Reliance on Predicative Units as a Method of Analysing and Translating Amharic Written Texts

In analysing and translating Amharic texts, most foreign students have experienced major problems while trying to ‘redirect’ the rigidly leftbranching syntax of Amharic into the predominantly right-branching syntax of most European languages. The way out of this difficulty proposed by some teachers of Amharic consists in the so-called ‘translating from the end’ principle: the student begins to decipher the structure of an Amharic sentence from the finite verb form at its very end and gradually proceeds towards the beginning of the sentence, untangling—one by one—the syntactic structures involved. In the course of teaching Amharic, I have found this method largely inadequate for the purpose it is supposed to achieve. As an alternative to the ‘translating from the end’ method the author proposes another strategy which could be termed ‘reliance on predicative units’. In using this strategy, the student should, first of all, single out verb forms which are likely to perform the function of (final or dependent) predicates. The second step consists in delimiting groups, or units, headed by every such verb form. The third step is to provide a rough, working translation of every such unit without taking into consideration its relations to the other units in the sentence. The fourth, and final, step consists in joining the translations of the predicative units together; at this stage, detailed knowledge of Amharic morphosyntactic rules is very much required.


Introduction
It is common knowledge among linguists that the verb constitutes the semantic and communicative centre of the sentence. William Foley has aptly compared the grammar of any language to a theatrical play: Consider for a moment the metaphor of the grammar of language as the performance of a play. In this play the nouns are like the actors and the props: the various persons and objects involved in the production. But verbs are actually the script: the way in which the various actors and props interact. 1 Any event occurring in extra-linguistic reality can be linguistically encoded as the combination of a predicate (the word denoting this very event) and its arguments (i.e. the nominal phrases or clauses denoting the participants in this event). Any given predicate in a given language always has a fixed 'configuration' of these participants, namely their number and the ways they interact in the course of the event (their 'syntactic roles').
In many languages, certain types of predicate are almost always (or as a general rule) left unexpressed on the surface level of sentence structure. In Russian, for example, if the event described in the sentence relates to the present time and the sentence itself is a copular sentence, the copular predicate is not expressed by any linear unit, in other words, it has a zero surface representation. Likewise, in Russian, if the sentence describes the condition of someone or something being situated in a place or at some definite point  The writing of this article has been aided by a grant from the Russian Foundation for Fundamental Research (РФФИ) no. 17-06-00391. The author would like to express his gratitude to Maria Bulakh (Russian State University for the Humanities, Moscow) and Magdalena Krzyżanowska (Universität Hamburg) for having read preliminary versions of this article and shared their comments. The responsibility for all mistakes and inaccuracies in the article remains solely with the author. 1 Foley 1993, 133. on the time axis, and if this situation relates to the present time, then we again have a zero predicate on the surface level.
In many languages certain types of predicate are generally expressed not by verbs, but by some other part of speech, most often by adjectives. The relations between prototypical verbs and prototypical adjectives constitute a well-known issue in typological linguistics. 2 In Amharic, prototypically adjectival meanings are conveyed either by lexemes whose syntactic behaviour resembles that of nouns (e.g. the names of colours or such notions as 'big', 'small', 'tall', 'short', and so on) 3 or by relative forms of stative verbs. 4 Important for our topic is the fact that both 'nominal' and 'verbal' adjectives, when used predicatively and not attributively, obligatorily take one of the copulas used in the language.
I maintain that, firstly, in Amharic, copular and copula-like predicates are always present both in the deep and in the surface structures of a sentence (unlike non-Semitic or Semitic languages, such as Russian or Arabic); and, secondly, Amharic has a fairly pronounced tendency to express stative predication either by means of nominals or by means of (relative) verbs-in both of these cases, a copula expressing predication is obligatory. Generally speaking, nearly every Amharic sentence contains a word whose part-of-speech characteristics is perceived as unambiguously verbal, the notable exception being constituted by spontaneous responses to cues in dialogues, 5 for example Däbdabbewǝn yäṣafäw man näw? Wändǝmme.
'Who is the one who wrote the letter?' 'My brother.' 2 On this issue see, for instance, Dixon 1977 andDixon 2006. 3 Amharic 'nounlike' adjectives are actually syntactically distinguishable from true nouns: they cannot occur as the subject of a clause unless they take the definite article. See Krzyżanowska 2017, 217. 4 Cf. such examples as yäbässälä, 'ripe'; yäbäsäbbäsä, 'rotten' (alongside the 'nominal' adjective bäsbassa); yämmimoq, 'warm' (alongside the 'nominal' adjective muq); yäqäzäqqäzä, 'cold' (alongside the 'nominal' adjective qäzqazza); yämmiyadallǝṭ, 'slippery' (alongside the 'nominal' adjective nominal adjective ); and many other similar instances. As we can observe from these examples, cases when a notion can be expressed both by a 'nominal' adjective and by a relative form of a stative verb are not infrequent in Amharic. 5 Similar to this kind of usage is the occasional omission of the verbal element of predication in elliptical sentences which can be considered as spontaneous responses only within the broader context of the actual speech act.

