Journal of Pragmatics 128 (2018) 102e115

Contents lists available at ScienceDirect

Journal of Pragmatics
journal homepage: www.elsevier.com/locate/pragma

Evidential commitment and feature mismatch in Spanish

estar constructions
Victoria Escandell-Vidal
Departamento de Lengua Espan ~ ola y Lingüística General, Universidad Nacional de Educacio
n a Distancia (UNED), Senda del Rey,
7 -708A, 28040 Madrid, Spain

a r t i c l e i n f o a b s t r a c t

Article history: Some Spanish constructions with the copula estar plus an adjective, like María esta  muy
Available online 28 October 2017 guapa (María be-ESTAR.PRS.3SGvery pretty), are interpreted systematically, without any
particular contextual requirements, as indicating that the predication is based on the
Keywords: speaker’s direct perception. Interestingly, this phenomenon obtains only for a subset of
ser/estar estar þ adjective constructions.
Direct experience
The aim of this paper is to account for the conditions under which this entailment of direct
Evidentiality
experience appears. More specifically, I argue that it is the result of an interpretive,
Mismatch
ILP/SLP
inferential solution triggered to solve a feature mismatch between the semantic re-
Commitment quirements of the copula estar and the semantic properties of individual-level adjectival
predicates. The evidential commitment is added to provide a situational anchor for the
estar predication when the adjectival predicate is unbounded.
The approach presented here has implications for our view on commitment and evi-
dentiality, and can cast some light on the issue of copula selection in Spanish.
© 2017 Elsevier B.V. All rights reserved.

1. Introduction. The facts

Languages with more than one copular verb, such as Spanish, Catalan or Portuguese, pose special challenges for gram-
matical description (for an overview, see Leonetti, 1994; Fern
andez Leborans, 1999; Batllori, 2006; RAE, 2009: x37.7; Camacho,
2012; Marín, 2015; Pe rez Jime
nez et al., 2015). Copula choice is a much discussed, though not yet well understood, issue.
To add even more complexity, other phenomena seem to interact with copula choice. This is the case of the facts dealt with
in this paper. Consider the estar þ adjective constructions in (1):

(1) a. María esta  muy guapa.

María be-ESTAR.PRS.3SG very pretty.
‘María looks very pretty.’
b. ¡La comida del gato 
esta deliciosa!
The food of-the cat be-ESTAR.PRS.3SG delicious
‘The food of the cat tastes delicious!’

E-mail address: vicky@flog.uned.es.

https://doi.org/10.1016/j.pragma.2017.10.004
0378-2166/© 2017 Elsevier B.V. All rights reserved.
 V. Escandell-Vidal / Journal of Pragmatics 128 (2018) 102e115 103

In addition to the predication expressed, the examples in (1) also convey the idea that the speaker bases her assessment on
direct experience. Thus, the sentences in (1) carry two different commitments (on this notion, see De Brabanter and Dendale,
2008). By uttering (1)a, the speaker commits herself (i) to the belief that María looks pretty, and (ii) to having direct evidence
for her assertion. In a similar way, (1)b commits the speaker (i) to believing that cat food is tasty, and (ii) to having tasted it.
The second commitment is of an evidential nature since it relates the assertion to the source from which the information
has been obtained (Aikhenvald, 2004, 2014; cf. De Haan, 2005a, 2005b, 2005c; Speas, 2010; Kalsang et al., 2013). In languages
with evidential markers, “uttering a sentence S containing an evidential marker [ev] commits the speaker to the existence of a
situation in which she receives evidence for [[S]]” (Davis et al., 2007: 73).
What is interesting about the examples in (1) is that the evidential commitment obtains without any specific linguistic
operator to this effect. In this sense, it represents a challenge to compositionality, since the meaning of the sentence seems to
be not fully derivable from the meaning of its parts. The evidential commitment does not look as a conversational implicature
either, given that no specific set of contextual assumptions is required for it to arise. The robustness of the direct experience
commitment can be made visible by minimal pairs like the following:

(2) a. Ayer vi a 
María. Esta muy guapa.
Yesterday see.PST.1SG to María. Be-ESTAR.PRS.3SG very pretty
‘Yersterday I saw María. She looked very pretty.’
b. #Hace tiempo que no veo a 
María. Esta muy guapa.
Make.PRS.3SG time that not see.PRS.1SG to María. Be-ESTAR.PRS.3SG very pretty
‘I haven’t seen María for a while now. She looks very pretty.’

The sequence in (2)a is perfect, whereas that in (2)b is considered “odd”, “incongruent”, “inconsistent” by native speakers. I
have tested this contrast with dozens of them and they are unanimous in their judgement. The reason is that by uttering (2)b
the speaker commits herself to entertaining contradictory thoughts, claiming at the same time (i) that she has not seen María
for a while, and (ii) that she has direct experience that María looks pretty now.
The examples in (3) provide further examples of utterances conveying an evidential commitment of direct experience:

(3) a. La Real estuvo inteligente y caritativa

The Royal be-ESTAR.PST.3SG intelligent and charitable
‘The Royal [Society of San Sebastian (a football team)] was intelligent and charitable.’
[They played in an intelligent way and did not humiliate their rival team]
(E. Rodrigalvarez. El País, 16/09/1998)

b. El rey estuvo simpatico con los periodistas.

The King be-ESTAR.PST.3SG friendly with the journalists.
‘The King appeared friendly with the media’
n.com, 1/6/2011)
(J. Bono. La informacio
c. El analisis de 
Onega: “El Banco de ~ a es
Espan culpable, pero
The analysis of 
Onega: The Bank of Spain be-SER.PRS.3SG guilty, but
Bruselas estuvo torpe y ciega.
Brussels be-ESTAR.PST.3SG clumsy and blind.

‘Onegas’s analysis: the Bank of Spain is to blame, but Brussels behaved in a clumsy and blind way’
(Ivoox.com, 17/06/2014)

In all three cases, the utterances entail that the speaker has had direct experience for his assessment. In (3)a the sports
caster makes this assertion after having attended the football match; in (3)b a politician who was with the King comments on
their meeting with the media; in (3)c an economic analyst judges the way in which the EU has dealt with a certain problem.
The aim of this paper is to account for the commitments of direct experience that are found in (1)e(3). The main questions
I want to address are the following:

(i) When does the evidential commitment arise?

(ii) Why does it arise?
(iii) What are the implications for a theory of language?

The paper is organized as follows. In Section 2, I will briefly review and comment on several representative approaches on
experiential requirements in copulas and adjectives. Section 3 is devoted to presenting my proposal about when, why and
how evidential commitments arise in estar þ adjective constructions. In Section 4, I will discuss some consequences of my
proposal. The main conclusions will be gathered in Section 5.
104 V. Escandell-Vidal / Journal of Pragmatics 128 (2018) 102e115

