BASÀ CLAUSE STRUCTURE; A Minmalist Approach

By

DANSABO, Nyizo Friday

0819705010763

Department of Languages and Linguistics, Faculty of Arts, Nasarawa

State University Keffi.

1
 CHAPTER TWO

Review of Related Literature

2.0 Preamble

This chapter reviews work done by other researchers and scholars that are related to the topic of

this present research. Work or literature on clausese: types, functions and conceptual issues in

minimalist grammar as they apply to clauses are discussed. The theoretical phrame work which

is based on chosky’s approach to syntactic analysis is examined in detail. This is achieved

through the assessment of the historical trend of Chomsky’s universal grammar so as to develop

an appraisal of the minimalist program to determine what led to its introduction. This review is

also designed to serve as a basement on which this work rests.

2.1 Conceptual Review

This section reviews literature on clauses, accessibility hierarchy, types, structure, functions of

clauses as well as conceptual issues in minimalist program as they apply to clauses are discussed.

2.1.1 Clause

A clause according to Jowitt (2015), is usually defined as a group of words containing a subject

and a predicate; it functions as part of a sentence. It is obvious that the definition of a clause is

more like that of a sentence. Jaymz (2018), states that a clause is a group of words that includes a

subject and a verb and forms a simple sentence or part of non-simple sentence. The finite verb is

most central to what constitutes a clause Omachonu (2003) , A clause has the characteristics of a

sentence which has to do with a subject and a verb. The only has two obvious differences that

2
exist between a sentence and a clause is that a sentence expresses complete thought; a clause

does not necessarily express complete thought as expressed in the following examples:

4. i. That the forecaster (N) predicted (V)

ii. Whenever Lucas (N) feels (V) unhappy

iii. Because the plane (N) departed (V)

Clauses form the largest grammatical unit in linguistic analysis. Nweya (2018), says the nature

and structure of clauses have attracted tremendeous research interest across languages. As stated

in the previous chapter, generally, a clause is fundamentally structured in three domains: the

Verb Phrase (VP) domain, the Tense Phrase (TP) domain and the Complementiser Phrase (CP)

Domain 2 (see Rizzi, 1997, Shlonsky, 2007, Radford, 2009 and Haegeman, 2012). Each domain

is associated with different functional categories. Contemporary studies show that the

distributions of functional categories in the clausal domains provide information about the

structure of the basic clause.

2.1.2 Main/Independent Clause

A clause that can stand on its own as a sentence is called independent or main clause. Every

sentence contains one or more independent clause(s); which include subject, verb, singleword

modifiers, and phrase modifiers. Main clause according to Jamz (2018) is a major clause in

English. It expresses a complete thought and can stand alone as a sentence. The main clause is

also known as an independent or alpha clause. This clause does not depend on another clause to

fully express itself as shown in example (5) below:

5.i. James’ sister teaches English at the college.

ii. Have you seen Sam’s ball?

3
In the same vein, Rosa and Eschhol (1994) say an independent clause is another name for a

simple sentence. An independent clause has a subject and a verb, expresses a complete thought,

and can stand alone as a complete sentence because it doesn't depend on anything else to

complete the thought. This they explained with the following examples as in (6):

6.i. Thomas Jefferson and Sally Hemmings lived in Paris for a time. subject verb

ii. Martha Jackson was a remarkable woman. Subject verb verb

iii. Socrates drank hemlock and died. subject verb

iv. The cat clawed its way to the top of the tree.

2.1.3 Subordinate Clause

This can be simply defined as a clause that cannot stand on its own in a sentence. Jowitt (2015)

sees it as the clause that describes the subject of the main clause. Umaru (2016), states that

subordinate clauses are categorized into three major types manely: noun clauses, adverbial

clauses and relative clauses.

2.1.3.1 Noun Clause

Noun clauses, he said may function as subject of a sentence, subject of a verd, object of a

preposition, complement of the subject, complement of the object as well as complement of the

preposition. Opega (2005), says noun clauses in English are introduced by such interrogative

words such as; what, how, whom, why, whether and whomever. The following examples of

English clauses explain his definition:

7.i. That she was accussed of theft was not suprising (subject of the sentence)

ii. He gave us a promise (subject of verb)

iii. I have no doubt in what he says (complement of the subject)

iv. Helping the poor is what he does (complement of subject)

4
v. Her father gave her a present (complement of object)

vi. we give honour to whoever deserves it (complement of preposition).

The underlined words specify different functions as indicated against each sentence in the above

examples (7i - vi).

A noun clause is a dependent clause. A noun clause can be a subject, a direct or indirect object,

or an object of a preposition. Noun clauses can begin with "wh- question" words (what, which,

when, where, who, whom) and question words like (how, if, that). The examples in (8) below

explain this:

8. Noun Noun Clause Subject

i. His whereabouts are unknown. Where he lives is unknown. Direct Object

ii. I don't know that man. I don't know who he is. Indirect Object

iii. The man gave the students the key. The man gave whoever wanted it the key. Object of Prep.

iv. He isn't interested in geometry. He isn't interested in what the class is studying.

A noun clause begins with a subordinator that connects the clause to the main clause. The

following are subordinators used to introduce noun clauses: how, where, what, whatever whose,

whosever that whether, if, who, whoever, which, whichever, when, why, whom and whomever

and so on.

It is interesting to note that the subordinator is in italics. The noun clause including the verb and

any helping verb is underlined. The laboratory aide reported that all the students had completed

the experiment. The students asked when the psychology reports were due. Mrs. Peterson asked

whether the secretaries had ordered the office supplies yet.

5
2.1.3.2 Adverbial Clause

Adverbial clauses on the other hand accoding to Reinking and Osten (2005), are elements that try

to answer questions in sentences. They modify verb, adverb, adjective and sentences thereby

providing answers to questions asked by ordinary adverbs. Quirk and Greenbaum (1973) identify

different types of adverbial clauses such as: clauses of time, clauses of cause, clauses of

condition and clauses of concession, clauses of reason, clauses of manner and clauses of

comparison and clauses of proportion and preference. According to Dryer (2011), adverbial

clauses are subordinate clauses in the sense that their occurrence usually depends on the main

clause. However, not all languages mark the distinction between dependent and independent

clauses in the same way. Certern devices have been identified to be commonly found among

languages:

(i) adverbial subordinators, either in the form of separate words such as before and where in

English or verbal suffixes as in Kiowa (a language in central United States)

(ii) special verb forms such as infinitives and participles as in English and Latin

(iii) special word order as in German where independent clauses usually uses verb-second word
order whereas subordinate clauses uasually use verb-final word order (Dryer, 2011; Thompson et
al., 2007).
9. à-dę̀·k’ɔ́·-àl hɔ́n àn à-dę̀·hę́·m-ô
1SG-lie-although NEG HAB 1SG-sleep-NEG
‘Although I lie down, I can’t fall asleep.’
(Dryer, 2011 from Watkins, 1984: 242)

10. Inflating her lungs, Fiona screamed.

11. Wir wohn-ten auf dem Lande, wie ich dir schon gesagt habe
we live-PAST on ART(DAT) land as I you already told have(SG)
‘We lived in the country, as I have already told you.’ (Thompson et al., 2007: 239)

6
Adverbial clauses may also be expressed in verbless form, as the English example in (12). Such

clauses typically do not specify in which way (temporally, causally, conditionally, etc.) they

modify the state of affairs described in the main clause, and thus are much more challenging for

interpretation than most prototypical adverbial clauses.

12. Alone in his room, she switched on the light.

Subordinators in the form of separate words (such as that and if in English) are the most

commonly way to mark adverbial clauses (Dryer, 2011). Yet the presence of a subordinator does

not guarantee that the relevant subordinate clause can be classified as an adverbial clause, or a

relative or complement clause. As illustrated in (13), the word that is used as an adverbial

subordinator in (13a), but a complementizer in (13b) and a relativizer in (13c):

13. a. He talked so fast that most people couldn’t follow.

b. He said that most people couldn’t follow.

c. The talk that most people couldn’t follow was given by a colleague of mine.

The sentences in (13) show that even in individual languages there may exist no formal

differences between the three major types of subordinate clause. Rather, we need to refer to their

function to determine the type: an adverbial clause is a modifier of the main clause and thus

optional; a relative clause is a modifier of a noun (phrase) and thus optional; a complement

clause is the core argument of a predicate and thus not omissible (Diessel, 2001).

2.1.3.3 Relative Clause

Relative clauses, which are characterized as a kind of noun phrase postmodifier, are generally

classified into different categories. Diessel and Tomasello (2000:132) defined a relative clause as

a “subordinate clause that modifies a noun or noun phrase in an associated main clause”.

7
In English grammar, relative clauses fall under the general area of modifiers which encompass

determiners, adverbs, adjectives, nouns, phrases and clauses (Ebest et al., 2000; Quirk &

Greenbaum, 2000; Leech & Svartvik, 2013). In most cases, grammatical elements such as

“articles” and “determiners” – a, an, the, few, a few, some, each, each other and one another –

are also added to the class of modifiers (Rao & Ado, 2010). Modifiers may be defined as “any

word or group of words used to describe a noun” (Sekyi-Baidoo, 2003:503); or as “words

(adjectives and adverbs) which help us clarify our ideas and make them more vivid” (Long,

2007: 647); or further as “determiners which operate as premodifiers in the structure of a noun

phrase” (Thakur, 2007:32). Within the group of modifiers, the relative clause component

presents challenges to language users in both the reading and writing processes (Gibson et al,

2013). Indeed, although second language users are able to recognize relative clauses in texts,

difficulties arise when they have to create or compose texts requiring this grammatical structure

especially in defining and non-defining contexts (Felser et al., 2003; Celce-Murcia & Larsen-

Freman, 1999; Frenck-Mestre & Pynte, 1997). The difficulties appear to relate to (a) deciding if

the relative clause is essential or nonessential, and applying the appropriate punctuation; and (b)

positioning the relative clause in relation to the sentence element that it modifies.

Umaru (2016), says relative clause is a sentence formation higher than the simple sentence with

one or more depending on the other to make meaning; it is also one of the options for post-

modifying a noun phrase. Kroeger (2005), states that a clause occurring within an NP as either a

complement or a modifier, modifying the clause is a kind of an adjunct referred to as relative

clause. A relative clause is formed ‘when one clause is embedded into an NP of another clause’.

Finegan (2000), opines that when the NP that has the same referent occurs in two clauses, a

relative clauses can be formed by embedding one into another. Reltive clause in English and

8
many other language, Aderson (2002), says as relative clause is usually a constituent of the noun

phrase whose head it modifies. A relative cclause is an adjective modifier subordinated to an NP-

noun phrase, with a pronoun inside the NP (head of the clause) he explained this with the

following examples:

.14. The man [who ate fruits] died a horrible death.

In the example (14) above, the clause ‘who ate fruits’ is embedded in the main clause; ‘the man

……… died a horrible death’. The embedded clause is actually a modifier of ‘the man’ and

the relative pronoun ‘who’ is co-referent with ‘the man’ and also serves as a subject of its own.

15. The man who(m) I saw was big.

The above example (15) shows that ‘the man’ is the subject of the main clause, while the

coreferent ‘who (m)’ is the object of ‘see’ in the subordinate clause

2.2 Theoretical Review

Theoretical Studies

This sub-section reviews the analysis of clause structure in Generative Grammar. It focuses on

the previous clausal architectures proposed by scholars based on crosslinguistic evidence.

2.2.1 Analysis of Clause Structure in Generative Grammar

Chomsky’s (1957) Syntactic Structures is a revolutionary work that has changed the

understanding of linguistics and marked the introduction of Generative grammar. The work also

revolutionised the cognitive approach to linguistic analysis now construed as modern linguistics

even though that was not the intention. The intention was to construct a grammar that can be

viewed as a device of some sort for producing the sentences of the language under analysis. The

outcome is expected to be a theory, which can be used with no reference to a particular language.

His ideas gave rise to the term ‘generative grammar’. Hence, Chomsky (1957) using Phrase

9
Structure Grammar (PSG) demonstrates how phrases and clauses can be derived using the

‘Phrase Structure rule (PSR) or re-write rules’. These PS rules are used to generate the basic

structure called ‘kernel sentences’ while transformations re-arrange the sentences in various

ways to generate other kinds of structures such as negative, passive and interrogative sentences.

One of such popular transformation is the affix hopping rule. In addition,

Chomsky emphasised the need for grammar to be simple. Hence, he made a distinction between

shorter and longer derivations. In this approach, Chomsky (1957: 4) represents clause structure

as shown below:

(16) Sentence NP + VP

Where the constituents NP can be re-written as [Det+N] and VP as [V+NP PP] down to the point

where only lexical items such as N, V, ADJ, ADV and DET terminates the string. In PSG, as

could be observed above, AUX is not represented as an independent constituent but as a

constituent of the verb. The VP is re-written thus (See Chomsky, 1957:39):

(17) Verb Aux + V (where Aux C(M) (have + en) (be + ing) (be + en).

These ideas were modified to a great extent in Chomsky (1965) where he introduced the notion

of competence and performance. One of the main proposals of the study is the Standard theory

(ST) of Transformational Generative Grammar. As it relates toclause structure, Chomsky, in this

work, recognises AUX as an independent functional category. This assumption led to the

modification of the structure of the basic clause and this modification is represented below:

(18) S NP AUX VP

According to him, S is the initial symbol of grammar representing the category, sentence. AUX

includes modals and verbal auxiliaries (Chomsky, 1965:64-8) while NP and VP capture the

subject and verb with their complements respectively. Research in syntactic analysis found this

10
representation defective. Hence, there was a need to modify or reformulate the idea in the

subsequent models. It was observed that AUX failed to capture other elements such as AGR, T,

Mood and ASP. These elements were assumed to be part of AUX. Consequently, AUX was

replaced with INFL to capture the fact that AUX is specific to modals and modal auxiliaries. In

other words, they are just parts of elements within INFL.

In the early 1980s, Chomsky (1981) introduced the Government and Binding (GB) model of

grammar in which he explains, with some modifications, some of his leading ideas in the works

he presented at the Pisa lectures of April 1979. Modifications to the study followed from the

contribution of other scholars prior to its publication which hinges on the shortcomings observed

in the work. In the study, he represents the structure of the basic clause as in (19) below:

(19) S NP INFL VP

Where INFL = [ Tense], subject to the idiosyncratic properties of the adjacent lexical items. The

rule represents the clause structure of languages where the subject of the sentence is obligatory

such as Germanic and African languages. But having observed that due to certain parameters of

UG, the subject NP could be optional in some languages 2, the rule was reformulated. Hence,

Chomsky revised the clause structure of such languages as shown below (see Chomsky,

1981:28):

(20) S (NP) INFL VP

This period witnessed the introduction of X-bar theory to make up for the shortcomings of PSR.

The x-bar theory was formalised in Jackendoff (1977). It holds that phrases are headed by lexical

items often identified as the lexical head. The lexical head is labeled X 0. One of the main

advantages of the theory is that it accounts for the intermediate projection which PSR failed to

11
account for. Scholars observed that there is a projection bigger than the lexical item but smaller

than the maximal projection (Radford, 2009).

This constituent is identified as an intermediate projection in syntactic analysis. Hence, X-bar

has three bar levels-the lexical, intermediate and maximal level. This is represented in the

schema below:

Fig. 4

XP

Spec X1

X1 WP (Adjunct)

X0 ZP (Complement)

(Ouhalla 1999:115, Black 1999:5, Nwaye 2018:9)

In (fig.4), XP is the maximal projection while X 0 is the lexical head. It shows that WP (adjuncts

expands XI to another XI while ZP (complements) expands XI into X0. It demonstrates the

closeness of the complement to the head than the adjunct. This period witnessed the tradition of

projecting clauses following the X-bar theory of phrase structure rule. The theory was successful

to a great extent since it was able to capture parametric variation cross-linguistically especially in

relation to word order. It also eliminated the redundancy observed in re-write rule (See Chomsky

1970).

2.2.2 The Minimalist Program

The most radical reformation of transformational generative grammar according to Onuoha in

Mbah (2016), has been the Minimalist program of Chomsky (1995). The idea of minimalist

12
program can be traced back to Chomsky’s 1986 Introduction of the principle of Full

Interpretation. Chomsky (1986a) as cited by Cook and Mark (2010:242) opined that the principle

of Full Interpretation states that every element in a structure must be interpreted in some way,

which means that there is no superfluous element in the structure of language. Minimalist

program is the most current model of generative grammar as proposed in Chomsky (1993, 1995,

etc.). Minimalist program uses certain instruments to operate such as select, merge, and move,

with other hypotheses that it operates with are the DP-hypothesis (Abney 1987), Split-Infl

hypothesis (Pollock 1989), VP-internal subject hypothesis, etc. these instruments and hypothesis

are used to enhance the workings of its various operations. The minimalist program (MP

henceforth), just as other theories that came much earlier in the generative traditional grammar

see language as a faculty in the human cognitive system which makes use of diversified but

connected operations networked around the basic principles of Universal Grammar. According

Ilori (2010), Universal Grammar represents fixed general principles that obtain in all natural (i.e.

human) languages, which differentiate them from artificial languages such as Basic, Prolog, etc.

found in mathematics and computing. Universal Grammar (UG henceforth) also consists of

parameters which are assumed to be subject to language specific or typological variations which

may make particular languages A, B, and C differ in some respect from one to the other. This

ideology underlaid the concept of principles and parameters right from Chomsky 1981 to 1993,

when the minimalist ideology took over. In the same line, Hendrick (2003) says minimalism is

an attempt to rethink many syntactic phenomena in the hopes of extracting greater insight by

questioning the explanatory role of representations.

The minimalist programme came into being because Noam Chomsky felt or discovered that grammars are
not as complex as they were made to be in earlier generative theories or approaches like Government-
Binding (GB) or Principles and Parameters theory (PPT), which were characterized by postulations of

13
various complex theories and sub-theories of connected structures and principles. Rather, grammars are
minimally complex systems of some perfect optimal design controlled by a man-specific genetically
endowed language faculty or acquisition device, which creates syntactic structures designed to perfectly
interface with other components of the mind namely the speech and thought systems (Chomsky 2001,
2002). GB tries to explain a universal grammar theory (UG), which describes all languages. It assumes
that a large portion of grammar of any particular language is common to all languages (Black, 1999).
Consequently, “UG rules are to be found in English, Berber, and indeed in any human language”
(Ouhalla, 1999, p. 7). Its view is that UG can be broken down into two main components: levels of
representation and a system of constraints. GB refers to two central sub-theories of the theory:
government, which is an abstract syntactic relation and binding, which deals with the referents of
pronouns and anaphors. It was the first theory to be based on the Principles and Parameters model of
language, which also underlies the later developments of the approach described in Chomsky’s recent
book The Minimalist Program (1995).
The fact that UG is minimally complex, according to Chomsky, is the reason for the tremendous

ability displayed by young children in acquiring language in a remarkably short period of time,

which in itself fulfils the learnability criterion of adequacy for any linguistic theory (Ilori

2010:21). The basic reason for this is to make easy and simple the descriptive instruments of

grammars especially for language analysts. This thought or orientation made the minimalist

grammar significantly different from other grammars that came earlier in the generative tradition

due to the fact that it makes use of very small and simple theoretical instruments for the

description of basic syntactic derivation and others that may be relatively complex ones e.g.

derivations via movement. Only two structural levels of linguistic representation (i.e. the two

interface levels of Logical Form (LF) and Phonetic Form (PF) are recognized in MP. This ruled

out the D-structure and S- structure levels of GB and PPT. Similarly, phrase structure rules with

their notion of ‘consist of’ inherent in earlier generative models are jettisoned in favour of

minimal computational operations namely select, merge, and move which are used for syntactic

processing within the computational system. In the same vein, the projection principle which

relates lexical properties to the syntactic levels of D- and S-structures alongside the X-bar
14
structure was also removed and replaced with bare phrase structures which disallow vacuous

projections in syntax.

In conclusion, the basic guiding principle of the MP is: minimize the theoretical and descriptive

apparatus of grammar by avoiding unnecessary complex formulations in

2.2.3 The Structure of Minimalist Grammar

Architecture of the Minimalist Program In MP, the human mind is seen as a complex computer

machine that processes language. Therefore, it is a representation of the computational system of

human languages (CHL). Okechukwu (2018), states that the basic computational process selects

lexical item tokens and builds them into a structure by successive application of merge. The

derivation splits at the point of spell-out where the phonetically relevant materials are sent to PF

and the grammatical and semantically relevant features sent to LF. Minimalist program or theory

is basically structured around three systems that are interwoven or linked to each other as stated

by Johnson (2010). These three systems are the lexicon, the computational system, and the two

interface system of output representations, i.e. the semantic/logical forms (LF) component and

phonetic/PF component. These are assumed to be the primitive substantives of the grammar of

every natural language, at least within the context of MP. To form a clause within this grammar,

relevant word items are selected from the lexicon and merged or combined in a pair wise fashion

to form larger syntactic units or linguistic expressions which serve as inputs to the two interface

output components of LF and PF. The semantic component maps/converts the derived linguistic

expression into corresponding semantic form/representation, i.e. the output of grammar at the

meaning end, while the PF component produces the phonetic spell-out for the derived expression

by mapping them into a Phonetic Form representation which is the output of grammar at the

sound end. Derived linguistic expressions are said to converge or crash at LF and PF based on

15
the formal/grammatical features that each of the word items used in the derivation has which

must be in harmony with those of others. Spell-out is the stage or point where semantic and

phonetic features are separated for interpretation at PF and LF. Chomsky (1995:219) puts this

formally by saying,

… take L to be a generative procedure that constructs pairs (p , l)

that are interpreted at the articulatory-perceptual (A-P) and
conceptual-intentional (C-I) interfaces respectively as “instructions”
to the performance system. p is a PF representation and l and LF
representation, each consisting of “Legitimate objects” that can
receive interpretation …
This invariably means that the Lexicon feeds the computation component with word items and

relevant syntactic information about them by specifying their formal features/syntactic

properties. The computation system processes the information by selecting and merging the word

items appropriately to construct well-formed linguistic expressions which are then moved into

the working area to feed the two output interfaces (i.e. PF and LF) as specific instructions for

spell-out and interpretation. It is at this point that the formal/grammatical features of words in

any syntactic derivation are checked against one another. While any syntactic derivation whose

features check appropriately converges at PF and LF, structures where features of items fail to

check as they should simply crash at these twin output interfaces.

16
Fig. 5
Linear Order (LCA),
Spell-Out of Occurrences/Copies

TransferPF SM-Interface
(Spell-Out) (PF-Interface

Transfer LF CI – Interface
(LF - Interface )
Merge:
Internal merge Quantifier scope
External merge Reconstruction

Lexicon Colins (2016:24)

The above schema illustrates that the basic computational process takes selected lexical items

(LIs) with their peculiar properties (SEM-F, SYN-F, PHON-F) from the lexical items and

builds them into a structure by a succession of merge operations. Merge is external when LIs are

new in the derivation and internal when any of the items is already in the derivation and undergo

scrambling/displacement via head movement, subject movement, cyclic movement and/or

remnant movement to form a syntactic object (SO). At some point during computation, the

derivation splits into two where the phonetically relevant materials are transferred to SM/PF

interface and the grammatical and semantically relevant features are transferred to CI/LF

interface. In this model, there are two points of transfer; TransferPF and TransferLF which are

assumed to be separate operations taking place at the same point. TransferPF is equivalent to

Spell-Out. Nwaya (2018) identifies as (i) ordering of terminals for appropriate interpretation

following the principle of the linear correspondence axiom (LCA) (ii) and spelling out of

occurrences/copies. The rule for spelling out copies says that only the last created occurrence is

spelled-out. Another process that might take place at TransferPF is tonal modification as it

17
concerns languages like Igbo. At the TransferLF, part(s) of the occurrences are deleted to create

interpretable structures (i.e. structures that have meaning). Occurrences enable one to determine

the scope of quantifiers and source of moved items for the purpose of reconstruction. The

computational procedure continues to apply similar processes to build fully formed structure at

the PF and LF interfaces. At these points, the syntactic objects are presented for appropriate

interpretation. If the PF is phonetically interpretable and the LF semantically interpretable, the

derivation converges, otherwise, it crashes. In the earlier versions of the program, there are

economy motivated principles proposed to minimise the cost of derivations and make derived

syntactic representations as simple as possible.

2.2.3.1 The Lexicon

The lexicon is said to be a mental dictionary, so to say, where all possible word items of L are

listed in the minimalist grammar. Collins and Stabler (2016: 43) define it as “a finite set of

lexical items.” It is constructed on the basis of the set of features available in UG to enable them

to project fully in the computation process. The minimalist lexicon specifies the grammatical

properties (i.e. phonological, semantic and syntactic properties) which are characterized in terms

of formal features (i.e. features which play some kind of roles in morphology/syntax) for each

word item on its list. This is the alternative put in place in the MP to replace the concept of

structural case assignment. In other words, lexical items already have their case properties,

among other features, right from the lexicon. Therefore, lexical items in the MP are complexes of

features listed in the lexicon (Chomsky 1995b:243). Three major sets of grammatical/formal

features are carried by word items in the minimalist grammar, namely head-features, specifier-

features, and complement-features. Head-features provide information on the inherent

grammatical properties of a word item. These include its categorial designation e.g. N/D, V, T,

18
etc., case form e.g. Nom, Acc., etc., phi-features e.g. number, person, gender, agreement, and

other morphological/inflectional trappings that contribute to the syntax of such item. Specifier

features indicate the likely kind of specifier that such item allows just as complement features

supply relevant information on the possible kinds of complement it selects. All of these features

are checked in the process of computation and only linguistic expressions whose features check

properly can converge at LF and PF. Uninterpretable features, i.e. features with no semantic

content with which they can make contribution to the meaning of an expression, are eliminated at

LF. By implication, only interpretable features survive computation.

2.2.3.2 The Computational System

The MP concept of computation is that which has to interface successfully with other faculties

that impose their own restrictions on what goes on in the language faculty, known as Bare

Output Condition (Cook and Newson 2010:248). Computational system in minimalist program is

a system that relates sound form and meaning using minimal operations to construct a linguistic

relationship between the two. In order to construct a well-formed linguistic relationship between

sound and meaning, the computation system makes sure that the linguistic expression (form and

meaning) of language satisfies output conditions at PF and LF. This is actualized by making sure

that form and meaning, are compatible, i.e. their derivation is based on the same lexical choices.

Therefore, MP sees computation

Fig. 6
Sensorimonitor Phonetic computations Logical conceptual-intentional
System Form system Form system
(Cook & Newson 2010:248)

19
One vital implication of this is that, in the process of computation, CHL can access a particular

lexical item more than once to generate syntactic objects depending on the structure intended.

