Ewa Cieślicka, Arkadiusz Rojczyk University of Silesia in Katowice, Poland Self-reported vs. Self-rated Pronunciation in a Non-native Language A b s t r a c t The study investigates how their own accent in English is self-perceived by Polish learn- ers. More specifically, we compare how, and to what extent, self-reported pronunciation dif- fers from self-rated pronunciation prior to and after the exposure to one’s recorded speech. Previous research on non-native accent rating has concentrated on scores obtained from native speakers or other proficient speakers of English. In the current study, we concentrate on how learners evaluate their own accent in English for parameters such as pronunciation, articulation, and f luency. We also introduce an independent variable of proficiency to see if it interacts with the perception of learners’ pronunciation. Both quantitative and qualitative analyses were conducted and the result showed that there are no major differences between how learners report their accent in English and how they rate it from the recording of their own speech. It indicates that the general self-image of one’s accent is fairly stable and expo- sure to the sample of one’s speech does not change the overall self-perception. Keywords: psycholinguistics, phonetics, self-assessment of pronunciation, language learning Introduction Several studies have provided the rationale for self-assessment in language learning, namely: promotion of learning and increased learner motivation, a raised level of awareness, the positive impact of metacognitive awareness on goal-orientation, an expansion in range of assessment techniques, shared assess- ment burden and the autonomy of learners as a post-course effect (Oscarsson, 1989; Salimi, Kargar, & Zareian, 2014). The significance of accurate self-rating is a result of the assumption that standardized tests do not fully succeed in evaluating language competence, due to creating artificial contexts for language Theory and Practice of Second Language Acquisition vol. 3 (2) 2017, pp. 69–86 Ewa Cieślicka, Arkadiusz Rojczyk70 use (Spence-Brown, 2001). Self-assessment has been argued to constitute an in- dispensable part of autonomous learning (Grader & Miller, 1999; Harris, 1997). Hunt, Gow, and Barnes (1989, p. 207) emphasize that without self-assessment, “there can be no real autonomy.” The purpose of this study is to establish how Polish learners assess their own pronunciation skills at secondary and terti- ary levels of education. The research aimed to check whether any significant changes could be observed in the learners’ self-rating scores before and after they were exposed to the pre-recorded samples of their own productions. We aimed to tap the expectations students hold for their L2 pronunciation skills and see if—when confronted with reality—they believe they have met them. Another goal was to look for differences between the two levels of pro- ficiency to see how experience affects self-ratings. The following paper also aims to address the issue of learners’ self-judgment of their pronunciation, confidence in their own pronunciation skills and awareness thereof. We sought to find whether there arises the problem of either overestimation or underesti- mation of competence, of which, as MacIntyre, Noels, and Clément (1997, p. 279) speculate, “self-enhancement would probably facilitate language learning while self-derogation would impair progress.” In their research, MacIntyre et al. (1997, p. 281) “took ratings of perceived competence prior to the language tasks” and note that, “it also would be interesting to have self-assessments done after task performance. […] Post facto reporting might actually enhance the biases, because highly anxious students might focus on errors in performance and less anxious students […] on their communicative successes.” The present study can therefore be considered a follow-up investigation with a focus on accent rating. It combines qualitative and quantitative methods of research in pursuance of a deeper understanding of both the tendencies observed in self- rating and the possible reasons behind them. Previous Research Nowacka (2012) points to the importance of good English pronunciation as perceived by L2 learners. In her survey, almost all respondents (98%) agreed on its significance, while 89% of them believed that students should aim for native English pronunciation. The findings of a number of studies into the reliability of self-assessment suggest a pattern of overall agreement between self-assessments and ratings (Blanche, 1985; Oscarsson, 1978; Rea, 1981). Oscarsson (1978) notes that, when given scaled descriptions of performance as rating tools, adult learners demonstrate the capacity to accurately evaluate their linguistic ability. It has also been reported that over the course of second Self-reported vs. Self-rated Pronunciation in a Non-native Language 71 language practice and through feedback, specific student training and/or do- mestic immersion speakers display an improved ability to correctly self-assess their sound discrimination skills and oral performance (Chen, 2008; Dolosic, Brantmeier, Strube, & Hogrebe, 2016; Ross, 2006; Yules, Hoffman, & Damico, 1987). This indicates that more advanced learners should find self-assessment of their