Finite vs Non-Finite Verb Forms
In Amharic, as in the majority of the world's languages, we find both finite and non-finite verb forms. The history of the attempts at defining the distinction between these two sets of forms goes back to the grammatical tradition of ancient Rome, where a verbal form was considered finite if it was specified according to the categories of aspect/tense, mood, and subject agreement; accordingly, a form was considered non-finite if it lacked specification of these categories. As time went by and as many new languages came under the scrutiny of linguists, it turned out that none of the restrictive values mentioned above is strictly obligatory for a verbal form to be analysed as non-finite. For example, in Amharic as well as in most other Ethiopian Semitic languages, converbs and some other verbal forms whose prototypical position within a sentence is non-final may take personal agreement markers. 6  For the purposes of this paper (but by no means universally) it is sufficient, in my opinion, to define finite verbal forms as those which are capable of performing the function of the predicate of a syntactically independent clause. Correspondingly, non-finite forms are defined here as those which are incapable of performing this function and which perform the function of the predicates of dependent clauses. 8 Attempting to characterize Amharic morphosyntax in terms of balancing and deranking, 9 we arrive at the conclusion that, in contrast to Standard Average European, Amharic exhibits a strong tendency towards deranking 6 In Gǝʿǝz these are only subject agreement markers, while in other Ethiopian Semitic languages converbs may also take markers of personal agreement with the (direct or indirect) object, thus possessing a polypersonal type of conjugation. 7 Haspelmath 1995, 5 (emphasis in the original). 8 In another Ethiopian Semitic language, Tǝgrǝñña, the converb is able to perform the function of the main predicate of an independent sentence. However, this peculiarity can be best described in the following terms: in Tǝgrǝñña the (erstwhile) purely converbal form has split into two different usages-non-finite and finite-and this split has created two morphologically identical, but semantically and syntactically different, verb paradigms. 9 A verb form is said to be balanced if it is identical to forms used in independent declarative clauses; a subordinate verb form is said to be deranked if it cannot be used in independent declarative clauses. For more on these terms see, for instance, Croft 1990, 216-217. verbal forms in dependent clauses. In fact, most often, these dependent verb forms are not identical to those employed sentence-finally. It is this tendency towards deranking, together with the basic SOV word order which always places the main predicate at the very end of a sentence, that is instrumental in distinguishing finite and non-finite predicates in Amharic. The Amharic non-finites can be subdivided into three main types: (1) converbal forms, (2) conjunctional forms (including the relative subtype), (3) infinitival forms. However, it should be pointed out that it is not easy to draw a clear-cut boundary between the first and the second type of non-finite. However, since it is not the aim of this paper to offer theoretical guidelines for distinguishing between converbal and conjunctional forms, no attempt is made here to work out clear-cut criteria for such differentiation. 10 Instead, we can simply follow the traditional approach, according to which Amharic has only one converbal form at its disposal, the so-called 'suffixal converb' or 'gerund' (nägro), while almost all other subordinate verb forms are united under the term 'conjunctional forms'. This approach is not at variance with the practical aims of analysing and translating Amharic written texts.
Infinitival forms, on the other hand, are fairly easy to define, since the essential component is always the infinitive, used either alone or accompanied by additional morphemes such as possessive markers, the accusative enclitic, or adpositional combinations.

