palmer proposes that the historical explanation is that the verb to
have is from to take, and hence took (past) is have present; one of the verbs to
be is from to happen, and hence it happened (past) is it is (present).
Cushitic languages often have a verb to have: Dullay seeq, Iraqw and
Alagwa koom, but also Kemant (Appleyard 1975: 341-342) and Bilin (Palmer 1965)
have such verbs.

27. Selectors/indicator particles/INFL/sentence type markers
Cushitic languages are verb final, but many of them have an additional inflectional
element in the sentence that is separate from the verb and that has been termed in
various ways: selector in South Cushitic languages, indicator particle in Somali,
focus marker in Oromo. The prime function of selectors is to express elements of
information structure. The Cushitic languages that have selectors are Alagwa, Arbore,
Boni, Burunge, Dahalo, Dhaasanac, Dullay, Elmolo, Dirayta, Iraqw, Konso, Oromo,
Rendille, and Somali; the languages that have no selector are Afar, Agaw, Bayso,
Beja, Burji, Haddiyya, Kambaata, and Sidamo. In all languages that have a selector
either the selector has a function as sentence-type marker and these sentence types are
at least partly related to backgrounding information, or the selector expresses focus in
one way or another. The only exceptions are Elmolo and Dahalo. Nearly all languages
with a selector with the exceptions of Boni and Rendille also mark the subject in the
52
selector. These latter languages use independent subject pronouns where others use
subject suffixes or clitics. Rendille and Boni do, however, have an impersonal subject
marker that is integrated into the selector. They also have object pronoun clitics that
are part of the selector. Inflectional subject marking is typically a characteristic of the
verb. Thus selectors take up part of typically verbal functions. In this sense the
sentence-defining properties are divided over verb and selector.
While in some languages the position of the selector determines the scope of
focus, in a number of other languages the position of the selector is more fixed and
the position of the object vis--vis the selector determines the information value of the
object. Syntactically there are three types of selectors: (i) those that define the left
border of a syntactic unit such as the verbal piece in Somali; (ii) those that indicate
focus as a pro-clitic to the verb, and (iii) those that indicate focus by their position in
the sentence. Once the selector has a fixed position, then it also has a stronger
syntactic function in the semantics of the placement of complements in relation to the
selector; in addition to subject marking, these selectors also have the verbal quality of
valency. By developing more verbal functions, such a pivot attracts other inflectional
marking such as tense/aspect marking. There seems to emerge a division of marking
of grammatical roles in several of these languages: subject on verb, object in selector,
and others in the case clitics which have a fixed position between the selector and
the verb.
An overview of the categories that are expressed in and on selectors is given in
Table 12. The column Sub indicates whether the subject is indicated in the selector,
SenType whether sentence type is indicated, Mood whether questioning or negation is
indicated, Focus whether the selector has focus meaning, ImpS whether the selector
may contain an impersonal subject, Object whether the object can separate selector
and verb, Obj pro whether the language has an object pronoun series different from
the independent pronouns, Case whether adverbial case markers occur on the
selector. Deixis indicates direction marking and Tense/Aspect whether tense/aspect is
expressed on the selector (in addition to the verb).
Table 12: Overview of the properties of selectors
Sen
type
Mood Foc Sub Imp
Sub
Object Obj pro Case Deixis Tense /
Aspect
Arbore y y n y y y n y y y
Dhaasanac y y n y n y y y n n
Elmolo n - n y y n? y y y y
Dullay n n y y n y y y n n
Konso y y y y n y y n n n
Oromo n y y (y) n n n n y n
Boni n y y n y n y y y n
Rendille n n y n y n y y y n
Somali y y y y y y y y y n
Dahalo n y n n n y n y y y
Alagwa y y y y y y y y y y
Burunge y y y y y y y y y y
Iraqw y y n y y y y y y y
Source: Mous (2005).

53
Selectors with all the inflectional categories that can be expressed on them can
develop into quite extensive inflectional complexes, specifically in Dahalo, Alagwa,
Burunge, and Iraqw, as can be seen for Dahalo in (27) and for Iraqw in (28).

(27) b'-ka-v-ji ggwa (Dahalo, Tosco 1991: 71)
NEG-IRR-PAST-HAB love:E1:3M
He didn't love him.
(28) mu-s-tu-nd-a-y hanis (Iraqw, Mous 1993: 123)
QUES-REAS-IMPS-O.2.PL-PERF-DIR give:PAST
Why were you (plural) favoured?

In Alagwa, Burunge, and Iraqw there are different sets of selectors for main
clauses as opposed to consecutive clauses, and yet another set for object relative
clauses. In Arbore the selector indicates both sentence type, which is (definite)
indicative in (29), and subject, which is third-person singular in (29). Other sentence
types that are indicated in the selector are indefinite indicative future, indefinite
indicative present, jussive and negative.
5
The subject marking is either suffixed or
prefixed to these sentence-identifying selectors. Not every sentence, however, has a
sentence identifier (Hayward 1984a). In Dhaasanac too there are these two categories,
of sentence type (indicative and non-indicative) and subject. In Konso independent,
dependent, jussive, and negative clauses all require slightly different selectors.
(29) mo '-y kor kure (Arbore, Hayward 1984a:110)
man DEF.IND-3S tree cut:3SG.M:PERF
The man cut the tree.

In addition to the main distinctions in sentence type expressed in the choice of
selector, some languages also have mood prefixes that mark negative and prohibitive
sentences, as well as sentences questioning what; the marker for both is commonly
ma. Different ways of expressing questions exist as well, e.g., in the form of question
words; for negation, most languages have additional negation marking on the verb.
Additional mood distinctions in the selectors such as conditional and concessive are
made in Alagwa, Burunge and Iraqw. Some of these originate in grammaticalized
adverbs.
Focus is the central concept for the selectors. In some languages the selector
itself marks focus, most often verb focus or sentence focus. Several languages have
constituent focus markers that are separate from the selectors. For example, in
Dhaasanac topic and focus are the main organisational factors in syntax. Neutral
sentences have subject-case marking on the subject and no focus selector (30a). The
presence of a selector indicates verbal focus, i.e., the verbal focus marker
h
a in (30b);
there is an additional verbal subject pronoun as in (30b) or a full (subject) pronoun as
in (30c). When the subject is topicalized, a subject pronoun is used, as in (30d).
Subject focus is expressed by the addition of a nominal focus marker cliticized to the
subject NP (30e), as shown in Tosco (2001: 261-273).

5
The distinction definite/indefinite in Arbore refers to aspect marking, independent of the aspectual suffix to the
main verb.
54
(30) a. ar kufi
bull:S die:PF.A
The/A bull died. (neutral)
b. r
h
a
h
diyyime
bull FOC 3.VERB make.noise:IMPF.A
The bull is making noise. (verbal focus)
c. s
h
a y muura
meat FOC I cut
Ill cut the meat.
d. r
h
kufi
bull 3S die:PF.A
The bull died. (as answer to What happened to the bull?)
e. r=ru kufi
bull=FOC die:PF.A
The bull died. (subject focus)

Selectors differentiate at least between speech act participants and third
persons (Alagwa, Burunge, Iraqw) or make person distinctions without gender and
number differentiation (Dullay, Konso) or distinguish both person and number
(Arbore, Dhaasanac, Elmolo), and gender as well (for third person) in Somali.
Cushitic languages do not distinguish second singular and third feminine in the
subject agreement on the verb; and many of them also do not distinguish between first
singular and third masculine. Thus the subject marking in the selector resolves that
ambiguity.
Many of the Cushitic languages have a separate impersonal subject pronoun
that is used in passive-like sentences (see section 23 on pronouns above). The verb
form used in the impersonal construction is always in the third-person singular
masculine form (3M), except for Arbore where it is 3PL (Hayward 1984a: 305). For
those that have 3M one could also argue that the verb is simply not conjugated for
person. There are some indications that this impersonal subject marker is of a
different order than the subject pronouns. One such indication is that its structural
position is different from the subject pronouns in Somali, according to Svolacchia et
al. (1995).
6
In Alagwa, Burunge, and Iraqw the same markers, impersonal subject plus
object pronoun, are used for predicative adjective constructions (31).
(31) tluway ku er (Iraqw, Mous 1993: 203)
rain(m) O.3:IMPS:O.M insufficient:M
Rain is insufficient.

Object marking in the selector is not always compulsory but is dependent on
the position of the full noun object; it is also related to the information structure.
Those languages that have object pronouns in the selector complex (Elmolo, Dullay,
Rendille, Somali, Dahalo, Alagwa, Burunge, and Iraqw) have them as enclitics to the
selector, often replacing the fronted full object noun. Somali, Rendille, Boni and
Elmolo have a second series of object pronouns.

6
But not for Saeed (1993: 216).
55

28. Verbal Derivation
The common verbal derivations are causative s, middle d (also called reflexive,
intransitive, or autobenefactive), passive m, inchoative w and reduplication for
frequentative and habitual. The non-reduplicative derivations are suffixes. Only Afar,
Saho and possibly Burji have derivational prefixes in addition to suffixes. A number
of other derivations occur in a subset of languages only. The non-reduplicative verbal
derivations are reconstructed for East Cushitic by Hayward (1984b) and for Agaw by
Appleyard (1986). Konso and Dullay have a singulative or punctual derivation that is
formed by geminating the final root consonant, which indicates that the action is done
once or a little bit. The forms for the common derivations given here are not valid for
all languages; some Agaw languages have -st for passive, for example, and others use
these derivations in different functions; for example, Iraqw has a durative -m and no
passive; in the Agaw languages the reflex of the m derivation denotes reciprocity
(Appleyard 1986a: 17-18). Other derivations exist too; for example, Dullay has a
social derivation in some verbs such as dih-im give advice, hor-im settle an
appointment (Amborn et al. 1980: 118); several languages show signs of frozen
derivations that are no longer productive and to which no meaning can be attached,
e.g. -pp and -ott in Kabeena (Crass 2005: 145), -an in Dullay (Amborn et al. 1980:
118); -ab in Tsamakko (Sav 2005: 185) and -b in Dahalo (Tosco 1991: 48); -a in
South Cushitic for stative (Kieling 2002: 301). South Cushitic has a suffix -eel for
sound-expressing verbs derived from ideophones and onomatopoeia. In a number of
languages the vowel of the derivational suffix is analysed as a morphological
epenthesis; see Lloret (1987) for Oromo and Mous (1993) for Iraqw.

28.1. The Causative
The causative is marked by a suffix -s or -sh preceded by a vowel i which is
sometimes analysed as epenthetic; the causative is -d in Dahalo and Bilin. The
causative introduces an external causer to the state of affairs. The causative may be
doubled, or languages have two causative forms, a short and a long one. There are in-
depth studies of the Oromo causative (Owens 1985b, Lloret 1987, and Dubinsky et al.
1988). The picture that arises from these is that the number of causative morphemes
in a verb stem reflects the number of agents in the state of affairs; an unaccusative
(agentless) intransitive basic verb requires a double causative, while an active
(agentive) intransitive verb requires only one causative morpheme. This
generalisation requires that certain double causatives are recognized as reduplicated
causatives rather than double causatives and function as intensives. In the related
language Konso double causatives are also common, but there is no agent counting;
instead the double causative increases the indirectness of influence of the extra causer
onto the state of affairs (Mous 2004c). Wondwosen (2006) also reports for Oromoid
Dirayta that the distribution of single and double causatives is not along the lines of
counting underlying agents. Double causatives are very common and reconstructed
for East Cushitic by Hayward (1984b) in several combinations of the reconstructed
causative suffixes; double causatives are also reported for the Agaw languages
(except for Kemant) (Appleyard 1986), but are absent in South Cushitic and Dahalo;
these latter languages have an intensive meaning as one of the functions of the
causative; such a function is restricted to the reduplicated causative in Oromo (Lloret
1987).
56
The following Konso example shows that the derived causative in (32b) has the
external cause expressed in the subject and differs from the underived verb in the
expression of an external cause; interestingly the causative verb is still intransitive.
(32)a. i awd-
s3 bright-PF
It is midday/totally bright.
b. waaga i awd-ish-
god s3 bright-CAUS-PF
The weather is clear again [God has cleared up/caused brightness].
The indirect causative in Konso expresses that the external causer has less direct control of
the action expressed in the verb as in the Konso sentence (33b)
(33)a Mamm damta oorra dam-sh-
Mammo food people eat-caus-pf
Mammo fed the people.
(33)b Mamm oorra damta dam-aciis-
Mammo people food eat-icaus-pf
Mammo ordered the people to eat food.

