Corpus Pragmatics

, Volume 2, Issue 1, pp 57–82 | Cite as

Preference Organization and Cross-Cultural Variation in Request Responses: A Corpus-Based Comparison of British and American English

Original Paper

Abstract

This study compares request responses in two national varieties of English, American English (AmE) and British English. The responses are manually retrieved from two corpora of English, the British component to the International Corpus of English and the Santa Barbara Corpus of Spoken American English and analyzed both for their function and format. It is shown that request responses in both varieties pattern as preference structure would predict: compliant responses occur more frequently than non-compliant ones and can thus be understood as the preferred response type to requestive first pair parts. The analysis of co-occurrence patterns of first and second pair parts reveals that the directness level of the first pair part does not significantly influence the response type of the second pair part. It does, however, have an impact on the level of explicitness of the response. While cross-cultural differences in request responses are generally rare in the present database, they do surface in the higher numbers for implicit compliance in the AmE corpus. These results show the importance of a cross-cultural approach to the study of naturally occurring request responses retrieved from corpora of authentic speech.

Keywords

Request responses Preference organization American English British English Pragmatic variation 

Introduction

The present paper is a corpus-based contrastive investigation into verbal responses to requests in two national varieties of English: American English (AmE) and British English (BrE). With this research design, we aim at combining two research traditions which have mostly been carried out separately from each other. Firstly, while cross-cultural pragmatics (CCP) has investigated (and indeed found) regional and social differences in language use, it has focused heavily on first pair parts of adjacency pairs. This focus has (at least partially) been caused by the omnipresence of discourse completion task (DCT) data, which makes it almost impossible to study interactional aspects of language use. Secondly, the conversation analytic tradition, has, conversely, provided us with many insights on the sequential aspects of conversation, but has generally not included a cross-cultural perspective; neither has it attempted a quantitative evaluation nor a comparative perspective.

What we know about request sequences thus stems from studies which either (1) make use of data with questionable validity (DCT-based CCP) or (2) have offered accounts based on naturally occurring data which avoid cross-cultural comparisons altogether. The present study is an attempt to fill this research gap by employing a corpus-based contrastive approach. We rely on the conversational data retrieved from two corpora of English to obtain a database of request sequences large enough for both quantitative and qualitative analysis. Our objective is to answer the following research questions:
  1. 1.

    How are request responses realized functionally (in terms of preference organization) and formally (both in terms of explicitness and in lexico-grammatical terms) in naturally occurring conversations?

     
  2. 2.

    Is there a correlation between request and response, i.e. the realization of the second pair part contingent on the directness of the first pair part?

     
  3. 3.

    Do systematic cross-cultural differences occur between AmE and BrE request responses (from both a formal and functional point of view)?

     

We will first provide the reader with a short definition of requests and their responses, before reviewing studies investigating naturally occurring requests and their responses (see “Empirical Studies on Requests and Their Responses” section). We will then briefly outline the data collection procedure and justify the choice of corpora and their comparability (see “Methodology ” section) before discussing our results in detail (see “Results” section).

Empirical Studies on Requests and Their Responses

Requests and Request Responses

According to classic speech act theory, requests are defined as directive speech acts in which a speaker wants the hearer to do a future action which they would not have done without the production of the request (Searle 1976). While Austin’s (1962) original conceptualization of speech acts entails an interactive perspective (in that the illocutionary act has three effects including the production of an uptake and inviting of a response), Searle’s (1969) account of speech acts developed into a cognitive theory of speaker meaning. As a consequence, the interactional dimension of speech acts has been overlooked (cf. Sbisà 2013). The assumption that pragmatic meaning is created predominantly by the speaker is taken up by speech-act based approaches to verbal politeness. In Brown and Levinson’s (1987) theory, requests are conceptualized as intrinsic face threatening acts which are likely to be realized by indirect linguistic choices, including the use of politeness strategies.

As a result, both speech act theory and related theories of verbal politeness are unable to account for the interactional properties of requests. Thus, different explanatory frameworks, such as conversation analysis, need to be invoked. Conversation analysis (CA) has its roots in ethnomethodology and assumes that language is a social practice (Schegloff 1996; Drew and Curl 2008; Heritage 2008). From a conversation analytic perspective, requests are prototypically viewed as first pair parts of an adjacency pair which create the expectation of a second pair part (response) to occur (Schegloff 1972; Schegloff and Sacks 1973). Yet not all alternative second pair parts are equal in status; some might be preferred while others are dispreferred. Dispreferred second pair parts prototypically display structural markers of dispreference such as delays, qualifications and accounts (Pomerantz 1984; Levinson 1983; Walker 2013). Research on requests and their responses has shown unequivocally that indications of non-compliance are dispreferred second pair parts while indications of compliance are preferred (Levinson 1983; Davidson 1984; Lee 2011). Furthermore, the production of a second pair part can be an indicator that the addressee has understood (or failed to understand) the speaker’s intention (Sacks et al. 1974; Hutchby and Wooffitt 2006; Drew and Curl 2008). Second pair parts can thus serve as a monitoring device not only for speakers to check whether their intentions were identified correctly but also for researchers attempting to identify speech acts in authentic discourse.

Empirical Studies on Requests

While there is a wealth of request studies based on Discourse Completion Task data,1 there are only a few studies on requests which employ naturally occurring data. While there is some research based on parent–child interactions, institutional discourse or service encounters (e.g. Wootton 1981, 1997; Craven and Potter 2010; Pufahl Bax 1986; Vine 2009; Geluykens 2011; Félix-Brasdefer 2015) only three studies observe adult speakers in informal contexts. The earliest and most influential of these is Ervin-Tripp’s account of requests in (mostly) conversational American English (AmE). The author establishes a taxonomy of six request types based on formal, discursive, and social criteria. She distinguishes between need statements, imperatives, embedded imperatives, permission requests, request questions and hints which are used in different social situations and constellations. She reports that almost all request types can be used in a variety of contextual situations.2

In a more CA oriented approach to the study of requests, Curl and Drew (2008) compare requests in naturally occurring private telephone conversations to requests in telephone calls to a primary care physician’s practice in Britain. While their approach generally is a qualitative one, the authors report that two thirds of the requests in their conversational corpus are realized by modal verbs which either question or state the hearer’s ability or willingness to comply with the request (can/could/will) or imperatives. The authors find a different pattern of requesting in the institutional corpus (with modal expressions being almost absent and requests prefaced with I wonder being frequent) and conclude that speakers use the different request formats to “display or claim entitlement to make a request, and to display (or conversely, not acknowledge) an understanding of the contingencies associated with granting their request” (Curl and Drew 2008: 16). They argue that the use of modal verbs indicates that speakers treat their request as non-contingent in the sense that they “treat the conditions necessary for granting their request as fulfilled, and therefore their request as unproblematic” (Curl and Drew 2008: 27). In contrast, requests prefaced with I wonder if show the speaker’s awareness of the contingencies associated with the granting of the request. The authors conclude that by choosing a particular request format, speakers “make a claim as to what they believe themselves reasonably entitled to” (Curl and Drew 2008: 27) and thus display their social relationships through the use of certain language forms.

In a contrastive study comparing request for non-verbal action in American and British English (BrE) conversational data, Flöck (2016) also finds that imperatives and modal constructions which either question or state the hearer’s ability or willingness (‘preparatory’) constitute the most frequent request realization strategies for compliance. In her conversational data, imperative-based requests and elliptical structures (‘mood derivables’) alone account for 49.6% of all requests in the AmE and 39.6% in the BrE database (Table 1; cf. Flöck 2016: 121).
Table 1

Request strategies and directness levels in conversational requests

Directness level

Head act strategy

Example

AmE

BrE

N

%

N

%

1

Mood derivable

Do X

129

49.6

103

39.6

Obligation statement

You should/must do X

31

11.9

43

16.5

2

Performative

I request/ask you to do X

1

0.4

4

1.5

Want/need statement

I want/need you to do X

7

2.7

3

1.2

Preparatory

Can you do X?

