i
The Origin of Plural Gender in East-Cushitic Languages
Jeffrey Nijboer
MA thesis Linguistics, Leiden University, December 2019 s1166182
ii
Abstract
There are a number of Cushitic languages that have a gender system that includes a gender known as „plural gender‟. The existence of „plural‟ gender is a peculiar feature in these languages and for that reason this thesis aims to find out how this gender developed. This thesis analyzes three languages in each of two branches of East-Cushitic languages. For each of the two branches this thesis examines at least one language with plural gender and one without.
iii
Table of Contents
1. Introduction
12. Introduction to the Languages
52.1 Introduction to the Oromoid Languages 5 2.1.1 Konso 5
2.1.2 Oromo 6
2.1.3 Diraytata 8
2.2 Introduction to the Western Omo-Tana Languages 9
2.2.1 Arbore 9
2.2.2 Dhaasanac 10 2.2.3 Elmolo 10
3. The Oromoid Languages
113.1 Initial Proposal for Cognates 11
3.2 Proto-Oromoid Reconstruction 15
3.3 Proto-Oromoid Plural Suffix Reconstruction 17
4. The Western Omo-Tana Languages
184.1 Proposal for Cognates 18
4.2 Reconstruction of Plural Suffixes 20
5. Comparison and Discussion
215.1 Comparing Oromoid and Western Omo-Tana 21 5.2 Potential Scenarios 23
6. Summary
25References
26Appendix A. Black’s (1974) Phonological Reconstruction
27Appendix B. List of Konso Plural Gender Nouns
29Appendix C. List of Arbore Plural Gender Nouns
341
1. Introduction
There is a complex relation between the features of gender and number in Cushitic languages (Corbett & Hayward 1987; Mous 2008; Tsegaye et al. 2013). Gender in a number of Cushitic languages has either two or three values, depending on the interpretation of linguistic data; the two undisputed values are masculine (m) and feminine (f). However, there may be a third gender: plural gender (p) (Corbett & Hayward 1987; Mous 2008; Tsegaye et al. 2013).
Regardless of which analysis turns out to be correct, it is clear that there is something that is at the very least somewhat atypical about the number and gender systems of these
languages. This thesis aims to find out how this came about through a comparative historical analysis of the subfamilies of some of the languages that have this system.
The rest of Section 1 will show the discussion over whether plural gender is a valid
category. Section 2 will introduce the languages and their subfamilies that this thesis analyzes, Section 3 will feature the analysis of the languages of the Oromoid family, Section 4 will do the same for the Western Omo-Tana family and Section 5 will compare and discuss the conclusions drawn in Sections 3 and 4 and propose a history of how plural gender arose as a feature.
In an attempt to avoid confusion, this thesis will abbreviate plural gender to (p) (masculine and feminine to (m) and (f)) and plural number to pl (singular to (sg)). Furthermore, these plurals will always be referred to as either plural number (sometimes also plurative) or plural (as a) gender and not just „plural‟ when an abbreviation is not used. This is necessary, because although plural gender can be connected to plurality it ís a different category from plural as a number and referring to both as just plural will get confusing.
One interpretation of this system, followed by Corbett and Hayward (1987) and Corbett (2000), holds that Cushitic languages have two gender values and that words requiring (p)-agreement are marked with plural number, because of a general typological principle that holds that the values of two independent features cannot mix (Mous 2008; Tsegaye et al. 2013). This position will henceforth be referred to as Corbett‟s position.
Another position, held by most descriptive linguists working on Cushitic languages (Tsegaye et al. 2013), argues that the number system is separate from the gender system, as evidenced by the existence of separate agreement markers for the two on adjectives (Mous 2008; Tsegaye et al. 2013). This position will from now on be referred to as the traditional position (because it existed before Corbett‟s position did).
To better illustrate Corbett‟s position, Corbett and Hayward‟s (1987) argumentation against plural as a gender on the basis of Bayso will now be summarised. Bayso is spoken around Lake Abaya in Ethiopia and is closely related to Somali (Corbett & Hayward 1987). In Bayso, three different agreement markers are distinguished by agreeing elements, illustrated in the following sentences by markers on verbal predicates (Corbett & Hayward 1987):
(1) lúban gira lion is „There is a lion.‟ (2) kimbír gitta bird is „There is a bird.‟ (3) ilkoo giran tooth/teeth are
2
Example (1) shows a masculine agreement form, (2) shows a feminine agreement form and (3) shows a plural gender agreement form (Corbett & Hayward 1987). Other agreeing elements are associative particles and demonstratives, each with three distinct forms for (m), (f) and (p) (Corbett & Hayward 1987).
Four number values can be indicated on the noun: general number (lúban „(a) lion‟), plural number (lubanjool „lions‟), paucal number (lubanjaa „a few lions/some lions‟) and singular number (lubantiti „a/the particular lion‟) (Corbett & Hayward 1987; Corbett 2000: 10-11, 181). General number is used to refer to a noun when number is not considered relevant (Corbett 2000: 9-10). Like plural number, paucal number refers to more than one referent; however the referents of paucal number are fewer than those of plural number (Corbett & Hayward 1987; Corbett 2000: 22-23).
In many cases the change of number on Bayso nouns will also cause a change in
agreement. For example, all nouns will get (p) agreement when they have paucal number and many nouns take (m) agreement when they have plural number (Corbett & Hayward 1987). Nouns take the same agreement with singular number as they do with general number, which is either (m) or (f); nouns that take (p) agreement general number do not appear in singular number form (Corbett & Hayward 1987). Based on these changes there are eight different types of noun (Corbett & Hayward 1987); these are called agreement classes (Corbett & Hayward 1987; Corbett 1991: 147-150; Corbett 2013). Since all nouns take (p) agreement with paucal number and agreement is the same with general number and singular number, these different agreement classes are based on the difference between general and singular on the one hand and plural number on the other (Corbett & Hayward 1987); the agreements are as follows (Corbett & Hayward 1987):
(m) with (gn/sg) stays (m) with (pl) (m) with (gn/sg) becomes (p) with (pl) (m) with (gn/sg) becomes (f) with (pl) (f) with (gn/sg) becomes (m) with (pl) (f) with (gn/sg) becomes (p) with (pl) (f) with (gn/sg) stays (f) with (pl) (p) with (gn) becomes (m) with (pl) (p) with (gn) stays (p) with (pl)
These classes do not have equal size (Corbett & Hayward 1987). There are only eleven nouns in the last two groups combined (nouns that take (p)-agreement with general number); Corbett and Hayward (1987) claim that these nouns are lexically marked for (p)-agreement, making them exceptions similar to pluralia tantum nouns (nouns that appear only in plural form) in other languages. Because these nouns are exceptions, Corbett and Hayward (1987) dismiss the notion of plural as a gender.
The classes which do not take (m)-agreement with plural number are also very small (incidentally also eleven in total: seven take (p) agreement and four (f) agreement) (Corbett & Hayward 1987). Therefore Corbett and Hayward (1987) label these nouns lexical exceptions too, leaving them with a regular gender system that distinguishes two genders ((m) and (f)) with singular and general, while losing that distinction in other numbers, because all regular nouns take (p)-agreement with paucal number and (m)-agreement with plural number (Corbett & Hayward 1987).
Mous (2008), who adheres to the traditional position, does not agree with this analysis. He argues that there is a separate plural gender and that the number system is independent of the gender system, based mostly on one argument, illustrated by the following examples from Iraqw (Mous 2008):
3 (4) faʕa ka ħeer
food.F is.F insufficient:F(:S.R.)
„The food is insufficient.‟ (5) tluway ku ħéer
rain.M is.M insufficient:M(:S.R.) „The rain is insufficient.‟
(6) ħayso ki ququmaar tail.P is.P short:P(:S.R.) „The tail is short.‟
(7) ħaysee ka ququmat tails.F is.F short:F:M.R. „The tails are short.‟ (8) daaqay ku ququmát
boys.M is.M short:M:M.R.