L2 pronunciation less problematic. Previous findings also point to the importance of clear and explicitly stated assessment criteria which help im- prove the effectiveness and precision of self-assessment (Chen, 2008; Dolosic et al., 2016). Some learners have admitted to employing (among other cognitive strategies) phonetic self-evaluation as a tool in developing their pronunciation skills (Nowacka, 2008). However, other research suggests that problems with valid self-evaluation occur even in advanced L2 learners (Dlaska & Krekeler, 2008). Although the results of expert raters’ assessments of pronunciation and self-ratings coincided in 85% of all cases, the students only managed to identify a little less than a half of all the speech sounds which the raters con- sidered inaccurate. In their study, MacIntyre et al. (1997, p. 274) observe that “subjective, self-rated proficiency relates substantially to actual proficiency, but the two are not isomorphic.” Raasch (1980) and Ross (2006) note rather low correspondence between self-assessment and expert judgments. Nowacka (2008) also found that graduates from English departments in Poland tend to lack self-criticism, and are therefore incapable of accurate and objective self- evaluations of their pronunciation. Blue (1988) shows that nationality serves as an important factor, with some nationalities tending towards overestimation of their level and others likely to underestimate it. There is also evidence to suggest that students whose self-assessments are unrealistically high or low are more likely to give up on language learning than those who are able to assess their skills realistically (Blue, 1994). Previous research also shows that non-native speakers tend to be consistent in rating other non-native speakers’ pronunciation more critically than they rate their own (Episcopo, 2009). Breitkreutz, Derwing, and Rossiter’s (2001) study draws attention to the fact that in stand-alone pronunciation classes, some of the most popular CDs used in computer labs focus solely on phonetic segments. Consequently, it may be assumed that the learners’ awareness of the primacy of prosodic variables is low. In Derwing and Rossiter (2001), nearly 40% of ESL students were unable to identify specific problems with their pronunciation. Meanwhile, the remainder usually pointed to segmentals (84% of all problems mentioned). Only 10% of the acknowledged difficulties were related to prosodic aspects of pronunciation. Ewa Cieślicka, Arkadiusz Rojczyk72 The Current Study In the current study we look into how and to what extent self-reported pro- nunciation differs from self-rated pronunciation. By self-reported pronunciation we mean evaluating one’s pronunciation skills in general without referring to any particulars samples of one’s speech. By self-rated pronunciation we mean evaluating the recording of one’s pronunciation by listening to a recording of one’s speech. The main objective is to directly confront scores obtained for self-reported pronunciation with those obtained for self-rated pronunciation. Such confrontation allows us to look into the stability of the perception of one’s pronunciation skills. Logically, one of the following patterns is predicted to emerge in the results: 1. The scores for self-reported pronunciation are equal to the scores obtained for self-rated pronunciation, indicating that the awareness of one’s pronuncia- tion skills is stable and is not affected by specific performance in a given speech sample. 2. The scores for self-reported pronunciation are higher than the scores obtained for self-rated pronunciation, indicating that the awareness of one’s pronun- ciation skills is generally overrated and that it is verified negatively when listening to one’s recorded speech samples. 3. The scores for self-reported pronunciation are lower than the scores obtained for self-rated pronunciation, indicating that general self-perception of one’s pronunciation is underestimated and that it is enhanced when listening to one’s performance in a recording. Another categorical predictor used in the current study is the proficiency of speakers. Lower-proficiency students versus higher-proficiency students are hy- pothesized to perform differently in their ratings of self-reported and self-rated pronunciation; however, the exact pattern of differences cannot be predicted. The lower-proficiency speakers/raters had not received any specialized train- ing in pronunciation and phonetics. The higher-proficiency speakers/raters had completed a two-year course in pronunciation and phonetics. While the actual differences in pronunciation skills between the two groups are obvious, it is hard to predict how proficiency and pronunciation training will affect self-reporting and self-rating of their pronunciation. One or some of the following patterns are logically likely to emerge: 1. Higher-proficiency speakers/raters will self-report their pronunciation higher than lower-proficiency speakers/raters because they have confidence in their pronunciation skills as a result of longer exposure to English and the com- pletion of pronunciation and phonetics training. 