The subject of a stative object experiencer verb is an intensional Cause

. In the domain of experiencer verbs, little attention has been paid to the potential intensionality of object experiencer (OE) verbs (but see Cheung & Larson 2015). Relying on diagnostics identified in the literature (Dowty 1979; Schwarz 2020), I argue that the intensionality of OE verbs is correlated with their aspectual properties: they are intensional in their subject when stative but not when eventive. I then consider whether this finding can help us choose between competing theories of the argument structure of OE verbs (Arad 2000; Landau 2010). I will argue that the answer is negative.

In the second part of the paper, I consider the question whether this new finding can help us choose between competing theories of the argument structure of OE verbs.A widely held view is that the argument structure of stative OE verbs is different from that of eventive ones (Pylkkänen 2000;Reinhart 2002;Rothmayr 2009;Landau 2010;a.o.).Many authors assume that while eventive OE verbs realize a Cause and an Experiencer (following Pesetsky 1995), stative OE verbs realize a Subject Matter instead of a Cause, as outlined in (3).Specifically, Pylkkänen (2000), Reinhart (2002), and Landau (2010) all argue that the subject of stative OE verbs is a moved SM argument.I will refer to this as view 1.
(3) The argument structure of OE verbs under view 1 Eventive: [Cause, Experiencer] Stative: [Experiencer, SM] By contrast, there is an alternative view according to which the subject of OE verbs is invariably a Cause, and therefore an external argument, regardless of the aspectual distinction.On this view, all OE verbs are causatives (Arad 2000).I will refer to this as view 2.
(4) The argument structure of OE verbs under view 2 [Cause, Experiencer] The empirical finding that only stative OE verbs have an intensional subject seems to have a natural characterization under view 1.Given the well-known observation that the internal argument (i.e. the SM) of SE verbs is intensional, such a proposal would lead one to expect (correctly) that OE verbs pattern with SE verbs as regards intensionality only on their stative reading.Nevertheless, I will demonstrate that view 2 is equally able to account for the facts, once we take a more careful look at the referent of the Cause argument on the two readings.In particular, I will propose that while the subject of OE verbs is invariably a Cause, the Cause of stative OE verbs refers to a cause that is created upon Experiencer-internal cognitive activities (i.e.mind-internal cause), hence intensional.Meanwhile the Cause of eventive OE verbs refers to a cause that is external to such cognitive activities (i.e.mind-external cause), hence non-intensional.The organization of this paper is as follows.Section 2 will outline diagnostics for intensionality introduced in the literature.I will briefly review the conclusion in the literature that SE verbs are intensional in their internal argument (i.e. the SM).In section 3, using the same diagnostics, I will demonstrate that stative OE verbs exhibit the intensional properties in their subject, while their eventive counterparts show none.In section 4, I will explore how this finding can be reconciled with the competing views of the argument structure of OE verbs outlined above.While it undoubtedly presents a natural fit for view 1, I will argue that it can equally well be accounted for under view 2, according to which the subject of such verbs is invariably a Cause, therefore an external argument, regardless of the aspectual distinction.Section 5 concludes.

THREE DIAGNOSTICS FOR INTENSIONALITY.
A predicate is taken to be intensional if a sentence containing it exhibits at least one of the following characteristics: (i) substituting an expression with a coreferential term need not preserve truth-value; (ii) an expression which has no extension in the actual world (i.e.non-existential expression) need not induce falsity of the sentence; (iii) there appears a so-called non-specific reading (in addition to a specific reading) with an indefinite expression (Forbes 2000(Forbes , 2006;;Pearson 2015;Cheung & Larson 2015;den Dikken et al 2018;Schwarz 2020;a.o.).Let me expound on each of the diagnostics.

(i)
Substitution failure with a co-referential term Consider the examples in (5).Suppose that John does not know that the true identity of Batman is Bruce Wayne.Given this state of affairs, the two sentences in (5) can bear a different truth value: (5a) can be true while (5b) is false.
(5) a. John believes that Batman drives a Dodge Charger.b.John believes that Bruce Wayne drives a Dodge Charger.
Consider the examples in (6), in which the verb look for takes a simple noun phrase object.Assuming the same scenario as before where John is unaware of the fact that Batman is Bruce Wayne, (6a) can be true without (6b) being necessarily true.
(6) a. John is looking for Batman.b.John is looking for Bruce Wayne.
Non-intensional predicates are transparent to substitution of a co-referring term.This is illustrated in (7).The truth value of the two sentences does not vary: if (7a) is true, then (7b) should be true as well, regardless of John's knowledge state about the identity of Batman.