However, separate applications of operation select on an item must be distinguished at LF.

Basically, the CHL operation is recursive. It constructs syntactic objects, not only from two new

lexical items, but also by merging new items with already formed syntactic objects. Syntactic

objects are assumed to be some kind of re-arrangement of the properties of the lexical items used

to construct them. Three minimal operations are used by the computational system to achieve its

mandate; these operations are select, merge, and move.

2.2.3.4 Operation Select

This is a procedure that picks out word items from the lexical resources of L and introduces them

into a derivation. Items are not selected together at the same time but one after the other with

each serving as a syntactic object in the derivation. However, already constructed derivation is

also seen as a syntactic object to which another lexical choice can be merged. In the MP, it is

basically assumed that operation select is the starting point of derivation while the numeration

(N) is regarded as a pre-computational operation. Chomsky (2015: 208) notes that select is an

operation of the CHL and a procedure that selects LI from the numeration, reducing its index by

1, and introduces it into the derivation as SOn+1. While a numeration is a set of pairs (LI, i),

where LI is an item of the lexicon and i is its index, understood to be the number of times that LI

is selected. In other words, this operation involves the initial choice of the numeration of lexical

items needed for a particular type of construction from the lexicon. Each numeration consists of

the set of words to be used in a derivation and an indication of the number of times each word is

to be used (Al-Mutairi 2014:38).

20
2.2.3.5 Operation Merge

Merge is the most significant syntactic process that takes two linguistic items (LIs) α and ß, and

combines them into a set of (α,ß). Wales (2002), said that within the Minimalist Program, syntax

is derivational, and Merge is the structure-building operation. Merge is believed to have certain

formal properties constraining syntactic structure, and is implemented with specific mechanisms.

This invariably means a process where two syntactic objects are combined to form a new syntactic

unit (a set). It is an operation that combines two separately lexical items which are selected to

construct a new single unit of syntactic object. For example, objects α and ß can be combined

using merge to derive or form a new unit of syntactic object K in an event that K is fixed for the

set {α, ß}. In other words, K must have the form {γ, {α, ß}} where γ is a type of K. Chomsky

(1995b:243) puts this concept formally as:

K = {γ, {α, ß}} where α, ß are objects

and γ is the label of K.

This means that α and ß are the constituents of K or that K is constituted from the pair (α, ß).

However, γ is a label type of K which may turn out to be either α or ß depending on the one that

projects out of the two.

This is illustrated with the structural schema in 7a and b.

Fig. 7 a2 ß2

ß α1 ß1 α

In (7a), a projects as the item that determines the label of the new syntactic object a2.

However, b projects in (7b) to determine the label b2. The item that projects is regarded as the

head of the new construct.

21
In conclusion, therefore, merge is an operation that takes a pair of syntactic objects, SOi : SOj,

and replaces them by a new combined syntactic object, SOij

(Chomsky (1995:226).

2.2.3.6 Operation Move

Movement in the MP is radically different from the general concept of move a in other earlier

models of generative grammar (e.g. G-B and PPT). In the latter, move a is an operator that moves

any item anywhere subject to bounding or island constraints (Ross 1967) and other conditions

like government and the empty category principle (ECP). On the other hand, movement in the

MP is primarily motivated by feature checking which is subject to some minimal economy

conditions, e.g. the Minimal Link Condition (MLC). Basically, two movement operations are

assumed in the MP. These are Move- F(eature) otherwise known as the feature-checking theory

of movement, and O(perator) movement which incorporates A(rgument) and A-bar movements.

These operations are sketched out in the following sub-sections.

2.2.3.6.1 Move-F

This is a computational system internal movement mechanism which is triggered only by the

need to check the formal features of syntactic objects selected and merged for the processing of

syntactic derivation. Lexical items are assumed to move along with their features, which of

course are the primary target of Move-F, in order to check off those features against those of the

heads of target projections. As earlier mentioned, MP assumes that lexical items carry three sets

of grammatical features, namely head-features, which determine the intrinsic grammatical

properties (e.g. categorial feature) by which an item could be identified as different from every

other item; complement features which indicate the kinds of complements an item will select;

and specifier features which describe the kind of specifier that a syntactic object head takes.

22
Generally, features are assumed to be morphological, i.e. they are somewhat inflectional in

nature, and are described relatively in terms of strength, such that a particular feature of a

particular item may be strong or weak. For instance, categorial features are assumed to be strong.

Similarly, some of the formal features are assumed interpretable while others are not

interpretable, especially at LF. Interpretable features, which include categorial- and phifeatures,

are those that have semantic content by virtue of which they contribute to the meaning and

overall interpretation of the structural derivation. Uninterpretable features, such as specifier- and

complement-features, are the opposite. Essentially, a derivation that will converge at LF must

contain only the (semantically) interpretable features. Any uninterpretable feature present at LF

will make the derivation crash. In a nutshell, every convergent derivation must be made up of

syntactic objects that carry compatible features. Any incompatible feature disrupts structural

harmony, blocks convergence, and causes the resultant phrase/clause to crash. In order to ensure

compatibility of features carried by lexical items, the computation system checks their features

against one another by invoking the features-checking movement operation, which raises just the

F(eature) alone. However, in the process of raising F, the item carrying F is pied-piped to F as a

kind of minimal extra baggage required for convergence. As a minimal economy requirement,

especially for overt movement, F has the mandate to carry along just enough material needed for

convergence. Chomsky (1995b:226) gives a vivid description of how this operation works when

he said I assume, then, that the operation move raises F and derivatively raises FF (F) as well,

carrying along a phrasecontaining F only when the movement is overt,

as required for convergence… Thus raising without

pied-piping is more economical in some natural

sense, but that is irrelevant if the derivation does not converge …

23
if all features of some category a have been checked, then a is

inaccessible to movement, whether it is a head or some projection.

But if some feature F is as yet unchecked, a is free to move. Economy

conditions exclude ‘extra’ moves and anything more than the minimal

pied-piping required for convergence. In covert movement, features

raise alone.

There are two feature-checking configurations in the MP. These are the Spec- Head and Head-

Head checking relations. The Spec-Head feature-checking relation concerns the specifier feature

of a head which attracts the feature of another syntactic object from its original theta-marked

position into Spec-Head for feature-checking purpose. In the process, the specifier feature of the

head is checked by the moved item, and the case feature of the moved item is in turn checked by

the head. A good illustration of this configuration could be seen in the checking relation between

subjects of clauses and I heads of IPs. The subject is attracted from its VP-internal q-position to

Spec-I where it checks the spec-feature of I, and I in turn checks the nominative case-feature of

the moved subject, as shown in (8).

Fig.8.

XP

SPEC X’

X YP

In (8), the Spec-Head checking relation is mutual between the head and its specifier. It is this

same configuration that takes care of complement-feature checking of objects against V heads in

24
an operation in which the object of V moves to Spec VP to check the complement feature of V

and V in turn checks the accusative case-feature of the raised object.

On the other hand, Head-Head movement configuration does not involve specifiers or

complements. Rather it is an operation that moves one head item into another head position

simply because the item in the latter position possesses an unchecked feature that attracts the

unchecked feature of the former. This is used to take care of X0 categories only (i.e. N, V, D, I,

C, etc.), both lexical and functional. Two good examples of the Head-Head movement are found

in English I to C and V to I movements. For instance, the interrogative feature of C is assumed to

be strong in English, so it often attracts aux items from I(nfl) into COMP to derive polar

interrogative expressions like

(21).

[CP [C willi ] [IP You [I ti] [VP marry me ] ] ]13

The framework of the Head-Head configuration is informally represented as in (9).

Fig.9. XP

SPEC X’

X YP

t Y

Y W

25
Finally, it is imperative to note that grammatical features are checked in the course of every

computational derivation. In the process, all uninterpretable features are erased once checked

because they play no role in the interpretation of syntactic derivation since they have no semantic

content. This leaves only the interpretable features surviving the corresponding LF

representation. Phonetic features are assumed to be processed separately by the PF component,

hence they by no means whatsoever input into the LF component. As earlier hinted, the

uninterpretable features are specifier-, complement- , and case-features while the interpretable

are the phi-, tense-, and categorial- features. Radford (1997b:72) put these move succinctly:

If there is a match between checker and checked in respect of any given feature, the relevant
specifier- or complementfeature is erased (because specifier- and complementfeatures are
uninterpretable), and the corresponding headfeature is erased if it is uninterpretable (but is not
erased if interpretable). If there is a mismatch between checker and checked in respect of some
feature, the relevant feature cannot be erased from either.
It is also of great essence to state that elements with strong features (e.g. case) apparently move

to get those strong features eliminated before spell-out. Although, such apparent movement is

costly because it violates procrastinate, they are inevitable for convergence.

2.2.3.6.2 Economy Conditions on F-Movement

Study shows that initially based (Chomsky 1993), three general economy conditions were put in

place to guide movement in the feature-checking configuration. These are Greed, Procrastinate,

and Last Resort. Greed specifies that the only reason why constituents (syntactic objects) move

is to satisfy their own morphological requirement of checking own features only. Procrastinate is

a condition put in place to prohibit overt movement, as overt movement is assumed to be costly

while covert movement is cost free. Last Resort assumes movement operations are driven by the

necessity that the item involved must be moved otherwise computational derivation will not

26
converge. In other words, some morphological necessities make it imperative for certain features

to be checked in the checking domain of some heads, thereby forcing such features to move

along with the items carrying them to the checking domains. If not, the intended derivation will

fail to converge. In actual sense, Chomsky (1993) sees Greed as a self-serving Last Resort.

However, probably for the obvious reason of similarities in Greed and Last Resort, Chomsky

(1995) reformulates the feature-checking movement as Attract. Attract, which subsumes Greed,

Last Resort, and the minimal link condition, is therefore conceived as a movement of a set of

features from one category position to another given minimal closeness of the extraction and

landing sites, and feature compatibility between the head and the feature attracted. Chomsky

(1995b:297) codes Attract as:

K attracts F if F is the closest feature that can enter into

a checking relation with a sub label of K.
While the Last Resort component of Attract is evident in the fact that elements move only when
attracted by features of the target that must be checked, Minimal Link Condition (MLC) in
Attract is formally coded as;
K attracts a only if there is no b, b closer
to K than a, such that K attracts b.
(Chomsky 1995:311)
Generative procedures that involve movement are constrained by economy conditions such that

only minimal rules and conditions are required for linguistic computation. The economy

conditions discussed in the sub-sections below are last resort, shortest move and minimal link

condition (MLC). However, it is worthy to note that Greed and procrastinate have been re-

interpreted in terms of last resort. Collins (2001:50) points out that linguists have realised that it

is not in all cases that movement is motivated by the need to satisfy the morphological properties

of the moving element and movement is not often blocked by greed. More so, with the

27
introduction of phase, it is assumed that all movement operations must occur before spell-out

(Cook and Newson 2007). Hence, greed and procrastinate are not relevant for convergence.

2.2.3.7 Last Resort

This economy principle supposes that operations apply only if the derivation would otherwise
result in an ungrammatical representation at LF or PF due to lack of full interpretation. It
demands that every computational operation must serve a grammatical purpose. Chomsky
(2015:257) formalises last resort as follows:
Move F raises F to target K only if F enters into a checking relation with a sublabel of K. In this
regard, SO moves only if it is attracted by a feature of another SO that needs to be checked.
Therefore, displacement operations come into play to eliminate [-interpretable] feature to avoid a
crash. Other instances of operations that are subject tolast resort include the spelling out of
resumptive elements and do-support (Collins,2001:46). The former is applicable to Igbo-type
languages while the latter is applicableto English-type languages

2.2.3.8 Shortest Move and Minimal Link Condition

Shortest Move is an economy condition that requires all movement to be to the nearest possible

position in the computation procedure. Cook and Newson (2007:247) note that the range of

possible positions is determined by the properties of the moved element. Hence, a constituent

must move from its source position to the next hierarchically closest position of the right hand in

an upward movement direction. This is captured by the generalisation that movement to a pecific

kind of landing site does not skip landing sites of the same type. Generally, the principle favours

shorter movement over longer ones. It currently includes decision about which constituent is to

move when more than one position is available, in which case, it is canonically the closest item

to the landing site. (cf. Thráinsson, 2001:142 & Crystal, 2008: 435). Shortest move is not quite

different from Minimal Link Condition (MLC henceforth). MLC requires that the links between

elements in a movement chain be kept to a minimum. Chomsky (2015: 273) suggests that

28
formulating MLC is more natural if movement is reinterpreted as‘attraction’ or ‘Attract’ for short

where an element K attracts F for the purpose of feature checking. Chomsky (2015:.271)

captures MLC as follows:

Minimal Link Condition

…at a given stage of a derivation, a longer

link from α to K cannot be formed if there
is a shorter legitimate link from β to K.

This implies that MLC does not allow long movements where shorter movements are possible.

In the first instance, it seems that MLC and Shortest move are in conflict because the shorter the

movement, the more movement needed to cover the distance. A plausible solution is to posit that

movement chains are added to the structure as a single element in one derivational step to avoid

lengthening of the derivation by MLC. In MP, it is assumed that MLC captures the Relativized

Minimality (Rizzi, 1990) of GB. Hence, displacement operations triggered by Attract are subject

to MLC.

2.2.3.9 O (perator) and A(rgument) Movement

The term operator in the MP denotes word items that possess a syntactic property which triggers

inversion for interrogation or some kind of nominal internal negation. These include wh-items

and negative polarity items which are themselves called operator expressions. Operator

movement therefore is an operation that moves an operator expression into the specifier position

within CP driven by the operator feature (Chomsky 1995b:325). When it applies to expressions

that contain wh-operator, it is more specifically referred to as wh-movement, which derives

content-word questions (Radford 1997b:130-132).

29
According to Chomsky, the processing of this kind of movement is not at par with the feature-

checking movement, as raised operator expressions land in Spec-CP in a streak of movement that

appears superficially long distant. However, this movement does not violate the Minimal Link

Condition/Shortest-move of Attract since spec-CP is the smallest/shortest minimal/possible link

with which the operator expression can form a chain. The idea then is: move the smallest

constituent possible the smallest

2.2.4 Spell-out, Convergence, and the Interface

2.2.4.1 Spell-out

Spell-out according Ilori (2010), is a point/stage in the course of computation, after the

application of merge, where PF and LF representations of syntactic derivations are separated. At

that point, phonetic and semantic features are separated for processing by separate components

of the grammar i.e. the phonological component and the semantic component. The computation

operation at spell-out simply strips derivations of elements relevant to PF (p), leaving the

remaining elements, which are semantic, to be mapped directly to LF (l) by the ‘covert

component’. The stripped phonological elements are then mapped to PF (p) by the phonological

component. Generally, spell-out is the point at which phonologically relevant materials are

transferred to the SM or LF interface for appropriate interpretation. Chomsky (2005a:16)

describes it as transfer to the sound interface. At some point in the computational procedure, a

derivation is split into two and transferred to the interface levels via different operations:

TransferPF and TransferLF. TransferPF is equivalent to spell-out. In the recent Minimalist model, it

is a distinct operation from TransferLF where grammatically relevant materials are sent. With

regard to this, Hornstein, Nunes and Grohmann (2005:46) point out that the phrase markers that

feed LF and PF interfaces are structurally different, though they share common derivational

30
history; therefore, the computation must split. According to Cook and Newson (2007), it enables

the derivation to be spelled out in its outward physical form. Spell-Out, therefore, somehow

looks like an overt syntactic component of MP which appears analogous, but not identical, to the

S-structure of GB and PPT.

In essence, the computation operation runs on line directly to LF (i.e. N à l). Spell-out applies at

some point to sieve out the phonological elements and map them to PF (p) via the phonological

component, while the mapping of the N(umeration) to LF (l) continues via the semantic

component. Syntactic operations before spell-out are overt, but immediately after spell-out, all

operations must be covert. In addition to these, spell-out can apply anywhere in the course of

computation simply because it is not constrained by any rule, and a derivation can crash if any

wrong choice is made in the course of computation (Chomsky 1995b:189, Napoli 1996:391).

Practically, when an item is phonetically silent in the speech system output, such is said to have a

null spell-out or simply a null phonetic form (Radford 2004:358). In other words, it is possible

for an item to be syntactically and semantically present in the output of a syntactic derivation

without being phonetically realized.

2.2.4.2 Convergence

After spell-out, a given derivation must converge at the two interface levels of LF and PF. Any

derivation that fails to converge successfully at these two interfaces is said to have crashed. For a

derivation to converge, it must satisfy the principle of Full Interpretation (FI), which

… specifies that the representation of an expression

must contain all and only those features which are
relevant to determining its interpretation at the
relevant level.
(Radford 1997b:261)

31
In other words, the PF (p) and LF (l) representations of such derivation must contain ‘legitimate

objects’ only, where legitimate objects are those that carry the feature [+ interpretable].

Consequently, for a derivation to converge at LF, its LF representation must contain only

semantically interpretable features i.e. those features that determine the meaning of the

expression. Similarly its PF representation must also contain only those features, which

determine its pronunciation. When any of these requirements is breached in any structure, such a

structure simply fails to converge. A derivation converges fully only if it successfully converges

at the two interface levels. Convergence is determined by independent inspection for feature

compatibility among syntactic objects at the interface. Any feature mismatch in a syntactic

derivation makes such derivation crash. Finally, convergent derivations must also satisfy the

natural economy conditions e.g minimal link (MLC) and shortest move. Less economical

computations are blocked even when they converge, since it is not all convergent derivations that

become admissible at the end of the day. Chomsky (1995b:220) explained this:

The language L thus generates three relevant sets of computations:

The set D of derivations, a subset

of DC of convergent derivations of D, and a subset
DA of admissible derivations of D. FI determines
DC, and the economy conditions select DA.
(Chomsky (1995b:220)
This implies that it is not all sets of SDs that converge or become DC. Similarly, only a subset of

DC become admissible (DA), and while the principle of full interpretation determines which D

becomes DC, only the economy conditions select the DC that eventually becomes DA 14.

2.3 The Bare Phrase Structure Theory

Right from Chomsky (1981), the X-bar phrase structure model had been the standard for

syntactic analysis. X-bar was primarily motivated by the works of Zellig Harris (1951) and that

32
of Ray Jackendoff (1977), which argue for the recognition of an intermediate category assumed

to be higher than the lexical category but lower than the phrasal category. The structural

projection of X-bar is one in which every X (where X is a lexical head variable) projects through

a mandatory intermediate X’ en-route to the maximal XP category. Irrespective of whether X has

a complement, a specifier, or even an adjunct, these three positions are assumed to be structurally

present in every syntactic projection of X. As a result, an X projection without a filled specifier

or complement position is regarded as a non-branching projection (Radford 1988:177). The

schema in

(1) illustrates the X-bar phrase structure.

fig. 10. XP

Spec X’

X’ Adjunct

X Complement

However, beginning from Chomsky (1993), attention began to shift to the minimalist model of

phrase structure commonly called the bare phrase structure, which various scholars have referred

to as a modified version of X-bar. Although the Bare phrase structure is X-bar related, the basic

assumption underlying it is a bit different, as every phrase structure contains bare outputs which

disallow non-existent and superfluous projections in syntax. The theory retains the concept of

minimal, maximal, and intermediate projections of X-bar but defines their identities within the

inclusiveness condition, such that minimal and maximal categories are not identified by any

special markings or properties inherent to them. Instead, they are defined in terms of their

33
derivational and relational properties with other items in the structure being represented.

Accordingly,

a category that does not project any further is

a maximal projection XP, and one that is not a

projection at all is a minimal projection Xmin;

any other is an X’, invisible at the interface and

for computation.

(Chomsky 1995b:242)

This implies the following;

i. A minimal category x0 is any element or node that does not dominate any node though it can

project an XP.

ii. A node/element not dominated by any other node of the same label is a maximal category.

iii. The intermediate projection X’ is not automatic but optionally derived only when it is

required. In such a case, it must not be the maximal projection of X, but must combine with a

specifier node to form another projection of X, say X’ or XP.

In other words, it is possible for a single head to project two (or more) specifiers depending on

its feature composition. However, there must be a ‘feature-mediated’ relation between such a

head and its specifier before such a specifier can be licensed. The multiple specifier structure is

represented in (Fig. 11) following

34
Chomsky (1995a:432).

Fig. 11.
XP

Spec1 X’

Spec2 X'

H(ead) Complement

The specifier positions in (11) are not inherently stipulated for the H but derivable only when the

enabling feature composition of H allows them.

2.3.1 Key Assumptions

There are five major assumptions that have been incorporated into the minimalist program, and

these assumptions have become vital to the general understanding of its structure. These are the

VP-Internal-Subject-, Split-VP-, Split-Infl-, Split-CP-, and the DP- hypotheses. These

hypotheses are crucial to the MP because they directly influence, in one way or the other, its

theoretical outlook on issues relating to word order, nature and content of predicate structure, and

syntactic headedness in structural derivations. We present these assumptions, one after the other,

in the following subsections.

2.3.2 The Split-Infl Hypothesis

This hypothesis has its origin in the work of Pollock (1989). It claims that there is not just one

auxiliary position in the Infl phrase but two, namely T(ense) and Agr(eement). Prior to Pollock

(1989), the standard practice in generative grammar, based on Chomsky (1981, 1986), was that

the Infl category projects and heads the Infl phrase (IP= sentence) just as every other X0

category projects and heads its resultant XP. Structurally, the Infl category was believed to house

35
categories ranging from Agreement, Tense, Modal, Aspect and Neg(ator) traditionally called

auxiliaries, and all of them are structurally fused as one single/unit projection of the Infl

constituent. Pollock challenged this traditional claim and argued for the recognition of

Agreement and Tense as two independent functional heads in Infl, which possess the structural

ability to project their own phrases, namely AgrP and TP, in line with the X-bar structure.

Chomsky (1991, 1993) modified Pollock theory by arguing for two Agr(eement) Phrase

projections instead of one, such that AgrSP (i.e. Agreement subject phrase) and AgrOP (i.e.

Agreement Object Phrase) replaced the original AgrP of Pollock. Although Chomsky (1995b)

subsequently argued for the elimination of Agreement projections in clause structure on the

premise that they are bundles of phi-features which already form part of the formal features of

word items in the lexicon, the principle underlying the splitinfl hypothesis has continued to wield

considerable influence in generative grammar. For instance, the phrase projection scope has been

widened to accommodate NegP (Negator Phrase) and AspP (Aspect Phrase), which are phrasal

projections of former Aux items16. In line with this principle, Oduntan (2000) argued for NegP,

while Taiwo (2005) also recognizes a TaspP (Tense and Aspect Phrase) projection for Yorùbá.

2.3.2.1 The VP-Internal Subject Hypothesis

The standard assumption and practice right from Government-Binding (Chomsky 1981) up to

principles and parameters (Chomsky 1986, 1991) is that subjects of clauses occupy the Spec

position within IP and remain there because they are base-generated in that position. This implies

that Spec-IP was a q-marked argument position in the clausal architecture.

However, various scholars beginning with Koopman & Sportiche (1985, 1990), Kitagawa

(1986), and Kuroda (1988) among others, have since argued against this claim. Instead of the

traditional position, they favour a VP-internal generation of subjects. The core of their claim is

36
that subjects of clauses which normally occupy Spec-IP are not primarily generated there.

Rather, they are base-generated in a VP internally thetamarked position before being raised to

Spec-IP. That VP-internal position is assumed to be Spec-VP, and it is from there that the feature

checking configuration raises subjects to Spec-IP. Since then, much conceptual and empirical

evidence have been presented by various scholars in support of the VP-internal subject theory.

Such arguments include the case of floating quantifiers in French and English (Koopman &

Sportiche 1990), and facts from VP coordination (Burton & Grimshaw 1992) and the syntax of

clit

2.3.2.2 The Split Verb Phrase Hypothesis

This hypothesis which is also called the VP-shell or Light-verb analysis is advanced in

Chomsky (1995b). It holds that VPs canonically have a complex structure comprising an inner

core VP headed by a lexical verb and an outer vp shell headed by a strong null (Ø) light verb to

which the lexical V head of the inner/core VP adjoins when raised into vp to lexicalize v. The

Split VP hypothesis originated from the work of Larson (1988) which proposed a structural

analysis that conflicts with the standard analysis of dative constructions. According to Larson, it

is an implementation of a proposal suggested by Chomsky (1955, 1975). Using dative

complementation constructions, Larson demonstrated that the verb and its indirect object make

up a constituent that excludes the direct object as shown below.

(22) a. John [VP a letter [V’ send to Mary]]

b. John send [VP a letter [V’ t to Mary]]
In (22a) a letter is the subject while (to) Mary is the object within the inner VP. In (22b) the

structure is obscured by an operation of V raising. In the literature, this has become known as

Split VP or the VP shell hypothesis. In this proposal, it is assumed that external arguments

37
originate within the outer ʋP shell while internal arguments originate within the inner core. This

is captured in the schema below:

Fig. 12. VP

Exg. Arg. V1

V VP

DO V1
V IO

In the schema above, there are two VP shells, one headed by an empty head and the other headed

by the lexical verb. The empty ʋ is a place holder created following X-bar assumptions to enable

the lexical V move to the empty position and discharge its external θ-role and satisfy VPISH (see

also Hornstein, Nunes & Grohmann, 2005 Radford, 2009). Chomsky (1995) developed the idea

further by suggesting that the outer shell is not projected from an empty node as suggested by

Larson (1988: 382- 384), but rather, from a phonetically null ‘light verb’, whose meaning

depends largely on that of its complement. However, the proposal generally suggests that VPs

have complex structures and should be split into two distinct projections: an outer ʋP shell

headed by a null light verb and an inner core VP headed by the lexical verb. The emergence of

Split VP analysis solved the puzzles associated with analysing double object constructions

(DOC) across languages. In view of this, DOCs including applicative constructions shall be

analysed using the Split-VP hypothesis.

2.3.2.3 The Split Complementiser Phrase Hypothesis

With insight from Pollock (1989), the split projection hypothesis was also extended to the C-

system. This idea, championed by Rizzi (1997) and further developed in Rizzi (2001, 2003, &

38
2013), suggests a C(omplementiser Phrase)-system that is more articulated in structure, where

each of the elements traditionally associated with the CPlayer (within the GB-framework)

becomes the head of a maximal projection. This is to accommodate the complementiser and

other items that occur at the left edge of the sentence also called the left periphery. With regard

to this, Rizzi (1997) assumes that each of the force, topic, focus, and finiteness features that

appear in the left-periphery is the head of a functional projection ForceP, TopP, FocP and FinP,

which projects within the C-system. Rizzi (1997: 297) presents this proposal as shown in the

schema below:

Fig. 13.

