The effects of oral corrective feedback and language anxiety on pronunciation development

ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 Estudios de lingüística inglesa aplicada THE EFFECTS OF ORAL CORRECTIVE FEEDBACK AND LANGUAGE ANXIETY ON PRONUNCIATION DEVELOPMENT LOS EFECTOS DE LA RETROALIMENTACIÓN CORRECTIVA ORAL Y LA ANSIEDAD SOBRE EL DESARROLLO DE LA PRONUNCIACIÓN María Luquin Universidad del País Vasco/Euskal Herriko Unibertsitatea, Spain [email protected] Hanne Roothooft Universidad Pública de Navarra, Spain [email protected] DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 While a substantial body of research has shown that oral corrective feedback (CF) is effective for the acquisition of foreign language grammar, very few studies have focused on the impact of CF on pronunciation development. Moreover, there is a dearth of CF research which takes individual differences such as foreign language anxiety into account. The present study therefore investigated the effects of recasts and metalinguistic feedback on the pronunciation of the -ed ending, comparing 30 low- and high-anxiety learners with an A2+ level of English at a Spanish secondary school. A pre-test/post-test design involving a reading-aloud test and a storytelling treatment was applied to a recast (n=10), metalinguistic feedback (n=10) and control group (n=10). Each of the three groups was ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 The effects of oral corrective feedback and language anxiety… 42 further divided into a high-anxiety (n=5) and low-anxiety group (n=5). Significant differences were found between the recast and control group, confirming previous research showing that recasts are beneficial for pronunciation development. However, despite certain indications that lowanxiety learners benefited more from CF, and high-anxiety learners seemed to benefit especially from recasts, no significant differences between the anxiety-groups could be identified in this study. Key words: foreign language anxiety; corrective feedback; recasts; metalinguistic feedback; modified output; pronunciation Si bien una gran cantidad de estudios han demostrado que la retroalimentación correctiva oral (RC) es efectiva para la adquisición de la gramática en un idioma extranjero, muy pocos se han centrado en el impacto de la RC en el desarrollo de la pronunciación. Además, escasea la investigación de RC que tenga en cuenta las diferencias individuales, como la ansiedad en el aprendizaje de una lengua extranjera. Por lo tanto, el presente estudio investigó los efectos de las reformulaciones y la retroalimentación metalingüística sobre la pronunciación de la terminación de pasado -ed, comparando 30 estudiantes de ansiedad alta y baja con un nivel A2 + de inglés en una escuela secundaria de España. Un diseño pretest/post-test consistente en una lectura en voz alta y una narración fue llevado a cabo con un grupo de reformulación (n=10), un grupo de retroalimentación metalingüística (n=10) y un grupo de control (n=10). Cada uno de los tres grupos fue subdividido en un grupo de ansiedad alta (n = 5) y otro de ansiedad baja (n = 5). Se encontraron diferencias significativas entre el grupo de reformulación y el grupo de control, lo cual confirma investigación previa que demuestra que las reformulaciones son beneficiosas para el desarrollo de la pronunciación. Sin embargo, a pesar de ciertas indicaciones de que los estudiantes con ansiedad baja se beneficiaron más de la RC, y los estudiantes con ansiedad elevada parecieron beneficiarse especialmente de las reformulaciones, no se pudieron identificar diferencias significativas entre los grupos de ansiedad en este estudio. Palabras clave: ansiedad en el aprendizaje de una lengua extranjera; retroalimentación correctiva; reformulaciones; retroalimentación metalingüística; producción modificada; pronunciación. 43 María Luquin & Hanne Roothooft ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 1. Introduction While there is a considerable body of research which shows positive effects of oral corrective feedback (CF) on the acquisition of L2 grammar (e.g., Ammar & Spada, 2006; Doughty & Varela, 1998; Ellis, Loewen & Erlam, 2006; Lyster & Saito, 2010; McDonough 2007), few studies to date have investigated its impact on vocabulary learning (e.g., Dilans, 2010) and even fewer studies have focused on the impact of CF on the acquisition of pronunciation features (e.g., Saito & Lyster, 2012a,b). Pronunciation is nonetheless an important aspect of L2 oral skills, since it can affect overall intelligibility and the ability to communicate. Moreover, although several researchers have tried to determine if there is a difference between different CF-types, so far the question whether to use recasts or prompts (such as metalinguistic feedback or clarification requests) remains unresolved (Goo & Mackey, 2013; Lyster, Saito & Sato, 2013). Research on CF-effectiveness is further complicated by the