Information structuring in Yoruba(*).
As in many other languages, the focus particle ni in Yoruba is derived from the copula ni. The basis for both junctions of ni is a preconstructed domain, that is, a presupposed set of items out of which the speaker exhaustively selects the one (or more) that she assumes to be relevant in a given situation. The copula ni is used only in contexts to which the preconstructed domain applies. Thus, its functional range cannot be covered by the framework presented by Hengeveld (1992). The focus function of ni can only be adequately described in terms of identificational focus, because this focus type integrates the concept of the preconstructed domain. Focus theories such as Lambrecht (1994), Vallduvi and Engdahl (1996), and the thetic/ categorical distinction fail to adequately cover the functional range of Yoruba ni, because they concentrate on information focus and neglect identificational focus as recently discussed by Kiss (1998). However, the functional range of Yoruba ni goes beyond identificational focus in English and Hungarian (cf. Kiss 1998) inasmuch as its use is not limited to noun phrases. It also marks verbs and clauses. This can possibly be explained by the pervasiveness of the concept of the preconstructed domain, that is, by the fact that preconstruction is crucial not only for the focus construction but also for the copula construction with ni.
Information structure or information packaging has recently become the subject of many publications in the functional as well as the formal framework of linguistics. The aim of the present paper is to discuss information packaging in Yoruba within the context of copula and focus constructions. The idea of the preconstructed domain, that is, a domain of concepts that are presupposed by the participants in a dialogue, appears to be a key notion for a proper understanding of how information is structured in Yoruba. A fundamental distinction seems to be made between utterances that involve such a preconstructed domain and those that do not. The former are marked with the particle ni, while for the latter the verb is typically preceded by the high tone syllable. For a domain to be preconstructed, it does not need to be mentioned linguistically in previous discourse; it can also be presupposed due to specific situational knowledge or even to general, maybe culturally determined, background knowledge. As will be seen below, a preconstructed domain is the necessary condition for statements involving identification. The preconstructed domain is an explicitly or implicitly presupposed set of entities or states of affairs from which the speaker selects the ones that are relevant for her/his utterance.
It is a well-known fact that the copula is used in focus constructions in many languages. The different stages of grammaticalization taking place from copula to focus marker have been described by several linguists, such as Heine and Reh (1984: 181-182). What makes Yoruba particularly interesting in this context is the fact that the notion of preconstruction is also present at the level of the two copula verbs ni and je. If there is a preconstructed domain, ni has to be used; otherwise, the copula je is used. As we shall try to show, the fact that preconstruction is relevant even at the level of ni in its copula function yields important consequences for ni in the focus-marker function. As is to be expected from Kiss (1998), ni is used to mark identificational focus (Kiss 1998) in the focus position at the beginning of a sentence. This is true if the concept in the focus position is an entity. In contrast to English or Hungarian, Yoruba also allows states of affairs and predicates to occur in the focus position marked by ni. In these cases, the focus marker keeps its function of reference to a preconstructed domain, but it no longer implies identification and exhaustiveness to the same extent with all the different types of focus construction. Thus, the focus position in Yoruba is not limited to identificational focus as one would expect from reading Kiss (1998). This and the occurrence of ni with predicates and states of affairs may be due to the fact that preconstruction is crucial for ni in the copula function.
In the following section (section 2) we shall analyze the functions of the copula verbs ni and je. The classification of copula verbs proposed by Hengeveld (1992) is examined in light of data from Yoruba, and it will be seen that this classification is limited in that it does not account for the functional difference between the two copula verbs in Yoruba. Section 3 presents our account of focus constructions in Yoruba. The four types of Yoruba focus constructions will be discussed. We will try to show that their function is derived directly from the copular function of ni. Moreover, we shall introduce Kiss's (1998) account of identificational focus in Hungarian and English in this section. In section 4 we shall compare the focus system of Yoruba with the following other pragmatic theories dealing with focus: Dik's (1997) six types of focus, the thetic/categorical distinction (Kuroda 1972; Sasse 1987, 1995), information packaging discussed by Vallduvi and Engdahl (1996), and information structure as presented by Lambrecht (1994). As we shall try to show, Dik's (1997) approach is the only one with which the functional range of Yoruba ni can be adequately covered. The conclusion reached in section 5 is that a complete account of copula and focus constructions in Yoruba cannot be provided without recognizing the centrality of the notion of the preconstructed domain, as this article will attempt to demonstrate.
2. The function of the copula ni in contrast with je
We examine in this section utterances in Yoruba involving the copula verbs ni and je. As we shall see in what follows, these markers (along with the high tone syllable, which is used in conjunction with je) play an important role in the way that information is structured in the language, for their correct use depends on perceptions held by the participants in a speech situation regarding the different elements of their discourse -- that is, the entities, the events, and the states of affairs referred to. The difference between ni and je will be seen to be a question of PRECONSTRUCTION, that is, the existence of a domain of concepts presupposed by the speech-act participants. Where a certain domain is preconstructed the copula ni must be used, and where such is not the case, then je is employed. The discussion that follows in the next section on the function of ni as a focus marker will be seen to derive directly from its role as a copula.
The structure we are concerned with here can be represented as follows:
(1) X `BE' Y (where X, Y = NP).
This is an ambiguous structure as far as the Yoruba language goes, for it will be realized in two different ways depending on whether the speech-act participants presuppose a certain domain in the context of which the utterance is made. If a domain of concepts is presupposed and the copula ni is used, then X represents the predicate NP while Y is the argument NP. If je is used, in instances not involving presupposition, the roles are reversed: X is the argument NP while Y is the predicate NP. Example
(2) below involves the preconstruction of the domain "king of animals" whereas (3) does not involve any preconstruction.
(2) kiniun ni oba eranko. lion be king animal `The lion is the king of animal.' (3) kiniun je eranko nla. lion be animal big `The lion is a big animal.'
This notion of preconstruction seems to be a very important one in Yoruba, for it appears to be one of the primary conditions that determine the kinds of utterance that are possible within given contexts. This same notion was seen to be important for adequately accounting for the function of the perfect marker ti in Yoruba (Bisang and Sonaiya 1997). Reference to a preconstructed domain (RPD) explains the use of ti in this temporal function as perfect marker, for it signals that there is shared knowledge between the interlocutors in a speech situation that a particular event is "preconstructed," that is, it possesses a high potential of being realized. The marker ti allows the location in time of a preconstructed event with respect to a particular reference moment (typically the moment of speech, unless otherwise specified). Thus, in (4) below, the fact of Ayo's coming is held by both speaker and addressee to be potentially realizable, and what the question does is to request information as to whether this preconstructed event has indeed been realized as of that particular moment in time.
(4) se Ayo ti wa? Q Ayo HTS RPD come `Has Ayo come?'(1)
As such, this question would not be appropriate in a context where neither speech-act participant (or only one of them) is aware that Ayo is supposed to come. Furthermore, this same notion accounts for other, more pragmatic, uses of the marker ti, especially in the context of wh-questions and responses to them.
