When estar is not there: A cross-linguistic analysis of individual/stage-level copular sentences in Romance

2.1 On the ser-IL/estar-SL relation

In the last decades, the ser/estar alternation in Spanish, Portuguese, and Catalan has been related to the IL/SL dichotomy (Arche 2006, 2012, Bosque and Gutiérrez-Rexach 2009, 313–20, Brucart 2005, Cunha 2011, 2021, 598–600, Escandell and Leonetti 2002, Fábregas 2012, Fernández Leborans 1999, Fernández Leborans and Sánchez López 2015, Institut d’Estudis Catalans 2016, 862–71, Leonetti 1994, 2015, RAE and ASALE 2009, 2811–15, Ramos 2002, 2005–17, Raposo 2013, 1304–12, Silvagni 2017a, 2018, a.o.).^[6] Under this analysis, ser and estar are studied as syntactic exponents of IL and SL-predications: on one side, ser heads IL-CopPs, that is, predications that describe a characteristic, an intrinsic quality, of the subject (3); on the other side, estar heads SL-CopPs, that is to say, predications that denote a situation in which the subject is involved (4). For this reason, focusing on the subject, in the literature, estarPs are also referred to as structures that describe a condition, a state, or a manifestation of the subject.^[7]

As it is well known, the IL/SL distinction is embodied in a large variety of syntactic and semantic phenomena, which go beyond copular alternation and mainly concern modification, secondary predication and quantification, together with discourse effects (see Fábregas 2012, Fernald 2000 for an overview of the IL/SL dichotomy). On these bases, the relation between the phenomenon of copular alternation and the IL/SL aspectual distinction has been established by observing that serPs and estarPs behave as typical IL and SL-predications, respectively (Fernández Leborans 1999, Leonetti 1994, Silvagni 2017a, 2021a, a.o.).

More precisely, as far as modification is concerned, it is well known that only SL-predications, unlike IL-predications, can be modified by locatives and comitative event-related modifiers (Chierchia 1995, Kratzer 1988, Maienborn 2004, 2007, a.o.). In the case of serPs and estarPs, only the latter, as typical SL-predications, can be modified by locatives and comitatives (cf. [5] and [6]).^[8]

An important issue must be clarified at this point: only event-related, not frame-setting, locative and comitative modifiers can function as SL-diagnostics. On the one hand, frame-setting modifiers are sentential modifiers that provide a domain to which the whole proposition is restricted (i.e., a “frame”). They are merged VP-externally, in a topic position; hence, they are not sensitive to the aspectual nature of the base predication (VP) (Maienborn 2001). For this reason, frame-setting modifiers can appear with any kind of eventuality, that is, as far as we are concerned, with both IL and SL-predications, and are not a useful test for SL-ness (7). These modifiers are interpreted as a context within which the base predication is inscribed, and such an interpretation helps to distinguish frame-setting from true event-related modifiers when they appear post-verbally (7).

On the other hand, event-related modifiers are VP-adjuncts that specify where, with whom, how, etc. the event described in the VP-domain takes place: “there is an event, and that event happens in a place, in the company of someone, etc.” In other words, event-related modifiers are (non-selected) building blocks of the event that expand its denotation (see Maienborn 2001, Maienborn and Schäfer 2011 for a detailed analysis of the difference between frame-setting and event-related modifiers). Event-related modifiers are VP-adjuncts. Therefore, they are sensitive to the aspectual nature of the VP, and are licensed only by an SL-eventuality. This is the case of the examples in (6), where the modifiers receive the proper event-related interpretation: “there is a state of Ana lying down, and that state holds in her room/in the company of her teddy bear’. As expected, the same interpretation is ruled out with ILPs as in (5): # “there is a property of Ana being a lawyer, and this property holds in the library/with her teddy bear.” In fact, the grammaticality of sentences like (5) is bound to a frame-setting reading of the modifier: ‘whenever Ana is in the library/with her teddy bear, she says that she is a lawyer/she behaves as if she were a lawyer’ (see Maienborn 2004, 2011, 2019, Silvagni 2017a, 2021a for a detailed discussion on modification with IL and SL predicates).^[9]

As for secondary predication, only SL-predications allow for non-selected predicates, also known as “depictives” (Demonte and Masullo 1999, Hernanz 1988, Rapoport 1991, a.o.). When it comes to ser and estar, serPs reject depictives (8), while estarPs allow them (9).