interference of both external (e.g., CF-type, linguistic target, interactional setting) and internal variables (e.g., proficiency level, working memory, anxiety). The internal variables in particular, also referred to as individual differences, have received the least attention from CF-researchers (Ellis, 2010; Lyster, Saito & Sato, 2013). An individual difference which is particularly interesting to investigate is foreign language anxiety, since Sheen (2008) states anxiety is an important variable that may interfere not only in CF efficacy, but also in learners’ responses (i.e., uptake). Only few CF studies have ventured to relate anxiety to error correction (e.g., Sheen, 2008), and to the best of our knowledge, no studies have examined the impact of anxiety and CF on pronunciation development. To fill this gap, the current study analyzes the possible influence of learners’ anxiety on two types of error correction (recasts and metalinguistic feedback) directed at a specific pronunciation feature. 2. Literature Review 2.1. The Effectiveness of Oral CF Since Lyster and Ranta published their seminal paper on oral CF in 1997, researchers have carried out a great number of studies investigating the effects of oral CF on second language acquisition. Most of these studies focused on specific grammatical targets, such as question formation (e.g., ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 The effects of oral corrective feedback and language anxiety… 44 Mackey & Philp, 1998; McDonough, 2005), articles (Muranoi, 2000; Sheen, 2007) or the past tense (Doughty & Varela, 1998; Han, 2002; Yang & Lyster, 2010), to name a few. Several meta-analyses of these studies have been published, which have all concluded that oral CF has positive and durable effects on acquisition (Mackey & Goo, 2007; Russell & Spada, 2006; Li, 2010; Lyster & Saito, 2010). A few studies have looked at how oral CF impacts on vocabulary acquisition, and these have also concluded in favour of CF (e.g., Dilans, 2010). However, as mentioned in the introduction, its impact on L2 pronunciation has received very little attention. Before having a closer look at the few studies which have addressed this question, we need to discuss the different variables involved in CF research. While most researchers now agree oral CF is effective, there is still some debate about the most effective CF-type. As Lyster, Saito and Sato (2013) explain, oral CF-types can be classified according to two dimensions: explicit vs. implicit, on the one hand, and input-providing vs. output-pushing, on the other. Recasts, usually classified as implicit and input-providing, have been put forward as an effective technique by some researchers (Han, 2002; Ishida, 2004; Leeman, 2003; Mackey & Philp, 1998; Mcdonough & Mackey, 2006; Loewen & Nabei, 2007). However, many of these studies were carried out in laboratory settings and did not compare recasts to other types of CF. Comparisons between recasts and more explicit CF-types have been made, generally resulting in greater effects for the more explicit CF (e.g., Ellis, Loewen & Erlam, 2006; Sheen, 2007; Yilmaz, 2012). However, Li (2010) states that recasts seem to have greater effects on delayed post-tests, which means their impact may take longer to manifest itself. When comparing recasts to prompts, a group of output-pushing techniques such as elicitation or clarification requests, studies carried out in classroom-settings indicate a greater benefit for prompts (Lyster & Saito, 2010). The question of which type of CF is the most effective has not been fully resolved so far, and more and more researchers point out the need to focus on other variables which potentially interfere with CF effectiveness, such as proficiency, working memory or indeed, language anxiety. Moreover, recasts and prompts may affect different target structures in different ways (see for instance Ellis, 2007). Mackey, Gass and McDonough (2000) showed that learners do not tend to notice recasts on morphosyntactic errors, while recasts on phonological errors are more salient. As mentioned earlier, most research on CF has looked at grammatical targets. For pronunciation, it may well be that 45 María Luquin & Hanne Roothooft ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 recasts are a better choice, given the fact that they provide a correct model for the students. However, as we will see in section 2.2, hardly any studies have compared the effects of recasts and prompts on the acquisition of a phonological target, so this question remains to be investigated. 