Within the context of ni, reference to a preconstructed domain functions in a somewhat different manner. In the case of the copula, the preconstruction has to do with a relevant, clearly delimited domain, recognized by the interlocutors as the appropriate one with respect to which the information supplied by the predicate NP is being given. While ti is used for the temporal location of preconstructed events, the copula ni may be seen as a marker allowing the exhaustive identification of some particular entity (or entities) that are pertinent to the speech situation. Thus, in (2) above, the domain is clearly specified: the animal that may be regarded as "the king of animals." It is with regard to that domain that the choice of "lion" is made. This reference is not always so explicitly made as in (2), where it is given in the utterance itself; it may well be that the domain of preconstruction is available within the larger discourse context. For example, in a context in which the topic of professions is already under discussion, (5) will be understood as being made in respect of Ayo's profession:
(5) tisa ni Ayo. teacher be Ayo `Teacher, that is what Ayo is.'
Again, the point to be underscored is that such an utterance cannot be made if there is no prior knowledge, shared by the interlocutors, of the relevant domain to which the utterance applies. It is this shared knowledge that constitutes the common point between the PRD in respect of ti and in respect of ni. In other words, there is the existence of certain assumptions or presuppositions that go beyond the illocutionary character of the utterance itself.
A major difference between these two types of preconstruction, however, and a crucial condition for the felicitous use of the particle ni as both a copula and a focus marker, is the fact of CHOICE that is implied in its use. Once a domain is preconstructed, then a choice has to be made from among the members of that domain. To return to example (5) above, what is implied is that a selection is made by the speaker from the domain of professions (lawyer, teacher, engineer, etc.), and `teacher' is identified as the one that rightly qualifies Ayo. This is the reason that the feature of [exhaustiveness] is normally associated with this kind of construction. To select one (or more) members from the given domain is to do so exhaustively, cancelling out the existence of any alternative to the choice that has been made.
To preview an aspect of the analysis that will be presented in the discussion of ni as a focus marker, the three important features that are required for the proper use of the copula ni can be seen as [+preconstruction], [+identification], and [+exhaustiveness]. In other words, the basic condition without which ni could not be acceptably employed is that there be preconstruction.
However, as far as its use as a copula is concerned, the other two features are entailed by this preconstruction: once there is preconstruction, then there is identification (i.e. the selection of a particular entity from the group of potential candidates), and this identification is necessarily exhaustive in nature. As we shall see, this type of entailment is not involved in all cases of focus ni.
Example (3), with the copula je, does not involve any preconstruction, neither is any selection made from among a set of potential arguments that may fit the predicate NP slot. The copula je is used, rather, to make out-of-the-blue utterances, which do not presuppose any particular state of affairs about which the interlocutors share some knowledge. It is to be noted that the utterance would still not be correct with je but without the qualifier tiki `big':
(5') *kiniun je eranko. lion be animal
This means that in Yoruba, one cannot make an out-of-the-blue declaration that is in effect an identification or a kind of categorization, since identification is always specified with regard to a preconstructed domain for which the entity in the predicate NP exhaustively holds (also cf. examples  and  on definitions and classifications into genus types). The only way an out-of-the-blue declaration could be acceptable is for a qualification to be made, as in (3). With this utterance, the speaker is talking about the lion and simply states that bigness is one of the qualities that this animal possesses. However, if bigness were to be expressed within the context of preconstruction (i.e. the size of animals is what is being discussed), then we would have a construction with ni:
(3') eranko nla ni kiniun. animal big be lion `The lion is a big animal.'
We see therefore, that the sentence `The lion is a big animal' does have two meanings in Yoruba (notice that this is a gloss provided for example  as well). If the sentence is uttered out-of-the-blue, with the speaker simply desiring to say something about the lion, then the construction with je is used. However, if there is preconstruction regarding, for example, size, then ni is employed, as in (3') above.
The fact that identification is always specified with regard to a preconstructed domain does not only explain why (5') is impossible, it also accounts for the inacceptability of the inversion of (5), presented here as (6):
(6) *Ayo ni tisa. Ayo be teacher
Example (6) implies that being a teacher is exhaustively true of Ayo. This is impossible, since there are certainly many more teachers in the world. If the NP in the argument position implies a set that can be exhaustively delimited by the predicate NP, the construction with ni becomes compulsory. This is the case if the NP in the argument position contains a superlative as in example (6') or a restrictive relative clause as in example (6"):
(6') Ayo ni/*je tisa to dara ju. Ayo be teacher REL:3s good most `Ayo is the best teacher.' (6") Ayo ni/*je tisa to na Tolu. Ayo be teacher REL:3s beat Tolu `Ayo is the teacher who beat Tolu.'
Thus, in just the same way that the presence of ti in an utterance signals the existence of some state of affairs that needs to be taken into consideration, so also the use of ni clearly involves reference to some preconstructed domain that must be taken into account: either it is already somehow existent within the speech context or it is simultaneously introduced with the utterance. It is within the context of this domain that the set of possible NPs that may complete the argument NP is to be found. In other words, it is within this domain that the appropriate member of the set must be selected or identified. Since the use of ni involves making a choice, the domain serves to determine the space within which the choice is to be made.
Hengeveld (1992) presents a functional classification of copula verbs that is supposed to be valid for the generality of human languages. In what follows we shall briefly examine Hengeveld's proposal in the light of facts from Yoruba. It will be shown that the classification proposed by Hengeveld does not adequately account for the Yoruba data in that it fails to offer an explanation for the existence of two copula forms in the language, and it also does not account for the way in which they are functionally differentiated. The copula ni is used in all the four types of construction Hengeveld proposes. What this suggests is that the functional difference between ni and je is based on another kind of distinction, that is, the presence or absence: of preconstruction.
Hengeveld (1992) identifies two major types of function for copula verbs that are term predicates: they may be used either for identification or for classification. Each of these two types, it is claimed, can further be described as being either specifying or characterizing in its function. He thus proposes Table 1, and the four different structures possible can be illustrated with the sentences given below (Hengeveld 1992: 84):
(7) The capital of France is Paris.
(8) Paris is the capital of France.
(9) A bachelor is an unmarried man.
(10) A cat is an animal.
Table 1. The four major types of functions expressed by copula verbs (Hengeveld 1992: 84)
Specification Characterization Identification (7) (8) Classification (9) (10)
According to Hengeveld (1992), the predicate term is definite in cases of identification while it is indefinite in classification, since the latter deals with class membership. The difference between specification and characterization is explained in terms of the question that each seeks to answer: for the former, the question to be answered is "Who/What is X?", while the latter asks the question "What can you tell me about X?" In cases of specification, therefore, there is complete overlap or identity between the referent set of the argument term and that of the predicate term, and this fact is captured by the notion of exhaustiveness. Instances involving characterization, on the other hand, do not possess this property of exhaustiveness.