Force P

force
TopP*

Top*
FocP
Foc
Top P*
Top*
FinP

Fin

In this structure, each of the elements that manifests in the left-periphery projects the Csystem.

The asterisks that appear on the right side of topic indicate the recursive nature of topics in Italia

and are called kleene Star (see Crystal 2008:42). Rizzi argues that this structure could be used to

account for the ordering constraints involving the elements of the C-system. Insight from this

39
proposal has attracted scholarly research into the expanded or articulated left periphery across

languages. These studies provide pieces of evidence in support of this Hypothesis (See also

Haegeman, 2012.) However, not all languages’ CP has the expanded CP as the one proposed by

Rizzi in terms of features and their hierarchy. For instance, Abraham (1997: 39) cited in Jayeola

(2016: 42) argues that there is TopP or FocP between CP and IP. Similarly, Olaogun (2016)

argues that FocP dominates InterP in Ǹko̩ ̀ -kóo. Nevertheless, this study relies on the insight

from this hypothesis in studying the elements that appear in the left-periphery of Igbo clauses

which is part of the main idea pursued in this study.

2.3.2.3 The Determiner Phrase Hypothesis

DP hypothesis originally DP-analysis follows from a proposal by Abney (1987) within the GB

framework that NPs are headed by a functional element, D, identified with the determiner in a ay

that the structure of the noun phrase parallels that of the sentence which is headed by the INFL.

Abney provided theoretical and empirical argument for the idea that a functional category DP, is

the maximal category projected by the class of determiner elements and heads the noun phrase.

In other words, all definite expressions including those that do not have overt determiner have

the status of DPs. DP represents the extended and the maximal projection of the lexical head, the

determiner. Therefore, nominals such as the governor and proper name like Joseph share the

same status as DPs. The D is lexicalised by determiners, articles, pronouns, quantifiers, and

demonstratives. In some cases, the determiner may be null. For instance, proper names may not

occur with overt determiners as demonstrated below:

40
Fig.14. DP DP DP

D N D N Spec D1
The Joseph Ø Joseph D NP
The Spec N1

N
Joseph
According to Bernstein (2001), the proposal resolved the problems posed for X I theory by the

traditional characterisation of NPs and unified the treatment of noun phrases and clauses. In this

regard, this study assumes that Igbo NPs are complement of DPs so as to conform to the tenets of

MP and for ease of analysis. This assumption is captured in (14) where both are shown to be

maximal projections. However, the study didn’t include analysing the internal structure of DPs.

see (Anurudu 1999, Obasi 2011, Mbah 2011 and Obiamalu 2014, 2015) for various proposals in

respect to the DP analysis.

2.3.3 Key Minimalist Assumptions

This sub-sections discusses some key assumptions developed in earlier framework which were

incorporated into MP probably because of their empirical significance. In other words, they were

not developed alongside MP. These include the DP-hypothesis, the Predicate Internal Subject

Hypothesis (PISH), the Bare Phrase Structure (BPS) and the Split projection hypotheses such as

the Split-INFL, Split-VP and Split-CP hypotheses. These assumptions improved the

understanding of syntactic structures and are central to the analysis of clause structure which is

the primary focus of this study.

41
2.3.3.1 The Predicate-Internal Subject Hypothesis

The predicate-internal subject hypothesis (PISH) also known as VP-internal subject hypothesis

Nweya (2018:48), claims that subjects in sentences originate internally within the VP containing

the relevant verb and move from there to the spec-T. There were two motivations for this

hypothesis: the need to account for θ-role assignment to the external argument and the need for

the subject of the sentence to occupy [Spec, IP] as suggested in Chomsky (1981), Haegeman

(1994), Ouhalla (1991) a.o. Since it is impossible to assign external θ-roles under Head

complement configuration, it was suggested that all θ-roles must be assigned within the

projections of the H. Hence, external arguments get their θ-roles at the Spec of ʋP where they are

base generated. To account for the fact that subject must occupy [Spec, IP/TP], a movement

process, motivated by the need to check features, was instituted to raise the subject from [Spec,

VP] to [Spec, IP/TP]. The constituents value their [-interpretable] features in the process of

movement. In summary, this hypothesis was adopted in research in the mid-1980s (Jackendoff

1977). It provides a better way to capture the fact that non expletive subjects are external

arguments of verbs while the objects are the internal arguments and that θ-roles are assigned

under merge operation.

2.3.3.2 The Split Inflection Hypothesis

In the earlier versions of GB, prior to the development of the split INFL idea, it is assumed that

elements such as Auxiliary, Negative, Modal, Agreement markers, Tense etc. occupy the INFL

node. This tradition violates the Endocentricity Constraint which holds that every head projects a

phrase and all phrases have heads (cf. Haegeman 1991, Hornstein, Nunes and Grohmann, 2005).

In this regard, Pollock (1989) provides empirical arguments in favour of the view that INFL

should not be considered as one constituent with two different sets of features ([±Tense, ± Agr]),

42
instead, each of the set of features is the syntactic head of a maximal projection, AgrP and IP (the

latter he called TP in French language). He, therefore, argues that the INFL should be split into

Tense Phrase (TP), Agreement Phrase (AgrP) and Negative Phrase NegP. A suggestion which

Chomsky (1993) further extended by suggesting that Agr should be split into AgrSP and AgrOP.

Extending the proposal, Felser (1999a) in Radford (2009: 339) proposes AspP while Schütze

(2004) also in Radford (2009:342) proposes MoodP. Generally, the split INFL idea has been

implemented in many languages of the world such as Italia (Ouhalla, 1991), Yoruba, (Ajo̩ ngo̩ lo̩

, 2005), Iyinno Abimbola, Igbo (Obiamalu 2013), etc. Thus, all the INFL elements were given

autonomy to project their own phrase; a move that is compatible with the endocentricity

constraint. Chomsky (1995) later did away with Agreement projections on the ground that they

are not visible at PF and LF interfaces since they bear [-interpretable] feature.

2.4 Empirical Review

This section studies earlier research with emphasis on clause structure carried out in other
languages using minimalist program or related theories that precede MP. The section reviews the
empirical data from various scholars such as Ilorin (2010), Umaru (2016), and Nweya (2018).

2.4.1 Ilorin’s (2010) Account on Igala and Yoruba Clause Structure

The primary device used to build up the syntactic structures of the minimalist clause is

computation (CHL). Computation consists of two major operations, namely, merge and

attract/move, which interact overtly to derive syntactic objects from which convergent

derivations are formed.

Merge is an operation which somehow implies ‘select (fully inflected items from the lexicon)

and then merge.’ It combines two categories of words (either functional, lexical or phrasal) at a

time to form a larger structural unit which can in turn be merged with another item or another

43
already formed structural unit to form yet another larger unit, etc. In a nutshell, merge is a binary

operation that targets two and only two items at a time. For instance, in the derivation of the

Ìgálà expression in (23) below,

23. Ma á là úchu lé ̣
3PL Prog buy yam that
‘They are buying that yam tuber.’

The noun úchu is merged to D lé ̣ to derive the NP úchu lé ̣ which in turn is merged with

the V là to form the VP là úchu lé ̣ to which the Infl item á mergers to derive the I-bar

as in (Fig.15).

Fig. 15. I

I VP
á
V NP

là
N D

Úchu lẹ́

Similarly. Ilori (2010) gives detailed analysis of the operational steps found in Yorùbá clause as

seen in the expression (24) below. The only noticeable difference is that the merger operation in

this case involves a locative adjunct PP.

24a. Akín ra ata ní ọjà

----HTS buy pepper Loc market.
‘Akin bought pepper in the market.’

44
Fig. 16. TP

N T'
Akin
T VP
HTS V'

V PP

V N P N
ra ata ní ọjà

In this particular case, an already formed PP ní ọjà is merged with an already formed Vbar ra

ata to form another V-bar ra ata ní ọjà which projects VP. The VP is merged with T to form T-

bar HTS ra ata ní ọjà. Given the strong EPP feature of the HTS (T), the N Akin adjoins to

Spec-TP to form the convergent derivation Akín ra ata ní ọjà.

Using the same merger operation, Ilorin demonstrated how a CP (Complementizer Phrase) can

be derived in both Igálà and Yorùbá by simply merging a complementizer item like pé ‘that’ in

Yoruba and kì ‘that’ in Igala respectively, to the already formed convergent derivation in

(Fig.17):

Fig. 17. CP CP

C CIP C CIP

kì Ma á là úchu lé ̣ pé Akín ra ata ní ọjà

‘that they are buying that yam tuber’ ‘that Akin bought pepper in the market’

(Ilorin 2010:184)

45
as represented in the partial structural sketch in (17a&b). He noted that the examples are partial

representations of the overall syntax of Ìgálà and Yorùbá clause derivations. The only syntactic

information provided in the structural sketches above is categorial in nature. In other words, the

structural sketches of the merger operations in (17a&b) simply tell us the grammatical categories

(i.e. functional, lexical or phrasal) to which each of the items and their pairwise combinations

there belong. However, other operations and surface output conditions interact with the

computation process to fine-tune the outputs of merge before derivations could converge at LF

and PF. Any derivation that fails to converge at any of these two interface levels crashes i.e. such

a derivation is unacceptable or simply ungrammatical. The major operation that interacts with the

output of merge for convergent derivations is movement, popularly called attract or raising in

the MP. Movement in this instance is triggered by the need to check the formal features of

syntactic objects in derivations. Each word/item in every derivation carries three sets of

grammatical properties or formal features (FF) which must be checked in appropriate checking

relations and domains before the relevant derivation can converge. These features are head

features, which stand for the inherent or intrinsic grammatical properties of a word; specifier

features which indicate the properties or kind of specifier (or precisely ‘subject’ in case of V and

T) that a given word requires, and complement features which embody grammatical information

regarding the kind of complement a word allows or selects. Out of these sets of features, only

some head features, i.e. categorial and phi-features, are interpretable at LF. The rest - head case-

feature, specifier features, complement features and verb inflection features other than tense - are

uninterpretable (Radford 1997b:71). Uninterpretable features are those that have no semantic

content and as such play no role in the semantic interpretation of relevant derivations. All

uninterpretable features must be checked and erased because representations at LF should

46
contain only semantically interpretable features; if not, the principle of full interpretation would

be violated and the derivation would fail to converge. If a derivation contains one or more

uninterpretable features at LF as a result of mismatch between features of the checker and the

checked, the relevant feature(s) cannot be erased from either and such a derivation automatically

crashes (Radford 1997a:174 and 1997b:175).

With this as theoretical platform, he examined how formal features are checked in derivations of

Ìgálà and Yorùbá clause structures. Where he began by looking at feature checking in the Ìgálà

clause expression ma á là úchu lé.̣ Ilorin said that the items in this convergent derivation have

the following features:

Ma: Head features; pronoun-D (categorial), 3PL (phi-), Nominative (case)’

Sepcifier-feature; none.
Complement features; none.
á: Head-features; T (categorial), Present progressive.
Sepcifier-features; 3PL Nominative nominal phrase / expression.
Complement-features; Transitive finite V projection.
là: Head-features; V (categorial) Transitive, finite.
Specifier-features; 3PL Nominative nominal phrase / expression.
Complement-features; Accussative nominal phrase
úchu: Head-features; N(Categorial), Accussative (case)
Specifier-features; none
Complement-features; Lexical Determiner
lé:̣ Head-features; D (categorial)
Specifier-features; none
In the same vein, the items in the Yorùbá expression Akín ra ata ní ọjà according him have the
following features which are checked in appropriate spec-head domains:
Akin: Head-features; N (categorial), 3sg (Phi- ), Nominative (case).
Specifier-features; none.
Complement features; none.

47
HTS ( ´ ) : Head-featues; T(categorial), Present/Non-future
Specifier-features; 3sg Nominative nominal expression.
Complement-features; Tensed V-projection.
rà: Head-features; V (categorial), Tensed , Transitive.
Specifier features; 3sg Nominative nominal expression.
Complement-features; Accusative Nominal expression + optional adjunct PP.
ata: Head-features; N (categorial), Accusative (case).
Specifier features; none.
Complement features; none.

2.4.2 Umaru’s (2016) Account of Relative Clause in Idoma

Like Ilorin, Umaru (2016:82) states that the Idọ́mà relative clauses are introduced by the relative

pronoun /n/ ‘who’, ‘which’, ‘whom’, which is realized as a nasal sound /n/ which bears high tone

[+pronoun] feature. The sound following [n-] is determined by the initial sound of the vowel of

the succeeding word such as; né, in nẸnẹ́nchẹ̀, as in nÁléchénù, ní as in nígbánọ́ and nọ́ as in

nọ́ipẹ̀, núm as in múm mà. He clarifies this by saying that the realization is due to the fact that

most Idómà NPs are vowel initial. Also, ọnyé ‘who’ is used in situation where the relative clause

introducer is at the beginning of the relative clause structure. A relative clause in Idómà is an

adjective modifier subordinated to noun phrase (NP), with a pronoun inside the NP (the head of

the clause). For example in (25) below:

25. Ọchẹ̀ à [nó lé ikpọ́chí] kwú ikwú bọ̀bí

Man the [REL VERB NP] die death bad
‘the man [who ate the fruits] died a horrible death’

From the example (25) above, the clause nó lé ikpọ́chí ‘who ate the fruits’ is embedded in the

main clause; Ọchẹ̀ à ‘the man’ and the relative pronoun nó- ‘who’ is co-referent with Ọchẹ̀ a

‘the man’. This can also be noticed in the subject of the relative clause nó lé ikpọ́chí ‘who ate

48
the fruits’. Related to this, the head and the relative pronoun do not need to have the same

grammatical functions in their classes because referential and grammatical identities are not the

same thing. As illustrated in (26):

26. Ọchẹ̀ à [núm má] gbẹ̀

Man the [REL+ I see] big
‘the man who(m) I saw was tall’

Ọchẹ̀ à ‘the man’ in example (26) above, is the subject of the main clause, while the co-

referent núm (whom) is the object of má ‘see’, the subordinate clause. Núm má ‘whom I saw’

refers to ọchẹ̀ à ‘the man’. The referent is ọchẹ̀ à gbẹ̀ ‘the tall man.’

Umaru identified the characteristics of Idọ́mà relative clauses as being sensitive to tense, aspects

and mood. In his word Umaru (2016:89) to mark tense in Idọ́mà, reference is usually made to

time of the action, which has indicators such as ẹ́ẹ́, le, ya used to mark past tense. This he

exemplified in the following constructions:

27. Ochóhẹ́pọ́ nó kwú ẹ́nẹ́à é lè nyí íchẹ̀

NOM REL die yesterday they have plant today
‘Ochohepo who died was buried today’
The example in (27) above shows that kwú ‘died’ is post modified by the time of the action;

ẹ́nẹ́à ‘yesterday’ to state the tense, mention must be made of the time the action took place. The

verb kwú ‘died’ does not change its morphological structure, but retains its structure to either

express the present or past time. Present tense in Idọ́mà is also indicated by the time of the action

using nyá as illustrated in (28):

Ọ́yì ọ̀chẹ́nyílọ̀ nó yọ̀ nyá wé ọ̀í m

Child male REL is this BE child my
‘this boy is my child’

49
In the above data, nya ‘this’ indicates the present tense. This means that the relative clause nó yọ̀

nyá ‘who is here’ is stated in the present tense.

The future tense in Idọ́mà is formed by inserting the morpheme gé into the forms as shown in

(29) below:

29. Ọípẹ̀ óchènyá ná gé má ọ̀chọ́chi óchí a wẹ́ ọ̀ì m

Child female REL will VERB morning tomorrow the BE child my
‘the girl whom you will see tomorrow morning is my child’
The example in (29) above shows the futurity in the relative clause of Idoma. This is expressed y

the use of the future tense marker gé. It also indicates that the presence of the maker gé makes

the the whole construction futuristic; that the action is yet to take place.

On aspect, Umaru said a grammatical description of verbs referring primarily to the way the

grammar marks the duration or type of temporal activity denoted by the verb is attested in Idoma

relative clauses. The he demonstrated using the following data in (30) :

30. N jé elikà lẹ́ ọ́chẹ́nyà nó okáchè máà

I PERF ring PP NP REL NP PERF
‘I have given a ring to the lady whom Okche saw’
He explained that the verb is perfective, even though the morpheme máà ‘saw’ indicates both the

past and tense simultaneously in the grammar of the language. the example shows that the ring

was given to the recipient Okáchè saw earlier. In the progressive form of it, he states that there

would be addictive morpheme such as gé, yọ́ and so on used before the verb. The example given

in (31) below clarely explains that:

31. N jé alíkà lẹ́ ọ́chẹ́nyà nó okáchè yọ́ máà

I PERF ring PP NP REL NP PROG. See
‘I gave a ring to the woman whom Okache is seeing’
The data in (31) above shows a progressive aspect of the verb in the Ìdọmà relative clauses. The

morpheme yọ́ ‘is’ indicates that the activity of the verb is progressive, with the progressive

50
marker pre-modifying the verb máà ‘see’ which is a member of the relative clause nó okáchè

yọ́ máà ‘whom Okache is seing’. This, therefore, affirms the fact that relative clause marker in

Idoma is sensitive.

Loking at mood as one of the characteristics of relative clause in Idoma Umaru sates that mood

refers to the speaker’s perspective on the event in particular, whether the event described is a

possibility, probability, a necessity or an obligation which is expressed through modal auxiliary

verbs (Carnie, 2013). Major mood categories are declarative, imperative, and interrogative

otherwise known as statement, command and question (Kroeger, 200). Idoma relative clauses are

sensitive to mood as seen in (32):

32.a. Ádọ́yì nó wá á gè kwúchè nyọ́ fíà (definitive)

NOM REL BE TENSE MOOD return go soon
‘Adoyi who came will soon return back’

b. Ádọ́yì nó wẹ́ ádà kọ́yí á nó yágbè chíkà kó gáẹ̀ (obligation)

NOM REL BE NOM son the REL VERB MOOD have come
‘Adoyi who is the father of the injured girl should have come’

c. Adámà nó chíkà wá gé wá gílà (possibility)

NP REL MOOD VERB MOOD VERB possible
‘Adama who should have come will come’

The mood expressed in (32a) above is signified by gé ‘will’ which indicates definateness and

expressed within relative clause which shows that the person in question will surely return back.

The obligatory mood is exemplified in (32b) where we have a stacked relative clause, the mood

shows in the second construction nó yágbè ‘who was injured’ that relates to the first clause; that

is nó yágbè ‘who is injured’ modifies the the first clause nó wẹ́ ádà kọ́yí à ‘who is the father of

51
the girl’. Chika ‘should’ expresses the mood of the action (obligation). In example (32c) the

expression is that of possibility. Which is expressed in morphemes such as chíkà ‘should’ and gé

‘will’. The relative clause expression here shows that the speaker is not quite sure the person will

carry out the action.

2.4.3 Nweya’s (2018) Accounts of Igbo Clauses

Newya started by discussing the VP domain of the clause where she examined the derivational

path of the domain using the cartographic approach and the phase derivation theory in order to

arrive at a structure that captures all the significant elements in the domain. This domain

represents argument and event structure. Studies show that it provides answers to the way

arguments are mapped onto syntactic structures.

Starting with monotransitive constructions, Nweya (2018:94) said they are those constructions

where the verb is associated with an external and an internal argument. The internal argument

may be an object or a complement. Monotransitive verbs can be subjected to the VP shell

analysis, which identifies two parts of the VP layer: the outer shell headed by the light ʋ and the

inner core headed by the lexical verb. The lexical verb functions as the complement of the null

light verb ʋ and assigns theta role to the OBJ at the point of merge following θ-criterion. The

light ʋ is assumed to have a strong [vF] which triggers it to attract the head of the lexical V to

adjoin to it. In addition, it has a strong edge feature [EF] which enables it to attract the external

DP argument to the Spec, ʋP subject to PIC. It also values the accusative case of the object. At

the completion of the ʋP phase, its complement VP is transferred to the interface levels. The

Simple VP structure based on her study, it is taken to be verbs with only two arguments, an

internal and external argument; and project only a V head, whether simple, compound or

complex.

52
Consider the examples below:

33 a. Ùde gwò - rò ọgwù .

U. prepare-PST charm
‘Ude prepared a charm’

b. Ifee zà-rà ụlọ

I. sweep-PST house
‘Ifee swept the house’

c. Uchè nà-à-gụ akwụkwọ.

U. AUX-PART-read book
‘Uche is reading a book’

d. Ha gbù-rù oke
3PL kill-PST rat
‘They killed a rat’
e. Ngọzị tụfu-̀rù akwụkwọ
N. throw-PST book
‘ Ngozi threw away the book’ (Mbah 1999:142)

f. Ndị ahu gbù-rù agwo .

PersonPL DEM kill-PST snake
‘Those people killed a snake’

The structures in (33a-f) above contain monotransitive verbs with an internal and an external

arguments. For instance, (33a), Ude ‘personal name’ is the external argument while ọgwu ̣̀

‘charm’ is the internal argument. In some instances, monotransitive constructions may have

adjuncts which are optional constituents as shown in (34) below:

34. a. O mè- rè òfụma n’ùle

3SG do-PAST well P-exam
‘S/he did well in the exam’

b. Ha nà e-re anù òfụma n’ahịa ugbu ā

3PL FUT PART-sell meat well P-market now DEM
‘They sell meat affordably in the market now’

c. Ànyị jè-rè ozī n’Òshà

1PL go-PST message P-Ọsha

53
 ‘We went to deliver a message at Osha’

Observe the presence of adjuncts in (34) above. Data (34a) and (34b) have PP and AdvP as

adjuncts respectively. In (34b), the manner adverbial o f̣̀ ụma ‘well’ follows the DP object anụ

‘meat’ but precedes the PP adjunct n’ahịa ‘in the market’; while the temporal adverbial ugbu

‘now’ follows the PP ‘n’ahịa’. Data (34b) also shows that whenever temporal and manner

adverbs co-occur in a construction with a PP adjunct, the manner adverb may precede the

temporal adverb with the PP adjunct intervening between the two.

However, it is difficult to determine if temporal adverbs like gboo ‘early’ and ugbu ‘now’ always

follow PPs as these examples in (35) show:

35. a. Ha nọ n’ahịa ugbu a

3PL be P-market now DEM
‘ They are in the market now’

b. Ha nọ ugbu a n’ahịa.

3PL be now DEM P-market
‘They are now in the market’

In (35a), the temporal adverbial, ùgbu ‘now’, follows the PP place adverbial in (35a); conversely,

the PP follows the time adverbial in (35b) and the sentences converged.

Therefore, there seem to be a symmetric c-command relationship between time AdvP and PP

place adverbial in Igbo probably because they are both adjuncts. Nevertheless, they do not bring

about significant difference in the word order.

To show how monotransitive constructions are derived using (33a) above, computation begins

with the numeration of LIs56 as follows:

36. N ={ Ùde1, ʋ1, gwọ1, rV-PST1, ọgwù }

The numeration above which is an unordered set shows that LIs enter derivation with their

idiosyncratic properties, (SYN, PHON, SEM); afterwards, the combinatorial operation merge

54
takes a pair LIs from the numeration and joins them to derive a new structure reducing the index

by (1) as follows: gwọ ‘mix’ is merged with ọgwu ̣̀ ‘charm’ to derive VP at the point of which θ-

role THEME is assigned to the DP ọgwu ̣̀ ‘charm’ following the Theta-role Assignment Principle

TRAP stated below:

(37a) Theta-role Assignment Principle (TRAP)

θ-role can only be assigned under a merge operation
(Hornstein, Nunes and Grohmann 2005:54)

VI is not projected because the verb is monotransitive. The partially formed structure is merged

with an abstract/null causative light verb to form ʋI. The light ʋ is assumed to have a strong [vF]

which enables it to trigger the lexical verb to adjoin to it. Therefore, gwọ ‘mix’ moves from its

position in head V to adjoin to the null light verb via head to head movement to value the V-

feature of the light verb. In this position, the light verb serves as a transitive probe and searches

for an unvalued case feature in its c-command domain 60 and this is satisfied by the active DP at

the object position of V. Since Igbo does not permit the object to move out of the VP overtly, the

object covertly moves to the specifier of ʋP to value its case feature in a spec-head relationship as

stated in (37b) below:

(37b) Accusative Case Valuation

An unvalued case feature on a goal is valued as
accusative via agreement with a transitive probe
(Radford 2009:304).

Thus, the light ʋ values the case feature of the object. At this stage in the derivation, the strong

EPP feature on the light verb requires its spec to be filled. As a result, the DP subject Ude‘PN’ is

selected from the numeration and merged to the already formed structure to form ʋP. The DP

receives the θ-role [AGENT] from the head ʋ as stipulated by TRAP. At this stage, the unvalued

number and case features of the OBJ have been valued and the unvalued ones deleted (though

55
they remain visible at PF). In this way, merge yields the structure below with arrows showing

internal merge and the point of transfer.

Fig. 18.
vP phrase

DP V1
Ùde
Ʋ VP
gwo
V DP
gwo gwo
Having derived a propositional ʋP phase, operations TransferPF and TransferLF which are two

separate operations apply simultaneously to the c-command domain of the phase splitting the

derivation into two where phonetically relevant materials are sent to PF interface and the

grammatical and semantically relevant features are transferred to LF interface. From this point

forward, the complement of the ʋP phase becomes frozen in place and inaccessible to further

syntactic computation following PIC repeated here as

2.4.3.1 Phase Impenetrability Condition/PIC

The c-command domain of a phase head is impenetrable to an external probe (i.e. a goal which is

c-commanded by the head of a phase is impenetrable to any probe c commanding the phase).

(Radford 2009:380) In other words, ọgwụ ‘charm’ is spelt-out overtly at the PF interface. The

derivation converges because the representation meets the interface conditions (e.g. linear order,

scope, propositional status etc.). Recall that some monotransitive constructions may have AdvP

and PP adjuncts as shown in (33) and (38a) below.

