Applied linguistics in the late 1990s saw a revival of interest in the study of language play and linguistic creativity (Carter, Reference Carter1999, Reference Carter2004; Crystal, Reference Crystal1998). The topic had long received attention from researchers in children's first language (L1) development; however, at this time inquiry extended to L2 language play in adolescents and adults (Broner & Tarone, Reference Broner and Tarone2001; Cook, Reference Cook1997, Reference Cook2000; Kramsch & Sullivan, 1996; Lantolf, Reference Lantolf, Glass and Perez-Leroux1997; Sullivan, Reference Sullivan and Lantolf2000a, Reference Sullivan, Hall and Verplaetse2000b; Tarone, Reference Tarone, Swierzbin, Morris, Anderson, Klee and Tarone2000). With Vygotsky (Reference Vygotsky1978) as his theoretical framework, Lantolf (Reference Lantolf, Glass and Perez-Leroux1997) conceptualized language play as private speech that facilitates L2 development by allowing for rehearsal of new language forms. Cook (Reference Cook1997, Reference Cook2000) took a more expansive view of language play, including not only repetition of forms but their manipulation as well. His definition was also broader, recognizing that language play is not restricted to forms but often involves semantic and pragmatic play, including play with the functions of language. This type of play can involve the creation of joint fictionalizations (e.g., what if. . .?) and other types of play often associated with humor, such as banter and joking. In this way, Cook also acknowledged the sense of enjoyment that often results from such activity. Thus, while Lantolf focused on play with the L2, Cook's wider view included both play in and play with the L2. Play with the L2 consists essentially of the types of behaviors that are often referred to as wordplay, where language itself is manipulated. Although wordplay is generally associated with semantics (e.g., puns) and morphology, any level of language may be manipulated, including phonology, syntax, or pragmatics, for instance, by inventing a new word or subverting a ritual (e.g., “you're not welcome” as a response to “thank you”). Playing in the language involves the use of language to construct and engage in playful activities. This can include actual games, as well as less formalized activities, such as narratives and joking (for further discussion of this distinction, see Bell, Reference Bell2012; Vandergriff & Fuchs, Reference Vandergriff and Fuchs2009, in press). The majority of the empirical work that grew from these initial discussions seems to have been taken up largely in Cook's sense of language play.
The interplay between formulaicity and creativity in language use and acquisition is widely recognized (e.g., Myles, Hooper, & Mitchell, Reference Myles, Hooper and Mitchell1998; Myles, Mitchell, & Hooper, Reference Myles, Mitchell and Hooper1999; Nattinger & DeCarrico, Reference Nattinger and DeCarrico1992; Wray, Reference Wray2002). Language play, a specific type of creative language use, is often dependent on formulaic language. As Crystal (Reference Crystal1998) described it, “To play with language requires that, at some level of consciousness, a person has sensed what is normal and is prepared to deviate from it” (p. 181). We might also add the same for playing in (as opposed to with) a language. Wray's (Reference Wray2008) 11 criteria for determining whether or not a chunk is formulaic include two characteristics that relate to this. First, using irony as an example, Wray noted that in order to judge formulaicity one must be able to judge whether a particular string is associated with a particular situation or register, and that use outside of that context “does not reduce its formulaicity. Rather, its formulaicity is what ensures its ironic interpretation”(p. 117). Second, Wray noted that formulaicity must be recognized in creative expressions that have clearly been derived from prefabricated chunks.
