Mitigating strategies and politeness in German requests

2.1 Requests as face-threatening acts

According to Searle’s (1969) classification of speech acts, requests are directives. Thus, the illocutionary point is that a speaker S wants a hearer H to perform an action A in the future. Since S’s wants restrict H’s freedom of action, requests pose a threat to H’s negative face (in the sense of public self-image) according to Brown and Levinson’s (1987 [1978]) seminal work on politeness. Thus, requests are so-called face-threatening acts (FTAs). The potential face-threat caused by a request requires politeness which comes in the form of linguistic expressions that mitigate the face-threat on H. A central aspect of Brown and Levinson’s politeness theory is that it suggests a direct connection between a linguistic form (e.g., an indirect speech act) and politeness. In the case of requests, this means that S can use more or less explicit expressions depending on the contextual variables. If politeness is the less preferred strategy in a specific situation (e.g., if urgency prevails), S can perform the request by going “bald-on-record”, i.e., without taking care of H’s face. Imperative requests without mitigation are typical examples of this strategy. However, doing an act baldly, without any redress, hardly ever occurs in requests with a benefit for the speaker.^[2] In situations where face considerations play a role, S can go “on record” and make the communicative intent transparent, but the request requires redressive action in the form of positive or negative politeness. A speaker can be positively polite by paying attention to H’s needs (e.g., by making a compliment) or by marking H as an in-group member (e.g., by using terms of endearment such as darling). Negative politeness manifests itself, for example, in conventionally indirect speech acts such as the prototypical Can you VP? request. According to Brown and Levinson (1987: 70), they are a compromise “between (a) the desire to go on record as a prerequisite to being seen to pay face, and (b) the desire to go off record to avoid imposing”. Even though the speech act is not maximally explicit the intended meaning is undeniable here. The situation is different for “off record” strategies, where the communicative intent is not transparent. According to Brown and Levinson (1987) “off record” strategies such as negative state remarks (e.g., It’s cold in here) trigger conversational implicatures (e.g., +> close the window).

In their model, Brown and Levinson (1987: 68–83) specify further when politeness can be expected. It is assumed that the use of a certain strategy for the utterance of a potentially face-threatening act is determined by the situational value of the three variables “power”, “distance”, and “degree of imposition”.^[3] Power refers to the social status difference between S and H here. The greater this difference and the lower the relational power of S, the more linguistic mitigation is to be expected. Distance refers to the familiarity between interactants. As familiarity increases, the need for mitigating procedures decreases accordingly. Finally, the weight or degree of imposition, which is a feature of the requested action and can be defined by variables such as benefit, action type, or urgency, determines the use of linguistic politeness strategies: low degree of imposition requires less linguistic mitigation. For instance, requests for non-physical activities and/or a low urgency – such as a request for the confirmation of an invitation within the next few days – have a low imposition. These social factors and their complex interactions lead to the existence of a large variety of forms that are used to perform a request.^[4]

Despite its plausibility, Brown and Levinson’s (1987 [1978]) politeness theory has often been criticized – not least because of the face concept which is central to it – as showing a bias towards a too individualistic perspective that cannot be considered universal (for an overview see, e.g., Ehrhardt 2018; Flöck 2016: 24–28; Terkourafi 2012). Newer discursive approaches such as that of Locher and Watts (2005) do not view politeness as exclusively speaker-oriented but rather as something that is interactively negotiated, i.e., as “relational work” or “rapport management” (e.g., Spencer-Oatey 2005). These approaches do more justice to the fact that politeness is not primarily about the speaker’s need to mitigate face-threatening speech acts. In a similar vein, Terkourafi (2015) argues that “conventionalization” is more central to politeness than the notion of indirectness/implicitness. In her model, politeness is not understood as tentativeness, “but as providing outward displays of one’s familiarity with the norms governing the current exchange” (Terkourafi 2015: 15). Thus, she regards politeness as a matter of habit, not of rational calculation (of face needs). I also take the view that politeness should not be defined as a level of meaning communicated by means of a requestive utterance and cannot be primarily based on the notion of “indirectness”. Politeness is rather a matter of utterance interpretation that can but need not be achieved by H and is thus a perlocutionary effect of S’s utterance (see e.g., Searle and Vanderveken 1985 on this notion). However, I also assume that speakers are – or at least try to be – strategic to a certain degree when it comes to the formulation of a more or less polite utterance as they adapt their verbal requestive behaviour to politeness-sensitive social variables (such as degree of imposition). Thus, in line with Ogiermann (2009), I take the view that even though no utterance is inherently polite as such, there is nevertheless a great degree of consensus among members of a culture that allows to make predictions on the appropriateness/politeness evaluation of particular structures in a given context. Thus, the crucial concepts of Brown and Levinson’s framework such as “face-threat” or “negative politeness” still prove very useful for analysing requestive strategies. Or, as Kádár and House (2021: 81) put it:

any theory which argues that there is a definite relationship between form and politeness – or totally rejects this relationship – is potentially problematic if it does not take a contrastive perspective, simply because this relationship is always subject to linguacultural variation.

The experimental approach presented in Section 3 makes it possible to study the relationship between form and politeness in the sense of face-threat compensation within one language. Before addressing my own study, I will first take a closer look at the different components of requestive utterances and their specific politeness potential.

2.2 Linguistic mitigating strategies

2.2.1 Explicitness of the head act

Let us first look at the different requestive head act strategies that can be distinguished according to their “explicitness” before discussing the politeness effects attributed to them. Explicitness refers to “the degree to which the speaker’s illocutionary intent is apparent from the locution” (Blum-Kulka et al. 1989: 278). Unlike Brown and Levinson (1987 [1978]), Blum-Kulka et al. (1989), and the literature referring to them, I will not use the terms “indirectness” and “directness” in what follows – these are traditionally used to refer to different linguistic strategies for performing the requestive head act – preferring “implicitness” and “explicitness” (with the exception of “conventional indirect request”, which I use as a label for a certain category; see below). By doing so I want to avoid confusion, especially with speech act theoretical approaches where “(in)directness” has a special connotation and is not taken as scalar. Furthermore, the term “explicitness” is less ambiguous than “directness”: while “explicitness” only captures the apparency of S’s illocutionary intent, “directness” is sometimes also entangled with the expected degree of face-threat (see Decock and Depraetere [2018] for a critical discussion of the terminological and conceptual confusion regarding the notions of “directness” and “indirectness”).

In accordance with the literature on requestive strategies, I regard the terms “explicit” and “implicit” as abstract labels referring to the endpoints of an explicitness continuum (see, e.g., Blum-Kulka 1987; Blum-Kulka et al. 1989; Brown and Levinson 1987 [1978]; Leech 1983; see also Ruytenbeek 2021: 81–105 for a discussion). Following Blum-Kulka et al. (1989), I further assume three broad categories or levels of explicitness for the head act, labelled as “impositive”, “conventional indirectness” and “hint”.

So-called impositives exhibit the highest degree of explicitness in that the request is marked as such by the sentence type in the case of imperatives or by other elements (e.g., in the case of explicit or hedged performatives by the performative verb).^[5] Also locution derivables and want statements have a high degree of explicitness. This is evident, for example, from the fact that a response to the direct (i.e., assertive) interpretation of a want statement such as I want you to leave the room by replying Indeed or That’s right does not seem appropriate.

Conventionally indirect requests (CIRs) are considered less explicit. The strategies of this category are based on conventions of means and have a high degree of standardization in the sense of frequency of use (see, e.g., Clark 1979; Kasper 1995; Ruytenbeek 2021: 35–40). One such convention of means is connected to the speech act theoretical notion of preparatory conditions: In these strategies, also referred to as query preparatories, the speech act is performed by reference to a preparatory condition for the feasibility of the request. For example, ability questions such as Can you VP? target the first Searlean preparatory condition for requests: “S believes that H is able to perform the requested act” (cf. Searle 1969).^[6] Besides ability, reference to the willingness of H is a common strategy in some languages (e.g., Will/Would you VP?).^[7]

Finally, hints – also called “non-conventional indirect” strategies in the literature – are maximally implicit requests that trigger implicatures. In the case of strong hints, the request is made either by reference to the requested object or via contextual information. In the case of mild hints, the utterance contains no element which is of immediate relevance to the illocution or proposition. Since the communicated intention is not unambiguously clear to H, the meaning is more or less negotiable. Table 1 gives an overview of the three main categories and the different strategies for the realization of the head act.^[8]

Table 1:

Explicitness strategy types (according to the CCSARP, cf. Blum-Kulka et al. [1989: 278–281]).