With respect to the distinction made between identification and classification in terms of definiteness or indefiniteness of the term predicate, it is not certain that this is an important distinction to be made for Yoruba, since no formal difference is seen in sentences like (2) kiniun ni oba eranko `The lion is the king of all animals' (where the predicate term is supposed to be definite) and a sentence like eranko ni kiniun `The lion is an animal'.
With respect to identification, it is assumed that languages potentially make a distinction between utterances involving specification and characterization. While (7) is a response to the question `What is the capital of France?', (8) seeks to answer the question `What can you tell me about Paris?', and these two questions may be answered in different ways, formally. In Yoruba, however, the difference between these two questions is not reflected in the answers given to them. In both cases, the following answer is acceptable:(2)
(11) Paris ni olu-ilu ile Faranse. Paris be head_town country France `Paris is the capital of France.'
However, what is important, as far as Yoruba goes, is the fact that both of these questions involve preconstruction, though of somewhat different types. To ask for the capital of a country is to request that one particular town be identified, from among the set comprising all the towns in that country. In this case, ni is obligatory, and we have identification in an exhaustive manner.
With regard to the second question, `What can you tell me about Paris?', it is the kind of answer that one gives that will determine whether ni or je is used. If the answer given is an identifying type, that is, if one says something that is true only of Paris, then ni must be used, because the utterance as such involves a preconstructed domain. Thus, if one responds that it is the capital of [trance, then one expresses this in exactly the same way as if one were responding to the question `What is the capital of France?'. In a similar manner, if one wishes to respond by saying that Paris is the biggest city in France, (12), or the most beautiful city in the world, (13), then ni is; still used:
(12) Paris ni ilu ti o tobi ju ni ile Faranse. Paris be city REL 3s be_big most LOC country France `Paris is the biggest city in France.' (13) Paris ni ilu ti o dara ju ni gbogbo aye. Paris be city REL 3s beautiful most LOC all world `Paris is the most beautiful city of the world.'
Superlatives are a clear case of identification: something is perceived as being true only for the entity identified.
If, on the other hand (still in response to the question `What can you tell me about Paris?'), one expresses something that is not true of Paris alone, then this is done with je as in example (14).
(14) Paris ' je ilu ti o dara pupo. Paris HTS be city REL 3s beautiful plenty/very `Paris is a very beautiful city.'
Here we do not have identification, for no selection is being made. Some information is merely being given regarding Paris that is not exclusive to it; there may well be many other beautiful cities in the world.
We are able to see from the foregoing that the distinction made by Hengeveld between identification-specification and identification-characterization does not apply to Yoruba, for identification is always specified. What is needed in the framework is how to represent instances of characterization that do not involve identification; that is, je, the copula used in the nonpreconstructed context, needs to find a place in this classification.
The other type of copula construction, described by Hengeveld (1992) as involving classification, is also not differentiated in Yoruba according to whether it is specifying or characterizing. In this instance, too, the focus particle ni can be used in both cases. The sentences `A bachelor is an unmarried man' (corresponding to  above) and `A cat is an animal' (corresponding to  above) can be translated as follows:
(15) apon ni okunrin ti ko tii gbeyawo. bachelor be man REL NEG not_yet carry_wife `A bachelor is an unmarried man.' (16) eranko ni ologbo. animal be cat `A cat is an animal.'
From all that has been discussed so far, it is obvious why ni is required in these instances as well: both cases involve exhaustive identification. Such is the nature of definitions, (15), and classification into genus types, (16).
It is at the level of what Hengeveld (1992) calls characterization that there are problems with Yoruba data. One could say that Yoruba characterizes in two different ways: with ni if the characterization involves preconstruction and is therefore identificational in nature, and with je if such is not the case. Just as is the case for the identification-characterization type, the classification-characterization copula construction also yields two possibilities. Thus, in respect of (16) above the question asked is, `What can you tell me about a cat?', and (16) is indeed one way of answering the question. The reason that ni is used here is that the speaker chooses to say something that involves preconstruction: he presupposes the domain of genus type and selects the one to which the cat belongs. If, however, what is said does not involve any selection, but some information is simply given that is true of the cat but may equally be true for some other animals, then je is used:
(17) ologbo ' je eranko ti o ni itiju. cat HTS be animal REL 3s have shyness `The cat is a shy animal.'
To summarize our discussion of the Yoruba copula forms ni and je, discussed in this section, it can be seen that ni can be used in all the four copula types discussed by Hengeveld, whereas the use of je is limited to characterization.
The criterion that accounts for this fact cannot be found within the parameters presented by Hengeveld. It is the notion of preconstruction that determines which of the two copulae has to be used. Since specification implies preconstruction, only ni can be used in this context. With characterization, however, the situation is different. It can occur in contexts involving preconstruction as well as in those that do not. For that reason, both copula forms are allowed.
3. The copula ni in the function of a focus marker
This section is devoted to the four types of focus construction in Yoruba and to a discussion of Kiss (1998).
The four constructions of focus marking in Yoruba can be defined by the position and the scope of ni. If ni occurs after an argument and has scope over this argument that is moved to the beginning of the clause we shall talk about the argument-final construction. A construction in which ni is argument-final but its function extends over the whole clause will be called argument-final construction with clausal scope. Constructions in which ni is put after the predicate and has scope over the predicate are predicate-final constructions. If ni occurs at the end of the clause that is at the same time its scope we are dealing with a clause-final construction. We shall discuss each of these four constructions individually in this section. As we shall see, the argument-final construction can be described in terms of Kiss's (1998) concept of identificational focus. This concept consists of the three features of [[+ or -] preconstruction], [[+ or -] identification], and [[+ or -] exhaustiveness]. Among the above four constructions, only the argument-final construction and the clause-final construction will turn out to encompass all three features, whereas the argument-final construction with clausal scope and the predicate-final construction can be characterized only by [+preconstruction]. As we shall try to argue at the end of this section, the presence of all three features depends on the extent to which a certain utterance in the focus position can be conceptualized as an entity.
In constructions with argument-final hi, the argument marked by ni occurs in the focus position that is the first position of the clause. This position is obligatory for wh-words:
(18) a. ki lo ra?(3) what FOC:2s buy `What did you buy?' (19) a. ta lo ra aso? who FOC:3s buy clothes `Who bought clothes?'
Both of the above questions can be answered in two ways, either with the noun phrase denoting the entity asked for by the wh-word in situ, (18b) and (19b), or in the focus position followed by ni, (18c) and (19c):
(18) b. mo ra aso. I buy clothes `I bought clothes.' c. aso ni mo ra. clothes FOCI buy `I bought clothes.' (19) b. Ayo ' ra asp. Ayo HTS buy clothes `Ayo bought clothes.' c. Ayo lo ra aso. Ayo FOC:3s buy clothes `Ayo bought clothes.'