38. Ha bò - rò anụ ā ò f ụma n’ahịa.

3PL splay-PST meat DEM well P-market
‘They properly splayed the meat in the market’

56
To demonstrate how these kind of structures are derived, computation begins with the
numeration of LIs as follows:

39. N ={ Ha1, ʋ1, bọ-1, rV-PST1, anụ1, a1, ọfụma1, na1, ahịa1}

Similarly, operation merge takes a pair of LIs from the numeration and merges them to derive a

new structure reducing the index by (1) as follows: na ‘at’ is merged with ahịa ‘market’ to derive

PP at the point of which θ-role [LOCATION] is assigned to the DP ahịa ‘market’ in line with

TRAP. For the unvalued case feature of the DP to be valued, it covertly moves to Spec PP, i.e.

the checking domain where it enters into an agree relationship with P, after which its case is

valued as OBL. Igbo does not have adpositions as found in some languages (e.g. Zarma, see

Jayeola 2016), therefore, the PP complement cannot move overtly to value its case feature. The

derivation proceeds with the merging of the derived PP n’ahịa ‘at the market’ with the lexical V

bọ ‘cut’ to yield VI. VI is then merged with adverb ọfụma ‘well’ to form another VI; the derived

VI is internally merged with V bọ ‘cut’ to form another VI. At this point, the internal argument,

anụ ‘meat’ is merged in the structure yielding another V I. The derivation continues as VI is

internally merged again with the lexical V bọ ‘cut’ to yield VP after which the lexical verb

assigns θ-role to the object DP anu ‘meat’. The multiple VI is projected to introduce the AdvP

and the PP adjuncts in the structure. The partially formed structure is merged with an abstract

null causative light verb to form ʋI. The strong [vF] on the light ʋ triggers the lexical verb to

adjoin to it. The light verb probes for an unvalued case feature in its c-command domain and this

is realised by the active DP anụ ‘meat’ which moves covertly to the specifier of ʋP to value its

case feature in a spec-head relationship.

It is worthy to note that agreement between transitive verbs and their objects in Ìgbò is abstract

in the sense that it is not morphologically marked. At this stage in the derivation, the strong EPP

57
feature on the light verb requires the spec ʋP to be filled. Hence, the DP subject Ha ‘3SG’ is

selected from the numeration and merged to the already formed structure to form ʋP. The DP

receives the θ-role [AGENT] by the head ʋ as stipulated by TRAP. All the unvalued features of

the DPs have been valued and the unvalued ones deleted. Thus, merge yields the structure below.

Double Object Constructions

Another area of interest to this study that Nwanya (2018) discussed in her work is the Double

Object Constructions (henceforth DOCs). She said DOCs are known to have ditransitive verbs

which are verbs that licence two objects: a DO (theme) and an IO (goal). In Ìgbò, DOCs are

grouped into two based on the nature of the predicate: (i) those with zero extension and (ii) those

having extensional applicative morphemes. Since the emergence of generative grammar,

DOCs create some puzzles in syntactic analysis due to their behaviour crosslinguistically. In this

regard, the VP-shell analysis of Larson (1988) emerged in an effort to solve the puzzle created

by DOCs.

2.4.3.2 Simple Double Object Constructions

Simple double object constructions (SDOCs) are those constructions that involve verbs that

require two DPs in their object positions without any overt verbal morphology in order to satisfy

their categorial requirements.

One observation about Igbo SDOCs is that the IO precedes the DO as shown in the data below:

40. a. Àmàrà tù - rù Uchè òkwutē.

A. throw-PST U. stone
‘Amaka threw Uche a stone’
b. Chinekè za-ra Ugo ekpere.
C. answer-PST- U. prayer
‘God has done me good’.
c. Ọ gbà-rà Àmàrà mmiri
3SG kick-PST A. water
‘S/he splashed water on Amara’

58
In (40) above, one can observe that the verbs are subcategorised as V___DP DP. The examples

also show that the IO (GOAL) comes before the DO (THEME). For instance, in (40c), m is the

IO [GOAL] and precedes ụkwụ ‘leg’ the DO [THEME]. In other words, all the examples above

have the same linear order.

3.5 Some Pevious Works on Basà Grammar

Works of different kinds have been carried out on Basà but of interest and relevance to this work

are the earlier findings carried out in Basà noun phrase by Imoh (2011, 2012, 2014) and Dansabo

(2016). Based on the findings made by Imoh (2011), he establishes that the ordering of the

syntactic elements in Basà noun phrase is not in consonance with pattern of English as

exemplified below:

41. NP -> N + Det as in

a. yèu íke b. íjènè éte
| | | |
N DET N DET
child one ‘a child’ yam that ‘that yam’
Imoh (2012), states that in Basa, names occurring at the subject mark nominative case by

prefixing such names with the prefix ga-. The examples (42) below shed light on his findings:

42. a. gà-Ǹdázhaga
Nom name ‘Ndazhaga’

b. gè-Wèbìye
Nom name ‘Webiye’

c. gè-Peni
Nom-name ‘Peni’
(Imoh 2011:)
These findings on Basà NPs are basically introductory, as they do not unfold other intricacies of

the noun phrase, namely, the structure of Basà NP as well as its relationship with other phrases in

the language structure.

59
3.5.2 Morphosyntax

Imoh (2014:96) studied the the basic rules governing plural formation in Basà where he said in

Basa, plural formation processes are complex and arbitrary. The involve processes such as zero

affixation, attatchment of sh-prefix to the noun base, consonant alternation, reduplication, m/n

formation/internal mordification, supletion, elision (deletion), replaceive process (vowel

alternation), and paucity. Imoh proved these processes using the following examples:

42 a. meni, manebe, moko

‘water’, ‘oil’, ‘gum’
b. tama, tuwa, togba
‘pap’, ‘masqurade’ ‘vomit’
In Basa, Imoh (2014:97) said nouns that begin with /m/, /t/, /s/, /ʃ/ and /ʧ/ sounds are invariable

whether they are used to express a singular plural or sense of the word.

43. singular plural

a. akara ‘chair’ sh-kara ‘chairs’
b. aloma ‘hen’ sh-aloma ‘chikens’
c. akama ‘hook’ sh-kama ‘hooks’
singular nuons are prefixed with the morpheme sh- for the formation of their plural forms as

exemplified in 43 above.

3.5.3 Head of Noun Phrase

Dansabo (2020), while discussing the head of a noun phrase in Basà using the Xbar approach,
said it consists of a noun or pronoun. According to him, “going by the X-bar theory, N (noun)
which is the head of the phrase is followed by N¹ which in turn, is immediately dominated by NP
or N¹¹.” He explained this with the data in (43) demonstrate the head of a NP in Basà:

(43) a. alò̟mà i shenje

chicken AGR red
‘a red chicken’

b. Gò-Nyìzò gâ ˋ ’nyẹ̟ncẹ̟

NOM-name DEM clever
‘Nyizo is clever’
60
Fig. 19. Fig. 20

N¹ N¹¹

N A¹ N¹ A¹

AGR A NOM N Spec A

alò̟mà i shenje gò - Nyìzò gâ ˋnyẹncẹ

chicken AGR red/brown NOM-Nyizo DEM clever

‘The brown chicken’ ‘Nyìzò is clever’

The X-bar theoretic representation in fig. 3 recognizes an intermediate category which is larger

than a word but lesser than a phrase. The head of the phrase is alò̟mà chicken’ which also

functions as the minimal projection (N or N°). The N¹ is alò̟ma i shenje ‘red chicken’ is the

maximum projection while I (AGR) serves as the subject agreement marker and the adjective

shenje serves as the complement of noun.

Fig. 20. Gives account of the structure in (43b) which shows that in Basa, the head noun gò-

Nyìzò ‘Nyìzò’ occurs before the specifier which is the demonstrative pronoun gâ ‘this’ followed

by the complement which the adjective ’nyẹncẹ ‘clever’. It is important to note that, the

nominative morpheme gà- is only used for proper names occurring in the subject position of a

given construction.

2.5.2 Grammatical Functions of Noun Phrase in Basa

61
The noun phrase performs basic functions as subject and object of a verb. Other functions it

performs are direct object, indirect object, subject complement and object of a preposition. In the

analysis that follows, some of the functions of NP in Basà are examined.

2.5.2.1 Noun Phrase as a Subject of the Verb

The person or item in a clause or sentence that performs the action of the verb is known as the

subject. It is the element in the sentence with which the verb has grammatical agreement. The

doer, performer, or what is responsible for carrying out the action in the sentence is referred to as

the subject. Structurally, the noun phrase as a subject of the verb, occurs before the verb phrase

in basic sentence. The following italicized word are examples of subjects in Basà:

(44). a. uluwu áta zhe nà ambè̟

cobra DEM go.PST PREP bush
‘That cobra is went into the bush’

b. Ijè̟nè̟ mi hweni wenèbe

yam 1Sg.POSS burn-PST yesterday
‘My yam got burnt yesterday’

c. Gà-Láshẹ swe mèni

NOM-Lashe drink.PST water
‘Lashe drank water’

In (44a), the noun phrase uluwu áta ‘that cobra’ serves as the subject of the verb in the sentence

because uluwu ‘cobra’ is the performer of the action of the verb, zhe ‘went’. In (b) íjènè mi

‘my yam’ is the subject of the verb hweni ‘burnt’. In (c) Gà-Láshe̟ ‘Lashe’ functions as the

subject of the verb swe (drank). These noun phrases all precede their respective verbs

syntactically.

2.5.2.3 Noun Phrase as Object of Verb

Direct objects follow transitive verbs, action verbs that require something or someone to receive
the action.

62
45 a. Gè- Sheneni po Gûre ùnwanpè̟

NOM-name hit.PST name slap

‘Sheneni slaped Gûre’

b. Bò̟ kwo̟ne̟ ùhwâ u gémbí

2SG Kil.Pst snake AGR big
‘s/he killed a big snake’ ‘

c. Gè - Jére keri yeu nè mèni

NOM-name throw.PST child PREP water
‘Jere threw the child in water’

d. Bò̟ zweni Tàshe̟lane̟ meni

3Sg Pour.PST name water
s/he poured Tashe̟lane water’

In (45a), Gûre serves as the object, that is, the receiver of the action carried out by Sheneni and

identified by the verb. It is important to note that in the construction above, the object NP occurs

between the verb po ‘hit’ and the object or recipient Gûre. In (8b), ùhwâ ‘snake’ is the receiver

of the action kwo̟ ne̟ ‘killed’; as such, it serves as the object of the verb. Again, since the object

NP comes immediately after the verb.

From the examples in (45c), we see that yè̟u ‘child’ and Tàshe̟ lane̟ are the direct receivers of the

action identified by the verb, keri ‘threw’ and zweni ‘poured’ that were carried out by Jére and

Bò̟ ‘S/he’. Syntactically, yè̟u and Tàshe̟ lane̟ which are both noun phrases occur after the

transitive verbs in each of the constructions.

2.5.2.4 Noun Phrase as an Indirect Object

Indirect objects are sometimes referred to as beneficiaries. These elements are typically animate.

The indirect object is the item in a sentence that receives the direct object. Beneficiaries are

found in sentences where the verb indicates, in some way, an action of giving or telling.

Sentences with an indirect object usually contain a direct object as well. For example:

46. a. Gò- Nyìzò peni Èzìmí ùkwuribi

63
 NOM-name give.PST name money
‘Nyizo gave Azumi money’

b. Aguma peteni Tàshe̟lane̟ ùfwòro

chief put on name cap
‘The chief crowned Tashilane’

The examples in (46a) show that structurally, in Basa, the indirect object which is the receiver or

benefactor occurs before the direct object. While money ‘ùkwuribi’ serves as the direct object,

Èzìmí is the benefactor or receiver of the object ‘money’.

2.5.2.5 Noun Phrase as Subject Complement

A noun that is equal to, or renames the subject, is called a subject complement.

47. a. Gò-Ímòh shẹ bìmbisegenì

NOM-name AUX teacher
‘Imoh is a teacher’. ‘.
b. Gà- Shìlàtò̟ shẹ la àguma
NOM-name be PST chief
‘Shilatò̟ was a chief’

In (47a) above, bìmbisegenì ‘teacher’ and àguma ‘chief’ are subject complements in (a) and (b).

They follow the linking verb she̟ ‘be’. They link the subjects gò-Ímìh and Shìlàtò̟ respectively.

The subject complement also completes the meaning of the sentences.

2.5.2.6 Noun Phrase as Complement of Preposition

The NP can also function as the object or complement of preposition.

48. Gò Dόmòzù ’ye̟me̟ bò̟janà nà akama

NOM-name catch.PST fish with hook
‘Domozu caught a fish with a hook’

In (48), bò̟janà ‘fish’ is the indirect object of the verb ’ye̟ me̟ ‘caught’, while akama ‘hook’ is

the object of the preposition nà ‘with’. The object of the preposition also functions as an

adverbial complement because it shows the instrument used in carrying out the action.

64
2.5 Summary

The first section of this chapter discussed the basic concepts of clause and its features as elored

by other schorlars. The second presents the theoretical studies and framework used in the study.

Specifically, it traced the analysis of clause structure from the introduction of generative

grammar to its current model which is Minimalism as well as alternative approaches to the

analysis of clause structure. It also discussed the architecture, components and basic assumptions

of MP and the cartography enterprise on which this study is based. The last sub-section, previous

studies related to the present work were reviewed by identifying the goals of the studies, major

findings and relationship with the present study. From the reviewed works, this study identified

some existing gaps to show that this study is still necessary despite the available studies. As well

as receive guides and focus on how to go about carrying out a prper study of the Basa clause

structure.

65
 CHAPTER THREE

Research Methodology

3.0 Preamble

This chapter discusses the research methodology applied by the researcher in the collection of

data for the research. This has to do with the instruments used, and the methods of data analysis

applied.

3.1 Area of Study

This study is concentrated on the Basa speaking communities in Toto Local Government of

Nasarawa, Bassa and Kogi LGAs of Kogi State. Toto Local Government Area of Nasarawa State

which is bounded on the south by River Benue, north by Nasarawa Local Government Area and

on the West by Abaji Area council of The FCT. Bassa Local Government area of Kogi state is

bounded on the south by Dekina Local Government Area and River Benue on the north, on the

west by River Niger while on the east, it is bounded by Ankpa Local Government Area (1moh

2016:18). Kogi Local Government Area is bounded on the It is the dialect that is widely spoken

and taken to be the central dialect given the fact that it is the dialect that has been reduced to

writing and is used in religeus activities as well as primers for elementary teaching and learning

of the language.

3.2 Research Design

The research design for this study is basically qualitative in approach employing interviews and

naturalistic observations. As a case study, we observe and describe the clause structure in Bassa.

3.2 Population of Study

66
This research will explore the structure clauses in Basà as such only native speaker of the

language will be consulted as the respondents. Ten of these native speakers will constitute the

study population of study. The ages of respondents will be taken into consideration as one

intends to consult respondents whose ages range from 40-77. Their education background will

also be put into consideration as well as their place of growth. The knowledge of the researcher

as a native speaker of the language will also help in attesting data in order to direct the course of

the research.

3.3 Study Area

The study basically is on the central dialect of Bassa language spoken in Nasarawa, Kogi, Benue

states and the FCT called bòbú. However, the research area is Toto Local Government Area of

Nasarawa state. Toto local government is situated in the western part of the state, it is bordered

by Abaji Area Council of FCT to the north, to its’ south is Nasarawa Local government, to the

East is River Benue while to the west it shares boundary with Kogi state.

Sampling is done within the Bassa speaking people occupying Toto Local Government area of

Nasarawa state. They are found in more than 20 villages in the local Government. Though,

emphasis is on Ugya town which is basically the central area of the Bassa people in the area.

3.5 Instrumentation and Data Collection

The data for this study will be collected through two sources; the first is the primary source and

the second is the secoundary source. The researcher intends going to the field which is the

communities where the language is spoken to collect data from native speakers during meetings

67
The researchers intuitive knowledge of the language will be of great help as he being a native

speaker of the language will be of great help. The secondary source has to do with written

documents in Bas language which the researcher will be going through to extract data that will

be of use to this work. These materials will include but not limited to; Basa bible, Basa primars,

Basa magazine, Basa short stories.

3.4 Sources of Data

Data for this study will be gathered from native speakers of Bassa within Toto Umasha axis of

Nasarawa state, Oguma and Sheria axis of Bassa LGA as well as Tawari and Goro axis of Koton

Karfe LGA of Kogi State. Some data will be obtained through observation of naturally occurring

speech of Bassa native speakers during tribal meetings and church services. Some will be

ignited through oral interviews and naturally-occurring observation while others will be

generated by the researcher himself. In the interview session, the researcher plans to obtain the

permission of the respondents to record their natural conversation with each other and with the

researcher. There is no intention to have any form of a structured interview questions for

respondents since what is being investigated here, is from the everyday conversation. What the

researcher intends to consentrate on is mostly to generate corpus of data for the research which is

the primary source of data collection.

3.6 Secondary Sources of Data

The collection of secondary data consisted of written documents like the Bassa Holy Bible,

Bassa Primers and Bassa articles published in magazines and works by Imoh (2011, 2013 and

2014; Dansabo 2016) will be consulted for information that will be of help and can be utilized in

68
this study. Importantly, the researcher also intends to make use of the library of the Nasarawa

State University, Keffi and the National library, Abuja.

3.7 Methods of Data Analysis

The descriptive method of analysis will be employed in the analysis and interpretation of data.

The minimalist programme will be adopted for analysis due to the fact that MP is a more

economical approach to the analysis of grammars of I-languages as revealed by Chomsky (1993,

1995, 2000, 2005, 2008, 2013 & 2015) and works by other linguists. Chomsky (2015: vii) states

that it is a program and not a theory. According to him, “it is a seamless continuation of pursuits

that trace back to the origins of generative grammar, even before the general biolinguistics

program, as it is now often called, began to take shape in the 1950s.” From the beginning, its

major goal is to clarify the concept “simplest grammar” and “to determine how to choose the

simplest grammar for each language.” As such, MP assumes that the basic principle (BP) of

language is that each language produces or has infinite number of hierarchically structured

expressions, each interpreted at two interfaces, conceptual-intentional (C-I) and sensorimotor

(SM). The two interfaces provide external conditions that the BP must satisfy (Chomsky,

1993:168, 2015: ix). MP makes use of basic computational operations to achieve its economy

driven goal. Since syntax is linked to these two systems, the syntactic model defines two

interface levels, one for each of them: Phonological Form (PF) is the interface to the articulatory-

perceptual system, and Logical Form (LF) is the interface to the conceptual-intentional system.

To derive a clause using this model of grammar, a set of lexical items with tokens will be

selected from the lexicon. The syntactic structure of the clause will be built up by taking words

out of the Lexicon one by one and merging them in the structure for a derivation to converge at

69
both PF and LF. At some point in the cause of deriving LF, spell-out is expected to take place.

Spell-Out refers to the process of deriving PF, the interface form to the articulatory-perceptual

(AP) system. PF basically contains the phonological and prosodic features of the clause. Apart

from LF and PF, other fundamental concepts in minimalist syntax are economy, the principle of

Full Interpretation, and features. Full interpretation (FI) holds that every element in a structure

must be interpreted in one way or the other. The discussion above shows that MP will be suitable

for a detailed description and analysis of an underveloped language like Basà if properly

followed and applied in the analysis of data.

70
3.0 Preliminaries
The syntax of every natural language uses, as building blocks, words, which belong to varied
syntactic classes or categories. A category in grammar could be defined as a class of expressions
or items of grammar which share a common set of core grammatical properties e.g.
morphological or syntactic (cf. Radford 1997b: 29). The fact that syntactic or grammatical
operations that generate clausal structures directly target categories of words or phrases makes
the issue of category inventory integral to any meaningful and principled description of the
grammar of any language. In the words of Awobuluyi (1978:1), The primary job of the
grammarian of any language is (1) to establish the number and types of syntactic classes of
words in the language, and (2) to describe how individual members of such classes of words are
combined to form acceptable sentences or utterances in the language. Therefore, our aim in this
chapter is to address ‘(1)’ by presenting a detailed but theoretically-based description of the
syntactic categories of words instantiated in Igálà and Yorùbá.

3.1 Universal Categories and Natural Classes

Before any adequate language-particular classification of word categories could be properly
carried out, it is pertinent to be aware of the available lexical categories in universal syntax.
Jackendoff (1977:2) puts this more succinctly by posing the following thought provoking
questions.
…what are the lexical categories (parts of speech) available
to linguistic theory? How are they associated with syntactic
structures, and what is the relationship between lexical categories
and the syntactic constructions in which they take part? What

71
relationships exist among lexical categories and how can they
be expressed? …

Traditionally, between eight and ten lexical categories were universally recognized. These
include Noun, Pronoun, Verb, Adjective, Adverb, Preposition, Conjunction, Article/Determiner,
and Interjection. Other categories recognized in traditional grammar were grammatical features
like case, number, person, mood, gender, tense, etc. These were however seen as mere
subordinate features of the contentive lexical categories (Awoyale, 1992: 70). In generative
grammar, two feature-based systems of categories exist. These are the Chomskyan system of [α
Nominal, β Verbal] and the Jackendovian [α Subject, β Object] system (Chomsky 1974
Jackendoff 1977). Among other things, these two systems defined categories in terms of their
assumed natural classes which are based on the shared syntactic features of the items concerned.
Chomsky (1974) postulates four universal lexical categories -- V, N, A, P -- defined as products
of the lexical features [α Nominal β Verbal ] as sketched out in table I.
Table I:
[+Verbal ] [-Verbal]
[-Nominal] V P

A N
[+Nominal]

(Chomsky 1974)
This system captures [V, P], [N, A], [V, A] and [N, P] but excludes [V, N] and [P, A] as natural
classes. The system however does not make provisions for closed-class items, i.e. functional
categories.
Although the Jackendovian system uses a different set of lexical features, i.e. [α Subject, β
Object], it also recognizes the same set of lexical categories sketched out below in table II.
Table II: [+Subject] [-Subject]
(Jackendoff 1977:33)
In addition to these, the Jackendovian system recognizes and represents closed-class items like
Modal, Article, and Degree words using the features [± Determiner]. V, N, P, A are scored as
[+Complement] while all other categories are [-Complement]. The [±Determiner] feature is

72
additionally used to further specify the [-Complement] items. Essentially, the Jackendovian
system captures [V, P] [N, A], [V, N] [P, A] and excludes the Chomskyan [V, A] and [N, P] as
natural classes. The fact that the Jackendovian system provides some new insights led Fukui
(1986) and Abney (1987) to modify the Chomskyan [α N, β V] by incorporating the [±
Functional ] feature in order to cater for closed-class elements. That eventually led to the
emergence of the standard category features of [α Nominal, β Verbal, g Functional] in current
generative grammar. Fukui recognizes three functional categories namely I(nflection),
C(omplementizer), and D(eterminer}, as corresponding heads for some lexical categories. I is the
functional head of V just as D is the head of N and C the head of CP (see table III).
Table III:
[+Verbal] [-Verbal]
(Fukui 1986:55)
Abney for his part recognized three functional heads – I, C, and D (=Det/Degree), as sketched
out in table IV. He however hinted that his classification is not exhaustive as it does not take care
of conjunctions (Conj) which according to him appears to be [+F].
Table IV: [-F] [+F]
(Abney 1987:63)
As a fallout of Chomsky (1974), Jackendoff (1977), Fukui (1986), and other related studies,
Chomsky (1986b) recognizes five distinct lexical and two functional categories as syntactic
universals. The lexical categories were Noun, Verb, Adjective, Adverb, and
I
C
D
V, Aux, P(?)
I, C
N, A, Q, Adv
D
[-Nominal]
[+Functional]
[- N]
[+N]

73
[+Nominal]

Pre/Post-position while the functional categories were complementizer (C) and Inflection (I).
Later works such as Abney’s DP-hypothesis (1987) and Pollock’s Split-Infl hypothesis (1989)
expand the domain of functional categories to include lexical determiners, pronouns, and all Infl
dominated elements. Pollock’s proposal particularly led to the elevation of elements previously
assumed to be dominated by Infl to the rank of independent functional categories, which
subsequently led to the proliferation of functional projections e.g. AgrP, TP, NegP, AspP, etc. in
generative grammar23.
In summary therefore, two major syntactic categories are assumed available in UG, namely
lexical categories (i.e. V, N, A, and P) and functional categories (i.e. Det, C, Deg, and I24).
While lexical categories are open-class contentive elements i.e. items with intrinsic descriptive
contents, functional categories are functors that belong to closed class sets of words, which head
functional projections in syntax.
3.2. Category Instantiations
Having established available syntactic categories within the context of UG, we shall now turn to
issues on language-specific instantiations of the noted categories, especially in respect of Igálà
and Yorùbá. This takes us to the question of whether or not all languages have or instantiate the
same set of categories, both lexical and functional, and whether the order of structural
projections of those categories is fixed for all languages or vary from language to language.
Generally, there appears to be a consensus on the universality of the lexical categories of N and
V in natural languages. In other words, all natural languages appear to attest nouns and verbs as
syntactic categories. For instance, none of the numerous studies on Yorùbá and the few ones
presently available on Igala grammar denies the existence of nouns and verbs as core syntactic
categories of their grammars. This claim however excludes the issue of what counts as a noun or
verb in a particular language, which requires separate language-internal evidence for proper
clarifications. Conversely, universal claims on other lexical categories such as adjectives,
adverbs and prepositions have continued to generate heated debates and controversies among
scholars. For instance, the issue of preposition as a syntactic category in Yorùbá is still being
debated. While studies like Awobuluyi (1967, 1978), Oyelaran (1989), Owolabi (1989),
Déchaine (1993), and Yusuf (1999, 2003) argue for its existence, others

74
including Bamgbose (1990), Awoyale (1994), Yusuf (1989, 1992), and Oduntan (2000) reject
such claim and analyzed items claimed to be prepositions by the former as verbs. Similarly,
scholars like Afolayan (1972) claimed the existence of predicative adjectives in Yorùbá but
others e.g. Ogunbowale (1970), Awobuluyi (1978), Bamgbose (1990), etc. refuted the claim.
They however agree that Yorùbá instantiate attributive adjectives. Déchaine (1993) raises an
objection by arguing that the syntactic category A (i.e. Adjective and Adverb) is totally absent in
Yorùbá. The Universality of functional categories is a by far more controversial issue. The bone
of contention is whether there is justification for claims that all proposed functional categories
have independent projections in some particular fixed orders, in all languages. For instance, is
there sufficient cross-linguistic evidence to prove that Agreement and Determiner projections are
universally attested in such a fixed manner that makes D items always head every nominal
projection as generally assumed in the minimalist program? Do all languages even instantiate all
of the noted functional categories?
Two separate hypotheses – the structural uniformity hypothesis (SUH) and the Limited Diversity
Hypothesis (LDH) – have been put forward in attempts to determine these in formal syntax.
There are two versions of the SUH. The strong version credited to Pollock (1989) and Chomsky
(1993) assumes clausal architecture is completely determined by UG in the sense that all clauses
in all languages have the same set of functional categories and their sequence (dominance
relation) is uniform. (Thráinsson 1996:225)
The weak version of the SUH presented by Iatridou (1990) and Rizzi (1993) assumes that
clausal architecture is determined by UG in the
sense that UG defines a set of functional categories,
{ F1, F2, …, Fn }, that languages select from. For any
functional categories, Fi and Fj , the sequence will be
uniform whenever they occur, i.e. if L1 and L2 each
instantiate both Fi and Fj and Fi c-commands Fj in L1
then Fi c-commands Fj in L2.
(Thráinsson 1996:225)
Basically, the structural uniformity Hypothesis (SUH) assumes UG defines sets of
functional categories and also fixes their dominance relations in all languages. The only
difference between the two versions is that the weak version assumes languages vary in

