The Status of Sì in Yoruba

This paper examines the syntax of sentential conjunction in the Yoruba language with the view of ascertaining the actual syntactic function of sì on which opinions have been polarized in the syntax of Yoruba grammar. This paper argues that the Yoruba language has a sentential/clausal conjunction element but its structural position is not the between the two clauses. The researcher conducted a series of structured interviews and also consulted existing works targeting the sì in compound clauses so as to generate the data for the research. The research adopts the Minimalist Program (MP) as its theoretical tool for the analyses presented in the article. MP views syntactic derivations as resulting from computational systems whose operations are based on operation Select and operation Merge. Syntactic structures are built in a par-wise fashion from bottom to top by putting two items together at a time. The research shows that there are three different sì in Yoruba occurring in seemingly similar environment but they are performing different syntactic functions. One sì denotes “emphasis” as a preverbal element; the second one denotes “consecutiveness of action” while the third one performs the function of sentential/clausal conjunction, an overt realization of the in the second clause. The overtly marked sentential/clausal conjunction also has a variant, which is an abstract realization between the two conjoined clauses.

In the Minimalist Program literature (hence, MP) there are several publications on the nature of sentence types and derivations (Pollock, 1989;Chomsky, 1991Chomsky, , 1993Chomsky, , 1995Rizzi, 1997;Ajiboye 2005;Ajongolo, 2005;Abimbola, 2014, Olaogun 2016. Compound structures have been identified in MP as projections of the conjunction head (hence, Conj 0 ) whose maximal projection is conjunction phrase (hence, ConjP). ConjP as a functional projection combines two or more Tense Phrases 1 (TPs, i.e. TP 1 and TP 2 TP n) in presyntactic computations together to form compound 1 Yoruba clause structure has been identified as having a tense head represented by T0, and TP re ten as a maximal category, Ilori (2010); Oduntan (2000); among others. Other abbreviations used in this work include; * -ungrammatical structure/sentence; MP -Minimalist Program; Conj0/Conj -conjunction head; ConjP -Conjunction phrase; ConjII -extended projection of ConjP; TP1 and TP2 -clause one and two in conjoined sentences; T1intermediate category/projection of tense phrase; sentence constructions. Compound sentences are known to consist of two different clauses joined by a conjunction, in addition to having the two clauses as separable. Within Yorùbá linguistics, opinions are still polarized concerning one of the sentence conjunctions in Yoruba.
In the traditional grammar of the language sì is regarded functionally as the sentence conjunction. Scholars who hold this view are Bamgbose (1990), Yusuf (1980Yusuf ( , 1999 among others. We shall regard this hypothesis as the first school of thought. But some scholars do not agree on the presence of sentential conjunction in Yoruba and out rightly disagree with the first school of thought on sì as the actual sentential/ clausal marker of conjunction. Awobuluyi (1978;2001;2008;2013) and Ilori (2010) hold the view that Yoruba does not have sentential conjunction, constituting the second school of thought. But there is no known work that has considered the status of sì in Yoruba grammar. This is the gap this research intends to fill using the Minimalist Program.
Based on the behavior of conjunctions, conjoined constituents are said to be on either sides of the conjunction element. It is observed, however, that sì behaves differently and hence we investigate the behavior of the item if truly it is actually a conjunction as it has been classified traditionally and speculated in the language or whether it is a preverbal adverb as first mentioned by Awobuluyi (1978 Figure 3. A mono-clausal construction with Sì. Figure 2 represents the two clauses in Figure 1 where the bracketed TP 1 is clause1 and clause2 is TP 2 and sì is regarded as the linker between the two clauses, an item which is noticeable to occur consistently after the subject DP. Figure 3 is an example of mono-clausal construction where is occurs independently without the second clause. 2

Aims of the Research
It should be noted that the status of sì in the grammar of Yoruba language is still unknown. This paper aims to fill this vacuum. It will also examine the derivation of sentences involving clausal conjunction. Apart from the aforementioned goals, it will also investigate the possibility of the item being two in Yorùbá. To archive the aforementioned aims, the following research questions are developed to guide the analysis presented here: • What are the inadequacies in the analyses of conjoined sentences? • Does the Yoruba language have an overt realization of sentence conjunction marker? • If the language has a sentence conjunction marker, what is its form?
• Is sì a sentence conjunction marker in the Yoruba language? • What is the function of sì in the Yoruba language?