2. Lower-proficiency speakers/raters will self-report their pronunciation higher than higher-proficiency speakers/raters because they are not acquainted with Self-reported vs. Self-rated Pronunciation in a Non-native Language 73 detailed intricacies of English pronunciation and are unaware of mistakes they make. 3. When listening to the recording of their own speech, higher-proficiency speakers/raters will self-rate their pronunciation higher than lower-proficien- cy speakers/raters because, again, they will have more confidence in their performance. 4. When listening to the recording of their own speech, lower-proficiency speakers/raters will self-rate their pronunciation higher than higher-profi- ciency speakers/raters because they will be unaware of and will not notice their pronunciation mistakes and problems. The last issue to resolve is if self-reported pronunciation and self-rated pro- nunciation will interact significantly between the two groups. We hypothesized that the higher-proficiency speakers/raters would self-rate their pronunciation from the recording higher than in self-reports, because the knowledge of pho- netics and pronunciation details may decrease general self-perception of their pronunciation, however, when they were exposed to the real samples of their speech, they would be likely to appreciate it with higher scores. On the other hand, lower-proficiency speakers/raters were hypothesized to self-rate their pronunciation in the recording lower than in self-reports, because the exposure to their own real pronunciation would make them realize that it did not match the target native pronunciation. However, these hypotheses are not a very strong bet and other patterns are not much less likely to emerge in the results. Together with the results for pronunciation rating, we also report results for other aspects of speech, such as articulation and fluency. Moreover, we decided to include descriptive qualitative questions about the strong and weak points of the speakers/raters pronunciation. Participants A total of 66 learners of English took part in the study. The lower-profi- ciency group consisted of 35 participants (20 females; 15 males) with a mean age of 18 years. They were all second-grade students of the International Baccalaureate Diploma Programme (classes taught exclusively in English) at the 1st Liceum Ogólnokształcące im. Edwarda Dembowskiego in Gliwice. Their estimated proficiency was B1–B2 in the Common European Framework of References for Languages (CEFR). They had had no separate dedicated train- ing in pronunciation or phonetics. The higher-proficiency group consisted of 31 participants (24 females; 7 males) with the mean age of 24.9 years. They were fifth-year students of English recruited from the Institute of English, University of Silesia in Katowice. Their proficiency was C1–C2 in the CEFR. They had completed a two-year course in pronunciation and phonetics which had covered Ewa Cieślicka, Arkadiusz Rojczyk74 both segmental and suprasegmental aspects of English pronunciation. None of the participants in either group reported any speech or hearing disorders nor had any indication of such. Materials The materials included rating of selected aspects of pronunciation and speech production using a 1–9 Likert scale. In the first part of the experiment, the aspects to rate referred to the participants’ self-perception without any ex- posure to the recordings of their speech. The same aspects were used in the second part of the experiment, when the participants rated recordings of their own speech. The following parameters were used for rating: 1. How do you rate your pronunciation? [(1) strongly Polish-accented—(9) native-like]. 2. How do you rate your articulation? [(1) very unclear—(9) very clear]. 3. How do you rate your fluency? [(1) very poor—(9) very good]. We also decided to include a more subjective, emotionally-loaded parameter of expected changes and improvement in the participants’ pronunciation: 1. How much do you want to change in your pronunciation? [(1) nothing—(9) everything]. Additionally, two descriptive questions for qualitative analysis were pro- vided: 1. What do you consider to be the strengths of your pronunciation? 2. What do you consider to be the weaknesses of your pronunciation? Procedure In the first part of the experiment, the participants filled in a questionnaire with afore-described points relating to their self-perceived pronunciation. They were instructed to objectively rate and describe how they perceived their pro- nunciation and accent in English. In order to reduce social desirability bias, the personal details provided by the respondents were kept to a minimum—initials and dates of birth—for proper assignment of the recordings in future stages of the research. After completing the questionnaire, the participants were recorded reading a short passage in English: I entered the hotel manager’s office and sat down. I had just lost $50 and I felt very upset. “I left the money in my room,” I said, “and it’s not there now.” The manager was sympathetic, but he could do nothing. “Everyone’s losing money these days,” he said. He started to complain about this wicked Self-reported vs. Self-rated Pronunciation in a Non-native Language 75 world but was interrupted by a knock at the door. A girl came in and put an envelope on his desk. It contained $50. “I found this outside this gentle- man’s room,” she said. “Well,” I said