Junghui Su is a recent graduate from the Department of Education, University of Oxford, with her Master’s dissertation on factors influencing native Mandarin speakers’ L2-English spoken fluency. She is currently working as a teacher of (GCSE/A level) English at Longpanhu International School, China, teaching GCSE/A Level English as well as test-oriented courses  (IELTS and SAT). In addition, she is appointed as the teacher trainer for the Department of English at the school, holding weekly seminars and training sessions for English teachers. Her research interest is in the development of L2 learners’ oral proficiency, particularly that of L2 English and L2 Mandarin learners. In addition to language acquisition, she is interested in research methodology and language assessment, especially designs for speaking tests.

Video Abstract

Written Abstract

The majority of Taiwanese university students obtain high proficiency in all domains of L2-English but speaking. Research often attributes this imbalance to insufficient opportunities to speak in the general context where most Taiwanese students learn English. Indeed, L2 classrooms in Taiwan have been criticised for neglecting oral skills, and students are often anxious about their lack of L2 speaking competence. Concerning the worrisome phenomenon, rarely has research sought to explore students’ self-efficacy beliefs developed in this learning context or examine the impact of self-efficacy on speech production. The present study therefore targeted 45 Taiwanese university students who had received typical L2-English education in the country to explore their self-efficacy and how these perceptions influenced L2 oral fluency. It used a mixed-methods approach, incorporating vocabulary, grammar and speaking tests, an interview and a questionnaire to examine the relationships between actual linguistic knowledge and spoken fluency, actual and perceived linguistic competence, and self-efficacy beliefs and spoken fluency. The primary findings were based on statistical models such as correlation and regression, as well as on qualitative interpretations of participant feedback. Through these empirical investigations, this study identified two variables: actual vocabulary knowledge and vocabulary self-efficacy, as statistically significant predictors of L2 spoken fluency. That is, these two factors made independent contributions to students’ spoken fluency even when all other variables were controlled for. These findings contribute to our understanding of the nature of self-efficacy and its influence on speaking. By further implication, this study provides a foundation upon which pedagogical insight regarding the teaching of L2 speaking in Taiwan could be drawn.


Taiwanese EFL students’ L2 speaking, compared with other domains of their L2 proficiency, is often considered weaker (e.g., Huang, 2012; Teng, 2005) and attributed to insufficient opportunities for practice in the general context where most of them learn English (Hsu & Chiu, 2008; Huang, 2012; Lai, 2009; Wu, 2012). A series of national centralised exams, with English included as a major subject covering domains of listening, reading and writing, have shaped English curricula in the Taiwanese education system and, inevitably, led to a trend where L2 proficiency is measured by paper-based performance and where communicative activities are nearly absent. It is often not until tertiary education that speaking is officially incorporated into the classroom. Except few who have received training at private institutions or abroad, most Taiwanese students enter university with little experience in L2 oral communication despite having achieved considerable levels of proficiency in other domains of the language after ten years of compulsory EFL education. In university settings, these students become anxious about their lack of L2 speaking competence (Liu, 2005; Yu, 2008) under common practice of CLT approaches and graduation requirements, both of which consider English speaking an essential component. Despite the situation, rarely has research sought to explore solutions to the disadvantaged development of these students’ L2 speaking competence in the given context.

Studies on Taiwanese students’ L2 proficiency often overlook the considerable imbalance between productive and receptive skills, specifically between oral competence and linguistic knowledge. Very few have yet investigated the claim that psychological self-beliefs of ability also affect fluency (e.g., Graham, 2006). The present study therefore targets students under the described circumstances with a focal attempt to examine factors relevant to L2 spoken fluency. In particular, it looks into how self-efficacy beliefs directly influence, and mediate the impact of L2 linguistic knowledge on, L2 spoken fluency. Through empirical investigation of the relationships between linguistic knowledge, self-efficacy and spoken fluency, this study contributes to our understanding of the nature of self-efficacy and its influence on L2 performance. Its ultimate aim is to provide a foundation upon which pedagogical insight into the teaching of L2 speaking in Taiwanese universities could be drawn.

Literature review

Understanding fluency of speech

With its nature primarily bound to time, fluency is a real-time performance skill distinguishable from knowledge and other proficiency components (Lennon, 1990; Schmidt, 1992). Although it is agreeably a construct widely adopted in the SLA literature, little consensus has been reached on its precise definitions (Hieke, 1985; McCarthy, 2009). There does not exist a uniform or cross-contextual measure of fluency given the diverse perspectives taken to approach it. Some studies use temporal measures to examine fluency as a monologic performance, while others take into account social aspects of speech and resort to human judges to assess fluency (McCarthy, 2009). Koponen and Riggenbach (2000) listed four categories of conceptualisation which respectively regarded fluency (a) a performance phenomenon captured in temporal characteristics; (b) the automatisation of oral production processes; (c) hearer-based perceptions of speech flow; and (d) a pedagogical element of oral proficiency as opposed to accuracy. On this subject, Fillmore (1979) also identified four levels of spoken fluency, including (a) the temporal flow of speech; (b) the logical coherence and comprehensibility of speech; (c) the sociopragmatic appropriation of speech; and (d) the originality and wittiness of speech.

Many researchers consider temporal characteristics of speech performance as indications of the level of automaticity and therefore carry out calculations using a variety of means. For instance, speaking rate and mean length of run are two common measures for speed of delivery. Recent research has also proposed taking the distribution, the length and the nature of pauses in utterances as important temporal measures (e.g., Chambers, 1997; Kormos & Dénes, 2004; Mochizuki & Ortega, 2008; Riggenbach, 1991; Towell et al., 1996; Wolf, 2008).

Although widely supported, fluency as calculated by temporal measures is not without debate. Emerging evidence across disciplines suggests that speakers’ speech rate and pausing depend not only on automaticity, but largely on interlocutors, given contexts and speech genres (e.g., Giles, Coupland & Coupland, 2010). This indicates that speed of delivery and pausing, sensitive to contextual conditions, are complex phenomena (Goldman-Eisler, 1972) which may result in increased ‘flow’ of speech in communication (i.e., fluency). For the above reasons, as Housen and Pierrard (2005) pointed out, it remains an ongoing quest for researchers to draw boundaries which distinguish sped-up processing and automaticity.