The Minimalist Program
The Minimalist Program (hence, MP) propounded by Chomsky (1995Chomsky ( , 2000 among others, is adopted for the analysis of this work. MP is a theory that builds on economic principles in the derivation of syntactic structures. It is a more natural and general approach to language (Ouhalla, 1999). MP views the human cognitive system as a computational system similar to that of a computer and uses a limited set of mechanisms and constraints to provide adequate explanation to language structures. MP's assumptions significantly deviated from its offshoot -Principles and Parameters' theory and other models of Generative Grammar -by eliminating theory internal levels of representation such as D-structure and S-structure. The Computational System of Human Language (C HL ) in MP is regulated by a principle called the Inclusiveness condition, which assumes that a derivation of a syntactic structure can be built only from the items specified in the numeration (Radford). The numeration serves as the spring-board for the computation where items are first selected into from the lexicon. The lexicon consists of lexical items (LI) used in the computational processes.
MP adopts minimal operations for computational processes: Operation Select, and Operation Merge, Agree, Transfer and Spell-Out. • Operation Select is used to choose from the list of Lexical Items (LIs) available in the lexicon into the numeration for further computation. • Operation Merge is a binary operation that combines two LIs to derive a Syntactic Object (SO). The by-product of merge operations is endocentric, so that when Merge groups two LIs, one of them projects maximally; for instance, if merge combines two elements X and Y, the resulting phrase takes its label from either X or Y (Collins, 2011) e.g. merge (X, Y) = {X, Y} = XP. Merge can be divided in two: internal and external merge. Internal merge is concerned with SOs that enter the derivation but have to undergo another merge operation. External merge only targets merger of SOs that enter the derivation from the lexicon. • Agree is the mechanism responsible for the matching of features for onward valuation of unvalued features, i.e., a situation where a head looks for matching features in its c-commanding domain. • Transfer is an operation in narrow syntax that ships derivation to the interfaces, i.e., LF and PF. • Spell Out is the point of interpretation at either the PF or LF. • This theory determines the structural representations drawn in this work. Also, the data analyzed were subjected to theoretical analysis.

Results
The materials for the study include: a) a primary data source where structured interviews were conducted in English and the responses were given in Yoruba by randomly selected Yoruba speakers; b) secondary sources were also consulted, including existing works on Yoruba grammar.

A Review of Existing Works on Sì
Scores of Yorùbá grammarians and linguists have examined the sì and opinions remain polarised on the conjunction sì in the language. Scholars like Yusuf (1980), Bamgbose (1980), Awobuluyi (1978) and Ilori (2010) among others have discussed conjunctions in Yorùbá. Conjunction has been variously defined, it is said to link two or more items together. In Yorùbá, there are basically two types of conjunctions regarding the items they can conjoin. Ogunbowale (1970, p. 95), cited in Yusuf (1980, p. 2), defines conjunction as, "a word which links part of speech or sentences together". This means that conjunctions that are used to connect two words, phrases, or sentences together do occur in between the words, phrases, or sentences that are being connected. As implied, from this analogy, conjunctions acts like the concrete between two bricks holding them together. Consider the following examples; As evident from the data presented above in Figure  4 and Figure 5, one could deduce that the conjunction àti/pèlú are DP conjunctions and cannot be used to connect DP and a clause/sentence together as seen in Fig. 6, any attempt to make àti/pèlú function in similar structures as Figure 6 would yield ungrammatical sentence structure in Yoruba language. Yusuf (1980) is one of the earliest scholarly works found on sì and his view has some theoretical implications and shortcomings as first observed by Ilori (2010). First, the lowering of sì to the position after the subject DP is rightward. This is not theoretically possible in Minimalist assumptions. This is because MP does not allow rightward movement. Secondly, clauses are generated from the list of lexical items selected from the lexicon, made available in the lexical sub-array, i.e., where computation begins, resulting in clausal architectures built from the Merge operation of two basic elements {α and β} with projection towards the leftward movement or Left periphery. This being so, there is no point where an item that enters late in the derivation can 'move down' to what has already being computed in the derivation. Judging by the ways conjunction elements in the language behave (even those Awobuluyi (1978) referred to as disjunctions) usually stand in between the elements being conjoined together and none of them moves rightward. Yusuf's representation is provided below in Figure 7 and Figure  8.

