1. Introduction
The clause is commonly seen as consisting of several structural layers, each with different properties.Footnote 1 For example, Rizzi (Reference Rizzi and Liliane Haegeman1997) talks about (i) the lexical layer, where 
$\unicode{x03B8}$-roles are assigned, (ii) the inflectional layer, where argumental features such as case and agreement are licensed, and (iii) the complementizer layer – also known as the left periphery – which hosts topics and various operator-like elements, such as interrogative and relative pronouns, focused elements, and so on. This observation is also captured in Grohmann’s (Reference Grohmann2003) clausal tripartition into “prolific domains”: the 
$\theta-domain$, the 
$\varphi-domain$, and the 
$\omega-domain$, the last of which Grohmann defines as part of the derivation where discourse information is established. Rizzi (Reference Rizzi and Liliane Haegeman1997) views the left periphery as the interface between a propositional content expressed by the IP and a superordinate structure (a higher clause, or the articulation of discourse). Crucially, he does not consider the left periphery to be part of the extended projection of the verb, but a layer fundamentally distinct from the IP. For Rizzi (1997: 325), the topic and focus field in the left periphery expresses the “topic-comment and focus-presupposition articulations”, respectively, and elements that occupy the left periphery are endowed with [Topic] and [Focus] features. Most subsequent work, especially in the cartographic tradition, has assumed that the left periphery encodes primarily information-structural content of the utterance, though it has also been noted that the left-peripheral ‘focus’ position appears to be, at least in some languages, tied not to discourse prominence in general, but to exhaustivity, a property such structures share with English-style clefts (see, e.g., É. Kiss Reference Kiss1998 and Horvath Reference Horvath, Karimi and Wilkins2007 on Hungarian).
This paper investigates a phenomenon found in some languages with a rich left periphery (so-called discourse-configurational languages), where nominal predicates in discourse-neutral contexts occur in apparently the same position as that otherwise occupied by focused/exhaustified constituents. In the West Atlantic Niger-Congo language Wolof, one of the languages investigated in this paper, wh-words and exhaustified constituents A
$'$-move to the specifier of the A
$'$-movement complementizer (l)a, as in (1). Nominal predicates are found in the same position, shown in (2). I here refer to clauses with nominal predicates as copular clauses for simplicity, even when sentences at the centre of our investigation do not have a verbal copula.
Due probably to an Indo-European bias, both the descriptive and the formal literature commonly describe such clauses, especially when it comes to African languages, in the following way. Elements such as (l)a in Wolof, that I consider a complementizer, are termed ‘copulas’, even if they do not display any verbal characteristics, such as the ability to carry inflectional morphology, like tense or aspect marking, or to be modified by an adverb.Footnote 2 Additionally, many of these languages do have copulas which do exhibit verbal properties, as the Wolof example with subject exhaustification in (3) shows.
If (l)a in (1) is analyzed as a verbal copula, then it is argued that this sentence is a biclausal cleft. This requires positing a null expletive subject, and the remainder of the clause is treated as a relative clause; there is clear evidence against this view in Wolof. The motivation for this approach also comes from the interpretation of sentences such as (1), which, as the translation indicates, is close to that of English clefts, and is probably partially due to the fact that sentences as in (2) have the same element (l)a, which, through the lens of Indo-European languages, must then be similar to be.
In this paper, I discuss A
$'$-fronting for exhaustivity/focus and copular clauses in two typologically distinct languages: Wolof, previewed above, and Igala, a Yoruboid language spoken in Nigeria. We shall see that there is solid morphosyntactic evidence supporting the claim that structures as in (1) are monoclausal in Wolof. In Igala, on the other hand, such constructions have mixed monoclausal and biclausal properties, suggesting a possible ongoing grammaticalization process. Crucially, clauses with nominal predicates as in (2) can be formed in the left periphery in both languages, without being accompanied by the typical information-structural effects associated with left-peripheral elements or cleft constructions. I discuss the repercussions of these findings for our understanding of the left periphery in the architecture of the grammar, and propose that the prevalent view, according to which the left periphery is primarily the locus of discourse-related information, is at the very least incomplete. I propose that languages such as Wolof and Igala, where nominal predication and fronting for exhaustivity/focus overlap, are the key to furthering our understanding of the left periphery and its relationship with the extended projection of the verb. My primary goal here is to describe the empirical landscape, and offer some suggestions as to what an analysis should be able to explain; I do not offer a new account of the left periphery at this point.
The article is organized as follows. In section 2, I briefly address a few relevant examples of clefts and cleft-like constructions, focusing on African languages, and the morphosyntactic similarities between those structures and copular clauses. Sections 3 and 4 lay out the properties of the left periphery in Wolof and Igala, respectively, and give evidence that clauses with nominal predicates in both languages involve A
$'$-movement of the nominal predicate and the topicalization of the subject. Section 5 discusses the issues presented by the phenomenon, considers existing analyses, and lays out an informal proposal for what a comprehensive analysis ultimately needs to account for.
2. Clefts in Niger-Congo
Clefts are cross-linguistically a common structure, usually described as separating the ‘discourse prominent’ constituent from the background; that is, they are considered to be tied to the expression of ‘focus’ (the term is generally used fairly informally and with inconsistent meaning in the literature). I here use cleft to designate biclausal structures similar to English clefts, involving a copular clause hosting the clefted constituent, and a constituent that has properties of a relative clause, as in (4).
We have evidence that, in English, the clefted constituent, beans, arrives at its surface position via A
$'$-movement, but there is no agreement on what the structure of English clefts is (for a summary of various proposals, see, e.g., Hartmann and Veenstra Reference Hartmann, Veenstra, Hartmann and Veenstra2013).
Across Niger-Congo, cleft-like constructions involving A
$'$-movement are common. In some languages, the structures in question indeed look very similar to English-style clefts, as the example in (5) from Lubukusu (Bantu) shows: there is a copular element (ba) which exhibits agreement, and a complementizer-like element that is also found in (some) relative clauses (ni).
‘It was beans that Nangila cooked.’