Reliance on Predicative Units: The Essence of the Method
In analysing and translating Amharic texts, most students have experienced major problems while trying to 'redirect' the rigidly left-branching syntax of Amharic into the predominantly right-branching syntax of most European languages. One way out of this difficulty proposed by some teachers of Amharic consists in the aptly named principle of 'translating from the end': the student begins to decipher the structure of an Amharic sentence from the finite verb form at its very end and gradually proceeds towards the beginning of the sentence, untangling-one by one-the syntactic structures involved.
Having tried this method myself and tested it on quite a number of occasions on my students, I must say that I have found it largely inadequate for the purpose it is supposed to achieve. Generally, by the time the student reaches the beginning of the sentence, he/she has forgotten what there was at the end.
As an alternative to the 'translating from the end' method I venture to propose another strategy which could be termed 'reliance on predicative units'. This strategy is essentially a step-by-step process aimed at creating a full and correct translation of the sentence analysed into the mother tongue of the student.
1) The first step in this analysis would consist in singling out all those verb forms, both finite and non-finite, which can perform the predicative function.
2) The second step would be to delimit predicative units, that is, clauses headed by the verb forms singled out in the first step. In view of the general left-branching character of Amharic word order, this task is relatively easy to perform: in the overwhelming majority of cases, the whole span of the sentence is simply divided into segments, the right-hand boundary of each of them being constituted by the form of its predicate.
3) The third step is to attempt to create a rough, working translation (into the mother tongue of the student) of each of these units, preferably (but by no means rigidly) moving rightward, from the beginning of the sentence to its end, without taking into consideration the precise syntactic-semantic relations of each unit to the other units. It means, as a general rule, that the verb form at the end of every such unit has-at this particular stage-to be translated in its 'pure' lexical meaning, as if it were devoid of any markers of morphological categories whatsoever. At this stage, the student is strongly advised to make written notes of these preliminary translations.
4) The fourth and final step consists in 'binding' the translations of the predicative units together. This stage involves the careful and precise application of the student's knowledge of Amharic morphosyntax; thus, the entire methodology proposed in the present paper is practicable only at advanced levels of study.

Reliance on Predicative Units: Difficult Cases
Before we proceed to exercises in applying the method outlined above, we must examine a number of cases which might present difficulties in applying this method. These cases are (1) the predicate expressed by several coordinated verb forms; (2) the cleft sentence; (3) the predicate expressed by a relative verb form having a head noun; (4) the predicate expressed by a relative verb form having a head noun, the relative form being obscured by a preposition; (5) the predicate expressed by an analytical verbal construction; (6) the predicate expressed by an infinitival construction with a compound conjunction structurally analogous to a compound adposition; (7) the infinitive standing alone, without any accompanying morphological markers, or with the preposition lä-; (8) the predicate expressed by a finite verb form not at the end of a sentence, but followed by a verb of speaking in a finite or non-finite form; (9) the (apparent) non-finite predicate expressed by a converb whose meaning is highly lexicalized so that this converb performs the function of an adverbial modifier of its main verb.

The Predicate Expressed by Several Coordinated Verb Forms
Whenever a student sees a verb form with the conjunction -nna not in sentence-final position, 11 he/she should start looking for another (or possibly several other) verb forms whose morphological make-up is analogous to the one with the conjunction. 12 These will be coordinated verb forms constituting one single predicative unit, the right boundary of which coincides with the rightmost of these coordinated predicates.

The Cleft Sentence
The student should learn to recognize the three constitutive elements of a cleft as well as their alternative orders. The constituents of the cleft sentence are (1) relative clause-focus-copula, or (2) focus-copula-relative clause. 13 If the student is to recognize the relative clause in both orders, he/she should notice that the relative verb form at its end does not have any head noun immediately to its right: in the latter order (2) the relative verb simply stands at the very end of a sentence, while in the former order (1) there is imperfective and only the sentence-final predicate will be in the compound imperfe ', 'the guests are eating, drinking, singing, and dancing'. 13 There is another peculiarity concerning word order in cleft sentences: the relative clause can be split by the focal part of the utterance and the copula, e.g. 'sǝlä hǝywät tar^ku bäṭam tǝnnǝšnäw yämmǝnnawqäw', 'there is very little that we know about his biography' (the relative clause in bold). However, this type of clause order is comparatively rare, both in oral speech and in writing.
usually no nominal part of speech in its immediate right-hand vicinity. 14 Having discovered this, the student should realize that, in all likelihood, it is a cleft sentence. If, going further toward the end of the sentence, he/she notices any copular form, whether free or conjunct, his/her initial suspicion is confirmed. Once the cleft sentence has been recognized, the translation must highlight the focal part, retaining it as the communicative centre of the sentence.