28.2. The Middle
The middle is very common in Cushitic. In many languages the productive meaning
of the middle is to express that the action is benefactive (or occasionally malefactive)
to the subject and hence it is called autobenefactive; the term subject-reflexive is
also used. This sense of the middle meaning is applicable to a wide range of verbs,
and once the autobenefactive sense of the middle develops in a language, it becomes
productive. The autobenefactive sense possibly spreads through language contact,
since it is absent in the geographically distant South Cushitic languages; Hayward
(1975: 221) considers the autobenefactive function as a good isogloss. Hayward
distinguishes between middles of agentive verbs that have either reflexive or
autobenefactive meaning and middles of patient-type that are always intransitive and
often derived from nouns or adjectives (Hayward 1984b: 83-84). He also points to the
formal complexity of the middle derived stems often showing alternation of
consonants in different persons of the paradigm in conjugation. The Cushitic
languages show the centrality of the body in the semantics of middles. The evidence
for this lies in the presence of the subcategories of the body in all the languages in
Table 13 and in the fact that a number of derived verbs are used for actions performed
by the body as opposed to the individual. The fact that the marking is derivational
rather than inflection allows for a more lexical or concrete and a less grammatical
meaning, when compared to inflectional middles and to syntactic constructions as in
Kemmers (1993) study of the middle.
The spontaneous action middles are well represented in Cushitic middle
derived verbs and this semantic aspect should actually be considered as part of
prototypical middle meaning. The facilitative use which is closely connected to the
spontaneous action middle (Kemmer 1993: 148) is not or is rarely present in Cushitic
languages. In the Cushitic languages there is relatively little use of the inherent
57
reflexive and inherent reciprocal meanings of middle markers, which is not surprising
given the presence of reciprocal/reflexive pronouns. Remarkable is the near absence
of certain lexicalisation patterns: There are virtually no commissive, intentive middles
and relatively few emotional speech middles. Other lexicalisation patterns emerge in
Cushitic: the middle marking of verbs for to hide, to remain, stay, and a middle
denominal verb to work. See Hayward (1975) for a historical overview of middle
marking, Mous (2001) for a typological overview of the middle in Cushitic, Saeed
(1995) for the middle in Somali, Mous and Qorro (2000) for the middle in Iraqw and
Mous (2007) for the middle in Konso.
The body orientation for the subject of middle derived verbs is evidenced by
the Iraqw example (34).
(34a) yae-r-e a-ga tunqula
leg-F-my O.F-PF sprain:1SG
I sprained my ankle.
(34b) yae-r-e aa tunqulu-t
leg-F-my S3:PF sprain-MIDDLE:3F
My ankle sprained.
The middle is also used to express lack of control by external force as in the following Konso
examples (middle is expressed by the suffix -ad.
(35) pisha i urqn water flowed
pisha i urq-ad-n water found its way by itself
i sukum he rolled
i sukum-ad- he rolled by himself

The benefactive reading of middle derived verbs is productive in Konso, (36).
(36) i-aan-ad- he tavelled for his own purpose
i-kod-ad- he worked for himself
alal-ad- he chewed for himself
One of the differences between the middle and the passive is that in the middle situtation
no agent or cause/source of the action is present or imagined, compare (37b) with (37a)
in Iraqw.
(37a) inqwari kaa kunjuu-s
sheet.F O3:IMPS:O.F:PF fold-CAUS:IMPS
The sheet has been folded.
(37b) inqwari aa kunjut
sheet.F S3:PF fold:MIDDL:3SG.F:PAST
The sheet is folded in a crooked way (you dont know how it got folded).

58

Table 13: Semantic subdomains of middles in some Cushitic languages.
subcategory Iraqw Somali Oromo Afar
body care (groom and wear) + + + +
body motion (nontranslational motion) + + + -
motion of hands + + + -
body activity + + + +
(negative) body state + + + +
(change in) body posture + + + (+)
hide oneself + + + +
remain-stay + +? + -
body focused displacement/transl. motion + + + +
(negative) state of mind (emotion) + + + +
(complex) cognition + + + -
commissive, intentive - - - -
(emotional) speech - (+) - +
(inchoative) non-control/spontaneous action + + + +
(evaluative) facilitative - - - -
inherent reciprocal - + - -
autobenificient - ++ ++ ++
logophoric - - - -
work - ? + +
intensive - - - -
separate + - + +
negative connotations + - + +

28.3. The Passive
The third valency-changing derivation is the passive. The passive has the patient of
the base verb as subject and the agent is no longer expressed. However, with the
passive verb an agent is assumed to exist even if it is not expressed; the passive differs
in this respect with the middle. Not all Cushitic languages have a derivational passive;
for example, the South Cushitic languages, Dhaasanac, and Oromo have no
derivational passive. Several Cushitic languages have an impersonal construction,
some of them in competition with the passive (e.g. Somali, Afar). The form of the
passive is -am (Somali, Konso, Dullay, Afar) or -s(t) (Khamtanga, Bilin).
7
In the
Dullay languages there is overlap in function between the passive and the middle
(Amborn et al. 1980: 119; Sav 2005: 180-182). In Kabeena there are two
morphemes -ta (probably etymologically related to the middle, Hayward 1984b: 94)
and -am; in addition there is a fixed combination of middle and passive, -akk-am to
express the reciprocal (Crass 2005: 143-145). Rendille has two derivations: the
neuter-passive -am which does not imply the presence of an outside agent, e.g. fur-m-
a get opened; and the true passive -nam which does imply an outside agent which is,
however, never expressed, fur-nam-a be (able to be) opened (Pillinger and Galboran
1999: 32-33). Hayward (1984b) reconstructs the Agaw languages with a derivation
in -t that expresses both middle and, in a small number of verbs, also passive; while

7
The Dahalo passive is -ikud (Tosco 1991: 46).
59
the combined causative-middle -st is a productive passive derivation, Appleyard
(1986: 8, 15)
The semantic restrictions on the base for the passive differ from language to
language. In Somali the base verb must be semantically causative and the passive
cannot be used with experiencer verbs or with activity verbs such as eat, carry
(Saeed 1999: 138). In Konso there are no general semantic restrictions on verbs to
form a passive; the passive can apply to all kinds of intransitive verbs: unergatives
such as go, swear, lie, live, and to an unaccusative intransitive verb such as be
satiated, satisfied which describes the (resultant) state/quality of the subject, and non-
volitional body actions such as burp, laugh, be smothered (see Mous 2007).

The passive requires the patient to have subject function and the agent need no longer
be expressed as in the Konso sentence pair (38). But in Konso even intransitive verbs
can be passivized (as is common among the languages of Ethiopia) as in example
(39).
(38) anti inna kataata in erg-
I boy food 1 send-PF
I sent the boy food
b. kataata inna i rg-am-t-
food.F boy 3 send-PAS-F-PF
Food was sent to the boy
(39) urmalaa i an-am-
market 3 go-PAS-PF
The market was frequented.
An impersonal (unspecific) subject construction using the third person plural is
commonly used and constitutes a functional competitor to the passive derivation. In
Konso sentences describing pictures to elicit expressions for locational relations make
abundant use of impersonal constructions to describe situations. Not only the agent
but in fact the action is irrelevant in these sentences, (40).
(40) mataafaa shelfeeta kara xaay-e-n
book shelf on put-pf-pl
A book is on the shelf. (They've put a book on the shelf).

28.4. The Frequentative, Habitual
Most languages have a derivation by reduplication that expresses plural action such as
continuous, repetitive, or iterative action, or intensive or quick action. Plurality of the
subject of an intransitive verb or of the object of a transitive one is a factor that may
trigger the use of this derivation but there remains a choice for the speaker to indicate
such plurality in this way or not. The reduplication applies to the initial syllable of the
verb stem and can take several forms across and within languages: C
1
V
1
C
1
- forming a
geminate as second radical in the derived verb (except for consonants that do not
occur as geminates), e.g. Somali duudduub from db fold but jajab from jb
break (Saeed 1999: 49-50, Banti 1988b). In Somali the vowel length in the
reduplicant is identical to that in the original syllable, but most other languages, e.g.
60
Oromo (Owens 1985a: 84) and Dahalo, require this vowel to be shortened, e.g.
Dahalo gagaalij from gaalij go home (Tosco 1991: 48). A second type of
reduplication is C
1
V
1
C
2
-. Rendille has both, e.g. furfura from fura be open and
diddiiba from diiba hand over. In addition Rendille has aC
1
- geminate forming an
alternative derivation, e.g. ahhida or hidhida from hida (Pillinger and Galboran 1999:
33). The third common type of initial reduplication is C
1
V
1
-. Note that this can also
occur as a variant of the other reduplications where the reduplication would lead to
inadmissible geminate consonants or consonant clusters. Nevertheless, a separate
C
1
V
1
- reduplication has to be recognized. It occurs, for example, in Dhaasanac where
the vowel in the reduplicant must be long, e.g. fafa from f (Tosco 2001: 142), and
in Boni where the vowel has to be short, e.g. sisii from sii give and duduud from
duud consider (Heine 1977: 280-281). South Cushitic has C
1
V
1
- and C
1
V
1
C
2
-
reduplication with a difference in meaning, the former for frequentative and the latter
for distributive/frustrative action (Kieling 1992: 192). Afar has a different type of
reduplication for intensive action and in the Aussa dialect for frequentative action:
The onset and the shortened rhyme of the final syllable is reduplicated and closed by
the onset consonant giving rise to a geminate, thus C
x
V
x
C
x
- where x is the
penultimate consonant and inserted before the final syllable, e.g. usussuul laugh
heartedly from usuul laugh, biyayyaak from biyaak hurt. If C
x
is a geminate in the
base, we get successive geminates, e.g. iggiggif kill brutally from iggif kill. An
exception to this pattern is camcamm throw hard from camm throw (Bliese 1981:
127-128). Final reduplication is very common in South Cushitic where the last root
consonant is reduplicated but only if the verb is derived or has a form that is identical
to a derived verb; de facto, the penultimate consonant is reduplicated with an
epenthetic a if the ultimate consonant is m, s or t, that is, a consonant that appears in
one of the segmental verbal derivations; see (41) which also contains an example of
initial frequentative derivation, long aa as epenthetic vowel to express habitual
(Mous unpublished).
(41) Alagwa imperfective derivation (Mous unpublished).
ag eat
ag-im future
ag-amim durative
ag-ag-im present progressive
ag-aag-im habitual
aga-agim frequentative

A number of different reduplications are used to derive verbs in the
imperfective domain; in some languages several reduplication types are used for the
same broad function; in others, distinctions in function can be made for the various
reduplications. There are also languages that use segmental means for the same
functions. Kabeena has -ans, a fixed combination of passive and causative, for
repetitive, iterative and frequentative (Crass 2005: 146); Dahalo has -ameemit for
frequentative (Tosco 1991: 47), similar to the South Cushitic fixed
combinations -maamiit, -maamiis, -aamiim for habituals.
Dullay and Konsoid have a singulative derivation that consists of gemination
of the final consonant which expresses that the action is done once or a bit. Some
examples from Tsamakko are ug to drink: ugg to sip, kad to climb: kadd to
climb with one movement, ka to get up: ka to get up suddenly, cox to milk:
61
coxx to squeeze the udder once. The singularity can refer to the number of objects,
ab to tie: abb to tie one thing at one time, di to plant: di to plant one plant
at one time. Singulatives are preferred in imperatives (Sav 2005: 187). The
derivation cannot apply to verbs that have a final geminate consonant, but some of
those originate in this derivation. In Dullay there is free variation for a number of
lexemes between single and double final consonant, far ~ farr to die (Amborn et al.
1980: 117).

28.5. The inchoative and verbalizers
The inchoative, or inceptive, derivations are de-nominal or de-adjectival verbalizers.
Several forms are attested within one language and across languages. The inchoative
derivation is -aw(w), -uy or -um in Dullay (Amborn et al. 1980: 117-118), -ow in
Somali (Saeed 1999: 135-136), -ow or -aw in Rendille (Pillinger and Galboran 1999:
33), -uw in South Cushitic, -um in Konso, -a, -aaw, or -ee in Kabeena (Crass 2005:
146-147).
Verbalizing derivation in general contains verbal derivational suffixes such as
causative or middle, depending on the meaning of the resulting verb, and often
containing different vowels, oo or ee. The difference in the vowel is a relic of fusion
with a preceding inchoative derivation. The typical verbalizing suffixes are -ees
(Kabeena, South Cushitic), -ood (South Cushitic). In Kabeena there is also -aar
(Crass 2005: 153).

