Exclamations as emotive assertions: more questions (than answers)

1 Trotzke & Giannakidou’s proposal in a nutshell

The purpose of Trotzke and Giannakidou’s target paper entitled “Exclamation, intensity, and emotive assertion” is to present a novel analysis of exclamations. The main idea put forth by Trotzke and Giannakidou (henceforth T&G) is summarized in the following sentence: exclamations are emotive assertions, a specific kind of assertion that “asserts emotion, is veridically committed to the truth of the sentence, and presupposes intensity” (p. 5). In a nutshell, in this contribution, T&G argue against the illocutionary force hypothesis, according to which exclamations have a distinctive illocutionary force, an idea that is generally encoded in semantic analyses as a force operator, responsible for conveying the emotive/exclamatory component in the semantics of exclamative clauses. Instead, T&G claim that this emotive component is expressed via a syntactically represented null emotive predicate. More specifically, they propose that exclamatives are semantically equivalent (i.e., identical in terms of what they assert and presuppose) to sentences consisting of an emotive predicate and an embedded clause, (1) (their examples (43) and (44)).

In their own words (p. 26):

They both assert that the speaker has the emotion of amazement towards the believed proposition ‘Eliud Kipchoge was extremely fast’, and presuppose that the speaker has this very belief.

The emotive predicate (be amazed/surprised) is central in this analysis. A second and related meaning component is subjective veridicality, which replaces factivity, one of the hallmarks of exclamatives since Elliott (1974) and Grimshaw (1979). Subjective veridicality is a presupposition of proposition-embedding emotive predicates, as argued for in Giannakidou and Mari (2021), according to which the subject of the emotion believes that the embedded proposition is true. In other words, this presupposition is weaker than factivity, which they call objective veridicality, characteristic of a predicate like know. While in proposals such as Zanuttini and Portner (2003) factivity was conceived of as a formal feature represented in the left periphery of a matrix clause, here subjective veridicality naturally follows from the lexical semantics of a null V-emotive predicate. To support this claim, novel data from Greek exclamatives are presented, whereby the emotive complementizer pu is shown to typically follow the wh-phrase in a wh-exclamative. Crucially, Greek has three complementizers, but pu is the one selected by V-emotive predicates when they select propositions.

A final – but differentiated – meaning component is that of intensity, which is triggered by the combination of counterexpectation (conveyed by the V-emotive predicate) and extreme degree (represented in the propositional content of structures including so and a gradable predicate, here named SO-degree: I am surprised / amazed that Kipchoge ran so fast.).

Syntactically, exclamative sentences are not instances of insubordinated constructions (truncated main clauses), but rather are embedded structures under an elided predicate. The data from German dass-exclamatives, which contrast with main clauses introduced by ob ‘whether’, argue in favor of this analysis. Dass ‘that’ being the default complementizer in proposition-embedding predicates also supports the claim that exclamations are nothing but covert embedded structures under a higher assertive predicate.

A natural corollary of this analysis is that the distinction between sentence exclamations and wh-exclamatives (Rett 2011) merely points at differences in the embedding structures; exclamatives have SO-degree as their descriptive content, whereas sentence exclamations are embedded propositions that do not describe extreme degrees. They both make emotive assertions, though. In relation to this, the expression of extreme degree, which has been another hallmark in the characterization of wh-exclamatives (Castroviejo 2006, 2007; Rett 2008, 2009, 2011; Zanuttini and Portner 2003), is viewed in this proposal as the trigger of intensity, but is not a critical component in exclamations.

2 Overall assessment

I would like to highlight three main merits of this proposal as well as point out related questions that may need further clarification.

First, beyond proposing that the emotive component in exclamations is an attitude rather than a speech act, T&G are explicit about how emotivity is unpacked by relating to the previous literature on emotive predicates. Moreover, in doing so, subjective veridicality overcomes factivity, and with it, the idea that exclamatives are factive is now qualified. By establishing this close relationship between overt emotive predicates and exclamations, it is possible to make precise predictions regarding what is presupposed in exclamations and how these presuppositions are triggered.