2.2. Oral CF and Pronunciation So far, there is a dearth of research looking at the effects of CF on students’ accurate pronunciation. A series of studies carried out by Lyster and his colleagues combined form-focused instruction with CF on the pronunciation of different consonant and vowel sounds. Saito and Lyster (2012a) and Saito and Lyster (2012b) only investigated the effects of recasts on pronunciation, while Gooch, Saito and Lyster (2016) is one of the rare studies comparing the effects of recasts and prompts on pronunciation development. Saito and Lyster (2012a) found that a combination of form-focused instruction and recasts had positive effects on the pronunciation of /ɹ/ by adult Japanese ESL learners, but recasts were not compared with any other types of CF. Saito and Lyster (2012b) used data from the same study involving adult Japanese learners, but focused on the pronunciation of vowel sound /ae/. As for /ɹ/, they concluded that recasts positively affected the pronunciation of /ae/, both in words the students had previously practiced and in a new word. Using the same method as Saito and Lyster (2012a,b), Gooch, Saito and Lyster (2016), compared the effects of recasts and prompts (a combination of clarification requests and elicitation) on Korean adult EFL learners’ pronunciation of /ɹ/. They concluded that both types of oral CF appeared to have positive effects, although recasts were found to be more helpful for controlled production and prompts benefited both controlled and freer production. To the best of our knowledge there is only one other study which has compared prompts and recasts with regard to pronunciation learning. Mohammadi (2014) focused on the influence of recasts and prompts on Iranian high school students’ performance in final ending –s/-es pronunciation. In this study, the pronunciation rules were first taught to students. They then practiced the target structure in more controlled tasks (fill-in-the blank with the correct sound and read-aloud tasks) and a communicative task in which they had to tell a picture story about a ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 The effects of oral corrective feedback and language anxiety… 46 student’s typical day. Mohammadi (2014) observed that recasts were more effective than prompts, but it is not clear how exactly prompts were operationalised in this study. The few studies that have been carried out on CF and pronunciation have thus combined explicit rule teaching with practice and feedback, and have focused on a limited number of phonemes. 2.3. Foreign Language Anxiety and Oral CF Apart from comparing the effects of different types of CF on different target structures, researchers have also stressed the need for taking individual learner factors into account when studying the effectiveness of CF. As argued by Sheen (2008), one of the individual factors that can have a significant impact on error correction is foreign language anxiety. Not only is anxiety often related to oral production (Awan et al., 2010; Hashemi & Abbasi, 2013; Horwitz et al., 1986; Ortega, 2009; Rassaei, 2015; Sheen, 2008) and pronunciation (Szyska, 2016; Vitanova & Miller, 2002), some scholars also claim oral CF may increase anxiety and should therefore not be used (Krashen, 1982; Truscott, 1999). Studies on teachers’ beliefs about oral CF moreover show that teachers tend to be concerned about possible negative effects of CF on students’ motivation and self-esteem (Lasagabaster & Sierra, 2005; Yoshida, 2010). It is therefore surprising that so few studies on oral CF have taken the variable of anxiety into account. DeKeyser’s (1993) research is pioneering in this respect. He conducted a study with high-school learners of French in an attempt to evaluate the impact of CF in connection with individual differences. He found that CF was the most beneficial for students with high previous knowledge, high aptitude, and low anxiety, but DeKeyser’s analysis did not separate recasts from other types of error correction, nor did he analyze any particular language pattern. Rassaei (2015) investigated the impact of anxiety on recasts and metalinguistic feedback aimed at the definite and indefinite English article. The results indicated that although low-anxiety learners improved with both types of CF, they profited more profoundly from metalinguistic corrections, while high-anxiety learners responded better to recasts. A possible explanation put forward by the author is that recasts do not seem to interfere with communication, and imply less cognitive effort. 47 María Luquin & Hanne Roothooft ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 Similar results were those evidenced by Sheen (2008), who only analyzed the effects of recasts on English articles and found that lowanxiety students from the experimental group performed significantly better than high-anxiety learners, and even outperformed low-anxiety learners from the control group, while no significant differences were found between the high-anxiety experimental group and the high-anxiety control group. Moreover, she found that the low-anxiety recast students were also able to produce a higher number of learner uptake moves, or modified output. 