75
the inventories of functional categories they select from UG while the strong version assumes the
same sets for all. On the other hand, the Limited Diversity Hypothesis (LDH) as proposed by
Thráisson (1996) claims that clausal architecture is determined by UG in the sense that UG
defines a set of functional categories, {F1, F2 … Fn }, that languages select from. Crosslinguistic
and intra-linguistic variations are limited to the following:
- it is not the case that all FCs are instantiated in alllanguages.
- The FCs selected by a given language may not be
present in all clause types of that language.
- The sequence (c-command relations) of those
functional categories (dominance relations between
the functional projections) that are directly related to
morphological distinctions may vary from language
to language.
(Thráinsson 1996: 257)

The verdict here is glaring: UG defines sets of functional categories but cross-linguistic and
intra-linguistic factors determine those categories selected as well as their dominance order in a
given language. Giorgi and Pianessi also contest the claims of the structural uniformity
hypothesis. They assume that the universal component of grammar contains two sub-
components, namely, a set of universal features and the mechanisms that regulate the projection
of those features in syntactic derivation28. The latter in turn contain two sub mechanisms. The
first of the sub-mechanisms is called the Universal Ordering Constraint. It assumes Features are
ordered so that F1 > F2, the checking of F1 precedes the checking of F2. The second sub-
mechanism is the feature scattering principle, and it simply assumes: Each feature can head a
projection.(Giorgi & Pianessi: 1997: 14-15).
These two sub-mechanisms interact to yield hybrid and isolating categories that make up clause
structures. For instance, if F1> F2 > F3 is an ordered set of features. Each one of them can either scatter
and project separately to yield three isolating categories or fuse together in various forms to yield
different hybrid categories. In other words, the three can fuse to form a single hybrid category. Similarly,
F1 and F2 can also fuse to form another hybrid category that isolates and dominates F3. Lastly, F2 and F3
can fuse to yield another hybrid that leaves out F1 as a separate category. In all of these combinations, it
is the linear order of the projections that determines their dominance relations. There is no universally
fixed order for projections and c-commands depend on the natural linear arrangements permitted in the
particular language concerned. The implication of Giorgi & Pianessi’s claim is that the inventory of
categories and their structural ordering in syntax is not universally fixed; rather, the two vary from

76
language to language. As a result of this, language internal evidence must be presented to support claims
on issues of category instantiation and structural projections. Interestingly, this view is not entirely new as
earlier studies such as Awobuluyi (1978) have expressed similar opinion.
There is no fixed number of parts of speech that must be
found in every language. In order words, some languages
have more parts of speech than others. This being the case,
it is no use for someone wanting to find out how many parts
of speech there are in Yoruba to look to Hausa, or Igbo, or
English for guidance. Such a person must find guidance for
himself within the Yoruba language itself and there only.
(Awobuluyi 1978:1)

Therefore, following Awobuluyi (1978), Thráinsson (1996), Giorgi & Piasnessi (1997) among
others, the present study assumes UG sets of both lexical and functional categories from which
separate inventories are proposed for Igálà and Yorùbá on the strength of empirical intra-
linguistic evidence.
3.2.1 Traditional Assumptions on Igálà Categories
Previous writers on Igálà grammar appear unanimous in assuming the eight traditional ‘parts of
speech’ as syntactic categories for Igálà. These are noun, pronoun, adjective, verb, adverb,
preposition, conjuction, and interjection. Generally, in those previous studies, assumed
categories were notionally defined and words were assigned to categories mostly on the basis of
meaning or semantic interpretation. For instance, guiding definitions are of the following type.
A noun is the name of animals, places, objects …
pronouns are words that stand in place of nouns…
verbs are action, doing or saying words, … etc.

As expected of every traditional approach, most of the claims and conclusions drawn concerning
Igálà category inventory by those previous studies are spurious and often structurally misleading.
For instance, óṃ a ‘child’ and àtá ‘father’ are names (i.e. nouns, at least going by the traditional
definition). However, they were also, analyzed as adjectives in expressions like the one in (24).

77
24. Óṃ a fẹ èḍ ò ̣ àtá un (fèḍ ò)̣
Child want heart father 3sg-gen
‘The child loves his/her father.’

Similarly, already assumed pronouns such as mi ‘my’ and ma ‘’their’ were also analyzed as
possessive adjectives in expressions like those in (25a-b).

25a. Òṭ ákádà mi dé ̣ ì

Book my be this
‘This is my book.’
b. Ì la èḍ à ma
3sg buy shoe their
‘S/he bought their shoes.’

Though it may appear as if such works assume that some items can function in two or more
categories, they however fail to state that in any clear terms. The attendant structural lapses and
contradictions in those earlier studies constitute part of the motivation for the present study.
Their unanimous claim on interjection as a word category in Igálà exposed the depth of the
traditionally misleading outlook of those earlier studies. For instance, items such as âh!, ôh!,
héè!, íyóò, etc. were said to be examples of interjections that determine the exclamatory
interpretation of the Igálà
expressions in (26a-e).

26a. Âh! u mà mẹ
? I know 2pl.
‘I know you (pl.)!’
b. ôh! Ì lè kwú
? s/he go die
‘S/he died!’
c. Héè! éngínì mi che nyò ̣

78
? guinea-brocade my be good
‘My guinea-brocade is beautiful!’
d. Íyóò! Ì chánè ̣ mẹ
? 3sg do-ground 2pl
‘It is your (pl.) land!’
e. Ój̣ ó ̣ mi!
God my
‘My God!’ (Cited from Okpanachi 1996:29.)

This claim cannot be true anyway because the feature that determines whether a given expression
is exclamative, interrogative, imperative or even declarative is not dictated by the presence or
absence of the so-called interjection in (26). Rather, we assume that it is the illocutionary force
of an expression, i.e. the kind of speech act an expression is used to perform, that determines that
(Grice 1975, Radford 1997b). The fact that (26e) was assumed exclamative despite the absence
of the so-called interjections makes our claim more plausible. Similarly, varied forms of the so-
called interjection are items commonly used for non-linguistic acts like crying, sobbing or
weeping in most natural languages. It then implies that such a syntactic category should be
universal if at all it truly exists. Since that is never the case, we consider the claim for Igálà
untenable. Other cases of traditional assumptions on Igálà investigated in this study include the
so-called Igálà plural nouns, singular/plural verbs, place and time adverbs, and the socalled body
and place prepositions. We also investigate the status of some traditional adjectives. For instance,
we attempt to clarify whether items such as èṇ yò ̣‘good’ , èṛ ínyò ̣‘sweet/sweetness’, èḅ íéṇ ẹ
‘bad’, èḳ a, ‘sour’, etc. as used in (27) below truly qualify asdescriptive/attributive adjectives as
traditionally claimed.

27a. óḳ ò ̣ èṇ yò ̣ ‘good vehicle’

b. óbó(bó) èṛ ínyò ̣ ‘sweet soup’
c. Ukóḷ ó ̣ èḅ íéṇ ẹ ‘bad work’
d. ómi èḳ a ‘sour water’
e. úchu èḥ i ‘cooked yam’

79
Similarly, can we truly claim that Igálà words such as èmi ‘here’, òṃ ọ ‘there’, óg̣ ánè ̣‘down on
the floor’, etc. are adverbs structurally? All these and other Igálà related traditional assumptions
are critically but objectively examined in the present study.

3.2.2 The Present Approach

The approach to categorial classifications in the present study is functionally based. By this, we
mean that classifications of lexical items into categories are based on examining the universal
(primary) syntactic functions or properties of each category group vis-à-vis the syntactic
position(s) that words in each of the category groups notably occupy in the clause structure. In
addition to these, language particular properties of words in each group were established and
used to motivate the actual assignment of particular words into appropriate particular categories
in the language concerned. This approach is informed by the fact that words in particular
languages do have language specific properties or features which serve as some kind of
indicators that help in determining the word group to which they belong, universal primary
properties notwithstanding. For instance, the primary property of adverbs as lexical adjuncts
universally determines the Adverb category. However, it is on account of language particular
properties that words ending with the -ly suffix e.g. sadly, intelligently, speedily, etc. were
assigned to the adverb category in English. Interestingly, that same property of adverbializing -ly
suffix is also used as a criterion that identifies adjectives in English, as adjectives remain the
only set of items that permit the –ly suffix to derive adverbs in English (Radford 1997b:32).
However, using this same properly to identify adverbs or adjectives in some other languages, say
Yorùbá or Igálà, would no doubt be structurally misleading simply because that property is
English-specific, and languages differ within various limits from one to the other. The only way
out is to painstakingly identify those language-specific properties common to adverbs and
adjectives in either of these other two languages before any structurally meaningful assignment
of words into those two categories could be done for them.
Similarly, various Scholars (Awobuluyi 1967, 1978; Bamgbose 1967, 1990; Owolabi 1989; etc.)
assigned words such as bú ‘to abuse’, sè ‘to cook/boil’, lọ ‘to go’, etc. to the verb category in
Yorùbá given their unique ability to partially reduplicate for focus and relative structure
formation, as in lọ à lílo bú à bíbú, sè à sísè, etc.

80
However, verbs cannot be so identified in Igálà or English because the property of V in focus
there is peculiar to Yorùbá. On the contrary, lexical items may be assigned to the V category in
English based on their tense inflection properties, which is another English specific feature of
verbs. Therefore, the underlying assumption in the present study is that universal properties of
words only constitute the basis for distinguishing one category from another, while language-
particular properties, i.e. those which do not cut across all languages but are peculiar to each
language, play invaluable roles in the actual assignment of words to specific and appropriate
syntactic categories in every language.
3.3 Lexical Categories
There is ample structural evidence that Igálà and Yorùbá like other natural languages, have clear-
cut distinctions between nouns and verbs as separate syntactic categories30. In addition to these
two primary contentives, we shall argue against claims to the contrary that Adjective, Adverb,
and Preposition are part of the lexical inventory of both Yorùbá and Igálà. We shall also present
conceptual evidence to argue that words traditionally analyzed as place/time adverbs and
place/body prepositions in Igálà are nouns given their general and language-specific syntactic
properties. Because of the central role played by verbs in clause architecture, especially within
the minimalist framework adopted for this study, we shall begin our discussion on lexical
categories with verbs.
3.3.1 Verbs
Within the context of the minimalist program, a verb is a lexical item that assigns or licenses
theta roles performed by arguments, i.e. noun phrases, in clause structure. As we hope to show in
this chapter and chapter four, this definition takes care of both the syntax and semantics of verbs
in Igálà and Yorùbá clause structures. The term theta/thematic role and its assignment (or
licensing in MP) is better explained within the context of argument structure. The concept of
argument structure takes every simple clause or sentence to be a proposition comprising a
predicate and a set of structural arguments. A predicate (in this case, a verb)32 is an expression
that denotes a particular event or activity while an argument refers to the expression that denotes
a participant in the activity or event denoted by the predicate in a given proposition. The event
denoted by a predicate is assumed to comprise some inherent participatory roles (i.e. theta roles)
which the predicate assigns to its participant positions(s) (i.e. argument position(s)), for the
participant items taking part in the event or activity in question to saturate. In standard generative

81
practice (MP inclusive), every verb is assumed to have an argument structure where its primary
function is to assign Ө-roles to argument positions licensed or allowed in the structure, where
arguments (nouns/nominal items) are mapped to saturate such roles. Theta/Ө-roles are semantic
roles such as AGENT, THEME/PATIENT, GOAL, BENEFACTIVE, EXPERIENCER, etc.
assumedly performed by arguments in relation to their predicate within a given argument
structure. A verb may be a one-, two-, or three- place predicate depending on the number of
arguments it has. It is also pertinent to note that verbs do not assign theta-roles to nonargument
positions, i.e. syntactic positions that do not (directly) participate in the event or activity denoted
by the V. Two such positions are Adjunct and Spec-CP positions. Let us illustrate these
assumptions with the Igálà expressions in (28) and (29) below:
28a. Ányájá kpa ẹla
Trader kill animal
‘The trader killed an animal.’
b. *Ányájá kpa
Trader kill
c. Ój̣ óṇ ùgwà jẹ òj̣ è ̣ òṇ álé ̣
Ój̣ óṇ ùgwà eat food yesterday
‘Ój̣ óṇ ùgwà ate (some food) yesterday.’
d. *Ójóṇ ùgwà jẹ cf. Ójóṇ ùgwà jẹ/nmọ ( ___ )
Ój̣ óṇ ùgwà eat ‘Ój̣ óṇ ùgwà ate/drank.’
e. *Ójóṇ ùgwà jẹ òṇ álé ̣
Ój̣ óṇ ùgwà eat yesterday
‘*Ój̣ óṇ ùgwà ate yesterday (as food).’
29a. Íjóṇ ì lè (w)á
John go come
‘John came.’
b. *Íjóṇ ì lè (w)á òṭ ákádà mi
John go come book my
‘*John came my book.’
c. Íjóṇ ì lè (w)á únyí lè ̣
John go come house that

82
‘John came to that house.’
The argument structure of the V kpa in (28a-b) specifies it is a two-place predicate that provides
for two argument positions to which it assigns two participant roles: AGENT, i.e. an entity that
causes or instigates the event denoted by kpa ‘to kill’, and PATIENT, i.e. the entity directly
affected by the event. It is therefore expected that two structurally appropriate participating
arguments would be mapped to the two argument positions to saturate the Ө-roles assigned.
These are structurally obligatory for the projection of kpa as a predicate in Igálà to be
syntactically well-formed. In (28a) where the two licensed argument positions and Ө-roles were
appropriately saturated by Ányájá ‘trader’ in the grammatical function of subject as AGENT and
ẹla in the accusative (object) position of V as PATIENT, the resultant syntactic derivation is
adjudged well-formed while in (28b) where the PATIENT argument is missing, the derivation is
adjudged to be ill-formed. The same explanation accounts for the well-formedness of (28c) and
the possible illformedness of (28d) which could also be also be interpreted as a well-formed
elliptical sentence where the object of jẹ is not phonetically realized but structurally present and
understood. (28e) is a little bit different, as òṇ álẹ́ ‘yesterday’ may be superficially mistaken for
the PATIENT argument of the V jẹ ‘to eat’ there. However, òṇ alé ̣has no direct structural object
association with jẹ there, as it is not a participant in the eating event. It only provides additional
information on the event by specifying the exact day in which the eating took place. We
therefore submit that òṇ álé ̣occupies a structural adjunct position which does not constitute an
argument position of jẹ in (28e). Conversely, the V wá as used in (29) is a one-place predicate
that licenses only the AGENT/ACTOR marked argument (i.e. an entity/person who causes some
state of affairs to come about, sometimes deliberately.) position, which is saturated by Íjóṇ ì as
the grammatical subject of wá in the well-formed (29a & c). The ill-formedness of (29b) results
from the erroneous projection of two arguments for wá which structurally licenses only one Ө-
marked argument. In a nutshell, the second argument òṭákádà mi ‘my book’ in (29b) is
redundant and thereby leads to the ungrammaticality noted there. It is also pertinent to remark
that únyí lè ̣‘that house’ in (29c) is not an argument of wá. It simply performs a function similar
to that of òṇ álé ̣in (28c) by providing additional information on the activity denoted by wá. The
information this time around has to do with the place/goal of the event denoted.

3.3.1.1 Simple Verbs

83
These are regarded as simple syllable verbs which possess among other things the structural ability to
function independently in a simple clause construction without the assistance of any other verb. They
include the traditional transitive and intransitive verbs such as kpa ‘kill’, wá ‘come’, jẹ ‘eat’, rà
‘buy’, etc. already exemplified in (28-30) above; verbs ‘to-be’, such as Igálà che ‘do/be’, dé ̣ ‘is’, jọ
‘are’, and Yorùbá ṣe ‘be/do’ and jé ̣ as used in (31) and (32) below.
31a. Òmi che ítíchà ( ù che ) (Igálà)
I do/be teacher ( I be )
‘I am a teacher.’
b. Áúdù dé ̣ èmi
Aúdù be here
‘Audu is here.’
c. Àm̀ bógìjo jò ̣ únyí ónú
Old-folks be(pl.) house chief
‘The old people are at the chief’s house.’
32a. Olówó kì í ṣe Ọlóṛ un (Yorùbá)
The-rich neg prog be God
‘The rich is not God.’
b. Èdè jé ̣ àmì ìdánimò ̣ èḍ á
Language be sign identification human
‘Language is a sign for human identification.’
Other verbs in this group are attributive verbs, cognate object verbs, and those called
interrogative verbs in relevant literature. Attributive verbs, as the name implies, are verbs
with attributive meanings that Awobuluyi (1978:56) called adjectivisable verbs.
Examples of this are those items highlighted in (33) and (34) below.
33a. Àtá un nya nána (Igálà)
Father 3sg short very
‘His father is very short.’
b. Óỵ à mi nyò ̣ gbèẹ̀ ̣
Wife my beautiful very
‘My wife is very beautiful.’
c. Ómi úchu lẹ ka
Water yam the sour
‘That yam water is sour.’
d. Álèmu kì ì kpa
Orange that 3sg ripe
‘Orange that is ripe.’
e. Àfè ̣ íye mi gbẹ
Shirt mother my dry
‘My mother’s shirt (top dress) is dry.’
34a. Ọkùnrin náà ga (Yorùbá)
Man the tall
‘The man is tall.’
b. Àṣeju àwọn olóṣèlú Nàìjíríà pò ̣
Excess they politician Nigeria many
‘The excesses of Nigerian politicians are many.’
c. Àgbá epo náà kún déṃ údéṃ ú
Drum oil the full to-the-brim (ideophonic)
‘The oil drum is full to the brim.’

84
d. Ara ejò náà ń dán
Body snake the prog shine
‘The body of the snake is shining.’

3.3.1.2 Complex Verbs

This term, as used here, refers to Igálà and Yorùbá verbs that have more than one
syllable. They are sub-divided into four: those that are fusions or combinations of two
simple verbs, i.e. V +V à V; those that are combinations of a verb plus a noun object or
noun object incorporation, i.e. V+N à V; those that have more than one syntactic
function e.g. N/A that also functions as V, i.e. N/A à V; and those verbs that are
outright loans from other languages. The criteria used for classifying verbs into this
group are based on the internal structures and function(s) of the items concerned.
3.3.1.2.1 V + V Combinations
There are two subgroups of this type of verbs. The first contains verbal
combinations that have the structural ability to split into halves to sandwich their nominal
object. These types of verbs are those commonly called splitting verbs in the Yorùbá
linguistic literature. Every verb in this subgroup usually splits into two halves to
64
sandwich its object. One property of this subgroup of verbs is that when they split, each
half of the phonological sequence is often rendered meaningless, and where such a half
has any meaning at all, such a meaning is often literal and incoherent with the overall
logical interpretation of the whole verb sequence. According to Awobuluyi (1978:53),
many of the Yorùbá splitting verbs have idiomatic meanings e.g. bá...wí ‘to scold’,
bà...jé ̣ ‘to spoil’, dì…mú ‘to hold’, ré.̣..jẹ ‘to cheat’, bè/̣yè ̣…wò ‘to inspect or pay a
visit’, pa…dà ‘turn around’, gbá…mú ‘to lay hold or catch’, gbà…gbó ̣ ‘to believe’,
etc., as they occur in (38) below.
38a. Àwọn olóṣèlú ń ré ̣ ará ìlú jẹ (Yorùbá)
They politician prog cut folk town eat
‘Politicians are cheating the populace.’
b. Olùkó ̣ bá Ṣọlá wí
Teacher meet -- say
‘The teacher scolded Sola.’
c. Tóḷ á ba fóònù mi jé ̣
? phone mu ?
‘Tola spoilt my telephone set.’
d. Mo gba Ọlóṛ un gbó ̣
I accept God hear
‘I believe in God.’
e. Wóṇ di déṛ éḅ à náà mú
They-HTS tie driver the catch
‘They hold the dirver.’
f. Mo gbá ọ mú ní oní
I ? you catch at today
‘I caught you today.’
The structural behaviour of Igálà splitting verbs such as kó…̣ dá ‘to cut’, kpa…bíé,̣ ‘to
spoil’, dú…ngà ‘to show’, gu…jo ‘to burn’, che...kpa, ‘to heal’, rì…dà ‘to turn’,
bò…má ‘to cover’, etc. is not very different from those of the Yorùbá verbs highlighted

85
in (38). (39a-g) are examples of their use in the Igálà clause structure.
65
39a. Ṣúáíbù kó ̣ úchu lẹ dá (Igálà)
? yam the ?
‘Ṣuaibu cut the yam.’
b. Ì kpa àfè ̣ mi bíé ̣
3sg ? shirt my spoil
‘S/he spoilt my shirt.’
c. Àmachukóḷó ̣ gu únyí lẹ jó
workers put house the burn
‘The workers burnt the house.’
d. Ój̣ ó ̣ che òg̣ á mi kpa
God do sickness my kill
‘God healed my sickness/disease.’
e. du òṭákádà lẹ ngà mi
bring book the show me
‘Show me the book.’
f. Aládi rì òṭ ákádà dà
? book ?pour
‘Aladi turned the book (page).’
g. Ù bò ùchà má
I cover pot ?
‘I covered the pot.’
Unlike the splitting verbs, the other subgroup of the V+V combinations contains
verbs which do not split to sandwich their object, but fuse together as V units to bring out
the intended meaning(s). Examples of such verbs in Igálà are réwá ‘to remember’, gúdú
‘to enter’, gwùgwú ‘to sit’, tèg̣ wù ‘to climb’, jàdù ‘to save’, bíéṇ ẹ ‘to be bad’, dágo ‘to
stand’, kpalu ‘to clean’, dàchí ‘to lie down/sleep’, etc., as in (40).
40a. Ù réwá fàífàí (ré + wá) (Igálà)
I remember now-now ? come
‘I have just remembered.’

complex verb.
3.3.1.2.2 V+N Combinations
Structurally, this verb type contains a sequence of simple verb and its licensed
noun object complement. A group of such verbs already have their noun object
incorporated into the V function to such an extent that the output of the combination is
semantically inseparable with somewhat idiomatic interpretation that can no longer be
easily deduced from the literal meanings of the V+N input. Yorùbá verbs in this group
are given in (42) and exemplified in (43).
42. V + N V (Yorùbá)
a. jé ̣ ọwó ̣ à jéẉ ọ́
answer hand ‘confess’
b. pa èḷé ̣ à pèḷé ̣ cf. pa gúló-̣ gúló ̣
? ease ‘Sorry/easy please.’ ‘?to cringe’
c. rán36 etí à rántí
repeat ear ‘remember’

86
3.3.1.2.3 Function Shifting Verbs
This term is used in this thesis to denote verbs that perform other
lexical/categorial function(s) in the grammars of Igálà and Yorùbá. Such verbs are
regarded as complex because they have more than one syllable, each. In Yorùbá, most of
the verbs in this group are derived from nominal roots through conversion, i.e. N à V.
These are wàhálà ‘trouble à to bother someone’, gààrí ‘cassava grits à to cater for’,
pàtàkì ‘importance à to make oneself feel important’, and derived attributive items like
73
funfun ‘white àto be white’, dúdú ‘black à to be black’, pupa ‘red à to be red’,
tuntun ‘new à to be new’, etc. as used in expressions like those in (50)42.
50a. Ọmọ náà wáhálà mi
child the trouble me
‘The child troubled me.’
b. Ìwọ kó ̣ ni o ń gààrí mi
You neg Foc 2sg Prog cater 1sg
‘You are not the one that caters for my needs.’
c. Mo yára pàtàkì ara-mi
1sg fast-body made-important body-my
‘I quickly made myself important.’
d. Mótò náà án funfun/dúdú/pupa/tuntun
car the HTS white/black /red / new
‘The car is white/black/red/new.’
It is pertinent to mention that words called derived attributive items here also function as
adjectives in Yorùbá because their attributive features narrow the semantic scope of any
noun that they modify, as in (51). We are however of the view that the adjective forms of
these items derive from already formed nominal roots through N à A conversion.

3.3.2 Nouns
The term noun, as used here, refers to R-expressions, i.e. contentive nominal word items. It excludes
functional nominal words which are generally [+referential] because they are non-contentives.
Generally, there are two properties identified with nouns within the context of UG in the MP. The
first, closely tied to the primary function of V as thematic role assigner, has to do with the syntactic
positions that nouns occupy in the argument structure vis-a-vis the function they perform in such
positions. In the minimalist framework, argument positions are the preserves of noun phrases – DPs
or NPs –, which serve as participants in the thematic structure of the predicate. In other words,
noun/nominal projections are the participants/arguments that saturate or realize the Ө-role(s) licensed
by the verb. Thematic roles e.g. agent, theme, experiencer, etc. represent the semantic functions
performed by nominal projections while their syntactic function is determined by the exact syntactic
position they occupy in the base form of a syntactic derivation. For example, given the VP-internal
subject hypothesis, which assumes the grammatical function subject (nominative) and object
(accusative) positions are basegenerated within the VP shells, two derivationally linked but
syntactically defined subject positions have emerged in the minimalist clause architecture viz. Spec-
vP and Spec-IP. The subject argument is assumed to be base-generated in spec-vP and get raised to
spec- IP to check its formal features against the Infl head. Similarly, the object argument is assumed
to be base-generated in spec-VP and raises to a spec-of-v to check its accusative case feature
(Chomsky 1995). Although following Chomsky (1993) AgrSP and AgrOP, some other studies have
argued that the accusative argument raises to the spec of a VPinternal functional projection with

87
different suggestions on the exact label of the functional projection, e.g. Agreement (Koizumi 1995,
Fujita 1996; Radford 1997b), Aspect (Travis 1991, 1999; Slabakova 1997), and Predicate (Bowers
1993), the fact remains that all of them were unanimous on Spec-VP as the base source of accusative
argument.
Therefore, two sub-classes of noun/nominal words are recognised in this study:
lexical nouns which are usually called ‘nouns’ and functional noun/nominal words often
called ‘functional D items’ or ‘functional nominal words’ in this thesis. The latter are
fully discussed under functional categories (§3.4.2.) and chapter six. The diagrammatic

sketch in (70) depicts the group members of these sub-classes of noun/nominal words in
Igálà and Yorùbá.