29. Structure of the simple clause
The coding of relationships between noun phrases and the predicate include coding on
the verb, subject and object pronouns in an inflectional complex, case marking on the
noun phrase, preverbal adverbial case clitics, postpositions, linear order with respect
to an inflectional complex, incorporation by the verb, by verbal derivation, by pause
and other intonational means.
The syntax of many Cushitic languages is primarily governed by pragmatic
principles. In a number of Cushitic languages there is a subject clitic (selector, see
section 27) which is independent of the verb and which plays a role in focus. For
some of these languages the subject clitic marks the beginning of what has been
termed the verbal piece or the verbal complex and which ends with the verb. Even
though this syntactic unit may contain the object, it is not comparable to a verb phrase
for various reasons in the different languages. One such reason is that non-objects and
sometimes subjects also occur within the verbal piece. Languages in which such a unit
can be recognised are Somali and the South Cushitic languages. These languages
show signs of polysynthicity.

30. Syntactic categories and relations
The syntactic relation of subject is expressed by inflection on the verb. The object is
less uniformly encoded. Some languages have an enclitic object pronoun. In most
Somali dialects this pronoun is empty for the third person. Despite gender and person
(and sometimes number) agreement of the subject on the verb and, in some cases, of
an object in an object pronoun, the subject and object are not always uniquely
distinguished. Independent pronouns often do not distinguish between the subject and
object and are not obligatory. In the northern languages the subject and object may be
62
distinguished by word order. In the Lowland East Cushitic and South Cushitic
languages word order is determined by information structure.

31. Position of the verb in the clause
The verb tends to be final in the sentence. Most languages allow for material to appear
after the verb for pragmatic functions, not only as an afterthought, e.g. Somali,
Alagwa, Burunge. The position of the verb does not distinguish subject and object.
The position immediately before the verb is one for out of focus. Nouns can form a
phonological and intonational unit with the verb for pragmatic reasons showings signs
of noun incorporation; see Sasse (1984) for Boni and related languages, Kooij and
Mous (2002) for Iraqw, Tosco (2004) for Somali, and Kieling (2007) for Alagwa.
See also section 37 on topic and focus.

32. Coding the second argument (object)
The coding of the second argument, the object, is most straightforward for those
languages that have accusative case. For example, in Awngi the object is marked by
the accusative case suffix:

(42) nn una iws nkants
this:FEM woman his/her:MASC.GEN:DAT lover:DAT
ost rg digay
milk:ACC:and honey:ACC she:presented:PREDISTINATIVE:DURATIVE
This woman used to treat her lover to milk and honey. (Hetzron 1976: 34)

However, the majority of the languages have a subject case system in which
the object is not marked. In such languages objects may be defined on the basis that
they can undergo passivization. Such is the case in Oromo: me in (43a) is an object
because it can become the subject of the passive verb of (43b), but house in (44a) is
not an object, because it cannot become the subject of a passive verb (as in 44b)
(Owens 1985a: 167).

(43) a. inni n arke
he me saw
He saw me.
b. an n-n ark-am-e
I FOC-I see-PAS-PAST
I was seen.

(44) a. inni man deeme
he house went
He went to the house.
b. *na-ni n deem-am-e
house-NOM FOC go-PAS-PAST

In other languages objects are the arguments that do not trigger agreement and
thus are not subjects and that are not marked by case clitics or adpositions. Thus, the
object is negatively defined among the arguments (Sasse 1984: 245). Grammatical
relations tend not to be the most central organisational principle in Cushitic syntax.
63
In a number of languages objects are the arguments that can be referred to
with object pronouns. Such is the case in the South Cushitic languages. For example,
in Iraqw the feminine object pronoun a agrees with the object beer (45), and
replaces an understood object in (46).

(45) buura a-ga wh
beer O.F-PERF drink:1.SG
I drank beer.

(46) g-a-na alhhe'es
O.3-O.F-PAST finish:3.SG.M:PAST
He finished it (i.e. the field (f)). (Mous 1993: 244)

33. Coding a third argument (adverbial case, postposition)
There are two different ways to code a third argument in Cushitic. One is by means of
an adverbial case clitic. This is either linked to the noun or syntactically linked to the
verb in which case it may end up on the wrong nominal (anti-iconicity); see section
15 on non-core cases and clitics. The other manner is by means of an adposition (see
section 16 on adpositions) which is sometimes a clitic.
The semantics of a case clitic can be quite diverse. For example, the case clitic
=nu in Tsamakko marks the beneficent (47), the representative (48), the goal (49),
the purpose (50), the locative direction (51), and the basis for comparison (52) (Sav
2005: 103-107).

(47) baare abba kaayu=nu pa-o qod-as-i
Baare father PRON.M.1SG.M.POSS=from field-M plough-Caus1-3SG.M.UNM
Baare ploughed the field to the benefit of my father.

(48) bogol-k-o=nu qol-e cox-inda
king-SG-M=from cattle-P milk-PLUR.IMP.B
Milk the cattle on behalf of the king!

(49) laabl-e gaan-t-e=nu eed-i
cloth-F woman-SG-F=from bring-1SG.UNM
I brought the cloth to the woman.

(50) korkor-o=nu gor-e ergad-e qets-inki
house.wall-M=from people-P assemble-3PL.UNM cut-3PL.CONS.A
The people assembled and cut (wood) for (building) the wall of the house.

(51) inka=nu kol-i
Jinka=from return-1.SG.UNM
I returned from Jinka.

(52) baqqala mia=nu qarra ki dal-ad-i
Baqqala Mia=from before Sent.3 give.birth-MID-3SG.M.UNM
Baqqala was born before Mia.

64
Locatives are often expressed through a combination of locative nouns and
adpositions or case clitics. In the following Konso example, the locational noun xati is
followed by the clitic pa and again by a directional marker.

(53) goyra tika kap-ee deh-e ma xati-pa-xa ca
tree house near-3 grow-PF but down-DEST-downwards be:IPF
The tree grows near the house, but further down. (Daudey and Hellenthal
2004: 87)

34. Clause chaining
There are several strategies to link clauses. One common strategy is to have a series of
subordinate predicates to the final main verb. An example from a long stretch of such
subordinate verb forms or converbs is the following string in Awngi.

(54) ndesk n aq lngiswa berwa
and-from-that that man both(-ACC) oxen-(ACC)
keseram ntakntak zursi,
lost-and both-here-and-there while-he-was-in-the-state-of-turning,
ji lngigi kecernuda btda lngiso
they both in-that-they-fixed in-place both(-ACC)
berwa aska kaskam aredkam kam
oxen-(ACC) they-took-and they-went-and they-slaugthered-and they-ate-and
widnids falleng, ndegena demeka
from-finishing after, again once-more
tay dade kans! tn.
sheep-(ACC) to-steal let-us-go! they-said-to-each-other.
(Hetzron 1969: 11)
Then while the man, having lost both oxen, was turning here and there, they
both (the thieves) went to the fixed place taking both oxen with them, and
slaughtered and ate them; after they finished, they said again to each other:
Let us go to steal sheep!.

Another strategy is to concatenate clauses with a coordinating particle. This
strategy is common in Konso stories. The clause-coordinating clitic -ka appears in the
position after the subject in the second clause (see 55).
(55) [isheeta i xaa-t-i-][ka dagint-aadd-i yag-at-i-]
she 3 wake-F-PF-and body-3SG.POSS-3 wash-MID:F-PF-
[ka hapurss-at
i
] [nes-att-i]
and dress-MID:F-PF rest:MID:F-PF
[tik (kara) saha-t-i][-ka sekkammaa-yy sook-t-
i
]
house (inside) clean-F-PF-and here.after-SET leave-F-PF
She got up, washed herself, got dressed, cleaned the house and went out.
(Mous 2006)
(56) [arp-oo-se ana turaa xa'-ad-e] [ka aan-ee] [takal-ee pi'-e]
elephant-REF-DEM me front flee-MID-PF and go-PF cliff-SET fall-PF
[ka qeq-qep-e]
and INT-break-PF
65
[ka xosaltaa paay-e] [ka oppaa-ee-w paq-e] [ka twee].
and laughter start-PF and on-SET-too burst-PF and die:PF
That elephant fled from me and left and fell into the ravine and broke into
pieces. And he (bedbug) started to laugh and likewise burst on it and (in doing
so) died. (Mous 2006, example from Korra Garra 2003)

Yet another common strategy is tail-head linking, which is common, for
example, in Alagwa. In a story new entities are usually introduced in the post-verbal
position, as is the case in the first sentence of (57); in the next sentence this previously
introduced entity, troughs, now appears sentence-initially and with a referential
demonstrative, while the new entity, milk, appears in the post-verbal position; in the
next sentence this information in repeated and the sentence is marked as being
background information. Such sequences and repetitions for cohesion are typical for
narrative style (Mous 2001).
(57) i-n hts-is mlambabee;
S3-PF full-CAUS:3M troughs;
mlambabee-w-d i-yaa hts-ir ilibaa.
troughs-P-DEM S3-PST full-3PL milk.
ilibaa ki hats-ir-i; ...
milk DEP-S3 full-3PL-BGND
He filled troughs. Milk filled those troughs. The troughs being filled with
milk, ...

35. Negation
Negation is marked in several different ways. Negation may be marked in the
selector, i.e., the preverbal inflectional complex. This is the case in Arbore,
Dhaasanac, Somali, Boni (58), Dahalo, Iraqw, and Konso (59); and in South Cushitic
specifically for prohibitive use (60). Negation may also be expressed by using a
specific negative verbal conjugation, as is the case in most languages. The two
options may be both present in the same language, as is the case for Konso and Iraqw.
In Oromo negative verbs are formed by prefixing a particle hin to the verb which
receives a high tone on the first syllable and the dependent suffix is used for the
imperfective (61) (Owens 1985a: 66-67). Dullay uses subjunctive paradigms for the
negative; in Konso one of the negative paradigms has the subjunctive ending o but
differs from the subjunctive tonally. Zaborski (2005) provides an overview of such
negative conjugations in Cushitic and discusses such negative paradigms for Beja,
Afar, Rendille, and Arbore.

(58) idohodi hu-dtto hki
women:DEF NEG-go there
Women do not go there (while men are allowed to). (Boni, Sasse 1981: 280)

(59) n-kkin-nean-c
1SG:NEG-drink-NEG-AM
I dont drink. (Konso, Bliese and Sokka 1986: 22)

66
(60) mi-ti taa-aar
PROH-us beat-NEG.IMP
Dont beat us! (Iraqw, Mous 1993: 165)

(61) hin-dem-u
NEG-go-DEP
He is not going. (Oromo, Owens 1985a: 66)

Negative verb paradigms may develop out of a periphrastic construction
involving a negative auxiliary verbs such as rib to refuse in Beja, wee to lack and
hinna not be in Afar (Zaborski 2005: 697), and kaa be absent in Iraqw and ba
be without in Alagwa and Burunge (Kieling 2002: 382-389).

36. Questions
There are several ways to form questions. Iraqw can be taken as an example of a
language that has three different kinds of question formation. Questions are often
formed by questioning intonation with or without additional segmental material. In
Iraqw yes/no questions are formed by questioning intonation (rise in pitch followed
by an incomplete fall) and the addition of a predicative suffix to the verb which is
usually the final element of the clause (62). Content questions are often formed by the
use of a question word. In Iraqw these are sentence-final as complements of a cleft
construction with a general word as head of the relative clause sentence-initially (and
the complement of the cleft can be left out) (63). Another manner of question
formation is by prefixing m to the selector, the preverbal inflectional complex. This
asks for an object of the verb or of the case clitic, e.g. (64) and (65) in Iraqw.

(62) loos ga do-i
beans O3:O.F cultivate:3M:INTER-3:PRED
Does he cultivate beans? (Mous 1993: 287).

(63) he kung u axwes (a heem)
man:CON you.M O.M talk:3M (COP who)
Who is talking to you? (Mous 1993: 283)

(64) laar m-a ay-an
today QUES-O.F eat-1PL
What are we eating today? (Mous 1993: 287)

(65) m-a-s aaam-in
QUES-O.F-reason cry-DUR:2SG
Why are you crying? (Mous 1993: 287)

Kabeena (Crass 2005: 284) and Oromo (Stroomer 1988) combine questioning
intonation with a full realisation of the final whispered vowel. Kabeena may have an
additional question suffix ndo for a leading yes/no question (66) and the question
word in situ for information questions and no question intonation is needed (67). In
the Agaw languages the questioned element is marked by the particle ma in yes/no
questions; the interrogative pronoun in content questions is either sentence initial or
67
precedes the verb; in addition a particle is added sentence-finally (Hetzron 1976: 38-
39).