to the manager, “there is still some honesty in this world!” They were instructed to read it with a natural tempo in the most natural way. The lower-proficiency students were recorded in a quiet room with a dy- namic Shure SM58 microphone connected to the Quad-Capture USB interface (Roland). The higher-proficiency students were recorded in the Acoustics- Phonetics Laboratory at the Institute of English, University of Silesia in Katowice, in a sound-proof booth. The signal was captured at 44100 Hz (24 bit quantization) through a headset dynamic Sennheiser HMD 26 microphone fed by a USBPre2 (Sound Devices) amplifier. All recordings were saved as .wav files. After two months, the participants were invited to take part in the second part of the experiment in which they were asked to listen to their own recording and rate it using the same parameters as in the first part. A two-month period guaranteed that the participants did not remember how they had rated their pronunciation in the self-reporting stage. The recordings were played through headphones from a laptop. All participants easily recognized their voice and were aware of the fact that they were to rate their own pronunciation from the recording. Analysis and Results Both quantitative and qualitative analyses were conducted. The quantitative data were analyzed in a mixed 2x2 ANOVA with a between-subject independent variable of proficiency (lower/higher), a within-subject independent variable of task (self-rate your pronunciation/rate your recording) and a dependent variable of rating (1–9). The qualitative data were analyzed from descriptive answers concerning the strengths and weaknesses of the participants’ self-reported pro- nunciation and from the exposure to the recordings. Quantitative analysis Rated pronunciation. The mean rating for self-reported pronunciation in the lower-proficiency group was 5.6 (SE = 0.2) and 5.5 (SE = 0.2) in the higher- proficiency group with the insignificant difference [F(1, 64) = .15, p > .05], Ewa Cieślicka, Arkadiusz Rojczyk76 indicating that both groups self-reported their pronunciation almost alike. The ratings between 5 to 6 suggest that both groups perceived their pronunciation as only a little higher than average. When exposed to the recording of their own speech, the participants rated their pronunciation almost identical as in the self-reporting stage. The mean rating was 5.5 (SE = 0.2) for the lower- proficiency and 5.8 (SE = 5.8) for the higher-proficiency group respectively. The difference between the groups was not significant [F(1,64) = .88, p > .05]. Predictably, there were no significant within-subject differences between self- reported pronunciation and pronunciation rated from the recording in either the lower-proficiency group [F(1, 34) = .07, p > .05] or higher-proficiency group [F(1,30) = 2.6, p > .05]. Rated articulation. The self-reported mean scores for articulation were very similar to the ones for pronunciation. The mean rating in the lower- proficiency group was 5.6 (SE = 0.2) compared to 5.4 (SE = 0.2) in the higher- proficiency group. The difference was not significant [F(1, 63) = .79, p > .05]. When exposed to the recording, the lower-proficiency students had a mean score of 5.2 (SE = 0.3). The mean score in the higher-proficiency group was 5.6 (SE = 0.3). Again, the difference was statistically insignificant [F(1, 64) = 1.14, p > .05]. Within-group comparisons for each group between scores obtained from self-reporting and rating the recording revealed that the differences were not significant in either group. In the lower-proficiency group the difference was close to significant [F(1, 33) = 3.15, p = .08], pointing to some trend that listening to the recording decreased ratings of articulation (5.6 to 5.2), however, again, the difference was not statistically significant. In the higher-proficiency group the trend was reversed (5.4. to 5.6), but it was also statistically insignifi- cant [F(1, 30) = .74, p > .05]. Rated fluency. Ratings for fluency did not differ much from the already reported ratings for pronunciation and articulation. Self-reported scores did not differ significantly between the lower-proficiency (M = 5.9; SE = 0.2) and higher-proficiency (M = 5.7; SE = 0.2) groups [F(1, 64) = .53, p > .05]. A sig- nificant difference emerged when the participants listened to the recording. The lower-proficiency students rated their fluency significantly lower (M = 4.9; SE = 0.3) than higher-proficiency students (M = 6.2; SE = 0.3) [F(1, 64) = 9.53, p < .01]. A mixed analysis with two categorical predictors of proficiency (lower / higher) and task (self-rate your pronunciation / rate your recording) revealed a significant interaction of proficiency and task [F(1, 64) = 14.5, p < .001]. Figure 1 shows that listening to the recording triggered different tendencies in rating fluency depending on proficiency. Self-reported vs. Self-rated Pronunciation in a Non-native Language 77 Figure 1. Interaction plot between proficiency and task for rated fluency. While the higher-proficiency students increased their mean ratings from 5.7 to 6.2, the lower-proficiency students decreased their mean ratings from 5.9 to 4.9. The post hoc Bonferroni tests showed that the significant main effect was largely contributed to by the