Noun/Nominal Words
Lexical Nouns
(Contentives)
Functional Determiners (D)
(Non-Contentives)
Canonic Determiners
Pronouns
Articles Demonstratives
Definite Indefinite
Personal Anaphor Quantifier interrro- Relative Genitive
gative
Mono- Bi- Mono- Bi- reflexives reciprocals +HUM –HUM +HUM
syllabic syllabic

3.3.2.1 Iglálà Nouns

The two universal properties of nouns enumerated above readily admit most of
the items that have been analyzed as nouns in previous studies on Igálà grammar. These
include items like ónèḳ èḷẹ ‘man/male’, ónóbùlẹ ‘woman/female’, éḷa ‘animal’, óṃ a
‘child’, ùfèḍ ò ̣ ‘love’, ányò ̣‘joy’, ólí ‘tree’, únyí ‘house’, ómi ‘water’, ébútú ‘sand’, etc.
These are used as case-marked thematic arguments or lexical heads of thematic
arguments in the Igálà clause structure. For example in (71), the nouns óṃ a and únyí
function as lexical heads of the bracketed nominative and accusative arguments of the
verb kó.̣
71a. [Óṃ a un] kó ̣ [únyí titọ]
child 3sg build house new
‘His/her child built a new house.’
In addition to these, there remain some Igálà lexical items which the present study
considers to be nouns but which were traditionally grouped into some other categories
based mostly on their semantic interpretations. Such items include òṇ à ‘tomorrow’, òṇ álé ̣
‘yesterday’, òṃ ọ ‘there’, èmi ‘here’, etc. analyzed as adverbs because they translate
English adverbs and occupy the canonical adjunct position of adverbs as shown in (72).
72a. Ádágba á che ukóḷ ó ̣ òṇ à
---- Fut do work tomorrow

88
‘Adagba will work tomorrow.’
b. Ifidé hi úchu òṇ álé ̣
Fidelis cook yam yesterday
‘Fidelis cooked yam yesterday.’
c. Àtá mi dé ̣ òṃ ọ/èmi
Father my be there/here
‘My father is there/here.’
Similarly, items such as éfù ‘stomach, inside,’ étí ‘ear, side’, etc were analyzed as
prepositions in the context typified in (73).

3.3.4 Adverbs
Structurally, adverbs are VP lexical adjuncts typically used to specify manner or
degree of the event or activitiy denoted by the V head of VP. Because adverbs are
adjuncts, they are optional elements within the clause structure and can therefore easily
be done away with without affecting the grammaticality of the sentence concerned. They
often occur as clause final adjunct words. For example, gbíti and nyòṇ yò ̣ are adverbs
modifying the V che ‘to do’ in (119a). However, their absence in (119b) does not render
the sentence ungrammatical in anyway.
119a. Adaga che ukóḷó ̣ gbíti/nyòṇ yò ̣
------ do work hard/well
‘Adaga worked hard/well.’
b. Adaga che ukóḷ ó ̣
------ do work
‘Adaga worked.’
There are other adjuncts within the VP which are not lexical. They are combinations of
items which are traditionally called adverbials, which name location or direction of
events/activities and proceed to identify such location/direction with a goal or space noun
phrase. Such phrasal adjuncts are now called prepositional phrases (PPs), which are fully
discussed later in §3.3.5.
3.3.4.1 Adverbs in Igálà
Two types of adverbs are traditionally identified for Igálà based on their location
in relation to the verb they modify in the Igálà clause structure. The first group contains
híká ‘quickly’ and tákpa ‘early’ which often linearly precede the verb that they modify.
Their use is illustrated below in (120) and (121).
120a. Ùwẹ híká wá kì á ló
2sg quickly come that 1pl go
‘You come quickly and let us go.’
b. Ù híká che ukóḷ ó ̣ mi
1sg quickly do work my
‘I quickly did my work.’
114
121a. Ì tákpa wá èmi (wémi)
3sg early come here
‘S/he came here early.’
b. Énè ̣ dé ̣ éfù tákpa wá
Person be inside early come
‘The person inside came early.’

89
Although these items translate English manner adverbs, they share some characteristic
properties of Igálà verbs. In fact, they behave like verbs in serial verbal constructions in
(120&121). This is more obivious in (122) and (123) where they are imperatively used.

3.3.5.2.1 Pèḷú: A Preposition or Conjunction?

Almost all of the studies that assumed preposition for Yorùbá analyzed pèḷú
‘with/and’ as a member of the Yorùbá P category (Awobuluyi 1978, Bamgbose 1990,
Yusuf 1999, etc.). However, this view appears to be informed by the literal meaning of
pèḷú in the following expressions.
153a. Ó ṣe é pèḷú u-túláàsì (Awobuluyi 1978:99)
HTS do it with force
‘He did it by force.’
b. Olóyè dé pèḷú ìlù àti ariwo (Yusuf 1999:22)
Chief arrive with drum and shout
‘The chief arrived with drums and loud ovation.’
c. Adé lọ pèḷú u-wọn
---- go with 3pl
‘Ade went with them.’/ ‘Ade went to join them.’
It is pertinent to mention that pèḷú is also generally accepted in the the literature as a
nominal conjunction based on its meaning in contexts like those in (154).
154a. Mo rí Akin pèḷú (u-)Ṣadé
1sg see ---- with/and ----
‘I saw Akin with/and Ṣade.’
b. Èmi pèḷú u-Táyé
‘I with/and Táyé.’
However, morphological and syntactic evidence suggest, quite contrary, that pèḷú is
neither a preposition nor a conjunction but a VP with the V+N structure, i.e. pa ‘to make
ready’ + è-̣lú ‘mixture’ à pèḷú ‘to mix with’. The morphology of pèḷú becomes clearer
when it is compared to other items derived in the same manner, as done in (155).
155a. pa + è-̣ lú ‘mix-ure’ à pèḷú ‘to mix with’
b. pa + è-sè ‘provision’ à pèsè ‘to make provision’
c. pa + è-̣dín ‘reduction’ à pèḍ ín ‘to reduce’
d. pa + è-lé ‘increase’ à pèlé ‘to increase’
e. pa + èḷé ‘?easy’ à pèḷé ‘?to take it easy’
f. pa + èṣ è ̣ ‘calm’ à pèṣ è ̣‘to be calm’
134
g. pa + èḳ a ‘branch à pèḳ a ‘to make/have branches’
h. pa + àtẹ ‘wares’ à pàtẹ ‘to make ready or display wares’
As evident in (155a) it is the noun object of pa, i.e. èḷú, that possesses the ‘(mix-) with’
interpretation generally associated with pèḷú.
Sytactically, any consonant initial nominal item that follows pèḷú usually takes
the prosthetic vowel or MTS as evident in (153a&c) and (154a&b). This is clear evidence
that the item immediately preceding such nominal is itself a nominal item, i.e. èḷú.
Another piece of syntactic evidence that shows that pèḷú is actually a VP is found in
expressions like
156a. kí Olúwa kí [ ó pèḷú u-yín ]
Comp Lord Comp HTS mix-with 2pl

90
‘may the Lord be with you.’
b. Akín pa-èḷú à Akínpèḷú
Akin-HTS mix-with a Yorùbá name
‘Akin was with them’
In (156), the HTS which signifies that an expression is a clause rather than a phrase
immediately precedes pèḷú. The implication of that is that pèḷú in those expressions is a
VP. It logically follows that contexts where another V linearly precedes pèḷú as in (153ac)
are serial verbal constructions64. The only way to sustain the preposition and
conjunction claims for pèḷú is to say that the VP pa èḷú (cf. also pa àdánù) takes up
those functions through conversion.
3.3.5.2.2 ti: A Verb or Preposition?
Two claims exist on ti. First, it is assumed to be a preverbal preposition
(Awobuluyi 1978, 1982; Fágbóṛ ún 1998; and Yusuf 1999) in the contexts like (157).
157a. Adé ti ìbè ̣ jáde
---- from there come-out
‘Adé emerged/came out from there.’
______________
64. Awobuluyi (1978:99) expressed similar doubt on the P status of pèḷú, though he called it a preposition.
He later called it a verb (Pg.100) given its structural behaviour in ó ṣe é pèḷú u-túláàsì’ (153a) where
túláàsì was assumed to be a noun object of the V pèḷú. See also Awobuluyi (2008a:164).
135
b. Mo ti ibè ̣ dé ní àárò ̣ (Awobuluyi 1978:98)
1sg from there arrive at morning
‘I came back from there in the morning.’
c. Yẹmi ti Ìbàdàn lọ sí Èkó
---- from ---- go Loc Lagos
‘Yemi left for Lagos from Ibadan.’
d. ó ti àárò ̣ bèṛ è ̣ ìṣekúṣe (Awobuluyi 1978:98)
HTS from morning begin bad-act/behaviour
‘He began behaving badly from his youth.’
Secondly, ti is assumed to be a locative verb in serial verbal construction in similar
structure (Ọduntan 2000: 128):
ti is a regular lexical item, which some have
called a preposition, and which the present
study takes to be a verb … ti is a regular verb
that licenses locative or time NP.
e.g.
158a. Adé ti oko wá
--- V farm come
‘Ade came from the farm.’
b. Adé ti ilé rè ̣ lọ sí kíláàsì
---- V house 3sg-Gen go v-goal class
‘Ade left for class from his house.’
c. Adé ti orí àga gbé oúnjẹ
---- v-Loc top chair carry food
‘Adé carried the food from the chair.’
First and foremost, the locative-V claim on ti is suspect. The so-called V-ti is highly
restrictive functionally as it can only be used in a serial verbal construction. It does not
function alone as a verb in a sentence. In addition to this, ti as a V does not possess any

91
known UG and Yorùbá language specific properties of V. For instance, it cannot be
partially reduplicated to derive nouns for focusing or relativization unlike other Yorùbá
verbs. In other words, there is nothing like *títi derived from the so-called V ti in
136
Yorùbá. Witness the illformedness of (159a-d) where ti is assumed to be a V and the
well-formed true lexical verbs are used in simlar contexts of serial verbal constructions.
159a. * Títi ni Adé ti oko wá
? foc Ade V farm come
b. *Títi-wá ni Adé ti oko wá
? foc Adé V farm come
c. *Títi/ Títi-gbé ni Adé ti orí aga gbé ounjẹ
? ? foc Adé v top chair carry food
d. *Títi/Títi-wá tí Adé ti oko wá
? ? rel Ade V farm come
e. Jíjí/Jíjí-jẹ ni wóṇ jí ẹran jẹ
stealing/stealing-to-eat foc they steal meat eat
‘It was a fact that they stole and ate the meat.’
f. fífi/fífi-kún tí wóṇ fi kún owó epo
taking/adding rel they take add money oil
‘The fact that they increased the price of oil’
h. fífi/fífikún ni wóṇ fi kún owó epo
taking/adding foc they take add money oil
‘It was the fact that they increased the price of oil.’
We therefore conclude pending further light on the issue that ti is better analyzed as a
preposition rather than a verb.
3.3.5.2.3 ní and sí as Prepositions
Various studies have argued in favour of ní and sí as items that qualify as
syntactic prepositions in Yorùbá (Awobuluyi 1978, Oyelaran 1989, Owolabi 1989,
Déchaine 1993, etc.). We concur with this view, and take ní and sí to be instantiations of
the Locative/Goal preposition in (160a–d).
160a. Tádè lọ sí ọjà
---- go to market (Place Location/Goal)
‘Tade went to the market.’
137
b. Ọmọ mi wà ní ilé-ìwé (Place Location)
Child my be at/in school
‘My child is in school.’
c. Ó dé ní àárò ̣ (Time Location)
HTS arrive at/in morning
‘He arrived in the morning.’
d. Wóṇ fi ìpàdé sí ìròḷ é ̣ (Time Locative)
3pl-HTS put meeting to evening
‘They fixed the meeting for evening.’
Although, studies like Awoyale (1975, 1995) and Ọduntan (2000) have argued that ní and
sí are V-locative and V-goal verbs, the universal and langauge-internal properties of these
two items clearly differentiate them from verbs in Yorùbá. Firstly, unlike verbs, they do
not occur as predicators as they never follow any Aux/Infl items directly. Secondly, they
do not undergo the morphological process of consonant prefixation commonly used to

92
nominalize Yorùbá verbs for focusing and relativization. For instance, (161a-d) are not
well-formed because ní and sí were made to undergo that process reserved for verbs only
in Yorùbá.
161a. *Níní ni ọmo mi wà ní iléèwé
? Foc child my be V-Loc school
b. *Níní tí ọmọ mi wà ní iléèwé
? rel child my be V-Loc school
d. *Sísí tí Tádé lọ sí ọjà
? rel ----- go V-Goal market
Thirdly, the projection that sí and ní head in the Yorùbá clause structure is an adjunct
phrase which is optionally selected in the VP. This is shown by the fact that this post-V
adjunct phrase can be moved into Spec-CP for emphasis, as illustrated in (162).
162a. Mo wà [ní Èkó]
1sg be Loc Lagos
‘I am in Lagos.’
b. [(ní) Èkó]i ni mo wà ti
Loc Lagos Foc 1sg be
138
‘I am actually in Lagos.’
c. Mo rán Ṣadé [ sí ilé ]
1sg send ----- Loc/Goal house
‘I sent Ṣadé to the house.’
d. [(*sí) ilé ] ni mo rán Ṣadé (sí)65
(*to) house Foc 1sg send ----- (to)
‘It was the house that I actually sent Ṣadé.’
On the other hand, verb projections cannot be moved in Yorùbá e.g.
163a. *[V rán Ṣadé]i mo ti sí ilé
b. *[V wà ]i ọmọ mi ti ní iléèwé
All of these in addition to the fact that P is universally defined by the semantic/syntactic
features of GOAL and LOCATION (Emond 1985:30; Déchaine 1993:78) which is
clearly fulfilled by sí and ní support the claim that si- and ní-phrases in Yorùbá are not
VPs, but PPs.
It is pertinent to point out that Ọduntan (2000) in his bid to justify his claim that sí
and ní are V-goal and V-locative items, structurally confused the antifocus particle ní
with the Locative P ní, and that appeared to have informed his conclusion that Yorùbá níphrase
is a VP internal focus item66. While we are not making any statement on the
appropriateness or otherwise of the internal focus function of the antifocus ní, there is no
doubt that the Locative ní and Loc/Direction sí head a P and not a V projection in
Yorùbá.
There is also another sí ‘be’ item which is a negative polarity verb in Yorùbá,
illustrated in (164) below.
164a. Oúnjẹ (k)ò sí
Food neg be ‘There is no food.’

3.4 Functional Categories

These are categories of syntax that have no descriptive content as such but
primarily serve as carriers of information coded in the grammatical properties of content

93
words, i.e. lexical categories, in the clause structure. Such properties are number, person,
gender, case, etc. One inherent characteristic feature of functional categories is that they
have closed membership. In other words, new entrants, either through lexical borrowing
or morphological derivation, are not permitted into their ranks.
In the minimalist grammar, words outside the domain of V, N, A, and P belong to
one functional category or the other. Three main layers of functional categories are
identified in the minimalist framework. These are the Infl layer, which comprises
elements like Tense, Aspect, Agreement, Negation, etc.; the Complementizer layer which
houses complex clause introducers such as focus, relative and conditional clause markers;
and the functional D layer which subsumes determiners, demonstratives, quantifiers,
personal pronouns, degree words, and probably conjunctions68. Given the assumption of
the Split-Infl hypothesis, which assumes each of the functional categories has separate
and independent projections, each functional category shall be treated independently, one
after the other. Our aim in this section therefore is to highlight and structurally justify,
using empirically motivated evidence, those functional items as attested at the functional
layers of Igálà and Yorùbá.
3.4.1 The Infl Layer
Membership of the Infl layer includes finite and non-finite auxillary items like
Tense, Aspect, Agreement, Negation, etc. Our concern is to identify items attested at
these syntactic points and how they are structurally patterned in the syntax of Yorùbá and
Igálà.
3.4.1.1 Tense
This is a grammatical element of Infl that relates the time of speech to the actual time of
the event or activity denoted by the verb. A binary time-reference contrast or tense
distinction is generally assumed in syntax to describe an event taking place in the present
(or future) – i.e. present tense – as against one that has already taken place some time in
the past – i.e. past tense. Verbs are said to carry either the past or non-past tense feature
of tensed clauses in the minimalist framework.
3.4.1.1.1 Tense in Igálà
Igálá appears to mark tense in two relatively distinct ways. These are the nonfuture
and the future tenses. The non-future tense comprises both the present and past
tenses which do not appear to have any visible morphological or phonetic content. Hence,
they are often realized null, superficially, at Infl. The null non-future tense in Igálà is
illustrated in (165).
165a. Ì la úchu kâ (òṇ álé/̣òdùdu-í ) (Past)
3sg (?) buy yam one (yesterday/morning-this)
‘S/he bought a piece of yam (yesterday/this morning).’
b. Ì gba òṭ ákádà lé ̣ (ìkò lé ̣)
3sg (?) read book the (time that)
‘S/he read the book (that time).’
141
c. Àtá du óḳ ó ̣ nwú ma (èg̣ bà kì wá )
father (?) take money give 3pl (when comp-3sg come)
‘Father gave them money (when he came).’
d. Ma jọ òṃ ọ (fàí-fàí) (Present)
3pl (?) be there (now-now)
‘They are there (now).’
e. Íye un dé ̣ ájá (ój̣ ó ̣ dúú)

94
Mother 3sg (?) be market (day every)
‘His mother is usually in the market (every day)’
In (165), no tense item is phonetically visible at Infl, yet the verbs there are interpreted as
past (165a-c) and present (165d&e). Optionally, the time notion in these expressions can
be complemented to reflect the (near) exact time of event by using past/present VP
adjuncts like those highlighted there. One piece of evidence which shows that tense in
Igálà expressions like those in (165a-e) is non-future is found in the fact that such
expressions exclude future time adjuncts. For instance, (166a-c) are ill-formed for this
reason.
166a. *Ì la úchu kâ òṇ à
3sg (?) buy yam one tommorrow
‘*S/he bought a piece of yam tommorow.’
b. *Àtá du óḳ ó ̣ nwú ma òṇ à
Father (?) take money give 3pl tomorrow
‘*Àtá gave them money tommorrow.’
c. *Ma jọ òṃ ọ òḍ ẹdẹ òṇ à
3pl (?) be there all tommorow
‘They are there all through tommorrow.’
At this juncture, it is pertinent to remark on the status of the item lè, which earlier studies
on Igálà grammar have called a past tense marker in the following expressions.
167a. Ijóṇ ì lè (w)á
John ?Pst come
‘John came.’
142
b. Ónú lè kwú ß Ónú lè (tú úgbó ú-)kwú
chief pst-go die chief go to forest death
‘The chief died.’
First, the syntax of the idiomatic base form of (167b) shows that lè is a verb meaning ‘to
go’. Second, sructural facts from other clause expressions where lè is used, e.g. (168a-b),
equally indicate that lè is a verbal element, as it occurs as the only verb there.
168a. Ma lè tú ájá
3pl go to market
‘They went to the market.’
b. Ì lè mèẹ́ ̣
3sg go complete/finish
‘S/he has gone.’
c. élè kì lè tú ájá
going comp-3sg go Loc market
‘His/er going to the market’
Thirdly, lè can be nominalized with the gerundive prefix é- as in (168c) just like other
Igálà verbs. We therefore conclude that lè in the expressions in (167a-b) is not a past
tense marker but the first in a sequence of two verbs in a complex VP. The past tense
element in such expressions is phonetically null, as we have earlier observed.
On the other hand, the future tense in Igálà is marked at Infl by á ‘will’ and/or á
nyá ‘will be going’, and we strongly believe that nyá is a combination of the V nyú ‘go’
+ á ‘progressive aspect’). This is illustrated below in (169-171).
169a. Achílè ̣ á rọ ibóḷ ù cf. Achílè rọ ibóḷ ù
----- will play ball ----- play ball

95
‘Achile will play ball.’ ‘Achile played ball.’
b. Achílè á nyá rọ ibóḷ ù
----- will go-Prog play ball
‘Achile will be playing ball.’
c. (Ù)nà á ló cf. Ù/*Nà ló
1sg will go 1sg go
‘I will go.’ ‘I went.’
143
d. (Ù)nà á nyá ló
1sg will go-prog go
‘I will be going.’
170a. À á gbó ̣ cf. À gbó ̣
1pl will hear 1pl hear
‘We will hear/listen.’ ‘We heard.’
b. À á nyá gbó ̣
1pl will go-prog hear
‘We will be hearing/listening.’
171a. Mà á gbó ̣
3pl will hear
‘They will hear/listen.’
b. Mà á nyá gbó ̣
3pl will go-prog hear
‘We will be hearing/listening.’
We therefore, conclude that Igálà projects TP at the Infl layer using two syntactic
elements. While the non-future tense is projected by an empty category Ø, the future
tense is projected by á. It should be noted that Igálà has another á which marks the
progressive aspect. It is this aspect that occurs as a part of nyá in (169-171). The fact that
the future á and the aspectual á can co-occur is evidence that the two are separate items.
The progressive aspect á is treated in detail later in §3.4.1.2.1.
3.4.1.1.1.1 Modals in Igálà
Modals express notions of possibility, permission, intention, and necessity. Three modal
items which appear to form part of the Igálà Infl structure are gbéḍ ò ̣ ‘must’, neke ‘can’,
and múdà ‘may’, as illustrated in (172).
172a. Ì múdà á ló
3sg may fut go
‘He may go.’
b. Ma múdà kéḍ òṇ ó
3pl may believe
144
‘They may believe.’
c. Ma múdà á nyá jòj̣ è ̣
3pl may prog fut eat-food
‘They may go to eat the food.’
173a. Ù gbéḍ ò ̣ ch’ukóḷ ó ̣ gbíti
1sg must do-work hard
‘I must work hard.’
b. Àlí gbéḍ ò ̣ jùjè ̣ lẹ nô/n
---- must eat-food the not

96
‘Ali must not eat the food.’
c. Ìdókò neke ch’ukóḷ ó ̣ i
----- can do-work this
‘Idoko can do this work.’
d. Wẹ neke gá àfè ̣
2sg can sow shirt
‘You can sew a shirt.’
3.4.1.1.1.2 The So-called Infinitive Tense in Igálà
Traditionally, Igálà is believed to attest the infinitive (i.e. tenseless) clause which
is assumed to be marked by either of two high-toned vowels, á and é, as exemplified in
(174) and (175).
174a. Ì d’óṃ ọ (dé ̣òṃ ọ) á kó ̣ ma
3sg be-there ?to teach them
‘S/he is there teaching them.’
b. Àlí dé ̣ ùbì únyí á rákwú (rọ ákwú)
--- be back house (?to) cry (cry sob)
‘Ali is at the back of the house crying.’
c. Óṃ a mi nyú/nyí óko á gwa úchu
Child my go farm (?to) dig-up yam
‘My child went to the farm to harvest yam.’
145
175a. Nàá ténè é ló t’únyí mi
1sg-will want to go to-house my
‘I want to go to my house.’
b. Ma kò ̣ é ch’ukóḷ ó ̣ ma
3pl refuse to do-work their
‘They refused to do their work.’
c. Ch’ánè ̣ é gba
do-ground to read
‘Begin to read.’/‘Start reading.’
Our first observation is that the highlighted á in (174a-c) appears similar both in form and
meaning to the progressive aspect highlighted in (176) below.
176. Óṃ a un á rakwú
Child 3sg prog cry-sob
‘His child is crying.’
As such, (174a-c) could be argued as having a complex clause structure in which two
simple clauses that share a single subject are knitted. The subject originated in spec-vP of
the second clause containing the continuous aspect á (i.e. AspP) but got raised to the
subject position of the first clause (i.e. TP) through successive cyclic movements. First, it
raised to spec-AsP of the second clause, next to Spec-VP of the first clause and
eventually to spec-TP of the first clause. This view is demonstrated below in (177).
177a. [TP Ì [T Æ] [VP tTheme dé òṃ ọ ] [AspP t [Asp á ] [vP tSUBJ kó ̣ ma ] ] ]
If this observation is true, then á in Igálà structures like those in (174a-c) is not an
infinitive tense marker, but an occurence of the progressive aspect.
Similarly, the claim that é in the structure type exemplified in (175a-c) marks the
infinitive in Igálà is debatable. As we have earlier noted (§3.3.1.3), Igálà has an é- gerund
prefix, which attaches to Vs/VPs to derive gerundive nominals. In the context
exemplified in (175), é immediately precedes a V/VP and that made us suspect that the

97
so-called é-infinitive is one and the same with the gerund prefix. The implication of this
is that the traditionally assumed surbodinating infinitive clause is not a clause afterall but
a nominalized V/VP which forms the direct complement of the main V that immediately
146
precedes é-. Therefore, clauses like those in (175a-c) would be better analyzed as done in
(178a-c), which agrees with the interpretation that native speakers of Igálà actually have
in mind when using such expressions.
178a. Nàá ténè éló t’únyí mi
1sg-will want going to-house my
‘I want to go to my house.’
b. Ma kò ̣ éch’ukóḷ ó ̣ ma
3pl refuse doing-work their
‘They refused to do their work.’
c. Ch’ánè ̣ égba
do-ground reading
‘Begin to read.’/‘Start reading.’
This observation has one interesting cross-linguistic implication, as it helps to shed some
light on similar clause expressions and issues in Yorùbá grammar. Some grammarians
(Crowther 1952, Bamgbose 1966, 1971, Rowland 1969, and Awoyale 1983) have
analyzed the bracketed Yorùbá expression type in (179) below as infinitive clauses.
179a. Adé fé ̣ [ é ̣ rí Akin ] (/ [rírí Akin] )
---- want to see ----- ( seeing ---- )
‘Ade wants to see Akin.’
b. Ó ṣe [é ṣe] (/ [ṣíṣe])
HTS do to do ( doing)
‘it is possible.’
c. Mi ò ní [ í gbà ] (/[gbígbà])
1sg neg have to accept ( accepting)
‘I will not agree.’
While a gerundive vowel prefix has not been identified for Yorùbá, the behaviour of the
highlighted item í in (179a-c) is quite similar to that of Igálà é- in (175 & 178). Evidence
from relevant data in Olumuyiwa (2006:117) indicates that the options in parenthesis in
(179) are forms of the bracketed expressions as rendered in the Ekiti and Moba dialects.
Following Awobuluyi (2001 and 2008), this study claims that Yorùbá gerund nouns are
derived through a V/VP root consonant copy and prefixation (§5.1.1.2.2) which involves
147
an i insertion rule that breaks unallowed clustering of consonants. This derivation is a bit
different from that of Igálà which simply uses vowel prefixation for the same purpose.
Putting these pieces of information and evidence together, it appears logical to
conclude that what looks like an infinitive clause in SY is actually a gerund
nominalization, as both cross-linguistic and language-specific dialectal evidence suggests
(cf. Abraham 1958:xxvi and Awobuluyi 1967:29, 1970:28-38). While Igálà provides
evidence in the similarity of (179) to (178) and vowel í to é- (e.g. both are high front
vowels), the CY dialects mentioned provide evidence which shows that the consonant
prefix of the gerund noun base output is phonetically null in SY expressions of the type in
(179a-c). We therefore conclude that Yoruba, like Igálà, has no infinitive clause type.
3.4.1.1.2 Tense in Yorùbá
Yorùbá like Igálà has a two-way tense distinction between future and non-future.

