3.1. The language

Caboverdean is the mother tongue of virtually all inhabitants of the West African Republic of Cabo Verde (around half a million) and of most Cabo Verdeans living abroad (estimated 1 million)6.

Most of its lexicon is of Portuguese origin (see Quint & Moreira Tavares 2019, and references therein), with only a small part inherited from the West African languages spoken by the continental slaves who were violently brought to the archipelago by the Portuguese colonists. Despite this historical and lexical relation, many Caboverdean properties naturally differ from European Portuguese, such as: it prohibits null referential subjects in most contexts; it lacks expletive subjects; it lacks subject-verb agreement; and it displays a small set of temporal morphemes, with the bare form of dynamic verbs denoting past situations (more on this in the next subsection).

The internal variation in the language – mostly concerning the phonology of lexical items, but also the forms of some functional morphemes – is traditionally taken as corresponding to two main groups of varieties: the Sotavento varieties, which started to develop in the 16th century in Santiago, and then in other southern islands (Jacobs 2010, and references therein); and the Barlavento varieties, which developed later in the northern islands (firstly in Santo Antão and São Nicolau, in the 17th century; in São Vicente, now the most prominent of these northern islands, the first community of speakers only settled in the early 19th century – see Swolkien 2015, and references therein). This traditional divide is now commonly associated with the most populated islands of the country, Santiago and São Vicente, respectively – and all these examples are from these two; they belong to the oral corpus LUDViC (Pratas 2020), which includes almost 150,000 words so far, obtained in semi-informal interviews7.

Additionally, Caboverdean is not an official language in its own country (it is in their plans, but these have been delayed by several factors), which means that there is no standard variety – and the attempts to settle an official spelling are sound but still limited8. Besides a few experiments of bilingual education, most teaching is still in Portuguese. These facts bring about very interesting challenges and, at the same time, some potential advantages for linguistic studies. One of these challenges is that, in order to register the oral data they want to work with, linguists must take many orthographic decisions themselves, and that is what happened in the oral corpus LUDViC (see the manuals available in the website) – some interesting points of variation at the phonological level, although not discussed in this study, are reflected in these scholars’ spelling, and these options are also adopted here. Another challenge, with a potential advantage, is that Caboverdean internal variation is not officially contested/contradicted (there is no enforcement at school of any standard rules), which means that it runs quite spontaneously.

3.2. Caboverdean states

The list of the language’s temporal morphemes and my current analysis of their meaning can be summarised as follows9. The bare form of most verbs denotes a past situation in what indeed seems a kind of non-periphrastic perfect, while the progressive is marked by an overt preverbal functional morpheme which appears mostly in the allomorphs sata (Santiago) and tita (São Vicente)10 – the salient aspectual opposition is thus between the perfect and the progressive, rather than one between the perfective and the imperfective. And then the preverbal morpheme ta marks habituals and generics, as well as some subsequent situations11.

So, basically three types of states are obtained here, which – each in its own way – temporally relate to a topic time (in the terms of Klein 1994), and this by default coincides with the time of utterance. These three types of states are: (a) the resultant state of a past situation (perfect), (b) the state expressed by the progressive, and (c) the state expressed by the habitual. And this Caboverdean reliance on states of different types to express a variety of temporal meanings constitutes a general ground of paramount importance for the current proposal on clefts.

Note that while the progressive and the habitual behave like states themselves, the perfect is more complex, in that it involves a dynamic situation (that is, non-stative) plus the referred resultant state (not to be mistaken for the result state described in Moens & Steedman 1988; Smith 1997). So while progressives and habituals, like other stative constructions, do not provide a temporal advancement when they occur in a narrative sequence, a string of sentences in the perfect temporally locates one situation after the other, which is typical of dynamic predicates. This is so because the Caboverdean state involved in the perfect, while being an important temporal anchor, is not part of the situation itself; this resultant state is rather “an abstract state of the event’s ‘having occurred’” (Portner 2011: 1230), a “post-time of the situation described by the predicate” (Klein 2014: 962). And this abstract, resultant state is also a central character in the current story for clefts, as discussed in Section 5.

