The enclitic -mm in Amharic: reassessment of a multifunctional morpheme (1).

Demeke, Girma A. ; Meyer, Ronny

Abstract

In this article we examine the enclitic -mm in Amharic and its claimed functions as topic marker and coordinative conjunction. The enclitic -mm has multiple translations into English, and would thus seem to constitute a multifunctional morpheme. In contrast to previous analyses, we argue that the several interpretations of enclitic -mm in discourse are due to different context-dependent readings of its single function, namely as contrastive focus marker.

1. Introduction

The enclitic -mm in Amharic has been claimed to fulfill several discourse-pragmatic and grammatical functions. On the one hand it is referred to as a marker of topic, topic change or focus (Blejer 1986: 186; Bliese 1988: 619; Gasser 1983: 114, 128; Hetzron 1973). On the other hand it is also referred to as a coordinating conjunction with noun phrases and as a marker of indefinite pronouns (Blejer 1986: 185ff.; Leslau 1995: 882ff.). Kapeliuk (1978, 2001) and Gasser (1985) analyze the enclitic as a connective element on the discourse level. Hetzron (1973: 4-5, 8), furthermore, proposes that the same enclitic -mm is also involved in the circumfixal negation of matrix clauses in the indicative mood. According to Hetzron it also functions as a prefixal nominalizer with relative verbs in the imperfective aspect.

Amharic is an Ethiosemitic language spoken in Ethiopia by approximately 80% of the population (cf. Meyer and Richter 2003: 40; Girma 2001). Structurally it belongs to Heine's type D languages (Heine 1975: 36f.), i.e., it is a subject-object-verb (SOV) language with modifying elements preceding the head. Amharic is not a language that possesses obligatory focus markers in syntax like, for instance, Somali (cf. Saeed 1984) or Zay (cf. Meyer 2002, 2005: 291). Focus in Amharic, or more generally the pragmatic status of a sentential constituent in the sense of Chafe (1976: 27), is mainly controlled by pragmatics and not by grammar, i.e., its expression is optional. (2) Information structuring applies both on the sentential and on the extra-sentential level. While on the sentential level a single sentence is the frame of reference for discourse markers, on the extra-sentential level constituents of at least two sentences are set into relationship to one another.

2. Interactions between sentence topic and the morpheme -mm

According to Hetzron (1973) and Blejer (1986), one of the main functions of-mm is to mark the topic of a sentence. Kapeliuk (1978: 275) and Girma and Meyer (2007), in contrast, do not consider it a topic marker because topics in Amharic are marked syntactically by fronting them into the sentence-initial position. In an SOV language like Amharic the sentence-initial position is usually filled by the subject. However, any nonverbal phrase can occur in sentence-initial position and become the topic of the sentence if it is definite in the sense of being identifiable in the discourse, as in the following examples: (3)

 (1) a. lelocc-u bere-wocc necc-u-n bere
 other:p-DEF ox-p white-DEF-ACC ox
 abarrer-u-t.
 send.away:PV-3p-AgrO:3sm
 'The other oxen sent the white ox away.'

 b. bere-we-ne-mm geb-occ bell-u-t.
 ox-DEF-ACC-MM hyena-p eat:PV-3p-AgrO:3sm
 'As for this ox, the hyenas ate him.'

 (2) a. be-sostenna-w qen mammo ye-qen sera
 at-third-DEF day Mammo GEN-day work
 agenne.
 find:PV:3sm
 'On the third day Mammo got work for one day.'

 b. le-serra-w sera-mm ye-gulbet waga
 for-work:PV:3sm-DEF work-MM GEN-hard.work price
 yemm-i-hon saga and habtam
 REL-3sm-be:IPV meat one rich.person
 sette-w.
 give:PV:3sm-AgrO:3sm
 'As for the work he did, a rich man gave him meat as payment
 for the hard work.'

In example (1) the phrase berewen 'the ox (ACC)' is the topic of sentence (b) even though it is the direct object in the sentence. It represents background information by virtue of referring to the noun neccun bere 'the white ox (ACC)' in sentence (a). In sentence (b) from example (2), the prepositional phrase leserraw sera 'for the work he did' is the topic of the sentence. The head noun sofa 'work' represents background information because it has already been introduced in the preceding sentence. In both sentences we can observe a change of the word order from basic SOV to OSV. The initial phrase of the respective (b) sentences is a direct object explicitly marked with accusative in (1) and an adjunct in (2). An attempt to front the subject and habtam 'a rich person' in the (b) sentence from example (2) so as to precede the direct object ye-gulbet waga yemmihon sera 'meat as payment for the hard work' would yield an odd sentence.

Although the enclitic element -mm follows the topicalized constituent, it does not itself have the function of a topic marker. In fact, the enclitic element -mm can also be found cliticized to constituents which are not topics in positions other than sentence-initial (cf. example (3) below). Rather, the optional suffixation of enclitic -mm to a sentence-initial constituent is due to its discourse-pragmatic function as a focus marker. (4) The definite phrase marked by enclitic -mm at the beginning of a sentence, therefore, conveys two pieces of pragmatic information at the same time: it is the topic or contains the given information, and it indicates contrastive focus as well (of. also Gasser 1985: 52). We understand contrastive or identificational focus in the sense of Drubig and Schaffar (2001: 1080) and Payne (1997: 269f.) as a device used by the speaker to emphasize a constituent which is part of the presupposed knowledge shared between speaker and addressee. (5) According to Dik et al. (1981: 60) the term contrastive focus encompasses various subcategories like selective, expanding, restricting, replacing and parallel focus. Selective focus denotes "one item from among a presupposed set of possible values" (Dik et al. 1981: 62), while expanding focus adds items to a presupposed set of values (Dik et al. 1981: 65). It is not always possible to make a clear-cut distinction between these two kinds of focus because the presupposed set is in fact often salient knowledge shared between speaker and hearer. It depends, thus, on the interpretation of the hearer if the speaker singles out one item from a given set of values (selective focus) or if he adds an additional item to this set (expanding focus).

