2.1.1Indirectness and non-literalness

According to Searle’s (1969) speech act theory (SAT), the assignment of a speech act (or illocutionary act) to a particular utterance depends on the mood of the sentence uttered.11.Following the speech act theoretic tradition, I will use the phrase illocutionary act or, for short, speech act, to refer to the action performed by a speaker who utters a sentence with an illocutionary force and a propositional content. It is the morpho-syntactic features of the sentence (the illocutionary force indicators) that determine illocutionary force (Searle 1979: 30; Vanderveken 1990: 15–16). This theory is often called “literalist” because it postulates a one-to-one correspondence relationship between sentences containing such illocutionary force indicators and their illocutionary meaning when they are used in the performance of literal SAs. In other words, for SAT, sentence-types encode illocutionary forces — a view called the Literal Force Hypothesis. (For more details, see Ruytenbeek 2021, Chapter 1 and Meibauer 2019)

In SAT, in addition to the SA that can be predicted on the basis of the sentence-type, an utterance can result in the performance of another SA. For example, the interrogative sentence in (1) would constitute, at the literal level, the performance of a request for information about the addressee’s (A) ability to close the door.

In many situations, this interrogative construction is uttered by the speaker (S) to indirectly request that A close the door, without S actually intending A to provide them with an answer to the question asked. In that case, the only thing that would matter to S is whether A will comply with their request for action. According to SAT, the request achieved by means of the utterance in (1) is regarded as indirect in the sense that it is performed by means of, and in addition to, the literal SA associated with the interrogative sentence type instantiated by (1), namely a polar question about A’s ability to close the door. To infer that S did not want them to take the utterance of (1) as a literal question about their abilities, A has to use Gricean principles of conversational cooperation, as well as contextual information (Grice 1975). This inferential reasoning allows A to come to the conclusion that the literal question about their abilities is irrelevant in the current conversational context. A will then use the literal meaning of S’s utterance as a starting point to uncover S’s illocutionary intention beyond that literal meaning (Searle 1975, 1979: 113–114).

The original speech act theoretic definition of indirectness thus is closely related to the notion of non-literalness. It cannot, however, satisfactorily be equated with it (Grainger and Mills 2016: 35–37). In SAT, indirectness constitutes a subset of the category of non-literalness. Non-literal language use — for example, irony and metaphor — and indirectness are similar in the sense that the content of the SA that S primarily intends to perform is not identical with the content of the SA encoded by the sentence-type of the utterance (sentence mood). ISAs, however, are characterized by the double performance of a direct SA and an indirect SA, the latter being performed not instead of, but in addition to the former (Searle 1979: 114–115; 143–144). By contrast, in the case of a (non-literal) ironic or metaphorical utterance, S means something different from what they literally express. It is therefore possible for an utterance to convey both a non-literal SA and another, distinct, ISA. Accordingly, the interrogative utterance in (2) could be intended as a request in a context where A is making a lot of noise.

With (2), S would be requesting that A keep it down. According to SAT, such an ironic request would count two times as non-literal: S is not asking whether it is possible for A to make more noise, and neither are they actually requesting that A make more noise.

In SAT, there is a one-to-one relationship between each of the three sentence-types and the major English SA types (Searle 1979; Searle and Vanderveken 1985; Vanderveken 1990). According to the literalist view assumed by SAT, illocutionary forces are encoded at the level of sentence-types. Declaratives encode the illocutionary force of asserting, imperatives the force of requesting action, and interrogatives the force of requesting information. A straightforward consequence of this view regarding the scope of indirectness is that, as soon as the actual illocutionary force of an utterance departs from its direct illocutionary force, the SA counts as indirect. So, any request for action performed by means of a sentence that is not imperative, any request for information performed by means of a sentence that is not of the interrogative type, and any assertive SA that is not performed by means of a declarative sentence fall within the category of indirectness. For instance, the good wishes performed with the imperative utterance in (3) would be considered indirect, i.e., performed by means of, and in addition to, the literal directive SA encoded by the imperative sentence-type. (For a more detailed discussion of the indirect uses of imperatives, see, e.g., Ruytenbeek 2021)

As we will see in Section 2.2, in the aftermath of SAT, several scholars have extended the notion of indirectness to cases that do not fit the literalist dimension of the initial theory.