Returning to the presentation of the main Caboverdean temporal morphemes, then we have another one that combines with the values (a)-(c), above, to mark a different topic time. This other morpheme also shows up in different shapes, mainly as two allomorphs – the verbal suffix -ba (Santiago) and the preverbal tava (São Vicente, which can be pronounced as tá)12 – and (still following Pratas 2021) rather than simply marking past, it marks low accessibility: it can be used with past situations as well as others that are both non-past and non-present, and therefore the only certainty is that they all signal a topic time no longer coincident with the time of utterance. It can be either a point in time before that (for past interpretations) or an unspecific temporal location that is definitely neither in the past nor in the present (e.g., in some conditionals, among others). Note that these Low Accessibility Morphemes (LAMs) are unrelated to the degree of time remoteness. In other words, they do not denote bigger or smaller temporal distances, rather referring to the perception/presentation of a given temporal location as not being fully accessible – in the sense of us, from our own temporal coordinates, from this worldly perspective we are conscious of (see Buonomano 2017; Rovelli 2018), feeling that it is only available through our memory13, or through other cognitive processes like imagination or desire.

Finally, while for most verbs this notion of low accessibility is expressed through the mentioned suffix -ba (Santiago) or the preverbal allomorph tava (São Vicente), as was also referred above, other verbal elements express the same value in their own lexical way, such as some suppletive verb forms inherited from Portuguese. This is for instance the case of the BE form era (from the Portuguese past imperfective of BE, third person singular)14, which in Caboverdean is also a low accessibility morpheme (LAM), in contrast with é (also third person singular), which takes the time of utterance as its topic time (so, being a state – not a dynamic verb – it expresses a present situation, as it does in Portuguese) – these two forms abound in both varieties of the language. Another suppletive BE form is foi, from the Portuguese simple preterit perfect (also third person singular), which in the São Vicente variety is used in very specific perfect contexts, like some passives (see Section 5 for an example and a note on the BE-less passives in the Santiago variety). The temporal import of all these elements will be illustrated over the next sections.

3.3. A variety of combinations

This subsection is devoted to presenting a list of Caboverdean clefts that can be found in the oral corpus LUDViC – a qualitative approach is therefore assumed in the treatment of these corpus data. From the versions attested crosslinguistically, the clefts available here have: (a) diverse word orders; (b) some clefts whose relativiser is a complementizer (corresponding to English it-clefts, but Caboverdean has no it in them) and some whose relativiser is a wh-element (corresponding to English wh-clefts, or pseudoclefts); (c) restrictions on which constituents can be cleaved (this description is in its first steps); (d) different combinations between the verb BE and the cleft-clause. The language displays no subject-verb agreement, nor morphological distinctions for Case15, and thus points (e) and (f) from Section 2 do not apply16.

As was briefly mentioned above, pseudoclefts, which also exist in the language, are not included in this paper. This is so for two reasons: they are varied and display problems of their own regarding their wh-elements (including their relation to questions, etc.); and they do not share the one characteristic that is relevant in the current study – the lack of an overt BE form. Therefore, only Caboverdean clefts corresponding to English it-clefts are examined here, where they go simply by the name of clefts.

From the components (a)-(d), above, Caboverdean productively exhibits at least four types of sentences, with some geographical variation: standard clefts with past and present BE, from Santiago (1); standard clefts with present BE, from São Vicente (2); clefts where present BE appears twice (pre- and post-cleaved constituent), from São Vicente (3); inverted clefts, from São Vicente (4); and clefts that exhibit no BE whatsoever, in both varieties (5).

Other questions may certainly arise about other elements, such as which cleaved constituents are available in each variety – and these will have their own future studies. For now, the goal that guided the qualitative selection of clefts from the corpus concerns the verb BE, and so do the questions summarised in the next subsection.

3.4. The absence of BE and related questions

A brief note is required here adding that clefts with no overt BE, corresponding to the ones in (5), are also attested in Brazilian Portuguese (BP), as is briefly mentioned in Assis (2001: 5) and Kato (2009: 383), among others. The illustration of these BP cases often comes along with one “standard” cleft (in the literature on Portuguese cleft constructions, this is the type matching with English it-clefts; here, again, I call them just clefts) which contains a “grammaticalised” BE form, é (6b); the whole set in (6) is adapted from Kato’s (2009) example (49).