(66) samaan
i
taafaa udulanu manc
u

sky teff:ACC thresh:IMPFV:3M:REL man:NOM
lallane-he-ndo?
show:IMPFV:3M-2SG.OBJ-EMPH
Dont you see the man there in the sky who is threshing teff? (Kabeena,
Crass 2005: 284)

(67) wombisanaan
i
komboosana yu ma
Kabeena:INSTR komboosana:ACC say:VN:NOM what:ACC
yoh
a

say:VN:ACC:COP.M
What is komboosana in Kabeena? (Kabeena, Crass 2005: 285)

(68) kt al gaytir-a
you where you:will:settle-QUES
Where will you settle? (Agaw, Hetzron 1976: 39)

Question words are sentence initial and marked as topics in Somali and Oromo
(69) (Sasse 1977: 348-349).

(69) eenn-tti arge nam-tti arge
who-TOP see:PF:3M man-TOP see:PF:3M
Who saw it? It was the man who saw it. (Borana Oromo, Sasse 1977: 348)

37. Topic and Focus
The syntax of Cushitic languages is primarily pragmatically organised. Focus
constructions are common and often involve cleft constructions, as in (70) for
Khamtanga. Appleyard (1989) points out that this is an areal phenomenon for
Ethiopia and shared with Amharic and Tigrinya.

(70) wmbriz digil g
w
yyrd an
chair:DEF:OF TOP:DEF:on (SUB)REL:1SG:sit 1SG COP
It is I who am sitting on the chair. (Khamtanga, Appleyard 1989: 301).

One of the possible functions of the inflectional complex, the selector (see
section 27 above), is that of indicating the (type of) focus. In Somali the selector, or
indicator particle, is attached to the focus marker. In the following examples the type
of focus marker indicates the type of focus: subject focus (71), verb phrase focus (72)
or complement focus (73).

(71) naag baa libaax aragtay
woman FM lion saw:she
A WOMAN has seen a lion.

(72) Cali moos w-uu cunay
Ali banana FM-he ate:he
68
Ali HAS EATEN a banana.

(73) Cali wax-uu cunay moos
Ali FM-he ate:he banana
Ali has eaten a BANANA.

Languages with a separate inflectional complex preceding the verb have the
option to utilise the position between the inflection complex and verb for
backgrounding or out-of-focus expression. Iraqw is such as language; compare (74a)
and (74b) where the coffee is backgrounded in (74a).

(74)(a) a kahaw wh (b) kahawa u wh
S.1/2 coffee:CON drink:1.SG coffee O.M drink:1.SG
I use coffee; I am a coffee drinker. I drink coffee

This phenomenon comes close to object incorporation, although true object
incorporation is still different in Iraqw, as it requires a bare noun object without the
construct-case marking (as in (75)). The properties of Iraqw noun incorporation are
discussed in Kooij and Mous (2002). Noun incorporation for Somali is discussed in
Tosco (2004). Sasse (1984) shows the out-of-focus function of noun incorporation in
Bayso, Burji and Boni (see also Sasse 1981). The examples in (76) show the different
focus types in Boni where the non-focus position in immediately before the verb.

(75) a-ga hee gas
S.1/2-PAST man kill
I committed manslaughter.

(76)a. hc-idohoo biyo=taaka
SGLTV-woman water=drink:IMPFV:3F
The woman drinks water.
b. hc-idohoo biy- taaka
SGLTV-woman water-NOUN.FOCUS drink:IMPFV:3F
The woman drinks WATER.

c. hc-idohoo biyo -taaka
SGLTV-woman water verb.FOCUS-drink:IMPFV:3F
The woman DRINKS water. (Boni, Sasse 1984: 252-253)

Cushitic languages make ample use of focus clitics to indicate several types of
focus/contrast on specific phrases. In Oromo, for example, the preverbal clitic hin
indicates that both the subject and the predicate are focussed (77), whereas a post NP
clitic -tu indicates contrast (78); the particle da is used for contrast on PPs (79) (see
also Clamons et al. 1993).

(77) Tlluu-n hin-duf-a
T-NOM FOC-come-3M-IMPF
Tulluu will come. (Baye 1988: 368)

69
(78) Tulluu-tu hoolaa bit-e
T-CONTRAST sheep buy:3M-PF
It is Tulluu who bought a sheep. (Baye 1988: 372)

(79) (Tulluu-n) eeboo-da-n leena aee-s-e
T-NOM spear-FOC-with lion kill-CS-3M-PF.
It is with a spear that Tulluu killed a lion. (Baye 1988: 379)

There is a topic position preceding the sentence and followed by a pause in,
for example, Iraqw and Somali. In the Iraqw example (80) the first noun phrase, the
road (that was magically cut in the lake) is the topic but it does not reappear as the
subject (lake) or the object (them, the cannibal clan) in the remainder of the sentence.

(80) balbal-d, tlawi gi-na bara-d harak
road-DEM4 lake O3:O.P-PAST in-DEM4:DIR return:3SG.F
About that road, the lake returned them into it. (Iraqw, Mous 1993: 274).

In Somali a subject that is not in focus is realised as a left-hand (sentence-
initial) external (extra-sentential) topic (Frascarelli and Puglielli 2007: 123):

(81) Cali MOOS buu cunay.
Cali banana FM.SCL3SGM eat.PAST.3SGM
As for Cali, he ate a BANANA.

Sentences may be marked to have no pragmatically motivated internal
structure. Tosco (2001: 263-266) shows that such topicalized sentences are
characterised by the use of a subject pronoun (and not a focus subject pronoun) in
Dhaasanac. Such sentences are characterised by the use of waa in Somali (see Ajello
1995).

38. Complex sentences
Several events are often combined into one sentence in which the final verb is the
main verb and the preceding verbs are converbs, that is, they are less finite, reduced in
person and/or tense marking and possibly marked for subordination. For example, in
Oromo pre-final perfect verb forms with the same subject tend to be marked either
prosodically by a High tone or by a suffix -ti plus vowel lengthening of the preceding;
with different subjects a gerund/converb/perfective in nan is used (Banti 2006).

(82) a. inni as duf makiina bit gale
he here come:3M:PF:H car buy:3M:PF:H return:3M:PF
He came here, bought a car and returned. Owens (1985a: 215)
b. heddu ofnan polis-ni n daabe
much drive-PER police-NOM me stopped
Because I was driving fast, the police stopped me. Owens (1985a: 151)

The Dullay and South Cushitic languages use clauses with consecutive tenses
following the main clauses instead of such converb constructions, as is clear from
the first lines of a Burunge story:
70
(83) wakailee kwai haa daw hinga
once hare and elephant 3:REFL:PRET
slaaaslaiyaya letu wak
a
kwai higi
like:FREQ:INT:3:IMPFV:3PL day one:F hare 3:SEQUEN
kaah
i
sa daw...
say:3PF for elephant
Long ago Hare and Elephant were good friends. One day Hare told
Elephant...
(Burunge, Kieling 1994: 165)

Nominalized verbs retain the ability to have an object in Iraqw. In (84) the
verbal noun is within the verbal complex in object position, but its logical object
precedes the verbal complex and is referred to with an object pronoun, which is
excluded if an object of non-verbal origin precedes the main verb.
(84) anng ayto'o a door a
1SG maize O.F cultivating:F:CON like
I would like to cultivate maize. (Iraqw, Mous 1993)