One might assume now – as was partly done in earlier approaches (e.g., Leech 1983) – that greater implicitness implies increased politeness, i.e., that there is a linear relationship between the two. However, the relation between explicitness and politeness is more complex.^[9] As Blum-Kulka (1987) has shown in her pioneering study, more implicitness is no guarantee for more politeness: the most implicit request strategies (i.e., hints) were not perceived as most polite but conventionally indirect requests were, particularly query preparatories. As mentioned above, CIRs are said to be both communicatively effective due to the low level of ambiguity and interactionally safe since S takes care of H’s emotional costs. Hints, on the other hand, are said to involve higher communicative risks since they might carry a multiplicity of meanings in context (Blum-Kulka 1989).

As previous experimental studies on German have shown, Can you VP? is indeed the most frequently produced requestive strategy in this language (see, e.g., Ackermann 2021; Blum-Kulka and House 1989). When it comes to the politeness potential of this query preparatory there are, however, two major hypotheses as Ruytenbeek (2020: 119) mentions “On the one hand, Can you VP? is used to communicate extra politeness effects absent in imperative directives. On the other hand, Can you VP? enables speakers to avoid being perceived as impolite, while making their directive intent clear enough”. Production task experiments on German provide evidence for the first or “extra politeness” hypothesis since Can you VP? occurs more frequently in contexts where politeness plays a bigger role, especially in situations with a high degree of imposition (see, e.g., Kranich et al. 2021; Lochtman 2022). If Can you VP? could be considered a “baseline” for politeness evaluations (and enable S to avoid the risk of being perceived as impolite) it should be used equally frequently in different social contexts – a result that Ruytenbeek (2020) reports for Belgian French e-mail communication.

3.1 Tested variables and hypotheses

In the experimental study which is described in this section I use an online survey with both a DCT and appropriateness ratings. The DCT – which is more central to this study – contains two situations in which the participants were asked to formulate a request, with the degree of imposition manipulated. As empirical studies on German (and English) have shown, a higher degree of imposition is particularly associated with more polite requests (see, e.g., Biesenbach-Lucas 2007; Brown and Gilman 1989; Holtgraves 2005; Kranich et al. 2021 among others).^[15] Moreover, it has been shown that features of the requested action (e.g., weight of imposition) are less sensitive to language change and even seem to override social relational factors such as power or distance (see Blum-Kulka and House 1989; Kranich et al. 2021).^[16] Thus, manipulating the factor “weight of imposition” seems to be promising for investigating politeness.

It is also expected that verbal mitigation strategies will be observed in both situations in the DCT test, since in both cases participants had to produce a face-threatening request with a benefit for them. However, it is further assumed that the need for politeness increases with the degree of imposition. Thus, a tendency towards unmarked behavior is expected in the low-imposition task or – in Watts’ (2003: 257) terms – “politic behavior” that is defined as “linguistic and non-linguistic [behavior] which the participants construct as being appropriate to the ongoing social interaction”. Increased politeness, i.e., linguistic strategies that indicate the speaker’s attempt to save the intended addressee’s face, is more likely in the high-imposition context. This leads to the following hypotheses regarding the realization of the requestive utterances in the production data.

3.2 Test design

3.3 Data collection and participants

The data collection took place between December 2018 and March 2019. The link to the online survey, which was created via Soscisurvey,^[21] was distributed via various mailing lists and social media channels. After the data collection, coding of the requests from the DCT was done according to the coding scheme found in the appendix, which is based on the CCSARP.^[22]

A total of 578 native speakers (plus 7 non-natives, which were excluded) from Germany (465), German-speaking Switzerland (74) and Austria (39) completed the online questionnaire.^[23] The participants were between 15 and 91 years old (Ø = 39.8). 376 people said they were female, 186 male; 16 participants did not assign themselves to any gender.