2. Experiential requirements

2.1. Estar, an experiential copula?

It would be tempting to suppose that it is the copula estar that has some built-in meaning component that relates the
predication to the speaker’s direct experience. In fact, this idea has been suggested in the literature under different terms. In
this section I will briefly review some of these approaches and argue that this proposal does not cover all the facts in a
satisfactory way.
Gili Gaya (1961) is among the first grammarians to relate estar predications to the notion of ‘immediate experience’:
 fría
We use estar in the judgments that depend immediately on our experience (…); in order to say Aquella nieve esta
[‘That snow is cold’] I need to touch it now (Gili Gaya, 1961: 57-58; my translation, VE-V)
The main difference between ser and estar in Gili Gaya’s view has to be accounted for in aspectual terms: ser-predications
are imperfective (unbounded) and attribute a property to a subject, taking into consideration the property in itself; estar-
predications, in contrast, are perfective (bounded), and present the property “as the result of an action, a transformation or a
change that we know, or suppose that has taken place, is taking place or will take place” (Gili Gaya, 1961: 62; my translation,
VE-V). What is relevant to the current discussion is that this change-of-state has to be perceived by the speaker in order for
the experiential effect to obtain:
In order to know whether the action or the change has taken place, we usually resort to experience. I see that a boy has
grown and I say esta alto [be-ESTAR.PRS.3SG tall]; I drink my coffee and say esta
 frio [be-ESTAR.PRS.3SG cold]. In other words,
we use estar in the judgements that depend immediately on our experience. […] It is the experience that gives a
perfective sense to the enounced quality (Gili Gaya, 1961: 62; my translation VE-V).
A similar approach is suggested in Querido (1976) for Portuguese. For him, “estar is the appropriate copula to report a first
sensorial experience.” (Querido, 1976: 354; apud Maienborn, 2005; Roby, 2007). When one cannot tell whether a perceived
property is a permanent or a transitory one, one should opt for estar. The notion of change-of-state seems to play no role at all
in Querido’s analysis: having ‘first sensorial experience’ is enough to license the use of estar, even when the speaker has no
grounds for claiming any change with respect to a previous situation.
Roby (2007: 97) endorses a weak version of Querido’s account in which “estar is preferred over ser when reporting a first
sensorial experience and necessitated when the speaker cannot assume the usual state of what is being described”.
Maienborn (2005) also relates estar to personal experience, even overriding the alleged correlation between ser and
inherent properties: “If the speaker’s claim is based on fresh evidence, estar may also be used to express essential properties”
Maienborn (2005: 160). In her approach, the use of estar is felicitous only if the context can provide a contrasting topic
situation, i.e., one for which the predication does not hold. Thus, in (4) the use of estar imposes the need to identify a topic
situation contrasting with other possible topic situations to which the predicate does not apply:

(4) La carretera 
esta ancha.
The road be-ESTAR.PRS.3SG wide.
‘The road is wide [now/here/as far as I know]

This contrast can be obtained with respect to three different dimensions: temporal, spatial and epistemic, depending on
whether the contrast involves other times, other places or other situations of (un)certainty. It is the epistemic dimension that
is relevant to our discussion: estar is licensed by fresh evidence, i.e., in a situation contrasting “with other topic situations that
do not allow us to decide whether the predicate applies to the subject referent or not” (Maienborn, 2005: 172).
The proposals reviewed so far emphasize the experiential nature of estar, either related to a change-of-state, to a first
perceptual experience, to direct experience or to uncertainty. Now, though it is clear that there are connections between estar
and personal experience, this is not necessarily the case in all the situations.
Consider the example in (5):

(5) Este jamo n serrano esta  fenomenal.

This ham Serrano be-ESTAR.PRS.3SG wonderful.
‘This Serrano ham tastes wonderful.’

n serrano did not taste good prior to its being tasted

As Roby (2007: 27) outs it, in (5) “there is no implication that the jamo
and evaluated by the speaker”. Change-of-state, then, is not a necessary condition for estar predications.
As for the ‘first sensorial experience requirement’, it does not give the right generalization either. If this were the case, the
immediate prediction would be that repeated perceptual experience will cause the copula shift to ser. However, this is not
what Spanish speakers do. After bathing into a Swedish river, the speaker will always say (6), no matter how many times she
has had this experience before:
 V. Escandell-Vidal / Journal of Pragmatics 128 (2018) 102e115 105

(6) El agua 
esta muy fría.
The water be-ESTAR.PRS.3SG very cold
‘The water is very cold.’

Finally, having fresh, direct, experiential evidence is not a necessary condition either, as shown by the examples in (7):

(7) a. Mario esta  cansado.

Mario be-ESTAR.PRS.3SG tired.
‘Mario is tired’
b. Las tiendas n
esta cerradas.
The shops be-ESTAR.PRS.3PL closed
‘The shops are closed.’

In order to assert (7) felicitously, the speaker does not need to have direct evidence for Mario’s tiredness, or for the shops
being closed. In these cases, the estar þ adjective constructions do not carry any entailment that the speaker is basing her
assessment on direct perception. Similarly, if we substitute the adjectival predicate guapa (‘pretty) in (2) by enferma (‘ill’), as in
(8), the evidential commitment does not arise, and therefore both (8)a and (8)b are perfectly congruent:

(8) a. Ayer vi a María. Esta muy enferma.

Yesterday see.PST.1SG to María. Be-ESTAR.PRS.3SG very ill
‘Yersterday I saw María. She is very ill.’
b. Hace tiempo que no veo a María. 
Esta muy enferma.
Make.PRS.3SG time that not see.PRS.1SG to María. Be-ESTAR.PRS.3SG very ill
‘I haven’t seen María for a while now. She is very ill.’

In (8)a the speaker has met María, but in order to assert that María is ill, it is not necessary for the speaker to have
perceived that María is ill; it is perfectly possible that María herself has told the speaker about her invisible illness. Similarly,
in (8)b the speaker may have learnt that María is ill from others, and this is a plausible explanation for the fact that they have
not met for a while. This shows, then, that with a predicate like estar enfermo (‘to be ìll’), the direct experience commitment
does not obtain.
The previous considerations thus indicate that the generalization that estar is the copula chosen to indicate first immediate
impressions does not capture the facts under discussion in an adequate way: it can predict neither the situations in which
estar will be used, nor the interpretations that the utterances with estar will receive.

2.2. Direct experience and predicates of personal taste

If the copula is not the origin of the evidential commitment found in (1)e(3), one then could suppose that it is the
adjectival predicate that is responsible for this interpretation. In this section, I will consider this possibility.
Recent research has shown that the direct experience requirement can be found as a lexical feature of some predicates:
these are known as ‘predicates of personal taste’ (PPT) (Lasersohn, 2005, 2009; Stephenson, 2007; Pearson, 2013; Ninan,
2014; Bylinina, 2014, 2017):
In order to assert that ‘x is P’ for some taste predicate P, one typically must have direct sensory experience of the
relevant kind on the basis of which to judge whether x is P. For tasty, for example, I must have tasted the object I am
talking about (Pearson, 2013: 118).
Thus, the direct experience requirement found in (1)b has to do with the adjectival predicate, rather than with the copula
estar, as shown by the fact that also in English e a language without copula alternation, the speaker who utters the equivalent
to (1)b commits herself to having tasted cat food. This approach can be extended to other adjectives indicating personal
judgments, such as divertido (‘funny’) and genial (‘great’) in sentences like the following:

(9) a. La película esta  divertidísima.

The movie be-ESTAR.PRS.3SG very-funny.
‘The movie is very fun.’
b. El disco 
esta genial.
The record be-ESTAR.PRS.3SG great.
‘The record is great.’
106 V. Escandell-Vidal / Journal of Pragmatics 128 (2018) 102e115

(9)a requires that the speaker has watched the movie, and (9)b is felicitous only after listening to a musical record.
The experiential commitment in these cases cannot be cancelled without contradiction, as shown by the oddity of (10):

(10) #Esto esta  delicioso, pero yo no lo he probado.

This be-ESTAR.PRS.3SG delicious, but I not it have.PRS.1SG taste.PP
‘This is delicious, but I haven’t tasted it.’