2.1.2Indirectness and the locutionary/illocutionary distinction

The speech act theoretic notion of indirectness introduced by Searle (1975) can be traced back to Austin’s (1962) distinction between locutionary and illocutionary acts. To simplify somewhat, according to Austin (1962: 94–98), performing a locutionary act boils down to producing a meaningful utterance; the performance of an illocutionary act necessarily entails the performance of a locutionary act. An example is (4), which, at the illocutionary level, would be a request to shut the door.

The locutionary act performed with (4) can be referred to by using reported speech, as in S said that A could close the door (Austin 1962: 95). In the case of the imperative in (5), the locutionary act would be that S told A to shut the door:

Thus, according to Austin, the three major English sentence-types, i.e., declarative, interrogative, and imperative encode different types of locutionary acts (see also Recanati 2013). While different SAs are possible at the illocutionary level, such as the assertion that A could shut the door for (4), only one locutionary act of saying is associated with the utterance.

As in Austin’s (1962) approach, the definition of indirectness proposed by Recanati (1987) builds on the distinction between locutionary and illocutionary acts associated with the utterance of a sentence. For Recanati, it is crucial to discern the illocutionary act indicated by an utterance, or its potential illocutionary force, and the illocutionary act that is actually performed by means of the utterance. According to Recanati, the locutionary act corresponds to the illocutionary act indicated by an utterance. In this view, it is possible for the illocutionary act indicated by a sentence (locutionary act) and the illocutionary act that the utterance is meant to perform (actual illocutionary act) to differ. This is what happens with ISAs. In (4), for example, at the locutionary level, S is saying that A is able to shut the door. The illocutionary act that is indicated by the utterance of the sentence (4) is thus an assertive. However, on some occasions, in uttering (4), S performs an illocutionary act that does not match the one indicated by the sentence uttered, i.e., a request that A shut the door.

Another way to approach indirectness in reference to the locutionary/illocutionary opposition has been proposed by Kissine (2013: 118–122). This time, the distinction concerns the propositional content of the locutionary act and of the illocutionary act, respectively. As Kissine (2008, 2013) suggests, in the case of an ISA such as (4), the propositional content of the locutionary act of saying that (i.e., that A is able to shut the door) is different from the propositional content of the illocutionary act of requesting, i.e., that A shut the door.

For Kissine (2013: 98–100, 118–122), the illocutionary act performed by uttering a sentence should be considered indirect only if its propositional content is not identical to that of the locutionary act that is performed with the utterance.

2.1.3Indirectness and secondariness

Thus far, we have discussed definitions of indirectness from a formal point of view, i.e., by differentiating levels of meaning associated either with the type of sentence used or with the type of SA performed or indicated. Now I would like to consider ISAs from a more cognitive point of view, by giving a central role to distinctions involving the perspective of the interpreter. One such distinction concerns the primary vs. secondary status of the SAs that can be performed by the uttering of a sentence. For instance, according to Recanati (2004: 74–75), an illocutionary meaning (or SA) would be secondary in the sense that its interpretation requires the derivation of another, more basic, primary meaning that is implied by the utterance act. Understanding the secondary meaning of an utterance thus involves an inferential path taking the primary meaning as a starting point. Secondary illocutionary meanings meet Recanati’s (2004: 42–44) availability condition, according to which language users are aware of the distinction between primary and secondary meanings, and of the inference from the primary meaning to the secondary meaning. Whether an IR performed by means of, for example, (1)–(4) would meet the availability condition is unclear.

This question boils down, in part, to asking whether the interpreters would also infer the meaning of a question (1) or statement (4) about their ability to shut the door. If the availability condition is not met, the SA will qualify as primary. (The experimental evidence discussed in Section 4 suggests that this is indeed often not the case and hence, that many of such SAs qualify as primary.) By contrast, it seems safer to hypothesize that, in the case of the remark about the negative state of affairs in (6), the primary meaning of a statement would be available to the interpreter.

The idea is that A would be able to reflect afterwards on the reason why S made a comment about the temperature in the room. The IR in (6) would thus be considered secondary in the sense that it meets the availability condition.

In Ruytenbeek (2021: Chapter 1), I pointed out an ambiguity in Recanati’s primary/secondary distinction. Based on Recanati’s writings, it is possible to interpret the distinction in terms of a logical relationship. According to another interpretation, the distinction concerns the temporal relationship between the two levels of meaning: an illocutionary meaning would be secondary because it is derived after the primary illocutionary meaning. This is precisely the definition that Kissine (2013) adopted. According to him, a SA counts as secondary if its understanding necessitates the prior interpretation of another, primary SA. Under this chronological definition, the IR performed with the remark in (6) would be considered secondary because A first has to interpret the remark as a statement about the temperature before being able to understand it as a request.