In the same vein, only SL-predications, unlike IL-predications, can be quantified (Chierchia 1995, Cunha 2011, 47, De Hoop and De Swart 1989, Farkas and Sugioka 1983, Kratzer 1988, a.o.). As far as ser and estar is concerned, only estarPs (11), unlike serPs (10), can be quantified.

Moreover, SL-predications are interpreted as “happenings,” while IL-predications are interpreted as inherent characteristics (or properties) of the subject, which are not restricted to a specific spatiotemporal location. This can be appreciated by paying attention to the interpretation that predicates receive in different tenses (Cunha 2011, 51, Pereira Oliveira 2001, 72). In the present tense, SL-predications are interpreted as occurring “here and now,” that is, as being restricted to the time of the utterance, while IL-predications lack the “here and now” reading and are interpreted as a characteristic of the subject. The same happens among CopPs: estarPs are interpreted as happenings restricted to the time of the utterance (13), while serPs are interpreted as properties of the subject and, as such, independent of the specific spacetime point at which they are uttered (12).^[10]

In the past tense, SL-predications can be interpreted as if they happened only once or on several occasions.^[11] Oppositely, IL-predications can hardly be found in the past and, provided that they are appropriate in the past, they are interpreted as properties, which apply to the subject over the entire timespan referred to by the predication. Again, serPs behave as IL-predications (14) and estarPs behave as SL-predications (15).

The “happening” interpretation of SL-predications can also be tested by referring to the predicate through eventive verbs, such as to happen, to occur (Cuartero Otal 2011, 103, De Miguel 1999, 3011–2, Wanner 1999, 15). As expected, these verbs can take up only SL, not IL-predications. In the case of CopPs, estarPs, as happening-descriptive (that is, SL) CopPs, can be easily referred to by event-denoting verbs (17), unlike serPs, because of their property-descriptive (IL) nature (16).

The fact that serPs behave as canonical IL-predications, while estarPs behave as typical SL-predications, has served in the literature as empirical evidence that the copular alternation phenomenon is an overt manifestation of the IL/SL distinction.^[13] In other words, in Spanish, Portuguese, and Catalan ser and estar copulas can be taken as the ultimate instrument to grasp the IL or SL identity of non-verbal predications.

2.2 The case of mono-copular languages: French, Italian, and Romanian

The clear-cut distinction between IL-CopPs (serPs) and SL-CopPs (estarPs) observed in Spanish, Portuguese, and Catalan is not shown in the case of mono-copular languages. Here, the lack of an estar-like SL-copula entails that all predicates are linked to one and the same copula: essere (Italian),^[14] être (Frances), a fi (Romanian) (cf. [18] and [3]–[4]).

At this point, it is particularly important to keep in mind that the existence of derivates of Latin stare in Italian (stare) and Romanian (a sta) should not be confused with the existence of an SL-copula equivalent to Spanish, Portuguese, and Catalan estar. As mentioned in Section 1, estar is a fully grammaticalised form of Latin stare that behaves as a pure copula. In fact, estar is the verb that allows any SLP to be attributed to its subject (estar/*ser ebrio [Spanish] ‘To be drunk’, see also Section 3.1). Neither standard Italian nor Romanian nor French have this kind of SL-copula. The derivates of Latin stare in Italian and Romanian are aspectual verbs, which could rather be analysed as pseudo-copulas. They are not on a par with the SL-Cop estar, mainly for two reasons: first, they add aspectual information to the predication (namely, control, action, duration, etc.), giving rise to predications that are aspectually more complex than pure attributive copular phrases; second, they are not compatible with any SLP. Like all aspectual verbs, they can only combine with a particular kind of SLPs (in this case, those predicates that denote a state that can be controlled by the subject): stare tranquillo/seduto (Italian) or a sta liniştit/aşezat (Romanian) ‘To stay calm/remain seated’, but not *stare sporco/pieno (Italian) or *a sta murdar/plin (Romanian) ‘To stay dirty/full’.