Two assumptions can be inferred from this set: one is that the absent BE form in (6c) is the “grammaticalised” é (which in its plain use means present), and not the simple preterit perfect, foi, that is used in European Portuguese in a similar context; the other assumption is that that BE form is missing from a particular position in the sentence – this latter assumption is inferred here because examples like (6c) appear in the literature associated with the word order in (6a) and (6b), and not with inverted clefts of the type illustrated in (7), which are the so-called é que-clefts (see footnote 18).

Since a lexical item does not need to be “grammaticalised” to be silenced – silent elements abound in natural language with no prior process of grammaticalisation (e.g., an assorted list of constituents in ellipsis, etc.) – this possible deletion is certainly no argument for the grammaticalisation hypothesis. The reasons behind it have rather been connected to its invariable form and to some syntactic limitations (like never being negated or modified); under this view, é has therefore been considered a non-verb, just an element of the functional unit é que: “it is plausible to consider that é que ‘is-that’ clefts (unlike standard clefts, wh-clefts and pseudoclefts) are simple clauses, in which é que is a lexicalized expression that occupies a functional category in the left periphery of the clause” (Lobo & Santos & Soares-Jesel 2012: 7; see also some of the references therein). So the meaning of é is simply discarded. Curiously, however, é is not (yet) allowed to go missing in European Portuguese: (6c), as well as (6b), indeed appear in no variety of EP (see Vercauteren 2010). On top of this, for the cases where it is silent in Brazilian Portuguese, such as (6c), it is (as was mentioned above) connected in the literature with a “standard” cleft, not an é que one, and so another grammaticalisation process (dissociated from que) should be defended.

Independently of the solutions that have been essayed to account for (6) and (7), the truth is that no attention has been given to the meaning that é carries into the sentence – it is simply dismissed as just a “grammaticalised” element whose contribution has been consistently dismissed. The current study, however, sees é in Caboverdean clefts as a verb form, whose meaning is of the highest importance.

This option solves the problem about whether the examples in (5) are monoclausal constructions (they are not), which is unrelated to the position of the silent BE – that is, the point is not about which type of cleft it has been deleted from. So far, the corpus contains no proof of inverted clefts with present BE in Santiago, but as they are available in São Vicente (see (4)), we may say that, anyhow, the examples in (5) are ambiguous as to the position of this silent BE.

Instead, the proposal put forward here is centred on another kind of issue – the temporal meaning of clefts. This path is an effect of the following reasoning: the function of BE in clefts is such that, in (5), the meaning of each sentence is intact. Any good account for the existence of clefts in natural language must make sense of this, besides all their other syntactic and semantic idiosyncrasies.

The most auspicious clue in this respect came, as was already pointed out, from the analysis in Delin & Oberlander (2000) that a cleft includes an original eventuality (in the cleft-clause) and a created state (expressed by BE); this state plays the key role in the discourse structure that may explain why speakers opt for clefts rather than the simple declarative sentences (following some important points in Prince 1978 and in Hedberg 1988). This notion of a stative layer that distinguishes clefts from their simple sentence counterparts is nicely coherent with the relevance of states in the overall picture of Caboverdean temporal meanings (see 3.2, above). On the other hand, the cleft states are certainly of a specific nature, since those other Caboverdean states described earlier do not exhibit any form of BE. In those cases, we know: (a) that the resultant state serves as a temporal anchor for the perfect, expressed by the whole clause, which includes a bare form of a dynamic verb; and that both the (b) progressive and (c) the habitual are indeed always marked by temporal morphemes with roots in a Portuguese locative verb, estar (BE stage-level). Note that estar is certainly stative but is distinct from ser. And so the big question now is about the exact role of the ser forms that appear in Caboverdean clefts. This is the issue explored in the next two sections.

