Studies in Second Language Learning and Teaching Department of English Studies, Faculty of Pedagogy and Fine Arts, Adam Mickiewicz University, Kalisz SSLLT 6 (4). 2016. 587-617 doi: 10.14746/ssllt.2016.6.4.3 http://www.ssllt.amu.edu.pl Variations in developmental patterns across pragmatic features1 Qiong Li Carnegie Mellon University, Pennsylvania, USA [email protected] Abstract Drawing on the findings of longitudinal studies in uninstructed contexts over the last two decades, this synthesis explores variations in developmental patterns across second language (L2) pragmatic features. Two synthesis questions were ad- dressed: (a) What are the variations in developmental patterns across pragmatic features?, and (b) What are the potential explanations for the variations? In re- sponse to the first question, previous studies showed that L2 pragmatic develop- ment is a non-linear, dynamic process, with developmental paces varying across pragmatic features (Ortactepe, 2013; Taguchi, 2010, 2011, 2012; Warga & Scholmberger, 2007). These studies revealed that some aspects of pragmatic fea- tures (e.g., semantic strategies of speech acts) develop faster than others (e.g., lexical features such as mitigators). In response to the second question, three po- tential explanations were identified to account for the developmental variations: (a) language-related, (b) situation-dependent, and (c) learner-related explanations, with three subcategories for the language-related explanation: (a) the functions of pragmatic features, (b) the frequency of availability of target features, and (c) the similarity and difference between languages with respect to the target feature. Keywords: L2 pragmatic development; variations; longitudinal perspective; non-linearity 1I am grateful to Dr. Naoko Taguchi for her guidance for this synthesis. I also appreciate the comments from three anonymous reviewers on an earlier draft of this article. 587 Qiong Li 1. Introduction Pragmatic competence in a second language (L2) involves mastering pragma- lingistic and sociopragmatic knowledge (Thomas, 1983), as well as developing efficient control of both knowledge bases when encoding and decoding lan- guage functions in a sociocultural context (Taguchi, 2010, 2011). Pragmalinguis- tics refers to the linguistic resources available to perform language functions, while sociopragmatics refers to a language user’s assessment of the context in which those linguistic resources are implemented (Leech, 1983; Thomas, 1983). Research on L2 pragmatics encompasses two primary areas: the study of L2 use and the study of L2 learning (Kasper & Rose, 2002). The former investigates how non-native speakers comprehend and produce speech actions in the target lan- guage. The latter examines how L2 learners develop the ability to understand and perform those actions in the target language. Additionally, both L2 use and learning are constrained by social conventions and contexts in which learners decide between different pragmatic meanings (Kasper & Rose, 2002). Therefore, L2 pragmatic development involves acquisition of a complex interplay among language, language users, and social contexts (Taguchi, 2012). Two methods have been adopted to study pragmatic development in un- instructed contexts: cross-sectional and longitudinal. Cross-sectional studies compare pragmatic performance among different groups based on different proficiency levels or length of stay. The group differences of pragmatic perfor- mance are perceived as changes that learners exhibit at different learning stages and thus indirectly demonstrate development. In contrast, longitudinal studies trace the trajectory of L2 pragmatic development over a period of time with the same learner(s). The changes captured over time provide insights into the de- velopmental patterns and the factors affecting pragmatic development. This longitudinal perspective directly addresses the developmental process of prag- matic competence, which contrasts with cross-sectional studies’ indirect ways of demonstrating development. Compared with cross-sectional studies, longitudinal investigations are still under-explored in the area of L2 pragmatics, as acknowledged by many re- searchers (e.g., Bardovi-Harlig, 1999, 2000; Kasper & Rose, 2002; Kasper & Schmidt, 1996; Taguchi, 2010, 2011, 2012). It is only within the past two decades when the number of longitudinal studies started to increase. This body of stud- ies have examined various pragmatic features: address terms (e.g., Hassall, 2013), implicatures (e.g., Bouton, 1992), formulaic expressions (Taguchi, Li, & Xiao, 2013), discourse markers (e.g., Polat, 2011), and speech acts such as apol- ogies (e.g., Warga & Scholmberger, 2007), compliments (e.g., Félix-Brasdefer & Has- ler-Barker, 2014), disagreement (e.g., Bardovi-Harlig & Salsbury, 2004), indirect 588 Variations in developmental patterns across pragmatic features opinions (e.g., Taguchi, 2012), offers-refusals (Barron, 2003), refusals (e.g., Ren, 2012), requests (e.g., Li, 2014), and suggestions (e.g., Matsumura, 2001). These studies showed that the developmental paces varied across prag- matic features or aspects. For example, Taguchi’s (2007, 2008) studies revealed that learners’ comprehension developed faster and more accurately for indirect refusals than for indirect opinions. Several studies on speech acts (e.g., Schauer, 2009) uncovered that semantic strategies used to structure a speech act pro- gressed faster than lexical and syntactic forms. These findings suggest that prag- matic development is a non-linear, dynamic process with uneven developmen- tal paces in developmental patterns (e.g., Ortactepe, 2013; Taguchi, 2012; Warga & Scholmberger, 2007). Despite this generalization, with the notable ex- ception of Taguchi’s (2010) synthesis, very few studies have systematically ad- dressed the uneven developmental paces or explored the potential explana- tions for variations in these longitudinal studies. To fill these gaps, this synthesis seeks to describe and explain the varia- tions in developmental patterns across pragmatic features in uninstructed con- texts. Two questions guide this investigation: 1. What are the variations in developmental patterns across pragmatic features? 2. What are the potential explanations for the variations? Before addressing these two questions, the following section will first briefly in- troduce the background guiding this synthesis. 2. Background In interlanguage pragmatics, research has revealed that pragmatic development shows uneven developmental paces across pragmatic aspects and features. This pathway aligns with the non-linear perspective in complex dynamic systems the- ory (CDST), which claims that language development is a non-linear, complex process in which changes emerge at uneven paces. Some changes emerge grad- ually over time, but others occur suddenly and radically (Larsen-Freeman & Cameron, 2008). In language development, different aspects of language abili- ties have different rates of changes and changes in one aspect do not necessarily cause changes in others. As part of language ability, pragmatic competence also shows uneven paces in the developmental process, which is called developmen- tal variations in this synthesis. Several studies of speech acts have revealed that, within the same period of time, the developmental pace with semantic strate- gies was faster than that with pragmalinguistic expressions (e.g., Felix-Brasdefer & Hasler-Barker, 2014; Li, 2014; Schauer, 2004, 2009). The characteristic of non- 589 Qiong Li linearity in pragmatic development and the empirical evidence provide justifi- cation for the exploration of variations in L2 pragmatic development. In addition to non-linearity, the temporal dimension is another key feature in tracing and observing language development in general, and L2 pragmatic de- velopment in particular. The critical role of “time” is emphasized by Ortega and Iberri-Shea (2005): “Many, if not all, fundamental problems about L2 learning that second language acquisition (SLA) researchers investigate are in part prob- lems about ‘time’” (p. 26). Similarly, CDST also stresses this time scale in lan- guage development (de Bot, Lowie, Thorne, & Verspoor, 2013; Ortega & Bynes, 2008). The observation of developmental processes is not limited to static phases themselves, but also involves the transitions between phases within a period of time (de Bot et al., 2007). This time-based view underlines the im- portance of the longitudinal perspective on language development. To capture the developmental variations in pragmatic patterns, the present synthesis fo- cuses on longitudinal investigations of L2 learners’ pragmatic competence. To date, research does not only describe the developmental process but also explains pragmatic development from diverse perspectives. Two categories of theories have been applied to explain pragmatic development: theories with an individual-psychological focus (e.g., cognitive-processing models) and theo- ries with a social practice focus (e.g., sociocultural theory and language sociali- zation; Kasper & Rose, 2002, p. 61). Both approaches are valuable in explaining pragmatic development since the development of pragmatic competence is characterized by the interactions between “cognitive, social, and environmental factors” (Taguchi, 2012, p. 66). The diversity of explanations for