Responses to Thanks in Ireland, England and Canada: A Variational Pragmatic Perspective

The present study investigates responses to thanks across the varieties of Irish English, English English and Canadian English. Data are taken from the Lueneburg Direction-Giving (LuDiG) corpus, a specialised corpus of spoken direction-giving exchanges across pluricentric varieties constructed using Labovian-style methods. The analysis centres on the cross-varietal pragmatic choices made in responding to thanks on the level of tokens, types and strategies. Findings point to the broad universality of realisations of responses to thanks across the pluricentric varieties at hand. Variety-preferential choices are, however, also recorded on a national level, particularly in type and strategy preferences. While all varieties use a ‘minimising the favour’ strategy extensively, this strategy is employed to a comparatively higher degree in the Irish English and English English data. In contrast, the speakers of Canadian English use an ‘expressing appreciation of the addressee’ strategy to a comparatively larger extent. Speakers of Canadian English are suggested to orient more strongly to positive face needs, and speakers of Irish English and English English more strongly towards negative face needs. The paper also discusses the methodological challenges of contrasting spoken interactional data for cross-varietal pragmatic speech act analyses and shows some strengths of specialised corpora in this regard.


Introduction
Pluricentric varieties, such as Irish English (IrE), English English (EngE) and Canadian English (CanE), although sharing a common language, show evidence of synchronic variation. The study of such variation was long focused on the level of 1 3 pronunciation, vocabulary and grammar. In recent years, however, there is a heightened awareness that pluricentric varieties also display considerable synchronic pragmatic variation (cf. Schneider & Placencia, 2017 for an overview). Cross-varietal pragmatic variation across pluricentric varieties is seen in conventions of language use specific to one variety relative to another variety/varieties (variety-specific features) and also in particular conventions of language use preferred in one variety over another (variety-preferential features) (cf. Barron & Schneider, 2009; cf. also Barron, 2021b).
Responses to thanks are among the speech acts which have been shown to exhibit variety-specific and variety-preferential variation across pluricentric languages. Responses to thanks sequentially follow expressions of gratitude and function to address an imbalance in the thanker/thankee relationship caused by this expression of gratitude. The speech act of responses to thanks is highly routinised, a fact which can be explained by their frequent recurrence and thus by societies' need for an efficient way to deal with this communicative situation efficiently (cf. Barron, 2003;Coulmas, 1981; Wray, 1999 on pragmatic routines). As cultural answers to a communicative need (Coulmas, 1985: 53), it is little surprising that pragmatic routines realising responses to thanks have been found to reveal variation across pluricentric varieties (Bieswanger, 2015;Mulo Farenkia, 2013;Schneider, 2005;Schröder & Schneider, 2018). To date, however, cross-varietal research has been based on either written production questionnaire data (Mulo Farenkia, 2013;Schneider, 2005;Schröder & Schneider, 2018) or on field notes (Bieswanger, 2015). There exists to date no crossvarietal study of responses to thanks based on transcripts of spoken interactional data. Production questionnaire data are written data in which informants are asked to write what they would say in a particular situation. They are suited for eliciting prototypical language use and for providing an insight into pragmalinguistic knowledge. They cannot, however, be seen to replicate interactional data and to reflect what speakers would actually say in a particular situation (cf. Ogiermann, 2018;Sweeney & Hua, 2016). Field notes come nearest to such data but are limited given a bias towards salient and expected responses, the impossibility of verifying responses and also short-term memory restrictions, the latter which are particularly relevant where the researcher themselves is involved in the interaction (Golato, 2017: 23).
Apart from culture, genre has also been found to influence realisations of responses to thanks (cf. Schneider, 2007). Research on CanE has been conducted on a range of genres, including everyday conversational discourse (Mulo Farenkia, 2012Schneider, 2017) as well as responses to thanks in the context of the genre of direction-giving (Bieswanger, 2015;Dinkin, 2018) and in the context of interviews (Talla Sando Ouafeu, 2009). Descriptions of responses to thanks in IrE and EngE are, however, limited to research on everyday conversational discourse (Schneider, 2005).
The present study analyses realisations of routine responses to thanks across CanE, EngE and IrE with a focus on identifying potential variety-specific and variety-preferential uses. The study uses recorded spoken interactional data elicited using Labovian-style elicitation methods in the context of the genre of direction-giving. It thus adds to the scholarship on responding to thanks by addressing the research gaps set out above, namely the need for spoken interactional data in cross-varietal analyses of responses to thanks and the need for analyses in genres other than everyday interaction. The study is situated in the area of variational pragmatics, a field of research focused on the study of pragmatic variation within a single language according to macro-social factors, such as region, gender, age, socio-economic class and ethnic identity (Barron & Schneider, 2009;Barron, 2014Barron, , 2017aBarron, , 2021bSchneider, 2010Schneider, , 2021Schneider & Barron, 2008). The independent variable in the present study is region on a national level. Further macro-social factors are controlled in order to enable maximal equivalence across varieties. The analysis is based on a sub-section of the Lueneburg Direction-Giving (LuDiG) corpus, a specialised corpus designed for the purpose of the present cross-varietal study. The analysis centres on lexicalised verbal responses to thanks and specifically on the tokens, types and strategies used in realising a response to thanks across the three pluricentric varieties of CanE, EngE and IrE (cf. "Section Responses to Thanks: Description and Realisation").
In the following, I first sketch the nature of responses to thanks ("Section Responses to Thanks: Description and Realisation") and the scholarship on responses to thanks across the varieties of English ("Section Responses to Thanks Across the Varieties of English"). I then discuss methodological comparability in cross-varietal pragmatic comparisons of spoken interactional data ("Section Comparability of Context") and in this context consider the use of specialised corpora ("section The LuDiG Corpus: A Comparable Specialised Corpus"). Focus then turns to the LuDiG corpus, the specialised corpus employed in the present study, and to the specific dataset underlying the present analysis ("section The LuDiG Corpus: A Comparable Specialised Corpus") before I then consider the steps needed to identify responses to thanks in the database in a manner which ensures maximal comparability of the pragmatic feature across varieties ("Section Preparing for Analysis: Ensuring Comparability of the Pragmatic Feature"). The analysis follows ("Section Findings"), and then a discussion of the findings ("Section Discussion"). The paper closes with a summary of findings with suggestions for future potential research ("Section Conclusion").