4.1 Production data

The sample contains 1,006 observations: 559 for the low- and 447 for the high-imposition context. This is because 23% of the participants refused to respond to the high-imposition situation – about half of them indicating a perceived discomfort in the comment field (e.g., käme mir sehr unhöflich vor ‘would seem very rude to me’). In contrast, only 3% of the participants could not identify with the low-imposition situation.^[24] This shows that the imposition for the addressee in the high-imposition context was considered (too) high to imagine a possible requestive strategy by a substantial number of participants.

In order to assess which potential mitigation strategies occur significantly more often in the high-imposition context, the data have been analysed with a binary logistic regression model.^[25] The variables head act strategy, subjunctive, downtoner, understater, embedding, politeness marker, vocative alerter, and external modification were integrated in the multifactorial model. I started with a maximal model. Independent variables that had no significant influence were removed from the model in a step-down procedure based on the Akaike Information Criterion (AIC). The only variable removed was embedding as it did not improve the model quality. The specification of the final model is set out at (5).

Overall, the main effects predict the context very well (Nagelkerke pseudo-R² = 0.876). As the concordance index indicates (C = 0.984), the model discriminates extremely well.^[26] Table 3 gives an overview of the significant variables integrated into the model. In addition, the effect sizes (log odds) including 95% confidence intervals, the standard errors, the Wald test statistic (a ratio of the estimate and the standard error), the p-values, and the Variance Inflation Factors (VIF) are listed.

The effect coefficient – in contrast to the p-value – indicates the direction of the effect as well as its strength. Log odds are centered around 0 and reach from + Infinity to – Infinity. Thus, a coefficient of 0 would mean that there is no difference between the levels of the variable with regard to the choice between the high- and the low-imposition contexts. Positive coefficients indicate that the specified level (e.g., occurrence of a vocative) occurs more frequently in the high-imposition context in comparison with the base level (no vocative). The right-most column in Table 3 shows the VIFs. These VIFs (all < 3) indicate that there is no critical multicollinearity between the factors. In addition, bootstrapping was used to test whether the model overfits the data, which is not the case.^[27]

Overall, the statistical model confirms the hypotheses. As the positive effect coefficient indicates, CIRs (i.e., Can you VP?) are more typical for the high-imposition context compared with impositives (H1). As Figure 1 further visualizes, they are the dominant requestive strategy in both the low- (66%) and the high-imposition context (87%). However, while Can you VP? is nearly the only option in the high-imposition context, impositives also play a substantial role in the low-imposition context (28%).

Figure 1:

Observed frequencies for the strategies “impositive” “conventional indirect” and “non-conventional indirect” by context.

Contrary to H1, H2 is not confirmed by the data. As the results of the regression model indicate, implicit requests with a lower degree of standardization such as willingness questions (e.g., Magst du kurz halten? ‘Would you like to hold this for a second?’) show no significant effect (6% in the low-imposition context versus 9% in the high-imposition context). Thus, the data do not support the assumption that non-conventional indirectness is used by the participants to be extra polite in the high-imposition context investigated here.

In line with expectations, the subjunctive tends to occur more often in the high-imposition context (H4). As the coefficient indicates, the effect is small. In contrast, the politeness marker bitte and especially understaters (e.g., kurz ‘quick’), which have the highest coefficient, are characteristic of the low-imposition context, which confirms hypotheses H6 and H7. Both hypotheses H8 and H9 are also confirmed by the production data: single and especially multiple external modifiers as well as vocatives occur more frequently in the high-imposition context than in the low-imposition context.

As mentioned above, the variable embedding did not improve the model quality and is therefore not included in the final model. H5 is thus not confirmed. The non-significant impact may be due to the fact that embeddings are overall very rare in the production data (only 31 observations in total). One result that even contradicts the hypothesis concerns downtoners (H3). The negative coefficient indicates that the occurrence of downtoners (and here especially mal, eben, grade ‘just, once’) decreases the chance of high-imposition context. Thus, they are an indicator for the low-imposition context and not, as expected, for the high-imposition context. A discussion of the implications of these results – where also a more fine-grained analysis of downtoners is proposed – is given in Section 5. First, the results of the rating section are reported in the next section.

