An investigation into oral fluency perceived by teachers and students—in a Vietnamese context of English education

This study attempts to investigate how four groups of participants at Tra Vinh University perceive oral fluency. This study is conducted in the hope to obtain more perceptions of oral fluency so future rubrics can be developed based on such perceptions. The study used a set of items stating what oral fluency should be like, which was developed based on the Qualitative aspects of spoken language proposed by the Council of Europe (2019). To test if the items were accepted, the study invited 33 senior English-majored students, 20 English teachers with a BA, 22 English teachers with an MA, and 10 English teachers with a doctorate degree for the study. A five-point Likert scale questionnaire consisting of 18 closed-ended questions and an open-ended question was employed to collect the participants’ perceptions of oral fluency. The results showed that the participants with an MA and the participants with a doctorate degree tend to agree more with the items in the questionnaire while the other groups slightly agreed with most of the items. Furthermore, many of the participants provided the study with their own ideas about oral fluency, which were finally synthesized. Some limitations and recommendations of the study were also included.

linguistic features in case of learning a foreign or second language. According to Derwing et al. (2004), more research on factors influencing perceptions of L2 fluency should be investigated. They were concerned about reliable judgments of fluency, and different judgments between untrained listeners and trained listeners on L2 learners' fluency.
With its importance, this study hopes to make it clear about what oral fluency should be like so the language learner, the language teacher, and the language assessor can think alike. From there, more understanding of the criteria used to rate oral fluency can be reached. This study used a set of ideas about oral fluency which were developed based on the Qualitative aspects of spoken language use-CEFR developed by the Council of Europe (2019). This set has not been synthesized in the previous studies and in this study the authors hope to see if this set of ideas is approved by the participants. Furthermore, this study also attempts to collect more open ideas about oral fluency from the participants. Their perceptions of the oral fluency found in this study can be very significant as the authors would like to report the results to the faculty for consideration of the items in the questionnaire that received agreement and strong agreement used to assess oral fluency and from there, the teachers can design a new set of criteria for assessing oral fluency in the English department.

Theoretical background
With clear and correct perceptions, human can avoid doing things through emotions. With clear and correct perceptions, the human can complete their work at best. So, what is perception? As defined by Ou (2017), perception means the listener perceives something by understanding and raising awareness through sensory information; it is the process of receiving and collecting the action of taking possession. Similarly, Colman (2006, as cited in Khau et al. 2022) considered perception as a process or product of perceiving things via the human senses. According to Cambridge University Press (2022), perception is the ability to see, understand, etc. clearly. With the importance of perception, making all related stakeholders in an area perceive things similarly is indispensable.
As this study also aims to create a clear picture of oral fluency, the authors would also like to define assessment rubrics to prevent fluency from being judged or rated unfairly. According to Reddy and Andrade (2010), rubrics help students understand the targets for their learning and the quality standard required for the completion of a task. Campbell (2005) said rubrics can be also a form of e-assessment which was advocated by the instructors who said that rubrics help them grade more consistently, reliably, and effectively. Andrade and Du (2005) said that the students were found to satisfy with rubrics that are transparent and fair.
More specifically, this study hopes to see how both teachers and students perceive oral fluency so that more suitable criteria can be brought into assessing oral fluency. Thus, what is oral fluency? Fillmore (1979) stated that to talk fluently, a speaker has to be able to talk at length with few pauses and has to be able to fill time allowance with words. Another definition was suggested by Lennon (1990, as cited in Wood, 2010, who said that level of fluency is not from the speaker themselves, but from the listener's part. Similarly, when assessing oral fluency, De Wolf et al. (2017) considered the following aspects as necessity for oral fluency: speech rate, articulation rate, number of pauses per minute, phonation-time ratio, filled pauses, dysfluencies, length of utterance and pause duration. What is more, fluency is when someone is good at what they are doing (Foster, 2020). In many cases, assessors perceived that fluency would be assessed based on native speakers' perception of fluency in L2 speech production (Mora & Valls-Ferrer, 2012;Riggenbach, 1991).