Yusuf's (1980) View on Sì
As evident in Figure 7 and Figure 8 Figure 7 is not endocentric, i.e., it is not the projection of the conjunction head and so it is not allowed in MP. Figure 8 does not have a head of its own; besides the use "S" is alien to MP.

Awobuluyi's View (1978 & 2008) on Sì
Traditionally, àmọ, ṣùgbọń and sì were identified as the clausal conjunctions in the language. But Awobuluyi (1978, p. 104) taking insight from the behaviour of these items noted that two separate classes are combined together in one class called conjunction in the Yorùbá traditional grammar. On the one hand, there is the disjunction which he defines as showing two or more elements are alternatives, and a person can only choose one'; on the other hand, there are the conjunctions which 'show that two or more things go together or are united. In sum, both conjunctions and disjunctions 'relate elements of the same functional class'. In his work sì is not classified as a sentence conjunction rather he classifies the item as pre-verbal adverb which marks 'consecutive action' (Awobuluyi, 1978, p. 69). He assumes that there is no sentence conjunction that behaves as phrasal conjunction or disjunctions in the language. In his discussion on the HTS in his (2008)  "He is the one that went there" It should be noted that Awobuluyi is the first scholar who observed and thus rejected the ideal that sì is a sentence conjunction. However, the reason for grouping the item with pre-verbs is unknown. The grouping is not enough to say that the item not a conjunction.

Ilori's (2010) View on Sì
According to Ilorio, the structural position occupied by sì within two sentences or clauses conjoined together seemingly questions the traditional or popular view. Ilori observed that sì regularly occur after the subject of clause2 of such compound clauses. As evident in Figure 11 below: [ TP Adé jẹun ó sì yó bámú] [ TP Ade eat-thing he sì full to-the-brim] "Ade ate and he was filled to the brim" Figure 11. An illustration of the position of sì after the subject of the second clause.
His view is, however, unlike most of his predecessors. He claimed that sì is a kind of adverb, "that adjoins to the pre-VP adjunct position in the second IP of the compound clause. It is from that position that the subject raises across sì/dẹ̀ to spec-IP to derive the surface syntax in which the subject of the clause2 linearly preceedes sì/dẹ". 4 His view corroborated Awobuluyi's (1978) view that sì is a pre-verbal adverb which indicates "consecutive action". In view of this claim he proposed the following architecture in Figure 12 below; His claim assumes that sì occurs as syntactic adjunct, i.e. adverb, which originates in pre-VP or pre-ASP within the second clause. 5 Ilori (2010, p. 176) concluded that Yorùbá does not have clausal co-ordinating conjunction similar to disjuncts like àmọ́ and ṣùgbọń "but" which structurally occupy the central position between the clause1 and clause2 in the language. We disagree with Ilori (2010) on this c.aim that although it may not be possible to have overt realisation of the conjunction marker between clause1 and clause2, a non-overt counterpart is realised in the structure by vertue of its feature.
One may quickly want to jump to a conclusion based on Ilori (2010) that truly Yorùbá does not have a clausal conjunction as he opined. But there are some notable shortcomings in his view notable as follows: (i) from the theoretical view, the Minimalist Program does not allow for traces 6 , movement is specified by the need of a Lexical Items (LIs) to satisfy some features otherwise it is frozen as Activity Condition specifies as Figure 13 below; (ii) adjunction as used in Ilori's work is theoretically anomalous. Before adjunction of constituents can take place the item must have: (a) at one time being adjoined in the lower level of the tree before it could be raised to spec-VP (as he assumes), which is suggestive of the fact that there is a fusion or merger of two different theoretical models in his work; (b) adjunction extends a MaxP into another MaxP, but this is never the case in his representation. As evident in the work, he assumes sì/dẹ̀ is adjoined to spec-VP as if it is DP or a MaxP, whereas it is not. That point shows that the item was merged at spec-VP. sì/ dẹ̀ did not originate from anywhere in the derivation but merged at spec-VP which shows that sì/dẹ̀ cannot enter the derivation there because it does not have MaxP projection. (iii) The assumption that there is no conjunction in such clauses is unacceptable. There is actually a connective abstract/ non-overt element which connects TP 1 to TP 2 as shown below in Figure  14.   We draw cross-linguistic evidence and present theoretical evidence from the Minimalist point of view on conjunctions and sì.
As one could have observed so far, assuming the previous version of generative theory is problematic towards ascertaining the function of sì but the problem becomes more vivid as mere over generalization of meaning carried over from the traditional view to the generative theory. Yet the earlier generative version cannot handle the derivation of such constructions accurately which is why it is confortable to assume (even with GB) lowering of the clausal conjunction. The question one is likely going to ask is this, 'what actually connect the two clauses?' Is it the case that there is no connective morpheme/ Lexical Item (LI) as assumed and shown by Ilori (2010) and implicit Awobuluyi's (1978) claim? In the next section we present the various readings of sì in Yorùbá.