References

Ajello, Roberto. 1995. La focalizzazione in somalo. Scritti linguistici e filologici in
onore di Tristano Bolelli, ed. Roberto Ajello and Saverio Sani, pp. 1-28. Pisa:
Pacini.
Ali, Mohammed and Andzrej Zaborski. 1990. Handbook of the Oromo language.
Wroclaw: Zaklad Narodowy im. Ossolinskich Wydawnictwo Polskiej Akademii
Nauk (Aethiopistische Forschungen, 30) Stuttgart: Steiner.
Amborn, Hermann, Gunter Minker and Hans-Jrgen Sasse. 1980 Das Dullay:
Materialen zu einer ostkuschitischen Sprachgruppe. (Klner Beitrge zur
Afrikanistik, 6.) Berlin: Dietrich Reimer.
Anbessa Teferra. 2000. A Grammar of Sidaama. Ph.D. dissertation, The Hebrew
University of Jerusalem. Jerusalem.
Anbessa Teferra. 2002. Phonological government in Sidaama. In New trends in
Ethiopian Studies. pp. 1085-1101.
Andrzejewski, B.W. 1960. The Categories of Number in Noun Forms in the Borana
Dialect of Galla. Africa 30: 62-75.
Andrzejewski, B.W. 1970. The Role of Tone in the Borana Dialect of Galla. In
Proceedings of the II international conference of Ethiopian studies, Addis Ababa
1966, II, pp. 88-98. Addis Ababa: Institute of Ethiopian Studies.
Andrzejewski, B.W. and I.M. Lewis. 1964. Somali poetry: an introduction. Oxford:
Clarendon Press.
Andrzejewski, B.W. with Sheila Andrzejewski. 1993. An anthology of Somali poetry.
Bloomington: Indiana University Press.
Appleyard, David. 1975. A descriptive outline of Kemant. BSOAS 38: 316-350.
71
Appleyard, David. 1984. The morphology of the negative verb in Agaw. Transactions
of the Philological Society 1984: 202-219.
Appleyard, David. 1986a. The radical extension system of the verb in Agaw. In
Ethiopian Studies: Proceedings of the Sixth International Conference, Tel Aviv,
14-17 April 1980, ed. Gideon Goldenberg, pp. 1-23. Rotterdam: A.A. Balkema.
Appleyard, David. 1986b. Agaw, Cushitic and Afroasiatic: the personal pronoun
revisited Journal of Semitic Studies 31(2): 195-236.
Appleyard, David. 1987a. A grammatical sketch of Khamtanga I, II. BSOAS 50: 241-
266, 470-507.
Appleyard, David. 1987b. Reinischs work on Agaw and its significance today. In Leo
Reinisch: Werk und Erbe (sterreichische Akademie der Wissenschaften,
Sitzungsberichte der phil.-hist. Klasse, 492.), ed. Hans Mukarovsky, pp. 97-106.
Wien: Verlag der Akademie.
Appleyard, David. 1987c. The vowel system of Agaw: reconstruction and historical
inferences. In Proceedings of the fifth International Hamito-Semitic Congress, ed.
Hans Mukarovsky, vol. 2, pp. 12-38. (Beitrge zur Afrikanistik, 57). Vienna: Afro-
Pub.
Appleyard, David. 1988a. The Agaw languages: a comparative morphological
perspective. In Proceedings of the Eighth International Conference on Ethiopian
Studies (vol 1.), ed. Taddese Beyene, pp. 581-92. Addis Abeba: Institute of
Ethiopian Studies.
Appleyard, David. 1988b. The definite article in Xamtanga. ALC 1(1): 15-24.
Appleyard, David. 1989. The relative verb in focus constructions: an Ethiopian areal
feature. Journal of Semitic Studies 34(2): 291-305.
Appleyard, David. 1990. Prepositional particles in Somali and their cognates in other
Cushitic languages. African Languages and Cultures 3(1): 15-31.
Appleyard, David. 1991. The rle of tone in some Cushitic languages. York Papers in
Linguistics 15: 5-32.
Appleyard, David. 1992. Vocalic ablaut and aspect marking in the verb in Agaw.
JAAL 3: 126-150.
Appleyard, David. 1994. A Falasha prayer text in Agaw. In Semitic and Cushitic
Studies, ed. Gideon Goldenberg and Shlomo Raz, pp. 206-251. Wiesbaden:
Harrassowitz.
Appleyard, David. 1996a. Kalia a "new" Agaw dialect and its implications for
Agaw dialectology. In Voice and Power. The Culture of Language in North-East
Africa, ed. R.J. Hayward and I. Lewis, pp. 1-19. London: SOAS.
Appleyard, David. 1996b. The position of Agaw within Cushitic. In Studies in Near
Eastern languages and Literatures. Memorial Volume for Krel Petrtek, ed. Petr
Zemnek. pp. 1-14. Praha: Academy of Sciences of the Czech Republic Oriental
Institute.
Appleyard, David. 1999. Language Death the case of Qwarenya (Ethiopia). In
Endangered Languages in Africa, ed. Matthias Brenzinger. pp. 143-161. Cologne:
Rdiger Kppe.
72
Appleyard, David. 2001. The verb 'to say' as a verb "recycling device" in Ethiopian
languages. New Data and New Methods in Afroasiatic Linguistics: Robert Hetrzon,
in Memoriam, Andrzej Zaborski, ed., pp. 1-11. Wiesbaden: Harrassowitz.
Appleyard, David. 2006. A comparative dictionary of the Agaw languages. Cologne:
Rdiger Kppe.
Argaw, Makonnen and Gerard Philippson. 1988[1991]. Essai de description du parler
maccha de la langue oromo. Bulletin des tudes africaines 8(16): 5-51.
Arvanites, Linda. 1991 The Glottalic Phonemes of Proto Eastern Cushitic. PhD
dissertation. UCLA , Los Angeles.
Azeb Amha, Maarten Mous and Graziano Sav (eds.). 2006. Omotic and Cushitic
Language Studies: Papers from the Fourth Cushitic Omotic Conference, Leiden,
10-12 April 2003. Cologne: Rdiger Kppe.
Banti, Giorgio. 1984. The Morphology of the Nominative Case in Somali. In The
Fifth International Phonology Meeting, June 25-28, 1984, Eisenstadt, Austria,
Discussion Papers, ed. Wolfgang Dressler, Oskar E. Pfeiffer and John R.
Rennison, pp. 27-31. (Wiener Linguistischer Gazette, beiheft 3). Vienna: Institut
fr Sprachwissenschaft des Universitt Wien.
Banti, Giorgio. 1987a. Evidence for a second type of suffix conjugation in Cushitic.
In Proceedings of the fourth International Hamito-Semitic Congress, ed. Herrmann
Jungraithmayr and W. Mller, pp. 123-168. (Current Issues in Linguistic Theory,
44). Amsterdam: John Benjamins.
Banti, Giorgio. 1987b. Invariable Verbal Paradigms in some Omo-Tana Languages.
In Proceedings of the fifth International Hamito-Semitic Congress, ed. Hans
Mukarovsky, vol. 2, pp. 29-47. (Beitrge zur Afrikanistik, 57). Vienna: Afro-Pub.
Banti, Giorgio. 1988a. Two Cushitic systems: Somali and Oromo nouns. In
Autosegmental Studies in Pitch Accent, ed. Norval Smith and Harry van der Hulst,
pp. 11-49. Dordrecht: Foris.
Banti, Giorgio. 1988b. Adjectives in East Cushitic. In Cushitic - Omotic: Papers
from the International Symposium on Cushitic and Omotic Languages, Cologne,
January 6-9, 1986, ed. Bechhaus-Gerst, Marianne and Fritz Serzisko, pp. 205-259.
Hamburg: Helmut Buske.
Banti, Giorgio. 1994. Some Concord-less Verbal paradigms in Omo-Tana. Sprache
und Geschichte in Afrika 12/13: 9-40.
Banti, Giorgio. 2001. New perspectives on the Cushitic Verbal System. In
Proceedings of the twenty-seventh annual meeting of the Berkeley Linguistics
Society, March 22-25, 2001. Special Session on Afroasiatic Languages, ed.
Andrew Simpson, pp. 1-48.
Banti, Giorgio. 2006. Remarks on the typology of converbs and their functional
equivalents in Cushitic. Paper presented at Workshop on converbs, medial verbs,
clause chaining and related issues, Leiden University, 8.12.2006.
Baye Yimam 1988. Focus in Oromo. Studies in African Linguistics 19(3): 365-384.
Baye Yimam. 1987. Relative clauses in Oromo. Journal of Ethiopian Studies 20: 60-
74.
73
Bechhaus-Gerst, Marianne and Fritz Serzisko (eds.). 1988. Cushitic - Omotic: Papers
from the International Symposium on Cushitic and Omotic Languages, Cologne,
January 6-9, 1986. Hamburg: Helmut Buske.
Berger, Paul with Roland Kiessling (ed). 1998. Iraqw texts. (Archiv afrikanistischer
Manuskripte, 4). Cologne: Rdiger Kppe.
Biber, Douglas. 1984. The diachronic development of preverbal case markers in
Somali. Journal of African Languages and Linguistics 6: 47-61.
Black, Paul and Shako Otto. 1973. Konso dictionary. Ms
Black, Paul. 1973. Draft sketch of Konso Phonology, Morphology and Syntax. Ms.
Black, Paul. 1974a. Lowland East Cushitic: Subgrouping and Reconstruction. PhD
dissertation, Yale University.
Black, Paul. 1974b. Regular metathesis in Gidole. Folia orientalia 15: 47-54.
Blaek, Vclav and Mauro Tosco. 1994. Between South and East Cushitic:
reconsidering the Position of Dahalo. Paper presented at the XI. Afrikanistentag,
Cologne, 19-21 September 1994.
Blaek, Vclav. 2003. Fauna in Beja lexicon. A fragment of a comparative-
etymological dictionary of Beja. In Studia Semitica: Festschrift for Alexander
Militarev, ed. L. Kogan, pp. 230-294. Moscow: Russian State University for
Humanities.
Bliese, Loren and Sokka Gignarta. 1986. Konso exceptions to SOV typology. Journal
of Ethiopian Studies 19: 1-40.
Bliese, Loren F. 1981. A generative grammar of Afar. Dallas: SIL.
Bynon, James (ed.). 1984. Current progress in Afro-Asiatic linguistics: Papers of the
Third International Hamito-Semitic Congress (Current Issues in Linguistic Theory,
28.) Amsterdam: John Benjamins.
Bryan, Margaret. 1959. The T/K languages: a new substratum. Africa 29: 1-21.
Carlin, Eithne and Maarten Mous. 1995 The back in Iraqw: Extensions of meaning
in space. Dutch Studies-NELL 2: 121-133.
Castellino, G.R. 1975. Gender in Cushitic. In Hamito-Semitica: proceedings of a
colloquium held by the Historical Section of the LinguisticsAssociation (Great
Britain) at the School of Oriental and African Studies, University of London, on the
18th, 19th and 20th of March 1970, ed. James and Theodora Bynon, pp. 333-359.
The Hague: Mouton.
Cerulli, Enrico. 1938. Studi Etiopici 2: La lingua e la storia del Sidamo. Roma:
Istituto per lOriente.
Clamons, Cynthia R. 1992 Gender in Oromo. PhD thesis University of Minnesota.
Clamons, Cynthia Robbin, Ann E. Mulkern and Gerald Sanders. 1993. Salience
signaling in Oromo. Journal of Pragmatics 19(6): 519-536.
Clamons, Robbin. 1999. How recent contact erased ancient traces in the gender
systems of the Oromo dialects. Berkeley Linguistic Society 21: 389-400.
74
Cohen, David. 1988. Coushitique-Omotique. In Les langues dans le monde ancien et
moderne, ed. J. Perrot; Part 3: Les langues Chamito-Smitiques, ed. D. Cohen, pp.
243-295. Paris: ditions du CNRS.
Cohen, David, Marie-Claude Simeone-Senelle and Martine Vanhove. 2002. The
grammaticalization of 'say' and 'do': An areal phenomenon in East Africa. In
Reported discourse: a Meeting ground for different linguistic domains, ed. Tom
Gldemann and Manfred von Rocador, pp. 227-251. Amsterdam: John Benjamins.
Comrie, Bernard. 1981. Language Universals and Linguistic Typology. Oxford:
Blackwell.
Corbett, Greville G. 1991. Gender. Cambridge: Cambridge University Press.
Corbett, Greville G. 2000. Number. Cambridge: Cambridge University Press.
Corbett, Greville G. and Richard Hayward 1987. Gender and number in Bayso.
Lingua 73: 1-28.
Crass, Joachim. 2005. Das Kabeena: deskriptive Grammatik einer
hochlandostkuschitischen Sprache. (Kuschitische Sprachstudien, 23). Cologne:
Rdiger Kppe.
Daudey and Hellenthal. 2004. Some morpho-syntactic aspects of the Konso language.
MA thesis African Linguistics, Leiden University.
Diriye Abdulahhi, Mohamed.2002. A linguistic Biliography of the Somali Language.
htttp://www.geocities.com/mdiriye/biblio_somali.html. Consulted 8.1.2007.
Dolgopolsky, Aron B. 1987. South Cushitic Lateral Consonants as compared to
Semitic and East Cushitic. In Proceedings of the fourth International Hamito-
Semitic Congress, ed. Herrmann Jungraithmayr and W. Mller, pp. 195-214.
(Current Issues in Linguistic Theory, 44.) Amsterdam: John Benjamins.