Responses to Thanks
In the following I first describe the nature of responses to thanks ("Section 2.1, Responses to Thanks: Description and Realisation"). 1 An overview of research on responses to thanks across the varieties of English follows with a particular focus on IrE, EngE and CanE ("section Responses to Thanks Across the Varieties of English").

Responses to Thanks: Description and Realisation
Responses to thanks form the second part of an adjacency pair initiated by an expression of gratitude. This adjacency pair follows in response to a previous favour or face-enhancing act termed the reference action (Wunderlich, 1980: 295), which the thankee (the individual responding to thanks) has performed for the benefit of the thanker. In expressing gratitude, a thanker communicates their appreciation of the reference action and in doing so builds up the thankee's positive face, i.e. the desire to be accepted by others (Brown & Levinson, 1978, 1987. At the same time, however, the thanker communicates his indebtedness to the thankee and so threatens their own face. In this context, responses to thanks then function to restore any imbalances in the thanker/thankee relationship resulting from the thanker's hearersupportive expression of indebtedness. As such the function of responses to thanks is "to minimize the thanker's indebtedness" (Schneider, 2005: 103). From a Searlean perspective, responses to thanks may be categorised as expressive speech acts given that in responding to thanks, a thankee expresses their view of the situation at hand (Searle, 1976;cf. Rüegg, 2014).
Responses to thanks may be realised verbally and/or non-verbally. Non-verbal realisations include nods (Bieswanger, 2015: 537) or potentially also smiles. Verbal realisations may be lexicalised or non-lexicalised, the latter including realisations such as uh-huh, mh-hm. 2 Lexicalised responses to thanks have been analysed using the structural notion of head moves and supportive moves (Schneider, 2005: 113). A response to thanks head move can be understood as the minimal unit in an utterance which realises a response to thanks. Response to thanks head moves are defined by Schneider (2005: 113) as realisations of responses to thanks using standard routine formula, such as no bother or you're welcome in examples (1)-(2) respectively. Multiple head moves are also possible, as in (3) where the head move alright is combined with the head move no problem. Supportive moves may accompany head moves or may be used without a head move (cf. Edmondson & House, 1981;Schneider, 2005;Barron, 2021c). 3 Bye in (4) is an example of a supportive move accompanying the response to thanks head move no problem. 4

3
Responses to Thanks in Ireland, England and Canada: A Variational… (1) Direction-seeker: Thanks.
( The present analysis focuses on lexicalised realisations of response to thanks head moves (cf. Barron (2021c) for an analysis of uses of supportive moves in CanE, EngE and IrE). The question posed is how these moves are realised across the varieties of CanE, EngE and IrE. The analysis of head moves centres on the analysis of tokens, types and strategies of responses to thanks. These concepts are explained in the following; the hierarchical relationship between them is visualised in Fig. 1. Response to thanks head move tokens represent the actual routine formulations employed by speakers. Examples from the present data include no bother in (1), you're welcome in (2), no problem in (3) and (4), alright in (3), that's alright in (5) and you're alright in (6). On an abstract level, these tokens are then grouped into types (cf. Fig. 1) according to lexical and semantic similarities. Head move types are given in upper case in the present context. Thus, due to semantic similarities, the head move tokens no bother in (1) and no problem in (3) and (4) all realise the head move type NO PROBLEM. 5 Due to lexical and semantic similarity, the head move tokens alright in (3), that's alright in (5) and you're alright in (6) all realise the head move type ALRIGHT. On the most superordinate level, head move types are grouped into head move strategies (cf. Fig. 1) based on underlying common rhetorical strategies which they employ in functioning as responses to thanks, i.e. in minimising any indebtedness caused by the act of thanking and the reference act carried out by the thankee. Thus, NO PROBLEM and ALRIGHT Fig. 1 Relationship between the tokens, types and strategies realising a response to thanks head move 5 The choice of type label was originally based on frequencies of occurrences (cf. also Schneider, 2005: 136). In an attempt to increase comparability across studies, established labels are currently employed irrespective of frequencies.
both realise a 'minimising the favour' strategy. These head move types have in common that both attempt to minimise any indebtedness by playing down the efforts involved in carrying out the reference action. A further example is the 'expressing pleasure' strategy. This Schneider (2005: 106) describes as minnimising indebtedness by "claiming a complementary benefit to the thankee". Head move types such as PLEASURE, made up of tokens such as (my) pleasure, and COOL, consisting of head move tokens, such as cool (assumed elliptical form: I find that cool), realise an 'expressing pleasure' strategy. Other head move strategies include an 'expressing appreciation of the addressee' strategy (e.g. WELCOME), a 'verbally acknowledging thanks' strategy by which informants acknowledge the expression of gratitude (e.g. YEAH) and a 'returning thanks' strategy (e.g. THANKS) (cf. Aijmer, 1996;Bieswanger, 2015;Schneider, 2005) Table 1 shows the head move strategies employed in the present study together with the head move types realising them in the data. This categorisation broadly follows previous categorisations (e.g. Aijmer, 1996;Bieswanger, 2015;Schneider, 2005). 6 Only the categorisation of CHEERS as realising a 'returning thanks' strategy differs from previous categorisations which code this type as an 'expression of appreciation of the addressee' strategy (cf. Staley, 2018: 131). The categorisation in the present context as a 'returning thanks' strategy is supported by the Cambridge Dictionary (cheers, s.v.) which lists cheers as informally "used to mean 'thank you'" (cf. also Quirk et al., 1985: 852-853, who list cheers as an expression of gratitude).