5.1. BE and temporal meaning

In his “brief” history of the verb ‘to be’, Moro (2017) disputes the popular view that there are various meanings for it: “this idea that the verb to be ambiguously expresses identity one minute and predication the next has now, in fact, taken hold everywhere” (Moro 2017: 55). The author rather proposes that in English and in Italian (at least), BE is a raising verb, in the sense that the other sentence constituent(s) are generated to its right – the word order that is visible is then the result of raising one of these constituents to the preverbal position (when these constituents are two NPs, the interpretation varies according to which NP is raised: contrast, for instance, ‘A picture of the wall was the cause of the riot’ and ‘The cause of the riot was a picture of the wall’; the subject is raised in the first case; the predicate is raised in the second case). This also means that Moro (2017) assumes the Aristotelian view that the verb ‘to be’ is not a predicate – the predicate function is, as was just said, performed by another element in the sentence.

These considerations about syntactic structure are not further developed here19, but another point in Moro (2017) is crucial for the current purposes. “[F]or Aristotle the verb to be is not a predicate, but expresses tense when the predicate does not consist of a verb” (Moro 2017: 23). And this seems indeed about BE carrying temporal meaning (note that Moro 2017 refers to this through the strictly linguistic function of “grammatical tense”, which is coherent with Comrie’s 1985 definition that grammatical tense is about morphological markings, not about interpretation). Let us see Moro’s words: “[grammatical tense] constitutes an interpretive structure specific to this ‘phenomenon’: present, past, future, aorist (or indefinite tense), and verbal aspect aren’t notions that we use to measure the reality of our world; rather, they’re notions that we use to project onto the reality of our world through grammar” (Moro 2017: 231). The important point is that the author is referring to the expression of temporal notions in natural language, and BE is one key element for this in many constructions around the world, even if some languages dispense with BE altogether, and others prohibit its use in the present (e.g., Russian and Maltese); in other words, “there are languages in which to say John is a teacher, you say something like John a teacher” (Moro 2017: 3).

In the majority that have BE, however, the temporal notions that it expresses are clearly not restricted to past, present or future; other values are available as well, as demonstrated by the pair John reads and John is reading. In this paper, the most relevant point is that BE can even express atemporality – this is explained in the next subsection.

5.2. A state depicted as a fact

Regarding the central topic of this article, it is here defended, with Delin & Oberlander (2000), that the function of BE in clefts is to create a state, which must play an important discourse role. This created state seems uncontroversial – BE is, after all, the quintessential stative verb. Even from a superficial observation, this analysis is therefore coherent with the set of Caboverdean examples just presented: they express this cleft-state through an overt BE, be it era or é, and they keep expressing it even with no overt BE.

Now, a detail in which the analysis put forward here differs from that in Delin & Oberlander (2000) is this: regarding the known overlapping of states regarding dynamic situations in a narrative, they observe that, in clefts, “what is overlapped seems to be the reference time of the embedded [...] event”, that is, the one expressed in the cleft-clause. And what is proposed here instead is that this created state fits into the discourse structure as follows. This state created by Caboverdean clefts is indeed a (stative) situation that is depicted as a fact, and therefore freely combines with the present, regardless of the temporal location of the embedded eventuality (in the cleft-clause) – that is, it easily takes the time of utterance as its default topic time. This notion about facts is rooted in philosophical studies20. The working definition of fact used here is found in Vendler (1967: 144):

To use another common example (always possible with a variety of historical characters), we may say: ‘The Queen’s death was an event’, but also ‘The Queen’s death is a fact.’ The first formulation refers to a dynamic situation located in September 2022, which is no longer happening. The second formulation refers to something (the nature of this something is a matter for metaphysics) that still is. Note that this latter case may also be expressed in the past, if for some discourse reason the relevant topic time is prior to the utterance time; more specifically, one may say ‘In December 2022, the Queen’s death was a fact.’ What is relevant here is that, as was said above, the Caboverdean clefts’ states may be depicted in the present regardless of the temporal locations defined by the cleft-clauses.