3. Focus marking by -mm

3.1. Focusing constituents

Phrases or constituents marked by the enclitic -mm also occur in nonsentence-initial position. In this case it can be shown clearly that the enclitic element -mm functions only as a focus marker: (6)

 (3) hesan-u be-yye-gize-w le-enat-u
 baby-DEF at-DISTR-time-DEF to-mother-poss:3sm
 yeteleyayye-w teyaqe-mm ye-teyyeq-at nebber.
 various-DEF question-MM 3sm-ask:IPV-AgrO:3sf aux(past)
 'The small child was asking his mother VARIOUS QUESTIONS every
 time.'

 (4) lemen ke-zih ke-mme-te-bey-ew ferafere
 why from-this from-REL-2sf-eat:IPV:3sf-DEF fruit
 le-ene-mm tennes tennes
 for-I-MM few few
 atte-werewer-i-lle-nne-m?
 NEG:2sf-throw.down:IPV-2sf-for-AgrO: 1s-NEG
 'Why don't you throw down some fruits FOR ME from among those
 which you are eating?'

 (5) be-akkababi-w be-astesasseb beslet-u-nna
 at-vicinity-DEF at-thinking cleverness-poss:3sm-and
 be-asteraqinnet-u ye-tawweq-u and terecce
 at-being.mediator-poss:3sm REL-be.known:PV-3p one Tiriche
 ye-tebal-e ye-tota-wocc aleqa-mm nebber.
 REL-be.named:PV-3sm GEN-monkey-p chief-MM aux(past)
 'There was A CHIEF of the monkeys who was named Tiriche and
 who was known for his maturity in thinking, his cleverness and his
 abilities as mediator in the vicinity.'

In examples (3) to (5) the constituent marked by the enclitic -mm does not have a topic function at all. It is used to change the neutral status of a phrase or constituent to a status of contrastive focus. In examples (3) and (4) the constituents to which -mm is suffixed express expanding focus, i.e., they add a single item to a presupposed set of values. Although not mentioned overtly, the presupposed set seems to be salient information shared between speaker and hearer. In sentence (5), by contrast, the enclitic -mm singles out the item yetotawocc alexia 'chief of the monkeys' in terms of his aforementioned qualities yastesasseb beslet 'maturity of thinking' and asteraqinnet 'ability as mediator'. The enclitic -mm in sentence (5), therefore, expesses selective focus.

Focus marking via enclitic -mm is not restricted to nonverbal constituents but may also appear on verbs. The usage of the enclitic with main verbs, however, is very rare; an example is (6):

 (6) yeh-en gudday le-enat-u y-awrall.
 this-ACC matter to-mother-poss:3sm 3sm-tell:IPV:aux(np)
 y-asay-all-emm.
 3sm-show:IPV-aux(np)-MM
 'He tells this matter to his mother. HE ALSO SHOWS (it to her).'

 (7) aheya-we-n gend lay asro-mm encet
 donkey-DEF-ACC trunk above tie:CONV.3sm-MM wood
 leqema gebba.
 picking enter:PV:3sm
 'AFTER HE TIED Tim DONKEY TO A TRUNK he went to pick up wood.'

 (8) betam sele-rab-accewe-mm
 very because-be.hungry:PV:3sm-AgrO:3p-MM
 ke-wasaccew ye-wett-u-t qemse-w
 from-cave:poss:3p REL-come.out:PV-3p-DEF taste:CONV-3p
 l-i-mmelles-u ne-w.
 INGR-3p-return:IPV-p COP-3sm
 'SINCE THEY WERE VERY HUNGRY, they left their cave intending to
 return after they had eaten something.'

 (9) tennes ende-bella-mm
 little that-eat:PV:3sm-MM
 ye-metta-bbe-te-n gudday
 REL-come:PV:3sm-at-AgrO:3sm-ACC matter
 ende-mm-iy-awq-ew le-asser aleqa-w
 that-REL-3sm-know:IPV-AgrO:3sm to-officer-DEF
 negger-e-w.
 tell:PV-3sm-AgrO:3sm
 'JUST AFTER THEY HAD EATEN A LITTLE BIT, he told the officer that
 he knew the matter for which he came.'

(10) sew-occ yeh-ecce-n qongo le-magbat negus-u-n
 man-p this-f-ACC beauty for-marriage king-DEF-ACC
 b-i-teyyeq-u-mm l-i-ssaka-ll-accew
 if-3p-ask:IPV-p-MM INGR-3sm-be.successful:IPV-for-AgrO:3p
 al-cal-e-m.
 NEG-be.able:PV-3sm-NEG
 'EVEN IV the people ASKED the king to marry this beauty, they
 could not succeed with it.'

In sentence (6) the action of showing, which is marked by the enclitic -mm on a main-clause verb, is an expansion of the information provided in the previous predication. In sentences (7) to (10) a subordinate verb/ predication marked by the enclitic -mm attracts the particular attention of the hearer. The discourse-neutral variants of these sentences would have been uttered without the attached enclitic. In all these sentences the subordinate phrase provides specified and emphasized information which modifies the main predication in a certain way. These subordinate clauses are, therefore, marked for selective contrastive focus.

3.2. Multiple focusing?

If our analysis of the enclitic -mm as a contrastive focus marker is correct, then Amharic should represent a language in which multiple focus marking is possible. The following example illustrates this very situation:

(11) negus-u-mm betam b-i-bbesacc-u-mm
 king-DEF-MM very if-3p-be.annoyed:IPV-p-MM
 ke-meggedel at-zell-u-m
 from-being.killed NEG:2p-jump:IPV-p-NEG
 ble-w ...
 say:CONV-3sm ...
 'EVEN IF THE KING WAS very ANNOYED he said that they wouldn't
 escape from being killed and ...'