The Predicate Expressed by a Relative Verb Form with a Head Noun
When a relative verb has a head noun-any nominal phrase (NP), simple or compound-we have just another type of subordinate clause, and the general rules of delimiting predicative units laid out above also apply here. However, in this case, the student should be especially observant, as in most European languages relative clauses are generally placed to the right of their antecedents. The method here consists, firstly, in delimiting the boundaries of, and translating, the head NP and, secondly, in translating the relative clause situated to the left of, and headed by, the relative verb form, thereafter, placing the resulting translation to the right of the translation of the head NP.

The Relative Form being Obscured by a Preposition
The concatenation of a relative clause and a head noun can become even more difficult to single out and translate in cases where the relative prefix in a perfective or imperfective relative form is obscured by a preposition, as in the following example: bäräṭṭäbä ärq ṭärä ezawǝn ṭärrägäčč 'She wiped the dining table with a wet (lit. 'that-which-is-wet') piece of cloth.' In my opinion, the most important point here is to remember the grammatical rule according to which the relative morpheme is deleted (fully in the case of a perfective relative or partially in the case of an imperfective relative), becoming a phonological zero if immediately preceded by any preposition. This done, the analysis that comes next will take into account the nominal antecedent of the relative verb, and translate the relative clause plus its antecedent, as in § 3.3.

The Predicate Expressed by an Analytical Verbal Construction
Before embarking on analysing a sentence according to the method described, the student should learn the inventory of the Amharic analytical verbal constructions, their formations, and semantic properties; otherwise he/she could easily mistake such a construction for a concatenation of two separate predicates, each of them constituting a predicative unit of its own. This would inevitably result in an incorrect translation. However, it is also possible and even advisable to introduce sentences containing this feature at relatively early stages, if the teacher is sure that the sentences offered contain no analytical constructions with which the students are as yet unfamiliar.

Infinitival Construction with a Compound Conjunction
The student should learn to distinguish between adpositions (combining with nominal parts of speech) and homonymous conjunctions (combining with verbal forms). Here are two examples highlighting this contrast, with a combination of a preposition and a postposition in (1) and with a homonymous conjunction in (2): 1) käsäʾat bäh w ala wädä sinima bet ǝnnǝhedallän 'In the afternoon we will go to the cinema.' 2) mǝsa käbälla bäh w ala tǝnnǝšǝräft fällägä 'After eating lunch he wanted [to take] a little rest.' Adpositional combinations do not signal the right boundary of a predicative unit, whereas conjunctional combinations do.

The Infinitive Standing Alone or with the Preposition lä-
The infinitive used alone (and not forming a part of any analytical verbal construction, see § 3.5) should not be considered as a separate predicate when it plays the role of subject in its clause. Compare the following example: ǝndäzzih männagär ṭru aydällämm 'It is not good to speak like this.'

The Predicate Expressed by a Finite Verb Form Followed by a Verb of Speaking in a Finite or Non-Finite Form
If a finite verb form is found not at the end of a sentence, but is directly followed by a verb of saying alä in a non-finite form followed directly by a verb of speaking, such a sequence signals direct speech, that is, a clause whose predicate is represented by the sentence-medial finite verb form, for instance: ǝrsu nägä almäṭamm bǝlo mälläsällǝňň 'He answered me, saying: "I won't come tomorrow".' In the above example the sentence-medial finite verb form is in bold letters while the verb of speaking following it is underlined.

The Non-Finite Predicate Expressed by a Converb functioning as an Adverbial Modifier
In Amharic we quite often encounter forms which, being converbal by their origin, have become highly lexicalized in today's language. It means that they have ceased performing the function of a predicate in a dependent clause and have been transformed into pure adverbs, modifying in one way or another the event designated by the verb coming after them. Such 'converbal adverbs' are, for instance, dägmo, 'again', 'also', 'on the other hand'; fäṣṣǝmo, 'absolutely', 'at all'; ablǝṭo, 'more', 'best'; qäṭṭǝlo(mm), 'next', 'afterwards', 'later on'. Most of them can either retain full personal agreement with the subjects of their main verbs or occur in a frozen form, meaning that they always remain in the third person masculine singular.
The numerous instances of the copula or its contextual analogues in the function termed 'les verbes redondants' 15 by Olga Kapeliuk also belong here. These redundant verbs perform the function of connecting a noun or a verb with its qualifiers, for example: 1) nǝṣuh yähonä säw 'an innocent person' 15 For further details on this type of predication see Kapeliuk 1983. Generally speaking, the Amharic language has a strong tendency to express overtly a number of syntactic relations which, in most European languages, usually receive zero expression at the surface syntactic level. Any teacher of Amharic should be thoroughly familiar with this tendency and be able to explain it clearly to the student.
When dealing with similar forms it is worthwhile bearing in mind that a literal translation, however clumsy it may sound in the target language, may well prove helpful in understanding the meaning of the whole nominal or verbal phrase. Thus, the above examples can be translated word for word as, respectively, 'innocent who-is person', 'of-qǝne teacher he-having-become he-graduated', and 'well I-doing I-know-him'. The redundant nature of such predicates becomes clear in the literal translations, and the student can omit them in the finalized version of his/her work.

