Definitions of Formula

As discussed in the first section of this article, definitions of formula in pragmatics include components of recurrent sequences, social contract, and importance of context. The most common definition in pragmatics is Coulmas's definition given in the first section. In addition, Bardovi-Harlig (Reference Bardovi-Harlig2009, Reference Bardovi-Harlig, Kasper, Nguyen, Yoshimi and Yoshioka2010) also adapted the relevant characteristics outlined by Myles, Hooper, and Mitchell (Reference Myles, Hooper and Mitchell1998). Pragmatic routines are generally (a) at least two morphemes in length; (b) phonologically coherent, that is, fluently articulated and nonhesitant; (c) used repeatedly and always in the same form; (d) situationally dependent; and (e) community-wide in use (in keeping with the social focus of pragmatic routines, the psycholinguistic criteria are not listed; see Myles et al., Reference Myles, Hooper and Mitchell1998 for the remaining three criteria). The first two characteristics apply to the shape of the string (the phonological criteria being more characteristic of expert users rather than novices); the third speaks to its relatively consistent form; and the last two speak to the more pragmatic features determined by the situation and community-wide use, which speaks to the social contract.

While this definition includes context, other definitions link formulas explicitly to their pragmatic roles. Kecskes (Reference Kecskes2010) made a distinction between conversational routines and SBUs. For Kecskes, conversational routines are the broader category (and include SBUs as a subset) and are functionally bound, whereas SBUs are situationally (or contextually) bound. Examples of conversational routines include after all and to tell you the truth, which can be used in a variety of social contexts and topics. In contrast, SBUs such as How do you do? and Welcome aboard are linked to specific pragmatic contexts, namely introductions in the case of How do you do? and either literally boarding a ship or a plane or more figuratively, starting a new position in the case of Welcome aboard. In addition, following Blum-Kulka (Reference Blum-Kulka, Blum-Kulka, House and Kasper1989), Reiter, Rainey, and Fulcher (Reference Reiter, Rainey and Fulcher2005) described formulas as illocutionarily transparent.

Definitions are tested in the course of empirical research in at least two ways: by how the definitions are operationalized and how they function as guides for the identification of formulas in language samples. The next two sections consider each point in turn.

How Formula Is Operationalized

The general agreement that formulas in pragmatics are conventional expressions representing ways of saying things agreed upon by a speech community gives way to differences when we consider how formulas are characterized. Some studies operationalize formula as a specific sequence of words, as in you know (House, Reference House2009; Pilcher, Reference Pilcher2009), no worries and how are you going? (Davis, Reference Davis2007) and can I leave a message? and for here or to go? (Roever, Reference Roever2005). Other studies distinguish sequences with less variation from sequences that exhibit more (e.g., Nice to meet you vs. {Good/great/nice}{to see/seeing} you, Bardovi-Harlig, Reference Bardovi-Harlig2009).

Still other studies operationalize formula as a type of recipe at various linguistic levels. One of the first studies in empirical pragmatics, Manes and Wolfson (Reference Manes, Wolfson and Coulmas1981), identified three compliment formulas which were expressed (in order of frequency) as

1. a. NP {is/looks}(really) ADJ

2. b. I (really) {like/love} NP

3. c. PRO is (really) (a) ADJ NP

Yu's (Reference Yu2011) study of Taiwanese Mandarin speakers and ESL (English as a second language) learners added two additional formulas to Manes and Wolfson's: You V (NP) ADV (e.g., “You're doing very well.”), ADJ NP (e.g., “Great job!”). Wee and Tan (Reference Wee and Tan2008) investigated expressions that are captured as “So TIME” as in That's so last year and Email is so five minutes ago.

Less syntactically constrained is Overstreet and Yule's (Reference Overstreet and Yule2001) characterization of disclaimers in the form not X or anything, but Y, as in I don't want to sound like your mother or anything, but I think you should wait and We don't mean to cause a fuss of anything, but this isn't what we ordered.