Status of Sì in Yoruba Grammar
In this section we examine the various meaning of sì in Yoruba and also to ascertain among other things: [ TP Ó lọ sí ibẹ] [ TP He/she went prep there] "He went there" Figure 18. A simple sentence without sì.
In Figure 15 above, the sentence does not overtly mark any emphatic reading as did Figures 16 and 17. Figure 15 is just a simple declarative without any other information excerted in it. The addressee is actually aware of what the speaker is talking about before he stressed the point in the information. There is nothing else marking emphasis in Figure 16 except sì and nothing seems to be connected to that sentence as clause 2. In essence, Figure 15 is not adjoined to any sentence. Figure 18 is a simple declarative sentence, which does not exert any emphatic force compared to Figure 19 where it is as a matter of fact that the said action was carried out by the ó, the Ex-DP (subject). The addressee is unaware of the fact that the person being talked about actually did what was done. • As observed earlier, Awobuluyi (1978, pp. 73-74) notes that sì function is not according to the traditional view rejecting that sì is a conjunction in Yorùbá 9 , he claims that sì marks consecutive action as against the conjunction in linking two clauses together. Consecutive actions denote that events in a given sentence occur in succession or one after the other. One action occurs in precedence to the next in sequence and order of the time of occurrence without interruption. For examples: [ TP Tọṕẹ́ ra aṣọ tuntun, ó sì wọ̀ ọ] [ TP Tope buy-pst clothe new, he CA wear it] "Tope bought a new clothe and wears it" [ TP Ó gé ẹran, ó sì sè é] [ TP He cut-pst meat, he CA cook it] "He cut the meat and cooked it" Figure 21. A compound sentence with consecutive action.
9 Although Awobuluyi did give any explanation regarding such claim in his work. It is however the focus of this paper to show how such constructions has consecutive actions and also shows the conjunction linking the two clauses together. 10 CA denotes "Consecutive Action".
[ TP Ó mu ọtí, ó sì yó kánrin] [ TP He drink:pst wine, he CA brim excess] "He drank wine and he was drunk a lot" Figure 22. Consecutive action in compound sentence.
In Figure 20, the action of the buying of clothes, which is in clause1, precedes that of wearing it signified by sì in clause 2. 12 Similarly the action in clause1 of Figure 21 occurred before that of clause 2 invariably, the events thereof occurred in succession. There is no way where the cooking of the meat could occur before the buying of the same meat. Also, the events in described in Figure 22 were in sequence. The event of clause one, i.e. drinking of wine, occurred before the man got drunk. He could not be drunk before drinking of the same wine, which got him drunk. Nothing else informs this sequential ordering of events denoted in Figure 22 than sì. From the foregoing, one could conclude that which conjoins the two clauses is overtly not marked in Figures 20, 21 and 22 above. The question arising from this is, if sì is not conjoining the two clauses what is the conjunction? We shall come back to this question in the next section.
• Although the pool of evidence raised from the last two points could actually make one jump to a conclusion that sì is not a conjunction in Yorùbá. There are some expressions where one would be missing words to look further for the conjunction other than sì. For example: [ TP Ó sanra, ó sì tóbi] [ TP He be-fat, he conj be-big] "He is fat and he is big" Figure 23. A compound sentence with sì as conjunction.
11 You can have the same sentence rendered without consecutiveness in the action expressed like: Ó mu ọtí, ó yó kánrin where the clauses are only just conjoined together with an abstract conjunction, orthographically represented with that 'comma'. 12 Sì in sentences of this sort could be replaced with nítorí "because" or consequence of what has happened then, another event also occurred. For instance, Figure 20 Tọṕẹ́ ra aṣọ tuntun, ó sì wọ̀ ọ́ could be Tọṕẹ́ wọ̀ aṣọ tuntun, nítorí ó rà á (aṣọ tuntun) "Tope wore a new clothe because he bought a new clothe". It is a case of one event bringing about the occurrence of another in succession. Consider this other example; Èmi yóò dìde, èmi yóò sì tọ baba mi lọ, èmi yóò sì wí fun pé bàbá… , "I will arise and go to my father, and tell him father …" (Diglot Bibelì mímọ́ elédè méjì, Luke, 15:18, p. 1684).
The order of event came sequentially. The first thing is getting up, followed by going to the father and telling him that …, getting up cannot be preceded by going to the father or telling the father all that has happened. Nothing else is responsible for the ordering of the event other than sì. [ TP Ó pupa, ó sì lẹẃà] [ TP He be-fair, he conj be-beautiful] "He is fair and he is beautiful" Figure 25. A compound sentence showing sì as a conjunction.
The sequences of events in Figures 23, 24 and 25 above are not in sequential or consecutive order; rather the events were completed and not related to, overtly, the next event. It is not a consequence of being "fat" that leads to being big. It is quite relative to the individual who is fat. One that is fat does not necessarily have to be big. In Figure 24, one could go out without crying and one could stay indoors without crying. Invariably, neither of the two clauses could trigger emphasis nor consecutive actions. Similarly, one could be fair without being beautiful and one could be beautiful without being fair. Neither of clause1 nor clause2 is taking precedence, which informs why there is no consecutive reading in the events. There is no emphasis placed on the clauses, therefore sì is not in any way connected to emphasis and consecutive actions rather it is conjoining the clauses together.
The two clauses linked together in Figures 23, 24 and 25 above are linked by overt conjunction marker sì. But how is it that sì is a conjunction in some expressions, emphatic marker in some others and consecutive action marker?
It is logically plausible to assume that there are three different sì in the language, which have the same occurrence position from the Minimalist point of view. MP assumes that LIs are fully specified in the Lexicon with all the required syntactic, phonological and semantic properties needed for LIs in the derivation. If a lexical item is identical with another LI but different functionally, then they are two different lexical items specified in the lexicon. Variants are treated as individual lexical items in the Lexicon. From a logical point of view, there is a sì which marks emphasis, another one marks consecutive action, while the third denotes clausal conjunction. The three lexical items are specified in the lexicon with their features.