As a way to explain this behavior, it has been suggested in the literature that PPTs contain a slot for an experiencer in their
argument structure (Pearson, 2013; McNally and Stojanovic, 2014). This experiencer argument is identified with the speaker s
in autocentric contexts, as proposed by Ninan (2014) in his Acquaintance Principle:

(11) In autocentric contexts c, sc knows (at tc in wc) whether [[o is tasty]]c is true, only if sc has tasted o prior to tc
in wc. (Ninan, 2014: 13)

In addition, PPTs can be characterized as expressing subjective opinions: the truth of a sentence containing a PPT does not
depend only on the state-of-affairs, but also on who the speaker is. If I utter (1)b or (9)a, I am expressing a personal judgement,
not a universal truth. This means that tastiness and fun are not intrinsic properties of entities or events; rather they are
relative to subjects of experience. This is why different subjects can have different opinions without this disagreement giving
rise to contradiction (a phenomenon known as ‘faultless disagreement’; Ko €lbel, 2004).
But this is not the whole story either. Consider the minimal pair in (12):

(12) a. El jamon iberico esta riquísimo.

The ham Iberian be-ESTAR.PRS.3SG delicious.
‘The Iberian ham tastes delicious.’
b. El n ib
jamo erico es riquísimo.
The ham Iberian be-SER.PRS.3SG delicious.
‘Iberian ham is delicious.’

Both sentences are possible, but their meaning and commitments are different: (12)a, with estar, triggers the direct
experience interpretation, so it can be felicitous, for instance, after having tasted a specific piece of Iberian ham; (12)b, in
contrast, is not a perceptual report about a recent experience with a specific ham, but rather a generic assessment about
Iberian ham in general. The experiencer is present in either case, because its occurrence is a built-in lexical feature of the
predicate, but the need to have a direct exposure is retained only with estar. With ser, in contrast, the experiencer argument is
bound by the generic quantifier GEN (Chierchia, 1995), hence the genericity of the statement. Uttering (12)b is compatible
with the speaker having tasted Iberian ham in one or several different occasions, but would not be not contradictory if no
particular situation can be identified where the information has been directly acquired.
Thus, the generalization about PPTs can account for direct experience commitments found in sentences with prototypical
PPTs such as tasty and fun. It can be extended to other evaluative adjectives like delicious, pleasant, pretty, boring and
disgusting, and also to moral predicates, aesthetic predicates and even to gradable adjectives like tall and rich, as has been
suggested in the literature (Moltmann, 2010). It is difficult, though, to accommodate other predicates into this picture. In
addition to examples like those in (3) above, one can find cases such as those in (13):

(13) a. Aquel día Juan estuvo tico.

democra
That day Juan be-ESTAR.PST.3SG democratic.
‘That day Juan behaved as a true democrat.’

b. Pedro esta  hoy muy franc

es.
Pedro be-ESTAR.PRS.2SG today very French!
‘Pedro is [behaving as] Frenchmen today!’

Democra tico (‘democratic’) and franc es (‘French’) are relational adjectives, not PPTs. They are typically used with ser to
classify entities, not to express personal opinions. Interestingly, the sentences in (13) carry the same direct experience
commitment, namely, that the speaker has witnessed Juan’s or Pedro’s behavior and, on the basis of this perceptual evidence,
she can assert the propositional content expressed in (13). It is true that the sentences in (13) are stylistically marked (see
Section 4.3 for more details), but their interpretation shows that being a PPT does not seem to be a necessary condition for the
evidential commitment to arise.
 V. Escandell-Vidal / Journal of Pragmatics 128 (2018) 102e115 107

3. A proposal

In the previous sections I have examined two possible accounts for the direct experience commitments illustrated in
examples (1)e(3). I have tried to show that none of these approaches covers the data in an adequate way: on the one hand, the
evidential commitment does not arise with all the adjectives when combined with estar; on the other hand, not all the
adjectives that give rise to the experiential commitment belong to the class of PPTs. In this section I want to put forward a
proposal that can explain and predict when and why the evidential commitment of direct experience arises in some
estar þ adjective constructions. My analysis elaborates on the proposal put forward in Escandell-Vidal and Leonetti (2002)
and offers a new answer to the question of why the evidential commitment of direct perception arises.

3.1. The distribution of the phenomenon

The first step is to offer a principled explanation for the fact that the evidential commitment arises only with a subset of
estar constructions. We have found it when estar combines with predicates such as guapo (‘pretty’), delicioso (‘delicious’),
tico (‘friendly’), torpe (‘clumsy’), genial (‘great’), and even democra
inteligente (‘intelligent’), caritativo (‘charitable’), simpa tico
(‘democratic’) and franc es (‘French’). It does not obtain, in contrast, when estar combines with enfermo (‘ill’), cansado (‘tired’),
cerrado (‘closed’). Therefore, the key to the problem is to be found in the class of adjectives.
The relevant difference between the classes of guapo (‘pretty’) and enfermo (‘ill’) has to do with their aspectual features:
guapo is an ‘individual-level predicate’ (ILP) whereas enfermo is a ‘stage-level predicate’ (SLP). ILPs are unbound and express
properties of individuals (i.e., of objects and kinds), while SLPs are bound and express properties of stages (i.e., spatiotemporal
“slices” of individuals) (Milsark, 1974; Carlson, 1977; Kratzer, 1989, 1995, among others). The ILP/SLP distinction, though not
uncontroversial, is a well-established one in the literature and is independently needed to account for a large number of
grammatical facts (for a critical overview, see Leonetti, 1994; Arche, 2006; RAE, 2009; Marín, 2010, 2015; Camacho, 2012;
Fabregas, 2012; Silvagni, 2017).
If this line of reasoning is accepted, then it is easy to see that all the adjectives that give rise to the evidential commitment
in the examples considered so far happen to be ILPs. The descriptive generalization that emerges from this is, therefore, that
the entailment of direct experience appears only when estar is combined with an ILP (cf. Escandell-Vidal and Leonetti, 2002).
In contrast, if estar occurs with a SLP no experiential interpretation is imposed on the construction.
It is, therefore, the combination estar þ ILP that triggers the evidential commitment. We have, then, an answer to the
question about the distribution of the phenomenon, so we are able to determine and predict when the systematic
commitment of direct experience will obtain. This account also has the advantage of maintaining the intuition found in the
literature that estar is the choice copula for giving assessments based on direct perception, without overgeneralizing this
effect to all the estar constructions. My proposal can also explain why and how the interpretation of PPTs (a subclass of ILPS;
Chierchia, 1995) changes depending on whether they occur with ser or with estar: only in the latter case the direct experience
commitment arises, whereas with ser a generic interpretation is preferred.