Responses to Thanks Across the Varieties of English
There exists a growing number of variational pragmatic studies on responses to thanks in a range of regional and social varieties of English. Some of these contrast responses to thanks across national varieties (Bieswanger, 2015;Mulo Farenkia, 2013;Schneider, 2005Schneider, , 2017Schröder & Schneider, 2018), others contrast responses to thanks within one variety across social parameters (Dinkin, 2018;Rüegg, 2014;Staley, 2018). These studies have highlighted many consistencies in 1 3 Responses to Thanks in Ireland, England and Canada: A Variational… head move token and type uses and with regard to head move strategy uses across national varieties. Variety-specific head move tokens and types, while they exist, are rare. For nothing is one example of a token reported to be used in Cameroon English but not yet recorded elsewhere (cf. Tallo Sando Ouafeu 2009;Mulo Farenkia, 2013). In contrast, variety-preferential uses of head move types and strategies have been frequently reported. Schneider's (2005) production questionnaire study involving involving American English (AmE), EngE and IrE, for instance, revealed clear profiles in head move type preferences across varieties. While WELCOME (53.5%), NO PROBLEM (20.2%) and ANYTIME (18.6%) were the preferred heads in his AmE data, OKAY (51.2%), WELCOME (16.3%) and ANYTIME (12.4%) were preferred in the EngE data and WELCOME (34.2%), ANYTIME (25.5%) and NO PROBLEM (24.2%) in the IrE data. Thus, OKAY was particularly frequent in the EngE data and WELCOME, followed by NO PROBLEM, in the AmE and IrE data. Viewed from the perspective of head move strategies, IrE and AmE informants in Schneider's (2005) data preferred an 'expressing appreciation of the addressee' head move strategy followed by a much lower use of a 'minimising the favour' strategy. EngE informants, by contrast, preferred a 'minimising the favour' head move strategy followed by a much lower use of an 'expressing appreciation of the addressee' strategy. Schneider's (2005) analysis also provides some insights into the relationship of response tokens and formality levels across varieties. Formality levels were incorporated into his study by the choice of situations, and also by the choice of language realising the expression of gratitude given in each situation. In Schneider's (2005) informal situation a response to thanks was elicited in response to the prompt, Thanks for the coffee, while in the more formal situation the response followed a more elaborate first pair part of the form Thank you very much for the lift. Schneider (2005) suggested WELCOME to be a formal head move type in IrE and for OKAY to be a potentially informal type in IrE. However, given low levels of OKAY in the IrE data, this suggestion needs testing.
Also on the subject of variety-preferential uses, but turning to CanE, Schneider (2017), a study using the same questionnaire as in Schneider (2005) but with CanE informants, shows CanE informants to prefer NO PROBLEM (44.2%) and WELCOME (27.9%). Broadly reflecting these findings, Mulo Farenkia (2013), also employing production questionnaires, albeit with different situations, reports NO PROBLEM (47.97%) and WELCOME (22.76%) to be most preferred in CanE (cf. also Mulo Farenkia, 2012). The substantial use of WELCOME in CanE as reported by Schneider (2017) and Mulo Farenkia (2013) is also seen in Bieswanger' (2015) study adopting a rapid anonymous design with field notes. However, Bieswanger (2015) finds uses of NO PROBLEM to be much less frequent than reported in Schneider (2017) and Mulo Farenkia (2013). In addition, and divergent from the other studies, he finds YEAH to be used very frequently in CanE. Indeed, also related to this finding, Bieswanger (2015) reports the strategy of 'verbally acknowledging the thanks' to be used most frequently in CanE, followed by 'expressing appreciation of the addressee' strategy. 'Minimising the favour' was used to a more limited extent. Also focused on CanE, Dinkin (2018) adds a social and apparent-time change perspective in the context of a rapid anonymous study using field notes (Labov, 1972). The paper shows firstly that, similar to the case of WELCOME suggested above by Schneider (2005) for IrE, YOU'RE WELCOME is viewed as a more formal response to thanks in CanE. It is used across age groups in response to rather more elaborate expressions of gratitude, such as thank you very much and thank you. Furthermore, the same research suggests that in contrast, NO PROBLEM, is a more informal head move in CanE. Dinkin (2018) shows in addition, however, that in apparent time, NO PROBLEM along with NO WORRIES, are replacing YOU'RE WELCOME, with younger speakers using NO PROBLEM in situations which would previously have demanded the more formal option YOU'RE WELCOME. At the same time, the type NO WORRIES is increasingly employed by younger speakers as an informal type relative to NO PROBLEM. Overall then, Dinkin (2018) shows an increase in uses of NO PROBLEM and NO WORRIES and decreases in uses of YOU'RE WELCOME. In addition, the paper shows change in stylistic stratification, with NO PROBLEM functioning as a less formal response for older speakers. For younger speakers, in contrast, NO PROBLEM is a general-purpose response, usable in both formal and informant contexts.
In sum then, the literature review shows that cross-varietal scholarship on the varieties at hand is limited to studies using production questionnaire data and field notes. In addition, research contrasting EngE and IrE is limited to a single study by Schneider (2005). Although additional studies on responses to thanks exist for British English (cf. Aijmer, 1996 based on the London-Lund corpus and Edmondson & House, 1981 based on roleplay data), Schneider (2005) is the only study of responses to thanks in IrE. Findings to date have revealed many variety-preferential uses on the levels of head move type and strategy. Finally, formality differences in the use of response to thanks types have been suggested for IrE and CanE. Further research is, however, needed here, particularly for the case of EngE and IrE. The present study takes up these issues in a study focused on the tokens, types and strategies in responses to thanks in interactional spoken data in CanE, EngE and IrE. It investigates variety-specific and variety-preferential uses and compares findings to previous scholarship.

3
Responses to Thanks in Ireland, England and Canada: A Variational…

Methodology: Contrastivity and Comparability Across Varieties with Specialised Corpora
The present study is situated within variational pragmatics (Barron & Schneider, 2009;Schneider & Barron, 2008). Following variational pragmatic scholarship, it observes the central and related principles of empiricity, contrastivity and comparability (Barron & Schneider, 2009;Barron, 2017aBarron, , 2021bSchneider, 2010Schneider, , 2021. The requirement for these principles is based on the assumption that it is only in contrasting varieties that it becomes clear which features are unique to, preferred over, related to or shared with other varieties. Contrastive analyses require empirical research with comparable data across varieties. Specifically, contrastivity in a pragmatic context necessitates a) comparability of context and b) comparability of the pragmatic feature analysed across varieties (cf. Barron, 2021b, c). We deal with the steps taken to ensure such comparability of context in "Section Comparability of Context". "Section The LuDiG Corpus: A Comparable Specialised Corpus" then sketches the specialised corpus designed to maximise comparability of context in the present study. Comparability of the pragmatic feature is discussed in "Section Preparing for Analysis: Ensuring Comparability of the Pragmatic Feature" in preparation for the analysis in "Findings".