In a study of responsibility statements that appear in acknowledgments in published articles, Bing and Ruhl (Reference Bing and Ruhl2008) identified the semantic value of components. They suggested that the basic formula for statements such as Naturally, I remain responsible for any mistakes still present and The views expressed are solely those of the authors is Acknowledge the convention + Content + Shortcomings + Responsibility of the author + No responsibility of others. Each category is illustrated with content, but is not linked to syntactic form (as their appendix of 165 authentic expressions shows): Acknowledge the convention (of course, as usual) + Content (final product, opinions) + Shortcomings (deficiencies, remaining errors) + Responsibility of the author (only the author, mine alone) + No responsibility of others (They are not responsible, None of the above should be blamed). They note that although this is the full formula, all parts of the formula are optional although some components occur more frequently than others (e.g., 72 percent of all such responsibility statements contain some reference to errors, and 42 percent to the conventional nature of responsibility statements.)

Culpeper (Reference Culpeper2010) identified conventionalized impoliteness formulas as choices among words (following pattern grammar, see Hunston & Francis, Reference Hunston and Francis2000). For example, under insults (one of nine categories), Culpeper listed personalized negative vocatives [you] [rotten/dirty/fat/little/etc.] [moron/pig/bastard/loser/brat/etc.] [you] (which I summarize as You *!#@&*, you! As in You rotten liar, you!) and personalized negative assertions [you] [are] [so/such a] [stupid/hopeless/pathetic/terrible/fat/ugly/hypocrite/disappointment], as in You are so pathetic! and You are such a hypocrite!

As this review of the ways in which formulas are operationalized shows, some are very specific, some are rather underspecified (but described by their authors in the text), and some are stated without generalization as patterns of words. What they all share, however, is an attempt to describe a repeated pattern relevant to language use in the area of pragmatics. The next section surveys the means by which formulas are identified in pragmatics research.

How Are Formulas Identified?

How formulas are identified depends in part on the goal of the specific inquiry. Some studies start with the formula (in that case the formula is preidentified), and others start with a context and identify formulas used therein. Studies of production, the dominant area of investigation in pragmatics, may take either approach, but studies of comprehension, recognition, or attitude necessarily begin with the former (see, e.g., Bardovi-Harlig, Reference Bardovi-Harlig and Han2008, Reference Bardovi-Harlig, Kasper, Nguyen, Yoshimi and Yoshioka2010; Davis, Reference Davis2007; Roever, Reference Roever2005, Reference Roever2006, Reference Roever2007; Sifianou & Tzanne, Reference Sifianou and Tzanne2010; Taguchi, Reference Taguchi2011). Production studies that begin with a formula generally investigate the contexts in which it occurs and may also include variations in pronunciation and meaning (House, Reference House2009; Ohashi, Reference Ohashi2010; Overstreet & Yule, Reference Overstreet and Yule2001; Pilcher, Reference Pilcher2009; Vergaro, Reference Vergaro2008; Wee & Tan, Reference Wee and Tan2008). Frequency is rarely a methodological issue in this approach, as the value of the expressions is asserted or assumed.

The second approach to formulas is to identify them in context. In this approach, context may be defined by illocutionary force, speech acts, or interactional setting or speech event. The studies are concerned with identifying recurrent sequences (defined earlier), but are variously interested in establishing situational dependence or frequency of occurrence.

Let us first consider a classic study as an example. Manes and Wolfson (Reference Manes, Wolfson and Coulmas1981) began with the compliment speech act and collected 686 examples from spontaneous conversation. They identified three main recurrent patterns in the data, one of which, NP {is/looks}(really) ADJ, accounted for 53.6 percent of the compliments; two additional forms—I (really) {like/love} NP and PRO is (really) (a) ADJ NP—together constitute 30 percent of the compliments, and the three account for 85 percent of the 686 compliments collected by the study. In this study Manes and Wolfson identified (a) recurrent sequences in (b) a well-defined context and (c) established their rate of use.