Related studies
As the current study attempts to test how the stakeholders in the EFL context, specifically at Tra Vinh University, Vietnam, perceive oral fluency, the authors would like to read through many of the studies related to the topic. The findings and ideas can help the authors have appropriate research questions and suitable data collection tools for the study.
First, oral fluency can be perceived related to pauses, hesitation, speed rate, and dysfluency. Kormos and Dénes (2004) collected speech samples from 16 Hungarian L2 learners at two levels of proficiency assisted with computer technology. Three experienced native and three non-native teachers were chosen as judges. The source was logical cartoon strips extracted from popular English course books and the speakers were told to narrate in 2-3 min plus a 1-min plan. It uncovered that fluency is best conceived of as fast, smooth, and accurate performance. Interestingly, the number of filled and unfilled pauses and other disfluency phenomena were not seen to affect fluency perceptions. Time needed for preparation should be brought into discussion prior to having the speaker talk and it seems that the listeners see pauses and other disfluency contexts as natural phenomena. In the same way, Sato (2014) used an individual task and an interactional task to test their oral fluency and to test their perceptions of oral fluency criteria. Four native speakers of English in the field were recruited to rate these students' tasks. A questionnaire was used to test their perception of oral fluency and then the author continued to test their oral fluency based on four empirically based oral fluency scales. The scores were given to unpruned speech rate, pruned speech rate, individual perceived fluency, and interactional perceived fluency. The assessors used the verbal protocol method to rate 16 individual tasks and 8 interactional tasks: band 1 (e.g., In individual tasks, the speaker speaks very slowly and haltingly with long pauses-within-word and betweenword, false starts, reformulations, and/or fillers. Each utterance is short, often consisting of a single word) to band 4 (e.g., The speaker produces fairly long stretches of language and each word is produced quickly-within-word. Pauses are noticeable but the number of pauses between-word is small. In regard to individual performance, the assessors perceived pauses as indications of inability to phrase utterances, insufficient time to formulate sentences, or an inability to get things right the first time. This study has provided the field with clear ideas of criteria used to rate fluency and this set of ideas can be used as a reference. Similarly, Préfontaine et al. (2016) had 40 L2 voluntary participants with different levels of proficiency in French for the study. Eleven French native speakers were chosen as raters. In task 1, the participants told a story with six random pictures, allowing creative performance. In task 2, the participants retold a story about a horseback riding accident from a short text in English. In task 3, an 11-cartoon strip was provided and the participants narrated the story with a clear event sequence. All the three speech samples were analyzed using PRATT, tracking utterance fluency temporal variables. The assessors referred to L2 fluency Assessment Grid (Préfontaine et al. (2016)) to rate the speech samples. It found that raters' judgments of fluency according to the descriptors in the grid, speed, and pausing were influenced much similarly to the utterance fluency variables, of which mean length of runs and articulation were found to be the most influential factors in raters' judgments. Again, with clear rating scales, the assessors can do their best and with support from the speech analysis software more evidence can be collected for perception of oral fluency. Van Os et al. (2020) tested the effect of speed rate and delay between questions and answers (various gaps and overlaps) in a dialogue format on fluency judgment for both native and nonnative speakers' answers to questions. It found that more fluent speakers could deliver faster speech while less fluent speakers delivered slow speech. It also uncovered that an interaction effect between speech rate and delay step. In terms of fast speech judgment, overlaps with an interlocutor was rated as less fluent than gaps (a case of native speakers) while regarding slow speech, overlaps were rated as more fluent than gaps (a case of both native and nonnative speakers). Nonetheless, it can be also noted that when speakers are interacting with each other, one can influence the other's ideas, so speed rate produced by involved speakers can be influenced, too. Suzuki et al. (2021) conducted a meta-analysis to examine the relationship between utterance fluency and listener-based judgments of perceived fluency by analyzing primary studies. They analyzed 263 effect sizes from 22 studies to calculate the mean effect sizes of the links between utterance and perceived fluency. Perceived oral fluency was strongly associated with speed and pause frequency, moderately with pause duration, and weakly with repair fluency. Moderator analyses uncovered that the utterance-perceived fluency is affected by methodological variables related to how speech samples are prepared for listeners' judgments and how listeners' attention is directed in evaluations of fluency.