53

20.4

65

25.0

Suggestory formula

How about doing X?

28

10.8

26

10.0

3

Hint

No codified form

11

4.2

16

6.2

Total

260

100.0

260

100.0

Similar to Curl and Drew (2008), modal constructions (“preparatory”) also belong to the most frequently employed requesting strategies. However, they occur with markedly lower frequencies than the imperative strategies (20.4% of all requests in AmE and 25.0% in BrE).

Empirical Studies on Request Responses

While the majority of research has been conducted on the first pair part, responses to requests have attracted at least some attention from researchers working in different paradigms. While some of these studies are more cognitive in nature (cf. e.g. Clark 1979; Clark and Schunk 1980), the ones that are most relevant to our purposes investigate responses sequentially as adjacency pair second parts in a CA framework. One of the few such response studies based on conversational non-institutional data in English is Rauniomaa and Keisanen (2012), who analyze requests and responses in naturally occurring casual conversations. They report that out of 69 request sequences they analyze, 56 are complied with and focus on this type of responses in their analysis. They report on two different multimodal formats of request responses. The first format consists of the request’s fulfilment only (without there being a verbal response). The vast majority of all favorable responses in their corpus follows this structure. The second format consists of two components, an acceptance (affirmative response token) and the actual fulfilment of the request. The authors report that, as some requests cannot be fulfilled immediately, the affirmative response tokens might indicate future rather than immediate compliance with the request.

Stivers and Rossano (2010) also investigate responses to requests from a CA perspective but focus on the relation between first and second pair part. They find that different first pair parts mobilize responses to different degrees. Their analysis shows that off-record request strategies are less likely to evoke a response by the addressee than more direct strategies. The authors invoke Brown and Levinson’s (1987) concept of a priori payoffs for off-record language use. They claim that speakers who use off-record request strategies are not only “less coercive of a particular type of response, but they are also less coercive of any response” Stivers and Rossano 2010: 26; original emphasis). Consequently, they conclude that the indirect nature of the off-record request has a lower mobilization potential than more direct first pair parts. This claim, however, still needs further investigation since the authors do not quantify their results and only present isolated examples as evidence.

Thompson et al.’s (2015) study contains the most elaborate account of request responses in naturally occurring conversation to date. The authors analyze responses in AmE request-for-action sequences and develop a taxonomy of response forms for request sequences. The authors distinguish between five different responsive actions in request-for-action sequences: (1) particle, (2) minimal clausal, (3) expanded clausal, (4) graded clausal and (5) unrelated clausal. The different responsive actions are distinguished from each other on formal grounds and differ in their degree of autonomy from the first pair part, the degree of the requestee’s accountability, and assumed agency (Table 2 gives an overview). While particle responses display the lowest degree of turn autonomy, unrelated clausal responses are (as the name suggests) completely unrelated to the prior request grammatically and thus display the highest degree of turn autonomy.
Table 2

Requestive action forms (partly based on Thompson et al. 2015: 223–224)

 

Particle

Minimal clausal

Expanded clausal

Graded clausal

Unrelated clausal

Linguistic form compliance

alright

okay

sure

I will, I’d be happy to, etc.

I won’t

I’ll X

I won’t X

[X = requester’s formulation of action]

I’ll Y

I won’t Y

[Y = alternative formulation of action]

[preparatory steps]

Linguistic form non-compliance

no

I’m not

I don’t want/like to, I like to, etc.

I’m not Xing

will X, I am Xing

[X = requester’s formulation of action]

n.a.

[reasons for non-compliance]

Turn autonomy

Agency

Claim to deontic rights

The authors dismiss both preference-organizational and minimization accounts as explanatory models for the relation between request and response, as both accounts are unable to explain the relatively high numbers for elaborate compliance responses in their AmE data. Instead, they employ a deontic rights approach to explain how first and second pair part are related. They operationalize deontic rights by invoking the concepts of entitlement and contingency (cf. also Lindström 2005; Heinemann 2006; Curl and Drew 2008; Craven and Potter 2010). While entitlement refers to how speakers position themselves with respect to having the right to make their addressee do an action, contingency relates to the requester’s awareness of the possibility that there may be factors which prevent the request from being complied with. Thompson et al. (2015) maintain that strong claims to deontic rights translate into a high display of entitlement and a low display of contingency. In turn, weak claims to deontic rights correspond to a low display of entitlement and a high display of contingency. Following Stevanovic and Peräkylä (2012), Thompson et al. (2015) suggest that participants in conversation strive to achieve a situation of deontic congruence in which a requester’s strong deontic rights are met with a display of weak deontic rights on the part of the requestee. As the authors relate the display of agency in responding to requests to a display of strong deontic rights, the response forms identified can be correlated with the pattern established. As particle responses display the lowest degree of agency, they also display the weakest claim to deontic rights and are thus the prime candidates for responding to a request with high display of deontic rights. Unrelated clausal forms, with their high display of agency and thus strong deontic rights, should then be expected to occur after requests displaying weak deontic rights (cf. Table 2; dark hue indicates strong realization of a criterion).

When it comes to the frequency distribution of the different response strategies, Thompson et al. (2015) find that particle responses are overall the most frequent responses (N = 40) followed by unrelated clausal responses (N = 27). Minimal clausal (N = 12), expanded clausal (N = 10) and graded clausal responses (N = 7) occur with markedly lower frequencies. In the two most frequent response types, a clear functional contrast can be identified: Whereas particle responses are predominantly used to signal compliance (36 compliance vs. 4 non-compliance strategies), unrelated clausal responses are mainly used to express non-compliance (9 compliance vs. 18 non-compliance strategies). For the other response types, the patterns are either not as clear-cut or the numbers are too low for generalizations.

Flöck’s (2016) study on naturally occurring data, finally, focuses primarily on the analysis of requests (first pair part), but uses responses as an identification criterion and offers a brief cross-cultural analysis of request responses in AmE and BrE. In contrast to Rauniomaa and Keisanen (2012), she finds that out of the 520 requests she analyzes, the majority are responded to verbally. With a focus on verbal responses only, she distinguishes between two response formats (compliance vs. non-compliance) and within each category between explicit and implicit responses, adding up to four different permutations. Flöck (2016) counts the production of agreement with the literal meaning of the request (e.g. yeah or yes) and verbal indications of compliance (e.g. I will) as explicit compliance strategies. Implicit compliance is mostly realized by responding with a request for specification. While explicit non-compliance includes disagreement (e.g. no) and verbal indications of non-compliance (e.g. I won’t), implicit non-compliance is achieved by providing reasons for the non-compliance, challenging the appropriateness of the requested action or the requester’s entitlement for the production of the request. In a brief cross-cultural analysis of request responses, she finds mild differences between AmE and BrE responses in that BrE speakers tend to use more explicit compliance strategies than their AmE counterparts. While the author does not include a systematic formal analysis of request responses, her functional analytical categories, in particular her distinction between explicit and implicit responses, also form the basis for the present paper.

To summarize, studies based on naturally occurring data have provided us with findings about the surface structures both requests and their responses can take. In contrast to disciplines such as cross-cultural pragmatics, existing studies to date have not studied request sequences from a contrastive perspective (with the exception of Flöck 2016). Thus, there is a regrettable lack of research on cross-cultural differences in the realization of both naturally occurring requests and their responses.