98
Unlike Igálà which realizes non-future as null at Infl, Yorùbá appears to mark non-future
with a high toned vowel that shows up between the subject and the verb. This high tone
syllable (HTS) often cliticizes to the final syllable of preceding subject nouns in most
Yorùbá clauses that are not marked for future tense, except in negative clauses. Whenever
a subject noun ends in a low-toned vowel (e.g. ilè ̣‘ground’), the HTS shows up as a copy
of that final vowel (as in ilè ̣é ̣). When the subject ends in a mid-tone vowel (e.g. Akin),
the HTS simply cliticizes to that final vowel and supercedes its mid-tone (as in Akín).
However, if the subject ends with a high-toned vowel, the HTS is simply absorbed by the
preceding vowel that ends the subject (e.g. Adé + HTS à Adé ). In such instance, the
HTS is only covertly present in the syntax.
The exact form and function of the HTS have been a source of controversy in
Yorùbá syntax. Various scholars have called the HTS different names based on their
perception of its function: Bamgbose (1967) called it a ‘predicate junction marker’;
Oyelaran (1970) and Fresco (1970) said it is a subject marker; Stalhke (1974) argued it is
a ‘subject concord prefix’; and Awobuluyi (1975, 1978, 1992) assumed it is a marker of
the non-future tense. One strong piece of evidence in favour of HTS as a marker of nonfuture
tense is visible in its mutually exclusive occurrence with the future tense markers
(i.e. máà, á, and yóò). The occurrence of the HTS exemplified below in (180) clearly
148
excludes yóò but optionally admits á and obligatorily admits máa (cf. Déchaine
1993:483).
180a. Akín ra ilé
Akin-HTS buy house
‘Akin bought a house.’
b. Akín á/máa ra ilé
Akin-HTS Fut buy house
‘Akin will buy a house.’
c. Akin á ra ile
--- Fut buy house
‘Akin wil buy a house.’
d. *Akín yóò ra ilé
Akin-HTS Fut buy house
Déchaine (1993:483) thought the HTS totally excludes yóò and á and that made her claim
that the HTS is a default Agr marker though not in the sense of a separate functional head
but as a relating item between two syntactic positions (predictably subject and predicate).
One thing that is however certain in all of these is that the Yorùbá HTS is an Infl item. It
cannot be a subject concord marker because Yorùbá does not mark subject-verb
Agreement common in English and neither does it attest the subject-Infl Agreement of
the type in Hausa (see Amfani 1996, 2004, 2005).
Although one of the characteristic features of the Yorùbá HTS centres on the
peculiar position that it occupies in the Yorùbá clause structure, i.e. the junction between
the subject and predicate, that cannot be its function. It only helps in differentiating
between a sentence and a noun phrase in somewhat similar structures like those in (181).
181a. Òg̣ á àgbà téẹ́ ṛ é ̣ (noun phrase ( -HTS))
Boss old slim
‘The old slim boss.’
b. Òg̣ á àgbàá téẹ́ ṛ é ̣ (clause (+HTS))
Boss old-HTS slim

99
‘The old boss is slim.’
149
It fails to predict and neither does it explain the obligatory absence of the HTS in
structures containing the future tense marker yóò and other items like the negator kò and
the question verbs dà And ńkó.̣
182a. *Akín yóò/kò lọ
Akin-HTS Fut/Neg go
‘Akin will go.’/‘Akin did not go.’
b. *Ṣég̣ un-ún dà/ńkó?̣
Ṣég̣ un-HTS Qst
‘Where is Ṣẹgun?’/‘What about Ṣẹgun?’
On the other hand, the fact that the non-future tense claim predicts the non-occurence of
the HTS e.g. with the future tense marker yóò, makes this latter claim a better, though not
necessarily a perfect, one, as it can occur with á and máa which are also future tense
markers. The future tense in Yorùbá is overtly marked at Infl by auxillaries like yóò,
máa, and á, as in (183a&b).
183a. Olú yóò/á/máa wá
Olu Fut come
‘Olú will come.’
b. Màá69 wá
1sg-will come
‘I will come.’
c. Olú (k)ò ní í wá (ní wíwá)
Olú neg has pref come (has coming)
‘Olú will not come.’
d. wíwá-wíwá, kò ní í wá (ní wíwá)
coming-coming neg has pref come (has coming)
‘as for coming, he won’t.’
A look at negative future tense expressions like (183c) may appear to suggest that níí is a
negative future tense marker. We are however of the opinion that the tense marker in
_______________
69. Thisitem is a fusion of the ‘1sg’ pronoun and the future tense marker á (i.e. mi + á à màá). The
derivation involves vowel assimilation and tonal contrast: mi + á (regressive vowel assimilation)
à ma á (tonal contrast) à màá. Cf. wọ + á à wa á à wàá ‘you will’, and o + á à o ó à òó
‘you will’.
150
such expressions did not make it to spell-out. In other words, it is realized null in the
convergent structure. The í following the V ní in such expressions is the high toned
vowel used in gerund nominalization (cf. 183d). We, therefore, conclude that Yorùbá has
descretely lexicalized future tense markers at the Infl layer which are realized null in the
negative future.
3.4.1.2 Aspect
Aspects denote the duration of the event or activity described by the verb in a
given clause. It shows whether such event or activity is ongoing (i.e. progressive) or
already completed (i.e. perfective).
3.4.1.2.1 Aspects in Igálà
There is evidence to show that Igálà marks the progressive aspect in Infl by the
use of the item á. This can be seen in (184).
184a. Àmímú á ló ̣ wẹ

100
Mosquitoes Prog bite you
‘Mosquitoes are biting you.’
b. Éṛ è ̣ mi á wó ̣ mi
Leg my Prog pain me
‘My leg is hurting.’/ ‘My leg hurts.’
c. Ókótobò á ténè kì jẹ ónú
Squirrel Prog ask comp-3sg eat king
‘The squirrel wants to become king.’
d. Ádéjọ á ró ̣ óḷ èṃ í gbèẹ̀ ̣
------ Prog tell lie much
‘Adejọ is fond of telling lies.’
By contrast, Igálà perfective aspect is not overtly marked in Infl70. Post-V manner
adverbs are often used to emphasize or indicate the completion/perfection of events
________________
70. Notethat when an item is realized nul, it only means such item has no phonetic form or content, and so
inaudible. However, the item still retains its grammatical features which make the presence of its
syntactic function felt in the structure concerned.
151
denoted by verbs in the clause structure. For instance, mèẹ́ ̣is an adverb used to indicate
that the event of ‘going’ in (185) below had been completed.
185. Ì lè mèẹ́ ̣
3sg go finish/complete
‘S/he has gone.’
3.4.1.2.2 Yorùbá Aspectuals
Three different types of aspects functionally lexicalized at Infl are noted by
scholars for the Yorùbá language. These are the progressive ń, perfective ti, and the
habitual aspect máa ń, which is clearly a combination of two items. These are
exemplified below in (186-188).
186a. Tádé ń ṣe ìdánwò
---- Prog do examination
‘Tádé is writing an examination.’
b. Lágbájá ń fọn fèrè
------- Prog blow trumpet
‘Lagbaja is blowing the trumpet.’
187a. Mo ti jẹun
1sg Perf eat-thing
‘I have eaten.’
b. Oúnjẹ ti jinná
Food Perf sink-fire (be-ready)
‘The Food is done.’
188a. Bóḷ á máa ń kàwé
--- Hab read-book
‘Bọla habitually reads.’
b. Mo máa ń kọ orin
1sg Hab sing song
‘I sing (habitually).’
Ọduntan (2000:136-140) challenged the claim on máa ń as a marker of the habitual
aspect. He argued that it is a combination of two aspects: máa which he called a durative
152

101
aspect that is homophonous but syntactically and semantically different from the future
tense marker máa ‘will’, and ń the progressive aspect. He adduced various reasons to
show that the durative máa is quite different from the future marker. One fascinating
piece of evidence he provided to back this claim is on the co-occurrence ability of máa
with other aspects. First, the aspect máa can co-occur with the perfective ti and the
progressive ń or both without having the future tense interpretation. This syntactic
behaviour of máa is exemplified in (189 -190) adapted from Ọduntan (2000:137).
189a. Adé ti máa jẹ ìrẹsì
---- Perf Dur eat rice
‘Adé would have eaten (i.e. finish eating) rice.’
b. Adé máa ń jẹ ìrẹsì
---- Dur Prog eat rice
‘Ade habitually eats rice.’
c. Adé ti máa ń jẹ ìrẹsì
---- Perf Dur Prog eat rice
‘Adé had been eating rice (before).’
Secondly, the durative aspect máa co-occurs with the future tense markers i.e. yóò, á, and
even máa with which it is homophonous. This is demonstrated in (190).
190a. Adé yóò/á máa jẹun
---- Fut Dur eat-food
‘Adé will be eating.’
b. Adé máa máa jẹun
---- Fut Dur eat-food
‘Adé will be eating.’
Ọduntan therefore explained that the state of the event/activity represented by the
combination of the durative máa and progressive ń is generally marked as being
incomplete, progressive, and not yet terminated, which actually gives, the the habitual
interpretation to the combination. Although Ọduntan’s arguments are compelling, we
shall not pursue their validity or otherwise but simply adopt them where necessary, since
the main issue has no serious bearing or consequences for the present study. What is
153
important to us is the fact that Yorùbá has aspect projections headed by functionally
lexicalized and phonetically visible items at Infl.
3.4. 1. 3 Negators
A negator is a functional element used to deny a proposition. In the words of Dahl
(1979:80), Neg(ator) is used
… for converting a sentence S1, into another
sentence S2, such that S2 is true whenever S1
is false, and vice-versa.
Generally, negators in most languages are Infl items which linearly precede the predicate
that they are used to deny. However, there are other languages where Neg is not solely
realized in Infl71.
3.4.1.3.1 Negators in Igálà
There are two Neg elements in Igálà. The first is a double particle item
comparable to, but different from, the French ne…pas exemplified in (191).
191. Je ne sais pas
1sg neg know neg
‘I do not know.’

102
The Igálà double-particled neg is realized partly in two structural positions. Its first part is
a high tone realized at Infl as a kind of prosodic modification on the subject noun phrase
while the other part, ń/nóò, regularly occurs clause finally. This superficially makes this
other part look like some kind of adverb. However, the fact that ń/nóò is obligatory in
every Igálà neg clause is an indication that, unlike adverbs, it cannot be a VP adjunct.
The prosodic neg particle is a high-tone that targets and phonemically supercedes
the tone of the final vowel of the subject noun phrase by changing it to an obligatory
high-tone irrespective of the original tone of such a vowel. With this, it prepares the
ground for the over-all negation of the clause by switching off the affirmative notion of
the declarative propostion before ń/nóò completes the work. For instance, if the subject
_______________
71. For instance, English not is now assumed to be a VP adjunct that is adjuncted to vP (Chomsky
1995:327-330; Radford & Atkinson et al. 2001:342-344). See Östen Dahl (1979) for other types
of negators and negation structures.
154
of an affirmative clause is a noun or noun phrase whose final vowel carries the mid- or
low-tone, the tone obligatorily changes to high for negation. This syntactic tonal
modification is evident in the following examples.
192(a)i. Àmbímọtọ á kọ éli
Children Prog sing song
‘Children are singing.’
ii. Àbímọtó ̣ á kọ éli ń/nóò
Children-neg Prog sing song not
‘Children are not singing.’
(b)i. Òḅ àlà nmọ ómi
Cat drink water
‘The cat drank water.’
ii. Òḅ àlá nmọ ómi ń/nóò
Cat-neg drink water not
‘The cat did not drink water.’
(c)i. Àfè ̣ wẹ che ìwà/àtìtì
Shirt 2sg do dirt
‘Your shirt is dirty.’
ii. Àfè ̣ wé ̣ ch’ìwà/ch’àtìtì ń/nóò
Shirt 2sg-neg do-dirt not
‘Your shirt is not dirty.’
(d)i. Íye Aládi nyú/nyí ájá éjulè
Mother Aládi go market éjulè
‘Aládi’s mother went to the éjulè market.’
ii. Íye Aládí nyú/nyí ájá éjulè ń/nóò
Mother Aládi-neg go market éjulè not
‘Aládi’s mother did not go to the éjulè market.’
(e)i. Éṇ è ̣ dúú dé ̣ òṃ ọ
Person all be there
‘Everybody is there.’
155
ii. Ẹnè ̣ dúú dé ̣ òṃ ọ ń/nóò
Person all-neg be there not
‘Nobody is there.’/ ‘Everybody is not there.’

103
Similarly, where short pronouns are used as subjects, the mid- or low-tone on their single
syllables changes to high for negation as in (193).
193(a)i. Ù jẹ òj̣ è ̣ úchu lè ̣
1sg eat food yam the
‘I ate the pounded yam.’
ii. Ú j’òj̣ è ̣ úchu lè ̣ ń/nóò
1sg-neg eat-food yam the not
‘I did not eat the the pouded yam.’
(b)i. Ù gbó ̣ ẹnwu kì ẹ kà
1sg hear thing that you say
‘I heard what you said.’
ii. Ú gbó ̣ ẹnwu kì ẹ kà ń/nóò
1sg-neg hear thing that you say not
‘I did not hear what you said.’
(c)i. Ì che ukóḷ ó ̣ gbíti
3sg do work hard
‘S/he did work hard.’
ii. Í che ukóḷ ó ̣ gbíti ń/nóò
3sg do work hard not
‘S/he did not work hard.’
(d)i. À á k’óc̣ hẹ éli
1pl Prog learn-work song
‘We are learning to sing.’
ii. Á72 k’óc̣ hẹ éli ń/nóò
1pl-neg-Prog learn-work song not
‘We are not learning to sing.’
__________________
72. This
item is a contracted form comprising the 1pl pronoun à, the high tone neg particle, and the
progressive aspect á. In other words, (193dii) = [à ´ á à á] k’óc̣ hẹ éli ń/nóò.
156
(e)i. Òmi déẹ̀ ̣
1sg be-this
‘It is me.’/‘This is me.’
ii. Í ch’òm(i) ń/nóò
3sg-neg be-me not
‘It is not me.’
iii. *‘Òmí déẹ̀ ̣ ń/nóò
1sg-neg be-this not
(f) i. Òṭ ákáda mi dé ̣ ì
Book my be this
‘This is my book.’
ii. Í che òṭ ákadá mi dé ̣ ì ń/nóò
3sg-neg be book my be this not
‘This is not my book.’
The foregoing examples show that the high tone neg particle is a reality in Igálà syntax.
It is equally pertinent to mention that ń (pronounced /ŋ/ ) is the most regular form of the
clause final neg particle. The nóò form is used only when the speaker intends to be
emphatic. We, therefore, analyze the -óò part of nóò as an emphatic clitic similar to
Yorùbá (o)ò in expressions like orí ì mi (o)ò! ‘ooh/ouch, my head!’ and oo in Ìgbò

104
nwáne m oo! ‘ooh my mother!’. The only difference is that the Ìgbò and Yorùbá
emphatic clitics are used for exclamation while that of Igálà is used for negation.
The second Neg element in Igálà is má. This item is also realized at Infl and
obligatorily followed by the clause final neg particle, ń/nóò. The neg má is exclusively
used in conditional clauses of the type in (194).
194a. Ì má gbà ń/nóò, ẹ mu du wa
3sg neg take not 2pl take-3sg bring come
‘If s/he refuses to take it, you bring it.’/‘Bring it if s/he rejects it.’
b. Ìchẹnwu kì má wa ń, é ̣ lè ń/nóò
If comp-3sg not come not, 2pl go not
‘If s/he does not come, don’t go.’
157
As evident in (194), the high tone neg particle is excluded whenever má is used. This
may be an indication that má and the high tone neg particle are occupying the same neg
head position in Infl. The syntax of negation in Igálà is fully discussed in chapter four.
3.4.1.3.2 Yorùbá Negators
Four negators are commonly identified for Yorùbá. These are (k)ò, (k)ì, má(à),
and kó.̣ These items are free morphemes that are descretely realized at Infl. This claim is
borne out of the fact that each of these neg items occurs immediately after the subject and
they co-occur with other Infl elements, as evident in the following clauses.
195a. Ọbásanjó ̣ (k)ò gbé Àbújá mó ̣
------ neg live ----- again
‘Ọbasanjọ no longer lives in Abuja.’
b. O (k)ò gbọdò ̣ fún un
2sg neg must give 3sg-Acc
‘You must not give it to him/er.’
c. Àwọn Yorùbá (k)ì í/ń ṣe òḷ ẹ
3pl ------- neg prog do indolence
‘Yorubas are not lazy.’
d. Èmi (k)ì í/ń ṣe ọba, èỵ in náà sì mò ̣…
1sg neg prog be king, 2pl too Adv know
‘I am not a king, and you ar perfectly aware of that…’
(Baṣọrun Gáà, Adebayọ Falétí 1972:5)
196a. Mo lè má(à) tẹ ibè ̣
1sg may neg step there
‘I may not go there.’
b. má(à) mú owó yẹn
Neg take money that
‘Don’t take that money.’
c. Ìwọ kó ̣
2sg neg
‘Not you.’
158
d. Ẹjó ̣ o-wọn kó ̣ [ni Adájó ̣ pè]
litigation their neg foc Judge call
‘It was not their case that the Judge called.’
Apart from kó ̣ commonly referred to as a nominal constituent negator which carries a
high-tone final vowel, the remaining Yorùbá negators have low-tone final vowels.

105
One other Yorùbá negator which often occurs in combination with other
preceding negative polarity items is ì (/rì in most of the CY dialects73) as it occurs with
(k)ò, tí, and má(à) in (197).
197(a)i. Olú (k)ò tí ì jẹun
---- neg ? neg eat-food
ii. Olú (k)ò ì jẹun
---- neg neg eat-food
‘Olu has not eaten.’
(b)i. Má(à) tí ì lọ
Neg ? neg go
ii. Má ì lọ
Neg neg go ‘Don’t go yet.’
(c)i. Owó mi (k)ò tí ì pé
Money my neg ? neg complete
ii. Owó mi (k)ò ì pé
Money my Neg Neg complete
‘My money is yet to be completed.’
The first question that needs to be answered is on the status of tí in structures like (197ac).
Ọduntan (2000:178) claimed the combination ‘tí ì’, is a monomorphemic negative
polarity aspectual tíì because it carries a neg meaning and requires a preceding ccommanding
Neg to license it as shown in (198).
______________________
73. This is the same Neg item in àì (/àrì in Èkìtì and Mòḅ à e.g. Wóṇ pá pá pá ké tí rì kú ‘They have
beaten him/er/it continuosly but s/he/it is not dead yet’). In this example, tí is a perfective aspect and
rì is a negator. See §5.1.1.2.1 and Olumuyiwa (2006: 42-44) for additional information.
159
198. Adé gbà láti máà tíì /*ti jókòó kọrin.
---- accept Inf. Neg Perf-Neg / Perf sit sing
‘Adé accepted to NOT sit yet while singing.’
‘Ade accepted to NOT sit and NOT sing yet.’ (Ọduntan 2000:178).
This claim can not be true for two reasons. First, it is evident in the gloss of (198), that tíì
is a fusion of the perfective aspect ti (= tí) and the Neg particle ì. Second, ì can occur
with its neg interpretation withouy tí as shown in (197aii-cii). However, there is still the
need to explain how ti = tí given that the fact the perfective aspect ti is usually midtoned.
Two pieces of evidence corroborate our claim.
First and foremost, tí in (197) and (198) has the same meaning/interpretation
with the perfective ti, and that simply implies they are two somewhat different forms of
the same free morpheme. In order words, they have/share a common semantic
distinctiveness (Nida 1949, 1965:7). This fact becomes evident when one looks at this
perfective tí in similar contexts in other Yorùbá dialects apart from SY. For instance, in
the Èg̣ bá dialect spoken in Ogun State, the mid-tone commonly associated with the
perfective ti is retained in expressions like those in (197) where it carries the high tone in
SY. The Èg̣ bá forms of (197a-c) are given below in (199a-c).
199a. Olú (k)ò ti ì jẹun
--- Neg perf neg eat-food
b. Má(à) ti ì lọ
Neg Perf Neg go
c. Owó mi kò ì ti ì pé
Money my Neg Neg Perf Neg complete

106
Similarly, the mid-toned perfective ti can be structurally noted in the Ìjè ̣ṣà and Ifè ̣
dialectal expressions below in (200).
200a. Mí tì gbó ̣ ß Mí ti ì gbó ̣
1sg-HTS Perf-Neg hear 1sg-HTS Perf Neg hear
‘I have not heard.’
b. Mé ̣ tì ghá ß Mi-é ̣ ti ì ghá
1sg-HTS Perf-Neg come 1sg-HTS Perf Neg come
‘I haven’t come yet.’/‘I am not ready to come yet.’
160
The output of ti + ì = tì in (200) also shows that the perfective mid-tone ti is used in the
imput. If it were the high-toned tí, the output would have been anomalous (i.e. tí +ì à tí )
in those dialectal contexts. These observations informed our thinking that the tí in tíì is
actually an occurrence of the perfective aspect, ti (cf. Èkìtì/Mòḅ à perfective tí in foot
note 71). The tonal change to high may have resulted from a structural tonal contrast with
the immediately following low-toned neg item, ì, for some as yet unidentified reason74.
3.4.2 The Functional Determiner (D) Layer
This is a [+F, +N] layer which comprises articles, demonstratives, and pronouns -
personal, quantitifier, interrogative, genitive, etc. The term Determiner in this context is
used in the minimalist framework as a cover term for the groups of items whose syntactic
features/properties, i.e. [+F, +N], indicate that they are functional items within noun
phrase. In other words, their syntactic domain is within the noun phrase where they
interact with other [+N] elements to project the noun phrase. Although, studies like
Abney (1987) and Longobardi (1994) have speculated that D instantiations probably
include conjunctions, such a proposal would be difficult to defend because conjunctions,
unlike other D instantiations, are never referential and neither are they restricted to the
noun phrase. For instance, there are exclusive clausal conjunctions, which clearly do not
share in the [+N] feature. Given the veracity of these and other related observations, we
propose that conjunctions are likely to be instantiations of C and not D (see § 3.4.3.3).
Basically, the assumption in minimalist grammar is that the functional Determiner
(D) technically heads every nominal projection in natural language. We shall, however,
delay discussions on this and other attendant issues on nominal elements and their
projections till chapter six where nominal projections are fully treated.
3.4.3 The Complementizer (C) Layer
Complementizers are clause introducers which are assumed to constitute a distinct
functional class that projects the functional phrase referred to as the Complementizer
Phrase (CP) in generative parlance. In the PPT theoretical framework, Comp is a
_________
74. Cf. Ọ̀yó ̣Yorùbá: Àwa ì bá tí wá = Àwa kì bá ti wá ‘We would not have come.’
161
mandatory pre-subject position associated with all clauses. The position is lexicalized
when a clause has a Comp item and left unrealized/suppressed otherwise. Hence, all
clauses in GB/PPT framework were assumed to be CPs underlyingly. In other words, all
clauses were assumed to be base-generated as CPs in that theoretical framework (Radford
1988, Haegeman 1991, Yusuf 1999).
In the mimalist framework, the complementizer projection is retained but it is not
assumed for all clauses. Only the minimal projections needed for each clause structure
are allowed, as the framework provides no room for any superfluous elements or
projections.
3.4.3.1 Complementizers in Yorùbá
107
Yorùbá attests, at least, four complementizer items which have been noted in
different ways by various scholars. The items are kí ‘that, before’, pé ‘that’, the invariant
relative clause introducer tí ‘who, that, which, etc., and the focus marker ni, as in the
following expressions.
201a. Adé rí wa [ kí [ á tóó lọ] ]
---- see us comp we-HTS before go
‘Adé saw us before we went.’
b. Ó yẹ [ kí [ ẹ sòṛ ò ̣ ] ]
HTS appropriate comp 2pl say-word
‘It is appropriate that you say something.’
/‘You ought to have spoken.’
202a. Ó dùn mí [ pé [wóṇ gbó ̣ ] ]
HTS pain me comp 3pl-HTS hear
‘It pains me that they heard.’
b. Abòṛ ìṣà gan-an mò ̣ [ pé [Ọlóṛ un wà ] ]
Idol-worshiper too know comp God be
‘Idol- worshipers too know that there is God.’
203a. Èmi ni bàbá onírungbòṇ yéụ́ ké ̣
1sg comp father owner-of-beard bushy
‘I am the old man with bushy beard.’
162
b. Ẹni tí ó ń gbé ibi gegele òkúta
Person comp HTS Prog live place top stone
‘The one that lives in the topmost part of the mountain (/stone)’
(Fágúnwà 1949:76)
One other item regarded as complementizer in this study is bí ‘way, how, if’ as used in
(204).
204a. Mo fé ̣ é ̣mò/̣mọ bí ó ti rí
1sg want know comp HTS Perf be
‘I want to know how it looks.’
b. Kò níí dára bí ó bá gbó ̣
Neg be be-good comp HTS ? Hear
‘It would not be alright if s/he hears.’
c. Màá kóḷ é bí mo bá lówó
1sg-Fut build-house comp 1sg ? have-money
‘I will build a house if I have money.’
It should however be noted that there is another bí in Yorùbá which studies like
Awobuluyi (1978:8-9, 16-17) and Déchaine (1993:92) have analyzed as a noun on the
strength of expressions like those in (205).
205a. Mo fé ̣ é ̣dà bí i-Jésù/rè ̣
1sg want be like Jesus/3sg-Gen
‘I want to be like Jesus/him.’
b. Ó ń ṣe bí i-mánéjà
3sg prog do like manager
‘s/he is behaving like the manager.’
c. Mo ṣe bí ó ti wí
1sg do like HTS Perf say
‘I did as s/he says.’