On the other hand, a number of researchers suggest that fluency be defined based on hearers’ perceptions. With regard to approaches, some appoint professional teachers or examiners for fluency assessment (e.g., Derwing, Rossiter, Munro & Thomson, 2004; Kormos & Dénes, 2004), while others seek this from non-professionals (e.g., Chiswick & Miller, 1998; Yeh & Inose, 2003). There is also research into features of speech that influence listeners’ perceptions of fluency. To determine the best predictors of perceived fluency, studies have combined temporal and perception measures by comparing ratings given on a fluency scale and the statistical analysis of temporal variables (e.g., Freed, 1995; Lennon, 1990; Riggenbach, 1991). However, a large number of studies have neglected to place restrictions on definitions of fluency, thus resulting in diverse conclusions of which temporal factors contributed to perceived fluency.

Rublik (2006) in an ethnographic case study pointed out that factors affecting fluency can fall under linguistic, contextual, cultural, or affective dimensions. To start with, robust empirical evidence has been shown in favour of the contribution of vocabulary knowledge to L2 speech production (Daller & Xue, 2007; Hilton, 2008; Koizumi & In’nami, 2013; Laufer & Nation, 1995; Vermeer, 2004). In contrast, grammar knowledge has shown a limited impact on fluency (e.g., Hilton, 2008); only a few studies have exhibited the influence of accuracy, arguably attributable to grammar knowledge, on perceived fluency (e.g., Kormos & Dénes, 2004; Riggenbach, 1991).

Socio-cultural factors influence oral performance by giving rise to a variety of internal and external pressures under which L2 speakers are placed. Contextual factors such as interlocutors, task types, planning time also affect fluency. Involvement and backchannel responses from an interlocutor correlate with L2 speakers’ speech rate (Wolf, 2008) and narrative skills (Bavelas, Coates & Johnson, 2000); task types provide frameworks within which L2 speakers perform orally and therefore have a direct impact on fluency of speech; time for pre-task planning increases fluency in oral production (Foster & Skehan, 1996; Mehnert, 1998; Ortega, 1999). As for affective variables, research has shown their predictive power in determining the ultimate success in L2 mastery (e.g., Gardner & Lambert, 1972; Krashen, 1982; Rajab, Far & Etemadzadeh, 2012; Vandergrift, 2005; Yashima, 2002).

Among a wide range of factors affecting oral fluency, research has rarely examined the independent contributions of affective factors to oral fluency using monologic task types, where a good command of speech manifests itself. In addition, the majority of these studies merely adopt temporal measures of fluency, neglecting the role of listeners in L2 speaking. The present study, therefore, aims to keep in view aspects of linguistic discourse, sociolinguistic awareness and strategic competence in fluency discussions (Fillmore, 2000). It situates in between the dichotomy posited by McCarthy (2009) and adopts Fillmore’s first two levels of fluency. Using qualitative judgements, it focuses on perceived fluency of speech in terms of its smoothness (operationalised as hearers’ perceived speech flow) and comprehensibility (Fillmore, 2000).

Self-efficacy as a construct

Bandura’s (1986) social cognitive theory introduces the concept of self-efficacy, defining it as an individual’s perceptions of self-competence in accomplishing certain levels of performance. Self-efficacy is situation-specific and determines the courses of action which an individual exerts to attain goals (Bandura, 1995). In academic settings, self-efficacy refers to the perceived self-abilities with which students approach tasks (Pajares & Johnson, 1996) and has received extensive support with respect to its role in mediating the effects of aptitude on performance (e.g., Multon, Brown & Lent, 1991; Pajares, 1996; Pajares & Miller, 1994, 1997).

Bandura (1997) suggested four primary sources of self-efficacy. First, mastery experiences referred to experiences which an individual considers successful, serving as authentic confirmations of one’s abilities in a specific area. According to Bandura, both experience and the interpretations of experience had a large impact on the shaping of self-efficacy. Second, modelling, or vicarious experiences, denoted comparisons made by an individual to others concerning competence and attainments (Bandura, 1997). In terms of the models with whom the individual was compared, similarity in performance and personal attributes (e.g., gender, race, educational background, socioeconomic status) determined the individual’s appraisals of the comparative information (Bandura, 1995).

Third, verbal persuasion pertained to others’ evaluative feedback given to an individual regarding performance and capabilities (Bandura, 1997). The strength of persuasory influence on self-efficacy was determined by the knowledge and credibility of the commentators (e.g., De Saint Léger, 2009; Graham, 2006). Fourth, physiological and affective states concerned biological factors that exerted influence on performance (Bandura, 1986). This source was also identified in Krashen’s (1982) affective hypothesis, which posited that variables such as motivation, confidence, and anxiety significantly impact L2 learning.

In essence, the construct of self-efficacy constitutes perceptions of self-abilities with which an individual approaches new tasks, and it continues to be reconstructed on the basis of relevant socio-psychological influences and prior experiences. Bandura (1997) and a growing body of research have demonstrated that in academic settings, students’ self-efficacy beliefs correlate more with accomplishment (e.g., Lane, Lane & Kyprianou, 2004) and strategy use (e.g., Cotterall, 1999; Yang, 1999) than does actual competence.

Research has brought attention to the importance of self-efficacy beliefs as motivational factors and powerful predictors of achievement in L2 performance (Pajares & Johnson, 1996). Existing SLA studies have almost exclusively pertained to self-efficacy regarding its relationship with motivation, strategy use (e.g., Cotterall, 1999; Klassen, 2002; Yang, 1999), or willingness to communicate (e.g., Fushino, 2010; Zhong, 2013). As for those examining relationships between self-efficacy and language performance, the majority concern the domain of L1 or L2 writing (e.g., Graham & Sandmel, 2011; Klassen, 2002; Pajares & Johnson, 1996). There is, thus, an observable dearth in the literature where the influences of self-efficacy on L2 speech performance remained inconclusive. Furthermore, Bandura (1986) noted that the assessment of self-efficacy ought to correspond with that of performance in research methodology to avoid inaccurate findings (Bandura, 1986; Pajares & Miller, 1997); however, a preponderance of studies have neglected to tailor measures of self-efficacy in accordance with the criteria they used to assess performance or vice versa.

Filling the gap: the present study

The present study seeks to address the above gaps in the literature by incorporating self-efficacy into the study of L2 spoken fluency. It attends to the aforementioned methodological as well as definitional issues in the literature. By viewing L2 fluency a procedural skill on the speaker’s side and measuring it as a perceptual phenomenon on the hearer’s part, the present study draws merits from both perspectives of fluency in terms of definition.