4.2 Appropriateness ratings

The results of the appropriateness ratings provide some information on the perception of different FTA strategies in standard situations. As the boxplots in Figure 2 show, participants prefer Could you VP? requests over imperatives and especially over negative state remarks (which were only tested in the breakfast situation). Even in situations with low social distance, no status differences, and a low rank of imposition, bare imperatives in particular are not considered fully adequate on average. In the breakfast situation 18% (105) of the participants made a politeness-relevant comment on the imperative item: 11% (63) noted that a bitte ‘please’ is missing, 4% (24) would formulate the request as a question and 3% (18) remarked that a combination of a question + bitte would sound more appropriate/polite. In the rubbish situation the imperative is modified by the appreciative subjectivizer sei so gut ‘be so good’. However, this does not give the explicit strategy a better rating on average. The lack of bitte ‘please’ or an interrogative strategy is only mentioned by 5% (29) of the participants and thus less frequently complained about in the comments here. However, seven participants (1%) noted that sei so gut ‘be so good’ sounds a bit old-fashioned, mummy-like or too demanding.^[28]

Figure 2:

Boxplots for different head act strategies in the breakfast situation (left) and in the rubbish situation (right); n = 578.

Overall, the CIRs are rated better in both situations (median=4). In the breakfast situation, nevertheless, 7% (43) of the participants complained that bitte ‘please’ is missing in the Could you VP? request. In contrast, only 2% (13) of the participants noted that the CIR – and here especially the subjunctive – seems too polite to them in such a familiar situation. The CIR in the rubbish situation was commented on less frequently overall, but similarly 5% (28) commented on bitte, 1% (7) found it too polite.

Remarkably, the negative state remark is rated worst (median=1). Here, 7% (38) of the participants either formulated/added a CIR (33) or a more explicit request (5) in the comment field. 3% (18) of the participants explicitly noted that the utterance is too indirect and thus even impolite for them (e.g., zu indirekt, unhöflich ‘too indirect, impolite’; wirkt so allein etwas schroff ‘seems a bit harsh on its own’). In line with previous studies, the results indicate that highly implicit request strategies such as negative state remarks are not perceived as more appropriate or politer than highly conventionalized ones in standard requestive situations with a low face-threat as the one I tested here.

5.1 Requestive head act strategy and internal modification

The query preparatory Can you VP? is the dominant requestive strategy in both the low- and the high-imposition context. This CIR also scores best overall in the adequacy ratings. However, Can you VP? is almost obligatory in high-imposition contexts, while impositives are also an option in requestive standard situations (H1). The low number of non-CIRs (especially in the high-imposition context) could partly be explained by the test design since it is plausible that “requests that in natural discourse might have been negotiated across several turns, are here confined to one turn. To achieve maximum efficiency, the requestive move in this case has to have the advantages of both a prequest and a request proper, an efficiency achieved by conventional indirectness” (Blum-Kulka and House 1989: 132).

The question now is to what extent Can you VP? contributes to the mitigation of the face-threat. As I would argue, there is not only one Can you VP? construction but two. These two constructions are usually lumped together under the term “conventional indirectness” due to their formal identity. However, high- and low-imposition Can you VP? requests differ with regard to their illocutionary meaning. The crucial difference lies in the degree of seriousness with which the “question for information” meaning of the utterance is intended. Especially in less standardized requestive situations, e.g., in high-imposition contexts, Can you VP? is often a real or “direct question for information” about H’s ability while it usually is a highly standardized and conventionalized “request for action” in low-imposition contexts. Thus, both constructions differ regarding their degree of explicitness.

From a constructionist perspective, standardized Can you VP? – which typically occurs in low-imposition contexts – only has the form of a question but means a request. Thus, the requestive potential is part of the pragmatic meaning of the construction; the semantics of können ‘can’ is bleached. The meaning of the construction is completely conventionalized and standardized, and S does not question the ability of H to fulfil the requested action (see, e.g., Stefanowitsch 2003 for a construction-based approach).^[29] In non-standard or high-imposition situations Can you VP? often – but of course not always – not only has the form of a question but is also uttered to express a question for information, which is less explicit than its more standardized counterpart with unambiguous requestive meaning.^[30] As I will discuss in the following, several internal modifiers are used to specify the intended pragmatic meaning/illocutionary point of the utterance.