In standard Italian, stare means that the subject stops or controls his movement or behaviour, and the verb is interpreted as ‘to stay’, ‘to remain’ (see Catalani 2004, Fratteggiani Tinca 1985, and Rainer 1984 for a detailed description of Italian stare). The same holds for Romanian, where a sta is a permanence verb equivalent to ‘to stay’, ‘to remain’ (Avram 1984, Mincă 2019, Pană Dindelegan 2013). Moreover, Romanian a sta differentiates from Italian stare in that it still has a locative meaning (‘to remain/to stand in a place’), which is typical of early stages of grammaticalisation of Latin stare (Copceag and Escudero 1966, 344–8) (see also Boioc Apintei and Dragomirescu 2021 for a general description of a sta).

In this scenario, in which the IL/SL distinction is not overtly marked in CopPs, it is necessary to study to what extent mono-copular languages are sensitive to the IL/SL paradigm: Do they lack the IL/SL distinction or is the distinction encoded in these languages despite the lack of copular alternation? Even though we expect mono-copular languages to possess the IL/SL distinction, as suggested by previous studies in the field (Carlier 2005, Martin 2008, 2009, Munaro 2016, Vinet 1991, a.o.) and taking into account the cross-linguistic scope of such a paradigm,^[15] it is worth investigating to what extent the IL/SL dichotomy is relevant for the non-verbal domain in these languages. As in the case of ser and estar, the aforementioned IL/SL-diagnostics can serve as a useful tool to test CopPs.

If we apply the tests to CopPs in French, Italian, and Romanian, we observe that the same difference examined between serPs and estarPs is found among CopPs in mono-copular languages. As far as event-related modification is concerned, only a group of CopPs allows locative and comitative modifiers, as typical SL-predications do (20). The other group rejects them (19).

The same happens with secondary non-selected predicates, as shown in (21) and (22).

In the case of quantification, the group of CopPs that allow event-related modifiers and secondary non-selected predicates can be quantified (24), unlike those CopPs that reject locatives and comitative modifiers and depictives (23).

When it comes to the interpretation of CopPs in the present tense, those CopPs that allow for event-related modifiers, depictives, and quantification are interpreted as occurrences restricted, at least, to the time of the utterance (i.e., “here and now”) (26). On the contrary, those CopPs that do not pass the previous tests are not interpreted as happening at the time of the utterance: they describe properties, characteristics, of the subject (25).

In the past tense, CopPs that refer to a happening are interpreted as occurred only once or on several occasions (28). By contrast, CopPs that describe properties are interpreted by default as being true for the subject over the timespan to which the predication is confined (27).

Finally, happening-descriptive CopPs can be replaced by happening verbs (30), unlike property-descriptive CopPs (29).

The aforementioned data show that CopPs in mono-copular languages are split into two groups: one group complies with all the tests for SL-predications (e.g., to be tired, to be alone, to be sad, to be angry, to be lying down), while the other group does not show any SL-property (e.g., to be a lawyer, to be Parisian). In conclusion, the lack of copular alternation does not prevent the IL/SL distinction: mono-copular languages do encode the IL/SL paradigm, and a distinction between IL-CopPs and SL-CopPs can be drawn.

It is also worth observing that the IL/SL-CopPs split of mono-copular languages coincides with the distribution of ser and estar in Spanish, Portuguese, and Catalan. This reinforces the idea that the IL/SL distinction is a universal, cross-linguistic paradigm, while the phenomenon of copular alternation is just typological evidence of it.