(Wasike Reference Wasike2006: 144)
There are also languages with A
$'$-movement in structures expressing focus/exhaustivity where we do not appear to be dealing with biclausal cleft constructions. This is the case in Hausa (Afro-Asiatic), which has an element that occurs with fronted constituents, nē in (6), variably analyzed as either a left peripheral head (Green Reference Green2007), or a focus-sensitive exhaustivity marker (Hartmann and Zimmermann Reference Hartmann, Zimmermann, Enoch Oladé Aboh, Hartmann and Zimmermann2007). (Throughout the article, I use the English cleft as the translation of all the constructions that have been argued to encode exhaustivity for simplicity and clarity, even if they do not involve biclausal clefting.)
‘It is food that the children bought.’
(Green Reference Green2007)
Green (Reference Green2007) shows that sentences as in (6) involve A
$'$-movement to a left-peripheral position, and argues that there is no morphosyntactic evidence that would justify treating them as biclausal clefts of the English type.
Then there are languages that seem to be somewhere in-between. Gatchalian (Reference Gatchalian2023, Reference Gatchalian2025) investigates cleft constructions in Kirundi (Great Lakes Bantu), which contain the same element ni found in Lubukusu and many other Bantu languages, with the crucial difference that it precedes the clefted constituent, as in the object cleft in (7), making it look less like a head to whose specifier A
$'$-extracted elements move.
‘It is the book that Yohani read.’
(Gatchalian Reference Gatchalian2025: 1)
Gatchalian argues that clefts in Kirundi are biclausal, relying on various morphosyntactic properties of the putative relative clause in (7), which align with those of embedded, and not matrix, clauses: embedded verbs occur with a distinct tone melody, negation differs between embedded and matrix clauses, and the conjoint/disjoint alternation is available only in matrix clauses. According to all these properties, the part of the clause following the clefted element igitabu ‘book’ behaves like an embedded clause.
The view that structures as above are parallel to English clefts also comes from the fact that the same morphology that occurs in focus/exhaustivity constructions is commonly found in copular sentences in these languages. Examples (8)–(9) show copular sentences in Kirundi and Hausa, with the elements ni/nē.
Kirundi sentences as in (8) cannot contain any inflectional morphology or agreement, and they can only be used for third-person singular subjects; a verbal copula is used in all other cases. Therefore, even if Kirundi clefts are biclausal, in the sense that the clause following the clefted element in (7) behaves like an embedded clause, the matrix ‘clause’ is not a full, finite clause. This is essentially what Gatchalian (Reference Gatchalian2023) argues for in the end – that ni is a non-verbal head, Pred, and that it does not have an extended projection, but forms some sort of a reduced clause; I return to this in more detail in section 5. Green (Reference Green2007), on the other hand, argues that focus/exhaustivity constructions are monoclausal in Hausa, and proposes that copular clauses as in (9) involve A
$'$-movement of the predicate to the left periphery and the topicalization of the subject. In the following two sections, I argue that copular clauses in Wolof and Igala are built in the same way.
3. The left periphery and copular clauses in Wolof
In this section, we first familiarize ourselves with the relevant aspects of Wolof clause-structure and the properties of its left periphery, and then look at how copular clauses pattern with respect to other structures in Wolof. The main goal of this section is to establish that Wolof copular clauses are built in the left periphery, with the subject being a syntactic topic, co-indexed with a clause-internal clitic, and the predicate A
$'$-moving to Spec,CP.
3.1 The structure of the left periphery in Wolof
Finite clauses in Wolof are built fairly ‘high’; they all have an overt complementizer layer hosting a variety of so-called sentence particles (e.g., Dunigan Reference Dunigan1994; Martinović Reference Martinović2015, Reference Martinović2023a).Footnote 3 In finite indicative clauses in non-extraction contexts, a non-clitic subject is obligatorily in a left-peripheral position, and co-indexed with a clitic pronoun clause-internally, as in (10) and (11). The non-clitic and the clitic subject can never occupy the same syntactic position in the clause, and in these structures, the non-clitic subject is optional; only the clitic is obligatory.
Evidence that the clitic in (10)–(11) is not agreement comes from A
$'$-extraction constructions, where the clitic and the non-clitic subject occur in complementary distribution clause-internally (below C), as shown in (12).
Either the clitic or the non-clitic subject is obligatory, showing that the fact that the non-clitic can be omitted in non-extraction contexts does not indicate that the language is pro-drop, but is readily explained if the non-clitic subject is in a left-peripheral position; for extensive argumentation and more details on the differences between non-extraction and extraction clauses, see Martinović (Reference Martinović2015, Reference Martinović2023a). Note that subjects commonly occur in left-peripheral positions in finite indicative non-extraction/matrix clauses across the Niger-Congo language family (a.o., Schneider-Zioga Reference Schneider-Zioga2000, Reference Schneider-Zioga2007; Baker Reference Baker, Andrew Carnie, Harley and Willie2003; Henderson Reference Henderson2006; Pietraszko Reference Pietraszko2021).
A
$'$-constructions come in several guises, their surface morphosyntax differing in various sentence-types (see Dunigan Reference Dunigan1994; Torrence Reference Torrence2012, Reference Torrence2013a, Reference Torrence, Katharina Hartmann and Veenstra2013b, Reference Torrence2013c; Martinović Reference Martinović2015, Reference Martinović2017, Reference Martinović2023a). We are here interested only in the constructions featuring the element (l)a, as in (12), argued in Martinović (Reference Martinović2015, Reference Martinović2017) to be an A
$'$-movement complementizer. The complementizer exhibits a subject/non-subject asymmetry (Martinović Reference Martinović2017), surfacing as la in the extraction of a non-subject, and as a in subject extraction, shown in (13).
It is also common for a subject pronoun to surface in front of a in subject extraction, making (14) parallel to (13a).
There is no evidence that clauses in (14) are structurally different from clauses in (13): both constructions pass the standard movement tests, topicalized constituents occur before the question word in both cases, and nothing can intervene between the question word and C or the pronoun-C complex. Structures as in (14) possibly involve agreement between the element in Spec,CP and C; I leave this issue aside here, as all speakers I consulted use a and pronoun-a interchangeably in all contexts. I use both types of example in the article.