The Form of the Conjunction. The Abstract Clausal Conjunction in Clauses with Sì as Emphasis or Consecutiveness Marker
As noted earlier, in some constructions sì often marks consecutiveness and/or emphasis which invariably implied that there is a conjunction present in them, overt or not. In such constructions, a nonovert clausal conjunction conjoins clause1 and clause 2 together and as a functional head it has the ability to project maximally. In such compounds, the nonovert conjunction is the head of the ConjP contrary to Yusuf's (1980)   Figure 27 above shows that the head of the projection is the Conj 0 represented by "α". The Conj 0 selects into its complement position TP 2 , and selects TP 1 into its spec position. Both TP 1 and TP 2 are convergent TPs which have been computed before been selected here and merged to Conj 0 in the narrow syntax. As said earlier, Conj 0 is headed by a strong nonovert in the ConjP. This is why it is able to select the two clauses into its complement and spec positions. This is further illustrated on the schema in Figure 28.  Conj 0 and TP 2 were the first to enter the derivation under merge and both of them project Conj 1 . TP 1 was selected into spec position, such that Conj 1 and TP 1 were merged together to project ConjP. Sì enters the derivation at spec-vP for either "emphasis" or "consecutive actions". But the Ex-DP has to value its unvalued features, it is a goal. The Probe head is T 0 which has to value its φ-features. It probes down its domain and found a matching goal ó at spec-vP. It attracts the goal to its spec and values its φ-features which explains the ordering of sì after the Ex-DP in clause2.
Only the highest copy of the Ex-DP was linearized. 13