pragmatic de- velopment provide insights into developmental variations across pragmatic fea- tures, which can be explored from various perspectives. The non-linear and time-based traits of language development guide this synthesis to address the developmental variations (uneven paces) across prag- matic features investigated in longitudinal studies, as well as to explore the po- tential explanations for these variations. Below, the methodology section will define longitudinal studies, which is followed by the literature search proce- dures used to locate the primary studies. It will also present the inclusion/exclu- sion criteria used to screen the studies and the coding process of categorizing the studies. After this, operationalization of key terms (e.g., developmental pat- terns, variations, pragmatic features) will be explained. Then, in the findings sec- tion, the answers to the two synthesis questions will be presented. Finally, the conclusion section will summarize the synthesis findings and provide implica- tions for future research directions. 590 Variations in developmental patterns across pragmatic features 3. Methodology The present synthesis adopts the criteria proposed by Ortega and Iberri-Shea (2005) and later revised by Taguchi (2010) to define longitudinal research. Ac- cording to Ortega and Iberri-Shea, longitudinal research is the observation of the same participant(s) over an extended period of time. They presented four definitional features of a longitudinal study: (a) the specific time length, (b) the presence of multi-wave data collections, (c) the conceptualization of capturing changes over time, and (d) the focus on establishing contextualized antecedent and consequent relations by tracing the phenomenon in its context, instead of with experimental controls. Drawing on these characteristics, a literature search was conducted using four electronic databases: LLBA, ERIC, Google Scholar, and the Carnegie Mellon University (CMU) library database (a university database). Key words including interlanguage pragmatics, pragmatic development, second language learn- ing/acquisition, andlongitudinal study were used to carry out the bibliographic search. This search, conducted in 2014, yielded 17 hits in LLBA, 286 in ERIC, 453 in Google Scholar, and 308 in the CMU library database. After cross-examining the studies, a total of 765 articles were identified as unique studies. Then, the following inclusion and exclusion criteria2 were used to screen the studies: 1. The study was empirical and was published in a journal or book.3 2. The study focused on adult participants. 3. The study traced the changes of specific pragmatic features4 over a pe- riod of time (it was a longitudinal study). 4. The data in the study were collected systematically over time (e.g., pre- and post-results). 5. The study included pragmalinguistic analysis.5 6. The study focused on naturalistic development: It did not involve in- structional intervention or other types of training.6 2 These criteria are based on those in Taguchi’s (2010) synthesis of longitudinal studies. 3 Doctoral dissertations (e.g., DuFon, 1999) were not included because they were not pub- lished in journals or books. 4 Some longitudinal studies (e.g., Dings, 2014; Shively, 2013) did not meet this criterion be- cause they did not focus on specific pragmatic features. 5 Some ethnographic studies (e.g., Siegal, 1995) did not meet this criterion because they did not code the data for pragmalinguistic analysis. Neither did this study present the changes of certain pragmatic features over the 18 months 6 Some longitudinal studies such as Cohen and Shively (2007) and Shively (2011) did not meet this criterion because their design involved instruction or training. For example, 591 Qiong Li 7. The study showed different paces of development of particular prag- matic features, meaning that within the same period of time and within the same group of learner(s), some pragmatic features/aspects devel- oped faster while others did not develop or developed more slowly. After applying these inclusion and exclusion criteria, 26 primary studies re- mained for the synthesis. 