Conclusion
The method outlined above has, in my opinion, the advantage of leaving purely formal considerations to one side, and of being clear to the student even when he/she is not very advanced; it gives them the possibility of, firstly, figuring out the semantics of the individual sentence units and, finally, the meaning of the sentence as a whole. It should be emphasized that the proposed method requires that the student be thoroughly acquainted with the nominal and (especially) verbal inflectional morphology of Amharic. He/she should also possess a sound knowledge of the inventory of Amharic analytical verbal constructions, both with respect to their structural properties and their semantics. However, the teacher should feel free to give any sentence containing material that is already known to the student at a particular stage of learning; in this way, it will become possible to start applying this method at a relatively early stage of studying Amharic.
It goes without saying that the method discussed in the present article is not applicable when the translation is performed orally, be it synchronous or consecutive translation. For these kinds of translation, the interpreter should be sufficiently well-trained in mentally grasping comparatively large blocks of information and retaining them in his/her memory until the process of oral translation of a given information unit is completed. 16 Therefore, this method should be reserved for teaching students how to analyse and translate Amharic written texts either in class or at home.

Exercise 1
For a first acquaintance with the method proposed above, let us take a very easy complex sentence: gǝddǝbu yämängǝst ǝqqǝd bämähonu sǝraw yǝqäṭṭǝlall 'Since [the construction of] the dam is the plan of the government, the work will continue.' This sentence is made up of only two clauses: the main one, headed by the sentence-final form yǝqäṭṭǝlall, and the dependent one, headed by the form bämähonu. Consequently we conclude that the boundary between the dependent and the main clause is to be drawn between bämähonu and sǝraw. Once this conclusion has been arrived at, it becomes largely a matter of individual choice for the student as to which of the clauses to translate first. In such a relatively simple and short sentence I personally prefer to begin translation with the main clause ('sǝraw yǝqäṭṭǝlall'), the more so, inasmuch as, in this particular case, this clause consists of only the subject and the predicate. The only difficulty that might arise here is the choice of the precise meaning of the verb. The verb qäṭṭälä, as can be observed from its dictionary entry, has both a transitive and an intransitive meaning. 17 This ambiguity can be avoided when one realizes that the noun preceding it is used with the definite article, but without the direct object enclitic -n; thus, it cannot be the object in the sentence, it can only be the subject of the verb. Having found the lexical meaning of the noun sǝra, we now are able to rule out the possibility of the transitive meaning for yǝqäṭṭǝlall, as 'work' cannot be conceived as a volitional agent performing an action. Furthermore, the English verb 'continue', like qäṭṭälä, is both transitive and intransitive, which makes it possible and expedient to translate the main clause here as 'the work will continue'.
Proceeding to the subordinate clause ('gǝddǝbu yämängǝst ǝqqǝd bämähonu'), we notice the clause-final verb form bämähonu. Here the student will conclude that this is a form of the verb honä with its two main lexical meanings: 'to become' and 'to be'. Let us remember that at this stage we should ignore the precise grammatical form of the verb; by default, we may translate this form simply as 'is' or 'becomes'. Having found the rest of the words in a dictionary, we get the following rough translation: 'The dam the government's plan is/becomes'. Now we can attempt to connect the two clauses, using our knowledge of Amharic verb morphology: the student should be aware that the combination of the preposition bä-with the infinitive followed by a possessive marker serves as one of the standard ways of expressing the causal semantic relation. Consequently, this causal clause should be translated as 'Since/as the dam (evidently, 'the construction of the dam') is the plan of the government'; then we join the translation of the main clause: 'the work will continue'.