Culpeper (Reference Culpeper2010) also employed a 50 percent usage criterion to identify impoliteness formulas. Culpeper first identified contexts in which impoliteness was likely to be in evidence. He selected television reality shows known for impolite talk, 51 graffiti dialogues collected from university library desks, and 100 diary accounts of impolite encounters. Next he identified candidates for impoliteness formulas by identifying interactions in which the addressees responded in a way that indicated they understood the utterance to be impolite. Indications that addressees found utterances to be impolite were identified in co-text, retrospective comments (a feature of the TV shows in which participants are interviewed about their experiences and other participants), and nonverbal reactions. Culpeper's final step involved establishing the frequency of use of the candidate impoliteness formula. He compared the candidate formulas against their occurrences in the Oxford English Corpus (OEC). In order for a candidate formula to be included in the final list of impolite formulas, at least 50 percent of the occurrences of any one variant of a formula had to involve impoliteness in more than 50 percent of its occurrences in the OEC. Culpeper explained:

Thus, because less than one-half of the uses of shut up could be shown to be impolite, it was not included. In contrast, because shut the fuck up was used predominately in cases involving impoliteness, it was considered to be an impolite formula.

A third investigation to link recurrent strings to context and frequency is Bardovi-Harlig (Reference Bardovi-Harlig2009; Bardovi-Harlig, Bastos, Burghardt, Chappetto, Nickels, & Rose, Reference Bardovi-Harlig, Kasper, Nguyen, Yoshimi and Yoshioka2010). Bardovi-Harlig and her students observed naturally occurring conversations in which conventional expressions occurred. These conversations then provided the models for the development of the contexts (called scenarios) for an oral discourse completion task (DCT) designed to elicit conversational formulas. The task was piloted twice with revisions, retaining only those contexts that elicited a single expression in at least 50 percent of the responses. In the final task, native speakers of English and learners of English, all residents of the same university community, responded orally to a timed, computer-delivered DCT (for details, see Bardovi-Harlig, Reference Bardovi-Harlig2009), The longest recurring string used in the context was identified. For example, thank you might occur in some contexts, but thank you so much or thank you for in others. Twenty-two of the scenarios yielded a single conventional expression that met or exceeded the 50 percent cutoff used by Culpeper (Reference Culpeper2010) and was consistent with the rate of use of the dominant compliment formula found by Manes and Wolfson (Reference Manes, Wolfson and Coulmas1981). In fact, in some contexts native speakers clearly preferred one expression to its closely related variants. In thanking a teacher for her time, native English-speaking undergraduate students used Thank you for your {help/time} in 94 percent of their response. In thanking a teacher for a make-up test, Thank you so much dominated the native-speaker undergraduate responses at 89 percent use.

Each of these three studies identified a recurrent pattern in a specified context and established a rate of use. A greater number of studies identify a recurrent pattern, specifying either the context or the frequency. Given the importance of context to pragmatics, more studies emphasize context over frequency and have identified a range of social and linguistic contexts, analyzing both authentic oral and written language samples. Sharifian's (Reference Sharifian2008) analysis of responses to compliments in Persian relied on her knowledge of Persian to identify formulas, which was later supported by informal quantitative descriptions of use (in some cases reported as “several uses”; in others, the number of uses is given). Two of the most frequently used formulas are used in more than one scenario by multiple speakers: loft daarin (roughly, “you are kind” or “that's so kind of you [to say that]”) occurs in at least six (of 10) scenarios in about one third of the responses; ghaabeli nadaareh (it is not worthy of you) also recurs but at a lower rate and in fewer contexts (four scenarios). The formulas only appear in the same contexts twice. Although the rates of use are quite a bit lower than the 50 percent cutoff used by Bardovi-Harlig (Reference Bardovi-Harlig2009) and Culpeper (Reference Culpeper2010), the appearance of the recurring sequences in multiple compliment contexts with multiple users helps to empirically support their status as social formulas, especially for readers who have no knowledge of the language. Other studies that focus on contexts and the formulas used in them include the investigation of statements of responsibility in acknowledgments in academic articles (Bing & Ruhl, Reference Bing and Ruhl2008) and concessive expressions in business letters (Vergaro, Reference Vergaro2008).