Methodology

Speech Acts and Corpus Linguistics

Corpus linguistic methodology originally was developed with the aim of electronically accessing representative samples of linguistic forms in large language databases. In contrast to most speech act studies, corpus linguistic research prototypically takes a form-to-function perspective, where linguistic forms constitute the starting point and can be used as search tokens for electronic searches. However, as all studies on requests have shown, this speech act can take numerous forms, ranging from realization strategies which are routinely associated with certain linguistic forms (e.g. imperatives) to those which are not codified at all (e.g. hints). Moreover, most of the forms realizing requests can also be employed to encode intentions other than requesting. Automated searches run the risk of not retrieving some requestive realizations while at the same time locating forms which are used with different intentions. Consequently, in their study on compliments in the British National Corpus (BNC), Jucker et al. claim that speech acts “are not readily amenable to corpus-linguistic investigations” (2008: 273). In a review on the use of corpora in pragmatics research, Rühlemann (2010: 289) comes to the same conclusion and adds that contextual information is only recorded “crudely” in corpora.

Thus, all corpus-based studies on speech acts so far have limited themselves to the investigation of their conventionalized or even formulaic linguistic realizations and were thus able to use automated searches. Such studies have thus focused on lexicalized stems of speech acts like apologies (Deutschmann 2006), suggestions (Adolphs 2008) expressions of gratitude (Jautz 2008; Wong 2010) or the bathroom formula (i.e. requests to leave an ongoing activity to go to the bathroom; Levin 2014). The authors acknowledge that while they were able to retrieve conventionalized realizations of the speech acts, they had no means of locating non-conventionalized strategies.

Despite the obstacles of automated corpus searches, we claim that corpora (or more specifically, their textual basis) do have a major potential even for research on non-conventionalized speech act realizations. As requests (like many other speech acts) can be considered to be low frequency items, only large databases can provide sufficient numbers for representative analyses. As a number of corpora have been specifically compiled for contrastive analyses, they are also well suited for cross-cultural pragmatic research. Corpus series like the International Corpus of English (ICE)—a corpus project which aims at representing different varieties of English and which is designed to facilitate contrastive corpus-based analyses between them—spring to mind most readily.

However, the problems with automated searches and the need for large databases in pragmatics research call for the development of pragmatically annotated corpora. The only corpus with pragmatic annotation we are aware of is the Irish component to ICE. The annotation system used (SPICE, Systems of Pragmatic Annotation in the Spoken Component of ICE-Ireland) includes corpus mark-up for speech act functions, utterance tags and discourse markers. The Searlean classification of illocutionary types is used as the basis for speech act annotation (cf. Kallen and Kirk 2012; Kirk 2016). While this system may still not be sufficient for employing automated searches for each pragmatic phenomenon, it provides a direction in which pragmatic corpus annotation may develop. For the time being, researchers interested in the study of non-codified pragmatic phenomena in varieties other than Irish English (including the present authors) have to be content with manual searches of corpus material.

Corpora Used and Coding Scheme

The database of the present paper consists of verbal responses to requests retrieved from two corpora of spoken English in manual searches (what Kohnen 2008: 296 calls a “genre-specific micro-analytic bottom-up” approach). The request sequences were identified not just by searching for the first pairs; rather, the interpretation of an utterance by co-participants was taken into consideration. Only those sequences were included in the present analysis where a verbal uptake was provided by a participant that marks the first pair part as a request. Overall, 171 AmE and 201 BrE request sequences with verbal responses were retrieved from the databases. Non-verbal utterances (of the type frequently found by Rauniomaa and Keisanen 2012) were not included in the analysis.

The AmE request responses were retrieved from a subcorpus of the Santa Barbara Corpus of Spoken American English (SBCSAE) while the BrE data were extracted from the conversational part of the British component to the International Corpus of English (ICE-GB). Each component of ICE consists of 500 text samples (300 spoken, 200 written) of 2000 words each, adding up to a total of one million words. ICE-GB is not only POS-tagged but also parsed. Further annotation includes the indication of overlap. The SBCSAE consists of 60 transcripts sampled with the aim of providing a source for researchers “interested in the nature of spoken American English, and more generally, of spoken language” (Chafe et al. 1991: 65) in descriptive, theoretical or pedagogical contexts. The SBCSAE does not only offer orthographic transcriptions of the data but also intonational mark-up, overlap and pauses.

While there are larger and more up-to-date corpora available (such as the British or American National Corpus or The Bank of English), the two corpora were chosen for a number of reasons for the purposes of the present study. Unlike many other corpora, ICE-GB and the SBCSAE provide researchers with a wealth of demographic background information on speakers, enabling them to control for many macro-social factors. This is a crucial prerequisite for pragmatics research as it has become clear to researchers that many pragmatic phenomena are sensitive to macro-social variation. As the two corpora (ICE and SBCSAE) are not per se comparable when it comes to the discourse types included, subcorpora were sampled to guarantee maximum comparability between the language samples included (cf. Table 3). While only the private dialogue transcripts of ICE-GB were chosen for analysis (N = 100), all transcripts containing scripted language material (such as sermons or guided tours) were excluded from the SBCSAE.
Table 3

Subcorpora used in the present study

 

AmE: Subcorpus SBCSAE

BrE: Subcorpus ICE-GB

Medium

Spoken

Spoken

Variety

American English

British English

Size

211,932 words

213,659 words

Sampling period

1990s

1990–1993

Discourse types

Informal conversations: face-to-face (96%), telephone (4%)

Informal conversations: face-to-face (90%), telephone (10%)a

aIn order to rule out influences of the different proportion of discourse type in the two subcorpora, the distribution of response types and strategies is compared across the face-to-face and complete database. The statistical testing reveals no significant differences between the face-to-face request responses and the whole database in either variety (AmE: χ2 (3) = 0.111, p > 0.5; BrE: χ2 (3) = 0.037, p > 0.5). Thus we conclude that the discourse type does not have a significant impact on the realization of the request response. As the number of telephone request responses is too low in both subcorpora (AmE N = 6, BrE N = 18), no statistical tests were run on those data

With the help of the demographic information provided, the subcorpora sampled are kept comparable not only for discourse type included but also for macro- and micro social factors. Overall, the subcorpora are balanced for gender, age, ethnicity and educational level of the speakers and include conversations with equal power and low social distance speaker constellations.3 Overall, we can thus conclude that the databases are comparable and that any pragmatic variation found must be due to cross-cultural variation.

The coding scheme employed in the present study is based on the analytical categories established in previous studies in request responses. More specifically, we have incorporated the analytical categories used in Flöck (2016) and supplemented them with the formal response categories established in Thompson et al. (2015). We thus distinguish between two response types (compliance vs. non-compliance) which each include two categories (explicit vs. implicit). Within these categories, we further distinguish between different response strategies which can then, in turn, be classified according to the formats that Thompson et al. have reported. Table 4 provides an overview of our coding scheme and gives an example for each category used in our database.
Table 4

Coding scheme of request responses applied in the present study

Category

Strategy

Format

Example

Compliance

Explicit

Agreement

Particle

Yeah

Minimal clausal

A: We should go soon B: We should

Unrelated clausal

A: We can put a pillow over her head.—B: It’s for her best

Indication of compliance

Minimal clausal

A: Make sure you don’t even have a cup of coffee. B: I won’t

Expanded clausal

A: You wanna open that or wait until later? B: I’ll wait until later

Graded clausal

A: Check it out. B: I’ll probably go between coastal

Unrelated clausal

A: You’ll have to [show a picture]. B: I have one somewhere

Implicit

Request for specification

Unrelated clausal

A: Cut me a slice there. B: This big?

