2.1 Cashing out dismissal

We analyze THROW as an operator that communicates discourse-managing intentions of the person gesturing, in the spirit of Romero & Han (2004), Farkas & Bruce (2010), and Repp (2013), among many others. We also draw on Rett’s (2021) approach to emotive markers (e.g., alas, unfortunately); like these markers, THROW signals an agent’s evaluative attitude towards something in the discourse.⁶ However, we do not claim that THROW is necessarily emotive. We rather assume a notion of evaluativity in the spirit of, e.g., Ernst (2009: 512), where evaluatives include emotive expressions such as unfortunately and amazingly as well as less clearly emotive expressions such as mysteriously, conveniently, or appropriately – and see Degaetano-Ortlieb (2015) on a discussion of importantly and notably in the context of evaluativity.

Our proposal is that THROW signals a negative evaluation that amounts to dismissal. At its core, we take dismissal to consist of marking something as unimportant. There are in principle several ways of thinking about unimportance in discourse – for example, in terms of low expected utility (e.g. van Rooy 2003; 2004; Bergen et al. 2016; see Daniels 2019), or irrelevance to conversational goals, among other possibilities. Here, we will treat unimportance as a primitive.

2.2 Formal rendition of the analysis

We propose that the use of THROW is governed by the discourse conditions in (4). Note that the utterance u_n hosting THROW can be devoid of accompanying speech (which we write as …_THROW), i.e., THROW has a speech-replacing (or pro-speech) use, as well as a speech-accompanying (co-speech) use, using the terminology introduced in Schlenker (2018a; b).

The analysis in (4) has two core components: a target question (Q) that THROW retrieves from the preceding discourse context, (4-i), and the attitude that THROW communicates regarding this target question, (4-ii). This question-based approach builds on the well-established view that utterances by and large (and, in particular, declarative utterances) evoke alternatives that amount to a question that is being answered (see, e.g., Roberts 2012; Büring 2003; Beaver & Clark 2008: 37).⁷ We remain agnostic as to whether the unimportance evaluation in (4-ii) is limited to the author, or whether the author conveys that people in general would consider this question unimportant – an issue that also arises for predicates of personal taste such as tasty or fun (see Lasersohn 2005; Stephenson 2007; Pearson 2013).

We will discuss both of the components in (4-i) and (4-ii) in turn, in Sections 2.3 and 2.4, respectively. Before doing so, we wish to highlight an important distinction between an utterance u_k that provides the target question Q vs. an accompanying utterance u_a, which combines with THROW to yield u_n (i.e., u_n = u_a + THROW), illustrated in (5B₁/5B₂). When THROW occurs on its own, as in (6B), u_n = THROW.

The only utterance that is relevant for whether the discourse conditions in (4) are fulfilled is the preceding u_k utterance, which we focus on in Sections 2.3–2.5. The absence of the accompanying u_a utterance from (4) is by design; as we will see in Section 2.6, u_a makes its discourse contribution independently from THROW, with the main restriction that the discourse contribution of u_a cannot be at odds with the dismissal that THROW communicates, as we saw in example (3).

It is worth pointing out that a co-speech realization of THROW, where the gesture is synchronous with the accompanying utterance, can be hard to distinguish from a pro-speech (or pre-speech) realization where THROW precedes the accompanying utterance. These realizations are most easily distinguished in examples where the accompanying speech has a prosodic peak near the end of the sentence, as in (7). Here, a pro-speech THROW will be realized before I, as indicated in (7B₁), whereas the stroke of a co-speech THROW will be aligned with the word containing the nuclear accent, baNAnas, as in (7B₂) (see, e.g., Ebert et al. 2011). The two realizations appear to have identical effects in (7B₁–B₂).

Going forward, when we probe for the semantic contribution of THROW, the most conservative approach is to look at THROW without an accompanying utterance, as the presence of an accompanying utterance will often trigger additional inferences. For instance, in (5B₁), we infer that B’s dismissal is justified by the accompanying assertion that tomorrow is Sunday, which suggests that it does not matter if it gets late on a Saturday night. By contrast, in (5B₂), we infer that B’s dismissal is justified by the accompanying assertion that it’s only 3am, which suggests that 3am does not count as late. These inferences are not contributed by THROW, nor does THROW operate on them in any meaningful way. They simply coalesce with the THROW contribution in the pragmatics. The inclusion of an accompanying u_a utterance thus introduces a confound into an investigation of THROW proper. In Sections 2.3 and 2.5, we only consider uses of THROW without an accompanying utterance; in Section 2.4, we revisit predictions on the u_k-u_a interplay, and we return to u_a in Section 2.6.