In sentence (11) the subject nogusu 'the king' is the topic of the sentence. It is also marked for selective focus by suffixation of the enclitic element -mm. In addition, the conditional clause betam bibbesaccu 'even if he (polite) feels annoyed' appears to be marked for focus because the enclitic element -mm is suffixed to it as well. This clause has the reading of a concessive conditional. According to Konig (1991: 85), concessive conditional clauses very often consist of a conditional or temporal clause and a focus particle meaning "even" or "also". Sentence (11) provides, therefore, an example with two focused elements in a single sentence. The two focused constituents can coexist because they operate on two different levels. In Zay, a language that is genetically related to Amharic, this kind of multiple focus marking would have led to an ung-mmmatical sentence (Meyer 2005: 290ff.). Amharic differs considerably from Zay in this respect. The main reason for the different behavior of Zay and Amharic seems due to different degrees of obligatoriness of the focus elements. There exist two types of focus markers in Zay: the first type is syntactically obligatory in forming a grammatical sentence; the second type is optional. While several of the optional focus markers may co-occur in a single Zay sentence, it is only possible to apply one obligatory focus marker per sentence. Amharic does not have obligatory focus markers but only optional ones, and these can co-occur in a sentence.

4. Scope of the enclitic -mm

Thus far we have paid attention to constituents focused by the enclitic -mm as a whole. Now we examine the scope of the enclitic -mm within such a constituent. It appears that only the constituent immediately to the left of the enclitic is in its scope:

(12) and wetadder ye-negus-u-n mist
 a soldier GEN-king-DEF-ACC wife
 ayyat.
 see:PV:3sm:AgrO:3sf
 'A soldier saw the wife of the king.'

(13) and wetadder ye-negus-u-ne-mm mist
 a soldier GEN-king-DEF-ACC-MM wife
 ayyat.
 see:PV:3sm:AgrO:3sf
 'A soldier saw the wife of the KING.'

(14) and wetadder ye-negus-u-n mist-emm
 a soldier GEN-king-DEF-ACC wife-MM
 ayyat.
 see:PV:3sm:AgrO:3sf
 'A soldier saw the WIFE of the king / A soldier saw THE WIFE OF
 THE KING.'

In sentence (12) no constituent conveys special discourse-pragmatic information; the sentence is neutral in regard to focus. In sentences (13) and (14) the enclitic -mm as focus marking device is attached to the genitive construction yenegusu mist 'the wife of the king', which itself is in object position. The two sentences are very similar in meaning, but they differ in the scope of the focused elements. In sentence (13) only the noun negus 'king' is in the scope of selective focus. That means that a presupposed set of items exist which have the quality of being wives. One wife is singled out from this set: the wife of the king (as opposed to the wife of a peasant or soldier, etc.). In (14), however, the interpretation is not as easy as in (13) because an ambiguity exists on the syntactic level. One reading of the sentence is that inside the phrase yenegusu-n mist only the head noun mist 'wife' is contrasted against several possible persons who may belong to the king, e.g., yenegusu-n legg, 'the child of the king', etc. In this case the modifying genitive 'of the king' remains constant and is out of the scope of the focus. The phrase in (14) has a second reading, however, where the modifying genitive noun and the head are taken as a single constituent. In this case the whole phrase yenegusu-n mist is contrasted against another entity, e.g., aster-en 'Aster (female name)', etc. In this reading the enclitic -mm does not assign selective but expanding focus to the whole genitive construction.

The interpretation of the scope of the enclitic -mm depends on both the particular discourse situation and the syntactic structure. When a genitive construction or an adverbial phrase or clause is the first constituent of a sentence, the position of the enclitic -mm is crucial for the interpretation of the phrase. A reading of a whole multiword phrase as the focused constituent is only possible when the enclitic is attached at the end of the entire phrase.

(15) a. ye-bet ensesat and lay hone-w wede
 GEN-house animal togehter be:CONV-3p towards
 lela ager hed-u.
 other country go:PV-3p
 'The domestic animals went together to another country.'

 b. tennes ke-tegwaz-u behwala-mm betam
 little after-travel:PV-3p after-MM very
 sele-dekkem-accew be-mekina le-mehed
 because-be.tired:PV:3sm-AgrO.3p by-car for-going
 tesmamm-u.
 agree:PV-3p
 'AFTER THEY HAD TRAVELED A LITTLE BIT they agreed to go by
 car because they were very tired.'

The adverbial clause tennes ketegwazu behwala-mm 'after they had traveled a little bit' in sentence (15b) contains background information about leaving a country which has already been mentioned in sentence (15a). The adverbial clause stands in sentence-initial position and is, therefore, the topic of the sentence. Furthermore, it is focused because it is followed by the enclitic -mm. The scope of the focus is the whole adverbial clause. If the enclitic -mm were suffixed to an element standing in nonfinal position, then only the respective constituent of the clause would be focused but not the entire clause. Thus, in sentence (16) the scope of the enclitic -mm comprises only the verb ketegwazu 'after they traveled'.

(16) tennes ke-tegwaz-u-mm behwala betam
 little after-travel:PV-3p-MM after very
 sele-dekkem-accew be-mekina le-mehed
 because-be.tired:PV:3sm-AgrO.3p by-car for-going
 tesmamm-u.
 agree:PV-3p
 'AFTER THEY HAD TRAVELED a little bit they agreed to go by car
 because they were very tired.'