Torrence (Reference Torrence2013a) shows that structures with (l)a pass the majority of the standard movement tests: sensitivity to strong and weak islands, reconstruction effects (idioms, Principle A, Principle C, quantifier binding), and strong crossover.Footnote 4 They do not have properties of biclausal clefts: there is no pronominal subject (and there would obligatorily need to be one, as Wolof is not a pro-drop language, as we saw above), and there is also no relative complementizer, which is otherwise obligatorily overt in Wolof (see Torrence Reference Torrence2005, Reference Torrence2012, Reference Torrence2013c; Martinović Reference Martinović2015, Reference Martinović2017); compare (15) and (16):Footnote 5
Additionally, no inflectional morphology can occur in the supposed copular clause of the cleft. There is no way to say something like Who was it that Moussa saw?, for example. Wolof has one optional past tense morpheme, (w)oon (Bochnak and Martinović Reference Bochnak and Martinović2019), which is usually affixed onto the verb, but it can also occur in clauses with no verbal element, as is the case in copular clauses, as we shall see below. In clauses with A
$'$-fronting, it can only be suffixed onto the lexical verb, and cannot occur in the higher part of the clause.
In addition to the position for A
$'$-moved elements, Wolof also has a left-peripheral position for topicalized constituents; (18) contains a topicalized object and an A
$'$-extracted subject. Topicalized elements are obligatorily co-indexed with clause-internal clitics (Russell Reference Russell2006).
‘The horse, it’s me who bought it.’
(Russell Reference Russell2006: 127)
The order of the topicalized and the A
$'$-extracted constituent is fixed: as (19a) shows, the wh-word cannot preceed a topicalized constituent, it can only follow it, as in (19b). Any constituent can be topicalized in Wolof, and multiple topics can occur, as in (19c).
A common property of topics is that they cannot be bare quantifiers (Rizzi Reference Rizzi, Osvaldo Jaeggli and Silva-Corvalan1986, Reference Rizzi and Liliane Haegeman1997); this is also the case in Wolof, and will be important in diagnosing the structure of copular clauses. Example (20) shows that a bare quantifier cannot precede an A
$'$-extracted element.
To complete the picture of the structure of the left periphery, the example in (21) shows the embedding complementizer ni (or ne for some speakers), which is distinct from sentence particles and can embed any finite clause with any sentence particle; in fact, the whole left periphery can be embedded in Wolof. I follow Rizzi (Reference Rizzi and Liliane Haegeman1997) and place the embedding complementizer in Force.
Based on the data presented above, I assume that the left periphery in Wolof has the structure in (22). I do not employ a Focus projection here, but label the projection to which elements A
$'$-move as CP.
Before moving on to the structure of copular clauses, I wish to briefly address the information-structural properties of A
$'$-extraction constructions discussed above. As mentioned in section 2, constructions such as these are commonly taken to express ‘focus’. Information-structure of these clauses is not at the centre of this article, but it is important to point out several of their properties that will eventually need to be explained in light of the data on copular clauses. First, it has been clear for some time that A
$'$-fronting of the type we are interested in here cannot be straightforwardly tied to focus. For example, both subject fronting, in (23a), and object fronting, in (23b), are felicitous as answers to a question meant to elicit a broad sentence focus, such as Why are you angry?. Additionally, an answer with a structure that does not involve A
$'$-extraction and is commonly used for predicate focus, in (23c), is also felicitous.
Clauses as in (23c) are often called explicative in the descriptive literature, as they are appropriate answers to why-questions. They are also felicitously used in contexts eliciting predicate focus, and are generally not felicitous in out-of-the-blue situations. This, however, holds only of eventive verbs; with stative verbs, they are commonly the speakers’ first choice in an out-of-the-blue context. This speaks strongly against functionalist analyses such as the one by Robert (Reference Robert and Bernard Caron2000, Reference Robert and Isabelle Bril2010: 473), claiming that ‘focus’ is grammaticalized in Wolof. The few examples here already contradict this view. It is, of course, difficult to know what ‘focus’ means in such works, as it is not formalized; for example, Robert (Reference Robert and Isabelle Bril2010: 473) says: “in my terminology, the notion of ‘subject focus’ applies to a morphological form which prototypically indicates that the syntactic subject is the rheme. In discourse, however, focusing forms may have a variety of uses in addition to this prototypical one indicating that the lexical content of the focused constituent is the informative part of the sentence, in particular for the verb which is a syntactic constituent having the special status of predicate.” Rheme for Robert simply means ‘new information’; I have shown above that ‘subject focus’ can be used even when the subject is given. Robert goes on to say that subject focus constructions encode: (i) Identification or qualitative designation of the subject, (ii) Definition or explanation of the current situation, (iii) Exclamation with an intensification of the verb. It is difficult to understand how subject focus is “grammaticalized” given this range of meanings, or what that would even mean.
Fanselow and Lenertová (Reference Fanselow and Lenertová2011) (see also Branan and Erlewine Reference Branan and Erlewine2023) call this subpart of focus fronting, and based on German and Czech argue that the left periphery is not confined to topics and foci, and that movement to the left periphery therefore cannot be triggered or licensed by features related to information structure, but is primarily formal fronting.
While the data from the literature on this phenomenon, as far as I can tell, always appear to involve the movement of a subpart of a focused constituent, this is not always the case in Wolof. For example, an answer in which the subject is fronted is felicitous in Wolof even in cases when the subject is given, as in the following example (answers with object fronting and no fronting are also felicitous here).
While understanding and analyzing the information-structural properties of A
$'$-movement constructions is not one of the goals of this article, a future analysis should be able to account for this variety in meanings, and relate them to the interpretations available in copular sentences, discussed in the following subsection.
The second important property of these constructions is something we may want our analysis of A
$'$-extraction in cleft-like constructions and copular clauses to capture. As noted by É. Kiss (Reference Kiss1998), and further elaborated on by Horvath (Reference Horvath, Karimi and Wilkins2007), fronting to the preverbal position in Hungarian results in an exhaustive reading of the fronted element. This is also the case in Wolof, illustrated in the dialogue in (24). The first answer (A1) is felicitous in a situation where the market is just one of the places where one can buy dried hibiscus flowers, maybe the most convenient one. If, however, the second answer (A2) is used, where ja ba ‘the market’ is A
$'$-fronted, the meaning is exhaustive: the hearer understands that the market is the only place where one can buy dried hibiscus flowers. It is not felicitous to follow up with a sentence like You can also buy them in the store.