3.2. Estar and classes of predicates

Predicting when the experiential commitment will arise is a step towards understanding the phenomenon under dis-
cussion. But a complete explanation of what is at stake here crucially requires giving an answer also to the second question,
namely, why the evidential commitment arises at all. To do so, some further considerations about the semantics of the copulas
and the ILP/SLP distinction are in order here.
As for the copulas, I assume that each copula has a semantics of its own. The differences can be cast in various terms,
depending on one’s favorite theoretical framework. For Delbecque (1997, 2000), ser attributes a category, is unbounded and
[spatial]; estar, in contrast, attributes a state, is bounded and [þspatial]. For Maienborn (2005), the copula estar contains an
additional meaning component with respect to ser: the requirement to link the predication to a specific topic situation. Thus,
estar needs a spatiotemporal anchor, whereas ser does not include such a requirement. More specifically, in Maienborn’s
(2005) terms, estar carries the additional presupposition that the referential argument z is related (via a free variable R) to
a specific discourse situation si:

(14) ser: lP lx lz [zQ[P(x)]

estar: lP lx lz [zQ[P(x)]y[si | R (z, si)]] (Maienborn, 2005: 168)

Estar contains a specificity presupposition on the topic situation:

…by using estar speakers restrict their claims to a particular topic situation they have in mind; by using ser speakers
remain neutral as to the specificity of the topic situation” (Maienborn, 2005: 168).
This view is also compatible with more syntactically oriented proposals about the ser/estar distinction, such as the one in
Brucart (2012). According to Brucart’s view, copular verbs in Spanish either have or lack an interpretable feature of terminal
coincidence [iRT]:
108 V. Escandell-Vidal / Journal of Pragmatics 128 (2018) 102e115

(15) Estar: [þiRT]

Ser: [iRT]

The particulars of the analysis can thus be implemented in different ways. What matters for the present discussion is that
estar forces the predication to be relativized to a situation e i.e., the one to which the speaker limits her commitment.
Let’s consider the ILP/SLP distinction in more detail now. As mentioned above, this distinction is responsible for contrasts
as those illustrated in example in (16) (cf. Escandell-Vidal and Leonetti, 2002). Only SLPs seem natural in perception reports
(cf. (16)a), depictive secondary predicates (cf. (16)b), absolute constructions (cf. (16)c), small clauses headed by con (‘with’) (cf.
(16)d), and small clause complements of impersonal haber (‘have’) with a predicate/subject order (cf. (16)e):

(16) a. Lo noto {cansado/*inteligente}.

He.ACC feel.PRS.1SG {tired/intelligent}
‘I found him {tired/*intelligent}’
b. 
Llego {descalzo/*astuto}.
Arrive.PST.3SG {barefoot/*clever}
‘He arrived {barefoot/*clever}’
c. {Estirado/ *alto}, Juan logra tocar el techo.
Stretch.PP/ *tall}, Juan manage.PRS.3SG touch the ceiling.
‘{Stretched/*tall}, Juan can reach the ceiling.’
d. Con el coche {estropeado/ *japon
es}, no pudimos proseguir.
With the car {break.PP/ *Japanese}, not can.PST.1PL continue.
‘With the car {broken down/*Japanese}, we couldn’t continue.’
e. Hay [SC [disponibles] [bomberos]]/* Hay [SC [altruistas] [bomberos]]
Have.PRS.3G available firemen/Have.PRS.3G altruistic firemen
‘There are firemen available/*There are firemen altruistic.’

In addition, the ILP/SLP distinction shows a remarkable cross-linguistic robustness and has been invoked to account for a
wide variety of phenomena in different languages, including the possibility of there-insertion in copular sentences in English
(Milsark, 1974), the interpretation of bare plural subjects and their acceptability in perception reports (Carlson, 1977), re-
strictions on absolute constructions (Stump, 1985), compatibility with modifying adverbials and the conditions governing
quantifier split in German (Kratzer, 1995), and the occurrence of adjectives as depictive secondary predicates, among others.
Various proposals have been put forward in the literature to account for these contrasts (see Fa bregas, 2012 for a critical
overview). Kratzer (1995) and Diesing (1992) argue that ILPs and SLPs differ in argument structure: SLPs have an extra
argument position for events or spatiotemporal locations (i.e., they require the property that they denote to be linked with an
external situation), whereas ILPs lack this position. Predications based on SLPs can describe events relativized to a specific
topic situation, but those made up of ILPs cannot. This view is again compatible with recent syntactic approaches. Following
Brucart (2012), predicates can be divided into two classes depending on whether or not they contain an uninterpretable
feature of terminal coincidence [uRT]: predicates with this [uRT] feature correspond to SLPs, whilst those lacking it are ILPs.

3.3. Feature mismatch in copular constructions

With the previous considerations in mind, I can now go into the second part of my proposal. In a nutshell, the idea I want to
put forward is the following: when estar is combined with an ILP the semantic requirements of the copula and those of the
adjectival predicate do not match. It is precisely this mismatch that triggers an inferential repair mechanism by adding the
evidential commitment as a way to satisfy the missing requirements of the copula and hence accommodate a matching
interpretation (cf. Escandell-Vidal and Leonetti, 2002, 2011).
Consider the various possible combinations between ser and estar, on the one hand, and ILP/SLP, on the other. When an SLP
is combined with estar, the two constituents share a crucial semantic feature: estar presupposes a spatiotemporally restricted
situation and SLPs contain a position for a spatiotemporal argument, so the features of the copula and those of the adjectival
predicate match. Similarly, when ser is combined with an ILP, the semantic properties of the two elements match, in the sense
that neither imposes any kind of spatiotemporal anchoring. This is the reason why estar and SLPs, on the one hand, and ser and
ILPs, on the other, seem to belong together.
In syntactic terms, when estar is combined with a SLP, the uninterpretable feature [uRT] in the predicate is licensed by
agreement with the corresponding interpretable feature [iRT] of the copula estar; in this case, the features in the copula and
the predicate match or agree. In contrast, when estar is combined with an ILP, which does not contain an uninterpretable
feature of terminal coincidence, the [iRT] feature of the copula cannot agree with the predicate; the copula maintains its
 V. Escandell-Vidal / Journal of Pragmatics 128 (2018) 102e115 109

interpretable feature [iRT] and imposes a delimited reading on the predicate, giving rise to constructions with a more or less
pronounced marked flavor. In any event, the result is not ungrammatical, since this agreement relation is a not formal
requirement of the copula.
Thus, the combination of estar (the situation bounded copula) with an ILP (an unbounded predicate) contains a mismatch
between the features of the copula and those of the predicate. The notion of ‘mismatch’ is therefore central to the current
discussion. A mismatch is a conflict or an incompatibility between the features of two linguistic items (Francis and Michaelis,
2003; Pylkk€ anen, 2008; de Swart, 1998, 2003, 2011; Moravcsik, 2010; Katsika et al., 2012). Mismatches have been identified in
the literature as the source of enriched interpretations. Additional meaning components, which would otherwise represent a
challenge for the compositionality hypothesis, can be introduced as a repair strategy when linguistic items with conflicting
requirements are composed with each other. In this way, apparent violations of compositionality are accounted for as the
result of specific interpretive repair strategies.
This view makes it possible to explain why the evidential commitment in estar þ ILP constructions is not merely a free
inference, or a contextually determined one; rather it is compulsory. This is so because it is triggered by the linguistic form of
the sentence itself, namely, by the feature mismatch, not by any sentence-external, contextual factors.
So far, then, my proposal can be summarized as follows:

e The systematic experiential commitment found in some estar constructions arises only when estar is combined with an
ILP.
e The combination estar þ ILP contains a semantic mismatch between the requirements of estar (the bounded copula
requiring a situational anchoring) and those of the ILP (an unbounded predicate).

We have now an answer for the question of why the evidential commitment obtains only in estar þ ILP constructions: it is
due to the existence of a feature mismatch. But once this question has been answered, a new one arises, namely, why
repairing the feature mismatch involves adding a direct experience commitment. This is where the notion of evidentiality
comes into play.