2.4. Modified Output Following Lyster and Ranta (1997), we defined uptake as a learner’s immediate response to a teacher’s CF. It can be a simple acknowledgement (“yeah”, “ok”, “yes”, “oh”); a repetition of the original mistake; a correction of the original mistake; or a partial correction. However, Sheen (2008) makes a distinction between “learner uptake” and “modified output”, arguing that the former can consist of any type of response, while the latter encompasses any attempt to repair the ill-formed utterance, be it repaired or not. As she puts it “learners might produce uptake but not necessarily modify their output, whereas even when they do produce modified output, they might not repair their original error” (p. 841). Even though it is important to note that uptake is not the same as acquisition (Mackey & Philp, 1998), some studies have found that it can be an important predictor of acquisition (e.g., Loewen, 2005; McDonough, 2005). Lyster (1998) discovered that pronunciation- and lexical-focused recasts triggered more instances of modified output than recasts directed at morphosyntactic errors. This finding is supported by Mackey et al. (2000), who noted that students perceive and respond to morphosyntactic recasts to a lesser extent than to recasts directed at lexical or phonological errors. 2.5. Research Questions Given that the relationship between CF and foreign language anxiety has received little attention and that to the best of our knowledge no study has been carried out so far on the acquisition of pronunciation features through CF taking foreign language anxiety into account, the aim of this paper is to ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 The effects of oral corrective feedback and language anxiety… 48 analyze the extent to which groups of high- and low-anxiety learners acquire a pronunciation feature through recasts and metalinguistic feedback, as well as to examine their uptake moves. More specifically, the following research questions underlie the present study: 1. What is the effect of recasts and metalinguistic feedback on the pronunciation of the -ed ending, and is there a difference between the two types of CF? 2. Do high-anxiety and low-anxiety students benefit from CF on their pronunciation of the -ed ending, and is there a difference between the two groups of students? 3. Do high-anxiety and low-anxiety students modify their output after oral CF directed at their pronunciation errors, and is there a difference between the two groups? 3. Methodology 3.1. Design The current study followed a quasi-experimental design which included a pre- and a post-test administered to three intact classes (n=64) that were each randomly assigned to two experimental groups and a control group. Afterwards, based on their responses to a foreign language anxiety questionnaire (see Appendix), the students in the three classes were divided into a group of high-anxiety (n=15) and low-anxiety learners (n=15). On a 4-point likert-scale, the mean score of all the questionnaires together was 2.00 with a standard deviation (SD) of 0.50. Following Sheen (2008), learners who scored over one SD above the mean were classified as “highanxiety learners”, while those who scored more than one SD below the mean were classified as “low-anxiety learners”. From each anxiety group, the five students with the highest and the lowest scores were selected, and those students whose score fell within one SD of the mean were excluded from the analysis because they can be said to belong to the middle range of anxiety, rather than clearly exhibiting high or low anxiety (Rassaei, 2015; Sheen, 2008). This means the final number of participants amounted to 30 students in the high- and low-anxiety groups. As a result, these six groups were formed: 49 María Luquin & Hanne Roothooft ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 1) Recasts (high-anxiety) (n=5) (class C) 2) Metalinguistic (high-anxiety) (n=5) (class A) 3) Control (high-anxiety) (n=5) (class B) 4) Recasts (low-anxiety) (n=5) (class C) 5) Metalinguistic (low-anxiety) (n=5) (class A) 6) Control (low-anxiety) (n=5) (class B) Figure 1. Study design. Figure 1 represents the design of the present study. Table 1 shows the mean language anxiety scores of the six groups. The anxiety questionnaire and the pretest were administered during the same session, four days before the treatment session. Three days after the treatment, the learners were given an immediatepost-test. Unfortunately, due to time constraints, a delayed post-test could not be conducted. ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 The effects of oral corrective feedback and language anxiety… 56 differences between the groups (F=2.9, p=0.04). However, as it was noted that the high-anxiety control group performed much worse than all other groups, another one-way-ANOVA was carried out without this group, only comparing the 5 others at the pre-test. This yielded no significant differences (F=1.83, p=0.16). It was therefore decided not to include the control group in the subsequent analysis of the impact of anxiety on CF-effectiveness. For the high- and low-anxiety groups, a repeated-measures ANOVA on the scores of the high- and low-anxiety metalinguistic feedback and recast groups indicated no significant difference for group, F=2.49, p=0.1, but a significant difference for time, F=24.9, p=0.0001, meaning that there was no evidence that anxiety has an influence on the student’s performance, but as mentioned in section 4.1, there were significant differences between the pre- and the post-tests. Pre-test Post-test Gain scores Mean SD Mean SD Mean SD Recasts (high-anxiety) 22.2 3.9 40.8 16.3 18.6 16.5 Metalinguistic (high-anxiety) 22.2 7.8 31.2 19.7 9 13.1 Control (high-anxiety) 13.4 10.4 13.6 9.2 0.2 4.7 Recasts (low-anxiety) 32.2 8.2 62.4 21.5 30.2 14.9 Metalinguistic (low-anxiety) 21.1 7.3 40 24.7 18.9 22.7 Control (low-anxiety) 30 12.8 31.2 16.6 1.2 4.8 Table 3. Pre- and post-test scores of the anxiety groups Although there were no statistical differences between the low- and high-anxiety groups in this study, certain trends can nonetheless be identified. The bottom right half of Table 3 shows the low-anxiety post-test results, which are considerably better when compared to their corresponding highanxiety results. For example, the low anxiety recast group had a mean posttest score of 62.4%, while the high anxiety recast group only scored 40.8% 57 María Luquin & Hanne Roothooft ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 on average. The highest score was obtained by the low-anxiety recast group, with a gain score of 30.2%. Moreover, when comparing the two CF-types, we note that the high-anxiety recast group improved twice as much (gains of 18.9%) as the high-anxiety metalinguistic group (gains of 9%). Figure 2. Pre- and post-test results of the different anxiety groups These tendencies might be more clearly visualized in Figure 2, which shows the pre- and post-test results of the different anxiety groups. As can be seen in Figure 2, the low-anxiety groups tended to benefit more from CF than the high-anxiety ones, and the greatest progress was made by the low-anxiety recast group. We can also observe from Figure 2 that the low-anxiety metalinguistic group had a similar post-test score to the highanxiety recast group, even though recasts were generally more effective. Finally, Figure 2 illustrates the greater progress made by the high-anxiety recast group in comparison with the high-anxiety metalinguistic group. 4.3. Learner Uptake To identify the learner uptake moves that occurred during the treatment, the students’ responses following CF were classified into four categories (repetition of the original mistake; repair; different mistake; and ignore). ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 The effects of oral corrective feedback and language anxiety… 58 Modified output No modified output Total corrections Recasts (high-anxiety) 30 5 35 85.8% 14.3% 100% Metalinguistic (high-anxiety) 34 9 43 79% 21% 100% Recasts (low-anxiety) 21 8 29 72.4% 27.6% 100% Metalinguistic (low-anxiety) 23 5 28 82.1% 17.9% 100% Table 4. Frequency of modified output Turning now to the experimental evidence on learner uptake, the number of feedback-moves and all the instances of modified output were counted (see Table 4). Out of a total number of 64 recasts, 35 were aimed at the high-anxiety group and 29 at the low-anxiety group. Similarly, out of 71 metalinguistic feedback corrections, 43 occurred in the high-anxiety group and almost half of them (28) in the low-anxiety group. Concerning modified output, Table 4 shows that there is not much difference between the four groups. Table 5 illustrates the rate of learner repair and no repair in the modified output responses for the low- and high-anxiety recast and metalinguistic feedback groups. As can be seen from Table 5, the high- (69%) and low-anxiety (66%) recast groups together produced a considerably higher rate of repair (43 repairs out of 51 modified output moves) than the high- (28%) and low-anxiety (46%) metalinguistic feedback groups (25 out of 57). As for differences in language anxiety, we can observe that the non-anxious students from the metalinguistic group repaired their errors to a greater extent (46%) than the anxious students from the same group (28%), while the difference between the high- and low-anxiety groups from the recast group is barely notable (69%, 66%, respectively). 59 María Luquin & Hanne Roothooft ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 Repair No repair Total modified output Recasts (highanxiety) 24 6 30 69% 17% 100% Metalinguistic (high-anxiety) 12 22 34 28% 51% 100% Recasts (lowanxiety) 19 2 21 66% 7% 100% Metalinguistic (low-anxiety) 13 10 23 46% 36% 100% Table 5. Frequency of learner repair A series of four chi-square tests of independence comparing the number of modified output and repair moves of the different groups revealed that there were no significant differences between the high-anxiety recast and the low-anxiety recast group (chi=1.03, p=0.6) or between the high-anxiety metalinguistic and the low-anxiety recast group (chi=2.51, p=0.3). However, significant differences were found between the highanxiety recast and the high-anxiety metalinguistic group (chi=12.94, p=0.0015) and between the low-anxiety recast and the low-anxiety metalinguistic group (chi=6.38, p=0.04). 