Additionally, elements in Spec,CP are incompatible with focus sensitive operators ‘also’ or ‘even’; they are exclusively compatible with ‘only’, as shown in the contrast between (25b) and (26).Footnote 6
Klecha and Martinović (Reference Klecha, Martinović, Anna E. Jurgensen, Sande, Lamoreux, Baclawski and Zerbe2015) propose that understanding how exhaustivity arises in sentences such as these also helps us understand the fact that, as we shall see below, nominal predicates move to the same position as exhaustified constituents. I return to this in section 5.
3.2 Copular clauses
Clauses with nominal predicates involve the element (l)a with the predicate preceding it and the subject obligatorily to the left of the predicate, co-indexed with a clitic which immediately follows (l)a. Clauses as in (27a) are the most natural kind of a sentence with a nominal predicate in Wolof; it is speakers’ first choice in out-of-the-blue contexts. This sentence structure can also be used when the predicate is exhaustified, as in (27b).
I here reproduce evidence presented in §2 of Martinović (Reference Martinović2023b: 251–255), showing that (l)a in clauses as in (27a) is the A
$'$-movement complementizer, that the predicate moves to Spec,CP, and that the subject is a syntactic topic, as in (28).
First, Wolof is an obligatory wh-movement language; wh-in situ can only be interpreted as an echo question (Dunigan Reference Dunigan1994, Torrence Reference Torrence2013c). If my proposal in (28) is correct, then a predicate wh-question corresponding to the declarative clause in (27a) should have the same structure, with the wh-word occupying the position to the left of (l)a, and the subject being to its left and co-indexed with a clitic following (l)a. This is indeed what we find; left-dislocation of the subject is obligatory in this structure, as shown in (29); the subject cannot stay clause-internal, as in (29b).Footnote 7
More evidence that (l)a in copular clauses is indeed the A
$'$-movement complementizer comes from long-distance movement. Long-distance extraction in Wolof requires the complementizer (l)a in the embedded clause (Dunigan Reference Dunigan1994; Martinović Reference Martinović2015, Reference Martinović2017), and resumptive pronouns are not grammatical in such cases, as shown in (30a). An A
$'$-dependency into a complement clause headed by a different sentence particle, for example na, can be established, but it requires the presence of a resumptive pronoun, as shown in (30b) (data from Torrence Reference Torrence2013c); I assume that these constructions do not involve movement (Martinović Reference Martinović2024b).
(Torrence Reference Torrence2013c: 108)
(Torrence Reference Torrence2013c: 108)
Predicates in Wolof are pronominalized with the object pronoun ko, as (31) shows. Wolof does not have VP ellipsis, therefore the pronominalization of the predicate is obligatory.
In this example, a different clause structure, one involving the overt copular verb nekk ‘be’, is used when the pronominal predicate occurs, as negation only occurs as a verbal suffix. In this clause, there is no A
$'$-movement; its syntax is parallel to the sentence in (10) (Martinović Reference Martinović2015, Reference Martinović2023a). As seen in (32), establishing a long-distance dependency into (31) is possible, but, just as in (30b), given that there is no A
$'$-movement complementizer, it requires the predicate nominal to be resumed with the object clitic.
Long-distance extraction from the sentence in (29), on the other hand, involves no resumption, which is to be expected if (l)a is the A
$'$-movement complementizer, and the predicate nominal can move long-distance because it already occupies Spec,CP of the embedded clause, demonstrated in (33).
The subject of copular sentences is syntactically a topic. First, it is obligatorily co-indexed with a clause-internal clitic, as are all topicalized DPs in the language (Russell Reference Russell2006). If it is a pronoun, it can only be one from the strong series, as is expected for pronouns in the left periphery (Cardinaletti and Starke Reference Cardinaletti, Starke and Henk van Riemsdijk1999, Martinović Reference Martinović2015). Clause-internal pronouns can only be strong if they are complements of prepositions, or if they are focused in-situ. There is no focusing involved in (34).
Second, subjects in copular sentences cannot be bare quantifiers (Rizzi Reference Rizzi, Osvaldo Jaeggli and Silva-Corvalan1986, Reference Rizzi and Liliane Haegeman1997), which we saw is true of topics in Wolof in general. In order to express the meaning in (35a), the quantifier must be embedded inside a DP (e.g., someone in this group; someone amongst us). Clause-internal subjects can be quantifiers, as in (35b).
It is also impossible to question the subject in the copular construction we are investigating here, shown in (36a). This is expected if the subject is a topic. If any element other than the predicate is extracted, a different construction is used – one with an overt copula, with the predicate nominal following it, as in (36b).
The contrast between copular clauses in the left periphery and those with the copula di drives home the point that the constituents of the former kind occupy a different layer of the clause and that such structures are verbless. Example (36b) is a regular clause with wh-movement of the subject and a verbal copula, di, which takes verbal morphology such as negation, as in (37a). Negation is obligatorily a suffix on a verbal head, and cannot occur in the copular clauses we have been investigating, as in (37b).Footnote 8
The position of the optional past tense morpheme (w)oon (Bochnak and Martinović Reference Bochnak and Martinović2019) also shows that copular constructions without a verb are built higher in the clause. The past tense morpheme is usually a suffix on the verb, but not obligatorily – it can occur in verbless clauses as well. In the copular clauses in the left periphery, it follows (l)a and the subject clitic, as in (38).
There is one other structural configuration in Wolof where the past tense morpheme is not affixed onto the verb. It happens in non-extraction configurations, when the verb is suffixed with negation and moves to C. The past tense morpheme remains stranded below the sentence particle and subject clitic, by hypothesis inside the TP (Torrence Reference Torrence2003, Martinović Reference Martinović2019). (For evidence that the verb is in C in (39a), see Martinović Reference Martinović2015, Reference Martinović2019, Reference Martinović2023a.) When there is no negation, the past tense is affixed onto the verb, as in (39b).
The reason for the different position of oon is not relevant here (see Torrence Reference Torrence2003, Martinović Reference Martinović2019 for different analyses): what matters is that the past tense morpheme is not necessarily affixed onto the verb, and that when it is not, it follows the C-subject clitic complex. The position of the past tense in copular clauses as in (38) therefore confirms that the two nominals are above the TP domain.