Positive evaluation

Unrelated clausal

A: Let’s get away. B: I’d love to

Follow-up request

Unrelated clausal

A: Yeah, I noticed on the case for the tape recorder too, that it’s all covered today. Look right here. B: Look at your cigarettes even

Delegating request

Unrelated clausal

A: Can we take this to the table? B: Take everything to the table, C

Indication of postponing

Unrelated clausal

A: Put the rest in there. B: Just a minute

Non-compliance

Explicit

Disagreement

Particle

A: Would you turn on the light? B: No

Unrelated clausal

A: Just shut up and listen to me. B: I don’t think there is that much

Indication of non-compliance

Minimal clausal

A: Can’t you stand up and hold it? B: I can’t a

Graded clausal

A: Look it up. B: I can’t be bothered

Implicit

Reason for non-compliance

Unrelated clausal

A: Aren’t you guys gonna stick up for me? B: He’s bigger than I am

Challenge

Unrelated clausal

A: Turn em [hearing aids] up. B: Don’t tell a great man what to do

Alternative action

Unrelated clausal

A: Do you want to go over X with me? B: Actually, let’s go over Y

aThe only examples of these strategies in the database are prefaced with a disagreement particle and are counted as combination strategies in our analysis. They are listed without the particle for illustration purposes here

Statistical Analysis

For the statistical analysis we used Generalized Linear Mixed Effects Models (GLMMs). We used the RStudio 1.0.136 (RStudio Team 2015) built on the R software (R Core Team 2013, version 3.3.2) and the R package “lme4” (Version 1.1.12) with the function “glmer” to perform GLMMs on our data to model both participant- and item-variability (Bates et al. 2015). GLMMs allow for differences between participants and items.

GLMMs are particularly suited for the present study as they are an extension of multiple regression and thus follow the aim to predict categorical outcomes based on predictor variables (fixed factors) and their interactions (cf. Bolker et al. 2009). GLMMs also allow for the inclusion of random variables in the model to control for effects that stem from either the language material included or speaker variance and thus are able to avoid what has been labelled the “language-as-fixed-effect fallacy” (Clark 1973). As early as 1973, Clark (1973: 335) criticizes that researchers have all too often generalized their findings “beyond the specific language materials they have chosen” and thus ignored the influence of variables like individual speakers on the whole data set. In a similar vein, Manning (2007: 1) argues that ignoring random effects by not modelling “the often significant correlation between data coming from one speaker” can lead to underestimation of standard error estimates and thus invalid significances.

We used two models for our analysis: While response type (compliance vs. non-compliance) served as the dependent variable in the first model, response category was used as the dependent variable in the second model. Possible fixed factors included variety (AmE vs. BrE), directness of the request, response type and the interaction between variety and response type. For random factors, participant and item were selected. In order to arrive at the best model, we first started out with a model that had a single random factor and then added additional random factors and random slopes. The model fit in terms of inclusion of random effects and fixed factors was assessed by comparing the Akaike-Information-Criterion values of fitted models (AIC; Akaike 1998) using Likelihood ratio tests (utilizing the anova-function). A decrease of two points in AIC values indicates a significant difference between two minimally differing models. The generalized mixed effects models were fitted with the Gauss-Hermite quadrature. We first contrast-coded the two levels of response type and variety (−1 and +1 for the two conditions) and the three levels of request directness (−1, 0, +1 for the three conditions) to allow for testing the overall effect of the fixed factors. Additionally, we calculated the variety × response category interaction in order to assess whether variety has a different impact than response category.

In the end, the following models were established to have the best fit and were thus selected for the analysis: In Model (1) response type (compliance vs. non-compliance) served as the dependent variable and variety (AmE vs. BrE) and directness of the request (as operationalized in three directness levels, cf. “Empirical Studies on Requests” section) were included as possible fixed factors. In Model (2), response category (explicit vs. implicit) served as the dependent variable with variety, response type, directness of the request and the interaction between variety and response type as fixed factors. In both models, the random factor of participant was included whereas item did not increase model fit.

Post-hoc comparisons on the levels of the interaction (in terms of the interaction between variety and response type in Model 2) were carried out using the R-package ‘phia’ on the basis of multiple comparisons of factor contrasts (de Rosario-Martinez 2015). To adjust for multiple comparisons, the Holm–Bonferroni method was applied. The Holm–Bonferroni correction reduces the possibility of obtaining significant results (Type-I errors) when performing multiple tests.

Results

Response Types: Preference Organization

As a first rough estimation of preference organization, we have investigated the relative frequencies of preferred (i.e. compliant) versus dispreferred (i.e. non-compliant) responses in our database. Figure 1 shows the frequencies for both the British and American data.
Fig. 1

Distribution of response types: compliance versus non-compliance

Not very surprisingly, in line with preference organization expectations and Thompson et al.’s (2015) findings, the results show a clear preference for compliance over non-compliance: over 70% of requests are complied with in both data sets. While the frequency of compliance is slightly higher in the American than in the British data (72.5 and 70.6%, respectively), the factor of variety does not have a significant impact on the choice of response type (β = 0.07185, SE = 0.25738, Z = 0.279, p = 0.7801). In other words, no cross-cultural variation is found with regard to preference organization.

Things change slightly when one takes into account the response categories, i.e. whether a response is uttered explicitly or implicitly (see Table 4 above). Explicit responses signal (non-)compliance in a semantically unequivocal and transparent manner. Signaling compliance explicitly can be achieved either through an explicit agreement signal, as in (1), or through another overt and non-ambiguous compliance signal, as in (2):

(1)

MARILYN: .. Peter, would [you like to] … string the beans?

 

     PETE:                                             [What can I do]. Sure a.

aThe transcription detail has been simplified in the examples. Transcripts with (anonymized) participant names are taken from our AmE subcorpus.

(2)

B: Well make sure you don’t even have a cup of coffee before you set foot on it.

 

     A: Well I won’t because it will weigh extra won’t it

Indicating non-compliance explicitly can likewise be done through the use of an overt disagreement signal, as in (3), or another clear semantic indication of non-compliance, as in (4):

(3)

MARLENA: Would you turn on the light?

 

   KITTY:         No,

 

   MARLENA:  …     Why.

(4)

A: Look it up

 

    C: Triple the price

 

    C: No I can’t be bothered

Implicit responses can be realized through a variety of different strategies (cf. Table 4 above for an overview), What these strategies have in common is that there is no clear, overt signal of (non-)compliance in the response; rather, the requester has to work out whether the requestee is complying or not based on contextual cues as well as on the response itself, through a conversational implicature. In example 5, for instance, the requestee utters a request for further specification as a response to the original request; in doing so, he implies that he is willing, in principle, to comply with the request once this specification query is resolved:

(5)

MARILYN: You could use the lettuce washer cause Pete’s using the colander.

 

   ROY:          .. Where’s the lettuce washer.

In example 6, the requestee implies non-compliance by giving the reason why compliance will not be forthcoming; though there is no overt non-compliance signal in the response, the requester should be (and as the example shows is indeed) able to work out the dispreferred nature of the response through an implicature:

(6)

A: Well do shut the door

 

   B: He’s coming back

 

   A: Yeah I know. But … sound travels

Politeness considerations would predict that non-compliant responses are preferably done implicitly, since refusing to comply with a request constitutes a challenge to the requester, which is a face-threatening act in its own right. Compliant responses, on the other hand, do not constitute such a face threat, which should allow the responder to formulate them in a fairly direct and explicit manner.

Figure 2 does indeed show a clear correlation between the type of response (compliance vs. non-compliance) and the strategy type (explicit vs implicit) employed. In both the British and the American data sets, non-compliant, dispreferred responses are overwhelmingly realized through an implicit strategy (87.2 and 93.2%, respectively), while preferred responses are most often realized through an explicit response strategy.
Fig. 2

Distribution of response categories: explicit and implicit responses

This pattern does not apply to the same degree to both data sets: whereas British responders overwhelmingly opt for an explicit strategy (86.6%), Americans do so to a lesser extent (62.9%). In other words: American responders tend to opt more frequently for an implicit strategy (37.1%) when complying with a request than their British counterparts (13.4%). The statistical analysis reveals that the factor of variety (AmE vs. BrE) (β = −3.7382, SE = 0.4341, Z = −8.612, p = 0.0001) and its interaction with response type (compliance vs. non-compliance) (β = 2.3211, SE = 0.2559, Z = 9.072, p = 0.0001) have a significant influence on the choice of response category (explicit vs. implicit).4 The post hoc comparisons on two levels of the interaction (AmE and BrE on compliance and AmE and BrE on non-compliance) show that it is the influence of the compliance strategies that affect the choice of response type significantly [compliance: χ2 (1) = 15.7395, p < 0.001; non-compliance: χ2 (1) = 0.8568, p > 0.05]. We thus find significant cross-cultural variation in the preference for explicit or implicit responses.