The choice between (18b) and (19b) or (18c) and (19c) is determined by the situation. From contrasting these examples one can conclude that the focus particle ni is used to mark identificational focus as defined by Kiss (1998):
An identificational focus represents a subset of the set of contextually or situationally given elements for which the predicate phrase can potentially hold; it is identified as the exhaustive subset of this set for which the predicate phrase actually holds (Kiss 1998: 245).
If there is a preconstructed set of potential entities from which the speaker has to select the one (or more) entities that actually hold(s) for a particular context, the construction with argument-final ni will be used. Thus aso `clothes' in (18c) and Ayo in (19c) belong to a set of entities known by the speaker and the hearer. If there is no preconstruction, that is, if the speaker just provides the information asked for by supplying the appropriate entity, the unmarked construction with the focus in situ will be selected. Consequently, aso `clothes' in (18b) and Ayo in (19b) are not presupposed; the speaker simply introduces them to the hearer in order to fill her/his information gap. Since the unmarked construction does not imply exhaustiveness, it can also be used to build up an open list of entities that all provide a possible answer to the hearer's question:
(18) b'. mo ra aso, mo ra bata ... I buy clothes I buy shoes `I bought clothes, I bought shoes, etc.' (19) b'. Ayo ' ra aso, Olu naa' ra. Ayo HTS buy clothes Olu too HTS buy `Ayo bought clothes, Olu bought some, too.'
Some questions such as those presented in (18a) and (19a) leave room for a choice between the focus construction and the unmarked construction. If the situation already implies preconstruction, only the argumentfinal construction is possible. This is the case in the following two examples:
(20) a. ta lo fo awo yii? who FOC:3s break plate this `Who broke this plate?' b. Ayo ni. Ayo FOC `It was Ayo.' c. #Ayo ' fo awo yii. Ayo HTS break plate this `Ayo broke this plate.' (21) a. ki lo di sowo? what FOC:2s hold LOC:Hand `What do you have in your hand?' b. owo ni. money FOC `It's money.' c. #mo di owo sowo. I hold money LOC:hand `I have money in my hand.'
In (20a) the demonstrative in awo yii `this plate' creates a situation of preconstruction in which the speaker cannot but identify exhaustively who did the action of breaking the plate. For that reason, only (20b) is pragmatically acceptable. In a similar way, the fact that what is asked for is in the hand of one of the speech-act participants creates a preconstructed situation in which only the focus construction in (21b) can be used, because only this construction denotes exhaustively what can potentially be in that hand.
The use of argument-final ni is not limited to selecting a particular entity (or possibly more than one) out of a preconstructed set. It can also be used to identify a whole state of affairs if this is what is supposed to be supplied by the speech situation. Thus, a sentence such as (22) can be used either in a situation where a group of presupposed persons is eligible and Ayo is the one who actually came or in a context where the whole state of affairs is asked for by a question such as `Is anything the matter?'. With the latter interpretation, (22) is analyzed as an argument-final construction with scope over the clause.
(22) Argument-final ni Ayo ni o wa lanaa. Ayo FOC 3s come at:yesterday `Ayo (is the one who) came yesterday.'
In the argument-final construction with scope over the clause, ni is in contrast with the pragmatically unmarked clause structure characterized by the high tone syllable (HTS). The unmarked clause structure is used if there is no particular preconstruction against which the utterance can be mirrored. This is certainly not the case if there is a preconstructed domain created by a question such as `Is anything the matter?'. The next question then is why ni occurs after the subject. According to Bisang and Sonaiya (1999) the high tone syllable marks that the relationship between a subject X and a predicate p in a state of affairs <X, p> is positively realized. In this sense, the HTS actualizes or positively constructs the relationship between a subject X and a predicate p. In a context in which there is no need to construct such a relationship between the subject and the predicate, that is, in a context in which it is the state of affairs as a whole including the relation between subject and predicate that is being referred to, the HTS is of no use.
The predicate-final construction is used in contexts where the information lacking is the predicate. Example (23b) is an answer to the question of what Bola is doing, (23a). The entity to which the predicate applies is known and is a part of that question. In such a context, it is not possible to mention that entity again in the answer.
(23) a. ki ni Bola n se? what FOC Bola PROG do `What is Bola doing?' b. o n sun ni. he PROG sleep FOC `He is sleeping.'
The use of this construction is limited to the predicate itself. As we have pointed out above, it is not possible to have an overt noun in the subject position in this construction. The pronoun o `s/he' is obligatory and functions like an agreement marker (cf. Dechaine 1993; Bisang and Sonaiya 1999). The same applies to objects with transitive verbs. They can only be referred to by an object pronoun as in (24b):
(24) a. se o fe fo aso yen ni? Q 2s want wash dress that FOC `Do you want to wash that dress?' b. rara, mo fe lo o ni. no I want iron it FOC `No, I want to iron it.' c. #rara, mo fe lo aso yen ni. no I want iron dress that FOC
VO compounds and emotional-verb constructions are two exceptions to this rule. In VO compounds the noun and the verb form one single compound. As was pointed out by Sonaiya (1990; also cf. Laniran 1992), there are two different processes of compounding nouns and verbs. In both cases, the verb has the structure CV and the noun has the structure VCV. In one process of combining, the first vowel of the noun is deleted. Thus, the noun iwe `book' loses its initial vowel and forms a compound with the verb ka `read': kawe `read'. Similarly, pgn `fetch' plus omi `water' yields ponmi `fetch water'. In the second process, the noun keeps its initial vowel but the verb loses its vowel. This, pa `kill' plus eran `meat' yields peran `kill an animal'. VO compounds in which the vowel of the noun is deleted (V2 deletion) are more easily acceptable with the predicate-final construction (examples  and ) than those in which the vowel of the verb (VI deletion) is lost (example ):
(25) o n kawe ni. s/he PROG read:book FOC `He is reading.' (26) o n ponmi ni. s/he PROG fetch:water FOC `S/he is fetching water.' (27) ?o n peran ni. s/he PROG kill:meat FOC `S/he is killing an animal.'
Sonaiya (1990) analyzes V2 deletion as a lexicalization strategy in Yoruba, which explains why it can be used more readily with predicate final ni. VO compounds with VI deletion do not form such a unit. The kind of context in which (27) would be acceptable is one in which, say, pictures of Bola performing different activities are being shown to children in a classroom. For each one, the teacher asks "What is Bola doing?" and the children give the appropriate answer.