3.4. Evidentiality

Evidentiality is usually conceived of as “a grammatical category that has source of information as its primary meaning e
whether the narrator actually saw what is being described, or made inferences about it based on some evidence, or was told
about it, etc.” (Aikhenvald, 2004:1). For many languages (Quechua, Tibetan, Tuyuca, etc.) it is compulsory to mark the source
of the information, together with the information itself: such languages are said to have ‘evidentials’ (Aikhenvald, 2004, 2014;
De Haan, 2005a,b; Speas, 2010; Kalsang et al., 2013). So, for example, in languages with complex evidential systems, if the
speaker has witnessed the communicated event (i.e., if she has direct, first-hand sensory experience), a special mark, a direct
evidential, has to be used. For direct evidentials, the felicity condition is that the situation described must have been directly
observed by the origo: this is what Garret (2001: 53) calls the ‘observability restriction’. So, if a marker of direct evidence
occurs, the speaker is committed not only to the truth of the information conveyed, but also to having direct experience for it.
Similarly, if the information conveyed by the speaker is the result of her thoughts (from logical deduction to mere supposition
and guessing), she has to indicate it as well by using the adequate inferential indicator. Finally, when the speaker has acquired
the information from others (including report, hearsay, rumor and folklore), this has to be explicitly communicated by means
of reportative markers. In languages with grammatical evidentials, the indication of the source is obligatory and the infor-
mation encoded by evidential indications cannot be cancelled.
The approach to evidentiality I will rely on is that put forward in Speas (2010) and Kalsang et al. (2013), which takes a
slightly different perspective on the topic. Instead of viewing evidential indicators as encoding the kind of source from which
the information asserted was received (whether from direct perception, inference or hearsay), in Speas’s account what
languages with grammatical evidentials encode specific sets of relations between three situations:

(i) the ‘evaluation situation’ (ES): the minimal situation of which p is true,
(ii) the ‘information situation’ (IS): the minimal situation in which the speaker came to know that p, and
(iii) the ‘discourse situation’ (DS): the speech event.

These three situations are linked to each other in terms of relations of precedence (>) and inclusion (3), in a way clearly
reminiscent of Reichenbach’s account for tense and temporal relations. For instance, direct evidentials indicate that the
evaluation situation ES (the one of which p is true) is included in the information situation IS (the one in which the speaker
came to know that p), and that they both are accessible from the discourse situation DS (Speas, 2010; Kalsang et al., 2013). This
can be represented as in (17):

(17) ES 3 IS
ES > DS
IS > DS
110 V. Escandell-Vidal / Journal of Pragmatics 128 (2018) 102e115

The inclusion of the ES in the IS captures the fact that the speaker has had direct evidence: by uttering a sentence with a
marker of direct evidence, the speaker is committed to the existence of a situation in which she acquired the information she
communicates. This view on evidentiality provides the basis for the account I want to put forward in what follows.
I have argued that estar þ ILP constructions contain a mismatch because estar presupposes the existence of a spatio-
temporally bounded situation, but ILPs are unbounded, so estar is unable to find a matching spatiotemporal anchor in the
local domain of its predicate. Now, my claim is that the direct experience commitment we find in (1)e(3) is the effect of an
abstract operation by which an information acquisition situation (IS) is inferred that includes the assertion (ES). This is a
means of satisfying the semantic requirements of the copula estar by binding it to another situation. Since the possibility to
trace back an assertion to its information source is available for any assertion, the strategy of relating the ES to the IS is a very
easy and accessible way to anchor a predication, whenever needed.
The two commitments that are conveyed by estar þ ILP constructions correspond to the two situations related, the ES and
the IS, as shown in (18):

(18)

In most cases, evidential indications give rise to binary constructions, where the propositional content and the evidential
commitment are independent from each other (Higginbotham, 2009). Now, not surprisingly, the experiential commitment
found in estar þ ILP constructions shares many of the properties with direct evidentials. For instance, the experiential
commitment in estar þ ILP constructions projects over negation. Consider (19), the negation on (1)a:

(19) María no 
esta guapa.
María not be-ESTAR.PRS.3SG pretty.
‘María doesn’t look pretty.’

The sentence in (19) can only mean that the speaker does not find María pretty, not that the speaker has not direct ev-
idence for that.
The evidential commitment in estar þ ILP constructions is also insensitive to disagreement, which targets only the
propositional content. So, if I utter (1)b, and you say That’s not true, the disagreement affects our opinions about cat food
tastiness, not the fact that I have tasted cat food (and, for what matters, the disagreement also entails that you have tasted it
too!). Finally, the commitment of direct experience cannot be cancelled by the speaker herself either: this is the reason of the
incongruity of the cases in which this commitment is explicitly denied. When the immediate context blocks the possibility of
inferring the speaker’s direct experience by denying the existence of the IS, estar þ ILP will be odd, due to the incompatibility
of two contradictory assumptions, whereas an estar þ SLP combination will still sound natural, as shown in (2) and (8) above.
(2)b is odd because the statement that the speaker has not seen María for long is in contradiction with a present-tensed
estar þ ILP sentence., i.e., with a grammatical environment carrying an inference of direct perception. (8)b, in contrast, does
not involve any feature mismatch and hence does not need any evidential repair mechanism; María’s illness can be inter-
preted, for example, as the reason why she does not go out as often as before.
In this section, I have presented an account for the direct experience commitment that occurs systematically whenever the
copula estar is combined with an ILP. The situation of information acquisition is a way to repair the mismatch by adding a
situational anchor to the assertion. This view has some implications and consequences. In the next section I will briefly point
out some of the most important ones.

4. Implications and consequences

4.1. Interactions between evidentials: estar þ ILP and the conjectural future

If the idea that estar þ ILP constructions systematically entail that the speaker has direct evidence for her assertion, the
prediction is that there will be interactions with other kinds of evidential indications. If so, this would provide very strong
support for the suggested analysis. The Spanish future tense offers, I think, a good case in point.
 V. Escandell-Vidal / Journal of Pragmatics 128 (2018) 102e115 111

In Escandell-Vidal (2014) it is argued that the Spanish inflectional future does no longer indicate future time reference, but
conjecture. Indicating conjecture is a kind of evidential indication: the speaker presents a state-of-affairs and, at the same
time, indicates that she has no better grounds than her intuition (i.e., the mode of knowing is intuitive inference).
Now, if this analysis is on the right track, the immediate prediction is that the estar þ ILP construction, which presents the
information as based on the speaker’s experience, will be incompatible with the simple future, which indicates that the
information is based on the speaker’s intuition. This is, in fact, what we find. Consider the following dialogues:

(20) A: -El rey 

esta bromeando con los periodistas.
The King be-ESTAR.PRS.3SG joking with the journalists.
‘The King is joking with the journalists.’

B: 
-Estara contento /de buen humor.
Be-ESTAR.FUT.3SG happy /of good humor
‘He must be happy/in good spirts.’
C: 
-#Estara simpatico.
Be-ESTAR.FUT.3SG friendly
# ‘He must be friendly.’

(21) A: -No lo consigue. ¿Qu

e le pasa?
Not it.ACCget.PRS.3SG. What he.DAT happen.PRS.3SG?
‘He can’t make it. What’s the matter with him?’
B: -No e…
s 
Estara dormido/… cansado/ drogado
Not know.PRS.1SG… Be-ESTAR.FUT.3SG asleep/ tired/ drugged
‘I don’t know. He must be asleep/tired/drugged.’
C: - #No s e… 
Estara torpe/ tonto/ lento...
Not know.PRS.1SG… Be-ESTAR.FUT.3SG clumsy/ goofy/ slow
# I don’t know. He must be clumsy/goofy/slow.

Answers B and C are meant to express conjectures about the content expressed in A. When the conjecture uses estar þ SLP,
as in the replies given by speaker B, the future is ok, since no contradiction arises between the future and the copular
predication. In contrast, when the future is combined with estar þ ILP, as in speaker C’s replies, the result is clearly odd. This is
due to the contradiction between what the estar þ ILP predication entails (namely, that the speaker has direct evidence) and
what the future indicates (namely, that the speaker has indirect, conjectural evidence). The contrasts illustrated in (20) and
(21) are systematic and there is common agreement on the oddness of C’s versions.
This phenomenon would be totally unexpected for other approaches, but comes as a natural consequence if my proposal is
accepted(see Escandell-Vidal, forthcoming, for details). Only if we accept the idea that the evidential commitment arises as a
way to repair the mismatch between estar and ILPs, and only if we accept that the simple future in Spanish has a basic
evidential semantics, would we be able to understand how and why estar þ ILP in the future tense gives rise systematically to
incongruent sentences.