Something that may need more elaboration would be how these embedding predicates compose with a wh-complement. In fact, in the paper there are no Greek and German examples of embedded wh-clauses under V-emotive predicates. Instead, the Greek and German data revolve around embedded that-complements, where the V-emotive takes a proposition as an argument. In relation to this, as far as I can tell, by positing SO-degree as the proposition denoted by exclamatives, the authors opt to not treat wh-exclamatives as semantically interrogative. In other words, the wh component is not given a semantic correlate or even a particular relevance for exclamative building. The same strategy was followed in Castroviejo (2006) for Catalan wh-exclamatives, where the burden of the degree component also fell on tan ‘so’. This could make sense given the existence of exclamative-only wh-words and phrases (Greek ti + ADJ being a case in point). However, in adopting this strategy, non-degree wh-clauses embedded under V-emotive predicates (e.g., It surprised Sonia who won the election.) have to receive a different compositional derivation from degree wh-clauses. Whether preserving the commonalities of the wh-family is a desirable goal can be a matter of debate.^[1] Another related question that this embedding analysis raises is how T&G would interpret wh-exclamatives embedded under cognitive factives (e.g., Bill knows how extremely slow this car is.). If how extremely were to be analyzed as the SO-degree component, we could conclude that such structures denote extreme degree but not counterexpectation, possibly an accurate depiction of the facts and a desirable outcome of T&G’s proposal. Certainly, this phenomenon is not T&G’s concern, since their goal is to discuss exclamations rather than exclamatives, but this could be seen as a relevant question for further research.

In relation to this last point, another neat aspect in this proposal is that it disentangles extreme degree from emotion by positing intensity as a third – but interconnected – meaning component. In this view, the intensity of an exclamation relies on how high the degree of some property is with respect to the speaker’s expectation. A consequence of this claim is that sentence exclamations that are devoid of this degree component (2), being embeddings under a V-emotive predicate (3), should in principle be “neutral in intensity” (p. 35).

At the same time, it is said that intensity in degree-less declarative exclamations involves intensity due to the counterexpectation of the null V-emotive verb (p. 36). I agree that, depending on the expectations of p being the case, when the speaker learns that p is true, intensity may vary; in this sense, it is present even in the absence of SO-degree. Hence, It’s raining in Barcelona! uttered by the same speaker in (2), may be less intense. It seems that this proposal is flexible enough to derive intensity through two different strategies, but the non-degree strategy is left more unspecified and we lack a clear parallelism between overt and null embeddings. While I share the intuition that (3) can be neutral in intensity, it seems that neutrality in intensity in the case of (2) would yield a non-emotive assertion. Could it be that degree-less sentence exclamations necessarily express some degree of intensity?

Third, another valuable contribution of T&G is their compilation of empirical data from recent literature, which challenges the empirical tests that had formed the basis of earlier theoretical analyses of exclamatives. Specifically, the following:

1. The descriptive content of a wh-exclamative cannot be directly refuted (Rett 2008, 2011).

2. Wh-exclamatives cannot be used to answer an information-seeking question (Grimshaw 1979; Zanuttini and Portner 2003).

Rejecting these two claims through experimental work, T&G challenge earlier accounts and provide an opportunity to gain new insights into exclamations. At the same time, though, they bring up an interesting reflection on how to interpret the results of diagnostics such as refutation and answerhood. Further research could explore whether these tests identify at-issue meaning, a commitment by the speaker, or a foregrounded proposition introduced by the exclamative clause.

Summing up, this is a bold proposal with strong theoretical and empirical foundations. Its explicit analysis invites readers to critically examine the predictions it generates. As such, this contribution provides the necessary tools to challenge established assumptions, refine previously accepted premises, and foster new debates that will undoubtedly enhance our understanding of exclamatives and exclamations.

Building on these previous points, the remainder of this commentary speculates on two particularly inspiring aspects of the proposal: the status of the various layers of meaning in emotive assertions (with respect to refutation and answerhood) and the capabilities of V-emotive predicates to convey emotions in exclamations.

3 Layers of meaning

I would like to start by reflecting on the predictions made by T&G’s proposal. To do so, I will informally break down the layers of meaning in their analysis on the basis of example (1) (leaving aside the presupposition of intensity).

Notice that this analysis is meant to cover the interpretation of a variety of exclamatory types from a syntactic perspective; as said, unlike previous accounts, wh-clauses, in the case of wh-exclamatives, do not receive a question analysis (Zanuttini and Portner 2003) or a degree property analysis (Rett 2011). The exclamative clause denotes a proposition, which is embedded under an emotive predicate. On another note, recall that p ₄ in (4) is not treated as a factive presupposition. Rather, p ₄ is not analyzed independently from p ₁; they are both affirmed and also asserted. This might explain why most of the assertivity tests (refutation and answerhood) are strictly directed towards p ₄. The so-called “descriptive content” of the exclamation is identified as p ₄.

In the following two subsections, I discuss refutation and answerhood of the descriptive content of exclamations as assertivity diagnostics.

4 Emotions and contexts that license exclamations

Teasing apart the emotive component from the extreme degree component in exclamatives has been an important challenge when trying to account for the semantics of exclamatives. As noted earlier in this commentary, one of the strengths of T&G’s proposal is to provide a precise decomoposition of the emotive meaning underlying in exclamatives by appealing to subjective veridicality and to the negative nonveridical presupposition of contrary emotives (Giannakidou and Mari 2021).