To further explain the depiction of this state as a fact (in this sense, facts are a subset of states), we can appeal to the definition of a resultant state used in 3.2. The whole affirmation by Portner (2011: 1230) is this: “The resultant state is to be distinguished from a result state. A resultant state is not an ordinary state which has been caused by the past event described by the sentence, but rather a kind of abstract state of the event’s ‘having occurred’.” Although Portner was writing about the perfect, this resultant state seems to correspond to the “post-time of the situation described by the predicate”, in the words of Klein (2014: 962). And the point made here for Caboverdean clefts is that the created state is thus presented as a resultant state, an “abstract state of the event’s ‘having occurred”, the “post-time of the situation”, which – for discourse reasons – the speaker may well choose to express by present BE. This nicely explains why we may get é in the main clause even though the situation in the cleft-clause occurred prior to the time of utterance21. And this study further argues that this is also what happens in clefts with no overt be, for the present value as the default interpretation for facts is always available.

As independent evidence that the time of utterance is the default topic time in Caboverdean for a range of constructions, the Santiago variety builds passives with no resort to BE; they consist only of the verb affixed by a passive morpheme, -du, which can be considered a participle (8)22.

Now we may check this for all the relevant cleft examples: the clefts with present é and a past embedded eventuality are here repeated for convenience (incidentally, with the goal of presenting them with an ‘is a fact’ statement, these paraphrases involve a wh-cleft – a pseudocleft).

An obvious question that arises at this point is whether one can account for this state presented as a fact and, at the same time, concede that BE is used in many other constructions where only regular stative situations are at stake, such as (13).

The next subsection defends how, on philosophical grounds, there is no conflict in that.

5.3. More about BE in philosophy

The argument that the created state expressed by BE in Caboverdean clefts is presented as fact – and not simply as any other stative situation – must encounter some fundament in what we know about the meaning of BE.

Although many different meanings for BE have been traditionally defended for a long time, Aristotle indeed had argued against that. It “is ‘said in many ways’ but it is not merely (what he calls) ‘homonymous’, i.e., sheerly ambiguous. Rather, the various senses of ‘being’ have what he calls a ‘pros hen’ ambiguity – they are all related to a single central sense (The Greek phrase ‘pros hen’ means ‘in relation to one.’)” (Cohen & Reeve 2021).

In a paper on the various contributions by Hintikka to the interpretation of Aristotle’s ontology, Calvo (2014) also contends that “Aristotle accepts and often emphasises that the verb εἰμί [BE] has several uses or several semantic forces. Nevertheless, they all are components of one unitary meaning” (Calvo 2014: 46). And it is Hintikka himself who wrote: “Everybody agrees that on different occasions verbs for being have different uses. They can express (among other things) identity, predication, existence and subsumption. Hence it is natural to say that on different occasions they have a different force or a different sense” (Hintikka 2006: 43)23. And this philosopher later gives an English example of a sentence containing different uses of BE: “‘X is a certain substance Z, and Z is Y.’ [...] In a way, it must therefore be in principle possible to separate the is of identity from the is of predication in such a simple statement” (Hintikka 2006: 48).

In the words of another influent philosopher, Susan Stebbing: “Any attempt to classify kinds of being will show that a much more thoroughgoing distinction than is usually made must be recognised between the existent and the subsistent” (Stebbing 1917-1918: 585)24. Stebbing then specifies that the existent, which includes the real and the unreal (the latter being like products of our imagination, which have their own existence as such), is spatial, temporal, or spatiotemporal, while the subsistent is non-temporal and non-spatial.

And, finally, when Fine (1982: 50), in a citation about facts which is equally quite essential for the current purposes, distinguishes between existing and being, it is – presumably – to emphasise two actual uses of BE: “I am inclined to think that there is an important ontological distinction between what one might call existence in the narrow sense and being in the broad sense, and it may then be true that particulars exist but facts have only being.” This is significant here because the persistence of facts is emphasised, just as it was in Vendler (1967), cited in 5.2.

From all that has been discussed so far – including the Caboverdean reliance on states of different types to express a variety of temporal meanings, as was discussed in 3.2 – the core proposal advanced here is that the states created in clefts by Caboverdean BE are depicted as facts, which belong to the realm of the subsistent, while other stative situations expressed by BE may belong to the realm of the existent. The former are therefore understandably compatible with present BE, regardless of the temporal location of the situations depicted in the cleft-clauses. In some cases, this default present value is obtained by a silent BE.