Empirical and theoretical arguments against equating exclamatives and emotive assertions

1 Introduction

In their target article entitled “Exclamation, intensity, and emotive assertion,” Trotzke and Giannakidou (in the following T&G) argue that exclamatives are emotive assertions. Correspondingly, examples such as (1a) are considered equivalent to subordinate clauses embedded under an emotive matrix predicate (1b) and also to declarative exclamations (1c) [See T&G (Sect. 4) for examples analogous to (1).].

Intuitively, all structures convey that the speaker is surprised at Ada’s speed. The proposal captures this intuition within a uniform approach and thus avoids the assumption that exclamatives form a separate sentence type with its own illocutionary force. In the following, I will nevertheless take issue with the given equation on both empirical and theoretical grounds. Focusing on German, I will first discuss the external combinatorics of exclamatives (Sections 2 and 3), then turn to their internal make-up (Section 4), and finally dwell on the interpretation of the relevant attitude (Section 5).

2 Pairing of questions and answers

T&G (p. 38f. and Sect. 2.3) argue that exclamatives pattern with emotive assertions in question–answer pairs such as (2) [= T&G (68)]. This is tied to the more general assumption that exclamatives can answer information-seeking questions if the discourse conditions are appropriate, as in (3) [= T&G (20), slightly adapted] with a question that supports wide focus.

The authors use such examples to show that exclamatives are regular emotive assertions. I disagree with this conclusion for the following reasons.

Let us proceed from the alternative assumption that exclamatives convey demonstrations of attitudes toward a state of affairs instead of corresponding assertions.^[1] From this perspective, (3b) and (3c) are possible answers to A’s question in (3) because the demonstration of surprise at Eliud’s speed is an indirect source for information about the quality of his performance. Notably, T&G also need to access the relevant information indirectly. As they assume that exclamatives are headed by silent emotive predicates, the quality of Eliud’s performance is merely an embedded piece of information. That is, examples such as (3) do not distinguish between the demonstration of an attitude and its assertion.

Similarly, (2b) may be possible because the demonstration of an attitude is a potential reaction to the question of which attitude an addressee has. That is, by demonstrating surprise, the addressee can let the questioner know that he is surprised. Moreover, I doubt the given judgment. To my ears, answering by demonstration is a marked strategy that addresses the given question in an indirect way. It seems to be felicitous only to the extent that the corresponding question is understood in relatively loose terms; see A’s use of a how-question in combination with the predicate think. This is different in example (4):

In this case, A’s question is tailored more clearly to a regular assertion. Therefore, B should commit to the truth or falsehood of the proposition that he is surprised, which renders answering by demonstration less appropriate. Correspondingly, the exclamative in (4b) is worse than the assertion in (4a). This difference does not follow from the assumption that exclamatives are elliptical emotive assertions. Furthermore, T&G misrepresent the position of Truckenbrodt (2013) on this issue. They suggest that there is a difference between deliberative ob-interrogatives on the hand and exclamatives on the other, with only the former being at odds with an elliptical analysis; see (5) [= Truckenbrodt 2013 (8), slightly adapted]. However, Truckenbrodt (2013) makes the very same argument for exclamatives; see his example (6) [= Truckenbrodt 2013 (10), slightly adapted].

Truckenbrodt’s judgment thus runs counter to the assessment of (2) in T&G. Rather, it sides with my position that, at best, exclamatives are substitutes for emotive assertions without being such assertions.

3 Attachment sites and anaphoric relations

Examples such as (7) [= T&G (70)] indicate that adverbial clauses cannot take scope over the surprise associated with exclamatives.

For Grosz (2012), this argues against a silent emotive matrix predicate within exclamatives. If exclamatives involved subordinating predicates, adverbial clauses should easily attach to these predicates. T&G (p. 40) disagree with this conclusion by attributing the observation in (7) to the fact that the speaker’s surprise is not discourse-given, but asserted. For the following reasons, I side with Grosz’ original position.

First, I do not understand the logic of the counterargument. If emotive predicates were present at the syntax–semantics interface, attachment sites should not be affected by whether the emotion is given by the preceding discourse or not. For one, the example in (8) shows that adverbial clauses can scope over new content that is asserted. Furthermore, in (9), the surprise is discourse-given. Nevertheless, scope of the adverbial over this attitude remains as bad as in (7).

Second, various constraints on anaphoric relations corroborate Grosz’ position from a different angle; see the following minimal pairs for exemplification:

The pronoun es ‘it’ can pick up an emotive predicate with emotive assertions, but not with exclamatives (10). Furthermore, the pronominal adverb deshalb ‘therefore’ can easily relate to the speaker’s surprise only with emotive assertions (11). The same holds for introductory wh-pronouns in continuative relative clauses (12). Notably, these contrasts are independent of discourse conditions. They all favor the assumption that the attitude associated with exclamatives is introduced in a way that is structurally different from its introduction within emotive assertions.