„The boys are short.‟ (9) na‟ii ki ququmat
children.P is.P short:P:M.R.
„The children are short.‟
These examples show that the agreements with gender and number are separate on adjectives, which in Corbett‟s analysis would lead to double marking of number whenever plural gender marking is involved (Mous 2008). The reason Corbett and Hayward missed this is because the two Cushitic languages they analysed (Bayso and „Afar) do not have such adjectival
agreement systems (Mous 2008).
Another argument by Mous (2008) is that when plural gender is taken to be a value of number a problem arises; there is now a divide between two types of nominal lexemes: one type specifies gender with singular but not with plural number, the other type specifies gender separately in singular and plural number, because gender with plural number does not have to be the same as the gender in the singular (Mous 2008). This is not a problem for Mous‟s interpretation, because the traditional position considers number to be a derivational feature in Cushitic languages, which allows for number to be able to change the gender of a noun (Mous 2008).
Mous‟s (2008) arguments are not necessarily the only arguments in favor of the traditional position however. Possible arguments in favor are also presented by Tsegaye et al. (2013), who did a psycholinguistic study on Konso, a Lowland East Cushitic language spoken in the south-west of Ethiopia. In this study, native speakers of Konso were asked to participate in one of two different experiments, where they were asked to name pictures on a laptop screen, while their reaction times were measured (Tsegaye et al. 2013). Before the start of the actual experiment, the participants were familiarised with the pictures that were part of the
experiment, after which they practiced naming the pictures (Tsegaye et al. 2013). During the experiment the participants had to name the pictures as quickly as possible, while ignoring auditorily presented distractor words or pink noise (Tsegaye et al. 2013).
In the first experiment the participants were asked to produce only the bare noun while naming the pictures (Tsegaye et al. 2013). This was done to find out whether or not the
4
participants would react faster when naming a word with plural gender if they heard a distractor word that is also a word with plural gender, than if they heard words with either masculine or feminine gender (Tsegaye et al. 2013). If so, that would be strong evidence in favor of plural as a gender, because it is expected that people will react faster while naming pictures if they hear a noun that has the same gender as the intended noun for that picture (this is called the gender congruency effect) (Tsegaye et al. 2013). If not, that could interpreted as evidence that the value (p) does not belong to the feature gender (Tsegaye et al. 2013).
The results from this experiment were that the average reaction times of the participants were 19 milliseconds slower in the gender incongruent condition (i.e. when the participants heard a word with a different gender than the word they had to name) than in in the gender congruent condition; however this result was not significant except in the subject analysis of the factor Target Gender (masculine, feminine and plural) (Tsegaye et al. 2013). Tsegaye et al. also found a gender congruency effect of 35 milliseconds in the production of plural gender nouns and 20 milliseconds in the production of masculine nouns, but they found no effect in the production of feminine nouns (only one millisecond) (Tsegaye et al. 2013).
In the second experiment the participants had to answer with nouns that were suffixed with definite markers to name the pictures they were presented with (Tsegaye et al. 2013). In Konso, plural gender words are marked with a definite marker -siniʔ and both masculine and feminine nouns are marked with the definite marker -siʔ (Tsegaye et al. 2013). As with the first experiment, a gender congruency effect is expected when a plural gender noun picture is combined with an auditorily presented plural gender noun, if (p) is indeed a value of gender (Tsegaye et al. 2013).
The results of the second experiment show an average reaction time for the gender incongruent condition that was 13 milliseconds slower than the reaction time for the gender congruent condition, but this result was insignificant in all conditions (Tsegaye et al. 2013). The gender congruency effect with the production nouns with plural gender was 32
milliseconds; however, for the production of nouns with another gender a non-significant -6 effect was found, i.e. reaction times were slower in the gender congruent condition than in the gender incongruent condition (Tsegaye et al. 2013).
The congruency effects for plural gender nouns of 35 milliseconds in the first experiment and 32 seconds in the second experiment suggest that (p) may be a value of gender (Tsegaye et al. 2013). However, the results are not conclusive, because they were insignificant in many cases and no congruency effect was found for feminine nouns in the first experiment and nouns without plural gender in the second experiment (Tsegaye et al. 2013). In fact, the results of these experiments, coupled with the results of experiments on other languages may indicate that a gender congruency effect does not exist (Tsegaye et al. 2013). Other possible explanations for the inconclusiveness of these experiments are reaction times that were slow overall, the fact that gender is not marked on the noun or that there were flaws in the set up; more specifically, some of the target words might not have been good candidates for being target words, since the participants reacted to some words very slowly, the potential reasons varying from unclear pictures to not being very common words to interference from other words with similar semantics (Tsegaye et al. 2013). Because of these issues the results of these experiments should be viewed with caution.
5
2. Introduction to the Languages
To answer the question of the origin of plural gender, this thesis will look at languages
belonging to the Lowland East Cushitic language family. The Lowland East Cushitic family is defined by Paul Black (1974: 6) as the group of languages descended from the latest common ancestor of Saho-Afar, Somali and Oromo (Proto-Lowland-East-Cushitic or PLEC).
Specifically, this thesis will analyze two subfamilies of this group: the Oromoid languages, represented in this thesis by (Borana) Oromo (referred to by Black (1974) as Galla), Konso and, to a lesser extent, Diraytata (also known as Gidole), and the Western Omo-Tana languages, consisting of Arbore, Dhaasanac and Elmolo. These groups were chosen for this analysis, because both groups contain at least one language that has (p) and at least one that does not.
2.1 Introduction to the Oromoid Languages
The Oromoid languages are subdivided into a branch consisting of the various dialects of Oromo and a branch consisting of the Konsoid languages, which forms a dialect continuum containing the various dialects of Konso, Dirayta, and Bussa, which are considered mutually unintelligible with the most common dialect of Konso (Black 1974: 12).
2.1.1 Konso
Konso is spoken by about 250,000 people in southern Ethiopia, specifically in the Segen Area Peoples‟ Zone in the state of Southern Nations, Nationalities and Peoples‟ Region (SNNPR). The Konso are farmers known for their terracing system, growing maize, wheat, barley, various types of beans and sorghum, cotton, and more. The Konso live in densely populated villages established on hilltops (Ongaye Oda 2013: 1-2).The Konso people intermarry with the Borana (Stroomer 1995: 195-196). The language has four dialects: Faashe, Karatte, Tuuro and Xolme (Ongaye Oda 2013: 2).
Konso is the central language in the analysis of plural gender in the Oromoid languages, because of the relevant languages it is the only one that has true plural gender and also has enough information available on it. Interestingly, Konso nouns with (p) always end in -aa.
Gender in Konso is marked on verbs and adjectives and is used to determine which
allomorph of definite and demonstrative markers is used. Verbs are marked with the suffix -ay with masculine third person subjects, with the suffix -t with feminine third person subjects and with -n with plural gender third person subjects. On attributive adjectives masculine and feminine are marked with -a, while plural is marked with -aaʔ. The definite suffix for masculine and feminine is -siʔ and for plural gender it is -siniʔ. Finally, masculine and
feminine nouns get -oosiʔ, -asiʔ or -siʔ as demonstrative suffixes, while (p) nouns get -oosiniʔ (Ongaye Oda 2013: 73-79, 97). See Table 1 for a summary of Konso Gender Agreement
KONSO GENDER AGREEMENT Masculine Feminine Plural Gender
Verbs -ay -t -n
Adjectives -a -aaʔ
Definite Suffix -siʔ -siniʔ
Demonstrative Suffix -oosiʔ/asiʔ/siʔ -oosiniʔ
Table 1. Konso gender agreement .
Number is marked on nouns and adjectives in Konso. Plural number on adjectives is indicated by reduplicating the root. On nouns, plural number is usually marked with a suffix.