Carter (Reference Carter2004) drew attention to the relationship between language play and formulaicity by referring to creativity that involves pattern re-forming, where speakers manipulate existing strings of prefabricated language to construct new, creative sequences. He contrasted this with pattern-forming creativity, which emerges over time, as speakers echo one another's speech patterns, aligning themselves through recycling and repetition of words and phrases. Cook (Reference Cook2000) also noted the importance of pattern formation in his description of the features of language play, emphasizing the creation of rhythm and grammatical parallels. The resulting discourse patterns that emerge during interaction may remain local patterns of language use, such as the nicknames or in-jokes that develop in intimate relationships. However, in some cases they may spread and become larger patterns that change the way people speak more broadly. For instance, the reality television show Survivor, which originated in Sweden in 1997 and has since appeared in numerous other countries, spawned the phrase to “vote someone off the island.” Given the popularity of the series, this string has quickly become a formulaic way to describe a situation in which a particular group of individuals is winnowed down, resulting in a winner. At present it seems to be used almost exclusively in a joking way.Footnote 1 However, it is easy to imagine a future in which this phrase becomes a common, nonplayful way of referring to instances in which a field is narrowed.
The discussion thus far has not focused specifically on L2 users, but on examining the relationship between language play and formulaic language in general. Yet, this relationship clearly has consequences for both L2 teaching and research. The interest in these topics, as well as recent conceptualizations of language itself and of the learning process as nonlinear, potentially signals the type of major shift that occurred when Hymes's (Reference Hymes, Pride and Holmes1972) notion of communicative competence turned our attention from grammatical correctness to appropriateness, from form to function. Historically, applied linguistics has tended to favor linguistic creativity in the sense that Chomsky used it: Competent language users have the ability to construct and understand an infinite variety of new utterances. This conception has had the effect of sidelining the teaching and investigation of playful and creative language, as well as of formulaic language, which was similarly disregarded (cf. Wray, Reference Wray2002).
In the following sections, I turn to L2 issues, focusing as much as possible on recent work. However, given the fairly recent resurgence in interest in this topic, particularly relevant or illustrative works from earlier will also be included. It is also important to note that while the authors of most of the research and examples presented in this review have not explicitly framed their discussions of language play around the concept of formulaic language, or have done so only tangentially, its presence is evident in many of their examples. Because of that, this review differs somewhat in format from others in this volume. Rather than discussing individual work, I attempt to draw out some initial generalizations regarding formulaic language and language play from parts of many studies, and thus I will largely refer to them in the aggregate. Finally, a major function of this review will also be to identify avenues for future investigations examining the relationship between formulaic language use and language play.
TYPES OF PLAY WITH FORMULAIC LANGUAGE
In this section I draw on two of the ways of organizing types of play already discussed to provide examples of the ways in which L2 users have been observed to play with language. First, the distinction between play with and play in the L2 was noted (Bell, Reference Bell2012; Vandergriff & Fuchs, Reference Vandergriff and Fuchs2009, in press), where the former deals with cases in which the language itself is manipulated for play, and the latter with instances in which the language is used to engage in play. Second, Carter's (Reference Carter2004) discussion of some creative language as pattern re-forming helps us focus on play with formulaic sequences, specifically. Pattern re-forming play disrupts familiar conventions, using recognizable sequences in novel ways. (In what follows, for illustration purposes, I take some liberties with the standard conception of formulaic language by including play that involves manipulations of single words).
Play with the L2 occurs less frequently than play in the L2. Kramsch (Reference Kramsch2009), however, provided an excellent example of intermediate French students in a U.S. college engaging in play with the L2 in an online chat. One student writes “vous n'etes pas des students mais des stupidents” (you are not students, but stupidents). This is bilingual play, not only due to the insertion of the English “students,” but also in the coinage of the playful term “stupidents.” By infixing -pi- the writer manages to jokingly insult his peers by describing them as stupid, while also inventing a word that cleverly echoes both the English word students and its French equivalent, étudiants. Bilingual play with conventional strings is also illustrated among the 4-year-old English learners in a Mandarin-English bilingual preschool. In this example, the words in italics were spoken in Mandarin:
Here, the learners begin by manipulating the formulaic recitation of the days of the week in a way that seems to be playful for them (they are smiling), even though they are following the rules by inserting the appropriate days in the correct order. Ken then changes the game by adding morphological play, incorporating the previous speaker's name to imagine it as the name of a day of the week. Additional examples of adults playing with the L2 can be found in Bell (Reference Bell2012), Pomerantz and Bell (Reference Pomerantz and Bell2007), and many of the studies involving children discussed later. Both of these examples illustrate pattern re-forming language play, although it is possible that they could become pattern-forming play, given repeated use of these new patterns.