2.3 The target of THROW

We began with the informal intuition that THROW comments on something that has been said in the preceding discourse. There are in principle several ways to cash out this idea formally: we could say that THROW targets i) a proposition, ii) a question, or iii) a speech act. In (4), we have chosen option (ii), for the following reasons.

2.4 The evaluative attitude expressed by THROW

Let us review the details of how the proposal applies when THROW responds to a declarative. In (3), repeated here as (18), A’s assertion of I think I forgot to pay my credit card bill addresses the question {A forgot to pay A’s credit card bill, A didn’t forget to pay A’s credit card bill}. B₁’s response of It’ll be fine contextually entails that whether or not A paid their bill is not a serious issue. These are precisely the discourse conditions that make THROW felicitous in (18B₁). On the other hand, the response of That’s not good in (18B₂) affirms that forgetting to pay a bill is a serious matter; this is in conflict with what THROW requires, hence THROW’s infelicity. While some combinations of THROW and accompanying utterances may simply trigger a search for an explanation of why the gesturer uttered the accompanying speech (see example (29B₂) in Section 2.6), such an explanation is out of reach in (18B₂) due to the incompatibility of the gestural content and the speech content. When THROW is not accompanied by speech, as in (18B₃), it contributes an underspecified dismissive attitude non-exhaustively paraphrasable as don’t worry or who cares.

When the preceding utterance is a polar question, as in (19), repeated from (9), the analysis works exactly the same way, except that the question dismissed by THROW is overtly pronounced by A. Here, THROW dismisses A’s question rather than answering it; B’s gesture cannot be construed as providing a negative answer to A’s question (i.e., THROW cannot mean ‘no’ in (19)).

As in (19B₃), the precise flavour of the dismissal contributed by THROW is underspecified in the absence of speech. In (19B), in particular, the source of unimportance is underspecified; it could be that B does not care about B’s husband, but it could also be that B does not care about parties, among other conceivable readings.

The proposed analysis predicts that THROW will generally be incompatible with accompanying u_a utterances expressing sincere interest or concern about the targeted issue. This is exemplified by the infelicity of THROW in (20B) below.

Likewise, THROW is inappropriate (or at best extremely rude/flippant) in situations where weighty matters are under discussion and discourse participants are expected to treat what is said as important. For example, consider (21), where A is a police officer interviewing B as a suspect in a serious crime,¹⁰ or (22), where A is performing a sincere, well-planned marriage proposal to their partner B.¹¹

In this section, we have already raised the issue of underspecification and interpretational flexibility when it comes to the sources of unimportance. We now present a case study that further elaborates on these.

2.5 Underspecification/flexibility of unimportance

There are many possible sources of unimportance. The proposed analysis is compatible with this flexibility; a THROW dismissal may take on flavours of indifference, (indirect) denial, etc. depending on the context. For example, B’s use of THROW in (23) will, all else being equal, most naturally be interpreted as dismissing the question addressed by A’s assertion on the grounds that B believes A’s answer to be obviously false, and thus highly uninformative (cf. an utterance of nah, no way, or whatever, man).

If, however, we adjust the context as in (24) so that B has, exceptionally, allowed the dog into the bedroom, B’s use of THROW can instead be interpreted as dismissing the question of whether the dog is on B’s bed as unimportant because both answers are compatible with B’s rules that day, and thus neither answer is noteworthy (cf. an utterance of yeah, I know or it’s fine).

Crucially, when THROW communicates the denial of p (as in (23)), we predict that it does so by way of dismissing p – e.g., on the grounds that it is obviously false. This is supported by the observation that THROW cannot be used to provide a direct negative answer (≈ no) to a polar question in examples such as (9) above, as spelled out explicitly in (25). To the extent that (25B) may imply that Sam is not coming, this would once again be a case of dismissing nonsense (≈ whether Sam is coming or not is unimportant, because it is obviously false), parallel to (23).