4 Verb position, complementizer dass, discourse particle aber auch

The distributional observations discussed so far concern the external combinatorics of exclamatives in German. Let us now turn to their internal make-up. T&G draw on three aspects in particular: the verb position, the restriction to the complementizer dass ‘that,’ and the role of the discourse particle aber auch (lit. ‘but also’). In the following, I would like to show that the respective arguments are not conclusive.

5 On the interpretation of the attitude in exclamations

It is characteristic of the emotive attitude in exclamations that it is bound to the speaker, to the time, and to the place of the utterance, in short, to its origo. This orientation toward the origo follows for free from the assumption that exclamations are demonstrations of emotions, as demonstrations are bound to the one who performs them, and to the time and place of the performance; see the deictic nature of speech acts more generally. However, it does not follow for free from the contrary assumption in T&G that exclamations are emotive assertions. The emotive attitude would hinge on a regular predication of a subject, as a consequence of which the emotion could in principle be attributed to any subject at any time and any place. In order to remedy this problem, T&G (p. 28f.) argue that the observed orientation toward the origo is rooted in a pragmatic implicature. Specifically, based on Gricean manner and quantity maxims, hearers would interpret an unspecified subject feature [±origo] as the “simplest unmarked [+origo].” I am not convinced by this explanation for the following reasons.

To begin with, I am unsure about the relevant theoretical underpinnings. Is there independent evidence for the assumption that silent subjects in German bear an unspecified origo feature? Furthermore, subjects in German are not specified for tense. This suggests that the proposal also requires an unspecified origo feature for the verb, which calls for further independent evidence. Finally, I do not understand why [+origo] is supposed to be simple and unmarked in comparison to [–origo].

There are also open empirical questions. For one, the orientation of exclamations toward the origo does not behave like an implicature. Specifically, the putative implicature cannot be canceled by an appropriate context; see (20), where A’s second utterance in (20c) should be specified as [–origo] in light of A’s first utterance in (20a) and B’s reaction in (20b). However, this is completely impossible for the exclamative in (20c).

It is also worth taking a closer look at a context in which the origo of exclamatives shifts systematically, namely, at free indirect discourse (= FID); see Eckardt (2014: Ch. 6) for an extensive discussion based on Rett’s approach to exclamatives. The interpretation of the following example is revealing for our case.

The exclamative in (21a) supports the interpretation as FID. That is, one interprets the exclamative as if Adam himself expresses his surprise at how Olga’s singing has just touched him. So the origo shifts to Adam and his time and place despite the use of preterit tenses and 3rd person pronouns. This tension between form and function is the hallmark of FID. Under the assumption that exclamatives are demonstrations of emotions, the observed shift follows from general properties of FID, namely, from the shift of the illocutionary responsibility to a character. The approach in T&G, by contrast, faces a serious challenge here. As the authors assume a silent emotive assertion on top of the subordinate structure, (21a) should be equivalent to (21b). Specifically, if the subject in emotive assertions had an origo feature, it should be shiftable in the given context, regardless of 3rd person pronouns and preterit tenses. However, this is not the case: (21b) differs from (21a) exactly by not triggering the effect of reading Adam’s mind. Rather, (21b) is interpreted as the narrator’s report on Adam’s feelings. In addition, this puzzle cannot be solved by assuming an emotive assertion with the 1st person pronoun and present tense; see (22), which cannot convey that Adam is surprised at his time and place because ich ‘I’ and ihn ‘him’ need to refer to different persons.

I conclude with a note on the kind of attitude relevant for exclamatives. T&G assume that the relevant predicate presupposes counterexpectation and thus leads to surprise as the relevant attitude. However, according to Rett (2011) and Eckardt (2014), exclamatives rather express that the speaker considers a state of affairs remarkable; see (23) [= Eckardt 2014: Ch. 6, (15), slightly adapted], where the relevant state of affairs is expected by the speaker.

From this perspective, surprise is merely the most typical reason for why a state of affairs is considered remarkable. This detail may be pedantic. However, it points to a more general question: why should exclamatives be based precisely on emotive predicates for surprise? If the attitude in exclamatives were rooted in a descriptive emotive predicate, one could expect more variation. For instance, be happy about or be angry about might be suitable as well. However, these options are not at stake. By contrast, the relevant attitude is abstract, the more so if Rett’s and Eckardt’s qualification is accepted. This fits better with deriving the attitude from the interaction of given contents with nondescriptive functional properties than from their interaction with descriptive predicates. Isn’t the exclamative accent a fairly intuitive means for marking given contents as remarkable?^[5]