More commonly, patterns may be exploited to play in the L2. In the following example, Daoud and Hassan are both intermediate ESL (English as a second language) students in an intensive English program. Daoud has just asked Hassan whether he would be willing to be cloned. The teacher arrives and begins to tease Hassan, joking that she would not want two of him:
The instructor's teasing affords Hassan the opportunity to take on the persona of a person sorely wronged, and after repeated teasing from the teacher and his classmate (omitted), he uses the formulaic sequence “I[’d] like to be alone right now.” In this case, the string itself is not manipulated, but the selection of this particular string allows him to express an exaggerated sense of gloom and, in this way, strike back at his attackers. Additional examples of formulaic play in the L2 can be found in such works as Bushnell (Reference Bushnell2009) and Pomerantz and Bell (Reference Pomerantz and Bell2007, Reference Pomerantz and Bell2011), as well as many other of the articles cited here.
These two types of play—using formulaic sequences to play with the L2 and using them to play in the L2—are evident across the literature on L2 language play. In the majority of the studies, what is exhibited is pattern re-forming play, whereas evidence of pattern-forming play is much less common (although see Evaldsson & Cekaite, Reference Evaldsson and Cekaite2010; Lytra, Reference Lytra2007). This may be largely due to the relative lack of longitudinal studies of L2 development. This recent work has enriched our understanding of how L2 users tend to play with language, but clearly further systematic investigation into this matter is called for, with naturalistic studies needed to complement the growing number of classroom studies.
FUNCTIONS OF LANGUAGE PLAY IN RELATION TO FORMULAIC LANGUAGE
It would seem plausible that formulaic and creative or playful language would perform very different roles in interaction, yet they may in fact be used for similar ends. Formulaic language facilitates efficient communication, from both a psycholinguistic and sociolinguistic standpoint (Schmitt, Reference Schmitt2005; Wray, Reference Wray2000, Reference Wray2002; Wray & Perkins, Reference Wray and Perkins2000). The use of prefabricated chunks of language allows speakers to be more fluent and eases the task of the listener, who is helped by being able to predict what is likely to come next. Formulaic language smooths social interaction, presenting potentially face-threatening acts in socially acceptable ways and constructing individual and group identities. In contrast, language play introduces at least short-term inefficiency. Unusual language, including malformed collocations, takes longer to process (e.g., Millar, Reference Millar2011). Such language also carries with it some social risk, as playfulness may be seen as disruptive or inappropriate. Language play can, however, also result in long-term rewards.
The ability to play with formulas (use language creatively) and to construct new (playful, creative) formulas has the potential for rich rewards in the social realm, allowing us to perform a range of social actions. These include constructing and reaffirming (group and individual) identities, negotiating shifting alignments, mitigating face-threats, and expressing changes in stance and footing (Cekaite & Aronsson, Reference Cekaite and Aronsson2005; Evaldsson & Cekaite, Reference Evaldsson and Cekaite2010; Huth, Reference Huth2010; Lytra, Reference Lytra2007; Pomerantz & Bell, Reference Pomerantz and Bell2007; Talmy, Reference Talmy, Kasper, Nguyen, Yoshimi and Yoshioka2010; van Dam, Reference van Dam and Kramsch2002). Evaldsson and Cekaite's (Reference Evaldsson and Cekaite2010) article, for example, presents several extracts demonstrating the sensitivity of a multilingual group of school children to incorrect Swedish formulations and how they reacted swiftly, teasing the offender and thereby upholding monolingual Swedish norms. For these students, formulaic ritual insults (e.g., “eat pooh”), delivered in their home languages, served as a means of getting around classroom regulations regarding so-called good language use. In an example that points to the impact on learning that sanctioned language play activities can have, Pomerantz and Bell (Reference Pomerantz and Bell2007) described an event in which one learner, who was deeply invested in his identity as a good-natured class clown and often engaged in off-task behaviors, became highly focused on accurately translating the formulaic language of mathematical story problems into Spanish. His work with the L2 was entirely serious, and in doing it, he expanded his vocabulary, while at the same time, as it turned out, creating a trick question to pose to the other team, thus maintaining his salient identity in this class.