Dubinsky, Stanley, Maria-Rosa Lloret and Paul Newman. 1988. Lexical and syntactic
causatives in Oromo. Language 64(3): 485-500.
Ehret, Christopher. 1980. The historical reconstruction of Southern Cushitic. (Klner
Beitrge zur Afrikanistik, 5) Berlin: Dietrich Reimer.
Ehret, Christopher. 1987. Cushitic Reconstruction. Sprache und Geschichte in Afrika
(SUGIA) 8: 7-180.
Ehret, Christopher 1998. An African classical age: eastern and southern Africa in
world history, 1000 B.C. to A.D. 400. Charlotsville: University of Virginia Press.
Ehret, Christopher and Derek Nurse. 1981. The Taita Cushites. Sprache und Geschichte
in Afrika (SUGIA) 3: 125-168.
Esser, Otto. 1991. Struktur, Realisation und Einsatz eines phonetisch-orientierten
Datenbanksystems, ein System zur Erfassung Verarbeitung und Auswertung
akustisch-phonetischer und linguistisher Daten auf dem PC unter Einbeziehing
vernakulrsprachliger Korpora. PhD Cologne.
Fallon, Paul D. 2006. Consonant Mutation and Reduplication in Blin Singulars and
Plurals. In Selected Proceedings of the 35th Annual Conference on African
Linguistics, ed. John Mugane et al., 114-124. Somerville, MA: Cascadilla
Proceedings Project.
75
Ferguson, Charles A. 1970. The Ethiopian Language Area. Journal of Ethiopian
Studies 8(2): 67-80.
Flemming, Harold C. and Marvin L. Bender 1976. Non-Semitic Languages. In
Language in Ethiopia, ed. by Marvin L. Bender, J.D. Bowen, R.L. Cooper, C.A.
Ferguson, pp. 34-62. London: OUP.
Frascarelli, Mara and Annarita Puglielli. 2007. Focus Markers and Universal
Grammar. In Omotic and Cushitic Language Studies: Papers from the Fourth
Cushitic Omotic Conference, Leiden, 10-12 April 2003, ed. by Azeb Amha,
Maarten Mous and Graziano Sav, pp. 119-134. Cologne: Rdiger Kppe.
Gasparini, Armido. 1983 Sidamo-English Dictionary. Bologna: Editrice Missionaria
Italiana.
Gasparini, Armido. 1994. Grammatica gedeo. Trieste: Univ. degli Studi di Trieste;
Roma: Missionari Comboniani.
Gebre-Tsadik, Abebe; M.G. Sim, R.J. Sim, C. Wedekind, K. Wedekind. 1985. The
Verb morphophonemics of five Highland East Cushitic Languages Including Burji.
AAP supplement 2.
Gensler, Orin. 2000. Proto-Afroasiatic as a Marked Nominative Language. Paper
presented at WOCAL3, Lom (Togo).
Gordon, Raymond G., Jr. (ed.), 2005. Ethnologue: Languages of the World, Fifteenth
edition. Dallas, Tex.: SIL International. Online
version: http://www.ethnologue.com/.
Gragg, Gene 2001. Cushitic Languages In Historische semitische Sprachwissenschaft,
ed. Burkhart Kienast, pp. 574-613. Wiesbaden: Harrassowitz.
Gragg, Gene B. 1982. Oromo dictionary. (Monograph / Committee on Northeast
African Studies, 12) East Lansing, Michigan : African Studies Center, Michigan
State University
Griefenow-Mewis, Catherine and Rainer M. Voigt. 1996. Cushitic and Omotic
Languages: Proceedings of the Third International Symposium; Berlin, March 17-
19, 1994. Cologne: Rdiger Kppe.
Gulliver, P.H. 1958. Counting with the fingers by two East African tribes. Tanganyika
Notes and Records 51: 259-262.
Hayward, K.M. and R.J. Hayward. 1989. Guttural: Arguments for a new distinctive
feature. Transactions of the Philological Society 87(2): 179-193.
Hayward, Richard J. 1974. The segmental phonemes of Afar. BSOAS 37(2): 385-406.
Hayward, Richard J. 1975. Middle voice verb forms in Eastern Cushitic. Transactions
of the Philological Society 1975: 203-244.
Hayward, Richard J. 1976. A question in Oromo morphophonology. Folia Orientalia
17: 29-40.
Hayward, Richard J. 1978a. Bayso revisited: Some preliminary linguistic observations
I. BSOAS 41(3): 539-570., II BSOAS 42(1): 101-132.
Hayward, Richard J. 1978b. The prefix conjugation in Afar. In Atti del seconda
congresso internazionale di linguistica camito-semitica, Firenze, 16-19 aprile
76
1974 (Quaderni di semitistica 5), ed. Pelio Fronzaroli, pp. 355-368. Firenze:
Istituto di linguiatica e di lingue orientali, Universit di Firenze.
Hayward, Richard J. 1978c. The stative conjugation in Afar. Annali dellIstituto
Orientale di Napoli 38(1): 1-39.
Hayward, Richard J. 1980. Participles in Afar: Evidence for the restructuring of Verb
Suffixes. BSOAS 43(1): 123-131.
Hayward, Richard J. 1981. Nominal Suffixes in Dirayta (Gidole). BSOAS 44: 126-
141.
Hayward, R.J. 1983. Some aspects of the phonology of ultimate vowels in Saho-
Afar. In Ethiopian Studies Dedicated to Wolf Leslau, ed. S. Segert and
A.J.E.Bodroglietti, pp. 221-231. Wiesbaden: Harrassowitz.
Hayward, Richard J. 1984a. The Arbore language: A First Investigation Including a
Vocabulary. (Kuschitische Sprachstudien, 2.) Hamburg: Helmut Buske.
Hayward, Richard J. 1984b. A reconstruction of some root extensions of the Eastern
Cushitic verb. In Current progress in Afro-Asiatic linguistics: Papers of the Third
International Hamito-Semitic Congress, ed. James Bynon, pp. 69-109. (Current
Issues in Linguistic Theory, 28). Amsterdam: John Benjamins.
Hayward, Richard J. 1986. Some observations on Dirayta (Gidole) Pronouns. In
Ethiopian Studies: Proceedings of the Sixth International Conference, Tel Aviv,
14-17 April 1980, ed. Gideon Goldenberg, pp. 275-294. Rotterdam: A.A. Balkema.
Hayward, Richard J. 1987. Terminal vowels in Ometo nominals. In Proceedings of
the fourth International Hamito-Semitic Congress, ed. Herrmann Jungraithmayr
and W. Mller, pp. 215-231. (Current Issues in Linguistic Theory, 44).
Amsterdam: John Benjamins.
Hayward, Richard J. 1988. Is there a language with an indefinite nominative - Burji?
In Proceedings of the international conference of Ethiopian Studies, vol. 1., ed.
Taddese Beyene, pp. 679-691. Addis Ababa: Institute for Ethiopian Studies.
Hayward, Richard J. 1991. Tone and accent in the Qafar noun. York Papers in
Linguistics 15: 117-137.
Hayward, Richard J. 1992. Underspecification theory in the analysis of the feminine
genitive in Qafar. Israel Oriental Studies 12: 139-156.
Hayward, Richard J. 1996. Compounding in Qafar. BSOAS 59(3) : 524-545.
Hayward, Richard J. 1998. Qafar (East Cushitic). In The handbook of morphology, ed.
Andrew Spencer Arnold M. Zwicky, pp. 624-647. Oxford : Blackwell.
Hayward, Richard J. 2002. Case or postposition. Afrikanistische Arbeitspapiere 71:
55-73.
Hayward, Richard J. and G.G. Corbett. 1988. Resolution rules in Qafar. Linguistics
26: 259-279.
Hayward, Richard J. and M. Orwin. 1991. The prefix conjugation in Qafar-Saho: the
survival and revival of a paradigm, part 1. African Language and Culture 4(2):
157-176.
77
Heine, Bernd. 1972/73. Vokabulare Ostafrikanischer Restsprachen 1. Elmolo. Afrika
und bersee 56: 276-283.
Heine, Bernd. 1975. Notes on the Yaaku Language (Kenya). Afrika und bersee 58:
27-61, 119-138.
Heine, Bernd. 1975/76. Notes on the Rendille language (Kenya) Afrika und bersee
59: 176-223.
Heine, Bernd. 1976. Bemerkungen zur Elmolo-Sprache. Au 59: 278-299.
Heine, Bernd. 1977. Bemerkungen zur Boni-Sprache (Kenia). Au 60: 242-295.
Heine, Bernd. 1979. The Sam Languages: A History of Rendille, Boni and Somali.
AAL 6/2.
Heine, Bernd. 1980. The Non-Bantu Languages of Kenya. (Language and Dialect
Atlas of Kenya, 2.) Berlin: Dietrich Reimer.
Heine, Bernd. 1981. The Waata dialect of Oromo: grammatical sketch and
vocabulary. (Language and Dialect Atlas of Kenya, 4). Berlin : Reimer.
Heine, Bernd. 1982. Boni Dialects. (Language and Dialect Atlas of Kenya, 10.)
Berlin: Dietrich Reimer.
Heine, Bernd and Mechthild Reh. 1984. Grammaticalisation and Reanalysis in
African Languages. Hamburg: Helmut Buske.
Heine, Bernd, Franz Rottland and Rainer Vossen. 1979. Proto-Baz: some aspects of
early Nilotic-Cushitic contacts. Sprache und Geschichte in Afrika 1: 75-91.
Hetzron, Robert. 1967. Agaw numerals and incongruence in Semitic. Journal of
Semitic Studies 12(2): 169-197.
Hetzron, Robert. 1969 The verbal system of Southern Agaw. Berkeley and Los
Angeles: University of California Press.
Hetzron, Robert. 1972. Phonology in syntax. Journal of Linguistics 8: 251-265.
Hetzron, Robert. 1976. The Agaw languages. AAL 3/3
Hetzron, Robert. 1978. The nominal system of Awgni (Southern Agaw) BSOAS 41:
121-41.
Hetzron, Robert. 1980. The limits of Cushitic. SUGIA 2: 7-126.
Hetzron, Robert. 1989. Typological Pecularities in Somali. Folia Linguistica 23: 7-26.
Hetzron, Robert. 1995. Genitival agreement in Awgni: variation on an Afroasiatic
Theme. In Double Case: Agreement by Suffixaufnahme, ed. Frans Plank, pp. 325-
335. New York: Oxford University Press.
Hodge, Carleton T. 1971. Afroasiatic: a survey. (Janua Linguarum, Series Pratica
163) The Hague: Mouton.
Hudson, Grover. 1986. Highland East Cushitic Dictionary. (Cushitic Language
Studies, 7.) Hamburg: Buske.
Hudson, Grover. 1989 Highland East Cushitic Dictionary. (Kuschitischen
Sprachstudien 7). Hamburg: Helmut Buske.
78
Hudson, Grover. 2006. Bibliography of Highland East Cushitic.
http://www.msu.edu/~hudson/HECregs.htm. Consulted 8.1.2007.
Hudson, Richard Anthony. 1964. A grammatical study of Beja. PhD thesis SOAS,
London.
Hudson, Richard Anthony. 1974. A structural sketch of Beja. African Language
Studies 15: 111-142.
Hudson, Richard Anthony. 1976. Beja. In The Non-Semitic languages of Ethiopia, ed.
M.L. Bender, pp.97-132.
Hulst, Harry van der and Maarten Mous. 1992. Transparant consonants. In
Linguistics in the Netherlands 1992, ed. Reineke Bok-Bennema and Roeland van
Hout, pp. 101-112. Amsterdam: John Benjamins.
Hyman, Larry M. 1981. Tonal accent in Somali. SAL 21(2): 169-203,
Ishetu Kebede. 1989. The copula in Oromo. In Current Approaches to African
Linguistics 5, ed. Paul Newman and Robert D. Botne, pp. 85-93. Dordrecht: Foris.
Johnson, John W. 1974. Heellooy, Heelleellooy : the development of the genre heello
in modern Somali poetry. (African series, 5). Bloomington : Indiana University.
Joswig, Andreas. 2006a. The Status of the High Central Vowel in Awngi. In
Proceedings of the XVth International Conference of Ethiopian Studies, Hamburg,
July 20-25, 2003, ed. Siegbert Uhlig, Maria Bulakh, Denis Nosnitsin and Thomas
Rave, pp. 786-792. Wiesbaden: Harrassowitz.
Joswig, Andreas. 2006b. The Phonology of Awngi. Ms. Addis Ababa: SIL Ethiopia.
Kawachi, Kuzuhiro. 2007. A Grammar of Sidaama (Sidamo), a Cushitic Language of
Ethiopia. PhD thesis University at Buffalo SUNY.
Kemmer, Suzanne. 1993. The Middle Voice. Amsterdam: John Benjamins.
Kieling, Roland. 1994. Eine Grammatik des Burunge. (Afrikanistische Forschungen,
13) Hamburg: Research and Progress Verlag.
Kieling, Roland. 2001. South Cushitic links to East Cushitic. In New Data and New
Methods in Afroasiatic Linguistics: Robert Hetzron in memoriam, ed. Andrzej
Zaborski, pp. 95-102. Wiesbaden: Otto Harrassowitz.
Kieling, Roland. 2002. Die Rekonstruktion der sdkuschitischen Sprachen (West-
Rift): Von den systemlinguistischen Manifestationen zum gesellschaftlichen
Rahmen des Sprachwandels. Cologne: Rdiger Kppe.
Kieling, Roland. 2003. Infix genesis in Southern Cushitic. In Afrasian: Selected
Comparative-Historical Linguistic Studies in Memory of Igor M. Diakonoff
(LINCOM Studies in Afroasiatic Linguistics 14), ed. M. Lionel Bender, Gabor
Takacs, and David L. Appleyard, 109-122. Munich: Lincom Europa.