6
All nouns with plurative marking have plural gender and all nouns with singulative marking have Konso has the following plurative suffixes (Ongaye Oda 2013: 80):
-ɗɗaa (27%)1 -wwaa (22%) -ɗaa (16%) -ayaa (7.5%) -iyyaa (5.5%)
In addition to forming them with these suffixes, pluratives can also be formed by reduplicating the base final consonant followed by -aa (e.g. hawla „tomb‟ > hawlallaa „tombs‟), by geminating the final consonant and lengthening the final -a (e.g. tika „house‟ >
tikkaa „houses‟). Some nouns have irregular or suppletive pluratives. Finally, it is possible to
further derive the reduplicated and geminated pluratives into double pluratives with either the -ɗɗaa suffix (with reduplicated pluratives) or the -ɗaa suffix (with geminated pluratives) (Ongaye Oda 2013: 83-86).
Singulatives are marked with the suffixes -ayta (m), -ta (m/f), -itta (m) or -teeta (f). Singulatives are always derived from nouns that are base pluratives (Ongaye Oda 2013: 87-88).
2.1.2 Oromo
Oromo is spoken by more than 25,500,000 people, mostly in western, central and southern Ethiopia, but also in northern Kenya and in Somalia (Ethnologue/Lewis et al. 2016). It is made up of several dialects, which are relatively homogeneous (Black 1974: 11-12). Oromo used to be a lingua franca and because of its many speakers is still an important language in the region; this also means that there are multiple dictionaries available. Especially important for the purposes of this thesis is the Borana dialect, spoken in southern Ethiopia, because both Konso and Diraytata are spoken in the vicinity of this dialect.
(Borana) Oromo does not have plural gender; it only has masculine and feminine gender. Generally, feminine nouns end in long vowels and masculine nouns in short vowels.
Masculine nouns that do not end in short vowels however, usually end in -aa. Nominal gender is marked on verbs, adjectives, demonstratives, possessives, singulative suffixes and subject suffixes on nouns (Stroomer 1995). Third person singular verbs with a masculine subject take the suffix -a, but if the subject is feminine they take the suffix -ti2. On adjectives, the form taken by gender marking is lexical; there are five different sets of adjectives, one of which does not mark gender and the other four each have their own set of masculine and feminine endings: -aa (m) vs -oo (f), -aa (m) vs -tuu (f), -ca (m) vs -ttii (f) and -sa (f) vs -ttii (f)
(Stroomer 1995: 53-55). Proximal demonstratives have different base forms depending on the gender of the head noun: kana for masculine nouns and tana for feminine nouns (Stroomer 1995: 62). Head noun gender on possessives is only indicated on first and second person possessives, with the prefix k- for possessives agreeing with masculine nouns and t- for feminine nouns (Stroomer 1995: 68). Singulative suffixes for masculine nouns are -ca or -sa, and for feminine nouns -ttii (Stroomer 1995: 45-46). Finally, Borana has three subject
suffixes: -ni, for nouns or adjectives ending in a long vowel, -ii, for adjectives or masculine
1
The percentages indicate how often these suffixes appeared in a sample of about 470 nouns (Ongaye Oda 2013: 80)
2 These are the forms for present tense main clause affirmative verbs; on other inflections the vowel will change.
7
nouns ending in a short vowel, and -tii, for feminine nouns ending in a short vowel (Stroomer 1995: 91). For a summary of gender agreement in Borana Oromo, see Table 2.
BORANA GENDER AGREEMENT Masculine Feminine
Verbs -a -ti Adjectives -aa -oo -tuu -ca -ttii -sa (Proximal) Demonstratives k- t- Possessives k- t-
Singulative Suffixes -ca/-sa -ttii
Subject Suffixes -ii, -ni3 -ni, -ttii4
Table 2. Borana Gender Agreement.
Borana nouns are usually unspecified for number and can be specified with either plurative or singulative suffixes. Generally, nouns affixed with a plurative suffix are feminine, although natural gender causes some exceptions (Stroomer 1995: 39). The nouns on the following list reproduced from Stroomer (1995: 40) form another exception; they are of masculine gender, always have plural number concord on adjectives and verbs and do not have equivalent forms that are non-plural in number:
aanani “milk”
afaani “mouth, language” bisaani “water”
buusani “Pleiades” c’id’aani “penis”
elellaani “cowrie shells” finc’aani “urine”
fooni “meat” funnaani “nose” halkani “night” ilkaani “tooth, teeth” imimaani “tears” hinjiraani “louse, lice” k’arrifaani “canine teeth” k’oraani “firewood” looni “cattle”
miciraani “stone(s) of a fruit” mid’aani “seeds, grains, food”
seep’ani “a leather straps on milking gourds” udaani “excrements”
Borana nouns can have their number pluralized by the following set of suffixes, which may not be exhaustive (Stroomer 1995: 41-44):
-oollee
3 If the final vowel is long (final vowels of masculine nouns are usually short) 4
8 -llee -oota -ootee -ii -yyii -yyee -(a)ani -(o)wa(a)ni -oo -eeni
The singulative suffixes -ca and -ttii are synchronically only productive with ethnonyms. These suffixes also appear on other nouns, but in those cases have lost their singulative meaning and are unspecified for number. The singulative suffixes -sa and -ttii are only found on a limited set of nouns that often have a root ending in -ee (Stroomer 1995: 45-48).
On adjectives, number can be expressed by reduplication of the first vowel and consonant. This is not obligatory and there is a semantic difference between reduplicated and
non-reduplicated adjectives. Specifically, non-reduplicated adjectives indicate that the described nouns are scattered. Additionally, adjectives ending in sa (m) and ttii (f) have plurals in yyii or
-yyee. Finally and interestingly, adjectives ending in -aa (m) and -tuu (f) have an ending -oo
that agrees with the previously mentioned set of masculine nouns without non-plural equivalents.
2.1.3 Diraytata
Diraytata is spoken by about in the Dirashe Special District in southern Ethiopia, slightly to the north of Konso (Wondwosen 2006: 2). According to Wondwosen (2006: 3) the speakers of Diraytata (the Dirasha) are bilingual in Oromo. The Dirasha are pastoralists and cultivators (Wondwosen 2006: 3). Diraytata is not very well described and will therefore only be used in a supporting role in this thesis.
There are two different analyses concerning plural gender in Diraytata: according to Hayward (1981) there is (p) in Diraytata. However, Wondwosen (2006) mentions only masculine and feminine. In addition, the set of nouns without forms for plural number
mentioned in Hayward (1981) does have forms for plural number according to Wondwosen‟s (2006) informants. This is also relevant because some of the forms in Hayward‟s (1981) list of nouns without pl forms are marked as (p) nouns in that list. Another feature analyzed differently by these two sources is the set of nouns which according to Hayward (1981) is a set of plural numbered nouns ending in -a which need to be derived with -itt to form
singulars, but Wondwosen (2006) treats those singulars as the base form that form plurals with a suffix -a.
Gender marking is not visible on nouns themselves, however, number marking is. Wondwosen (2006) gives the following ways to form a plural for nouns: -aɗa, -awwa, reduplication of the final consonant (e.g. lukkala „fowl‟ > lukkalla), -a, -(y)ya and -alla.
Agreement with gender and number exists for adjectives and verbs. On adjectives, feminine gender is marked by -at, while masculine gender is unmarked; plural number is marked by reduplication. On verbs gender is marked by a -t for feminine (singular), while masculine is unmarked. Plural number (third person) is marked with -an or -en, depending on the aspect of the verb (Wondwosen 2006: 96-97).
9
2.2 Introduction to the Western Omo-Tana Languages
The Western Omo-Tana language family consists of Arbore, Dhaasanac and Elmolo. There are grammars available for Arbore and Dhaasanac, but not for Elmolo, although it should be said that the grammar on Arbore is based on merely six weeks of fieldwork (Hayward 1984: VII). The relative lack of information on Elmolo means that, like Diraytata, Elmolo takes on a supportive role, rather than a central role in this analysis.