In clauses with the copula di, (w)oon is suffixed onto the copula, as in (40a). Additionally, the position of VP-adverbials illustrates the structural differences between a copular clause with a predicate in Spec,CP, and one with a subject in Spec,CP: in the former, demonstrated in (40b), the adverbial must be clause-final and cannot intervene between any of the constituents, which is expected if they are all higher than the VP. In a subject extraction sentence, on the other hand, the adverbial can occur between the copula with the past tense morpheme and the nominal predicate, showing that the predicate DP in the presence of a copula is in some lower, VP-internal position.
Finally, these examples also show that copular clauses in which both major constituents are in the left periphery are not in any way reduced, but are finite clauses that can contain tense morphology. This is an important point, in light of Gatchalian’s claim that in Kirundi, a copular clause that involves elements also found in A
$'$-movement constructions is not a full finite clause.
To summarize: clauses with nominal predicates in Wolof are finite clauses, that involve A
$'$-movement of the nominal predicate to Spec,CP, and topicalization of the subject.
4. The left periphery, clefting, and copular clauses in Igala
This section explores the same two constructions as section 3, focusing on Igala, a minority Yoruboid language spoken in Nigeria by about 1.6 million speakers (Eberhard et al. Reference Eberhard, Simons and Fennig2024). Igala is a heavily understudied language, so most information presented here is novel. We first establish the structure of focus/exhaustivity constructions in Igala. We shall see that while they have a more biclausal, cleft-like appearance, the ‘copular’ portion of the cleft still has some left-peripheral properties. Copular clauses themselves are very similar to those in Wolof and can be formed in the left periphery. The data presented in this section have been collected from three speakers of the Dekina variety of Igala, so the discussion and conclusions reached here are preliminary. My objective in discussing Igala is to make clearer the connection between predication and the left periphery, in using a language in which that part of the structure has mixed monoclausal and biclausal characteristics. Despite the fact that the left periphery has some attributes of clefts, copular clauses formed there do not have the information-structural traits of clefts/the left periphery, just as in Wolof. This strengthens the argument that the ability of the left periphery to host nominal predication should be at the centre of the exploration of its properties.
4.1 The structure of the left periphery in Igala
Igala is an isolating, tenseless, tonal language, with a variety of aspectual morphemes. Eventive verbs are interpreted as past tense and statives as present tense. Its surface word order is SVO, though it is possibly an SOV language underlyingly, as Koopman (Reference Koopman1984) argues is the case for some Kru languages. Example (41) illustrates a declarative, transitive Igala sentence, with progressive aspect, and (42) a sentence with a perfective auxiliary, with SOV word order. Igala has no agreement, and an overt subject – either a clitic or a non-clitic DP – is obligatory.
Igala has several positions to the left of the subject to which it can dislocate various elements. First, it is an obligatory wh-movement language (Martinović Reference Martinović, Jiayi Lu, Petersen, Zaitsu and Harizanov2024a); an object question is shown in (43a). In addition to question words, Igala, like Wolof, can place other elements in the pre-subject position; (43b), with a fronted object, is a natural answer to the question What did Ame eat?. The sentence-final particle yi optionally occurs in A
$'$-fronting constructions, regardless of which element is extracted.
In Martinović (Reference Martinović, Jiayi Lu, Petersen, Zaitsu and Harizanov2024a), I show that this kind of fronting has the properties of A
$'$-movement, as such constructions pass the standard movement tests, exhibiting strong and weak crossover, reconstruction for Principle A, idiom reconstruction, and island sensitivity; I do not repeat this evidence here.
Igala also has a position preceding the A
$'$-moved element for topicalized phrases. The example in (44a) contains a topicalized subject, which is obligatorily co-indexed with a clause-internal clitic, and an A
$'$-moved object. The topic is marked with a mid tone and the lengthening of the final vowel. Alternatively, the element chē must occur between the topicalized and the A
$'$-moved element, as in (44b). Chē ‘do’ occurs as a light verb in complex predicates, in copular clauses (as we shall see below), and between the topic and ‘focus’, as in (44b). I gloss it neutrally as che.
Other elements can be topicalized as well; (45) shows the two options for a topicalized object.
As is the case in Wolof, topics cannot be bare quantifiers. Examples in (46) contain the element ónè which can have the quantifier meaning in a non-topic subject position in (46a), but not in the topic position in (46b), where it means ‘the human’.
Unlike these kinds of examples in Wolof, structures as in (44b) have biclausal properties. First – and surprisingly, given the data that will follow – a cleft in which no inflectional material occurs in the ‘copular’ portion of the clause is not grammatical, as (47a) shows, with or without the relative clause complementizer kì, which is obligatory in relative clauses in (48). The only way to express the intended meaning is through the structure in (47b), where only the object is A
$'$-fronted.
Note again that Igala is not a pro-drop language; additionally, it has no verbless clauses either (which will become clearer when we look at copular clauses). I therefore take the absence of the expletive and the copula to indicate the absence of the structure that would host them. I propose that the sentence in (47b) has the structure in (49).Footnote 9
However, the occurrence of any inflectional material in the topic-focus part of the clause makes the cleft structure grammatical. I illustrate with aspectual morphology – the imperfective aspect á, and the future marker nyá. Both of them can occur in the ‘copular’ portion of the clause, and in the putative subordinate clause. Note that this is the only kind of finite non-matrix clause without an overt complementizer: relative clauses and subjunctive-like non-finite clauses obligatorily occur with the complementizer kí and kī;, respectively, and finite clauses embedded under bridge verbs are headed by the complementizer kàkíní. None of these elements can occur in (50).
These constructions are similar to clefts of the English type, but the ‘copular’ portion of the cleft retains some properties of the left periphery. For example, a topicalized constituent can occur instead of the expletive, as in (44b); note that subjects are otherwise never freely omitted. Example (51) shows the occurrence of aspectual morphology between the topic, in this case either the subject or the object, and chē. Note that the meaning is still that of a cleft; (51a) does not have the nonsensical meaning Chide will be the table that she will pull, nor does (51b) mean The table will be Chide who will pull it.
Even more interesting is that the topic can carry inflectional morphology, specifically negation. First, negation in Igala is bipartite, consisting of a high tone which docks onto the last vowel of the subject, and a sentence-final particle, as shown in (52) (Chaperon Reference Chaperon and To appearto appear).