The results for our American speakers are somewhat in line with those found by Thompson et al. (2015) for their American English data. The authors find that the majority of their compliant responses are realized by particle, minimal and expanded clausal responses (which roughly correspond to explicit compliant responses) but also report that unrelated clausal responses (i.e. implicit strategies) occur with some frequency.

Compliance Strategies and Their Linguistic Correlates

For our detailed analysis of the formal aspects of explicit compliance, we will correlate our main explicit compliance strategies (agreement, indication of compliance, or a combination of the two, cf. Table 4 in “Methodology” section) with the response formats identified by Thompson et al. (2015) (cf. Table 2 in “Empirical Studies on Requests” section). Table 5 gives an overview of the explicit strategies and their formal realization.
Table 5

Formats for explicit compliance strategies (following Thompson et al. 2015)

Strategy of explicit compliance

Format

AmE

BrE

n

%

n

%

Agreement

Particle

61

93.8

90

91.8

Minimal clausal

0

0.0

2

2.0

Expanded clausal

1

1.5

0

0.0

Graded clausal

0

0.0

1

1.0

Unrelated clausal

3

4.6

5

5.1

Subtotal

65

100.0

98

100.0

Indication of compliance

Minimal clausal

1

16.7

2

13.3

Expanded clausal

2

33.3

2

13.3

Graded clausal

1

16.7

6

40.0

Unrelated clausal

2

33.3

5

33.3

Subtotal

6

100.0

15

100.0

Agreement particle + indication of compliance

Minimal clausal

3

42.9

3

30.0

Expanded clausal

1

14.3

2

20.0

Graded clausal

1

14.3

4

40.0

Unrelated clausal

2

28.6

1

10.0

Subtotal

7

100.0

10

100.0

As we have seen in “Response Types: Preference Organization” section, explicit compliance is significantly more frequent in British English (N = 123) than in American English (N = 77). By far the most common strategy for explicitly complying with a request is simple agreement, which is most frequently realized through a single agreement particle in both data sets. Since no face work is necessary for such second pair parts, this strategy provides the responder with the most economical and transparent response option. For similar reasons, occurrences of unrelated clausal strategies are relatively rare.

The type of agreement particle chosen is to some extent dependent on the variety of English, as shown by Table 6 (which only shows those particles used in isolation, i.e. not the ones used in combination with some other form of compliance).
Table 6

Agreement particles used in the data sets

Agreement particles

AmE

BrE

n

%

n

%

yeah

20

32.8

34

37.8

yes

1

1.6

19

21.1

mhm/mm

8

13.1

10

11.1

right

1

1.6

7

7.8

combinations including yeah/yes

0

0.0

6

6.7

okay/ok

18

29.5

5

5.6

I know

2

3.3

4

4.4

sure

6

9.8

0

0.0

hm

3

4.9

0

0.0

Other

2

3.3

5

5.6

Total

61

100.0

90

100.0

By far the most common agreement token, generally speaking, is yeah: it is most common in both British and American English, though slightly more so in the former (38 vs. 33% of all particles, respectively). Next in line in terms of frequency are yes (N = 20), okay/ok (N = 23), and mhm/mm (N = 18), which exhibit some clear cross-cultural variation: while mhm/mm is used similarly in both varieties, yes is almost exclusively used in British English, whereas okay/ok is typically used by American speakers. All other particles occur relatively infrequently, though some appear to be (almost) exclusive to one variety (e.g. right for British English, sure for American English). Summing up, we can say that the choice of agreement particles exhibits clear cross-cultural variation.

As we saw in “Response Types: Preference Organization” section, implicit compliance is more frequent in our American data set. A closer look at the formal realizations (again using Thompson et al.’s (2015) categories) of this implicit compliance, and at its distribution over our various response categories (cf. Table 4 above) might shed some further light on its precise nature. Results can be found in Table 7.
Table 7

Formats for implicit compliance (following Thompson et al. 2015)

Strategy

Format

AmE

BrE

n

%

n

%

Request for specification (preparatory action)

Unrelated clausal

38

82.6

17

89.5

Positive evaluation

4

8.7

2

10.5

Follow-up request

2

4.3

0

0

Delegating request

1

2.2

0

0

Indication of postponing

1

2.2

0

0

Subtotal

46

100

19

100

Table 7 shows, first of all, that all implicit compliance is realized through an unrelated clausal format. It was established (in “Empirical Studies on Requests and Their Responses” section) that such forms, with their relatively high degree of agency, can be expected to occur after requests displaying weak deontic rights. This certainly seems to be the case for some strategies. When a responder indicates he/she wants to postpone the requested action, for instance, this is a clear signal that he/she has a strong deontic right; similarly, an expression of positive evaluation implies that the responder has a strong sense of entitlement, as he/she can afford to express his/her positive feelings about the requested action (if requester entitlement is strong, the responder’s evaluation of the request will be irrelevant).

However, as can be seen in Table 7, the vast majority of implicit compliant responses concern the strategy request for specification, in both data sets (83 and 89% of such responses, respectively). In these sequences, the responder counters the request with a request for further specification; a form of preparatory action is thus necessary for the successful carrying out of the initial request, as in:

(7)

KEVIN:    Cut me a slice there,

 

  KENDRA: This big?

Such responses were categorized as indirectly compliant, by virtue of the fact that, by asking for further specification, the responder implies his/her willingness to carry out the request eventually, as soon as the trouble spot caused by the lack of specification in the initial request is resolved (this interpretation is supported by closer inspection of the further interaction, which usually confirms that the request is complied with).

The implicit nature of the response, and its unrelated clausal format, in these exchanges appears to be caused by a trouble spot (i.e. lack of sufficiently specific information) in the immediate context of the request sequence, which needs to be resolved before the (deferred) requested action can be carried out. In other words, the successful completion of the request sequence is contingent on the resolution of the trouble spot. It seems difficult here to claim that the responder has strong deontic rights and that the speaker has a low level of entitlement; rather, the fact that the responder requests further specification seems to indicate that the responder does not question the deontic right of the speaker (i.e. his/her entitlement) to formulate the request in the first place. What we appear to be dealing with here is a situation with both high entitlement by the requester combined with high contingency. This would seem to argue against the inverse correlation between entitlement and contingency claimed by Thompson et al. (2015).

Non-compliance Strategies and Their Linguistic Correlates

Turning now to non-compliance, we have already established that this is mostly realized implicitly. Explicit non-compliance is both strongly dispreferred and highly face-threatening, and would indicate a strong sense of entitlement on the part of the responder, i.e. a signal that the speaker has no right to request the action. Even so, since non-compliance is dispreferred, one would expect the response to contain some redressive action in order to minimize the face threat. Explicit responses, especially those consisting only of a disagreement particle, would not be expected to occur often as non-compliance signals; rather, they are the prototypical compliance tokens, as shown in the previous section.