3.1. Coding for substantive and methodological features Each study was first coded for substantive and methodological features (Norris & Ortega, 2006). The substantive features involve target pragmatic features and learning contexts (second or foreign language context). The methodological fea- tures consist of six aspects: (a) length of study, (b) frequency of data collection, (c) measures used to capture development, (d) sample size, (e) participants’ L1(s) and the target language, and (f) participants’ proficiency. The appendix dis- plays all 26 studies coded for these features. As shown in the appendix, the pragmatic features examined include ad- dress terms, formulaic expressions, discourse markers, the sentence final parti- clene, conversational implicatures and speech acts. These studies fall into three broad categories of examination: pragmatic comprehension, awareness/per- ception, and production, with 17 studies concentrating on pragmatic produc- tion. Only seven studies investigated pragmatic comprehension and three stud- ies pragmatic awareness/perception.7 Additionally, most studies were con- ducted in the second language (SL) context. Seven involved the foreign language (FL) context, but only two were conducted exclusively in the FL context; the other five were carried out in both SL and FL contexts, comparing the learning outcomes between the two contexts. Finally, 16 of 26 studies had English as the target language. Other target languages included Spanish, Indonesian, Chinese, Japanese, French, and German. Regarding the length of study, there is little consensus about the optimal length of observation for a longitudinal study (Ortega & Iberri-Shea, 2005). The 26 studies confirm this claim, exhibiting a wide range of the study length (from four weeks to over four years). Most studies used convenience scaling based on institutional time (e.g., one academic year). The frequency of data collection points ranges between two and eight times, with a majority of studies involving Shively’s (2011) study involved an hour-long orientation to pragmatics and approximately 30 minutes of instruction on pragmatics before participants started doing their tasks. 7 Taguchi’s (2012) study was coded twice because it included two separate analyses of prag- matic comprehension and production. 592 Variations in developmental patterns across pragmatic features two or three points. The data were written or oral production elicited by instru- ments such as the discourse completion task (DCT) and questionnaires. A few studies collected authentic data by recording naturalistic conversations (e.g., Bardovi-Harlig & Hartford, 1993; Bardovi-Harlig & Salsbury, 2004; Polat, 2011) or administering role-plays (e.g., Bataller, 2010). 3.2. Coding for developmental variations and explanations In this step, each study was further coded for the aspects that directly answer the research questions: (a) pragmatic features that developed and did not de- velop/developed slowly, namely, developmental variations, (b) evidence (e.g., test scores, frequency) for faster and non-/slower development, and (c) poten- tial explanations for the developmental variations. The key terminology used in this coding process was defined. First,prag- matic features were defined as the operationalization of pragmatic competence in each study. Specifically, they refer to the units of analysis (e.g., address terms) researchers used to examine pragmatic competence. Second, themeasurement of developmentand non-development was adopted from previous studies that regarded target language speakers’ performance as the normative use of a cer- tain pragmatic feature. Within the same observation period, the changes that approximate the target-like norms indicate faster development, while the diver- gence from the target-like norms or no change represents slower or non-devel- opment.8 Third, developmental patterns were defined as paces or rates of changes of pragmatic features. These changes were traced in the performance of the same leaner(s) within the same period of time. Finally, variations were operationalized as the uneven developmental paces or rates across pragmatic features, which means that variations involved two aspects: faster and non- /slower development of the pragmatic feature.9 When coding for the potential explanations for the variations, I did not simply copy the authors’ explanations reported in their studies. Instead, I re- analyzed the study features and the findings/results reported in each study. This 8 The distinction between faster and non-/slow development was based on the evidence (e.g., test scores, frequency) presented in the original studies. For example, the accuracy scores in Taguchi’s (2012) study showed significant differences of low-imposition speech acts between the first and the second data collection points, as well as between the second and the third data collection points. However, high-imposition speech acts showed signifi- cant differences only between the second and the third data collection points. Thus, low- imposition speech acts developed faster than high-imposition speech acts. 9 Since variations cover these two aspects, the corresponding explanations should also ad- dress the two aspects at