In contrast to the studies that primarily examine context are studies that primarily report frequency. This profile is typical of corpus studies, which are still relatively rare in pragmatics (Schauer & Adolphs, Reference Schauer and Adolphs2006; Wong, Reference Wong2010). Wong investigated thanking expressions used by speakers of English in Hong Kong using the Hong Kong Interactional Corpus of English (HK-ICE) by extracting and examining 233 thanking expressions for their components. Lack of occurrence is particularly easy to document in a corpus, and Wong reported the absence of British thanking expressions cheers and ta in the conversation of Hong Kong English speakers.

In summary, although some studies have sought to establish empirically the status of the formulas they have investigated, many studies have not. The approach that researchers take depends in large measure on the research question that they intend to answer. Establishing the quantitative credentials of formulas occurs primarily when their status supports a higher-order argument: that compliments are formulaic (Manes & Wolfson, Reference Manes, Wolfson and Coulmas1981), that formulas are not restricted to the expression of politeness (Culpeper, Reference Culpeper2010), and that high-frequency formulas are the social norm in certain contexts and reasonable acquisitional targets for L2 learners (Bardovi-Harlig, Reference Bardovi-Harlig2009).

How Are Formulas Used?

Pragmatics research addresses the question of how formulas are used in two main ways: by characterizing the contribution of formulas to pragmatics in general and by demonstrating how individual formulas are used.

Terkourafi (Reference Terkourafi2002, Reference Terkourafi, Marmaridou, Nikiforidou and Antonopoulou2005) tied formulas to politeness. Terkourafi (Reference Terkourafi2002) characterized formulas as “ready-made solutions to the complex and pertinent problem of constituting one's own and one's addressee's face while simultaneously ensuring that one's immediate goals in interaction are achieved” (p. 196). She argued that in her Cypriot Greek data, comprising 115 hours of spontaneous conversations, formulas carry the burden of polite discourse, raising the possibility that formula use may be a significant feature of polite discourse in any culture. As Terkourafi pointed out, additional quantitative cross-linguistic research on polite language would be an appropriate test.

Culpeper (Reference Culpeper2010) discussed the roles of formulas in the realization of impoliteness, showing that impolite turns can be, and often are, formulaic. Thus, even if formulas carry the burden of politeness in a language, that is not their only job. The extension of their function may be seen as being illocutionarily transparent in the most general sense.

One example of how formulas may be used relates to speaker expectations. Reiter et al. (Reference Reiter, Rainey and Fulcher2005) investigated the relation of speaker expectations to the use of formulas, asking “Do higher expectations of request compliance lead speakers to employ formulaic requests rather than non-formulaic ones?” (p. 6). They found that informants reported significantly more certainty of compliance when using formulaic requests than when realizing nonformulaic ones.Footnote ⁴ Spanish informants reported more certainty of compliance when requests made during role-plays were formulaic than when they were not. In contrast, the requests of the British participants did not reflect the degree of certainty that addressees would comply. Although certainty differed, both Spanish and English requests employed the same type of pragmatic indirectness that social meanings of formulas may vary across cultures.

A second approach to determining how formulas are used is to explore one specific formula. Recent studies in English include conversational disclaimers (e.g., I wasn't trying to be nosey or anything, but I saw the letters on her desk; Overstreet & Yule, Reference Overstreet and Yule2001), academic responsibility statements (Needless to say, all errors and omissions are my own; Bing & Ruhl, Reference Bing and Ruhl2008), and So + TIME constructions (That's so last year!; Wee & Tan, Reference Wee and Tan2008). Ohashi (Reference Ohashi2010) investigated the augmentation of thanking and apology formulas by bowing in Japanese, the only study reviewed here to include a nonverbal component. Pilcher (Reference Pilcher2009) investigated the use of I don't know and I don't think in their nonlocalized (e.g., I dunno and I doØ think) and localized variants (e.g., I divn't knaa or I divn't think) in a Northern English dialect. She found that the use localized and nonlocalized variants are governed by different parameters. Whereas nonlocalized variants seem to carry specific functional loads, localized variants seem to function as social markers. These studies show that individual formulas may have very specific functions, and taken together they show that these functions are quite diverse. Studies of general functions of formulas find that they may serve to express both politeness and impoliteness, and they may also relate to speakers’ expectations about the listeners’ response.

How Widespread Is the Use of Formulas in Pragmatics?