2.6 Illocutionary independence of THROW

An important feature of our analysis that we have not yet explicitly discussed is the independence of THROW from the utterance u_a that it accompanies (if there is one). We propose that this is a virtue of our analysis, since THROW appears to encode an illocutionary act that is separate from any accompanying speech, giving rise to an interpretation along the lines of Krifka’s (2001) speech act coordination. THROW can be accompanied by u_a utterances with a variety of clause types, and it behaves uniformly regardless of what spoken content (if any) it accompanies. This is illustrated in (26); in each of B’s responses, THROW appears to dismiss A’s observation that B’s phone buzzed (the u_k) as unimportant.

Our analysis of THROW in (4) holds that THROW does not directly interact with or operate on the accompanying u_a utterance. Given the different semantic properties of, say, declaratives (26B₁), questions (26B₂), and imperatives/hortatives (26B₃) (see, e.g., Portner 2004), the flexibility of THROW to combine with more or less any type of u_a utterance corroborates this approach by making it implausible that THROW comments on the u_a utterance that it accompanies. THROW thus crucially differs from the iconic gestures that Schlenker (2018a; b) discusses in that it does not seem to give rise to any non-trivial interactions with accompanying speech. To see this, consider (27); the intended, but unavailable, reading is loosely modelled after Schlenker’s cosuppositional readings, where the contribution of THROW would be conditionalized on a positive answer to B’s question. However, such a use of THROW, in which it would compositionally interact with the question operator in the form of the non-trivial projective behaviour that (27B-i) spells out, is impossible.

The unavailability of the reading in (27B-i) sharply contrasts with the availability of that in (27B-ii). In the latter, THROW’s contribution is pragmatically conjoined with B’s request for information about whether something else is the case (namely: Is Sam coming?), giving rise to the speech act coordination [it is unimportant whether or not Alex is coming] and [I am asking you if Sam is coming]. We may intuit that B conveys a causal link along the lines of because Sam might be/is coming, it is unimportant whether or not Alex is coming, but this is plausibly an inference based on standard pragmatic reasoning, possibly involving discourse relations (see, e.g., Hobbs 1979; Lascarides & Asher 1993; Kehler 2002; Asher & Lascarides 2003; Jasinskaja & Karagjosova 2020; see also Lascarides & Stone 2009a; b; Hunter 2019 on the role of discourse relations in gesture-speech interaction). Crucially, no compositional interaction between THROW and the u_a is required to derive this effect. Parallel reasoning applies to the interpretation of (28), the declarative variant of (27).

Example (29) further illustrates indirect interactions between THROW and the accompanying utterance u_a. Here, the inferred discourse relations (see, e.g., Hobbs 1979; Lascarides & Asher 1993; Kehler 2002; Asher & Lascarides 2003; Jasinskaja & Karagjosova 2020) between A’s utterance and B’s utterance impact the combinability of u_a with THROW, as illustrated in (29). In (29B₁), the inclusion of only signals a disagreement between B and A, indicating the discourse relation Contrast; this disagreement is compatible with the dismissal that THROW communicates, and in fact supports B’s dismissal of A’s statement. By contrast, the spoken utterance in (29B₂) elaborates on A’s utterance without a clear sense of contrast, indicating the discourse relation Elaboration. We maintain that THROW is in fact acceptable in (29B₂) as well, but the orthogonal contributions of THROW and the accompanying u_a (= it’s 3am) trigger a search for an explanation: why did B elaborate on A’s statement by asserting it’s 3am while at the same time expressing dismissal by means of THROW? This gives rise to a slight oddness of (29B₂), indicated by the single question mark.

If THROW encodes a separate illocutionary act that is interpreted in conjunction with the illocutionary act associated with an accompanying utterance, this makes further predictions about the type of speech that THROW can accompany. Specifically, we expect THROW to accompany main clause material and not embedded speech. No such requirement holds for the preceding u_k utterance, as we saw in (14). Predicted exceptions are cases where THROW accompanies speech below the root-clause level that has illocutionary force in its own right, and thus hosts ‘embedded root clause phenomena’.