(ii) Truth with a non-existential term
The second property of intensional predicates is that they can take an argument that lacks an extension in the actual world without jeopardizing the truth of the sentence.Consider the examples in (8)-(9).Although there are no such things as werewolves, mermaids, unicorns, or vampires in the actual world, these sentences can be judged true as long as Mary believes in their existence.
(8) a. Mary thinks that a werewolf stole her dog's food.b. Mary believes that she invited a mermaid to her party.
(9) a. Mary wants a unicorn.b. Mary is hunting for a vampire.Now compare them to sentences with non-intensional predicates as in (10).The lack of extensions for werewolves and vampires in the real world results in the falsity of these examples.In other words, non-intensional predicates come with an existential commitment for their object.
(10) a. Mary sued a werewolf for theft.b. Mary hired a vampire as her gardener.
(iii) Availability of a non-specific reading with an indefinite object.
The final property of intensional predicates is that they allow a non-specific reading of an indefinite object. 1 Consider the examples in (11).(11a) can be interpreted as saying that Bill has a particular camera in mind, and he wants it.But it can also convey a situation in which there is no specific camera that he wants.Likewise, (11b) is ambiguous between the specific and non-specific readings of a member of staff that Bill is looking for.
(11) a. Bill wants a camera.b. Bill is looking for a member of staff.
The non-specific readings of these examples can be detected by looking at their compatibility with the phrase but no particular one (Zimmermann 2001).It can be added to the examples in (11) without contradiction, as shown in ( 12): (12) a. Bill wants a camera, but no particular one.b. Bill is looking for a member of staff, but no particular one.
Compare these examples to sentences with non-intensional predicates as in (13).Non-specific readings are unavailable in these examples, which is supported by their incompatibility with the phrase but no particular one.
(13) a. Bill sold a camera (, #but no particular one).b. Bill met a member of staff (, #but no particular one).).I will briefly go through applications of the diagnostics for intensionality to SE verbs, and reach the same conclusion reached in the literature that these verbs are intensional in their SM argument.

(i) Substitution failure with a co-referential term
1979 as well).Others however consider a richer environment, including where the object occurs with determiners like every, most, and even the definite the (Forbes 2006(Forbes , 2020)).Forbes (2020) claims that (ia) can express a situation in which Guercino is looking for every dog on Aldrovandi's estate but there are no particular dogs he is looking for (Forbes 2020).Likewise, the non-specific reading of the definite object in (ib) (i.e. the exit) is not hard to notice when imagining a situation where John is driving around an unfamiliar airport rental car lot, looking for the exit.As Forbes points out, in this case the driver has no particular exit in mind.
Guercino is looking for every dog on Aldrovandi's estate (, but no particular dogs).b.John is looking for the exit (but there is no specific exit that is being sought).
However, as Forbes also makes it clear, these extra determiners (other than a(n)) do not always work for this particular diagnostic.Intensional predicate draw, for example, admits a non-specific reading but only when its object is introduced with a(n), not with those other determiners suggested by him.Consider the contrast between (iia) and (iib).Forbes explains that the fact that (iia) has a non-specific reading is corroborated by the wall label for Guercino's The Aldrovandi Dog (ca.1625) in the Norton Simon Museum.The label says: 'this must be the portrait of a specific dog'.This suggests that there exists an alternative interpretation of the portrait, namely that it is not of a particular dog and that the painter just made one up (Forbes 2020).
Guercino drew a dog (, but no particular one).b.Guercino drew every/most/the dog(s) (, #but no particular one).
Given the somewhat more delicate situation with determiners other than a(n), I will make use of the indefinite singular a(n) for the non-specificity test throughout this paper, so as to avoid controversy.
When substituting an object of a SE verb with a coreferential term, the truth-value of the original sentence does not carry over to the sentence with the substituted object.Assuming the same scenario as before where John does not know that the identity of Batman is Bruce Wayne, (14a) can be true while (14b) is false.

(iii) Availability of a non-specific reading with an indefinite object
Finally, indefinite objects of SE verbs admit a non-specific reading, alongside a specific reading.Consider ( 16).The availability of non-specific readings in these examples is supported by the compatibility with the phrase but no particular one.