5. Discussion The analyses revealed that CF positively affected the students’ pronunciation of the –ed ending, since both experimental groups made significant improvements on the post-test, but contrary to expectations no significant difference between recasts and metalinguistic feedback was evident. The significant differences can only be found between recasts and control, confirming previous work on the positive effects of recasts on pronunciation (Lyster et al., 2013; Mohammadi, 2014; Saito & Lyster, 2012ab). However, although no statistical differences were found between the two CF types, ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 The effects of oral corrective feedback and language anxiety… 60 students receiving recasts appeared to perform better than those in the metalinguistic group. One of the few studies which has compared the effects of recasts and prompts on the acquisition of a phonological feature (Gooch et al., 2016) also found that both CF-types caused students to improve over time and only found differences with regard to the type of task used on the pre- and post-tests. While recasts in their study were particularly beneficial in controlled tasks, prompts were found to be helpful both in controlled and free production tasks. However, our study only included one type of task, a controlled reading-aloud test. Regarding the impact of anxiety, both high- and low-anxiety learners improved their pronunciation with the two CF types, and there was also some indication that low-anxiety learners benefited more, especially the low-anxiety recast group, which is in line with Sheen (2008). It also appeared that high-anxiety learners tended to respond better to recasts than to metalinguistic feedback, as was found by Rassaei (2015). Nonetheless, it needs to be stressed that no statistical differences between the anxietygroups were found. It may be that the tests lacked power because of the small sample size. When looking at modified output and repair, we noted that anxiety did not seem to play a role in the recast-group, where there was a high number of repairs in both low- and high-anxiety groups. However, students receiving metalinguistic feedback were able to repair their errors much more frequently if they were low-anxiety students, compared to the high-anxiety metalinguistic group, even though these differences were not statistically significant. The chi-square tests indicated that there were no significant differences between the high- and low-anxiety groups, but only between the type of CF. Given the fact that uptake after recasts usually involves a simple repetition of the correct form provided by the teacher, it is not surprising that recasts led to a higher number of repair. Gooch et al. (2016) also noted high levels of uptake after both recasts and prompts directed at pronunciation errors, but a higher level of repair after recasts. In our study, both the recast and metalinguistic groups tried to modify their output frequently after receiving CF, but the metalinguistic group had a more difficult task, since they did not receive a correct model and had to try to self-correct their errors. Nonetheless, as seen before, both types of CF led to significant improvements from pre- to post-test. Unlike in Gooch et al. (2016), we did not include form-focused instruction, but only provided feedback without focusing on the pronunciation rule beforehand, although a post-test questionnaire confirmed that students were aware of the rule for 61 María Luquin & Hanne Roothooft ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 -ed. This study thus shows that oral CF directed at pronunciation errors can be effective, even without providing explicit pronunciation instruction. Moreover, it is important to note that this study differs from those by Gooch et al. (2016), Rassaei (2015) and Saito and Lyster (2012ab) in that our learners were low-level secondary learners and not university/adult students. It is noteworthy that even pre-intermediate younger learners can improve their pronunciation as a result of oral CF in the form of recasts and metalinguistic feedback. 