lower-proficiency group which significantly decreased rating of their fluency when exposed to the recording ( p < .01). This difference was not significant in the higher-proficiency group ( p = .6). Desired change in pronunciation. The reported desired change in pro- nunciation prior to the recording was almost identical: 5.6 (SE = 0.3) and 5.5 (SE = 0.3) for the higher- and lower-proficiency group respectively. When exposed to the recording, the participants exhibited slightly different trends depending on proficiency. The higher-proficiency students decreased the rat- ing of desired change to 5.5 (SE = 0.3), compared to the lower-proficiency students, who increased their mean scores to 6 (SE = 0.3); however, the differ- ence between the two groups was not statistically significant [F(1, 64) = 1.58, p > .05]. Within-group comparisons revealed that the difference in rating for self-reporting and after listening to the recording was not significant in either the higher-proficiency [F(1, 30) = .01, p > .05] or lower-proficiency group [F(1, 33) = 1.43, p > .05]. Ewa Cieślicka, Arkadiusz Rojczyk78 Qualitative Analysis The participants’ answers provided for the two descriptive questions were assigned into one or more of four groups: as being related to pronunciation, articulation, tone of voice, and/or fluency of speech. Other answers, either un- related, unclear or difficult to categorize, were not taken into account for the purpose of this analysis, and were therefore omitted, but will be brought up in the general discussion. Differences in self-reporting. For question one, “What do you consider to be the strengths of your pronunciation?”, out of 31 respondents in the higher- proficiency group, 18 students provided answers which pointed to one or more of the four aspects of pronunciation mentioned above. The same was the case for 23 people in the lower-proficiency group of 35. The percentage share for each mentioned feature is presented in Figure 2. Figure 2. Bar graphs for answers assigned to the four aspects for both levels of proficiency, strength of pronunciation. It appears that there is a tendency for higher-proficiency speakers to ini- tially value their pronunciation and articulation slightly higher than the lower- proficiency group, but they are much less eager to consider their tone of voice satisfactory. Both groups reported to be equally content with their speech flu- ency. The response rate, however, is not to be considered very solid. For the second question, “What do you consider to be the weaknesses of your pronunciation?”, the response rate was considerable. Out of 31 respond- ents in the higher-proficiency group, assignable answers were provided by 29 students. In the lower-proficiency group, the question was answered by 30 of 35 students. Once more, the percentage share is represented in Figure 3. Self-reported vs. Self-rated Pronunciation in a Non-native Language 79 Figure 3. Bar graph for answers assigned to the four aspects for both levels of proficiency, weakness of pronunciation. The difference in how often pronunciation was mentioned is fairly consistent with what has been revealed in the first question—namely that lower-proficien- cy speakers are less satisfied with their pronunciation. Remarkably, identical disregard for their articulation and tone of voice was shown in both groups, but with tone of voice being mentioned less frequently. Lower-proficiency speak- ers were expressively less likely to refer to their fluency when discussing their weaknesses. Self-reporting vs. post-exposure rating in higher-proficiency partici- pants. The change in response rates between the self-reporting and post-expo- sure stages of the experiment was not substantial, with six more respondents answering the first descriptive question, and one fewer answering the second descriptive question (see Figure 4). Figure 4. Bar graph for descriptive answers assigned to the four aspects for higher-proficiency speakers, strength of pronunciation. Ewa Cieślicka, Arkadiusz Rojczyk80 The second stage, self-rating, resulted in a lower percentage share of pro- nunciation among the provided answers, although the change is not very sig- nificant. It may indicate the students’ initial—though moderate—overestimation of their skills. The change in fluency percentage, however, suggests that the exposure to the pre-recorded samples left the students pleasantly surprised by their own fluency of speech. There is only very minor variability in the rating of articulation and tone (see Figure 5). Figure 5. Bar graph for descriptive answers assigned to the four aspects for higher-proficiency speakers, weakness of pronunciation. Following the second part of the experiment, the only major changes in the descriptive answers were related to pronunciation and fluency. The higher- proficiency speakers were slightly more dissatisfied with their pronunciation, but their fluency was not mentioned as a weakness quite as often anymore. The frequency of their pointing to articulation and tone did not undergo any serious change. Self-reporting vs. post-exposure rating in lower-proficiency partici- pants. Similarly to what was the case for the university students, the difference in response rates between the two stages of study was not substantial, the only change occurring in