Emotional verbs are constructed with a noun denoting a feeling, such as eru `fear', or a body part, such as inu `inside, belly' or oju `eye, face', in the subject position and the experiencer in the object position, as in oju ti mi [face push/close me] `I felt bashful/ashamed'. Many of these emotional constructions are highly idiomatic. In constructions in which the noun in the subject position denotes a fully referential entity such as ese `foot' in (28), it can be part of a wh-question (cf. [28a]). In the answer to this question, however, the noun mentioned in the question must not be repeated. What remains is the verb plus its experiencer in the object position (and, of course, the pronoun of the third person singular as the minimum obligatory item in the subject position). This type of verb-experiencer construction is also allowed with predicate-final ni. In fact, this particular type of' construction is fairly frequent in spoken Yoruba.
(28) a. ki lo se e lese. what FOC:3s make 2s:OBJ LOC:foot `What's the matter with your foot?' b. o n dun mi ni. 3s PROG hurt 1s:OBJ FOC `It hurts.'
In the clause-final construction the whole clause, whose predicates and arguments have to be overtly present, is marked by ni:
(28) c. ese ' n dun mi ni. foot HTS PROG hurt 1s:OBJ FOC `It's because my feet hurt.'
Example (28c) is possible in a context where it is used as an explanation for a given situation. Thus (28c) may be an answer to a question such as (29).
(29) ki lo de ti o fi n rin bayen? what FOC:3s arrive COMPL you use PROG walk like_that `Why are you walking like that?'
The difference between the argument-final construction with clausal scope, as discussed in (22), with the clause-final construction, presented in (28c), is that the former simply supplies the answer to a question after a particular state of affairs, whereas (28d) clearly identifies the major source for a preconstructed situation. Consequently, the argument-final construction with clause-final scope in (28d) below is the answer to a question such as (30), which thus asks for more information in the form of a clause.
(28) d. ese lo n dim mi. foot FOC:3s PROG hurt 1s:OBJ `My feet hurt.' (30) ki lo se e? what FOC:3s make/do 2s `What's wrong with you?'
In another situation, the question in (30) may yield an answer such as (31):
(31) mama mi lo ku lanaa. mother my FOC:3s die yesterday `My mother died yesterday.'
In order to provoke an answer in the shape of a clause-final construction, more context is needed. For instance, one speech-act participant may look really sad and the other asks her/him, `You are looking really terrible, what's the matter with you?'. In such a context, one can expect a clause-final construction such as (32):
(32) o o mo pe mama mi sese ku ni. 2s NEG know COMPL mother my just die FOC `Don't you know, my mother has just died.'
As we can see from the following examples, (33) and (34), in questions where a whole state of affairs is under discussion, only the construction with clause-final ni is possible, (33). The same question in which the state of affairs to be discussed occurs in the argument-final construction is pragmatically odd, (34):
(33) se ese n dun e ni ti o fi n Q foot PROG hurt 2s:OBJ FOC COMPL 2s use PROG rin bayen? walk like_that `Is it that your feet hurt you that you walk like that?' (34) *se ese lo n dun e? Q feet FOC:3s PROG hurt 2s
Finally, we would like to point out a special form of the argument-final construction in which there is a nominalized verb in front of nj. The nominalizing pattern is the same for all verbs, `CV [right arrow] CiCV'.
(35) Argument-final ni with reduplicated verb didun lo dun mi. RED:hurt FOC:3s hurt ls:OBJ `Oh yes, it hurts indeed.'
The situation in which the construction in (35) is used differs significantly from the predicate-final construction as illustrated in (28b), in which the linguistic element in front of ni is simply a variable to a question in which the predicate is unknown. The argument-final construction with a reduplicated verb in front of ni is used to emphasize the predicate after a context for it has been constructed. Thus, the utterance in (35) is based on a context in which it is already known that `it hurts'. Consequently, the strategy of argument-final ni with reduplication is used to emphasize that `it really hurts indeed'. Similarly, the following example (36) may be uttered in a context in which it is known from the context that a child has been very ill. Thus, its mother may confirm and emphasize this fact by the following utterance:
(36) gbigbon lo n gbon. RED:shiver FOC:3s PRO(; shiver `And indeed, it was shivering [a lot].'
To conclude this section, we shall divide the concept of identificational focus as presented by Kiss (199g) into the three features mentioned at the beginning of this section, [[+ or -] preconstruction], [[+ or -] identification], and [[+ or -] exhaustiveness], where the latter two depend on the former in the sense that [[+ or -] identification] and [[+ or -] exhaustiveness] cannot exist without [[+ or -] preconstruction]. The first feature refers to the presuppositional status of what is within the scope of ni. Does what is marked by ni refer to elements given contextually or situationally or by world knowledge? The feature of [[+ or -] identification] marks whether the speaker identifies what is marked by ni with regard to a preconstructed domain or whether she/he merely provides relevant information. Finally, the feature of [[+ or -] exhaustiveness] denotes whether the speaker implies that what is marked by ni is exhaustive, that is, that there are no alternatives. Table 2 summarizes which of these features is relevant for each of the four constructions of focus marking.
Table 2. The four focus constructions in Yoruba and their semantic/pragmatic characteristics
Preconstruction Identification Exhaustiveness Argument-final + + + Argument-final with clausal scope + - (-) Predicate-final + - (-) Clause-final + + (+)
In the argument-final construction the function of ni is exactly parallel to its function as a copula, that is, it selects a particular entity from among a preconstructed set of possible candidates and exhaustively identifies it as the one that applies to a given situation. In the argument-final construction with clausal scope and in the predicate-final construction, what is marked by ni refers to a contextual or situational preconstruction without implying either identification or exhaustiveness. The speaker using ni in these two constructions simply provides the hearer with the information s/he has asked for -- the question of the hearer creating the framework or the preconstruction within which the speaker has to build up her/his reply. The bracketed minus sign (-) under exhaustiveness reflects the fact that according to Gricean maxims any answer to a question is supposed to follow the cooperative principle. Of particular importance in this context are the two maxims of quantity, which imply that the speaker's contribution is "as informative as required" and does not make her/his "contribution more informative than is required" (Grice 1975). This type of general exhaustiveness is not based on preconstruction and it is not based on a set-theoretical definition. In this sense, the concept of exhaustiveness adopted in Table 2 is stronger than the one based on the two maxims of quantity. In contrast to the former two constructions, not only does the clause-final construction provide information, it clearly identifies the major source or raison d'etre of a preconstructed situation and thus exhaustively states that this situation is exclusively due to the information presented by the clause marked by ni.
As is pointed out by Kiss (1998:261) for languages such as Hungarian or English, clauses, VPs, and predicative phrases cannot occur in the focus position.(4) This seems to be due to semantic reasons: "perhaps clauses, VPs, and predicative phrases cannot function as identificational foci because they do not denote individuals, which serve as the primary domain of quantification" (Kiss 1998: 261).
As we have seen above, the focus marker ni does occur with clauses and with verbs (predicate-final construction) in Yoruba, and it can even be found with copula verbs provided the copula is not itself ni. Thus, the copula je (cf. section 2) is used together with the focus marker ni in the following example:
(37) iku ati oorun egbe ni won je. death and sleep mate FOC 3p be `Death and sleep are mates/are equal.'