This study identifies potential confounding variables and minimise influences from fluency factors irrelevant to the target investigation. It controls for topic familiarity, anxiety and motivation. In addition, it controls for cultural and contextual variables by drawing participants from a culturally-homogeneous population in which students of the same ethnicity undergo similar classroom experiences and by assigning them an identical interlocutor, tasks, and planning time. Unlike prior studies which neglected Bandura’s (1986) guideline, the present study tailors the assessment criteria of self-efficacy to that of fluency performance. It also keeps track of the self-efficacy sources for in-depth interpretations of quantitative findings.


Research questions

The present study attempts to address the following questions:

  1. Is there a relationship between breadth of productive vocabulary knowledge and spoken fluency?
  2. Is there a relationship between grammar knowledge and spoken fluency?
  3. Is there a relationship between students’ actual and perceived vocabulary and grammar knowledge?
  4. Is there a relationship between students’ self-efficacy in linguistic knowledge and that in speaking?
  5. Does students’ self-efficacy in linguistic knowledge make an individual contribution to their spoken fluency?
  6. Does students’ self-efficacy in L2 speaking make an individual contribution to their spoken fluency?

Definitions of key constructs

In terms of oral performance, the present study falls in between the dichotomy posted by McCarthy (2009) in the sense that it views L2 fluency as an automatic skill of speakers (Schmidt, 1992), while measuring it based on listeners’ perceptions. With regard to linguistic knowledge, this study looks at vocabulary knowledge and grammar knowledge separately, each as independent and controlling variables.

The present study defines self-efficacy as students’ beliefs about their individual capacity in the breadth and utilisation of L2 linguistic knowledge (vocabulary and grammar knowledge) for communication purposes and in general L2 speaking competence. The construct does not include the element of anxiety as does self-confidence (Clément, 1980; Tremblay & Gardner, 1995), the level of which is separately measured and controlled for in the present study.

Research design

This study draws upon a combination of quantitative (QUAN) and qualitative (QUAL) methodologies since it offers a pragmatic alternative to choosing from a competing dichotomy between paradigms of the quantitative and qualitative research traditions (Tashakkori & Teddlie, 2003; Teddlie & Tashakkori, 2009). In accordance with DeKeyser’s (2014) call for mix-methods perspectives, this study addresses all the questions using QUAN data and further investigates them using QUAL data to corroborate general and idiosyncratic interpretations of the relations between linguistic knowledge, self-efficacy and spoken fluency.

Population and sample

The population is the specific group of Taiwanese undergraduate students who are ranked in the upper quartile in English in the national exams but have had very limited English-speaking opportunities throughout their L2 education. Although their English proficiency in reading, writing and listening is far above intermediate, these students struggle to communicate orally in English due to lack of practice.

Participants in the study were recruited from the top-ranked university in Taiwan because students accepted into this university conform homogeneously to the described population criteria (i.e., their English proficiency is far above intermediate). This study targets those among this group who have not been exposed to many opportunities to speak English. Thus, it puts out criteria to exclude students who: (a) have had study abroad experience in an English-speaking environment; (b) have native-level English-speaking parent(s); (c) have received any form of bilingual education; (d) have attended intense English oral training or courses at school or in other settings. Out of the sample (N = 45), there are 22 males and 23 females, aged between 18 and 22. These students all showed high motivation to improve L2 speaking by voluntarily contacting the researcher to participate in exchange for English speaking classes.

Research instruments

This study adopts Lex30, a test specifically designed by Meara and Fitzpatrick (2000) to estimate non-native test takers’ breadth of productive vocabulary knowledge in English. Walters (2012) strongly promoted Lex30 because of its valid prediction of test takers’ productive vocabulary at a high proficiency level. The test scores were transformed to fit a scale of 0 to 100, to match the scoring system of the grammar test. Studies have supported the concurrent validity (Fitzpatrick, 2007; Lemmouh, 2010), the pragmatic validity and reliability (Fitzpatrick & Clenton, 2010), and the internal reliability (Meara & Fitzpatrick, 2000) of Lex30 as a testing tool in a research context.

The grammar test chosen for this study was adapted from the English Placement Test provided by Oxford University Language Centre (OULC). The 100-point scale was the same onto which vocabulary scores was transformed. This grammar test was adopted due to its established validity and reliability in measuring the participants’ grammar knowledge (see Allen, 1992) for communicative purposes.

The self-efficacy questionnaire is designed mainly based on Bandura’s (1986) self-efficacy theory and comprises four parts. Part (a) contains measures of motivation and anxiety levels; part (b) investigates student’s self-perception of L2 linguistic competence; part (c) concerns the four resources of self-efficacy; part (d) collects students’ language education background.

The study operationalised linguistic self-efficacy as students’ self-judgments of their competence in vocabulary and grammar usage in speaking, and L2 speaking self-efficacy as their judgments of spoken fluency. The constructs were measured quantitatively through 62 items using 7-point Likert scales and twelve short-answer questions which provided idiosyncratic information in support of the numeric data.

The duration of the speaking test was approximately 10 minutes. The interview format was semi-structured and comprised three parts: the introduction, the monologue, and the short questions. The participant was given a verbal prompt and a silent preparation period of up to one minute before the monologue.

To control for potential confounding variables, the speaking test was designed based on the following rationales. First, the topic was general to avoid issues of topic familiarity. Second, the test removed interactional demand to maximise participants’ production and ensure minimal intervention from the interviewer, inducing participants’ authentic oral production which is intact from interactional variables.

The choice of raters as measures was justified since temporal measures could overlook the multifunctionality of pauses in human discourse (Færch & Kasper, 1983; Lennon, 1990). The overall fluency score was out of 10, with comprehensibility assessed on a scale of 1 to 3 and fluency on a scale of 1 to 7. Precautions were taken to prevent the bias effect, the halo effect and the leniency-severity effect. The raters were instructed to follow a detailed guideline strictly and gave ratings objectively based on analytic reasons. Strong inter-rater reliability was shown (α = .895, p < .05) in the pilot data.


A pilot test was conducted on 30 students who matched the criteria and shared similar L2 learning backgrounds with the target participants. Observations of the pilot participants’ performance were used to make necessary adjustments to the study design and instruments.

The researcher advertised the study at the university and invited eligible volunteers to participate in the study. Three native English-speaking monolinguals were personally invited by the researcher based on convenience to participate as fluency raters. These postgraduate students from UK universities had limited knowledge in foreign languages, linguistics, and language studies.