2.2.1 Arbore
According to the Ethnologue, the Arbore language had about 10,320 speakers in 2007, of which about 7,210 were native speakers (Ethnologue/Lewis et al. 2018).
Arbore has three genders, masculine, feminine and plural gender. In monosyllabic words, high tone indicates that the noun is masculine, while non-high tone indicates the noun is feminine (Hayward 1984: 131), seemingly implying that (p) nouns are never monosyllabic. There is gender agreement on verbs, adjectives and pronouns. When a noun is modified (e.g. by an adjective or determiner), gender is also marked on the nouns themselves. Table 3 presents the various forms of gender agreement in Arbore (Hayward 1984: 132-133, 184-185, 199-201, 227-229, 254)
ARBORE GENDER AGREEMENT Masculine Feminine Plural Gender
Verbs Prefixed y- t- y-
Suffixed Ø -t~Ø Ø
Nouns -h(a) -t(a(h)) -h(a)
Adjectives -á -á -o
Possessive Determiners -ás(s)ut -ás(s)et -ás(s)o
Interrogative Determiners búko bítoko toko
Possessive Pronouns h- t- to-h-
Deictic Pronouns Near há- tá- tó-
Remote h- t- to-h-
Table 3. Arbore gender agreement.
Hayward (1984: 159) describes three number categories for Arbore nouns: „singular/unit reference‟, „plural/multiple reference‟, and „singulative/singulative reference‟. The Arbore language has the following plural number suffixes (the gender they impose is given in brackets), some of which occur only on a few words (Hayward 1984: 166-178):
-mé (f) -aʔamé (f) -n (m) -má (p) -ó (p) - y (m) -mó (p) -omá (p) -mmé (f) -tonó (p) VVC-Cá (p) -té (f) -(C)é (p)
10
Singulative suffixes are as follows (Hayward 1984: 179-183) -č (m) -(i)t (m) -(i)n (m) -té (f) -n té (f) -an té (f) 2.2.2 Dhaasanac
The Dhaasanac language is spoken along the lower course of the Omo river and along the northeastern shore of Lake Turkana. Most Dhaasanac speakers live in Ethiopia, but their territory extends into Kenya (Tosco 2001: 1). Dhaasanac has about 60,730 speakers (Ethnologue/Lewis et al. 2018).
Dhaasanac only has masculine and feminine as values of gender and no plural gender. Gender agreement happens on verbs and optionally on adjectives. The gender of a noun can be determined from the form if there is a terminal vowel, formative or suffix. If the noun is in its basic form then gender cannot be determined from the form of the noun (Tosco 2001: 71).
Like in Arbore, three categories of number can be distinguished: basic, singulative and plural, corresponding with Hayward‟s (1984) unit reference, singulative reference and
multiple reference, respectively. The vast majority of plurals are of masculine gender. By far, most nouns have either only a basic form and a plural form or a singulative form and a plural form, but nouns that have a basic form and a singulative form are also fairly common. A few nouns have a form for all three categories. Finally, there is a group of nouns that only have basic forms, typically ethnonyms, names of Dhaasanac sections, clans and generation sets, names of social gender, or age groups or names of animals (Tosco 2001: 71, 73-74, 98-99).
Dhaasanac has the following plural suffixes (Tosco 2001: 83-92): -(C)u -u -a -i -(s)V(V)m (usually -am) -(C)anu -s -tu
Singulative suffixes are -(i)c/-(i)ɲ (rarely -ac/-aɲ), which are masculine and -ti, which is feminine (Tosco 2001: 75-78)
2.2.3 Elmolo
The Elmolo are fishermen living on the southeastern shore of Lake Turkana. In 1980, only eight speakers of the Elmolo language were reported, all over fifty years old, who used the language only as a secret code to exclude others and spoke Sampur otherwise (Heine 1980: 177). This means the language may well be extinct at this point.
The gender system of Elmolo has masculine and feminine nouns, but no plural as a gender. Gender is mostly unmarked the nouns themselves and is therefore determined by agreement
11
(Heine 1980: 181); the only exception to this is the suffix -te used to derive a singulative from an unmarked plural noun, as this suffix imposes feminine gender (Heine 1980: 183). Gender distinction is mostly irrelevant for plural nouns, as most plural nouns are masculine. Table 3 shows how agreement with gender is expressed (Heine 1980: 181):
ELMOLO GENDER AGREEMENT Masculine Feminine
Pronominal Possessives Ø t- Demonstratives Attributive Ø t- Predicative k- t- Verbal Subject Pronoun Prefix-Verbs y- t- Suffix-Verbs Ø -t
Table 4. Elmolo Gender Agreement.
Elmolo nouns can be divided into four sets, based on how they express number. One group of nouns suppletive plurals, i.e. they have completely different words for the singular and the plural version. Another set of nouns is number-indifferent or can only be identified as ether singular or plural by agreement. Number can also be expressed by suffixing the singulative -te to an unmarked plural. The last set of nouns forms plurals by suffixation, using the following set of plurals (Heine 1980: 182-183):
-e -o -ne -no -nu -a -ma
The last three of these plural suffixes occur only on a small set of nouns. The suffix -e occurs on monosyllabic nouns ending in a consonant. The suffix -o occurs on mostly feminine monosyllabic nouns ending in a consonant. -ne is used for that are mostly bisyllabic and mostly masculine. -no is used for bisyllabic and trisyllabic nouns ending in a vowel (Heine 1980: 182-183).
3. The Oromoid Languages
This section will analyze and compare the Oromoid languages. Section 3.1 will compare the three languages to each other, in order to find cognates for plural gender words. Section 3.2 will provide a reconstruction for these cognates. Section 3.3 will provide reconstructions of the plural number suffixes of the Oromoid languages.
3.1 Initial proposal for cognates
Starting with words from Konso that have (p) as their gender taken from Ongaye Oda (2013) and using Paul Black‟s (1974) reconstruction5
to predict what a cognate would be, I looked for cognates in dictionaries of two varieties of Oromo, one by Tilahun Gamta (1989) and one by Leus (1995). Table 6 shows the (possible) cognates found in this way.
5 Black’s (1974) table summarizing his phonological reconstruction is reproduced here in Appendix A in a slightly adapted form.
12
Konso translation Oromo Oromo translation Borana Borana
translation
(a)annaa milk aannan milk anani (n) milk
afaa mouth/language afa(a)n 1. mouth/beak 2.
language 3. talk afani
(n) mouth, tongue, language arasaa local beer made for
sale arafaa
foam, froth, lather, or
spume -
ɗarɗaa lie/untruth - dhara-ti lie/falseness
elalaa cowrie elella cowry elelani (n) cowry shells erkanaa message ergaa errand erga-ni (f) message
filaa comb filaa comb -
furaa padlock/key furu 1 ease, relieve 2 loosen
3 change, transfer fura-ta
to free, redeem, open, release, loosen. Improve hanʄufaa saliva (h)anc‟ufa 1 saliva/spittle 2 dung,
filth hancufa-ti
(human) saliva, spittle hiippaa riddle hibbo a game of riddle -
ipsaa light ibsa(/ibsaa) light ibse-ni/ibsa-ni (m) light
ʄorroʛaa eye discharge c‟orrok‟a
1. puddle (i.e. a small pool of dirty, mud-colored, stagnant water
2. mud-colored dirty water
-
leʛaa loan - lakawa-ni |
lakomsa-i
(f) counting, amount | counting, amount, number makkaa sickness maggaNNaa disease/illness
marʄaa hip mart‟ifaccu | mart‟o
to wear cloth around the waist/loin | cloth worn around the
waist/loin
mart'o-ni
cloth worn around the waist sewn together in
one piece miɗaa cabbage leaves mi(n)ɗan 1 crop 2 food midhani (n) grain, seeds,
fruit
ñaaññaa tomato - nyanya-ni (f) tomato
paankaa machete - banga-ni machete
pi aa water bisan/bi an water bisani (n) water pohaa contribution/tribune bu-a profit
aa aa stretcher - sabo-ni headrest, pillow,
any support
sinɗaa urine finc‟an urine fincani urine
sitaa tail of animal - fita-tta
to finish, end (noun:
fitisa-i/fita) teepaa rope teba/tep‟a
leather strap (for fastening things
together)
tepa-i small thin leather cord
utaa faeces udan faeces udani faeces
uwwaa dress uwwisa clothes/cash uwa name of the gorfo for women
xallaa kidney kale(e) kidney kale-ni kidney
xoffaa groin kofa thigh kofa-i
private parts of humans and
animals xoraa fine/punishment - qora-ni/qorsuma-ni judgement
13
Not all of these words are inherited from the proto-language however; words like
ñaaññaa/nyanya-ni „tomato‟ and paankaa/banga-ni „machete‟ are borrowed from Swahili,
other words could be borrowed by Konso from Oromo, which seems to have happened with
ipsaa/ibsa/ibsa-ni „light‟, which should have had /f/ instead of /s/ in the Oromo varieties or / /
instead of /s/ in Konso if both languages inherited it from the proto-language. It is theoretically also possible that Konso words have been borrowed by Oromo, but this is unlikely for social reasons and if it happened the borrowed word is only be present in Borana Oromo, which is the variety of Oromo that is in contact with Konso. In other cases the possible correspondences are semantically only loosely related; in those cases the words are either derived in one of the languages or the similarities are coincidental, for example the Konso word marjaa „hip‟ seems to have some relation to the Borana word mart’o „cloth worn around the waist, but at some point a derivation was involved; although these cases are
interesting, they do not offer any insight on the origin of plural gender.
However, there are seven words shown in Table 6 that are of particular interest. Those words are repeated in Table 7 below. They are interesting because they are likely direct cognates inherited from the proto-language in both languages (with the addition of one
morpheme in Oromo6: -n(i)) and because they also happen to appear on Stroomer‟s (1995) list of masculine nouns that always have plural number agreement (these words are considered neuter gender by Leus (1995)). This, combined with the fact that other (pl) nouns in Borana are usually feminine (Stroomer 1995), seems to indicate that this group of nouns is a
fossilized remnant of a formerly full-fledged class of nouns with plural gender.
Konso translation Oromo Oromo translation Borana Borana
translation
(a)annaa milk aannan milk anani (n) milk
afaa mouth/language afa(a)n 1. mouth/beak 2.
language 3. talk afani
(n) mouth, tongue, language elalaa cowrie elella cowry elelani (n) cowry shells miɗaa cabbage leaves mi(n)ɗan 1 crop 2 food midhani (n) grain, seeds,
fruit pi aa water bisan/bi an water bisani (n) water
sinɗaa urine finc‟an urine fincani urine
utaa faeces udan faeces udani faeces
Table 7. Konso/Oromo cognates of particular interest.
Searching for cognates of the remaining words on Stroomer‟s (1995) list in Black and Shako Otto‟s (1973) dictionary, which is more extensive than the dictionary in Ongaye Oda‟s (2013) grammar, the following additional correspondences were found:
Ko. ilkaa7 „teeth‟ and Or. ilkaani „tooth, teeth‟
Ko. lowaa „cattle‟ and Or. looni „cattle
Ko. qoraa „wood‟ and Or. k’oraani „firewood‟ Ko. soonaa/siinaa „nose‟ and Or. funnaani „nose‟ Ko. sowaa8 „meat‟ and Or. fooni „meat‟.
6
Or the deletion of this morpheme in Konso. 7
ilkaa does have a non-p singular form: ilkitta, which is masculine. ilkaa was given as an alternative plural number form by Ongaye Oda (2013).
8
14 Finally, there is also the following correspondence:
Ko. ukukkaa „egg‟ (hukukkaa in Black and Shako‟s (1973) dictionary) and Or. okokkaa (Tilahun Gamta 1989), okhokani (Leus 1995), okokaani (Stroomer 1995) „egg(s)‟.
This correspondence is not included in the previous lists, because these do not correspond according to Black‟s (1974) phonological reconstruction. Konso /u/ is expected to correspond to Oromo /u/ in most environments, including this one. However, these cognates are also of interest, because it is a singular number plural gender word in Konso and has the same ending in Borana as the words on Stroomer‟s (1995) list of words that always have plural number concord, while also being translated as both plural and singular in number and being assigned neuter gender by Leus (1995).
Now that the comparison between Konso and Oromo is made, it is time to also look at Diraytata. For Diraytata the following cognates of interest were found:
ra a „timber‟ (Hayward 1981)
si(i)tta „tail‟ as sitta (Hayward 1981) or siitta (Wondwosen 2006) sóha „meat‟ (Hayward 1981) (Wondwosen 2006)
ʔâanna „milk‟ (Hayward 1981) (Wondwosen 2006) ʔíilla „teeth‟ (Hayward 1981)
(ʔ)úku(u)kka „egg(s)‟ as úkuukka (Hayward 1981) or ʔukukka (Wondwosen 2006)
Unfortunately, the data Diraytata provides is not enough to either prove or disprove the notion that the plural gender feature was inherited from Proto-Oromoid.
Konso Konso Meaning Borana Oromo Borana Meaning Diraytata Diraytata Meaning
(a)annaa milk aanani milk ʔâanna milk
afaa mouth/language afaani mouth/language (kap) mouth/language
elalaa cowrie elellaani cowrie shells - -
ilkaa teeth ilkaani tooth/teeth ʔíilla teeth
lowaa cattle looni cattle - -
miɗaa cabbage leaves mid‟aani seeds, grains,
food - -
pi aa water bisaani water (hâk‟) water
puusaa Pleiades buusani Pleiades - -
ʛoraa wood k‟oraani firewood rayya timber
sinɗaa urine finc‟aani urine - -
sitaa tail (of an
animal) fita finish/end s(i)itta tail
soonaa/siinaa nose funnaani nose - -
sowaa/swaa meat fooni meat sóha meat
(h)ukukkaa egg okokaani egg(s) (ʔ)úku(u)kka egg(s)
utaa faeces udaani faeces - -
Table 8. Summary of cognates of interest. The Diraytata words kap and hâk’ are not cognates, but translations and are therefore given in brackets. The Oromo word fita might not be a cognate and only coincidentally look like sitaa/sitta, which is why it is in italics.
15
3.2 Proto-Oromoid Reconstruction
Now it is time to attempt a reconstruction of the Proto-Oromoid forms of the words in Table 8. Due to the fact that none of the Konso and Diraytata words have anything that points to a cognate of the -ni suffix in Borana, it is assumed that this suffix was a later addition. The following list shows the forms that appear to be clear cognates without any complications in the reconstruction:
*aana(a) „milk‟
*afaa „mouth/language‟ *miɗaa „(plant-based) food‟9 * i aa „water‟
*ʛoraa „wood‟ *sinɗ1aa „urine‟
*udaa „faeces‟
The rest of the reconstructions require some additional explanation. First to be discussed are the lowaa/looni and sowaa/fooni/sóha pairs. The following are the reconstructions of these pairs:
*loʔaa „cattle‟ *soʔaa „meat‟
The /w/ and /h/ reflexes in the Konso and Diraytata words for „meat‟ indicate a glottal stop according to Black‟s (1974) phonological reconstruction. However, Black (1974) does not give any reflexes of the glottal stop in Oromo. It appears that there are two possibilities for what happened in Oromo. One possibility is that the form was that in the reconstructions given above the glottal stop was dropped, leading to an intermediary form *foaa which assimilated into *foo. The other possibility is that these two words did not actually have the
-aa suffix in Proto-Oromoid or that that is not the form that Oromo inherited. In this case the
reconstructions for Oromo are *loʔ-ni and *foʔ-ni. The glottal stop would then have
disappeared with compensatory lengthening of /o/, resulting in the forms looni and fooni we see today. In this second case, the forms *loʔaa and *soʔaa are only valid for Proto-Konsoid and the reconstructions for Proto-Oromoid would be the following:
*loʔ- „cattle‟ *soʔ- „meat‟
Since no cognate for lowaa/looni was found in Diraytata, the above reconstruction with a glottal stop is not completely certain. However, since these forms are so similar to forms for „meat‟, and both Oromo forms deviate from the -(a)ani suffix that is the norm for the other masculine plurals in Oromo, it does not seem unreasonable to assume they are the same in this regard.