Comparability of Context
Our choices in language use are constrained by their context of use. This fact makes comparability across varieties difficult. So, for instance, whether you're welcome, okay or no problem is used depends on a range of contextual factors. Context is a multifaceted concept, with a social, linguistic, cognitive and socio-cultural component (cf. Fetzer, 2010;Fetzer & Oish 2011;cf. also De Saint-Georges, 2013). In cross-varietal analyses, an awareness of and attention to the influence of these facets of context is essential (cf. Barron, 2021c). Each of these four components of context is dealt with briefly in the following.
The social component of context involves the micro-social context and the macrosocial context in which interactions take place (cf. Barron, 2021c). The micro-social context includes interpersonal factors (e.g. social distance, social dominance, degree of imposition), physical surroundings, time of interaction, participant roles (e.g. doctor/patient, host/guest), while the macro-social context relates to the social parameters of the interactants (e.g. region, age, gender, socio-economic class and ethnic identity). Variation in any of these micro or macro-social factors may influence language use. Any of these factors may serve as an independent variable in a particular study. Having chosen this/these variable(s), analytical attention to or control of those remaining social features is necessary to achieve maximal equivalence within and across varieties.
Cognitive context refers to expectations and mental models of typical interactions. Cognitive context may vary, with mental models of how genres are negotiated differing. These cannot be controlled but it is necessary that the researcher has an awareness of them and their potential influence on the data (cf. Barron, 2021c).
Socio-cultural context encompasses culture-specific interpretations of social context, meaning that extra-linguistic variables, such as the status of a police officer, may differ across varieties. It also involves knowledge of social conventions with regard to appropriate use of language and knowledge of potential differences in cultural values (e.g. knowledge of what is a sensitive topic).
Finally, linguistic context involves genre, surrounding text and intonation. In written production questionnaires, the surrounding linguistic context can be easily controlled through prompts. However, in spoken interactional data, there may be considerable variation in what precedes and follows the pragmatic feature under analysis and such variation may influence language use, and thus also influence comparability across varieties. Careful attention is required on the analytical level. A rapid anonymous method in which one interactant employs an underlying script is one means of exercising some control and therefore also comparability over the linguistic context (cf. Labov, 1972;Bieswanger, 2015, cf. "Section The LuDiG Corpus: A Comparable Specialised Corpus").

The LuDiG Corpus: A Comparable Specialised Corpus
Specialised corpora are comparatively small corpora, generally composed of texts of a limited genre or register. They are composed with a particular intention in mind (cf. Koester, 2010). Comparable specialised corpora are corpora made up of two or more monolingual specialised corpora and designed according to the same principles. Such corpora hold many advantages for cross-varietal pragmatic analyses given that they enable a detailed knowledge of context, a potential delimitation of context, and the possibility of an in-depth analysis which takes linguistic context into account. In the present study, the Lueneburg Direction-Giving (LuDiG) corpus, a transcribed comparable specialised corpus, was specifically designed to ensure an analysis of responses to thanks across CanE, EngE and IrE with maximal contextual equivalence. The LuDiG is divided into three parts, and includes (a) IrE interactions recorded in Galway, (b) EngE interactions recorded in Bristol and c) CanE interactions recorded in Western Canada (Calgary in Alberta) and Eastern Canada (Halifax in Nova Scotia). 7 The LuDiG can be described as a structured corpus of directiongiving interactions which, in an effort to increase comparability across varieties (cf. "Section Comparability of Context"), exerts a high level of control over both linguistic and social context. 8 These controls are explained in the following. We turn first to the linguistic context.
Linguistic context was controlled in the specialised corpus in a first step via the choice of a single genre, the genre of direction-giving. This step also ensured some comparability of cognitive context in so far as it was expected that in direction-giving, direction-seekers in all varieties at hand expect direction-givers to provide them with information about how to get to wherever they want to go. Similarly, it was assumed that direction-seekers are expected to pay attention to directions and to be grateful for service. In a second step, the specialised corpus design followed a controlled rapid anonymous design (Labov, 1972) which had previously been employed by Bieswanger (2015) (cf. also Dinkin, 2018) and which allowed responses to be elicited following a single prompt (cf. also Myers Scotton & Bernsten, 1988). Specifically, the procedure involved an underlying script given to the direction-giver which included an expression of gratitude produced following direction-giving. In order to eliminate possible co-textual variation resulting from the form taken by the initiating expression of gratitude, all expressions of gratitude took the form of thanks (cf. Dinkin, 2018;Leech, 2014). This expression of gratitude had a double function, namely to (a) express gratitude for the directions given on the one hand and to (b) signal understanding of directions from the point of view of the direction-seeker and thus a willingness to close the exchange on the other (see Psathas & Kozloff, 1976). The informants' next turn, and the focus of analytical interest, was a spontaneous realisation.
Taking social context into account, the design of the LuDiG corpus introduced a certain level of control over relevant social variables to allow maximal equivalence across varieties: Apparent age of responders to thanks was set at apparent 30-50 year olds as judged by the direction-seeker. The age bracket 30-50 years was chosen in line with the peak age of standard usage (cf. Bieswanger, 2015;Holmes, 2013: 179). Turning to gender, the gender of the direction-seeker and thanker, and thus the recipient of the response to thanks, was a female student in all interactions. The thanker was the same student in the CanE and EngE data. She was a NS of CanE, a postgraduate student who had studied in Canada and in England. The IrE directionseeker was a postgraduate NS of IrE. Metadata was gathered on the apparent gender of the informants, whether masculine or feminine, as determined by informant voice and external appearance. On a micro-social level, levels of familiarity and the social role of the speaker/addressee were also controlled in interaction, with interactants socially distant to each other, and the person producing the response to thanks in the role of the direction-giver. Finally, interactions took place in an urban setting across all three varieties, in Bristol in England, Galway in Ireland and in Calgary and Halifax in Canada.
Interactions were audio-recorded in a public arena with no harm caused to informants. Following the interaction, the goals of the project were immediately explained and permission for use of the recording for research purposes was requested from each informant individually and anonymity guaranteed as outlined in the guideline for good practice recommendations of The British Association for Applied Linguistics (BAAL) (cf. also Economic and Social Research Council (ESRC) 2015: 31). Recordings allowed the observer paradox and the limitations of field notes to be avoided (cf. 1).
The present study of responses to thanks across CanE, EngE and IrE is based on a sub-set of the LuDiG corpus. The analytical focus takes region on a national level as the independent variable. Given this focus on region, particularly in light of the potential influence of gender on the conventions of language use (cf. Schneider & Barron, 2008), gender is controlled in the present analysis. Hence, the analysis only focuses on the masculine sub-corpus of the LuDiG corpus, a corpus of 29,713 words (cf. 6). Informant numbers in this masculine sub-corpus are displayed in Table 2. The CanE data is balanced across Calgary (36 informants) and Halifax (40 informants). Group size is oriented to recommendations for datasets of thirty informants in contexts of controlled samples in homogeneous settings. Such recommendations are based on research findings showing that responses of homogeneous groups tend to concentrate around a few subcategories (cf. Kasper & Dahl, 1991). Informants