Table 8 does indeed show once again that explicit non-compliance (especially realized through an isolated disagreement particle) is quite rare (10 out of 96 instances overall, or 10.4% of non-compliant responses). Only 4 instances of isolated disagreement occur (none in the British data) and, interestingly, in 2 of those cases the disagreement particle is in fact followed by a challenge of the response by the original requester, as in:
Table 8

Formats non-compliance strategies (following Thompson et al. 2015)

 

Strategy

Format

AmE

BrE

n

%

n

%

Explicit

Disagreement particle

Particle

4

66.7

0

0.0

Disagreement particle + indication of non-compliance

Minimal clausal

0

0.0

2

50.0

Graded clausal

0

0.0

1

25.0

Disagreement particle + reason for non-compliant

Unrelated clausal

2

33.3

0

0.0

Disagreement

Unrelated clausal

0

0.0

1

25.0

Subtotal

6

100.0

4

100.0

Implicit

Reason for non-compliance

Unrelated clausal

24

58.5

39

70.9

Challenge

Unrelated clausal

17

41.5

15

27.3

Alternative action

Unrelated clausal

0

0.0

1

1.8

Subtotal

41

100.0

55

100.0

(8)

MARLENA: Would you turn on the light?

 

    KITTY:        No.

 

    MARLENA: … Why.

In all other instances of non-compliance, the responder chooses to do so implicitly, either by stating the reason for not wanting to comply with the request, as in example 9, or by challenging the speaker’s right to formulate the request, as in 10:

(9)

B: We should maybe just leave a message here saying head over

 

     A: She won’t bother coming then though

(10)

     JO: ..  Now you--Have you got your hearing aids [in].

 

WESS:                                       [Ye]s dear.

 

JO:      Well turn em up. … That’s enough.

 

WESS: Don’t tell a great man what to do.

In terms of Thompson et al.’s (2015) deontic rights analysis, these two implicit strategies appear to be quite different. In challenging the speaker, the responder does of course appear to indicate that the former does not have the deontic right, or entitlement, to ask the addressee to carry out the requested action. Giving reasons for non-compliance, on the other hand, cannot be interpreted simply in terms of lack of the speaker’s deontic rights. While in some cases the reason for non-compliance might make reference to the requester’s lack of entitlement (and hence the responder’s deontic right to refuse to carry out the action), in other instances the reason seems to have to do with the responder’s ability to carry out the request (i.e. the epistemic dimension) or, more generally, the feasibility of the requested action. In example 9, for instance, the responder appears to indicate that complying with the request would be futile rather than to indicate that the requester is not entitled to ask for the requested action. One could also formulate this, once again, purely in terms of contingency: since the carrying out of the requested action is contingent on the reason(s) mentioned in the implicit refusal, the degree of contingency is so high that compliance is simply not an option (whether this is objectively the case is neither here nor there; it is rather the responder’s estimation of the non-feasibility which is the deciding factor here).

Overall, the detailed analysis of request response types, categories, strategies and their linguistic correlates shows that correlating the functional and formal levels yields results which might go unnoticed in a purely formal analysis. As we have discussed both for implicit compliance and non-compliance strategies, functionally diverse strategies (in terms of deontic rights) can be realized by unrelated clausal formats. The inverse correlation between entitlement and contingency postulated by Thompson et al. (2015) and its application to claims of agency in request sequences thus appears to be an oversimplification.

Correlation Between Directness of the Request and Response Type

As a final step in the analysis, we have investigated whether there is any link between the directness level of the request and the type of response it gets. Three directness levels are differentiated here (cf. Table 1 in “Empirical Studies on Requests and Their Responses” section): While directness level 1 (DL1) comprises the most direct head acts, directness level 3 (DL3) includes off-record realizations (hints). Directness level 2 (DL2) includes more indirect but still on-record realizations.

This part of the investigation has two components. First of all, we need to check whether directness levels have an influence on preference organization (Fig. 3). Put differently: does the level of directness of the request decrease or increase the likelihood of it being complied with? Figure 3 shows that (in-)directness (in contrast to what Stivers and Rossano 2010 find) has no influence whatsoever on preference organization: not only is compliance overall more frequent than non-compliance, it is also equally frequent for both directness levels. The Likelihood ratio testing revealed that leaving directness out of the model as a fixed factor to explain the choice of response type increased model fit. Preference organization, it can be concluded, is not influenced at all by the form of the request.
Fig. 3

Co-occurrence patterns of response type and first pair directness level

The statistical analysis further reveals that the directness level of the request does not have an influence on the choice of response category in terms of explicitness/implicitness. The Likelihood Ratio testing reveals that leaving directness and the interaction between directness and variety as fixed factors out of the model increased model fit. Entering directness as a single factor to the model did not show significant influence of directness on the choice of response type, either (β = 0.1358, SE = 0.1959, Z = 0.693, p = 0.488).

These findings are quite surprising as earlier studies (e.g. Stivers and Rossano 2010) predict that indirect requests would mobilize different kinds of responses. This claim is clearly not supported by our analysis. As we have analyzed verbal request responses only, we cannot say anything about the authors’ finding that indirect requests are furthermore less likely to be responded to.

Discussion

Our empirical analysis has revealed some clear correlations between request response format (compliance vs. non-compliance) and response type (explicit vs. implicit). First of all, in line with preference organization expectations, compliant responses are more likely to occur than non-compliant ones. Secondly, we have shown that, while compliant responses are likely to be realized explicitly, responders tend to non-comply in a more implicit manner. Given that non-compliance is a dispreferred option, and that it results in increased face threat, this also confirms our intuitions, and is in line with claims found in both the conversation-analytic and politeness literature.

However, our AmE responses do not follow this pattern as clearly as the BrE data in that American speakers comply significantly more often with requests in an implicit manner than their British counterparts; they also tend to be slightly more explicit when not complying. Apart from the deviance from preference structure predictions in AmE, further differences occur on the lexical level in the choice of agreement particle and in the individual strategies for realizing explicit and implicit (non-) compliance. Furthermore, our analysis of co-occurrence patterns between first and second pair part reveals that the form of the request does not appear to have an impact on the choice of response strategy.

The relatively high number of implicit compliance strategies in our AmE database is somewhat puzzling, and raises the question as to why responders would feel the need to avoid explicitness in a response which appears to be non-face-threatening (unlike non-compliance, which is intrinsically face-threatening). The explanatory model offered for the high number of implicit compliance strategies by Thompson et al. (2015), based on deontic rights, cannot fully account for the patterns found in our data. As was shown in “Results” section, the majority of implicitly compliant responses contain a request for specification. This indicates that complying with the request is contingent on the resolution of a trouble spot, rather than signaling low entitlement on the part of the speaker. Rather, the requester does have a high degree of entitlement, but the request is also highly contingent on other factors. Similarly, some non-compliant implicit responses cannot be interpreted solely in terms of lack of entitlement by the requester; rather they signal the responder’s inability to carry out the request (which may, again, be due to some extraneous, non-deontic contingency).

Thompson et al.’s (2015) deontic rights approach is further unable to account for our finding that the directness level of the first pair part does not have a significant impact on the type and strategy chosen for the second pair part (cf. “Correlation Between Directness of the Request and Response Type” section). A deontic rights approach would predict that direct first pair parts (i.e. requests that display strong deontic rights) should be more likely to be responded to with more implicit second pair parts as they display weak deontic rights.

The present study has mainly made use of a functional coding scheme, differentiating between response types (compliant vs non-compliant) and response category (explicit vs implicit). This was compared to the formal coding scheme employed by Thompson et al. (2015), which distinguishes between five formally identifiable categories which in turns are presumed to correlate with different levels of turn autonomy.

An examination of the similarities and differences between our functional response categories and Thompson et al.’s (2015) formal categories reveals some interesting results. First of all, as expected, the two coding schemes are congruent for implicit responses, in that they are all realized as unrelated clausal responses. It could hardly be otherwise, since all other categories in Thompson et al.’s (2015) classification generate responses which, in our functional terms, would be labeled explicit by definition.