3.1 Analytical grounds

The IL/SL distinction has been one of the most debated topics in Linguistics over the last forty years. Even if, at present, there is a broad consensus in the literature that the IL/SL paradigm is an aspectual contrast,^[16] linguists do not agree on the nature of the aspectual content involved in the distinction and its formal analysis (see Arche 2006, 2012, Fábregas 2012, Fernald 2000, Husband 2006, 2012 for an overview). The most common understanding of the IL/SL distinction is a vague description as a contrast between permanent (IL) and transient (SL) qualities. However, it is well known since Carlson’s seminal work (1977, 72–3) that the “duration” (permanent or transient) cannot be taken as a defining feature of IL-hood and SL-ness (see Arche 2012, Silvagni 2018, 22–5 on this topic).

In this work, I assume a theory of the IL/SL distinction that has recently been put forward in Silvagni (2017a) on the basis of the classical understanding of IL-predicates (ILPs) as “property-descriptive” predicates and of SL-predicates (SLPs) as “happening-descriptive” predicates (see Carlson 1977, 75, Milsark 1974, 211),^[17] as well as the common view that the IL/SL dichotomy is aspectual in nature (Arche 2006, 2012, Chierchia 1995, Diesing 1992, a.o.).

As for the aspectual nature of the IL/SL dichotomy, here the distinction is conceived of as the first parameter of the inner aspect,^[18] that is, as the basic distinction between non-eventive predicates (or States) and eventive predicates (or Events). According to this view, ILPs and SLPs differ with respect to the presence (SLPs) or the lack (ILPs) of inner aspect: ILPs, as non-eventive predicates, lack any aspectual content, while SLPs, as events, are characterised by inner aspect (also Hoekstra 1992).^[19] The aspectual primitive of SL-ness (or eventivity) is defined as a space-time point, understood as a non-discrete intersection of time and space,^[20] and is labelled as Stage for the sake of coherence (also Silvagni 2017b).

Syntactically, SL-structures are studied as AspPs (also Becker 2002, Bosque and Gutiérrez-Rexach 2009, 313–20, Fábregas 2012, Hernanz 1991), unlike IL-structures, which lack the Asp projection. According to this proposal, the Asp head delimits the domain of inner aspect and is responsible for the grammatical effects of SL-predications that were shown in Section 2;^[21] it is located between vP and VP (also MacDonald 2008, Travis 2010) and carries a “Stage” feature [S]. In short, under this theory, the IL/SL distinction is defined as in (31).

As for copular structures, here the copula is analysed as a V-head (also Fernández Leborans and Sánchez López 2015, Gumiel-Molina et al. 2015, Mikkelsen 2005, Heggie 1988) that selects for a PredP (Baker 2003, Bowers 1993, Svenonius 1994). Therefore, copular structures would be derived as follows (32):

When it comes to estar, a doubling phenomenon with respect to SL-ness is found in the derivation of estarPs. The first evidence is that lexical SLPs must appear with estar and are excluded from serPs, as shown in (33).^[22]

Therefore, estarPs are composed by two SL-elements: the SLP, which denotes a situation, and estar, which also denotes a situation (synonymous with to feel, to find oneself; also Roca Pons 1958, 371–4, Salvá 1830, 208–9). However, in estarPs, the locus of interpretation of SL-ness does not correspond 1:1 with its morphosyntactic manifestations (i.e., the predicate and estar): if that were the case, we would interpret two events, one on estar and another on the predicate. Oppositely, a single event is interpreted in estarPs from the two SL-positions in the structure (34).

Based on the Flexible Formal Feature Hypothesis (Zeijlstra 2008), according to which doubling effects with respect to a single semantic content involve formal features, estarPs are analysed as an instance of syntactic agreement between an uninterpretable [Stage] feature located on the SLP and its interpretable counterpart on the Asp head, which is morpho-phonologically realised as estar (35).^[23] The [uS] instance on the lexical SLP forces it to appear in a proper agree configuration with the [iS] instance on Asp_(estar), which satisfies [uS]s configurational needs.^[24] In sum, the derivation of estarPs would be as in (35).

Therefore, the inventory of aspectual elements in ser/estar languages is argued to be as in (36).

In contrast, no formal aspectual features are found in the derivation of serPs, which are analysed as VPs, as typical IL-structures (37).