6. Conclusion, Limitations, and Future Research The aim of this research project was to better understand to what extent foreign language anxiety plays a role in CF directed at pronunciation errors. To this purpose, the interference of anxiety in two types of CF (recasts and metalinguistic feedback) on a specific pronunciation feature as well as in their responses to CF was analyzed. The current findings supported the effectiveness of CF for the students’ pronunciation development. Furthermore, although recasts did not show a marked difference with respect to metalinguistic feedback, both experimental groups significantly improved from pre- to post-test. However, only the recast-group significantly outperformed the control group, confirming the positive effects of recasts on pronunciation found in previous studies. Regarding anxiety, even though statistical significance was not reached, we can observe certain trends in favour of low-anxiety learners, who obtained better results and were able to repair their errors to a greater extent than their more anxious counterparts. High-anxiety learners also appeared to benefit more from recasts than from metalinguistic feedback. A number of limitations of the present study need to be taken into account. First, as indicated earlier, a delayed post-test to investigate the longterm effects of the treatment was not administered due to time constraints. Second, the students were classified as either highly anxious or not anxious, while those whose score on the questionnaire fell within one standard deviation below or above the mean were excluded, which means that students with moderate levels of anxiety were not taken into account. What is more, the elimination of these middle range learners resulted in a small sample size for the different anxiety groups, which may be why statistically significant differences could not be identified. Fourth, the shortness of the one-hour ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 The effects of oral corrective feedback and language anxiety… 62 treatment for each CF type may have hindered a substantial positive effect on second language acquisition regardless of the anxiety level. Fifth, the fact that the verbs in the past were provided in the written form is likely to have triggered a prompting effect, making the students pronounce them the way they were written ([lɪved]). Finally, pronunciation was only tested through a controlled reading-aloud task, which means we cannot draw any conclusions about the effect of oral CF on students’ free production of the –ed morpheme. Future research should take these limitations into account, by comparing the effects of recasts and prompts in larger groups of low- and high-anxiety students, and by including both controlled and free production tasks. Different phonological targets should also be studied. Despite the aforementioned limitations, this study adds to the limited body of research focusing on the effects of different types of oral CF on pronunciation development, showing that both recasts and metalinguistic feedback are useful for improving EFL students’ pronunciation of the -ed ending, with a clear advantage for recasts. The present study also indicates that foreign language anxiety can play a role in the effectiveness of oral CF directed at phonological targets and that this issue is worthy of further investigation. References Ammar, A., & Spada, N. (2006). One size fits all? Recasts, prompts and L2 learning. Studies in Second Language Acquisition, 28(4), 543-574. https:// doi.org/10.1017/S0272263106060268. Awan, R. N., Azher, M., Anwar, M. N., & Naz, A. (2010). An investigation of foreign language classroom anxiety and its relationship with students’ achievement. Journal of College Teaching & Learning, 7, 33–40. https:// doi.org/10.19030/tlc.v7i11.249. Chowdhury, S. (2013). Hansel and Gretel. Retrieved March 2, 2017 from https:// prezi.com/lzjztgn3g1wd /a-poor-woodcutter-and-his-wife-had-two- children-named-hansel/ DeKeyser, R. (1993). The effect of error correction on L2 grammar knowledge and oral proficiency. Modern Language Journal, 77(4), 501-514. https:// doi.org/10.1111/j.1540-4781.1993.tb01999.x. 63 María Luquin & Hanne Roothooft ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 Dilans, G. (2010). Corrective feedback and L2 vocabulary development: Prompts and recasts in the adult ESL classroom. The Canadian Modern Language Review, 66(6), 787-815. https://doi.org/10.3138/cmlr.66.6.787. Doughty, C., & Varela, E. (1998). Communicative focus on form. In C. Doughty & J. Williams (Eds.), Focus on form in classroom second language acquisition (pp. 114-138). New York: Cambridge University Press. Ellis, R. (2007). The differential effects of corrective feedback on two grammatical structures. In A. Mackey (Ed.), Conversational Interaction in Second Language Acquisition: A Collection of Empirical Studies (pp.339-360). Oxford: Oxford University Press. Ellis, R. (2010). A framework for investigating oral and written corrective feedback. Studies in Second Language Acquisition, 32, 335-349. https:// doi.org/10.1017/S0272263109990544. Ellis, R., Loewen, S., & Erlam, R. (2006). Implicit and explicit corrective feedback and the acquisition of L2 grammar. Studies in Second Language Acquisition, 28, 339-368. https://doi.org/10.1017/S0272263106060141. Goo, J., & Mackey, A. (2013). The case against the case against recasts. Studies in Second Language