Intensionality of OE verbs.
As mentioned before, little attention has been paid to the role of intensionality in OE verbs, although it was briefly discussed by Cheung & Larson (2015: p135, fn. 12).These authors hold that OE verbs are not intensional in their subject, although they admit that this conclusion is tentative.The argument they offer for this conclusion is that subjects of OE verbs do not pass two of the three familiar diagnostics: they do not display substitution failure with co-referential terms, and non-specific readings are not found either.While they concede that OE verbs allow non-referring terms in the subject position, they diminish this evidence by casting doubt on the decisiveness of truth with non-existential terms as a test for intensionality, referring to Bennett (1974) and Dowty (1979).
The focus of Cheung & Larson's (2015) discussion of intensionality is to demonstrate that SE verbs are intensional in their object but that OE verbs are not, and they note that aspectuality is irrelevant to the intensionality of the object of OE verbs (Cheung & Larson 2015: p133-134, fn. 9).Remarkably, the authors do not pay any attention to the distinction between eventive and stative readings in their discussion of the potential intensionality of the subject of OE verbs.As I will show in this section, this was an unfortunate oversight, since a full-fledged investigation will reveal that the aspectual distinction in OE verbs is not orthogonal to the question of intensionality, and in particular that the subject of an OE verb is intensional if and only if the verb is stative.

(i)
Substitution failure with a co-referential term For my investigation I will use the OE verb frighten, which is ambiguous between an eventive (see ( 17)) and a stative reading (see ( 18)).It can be observed that there is a contrast between these pairs with respect to substitution of the subject.When the verb is eventive, its subject is transparent to substitution of co-referring terms.If (17a) is true, then (17b) must be true as well, regardless of John's knowledge state about the identity of Batman.By contrast, we do get substitution failure with the subject of the stative counterpart.(18a) can be true without (18b) being necessarily true, provided that John does not know that Batman is Bruce Wayne.Batman may have properties that cause John to feel fear whenever he thinks about him, but Bruce Wayne (who might be living next door to John) need not trigger the same reaction in John at all.

(ii)
Truth with a non-existential term I now turn to the second mark of intensionality, namely truth with a non-existential term.As was the case for substitutivity, the acceptability of non-existential subjects with OE verbs is sensitive to their aspectuality.Only stative OE verbs allow for non-existing subjects.Subjects of the eventive counterparts must have an extension in the real world just like arguments of non-intensional predicates.The contrast between (19a) and (19b) illustrates this point.(19a), where the adverb accidentally forces an eventive reading of frighten, is judged false since there are no such things as vampires or Nessie in the actual world, thus they cannot do anything (whether accidentally or not) that makes Mary feel frightened.By contrast, (19b) can be regarded as true on the reading in which vampires or Nessie have properties such that whenever Mary thinks about one or the other she experiences fear.

(iii) Availability of a non-specific reading with an indefinite object
The indefinite subjects of the eventive OE verbs as in (20) do not admit non-specific readings.
The only sensible reading of (20a), for instance, is that there is a specific guy who frightened John.The lack of non-specific readings in these examples is supported by their incompatibility with the addition of but no particular one.
Things are rather less clear for the subject of a stative OE verb.Although the indefinite subjects in ( 21) are most naturally interpreted as specific, a non-specific reading is not completely ruled out.Note that several factors disfavor the word order in the examples in (21) when they are uttered in an out of the blue context (cf.Silverstein's 1976 Animacy Hierarchy and Titov's 2012 Argument Prominence Hierarchy).However, they are much more acceptable in answer to a question like What sorts of things frighten/annoy John?, where the answers benefit from having a word order that parallels that of the question.Once these factors are controlled for, the indefinite subjects in (21) seem to allow a non-specific reading, at least marginally.
b.A member of Parliament annoys John (, ?but no particular one).
Examples of this type improve further when their subject contains a modifier, as shown in ( 22), where the non-specific reading of the subjects is accessed with relative ease.( 22) a.A hairy spider/an early morning lecture frightens John (, but no particular one).b.A badly brewed coffee annoys John (, but no particular one).
In summary, the three diagnostic tests on which I have relied indicate a clear split between the eventive and stative uses of OE verbs: eventive uses of OE verbs do not exhibit any properties of intensionality in their subject, whereas stative uses exhibit all three.I acknowledge that there may be speakers for whom the availability of the non-specific reading of the indefinite subject with the stative use is less than straightforward.However, it should be noted that passing all three intensionality diagnostics is not usually taken to be a necessary condition for a predicate to be considered intensional (Forbes 2006(Forbes , 2020)).This being so, I conclude that the split demonstrated here presents a pretty clear picture.
In the next section, I explore the question of whether this new finding can help us choose between competing theories of the argument structure of OE verbs.