The regression model has shown that speakers make use of very different internal modifiers depending on the context or face-threat. First, the adverb bitte ‘please’, which increases the illocutionary force salience of the utterance, is generally more frequent in the low-than in the high-imposition context (H6). This is not only because impositives, which require mitigation, occur more often in the low-imposition context. Bitte also occurs frequently, namely in 49% of the utterances with the real CIR (e.g., Kannst du bitte mal halten? ‘Can you hold this, please?’). Also in the adequacy ratings a frequent criticism was that a bitte ‘please’ was missing with CIRs in standard situations. This internal modifier not only specifies the requestive meaning but also marks an obvious request overtly as polite. Can you VP? + bitte is thus a good balance between clarity and politeness. Questions for information on the other hand do not allow for the occurrence of bitte since they don’t have a requestive meaning (e.g., ^? Kannst du bitte schon schwimmen? ‘Can you already swim, please?’). Indeed, bitte occurs significantly less frequently in the high-imposition context and in only 37% of the Can you VP? occurrences.^[31] In the high-imposition context an underspecified requestive meaning is thus preferred. The assumption that tentativeness is the preferred negative politeness strategy in face-threatening situations is further supported by the distribution of downtoners.

Downtoners indicating tentativeness such as vielleicht ‘possibly’, which signal the insecurity of S regarding the ability of H to fulfil the requested action, occur only in 2% of the Can you VP? requests in the low-imposition context. In the high-imposition context, downtoners such as (ganz) vielleicht, eventuell, ausnahmsweise ‘(quite) possibly, eventually, exceptionally’ occur with a higher frequency (24%, cf. 6a). As the logistic regression model shows, the probability of downtoners occurring in the low-imposition context is much higher overall (contrary to H3). However, it is especially the modal particles grade ‘just’, mal ‘once’, eben ‘simply/just’ (also in combination) which occur with high frequency in this context and especially with ability questions (69%, cf. 6b). These particles mark the casualness of the request rather than robbing the ability question of its point.^[32] Characteristically, these modal particles hardly occur in the high-imposition context (2.0%). Since downtoners that indicate tentativeness differ significantly in their occurrence from downtoners that indicate casualness within the same head act strategy (χ² (1) = 286.32, p < 0.001***, ϕ = 0.891), they should be treated separately in future studies.

The subjunctive, which occurs significantly more frequently in the high-imposition context, has a similar function to that of lexical downtoners that indicate tentativeness (H4).^[33] This morphological modifier further increases the optionality, i.e., the addressee’s freedom to decide whether or not to perform the action, and as a result the politeness.

As the data show, punctuation is another means of indicating illocutionary force salience that has rarely been discussed so far. Formally, CIRs are obviously interrogative sentences that should be marked by a question mark. However, we find deviations from this orthographically correct punctuation as (7) exemplifies.

Deviating from the norm, we find both full stops and exclamation marks (as well as combinations of question marks and exclamation marks). As one participant explicitly notes (7b), these are not (always) typos but at least sometimes deliberate decisions. Thus, these cases can be seen as an indication that the requestive meaning may override partial aspects of the form type of a sentence if the CIR is highly standardized.^[34] This non-interrogative punctuation moves CIRs a bit further towards explicitness, which should diminish their politeness value. Interestingly, we do not find this standard-defying punctuation equally frequently in both situations.

As Table 4 shows, full stops and exclamation marks occur slightly but nonetheless significantly more frequently than question marks in the low-imposition context (χ² (1) = 4.8405, p = 0.0278*, ϕ = 0.096).^[35] However, the effect size is extremely low since question marks, which would be in line with the written standard language, dominate in both contexts. A phonological experiment that contrasts intonation in questions for information with that in CIRs would be interesting and probably far more informative since intonation is not subject to codification. The fact that we can observe significant deviations from the norm in codified written language at all is nevertheless worth following up and would have to be further substantiated by corpus studies that investigate texts with a high amount of quasi-spontaneous speech.