Each participant first received the individual speaking test, followed by the linguistic tests and the questionnaire. All tests were administered under test conditions. The researcher anonymised all audio recordings and instructed the three raters to carefully listen to and evaluate the oral data following the guideline. They completed the rating task independently without any information about the participants’ backgrounds.


Intra-rater and inter-rater reliability in fluency scores

The study estimated the consistency between the judges’ ratings using the intraclass correlation coefficient (ICC), which turned out to be high at α = .868, F (44, 88) = 7.555, p < .001 for the overall fluency ratings. To ensure intra-rater reliability, each rater rated ten randomly-selected recordings a second time within a ten-day interval. Pearson’s r yielded values of .92, .94 and .89 for overall fluency, which were reasonable high reliability figures far above a r = .7 cut-off (George & Mallery, 2010; Pallant, 2001).

Internal consistency of constructs

Cronbach’s alpha was conducted to calculate the internal consistency of the constructs as measured in the questionnaire. It found high coefficients of .873 for anxiety level, .794 for self-efficacy in vocabulary knowledge, 813 for self-efficacy in grammar knowledge, and .877 for self-efficacy in L2 speaking based on an α = .7 cut-off (George & Mallery, 2010; Pallant, 2001).

Descriptives and parametric assumptions

Prior to conducting the data analyses, the present study checked and confirmed that the target variables met four basic parametric assumptions.

RQ 1

Pearson’s r was used to investigate the relationship between breadth of productive vocabulary knowledge and spoken fluency. The results indicated a strong, positive correlation between vocabulary knowledge and overall fluency (r = .529, p < .001) even with the small sample. The coefficient of determination (R2) indicated that vocabulary knowledge shared 27.98% of the variance in overall fluency.

The regression model resulted in significant prediction of the participants’ spoken fluency, F(1,43) = 16.709, p < .001. The simple correlation between vocabulary knowledge and fluency had a value of r = .529. An R2 = .280 showed that vocabulary knowledge accounted for 28.00% of the variation in spoken fluency (p < .001).

RQ 2

Pearson’s r was conducted, and the results showed no correlation between grammar knowledge and overall fluency (p > .05).

RQ 3

The relationships between students’ actual and perceived vocabulary knowledge, and between their actual and perceived grammar knowledge, were investigated separately. Pearson’s r showed a very weak and non-significant correlation (r = .023, p > .05) between perceived and actual vocabulary knowledge. Similarly, no significant correlation was found between perceived and actual grammar knowledge (r = .241, p > .05).

RQ 4

Pearson’s r showed a strong and positive correlation between vocabulary self-efficacy and self-efficacy in speaking using (r = .533, p < .001), with perceived vocabulary knowledge explaining 28.40% of the variance in their self-efficacy in L2 speaking. A strong and positive correlation between self-efficacy in grammar and that in speaking was also found (r = .559, p < .001). The participants’ perceived grammar knowledge explained 31.20% of the variability in their self-efficacy in English speaking.

A simple linear regression showed a strong and significant predictive power of the participants’ perceived vocabulary knowledge for self-efficacy in speaking, F(1,43) = 17.093, p < .001., indicating that perception of vocabulary knowledge made a significant (p < .001) contribution to predicting speaking self-efficacy. Similarly, the regression model suggested strong and significant predictive power of perceived grammar knowledge for self-efficacy in speaking, F(1,43) = 19.522, p < .001.

A hierarchical multiple regression analysis showed that self-efficacy in grammar (p < .01) and in vocabulary (p < .05) both made individual and significant contributions to predicting the outcome; 41.40% of the variance in self-efficacy in speaking was explained by the two predictors combined. The F-ratio had a significant value of 14.807, p < .05, suggesting that the model had significantly improved the ability to predict the outcome variable.

RQ 5

In a standard multiple regression, all variables were entered into the model in one block as predictors of overall fluency. The model including the five predictors explained 46.10% of the variance in spoken fluency and reached statistical significance (p < .001). The largest standardised beta coefficient was β = .527 for vocabulary score, p < .001, meaning that the measured breadth of productive vocabulary made the strongest unique contribution to explaining spoken fluency when the variance explained by all other variables in the model was controlled for. Self-efficacy showed the second next largest beta value, β = .312, p < .05. Neither grammar self-efficacy, β = .146, p > .05, nor actual grammar knowledge, β = -.023, p > .05, made a statistically significant contribution to the prediction of overall fluency scores.

In a hierarchical multiple regression, anxiety was entered at the first step, with vocabulary scores and grammar scores entered at the second step. Vocabulary self-efficacy and grammar self-efficacy were entered together as the third block. The model as a whole was found significant and predicted 46.1% of the variance in spoken fluency, F(5, 39) = 6.658, p < .001. The value of R2 change from models 2 to 3 suggested that linguistic self-efficacy explained an additional 14.2% of the variance in spoken fluency even when the effects of anxiety and actual linguistic knowledge were controlled for (p < .05). In particular, it was the participants’ self-efficacy in vocabulary that made a unique difference (β = .312, p < .05) in spoken fluency.

RQ 6

In the first hierarchical multiple regression, anxiety was entered at the first step; vocabulary and grammar scores were entered at the second step; ratings of vocabulary self-efficacy and grammar self-efficacy were entered together as the third block; ratings of self-efficacy in L2 speaking were entered at the final step. The final model reached statistical significance, explaining 46.20% of the variance in spoken fluency (F(6, 38) = 5.428, p < .001). When investigating the effect of self-efficacy in L2 speaking with all other variables controlled for, the predictor made no individual contribution to fluency (β = .045, p > .05). The change from model 3 to 4 (R2 change = .001, F change = .069, p < .05) suggested that self-efficacy in L2 speaking on its own did not provide additional explanation of the variance in spoken fluency.

In the second regression, anxiety was entered at the first step, followed by vocabulary and grammar scores entered as the second block. Next, ratings of vocabulary self-efficacy, grammar self-efficacy and L2 speaking self-efficacy were entered together as the final block. This regression demonstrated that when the effects of anxiety and actual linguistic knowledge were controlled for, self-efficacy predictor variables explained an additional 14.3% of the variance in spoken fluency (p < .05). Specifically, it was self-efficacy in vocabulary that had made a statistically significant contribution to the regression model (β = .297, p = .05).