Kieling, Roland. 2007. Alagwa functional sentence perspective and incorporation.
In Omotic and Cushitic Language Studies: Papers from the Fourth Cushitic
Omotic Conference, Leiden, 10-12 April 2003, ed. by Azeb Amha, Maarten Mous
and Graziano Sav, pp. 135-145. Cologne: Rdiger Kppe.
Kieling, Roland and Maarten Mous. 2003. The lexical reconstruction of West Rift
(Southern Cushitic). Cologne: Rdiger Kppe.
79
Kiflemariam Hamde. 1993. Blina Kolata ndba (Blin Dictionary). Sweden Blin
Language Research Group.
Knig, Christa. 2006. Marked Nominative in Africa. Studies in Language 30: 655-
732.
Kooij, Jan and Maarten Mous. 2002. Incorporation: a comparison between Iraqw and
Dutch. Linguistics 40(3): 629-645.
Korra Garra. 2003. Torraa Afaa Xonso 1. Leiden: Department of African Languages
and Cultures.
Kraska, Iwona. 1992. From verb to clitic and nominal suffix: the Somali -e, -o nouns.
Studies in the Linguistic Sciences 22(1): 89-106
Laitin, David D. 1977. Politics, language and thought: the Somali experience.
Chicago: University of Chicago Press.
Lamberti, Marcello. 1986a. Map of Somali dialects in the Somali Democratic
Republic. Hamburg : Helmut Buske.
Lamberti, Marcello. 1986b. Die Somali-Dialekte: eine vergleichende Untersuchung
mit 35 Karten und zahlreichen Tabellen. (Kuschitische Sprachstudien, 5).
Hamburg : Helmut Buske.
Lamberti, Marcello. 1986c. Somali language and literature. (African linguistic
bibliographies, 2). Hamburg : Helmut Buske Verlag.
Lamberti, Marcello. 1991. Cushitic and its classifications. Anthropos 86: 552-561.
Lamberti, Marcello and Eike Haberland. 1988. Ibaaddo ka-Ba'iso : culture and
language of the Ba'iso. (Studia linguarum Africae orientalis, 2) Heidelberg:
Winter.
Lecarme, Jacqueline. 1988. Subject Marking in Somali. In Proceedings of the Third
International Congress of Somali Studies, ed. Annarita Puglielli, pp. 69-77. Roma:
Il Pensiero Scientifico Editore.
Lloret, Maria-Rosa. 1987. The Morphophonology of the Causative in Oromo. JALL
9(2): 141-156.
Lloret-Romanyach, Maria-Rosa. 1988. Gemination and vowel length in Oromo
morphophonology. Ph.D. thesis Indiana University.
Lloret, Maria-Rosa. 1989. Final Vowels and Grammatical Marking in Oromo. In
Current Approaches to African Linguistics 5, ed. Paul Newman and Robert Botne,
pp. 73-83. Dordrecht: Foris.
Lloret, Maria-Rosa. 1994. On the representation of ejective and implossives. In
Phonologica 1992: proceedings of the 7th international phonology meeting, ed.
Wolfgang U. Dressler, Martin Prinzhorn and John R. Rennison, pp. 123-133.
Torino: Rosenberg and Selier.
Lloret, Maria-Rosa. 1995a. Implosive consonants: their representation and sound
change effects. Belgian Journal of Linguistics 9: 59-72.
Lloret, Maria-Rosa. 1995b. The representation of glottals in Oromo. Phonology 12(2):
257-280.
80
Lloret, Maria-Rosa. 1997. Oromo Phonology. In Phonologies of Asia and Africa
(Including the Caucasus), ed. Alan S. Kaye, pp. 493-519. Winona Lake, Indiana:
Eisenbrauns.
Luling, Virginia. 1987. Somali-English dictionary. Wheaton MD: Dunwoody Press
Militariev, A. 2005. Once more about chronochronology and the comparative method:
the Omotic-Afrasian case. In Aspects of comparative linguistics, v. 1. Moscow:
Rsuh publishers.
Moreno, Martino Mario. 1940. Manuale di Sidamo. Milano: Mondadori.
Morin, Didier. 1995. Des paroles douces comme la soie : introduction aux contes
dans l'aire couchitique (Bedja, Afar, Saho, Somali) (Langues et cultures africaines,
19) (SELAF 352). Leuven: Peeters.
Mous, Maarten. 1993. A grammar of Iraqw (Cushitic Language Studies, 9). Hamburg:
Helmut Buske.
Mous, Maarten. 1996. Broken plurals and syllable sequence restrictions in Iraqw. In
Studies in Afroasiatic Grammar: Papers from the Second Conference in
Afroasiatic Languages, Sophia Antipolis, 1994, ed. Jacqueline Lecarme, Jean
Lowenstamm en Ur Shlonsky, pp. 268-277. Den Haag: Holland Academic
Graphics.
Mous, Maarten. 2000. Ka baby kabbiyo: Riddling in Iraqw Afrikanistische
Arbeitspapiere 62: 39-69.
Mous, Maarten. 2001. Basic Alagwa syntax In New Data and New Methods in
Afroasiatic Linguistics: Robert Hetzron in memoriam, ed. Andrzej Zaborski, pp.
125-135. Wiesbaden: Otto Harrassowitz.
Mous, Maarten. 2003. The Making of a Mixed Language: The Case of Maa/Mbugu
Amsterdam: John Benjamins. .
Mous, Maarten. 2004a. The grammar of conjunctive and disjunctive coordination in
Iraqw Coordinating Constructions, ed. Martin Haspelmath. (Typological Studies
in Language, 58), pp. 109-122. Amsterdam: John Benjamins.
Mous, Maarten. 2004b. The Middle in Cushitic Languages In Simpson, A. (ed.),
Proceedings of the twenty-seventh annual meeting of the Berkeley Linguistics
Society, March 22-25, 2001: Special Session on Afroasiatic Languages, pp. 75-86.
Berkeley CA: Linguistic Society.
Mous, Maarten. 2004c. The Causative in Konso. In Kastenholz, R. and Storch, A.
(eds.), Sprache und Wissen in Afrika;Beitraege zum 15. Afrikanistentag, Frankfurt
an Main und Mainz, 30. September -2. Oktober 2002, pp. 213-228. Cologne:
Ruediger Koeppe.
Mous, Maarten. 2005. Selectors in Cushitic In Studies in African Linguistic
Typology , ed. F.K. Erhard Voeltz, pp. 303-325. Amsterdam: John Benjamins.
Mous, Maarten. 2006. Clause linking in Konso. Paper presented at the Clause linking
workshop, RCLT, Melbourne.
Mous, Maarten. 2007. The Middle and Passive derivations in Konso. In Omotic and
Cushitic Language Studies: Papers from the Fourth Cushitic Omotic Conference,
81
Leiden, 10-12 April 2003, ed. by Azeb Amha, Maarten Mous and Graziano Sav
(eds.), pp. 213-229. Cologne: Rdiger Kppe.
Mous, Maarten and Cabdelqaadir Ruumi. 2002 Hoe noem ik mijn kameel: over
bepaaldheid, getal en geslacht in het Somalisch (met Cabdelqaadir Ruumi) Paper
presented at Studium Generale, Leiden, 05.03.02
Mous, Maarten and Martha Qorro. 2000. Middle voice in Iraqw In Lugha za
Tanzania / Languages of Tanzania: Studies dedicated to the memory of Prof.
Clement Maganga, ed. K. Kahigi, Y. Kihore and M. Mous, pp. 141-159. Leiden:
CNWS.
Mous, Maarten, Martha Qorro and Roland Kieling. 2002. An Iraqw - English
Dictionary (Cushitic Language Studies,15) Cologne: Rdiger Kppe.
Murray, Robert W and Theo Vennemann. 1982. Syllable contact change in Germanic,
Greek and Sidamo. Klagenfurter Beitrge zur Sprachwissenschaft 8: 321-
349.(ego)
Nurse, Derek 1988. Extinct Southern Cushitic communities in East Africa. In Cushitic
- Omotic: Papers from the international Symposium on Cushitic and Omotic
Languages, Cologne, January 6-9, 1986, ed. by M. Bechhaus-Gerst and F.
Serzisko, pp. 93-104. Hamburg: Buske.
Oomen, Antoinette. 1978. Focus in the Rendille Clause. SAL 9(1): 35-65.
Oomen, Antoinette. 1981. Gender and Plurality in Rendille. AAL 8(1): 35-75.
Owens, Jonathan. 1980. Observations on tone in the Booran dialect of Oromo (Galla).
African Language Studies 17: 141-196.
Owens, Jonathan. 1982. Case in the Booran Dialect of Oromo. Au 65: 43-74.
Owens, Jonathan. 1985a. A grammar of Harar Oromo (Northeastern Ethiopia)
including a text and a glossary. (Kuschitische Sprachstudien, 4) Hamburg : Helmut
Buske.
Owens, Jonathan. 1985b. The Oromo Causative: Lexical Grammar without Lexical
Rules. Bloomington: IULC.
Palmer, F. 1959. The verb classes of Agau (Awiya) Mitteilungen des Instituts fr
Orientforschung 7: 270: 297.
Palmer, F. 1960. An outline of Bilin phonology. In Atti del convengo internazionale
di studi etiopici, pp. 109-116. Roma: Accademia Nazionale dei Lincei.
Palmer, F. 1965. Bilin to be and to have. African Language Studies 6: 101-111.
Palmer, F.R. 1957. The verb in Bilin. BSOAS 19: 131-159.
Palmer, F.R. 1958. The noun in Bilin. BSOAS 21: 376-91.
Palmer, F.R. 1970. Cushitic. In Current Trends in Linguistics 6, ed. T.A. Sebeok, pp. .
The Hague: Mouton.
Palmer, F. R. 1967. Word Classes in Bilin. Lingua 17:200-209.
Parker, E.M. and R.J. Hayward. 1985. An Afar-English-French dictionary (with
grammatical notes in English). London: SOAS.
82
Pia, J. Joseph. 1984. Multi-tiered vocalic inventories: an afroasiatic trait. In Current
progress in Afro-Asiatic linguistics: Papers of the Third International Hamito-
Semitic congress, ed. James Bynon, pp. 463-475. (Current Issues in Linguistic
Theory, 28.). Amsterdam: John Benjamins.
Pia, John Joseph. 1965. Somali sounds and inflections. PhD Indiana University. Ann
Arbor, Mich.: University Microfilms International). .
Pillinger, Steve. 1988. Tone and accent in Rendille: A preliminary investigation. In
Bechhaus-Gerst and Serzisko (eds).
Pillinger, Steve. 1989. Accent, Tone and Prosodic Structure in Rendille with
particular reference to the Nominal System. PhD thesis, London
Pillinger, Steve and Letiwa Galboran. 1999. A Rendille Dictionary Including a
Grammatical Outline and an Englsih-Rendille Index. (Kuschitische Sprachstudien,
14.) Cologne: Rdiger Kppe.
Plazikowsky-Brauner, Herma. 1959. Der Kausativ in den sogenannten kuschitischen
Sprachen. Anthropos 54: 129-140.
Plazikowsky-Brauner, H. 1960. Die Hadiya-Sprache. Rassegna di Studi Etiopici 16:
38-76.
Plazikowsky-Brauner, H. 1961. Texte der Hadiya-Sprache. Rassegna di Studi Etiopici
17: 83-115.
Plazikowsky-Brauner, H. 1964. Wrterbuch der Hadiya-Sprache. Rassegna di Studi
Etiopici 20: 133-182.
Plazikowski-Brauner, Herma and Ewald Wagner. 1953. Studien zur Sprache der Irob.
Zeitschrift der Deutschen Morgenlndische Gesellschaft 103 (NF 28): 378-393.
Praetorius, Franz. 1893. Zur Grammatik der Gallasprache. Berlin: Wolf Peiser.
Puglielli, Annarita. 1981. Sintassi della lingua somala (Studi somali 2) [Roma] :
Ministero degli Affari Esteri, Dipartimento per la Cooperazione allo Sviluppo:
Comitato Tecnico Linguistico per l'Universita Nazionale Somala.
Puglielli, Annarita. 1984. Aspetti morfologici, lessicali e della focalizzazione. (Studi
somali 5) [Roma] : Ministero degli Affari Esteri, Dipartimento per la Cooperazione
allo Sviluppo: Comitato Tecnico Linguistico per l'Universita Nazionale Somala.
Puglielli, Annarita et al. (eds.) 1985 Dizionario Somalo - Italiano. Rome: Gangemi
Editore.
Puglielli, Annarita. (ed.) 1998. Dizionario Italiano-Somalo. Rome: Carocci editore
and Universit degli Studi di Roma Tre, Dipartimento di Linguistica.
Reh, Mechthild 1999. Body, Back and Belly or On the antonyms of inside
and their conceptual sources Frankfurter Afrikanistische Bltter 11: 101-123
Roper, E.M. 1928. Tu BedawiE: an elementary handbook for the use of Sudan
government officials. Hertford: Stephen Austin.
Rottland, Franz and Maarten Mous. 2001. Datooga and Iraqw: A comparison of
subsistence vocabulary. In Von gypten zum Tschadsee - Eine linguistische Reise
durch Afrika: Festschrift fr Hermann Jungraithmayr zum 65. Geburtstag, ed.
Dymitr Ibriszimow and Rudolf Leger, pp. 377-400. (Abhandlungen zur Kunde des
83
Morgenlandes 53, 3) Wrzburg: Deutsche Morgenlndische Gesellschaft (Ergon
Verlag)
Saeed, John Ibrahim. 1984. The syntax of focus and topic in Somali. (Kuschitische
Sprachstudien 3) Hamburg : H. Buske.
Saeed, John Ibrahim. 1987. Somali reference grammar. Wheaton, MD : Dunwoody
Press.
Saeed, John Ibrahim. 1993. Somali Reference Grammar (second revised edition).
Kensington (Maryland): Dunwoody Press.
Saeed, John Ibrahim. 1995. The semantics of middle voice in Somali. ALC 8(1): 61-