the first descriptive question, which came to be answered by three more participants in the second part of the experiment (see Figure 6). Self-reported vs. Self-rated Pronunciation in a Non-native Language 81 Figure 6. Bar graph for descriptive answers assigned to the four aspects for lower-proficiency speakers, strength of pronunciation. After listening to their speech samples, the students were less likely to point to their tone of voice and/or fluency as one of their strengths, but appeared rather pleased with their pronunciation and articulation (see Figure 7). Figure 7. Bar graph for descriptive answers assigned to the four aspects for lower-proficiency speakers, weakness of pronunciation. It was not until the second stage of the experiment that the students seemed to widely acknowledge their lack of fluency as their weakness—the switch ap- pears to have been very rapid. Again, as a logical consequence of the respond- ents’ contetment with their pronunciation and articulation, those two features were mentioned much less frequently, in the case of articulation the frequency dropping to a near half of what was observed in the first stage. Ewa Cieślicka, Arkadiusz Rojczyk82 General Discussion The analysis of the quantitative results will be presented from two perspec- tives: the perspective of proficiency on self-perception of one’s pronunciation and the perspective of how listening to the recording of one’s speech changes the perception of one’s pronunciation. In the first perspective, what emerges as the most evident is the stability of self-perception of pronunciation irrespec- tive of proficiency. The fact that the higher-proficiency group was both more advanced in language competence and had had prior training in pronunciation did not affect their ratings—even though previous studies have shown that formal phonetic training substantially improves productive pronunciation skills (Nowacka, 2008). It is interesting to note that both groups rated their pronun- ciation between 5 to 6 on a 9-point scale, which points to the fact that they perceived their pronunciation as only slightly higher than the average between native and non-native. It may be concluded, at least in the light of the current results, that students, who should theoretically be much more confident in their pronunciation, are in fact not. Additionally, the ratings of the desired change do not show that less proficient students are more willing to change something in their pronunciation. It runs counter to initial intuition that lower-proficiency and lack of specialized pronunciation training should lead to the higher desire for improvement. Similarly surprising is the fact that the higher-proficiency stu- dents expressed relatively high desire for change, as high as the one expressed by the lower-proficiency students. It leads to the conclusion that proficiency and phonetic training do not guarantee more confidence in pronunciation. As Gardner et al. (1989) observe, anxiety levels normally tend to decline with the rise of experience and proficiency. Whenever a student expects to fail, anxiety results—which could in turn lead to self-derogation bias in the self-assessment of language skills. It could therefore be alleged that higher-proficiency students of English who should, in theory, be confident enough to perceive their pro- nunciation skills as advanced (due to their experience), and who at the same time tend to rate themselves as just slightly above average (between native and non-native) may suffer from a form of language anxiety. Similar results were obtained for articulation and fluency. As regards articulation, it may be argued that the participants did not fully differentiate between articulation and pronunciation. While it may be the case for the lower- proficiency students who had not received any training in pronunciation and phonetics, the higher-proficiency students are likely to have been aware of the difference between pronunciation and articulation. Nevertheless, the mean rat- ing for articulation is very similar to the one for pronunciation in both groups. It leads to the conclusion that articulation is taken to be fully integrated with Self-reported vs. Self-rated Pronunciation in a Non-native Language 83 pronunciation independent of how proficient students are or whether or not they have acquaintance with phonetics. The comparison of the results from self-reporting and after exposure to the recording points to a general conclusion that listening to one’s recording does not change the perception of one’s pronunciation. Listening to the recording did not change ratings for either pronunciation or articulation. The exception here is fluency, which exhibited the influence of task on ratings in an inter- acting pattern. After listening to the recording, the higher-proficiency students increased their mean scores, whereas the lower-proficiency students decreased their scores. It is not easy to explain why this parameter exhibited the effect of task. There are, however, two reasons to consider it as separate from both pronunciation and articulation. One reason is that fluency is somehow discon- nected from pronunciation. In common wisdom it is taken to mean how fast one speaks irrespective of specific correctness of pronunciation and articulation. It