The question “How widespread is the use of formulas in pragmatics?” is an important one lest we assume that all pragmatics, politeness, or impoliteness rests on formulaic language. Evidence from a variety of studies, languages, and data collection techniques suggest that formulas are only one type of pragmalinguistic resource among many others. Comparisons across languages show that situations, removed from their cultural contexts, do not inherently demand formulaic language. For example, shopkeepers and customers engaging in language-intense service encounters in shops in France and Syria differ in their use of formulaic language in the request solicitation-request-request uptake sequence, as they do in the verbal acceptance of goods (Traverso, Reference Traverso2006). Reiter et al. (Reference Reiter, Rainey and Fulcher2005) described contexts in which either a formulaic or nonformulaic response may be given. Moreover, Barron (Reference Barron2003) framed one learnability problem in L2 pragmatics in relation to differences in formula use in the L1 and L2: Difficulties may arise in “situations requiring a pragmatic routine in one language but an ad hoc formulation in the other; [and] situations requiring a pragmatic routine in one language but no remark in the other” (p. 186).

In the process of identifying contexts that were likely to produce the most consistent use of formulas in the speech of native speakers in their local university community, Bardovi-Harlig et al. (Reference Bardovi-Harlig, Bastos, Burghardt, Chappetto, Nickels, Rose, Kasper, Nguyen, Yoshimi and Yoshioka2010) found that candidate contexts did not represent speech acts equally. Requests, for example, showed low use of formulas, whereas thanking contexts showed reasonably high use. Even within thanking scenarios, there was variation from high rates of use of a single formula (89–94 percent) to lower rates of 66 percent in thanking contexts that elicited That'd be great, for example. (See Bardovi-Harlig, Reference Bardovi-Harlig, Félix-Brasdefer and Koike2012a, for a discussion of variation.) When concentrating on formulas in pragmatics, it is important to remember that pragmatics is not entirely formulaic.

Attitudes Toward Formulas

Some studies have begun to investigate attitudes toward pragmatic uses of formulas. Although in some sense beyond the traditional scope of pragmatics, the investigation of attitudes addresses some of the basic components of Crystal's definition of pragmatics, namely, the point of view of users and the effects that language has on other participants. Moreover, attitudes may also reflect participants’ view of the language community that uses the formulas and their desire to be a part of it.

Studies of hearers’ interpretations show that formulas are sometimes received with skepticism as to their sincerity and true intent. Greek respondents to a survey conducted by Sifianou and Tzanne (Reference Sifianou and Tzanne2010) reported that business uses of “elaborate formal constructions” (p. 675) such as Kαλημέρα, X (Good morning, this is X [last name]), the routine opening Πώς μπoρώ να σας εξυπηρετήσω (How can I help you?), and the closing routine Ευχαριστoύμε πoυ μας καλέσατε (Thank you for calling us) represent attempts at increasing business, rather than sincere politeness formulas. Moreover, 92 percent of the respondents reported that “these are not expressions of true politeness, but fixed, stereotypical expressions of formal politeness related to the company's marketing policy” (p. 675), and 66 percent thought that employees did not use these expressions spontaneously, but that they had been instructed and were obligated to do so as part of their job.

Davis (Reference Davis2007) investigated the attitude of potential formula users: Korean ESL students in Australia and Korean EFL (English as a foreign language) learners in Seoul were surveyed concerning pragmatic routines that could be called national formulas (e.g., for here or to take away? and how did you go? meaning “how did it go?”). Learners completed questionnaires that asked them to pick from multiple-choice responses that included Australian formulas, North American alternatives, and distracters; they also answered attitude questions. Korean students reported a reluctance to learn or use Australian formulas, stating a preference for North American or British English. If “using the expected formula is a strong indication of belonging, social identification, or acculturation” as Coulmas (Reference Coulmas and Asher1994, p. 1293) stated, then it appears that these learners may not want to belong to the community of Australian English speakers.