85.
Saeed, John Ibrahim. 1999. Somali. (London Oriental and African Language Library,
10.) Amsterdam: John Benjamins.
Sasse, Hans-Jrgen. 1977. A note on Wh-movement. Lingua 41: 343-354.
Sasse, Hans-Jrgen. 1979. The consonant phonemes of Proto-East-Cushitic (PEC): a
first approximation. AAL 7.
Sasse, Hans-Jrgen. 1980. Textproben der Boni-Sprache (West-Dialekt). Au 63: 79-
101.
Sasse, Hans-Jrgen. 1981. Basic Word Order and Functional Sentence Perspective in
Boni. Folia Linguistica 15(3-4): 253-290.
Sasse, Hans-Jrgen. 1981. Die kuschitischen Sprachen. In Die Sprachen Afrikas, ed.
B. Heine, Th.C Schadeberg and E. Wolff., Teil II, pp. 187-215. Hamburg: Helmut
Buske.
Sasse, Hans-Jrgen. 1982. An Etymological Dictionary of Burji. (Kuschitische
Sprachstudien, 1.) Hamburg: Helmut Buske. reviewed by Wolf Leslau (1988),
Zaborski in JALL
Sasse, Hans-Jrgen. 1984. Case in Cushitic, Semitic and Berber. In Current progress
in Afro-Asiatic linguistics: Papers of the Third International Hamito-Semitic
Congress, ed. James Bynon, pp.111-125. (Current Issues in Linguistic Theory, 28.)
Amsterdam: John Benjamins.
Sasse, Hans-Jrgen. 1984. The pragmatics of noun incorporation in Eastern Cushitic
languages. In Objects: Towards a theory of grammatical relations, ed. Frans Plank,
pp. 243-68. London: Academic Press.
Sasse, Hans-Jrgen. 1986. A Southwest Ethiopian Language area and its cultural
background. In The Fergusonian Impact, vol. 1: From Phonology to Society, ed.
Joshua A. Fishman et al., pp. 327-342. Berlin: Mouton de Gruyter.
Sasse, Hans-Jrgen. 1987. Kuschitische Sprachen. Studium Linguistik 21: 78-99
Sasse, Hans-Jrgen. 1987. Research on the Cushitic Language Family of East Africa.
In Recent German Research on Africa: Language and Culture, pp. 28-33. Bonn:
DFG. [description of Sasses research activities]
Sasse, Hans-Jrgen. 2003. Cushitic adpositions and their possible relatives in Semitic.
Afrasian: Selected Comparative-Historical Linguistic Studies in Memory of Igor
M. Diakonoff (LINCOM Studies in Afroasiatic Linguistics 14), ed. by M. Lionel
84
Bender, Gabor Takacs, and David L. Appleyard, pp. 123-142. Munich: Lincom
Europa.
Sasse, Hans-Jrgen and Helmut Straube. 1977. Kultur und Sprache der Burji in Sd-
thiopien: Ein Abri. In Zur Sprachgeschichte und Ethnohistorie in Afrika, ed. by
W.J.G. Mhlig et al., pp. 145-165. Berlin: Dietrich Reimer.
Sav, Graziano. 2005. A grammar of Tsamakko. (Kuschitischen Sprachstudien, 22).
Cologne: Rdiger Kppe.
Sav, Graziano and Mauro Tosco. 2000. A sketch of Ongota, a dying language of
Southwest Ethiopia. Studies in African Linguistics 29(2): 59-134.
Schlee, Gnther. 1978. Sprachliche Studien zum Rendille. (Hamburger philologisch
Studien, 46.) Hamburg: Helmut Buske.
Schneider-Blum, Gertrud. 12006. Beschreibung des Alaaba. PhD thesis, University of
Cologne.
Serzisko, Fritz. 1984. Der Ausdruck der Possessivitaet im Somali (Continuum :
Schriftenreihe zur Linguistik, 1) Tbingen : Gunter Narr.
Serzisko, Fritz. 1992. Collective and transnumeral nouns in Somali. In: Proceedings
of the first international congress of Somali studies , ed. Hussein M. Adam and
Charles L. Geshekter, pp. 513-525. Atlanta, GA: Scholars Press.
Serzisko, Fritz. 1982. Numerus / Genus - kongruenz und das Phnomen der Polaritt
am Beispiel einiger ostkuschitischer Sprachen. In Apprehension: Das sprachliche
Erfassen von Gegenstnden, II Die Techniken und ihr Zusammenhang in
Einzelsprachen, ed. Hansjakob Seiler and Franz Josef Stachowiak, pp. 179-200.
(Language Universals Series, 1/II) Tbingen: Gunter Narr.
Sim, Ronald J. 1981. Morphophonemics of the Verb in Rendille. AAL 8(1): 1-33.
Sim, Ronald J. 1989 Predicate conjoining in Hadiyya: a head-driven PS Grammar.
Ph.D. thesis University of Edinburgh.
Somali Bibliography: http://www.geocities.com/mdiriye/biblio_somali.html
Stroomer, Harry. 1987. A comparative study of three southern Oromo dialects in
Kenya : phonology, morphology and vocabulary. (Cushitic language studies, 6)
Hamburg : Helmut Buske.
Stroomer, Harry. 1988 On the base form and the non-base form of words in the
Boraana, Orma and Waata dialects of Oromo In Cushitic - Omotic: Papers from
the International Symposium on Cushitic and Omotic Languages, Cologne,
January 6-9, 1986, ed. by Marianne Bechhaus-Gerst and Fritz Serzisko, pp. 455-
471. Hamburg: Helmut Buske.
Stroomer, Harry. 1995. A Grammar of Boraana Oromo (Kenya). Phonology,
morphology, vocabularies. (Kuschitische Sprachstudien, 11.)
Svolacchia, Marco. 1987. Tipologia delle ostruenti labiali in afroasiatico e
implicazioni diacroniche e teoriche. In Atti della quarta giornata di studi camito-
semitici e indoeuropei, Bergamo, Istituto Universitario, 29 novembre 1985, ed.
Giuliano Bernini and Vermondo Brugnatelli, pp. 181-193. Milano: Unicopli.
85
Svolacchia, Marco, Lunera Mereu and Annarita Puglielli. 1995. Aspects of Discourse
Configurationality in Somali. In Discourse Configurational Languages, ed. Katalin
Kiss, pp. 65-98. New York: Oxford University Press.
Tosco, Mauro. 1991. A grammatical sketch of Dahalo: including texts and a glossary.
(Kuschitische Sprachstudien, 8). Hamburg: Helmut Buske.
Tosco, Mauro. 1994a. On Case Marking in the Ethiopian Language Area (with special
reference to subject marking in East Cushitic. In Sem, Cam, Iafet, ed. V.
Brugnatelli, pp. 225-244. Milan: Centro Studi Camito-Semitici.
Tosco, Mauro. 1994b. The historical syntax of East Cushitic: A first sketch. In
Perspectiven afrikanistischer Forschung. Beitrge zur Linguistik, Ethnologie,
Geschichte, Philosophie und Literatur, X. Afrikanistentag (Zrich, 23-25
September 1993), ed. by Thomas Bearth et al., pp. 415-440. Cologne: Rdiger
Kppe
Tosco, Mauro. 1996. The strange case of the Dirayta subject case. JALL 17: 27-45.
Tosco, Mauro. 1997. Af Tunni: Grammar, Texts, and Glossary of a Southern Somali
Dialect. (Kuschitische Sprachstudien, 13.) Cologne: Rdiger Kppe.
Tosco, Mauro (with Cabdulqaadir Salaad Dhoorre) 1998. Somali ideophones. Journal of
African Cultural Studies 11(2): 125-156
Tosco, Mauro. 2000a. Cushitic overview. Journal of Ethiopian Studies 33(2): 87-121.
Tosco, Mauro. 2000b. Is there an Ethiopian language area? Antrhopological
Linguistics 42(3): 329-365.
Tosco, Mauro. 2001. The Dhaasanac Language. (Kuschitischen Sprachstudien, 17).
Cologne: Rdiger Kppe.
Tosco, Mauro. 2003. Cushitic and Omotic overview. In Afrasian: Selected
Comparative-Historical Linguistic Studies in Memory of Igor M. Diakonoff
(LINCOM Studies in Afroasiatic Linguistics 14), ed. by M. Lionel Bender, Gabor
Takacs, and David L. Appleyard, pp. 87-92. Munich: Lincom Europa.
Tosco, Mauro. 2004. Between zero and nothing: between transitivity and noun
incorporation in Somali. Studies in Language 28(1): 83-104.
Tosco, Mauro with Cabdulqaadir Salaad Dhoorre. 1998. Somali ideophones. Journal
of African Cultural Studies 11(2): 125-156.
Treis, Yvonne. 2005a. Verbal Inflection in Kambaata (Highland East Cushitic /
Ethiopia) Paper presented on 2nd December 2005 at Leiden University.
Treis, Yvonne. 2005b. Does Kambaata have adjectives. Paper presented at the 35th
Colloquium on African Languages and Linguistics, Leiden, August 29-31, 2005. .
Treis, Yvonne. 2007. Towards a Grammar of Kambaata: Phonology, Nominal
Morphology and Non-verbal Predication. Unpublished PhD thesis, University of
Cologne.
Tucker, A.N. and M.A. Bryan. 1966. Linguistic Analyses: The Non-Bantu Languages
of North-Eastern Africa. Londen: OUP.
86
Vanhove, Martine. 2000. Notes sur les verbes statifs en afar de Tadjoura (Djibouti). In
Proceedings of the 2nd World Congress of African Linguistics Leipzig 1997, ed. H.
Ekkehard Wolff and Orin D. Gensler, pp. 773-786. Cologne: Rdiger Kppe.
Vanhove, Martine. 2004. A propos des nasales en bedja. In Langues et cultures:
terrains d'Afrique, Hommage France Cloarec-Heiss, ed. by P. Boyeldieu et P.
Nougayrol, pp. 271-279. Louvain: Peeters.
Vanhove, Martine. 2004. 'dire' et finalit en bedja: un cas de grammaticalisation.
Journal of African Languages and Linguistics 25: 133-153.
Vanhove, Martine and M.-T. Hamid Ahmed. 2004. Contrastive negation in Beja: the
auxiliary verb rib. Afrika und bersee 85: 81-98.
Voigt, Rainer M. 1984. Zu den stimmlosen Vokalen des Borana. Africana
Marburgensia 17(2): 59-69.
Voigt, Rainer M. 1985. Die beiden Prfixkonjugationen des Ostkuschitischen. Au
68: 87-104.
Voigt, Rainer. 1996 Zur Gliederung des Kuschitischen: Die Prfix konjugationen. In
Cushitic and Omotic Languages: Proceedings of the Third International
Symposium Berlin, March 17-19, 1994, pp.101-131. Cologne: Rdiger Kppe.
Voigt, Rainer. 2001. Bibliographie zur thiosemitischen und kuschitischen
Sprachwissenschaft. IV. 1999 mit Nachtrgen fr 1997 und 1998 Aethiopica 4:
182-190
Wedekind, Klaus. 1980. Sidamo, Gedeo (Derasa), Burji: Phonological Differences
and Likenesses. Journal of Ethiopian Studies 14: 131-176.
Wedekind, Klaus. 1990. Generating Narratives: interrelations of knowledge, text
variants, and Cushitic focus strategies. Trends in linguistics Studies and
monographs;52. Berlin: Mouton de Gruyter.
Wedekind, Charlotte, Klaus Wedekind and Abuzaynab Musa. 2003. Beja Pedagogical
Grammar. Ms Asmara, Aswan.
Winter, Christopher. 1979. Language Shift among the Aasx, a Hunter-Gatherer Tribe
in Tanzania. Sprache und Geschichte in Afrika 1: 175-204.
Wondwosen Tesfaye Abire. 2006. Aspects of Diraytata Morphology and Syntax: A
Lexical-Functional Grammar Approach. Ph.D. thesis Norwegian Institute of
Science and Technology, Trondheim.
Yri, Kjell Magne. 1990. A contribution to the Understanding of Metathesis and
Assimilation in Sidamo Afo (Sidamo). MA Oslo.
Zaborski, A. 1975. The Verb in Cushitic. Krakau: Nakadem Uniwersytetu
Jagiellonskiego.
Zaborski, Adrzej. 1986. The morphology of nominal plural in the Cushitic languages.
(Verffentlichungen der Institute fr Arikanistik und gyptologie det Universitt
Wien, 39: Beitrge zur Afrikanistik, 28) Vienna: Afro-Pub.
Zaborski, Andrzej. 1987. Insights into Proto-Cushitic Morphology. In Proceedings of
the fifth International Hamito-Semitic Congress, ed. Hans Mukarovsky, vol. 2, pp.
75-82. (Beitrge zur Afrikanistik, 57). Vienna: Afro-Pub.
87
Zaborski, Andrzej. 1989. Cushitic Independent Pronouns. In Proceedings of the
Eighth International Conference on Ethiopian Studies (vol 1.), ed. Taddese
Beyene, pp. 649-672. Addis Abeba: Institute of Ethiopian Studies.
Zaborski, Andrzej. 1997. Problems of the Beja present seven years later. Lingua
Posnaniensis 39: 145-153.
Zaborski, Andrzej. 2005. Negative Conjugations in Cushitic. In Memoriae Igor M.
Diakonoff; Babel unde Bibel 2: Annual of Ancient Near Eastern, Old Testament,
and Semitic Studies, ed. L.Kogan, N. Koslova, S. Loesov, and S. Tishchensko, pp.
687-698. (Orientalia et Classica, Papers of the Institute of Oriental and Classical
Studies 8). Winona Lake, Indiana: Eisenbrauns
Zaslavsky, Claudia. 1973. Africa counts. New York: Lawrence Hill Books.
Zelealem Leyew. 2003. The Kemantey Language: A sociolinguistic and grammatical
Study of Language Replacement. Cologne: Rdiger Kppe.
Zyhlarz, E. 1956. Die Fiktion der kuschitischen Vlker. Kush (Khartoum) 4: 19-33.