Furthermore, students exploit formulaic language for playful purposes in order to cope with classroom tasks and/or to resist or critique the pedagogical approaches used and the classroom norms of interaction (Blackledge & Creese, Reference Blackledge and Creese2009; Bushnell, Reference Bushnell2009; DaSilva Iddings & McCafferty, Reference DaSilva Iddings and McCafferty2007; Garland, Reference Garland2010; Lytra, Reference Lytra2007; Pomerantz & Bell, Reference Pomerantz and Bell2011). Blackledge and Creese (Reference Blackledge and Creese2009), for instance, documented strong resistance from a group of heritage language learners of Cantonese, who drew on formulaic sequences from the media (the one million pound question) and stylized overuse of conventional ways of providing feedback (uh-huh uh-huh) to disrupt a dictation activity. Similarly, Pomerantz and Bell (Reference Pomerantz and Bell2011) explored the ways that language play can serve as a kind of pedagogical safe house in language classrooms, allowing learners to express resistance to certain classroom practices. We show how the use of a formulaic sequence, such as me gusta in one Spanish class, is playfully exploited to signal adherence to classroom norms of L2 use and engagement with the activity at hand, while simultaneously mocking both of these.
Cognitively, too, language play may occupy a reciprocal role with formulaic language in L2 development and might allow for a resolution of some of the tension between analytic and formulaic learning. Although not explicitly naming formulaic language in a discussion of language play, Tarone (Reference Tarone, Swierzbin, Morris, Anderson, Klee and Tarone2000) touched upon this relationship, using Bakhtin's (Reference Bakhtin1981) notion of centripetal and centrifugal forces. In this view, language exists with the competing forces of normalization and creativity. The learning of prefabricated chunks contributes to normalization, that is, conventional ways of saying things. Play, on the other hand, as a type of creative language use, can contribute to the forces of change. Because of this, language play may aid in preventing fossilization (Tarone, Reference Tarone, Han and Odlin2006) and in restructuring and further developing the L2 system (Kim & Kellogg, Reference Kim and Kellogg2007; Tin, Reference Tin2011). There is growing evidence that an additional cognitive benefit of play that relates to the learning of formulaic language is that it acts as a memory aid (Bell, Reference Bell2012; Lucas, Reference Lucas2005; Tocalli-Beller & Swain, Reference Tocalli-Beller, Swain and Mackey2007) and thus will ultimately result in greater efficiency.
LANGUAGE PLAY AND FORMULAIC LANGUAGE AMONG ADULTS/ADOLESCENTS VERSUS CHILDREN
The study of language play among children has a much longer tradition than among adults, but early work tended to focus on L1 learners, with L2 learners receiving more attention only in recent years. Formulaic sequences and language play are quite common and appear to be closely linked in children—both L1 and L2 learners—from an early age. For young L2 learners, in particular, formulaic language is a rich resource for language play, providing children with slots that can be filled in unorthodox ways, allowing them an opportunity for both practice and analysis. On the other side of the coin, for these learners, language play may function as a way of building up formulaic sequences through experimentation. Often—perhaps more often than for adults—such play occurs as private speech (Lantolf, Reference Lantolf, Glass and Perez-Leroux1997). Examples of play with formulaic sequences can be found in Cekaite and Aronsson (Reference Cekaite and Aronsson2004, Reference Cekaite and Aronsson2005), Kim and Kellogg (Reference Kim and Kellogg2007), Lytra (Reference Lytra2007), Howard (Reference Howard2009), Wang and Hyun (Reference Wang and Hyun2009), and Evaldsson and Cekaite (Reference Evaldsson and Cekaite2010). Taken together and compared with the data reported in studies of adult L2 users, the results of these studies also suggest that, while children play with all levels of language, they may tend to engage in more play with language (vs. play in language) than adults, and particularly more phonological play. Recent work also shows young L2 learners playfully exploiting formulaic sequences to achieve a wide range of social actions (DaSilva Iddings & McCafferty, Reference DaSilva Iddings and McCafferty2007; Evaldsson & Cekaite, Reference Evaldsson and Cekaite2010; Lytra, Reference Lytra2007). Finally, although play activities tend to be more common in classrooms with younger learners, Gardner (Reference Gardner2008) provided one of the few descriptions of such an activity that involves play with formulaic language. Gardner observed a kindergarten teacher who transformed a traditional literacy activity into a game that, in requiring the students to justify their initial formulaic responses, pushed them into creative play.