The next reconstruction might only be valid for Proto-Konsoid, since the closest form found in Oromo, fita „finish/end‟, may only be coincidentally similar, and does not appear to be a plural form in Oromo. However, since it is a plural gender form in Konso and also in Diraytata according to Hayward (1981), it is included in the discussion:
9
16 *sitaa „tail‟
This form could also have been *siitaa, *sittaa or *siittaa, because it is impossible to decide which form is more like the proto-language with only two forms that are definitely
descendants of this reconstruction. However, the form *sitaa was chosen based on Oromo fita in case it does indeed prove to be cognate to sitaa and s(i)itta.
The reconstruction of Ko. soonaa/siinaa and Or. funnaani „nose‟ also requires further explanation. Several things are potentially problematic here: the alteration between /oo/ and /ii/ within Konso, a long vowel in Konso but a short vowel in Oromo, and that there is a short /n/ in Konso but a long /n/ in Oromo. Here, the following reconstruction is proposed:
*suʕnaa „nose‟
The *ʕ in this reconstruction is the *ʕ1/2 in Appendix A, because *ʕ was already gone in Proto-Oromoid. *ʕ in this reconstruction forms a possible explanation for the issues that involve length: in Konso the disappearance of *ʕ might have caused compensatory lengthening of the vowel, while in Oromo it might have assimilated into /n/. This
reconstruction does not explain the two different forms in Konso, however. In theory, an /i/ in Konso can correspond to an /u/ in Oromo if it was preceded by *ʕ, which would lead to **sʕunaa as an alternative reconstruction. The form **sʕunaa is problematic, however, because a syllable beginning with two consonants is not allowed in these languages. Another possibility might be *suʕunaa. In this case the Konso development would have been
*suʕunaa > *soʕinaa > *soinaa > soonaa/siinaa. However, for this to work, the second *u must have disappeared in Oromo before *ʕ did in order to explain the short vowel and geminated consonant.
Another reconstruction presenting difficulties is the reconstruction of
(h)ukukkaa/okokaani/(ʔ)úku(u)kka „egg(s)‟. Here, it is an issue that /o/ is found in Oromo ,
where /u/ is found in Konso and Diraytata. It is therefore a possibility that this word was borrowed in either Konsoid, Oromo or both. However, assuming that it is not borrowed, the following reconstruction might be the answer:
*ħukħukaa „egg(s)‟
The presence of *ħ in this reconstruction is based on the /h/ Black and Shako‟s (1973) and the spelling okhokani in Leus (1995). This reconstruction assumes that *u became /o/ in Oromo when it was preceded by *ħ. Although it could already be seen in the unreconstructed forms, this reconstruction shows even more clearly that this word is reduplicated.
Reconstructing the word for „teeth‟ from ilkaa/ilkaani/ʔíilla is also somewhat problematic. The best reconstruction that can be made of this is *ilkaa, but when one considers that this reconstruction should have resulted in Konso *ilxaa, it actually seems more likely that at least Konso ilkaa might be a borrowing from Oromo. The reconstruction *ilkaa might still be correct, however, if the Diraytata development went like this: Oromoid *ilkaa > Proto-Konsoid *i k aa > Pre-Diraytata *ʔiilha. The final step would then be that /h/ assimilates into /l/. This development, however is pure speculation.
The final two sets of words in Table 8 probably involve borrowing in at least one of the two languages. In the case of elalaa/elellaani „cowrie‟, the a/e difference in the second syllable could only be explained if the proto-form was **elaylaa, but this would raise the question of why *y is not present in the Konso word elalaa. It therefore seems more likely that this is a borrowing. Finally, in the case of puusaa/buusani „Pleiades‟, it is the presence of
17
/s/ in both languages that is problematic. An /s/ in Konso corresponds to /f/ in Oromo, while an /s/ in Oromo corresponds to Konso / /. /s/ can only appear in both languages at the beginning of a word before /a/ or /e/.
3.3 Proto-Oromoid Plural Suffix Reconstruction
A comparison should also be made of the nominal plural number suffixes of the Oromoid languages. Although they were given in chapter 2, the current forms will be listed here again:
Konso: -ayaa Diraytata: -a Borana: -(a)ani
-ɗaa -aɗa -eeni
-ɗɗaa -alla -ii
-iyyaa -awwa -llee
-wwaa -(y)yaa -oo
-oollee -oota -ootee -(o)wa(a)ni -yyee -yyii
Stroomer (1995) has already done some internal reconstruction, reducing the number of existing suffixes to the following elements:
-aa- -ee- -ii- -ll- -ni- -oo- -t-
The proposed Proto-Konsoid reconstructions are given in Table 9. These reconstructions are relatively straightforward, however there is a possibility that Konso -iyyaa is not fully cognate with Diraytata -(y)yaa and is instead by -ii (also found in Borana) and -aa or -(y)yaa. It should also be mentioned that the Konso -aa given in Table 9 is not a plural number suffix, but instead the ending found on virtually all Konso nouns that have plural gender.
Konso Diraytata Proto-Konsoid
-aa -a *-aa
-ɗaa/-ɗɗaa -aɗa *-ɗaa
-ayaa/-iyyaa -(y)yaa *-(y)yaa
-wwaa -awwa *-wwaa
Table 9. Proto-Konsoid plural suffix reconstructions.
Table 10 shows the proposed reconstructions for Proto-Oromoid. Despite the large number of plural suffixes or plural suffix components, Table 10 only contains three reconstructions, one of which is speculative. The reconstructed form *-aa is not actually new here, as it was already part of most of the reconstructed Proto-Oromoid words in section 3.2. This means that there is small possibility that Proto-Oromoid *-aa was not a true plural number suffix, but a
18
plural gender ending or suffix. However, the Oromo suffixes -aani and -owaani are used with other words than just the discussed set of words that are always plural number and *-aa is a part of all the Proto-Konsoid plural number suffixes. This suggests that it probably was a true plural number suffix. The speculative reconstruction is *-ii because it is not entirely clear that Konso -iyyaa is (partially) cognate with Oromo -ii. Finally, although the -ni plural number suffix element only occurs in Oromo, it should be considered that -n is recurring affix in the agreement of Konso plural gender. This means that there may have been a Proto-Oromoid suffix -n or -ni that indicated plural agreement which shifted to become a plural number suffix on the noun itself in Oromo. The opposite is also possible: The Proto-Oromoid suffix was a plural suffix on the noun, which spread beyond the noun in Konso to become an agreement suffix.
Konso/Diraytata/
Proto-Konsoid (Pre-)Oromo Proto-Oromoid
PK: *-aa -aa- *-aa
Dir.: -alla -ll- *-ll-
Kon.: -iyyaa -ii *-ii
Table 10. Proto-Oromoid plural suffix reconstructions.
4. The Western Omo-Tana Languages
As discussed in section 2.2, Arbore is the only language of the Western Omo-Tana languages that has plural as a gender. In order to find out whether Proto-Western-Omo-Tana also had plural gender, Arbore will first be compared to the other languages in this family in section 4.1. If PWOT did have plural gender, it can be expected that traces of it remain in the languages that lost it.
4.1 Proposal for Cognates
Table 11 shows all Arbore base plural gender nouns for which at least one cognate in one of the other two Western Omo-Tana languages could be found.
Arbore
Arb.