This prediction is confirmed by the distribution of THROW when it accompanies the antecedents of factual vs. hypothetical conditionals. Factual conditionals presuppose that someone believes that the antecedent proposition is true (Iatridou 1991, Haegeman 2003); on the basis of such observations, the antecedent of a factual conditional has been argued to possess illocutionary force separate from that of the consequent (see, in particular, Haegeman 2003: 334–336, who argues, building on Declerck & Reed 2001, that such antecedents “echo someone else’s speech act”). THROW is acceptable when accompanying the antecedents of factual conditionals, as illustrated in (30), but not in connection with the antecedents of hypothetical conditionals like (31), which lack this special property.¹²

In this respect, THROW resembles a range of speaker-oriented natural language expressions (see, e.g., Green 1976: 383–384; Repp 2013; Grosz 2020), which are restricted to root contexts in the sense of Hooper and Thompson (1973). An illustration is given in (32), where (32a) is a factual conditional, which allows for German schon in its discourse particle reading, while the hypothetical conditional in (32b) disallows it. The THROW examples (30) and (31) mirror the (32a)–(32b) distinction.

Note that in (30) THROW does not appear to dismiss the question addressed by B’s immediately preceding utterance of no one will notice (which also happens to be the question addressed by the antecedent proposition of the conditional uttered by A). Instead, A plausibly dismisses the question whether there is a typo in A’s handout or not, which A had addressed earlier.¹³ As a consequence, A’s use of THROW in (30) makes the same communicative contribution that B would make in (33), quite parallel to our earlier example (3). Here, the statement that no one will notice can be pragmatically construed as the explanation for why the presence/absence of a typo is unimportant.¹⁴

2.7 Non-at-issue status

Since THROW encodes a separate illocutionary move (making its own contribution to the discourse), we predict that it shares properties with other discourse-managing devices that have been given similar analyses. This prediction has already been confirmed by our comparison of THROW and German discourse particles in examples (30)–(33), which showed that both exhibit a restriction against accompanying—or occurring in—embedded (non-root) clauses. We proceed to show that, like German discourse particles, English emotive markers (e.g., Rett 2021) and a range of other expressives (e.g., Gutzmann 2013), the contribution of THROW cannot be targeted by direct denials, i.e., denials by virtue of No, … or That’s not true, … The response in (34A₁) can directly deny the truth-conditional content of (34B)’s speech, but it is not possible to use (34A₂) for directly denying the dismissal conveyed by the THROW gesture in (34B). However, the dismissal conveyed by THROW can be targeted in the same way in which other non-at-issue meanings can be targeted, e.g., by virtue of Hey, wait a minute! (see Shanon 1976 and von Fintel 2001), or its short form Hey!, in (34A₃); these have been argued to target “a wide range of non-at-issue meanings” (Potts 2012: 2521).

The unacceptability of (34A₂) confirms that THROW encodes non-at-issue meaning. In Rett’s (2021) terms, THROW encodes illocutionary non-at-issue meaning (in contrast to descriptive non-at-issue meaning), in that it encodes an agent’s attitude toward some descriptive content in the discourse.

3.1 Self-directed uses

The analysis spelled out in (4) does not require that the preceding utterance (u_k) is uttered by someone other than the individual who produces THROW. This is desirable, because THROW can be used in monologues like (35).

Here, as in the dialogues examined in Section 2, the speaker’s use of THROW with the second sentence dismisses the importance of the question addressed by the first. The speaker signals (to themself) that whether it is getting late or not is unimportant; this is presumably because, as they go on to assert, staying up for another half hour will not be harmful. In this way, the speaker raises a fact that could in principle serve as motivation for going to bed; upon consideration, they decide that the lateness issue is unimportant, and so THROW is licensed. The dialogue case in (36) seems to exhibit the same behaviour.

The distribution of THROW in monologues thus further supports our analysis; we now take this as our cue to explore what appear to be discourse-free uses in the next section.

3.2 (Apparently) discourse-free uses

The discourse conditions in (4) make reference to a prior utterance. We specified that there does not need to be an overtly expressed utterance u_a that accompanies THROW; this was necessary to account for cases of pro-speech THROW (i.e., cases where THROW is produced without accompanying speech). In this subsection, we note that the preceding utterance u_k that supplies the question dismissed by THROW may likewise be phonologically empty. To see that this is needed, consider an example like (37).