In studies of adolescents (Blackledge & Creese, Reference Blackledge and Creese2009; Broner & Tarone, Reference Broner and Tarone2001; Talmy, Reference Talmy, Kasper, Nguyen, Yoshimi and Yoshioka2010) and adults (Bell, Reference Bell2012; Bushnell, Reference Bushnell2009; Garland, Reference Garland2010; Pomerantz & Bell, Reference Pomerantz and Bell2007, Reference Pomerantz and Bell2011), the performative element of play with formulas seems to become increasingly important, in contrast to the private play that may be seen more in young children. In what may be a related development, playing in the language seems to occur much more often than playing with it (Bell, Reference Bell2012). Thus formulaic sequences are manipulated in the joint construction of fictionalized scenarios (e.g., imitation or creation of a particular persona or character) to a greater extent than in the construction of play with forms or semantic play. With greater life experience, older learners also have an increasingly broad range of resources from which to draw in constructing play with conventional language. For instance, both Talmy (Reference Talmy, Kasper, Nguyen, Yoshimi and Yoshioka2010) and Blackledge and Creese (Reference Blackledge and Creese2009) documented L2 students mocking lower-proficiency learners through the use of formulaic utterances, such as “we no English!” (Talmy, Reference Talmy, Kasper, Nguyen, Yoshimi and Yoshioka2010, p. 230), that index low L2 proficiency in a stigmatizing way. Similarly, adult social networks value different things from children's networks, so language play is used in the service of additional social functions. For example, Garland's (Reference Garland2010) study of mock translation contains examples of adult learners of Irish playing with formulaic sequences as a way of indexing their authentic relationship to the language, an example of which is provided later, in the section on identifying play with formulaic sequences.
LANGUAGE PLAY IN COMPUTER-MEDIATED COMMUNICATION
The scholarship cited thus far has almost exclusively involved work with oral L2 production; however, L2 language play has also attracted the interest of scholars of computer-mediated communication (CMC; Belz & Reinhardt, Reference Belz and Reinhardt2004; Kern, Reference Kern and Ayoun2008; Kramsch, Reference Kramsch2009; Vandergriff & Fuchs, Reference Vandergriff and Fuchs2009, in press; Warner, Reference Warner2004). This medium affords play with new types of language, such as acronyms, emoticons, and punctuation or other symbols. Asynchronous CMC can also give learners additional time to construct utterances, which may allow them to be more creative in their use of language. Communicating by computer, rather than face-to-face, might encourage greater linguistic risk taking as well. Perhaps for these reasons, CMC has widely been thought to encourage play (e.g., Kern, Reference Kern and Ayoun2008; Kramsch Reference Kramsch2009). Yet, by comparing the same task carried out face-to-face and in a chat room, Vandergriff and Fuchs (Reference Vandergriff and Fuchs2009) demonstrated that the medium itself does not result in a significant difference in the amount of play. However, it seems possible that the medium may encourage the playful manipulation of formulaic sequences with an eye toward constructing particular L2 identities online (Kramsch, Reference Kramsch2009).