is also connected with skills such as the generation of syntactic constructions and verbal fluency. Another reason is that rating fluency from the recording of a read text is not well-grounded methodologically. Fluency seems to be more connected with spontaneous speech, where it is inseparably coupled, at least in general conception of the term, with skills such as word finding and sentence construction. The qualitative analysis was conducted from two perspectives—the perspec- tive of differences in self-reporting among the two levels of proficiency, and the perspective of differences in descriptive answers pre- and post-exposure to one’s recording in each group of participants. Regardless of the level of proficiency, both stages of our study have come to show that out of the four aspects we grouped the descriptive answers into (pronunciation, articulation, fluency, and tone of voice), pronunciation was referred to the most often, both as a weakness and a strength. It could be argued that this is due to the im- portance students of all levels attach to attaining native-like model in accent, intonation, and stress. This conclusion is consistent with Nowacka’s (2012) aforementioned study. When enquired about their weaknesses, the students would frequently point to how “Polish” they sounded. In both groups, there were instances of participants misunderstanding the task and providing anwers which were irrelevant to pronunciation, but instead referred to their vocabulary, listening comprehension or grammar skills, which could be indicative of their poor understanding of what proper pronunciation stands for and equals to. It can be generally concluded that as it comes to fluency, the lower-proficiency students slightly overestimate their fluency whereas the higher-proficiency stu- dents underestimate it. A possible explanation for this is that generally, more is expected from university students in terms of fluency, and they could therefore repeatedly be made to feel that they did not meet certain standards. The only truly significant changes which could be observed in the qualitative questions Ewa Cieślicka, Arkadiusz Rojczyk84 were those related to pronunciation and fluency in the lower-proficiency group between the two parts of the experiment. For fluency, these results are in tact with what has been shown in the quantitative analysis, where the mean decrease in rating was statistically significant. In the case of pronunciation, however, the findings were not distincly reflected in the 1–9 scale ratings in the previous part of the analysis. Regardless of their level of acquaintance with transcription and phonetics, both groups used phonetic symbols to point to their specific articulatory problems—those instances, however, were still very rare. Many students saw their pronunciation as clear and easy to understand, and perceived this intelligibility as a definite strength. Another interesting aspect to note is the switch in response rates between questions concerning the self-reported strengths and weaknesses of the participants’ pronunciation. The increase in the number of answers for the weakness question may suggest that it is much easier for students to think and speak of the flaws in their language skills than of the strengths. Yet we must also acknowledge the possibility that the respond- ents might have been somehow intimidated by the prospect of being verified by comparing their reported strengths and their recording. Therefore, this par- ticular tendency could have more to do with experimental inhibitions among L2 learners than with how they actually perceived their pronunciation skills. Conclusions As may be concluded from the current results, the general perception of one’s pronunciation ability is stable, not affected by specific performance in a given recording and irrespective of proficiency. Students at both levels of proficiency express similar desire for improvement of their pronunciation skills—hence, experience and phonetic training do not seem to translate into confidence. Instances of the participants’ misunderstanding of the survey ques- tions were telling and are to be taken into account for future research, perhaps in the application of even more clearly defined criteria. The current findings that self-reported pronunciation does not differ from the pronunciation rated from the recording of one’s speech forms a starting point for our future investigation in which we intend to apply voice manipulation. The rationale is as follows: if one self-reports one’s pronunciation in the same fashion as they rate it from the recording of their voice, we are interested to see if ratings for the recording of one’s own pronunciation change if one does not know they are listening to their own speech. In other words, we plan to create a situation in which students will rate their own pronunciation, being unaware of the fact that they are actually listening to their own speech. In Self-reported vs. Self-rated Pronunciation in a Non-native Language 85 order to do that, we plan to manipulate voice parameters to the extent that the voice is reliably altered without manipulating other speech parameters that may contribute to accent rating. In order to set out with such a study, we needed to establish that students indeed do not differ in their self-reporting and rating their own recording when they know it is their own speech. The current study has established that self-reporting and the recording rating are fully connected. References Blanche, P., & Merino, B. J. (1989). Self-assessment of foreign-language skills: Implications for teachers and researchers. Language Learning, 39(3), 313–338. Blue, G. M. (1988). Self-assessment: The limits of learner independence. In Individualization and autonomy in language learning. Oxford and London: Modern English Publications in association with The British Council. Blue, G. M. (1994). Self-assessment of foreign language skills: does it work? CLE Working Papers, 3, 18–35. Breitkreutz, J. A., Derwing, T. M., & Rossiter, M. J. (2001). Pronunciation teaching practices in Canada. TESL Canada Journal, 19(1), 51–61. Chen, Y.-M. (2008). Learning to self-assess oral performance in English: A longitudinal case study. Language Teaching Research, 12(2), 235–262. Derwing, T. M., & Rossiter, M. J. (2001). ESL learners’ perceptions of their pronunciation needs and strategies. System, 30(2), 155–166. Dlaska, A., & Krekeler, C. (2008). Self-assessment of pronunciation. System, 36, 506–516. Dolosic, H. N., Brantmeier, C., Strube, M., & Hogrebe, M. C. (2016). Living language: Self- assessment, oral production, and domestic immersion. Foreign Language Annals, 49(2), 302–316. Episcopo, S. M. (2009). Non-native speaker attitudes toward non-native English accents (MA report, University of Texas, Austin). Gardner, D., & Miller, L. (1999). Establishing self-access: From theory to practice. Cambridge: Cambridge University Press. Gardner, R. C., Moorcroft, R., & Metford, J. (1989). Second language learning in an immersion programme: Factors inf luencing acquisition and retention. Journal of Language and Social Psychology, 8(5), 287–305. Harris, M. (1997). Self-assessment of language learning in formal settings. ELT Journal, 51(1), 12–20. Hunt, J., Gow, L., & Barnes, P. (1989). Learner self-evaluation and assessment – A tool for autonomy in the language learning classroom. In V. Bickley (Ed.), Language teaching and learning styles within and across cultures (pp. 207–217). Hong Kong: Institute of Language in Education, Education Department. MacIntyre, P. D., Noels, K. A., & Clément, R. (1997). Biases in self-ratings of second language proficiency: The role of language anxiety. Language Learning, 47(2), 265–287. Nowacka, M. (2008). Phonetic attainment in Polish university and college students of English: A study in the productive and receptive pronunciation skills. Unpublished doctoral disserta- tion. Lublin: Maria Curie-Skłodowska University. Ewa Cieślicka, Arkadiusz Rojczyk86 Nowacka, M. (2012). Questionnaire-based pronunciation studies: Italian, Spanish and Polish students’ views on their English pronunciation. Research in Language, 10(1), 43–61. Oscarson, M. (1989). Self-assessment of language proficiency: Rationale and applications. Language Testing, 6(1), 1–13. Oscarson, M., & Council for Cultural Cooperation, Strasbourg (France). (1978). Approaches to self-assessment in foreign language learning. Distributed by ERIC Clearinghouse. Raasch, A. (1980). Self-evaluation in adult education. Recherches et Echanges, 5, 88–99. Ross, J. A. (2006). The reliability, validity, and utility of self-assessment. Practical Assessment, Research, and Evaluation, 11(10), 1–13. Salimi, M., Kargar, A. A., & Zareian, A. (2014). Self-assessment of pronunciation and its impact on EFL learners’ pronunciation ability. International Journal of Language Learning and Applied Linguistics World (IJLLALW) 6(4), 605–622. Spence-Brown, R. L. (2001). The eye of the beholder: Authenticity in an embedded assessment task. Language Testing, 18(4), 463–481. Yule, G., Hoffman, P., & Damico, J. (1987). Paying attention to pronunciation: The role of self- monitoring in perception. TESOL Quarterly, 21(4), 765–768. Ewa Cieślicka, Arkadiusz Rojczyk Die Selbsteinschätzung der fremdsprachigen Aussprache vor und nach dem Abhören der Aufnahme der eigenen Stimme Z u s a m m e n f a s s u n g Die im vorliegenden Artikel geschilderte Untersuchung betrifft die Selbsteinschätzung der fremdsprachigen Aussprache bei polnischen Schülern der niedrigeren und höheren Lernstufe. Sie konzentrierte sich darauf, die Unterschiede zwischen der Selbsteinschätzung vor und nach dem Abhören der Aufnahme der eigenen Stimme zu vergleichen. In früheren Untersuchungen zur Ausspracheeinschätzung besprach man die den Schülern von Muttersprachlern oder fort- geschrittenen Sprachbenutzern gegebenen Noten, wohingegen in vorliegender Analyse die Einschätzung des eigenen Akzents hinsichtlich solcher Kriterien wie Aussprache, Artikulation und Flüssigkeit im Mittelpunkt steht. Die Forschungsergebnisse wurden dann einer qualitati- ven und quantitativen Analyse unterzogen. Es wurden keine relevanten Unterschiede zwischen der zu erklärten Ausspracheeinschätzung und der Selbsteinschätzung der Aussprache nach dem Abhören der Aufnahme von eigener Stimme festgestellt. Das lässt zum Schluss kommen, dass die allgemeine Wahrnehmung eigener Akzentqualität beständig ist und sie ändert sich nicht in Folge der Reaktion auf solche Aufnahmen. Schlüsselwörter: Psycholinguistik, Phonetik, Selbsteinschätzung der Aussprache, Fremd- sprachenunterricht