Qualitative data: interview and short-answer questions

From both ends of the fluency score scale, a fluent group composing the 10 most fluent speakers and the other made up of the 10 least fluent speakers were identified. These students’ test performances, questionnaire and interview responses were analysed in depth. The analysis procedures adopted were recursive, mainly including data entry, coding and developing categories (Johnson & Christensen, 2013) on learner strategies. Participant comments were coded based on recurrent themes; a category was created provided that over three comments shared a similar focus. The qualitative findings are presented in the following discussion, serving as explanatory resources from which theoretical and pedagogical inferences can be drawn.


Linguistic knowledge and spoken fluency

The present study disagree with general assumptions claiming that disfluency in spoken L2 production is generated in equal parts from deficiencies in, or the applications of, vocabulary and grammar knowledge. Instead, they echo studies which suggest that lexical knowledge plays a role of far greater potency than grammar knowledge in L2 speaking (e.g., Lennon, 2000; Wood, 2004).

Patterns of efficient planning are shown in the qualitative responses. When asked to describe the one-minute planning process, the majority of participants spoke of forming thoughts at varying degrees of precision (conceptualisation) and conducting mental searches for correspondent L2 vocabulary (formulation). Interestingly, most speakers who received the highest fluency scores also received the highest vocabulary scores. They reported structuring the monologue by key lexical items as well as formulaic speech units or frames. In contrast, lowest performers revealed effortful attempts to search for words and to formulate sentences which complied with morphosyntactic rules during the silent planning period. These findings echo those of Hilton’s (2008), Lennon’s (2000) and Wood’s (2004) studies, indicating that L2 spoken fluency is associated more with efficient retrieval and effective use of lexical knowledge than with those of grammar knowledge.

Linguistic knowledge and self-efficacy beliefs in the cultural context

There were no significant relationships between self-efficacy in vocabulary and productive vocabulary knowledge, nor between self-efficacy in grammar and grammar test performance. These mismatches between actual and perceived linguistic knowledge confirmed Rublik’s (2006) concerns about students’ weak reflective abilities in a similar cultural context, where lack of self-awareness may obstruct progression in productive L2 competence.

Perceptions of vocabulary and grammar knowledge both made a significant contribution to predicting self-efficacy in speaking. Further qualitative analyses of self-efficacy beliefs as measured in the questionnaire revealed that students with higher levels of self-efficacy have had experiences in either some form of in-class monologic L2 presentations, on varying topics and of different lengths, or L2 conversation with peers in academic settings (see also Cubillos & Ilvento, 2012). These students attribute the increase of vocabulary and grammar self-efficacy in speaking to their prior performance. Interestingly, none of them reported negative impacts from these types of activities on self-efficacy, irrespective of their perceived success or failure in tasks (see also De Saint Léger, 2009).

Most participants who had higher linguistic self-efficacy ascribed their levels of achievement in speaking to factors over which they had control, such as effort in preparation and use of strategies. They confirm Bandura’s (1995) theoretical suggestion that it is an individual’s causal attributions of success or failure, rather than achievement itself, that determine self-efficacy. They also indicated that good language learners (i.e., fluent speakers) are characterised by their use of metacognitive strategies (Macaro, 2001) and their evaluative skills regarding the effectiveness of learning activities (O’Malley & Chamot, 1990).

Students with higher vocabulary self-efficacy reported influences of verbal persuasion from credible commentators on their L2 speaking, particularly in vocabulary size and use. Students with higher self-efficacy in L2 speaking valued recognition of progress from trustworthy judges (e.g., teachers and native speakers) which induced positive effects on individual confidence and reduced performance anxiety.

All of the above participants, whose self-rated efficacy scores were high in at least one of the domains, emphasised the influence of prior successes or failures on their judgments (see also De Saint Léger, 2009). This confirmed the positive influence of mastery experiences on the development of L2 linguistic self-efficacy. Self-efficacy beliefs in vocabulary and in grammar are built incrementally upon achievements in specific tasks; students’ interpretations therewith have more power to shape self-efficacy than actual knowledge or performance (see also Graham, 2006). In the context of this study, students’ linguistic self-efficacy could be magnified in an encouraging learning environment where mastery experiences are cultivated and constructive feedback is provided even without immersion in L2.

Perception and performance: self-efficacy and spoken fluency

The predictive power of self-efficacy beliefs in L2 vocabulary, grammar and speaking on L2 spoken fluency is manifest in the findings. Specifically, beliefs in vocabulary knowledge are the one type of self-efficacy that actually makes a statistically significant contribution to L2 spoken fluency. Self-efficacy in L2 grammar or speaking on its own does not make an individual contribution to fluency when the effects of other variables are controlled for.

These findings confirm Bandura’s (1986, 1995) theory that self-efficacy is not less powerful than knowledge and skill with respect to the influence on academic performance. The theory’s core idea remains valid in the context of L2 speaking given that self-efficacy is an important predictor of L2 spoken fluency in the present study. As self-efficacy and spoken fluency develop simultaneously, speech generation becomes increasingly automatic. This helps to free speakers’ mental capacity, allowing for engagement in higher levels of processing, and in turn creates mastery experiences in which speakers identify an increased command of L2 speech production.

Fluent participants with higher vocabulary self-efficacy shared a few common features. First, they actively evaluated their performances and past experiences for the purposes of improving their use of L2 vocabulary in speaking. Second, their appraisals were based on self-improvement rather than comparisons with others, suggesting the power of mastery experiences over competitions in building self-efficacy (Bandura, 1997). Moreover, these participants’ spoken fluency was characterised by circumlocution and ‘a flow of confidence’, which elicited their high engagement in performing the task. These students focused on conveying messages in speech and usually reached the highest levels of speech comprehensibility. Thus, the present study promotes delayed instead of immediate feedback which interrupts learner talk (see Harmer, 2007; Hall, 2011).

Contribution to theory

Findings from this study suggested the powerful influence of vocabulary knowledge over grammar knowledge when it comes to Taiwanese university students’ L2-English spoken fluency. They also showed that self-efficacy beliefs were both constructors and products of past experiences (Bandura, 1986) and that they were important mediators between students’ L2 competence and spoken fluency. As vocabulary knowledge and vocabulary self-efficacy develop simultaneously, speech generation becomes increasingly automatic (i.e., more fluent).

This study not only supports Bandura’s (1986, 1995) theory that self-efficacy beliefs are interconnected but also brings in perspectives on the influences of the four sources of self-efficacy on L2 spoken fluency, recognising mastery experiences and credible verbal persuasions as important self-efficacy builders. Students who conducted self-evaluation on past and current performances based on these two sources experienced incremental changes in their perceptions of linguistic self-abilities. In addition, students who actively react to self-appraisals of past successes, failures and comments expand and improve L2 vocabulary in speech. Affective factors influenced not only self-efficacy by altering students’ interpretation of prior experiences, but also the processes of speech production. In contrary, the impact of modelling seems subtle on self-efficacy, arguably attributable to the lack of opportunities for observational learning as well as students’ tendencies to compare themselves competitively with the models, which according to Bandura (1995) contributes little to self-efficacy stability.