Drawing from T&G’s analysis of the emotive component, I propose two potential avenues for development. The first one concerns a set of contexts that license exclamations, but which have not raised a lot of attention. Leaving aside for now whether they constitute a clear example of exclamation, Michaelis (2001) discusses cases such as (12), in which the speaker invites the addressee to “share the speaker’s judgment” (p. 1039). The speaker was surprised when learning the news, and she thinks the addressee would also be, if he knew the value of who in the embedded clause (e.g., Tessa, the shiest member of the family).

I find this sort of examples especially appealing, because they are also present in matrix contexts in languages such as Spanish and Catalan, where they are frequently realized as DPs – as concealed exclamations (Grimshaw 1979). Consider for instance (13a) (Spanish). This sentence could be uttered by a worried and upset grandparent telling their friend about their forgetful grandchild, thus projecting surprise or shock in the addressee. Pending an empirical study, (13a) and (13b) appear to be interchangeable, except that (13b) contains an overt embedding predicate.

This set of predicates also includes you have no idea or you can’t imagine. Interestingly, they do not typically take that-complements, but they usually take wh-clauses, and also DPs in Romance languages.

If we now turn to the wh-version of (13a), i.e., (14a) (Spanish), it would be rather uttered in a situation more aligned with the depiction of contrary emotives in T&G’s proposal. Specifically, in (14a), the grandparent may have recently realized that their grandchild has not called in a long time, or they suddenly bring this idea to mind. They are not making any hypotheses about the addressee’s doxastic state or projecting the addressee’s future surprise. Hence, (14a) is not equivalent to (14b).

I do not want to convey that concealed exclamations are always uttered in situations where the speaker projects future or potential emotions in the addressee. These sentences can also make sense in contexts which are equivalent to an embedding under I can’t believe, in which the speaker is expressing that p ₄ is incompatible with what she thought possible. Intuitively, (15a), (15b), and (15c) (Catalan) can be felicitous in the same context.

Taken together – and, again, pending a deeper empirical study – this set of data may suggest that not all exclamatory types are felicitous in the same contexts (see also Castroviejo 2008), but also that exclamations may convey different messages associated with emotions; in an account where exclamations are embeddings under emotive predicates, this raises the issue of whether a unique semantic representation for the V-emotive predicate will suffice.

Related to this latter point, I would like to raise a second question, one that concerns the paraphrase of the V-emotive; more specifically, the differences between I am amazed and It’s amazing. T&G opt for the former as a paraphrase for the embedding verb in the asserted content of an exclamation (remember, p ₁ in (4)). One important difference between the two could be that the speaker maps onto the subject in the former, but onto the judge in the latter (see, e.g., Lasersohn 2005; Stephenson 2007; a.o.). Consider (16) briefly.

Observe that refuting (16a) may amount to negating the speaker’s amazement (as said above, something that challenges the sincerity condition of the speaker, highly marginal). However, refuting (16b) could be interpreted as an instance of faultless disagreement (It’s not amazing to me, see Kölbel 2004; a.o.). At first sight, if (16b) were a more promising approach than (16a), this small difference could be easily accommodated within the framework of the adopted theory of emotive predicates. I can identify two reasons to support this alternative solution. For one, it may straightforwardly explain why “exclamatives are always interpreted as conveying the surprise of the speaker (and not that of any other subject)” (p. 28). Additionally, it could fit in with the utterance of exclamations by speakers who are expressing their current state of amazement (illustrated in (16a) or by an embedding under I can’t believe), but also with exclamations that convey that some state of affairs deserves to be considered amazing (illustrated in (16b) or by an embedding under You wouldn’t believe). For instance, (17) (Catalan) could be uttered to express that the speaker is amazed or surprised by how much she loves the addressee, but it will more often be uttered when the speaker characterizes her loving towards the addressee as amazing, which amounts to saying she loves him a lot.

Consider now the example in (18) (Catalan). It illustrates another context where the speaker is usually not surprised or amazed (at the utterance time) by what a great time she and her friends have had. They are more frequently uttered to tell the addressee that they have had an amazing time.

Interestingly, because of the first person depiction of internal states (loving, having fun), in both (17) and (18), the speaker is supposed to be more opinionated than the addressee. Unlike (14a) above, these wh-exclamatives are typically uttered in the kind of contexts where the speaker wants to convince the addressee about SO-degree, so the extreme degree component seems more foregrounded than the emotive component.

To summarize this second point: it may be interesting to investigate whether the duality I am amazed versus It’s amazing underlies a more adequate characterization of exclamations, one that best captures the variety of contexts that license them with respect to the doxastic states of speaker and addressee at the moment of utterance of an exclamation.

5 Conclusions

Fortunately or not, exclamations and exclamatives remain far from fully understood. T&G take another significant step toward this ambitious objective. In this commentary, I have sought to honor their contribution by posing new questions that may help us advance toward that understanding.