Using a web-based questionnaire to investigate the emotional weight of I love you and its equivalents in other languages, Dewaele (Reference Dewaele2008) found that speakers could associate strong emotional weight with expressions of love in languages learned after the first. “Language dominance did not always coincide with perceived emotional weight of I love you” (p. 1770); neither was the language of the beloved or family members automatically associated with emotion weight. In contrast, uninstructed learners and learners who started late are unlikely to attribute high emotional weight to L2 expressions. Respondents’ reflections suggest that multiple factors contribute both to formula use and attitudes toward them.

Speakers’ attitudes toward the use of formulas may provide insight into how they are interpreted. Although Sifianou and Tzanne's (Reference Sifianou and Tzanne2010) investigation showed that listeners had somewhat negative attitudes toward certain formulas in institutional talk that led them to doubt the sincerity of the utterances in which they occurred, future research may also show positive influences in other settings. Attitudes also played a role in the identification of worthwhile targets in SLA.

Second Language Acquisition

Although early research on SLA addressed formula use in L2 pragmatics, (e.g., Edmondson & House, Reference Edmondson, House, Phillipson, Kellerman, Selinker, Sharwood Smith and Swain1991; House, Reference House1996; Scarcella, Reference Scarcella1979; Wildner-Bassett, Reference Wildner-Bassett1994; for a review, see Bardovi-Harlig, Reference Bardovi-Harlig, Bardovi-Harlig, Félix-Brasdefer and Omar2006), research on formulas in L2 pragmatics began to gain momentum again in the early 2000s. Kecskes (Reference Kecskes2000, Reference Kecskes2003) investigated interpretation and production. Kecskes found that learners more readily recognized literal meanings than idiomatic ones in expressions such as OK, shoot (go ahead), Get out of here (don't fool me), and Piece of cake (easy). Kecskes also reported that especially among the advanced Asian students, production was often grammatical and appropriate but not nativelike. Production data also showed that although native speakers could produce expressions that they probably had never used before (such as television announcements like Stay tuned, we'll be right back), advanced learners could not.

Barron (Reference Barron2003) investigated the development of pragmatic routines in the production of Irish learners of German during a study abroad year. This longitudinal study showed that the learners of German exhibited increased targetlike use of conventional expressions and decreased nontargetlike use of L1-driven expressions within a single academic year. As part of a larger cross-sectional study, Roever (Reference Roever2005) investigated learners’ ability to select an appropriate expression for a context using a 12-item multiple-choice task (the routines task) in which a brief description of a situation was followed by a question such as “What would Jack probably say?” The respondents then selected from among four choices: the target expression and three distracters that were either relevant but nonidiomatic or idiomatic but irrelevant. Roever found that even short-term exposure of up to three months led learners in the host environment to score significantly higher than learners without such exposure.

Building on the interpretation and selection tasks of Kecskes (Reference Kecskes2003) and Roever (Reference Roever2005), Bardovi-Harlig implemented a series of tasks to explore the relation of recognition of formulas to their contextualized use in SLA. The first task was derived from published formulas from other studies and was presented in written format (Bardovi-Harlig, Reference Bardovi-Harlig and Han2008). The second task selected formulas used by the community in which the learners resided and was delivered aurally in keeping with the mode in which learners would encounter them in conversation (Bardovi-Harlig, Reference Bardovi-Harlig, Kasper, Nguyen, Yoshimi and Yoshioka2010). The test included both authentic formulas and modified, but grammatical, counterparts. A cross-sectional study showed that as learners increase in level of proficiency, their acceptance of modified expressions decreased and their acceptance of authentic expressions increased. The authentic expressions were accepted as being heard significantly more often than modified responses.

The learners’ self-reports of recognition of formulas were then compared to their production of the same formulas in specific contexts (Bardovi-Harlig, Reference Bardovi-Harlig2009). The recognition task just discussed was paired with a computer-delivered oral DCT. Results showed that the learners’ production of conventional expressions correlated significantly with recognition. The results suggest that low production of conventional expressions by learners has multiple sources: lack of familiarity with some expressions; overuse of some familiar expressions, which subsequently reduces the opportunity to use more targetlike expressions; level of development; and sociopragmatic knowledge.