Although the studies already cited have largely focused on play with the L2, a second line of research has begun to examine language use while engaged in computer gaming or play in the L2. These studies seek to describe the interaction that occurs around such activities, as well as the affordances this type of interaction offers language learners. Piirainen-Marsh and Tainio (Reference Piirainen-Marsh and Tainio2009b), for instance, documented the use of other-repetition as a resource for participation by L2 users of English in a game with English voiceovers and narration. The young players in their study performed direct repetitions of characters’ phrases (e.g., piece of cake) and also recontextualized characters’ utterances, using them appropriately in new situations, sometimes even expanding upon them. Similarly, Zheng, Young, Wagner, and Brewer (Reference Zheng, Young, Wagner and Brewer2009) demonstrated how a new formulaic phrase (“hang on”) was noticed and subsequently used by a Chinese learner of English during online collaborative game-play with an English-speaking peer in the United States.
Leppanen and Piirainen-Marsh's (Reference Leppanen and Piirainen-Marsh2009) examination of the playful practice of fan fiction writing also demonstrates how L2 users play in English, through repetition and imitation of character utterances. They also found play with English, as writers manipulate typical formulas that are used in the games they write about, allowing them to parody that language. For additional work in this vein, see Piirainen-Marsh and Tainio (Reference Piirainen-Marsh and Tainio2009a), Piirainen-Marsh (Reference Piirainen-Marsh2010), and Thorne, Black, and Sykes (Reference Thorne, Black and Sykes2009). Having surveyed recent research, in the next two sections I address issues that researchers in this area will need to consider. I first discuss the challenges identifying play with formulas and finally turn to areas for future research.
IDENTIFYING L2 PLAY WITH FORMULAIC LANGUAGE
Both play and formulaic language can be difficult to identify, and examining the two together potentially complicates the situation even more. With respect to language play, Cook (Reference Cook2000) advocated looking for certain formal, semantic, and pragmatic features to identify the phenomenon. Examples of formal features might include repetition and pattern creation; semantic features include ambiguity and construction of alternate realities; and pragmatic features can be markers of performance or language that clearly functions to increase either intimacy or aggression. To identify formulaic language, Wray (Reference Wray2008) advocated the use of intuition as a first step and then proposed 11 criteria for evaluating intuitive judgments. Two of these criteria, as noted in the introduction, provide for the linking of both language play and formulaic language. Wray's criteria are also intended to be applied with a wide variety of speakers of varying degrees of fluency and proficiency, including nonnative speakers. For instance, criterion J allows the analyst to name a string as formulaic, based on intuition, but to also assess it as unintentionally misapplied to the context. Identifying both play and formulaicity creates new challenges. Is a formula being accidentally misapplied or intentionally played with? Interlocutor reaction is not necessarily going to help, as language play by L2 users is regularly misconstrued as error (e.g., Prodromou, Reference Prodromou2007). Looking at a few examples will help illustrate the issues involved. The first example comes from a 1998 recording between two high-intermediate L2 users of English and their American interviewer, taken from unpublished data collected as part of a dissertation study (Salsbury, Reference Salsbury2000). The three had been discussing U.S. foreign policy in the Middle East and especially the possibility of a U.S.-led invasion of Iraq in 1998. The interaction between the interviewer and one of the interviewees, who was from the Middle East, had become somewhat heated when the following took place:
Moussa's description of Saddam Hussein as a “bad, bad boy” seems clearly to be language play, demonstrating his understanding that this particular formulaic sequence, one that is normally used to refer to children, is unusual here. It would be difficult to argue that this is an error, as Moussa was well beyond the stage of learning where everyday language and terminology is something he struggled with. Similarly, given the uncomfortable amount of tension that had developed between the other two participants, it seems likely that this utterance was constructed intentionally, with the goal of lightening the atmosphere, which indeed it succeeded in doing, as the conversation turned to other topics. Not all sequences are as easy to identify and analyze.