The different scopes of the enclitic -mm are important for the discourse-pragmatic interpretation of circumpositional phrases. Circumpositions consist of a phrase-initial prefix and a modifying postpositional element at the end of the phrase. The postpositional element is optional in some circumpositional phrases. The presence of a modifying postpositional element is quite frequent in today's Amharic. There are, however, constructions in which the postpositional element is regularly omitted, as in phrases where a demonstrative pronoun (-zih 'this' or -ziya 'that') is the complement. Such phrases are illustrated as follows:

Circumposition Prefix

ke-ziya behwala ~ ke-ziya 'after that, then'
be-zih yetenessa ~ be-zih 'due to this'
be-zih gize ~ be-zih 'then'

The meaning of the corresponding circumpositional and prefix phrases is similar. Most often the discourse context enables the hearer to resolve ambiguous constructions (e.g., be-zih). When the circumpositional phrase appears in sentence-initial position, the enclitic -mm can be suffixed either to the demonstrative or to the postpositional element:

(17) ke-ziya behwala-mm negus-u wede-bet-accew
 from-that after-MM king-DEF towards-house-poss:3p
 temelles-u.
 return:PV-3sm
 'AFTER THAT the king returned to his house.'

(18) ke-ziya-mm behwala negus-u wede-bet-accew
 from-that-MM after king-DEF towards-house-poss:3p
 temelles-u.
 return:PV-3sm
 'After THAT the king returned to his house.'

When the enclitic is suffixed to the postpositional element at the end of the phrase, as in keziya behwala-mm 'after that', a reading as topic constituent in focus is achieved. The demonstrative pronoun refers to given information, and is also marked for contrastive focus. The scope of -mm, in this case, is not only the preceding postpositional element but the entire adpositional phrase. However, when the enclitic -mm is suffixed to the demonstrative pronoun keziya 'from that', only the demonstrative is in the scope of the enclitic -mm. In sentence (18), therefore, only the demonstrative pronoun -ziya 'that' is marked for contrastive focus of restriction by the enclitic -mm. When instead of a circumpositional phrase only a prefix is used, an ambiguous reading results:

(19) a. hulett-u gwadenna-mocc le-ceqeceq-accrew mensa[??]e
 two-DEF friend-p for-quarrel-poss:3p reason
 ye-hon-e-we-n neger
 REL-be:PV-3sm-DEF-ACC thing
 ya-negr-u-t-all.
 3sm-tell:IPV-p-AgrO-3sm-aux(np)
 'The two friends tell him the reason for their quarrel.'

 b. ke-ziya-mm mengedenna-w "yeh qelal aydelem"
 after-that-MM passenger-DEF this easy not.to.be:3sm
 ya-l-all.
 3sm-say:IPV-aux(np)
 'Then the passenger says that this is not easy.'

The prefix ke- followed by the demonstrative pronoun in (19b) may, on the one hand, refer to the event mentioned in the preceding sentence. In this case it can be considered as a topicalized constituent, which is also in focus. On the other hand, it is also possible to analyze it only as a focused constituent referring to the just-completed speech event itself, with no reference to the previous sentence, i.e., the passenger starts to talk only after the two friends have spoken to him and not before. This ambiguity seems to arise due to the omitted postpositional element.

5. On the concatenative function of -mm

The enclitic -mm can also be used as a coordinating device to join several nouns occupying the same syntactic position. In this function the enclitic -mm is suffixed to every noun in the respective enumeration:

(20) sew-u-mm negus-u-mm denegget.-u.
 man-DEF-MM king-DEF-MM be.surprised:PV-3p
 'The people as well as the king were both surprised.'

An alternative way of joining nouns is the usage of the coordinative conjunction enna ~ -nna 'and':

(21) sew-u-nna negus-u denegget.-u.
 man-DEF-and king-DEF be.surprised:PV-3p
 'The people and the king were surprised.'

However, examples (20) and (21), though similar in meaning, differ in the interpretation of the conjoined nouns. Whereas in (21) the nouns sew 'people' and negus 'king' are taken as one single set consisting of two items, in (20) these nouns must be analyzed as two different sets each consisting of only one item. (7) In Example (20) the nouns sew 'people' and negus 'king' are, therefore, not simply combined in the same single syntactic position. This becomes apparent when we have a closer look at the semantic behavior of the two morphemes:

(22) gebs-enna masella zerra.
 barley-and maize sow:PV:3sm
 'He sowed barley and maize.'

(23) gebs-emm masella-mm zerra.
 barley-MM maize-MM sow:PV:3sm
 'He sowed BARLEY and also MAIZE.'

In Example (22) one possible reading is that a peasant sowed barley and maize together on a plot of land, i.e., the nouns gebs 'barley' and masella 'maize' refer to a single set of crops which have been sown. This reading of mixing the two crops together can be manifested overtly by the usage of the modifying phrase addebalqo '(he having) mixed together':

(24) gebs-enna masella addebalqo zerra.
 barley-and maize mix:CONV:3sm sow:PV:3sm
 'He mixed barley and maize and sowed it.'

Due to the modifying phrase in (24) it is clear that both crops have been sown together. This cannot be the case in (23); using the modifier addebalqo '(he having) mixed' yields a semantically unacceptable construction:

(25) *gebs-emm masella-mm addebalqo zerra.
 barley-MM maize-MM mix:CONV:3sm sow:PV:3sm
 '*He mixed BARLEY and also MAIZE and sown it.'

The unacceptability of sentence (25) is due to the function of the enclitic -mm. It implies that the two nouns belong to two different sets and emphasizes both sets against each other so that we have two sets of nouns in the status of selective focus. Due to this contrastive focus marking, the two nouns cannot be taken as simply being conjoined. A lexical item, like addebalqo, which overtly emphasizes this "conjoinedness" will lead, therefore, to an ungrammatical structure.