As with many Niger-Congo languages, negation has two forms in Igala: the high tone occurs in declarative clauses, and in A’-movement constructions and nominalizations, the morpheme má occurs. The sentence-final particle is always obligatory. High tone negation can occur on the expletive element of the cleft, as in (53a). Negation in the lower clause is also possible, shown in (53b). The form of negation, má, which occurs in A
$'$-constructions, confirms that structures we have been investigating involve A
$'$-extraction, and are not some kind of a paratactic construction.
The data presented in this section are preliminary, and there is still much work to be done to understand the syntactic properties of clefts and the left periphery in Igala. For the purposes of this article, I will assume that Igala has both clefts and structures with a Wolof-style left periphery. Clearly the divide between the two is not clean, and it is likely that constructions with topics and foci are being grammaticalized into cleft constructions (or the other way around), but that this process is not complete. This results in a few peculiar properties of Igala clefts: that they are not grammatical if no inflectional morphology occurs, but are if such morphology is present; that the topic can occur in place of the expletive and carry inflectional morphology; and that the subordinate clause is the only such clause not containing an overt complementizer. This suggests that the position of the expletive is not, in fact, Spec,IP of a regular clause, but that it still has the syntactic properties of a topic. Even more variation is brought into play by the data on copular sentences, where we shall see that the part of the clause I will argue is the left periphery looks like what we saw in Wolof – the subject and the predicate exhibit the syntactic properties of topics and ‘foci’, but do not obligatorily exhibit their information-structural properties.
For the sake of completeness, I briefly show that A
$'$-extraction in Igala has the same basic information-structural properties as in Wolof. First, it is also not directly tied to focusing, shown by examples in (54), which illustrate subpart of focus fronting and parallel those from Wolof given in (23): (54a) involves no extraction, in (54b) the subject is extracted, and in (54c) the object.
Additionally, A
$'$-fronting in Igala also yields an exhaustive reading of the fronted element; for some speakers the exhaustivity is more prominent in the presence of the sentence-final particle yi. The answer in (55b), in response to (55a), indicates that the market is the only place where one can buy locust beans; it cannot be followed up with a sentence suggesting another place where one can buy them. If the market is just one of the places where you can buy locust beans, then a regular sentence without A
$'$-fronting is used, as in (55c).
We now move on to the copular clauses in Igala, which, I shall argue, are built in the left periphery, just like Wolof copular clauses.
4.2 Copular clauses
Copular clauses in Igala are strikingly similar to those in Wolof. A predicational sentence with a nominal predicate is given in (56):
This construction can also be used to answer a predicate question (What is Aladi?, as in, what is her profession), and in contexts in which the predicate is contrastively focused.
In the same context as in (56), speakers also accept the sentence in (57), though a common comment is that it is a bit curt, “not rounded up”.
In (57), the subject and the predicate appear to occupy the standard positions inside the IP, as I will show below. My focus here are sentences as in (56), which I argue have the syntax in (58). I will show that the subject nominal has the properties of a syntactic topic, and the predicate nominal A
$'$-moves to Spec,CP. And while, as noted above, this kind of a copular sentence can be used when the predicate nominal is focused, the key observation is that it occurs in contexts where that is not the case.
First, I present the same kinds of data we saw for Wolof to confirm that the subject and the predicate are in left-peripheral positions. Given that Igala is an obligatory wh-movement language, we should be able to felicitously substitute the predicate with a wh-word without changing the remaining structure, as we did in Wolof. This yields a grammatical predicate question, as shown in (59).
In the same vein, if the subject is a topic in this structure, it should not be able to be a question word; this is also born out, as shown in (60a). The question is felicitous if ì chē is left out, as expected if, in that case, the subject and predicate are in their canonical clause-internal positions, and the subject can move to Spec,CP (which is here string-vacuous, since Igala does not have an overt A
$'$-movement complementizer), as in (60b).
Just as in Wolof and many other languages that have strong and weak/clitic pronouns, left-peripheral positions can only be occupied by strong pronouns. This is the case in the copular clauses we are exploring here, as shown in (61): the clitic ì is impossible, only the strong pronoun ònwù is allowed. As in Wolof, (61b) does not require the subject to have particular information-structural properties, which is otherwise a requirement for the usage of a strong pronoun for the subject.
Additionally, just as we saw is the case for topicalized constituents in (46), the element ónè cannot be interpreted as a quantifier in the sentence in (62); it can only mean ‘the human’ (in a similar context as (46b)).
Aspectual marking in this kind of a clause is located inside the constituent marked here as IP, illustrated with the imperfective aspect and the future marker nyá, in (63), not in the part of the clause that I argue constitutes the left periphery.
Given the data we have seen in clefts, we may expect that these structures should show the same properties – that is, that they should be able to be interpreted as clefts. Example (63b) might then be infelicitous because of the mismatch in the aspectual marking in the matrix and embedded clause (i.e., it might mean something like It will be a teacher that Aladi is, which may be the cause of the ungrammaticality). Indeed, when the aspectual marking is identical in both parts of the clause, such examples improve, but they do not have the meaning of a cleft, as (64) shows.
‘It might be that Aladi will be a teacher.’Footnote 10
The speaker who offered the first interpretation also accepted (63b), under the interpretation It will eventually be discovered that Aladi is a teacher after all. What is interesting is that this is not a case of regular embedding in Igala. For example, the embedding complementizer kàkíní cannot be added anywhere in (64). When asked to give a sentence that means It will be the case that Aladi will be a teacher, speakers naturally offer the sentence in (65a). It is possible to also A
$'$-move the predicate in the embedded clause in a focusing context, as in (65b), but crucially, the subject Aladi cannot replace the expletive in the matrix clause. These are true biclausal structures.
The data on the left periphery in Igala are more complex than what we saw in Wolof, and more research is required for them to be well understood. ‘Focus’ constructions have some biclausal properties, but do not appear to be fully grammaticalized into clefts. Copular clauses, however, are quite similar to the ones in Wolof, with the subject and the predicate able to occupy positions in the left periphery. Crucially, in such constructions, the information-structural properties usually associated with the topic and ‘focus’ positions are not obligatorily present, but the constituents occupying those positions have the same morphosyntactic properties in both types of clauses: the element which is by hypothesis in the topic position exhibits a typical property of topics (e.g., it cannot be a quantifier), and the element which is by hypothesis in Spec,CP exhibits properties of an A
$'$-moved element. This drives home the point that the left periphery is not the locus of information-structural features, but that those properties of utterances must be derived in another way.