These same studies have also investigated learner variables, most notably proficiency, length of stay, and intensity of interaction. Proficiency was found to significantly influence ability of learners to distinguish authentic formulas from modified ones (Bardovi-Harlig, Reference Bardovi-Harlig, Kasper, Nguyen, Yoshimi and Yoshioka2010); the selection of conventional expressions by foreign language learners, but not for learners with host environment exposure (Roever, Reference Roever2005); and the speed of formula comprehension (Taguchi, Reference Taguchi2011). Increased length of stay led to improved formula production (Barron, Reference Barron2003), and study abroad experience affected the comprehension of routines (Taguchi, Reference Taguchi2011).

In a study of three variables—proficiency, length of residence, and intensity of interaction— Bardovi-Harlig and Bastos (Reference Bardovi-Harlig and Bastos2011) found that recognition of authentic formulas showed significant influence of intensity of interaction, and production of pragmatic formulas showed significant influence of both proficiency and intensity of interaction. Length of stay did not have a significant effect on either recognition or production. Intensity of interaction with other speakers in the host environment and, less directly, length of stay are important in the acquisition of pragmatics because they gauge the potential effect of input and interaction in contexts of authentic language use, presumably in just those settings in which formulaic language may occur and in which learners would need to use it.

A second area investigated in the SLA of formulas in L2 pragmatics is transfer. Sharifian (Reference Sharifian2008) found that intermediate-level Persian learners of English in Tehran often used translations of a Persian formula for responding to compliments even when they did not use the formulas in the same context in Persian. Barron (Reference Barron2003) found that transfer from the learners’ L1 decreased over time, with some variation across scenarios. Retrospective interviews showed that learners were aware of cross-cultural differences on the level of the discourse structure of offers (e.g., Irish English often uses a formulaic reoffer such as Are you sure?, whereas German does not), and that they often weighed the value of competing approaches to adopting an L2 practice that Barron characterizes as “When in Rome . . .” or maintaining L1 conventions as “But I want to be myself!”

Bardovi-Harlig, Rose, and Nickels (Reference Bardovi-Harlig and Han2008) found that L2 learners from different L1 language-culture backgrounds showed no significant difference in rates of use of conventional expressions in an instructed L2 environment. However, there was a scenario effect, and a significant difference was found for two of the eight expressions investigated. The results are very mixed in this area of inquiry, suggesting that some uniformity in task and general research design might eliminate some of the variability and clarify the L1–L2 relationship.

There is a clear advantage of target community exposure to the acquisition of formulas (Barron, Reference Barron2003; Roever, Reference Roever2005; Taguchi, Reference Taguchi2011) and proficiency (Bardovi-Harlig, Reference Bardovi-Harlig2009, Reference Bardovi-Harlig, Kasper, Nguyen, Yoshimi and Yoshioka2010; Bardovi-Harlig & Bastos, Reference Bardovi-Harlig and Bastos2011; Roever, Reference Roever2005; Taguchi, Reference Taguchi2011). L1 transfer may or may not be a mediating variable (Barron, Reference Barron2003; Bardovi-Harlig et al., Reference Bardovi-Harlig, Rose, Nickels, Bowles, Foote, Perpiñán and Bhatt2008).

Formulas in Pragmatics Pedagogy

Articles concerning the pedagogy of formulas in L2 pragmatics echo some of the same concerns found in the wider literature, namely, identification of formulas and level of learner knowledge, and introduce others such as development of instructional materials and the influence of instruction on pragmatic development.

The development of instructional materials has focused on thanking expressions, perhaps for their perceived usefulness. With the purpose of developing resources for teaching, Schauer and Adolphs (Reference Schauer and Adolphs2006) compared the efficacy of eliciting English thanking expressions via a DCT to identifying them in the five million word corpus, CANCODE (Cambridge and Nottingham Corpus of Discourse in English). Not surprisingly given our earlier discussion, they found that the focused contexts of the DCT yielded a range of thanking expressions not found even in such a large corpus as CANCODE. The responses to the DCT by native speakers of British English were consistent with earlier American studies on thanking (e.g., Eisenstein & Bodman, Reference Eisenstein and Bodman1986), with the exception of individual evaluating adjectives (such as wicked). The authors point to the advantages of the DCT data for teaching expressions in context, although the corpus data illustrate invaluable interactional features for students.