As we have also seen above, if the focus particle ni is used with predicates or with clauses, it loses part of the functional features it shows when used with nouns. This behavior of ni can also be explained by the semantic reason presented in the above quotation from Kiss (1998). If the focus particle ni marks a concept that does not denote an individual (or an entity in our terms), it loses some of the features that are prototypical for identificational focus. Thus, languages seem to have two options to deal with the semantic criterion of individuality. Either they do not allow categories with the feature [-individual] to occur in the position of identificational focus or they reduce the functional load of the focus particle if the concept marked by it does not denote an individual. The second option is the one at work in Yoruba. The fact that there is such an option is a strong case against the assumption that a special focus position such as the specifier position of a functional projection(5) is universally limited to the function of identificational focus as defined by Kiss (1998).
From Table 2 we can see that the function of ni is characterized only by [+ preconstruction] in the argument-final construction with clausal scope and in the predicate-final construction. This reduction of the functional load of ni is to be expected from the above explanation. What may look surprising is the fact that the clause-final construction is characterized not only by [+preconstruction] but also by [+identification] and to a certain extent by [exhaustiveness]. A possible explanation to this may be that it is easier to conceptualize a state of affairs as one single whole in the sense of an individual than its predicative subparts. The fact that states of affairs do not denote prototypical entities may then be seen as the reason for reinterpreting [exhaustiveness] in the sense that the state of affairs marked by ni is the source or raison d'etre of a preconstructed situation.
The predicate alone can only occur in the argument-final construction if it is nominalized by reduplication and thus conceptualized as an entity (cf. examples  and ). Moreover, it looks as if the predicate to be nominalized by reduplication has to be preconstructed in the sense that it is known to the speaker and the hearer. If argument-final ni with its general function of exhaustive identification is applied to a set consisting of only one single preconstructed predicate expressed by reduplicative nominalization, its reinterpretation in terms of emphasis seems to be quite straightforward.
4. The focus marker ni in Yoruba in the light of some current theories of information structure
In this section we shall look at the Yoruba focus particle ni from the point of view of some concepts of focus recently discussed in the literature. Most of this section will be dedicated to the focus theory presented by functional grammar (Dik 1997), because this seems to be the only theory that systematically integrates the concept of the preconstructed domain by distinguishing the feature [+ specific presupposition]. As we pointed out above, a preconstructed domain is a set of linguistically, contextually, or culturally presupposed items out of which the speaker selects the one (or probably more) that she assumes to be exhaustively relevant for a given situation. The other approaches discussed in this section, that is, the thetic/categorical distinction, information packaging as presented by Vallduvi and Engdahl (1996), and information structure as described by Lambrecht (1994), do not provide any concept such as that of the preconstructed domain and thus fail to systematically account for the cases in which the focus particle ni can be used in Yoruba.
Dik (1997) classifies the different types of focus by the features [[+ or -] contrastive], [[+ or -] specific presupposition], [[+ or -] corrective]. After having checked for each of these focus types whether it is compatible with ni, we shall try to give a general explanation for our findings.
In the completive focus, the speaker asks for a variable to which the hearer is supposed to present an answer. As we have pointed out in section 3, the focus particle ni has to be used if the entity supplied by the answer belongs to a preconstructed set from which the speaker has to identify the one that applies for a given situation exhaustively. If the speaker merely provides the information s/he has been asked for, the focus construction cannot be used. As we have also seen in section 3, the situation is different with predicates. In this case, the predicate-final construction has to be used irrespective of identification or exhaustiveness. What matters is only the fact that the question created a preconstructed domain for which the speaker's reply has to be relevant.
In the context of selective focus, the hearer is supposed to select a particular entity from a group of different entities or a particular predicate from a group of different predicates. In both cases we find ni: if what has to be selected is represented by an argument, the speaker has to use argument-final ni; if it is a predicate, predicate-final ni is needed.
(38) a. ewa lo fe je tabi iresi? bean FOC:3s want eat or rice `Do you want to eat beans or rice?' b. ewa ni. bean FOC `Beans, please.' (39) a. se o fe ya oko yii ni tabi o fe ra a? Q 2s want rent car this FOC or 2s want buy OBJ:3s `Are you going to rent or buy this car?' b. mo fe ra a ni. ls want buy 3s:OBJ FOC `I want to buy it.'
In the context of expanding focus, additional information is presented to some information that is already mentioned. Thus, a given argument can be expanded by naming more arguments (cf. example ) and a given predicate can be expanded by naming more predicates (cf. example ). In both cases ni or its negative counterpart ko ni is used as far as preconstruction goes. In addition, the predicate has to be reduplicated. The argument or predicate containing additional information, however, cannot occur with ni. In the case of the predicate, the verb that occurs in front of ni/ko ni has to be reduplicated. This seems to be triggered by the fact that the reduplicated verb is emphasized because of its being contrasted.
(40) a. Ayo bought coffee. b. kofi nikan ko lo ra, o ra ewa naa. coffee only NEG:FOC FOC:3s buy 3s buy bean too `He not only bought coffee, he also bought beans.' (41) rira nikan ko lo fe ra ewa, o fe buy:RED only NEG:FOC FOC:3s want buy banana 3s want ta a pelu. sell 3s:OBJ too/together `He is not only going to buy bananas, he's also going to sell them.'
In both cases of corrective focus, that is, in restricting and replacing focus, ni is used. In restricting focus, the speaker cancels some nouns or verbs out of a previously mentioned set of nouns (cf. example ) or predicates (cf. example ).
(42) a. You bought corn and beans. b. mi o ra agbado, ewa nikan ni mo ra. ls NEG buy corn beans only FOC ls buy `I didn't buy corn, I only bought beans.' (43) a. It seems Ayo grows and sells beans. b. o kan n-ta won hi. 3s only PROG-sell 3p FOC `He only sells them.'
In replacing focus, a given entity (represented by a noun) or a given predicate (represented by a verb) is replaced by another noun or predicate (cf. examples  and ). In the case of predicates, we find either the construction with argument-final ni plus reduplication, (45b), or the construction with clause-final ni, (45c).
[Figure 1 ILLUSTRATION OMITTED]
(44) a. You bought corn. b. agbado ko ni mo ra, ewa ni. corn NEG:FOC FOC 1s buy bean FOC `It wasn't corn that I bought, it was beans.' c. kii se agbado ni mo ra, ewa ni. it_is_not corn FOC 1s buy bean FOC `It wasn't corn that I bought, it was beans.' (45) a. John sells clothes. b. tita ko lo n-ta aso, sell:RED NEG:FOC FOC:3s PROG-sell clothes rira lo n-ra a. buy:RED FOC:3s PROG-buy 3s:OBJ `He doesn't sell them,, he buys them.' c. ko ta aso o, o n-ra a ni. NEG sell clothes EXCL 3s PROG-buy 3s:OBJ FOC `He doesn't sell clothes, he buys them.'