Another previously unpublished example from Salsbury's (Reference Salsbury2000) corpus, also involving Moussa, shows how the situation can quickly become more complex. Here Moussa described how his strategies for learning English have changed over time:
Wray (Reference Wray2008) noted that examples with errors, as Moussa's “slow by slow” seems to be, need special handling and should be assessed partly on the basis of attested forms in the data and partly on the basis of the form being targeted. It seems clear that Moussa's use of “slow by slow” was built on the formulaic sequence he used subsequently, “step by step.” He might even have been thinking of “little by little” and thus drawing on a broader hypothesis in which “x by x” is a sequence that can be manipulated in a variety of ways to describe things that take place slowly. However, is it play? Did Moussa intentionally manipulate this sequence in a nonserious way? These things are more difficult to ascertain. Neither he nor the interviewer take that sequence up playfully; however, neither does Moussa's change to “step by step” seem to be repair. The tone of the conversation is generally pleasant and light-hearted, but not overtly marked as play. Furthermore, taken from Lantolf's (Reference Lantolf, Glass and Perez-Leroux1997) view of play as private speech, it does not seem too far afield to suggest that Moussa's formulation could have been a type of rehearsal. In addition, it is certainly not unheard of for individuals to engage in language play that their interlocutor will not understand and that is intended only for the speaker's own private enjoyment.
Code-switching and translations can also be cues to play, but they may also add an additional layer of complexity to the identification of formulaic language. In the following example, adapted from Garland (Reference Garland2010, p. 33), Candice guessed at the formulation of “cool” in Irish, using mock translation:
Here we have play in the L1 (English), but also with the L2 (Irish). As Garland pointed out, even though Candice admitted a lack of proficiency by asking for the meaning of a word, she simultaneously demonstrated proficiency by showing knowledge of Irish constructions through her mock translation of them into literal English: In Irish, I like is constructed as is good with me. Another formulaic way of expressing emotion in Irish is to say that the emotion is on me. The author rightly noted that while this was play, it was impossible to identify whether Candice was playing with the to be + preposition formula in general, or with the “on me” one more specifically in saying that she was under cool. We also cannot determine whether she was deliberately manipulating these formulas (changing with me to to me, or changing on to under) or simply could not recall those forms. In either case we would identify her utterances as play; however, a deliberate transformation would signal a higher level of proficiency.
DIRECTIONS FOR FUTURE RESEARCH
As this review has indicated, investigations into language play and formulaic language have largely taken place without reference to each other. Linguistic creativity has been a concern of scholars examining formulaic language, but largely insofar as how to treat it when identifying and defining formulaic sequences, or to consider and compare analytic versus formulaic construction and comprehension of utterances (i.e., Chomsky's sense of creativity in language). Similarly, those studying language play have used the manipulation of recognizably formulaic language as a marker of play, without considering the relationship between the two. Yet, this review makes clear that this relationship needs to be examined. Rich descriptions of both phenomena exist, and both areas are moving toward explanatory models. Given the close and complementary connection between the two, a unified model that articulates the relationship between the functions and processing of formulaic language and language play should be developed to explain which drives the other, or to what extent is the relationship mutual.
From both a theoretical and pedagogical perspective, L2 proficiency is one factor that needs additional investigation. The ability to use formulaic language appropriately has been recognized as an important factor in L2 proficiency; thus rather than needing to encode or decode each utterance anew, a proficient L2 user can rely on formulaic chunks. In fact, skillful use of conventional language sequences can allow a learner to outperform his or her competence (Wray, Reference Wray2008). With regard to language play, Cook (Reference Cook2000) first suggested that it might be used to measure L2 proficiency. This idea has since received numerous mentions in the literature, but the extent to which Cook's proposal is accurate has not yet been subjected to systematic inquiry.