Meaning Dhaasanac Dha.
meaning Elmolo
Elm. Meaning
ʔeenú milk ʔeen-u milk (m) ‟énu milk (m)
ʔohó mouth,
language - - ‟ ho
mouth, language
ʔudú faeces ʔu faeces (m) útu faeces (m)
ʔummó children –
collectively ʔum-mu
children,
offspring (m) - -
bičé water bie water (m) píce water
bo water lily oɗ-ɗi sp. of water
plant (f) - -
haaló the Pleiades hála
name of a constellation
(Ursa Major?)(f)
- -
kalanó kidneys - - kâl kidney (m)
19
(f/m) mulkó iron anklets muulli tool, car,
bracelet mʊʊlle leg bracelet
ree á backbone - - rréka back
sin á urine sínna urine (m) sínna urine (m)
soonó nose sóon-o nose (f) sóóno nose
tummannó ribs, ribcage - - túmánte rib
Table 11. Arbore words of plural gender with cognates in Dhaasanac and/or Elmolo.
For most of the correspondences in Table 11, it is relatively straightforward to reconstruct the most likely forms of their Proto-Western-Omo-Tana ancestor, although there are some uncertainties regarding vowel and consonant length:
*ʔeenu „milk‟
*ʔoho „mouth, language‟ *ʔudu „faeces‟
*ʔummo „children (coll.)‟ *boɗ1ɗ1- „species of waterplant‟ *ha(a)l- „Pleiades (probably)‟ *kal „kidney‟
*kariso „molars‟ *mu(u)l- „(leg) bracelet‟ *re(e)ka „back(bone)‟ *sinɗ1a „urine‟ *soono „nose‟ *tum(m)an- „rib‟
However, the fact that it is not difficult to reconstruct the protoforms of these words does not mean that there is nothing more to be said about them. First of all, the reconstructions based on word correspondences without a Dhaasanac cognate cannot be automatically considered a Proto-Western-Omo-Tana reconstruction, since Arbore and Elmolo are more closely related to each other than to Dhaasanac (Black 1974, Tosco 2001, Tosco 2015). Another point of interest is *ʔeenu „milk‟, because a -nu plural number suffix exists in Elmolo and an -u plural number suffix exists in Dhaasanac10. This means it is possible (though not definitively proven) that *ʔeenu „milk‟ was a plural form in Proto-Western-Omo-Tana. However, two of the other reconstructions were almost certainly of plural number in the proto-language: *ʔummo „children (coll.)‟ and *kariso „molars‟, because they are still the plural number forms today (a suppletive form in the case of *ʔummo), although the Arbore form is considered the base form by Hayward (1984) with a singular formed through a singulative. Finally, although Black (1974) did not include it in his table, he does mention that /n/ (and also /y/) might be a reflex of *ɗ in Elmolo and Dhaasanac.
Two other words are more difficult to reconstruct. The first to be discussed here is the word for „faeces‟. Analyzing only the Western Omo-Tana forms, one could come to the conclusion that the reconstructed form should be **ʔu, and that Arbore and Elmolo added a suffix *-du at a later stage. However, no descendant of such a suffix has been found in any of the three languages and looking at the Oromoid cognates it is clear that at least /*d/ must have been a part of the word before the Proto-Western-Omo-Tana stage. This leads to the
conclusion that the correct reconstruction is most likely *ʔudu, and that Dhaasanac dropped the last syllable for some unknown reason.
The second reconstruction that requires more explanation is the reconstruction of the word for „water‟. The Arbore-Elmolo reconstruction is simple: * i e. However, when taking Dhaasanac into account something similar to what happened with the word for „faeces‟ appears to be going with „water‟ as well. The only difference is that here the final vowel *e
10Although this could also originally be an *-o since /*o/ and /*u/ merged into /*u/ word
20
has been preserved (or added again later) in Dhaasanac. The proposed reconstruction for Proto-Western-Omo-Tana is therefore also * i e
4.2 Reconstruction of Plural Suffixes
Since most of the current plural number suffixes appear in Table 12 below, the full lists from chapter 2 will not be repeated here, as was done for Oromoid in section 3.3. The only suffixes not present in Table 12 are Arbore -y and VVC-Cá, and Dhaasanac -s. It is important to remember that with the exception of -mé, -aʔamé, -n, -má and -ó, the Arbore plural number suffixes only appear on small sets of nouns. These reconstructions mostly speak for
themselves. There are two things, however, that might require some further explanation. The first is simply a reminder that */e/ and */o/ became /i/ and /u/ respectively at the end of a word in Dhaasanac.
The second is the reconstructed form *-m(a). It is not immediately clear whether the form should be m or -ma. At the Proto-Arbore-Elmolo stage it certainly appears to have been
*-ma. The question is whether Proto-Western-Omo-Tana already had *-ma and Dhaasanac
dropped *-a for some reason, or whether Western-Omo-Tana had *-m and
Proto-Arbore-Elmolo added the plural number suffix *-a to it. The second option seems most likely, because there appears to be no reason for *-a to be dropped in Dhaasanac and because,
looking at all the Arbore plural number suffixes, it would not be the only example in a Western Omo-Tana language of plural number suffixes being combined to form new plural number suffixes.
Arbore Dhaasanac Elmolo Proto-Western-Omo-Tana
-aʔamé -má -mé -mó -omá -mmé -(s)V(V)m -ma *-m -má -omá -Cá -a -a *-a -aʔamé -mé -mmé -té -(C)é -i -e *-e -ó -(C)u -o *-o
-tonó -tu - *-to
-n, -tonó -(C)anu -ne, -no, -nu *-n
-tonó -(C)anu -no *-no
21
5. Comparison and Discussion
This section will compare the Oromoid languages with the western Omo-Tana languages in section 5.1, in an attempt to find out whether the plural gender systems found within these two language families have a shared origin. Section 5.2 will discuss the various scenarios that are theoretically possible and will use the findings of this thesis to narrow down which
scenario is most likely.
5.1 Comparing Oromoid and Western Omo-Tana
In this section the Western Omo-Tana languages will be compared to the Oromoid languages. Mostly, Arbore will be compared with Konso. Consider Table 13:
Cognates that have plural gender in both languages
Konso Konso Meaning Arbore Arbore Meaning
aannaa milk ʔeenú milk
muklaa “pure” iron bracelets mulkó (set of) iron anklets
pi aa water bičé water
sinɗaa urine sin á urine
soonaa nose soonó nose
utaa faeces ʔudú faeces
Cognates that have plural gender in Konso, but not in Arbore
Konso Konso Meaning Arbore Arbore Meaning
ʛoraa wood or (m) wood
sowaa/swaa meat soʔ (m) meat
tiraa liver tirá (f) liver
xaxaa oath kakannó (f)11 oath
Semantically similar words that have plural gender in both languages
Konso Konso Meaning Arbore Arbore Meaning
afaa mouth, language ʔohó mouth, language
ɗilaa charcoal zanné charcoal
haarta artificial pond,
reservoir b n ḅ pond
hirrii aa eyelashes moy é eyebrow, eyelashes
ilmaamaa tears hi imá tears
ipsaa light hiiró sunlight/moonlight
kulaa testicles ʔedanó testicles
olsaa dream zan ó dream(s)
puusaa the Pleiades haaló the Pleiades
xarpaa door (woven from
wood) korbó
door made of interwoven lathes
Table 13. Comparison of Konso and Arbore.
The first point that should be remarked upon is that cognates that have plural gender in Arbore but not Konso are not included in Table 13. This is simply because no examples of this have been found so far. Another point that should be mentioned is that most of the Arbore words with plural gender are either mass or collective nouns (also see Appendix C). Konso on
11 Although Arbore kakannó has feminine gender, Hayward (1984: 166) includes it in a list of words that have only plural number forms.
22
the other hand, has proportionally fewer of those (see Appendix B). This is in part because Konso also includes many abstract nouns in this category.
Reconstructions of the cognates are provided below, beginning with the reconstructions of the cognates that have plural gender in both languages. The first thing that should be noticed here, is that, with the exception of *sinɗ1a(a), no ending can be reconstructed for these words.
Also, some reconstructions require some further explanation.