Here, THROW is perfectly felicitous despite the absence of an overt preceding u_k utterance. This can easily be understood as the self-directed use of THROW outlined in Section 3.1 combined with silent self-talk. If, as Eckardt & Disselkamp (2019) argue, speakers use language in their internal cognitive processes, it should not surprise us that THROW can target the resulting silent utterances. Plausibly, the speaker’s internal monologue in (37) looks something like this:

With the internal monologue filled in, A’s use of THROW is no longer surprising; it dismisses the current location of the sock as an unimportant issue.

3.3 Insincere uses

Like other meaningful expressions, THROW need not be used sincerely. Depending on the context and the content of the utterance that it accompanies, THROW can be deployed to produce irony, humour, or politeness. Since THROW is by its very nature a negatively evaluative expression, we expect irony to give rise to a positive meaning, much in line with ironic compliments (i.e., negative remarks such as you really did do an awful job that are used ironically for giving praise; see Kreuz & Glucksberg 1989; Kreuz & Link 2002, and Giustolisi & Panzeri 2021: 293 for recent discussion).

While THROW is not used to give compliments, one environment where THROW is particularly common is in response to compliments, as in (39).¹⁶ The intuition about such examples is that B accepts the compliment but uses THROW to display modesty.

In each of B’s responses, the use of THROW strictly speaking requires that B considers the question of whether they are great to be unimportant. If sincere, this dismissal of A’s compliment would be rather rude. However, it is most naturally understood here as a form of polite pretence; modesty demands that we minimize praise of ourselves. In this case, B’s use of THROW can be seen as a violation of Grice’s (1975) Maxim of Quality (part of the Cooperative Principle) to satisfy Leech’s (1983) Maxim of Modesty (part of the Politeness Principle).¹⁷

Another environment where ironic uses of THROW show up is in greetings where the pretended dismissal via THROW is interpreted as an intimacy-generating device, as illustrated in (40). Here, Sam’s use of THROW once again literally dismisses (the question raised by) Alex’s discourse-initial utterance; yet instead of creating a social rift, it seems to serve as a term of endearment. What happens here is very much like the interaction in (41), where Sam’s use of you old devil in (41b) should literally insult Alex but instead serves as a friendly epithet. Note that THROW can be added to this response quite naturally, as in (41c).

THROW differs from other gestures, including in particular facial expressions, which are known to signal that accompanying speech is to be interpreted ironically (Giustolisi & Panzeri 2021). Such an effect is illustrated in (42), where SICK-FACE stands for a facial expression signalling physical disgust (see Yoder et al. 2016: 303) which thus conveys that the accompanying assertion of tastiness is insincere. By contrast, in (39)–(41), the contribution of THROW itself is understood to be insincere – that is, the dismissal that THROW communicates is to be understood in an ironic way (as appreciation).

This reverse behaviour of gesture and speech with regards to irony can be made even clearer in (43), a variant of (39). Here, THROW does not mark that B’s statement that A is a darling is insincere. In fact, the opposite is the case: B’s statement that A is a darling serves to mark that B’s use of THROW is insincere. To our knowledge, this behaviour of THROW is systematic.

For present purposes, we here conclude our discussion of the behaviours of THROW and return to a separate issue, flagged in Section 1, namely the question of whether there are natural language counterparts of THROW, and how they compare in their meaning contribution and use.

5.1 Summary of main findings

This paper has analyzed the THROW gesture as an operator that marks discourse objects (namely questions) as unimportant. It contributes to a growing body of work that applies formal linguistic tools to the study of meaningful gesture (e.g., Lascarides & Stone 2009a; b; Ebert & Ebert 2014; Schlenker 2018a; b; Esipova 2019). However, most formal semantic work in this area has focused on deictic and iconic gestures, with notable exceptions including Ippolito’s (2019) work on the gestural marking of non-canonical questions in Italian, Francis’ (2021) work on gestural objection to discourse moves, and Laparle’s (2021) work on the gestural tracking of discourse topics. We thus take the present work to be part of a much needed filling-in of the theoretical landscape of pragmatic gesture uses. Such formal semantic work on gestures like THROW complements the rich work on pragmatic gesture that has long been underway in other traditions, including Bavelas et al. (1992), Kendon (1995), Müller (2004), Sweetser & Sizemore (2008), Wehling (2017), Cooperrider et al. (2018), and many others.