6. Adverbs and enclitic -mm

The enclitic -mm can also be attached to time adverbs like telant 'yesterday', zare 'today', ahun 'now', etc. In most cases this yields an emphasis on the time adverb, which can be interpreted as a corrective contrastive focus:

(26) zare-mm al-teressa-m.
 today-MM NEG-be.forgotten:PV:3sm-NEG
 'Even today it is not forgotten.'

In sentence (26) we can imagine a situation where the speaker supposes the hearer to assume that something has been forgotten at the present time. The speaker corrects this assumption by suffixation of the enclitic -mm to the adverb. The interpretation as a focused adverb also appears with ahun 'now':

(27) ahun-emm gudgwad k-agenn-e ye-gba!
 now-MM hole when-find:PV-3sm 3sm-enter:JUS
 'When he finds a hole JUST NOW, let him enter into it!'

In example (27) the speaker again focuses on the time adverb, but now not to correct a false assumption. Here the speaker is using selective focus. Depending on the context in which sentence (27) is used, an additional interpretation of ahunemm is possible. The combination of the time adverb ahun with the suffixed enclitic -mm has in fact been lexicalized to yield a new adverbial meaning 'again':

(28) be-zih gize ahunemm totit egg-wa-n
 at-this time now:MM monkey:DEF(f) hand-poss:3sf-ACC
 awetta-cc.
 stretch:PV-3sf
 'At this time again the small monkey stretched out his hands.'

In sentence (28) the phrase ahunemm can only have the reading 'again' and not 'now' because the time reference is given by the prepositional phrase bezih gize 'at this time'.

7. The enclitic -mm with quantifiers

Quantifiers like hullu 'all', bezu 'many', etc. can also be marked with the enclitic -mm. When it is attached to the quantifier hullu 'all', enclitic -mm does not emphasis each individual item encompassed by the quantifier, as is the case with numerals (see Section 8.1), but refers to a closed set of items as a single individuated entity which is contrasted against another set:

(29) geb-occ egegg eyye-tedesset-u hullu-n-emm
 hyena-p very CONT-be.happy:PV-3p all-ACC-MM
 bere-wocc cerres-w-accew.
 ox-p finish:PV-3p-AgrO:3p
 'The hyenas were very happy and finished ALL of the oxen.'

The phrase berewocc 'oxen' in sentence (29) refers to four oxen of different colors that have been previously mentioned in the discourse. The four oxen thus represent a given set of items. This entire set is put into a contrastive focus by the phrase hullunemm berewocc 'all oxen (ACC)' in the sense that not only one or two of the oxen have been eaten but all of them. The quantifier hullunemm 'all (ACC)' cannot overtly be marked by a definite article. Nevertheless it is referential and specific. It does not denote any number of oxen but it refers specifically to the aforementioned set of four oxen. The specific reference in this case can only be achieved by the enclitic -mm, because there is no definite article. This suggests that the enclitic may actually encode two functions: a) marking of contrastive focus and b) referring to a specific entity in the discourse. It becomes even clearer that the combination of quantifier and enclitic indeed refers to a specific set of items when we contrast a phrase with the enclitic -mm against a phrase without it:

(30) sew hullu / ?hullu sew
 man all all man
 'all men (everybody)'

(31) hullu-mm sew
 all-MM man
 'all the people'

First, we can observe a syntactic restriction. The quantifier marked by the enclitic -mm must appear directly before the head noun. By contrast, when enclitic -mm is not involved, the position of the quantifier is free in regard to the head noun. (8) Phrase (30) refers to a nonreferential set of men, i.e., it can either be all of mankind or all of the peoples of Ethiopia or all of the people in a classroom, etc. Phrase (31) differs in this regard from phrase (30) because due to either the general knowledge of the speaker and hearer or to the discourse, there is reference made to a specific, closed group of men marked by the enclitic -mm.

The following sentence (32) is part of a story about a donkey, a dog and a goat:

(32) sememmennet-accew hullu-mm ye-yye-ras-u-n
 agreement-poss:3p all-MM GEN-DISTR-self-poss:3sm-ACC
 hisab keflo wedasseb-u-bb-et
 bill pay:CONV:3sm towards:(REL)-think:PV-3p-to-AgrO:3sm
 bota le-medres nebber.
 place for-arriving aux(past)
 'Their agreement was that each of them would pay the bill by
 himself and then go to whatever place he wants.'

Here, the quantifier hullu-mm with the attached enclitic -mm refers to a known group of animals consisting of the abovementioned donkey, dog, and goat. If in sentence (32) the enclitic were not suffaxed to the quantifier then the interpretation would change slightly: the quantifier hullu would simply stand for an unknown and unspecific number of persons/animals who agreed to pay by themselves, with no reference to any particular entity in the common knowledge of speaker and hearer.

When the quantifier is involved in a comparison the enclitic -mm must always be used:

(33) keber ke-hullu-mm belay ne-w.
 honor from-all-MM above COP-3sm
 'Honor is above all.'

Here again it is clear that the enclitic -mm is functioning as a reference marker to a given, closed set of items in the world. This set need not be mentioned in the preceding discourse but can be part of the common knowledge presumed to exist between speaker and hearer. In sentence (33) the set is a set of qualities, of which keber 'honor' is seen as the most important.

8. Numerals and wh-words with the enclitic -mm

8.1. Numerals

A numeral always has a specific reference to a certain set of items when the enclitic -mm is attached to it. Beside this reference to a specific set of items, the enclitic also functions to distributively single out each item expressed by the numeral:

(34) anbessa-wa sost awre-wocc se-te-teyyeq
 lion-DEF(f) three animal-p when-3sf-ask:IPV
 sost-u-mm awre-wocc be-yye-tera be-sehetet
 three-DEF-MM animal-p at-DISTR-turn at-mistake
 menged merr-w-at.
 way lead:PV-3p-AgrO:3sf
 'When the lion (f) asked three animals, each one of the three
 animals in turn told her the wrong way.'