4.2. Changing the origo

The above considerations apply to autocentric contexts, in which the speaker is identified with the experiencer by default.
This does not mean, of course, that changes in the origo are not possible at all: they are possible, but must be overtly marked.
Consider the example in (22):

(22) Hace tiempo que no veo a los ez.

Pela
Make.PRS.3SG time that not see.PRS.1SG to the ez.
Pela
Por lo visto, su hijo 
esta altísimo.
Apparently, their son be-ESTAR.PRS.3SG tall.
‘I haven’t seen the Pelaez family for quite a while. Apparently, their son is very tall.’

The example in (22) is almost identical to (2)b. However, (22) is perfectly acceptable. The reason is the occurrence of the
marker por lo visto (‘apparently’), which has precisely the effect of indicating a shift in the origo, and so both the assertion and
the evidential commitment are attributed to another individual. In (22) the speaker is responsible neither of the assertion, nor
of the evidential commitment: she is merely reporting what another individual has asserted on the basis of his own direct
experience.
112 V. Escandell-Vidal / Journal of Pragmatics 128 (2018) 102e115

Changes of the origo do not represent a problem for my analysis. If an explicit indication is provided that other people (not
the speaker) are the source of the asserted content, this indication suffices to switch from the default, autocentric context to
an exocentric interpretation, where the source has to be linked to someone else. In any case, it is important to remark that the
change in the origo cannot affect only one of the two commitments: the assertion and the evidential indication must be linked
together to the same individual.
Some syntactic environments also favor changes in the deictic center. Consider the example in (23):

(23) María no se comio  el bocadillo. El salami estaba malísimo

María not CL eat.PST.3G the sandwich. The salami be-ESTAR.IMP.3SG very-bad
‘María didn’t finish her sandwich. The salami tasted bad.’

The assertion and the evidential commitment in the estar construction can be attributed to the speaker, as in the cases we
have examined before; but in (23), with the imperfective past, a change of deictic center is also acceptable, so this sentence
can convey María’s thoughts. When a new subject of consciousness is established, it can be the holder of both the assertion
and the evidential commitment. It is María’s voice that is being presented here, not the speaker’s, in an instance of polyphonic
discourse. Again, what is still impossible is to dissociate the assertion expressed from the evidential commitment.

4.3. Markedness

All estar þ ILP constructions contain a feature mismatch, but they are not of a kind: some of them sound quite natural,
while others have a marked flavor. Consider the sentences in (24)e(27) (the % indicates that the sentence sounds marked):

(24) a. El caf
e 
esta frío.
The coffee be-ESTAR.PRS.3SG cold.
‘I find the coffee (too) cold’
b. El ~o
nin esta alto.
The boy be-ESTAR.PRS.3SG tall.
‘[I find that] The boy is tall’
c. María estuvo inteligente.
María be-ESTAR.PST.3SG intelligent.
‘Maria was intelligent (on that occasion)’

(25) %Juan esta muy brita nico.

Juan be-ESTAR.PRS.2SG very British!
‘Juan behaves in a very British way.”

(26) -¿A qu

e hora sale vuestro tren? -le preguntaba Miguel a Zacarías.
What time does your train leave? -Miguel asked to Zacarias
s treinta.
-A las veintido
Twenty-two thrity.
s tú muy ferroviario.
-%Esta
be-ESTAR.PRS.2SGyou very “railroad”
‘You are talking like a railroad man.’
nchez Ferlosio: El Jarama, 1955)
(R. Sa

(27) %La red ha estado tica

democra
The net have-PRS.3SG be-ESTAR.PP democratic
‘The net has been democratic.’
(2016 Australian Open. Djokovic vs Federer match, 28/01/2016)

There is common agreement among the native speakers to whom I have presented the examples in (24)e(27) about their
status. The sentences (24)a-b sound totally natural; (24)c is still natural, though more marked; (25) and the final turn in the
 V. Escandell-Vidal / Journal of Pragmatics 128 (2018) 102e115 113

dialogue in (26) are clearly marked; (27) has definitely a very marked flavor. The robustness of these intuitions calls for an
explanation.
The account I have offered in the previous sections is built on the idea that the mismatch is repaired by inferring the
existence of a situation in which the speaker has experienced the event that is at the basis for her assertion. By default, the
speaker has the role of the experiencer in this situation. The degree of markednes, I argue, depends on the difficulty to retrieve
or accommodate a suitable information situation (IS). This has to do with both the kind of predicate and world knowledge.
The most natural cases are those with predicates of personal taste (PPTs): they already contain an experiencer slot in their
argument structure, so it is easy to identify it with the experiencer of the IS. A predicate like frío (‘cold’) is similar to a PPT in
that it also requires a judge to establish the standard according to the assertion is true, i.e., a coffee can be (too) cold or not
depending on one’s personal preferences. Adjectives like frío presuppose the existence of a judge responsible for establishing
the threshold of the attributed property. They are judge-dependent predicates (Bylinina, 2017). Judge-dependence favors the
identification of the judge with the experiencer in the acquisition situation. This reasoning also applies to other gradable
dimensional adjectives like alto (‘tall’), for which there may not be a totally objective standard.
Adjectives like inteligente (‘intelligent’) pose a further challenge. They do not operate on a physical dimension, but on a
more abstract kind of assessment; still they are gradable and subject to some extent to the estimation of a judge. Whereas
temperature and height are directly perceptible properties, intelligence has to be established indirectly, for instance, on the
basis of an individual’s performance or behavior. In the case of (24)c, then, the experiencer must have witnessed an event
from which she can have formed the belief that María behaved in a particularly intelligent way. Therefore, the accommo-
dation process here needs more steps than in the previous cases.
The predicates in (25)e(26) add a further level of complexity to the interpretation. They are relational predicates: they do
not attribute properties to entities in isolation, but rather establish connections between two entities; this connection is, in
principle, non-gradable. When these predicates are combined with estar, the identification of an acquisition situation (in
which the speaker is a direct experiencer of the relation) is more difficult to obtain. Forcing a gradable interpretation by
means of the insertion of muy (‘very’) seems to make things easier. Stereotypicity seems to play a significant role: it is easier to
imagine what counts as acting temporarily in an intelligent way, while it is more difficult e though not impossible e to figure
out what can count to behave in a British way or in a railroadman way.
An additional difficulty is found when the subject is not human, so one cannot invoke a particular behavior as the basis for
the assessment. The example in (27) was uttered by the sports caster during the Men’s Single Semifinal match of the 2016
Australian Open. What he wanted to convey is that the number of balls hitting the net and crossing into the opposite side
(which are considered as legal returns) has been almost equal for each player, so none of them was favored. This interpre-
tation is difficult to obtain not only because it needs more supporting contextual assumptions, but also because this kind of
net points are not a stereotypical instantiation of democracy. In any event, the only interpretation available for (27) is, as my
analysis predicts, the experiential one.
These contrasts show that some interpretations can be costlier than others, and that this depends on the difficulty to
identify the perceived event e something that has to do more with stereotypicity of situations than with grammar or
pragmatics. Nevertheless, this difficulties are not counterexamples to my account, so the interpretation proceeds along the
same lines in all the cases: the combination estar þ ILP triggers an evidential interpretation e one in which the speaker is a
direct experiencer of a situation where she acquires the assessment she expresses.