In an unrelated article, Bardovi-Harlig and Nickels (Reference Bardovi-Harlig, Nickels, Houck and Tatsuki2011) developed teaching materials and activities based on the data collected via the oral DCT used by Bardovi-Harlig (Reference Bardovi-Harlig2009; Bardovi-Harlig et al., Reference Bardovi-Harlig, Bastos, Burghardt, Chappetto, Nickels, Rose, Kasper, Nguyen, Yoshimi and Yoshioka2010). The native-speaker responses served as input, whereas the learner responses provided a needs assessment for instruction on expressions of gratitude in a range of contexts.

De Pablos-Ortega (Reference De Pablos-Ortega2011) conducted a survey of thanking expressions in 64 Spanish L2 textbooks. Unlike previous reviews of pragmatics in language textbooks (e.g., Bardovi-Harlig et al., Reference Bardovi-Harlig, Hartford, Mahan-Taylor, Morgan and Reynolds1991; Boxer & Pickering, Reference Boxer and Pickering1995; Vellenga, Reference Vellenga2004), de Pablos-Ortega's review found thanking to be reasonably well represented in the textbooks, although the number of examples is most robust at the lower levels, decreasing at the intermediate level and again at the advanced level. Twelve scenarios extracted from the textbook conversations from which the final thanking expression had been replaced by a generally cordial but nonthanking turn were presented to 100 native Spanish speakers who rated the appropriateness of the turn and supplied a replacement if necessary. Parallel to the findings of Schauer and Adolphs (Reference Schauer and Adolphs2006), the DCT responses were more varied than the textbook thanking tokens. Whereas the textbooks used only gracias, muchas gracias, and (muchas) gracias por, native speakers also used por favor (please) and por favor + gracias. Por favor (as in no por favor) was used particularly in situations equivalent to those eliciting no thanks in English.

Bardovi-Harlig (Reference Bardovi-Harlig, Houck and Tatsuki2011) demonstrated how an aural formula recognition task could be used as a preteaching needs assessment by teachers. The task was originally employed in Bardovi-Harlig (Reference Bardovi-Harlig2009, Reference Bardovi-Harlig, Kasper, Nguyen, Yoshimi and Yoshioka2010) to assess familiarity with conventional expressions in relation to production. Because nonuse of formulas by learners stems from two possible situations—that learners may be unfamiliar with a formula or that learners may know a formula but not know how to use it— the recognition task helps teachers assess whether to start instruction with basic input to introduce the formulas or to assume the formulaic sequence is familiar and to emphasize the use.

A second type of pedagogically oriented studies comprises investigations of the influence of instructional intervention on the acquisition of formulas. Although there are classic studies including House (Reference House1996), these are not well represented in the recent literature. Bardovi-Harlig and Vellenga (Reference Bardovi-Harlig and Vellenga2012) investigated the efficacy of using focused noticing activities guided by explicit metapragmatic statements to help students learn pragmatically situated formulas. The targeted expressions came from the oral DCTs used by Bardovi-Harlig in other studies; instructional input for noticing that matched the DCTs scenarios were culled from transcriptions of the TV series Friends and presented to learners.Footnote ⁵ Learners showed increased use of the formulas on which they received input, but one group also improved on another set of formulas. The authors hypothesize that adding production activities to the noticing activities might increase the gain scores.

In the five-year window considered in this article, the most pressing issue for the instruction of pragmatics is the development of evidence-based materials and activities. This mirrors concerns in pragmatics more generally. Given the interest in instructional effects in pragmatics (e.g., see Bardovi-Harlig, Reference Bardovi-Harlig, Gass and Mackey2012b; Rose, Reference Rose2005) and in other areas of language as well, more treatment studies can be expected once approaches to teaching formulas for pragmatic uses have been developed.

This section has reviewed five recurrent themes in recent pragmatics research in an attempt to characterize work addressing formulaic language. There are undoubtedly other worthwhile issues that may be identified by other readers of this same literature. The next section nonetheless changes focus to examine research on formulas and the language communities that use them.