Preparing for Analysis: Ensuring Comparability of the Pragmatic Feature
The analysis centered on the types, tokens and strategies employed by speakers across CanE, EngE and IrE to respond to thanks. The researcher had access to both the transcripts and to the original audio recordings. The analysis focused on the transcripts in the first instance, with audio recordings consulted where questions arose. The first step was to identify the responses to thanks in the corpus and to ensure comparability of that which was analysed across varieties. Recent discussions on the comparability of the pragmatic feature in (cross-varietal) speech act analyses increasingly recommend a definition of the pragmatic feature via multiple criteria (cf. Terkourafi, 2012;Barron, 2017aBarron, , 2017bBarron, , 2021bStaley, 2018). In the present analysis, the criteria employed included a) position as a second pair part following a sealing expression of gratitude realised by thanks, but also b) response to a final sealing thanks in the case of sequences of thanks and c) function to redress an imbalance in the speaker-hearer relationship caused by a H-supportive thanks. Thus, as a first step in the identification of the responses to thanks, the thanks realised by the direction-seeker following the direction-giving was identified using a simple search. It was classified further as a response to thanks only where it was clear that the utterance followed and did not occur simultaneously with the expression of gratitude (criterion a). Secondly, in transcripts with a sequence of multiple thanks realised by the direction-seeker following the direction-giving, the response to the final thanks in such sequences was chosen in order to come nearest to a sealing thanks across corpora (criterion b). Finally, the definition of the pragmatic feature according to function (criterion c) highlighted potential ambiguities for the researcher in analysing the corpus, as for instance where cheers followed an expression of gratitude, as in example (7).
Cheers in this example could potentially function as a routine response to thanks and/or, given the status of the initiating thanks as a sealing thanks. However, it could also potentially function as a leave-take. Given also the absence of video recording, it was not possible to disambiguate such instances. Consequently, there are three ambiguous instances of Cheers in the EngE data. This ambiguity was taken into account in the analysis and alternative figures presented for EngE which treat cheers as a leave-take rather than as a response to thanks. These are labelled EngE (alt). 10

Findings
The following overview of findings focuses on responses to thanks across the varieties of IrE, EngE and CanE. The analysis focuses on lexicalised realisations of responses to thanks and in particular on the choices of head move types and tokens ("Section Responses to Thanks: Variation in Head Move Types and Tokens") and on head move strategy use ("Section Responses to Thanks: Variation in Strategy Use"). Both analyses are based on the number of head moves per variety. These figures are displayed in Table 4. They are calculated based on the total number of speakers per variety using a head move over single and multiple head moves (cf. Table 3). 11 As set out in "Section Preparing for Analysis: Ensuring Comparability of the Pragmatic Feature", the ambiguities highlighted in categorising cheers in the EngE 11 Significant differences were recorded in the distribution of head moves across varieties. The reader is referred to Barron (2021c) for an analysis of the occurrences of head moves and their internal structure, as well as of the use of supportive moves across these varieties with the same corpus. Barron (2021a) discusses responses to thanks in the EngE data and in a selection of the CanE data (Calgary only) within the context of pragmatic competence in English as an International Language (EIL). 10 Similarly, a focus on function in the definition of the pragmatic feature also highlighted instances of upward intonation in the use of routines, such as alright? or okay?, which made it somewhat unclear whether the forms functioned as a response to thanks or as a confirmation check for the speaker themselves. In such cases, the criterion of uptake was employed and the hearer's response examined. This analysis revealed that in no case did the hearer orient to a confirmation check. Consequently, such cases were treated as responses to thanks in the analysis. data (cf. "Section Preparing for Analysis: Ensuring Comparability of the Pragmatic Feature") are addressed in the analysis. Alternative figures are presented for EngE (Eng (alt)) which treat cheers as a leave-take rather than as a response to thanks. Finally, the report of the findings below is supported by tests of statistical significance. Figures for Fisher's exact test are given. This test allows statistical significance to be calculated despite low cell frequencies.

Responses to Thanks: Variation in Head Move Types and Tokens
The analysis of the head move types identified in the data is given in Fig. 2 12 and Table 5. Figures are based on the total head moves per variety (cf. Table 4). Fig. 2 focuses on the distribution of head move types while Table 5 also provides information on these types plus on the tokens identified in the analysis to realise these head move types. The analysis of head move type yielded a total of 11 types in the corpus. Many of the head move types listed have been identified in previous analyses. To these were added YOU BET and COOL. 13 Given large occurrences of use and cross-varietal variation, the type OKAY (e.g. Schneider, 2005) is divided into two separate categories, namely into OKAY, a type encompassing those tokens sharing the lexical item okay only, and into ALRIGHT, a type including tokens sharing the form alright only (ALRIGHT). Table 5 shows that each variety employs a broad range of head move types, with most types recorded in CanE (10) and EngE (9). Only six head move types were employed in IrE; however, the number of head moves is also lower in this variety so that no generalisations can be made. In addition, many head move types, such Triple head --1    1 3 Responses to Thanks in Ireland, England and Canada: A Variational… as YOU BET, COOL or THANKS were employed only to a minimal extent across varieties, so that differences in the number of types used may well be due to random chance. Notable is, however, the use of CHEERS by 10% of the EngE informants, a head move type not used by speakers of either other national corpus. Table 6 summarises the primary head move preferences across varieties in order of decreasing frequencies (cf. also Table 5 and Fig. 2). In the CanE corpus, WELCOME and NO PROBLEM were the preferred head move types. In the IrE corpus, OKAY was the preferred head move type followed by NO PROBLEM and ALRIGHT. In the EngE corpus, ALRIGHT was the preferred head move type followed by OKAY. Across varieties, a glance at Table 6 and Fig. 2 shows differences in type preferences, in particular in the use of WELCOME, OKAY and ALRIGHT. Post-hoc tests show that CanE speakers preferred WELCOME to a significantly larger extent than IrE or EngE speakers, with as many as 34.2% of the CanE speakers employing this head move relative to only 4.5% in the IrE corpus and 6.7% in the EngE corpus (EngE alt: 7.4%) (IrE vs. CanE: p = 0.006; EngE vs. CanE: p = 0.003 (EngE (alt) vs. CanE: p = 0.006)). On the other hand, the frequency of OKAY is highest in the IrE corpus (IrE: 31.8%; EngE: 16.7% (EngE alt: 18.6%), CanE: 5.1%) and the frequency of ALRIGHT highest in the EngE corpus (EngE: 33.3% (EngE (alt): 37%), IrE: 18.2%; CanE: 5.1%). Tests of statistical significance show that OKAY is used to a significantly larger extent in the IrE corpus relative to the CanE corpus (IrE vs. CanE: p = 0.002). ALRIGHT is used in the EngE corpus to a significantly higher degree than in the CanE corpus (EngE vs. CanE: p = 0.000 (EngE (alt) vs. CanE: p = 0.000)). Finally, although absolute uses of NO PROBLEM were higher in the IrE and CanE corpora relative to the EngE corpus, these differences were not statistically significant.