In the literature on Spanish ser and estar, it has been observed that the association of non-verbal predicates and copulas follows a specific pattern, which is summarised as follows: SLPs combine exclusively with estar, while ILPs combine with ser and, under a proper discourse situation, can be coerced to an SL-reading by combining with estar (38) (Gallego and Uriagereka 2016, Silvagni 2015).^[26] The same distribution has also been indirectly described for Portuguese (Cunha 2021, 559, Raposo 2013, 1304–12) and Catalan (Institut d’Estudis Catalans 2016, 550–1, 867–71, Ramos 2002, 1960–6, 2005–17) (also Pérez-Jiménez et al. 2018, 160–71 on adjectives).

As we have already discussed, from a semantic point of view, ILPs can be described as those predicates that denote a property, while SLPs can be understood as predicates that refer to a situation. Some generalisations can also be made with respect to the relation between the IL/SL distinction and grammatical categories. Nouns and relational adjectives denote classes of entities; thus, they are exclusively ILPs; likewise, adverbs describe circumstances, hence attributive adverbs (e.g., bien ‘good’, estupendamente ‘very good’) are SLPs. Qualifying adjectives and PPs are split into the IL and SL classes. These predicates, in fact, can refer to a property (alto ‘tall’, con botones ‘with buttons’) or a situation (borracho ‘drunk’, de vacaciones ‘on vacations’). Most of SL-adjectives are participles, and in fact they have usually been referred to as ‘perfective’, or ‘resultative’, in the literature. As for SL-PPs, the majority are locative or positional PPs (en casa ‘at home’, de espalda ‘on his back’).^[27] These distinctions among non-verbal predicates have been extensively observed for Spanish (Fernández Leborans 1999, 2428–34, RAE and ASALE 2009, 2811–26, a.m.o.) and also for Catalan (Institut d’Estudis Catalans 2016, 550–1, 867–71, Ramos 2002, 1960–6, 2005–17) and Portuguese (Raposo 2013, 1304–12).

Anyway, under this theory, IL-hood and SL-ness are understood as a formal property of predicates, namely, the lack or the presence of a [uS] feature: those predicates that are endowed with a [uS] feature are SLPs; all the rest are ILPs. Once we assume this characterisation, the ultimate manifestation of the IL or SL nature of predicates happens to be their distribution with ser and estar: a predicate that cannot appear with ser is an SLP, while a predicate that can appear with ser is an ILP.

When it comes to the aforementioned analysis, it is worth pointing out that it correctly grasps the whole distribution of non-verbal predicates with ser and estar copulas (38). First, this analysis correctly prevents SLPs from appearing with ser (see data in [33]): in that case, the configurational need of the [uS] feature on the SLP to stand in a proper agree configuration with [iS] would not be satisfied, as displayed in (39).

Second, the analysis leaves open the possibility that the Asp_[iS] (estar) head merges on a PredP with an ILP (41), that is to say, it correctly predicts that ILPs could appear with estar (40).^[28]

Those constructions in which an ILP is attributed to a subject via estar have been studied as a case of “aspectual coercion,” because in these cases, a property-descriptive predicate is found in a situation-descriptive predication (Escandell and Leonetti 2002, Fernald 1999, a.o.).^[29] In (40), for example, the group 〈estar + ILP〉 describes a condition of the subject, who “finds himself/looks” beautiful (40a), unbearable (40b), or new (40c). Under the analysis assumed here, IL to SL coercion is studied as the consequence of merging estar _[iS] (i.e., Asp) above a PredP that lacks aspectual features, that is to say that coercion is conceived of as a syntactic operation triggered by the [iS] feature on Asp.^[30]

It is worth pointing out that coercion with estar is subject to several grammatical and pragmatical restrictions. On the one hand, only qualifying IL adjectives and PPs can appear with estar, while nouns, relational adjectives, and relational PPs are excluded from estarPs.^[31] Moreover, even if any qualifying ILP can be potentially coerced with estar, it has been observed that the degree of naturalness of the 〈estar + ILP〉 construction varies across Spanish dialects (Escandell and Leonetti 2016, see also footnote 26) and depends on speakers’ conceptual restrictions and frequency (Escandell and Leonetti 2002, 167–8, Fernald 1999, 43, 59–61). On the other hand, the use of estar with ILPs is always subject to evidentiality, that is, the speaker’s commitment to having direct perceptual evidence for the assertion made (Escandell 2018a, 2018b). In other words, using estar with ILPs always stems from an experiential act, which leads the speaker to describe a manifestation, a state, of the subject.