To return to the question of the definition of indirectness, we can conclude that the direct or indirect nature of an illocutionary act can be distinguished from the fact that it is primary or secondary. An indirect SA can thus be secondary, that is, interpreted without the utterance being understood as a direct (and primary) SA (Recanati, 1987: 165–167; see also Kissine 2013: 111–122). Accordingly, recent experimental work on the processing of indirect requests (IRs) such as Ruytenbeek et al.’s (2017) study indicates that the SA of requesting can be performed in an indirect manner without being interpreted as a secondary illocutionary meaning. (More on this in Section 4.1.)

2.1.4Conventional indirectness

The notion of indirectness that was originally proposed by Searle (1975) is associated with the distinction between conventional and non-conventional ISAs. As pointed out by Searle (1969, 1975), a general observation about ISAs is that the content of many sentences used in their performance is conceptually related to their felicity conditions. For instance, in the case of directives, the following IR constructions refer to different types of felicity conditions (cf. Searle 1975: 64–67, 71–72).

For instance, (1) concerns the preparatory condition for the performance of a directive SA, i.e., A’s ability to carry out the action expressed in the propositional content of the utterance. (7) refers to the propositional content condition according to which A will perform a future action. As a statement to the effect that S wants A to do the action of shutting the door, (8) concerns the sincerity condition for the performance of requests. Searle (1975: 66) noted that many sentences used in the performance of ISAs have a “generality of form.” What he was hinting at is that the syntactic structure of the constructions instantiated in, for example, (1), (7), and (8) includes a modal verb (i.e., can and will/would), the second-person pronoun you to single out the addressee and a verbal phrase (VP) corresponding to the action to be performed by A. In the literature, ISAs performed by uttering such sentences with a generality of form are called conventional (and also, albeit less frequently, conventionalized) ISAs (e.g., Searle 1975; Morgan 1978). What these constructions have in common is that they all instantiate a more abstract construction, i.e., modal verb + you + VP.

Clark (1979: 432–433) is more precise than Searle when it comes to explaining what it means for an ISA construction to be “conventionally” indirect. He explains that conventional ISAs involve different conventions of means, i.e., conventions about the strategies that can be used to perform ISAs. According to him, a convention of means “specifies a semantic device by which an indirect speech act can be performed” (Clark 1979: 433). An example of a convention of means is making a reference to the preparatory conditions that have to be met for the performance of the intended SA to be felicitous. From that perspective, the utterance in (9) would qualify as a conventional ISA — a conventional IR, to be more accurate — as it refers to a convention of means aligned with the preparatory condition for directives (cf. Searle 1975: 61–62).22. Panther and Thornburg (1998) reinterpreted Searle’s notion of conventionality of means through the lens of metonymic illocutionary scenarios, which outline how the propositional content of constructions used as IRs relate to the conceptual content of the SA of requesting. Within their cognitive linguistic framework, the components of the request scenario are a reinterpretation of SAT’s felicity conditions.

By contrast, unlike (9) and the constructions in (7)–(8), the polar interrogative sentence in (10) does not involve the convention of means about the preparatory conditions for directives (nor any other convention of means related to the felicity conditions for directives, more generally).

That being said, Clark claims that even though it is a non-conventional IR, (10), which expresses a question about the price of a fifth of Jim Beam, can still be used to request that A tell S how much a fifth of Jim Beam costs. IRs performed by means of sentences that do not express the expected action are also often referred to as “non-conventional” ISAs. An example is (6), which expresses a reason for A to perform some action.

Searle (1975: 66) pointed out that, even though it has not the generality of form discussed above, under proper circumstances (6) can serve as a request (for example, to shut the window or turn on the heating).