4.4. Temporal persistence

Before closing this section, some words are in order about the issue of temporal persistence. As mentioned before, a
common way to account for the distribution of ser and estar in Spanish relates the choice to the temporal or transitory nature
of the predicates. However, we can find examples of estar with both SLPs and ILPs with modifiers of duration, as in (28):

(28) a. Estuvo amabilísimo, como siempre.

be-ESTAR.PST.3SG very.kind, as always.
¡Es el hombre mas amable que he visto!
be-SER.PST.3SG the man more kind that have.PRS.1SG see.PP
‘[I find] he was very kind, as usual. He is the kindest man I’ve ever met!
b. Fue amabilísimo, como siempre.
be-SER.PST.3SG very.kind, as always.
‘He was very kind, as usual.
¡Es el hombre ma s amable que he visto!
be-SER.PST.3SG the man more kind that have.PRS.1SG see.PP
‘He was very kind, as usual. He is the kindest man I’ve ever met!
114 V. Escandell-Vidal / Journal of Pragmatics 128 (2018) 102e115

Despite the fact that (28)a uses estar, whereas (28)b uses ser, there is no difference at all with respect to the time span for
which the relationship between the subject and the predicate holds: in fact, as the continuations show, in both cases the
speaker is explicitly committed to a “permanent” attribution, regardless of the fact that the predication appears with ser or
with estar. Both examples allow the inference of temporal persistence, and the example with estar does not entail that some
boundary specifying the onset or the end of the interval over which the property holds. What is the difference then? Is this a
counterargument for my proposal?
The difference is, I think, a matter not of content, but of perspective: when ser is used, the property is presented as
classificatory, objective, not dependent on the evaluation of the speaker; with estar, in contrast, the property is presented as
anchored to the speaker, linked to a particular experience of hers or to the standards she has chosen. In any case, what (28)a
indicates is that a recent experience has confirmed a previous idea of the speaker, so no contrast (either temporal or locative)
is needed to explain why (28)a is felicitous with keeping both the evidential commitment and the indication of temporal
persistence.
This is fully compatible with my analysis because neither ser/estar nor the ILP/SLP distinction carry any systematic re-
striction about the (non-)permanent nature of the predication. The only requirement for estar is the existence of a spatio-
temporal anchoring, not a delimited span for the attributed property. The inclusion of ES in IS must not be a total inclusion,
but an overlap, as depicted in (18), with the property holding before and after the IS. Therefore, what is relativized to a
spatiotemporal situation here is not the property itself, but the grounds on which this property is attributed on the current
occasion.

5. Conclusions

In the previous sections I have analyzed some aspects of estar þ adjective constructions in Spanish. I have focused on those
cases giving rise to evidential entailments, i.e., committing the speaker to having direct perceptual evidence for the assertion
made. I have argued that indicating direct evidence is not part of the encoded meaning of estar, but rather an inference that
arises in very specific circumstances, namely when estar is combined with an ILP. In this combination, a semantic mismatch
occurs between the requirements of estar, which presupposes a situational anchoring, and those of an ILP, which crucially
lacks a matching feature. The existence of a situation of information acquisition works as a repair strategy to bind the
assertion and then satisfy the semantic requirements of the copula estar. The evidential interpretation is triggered by the
linguistic properties of the sentence and is hence obligatory and non-cancellable.
Though I have not discussed it directly, my proposal is based on a very specific and strong claim about the conditions
governing the distribution of the two copulas in Spanish. Central to my account is the idea that the ILP/SLP and the ser/estar
distinctions are not equivalent at all. This means that going with ser is not a criterial condition for ILPs, nor is it going with
estar a defining requirement for SLPs. Likewise, ser does not select ILPs and estar does not select SLPs e at least not in the
formal grammatical sense in which the word selection is used in the literature. The features of estar simply match with those
of SLPs, and the features of ILPs do not clash with those of ser. Mixed combinations are still possible, though they carry an
additional processing cost, which in turn is balanced with additional interpretive effects. Only if we assume that the ILP/SLP
distinction is autonomous with respect to the ser/estar distinction will we be able to obtain a predictive explanation about the
status of evidential effects.
There are many aspects still in need of discussion. Mismatches can involve more than two features, so the prediction is that
the more the features involved, the greater the interpretive effort. The role of tense and adverbs has to be established,
particularly with respect to the accessibility of the IS. More has to be said as well on issues such as dialectal variation (cf.
Delbecque, 2000; Escandell-Vidal and Leonetti, 2015). These are open topics for further research.
The analysis put forward here accounts for the occurrence of evidential interpretations, while maintaining a restrictive
stance about both the semantics of the two copulas and the aspectual features of the predicates. In this way, I have tried to
contribute to a better understanding of the meaning and distribution of the copulas, and the articulation of the semantics/
pragmatics interface.

Acknowledgements

The research underlying this paper has been funded by the Spanish Ministerio de Economia y Competitividad through
grants to the projects “Semantica procedimental y contenido explícito III” (SPYCE III; FFI2012-31785) and “The Semantics/
Pragmatics Interface and the Resolution of Interpretive Mismatches” (SPIRIM; FFI2015-63497). I am very grateful to the
audiences of various workshops where some preliminary ideas have been presented. Thanks to Aoife Ahern for checking my
English. Needless to say, any remaining shortcomings are my own.

References

Aikhenvald, A.Y., 2004. Evidentiality. OUP, Oxford.

Aikhenvald, A.Y., 2014. The grammar of knowledge in typological perspective. In: Aikhenvald, A.Y., Dixon, R. (Eds.), The Grammar of Knowledge. OUP,
Oxford, pp. 1e51.
Arche, M., 2006. Individuals in Time. John Benjamins, Amsterdam.
 V. Escandell-Vidal / Journal of Pragmatics 128 (2018) 102e115 115