Pedagogical implications

First, it is suggested that students develop strong and stable self-efficacy in learning environments where they are encouraged to work on self-improvement, rather than in those where they focus on outdoing peers (see Graham, 2006). The current L2 assessment culture in Taiwan is predominantly characterised by ranking systems, in which student academic performance is viewed in contrast with others by way of classroom, institutional and provincial rankings; thus, this study discourages the existing competition-oriented culture given its lack of and even detrimental effects on promoting students’ performance. Alternatively, it encourages the development of self-assessment tools at both cognitive and affective levels (see also De Saint Léger, 2009).

In terms of classroom practice, teachers should endeavour to observe and promote students’ self-efficacy in performance, particularly in productive vocabulary use. This is of particular relevance to the context of the present study considering the culture of humility, or self-deprecation, among Taiwanese students. When necessary, teachers should help students to value struggles and to avoid pessimism attributions of failure to incompetence, which could directly affect learning motivation, behaviour and performance (see Pajares, 1996; Schunk, 1991). Additionally, teachers should be careful about what they say and do while students are practicing. Pupils in Taiwanese L2 classrooms are generally quiet and inexpressive due to peer anxiety; in this scenario, teachers should strive to create a whole classroom environment where students are also supportive of each other when carrying out oral training.

As for activities, this study encourages tasks which require preparation to increase students’ productive L2 vocabulary and diverse L2 speaking activities such as impromptu speech and conversation, which allow students to practice dealing with apprehension in the classroom. In this case, teachers should incorporate appropriate and progressive levels of exercises in oral training. In addition, teachers should ultimately educate students to use their own initiative in building self-efficacy (Schunk, 1991; Schunk & Hanson, 1989), to persevere and to face challenges with problem-solving approaches since, as Bandura (1986) argued, self-efficacy and competence work in tandem.

Limitations and suggestions for future research

The sample size and the number of raters were both relatively small and specific. Future research is advised to recruit a larger number of participants and raters to improve the generalisability of the findings. The questionnaire was a self-report instrument, whose validity and reliability depended on its respondents. The speaking test could be unreliable with factors such as the interviewer’s varying tones and facial expressions, time of the day, and so forth. A video-recorded interview could be more objective. In addition, it could be argued that a student’s performance in one task was not sufficient to judge his or her spoken fluency. Multiple trials or incorporate additional measures could provide reliable assessments of spoken fluency. To provide a bigger and more detailed picture of the relationships being studied, longitudinal studies are recommended to explore how interactions between variables sustain, change or develop in different contexts in time.

Finally, future research should consider different ways in approaching the construct, collecting and analysing data, and interpreting the results. It should also take into account the effects of other variables (possibly e.g., speaking strategies, FS knowledge, personality) left out in the present study and aim to identify additional sources of self-efficacy, which can inform practitioners in their pedagogy.


The present study found that Taiwanese students’ self-efficacy in L2 vocabulary knowledge makes an independent contribution to the prediction of L2 spoken fluency, supporting Bandura’s (1986, 1995) theory about the role of perceptions (or self-efficacy) in mediating the effects of competence on academic performance. This study suggest that even in the absence of immersion experiences in L2 contexts, Taiwanese students’ L2 spoken fluency can be strongly improved when teachers and students strive to expand L2 vocabulary and cultivate self-efficacy. Not only did Taiwanese students’ self-efficacy beliefs significantly influence L2 spoken fluency and subsequent interpretations of the four sources of self-efficacy (i.e., mastery experiences, verbal persuasion, modelling experience, affective states), but they were considerably inaccurate in estimating actual competence (i.e., lack of self-awareness). As such, this study encourages researchers to identify how self-efficacy can be feasibly operationalised in classrooms (e.g., how teachers should go about ‘observing’ and ‘encouraging’).

Further, the study promotes the development of self-assessment tools including those at cognitive and affective levels. These tools allow each individuals to record and reflect their unique learning progress by drawing inspiration from past experiences and comments. The qualitative investigations in this study showed that fluent students with higher self-efficacy tended to conduct constant self-evaluations of past performances, striving to identify weaknesses that could be improved while reproducing successful experiences in further advanced tasks. These behaviours of more fluent students should be encouraged in everyone. This is particularly important in a prevalent culture of over-humility and self-deprecation in Taiwanese educational settings. Thus, teachers hold the responsibilities to encourage and help students gain experiences in speaking, make use of constructive feedback, conduct appropriate self-appraisals, and take lessons from both successes and failures.

In conclusion, this study suggests that the classroom focus be shifted from competitions and rankings to individual self-improvement. It encourages teachers’ efforts in developing pedagogical approaches that consider self-efficacy as well as promote the effective implementation of solid lexical syllabi. In addition, they suggest that L2 spoken fluency is not only affected by the size of L2 lexicon but also students’ self-efficacy in productive use of the lexicon in speaking. Teachers and students’ collective efforts in striving to identify causes of failure and reproducing successful experiences can increase students’ competence and self-efficacy simultaneously, resulting in incremental and solid progress in L2 spoken performance.


Allen, D. (1992). Oxford placement test. Oxford, England: Oxford University Press.

British Association for Applied Linguistics (BAAL) (2000). Recommendations for good practice in Applied Linguistics student projects. Available from: www.baal.org.uk/dox/goodpractice_full.pdf [Last accessed 29 May 2017].

Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice-Hall.

Bandura, A. (1995). Self-efficacy in changing societies. Cambridge, England: Cambridge University Press.

Bandura, A. (1997). Self-efficacy: The exercise of control. New York: W. H. Freeman.

Bavelas, J., Coates, L. & Johnson, T. (2000). Listeners as co-narrators. Journal of Personality and Social Psychology, 79(6), 941–952.

Chambers, F. (1997). What do we mean by fluency? System, 25(4), 535–544.

Chiswick, B. R. & Miller, P. W. (1998). English language fluency among immigrants in the United States. Research in Labor Economics, 17(9), 151–200.

Clément, R. (1980). Ethnicity, contact and communicative competence in a second language. In H. Giles, W. P. Robinson & P. Smith (Eds.), Language: Social psychological perspectives (pp. 147–154). Oxford, England: Pergamon.