Responses to Thanks: Variation in Strategy Use
The analysis of response to thanks strategy builds on the strategies set out in Table 1 (cf. "Section 2.1, Responses to Thanks: Description and Realisation"). Table 7 shows that preferences across varieties were broadly similar, with all varieties using a 'minimising the favour' strategy extensively. Despite these similarities, however, significant differences were recorded in the distribution of the 'expressing appreciation of the addressee' strategy and the 'minimising the favour' strategy across On the other hand, CanE speakers use the 'expressing appreciation of the addressee' strategy to a significantly higher extent relative to IrE speakers (CanE vs. IrE: p = 0.003). The CanE/EngE contrast in the use of an 'expressing appreciation of the addressee' strategy is also significant, albeit not for the CanE/EngE (alt) contrast which excludes the cheers tokens from the analysis (CanE vs. EngE: p = 0.020 (CanE vs. EngE alt: p = 0.052)).

Discussion
The analysis focused on choices of head move types and tokens ("Section Responses to Thanks: Variation in Head Move Types and Tokens") and on strategy use ("Section Responses to Thanks: Variation in Strategy Use"). I detail the findings in each of these areas in the following and discuss how they relate to the overriding question of how responses to thanks head moves compare across CanE, EngE and IrE and also how they relate to previous research. This latter step throws new light on previous findings and also the potential role of data type, as well as of further explanatory factors, such as formality levels and age.