3.2 Analysis

In this article, I put forward the hypothesis that the same formal feature [Stage] is at play in the derivation of SL-CopPs in mono-copular languages.

In order to test this hypothesis, I take into account three sets of evidence. First of all, we can go back to the IL and SL grammatical effects discussed earlier (Section 2) and take the split observed among CopPs in mono-copular languages (18)–(30) as evidence for a structural difference between IL-CopPs and SL-CopPs. Within the theoretical framework assumed here, CopPs would therefore be distinguished by the presence or the lack of an Asp projection (42)–(43).

Besides the evidence of a structural distinction at the VP level, we can analyse another set of data which tell us that, just as in ser/estar languages, in Italian, French, and Romanian SL-ness is rooted in non-verbal predicates.

As mentioned in Section 3.1, in the literature on ser and estar, it has been observed that SLPs can receive only the SL-reading (45), unlike ILPs, which can also be coerced into an SL-interpretation (44). This phenomenon implies the possibility of ILPs to appear with both copulas (44) and the impossibility of SLPs to appear with ser (45).

It is worth observing that the same happens in the case of mono-copular languages: property-descriptive predicates, apart from their default reading as a characteristic of the subject (46), can also receive an SL-interpretation (47); conversely, those predicates that describe a situation in which the subject is involved cannot be interpreted as a property of the subject, that is, as an ILP (48).

It is important to highlight that in the absence of an SL-copula (as in the case of mono-copular languages), evidentiality plays a decisive role in the IL/SL disambiguation of copular clauses with an IL attribute. In fact, coercion of an ILP into an SL-predication necessarily results from a speaker’s perception act. In addition, the SL-interpretation of the structure by the listener is subject to the correct retrieval of the relevant information about the same perception act. For example, the SL-reading of a predication like the one mentioned earlier (e.g., Anna è bellissima/elegante [Italian] ‘Anna is beautiful’) can be correctly grasped by the listener if and only if (s)he is able to link the assertion to a specific manifestation of the subject (Anna) in the world (for instance, Anna dancing on a stage, or getting out of a car). At the same time, perceiving the aforementioned situation is what pushes the speaker to build an SL-predication out of an IL-predicate (“beautiful,” “elegant”). Conversely, the lack of a perception act blocks the possibility of having an SL-predication: if the speaker does not perceive a specific manifestation of Anna, or the listener is not able to link the predication to such a perception act, the IL-predication arises (i.e., “Ana is characterised by beauty/elegance”).^[32]

In sum, in both mono-copular and bi-copular languages, evidentiality is involved in the IL to SL coercion phenomenon. However, contrary to bi-copular languages (where SL-structures are overtly marked by the estar copula), in mono-copular languages, evidential effects play a key role in the building and the right interpretation of SL-copular structures with an IL attribute.

Additional evidence for an IL/SL distinction among non-verbal predicates comes from secondary predication phenomena. As it is well known, SLPs not only allow for secondary non-selected predicates (see Section 2) but they can also function as free predicative adjuncts. In the NP-domain, it is observed that only SLPs can serve as NP predicative adjuncts, while ILPs can only be complements of N (Leonetti and Escandell 1991 on Spanish). For example, only SLPs can modify a strong DP, which cannot be further restricted by N-complements and only allows for adjuncts (49)–(50).^[33]

The same happens in the case of VP-adjuncts: only SLPs (52), unlike ILPs (51), can function as depictives.

The same phenomenon has been observed for perception verbs, which exclusively allow for SLPs in the predicative position (53).