Exercise 2
For our next exercise we shall look at a cleft sentence. When considering the structure of this sentence, the student should pay special attention to the relative verb form yastawwäqut. He/she should then see that the word immediately following this relative form is not a noun, but a preposition. This indicates that the relative verb here is not used in the attributive function. Following this, the student should be able to recognize the form of the copula at the very end of the sentence. By now he/she has singled out all three canonical constituents of a cleft, namely (1) the relative clause headed by the relative verb form, (2) the copula, and (3) the focal part of the sentence in between the relative verb and the copula, that is, the part which the speaker wants the listener to pay special attention to. 18 In contrast to Amharic, cleft sentences are not frequent in most European languages. The pragmatic connotations borne by clefts in such languages as Amharic are usually expressed in European languages by other means, most notably through word order, whereby what was the focal part in the Amharic sentence tends to be placed at the end of the translated sentence in the target language. Therefore the first step in translating the above sentence is the identification of the constituents, (1) 'ministru yǝhǝn yastawwäqut' and (2) 'näw', as outlined above, and bringing them together, as this constitutes the thematic part of the utterance. As a result of this first step we get the following translation: 'The minister communicated this'.
Next we can see that the focal part of the cleft is fairly long ('wädä sǝlṭan kämäṭṭu lämäǧämmäriya gize läʾagär wǝsṭ gazeṭäňňočč maksäňňo ṭǝrr 15 qän 2010 amätä mǝhrät bäṣǝhfät betaččäw gazeṭawi mäglä äňňočč maksäňňo a sisäṭu'). In fact, clefts with such prolonged focal parts are quite frequent in Amharic, especially in the written style. At this point the student's task is the analysis of the internal structure of the focal part. The analysis should be aimed first and foremost at singling out verbal forms capable of functioning as predicates. There are two such forms here: kämäṭṭu and sisäṭu. The form made up with the conjunction kä-and the perfect of the verb, mäṭṭu, can have more than one meaning: it can be understood as (1) a conditional subordinate form and as (2) a temporal subordinate form with the meaning 'since (a certain period of time)'. Consequently, the phrase 'wädä sǝlṭan kämäṭṭu' may be translated either as 'if he (hon.) came to power' or as 'since he (hon.) came to power'.
At this stage, the student is unable to choose between these two alternatives. To enable himself/herself to do that, he/she should proceed to translate the remaining part of the focus. At the very end of the focal part is the verb form sisäṭu; recognizing this, the student should conclude that this form has a temporal meaning usually rendered in English by the conjunction 'when'. Now the student can attempt to give a preliminary, word-forword translation of the focal part: 'to power since he (hon.) came for the first time to within-the-country journalists on Tuesday, the 15 Ṭǝrr, in 2010 Year of Mercy, in his (hon.) secretariat press release when he (hon.) gave'. Now we come to the concluding stage of our analysis: an acceptable translation which effectively communicates the focal part of the Amharic original. In this rendering, the thematic part consisting of the relative clause plus the copula should come first and the focal part should be placed toward the end of the sentence. Thus, we get the following translation: 'The minister communicated this when, for the first time since he had come to power, he gave a press release to local (as is evident from the context, 'Ethiopian') journalists on the 15 Ṭǝrr, in 2010 Year of Mercy'.