Head Move Tokens and Types
The analysis of head move tokens and types highlighted many commonalities across the varieties of English at hand. Variety-preferential uses of head move types were, however, also recorded (cf. summary of findings in Table 6). Specifically, a significant preference for the use of WELCOME in CanE was recorded relative to IrE or EngE. On the other hand, OKAY was used to a significantly larger extent in IrE visá-vis CanE, while ALRIGHT was employed to a significantly larger extent in EngE relative to CanE. In the following, the findings for each national variety are contrasted with previous research. Table 8 shows the preferences recorded in the present study and in previous studies. In order to achieve comparability across studies, the findings for OKAY and ALRIGHT in the present study have been merged in Table 8. We start with CanE. WELCOME and NO PROBLEM are used extensively in the present CanE corpus, a finding which reflects previous research. Table 8 shows these two head move types employed extensively in previous research on CanE by Mulo Farenkia (2012), Bieswanger (2015), Schneider (2017) and Dinkin (2018). Differences in relative preferences of the head move types WELCOME and NO PROB-LEM are, however, also evident. In Table 8, we see that uses of NO PROBLEM in CanE differ across studies. While Bieswanger (2015) records comparatively low use of this type, Schneider (2017) and Mulo Farenkia (2012) record high uses, also relative to the present study. Such differences across studies are suggested to relate to Responses to Thanks in Ireland, England and Canada: A Variational… Table 8 Preferred response to thanks head move types across varieties in the present study and across studies The total numbers in each study refer to the total of head moves in the data. The figures for Bieswanger (2015) displayed here have been calculated to include only lexicalised verbal realisations of head moves (n = 56). Also, Dinkin's (2018) data here has been re-calculated to omit the data for no responses. He groups sure, okay and yeah into an ACKNOWLEDGEMENT category, along with utterances and bye bye and have a good day, hence the grouping in the present table. The reader should note that in line with previous research, OKAY in Table 8 includes the types OKAY and ALRIGHT in the analysis in "Findings" IrE (n = 22) IrE (Schneider, 2005)  a) informality differences and to b) speaker age differences. Specifically, in relation to informality, Dinkin (2018) finds older speakers in his Toronto data to view NO PROBLEM as an informal type and to use NO PROBLEM to a much larger extent in response to an informal thanks relative to more complex expressions of gratitude (Dinkin, 2018: 204). WELCOME instead is used among the older age group in more formal settings. Thus, the relative (in)formality constructed in different data collection processes may have influenced the type chosen in different studies. Although both Bieswanger's (2015) and the present data-sets were collected in the same direction-giving setting, Bieswanger (2015) employed the more formal thank you as an expression of gratitude whereas the more informal thanks was the form used in the present study. Thus, a higher formality in Bieswanger's work may have led to a lower use of NO PROBLEM among the 30-50 year age group in his data relative to that of the present study. Differing speaker age differences is a further explanatory factor across studies. Dinkin (2018) finds that uses of the more informal types NO PROBLEM and NO WORRIES are increasing in apparent time and replacing the more formal WEL-COME. In other words, younger speakers are increasingly using forms, such as NO PROBLEM, also in situations where in the past, a more formal WELCOME might have been employed. Both Schneider's (2017) and Mulo Farenkia's (2012) informants from Cape Breton, Nova Scotia, were younger than those in either Bieswanger (2015) or in the present study. Schneider's (2017) informants are adolescents and Mulo Farenkia's (2012) students, 90% of whom were aged between 18 and 24. Such speaker age differences are suggested to explain the higher uses of NO PROBLEM in Schneider's (2017) and in Mulo Farenkia's (2012) data.
In the present CanE data, the type YEAH is also employed quite extensively at 19%. However, Bieswanger (2015) reports higher uses of YEAH (35.7%); indeed in Bieswanger's (2015) data, YEAH and WELCOME are the most frequently employed types. Dinkin (2018: 203) found acknowledgements, a category in which yeah tokens were included in his study, to be used more by men than women. Thus, the use of YEAH might have been expected to be higher in the present corpus, which focused exclusively on masculine speakers responding to thanks relative to Bieswanger's (2015) data which included both masculine and feminine speakers. The divergence of Bieswanger's (2015) data from all other data is difficult to explain and further research is needed. One possible explanation relates to the field note method employed. Yeah is multi-functional: it may function as a direct answer to a question or it may confirm unexpected additional information; it may also function as an "echoed yeah" by which interactants acknowledge an interactive unit in a general way (Jucker & Smith, 1998: 181). Bieswanger (2015: 531) defines his focus of interest functionally, terming it an imbalance reducer after thanks (IRAT). The complexities of interaction may, however, make it difficult for field researchers to focus on the exact function of a token following an expression of gratitude (cf. Barron 2021c). In other words, instances of yeah in field note data may potentially not actually function as responses to thanks but rather relate back to other features of the co-text, features which may not be noticed in the data-collection setting in which a researcher is participating and simultaneously focused on listening for realisations of a specific speech act.
Leaving the interpretation of the CanE data, we next compare the EngE and IrE responses to thanks with previous findings. We see that uses of OKAY (including OKAY and ALRIGHT in "Findings") reflect findings by Schneider (2005) for EngE, but not for IrE (cf. Table 8). In Schneider (2005), uses of OKAY are very low in IrE and thus contrast with the high levels of use in IrE in the present study. In addition, we see that uses of WELCOME are higher in Schneider's (2005) EngE and IrE data than in the present corpus (cf. Table 8). Possible explanations relate to the expression of gratitude employed and also to the data collection instrument. Schneider's (2005) production questionnaire included two initiative expressions of gratitude, one per situation. As set out in "section Responses to Thanks Across the Varieties of English", responses to thanks were elicited using two situations which differed in formality levels. Interestingly, while the use of WELCOME did not differ across these situations in Schneider's (2005) AmE data, the differences in the head move employed across these situations in the IrE data were notable, with most uses of WELCOME recorded in the more formal situation. Schneider's (2005) data, thus, suggests that WELCOME is a more formal head move type in IrE. In Schneider's EngE data, in contrast, WELCOME seems less frequent and more neutral, with 57.1% of the WELCOME head move types used being employed in the more formal lift situation and 42.9% in the less formal coffee situation. Turning to the present data and its relation to Schneider's (2005) findings, it would seem that the higher use of WELCOME in Schneider's (2005) IrE data relative to the present findings (and also the lower uses of OKAY in Schneider's, 2005 data) may be potentially explained via formality levels, with higher overall levels of formality in Schneider' (2005) questionnaire and lower levels of formality in the present setting (also communicated via a simple thanks). Other potential reasons for the diverging findings relate to the difference in data collection periods and the younger age of Schneider's (2005) informants who may have been potentially influenced by American uses of WELCOME as a widely used response to thanks head move type (Schneider, 2005) perhaps via television and film.

Strategies
In the present study, the grouping of head move types into strategies revealed all varieties to use a 'minimising the favour' head move strategy extensively. Significant differences in use were, however, recorded. The 'minimising the favour' head move strategy was used to a significantly higher degree in the IrE and EngE corpora relative to the CanE corpus. On the other hand, the 'expressing appreciation of the addressee' head move strategy was preferred in the CanE corpus to a significantly higher extent relative to the IrE corpus and in the EngE corpus (albeit not in the EngE (alt) corpus without cheers coded as a response to thanks).
The widespread use of a 'minimising the favour' and an 'expressing appreciation' head move strategy in the CanE context reflect findings by Bieswanger (2015). Compared to Bieswanger's (2015) findings, however, the use of the 'verbal acknowledgement of thanks' head move strategy was less important in the present context, as indeed set out in the discussion of the YEAH type above. Turning to the EngE corpus, findings broadly reflect those of Schneider (2005) which report a preference for a 'minimising the favour' head move strategy. By contrast, the present analysis of the IrE corpus diverges substantially from Schneider's (2005) analysis. Schneider (2005: 121-122) finds an 'expressing appreciation of the addressee' head move strategy to be employed most extensively in the IrE data followed by a 'minimising the favour' head move strategy. In the present context, use of the 'expressing appreciation of the addressee' strategy was found to be minimal. Instead, a 'minimising the favour' strategy was most commonly employed. The differences in findings relate closely to the differences in the use of the WELCOME type across studies -as discussed above. These diverging findings for CanE relative to EngE and IrE are suggested to potentially reflect cultural value schemes (cf. Barron, 2021a;Mulo Farenkia, 2012). High uses of the 'minimising the favour' strategy in IrE and EngE point to higher degrees of indirectness in the IrE and EngE corpora relative to the CanE corpus. In addition, the widespread use of the head move types OKAY and ALRIGHT in the IrE and EngE corpora, are, unlike 'expressing appreciation of the addressee' head move types, such as WELCOME, not specialised for thanking exchanges (Dinkin, 2018). Such findings also hint at a higher use of negative politeness in IrE and EngE and towards an orientation to the negative face of the addressee and thus to their need for independence and freedom of action in the IrE and EngE corpora (Brown & Levinson, 1978, 1987. In contrast, in CanE, the widespread use of WELCOME and the related 'expressing appreciation of the addressee' strategy point to a relatively higher use of positive politeness with considerably more responses to thanks addressed to the addressee's positive face and thus their need to belong and to feel accepted by others (see Brown & Levinson, 1978, 1987.