The numeral sostumm in sentence (34) refers to the noun phrase sost awrewocc 'three animals' in the preceding subordinate clause. This referentiality is achieved by virtue of the definite article -u, which indicates that the numeral has a referent which is identifiable in the discourse. The function of the enclitic -mm is to mark restrictive focus on the numeral, yielding a reading of three individuated items. The interpretation of sostumm is, therefore, not that of a selective focus nor of a topicalized construction ('as for the three animals'). The numeral sostumm must be translated as 'each of the three animals', i.e., we have here a selective contrastive focus construction referring to a closed set of items and singling out successively every single item of the set.

Numerals standing alone as head of a phrase and followed by enclitic -mm are always marked as identifiable items, either by the definite article or by a possessive pronoun:

(35) be-duro gize anbessa-nna lam gwadenna-mocc nebber-u.
 at-former time lion-and cow friend-p be:PV-3p
 hulett-u-mm and and legg nebber-accew.
 two-DEF-MM one one child be:PV:3sm-AgrO:3p
 'Once the lion and the cow were friends. Each of the two had one
 child.'

(36) and qen geb-enna aheya wede wenz
 one day hyena-and donkey towards river
 abrew werred-u, ke-ziya-mm
 be.together:CONV:3p go.down:PV-3p from-there-MM
 weha le-metettat wede wenz-u hulett-u-mm
 water for-drinking towards river-DEF two-DEF-MM
 tetegg-u.
 approach:PV-3p
 'Once a hyena and a donkey went together down to a river. And
 then each of the two were approaching the river in order to drink
 water.'

(37) yeh legg hulett-accehu-mm ye-ne ne-w
 this child two-poss:2p-MM GEN-I COP-3sm
 kalaccehu ...
 when:say:PV:2p
 'When each of you say: "This child is mine" ...'

In sentence (35) the numeral appears in subject position. It is marked as an identifiable item by the definite article and followed by the enclitic -mm. Its sentence-initial position and the anaphoric reference to the preceding sentence cause it to function as a salient topic. Besides this topical reading, the enclitic -mm emphasizes that each of the two women has one child. In sentence (36) the same number expression is used, but not in sentence-initial position. Here it has no reading as a topic; rather, the presence of-mm emphasizes that the number expression 'two' consists of two separate items (a hyena and a donkey) and that each of these two items is doing a certain action. The combination of numeral, possessive pronoun and enclitic -mm in sentence (37) has a meaning similar to that of the numerals in sentences (35) and (36). The definite article or the possessive suffix of the 3sm -u refers to two specific entities, which are emphasized as individuated, single entities in the discourse by the presence of the enclitic -mm. Interestingly, the numerals in (35), (36), and (37) must be marked by enclitic -mm; otherwise the expression would be ungrammatical. Numeral constructions, therefore, differ considerably from other constructions involving enclitic -mm. The reason for this behavior is not obvious. In addition to syntactic restrictions there is also a semantic difference involved in the interpretation of numerals with the enclitic -mm. The enclitic does not refer as contrastive focus marker to the entire set of items, but rather contrasts or highlights each item of the set against the other items of the same set.

A further interesting case is the behavior of the word and 'one' in Amharic. It fulfills two functions: it either denotes the number 'one' or it is used as a marker for an indefinite noun phrase (cf. Baye 1996: 57f.). Significantly, when and functions as an indicator for indefiniteness it can be marked by the definite article and-u, translated as indefinite pronoun meaning 'someone, anybody' into English (Baye 1996: 58). Thus the expression and-u cannot be indefinite because the definite article signals that and should be identifiable in the discourse or represents common knowledge shared between the speaker and hearer. The idea of indefiniteness of and-u has developed due to its nonreferentiality or nonspecificity. Although the item expressed by and-u may be identifiable in the discourse it is not known to which individuated item in the real world and-u refers.

The combination of and 'one', the definite article and the enclitic -mm has an identificational as well as a referential or specific reading: (9)

(38) ke-mehal-accew and-u-mm
 from-among-poss:3p one-DEF-MM
 'just one from among them'

The pronoun andu-mm in (38) is identifiable as belonging to the phrase kemehalaccew 'from among them' due to the definite article and the ablative preposition ke-. The enclitic -mm functions here, similar to the above numeral expressions, as a selective focus marker, i.e., it strongly emphasizes the numeral.

The interpretation of the pronominal and followed by the enclitic -mm is polarity-sensitive:

(39) telant andemm sew al-metta-m.
 yesterday not.a.single person NEG-come:PV:3sm-NEG
 'Yesterday not a single person came.'

In negative clauses this pronoun is interpreted as an indefinite pronoun 'not a single, nobody'. The negative meaning of andemm can be further emphasized by suffixing -acc to the numeral, which then becomes andaccemm 'not even a single'.

8.2. Wh-words

Wh-words change from interrogative pronouns to indefinite pronouns or quantifiers when the enclitic -mm is attached to them:

man 'who' mannemm 'nobody, everybody, whoever'
men 'what' mennemm 'nothing, everything'
mecce 'when' meccemm 'never, whenever'.

The interpretation of the pronoun interacts with the polarity of the verb in the same way as and. The following examples illustrate this semantic shift with wh-words:

(40) telantenna man metta?
 yesterday who come:PV:3sm
 'Who came yesterday?'

(41) telantenna mannemm al-metta-m.
 yesterday nobody NEG-come:PV:3sm-NEG
 'Yesterday nobody came.'

(42) mennemm madreg sel-al-cal-u ...
 nothing make because-NEG-be.able:PV-3p
 'As they could not do anything ...'

(43) men ye-dderreg, meccemm yehew
 what 3sm-be.done:JUS whenever voila
 wusegi-w!
 take:IMP:2sf-AgrO:3sm
 'What should be done, voile, take it whenever you want!'