Acquisition, 35, 127-165. https://doi.org/10.1017/ S0272263112000708. Gooch, R., Saito, K., & Lyster, R. (2016). Effects of recasts and prompts on L2 pronunciation development: Teaching English /ɹ/ to Korean adult EFL learners. System, 60, 117-127. https://doi.org/10.1016/j.system.2016.06.007. Guenther, L. (n.d). Little Red Riding Hood. Retrieved March 2, 2017 from http:// www.dltk-teach.com/rhymes/littlered/story.htm Han, Z. (2002). A study of the impact of recasts on tense consistency in L2 output. TESOL Quarterly, 36(4), 543-572. https://doi.org/10.2307/3588240. Hashemi, M., & Abbasi, M. (2013). The role of the teacher in alleviating anxiety in language classes. International Research Journal of Applied and Basic Sciences, 4(3), 640-646. Horwitz, E., Horwitz, B., & Cope, J. (1986). Foreign language classroom anxiety. The Modern Language Journal, 70, 125-132. ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 The effects of oral corrective feedback and language anxiety… 64 Ishida, M. (2004). Effects of recasts on the acquisition of the aspectual form of –te i (ru) by learners of Japanese as a foreign language. Language Learning, 54(2), 311-394. https://doi.org/10.1111/j.1467-9922.2004.00257.x. Krashen, S.D. (1982). Principles and Practice in Second Language Acquisition. Oxford: Pergamon Press. Lasagabaster, D., & Sierra, J.M. (2005). Error correction: students’ versus teachers’ perceptions. Language Awareness, 14, 112-127. Leeman, J. (2003). Recasts and L2 development: Beyond negative evidence. Studies in Second Language Acquisition, 25(1), 37-63. https://doi. org/10.1017/S0272263103000020. Li, S. (2010). The effectiveness of corrective feedback in SLA: a meta-analysis. Language Learning, 60(2), 309-365. https://doi. org/10.1111/j.1467-9922.2010.00561.x. Loewen, S. (2005). Incidental focus on form and second language learning. Studies in Second Language Acquisition, 27, 361–386. Loewen, S., & Nabei, T. (2007). Measuring the effects of oral corrective feedback on L2 knowledge. In A. Mackey (Ed.), Conversational Interaction in Second Language Acquisition: A Collection of Empirical Studies (pp.361- 377). Oxford: Oxford University Press. Lyster, R. (1998). Negotiation of form, recasts, and explicit correction in relation to error types and learner repair in immersion classrooms. Language Learning, 48, 183-218. Lyster, R., & Ranta, L. (1997). Corrective feedback and learner uptake. Studies in Second Language Acquisition, 19, 37-66. https://doi.org/10.1017/ S0272263197001034. Lyster, R., & Saito, K. (2010). Oral feedback in classroom SLA. A meta-analysis. Studies in Second Language Acquisition, 32, 265-302. https://doi. org/10.1017/S0272263109990520. Lyster, R., Saito, K., & Sato, M. (2013). Oral corrective feedback in second language classrooms. Language Teaching, 46, 1-40. https://doi.org/10.1017/ S0261444812000365. 65 María Luquin & Hanne Roothooft ELIA 19, 2019, pp. 41-70 DOI: http://dx.doi.org/10.12795/elia.2019.i19.03 Mackey, A., Gass, S.M., & McDonough, K. (2000). How do learners perceive interactional feedback? Studies in Second Language Acquisition, 22, 471- 497. https://doi.org/10.1017/S0272263100004022. Mackey, A., & Goo, J. (2007). Interaction research in SLA: A meta-analysis and research synthesis. In A. Mackey (Ed.), Conversational Interaction in Second Language Acquisition: A Collection of Empirical Studies (pp.407- 453). Oxford: Oxford University Press. Mackey, A., & Philp, J. (1998). Conversational interaction and second language development: Recasts, responses, and red herrings? The Modern Language Journal, 82(3), 338-356. https://doi.org/10.1111/j.1540-4781.1998. tb01211.x. McDonough, K. (2005). Identifying the impact of negative feedback and learners’ responses on ESL question development. Studies in Second Language Acquisition, 27(1), 79-103. https://doi.org/10.1017/ S0272263105050047. McDonough, K. (2007). Interactional feedback and the emergence of simple past activity verbs in L2 English. In A. Mackey (Ed.), Conversational Interaction in Second Language Acquisition: A Collection of Empirical Studies (pp.323-338). Oxford: Oxford University Press. McDonough, K., & Mackey, a. (2006). Responses to recasts: repetition, primed production, and linguistic development. Language Learning, 56(4), 693- 720. https://doi.org/10.1111/j.1467-9922.2006.00393.x. Mohammadi, M. A. (2014). Corrective feedback interventions and EFL learners’ pronunciation: A case of –s or –es ending words. International Journal of Learning & Development, 4(1), 40 – 58. Muranoi, H. (2000). Focus on form through interaction enhancement: Integrating formal instruction into a communicative task in EFL classrooms. Language Learning, 50, 617-673. https://doi.org/10.1111/0023-8333.00142. Ortega, L. (2009). Understanding second language acquisition. London: Hodder Education. Rassaei, E. (2015). Oral corrective feedback, foreign language anxiety and L2 development. System, 49, 98-109. https://doi.org/10.1016/j.system.2015.01.002.