What do our findings tell us about the argument structure of OE verbs?
As mentioned earlier, there are two competing theories of the argument structure of OE verbs.It has been widely assumed that the subject of an eventive OE verb is a Cause, while the subject of a stative OE verb is a moved SM argument (i.e.view 1, see ( 23)) (Pylkkänen 2000;Reinhart 2002;Landau 2010).The alternative view (i.e.view 2, see ( 24)), on the other hand, contends that the subject of OE verbs is invariably a Cause, therefore external, regardless of the aspectual distinction (Arad 2000).Do the findings of section 3 provide a strong argument for one or the other of these proposals?Undoubtedly, the finding that only stative OE verbs have an intensional subject is a very good fit with view 1, according to which the subject of a stative OE verb is a raised SM argument.Given the well-established conclusion that the internal argument (i.e.SM) of SE verbs is intensional, view 1 predicts that the subject of stative OE verbs should pattern with the object of SE verbs as regards intensionality.Given that on view 1 the eventive variant lacks the SM argument, it is also hardly surprising that this variant is not intensional in its subject.
We should, however, also explore the compatibility of our findings with view 2. It is important that we do so because, despite the fact that view 1 commands considerable support in the literature, it is not without problems.I will briefly discuss some arguments against it, all of which center on the rather weak support for the raising analysis for stative OE verbs.
First, an argument taken to support the raising analysis of stative OE verbs is tied to a phenomenon called backward anaphora illustrated in (25b), where the anaphor contained in the surface subject DP can be bound by the Experiencer in the object position (e.g.Belletti & Rizzi 1988).Reinhart (2002) claims that backward anaphora is only enabled when the subject is a moved SM, because in that case the anaphoric pronoun is bound by the quantified object in its underlying position.In (25a), backward anaphora is illicit, because the subject is a Cause, which merges externally.
b. [His1 health]2 worried [every patient]1 t2.(Reinhart 2002: 267) This approach, which relies on A-movement of the bound anaphora (Belletti & Rizzi 1988;Grimshaw 1990;Reinhart 2002), predicts that this phenomenon will not be found in analytical causatives.This is because in an analytical causative, an anaphora realized in the higher position (in the subject) is not an argument of the experiencer verb, but of the causative verb make or cause.Therefore, no proper c-command relation can hold between the antecedent (the Experiencer) and the anaphor, and backward anaphora should not obtain.However, this is contrary to fact, as pointed out by many authors (e.g.Pesetsky 1995; Bouchard 1995).Consider the examples in ( 26) -( 27).In the (b) examples, the bound pronoun is embedded in the subject DP 'his declining health', which is an argument of made and cause, respectively.It is therefore projected higher than the Experiencer argument.Nevertheless, the pronoun contained in it can still be bound by the quantified antecedent in the lower position.The contrast between the (a) and ( b) examples clearly cannot be accounted for by the purely structural approach that involves Amovement of the bound anaphora.Second, since stative OE verbs are assumed to involve A-movement of the SM argument, one would expect it to be possible for an indefinite SM argument to remain in situ and associate with an expletive in subject position.This is not the case, howevernot even in a language like Dutch where expletives are generally possible in structures lacking an external argument.2 (28) a. * There interested John a topic about the war.b. * Er interesseerde Jan een onderwerp over de oorlog.(Dutch) there interested John a topic about the war Consider (29) for a third counterargument.Once again assuming the raising analysis, the embedded clause of ( 29b) is an in-situ SM with the subject filled by an expletive (Reinhart 2002: p270).This should make the embedded clause of (29b) equivalent to that of (29a) since then the that- Since the verbs in (30) are stative, it is reasonable to assume that they are derived from a stative causative entry.However, once this is acknowledged, it becomes a moot point that the subject of stative OE verbs should be analyzed as a raised SM argument.
I now return to the question whether the findings of section 3 are also compatible with the argument structure suggested by view 2. I will demonstrate that this view is equally able to account for the facts, once we take a more careful look at the referent of the Cause argument on the two readings.
In philosophical and psychological work on emotion, it is widely held that the coming about of an emotion involves multiple causal factors (Lazarus 1982(Lazarus , 1991;;Nussbaum 2001;Scherer 2001Scherer , 2004;;a.o.).In particular, there is a fairly general view that contends that a central component of an emotional episode (i.e. a cycle that covers the start and the end of an emotion) is a cognitive process on the part of the Experiencer (Arnold 1960;Schachter 1964;Solomon 1976;Nussbaum 1990Nussbaum , 2001;;Scherer 2001Scherer , 2004;;a.o.).In somewhat simplified terms, an emotion arises in the following way: there is a stimulus presented to and perceived by the Experiencer.The Experiencer evaluates what is perceived, and the output of such an evaluation process leads to an emotional state.
I refer to the external stimulus that the Experiencer perceives and which provides the input for the subsequent evaluation process as a mind-external cause.Meanwhile, the output of such a cognitive process is dubbed a mind-internal cause, which directly leads to the Experiencer being in an emotional state.The mind-internal cause can remain in the Experiencer's mind even when the external cause that gave rise to it ceases to exist.For instance, if John's behavior during breakfast worried Mary (all day), it does not mean that Mary has to perceive John's behavior for the amount of time her worrying emotion holds.Rather, what the sentence means is that whenever Mary thinks about John's behavior at the breakfast table this morning, she experiences worry.
I propose that the subject of all OE verbs is a Cause, but that its referential properties differ depending on the aspectuality of the OE verb.Specifically, I claim that the Cause of eventive OE verbs refers to an Experiencer's mind-external cause while the Cause of stative ones refers to an Experiencer's mind-internal cause.
With this distinction in place, the variation in the subjects of OE verbs regarding intensionality follows.The subject of stative OE verbs is intensional since it refers to a cause that is internal to the Experiencer's mind.Hence, its nature depends on the Experiencer's cognitive state.The lack of intensionality in the subject of eventive ones follows as it refers to a non-evaluated stimulus, whose content is independent of the Experiencer's cognitive state.The unified causative analysis of OE verbs (view 2) is possible, then, with variation in intensionality understood as variation in the referent of Cause argument.