It seems to me that, despite the fact that sentences such as (6) do not explicitly mention the specific action that is requested from A, such utterances would nonetheless qualify as conventional ISAs: while unrelated to the felicity conditions of the SA, a relevant convention of means involves A’s reasons for complying with the directive (or S’s reasons for performing the directive). I fail to see why the notion of conventionality of means should be limited to the categories of sentences that both have a generality of form and refer to felicity conditions of a particular SA type. In Ruytenbeek (2021), I proposed to extend the scope of this notion not only to remarks such as (6) (the convention is the expression of a negative state of affairs), but also to imperative directives. In the latter case, the convention of means is expressing the content of the requested action using a force-dynamic pattern. I therefore believe that the right question to ask is not whether a construction used to perform an ISA is conventional or non-conventional; rather, it is about which convention of means is instantiated by a particular ISA utterance. Conventionality of means can thus be conceived of as a categorical criterion according to which a SA of a particular type can be performed by choosing from a list of strategies. In fact, this view does justice to Searle’s observation that not all the sentences of the group “Sentences concerning reasons for doing [the action]” have a generality of form. As will also become clear from the discussion offered in Section 2.2.1, whether an ISA construction is related to a convention of means for the performance of ISAs and whether it has a generality of form are two different questions that should not be confused. Because the bulk of research devoted to the relationship between SA performance and conventionality of means to date has focused exclusively on directives, another issue that remains open for further investigation is the extent to which the notion of conventionality of means can be applied to non-directive SAs.

2.2.1Standardization

In addition to the binary notion of indirectness and the categorical notion of conventionality (of means), what Clark (1979: 433) calls conventionality of form is also useful for distinguishing between the variety of indirect realizations of SAs:

There are […] conventions of form — conventions about the wording of indirect speech acts. Can you pass the salt? and Could you pass the salt?, for example, are highly conventional, or idiomatic, forms in English for requesting the salt. Is it possible for you to pass the salt? and Are you able to pass the salt? are less idiomatic, and Is it the case that you at present have the ability to pass the salt? is not at all idiomatic.

Within a single convention of means, constructions differ in terms of the degree to which they make their indirect illocutionary meaning available for interpretation. Consider the following two constructions, both of which concern A’s ability to shut the door.

While their literal meaning is identical (if one interprets modal can in the ability sense) and both can, in theory, felicitously be used as a request to shut the door, (1) is more strongly associated to the performance of an IR compared to (11). In other words, relative to (11), (1) has a stronger degree of standardization as an IR expression. Accordingly, ISAs performed by means of constructions such as (1) are referred to as standardized ISAs, while ISAs performed with constructions such as (11) are labeled non-standardized ISAs, referring to different positions on a cline of standardization. The notion of standardization relates to Morgan’s (1978) conventions of usage, according to which natural inferential schemes have “conventionalized” over time, giving rise to short-circuited inferences.

A similar view was advocated by Bach and Harnish (1979: 198), who proposed, following Morgan (1978), that the frequency of use of an IR construction such as (1) with a directive illocutionary meaning has resulted in the “compression of the inference” from the literal meaning of the utterance — the question about A’s abilities — to its indirect directive meaning. For them, considering that a given ISA construction is highly standardized for the performance of a type of ISA means that the inferential path from the literal to the indirect illocutionary meaning of the expression has been short-circuited; this enables, in the case of a Can you VP? IR, A to infer “[S’s] requestive illocutionary intent without having to identify the literal intent of questioning” (Bach and Harnish 1979: 198).

Following Bach and Harnish (1979), to avoid any confusion, I will not employ conventionality in two different senses. Rather, I will only use conventional(ity) to refer to conventions of means and standardization to refer to differences in the degree to which an ISA construction makes its ISA meaning more or less transparent for the interpreter.33.Note that Brown and Levinson (1987: 132; 290), following Morgan (1978), use the term conventionalization to refer to the diachronic process of standardization.

In addition to the two criteria that we have discussed above, i.e., conventionality of means and standardization, a third notion is required to satisfactorily account for the variety of constructions that can be used in the performance of ISAs. To illustrate, while the following two utterances involve the same convention of means about A’s abilities and are elaborations on the same Can you…? construction, they differ in the extent to which they make their indirect request illocutionary meaning salient.

This is what I called illocutionary force salience (Ruytenbeek 2021: Chapter 3). The idea is that, while the propositional content of these two utterances is identical, the presence of the request Illocutionary Force Indicating Device (IFID) please in (12) makes its IR meaning more salient compared to that of (1) (this prediction was experimentally demonstrated in Clark’s (1979) studies about IR interpretation by local merchants over the phone). Interestingly, the notion of illocutionary force salience both applies to direct and indirect SA realizations. For example, considering the pair of utterances in (5)–(13), it can be argued that the presence of please makes the request meaning of the construction more salient in the sense that the same utterance without the particle does not leave out other possible interpretations, such as a command or a piece of advice.