In summary, the data – which of course contain only a very limited set of social contexts – support previous findings that negative politeness expressed by CIRs can be considered as cultural ideal in Germany (as well as Austria and German-speaking Switzerland).^[36] In weaker FTAs, as in the low-imposition context tested here, the highly standardized Can you VP? request fulfils this need. Internal modifiers that minimize the costs for H are frequent in this context, such as understaters or adverbs such as bitte that overtly mark the utterance as a polite request (in contrast to an order or a demand) and thus increase the requestive force salience. In contexts that require extra politeness (i.e., requests with a high degree of imposition) it is not primarily the head act strategy that is varied since Can you VP? is the default formula in both situations of the DCT (but note, however, that explicit impositives are significantly less frequent here than in the low-imposition context). In the highly face-threatening DCT situation, increased negative politeness is primarily realized by several internal modifiers (i.e., downtoners, the subjunctive as well as punctuation) which support the question for information meaning of the Can you VP? construction by overtly signalling tentativeness. By these means the “requestive force salience” is decreased.

5.2 External modification and alerters

As shown in Section 4.1, the occurrence of single and particularly multiple external modifiers is more probable in the high-imposition context (H8). The experimental data thus confirms what Held (1995: 314) has shown for French and Italian and Kranich et al. (2021: 106) have recently shown for German (and English): “the higher the weight of imposition, the more [external] mitigation speakers use”. Even in requests elicited via written DCTs, which are said to be shorter than requests in naturally occurring speech, bare requests are very untypical in the high-imposition situation. As expected, grounders occur with highest frequency in both contexts. However, with 73% they are typical of the high-imposition context in contrast to the low-imposition context (22%). Also hedges/cost minimizers (e.g., hoffentlich hab ich dich nicht geweckt ‘I hope I didn’t wake you’) and repairs (e.g., tut mir leid, dass ich dich so spät störe ‘sorry to bother you so late’) are more typical if the imposition has a higher weight (34%); these external modifiers hardly ever occur in standard requests with low imposition (1%). Regarding the expression of gratitude there is no difference between both conditions (7–9%). In sum, one could speak of pragmatic iconicity here: the higher the weight of imposition, the wordier the request must be. Although extra politeness is partly conveyed via the head act and its internal modification in German, it is important to examine external modification as well since extra politeness is particularly achieved via devices beyond the head act strategy.

Finally, let us turn to the role of alerters. As the logistic regression model has shown, the high-imposition context boosts the chances for the occurrence of a vocative. This result also conforms to the expectations described above (H9). Nevertheless, the result should not be over-interpreted, as the two tasks compared not only differ in terms of the degree of imposition, but also in terms of the communicative situation. In contrast to the low-imposition situation, in the high-imposition situation there is no face-to-face interaction, as the request should be made via telephone, which elicits a greeting by S. As Günthner (2016) discusses with reference to Goffman (1971), onymic vocatives in particular are typical in greeting sequences since they are interpersonal rituals used to indicate a period of increased accessibility.^[37] Thus, the difference in medium could also cause the higher frequency of vocatives here besides the higher degree of imposition. Additionally, the varying social distance may have an influence on vocative production. The social distance between S and H may have been perceived as lower in the high-imposition context since here the addressee was defined as a family member and not as a friend as in the low-imposition context. Lower social distance may boost the use of in-group marking vocatives such as terms of endearment. Indeed, in the high-imposition context we find mainly kin(ship) terms such as Schatz ‘darling’ or Mami ‘mummy’ (15%), but also pronouns (du ‘you’, 9%), which only occur in 1% of the utterances in the low-imposition context. The kin terms in particular can be seen as in-group markers that convey positive politeness. However, pronominal vocatives are also noteworthy in this context. With a certain intonation – indicated in writing by some participants by vowel doubling (e.g., duuuu) or the insertion of <h> (e.g., duhu) – these alerters have similar functions as the devices indicating tentativeness discussed in the previous section. Mitigating the threatening nature of a request by displaying restraint thus seems to be a prominent negative politeness strategy in German realized through various morpho-syntactic, lexical, and graphemic (as well as intonational) devices.