The Clausal Architecture of Clauses with Sì as Conjunction Marker
Normally, sì often occur after the Ex-DP in clause 2 14 and this actually informs the perspective from which scholars have been looking at it since the 1980s. It should be noted that sì does not occur in between the conjoined clauses but rather in another position, which makes it difficult for some early generative theories and their adherents to capture its behaviors. A conjunction is a functional head linking two or more items/constituents of equal properties together. We shall use Figure 23 for our illustration (repeated here for convenience as Figure 29). The projection is given as Figure 30 below. Figure 29. A sentence for demonstration of sì as a conjunction.
In Figure 30 above, TP 2 enters the derivation before TP 1, which enters the derivation as the spec -ConjP. Ignoring all the details at the vP layer, the derivation proceeds with the merging of the Ex-DP argument ó "3 rd :sg" at spec-vP to project to vP which merged with T 0 pres, to project T I . T 0 is a probe head and it has unvalued features. It probes down its domain and finds a goal ó with matching features, it attracts the goal to its spec, where the φ-features were valued and deleted. Consequently, the Nominative case feature of ó was also valued and deleted. The Conj 0 overtly marked by sì, is merged with the derivation thus projecting Conj I . Sì at Conj 0 has some edge feature which must be satisfied thus projecting in multi-spec order. i.e. multiple spec which holds that [X n…-1 ] could project a multiple number of specs which reduced by immediate projection below it. To be precise, multispecs from the top of the tree to bottom reduces by one. Thus, ó at spec-TP 2 bares unvalued features, i.e. emphasis. By virtue of being raised to spec of Conj 0 it become prominent in TP 2 . The same person who has performed the event in TP 1 is also performing event depicted in TP 2 . The entity is singled out as the topic of the discussion while the rest of the clause is the comment about the same entity. 15 ó which now has two copies is occupying the spec-Conj II . Only the highest copy is linearized at PF. Conj II is merged with an already converged TP 1 and thus projects ConjP.
The results demonstrate that in the Yorùbá Lexicon there are three main sì: one which denotes "emphasis", the second one marks "consecutive action" and the third is used as "clausal conjunction".

Conclusion
From the Yorùbá traditional grammarians, there is a clausal conjunction which does not behave according to every other conjunctions in the language whereas, scholars like Awobuluyi (1978Awobuluyi ( , 2001, Ilori (2010) have assumed that the traditional view is wrong. Their assumptions polarized the view on Yorùbá clausal conjunction; hence, the status of sì is unknown. In this article, however, we have established that there may be three instances of sì in Yorùbá differentiated by their individual functions present in the Yorùbá lexicon. We also submit that contrary to Ilori's (2010) claim that sì enters the derivation through as an adverb in the light vP layer and concludes that when the sì that denotes conjunction is selected from the lexicon, it enters the derivation as a head. Additionally, it has a strong edge feature, which triggers the Ex-DP of TP 2 to move to its spec position thereby projecting multi-spec for Conj 0 . When s,ì which denotes "emphasis" or "consecutive action" is selected, the item will enter the derivation through the edge of the light vP as an adverbial item.