the same time. 593 Qiong Li re-analysis process relied on the information provided by the author(s) in the original studies and resulted in three potential explanations: (a) language-re- lated, (b) situation-dependent, and (c) learner-related explanations. Each single explanation was identified to account for both faster and non-/slower develop- ment simultaneously because developmental variations involve these two as- pects of development. Based on the language properties (e.g., frequency) in the studies, three sub- categories in the language-related explanation were further identified: (a) the functions of the target feature, (b) the frequency of availability of the target fea- ture, and (c) the similarity and difference between L1 and L2 for the target fea- tures. The situation-dependent explanation concerns the characteristics of the task items used in the study, specifically, the social variables involved in the task situations, that is, degree of imposition, social distance, and social status (Brown & Levinson, 1983). Finally, the learner-related explanation refers to the learners’ initial conditions (e.g., initial-level pragmatic knowledge) at the start of the study. When each study was categorized into a specific explanation, the explanation has to account for both faster and non-/slower development at the same time. Of the 26 studies, 20 studies were categorized into the language-related explanation, six studies into the category of the situation-dependent explana- tion, and two studies into the category of learner-related explanation. To im- prove the reliability of this categorization, three pragmatics PhD students re- categorized six different studies (about 20% of the 26 studies). Each of them selected two studies s/he had read from the 26 studies. None of them selected the same article. After this, they received the variation description found by the researcher in each study. The co-coders were required to come up with their own explanations for the variations they received. They were informed that the explanation that they came up with had to account for both faster and non- /slower development. None of the co-coders knew any of the explanations the researcher identified. The agreement rate between the researcher’s and the co- coders’ explanations was 83.3%, with one disagreement on a study. Agreement was achieved after the discussion of the variations in that study. 4. Synthesis findings This section synthesizes the findings of the 26 studies to answer the two ques- tions: (a) What are the variations in developmental patterns across pragmatic fea- tures and (b) What are the potential explanations for the variations? Since the variations found in these studies are associated with explanations, the findings for these two questions are presented together according to the three explanations: (a) language-related, (b) situation-dependent, and (c) learner-related. 594 Variations in developmental patterns across pragmatic features 4.1. The language-related explanation for developmental variations The first category of studies concerns the developmental variations caused by the language-related factors. Among the 26 primary studies, 20 studies fall into this cat- egory: one study of pragmatic awareness/perception, seven studies of pragmatic comprehension, and 12 studies of pragmatic production. As shown in Table 1, three specific types of language-related explanations were further identified: (a) the func- tions of pragmatic features, (b) the frequency of availability of the target feature, and (c) the similarity and difference of the target features between L1 and L2. Table 1 Studies in the category of the language-related explanation Target pragmatic Features/aspects that developed or Features/aspects that did not develop or devel- Studies features developed relatively quickly oped slowly Studies in the function subcategory (2 studies) Polat (2011) Discourse markers The use ofyou know andlike The use ofwell Sawyer (1992) Sentence final parti- The formulaic use of the particlene as The use of the affective particlene clene insoo desu ne‘that’s right’ Studies in the frequency subcategory (2 studies) Khorshidi (2013) Requests Direct strategies and conventionally Hints indirect strategies Ortactepe (2013)** Formulaic expres- Expressions such asI didn’t catch your Expressions such ascome again sions name andget out of here Studies in the L1-L2-similarities-and-differences subcategory (15 studies) Bardovi-Harlig & Suggestions and re- Frequency and the success rate of ini- The use of mitigators and aggravators Hartford (1993) jections tiated suggestions and refusals Bardovi-Harlig & Sals- Disagreements Indirect disagreement strategies (i.e., Lexical devices (i.e., overgeneralized use ofbut in bury (2004) the agreement-before-disagreement disagreement) strategy, postponing disagreement across turns) Barron (2003) Requests, offers, and Discourse structure in offers-refus- Linguistic behaviors (i.e., linguistic forms and in- refusals to offers als10 ternal modification in requests and offers) Bouton (1992, Conversational im- Relevance-based implicatures Pope questions, sequence implicatures, indirect 1994)** plictatures criticism, and irony Félix-Brasdefer & Compliments The range of compliment strategies The use of adjectives in compliments Hasler-Barker (2014) (e.g., an increase in the number of strategy types from four to seven) Hassall (2013)* Address terms Address terms in the vocative slot Address terms in the pronoun slot and the use of pronounsanda andkamu (both meaning ‘you’) Li (2014) Requests External modifiers Alerters and internal modifiers (e.g., mitigators) Ren (2012) Refusals Types of refusal strategies Use of adjuncts to strategies Schauer (2004, 2009) Requests External modifiers and request strate- Lexical and syntactic downgraders (e.g.,a bit, gies (fewer direct strategies and more wouldyou mind) conventionally indirect strategies) Taguchi (2007, 2008a, Conventional implica- Conventional implicatures (i.e., indi- Non-conventional implicatures (i.e., indirect 2008b, 2012)** tures rect refusals) opinions) and language-specific conventional implicatures (i.e., routines) Warga & Scholmerge Apologies Apology strategies (e.g., justifications, Linguistic forms (e.g.,tres ‘very,’vrainment ‘re- (2007) offers of repair) ally’) Note.* studies of pragmatic awareness/perception; ** studies of pragmatic comprehension (other studies focused on pragmatic production). 10 This study only focused on offers-refusals in the discourse structure. 595 Qiong Li 4.1.1. Functions of pragmatic features Pragmatic functions vary across features, with some serving simple functions and others affording complex functions. It is easier for learners to register the simple form-function mapping such as the one-to-one mapping of a pragmatic feature. Learning and using these features do not require much processing and retrieving workload. In contrast, the complexity of functions conveyed by the pragmatic features may slow down learners’ pragmatic development because learners have to establish the multiple form-function relationships and use the specific function based on their assessment of the context. One example illus- trating the function effect on developmental paces is the Japanese sentence fi- nal particlene investigated by Sawyer (1992). The study examined 11 Japanese learners’ use ofne in four semi-structured interviews over one year. The findings showed that learners developed a sufficient command of the formulaic use of ne as insoo desu ne‘that’s right/I see,’ but their progress was slow with the use ofneas an affective particle. The faster development of the formulaicne is likely due to the function it serves in the conversation. As Sawyer (1992) claimed, soo desu ne serves an all- purpose function. Interlocutors can use this expression to keep the conversation going without other responses. For L2 learners who have limited interactional competence, such a common, all-purpose expression is useful, which probably facilitates learners’ development of the formulaic use ofne. In contrast, the affec- tivene is difficult because it has complex interpersonal functions: to make a con- versation flow smoothly by showing consideration of the hearer or involving the hearer in the conversation (Sawyer, 1992). Thus, the use of the affectivene seems to be more demanding because L2 learners have to monitor the flow of conver- sation and identify the right timing to use the particle for interpersonal purposes. This function explanation is supported by Polat’s (2011) study, which traced the development of three discourse markers (DMs), that is, you know, well, andlike,11 by an untutored Turkish adult over one year. The analysis of 24 recordings of informal conversations between the researcher and the learner revealed different paces of changes across the three DMs. The use ofyou know andlike showed an approximation toward the target-like pattern. The frequency of you know declined by 50% over the year (from 4,122 to 2,019 occurrences per 100,000 words), even though this frequency was still higher than that in na- tive speakers’ (NSs) utterances (580 occurrences per 100,000 words). The use of 11 In the study, the analysis of a corpus, Santa Barbara Corpus of Spoken American English (SBCSAE), showed that the three DMs have similar frequency in the native speakers’ utter- ances (see details in Polat, 2011). 596