Batllori, M., 2006. Sincronia i diacronia de ser i estar, Sadurní Martí (coord.). In: Actes del 13e  Col$loqui Internacional de Llengua i Literatura Catalanes, vol. 2.
Publicacions de l’Abadia de Montserrat, Barcelona, pp. 43e54.
Brucart, J.M., 2012. Copular alternation in Spanish and Catalan attributive sentences. Revista de Estudos Linguisticos da Universidade do Porto 7, 9e43.
Bylinina, L., 2014. The Grammar of Standards (PhD). Univ. Utrecht.
Bylinina, L., 2017. Judge-dependence in degree constructions. J. Semant. 34, 291e331.
Camacho, J., 2012. Ser and estar: the individual/stage level distinction and aspectual predication. In: Hualde, J.I., Olarrea, A., O’rourke, E. (Eds.), The
Handbook of Spanish Linguistics. Blackwell, Oxford, pp. 453e476.
Carlson, G., 1977. Reference to Kinds in English (PhD). UMass, Amherst.
Chierchia, G., 1995. Individual-level predicates as inherent generics. In: Carlson, G., Pelletier, F.J. (Eds.), The Generic Book. University of Chicago Press,
Chicago, pp. 176e224.
Davis, Ch., Potts, Ch., Speas, M., 2007. The pragmatic values of evidential sentences. In: Friedman, T., Gibson, M. (Eds.), SALT XVII. Cornell University, Ithaca,
pp. 71e88.
De Brabanter, P., Dendale, P. (Eds.), 2008. Commitment. John Benjamins, Amsterdam.
De Haan, F., 2005. Semantic distinctions of evidentiality. In: Haspelmath, M., et al. (Eds.), The World Atlas of Language Structures. OUP, Oxford.
De Haan, F., 2005. Coding of evidentiality. In: Haspelmath, M., et al. (Eds.), The World Atlas of Language Structures. OUP, Oxford.
De Haan, F., 2005. Encoding speaker perspective: evidentials. In: Frajzyngier, Z., Rood, D. (Eds.), Linguistic Diversity and Language Theories. John Benjamins,
Amsterdam.
de Swart, H., 1998. Aspect shift and coercion. Nat. Lang. Linguist. Theory 16, 347e385.
de Swart, H., 2003. In: Francis, E., Michaelis, L. (Eds.), Coercion in a Crosslinguistic Theory of Aspect, pp. 231e258.
de Swart, H., 2011. Mismatches and coercion. In: Maienborn, C., et al. (Eds.), Semantics. DeGruyter, Berlin, pp. 574e596.
Delbecque, N., 1997. The Spanish copulas ser and estar. In: Vespoor, M., et al. (Eds.), Lexicon and Grammar in Cognitive Linguistics. John Benjamins,
Amsterdam, pp. 247e270.
Delbecque, N., 2000. Las co  pulas ser y estar: categorizacio  n vs. deixis. In: Maldonado, R. (Ed.), Estudios cognoscitivos del espan ~ ol. Revista Espan ~ ola de
Lingüística Aplicada, pp. 239e280.
Diesing, M., 1992. Indefinites. MIT Press, Cambridge.
Escandell-Vidal, V., 2014. Evidential futures. The case of Spanish. In: De Brabanter, P., et al. (Eds.), Future Times, Future Tenses. OUP, Oxford, pp. 219e246.
Escandell-Vidal, V., Leonetti, M., 2002. Coercion and the stage/individual distinction. In: Gutie rrez-Rexach, J. (Ed.), From Words to Discourse. Semantics and
Pragmatics of Spanish. Elsevier, Amsterdam, pp. 159e179.
Escandell-Vidal, V., Leonetti, M., 2011. On the rigidity of procedural meaning. In: Escandell-Vidal, V., Leonetti, M., Ahern, A. (Eds.), Procedural Meaning:
Problems and Perspectives. Emerald, Bingley, pp. 81e102.
Escandell-Vidal, V., Leonetti, M., 2015. Estar joven a los dos lados del Atl antico. In: Gonz alez, S
ainz, et al. (Eds.), Geome trica explosio n. Ca’ Foscari, Venice, pp.
65e78.
bregas, A., 2012. A guide to IL and SL in Spanish: properties, problems and proposals. Borealis 1 (2), 1e71.
Fa
Ferna ndez Leborans, M.J., 1999. La predicacio  n. Las oraciones copulativas. In: Bosque, I., Demonte, V. (Eds.), Gram atica descriptiva de la lengua espan ~ ola, vol.
2. Espasa, Madrid, pp. 2357e2460.
Francis, E., Michaelis, L. (Eds.), 2003. Mismatch: Form-Function Incongruity and the Architecture of Grammar. CSLI, Stanford.
Garrett, E.J., 2001. Evidentiality and Assertion in Tibetan (PhD dissertation). UCLA.
Gili Gaya, S., 1961. Curso superior de sintaxis espan ~ ola. Biblograf, Barcelona.
Higginbotham, J., 2009. Some preliminary distinctions. In: Stainton, R.J., Viger, C. (Eds.), Compositionality, Context and Semantic Values. Essays in Honour of
Ernie Lepore. Springer, Berlin, pp. 221e235.
Kalsang, K., Garfield, J., Speas, M., Devilliers, J., 2013. Direct evidentials, case, tense and aspect in Tibetan: evidence for a general theory of the semantics of
evidentials. Nat. Lang. Linguist. Theory 31 (2), 517e561.
Katsika, A., Braze, D., Deo, A., Pin ~ ango, M.M., 2012. Complement coercion: distinguishing between type-shifting and pragmatic inferencing. Ment. Lex. 7 (1),
58e76.
Ko€ lbel, M., 2004. Faultless disagreement. Proc. Aristot. Soc. New Ser. 104, 53e73.
Kratzer, A., 1989. An investigation into lumps of thought. Linguist. Philos. 12, 607e653.
Kratzer, A., 1995. Stage-level and individual-level predicates. In: Carlson, G., Pelletier, F.J. (Eds.), The Generic Book. University of Chicago Press, Chicago, pp.
125e176.
Lasersohn, P., 2005. Context dependence, disagreement, and predicates of personal taste. Linguist. Philos. 28, 643e686.
Lasersohn, P., 2009. Relative truth, speaker commitment, and control of implicit arguments’. Synthese 166, 359e374.
Leonetti, M., 1994. Ser y estar: estado de la cuestio n. Barataria 1, 182e205.
Maienborn, C., 2005. A discourse-based account of Spanish ser/estar. Linguistics 43 (1), 155e180.
Marín, R., 2010. Spanish adjectives within bounds. In: Cabredo Hofherr, P., Matushansky, O. (Eds.), Formal Analyses in Syntax and Semantics. John Ben-
jamins, Amsterdam, pp. 307e332.
Marín, R., 2015. Ser y estar. In: Gutie rrez-Rexach, J. (Ed.), Enciclopedia de Lingüística Hisp anica. Routledge, London.
Mcnally, L., Stojanovic I., 2014. Aesthetic adjectives. In: Young, J. (Ed.), The Semantics of Aesthetic Judgment. OUP, Oxford.
Milsark, G., 1974. Existential Sentences in English (PhD). MIT.
Moltmann, F., 2010. Relative truth and the first person. Philos. Stud. 150 (2), 187e220.
Moravcsik, E., 2010. Conflict resolution in syntactic theory. Stud. Lang. 34 (3), 636e669.
Ninan, D., 2014. Taste predicates and the acquaintance inference. Proc. Semant. Linguist. Theory (SALT) 24, 290e309.
Pearson, H., 2013. A judge-free semantics for predicates of personal taste. J. Semant. 30 (1), 103e154.
rez Jime
Pe nez, I., Leonetti, M., Gumiel-Molina, S. (Eds.), 2015. New Perspectives on the Study of Ser and Estar. John Benjamins, Amsterdam.
Pylkka €nen, L., 2008. Mismatching meanings in brain and behavior. Lang. Linguist. Compass 2. https://doi.org/10.1111/j.1749-818x.2008.00073.x.
Querido, A., 1976. The semantics of copulative constructions in Portuguese. In: Luj an, M., Hensey, F. (Eds.), Current Studies in Romance Linguistics.
Georgetown University Press, Washington.
RAE, 2009. [Real Academia Espanola y Asociacion de Academias de la Lengua Espanola] Nueva gram
~  ~ atica de la lengua espan ~ ola. Espasa, Madrid.
Roby, D., 2007. Aspect and the Categorisation of States. The Case of Ser and Estar in Spanish (PhD). University of Texas at Austin.
Silvagni, F., 2017. Entre estados y eventos. Un estudio del aspecto interno en espan ~ ol (PhD dissertation). Universitat auto noma de barcelona.
Speas, M., 2010. Evidentials as generalized functional heads. In: Disciullo, A.M., Hill, V. (Eds.), Edges, Heads and Projections: Interface Properties. John
Benjamins, Amsterdam, pp. 127e150.
Stephenson, T., 2007. Judge dependence, epistemic modals, and predicates of personal taste. Linguist. Philos. 30, 487e525.
Stump, G., 1985. The Semantic Variability of Absolute Constructions. Reidel, Dordrecht.