In the following section, I explore what the evidence presented thus far means for the architecture of the clause. I do not offer definitive answers in this article, but hope to identify the questions I believe arise from this, and how we may go about answering them.
5. Discussion
In the previous two sections we investigated the properties of the left periphery/cleft constructions in Wolof and Igala. The descriptive literature and the literature on grammaticalization have long noted that it is common for topic and focus markers to be reanalyzed into copulas (e.g., Li and Thompson Reference Li, Thompson and Charles N. Li1977, Heine and Reh Reference Heine and Reh1984), and this tendency is confirmed by the languages discussed here. Both Wolof and Igala A
$'$-extract nominals for exhaustivity, and both languages form clauses with nominal predicates in the left periphery, with the subject being a syntactic topic, and the predicate A
$'$-moving to Spec,CP. Additionally, the Igala left periphery has clausal characteristics: the same element that functions as a copula clause-internally is also found in the left periphery, between the topic and the A
$'$-extracted element, and the left periphery can host inflectional material. If the left periphery is being grammaticalized into an independent clause in Igala, the process does not appear to be complete.
Igala appears to be between languages like Wolof and languages like English, where we have no doubt that structures with A
$'$-movement for focus/exhaustivity are embedded under copular verbs in biclausal clefts. É. Kiss (Reference Kiss and Lunelle Mereu1999) argues that the cleft construction involves movement of the clefted constituent to the specifier of a Focus Phrase (which Brody Reference Brody1990 proposed for the Hungarian preverbal ‘focus’ position), which is then selected by a copula, demonstrated in (66). Note that in English, A
$'$-movement of non-wh XPs for exhaustivity (*each other they trust the most) is not licit when the clause is not embedded under a copula; this is not the case in Igala.
Another language that is somewhere in between Wolof and English is Kirundi, whose clefts also have biclausal properties in that the part of the clause following the A
$'$-moved constituent has the morphosyntactic characteristics of an embedded clause. Interestingly, the copular part of the clause does not exhibit clausal properties. Gatchalian (Reference Gatchalian2023, Reference Gatchalian2025) proposes that the left periphery has become an independent domain, albeit a very reduced one, without any verbal element and without an extended projection, as in (67).
If these analyses of Wolof, Igala, Kirundi, and English are on the right track, they are potentially capturing different stages in the grammaticalization of the left periphery into a copular clause (or the other way around; I have nothing to say about the direction of this change). According to this view, the left periphery would be a special structure used for what we may call ‘basic predication’, though we would, of course, need to precisely define what we mean by this, both syntactically and semantically. In order to shed some light on this puzzle, I focused on precisely those languages where the left periphery hosts nominal predicates. I believe the patterns discussed here suggest that understanding the relationship between movement for exhaustivity and nominal predication is crucial for a proper characterization of the left periphery and its position in the architecture of the clause.
Various works have touched on the intuition that what A
$'$-moves in the structures at issue is some sort of a predicate. For example, Wasike proposes that the complementizer-like element ni (see section 2) is a head in the left periphery in Lubukusu, which he terms “Pred”, but he leaves the identity of that head vague (“Our labels, particularly PredP, are neutral and can generalize across different construction types”; Wasike Reference Wasike2006: 142). While he argues that Lubukusu clefts are biclausal, this seems to be an attempt to capture the intuition that what we find A
$'$-moving has some properties of a predicate; however, without committing to a particular analysis for Pred, our understanding of these structures is not really improved. Wasike proposes that the structure for a Lubukusu cleft is as in (68). The clefted constituent A
$'$-moves to Spec,PredP (then continues to Spec,ForceP for reasons not relevant here), and then that entire clause is the complement of the copula in the higher clause.Footnote 11 The lighter-coloured constituents in the tree are traces of movement.
(Wasike Reference Wasike2006: 25)
É. Kiss (Reference Kiss, Valéria Molnár and Winkler2006) also addresses the intuition that some kinds of A
$'$-movement to the left periphery involve predication. The Hungarian preverbal position famously hosts exhaustified constituents, but also verbal particles and nominal predicates. She argues that Hungarian does not have a FocusP projection preverbally, contra Brody (Reference Brody1990, Reference Brody and István Kenesei1995) and much subsequent work (including, e.g., É. Kiss Reference Kiss2002), but a PredP projection, which can be filled by verbal particles that function as secondary predicates, or alternatively by an A
$'$-moved DP. The key observation that É. Kiss wants to capture is that the element in Spec,PredP is ‘interpreted as a predicate’. This observation is not formalized, however; É. Kiss states it only descriptively:
A preverbal constituent, whether simple or complex, occupies Spec,PredP, where it receives a predicate interpretation. As shown by Higgins (Reference Higgins1973), a nominal can function as a predicational, identificational, or specificational predicate. Any type of noun phrase can express specification. Predication (in the narrow sense), on the other hand, can only be expressed by a bare nominal or a nonspecific indefinite noun phrase (or by an adjective phrase).
(É. Kiss Reference Kiss, Valéria Molnár and Winkler2006: 187)
As É. Kiss notes, this intuition about different kinds of predicates goes back to Higgins’s (Reference Higgins1973) influential work on specificational pseudoclefts, but she does not go much beyond that. We do not know what it means, formally, to be a predicational or a specificational predicate, and consequently this does not tell us what it is about languages such as Hungarian, Hausa, Wolof, or Igala, that have A
$'$-positions in the left periphery to which they move DPs which are then interpreted either as exhaustified, or are nominal predicates in a copular sentence.