Exercise 3
For our third and concluding exercise let us select a prolonged sentence with a number of dependent clauses of various types. Such structural complexity is very common in written Amharic, both in the press and in fiction. The following sentence is taken from the opening passage of the novel Fǝqr ǝskä mäqabǝr by Haddis Alämayyähu.
We can see that the sentence consists of one main clause and no less than twelve dependent clauses of varying length. But there is more to it than that: most of the dependent clauses are dependent not directly on the main clause but on other dependent clauses.
Nonetheless, using the approach proposed in the present article we can arrive at an adequate translation of the sentence without paying too much attention to the complexity of its syntax. After all, this is an exercise in translation, and the student is not expected to draw syntactic trees of sentences he/she is seeing for the first time, and is not yet aware of their meaning. Our first task will be to delimit elementary predicative units within the sentence under analysis. The result of this operation will be as follows (boundaries between the units are indicated by double slashes): bäzzih huneta ǝkkul ǝdmeyaččäwǝn kasalläfunna // yämǝšt magbat fǝllagotaččäw ǝyyätäqännäsä kähedä bäh w ala // bägulmasannätaččäw gulbätam wädiyaw ṭenamma bämähonaččäw // räddat sayfällǝgu // rasaččäwǝn rädtäw // lämänor bičǝlumm // mǝšt kalagäbbu // ǝdmeyaččäw ǝyyägäffa // aqmaččäw ǝyyätäqännäsä sihedu // räddat yämmiyaṭu mähonaččäwǝn // wädaǧoččaččäw aṭbǝqäw sǝlämäkkäruwaččäw // käzziyaw kämankusa wǝddǝnäšbäṭamu yämmibbalutǝn // set agäbbu.
A few explanatory words are necessary to justify delineating some of these boundaries.
In the units 'bäzzih huneta ǝkkul ǝdmeyaččäwǝn kasalläfunna // yämǝšt magbat fǝllagotaččäw ǝyyätäqännäsä kähedä bäh w ala', we should be able to identify the verb form kasalläfunna, with both the subordinating conjunction k(ä)-and the coordinating conjunction -nna. It means that the form is not only subordinated by some other verb form, but also coordinated with yet another verb form. It would be natural if we presumed that such a coordinated form should likewise contain the same subordinating conjunction. Thus, we might detect the form kähedä a little below. Immediately after this second coordinated form we notice the word bäh w ala. At this stage we can already rule out the possibility of interpreting both these forms containing the conjunction k(ä)-as conditional, interpreting them instead as temporal with the meaning of 'after (some event)'. However, we also notice that the two coordinated forms have a different subject agreement, the former standing in the third honorific, while the latter is used in the third masculine singular. It means that these verb forms, although coordinated, have different subjects. The subject of the former form is not overtly expressed; covertly, it denotes the protagonist of the story who is consistently referred to with honorific forms. The subject of the latter form is overt-it is the noun fǝllagot(aččäw) which is modified by an infinitival attribute yämǝšt magbat.
The phrase 'yämǝšt magbat fǝllagotaččäw' can clearly be understood as meaning 'his (hon.) wish to marry a wife'. Here the infinitive standing alone, without any possessive suffixes, case markers, or prepositions but having a genitival attribute of its own, is to be analysed as performing the role of an attribute to the head noun fǝllagotaččäw in the clause whose left-hand boundary is constituted by the very beginning of the whole sentence, while the right-hand boundary is drawn after the predicate 'ǝyyätäqännäsä kähedä bäh w ala'. The student should already know that the analytical construction 'ǝyyätäqännäsä hedä' indicates the gradual development of a situation and is best translated into English with the help of such adverbs as 'gradually', 'by and by'.
By now we can offer a preliminarily translation of this part of the sentence as 'in such a fashion, after he had spent half of his lifetime and after his wish to marry [a wife] had gradually diminished'. The next predicative unit ('bägulmasannätaččäw gulbätam wädiyaw ṭenamma bämähonaččäw') can be roughly translated as 'to be/become strong and also healthy in his maturity'. Knowing that the construction consisting of the preposition bä-, the infinitive, and a possessive suffix has causal meaning, we can translate the phrase as 'because he was/became strong and also healthy in his maturity'.
The unit coming after ('räddat sayfällǝgu') has at its head the verb form sayfällǝgu. This particular form (conjunction s(ǝ)-+ negative simple imperfective) is known to perform the function of the negative correlate to the suffixal converb. Therefore, we can translate this small unit as 'without wanting/seeking a helpmate'.
In the following unit ('rasaččäwǝn rädtäw'), we should take into account the fact that the word ras, besides its principal meaning 'head', in combination with a possessive pronominal suffix with or without -n can function as the reflexive pronoun. Thus we arrive at the translation 'helping himself'.
Next, we can observe two adjacent verb forms: 'lämänor bičǝlumm'. The first of them is an infinitive with the preposition lä-while the second is a conjunctional form with the concessive meaning provided by the circumfix 'b(ǝ)-…-(ǝ)mm'. As čalä is a modal verb which can govern the infinitive of another verb with or without the preposition lä-, 19 the two verbal forms should be understood as a single complex predicate and translated together as 'although he could live'.
The following unit ('mǝšt kalagäbbu') can easily be translated as 'if he did not marry [a wife]', because the other conjunctional meaning of k(ä)-, 'since (a definite) time', is safely ruled out here.