In addition to providing us with a better picture of how language works in its full multimodal form, the particular evaluative meaning conveyed by THROW is not like other discourse-level meanings that have been described in the formal semantics literature as of now. This paper has presented what is, to our knowledge, the first semantic analysis of a gestural dismissal operator in natural language. Moreover, the proposed meaning is not only expressed gesturally; the striking parallelism between THROW and pah suggests that this meaning is also encoded in spoken language. This dismissal meaning is special among discourse-management operators in that it cannot interact with spoken material that it accompanies; THROW and pah operate on discourse objects (questions) but cannot comment on the discourse object that hosts them (i.e., on the accompanying speech). This obligatory illocutionary independence raises an important question for future research; why do dismissal operators have this property when other discourse-management operators (such as VERUM, discourse particles, alas and fortunately) do not?

5.2 Towards a broader understanding of dismissal operators

We have argued that THROW and pah convey a very general kind of dismissal. This is not the only dismissal meaning that natural language could encode; there could be other operators in this semantic field encoding more specialized flavours of dismissal (e.g., relating to the source of unimportance), targeting different kinds of semantic objects, combining dismissal with other meaning components, or making their contribution in an at-issue fashion. Future work that examines a broader array of dismissal operators will hopefully allow us to pinpoint which of the properties identified here are unique to THROW/pah and which belong to this semantic class more generally. For example, one might imagine that the puzzling illocutionary independence of THROW/pah is a consequence of their meaning. Perhaps it is not useful to dismiss one’s own utterance; if one believes that a question is unimportant, it may not be fully cooperative to address it. However, preliminary investigation suggests that the picture may not be so simple. For example, English shm-reduplication (e.g., table-shmable; see, e.g., Nevins & Vaux 2003; Grohmann & Nevins 2004) appears to comment on its base, in contrast to the illocutionary independence of question-targeting THROW and pah. This is illustrated by (54), where table-shmable seems to convey something along the lines of who cares about tables? (rather than who cares whether Sam is building a table?, though this may be implied). It is an open question whether B’s dismissal targets the stem table- in B’s own utterance or the previous occurrence of table in A’s utterance.

As we consider (54), we observe that THROW/pah have a parallel use. This paper explicitly limited its scope to uses of THROW that accompany or replace entire utterances, as these uses strike us as distinct from uses in which THROW attaches to a word or phrase (see footnote 2). Turning our gaze to the latter, we observe that both (pro-speech) THROW and pah have uses in which they follow a single word, and which pattern exactly like shm-reduplication, in (55B₁) and (55B₂), respectively. Here, THROW/pah seems to dismiss the noteworthiness of table in the given context, paralleling what we saw in (54). Numerous examples of the sort in (55B₂) can be found in large corpora such as Google Books, e.g., for strings such as “computers, pah”, “progress, pah”, or “intelligence, pah”, among many others. Such uses of THROW/pah are potentially distinct from the ones we analyzed in this paper in that their dismissal always seems to target the accompanying material (i.e., the word that precedes them); however, a unified analysis may still be possible, which treats (55B₁) on a par with the equally echoic (30), in that THROW in (55B₁) comments on the question is Sam building a table? Regardless, we conjecture, based on the parallels between (54) and (55), that the correct analysis of the semantics of shm-reduplication (whichever that may be) can be unified with the analysis of examples of the type in (55).¹⁹

Other items in spoken English that may plausibly encode dismissal include discourse markers such as whatever (see, e.g., Kleiner 1998; Benus et al. 2007), meh (see, e.g., Schultz 2019: 5), hmpf (see e.g. Hougaard 2019: 102), and the doubled response particle yeah yeah, to name a few. With an analysis of THROW in hand, we are now in a better position to conduct fine-grained comparisons between operators within this semantic field.

This paper represents a first step toward understanding this kind of meaning; there is, of course, much more to do. For example, we have restricted our investigation to cases where THROW responds to declaratives and questions; it remains to be seen how the analysis can be adjusted to deal with cases where THROW responds to other kinds of discourse moves (e.g., imperatives, exclamatives). While we have so far treated pah as a spoken equivalent to THROW, further investigation is needed to determine whether they do in fact pattern together in all environments. Finally, THROW’s status as a pragmatic manual gesture means that it is articulatorily possible for it to coincide with other gestures (e.g., facial expressions)²⁰ and spoken pragmatic operators (e.g., discourse and emotive markers); how the discourse-management labour is divided among these channels is an open question.