Cotterall, S. (1999). Key variables in language learning: What do learners believe about them? System, 27(4), 493–513.

Cubillos, J. H. & Ilvento, T. (2012). The impact of study abroad on students’ self-efficacy perceptions. Foreign Language Annals, 45(4), 494–511.

Daller, H. & Xue, H. (2007). Lexical richness and the oral proficiency of Chinese EFL students. In H. Daller, J. Milton & J. Treffers-Daller (Eds.), Modelling and assessing vocabulary knowledge (pp. 150–164). Cambridge, England: Cambridge University Press.

De Bot, K. (1992). A bilingual production model: Levelt’s “speaking” model adapted. Applied Linguistics, 13(1), 1–24.

De Saint Léger, D. (2009). Self‐assessment of speaking skills and participation in a foreign language class. Foreign Language Annals, 42(1), 158–178.

DeKeyser, R. M. (2014). Methodological considerations about research on language development during study abroad. In C. Pérez-Vidal (Ed.), Language acquisition in study abroad and formal instruction contexts (pp. 313–326). Amsterdam: John Benjamins.

Derwing, T. M., Rossiter, M. J., Munro, M. J. & Thomson, R. I. (2004). Second language fuency: Judgements on different tasks. Language Learning, 54(4), 655–679.

Færch, C. & Kasper, G. (1983). Strategies in interlanguage communication. London: Longman.

Fillmore, C. J. (1979). On fluency. In C. J. Fillmore, D. Kempler & W. S. Wang (Eds.), Individual differences in language ability and language behavior (pp. 85–101). New York: Academic Press.

Fillmore, C. J. (2000). On fluency. In H. Riggenbach (Ed.), Perspectives on fluency (pp. 43–60). Ann Arbor, MI: University of Michigan Press.

Fitzpatrick, T. (2007). Productive vocabulary tests and the search for concurrent validity. In H. Daller, J. Milton & J. Treffers-Daller (Eds.), Modelling and assessing vocabulary knowledge (pp. 116–132). Cambridge, England: Cambridge University Press.

Fitzpatrick, T. & Clenton, J. (2010). The challenge of validation: Assessing the performance of a test of productive vocabulary. Language Testing, 27(4), 537–554.

Foster, P. & Skehan, P. (1996). The influence of planning and task type on second language performance. Studies in Second Language Acquisition, 18(3), 299–323.

Freed, B. F. (1995). What makes us think that students who study abroad become fluent. In B. F. Freed (Ed.), Second language acquisition in a study abroad context (Vol. 9, pp. 123–148). Philadelphia, PA: John Benjamins.

Fushino, K. (2010). Causal relationships between communication confidence, beliefs about group work, and willingness to communicate in foreign language group work. TESOL Quarterly, 44(4), 700–724.

Gardner, R. C. & Lambert, W. E. (1972). Attitudes and motivation in second-language learning. Rowley, MA: Newbury House Publishers.

George, D. & Mallery, P. (2010). SPSS for Windows step by step: A simple guide and reference: 17.0 update (10th ed.). Boston, MA: Allyn and Bacon.

Giles, H., Coupland, J. & Coupland, N. (2010). Contexts of accommodation: Developments in applied sociolinguistics. Cambridge, England: Cambridge University Press.

Goldman-Eisler, F. (1972). Pauses, clauses, sentences. Language and Speech, 15(2), 103–113.

Graham, S. (2006). A study of students’ metacognitive beliefs about foreign language study and their impact on learning. Foreign Language Annals, 39(2), 296–309.

Graham, S. & Sandmel, K. (2011). The process writing approach: A meta-analysis. The Journal of Educational Research, 104(6), 396–407.

Hall, G. (2011). Exploring English language teaching: Language in action. New York: Routledge.

Hieke, A. E. (1985). A componential approach to oral fluency evaluation. Modern Language Journal, 69(2), 135–142.

Hilton, H. (2008). The link between vocabulary knowledge and spoken L2 fluency. Language Learning, 36(2), 153–166.

Housen, A. & Pierrard, M. (2005). Investigations in instructed second language acquisition. Berlin: Mouton de Gruyter.

Hsu, J. & Chiu, C. Y. (2008). Lexical collocations and their relation to speaking proficiency of college EFL learners in Taiwan. Asian EFL Journal, 10(1), 181–204.

Huang, H. Y. (2012). The effectiveness of formulaic sequence instruction on Taiwanese EFL learners’ oral fluency (Doctoral Dissertation). Alliant International University, San Diego.

Johnson, R. B. & Christensen, L. B. (2013). Educational research: Quantitative, qualitative, and mixed approaches (5th ed.). Thousand Oaks, CA: Sage.

Klassen, R. (2002). Writing in early adolescence: A review of the role of self-efficacy beliefs. Educational Psychology Review, 14(2), 173–203.

Koizumi, R. & In’nami, Y. (2013). Vocabulary knowledge and speaking proficiency among second language learners from novice to intermediate levels. Journal of Language Teaching & Research, 4(5), 121–136.

Koponen, M. & Riggenbach, H. (2000). Overview: Varying perspectives on fluency. In H. Riggenbach (Ed.), Perspectives on fluency (pp. 5–24). Ann Arbor, MI: University of Michigan Press.

Kormos, J. (2006). Speech production and second language acquisition. Mahwah, NJ: Lawrence Erlbaum Associates.

Kormos, J. & Dénes, M. (2004). Exploring measures and perceptions of fluency in the speech of second language learners. System, 32(2), 145–164.

Krashen, S. D. (1982). Principles and practice in second language acquisition. Oxford, England: Pergamon.

Lai, Y. C. (2009). Language learning strategy use and English proficiency of university freshmen in Taiwan. TESOL Quarterly, 43(2), 255–280.

Lane, J., Lane, A. M. & Kyprianou, A. (2004). Self-efficacy, self-esteem and their impact on academic performance. Social Behavior and Personality, 32(3), 247–256.

Laufer, B. & Nation, P. (1995). Vocabulary size and use: Lexical density in FL written production. Applied Linguistics, 16(3), 307–322.

Lemmouh, Z. (2010). The relationship among vocabulary knowledge, academic achievement and the lexical richness in writing in Swedish university students of English (Doctoral Dissertation). Stockholm University, Sweden.

Lennon, P. (1990). Investigating fluency in EFL: A quantitative approach. Language Learning, 40(3), 387–417.