The situation is different, however, for explicit responses. Thompson et al.’s (2015) deontic rights approach would predict that low deontic rights are signaled though the use of compliant responses with low turn autonomy and display of agency (i.e. particle responses), while explicit non-compliance should be indicated by forms displaying high turn autonomy (i.e. unrelated clausal responses). While in our data the majority of explicit compliant responses are indeed realized predominantly by particle forms (cf. Fig. 4), a significant portion are realized through so-called graded and unrelated clausal strategies (see also Table 5). According to Thompson et al. (2015), the latter two strategies are more autonomous and display a higher degree of agency, and would thus not be predicted for explicitly compliant responses, which should be associated with a display of lower deontic rights.
Fig. 4

Formal correlates of functional compliance response categories

In other words, in our data compliance may be signaled explicitly on a functional level while employing a formally ‘indirect’ formulation such as an unrelated clause. This is the case in example (11), where agreement is explicitly signaled through turn completion:

(11)

ANNETTE: We should just get one for both of our birthdays. Like maybe

                   [middle of the week]

 

ALICE:         .. sometime during the week, for both of our birthdays-. Cause

 

              I won’t be here for mine either.

The formal realization of the response here does not accurately reflect its functional response category. Put differently: formal realizations alone are not a reliable indicator of language functions (such as expressions of deontic rights). Moreover, we have already argued, when discussing non-compliance (in “Non-compliance Strategies and Their Linguistic Correlates” section), that deontic rights alone cannot explain the choice for a particular response strategy.

The role of deontic rights in request sequences, and in particular the respective impact of the notions of entitlement and contingency, certainly deserves further investigation. This also holds true for claims made on the basis of deontic rights about the influence of the first pair part on the second pair part. As empirical research on the interaction between request and request response is scarce, it is difficult to extrapolate from our present findings. More research on the interaction between first and second pair parts (regardless of specific illocutions) is needed in order to arrive at more robust generalizations.

Conclusion

To summarize, this paper has analyzed the functional and formal correlates of request responses from a cross-cultural perspective. While most of our findings are in line with expectations regarding preference organization and degree of explicitness, some results run counter to expectations. For instance, a higher number of implicit compliant responses was found than expected. We have argued that a deontic rights approach, such as the one proposed by Thompson et al. (2015), cannot fully account for such patterns. Other, non-deontic variables such as the contingency of the response on the resolution of some inherent trouble spot play at least a partial role.

In the present paper, we have employed a corpus-based approach with two (as we have argued) maximally comparable subcorpora for our contrastive research design. The corpus approach was specifically chosen in order to obtain enough request sequences for qualitative and quantitative analyses. For some low-frequency features, like indirect requestive first pair parts (i.e. hints), the numbers were still very low. In our analysis of co-occurrence patterns, these sequences could thus not be included systematically. This is unfortunate, as it is especially such low-frequency features which may reveal patterns which might not show in the dataset otherwise. In order to include such phenomena in a systematic analysis and still be able to control for participant homogeneity, even larger spoken language corpora would be needed that provide researchers with the detailed participant information needed for sociolinguistic comparisons. As pragmatic phenomena more often than not cannot be retrieved in automated searches, appropriate annotation systems such as SPICE are needed in order to obtain sufficiently large databases for analysis. In other words, much remains to be done in order to apply the advantages of a corpus approach to the study of language use.

Footnotes

  1. 1.

    For DCT-based studies on English cf. e.g. Blum-Kulka et al. (1989), Breuer and Geluykens (2007), Barron (2008).

  2. 2.

    The request taxonomy introduced by Ervin-Tripp (1976) is also used in Goldschmidt’s (1998) account of what she calls ‘favor asking’ sequences in American English.

  3. 3.

    See Flöck (2016) for a more thorough description of the database. As the current database of responses stems from the same subcorpora, the information given there also apply to the present study.

  4. 4.

    The factors response type (β = −0.5347, SE = 1.1437, Z = 0.468, p = 0.6401) and directness of the request (β = −0.1260, SE = 0.2395, Z = −0.526, p = 0.5988) did not have a significant effect on response category.