Parallel focus does not imply arty specific presupposition. This seems to be the reason why ni is not used in this context:
(46) Ayo.' ki mi daadaa, sugbon inu Ade o dun Ayo HTS greet 1s:OBJ well but inside Ade NEG pleasant si mi. PREP OBJ:1s `Ayo greeted me well, but Ade was not pleasant with me.'
From the above examples we can conclude that ni is used with all of the above subtypes of focus except with completive focus in the case of the argument-final construction, with the additional information provided by the expansive focus and with parallel focus. All of these cases can be characterized by the feature [-preconstructed]. This is straightforward with completive focus in the case of the argument-final construction. In the expansive focus type, the argument or predicate containing additional information does not belong to the preconstructed domain, that is, the domain of concepts presupposed by both speech-act participants. By adding these concepts, the speaker thus explicitly breaks up the preconstructed domain by saying that there is a potentially unlimited number of additional items to be subsumed within that domain. For that reason, it is not possible to use the focus marker ni with expansive focus. For similar reasons, also phrases do not occur in the identification focus position in Kiss's (1998: 252-253) theory. However, those concepts, which are contrasted with the concepts to be newly introduced, belong to the preconstructed domain, since they are shared knowledge of both speech-act participants. Consequently, they are marked by ni. In the case of predicate-final ni, the predicate even has to be reduplicated. In parallel focus the focus marker cannot occur because there is no preconstructed domain to contrast with. The concepts contrasted in this type of focus are in no way preconstructed, they are simply mirrored against each other.
Completive focus, which is the only noncontrastive type of focus, is still based on a preconstructed domain. This preconstructed domain is given by the question, which predetermines what kind of variables the speaker would like the hearer to supply. What the hearer is supposed to supply is a clear answer that does not allow for alternatives. This is sufficient for predicates to be always marked by predicate-final ni. As for arguments, argument-final ni is only allowed if [preconstruction] cooccurs with [identification] and [exhaustiveness]. The selective focus is probably the prototypical case of preconstruction. There is a certain domain that is known to both speech-act participants and the speaker exhaustively identifies the concepts that apply to a given context. In the context of restricting focus, those concepts of a preconstructed domain that actually hold for a given situation are confirmed as belonging to that domain. Since no new concept is coming up in this context, the concepts confirmed take the focus marker ni with arguments as well as with predicates. Moreover, the operator that is used to confirm the membership of a concept in the preconstructed domain is nikan `only' with arguments and kan `one, only' with predicates. As was pointed out by Kiss (1998: 265), this "quantifierlike element adjoined to an XP" represents "a special type of identificational focus," which is marked by ni in Yoruba. Finally, in replacing focus, one concept belonging to the preconstructed domain is replaced by another one. This is done in two steps. First, the concept to be cancelled from the preconstructed domain is mentioned. Since it belongs to the preconstructed domain, it will be marked by ni, or to be more precise, by its negative form, ko ni, or together with kii se `it is not that' in the case of arguments. In the case of predicates, the predicate will be negated, either by a normal negation or by marking the nominalized predicate by the negative focus marker. Second, the replacing concept will be introduced into the slot left open by the removed concept. Since there is no breaking up of the preconstructed domain, as in the case of expanding focus, the focus particle is used.
From the above findings we can see that the meaning of ni as a copula also decides which of the focus types as defined by Dik (1997) can take it as a focus marker. It is used to mark only those types of focus that are based on a preconstructed domain.
The thetic/categorical distinction was introduced from logic into linguistics by Kuroda (1972) and, more explicitly, by Sasse (1987, 1995). Categorical utterances are bipartite. They consist of an entity about which the predication is made, t]he predication base, and a predicate that says something about the predication base. Thetic utterances are monomial, that is, "no argument is picked out as a predication base; the entire situation, including all of its participants, is asserted as a unitary whole" (Sasse 1995: 4-5). As for Yoruba, we have seen above in example (22), repeated below as (47), that a sentence with argument-final ni can be a possible answer to the question `Is anything the matter?' or to the question `Who came yesterday?'. The former question calls for a thetic utterance, while the second question is bound to be answered by a categorical utterance.
(47) Ayo ni o wu lanaa. Ayo FOC 3s come at:yesterday Ayo (is the one who) came yesterday.'
Thus, the thetic/categorical distinction does not account for the opposition between the occurrence of both the unmarked construction and one of the focus constructions. Only if the concept of theticity is further confined to statements out-of-the-blue, as we typically find them in newspaper headlines of the type presented below in (48) (cf. Sasse's  annunciatives) and some other such statements, may we say that the neutral construction can be used for thetic utterances.
(48) awon ole ji oko gomina gbe. PL thief HTS steal vehicle governor carry `Thieves have stolen the governor's vehicle.'
Vallduvi and Engdahl (1996) divide the sentence into three items of information packaging. One of these items is the focus, which is defined "as the actual update potential of a sentence" (Vallduvi and Engdahl 1996: 469). The other two items are link and tail, which both belong to the ground, which is that part of the sentence that is "already subsumed by the input information state" (1996: 469). The link specifies where the focus should go in the information state, whereas the tail indicates how the focus fits there. Based on these three items, Vallduvi and Engdahl (1996) present four different types of instructions (link-focus instruction, link-focus-tail instruction, all-focus instruction, focus-tail instruction). We have tried to see whether these four types yield significant differences in the structuring of information in Yoruba. The result was negative. Yoruba uses the unmarked sentence structure with the high tone syllable for all four types of instruction. This doesn't come as a surprise, since there is no space for a presupposed set of items from which the speaker then makes a selection.
Similarly, Lambrecht's (1994)types of information structure are only based on the question of how information is organized in general, without asking the question of what happens to it if it has to be related to a preconstructed domain. In (4.9) we briefly present the four types postulated by Lambrecht (1994):(6)
(49) a. Topic-comment type (predicate focus): (What did the children do next?) The children went to SCHOOL. b. Identificational type (argument focus): (Who went to school?) The CHILDREN went to school. c. Event-reporting type (sentence focus): (What happened?) The CHILDREN went to SCHOOL. d. Background-establishing type: (John was very busy that morning.) After the children went to SCHOOL, he had to clean the house and go shopping for the party.
We have also tried to see what happens to these four types of information structuring if they are applied to Yoruba. The topic-comment type and the background-establishing type arc: both expressed by the unmarked clause structure with the high tone syllable. The identificational type can be expressed either by the unmarked construction or by the argument-final construction, depending on whether the entity that is asked for belongs to a preconstructed domain. The event-reporting type yields the argument-final construction with clausal scope. Thus, Lambrecht's (1994) theory yields some of the different focus constructions of Yoruba, but it does not cover the whole system of constructions. Moreover, there is the identificational type, which does not yield a coherent picture in Yoruba since it allows for two alternatives;.