More recently, Holtgraves and Robinson (2020) have shown that the presence of an emoji can help interpreters disambiguate utterances with a potentially sarcastic meaning. This suggests that emoji can function as IFIDs, thus increasing the degree of salience of the (possibly indirect) illocutionary meaning of utterances.

2.2.2Degrees of indirectness

At this stage, regardless of the exact definition of indirectness that we assume, a SA either is direct or indirect. However, there is also a graded notion of indirectness available in the literature (see, for instance, Leech 1983; Blum-Kulka and Olshtain 1984; Blum-Kulka 1987; and Blum-Kulka et al. 1989). According to this view, direct and indirect refer to different positions on a scale of (in)directness. If we apply this notion to the SA of request, for example, the direct SA constructions according to the original speech act theoretic definition, i.e., imperative requests, would be considered as “more direct” compared to the request forms originally classified as indirect, such as Can you VP?.

A scholar who adopted this view is Leech (1983: 108), who was interested in the effect of “using a more and more indirect kind of illocution” on the degree of perceived politeness.44.Contrary to what Blum-Kulka (1987: 133) claimed, while Searle (1975) suggested that the amount of interpretative work can be expected to be higher for requests performed by means of, for example, negative state remarks compared to ability questions, he never explicitly endorsed a graded notion of (in)directness. Nor did Brown and Levinson (1987) assume that indirectness comes by degrees. The focus was thus on how participants perceived varying degrees of (in)directness (Blum-Kulka 1987: 132). The rationale for a scale of request indirectness was that “forms [can be] ordered approximately according to […] the obviousness of the directive” (Ervin-Tripp 1976: 29) or to their “degree of illocutionary transparency” (Blum-Kulka 1987: 133). Accordingly, the longer “the inferential path needed to arrive at an utterance’s illocutionary point,” the more indirect the SA performed with this utterance (Blum-Kulka 1987: 133). A positive correlation is thus postulated between the degree of indirectness and the amount of inferential work for the interpreter. In the case of the most direct SA realizations, the illocutionary force is indicated, for example, at the level of the sentence-type used, while for the most indirect forms the illocutionary force has to be inferred.

Following the view that (in)directness comes in degrees, a variety of IR realizations were arranged on a cline of increasing indirectness, ranging from imperatives to negative state remarks, with standardized IR constructions of the Can you VP? type in between. While early proposals distinguished between only six request realization types (Ervin-Tripp 1976: 29–45), later classifications included eight (House and Kasper 1981: 163–164) or nine categories (Blum-Kulka 1987: 133).

Despite its apparent appeal, the operationalization of (in)directness in terms of relative degrees faces a number of problems. From the perspective of the amount of inferential work for the interpreter, some contrasts do indeed seem a priori plausible; imperative requests, for example, should be easier to interpret as requests relative to negative state remarks (cf. the category of hints in Ervin-Tripp (1976: 42–45) and House and Kasper (1981: 163)). For other contrasts, however, it is less clear whether one construction, e.g., ability interrogatives, should be perceived as more indirect than the other, e.g., want declaratives. In fact, the defendants of what can be called the graded indirectness hypothesis rarely made explicit the criteria they used to position different constructions on the (in)directness scale. This is understandable in Blum-Kulka’s (1987) study, as relative positions on the cline of (in)directness were determined on the basis of respondents’ directness ratings.

Directives set aside, the definition of (in)directness that has been applied to the SA of complaining in previous work (House and Kasper 1981: 160–161; Trosborg 1995: 315–372) proves even more problematic. As pointed out by Decock and Depraetere (2018), in addition to the notion of the transparency with which the illocutionary meaning is conveyed (assuming greater inferential work for more indirect SA realizations), the “assumed degree of face-threat” (2018: 35) for the recipient of the complaint also plays a role. For instance, according to House and Kasper (1981: 160–161), the complaint performed by means of (14) is more indirect compared to that performed with (15).