(44) meccemm seytan massasat-u ay-qer!
 whenever devil aberration-poss:3sm NEG:3sm-remain:IPV
 'The aberration of the devil will always remain!'

The wh-word man 'who' in (40) is used to refer to an individuated set of persons or a single person out of a set of possible persons who could have arrived. In (41), however, the pronoun mannemm 'nobody, anybody, somebody' does not refer to a single person out of a set but to the whole set of possible entities in general, as was the case with the quantifier hullu 'all'. These wh-words followed by enclitic -mm have a reading of an indefinite pronoun because they refer to an entire set without emphasizing its individuated entities.

These indefinite pronouns appear most often in negative sentence but they can also appear in affirmative sentences, although much less frequently:

(45) megeb-u-n mannemm ye-bla-w ...
 meal-DEF-ACC whoever 3sm-eat:JUS-AgrO:3sm
 'May whoever wants to, eat the meal!'

In expression (45) manemm simply stands for any person who wants to eat the meal from a known set of persons who probably can have this meal. Due to this unspecificity it seems that the speaker actually does not care who will eat the meal, which imposes a kind of negative connotation on the indefinite pronoun.

The use of the enclitic -mm with negative indefinite pronouns may lend support to Hetzron's (1973) hypothesis that there is a connection between the element -m (m) in negative matrix sentences and the enclitic -mm that expresses focus (cf. Girma 2003). At the present stage of research, however, the etymology of-m(m) with negative main verbs in the indicative mood still needs further investigation.

9. Conclusion

In the above description we have tried to illustrate the multifunctional usage of the enclitic -mm in Amharic. The two further functions of enclitic -mm as a nominalizer or a negative element, as suggested by Hetzron (1973), have not been taken into consideration in this article. Due to structural reasons we assume that the prefix mm- appearing with relative forms of the imperfective is a different morpheme (see also Gasser 1985 for the same conclusion).

In this article, we have argued that the common function of the enclitic -mm in all the various constructions is to mark contrastive focus. Contrastive focus, according to Dik et al. (1981), is a pragmatic category which comprises several subcategories such as focus of expansion or selection. As the enclitic -mm covers all these subcategories, its interpretation necessarily depends on the discourse situation and on the lexical items involved. In addition to contrastiveness, the enclitic -mm may express different kinds of referentiality. With numerals it can refer either to the individual elements of a given set or to the set itself. With all other lexical items, however, it seems to refer only to the entire set from which one item is selected or to which one item is added. Furthermore, certain constructions involving enclitic -mm have resulted in new lexicalized items. This is in particular the case with the adverb ahunemm 'again', and may also hold for phrases with a prefix as head and a demonstrative as complement.

Addis Ababa University

University of Mainz

Appendix. List of abbreviations

ACC accusative
AgrO agreement marker with
 the object
aux(np) non-past auxiliary
aux(past) past auxiliary
CONT continuative
CONV converb
COP copula
DEF definite article
DISTR distributive morpheme
f feminine
GEN genitive
IMP imperative
INGR ingressive
IPV imperfective
JUS jussive
NEG negative/negation
p plural
poss possessive suffix
PV perfective
REL relative clause marker
1s 1st person singular
2p 2nd person plural
2sf 2nd person singular
 feminine
2sm 2nd person singular masculine
3p 3rd person plural
3sf 3rd person singular feminine
3sm 3rd person singular masculine

Received 7 January 2004

Revised version received 20 January 2005

References

Baye Yimam (1996). Definiteness in Amharic discourse. Journal of African Languages and Linguistics 17(1), 47-83.

--(2004). The structure of Amharic noun phrases. Unpublished manuscript, Addis Ababa.

Blejer, Hatte Anne (1986). Discourse markers in early Semitic, and their reanalyses in subsequent dialects. Unpublished doctoral dissertation, University of Texas at Austin.

Bliese, Loren F. (1988). A discourse analysis of Amharic narrative. In Proceedings of the 8th International Conference of Ethiopian Studies, Taddese Beyene (ed.), 613-621. Addis Ababa: Institute of Ethiopian Studies, Addis Ababa University.

Chafe, Wallace L. (1976). Givenness, contrastiveness, definiteness, subjects, topics, and point of view. In Subject and Topic, Charles N. Li (ed.), 25-55. New York: Academic Press.

Dik, Simon C.; Hoffmann, Maria E.; de Jong, Jan R.; Sie, Ing Djiang; Stroomer, Harry; and de Vries, Lourens (1981). On the typology of focus phenomena. In Perspectives on Functional Grammar, Teun Hoekstra, Harry van der Hulst, and Michael Moortgat (eds.), 41-74. Dordrecht: Foris.

Drubig, Hans Bernhard and Wolfram Schaffar (2001). Focus constructions. In Language Typology and Language Universals, Handbooks of Linguistics and Communication Science 20.2, Martin Haspelmath, Ekkehard Konig, Wulf Oesterreicher, and Wolfgang Raible (eds.), 1079-1104. Berlin and New York: Walter de Gruyter.

Gasser, Michael (1983). Topic continuity in written Amharic narrative. In Topic Continuity in Discourse: A Quantitative Cross-Language Study, Talmy Givon (ed.), 95-139. Amsterdam and Philadelphia: John Benjamins.

--(1985). Amharic -M and -SS: morphology, theme and assumed knowledge. Lingua 65, 51-106.

Girma Awgichew Demeke (2001). The Ethio-Semitic languages: re-examining the classification. Journal of Ethiopian Studies 34(2), 57-93.

--(2003). The clausal syntax of Ethiosemitic. Unpublished doctoral dissertation, University of Tromso.