Conclusion
The analysis focused on responses to thanks across IrE, EngE and CanE using a corpus of spoken direction-giving interactions recorded via a rapid anonymous format. The study also highlighted the challenges of analysing language use in a crossvarietal manner, focusing in particular on the need for comparability of context and for comparability of the pragmatic feature. In this regard, it also showed the benefits of specialised corpora for cross-varietal pragmatic analyses. These benefits relate to a detailed knowledge of context, a potential delimitation of context, and the possibility of paying close attention in ensuring maximal equivalence in the identification of the pragmatic feature.
The present study also added to the cross-varietal scholarship on responses to thanks using spoken data and shed new light on previous findings. The analysis of head moves highlighted many common types and tokens across the varieties of English at hand. Variety-preferential preferences were, however, also recorded. The present findings, for instance, found CanE to prefer the head move type WELCOME to a significantly higher level than IrE or EngE. On the other hand, OKAY was preferred in IrE and ALRIGHT in EngE relative to their use in CanE. While some of these findings supported previous scholarship, some of the variety-preferential preferences recorded deviated from previous findings. OKAY was, for instance, found to be the preferred type in the IrE corpus followed by NO PROBLEM and ALRIGHT. These findings threw new light on previous research which had rather identified WELCOME as a preferred head move type in IrE and OKAY (including OKAY and ALRIGHT) as a dispreferred type (Schneider, 2005). The differences between the present data and previous findings were suggested to relate to the fact that WELCOME functions as a more formal head move type in the IrE context, a factor which would explain its higher use in previous research using more formal situations relative to the present context. In addition, uses of WELCOME in previous research were also suggested to possibly relate to the younger age group of previous studies and to their potential influence by AmE conventions. A further point of discussion in the data were uses of WELCOME and NO PROBLEM across corpora. These types were the preferred head move types employed in the present data but the extent of their use differed across studies. These findings were discussed in the context of previous research on apparent language change in CanE, with the age of the present informants (30-50 year olds) and relatively low formality levels, suggested to explain any differences in findings across studies (Dinkin, 2018). Finally, the present analysis also highlighted potential difficulties of field notes for data collection. Specifically, Bieswanger's (2015) report of high uses of YEAH in his CanE data were suggested to relate potentially to the difficulty of a focus on function in live interaction, particularly in the case of multifunctional tokens like YEAH. Hence, on occasion, co-text may have got in the way, and position rather than function may have been reported in the complexities of interaction.
Cross-varietal strategy use was also analysed. The present findings show a preference for a 'minimising the favour' head move strategy in the IrE and EngE corpora relative to the CanE corpus. In the CanE corpus, an 'expressing appreciation of the addressee' strategy was used to a larger extent relative to IrE and EngE. While these findings reflect previous research, the low use of the 'expressing appreciation of the addressee' strategy in the IrE corpus diverges from previous research, as also discussed in the context of WELCOME.
The present analysis has shown national variation in the conventionalised means of responding to thanks across IrE, EngE and CanE for the data at hand. Findings were most similar between IrE and EngE, both varieties preferring OKAY and ALRIGHT types and predominantly employing a 'minimising the favour' strategy. In CanE, in contrast, the WELCOME and NO PROBLEM types are most popular and although a 'minimising the favour' strategy is extensively used, so too is an 'expressing appreciation of the addressee' strategy. These findings were suggested to reflect a stronger orientation to the negative face of the addressee in the IrE and EngE corpora and to higher levels of positive politeness in CanE.
The present CanE corpus included responses to thanks from Eastern and Western Canada. A prior analysis of responses across these areas showed no pragmatic variation in responding to thanks. This high degree of uniformity in responses to thanks across different geographical varieties of CanE reflects previous research on CanE on the level of sound which suggests a general characteristic of urban CanE to be its relative homogeneity. As Labov et al. (2006: 217) put it, "Canadian English displays nothing like the dialect diversity of the United States, let alone that of Great Britain". Further research on sub-national variation, however, represents a desideratum for geographical varieties within England and Ireland before potential generalisations can be made concerning the uniformity of responses to thanks preferences within these nations.
The present study contrasted responses to thanks across three nations, Canada, England and Ireland. As such, it addressed the influence of the independent variable of region on a national level in realisations of responses to thanks. To this aim, it endeavoured to achieve maximal equivalence across sub-corpora on a contextual level ("Section Comparability of Context", "Section The LuDiG Corpus: A Comparable Specialised Corpus"). Such contextual control necessarily means, however, that the findings are not generalisable to all speaker groups across the three national varieties, but rather only to the specific sub-group analysed. Gender was one of the contextual variables which was controlled in the present context. Those responding to thanks in the present context were apparently masculine; the thanker and thus the person they responded to was a woman. Vocatives employed in the present masculine informant data, such as "my love" in example (5) and "darling" in example (7) would suggest a certain sensitivity to the gender of the thanker. 14 Such issues will be interesting to analyse in more depth and indeed future research with the LuDiG will contrast responses to thanks in the masculine and feminine sub-corpora.
Age was a further social factor controlled in the present context, with the age of responders to thanks set at the age range 30-50, the age of peak standard usage. The thankers for their part were students. The study could therefore neither examine potential apparent-time change nor age grading. Such are, however, potentially interesting issues. Dinkin (2018) has recorded apparent-time change in responding to thanks in CanE, for instance (cf. above); Further research is welcomed to investigate such issues in EngE and IrE.
Specialised corpora offer control over a broad range of features and as such are suitable for maximising equivalence across varieties as demonstrated in the present analysis. They might, however, beneficially be triangulated with metapragmatic data of language users from different varieties to yield a further insight into lay perceptions of the conventions of responding to thanks (cf. Schröder & Schneider, 2021). In addition, triangulation with data from more general cross-varietal comparable corpora, such as the International Corpus of English (ICE) corpus and the Corpus of Global Web-based English (GloWbE), represent an invaluable source of data. They provide naturally-occurring data and also important information as to the contexts of use of responses to thanks. Together with more specialised corpora which offer high comparability across varieties, they represent ideal resources for further investigations.

3
Responses to Thanks in Ireland, England  Word. Indicates falling, stopping tone-not grammatical Word, Indicates a "continuing" intonation-not grammatical Acknowledgements I would like to thank the special issues editors and two external reviewers for valuable input. Thanks also to Emily Black and Daniel Popp for support in the collection and coding of the present data and to Luisa Martens and Tessa Niels for support with formatting. All limitations remain the responsibility of the author.
Funding Open Access funding enabled and organized by Projekt DEAL.

Conflict of interest
The corresponding author states that there is no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.