Conclusion.
In this paper, I claimed that the intensionality of OE verbs in their subject correlates with their aspectual properties.Using the standard diagnostics for intensionality, I showed that the subject of an OE verb is intensional when the verb is stative, but not when the verb is eventive.While this finding is a good fit for the dominant view in the literature that the subject of a stative OE verb is a raised SM argument, I maintained that it is not at odds with a unified causative analysis of OE verbs, according to which all OE verbwhether eventive or stativehave a Cause as their external argument.The variation in the subjects of OE verbs with regard to intensionality can be understood in terms of variation in the referent of their Cause argument.Intensional properties shown in the subject of stative OE verbs are expected since the subject corresponds to the mind-internal cause, which is the output of the Experiencer's cognitive activity.The lack of intensionality in the subject of eventive OE verbs follows as it refers to the mindexternal cause.
ARGUMENT OF SE VERBS.It is well established in the literature that SE verbs are intensional in their internal argument (i.e. the SM argument;Larson et al 1997;Larson 2002;Forbes 2006Forbes , 2020;;Cheung & Larson 2015;den Dikken et al 2018; a.o. (14) a. John loves/disdains/fears Batman.b.John loves/disdains/fears Bruce Wayne.(ii) Truth with a non-existential term SE verbs can take an object that has no extension in the actual world without making the sentence false.Consider (15): (15) a. Mary loves a unicorn.b. Mary disdained a mermaid.c. Mary fears unobtainium.
in both examples are SMs.Unexpectedly on this assumption, the embedded clauses of (29a) and (29b) have different interpretations.(29b) presupposes the truth of the proposition expressed by the embedded clause, in other words, Mary did indeed say something stupid.(29a), by contrast, does not come with this presupposition.To illustrate this point, let us consider the following scenario: Mary arrived for a meeting very drunk.A few hours after the meeting, when she has sobered up somewhat, she cannot recall whether she said anything during the meeting and if she did what it was that she said.Suppose that in fact Mary was asleep throughout the meeting and so did not say a word.Only (29a), but not (29b) is felicitous in this context.The contrast between the two examples with respect to the scenario implies that only the that-clause in (29a) can be understood in relation to the Experiencer's knowledge state.This can be understood straightforwardly if the that-clause in (29a) is an SM, while the that-clause in (29b) is not.Instead, it is an (extraposed) Cause.