*ʕaan- „milk‟ Pr.Oromoid: *aana(a)
Pr.W.Omo-Tana: *ʔeenu
*bis- „water‟ Pr.Oromoid: *bi aa
Pr.W.Omo-Tana: *biče
*mulk-/*mukl- „iron (leg) bracelet(s)‟ Konso: muklaa
Pr.W.Omo-Tana: *mu(u)lk- *sinɗ1a(a) „urine‟ Pr.Oromoid: *sinɗ1aa
Pr.W.Omo-Tana: *sinɗ1a
*suʕn- „nose‟ Pr.Oromoid: *suʕnaa
Pr.W.Omo-Tana: *soon-
*ud- „faeces‟ Pr.Oromoid: *udaa
Pr.W.Omo-Tana *ʔudu
The first reconstruction requiring further explanation is *bis- „water‟. This is the reconstruction provided by Black (1974), based solely on his principal languages (Somali, Oromo and Konsoid). However, Black (1974: 174) also adds that the reflexes in his non-principal languages (which include Western-Omo-Tana) suggest a phoneme distinct from *s. The comparisons made in this thesis do not shed any further light on this issue, so it retains Black‟s reconstruction.
The other reconstruction that need extra explanation is *suʕn- „nose‟. This reconstruction is a possibility if Proto-Western-Omo-Tana underwent the same change as Konso (*u becomes *o before *ʕ). However, Black (1974) actually reconstructs this form without a pharyngeal. Instead, he reconstructs *s[i/u/o/a]n-, with the different vowel reflexes a result of the languages inheriting different forms caused morphological alternation of the vowel.
The reconstructions of the cognates that have plural gender in Konso, but not in Arbore, are fairly self-explanatory. However, one interesting outcome to notice is that it appears two of these reconstructions have an *-aa ending, which became the plural gender ending in Konso. Also of note is that unlike the previous set, this set contains two reconstructions that are made up of only one syllable.
*qor „wood‟ Pr.Oromoid: *qoraa
Pr.W.Omo-Tana: *qor *soʔ „meat‟ Pr.Oromoid: *soʔ-
Pr.W.Omo-Tana: *sóʔ *tira(a) „liver‟ Pr.Oromoid: *tira(a)
Pr.W.Omo-Tana: *tira *kaka(a) „oath‟ Pr.Oromoid: *kaka(a)
Arbore: kakannó
Finally, there are the words that are not cognates but semantically similar and that have plural gender in both Konso and Arbore. These pairs are interesting because they make up a relatively sizeable part of the Arbore plural gender words; specifically, they make up between a quarter and a third of all Arbore base plural gender words found in Hayward (1974). This
23
overlap is relevant because it shows that the semantic origins of plural gender are likely similar or possibly even identical. Alternatively, this could point to language contact
influencing the gender certain nouns belong to, although Konso and Arbore are not in direct contact with each other, which makes this scenario unlikely.
5.2 Potential Scenarios
Finally, it needs to be discussed which conclusions can be drawn from the analyses presented in this thesis. In this section, several possible scenarios will be discussed and compared to the results presented here.
Scenario 1:
None of the proto-languages discussed in this thesis, with the exception of Proto- Konsoid, had plural gender. The Konsoid languages and Arbore both developed plural gender fairly recently. This scenario requires that there are no traces of plural gender in any of the languages that currently do not have it.
Scenario 2:
Proto-(Southern-)Lowland-East-Cushitic did not have plural gender. Proto-Oromoid developed plural gender, but Western Omo-Tana did not, until Arbore developed it as well. Oromo lost plural gender again. This scenario requires that no traces of plural gender are found in Dhaasanac and Elmolo, while ideally some traces are found in Oromo.
Scenario 3:
Proto-(Southern-)Lowland-East-Cushitic did not have plural gender. Proto-Western-Omo-Tana developed plural gender, but Oromoid did not, until the Konsoid languages developed it as well. This scenario requires that no traces of plural gender are found in Oromo, while ideally some traces are found in Dhaasanac and Elmolo.
Scenario 4:
Proto-(Southern-)Lowland-East-Cushitic did not have plural gender. Both Proto- Oromoid and Proto-Western-Omo-Tana developed plural gender. Oromo, Dhaasanac and Elmolo lost plural gender again. In order to prove that this scenario is true, traces of former plural gender would have to be found in Oromo, Dhaasanac and Elmolo.
Scenario 5:
Proto-(Southern-)Lowland-East-Cushitic had plural gender. Both Proto-Oromoid and Proto-Western-Omo-Tana inherited plural gender. Oromo, Dhaasanac and Elmolo subsequently lost plural gender. This scenario requires that the plural gender systems of Konsoid and Arbore share some similarities and ideally some traces of former plural gender in the languages that lost it.
24
Proto-(Southern-)Lowland-East-Cushitic had plural gender. Proto-Oromoid inherited plural gender, while Proto-Western-Omo-Tana lost it. Oromo later lost plural gender as well, while Arbore redeveloped it. This scenario would be difficult to distinguish from scenario 5, because it could look like Arbore never lost plural gender if it redeveloped plural gender based on remaining traces of the old plural gender system.
Scenario 7:
Proto-(Southern-)Lowland-East-Cushitic had plural gender. Proto-Western-Omo- Tana inherited plural gender, while Proto-Oromoid lost it. Dhaasanac and Elmolo later lost plural gender as well, while the Konsoid languages redeveloped it. This scenario would be difficult to distinguish from scenario 5, for the same reasons as scenario 6.
Scenario 8:
Proto-(Southern-)Lowland-East-Cushitic had plural gender. Neither Proto-Oromoid, nor Proto-Western-Omo-Tana inherited plural gender. The Konsoid languages and Arbore later redeveloped plural gender. This scenario would be difficult to distinguish from scenarios 5, 6 and 7, for reasons similar to the reason scenarios 6 and 7 are difficult to distinguish from scenario 5. Losing plural gender in both branches and later redeveloping it in both branches also seems rather unlikely, but this scenario is included here for the sake of completion.
The presence of a set of words in (Borana) Oromo that are always plural and that form neat cognates with Konso plural gender words makes scenarios 1 and 3 less likely; although it could still be argued that the proto-language was more like Oromo than Konso, with a small set of nouns that are always plural instead of a full-fledged plural gender class. However, the case for arguing that no ancestor language of Oromo had plural gender becomes even weaker when taking into account the fact that (Borana) Oromo has a set of adjectives that take a suffix -oo, specifically for agreement with the nouns in -(aa)ni that are always plural. The existence of this agreement pattern essentially disqualifies scenarios 1 and 3 from further consideration.
Eliminating scenarios 1 and 3 was relatively simple. However, trying to determine which of the remaining scenarios is most likely is a bit more difficult. There are multiple reasons for this. The first is that trying to reconstruct what happened in the western Omo-Tana branch is made more difficult by three issues: one is the fact that Dhaasanac is a very innovating language, which means it is not unlikely that it would have lost any hypothetical traces of former plural gender; the second issue is that Elmolo is practically extinct and not very well described. The third issue is that Hayward spent a relatively short period of time working on Arbore, which means that it is quite possible that data relevant to this thesis was not collected. This makes it especially difficult to distinguish the two remaining scenarios without plural gender in Proto-(Southern-)Lowland-East-Cushitic (i.e. scenarios 2 and 4) from each other, because Dhaasanac and Elmolo are then the only sources for comparison to find out whether plural gender developed in Proto-Western-Omo-Tana or later, in Arbore.
Fortunately, those first two issues are less problematic for determining whether Proto-(Southern-)Lowland-East-Cushitic already had plural gender, since the Western Omo-Tana languages can also be compared to the Oromoid languages. Currently, the evidence seems to point to the absence of plural gender in Proto-(Southern-)Lowland-East-Cushitic. Looking at the nouns themselves, no clear pattern can be reconstructed that could point a plural gender. Similarly, comparing the agreement systems of Konso and Arbore shows almost no