References

  1. Adolphs, S. (2008). Corpus and context: Investigating pragmatic functions in spoken discourse. Amsterdam: Benjamins.CrossRefGoogle Scholar
  2. Akaike, H. (1998). Information theory and an extension of the maximum likelihood principle. In E. Parzen, K. Tanabe, & G. Kitagawa (Eds.), Selected papers of Hirotugu Akaike (pp. 199–213). New York, NY: Springer.CrossRefGoogle Scholar
  3. Austin, J. L. (1962). How to do things with words. Oxford: Oxford University Press.Google Scholar
  4. Barron, A. (2008). The structure of requests in Irish English and English English. In K. P. Schneider & A. Barron (Eds.), Variational pragmatics: A focus on regional varieties of pluricentric languages (pp. 35–67). Amsterdam/Philadelphia: Benjamins.CrossRefGoogle Scholar
  5. Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1), 1–48.CrossRefGoogle Scholar
  6. Blum-Kulka, S., House, J., & Kasper, G. (1989). Cross-cultural pragmatics: Requests and apologies. Norwood, NJ: Ablex.Google Scholar
  7. Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W., Poulsen, J. R., Henry, M., et al. (2009). Generalized linear mixed models: A practical guide for ecology and evolution. Trends in Ecology & Evolution, 24(3), 127–135.CrossRefGoogle Scholar
  8. Breuer, A., & Geluykens, R. (2007). Variation in British and American English requests a contrastive analysis. In C.-C. Pragmatics & I. Pragmatics (Eds.), Bettina Kraft and Ronald Geluykens (pp. 107–126). München: Lincom Europa.Google Scholar
  9. Brown, P., & Levinson, S. C. (1987). Politeness. Some universals in language usage. Cambridge: Cambridge University Press.Google Scholar
  10. Chafe, W., Du Bois, J. W., & Thompson, S. A. (1991). Towards a new corpus of spoken American English. In K. Aijmer, J. Svartvik, & B. Altenberg (Eds.), English corpus linguistics: Studies in honour of Jan Svartvik (pp. 64–82). London: Longman.Google Scholar
  11. Clark, H. H. (1973). The language-as-fixed-effect fallacy: A critique of language statistics in psychological research. Journal of Verbal Learning and Verbal Behavior, 12(4), 335–359.CrossRefGoogle Scholar
  12. Clark, H. H. (1979). Responding to indirect speech acts. Cognitive Psychology, 11, 430–477.CrossRefGoogle Scholar
  13. Clark, H. H., & Schunk, D. H. (1980). Polite responses to polite requests. Cognition, 8, 11–143.CrossRefGoogle Scholar
  14. Craven, A., & Potter, J. (2010). Directives: Entitlement and contingency in action. Discourse Studies, 12(4), 419–442.CrossRefGoogle Scholar
  15. Curl, T. S., & Drew, P. (2008). Contingency and action: A comparison of two forms of requesting. Research on Language & Social Interaction, 41(2), 129–153.CrossRefGoogle Scholar
  16. Davidson, J. A. (1984). Subsequent versions of invitations, offers, requests and proposals dealing with potential or actual rejection. In J. M. Atkinson & J. Heritage (Eds.), Structures of social action: Studies in conversation analysis (pp. 102–128). Cambridge: Cambridge University Press.Google Scholar
  17. de Rosario-Martinez, H. (2015). phia: Post-hoc interaction analysis. R package 0.2-1. https://CRAN.R-project.org/package=phia.
  18. Deutschmann, M. (2006). Social variation in the use of apology formulae in the british national corpus. In A. Renouf & A. Kehoe (Eds.), The changing face of corpus linguistics (pp. 205–222). Amsterdam: Rodopi.CrossRefGoogle Scholar
  19. Drew, P., & Curl, T. S. (2008). Conversation analysis: Overview and new directions. In V. K. Bhatia, J. Flowerdew, & R. H. Jones (Eds.), Advances in discourse studies (pp. 22–35). Routledge: Milton Park, etc.Google Scholar
  20. Ervin-Tripp, S. (1976). Is Sybil there? The structure of some American English directives. Language in Society, 5(1), 25–66.CrossRefGoogle Scholar
  21. Félix-Brasdefer, J. César. (2015). The language of service encounters: A pragmatic-discursive approach. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
  22. Flöck, I. (2016). Requests in American and British English: A contrastive multi-method analysis., Pragmatics & beyond new series Amsterdam: John Benjamins.CrossRefGoogle Scholar
  23. Geluykens, R. (2011). Politeness in institutional discourse: Face-threatening acts in native and nonnative english business letters. München: Lincom Europa.Google Scholar
  24. Goldschmidt, M. M. (1998). Do me a favor: A descriptive analysis of favor asking sequences in American English. Journal of Pragmatics, 29(2), 129–153.CrossRefGoogle Scholar
  25. Heinemann, T. (2006). “Will you or can’t you?” Displaying entitlement in interrogative requests. Journal of Pragmatics, Special Issue: Processes and Products of Negation, 38(7), 1081–1104.CrossRefGoogle Scholar
  26. Heritage, J. (2008). Conversation analysis as social theory. In B. S. Turner (Ed.), The new Blackwell companion to social theory (pp. 300–320). Oxford: Wiley.Google Scholar
  27. Hutchby, I., & Wooffitt, R. (2006). Conversation analysis: Principles, practices and applications. Cambridge: Polity Press.Google Scholar
  28. Jautz, S. (2008). Gratitude in British and New Zealand radio programmes. Nothing but gushing? In K. P. Schneider & A. Barron (Eds.), Variational pragmatics: A focus on regional varieties of pluricentric languages, 141–78. Amsterdam: Benjamins.Google Scholar
  29. Jucker, A. H., Schneider, G., Taavitsainen, I., & Breustedt, B. (2008). Fishing for compliments: Precision and recall in corpus-linguistic compliment research. In A. H. Jucker & I. Taavitsainen (Eds.), Speech acts in the history of english (pp. 273–294). Amsterdam/Philadelphia: Benjamins.CrossRefGoogle Scholar
  30. Kallen, J. L., & Kirk, J. M. (2012). SPICE-Ireland: A user’s guide. Belfast: Clo Ollscoil na Banriona.Google Scholar
  31. Kirk, J. M. (2016). The pragmatic annotation scheme of the SPICE-Ireland corpus. International Journal of Corpus Linguistics, 21(3), 299–322.CrossRefGoogle Scholar
  32. Kohnen, T. (2008). Tracing directives through text and time: Towards a methodology of a corpus-based diachronic speech-act analysis. In A. Jucker & I. Taavitsainen (Eds.), Speech acts in the history of English (pp. 295–310). Amsterdam: Benjamins.CrossRefGoogle Scholar
  33. Lee, S.-H. (2011). Managing non-granting of customers’ requests in commercial service encounters. Research on Language and Social Interaction, 44(2), 109–134.CrossRefGoogle Scholar
  34. Levin, M. (2014). The bathroom formula: A corpus-based study of a speech act in American and British English. Journal of Pragmatics, 64(April), 1–16. doi: 10.1016/j.pragma.2014.01.001.CrossRefGoogle Scholar
  35. Levinson, S. C. (1983). Pragmatics. Cambridge: Cambridge University Press.Google Scholar
  36. Lindström, A. (2005). Language as social action: A study of how senior citizens request assistance with practical tasks in the swedish home help service. In A. Hakulinen & M. Selting (Eds.), Syntax and lexis in conversation analysis (pp. 209–233). Amsterdam: Benjamins.CrossRefGoogle Scholar
  37. Manning, C. (2007). Generalized linear mixed models (Illustrated with R on Bresnan et Al.’s Datives Data). https://nlp.stanford.edu/manning/courses/ling289/GLMM.pdf.
  38. Pomerantz, A. (1984). Agreeing and disagreeing with assessments: Some features of preferred/dispreferred turn shapes. In J. Atkinson & J. Heritage (Eds.), Structures of social interaction studies in conversation analysis (pp. 57–101). Cambridge: Cambridge University Press.Google Scholar
  39. Pufahl Bax, I. (1986). How to assign work in an office: A comparison of spoken and written directives in American English. Journal of Pragmatics, 10(6), 673–692.CrossRefGoogle Scholar
  40. R Core Team. (2013). R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing. https://www.R-project.org.
  41. Rauniomaa, M., & Keisanen, T. (2012). Two multimodal formats for responding to requests. Journal of Pragmatics, 44(6–7), 829–842.CrossRefGoogle Scholar
  42. RStudio Team. (2015). RStudio: Integrated development for R. Boston: RStudio Inc.Google Scholar
  43. Rühlemann, C. (2010). What can a corpus tell us about pragmatics? In A. O’Keeffe & M. McCarthy (Eds.), The Routledge handbook of corpus linguistics (pp. 288–301). London: Routledge.Google Scholar
  44. Sacks, H., Schegloff, E. A., & Jefferson, G. (1974). A simplest systematics for the organization of turn-taking for conversation. Language, 50, 696–735.CrossRefGoogle Scholar
  45. Sbisà, M. (2013). Locution, illocution, perlocution. In M. Sbisà & K. Turner (Eds.), Pragmatics of speech actions (pp. 25–75). Berlin: De Gruyter.CrossRefGoogle Scholar
  46. Schegloff, E. A. (1972). Sequencing in conversational openings. In J. Gumperz & D. Hymes (Eds.), Directions in sociolinguistics. The ethnography of communication (pp. 346–380). Oxford: Basil Blackwell.Google Scholar
  47. Schegloff, E. A. (1996). Issues of relevance for discourse analysis: Contingency in action, interaction and co-participant context. In E. H. Hovy & D. R. Scott (Eds.), Issues of relevance for discourse analysis: Contingency in action, interaction and co-participant context (pp. 3–35). Computational and conversational discourse. Burning issues—an interdisciplinary account. New York: Springer.Google Scholar
  48. Schegloff, E. A., & Sacks, H. (1973). Opening up closings. Semiotica, 8(4), 289–327.CrossRefGoogle Scholar
  49. Searle, J. R. (1969). Speech acts: An essay into the philosophy of language. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
  50. Searle, J. R. (1976). A classification of illocutionary acts. Language in Society, 5(1), 1–23.CrossRefGoogle Scholar
  51. Stevanovic, M., & Peräkylä, A. (2012). Deontic authority in interaction: The right to announce, propose, and decide. Research on Language & Social Interaction, 45(3), 297–321.CrossRefGoogle Scholar
  52. Stivers, T., & Rossano, F. (2010). Mobilizing response. Language and Social Interaction, 43(1), 3–31.CrossRefGoogle Scholar
  53. Thompson, S. A., Fox, B., & Couper-Kuhlen, E. (2015). Grammar in everyday talk: Building responsive actions. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
  54. Vine, B. (2009). Directives at work: Exploring the contextual complexity of workplace requests. Journal of Pragmatics, 41(7), 1395–1405.CrossRefGoogle Scholar
  55. Walker, T. (2013). Requests. In M. Sbisà & K. Turner (Eds.), Pragmatics of speech actions (pp. 445–66). Handbooks of pragmatics, Vol. 2. Berlin: De Gruyter Mouton.Google Scholar
  56. Wong, May L.-Y. (2010). Expressions of gratitude by Hong Kong speakers of English: Research from the international corpus of English in Hong Kong (ICE-HK). Journal of Pragmatics, 42(5), 1243–1257.CrossRefGoogle Scholar
  57. Wootton, A. (1981). Two request forms of four year olds. Journal of Pragmatics, 5(6), 511–523.CrossRefGoogle Scholar
  58. Wootton, A. J. (1997). Interaction and the development of mind. Cambridge: Cambridge University Press.CrossRefGoogle Scholar

Copyright information

© Springer International Publishing AG 2017

Authors and Affiliations

  1. 1.Department of English and American StudiesUniversity of OldenburgOldenburgGermany

Personalised recommendations