Klecha and Martinović (Reference Klecha, Martinović, Anna E. Jurgensen, Sande, Lamoreux, Baclawski and Zerbe2015) attempt to provide a formal, uniform account of nominal predication and exhaustive identification – É. Kiss’s predicational predication and specificational predication – in Wolof. They analyze the complementizer (l)a as having a semantics whereby the unique individual satisfying the property denoted by its complement (the TP containing the trace of movement) has the property denoted by its specifier. The denotation of (l)a is shown in (69):
What this means is that (l)a binds the trace of movement (as in Sternefeld Reference Sternefeld, Hans Kamp, Rossdeutcher and Rohrer2001, Kobele Reference Kobele2010, Kennedy Reference Kennedy, Luka Crnič and Sauerland2014) to its specifier by an iota operator. This semantics requires the element in the specifier to be a property (i.e., of type 
$\langle e,t \rangle $), so if the specifier is occupied by a referential DP (of type 
$\langle {e} \rangle$), it must be type-shifted into a property. The analysis translates the sentences in (70a) and (71a) into (70b) and (71b):
The exhaustivity, which is imparted in (70a), is neutralized in (71a), because the property being exhaustified (the property of being a plurality identical to them) is already a singleton. (For details, see the original paper by Klecha and Martinović Reference Klecha, Martinović, Anna E. Jurgensen, Sande, Lamoreux, Baclawski and Zerbe2015). This analysis gets us a step closer to capturing the nature of exhaustivity, but this is achieved through specific semantics given to the A
$'$-attracting head (here, the complementizer (l)a), without it being related to any particular property of the left periphery.
There are several additional properties of predication in the left periphery that indicate that the key relationship is the one between the topic/subject and ‘focus’/predicate. First, in both Wolof and Igala the clause-internal subject (i.e., the subject below C) is a pronoun, and the non-pronominal subject is topicalized. In Wolof, this is obligatory – if the predicate is A
$'$-extracted, the subject must be dislocated and resumed, as shown by (72). Having a copula does not help (recall that a copula is necessary when both constituents are not in the left periphery, which occurs when a constituent other than the predicate is A
$'$-extracted; see (36b)).
In Igala, such a sentence is not ungrammatical, but it results in exhaustive reading of the predicate, as in (73).
The obligatory left-dislocation of the subject therefore appears to be an important property of predication in the left periphery that an analysis should capture. Klecha and Martinović (Reference Klecha, Martinović, Anna E. Jurgensen, Sande, Lamoreux, Baclawski and Zerbe2015) can account for this, but based on a technicality. Namely, when the predicative nominal moves to Spec,CP, its trace is of type 
$\langle{e}\rangle$, which should derive a type mismatch if the other DP in a predicational clause is also of type 
$\langle{e}\rangle$. They therefore stipulate that pronouns are of type 
$\langle{e, t}\rangle$, which avoids the type mismatch (see the original work for details), and forces a non-pronominal subject to be dislocated (since this is how the grammar can achieve co-indexation with the clause-internal pronominal subject). Clearly this cannot account for Igala, where the sentence with a non-pronominal, non-dislocated subject is grammatical, as in (73), but just has a different interpretation. The crucial relationship where predication is established is between the subject in Spec,TopP and the predicate in Spec,CP. Klecha and Martinović’s (Reference Klecha, Martinović, Anna E. Jurgensen, Sande, Lamoreux, Baclawski and Zerbe2015) analysis does not capture this.
Another piece of data confirms the importance of understanding the relationship between the subject and the predicate, and it is again shared between Wolof and Igala: equative sentences cannot be formed in the left periphery in either language. In Wolof, the sentence in (74a) is grammatical only if we are not equating the individual known as Clark Kent with the individual known as Superman, but trying to say that the individual Clark Kent has some characteristics of the individual known as Superman. A clearer example of this appears in (75), which can only mean that Demba sings as well as Youssou N’Dour, or in a manner similar to him (or has some other, very salient, property of Youssou N’Dour). This shows that the element in Spec,CP must be predicated of the topicalized element, and there appears to be no predicational relationship in equatives. The only way to express identity between two individuals in Wolof is by A
$'$-moving the higher of the two DPs to Spec,CP, and using a copula, as in (74b).
Similarly, equatives in Igala cannot be formed in the left periphery. As shown in (76), they are grammatical only when the two DPs are clause-internal.
It is important to note that the problem is not in having a referential DP in Spec,CP in such sentences, or two definite DPs occupying the Spec,TopP and Spec,CP, but the relationship they establish with each other. For example, specificational sentences like (77)–(78) are grammatical in the left periphery in both languages.Footnote 12
An additional quirk of predication in the left periphery, shown in (79b), is that definite descriptions are banned from Spec,CP in predicational copular clauses, in both Wolof and Igala:
That this is really a restriction on what can be a predicate in these kinds of structures is confirmed by Ch’ol, which has a designated predicate position, in which it places all predicates: verbal, prepositional, and nominal (Coon Reference Coon2017). Coon and Martinović (Reference Coon and Martinović2023) show that definite descriptions are banned from that position. A similar restriction is found in Scottish Gaelic (Adger and Ramchand Reference Adger and Ramchand2003). Whatever strategy enables definite descriptions to function as predicates otherwise (or, to put it more neutrally, to occupy apparent predicate positions), is not available in the left periphery.
The restrictions discussed above in Wolof and in Igala are present only in copular clauses; otherwise, any kind of a DP can A
$'$-move to Spec,CP, and be preceded by a topic. The relationship between the left-dislocated/topicalized element and the A
$'$-extracted element in Spec,CP (often informally referred to as the ‘topic-comment structure’) therefore needs to be captured by the analysis that aims at unifying fronting for exhaustivity and nominal predication.
In this section, I have laid out previous analyses that have in some way touched on the central issue discussed in this article: the fact that the left periphery in many languages is the locus of nominal predication. I have not proposed a new analysis here, but I hope to have laid out the issues that I believe a thorough analysis of the left periphery should capture, and proposed some directions for their further exploration. I have here not discussed languages which do appear to have a Rizzian left periphery, but which have not been reported to exhibit the properties discussed here, such as Italian. Ultimately, we need to understand whether we are dealing with syntactically universal layers of structure, with the variation being in other parts of the grammar (for example, the semantics of some of the left-peripheral heads), or if the left peripheries in languages like Italian, and the ones in languages like Wolof, are actually not akin to one another.
Acknowledgement
My deepest thanks go to my Wolof and Igala consultants for sharing their time and language with me, especially Achenyo Ejeba, Salem Ejeba, Dorcas Otu, Lamine Dieye, Mbaye Diop, Lamine N’Diaye, Demba Lô, Clarisse Sow, and others who wish to stay anonymous. I thank the audience at the Syntax/semantics reading group at the Department of Linguistics at McGill University for helpful comments and discussion, as well as Éric Mathieu and two anonymous reviewers. Research on Igala presented in this paper was funded by the SSHRC Insight Development Grant #430-2022-00486. All errors are my own.
Open access