Still, it seems reasonable to assume that with greater proficiency will come greater control of both formulaic language and the ability to play with it. In many instances, the ability to playfully manipulate—and understand the playful manipulation of—formulaic strings will indeed demonstrate that the user is “sensitive not only to what is possible, but also what is likely” (Wray, Reference Wray2008, p. 240). This ability allows the user to create or comprehend language that is “sufficiently right to be acceptable but sufficiently wrong to cause amusement” (Wray, Reference Wray2008, p. 240). However, we must keep in mind that both formulaicity and play exist on continua, and it is not the case that all formulaic language is equally common, nor that all language play is equally sophisticated (Carter, Reference Carter2004; Nattinger & DeCarrico, Reference Nattinger and DeCarrico1992). For instance, Davies (Reference Davies2003) described how one beginning English learner played with the formulaic greeting What's up? by inverting it to ask What's down? This provides an example of what I have referred to as “conventional creativity” (Bell, Reference Bell2005, p. 201) and suggests that not all demonstrations of ability to play with linguistic conventions will be considered on the same level.
Second language acquisition researchers regularly bemoan the lack of longitudinal studies of L2 acquisition, and indeed, this is another area in which such work would be welcome and would aid in addressing this question of proficiency. In addition, by tracing the development of language play and formulaic language, we could begin to understand the links between them and how each contributes to L2 learning. It is possible that language play can serve as evidence of the developmental path of formulaic language (cf. Durrant & Schmitt, Reference Durrant and Schmitt2010) and vice versa. This will also help in comparing age differences. Following Ellis, Simpson-Vlach, and Maynard (Reference Ellis, Simpson-Vlach and Maynard2008), we might ask what formulas learners play with most. Does play tend to occur more after first exposure, or only with chunks that have been firmly established as part of the learner's interlanguage? We should also consider whether and how this differs for various types of learners, such as classroom and naturalistic learners, or adults and children.
The study of formulaic language and L2 language play is complicated by challenges in identifying play with formulaic language versus language that is merely unusual or nonidiomatic, with no playfulness intended. Code-switching, too, complicates the landscape, as formulaic sequences from multiple languages may be combined in a single utterance. We need to examine the similarities and differences in the ways that L2 users play with formulas within and across their L1 and L2, as well as how these are played with multilingually (cf. Kramsch, Reference Kramsch2009). This work should include more studies outside the classroom (e.g., Bell, Reference Bell2005) and studies comparing play with formulaic language in a variety of contexts.
Finally, as with other aspects of L2 development, social factors should not be ignored. The most important may be the status of L2 users as such and the reception of their language use in different contexts and communities. In some situations, nonnative speakers find it difficult to participate in language play, as their manipulations of L2 conventions are often perceived as errors. Prodromou (Reference Prodromou2007) reported on one L2 speaker who felt “punished” (p. 20) when his language play was corrected. Such reactions may well inhibit further attempts at language play. Gender, too, stands out as a possibly fruitful area of investigation. Pomerantz and Bell (Reference Pomerantz and Bell2007), for instance, noted tangentially that the males in their study of a Spanish as a foreign language classroom in a university clearly engaged in more language play. On the other hand, Carter (Reference Carter1999) mentioned that in the CANCODE (Cambridge and Nottingham Corpus of Discourse in English) corpus of L1 data, women seem to be more creative. Both of these comments focus on language play, but in both instances we can wonder what the relationship was between language play and formulaic sequences.
Among the current perspectives on L2 development, complexity or dynamic systems theory (e.g., Larsen-Freeman & Cameron, Reference Larsen-Freeman and Cameron2008; van Geert, 2008; Verspoor, de Bot, & Lowie, Reference Verspoor, de Bot and Lowie2011), as well as emergentism (e.g., Ellis, Reference Ellis2008; O'Grady, Lee, & Kwak, Reference O'Grady, Lee, Kwak, Bhatia and Ritchie2009) each provide an appropriate lens through which to examine questions surrounding the twin phenomena of formulaic language and language play. Other approaches could, however, be used to examine the topic should scholars take an interest. Research that takes seriously the relationship between language play and formulaic language promises to lead us to a better understanding of language itself, as well as of language learning.