Moving further toward the end of the sentence ('ǝdmeyaččäw ǝyyägäffa // aqmaččäw ǝyyätäqännäsä sihedu'), the student can observe two verb forms with an identical subject: ǝyyägäffa and ǝyyätäqännäsä. Furthermore, he/she can also notice that the second of these forms is immediately followed by a temporal conjunctional form of the verb hedä. The student should be able to recognize the analytical verbal construction ǝyyä + perfect of the lexical verb + auxiliary hedä indicating the gradual development of the action designated by the lexical verb. By this time he/she is able to translate these two units containing coordinated predicates: 'when he gradually becomes advanced in years and when his strength gradually diminishes'.
The two units that come next ('räddat yämmiyaṭu mähonaččäwǝn // wädaǧoččaččäw aṭbǝqäw sǝlämäkkäruwaččäw') present certain special difficulties. The student should be able to discern here, in the first of the units under discussion, another analytical verbal construction consisting of the relative form of the lexical verb and the copula, the latter capable of occurring in any form required by the context. Accordingly, the copula occurs here in the form of the infinitive accompanied by a possessive suffix and the accusative marker. This form of the copula is preconditioned by its matrix verb sǝlämäkkäruwaččäw, 'because they advised him', in the second of the units. Consequently, in the predicative unit under consideration, one 19 There is an alternative option for čalä: it can also govern the l + imperfective form of a dependent verb. There are subtle but important semantic differences between the three alternative government possibilities for čalä; however, in the present context we do not intend to discuss them.
should discern the expression 'räddat aṭṭa', 'to lack a helper', 'to be in need of a helper'. At this stage the student can attempt to translate the two predicative units together. It is also assumed that he/she is familiar with the form aṭbǝqäw, which, although clearly converbal in its origin, no longer constitutes a separate predicate, but rather modifies the action designated by the form sǝlämäkkäruwaččäw. The translation will look something like 'since his friends strongly advised him that he would be without a helper'. This translation looks clumsy because in English the verb 'to advise' mostly presupposes by default that the content of the advice is positive. Therefore we could slightly modify our translation: ' since his friends cautioned him against remaining without a helper'.
The next non-finite predicate, which is also the last non-finite form in the whole sentence, is a relative form of the verb täbalä, 'to be said', 'to be named'. As the words preceding it are clearly a personal name, we can attempt to translate the remaining portion of the sentence as a whole, including the finite predicate at its end ('käzziyaw kämankusa wǝddǝnäšbäṭamu yämmibbalutǝn // set agäbbu'). The translation should be as follows: 'there, in Mankusa, he married a woman named Wǝddǝnäš Bäṭamu'. Now it is the turn of the student to put together the preliminary translations of all the predicative units. The resulting translation is quite rough and will need to be refined.
'In such a fashion, after he had spent half of his lifetime and after his wish to marry [a wife] had gradually diminished; because he was/became strong and also healthy in his maturity; without wanting/seeking a helpmate; helping himself; although he could live; if he did not marry [a wife]; when he gradually becomes advanced in years and when his strength gradually diminishes; that he should not remain without a helper; since his friends cautioned him; there, in Mankusa, he married a woman named Wǝddǝnäš Bäṭamu.' Now, for the meaning of the sentence to become transparent and for the translation to acquire a literary shape, the student should remember that Amharic is a rigidly left-branching language, that is, that nearly all dependent entities in this language are preposed to the entities governing them. A second thing well worth considering is that, while Amharic clearly prefers long periods containing a lot of embedded clauses, often with more than one level of embedding, English, as well as most other European languages, shows a tendency to evade such long and complex periods and to break them up into smaller independent sentences. It is at this stage that the student can attempt to establish the syntactic links between the Amharic claus-es and to connect the clauses in English (in fact, at the same time, disconnecting some of the clauses so that the resulting translation conforms to the stylistic norms of the target language).
'He spent half of his life in this way and his wish to marry gradually diminished. He could live without seeking a helpmate, helping himself because he was strong and healthy in his maturity. However, his friends convinced him that if he did not marry he would remain without a helpmate when he gradually became advanced in years and when his strength diminished. So he married a woman named Wǝddǝnäš Bäṭamu, from the same locality (as the one in which he lived), from Mankusa.' It is important that the student working on a literary translation be allowed to freely break up a large sentence like the one under analysis and construct several smaller sentences in the target language. The student should also be encouraged to find special words connecting the resulting sentences in the target language, such as the conjunction 'however' in the example analysed. Then, when the meaning of the Amharic sentence is completely cleared up, some of these clause connectors will inevitably be removed so that several shorter sentences will appear in place of the longer period in Amharic.