While, as House and Kasper proposed, in both (14) and (15) S indeed explicitly asserts something, the propositional content of (15) does not express the complainable, i.e., the negative event that S is complaining about. In face-threat terms, (15) is more likely to be perceived by A as being offensive — at least in Western cultures — (Decock and Depraetere 2018: 35), but that has nothing to do with the degree of illocutionary transparency or the hypothesized amount of interpretative work for A. In the same vein, Trosborg (1995: 314) draws a contrast between a “straightforward accusation” (direct complaint) and utterances that “only indirectly express [S’s] ill feelings towards [A].” Here, the propositional content of the intended complaint SA is different in both cases, which makes an (in)directness-based comparison irrelevant. The upshot of this discussion is that complaint (in)directness, as adopted by House and Kasper (1981) and Trosborg (1995), is a (largely unclear) combination of illocutionary meaning transparency and face-threat considerations. A possible solution would be to replace it, as Decock and Depraetere (2018) did, with a graded notion of illocutionary transparency of the complaint meaning.

A third issue is that the graded notion of (in)directness only seems viable if it is operationalized in a context-dependent manner. That is, unlike the traditional binary definition of indirectness, it is virtually impossible to determine a priori all possible contrasts in terms of relative (in)directness between candidate constructions. Moreover, the validity of a particular scale of (in)directness would require systematic empirical investigation. That being said, the view that different realizations of a given SA type can be ordered on a cline of (in)directness has made an important contribution by allowing testable predictions about the degree of interpretive effort associated with different SA realizations.

2.2.3Cross-cultural research on indirectness

Under the impulse of Brown and Levinson (1987) and Blum-Kulka et al. (1989), scholars have explored the (in)direct realizations of different speech acts across different languages. Most, if not all, of the studies that I will briefly address below adopt a graded view of (in)directness, whether at the level of individual SA realizations or at the level of groups of speakers. Using either authentic data from corpora or natural interactions or data from discourse completion tasks, these authors document speakers’ preference for more or less (in)direct speech act realizations. In these studies, indirectness is often approached as a type of politeness strategy (Brown and Levinson 1987; more on this in Section 3).

A tendency in cross-cultural pragmatic research on indirectness has been to include an English lingua-culture as a standard of comparison (e.g., Kerkam 2015 on Arabic; Ngor-To Yeung 2000 on Chinese; Larina 2008 on Russian; Ogiermann 2009 on German, Polish and Russian; Spees 1994 on Japanese; Tannen 1981 on Greek; Hidalgo-Downing et al. 2014 and Márquez Reiter et al. 2005 on Spanish). However, such a reference to English is not systematic and there also exist monocultural studies, such as Le Pair (2005) and Schouten (2007) on Dutch and Pizziconi (2009) on Japanese. In addition, the last few years have seen a growing number of studies adopting a cross-cultural comparative perspective, such as Chen and Wang (2021), who investigated indirectness in the speech of Chinese and Korean speakers, and Venuti (2020), who focused on IRs in German and Italian (see also Rygg (2012) on Norwegian and Japanese).

A large body of research specifically concerns (in)directness in directives (Benzdira 2023 on Algerian Arabic; Ngor-To Yeung 2000; Zhang 1995 on Chinese; Byon 2006 and Yu 2011 on Korean; Lwanga-Lumu 1999 on Luganda; Upadhyay 2003 on Nepali; Ruytenbeek 2019, 2020 and Manno 2002 on French; Marti 2006 on Turkish; Ruzickova 2007 on Spanish; Wierzbicka 2003: Chapter 2 on Polish). Here, too, cross-cultural comparisons can be found: De Geer and Tulviste (2002) documented IRs (and requests in general) in Swedish and Estonian, Veres-Guśpiel (2020) IRs in Polish and Hungarian, Márquez Reiter (2002) in Uruguayan and Peninsular Spanish, and Marsily (2018) in French and Spanish. Another example is Yu (2011), who analyzed the relationship between (in)directness and politeness in request realizations in English, Hebrew, and Korean. In Sifianou (1993), off-record indirectness was investigated both in Greek and in English. In the footsteps of Ogiermann (2009), Urbanik (2017) explored directives in Polish and Norwegian. As far as other SA types are concerned, in Chinese, for instance, indirectness has been investigated in general (Zhou and Zhang 2022), as well as in evaluations (Chen and Wang 2021), proposals, and disagreements (Ngor-To Yeung 2000). The indirectness of both requests and criticisms has been analyzed in Russian (Kulbayeva 2020).

Although not exhaustive, this overview shows that empirical studies on the topic of SA indirectness in the world’s languages are far from systematic. Hopefully, the coming years will see more comparative work between distant languages and cultures, as well as research targeting different types of SA within specific lingua-cultures.