108
Given its syntactic distribution and interpretation, we agree that this other bí in (205) is a
functional nominal item, which is quite different from the bí-clause introducer in (204).
This is because it has no inherent semantic content of its own and appears to be
referential like other [+F, +N] words.
163
3.4.3.2 Complementizers in Igálà
The complementizer layer of the Igala language contains two visible lexicalized
items. These are kì, which marks the relative clause, and kí ‘that’ which is usually
preceded by kà, as shown in the following expressions.
206a. Éṇ wu che kù ẹ kò?̣ (kì ùwẹ à kù ẹ)
Thing be that 2sg refuse
‘Why did you refuse?’
b. Ì kpa éjò úgbo kì á ch’ukóḷ ó ̣ lè ̣
3sg kill snake where rel-3sg Prog do-work the
‘S/he killed a snake where s/he was working.’
c. À gó abú kì chukóḷó ̣ un
We look how rel-3sg do-work 3sg-Gen
‘We watched how s/he did his/er work.’
d. Únyí kì kó ̣ dé ̣ ì
House that-3sg build be this
‘This is the house that s/he built.’
e. Ì kwô èmí èg̣ bà kì ój̣ ó ̣ á dú
3sg leave here period that day Prog black
‘S/he left here when it was getting dark.’
207a. Ù che é-kà kà kí ú ná wa ń (kí ú náàkúná)
1sg do saying say that 1sg-neg will come not
‘I have said that I will not come.’
b. Àtá un néjú kà kí igóṃ éṭì á ra óḳ ó ̣ un
father 3sg think say that government will pay money 3sg
‘His/er father thought that government will pay his money.’
c. Ù neke lí kà kí itíchà è ̣ che
1sg can see say that teacher 2sg do/be
‘I can see that you are a teacher.’
Usually, there is a phonological contraction at the word juncture between the comp kì or
kí and the following vowel if the item immediatelty following them is begins with a
vowel. That accounts for kì ù(w)ẹ à kù ẹ, kì ì à kì, kì ój̣ ó ̣à kój̣ ó,̣ and kí únáàkúná,
164
in (206) and (207). The deletion rule that dictates this contraction evidently targets the
vowel of the comp item.
In addition to kì and kí, there appears to be a structural need to recognize a null
topical complementizer (Ø) for Igálà. This view is informed by the fact that when a
topicalized item is moved into Spec-CP for emphasis in Igálà, there is no phonetically
visible C head element comparable to the Yorùbá focus marker ni or Hausa nè/cè (Yuka
2004, Amfani 2004) present in the CP to show that an item had been topicalized or
emphasized. For instance, (208b), a topic clause, is differentiated from (208c) which is a
relative clause containing a phonetically visible Rel clause marker kì.
208a. Ù kpa ẹja étí òhìmìnì
1sg kill fish side river

109
‘I caught a fish beside the river.’
b. Éj̣ ai ù kpa [ ti ] étí òhìmìnì
Fish 1sg kill side river
‘It is a fish I caught by the river side.’
c. Éj̣ ai kì ù kpa [ ti ] étí òhìmìnì
Fish that 1sg kill side river
‘The fish that I caught by the river side.’
We therefore conclude that in addition to kí and the rel clause marker kì, Igálà has a null
topical c-head, Ø, in CP.
3.4.3.3 Conjunctions as Complementizers: A Proposal
Traditionally, the term conjunction is restricted only to those connectives that
establish relationship of togetherness among conjoined items, while those that indicate
relationship of separateness or alternatives are often called disjunctions. For instance,
items corresponding to Yorùbá àti ‘and’ are conjunctions while those like tàbí ‘or’ are
disjunctions75. However, for the purpose of this study, we shall play down on the
propositional interpretation of these items and assume the two are conjunctions. This is
________________
75. Cf. Awobuluyi (1978:104) and Crystal (1985:65).
165
purely for the syntactic reason that both conjunctions and disjunctions structurally
coordinate items in syntax. Although, the syntactic classes/status of the elements they
connect or unite divide them into two groups of lexical/phrasal conjunctions and clausal/
sentential conjunctions.
For the reasons already stated in §3.4.2, we disagree with the speculation that
conjunctions are instantiations of the functional D category and hereby present evidence
in favour of their consideration as C instantiations.
First and foremost, Comp which now heads the CP is used in generative grammar
to capture the traditional subordinating conjunctions and clause introducers (Awobuluyi
1978, Mathews 1981, Crystal 1985, Radford 1988, 1997a&b). These non-controvercial C
items head a phrasal projection (which often embeds a clausal expression) that serves as
some kind of complement to the V or N that immediately precedes them (i.e. the item to
which they are directly merged). For instance, the C-bar of the relative clause often
complements or gives some kind of additional, but surbodinating, information on the N it
modifies (e.g. examples 203b, 206, and 207). Similarly, the direct complement of
conditional clause introducers like Yorùbá bí ‘if’ is usually a clause which gives some
kind of pre-requisite information that determines the proposition in the main clause (e.g.
207a-c).
A similar view of coordinated clauses joined by what we term core conjunctions
(e.g. kpàí/àti ‘and’ and àmáà/sùgbóṇ ‘but’ in Igálà/Yorùbá) suggests that their syntax
somehow parallels those of compound clauses where the traditional surbodinating
conjunctions are used. If we assume core conjunctions occupy the same C head position
just like surbodinating conjunctions, and the phrasal/clausal constituent that immediately
follows such conjunction is the direct complement merged to it, then coordinated phrases
and clauses can be structurally interpreted as some kind of CP, precisely ConjP. We
assume the first half of the coordination, either a clause or a noun phrase, is a ready-made
well-formed/convergent syntactic derivation (SD), which is merged as Spec to Conj-bar,
i.e. [Conj’ [C ] [IP/N] ], to derive ConjP = CP. This proposed grammar, which evidently
takes care of both clausal and phrasal syntactic coordination, using nominal/clausal
conjunctions, is sketched out in (209) and exemplified in (210).
110
166
209a. 209b.
è
210a.
210b.
211a. 211b.
è
211c.
è
ConjP
D/N
Ìwọ /Akin Conj N/D
‘You/
àti Tádé/èmi
and Tade/ I ’
Conj'
Conj'
Conj IP/N
ConjP
IP/N Conj'
Conj IP/N
ConjP
TP
Ù kà nwú u Conj NegP
‘I told him/er
You/ àmáà í gbọ ń/nóò
but s/he did not listen.’
listen.’
Conj'
CP
TP
Mà á kóḷ é C IP
‘I will buid a house
You/ bí mo bá lówó
if I have money.’
C IP C'
bí mo bá lówó
if I have money.’
C'
CP
TP
Mà á kóḷ é
‘I will buid a house
You/
C'
C IP C'
bí mo bá lówó
if I have money.’

111
C'
t
167
A comparison of (210b) with the conditional clause structure in (211b) shows that, apart
from their semantic interpretations which appear to differentiate them, the syntax of these
two clausal expressions is quite similar. This informs our proposal that conjunctions are
likely to be some kind of complementizer in UG.
3.4.3.3.1 Igálà Conjunctions
Igálà nominal conjunctions are kpàí ‘with, and’, àbè(̣kí) ‘or, and’ ti .. ti ‘and’ as
illustrated below in (212).
212a. Óḳ ọ kpàí óỵ à un (óỵ à/óỵ á un)
Husband and wife 3sg-Gen
‘The husband and his wife’
b. Ùwẹ kpàí òmi
‘You and I.’
c. Édúdú àbèḳ í éfufu
‘Black or white’
d. Òmi àbèḳ í àma (àbèḳ á ma)
‘I or they’
e. Òmi àbèḳ í ùwẹ
‘I or You’
f. t’ónẹkèḷe t’ónobùlẹ (ß ti óṇ ẹkèḷẹ ti ónobùlẹ)
and male and female
‘Male and female’/‘men and women’
g. t’ùwẹ t’òmì (ti ùwẹ ti òmi) á che ukóḷ ó ̣ lè
and You and I fut do work the
‘You and I will do the work.’
On the strength of expressions like (212g), some traditional studies claimed that the
conjunction there is tu…tu. However expressions like (212f-g) contrarily show the
conjunction in question is most likely to be ti…ti. This implies that the vowel /i/ of ti is
usually deleted whenever ti is contracted with a following N.
168
Igálà clausal conjunctions are cháí ‘unless, except’, àmáà ‘but’, and àbèḳ í ‘or’
(which appears to double as both nominal and clausal conjunction in the language)
illustrated below in (213).
213a. Ù dó ̣ (un) àmáà í dòhì ń/nóò.
1sg call 3sg but 3sg answer not
‘I called him/er but s/he did not answer.’
b. Íye un kà àmáà í gbọ ń/nóò
Mother 3sg talk but 3sg hear not
‘His/er mother warned but s/he did not listen.’
c. É ̣ neke ló ń cháí kí ù lè ̣ bé ̣ (bé ̣ß bá ùwẹ)
2sg can go not except that 1sg go meet
‘You can not go unless I go with you.’
d. Í che èéi ń cháí èélẹ
3sg be this not unless that
‘It can not be this except/unless that.’
e. Á ténè éṇ ẹ dúú ń cháí ùwẹ

112
1pl want person all not except 2sg
‘We don’t want anybody else except you.’
f. Ẹ é ̣ jẹ àbèḳ í é ̣ jẹ ń? (Ẹ é ̣ß Ẹ á …)
2sg fut eat or 2sg-fut eat not 2sg fut…
‘Will you eat or not?’
g. Ẹ é ̣ lè àbèḳ í é ̣ lè ń
2sg fut go or 2sg-fut go not
‘Will you go or not.’
There are some other items traditionally regarded as conjunctions in Igálà which
we have deliberately omitted here due to our conviction that they are not conjunctions, at
least, on syntactic grounds. For instance, tódú ‘because’ is an adverbial adjunct
comprising the preposition tẹ ‘that of’ and the noun ódú ‘name’. Items like taku/taki
‘then’ and dagbà ‘although, even though’ and ìchẹnwu/ìchẹun ‘if’, are complementizers,
some kind of conditional clause introducers. Lastly, items such as ábú ‘what, how’ (an
interrogative D-pronoun) and ìkò ‘time’ are nominal items simply because they have the
169
syntactic properties and distribution of nouns in Igálà. This can be seen in the following
expressions featuring the items in question.
214a. Ì fu-u che tódú Ój̣ ó ̣
3sg take-3sg do because God
‘S/he did it because of God.’
b. Ijésù lè kwú tódú wa
Jesus go die because 1pl-Gen
‘Jesus died because of us.’
c. Taku Ijésù kà nwú ma
Then Jesus speak give 3pl
‘Then, Jesus spoke to them.’
d. U lè tak(u/i) éṇ è ̣ lé ̣ lè
1sg go then person that go
‘I went, then that person went.’
e. Dagbà ómi á lò,̣ ẹ ló t’óko tila
Although water Prog rain 2sg go P-farm compulsorily
‘Even though it is raining, you must go to the farm.’
f. Dég̣ bà ì lè kwú, ì á dé ̣ éjú
Although 3sg go die 3sg fut be eye
‘Although s/he is dead s/he will live.’
g. Ábu kì ì gúdu tẹ éfù únyí Ọ́jọ́ ...
When that 3sg enter to inside house God
‘As s/he entered into the house of God ...’
h. Ábú ódú ẹ kọ?
What name 2sg-Gen Qst
‘What is your name.’
i. Ìchẹnwu kì ì dé ̣ ùwẹ èḍ ò ̣ : (ß Ì che ẹnwu ...)
If that 3sg be you heart : 3sg be thing ...
‘If it is something that pleases your heart ...’
j. Ìchẹnwu kì ì má wa ń, é ̣ lè ń
If that 3sg not come not, 2pl go not
170

113
‘If s/he doesn’t come, don’t go.’
k. Ù mà ìkò kì nyò ̣
1sg know time that-3sg good
‘I know the good time.’
We observe that dagbà/dẹgbà is a combination of da ‘?’ + èg̣ bà ‘time, when’ which
makes it look like a V+N combination structurally. Ìchenwu is a noun derived through
desententialization from ì che ẹnwu ‘It be something’ but idiomatically used as a
conditional clause introducer in expressions like (214i&j). Ìkò, as a noun, functions as
the direct object of the V mà (214k).
3.4.3.3.2 Conjunctions in Yorùbá
Yorùbá nominal conjunctions are àti/ti ‘and’, and (t)àbí ‘or’, as they occur in the
following expressions.
215a. Bàbá àti ìyá
Father and mother
b. Olùkó ̣ àti àwọn akéḳ òọ́ ṛ è ̣
‘The teacher and his/er students.’
c. Mo ń fé ̣ [móṭ ò tàbí ilé ]
1sg Prog want vehicle or house
‘I want a car or an house.’
d. [ti ibi ti ire ] ni a dá ilé-ayé
and evil and good foc 1pl create house-world
‘It is with [good and evil] that this world is created.’
It is important to note that (t) àbí ‘or’ is also used as a clausal conjunction in interrogative
expressions like those in (216) below.
216a. Mo rí ì (t)àbí mi ò rí i?
1sg see it or 1sg neg see it
‘Did I not see it?’
b. Wóṇ ti jéẉ ó ̣ (t)àbí wọn ò tí ì jéẉ ó?̣
1sg-HTS Perf confess or 3pl neg Perf neg confess
‘Have they confessed or not?’
171
In this expression type, clause 1 is usually positive while clause 2 is obligatorily negative.
Other Yorùbá clausal conjunctions are àmó ̣ ‘but’, ṣùgbóṇ ‘but’, bóyá...tàbí ‘whether...
or’, yálà…tàbí ‘either...or’ and àfi ‘except, unless’. These are briefly exemplified in
situational contexts below in (217)76.
217a. Mo gbó ̣ àmó ̣ mi ò gbà
1sg hear but 1sg neg accept
‘I have heard but I am not accepting the offer.’
b. Mo jẹun ṣùgbóṇ mi ò yó
1sg eat but 1sg neg be-full
‘I ate but I am not satisfied.’
c. Bóyá ó lọ tàbí kò lọ, àwa ò mò ̣
Whether HTS go or neg go we neg know
‘We do not know whether s/he went or not.’
d. Yálà wóṇ rí móṭ ò tàbí wọn ò rí, èmi ò mò ̣
Either they see vehicle or they neg see 1sg neg know
‘I don’t know whether they got a vehicle or not.’
e. Kò jéẉ ó ̣ àfi ìgbà tí wóṇ nà án

114
Neg confess except when that they beat 3sg-Acc
‘S/he did not confess unitl s/he was beaten up.’
The last, though not in anyway the list, of Yorùbá conjunctions (actually a disjunction) is
ańbòṣ ì/ańbòṭ órí/ańbèlèǹté ‘let alone’. This item conjoins only nouns, preposition
phrases or adverbials, and sometimes clauses with nominalizations e.g. as in (218a-d).
218a. Sọjí (k)ò ra ilè ̣ ańbọ̀sì (pé) yóò kó ̣ ilé
---- neg buy land let-alone that fut build house
‘Sọjí has no piece of land let alone building a house on it.’
b. Kì í jẹun ní alé ̣ ańbòṣ ì ní òru
Neg Prog eat in evening let-alone at night
‘S/he doesn’t eat in the evening, much less in the night.’
(Awobuluyi1978:108).
______________
76. See
Awobuluyi (1978) and Yusuf (1980) for a more comprehensive study of Yorùbá conjunctions and
disconjunctions.
172
c. kò ní ìyàwó ańbòṣ ì ọmọ
Neg have wife let-alone child
‘He has no wife much less children.’
d. Ṣé ó dáhùn ná ańbòṣ ì (pé) yóò lọ?
Q-m HTS answer yet let-alone that fut go
‘Has s/he answered at all let-alone that s/he would go.’
It is appropriate to mention that we do not consider òun ‘?and’ and sì ‘?and’ to be
conjunctions in this study, contrary to pupolar view. Following Ìlòṛ í (2004, 2006), we are
of the opinion that òun in the context where it is often called a conjunction as in (219) is
an occurrence of the 3sg pronominal items, òun77.
219a. Adé òun Dàda ß Adé òuni (àti) Dàda
--- 3sg-Emph ---- ---- 3sg and ---
‘Adé (he) and Dàda.’
b. Akin òun àwọn mòḷ éḅ í rè ̣ß Akini (àti) àwọn mòḷ éḅ í rè.̣
---- 3sg they family his ß ---- and they family his
‘Akin and members of his extended family.’
Although Taiwo (2005:80-83) presents data to argue that òun is a conjunction in the Ào
dialect, the fact that the dialect does not employ àti as a conjunction evidently cannot be
generalized to cover SY and other dialects of Yorùbá where àti is still in use as a
conjunction. It is pertinent to point out here that Ào is not the only Yorùbá dialect where
àti has gone out of use. Ìlàjẹ is another dialect where òun appears superficially like a
conjunction while àti is almost practically non-existent. Interestingly, Igálà, a cognate of
Yorùbá, still used ti ‘and’, a clear relic of Yoruboid àti ‘and’ as a conjuction while àti
itself is practically unknown as a word in the Igálà language of today. A good example of
the use of ti ‘and’ in Igálà is given in (220).
220. t’ùmi t’ùwẹ ß ti ùmi ti ùwẹ
and-1sg and-2sg and 1sg-Emph and 2sg-Emph
‘You and I.’
________________
77. See Yusuf 1(980) and Ilọri (2004, 2006) for more detailed discussion on this issue.
173
The implication of these observations is that Yorùbá dialects like Ào and Ìlàjẹ where àti
no longer exists may have completely lost the item over time just like Igálà, and

115
subsequently began to use òun as a default conjunction, given the irrecoverable absence
of the actual conjunction item. If this is true as we suspect it is, then our earlier claim that
òun is not a conjunction in Yorùbá, at least in SY and other dialects where àti is still in
active use as a conjunction, remains valid.
Sì/dè ̣is another item that various writers on Yorùbá grammar have analyzed as a
(clausal) conjunction. However, the syntactic position occupied by sì in Yorùbá
compound sentences calls the validity of this seemingly popular claim into question.
Unlike the other two Yorùbá clausal conjunctions, i.e. àmó ̣‘but’ and ṣùgbóṇ ‘but’ which
structurally occur in-between the two clauses/sentences being coordinated, sì regularly
occurs after the subject of clause 2 of such compound clauses. Examples;
221a. [ [S1 Mo gbó ̣] [S2 mo sì/è ̣ dáhùn ] ]
1sg hear 1sg ?and answer
‘I heard and I answered.’
b. [ [S1 Adé jẹun ] [S2 ó sì/dè ̣ yó bámú ] ]
---- eat-thing HTS ?and full to-the-brim
‘Adé ate and he was filled to the brim.’
Yusuf (1999:62, 2005:193) tried to explain this observation away by claiming that sì as a
conjunction underlyingly occurs in-between two coordinated clauses just like àmó ̣ and
ṣùgbóṇ before an obligatory transformational movement rule moves it into its surface
position in clause 2, as illustrated below in (222) adapted from Yusuf (1999:62) and
Yusuf (2005:193).
222a.
sì Infl
VP
S1
S
CONJ S2
NP INFL
174
b. Ọba dá ilé rè ̣ sí méjì sì ó kó òḳ an fún ijàpá.
King divide house his to two pack one give tortoise
‘The king divided his belongings in two and gave one part to the tortoise.’
There are two fundamental problems with this analysis. First, the rightward movement
rule on which the analysis was based appears strange and suspiciously unique as the
direction of movement rule in Yoruba is generally leftward. In addition to this, that
movement is not in line with generative syntax, especially the minimalist framework
which assumes raising (i.e. leftward movement) and not lowering (i.e. rightward
movement) as the principle underlying movement in syntactic derivation78. Secondly,
the claim that sì attaches to Infl does appear not to convey the absolute truth about its
distribution in that context. Sì in that particular context interacts with both verbs and Infl
items, particularly aspect, as it can immediately precede either of them. Examples;
223a. Olú wá ó sì/dè ̣ ti lọ
----HTS come HTS ?and Perf go
‘Olu came and he has left.’
b. Mo gbó ̣ mi ò sì/dè ̣ dáhùn
1sg hear 1sg Neg ?and answer
‘I heard and I did not answer.’
Interestingly, while sì can immediately precede every verb in such compound clauses, it
cannot similarly precede all Infl items. For instance, the negator (k)ò and the future

116
(y)óò ‘will’ cannot be preceded by sì/dè ̣ in that context. This is eviedent in (224-225).
224a. N óò lọ N óò sì/dè ̣ sọ fún un.
1sg will go 1sg will ?and say to 3sg
‘I will go and (I will) also tell him/er.’
b. *N óò lọ N sì/dè ̣ óò sọ fún un
1sg will go 1sg ?and will say to 3sg
_______________
78. Cf.
Coopman (1984) on V-to-I movement (raising), as against the traditional I-to-V movement of Affixhopping;
Emonds (1976) and Pollock (1989a) argument on V-to-Tense movement in French; and
Déchaine (1993) on Igbo V-movement. See Radford (1988) for similar argument.
175
225a. A lọ a (k)ò sì/dè ̣ sọ nǹkan kan
1pl go 1pl Neg ?and say something one
‘We went and we did not say anything.’/
‘We went and we said nthing.’
b. *A lọ a sì/dè ̣ (k)ò sọ nǹkan kan
1pl go 1pl ?and Neg say something one
What these facts suggest is that sì/dè ̣ is not a conjunction but some 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/dè ̣to spec-IP to derive the surface syntax in
which the subject of clause 2 linearly precedes sì/dè.̣ This structural view is presented
below in (226).
226.
For structures where sì/dè ̣ precedes aspects like ti, ń or máa ń, we propose that it
optionally adjoins to pre-AspP of the clause 2 just as it does with pre-VP.
Our claim that sì/dè ̣is a syntactic adjunct (adverb), which originates in pre-VP or
pre-AspP is not entirely new. Awobuluyi (1978:73-74) had hinted that sì is a preverbal
adverb with a semantic interpretation of ‘consecutive action’ i.e. an action which follows
another action that had earlier occur. In actual fact, the literal/one-on-one meaning of sì
is ‘in addition to, or additionally’, but its logical English interpretation is ‘and’, that
appears to be the genesis of the confusion and the attendant structural problem it has
generated in Yorùbá grammar.
I VP
I'
IP
sì/dè ̣ V
tSubj V'
V ….
Subj
176
From all syntactic considerations, it appears Yorùbá does not have the particicular
clause co-ordinatig conjunction (as opposed to disjunctions like àmó ̣ and sùgbóṇ ‘but’)
which corresponds to the English clause co-ordinating conjunction ‘and’ (cf. Déchaine
1993:194). The only item that would have qualified for that function is àti ‘and’ but it is
never used to co-ordinate clauses or VPs in Yoruba. For instance, the Yorùbá utterance
in (227) is ill-formed but its English gloss is well-formed simply because English has a
clause coordinating conjunction and which can coordinate clauses and VPs. However
such coordination is blocked in Yorùbá because àti ‘and’ is not a clause coordinating
conjunction in Yorùbá.

117
227. *Mo jẹun àti mo sùn dáadáa
1sg eat-thing and 1sg sleep well
‘I ate and (I) slept well.’
As a syntactic option, Yorùbá appears to have opted for another strategy to express such
clausal coordination by using the adverbs sì/dè ̣which is clearly base generated in the
second clause as a pre-VP/-AspP adjunct, and literally means ‘in addition to the previous
event/action denoted in clause 1.’ This device expresses the consecutive action that such
coordination conveys in Yorùbá. Even at that, the strategy applies to clauses only as no
two VPs could be so coordinated in Yorùbá. The following data corroborate our
observations and claim that sì/dè ̣often indicates consecutive action(s).
228a. Èmi óò dìde, èmi óò sì tọ baba mi lọ
1sg will arise 1sg will additionally go-to father my go
Èmi óò sì wí fún un pé …
1sg will additionally say to him that
‘I will arise and go to my father, and will say unto him…’
b. Ó sì dìde, ó sì tọ baba rè ̣ lọ …
HTS additionally stand-up, HTS additional go-to father his go
‘And he arose and came to his father…’ (Luke 15:18).
229a. Ọba dá ilé rè ̣ sí méjì, ó sì kó
King divide house his to two HTS additionally pack
òḳ an fún ìjápá.
one give tortoise
177
‘The king divided his belongings into two and (additionally/
consecutively) gave one part of that two halves to the tortoise.
(Yusuf 2005)
b. Aṣọ tí mo rà tí mo sì wò ̣ lánàá
Cloth that 1sg buy that 1sg in-addition wear at-yesterday.
‘The cloth which I bought and (consecutively) worn yesterday.’
(Bamgboṣe 1990:196)
It should be noted that dè ̣is also substitutable for sì in expressions like those in (228 &
229). We deliberately exclude it as an option to sì in those examples because they were
quoted directly from the acknowledged refernces. These observations informed our
submission that, purely on syntactic considerations, sì/dè ̣is not a conjunction but a pre-
VP/AspP adverb in Yorùbá.
3.5 Conclusion
We have discussed word category inventories of Igálà and Yorùbá in this chapter.
In addition to Noun and Verb, we showed that Adjective, Adverb, and Preposition are
part of the lexical inventory of Yorùbá and Igálà grammars. We argued that Igálà words
traditionally categorized as place and time adverbs, body and place prepositions are
nouns based on their phonotactic property and syntactic distribution which parallel those
of nouns in the language. Traditional adjectives like èṇ yò ̣ ‘good’, èṛ ínyò ̣
‘sweet/sweetnes’, èḅ íéṇ ẹ ‘bad’ etc. are reanalyzed as nouns for similar reasons. Manner
words like gbíti ‘hard’, gwóg̣ wálá ‘quickly’ piopio ‘immediately’, etc. are categorized as
adverbs because they do not share in the phonotactic property and syntactic distribution
of nouns in Igálà.
While we analyzed most of the post-V manner adverbs in Yoruba as nouns given
their morphology and syntactic distribution (cf. Awobuluyi 1978), pre-V adverbs are

118
retained as true adverbs in the language. Attributive words like dúdú, pupa, funfun, etc.
are categorized as nouns that additionally function as adjectives and verbs in Yoruba.
At the functional level, we identified Tense, Aspect, Neg, Det, Conjunction,
Focus, Topic, Rel, and Complementizer, as functional categories that are instantiated in
the grammar of Igálà and Yorùbá.

119