Although the relevance of preconstruction has been acknowledged in the literature, it is rather neglected in most theories about focus (cf. Kiss 1998:246 for a similar statement). Due to this fact, theories based on the thetic/categorical distinction (Kuroda 1972; Sasse 1987, 1995) as well as theories dealing with information packaging (Vallduvi and Engdahl 1996) or with information structure (Lambrecht 1994) do not yield the categories needed for adequately describing the criteria that determine the use of the four focus constructions of Yoruba presented under section 3.
The data on Yoruba show the importance of the notion of preconstruction, without which the focus system of this language cannot be understood. Not only is this notion applied to those elements of the language structure that prototypically denote entities, it is also used with predicates and with clauses where it is no longer linked to the functions of [identification] and [exhaustiveness]. Moreover, preconstruction is not only crucial for the focus system of Yoruba, it also forms the basis of the copula system of this language. As is well known, focus markers are often grammaticalized from copula verbs. The reason preconstruction is so pervasive in the Yoruba focus system may thus be due to the fact that it is also relevant for the copula. The same fact may also explain why the focus position in Yoruba is not limited to the function of identificational focus as defined by Kiss (1998).
Received 18 August 1997
Revised version received 25 October 1999
University of Mainz Obafemi Awolowo University, Ile Ife, Nigeria
(*) This article was supported by the Alexander yon Humboldt Foundation, which made it possible for Remi Sonaiya to stay at the University of Mainz from September 1995 to August 1996. We would like to thank the Alexander yon Humboldt Foundation for its generous and continued support, without which this and other papers would not have been possible. Correspondence address: Professor Walter Bisang, Institut fur Allgemeine und Vergleichende Sprachwissenschaft, Johannes Gutenberg-Universitut Mainz, Jakob-Welder-Weg 18, D-55099 Mainz, Germany. E-mail: wbisang(E-mail.uni-mainz.de.
(1) In our interlinear glosses, we use the following abbreviations:
COMPL = complementizer
EXCL = exclamatives
FOC = focus particle
HTS = high tone syllable
LOC = locative
NEG = negative
OBJ = object
PL = plural marker
PREP = preposition
PROG = progressive
Q = question marker
RED = reduplication
REL = relative marker
RPD = reference to a preconstructed domain
We use (*) to mark grammatically unacceptable clauses; (#) is used to mark pragmatically inappropriate utterances. The question mark at the beginning of an example is used to refer to utterances whose status is grammatically and/or pragmatically problematic.
(2.) In response to the question "What can you tell me about Paris?" a pronoun might also be used instead of the name of the town being repeated: oun ni olu-ilu ile Faranse.
(3.) Instead of ni o [copula 2s] and ni o [copula 3s] we very often find the contracted forms, i.e. lo [FOC: 2s] and 16 [FOC: 3si respectively.
(4.) Some lines above, Kiss (1998:261) is even more specific about the categories that cannot occur in the position of identificational, focus. She enumerates the following categories: that clauses, infinitival clauses, VPs, and predicative NPs/AdjPs.
(5.) Eun and Nishiyama (1997) propose an analysis for the focus position in Yoruba similar to the one Kiss (1998) proposes for English and Hungarian. The NP is moved to FP whereas ni is dominated by F'.
(6.) Lambrecht (1994) presents the following explanations:
"In the topic-comment type in [49a], the purpose of the assertion is to pragmatically predicate some property of an already established discourse referent. In the identificational type in [49b], the assertion has the purpose of establishing a relation between an argument and a previously evoked open proposition. In the event-reporting type in [49c], the purpose of the assertion is to express a proposition which is linked neither to an already established topic nor to a presupposed open proposition.... Finally, in the background-establishing type in [49d], a pragmatically presupposed proposition serves as a scene-setting topic for another proposition, which may itself be of any of the other three types" (Lambrecht 1994:126; the example numbers are ours).
Bisang, Walter; and Sonaiya, Remi (1997). Perfect and beyond, from pragmatic relevance to perfect: the Chinese sentence final particle le and Yoruba ti. Sprachtypologie und Universalienforschung 50 (2), 143-158.
--; and Sonaiya, Remi (1999). The functional motivation of the high tone syllable in Yoruba. Journal of African Languages and Linguistics 20, 1-19.
Dechaine, Rose-Marie (1993). Predicates Across Categories: Towards a Category-Neutral Syntax. Amherst, MA: GLSA.
Dik, Simon C. (1997). The Theory of Functional Grammar, part l: The Structure of the Clause, Kees Hengeveld (ed.), 2nd ed. Berlin and New York: Mouton de Gruyter. (First edition [ 1992]. Dordrecht: Foris.)
Eun Cho; and Nishiyama, Kunio (1997). Yoruba predicate clefts from a comparative perspective. Paper presented at the 28th American Conference of African Languages, Cornell University, 12 July.
Grice, Paul H. (1975 ). Logic and conversation. In Speech Acts, Peter Cole and Jerry L. Morgan (eds.), 41-58. New York: Academic Press.
Heine, Bernd; and Red, Mechthild (1984). Grammaticalization and Reanalysis in African Languages. Hamburg: Buske.
Hengeveld, Kees (1992). Non-Verbal Predication. Theory, Typology, Diachrony. Berlin and New York: Mouton de Gruyter.
Kiss, Katalin E. (1998). Identificational focus versus information focus. Language 74(2), 245-273.
Kuroda, S.-Y. (1972). The categorical and thetic judgment: evidence from Japanese. Foundations of Language 9, 153-185.
Lambrecht, Knud (1994). Information Structure and Sentence Form. Topic, Focus and the Mental Representation of Discourse Referents. Cambridge: Cambridge University Press.
Laniran, Yetunde Olabisi (1992). Intonation in tone languages: the phonetic implementation of tones in Yoruba. Unpublished Ph.D. dissertation, Cornell University.
Sasse, Hans-Jurgen (1987). The thetic/categorical distinction revisited. Linguistics 25, 511-580.
--(1995). "Theticity" and VS order: a case study. Sprachtypologie und Universalienforschung 48(1/2), 3-31.
Sonaiya, Remi (1990). Yoruba vowel deletion. Linguistique Africaine 5, 77-93.
Vallduvi, Enric (1992). The Informational Component. New York: Garland.
--; and Engdahl, Elisabeth (1996). The linguistic realization of information packaging. Linguistics 34(3), 459-519.
|Printer friendly Cite/link Email Feedback|
|Author:||BISANG, WALTER; SONAIYA, REMI|
|Publication:||Linguistics: an interdisciplinary journal of the language sciences|
|Date:||Jan 1, 2000|
|Previous Article:||The phonology of Classical Greek meter(*).|
|Next Article:||Pragmatics: Critical Concepts.|