Lennon, P. (2000). The lexical element in spoken second language fluency. In H. Riggenbach (Ed.), Perspectives on fluency (pp. 25–42). Ann Arbor, MI: University of Michigan Press.

Liu, G. Z. (2005). The trend and challenge for teaching EFL at Taiwanese universities. RELC Journal, 36(2), 211–221.

Macaro, E. (2001). Learning strategies in foreign and second language classrooms. London: Continuum.

McCarthy, M. (2009). Rethinking spoken fluency. Estudios de Lingüística Inglesa Aplicada, 9, 11–29.

Meara, P. & Fitzpatrick, T. (2000). Lex30: An improved method of assessing productive vocabulary in an L2. System, 28(1), 19–30.

Mehnert, U. (1998). The effects of different lengths of time for planning on second language performance. Studies in Second Language Acquisition, 20(1), 83–108.

Mochizuki, N. & Ortega, L. (2008). Balancing communication and grammar in beginning-level foreign language classrooms: A study of guided planning and relativization. Language Teaching Research, 12(1), 11–37.

Multon, K. D., Brown, S. D. & Lent, R. W. (1991). Relation of self-efficacy beliefs to academic outcomes: A meta-analytic investigation. Journal of Counseling Psychology, 38(1), 30–38.

O’Malley, J. M. & Chamot, A. U. (1990). Learning strategies in second language acquisition. Cambridge, England: Cambridge University Press.

Ortega, L. (1999). Planning and focus on form in L2 oral performance. Studies in Second Language Acquisition, 21(1), 109–148.

Pajares, F. (1996). Self-efficacy beliefs in academic settings. Review of Educational Research, 66(4), 543–578.

Pajares, F. & Johnson, M. J. (1996). Self‐efficacy beliefs and the writing performance of entering high school students. Psychology in the Schools, 33(2), 163–175.

Pajares, F. & Miller, M. D. (1994). Role of self-efficacy and self-concept beliefs in mathematical problem solving: A path analysis. Journal of Educational Psychology, 86(2), 193–208.

Pajares, F. & Miller, M. D. (1997). Mathematics self-efficacy and mathematical problem solving: Implications of using different forms of assessment. Journal of Experimental Education, 65(3), 213–228.

Pallant, J. (2001). SPSS survival manual: A step-by-step guide to data analysis using SPSS for Windows (Version 10). Crows Nest, New South West: Allen & Unwin.

Rajab, A., Far, H. R. & Etemadzadeh, A. (2012). The relationship between L2 motivational self-system and L2 learning among TESL students in Iran. Procedia, 66, 419–424.

Riggenbach, H. (1991). Toward an understanding of fluency: A microanalysis of nonnative speaker conversations. Discourse Processes, 14(4), 423–441.

Riggenbach, H. (2000). Perspectives on fluency. Ann Arbor, MI: University of Michigan Press.

Rublik, N. (2006). An investigation into the role of culture and the development of oral fluency: A case study involving Chinese learners of English (Unpublished Doctoral Dissertation). University of Western Ontario, Canada.

Schmidt, R. (1992). Psychological mechanisms underlying second language fluency. Studies in Second Language Acquisition, 14(4), 357–385.

Schunk, D. H. (1991). Self-efficacy and academic motivation. Educational Psychologist, 26(3/4), 207–231.

Schunk, D. H. & Hanson, A. R. (1989). Influence of peer-model attributes on children’s beliefs and learning. Journal of Educational Psychology, 81(3), 431–434.

Skehan, P. (1996). A framework for the implementation of task based instruction. Applied Linguistics, 17, 38–62.

Tashakkori, A. & Teddlie, C. (2003). Handbook of mixed methods in social & behavioral research. Thousand Oaks, CA: Sage.

Teddlie, C. & Tashakkori, A. (2009). Foundations of mixed methods research: Integrating quantitative and qualitative approaches in the social and behavioral sciences. Thousand Oaks, CA: Sage.

Teng, K. H. (2005). Perceptions of Taiwanese students to English learning as functions of self-efficacy, motivation, learning activities and self-directed learning (Unpublished Doctoral Dissertation). University of Idaho, ID.

Towell, R., Hawkins, R. & Bazergui, N. (1996). The development of fluency in advanced learners of French. Applied Linguistics, 17(1), 84–115.

Tremblay, P. F. & Gardner, R. C. (1995). Expanding the motivation construct in language learning. Modern Language Journal, 79(4), 505–518.

Vandergrift, L. (2005). Relationships among motivation orientations, metacognitive awareness and proficiency in L2 listening. Applied Linguistics, 26(1), 70–89.

Vermeer, A. (2004). The relation between lexical richness and vocabulary size in Dutch L1 and L2 children. In P. Bogaards & B. Laufer (Eds.), Vocabulary in a second language: Selection, acquisition, and testing (pp. 187–204). Amsterdam: John Benjamins.

Viberg, Å. (1993). Crosslinguistic perspectives on lexical organization and lexical progression. In K. Hyltenstam & Å. Viberg (Eds.), Progression & regression in language (pp. 340–385). New York: Cambridge University Press.

Walters, J. (2012). Aspects of validity of a test of productive vocabulary: Lex30. Language Assessment Quarterly, 9(2), 172–185.

Wolf, J. P. (2008). The effects of backchannels on fluency in L2 oral task production. System, 36(2), 279–294.

Wood, D. (2004). An empirical investigation into the facilitating role of automatized lexical phrases in second language fluency development. Journal of Language and Learning, 2(1), 27–50.

Wu, J. R. W. (2012). GEPT and English language teaching and testing in Taiwan. Language Assessment Quarterly, 9(1), 11–25.

Yang, N. D. (1999). The relationship between EFL learners’ beliefs and learning strategy use. System, 27(4), 515–535.

Yashima, T. (2002). Willingness to communicate in a second language: The Japanese EFL context. The Modern Language Journal, 86(1), 54–66.

Yeh, C. J. & Inose, M. (2003). International students’ reported English fluency, social support satisfaction, and social connectedness as predictors of acculturative stress. Counselling Psychology Quarterly, 16(1), 15–28.

Yu, M. C. (2008). Teaching and learning sociolinguistic skills in university EFL classes in Taiwan. TESOL Quarterly, 42(1), 31–53.

Zhong, Q. (2013). Understanding Chinese learners’ willingness to communicate in a New Zealand ESL classroom: A multiple case study drawing on the theory of planned behavior. System, 41(3), 740–751.