Girma A. Demeke and Meyer, Ronny (2004). [TEXT NOT REPRODUCIBLE IN ASCII] Die unauffindbare Nadel: Amharisch--deutsche Lesematerialen. University of Leipzig Papers on Africa, Language and Literature, No. 21. Leipzig: University of Leipzig.

--(2007). Topics and topicalization in Amharic. Journal of African Languages and Linguistics 28(1), 19-36.

Gutt, Ernst-August (1988). Toward an analysis of pragmatic connectives in Silt'i. In Proceedings of the 8th International Conference of Ethiopian Studies, Taddese Beyene (ed.), 665-678. Addis Ababa: Institute of Ethiopian Studies, Addis Ababa University.

--(1997). Concise grammar of Silt'e. In Silt'e-Amharic-English Dictionary (with concise grammar by Ernst-August Gutt), Eeva H. M. Gutt and Hussein Mohammed, 895-957. Addis Ababa: Addis Ababa University Press.

Heine, Bernd (1975). Language typology and convergence areas in Africa. Linguistics 14(4), 27-47.

Hetzron, Robert (1973). The element -mm in the Amharic verbal system. Annali dell'Istituto Orientale di Napoli 33, 1-10.

Hyman, Larry M. and Watters, John R. (1984). Auxiliary focus. Studies in African Linguistics 15(3), 233-273.

Kapeliuk, Olga (1978). Particles of concatenation and of reference in Amharic. Bulletin of the School of Oriental and African Studies 41(1), 272-282.

--(2001). Some suprasentential constructions in Amharic. In New Data and New Methods in Afroasiatic Linguistics: Robert Hetzron in Memoriam, A. Zaborski (ed.), 75-83. Wiesbaden: Otto Harrassowitz.

Kiss, E. Katalin (1998). Identificational focus versus information focus. Language 74(2), 245-273.

Konig, Ekkehard (1991). The Meanings of Focus Particles: A Comparative Perspective. London and New York: Routledge. Leslau, Wolf (1995). Reference Grammar of Amharic. Wiesbaden: Otto Harrassowitz.

Meyer, Ronny (2002). To be or not to be: Is there a copula in Zay? In Proceedings of the 14th International Conference of Ethiopian Studies, 6-11 November 2000, Addis Ababa, Baye Yimam, Richard Pankhurst, David Chapple, Yonas Admasu, Alula Pankhurst, Birhanu Teferra (eds.), 1798-1808. Addis Ababa: Institute of Ethiopian Studies, Addis Ababa University.

--(2005). Das Zay." Deskriptive Grammatik einer Ostguragesprache (Athiosemitisch). Cologne: Koppe.

Meyer, Ronny and Richter, Renate (2003). Language Use in Ethiopia from a Network Perspective. Frankfurt am Main: Peter Lang.

Payne, Thomas E. (1997). Describing Morphosvntax. Cambridge: Cambridge University Press.

Saeed, John Ibrahim (1984). The Syntax of Focus and Topic in Somali. Hamburg: Helmut Buske.

Notes

(1.) This article is part of ongoing research on language contact in Ethiopia within the framework of the multidisciplinary research project Kulturelle und sprachliche Kontakte: Prozesse des Wandels in historischen Spannungsfeldern Nordostafrikas/Westasiens at University of Mainz. The German Research Council has generously supported this research since 1997. We want to take this opportunity to express our gratitude to the University of Mainz and the German Research Council for their support. Furthermore, we want to thank Orin Gensler for his critical comments on this article and the English editing. We also express our gratitude to two unknown reviewers for their helpful suggestions for improving the article. Correspondence address: Ronny Meyer, SFB 295, Universitat Mainz, 55099 Mainz, Germany. E-mail: [email protected].

(2.) Hyman and Watters (1984: 242) point out that the "instance" which controls focus is of greater importance. This instance represents, according to them, a continuum between pragmatically controlled focus on the one hand and grammatically controlled focus on the other. In a pragmatically controlled focus system the speaker is free to choose between various focused and non-focused utterances according to the needs of the discourse situation. In a grammatically or syntactically controlled focus system the speaker must indicate focus on one of the constituents in any main clause.

(3.) Most of the cited sentences are taken from folk tales published in Girma and Meyer (2004).

(4.) This kind of topicalization corresponds to Chafe's (1976: 49) "English-style topics", meaning that "The so-called topic is simply a focus of contrast that has for some reason been placed in an unusual position at the beginning of the sentence."

(5.) Generally two types of focus constructions are distinguished in the literature: presentational (or assertive, information) focus and contrastive (or identificational) focus constructions. The two differ in respect to the presupposed knowledge between speaker and hearer. In presentational focus constructions no presupposed knowledge is assumed between speaker and hearer, while in contrastive focus constructions such presupposed knowledge exists (cf. Drubig and Schaffar 2001: 1079; Hyman and Watters 1984: 239f.; Kiss 1998). A distinction between these two foci on a morphosyntactic level seems useful in languages which possess obligatory focus markers. In Amharic this distinction is artificial at least in regard to morphosyntax.

(6.) This fact clashes with Gasser (1985) because the enclitic -mm is not used to indicate "discourse continuity".

(7.) This semantic difference in usage between the discourse-pragmatic particle -m(m) and a coordinative conjunction was already observed by Gutt (1988 and 1997: 942) for Silt'e, an Ethiosemitic language closely related to Amharic.

(8.) In standard Amharic it is more common to put the quantifier hullu after the head noun. In fact, only the quantifiers hullu 'all' and mullu 'full' can appear following the head noun. All other quantifiers and modifiers must precede their head noun. The uncommon phrase ?hullu sew 'all men' may represent a recent change in Amharic in which the quantifier hullu moves from an 'irregular' postnominal position to the regular prenominal position for quantifiers and modifiers.

(9.) See also Baye (2004) for the several interpretations of and 'one'.