To sum up, the data above bring to light a clear distinction between IL and SL non-verbal predicates across languages. Therefore, we can observe that, just as in the case of bi-copular languages, in mono-copular languages SL-ness is not only syntactically derived in copular sentences, but it is also lexically encoded in predicates.

Finally, even though SL-CopPs in mono-copular languages lack a visible doubling phenomenon between estar and the predicate, we can take the distribution of SLPs as a clue of evidence that they carry a [uS] feature also in French, Italian, and Romanian. As observed for Spanish (Demonte 1999, 189–90, Picallo 2012, 277), SL-adjectives canonically occupy a postnominal position across Romance languages (54).^[34]

Cinque (2010) suggests that adnominal adjective modification comes in two forms: through direct modification or through a reduced relative clause. According to Cinque, prenominal qualifying adjectives can only be direct modifiers, whereas qualifying adjectives of a reduced relative source always occupy a post-nominal position. If we follow Cinque’s (2010) proposal, we can interpret the necessary post-nominal position of SL-qualifying adjectives as a requirement to stand in a proper syntactic relation with a c-commanding element in the relative clause, that is, within the framework assumed here, as the consequence of an [iS]-[uS] doubling phenomenon (55).

Summing up all the above, when it comes to mono-copular languages (namely, Italian, French, and Romanian), even if they lack an SL-copula, a distinction between IL and SL-structures is found among CopPs. From a formal point of view, this means that a distinction between VPs and AspPs has to be drawn. Moreover, we see that, as in the case of ser/estar languages, such IL/SL distinction is lexically encoded in non-verbal predicates: SLPs are situation-descriptive predicates that differ from ILPs (property-descriptive) in that they carry a [uS] feature.

We can thus confirm the hypothesis that the [Stage] feature is at play in the derivation of SL-CopPs in mono-copular languages. Under this view, SL-CopPs in mono-copular languages would also result from a syntactic agreement operation between the SLP and an Asp head (56). Therefore, the only difference between mono-copular languages and bi-copular languages would lie in the lack of an SL-copula, that is, an overt morpho-phonological realisation of the Asp head of SL-CopPs (cf. [56] and [35]).^[35]

In sum, the inventory of aspectual elements in Italian, French, and Romanian would be as in (57).

As for IL-CopPs, no aspectual heads nor features would be at play in these structures (58), exactly as in the case of serPs in Spanish, Portuguese, and Catalan (37).

As in the case of bi-copular languages, this analysis also justifies those cases in which an ILP is used to describe a stage (not an intrinsic quality) of the subject (examples in [47]). In these cases, the Asp head is merged in the structure as follows (59):

From a cross-linguistic point of view, we conclude that the IL/SL paradigm is equally encoded across languages, both lexically and structurally: non-verbal predicates are clearly divided into two classes and SL-CopPs differ from IL-CopPs by an additional Asp projection. Unlike IL-CopPs, a [Stage] feature drives the derivation of SL-CopPs. SL-CopPs with an SL-base predicate are derived by means of an agreement operation between the SLP and the Asp head: SLPs encode a [uS] feature, that is, a requirement to be c-commanded by an [iS] instance, and the Asp head, carrying [iS], satisfies these configurational needs. This phenomenon is clearly seen in the case of Spanish, Portuguese, and Catalan, where the Asp head is overtly realised as estar, and thus, the semantic doubling with respect to SL-ness between Asp_(estar) and the predicate is visible (35). In contrast, Italian, French, and Romanian lack an estar-like copula and the agreement operation that drives the derivation of SL-CopPs is covert (56). As for SL-CopPs with an IL-base predicate, both in the case of bi-copular (41) and mono-copular languages (59), the Asp head (endowed with a [iS] feature) can be merged in the structure and coerce the ILP to an SL-reading.

We can thus conclude that the difference between bi-copular and mono-copular languages lies exclusively in the presence (Spanish, Portuguese, and Catalan) or the absence (Italian, French, and Romanian) of a morphophonological emergence of the Asp head, that is, an SL-copula (cf. [36] and [57]).