Second, oral fluency can be influenced by text structure and text complexity, and language proficiency. Skehan and Foster (1999) found task structure and processing conditions on narrative retellings influence oral fluency. They used series Mr. Bean as source for narrative tasks. Two tasks were chosen: a relatively structured narrative and a relatively unstructured narrative. To influence the processing load of the task, the two tasks were performed in four conditions from the most demanding to the least demanding. The degree of task structure was found to mainly influenced fluency; conditions of performing the task influenced complexity and task structure; task preparation affected accuracy. This can be true in reality as more demanding task structures can cause the speaker lots of hesitation or pauses for ideas and language choice and time for preparation prior to speak is not less important. Skehan et al. (2016) carried out a study on comparing first and second language fluency during narrative retelling tasks of varying degrees of tightness in structural organization and specifically investigated a distinction between discourse-based and clause-based fluency. The authors utilized four Mr. Bean video excerpts as the source: It started from no tight structure to the tightest structure and the strongest causal links among parts. Twenty-eight English-low intermediate NNSs and 28 NS speakers were the narrators who watched the videos and narrated. They were tested on fluency, structural complexity, and lexis. They found that if speakers (NNS and NS speakers) produced multi-clausal utterances, they tended to pause less often in the four tasks. In terms of complexity measures, only NSs tended to increase reformulation, repetition, and filled pauses, and mid-clause pausing when producing longer clauses. With regard to lexical measures, only NNS showed less 'repair' , generally and slightly increased clause-boundary pausing, producing greater fluency. For lexical sophistication, the NS group experienced less frequent lexical items, associated with more end-clause pausing. Again, text structures and text complexity were brought into discussion. Bui and Huang (2018) employed 58 participants speaking Cantonese with similar experience in studying L2 (reaching B2 English according to CEFR). They were asked to perform two very similar tasks about a discussion of a computer virus and a biological virus. Their tasks were recorded and coded for analysis. It uncovered that knowledge of the topic influenced how well they performed. Topic familiarity was predicted to affect their speed and mid-clause pausing. Nonetheless, it should have employed a prescribed rubric and the performance should be rated by trained raters. Zhang (2009) found the majority of Chinese learners of English were unable to speak fluently as they were not exposed to appropriate input and output during language learning. Moreover, while speaking they were thinking more about vocabulary and grammar, so they were not able to speak fast. Low language proficiency also influenced these learners. Most of them find interaction is unreal and does not facilitate them to speak frequently. This study has persuaded that material selection can have its role. Such idea aligns with Rossiter (2009), who said topic familiarity and linguistic aspects influenced the speed of talks.
Third, similar to many definitions of oral fluency, which consider that fluency can be perceived by the listener. Rossiter (2009) recruited 24 ESL intermediate learners and 6 native experts for the study. The learners included 15 novice native speakers and 15 advanced non-native speakers of English. The speakers were asked to complete a questionnaire on their language experience and an eight-frame narrative description task conducted and audio-taped at time 1 and again after 10 weeks, at time 2. The topic was about a couple moving to the country and finally returned to an easier life in the city. The results showed that the novice native speakers gave the highest fluency ratings to the speech samples, followed by the native expert group and the non-native speakers (respectively). Higher fluency ratings at time 2 were generally higher than those of time 1. It was explained that non-native speakers' judgments were influenced by their ESL teachers' ways, paying more attention to linguistic features. Then, this study can also suggest that the topic familiarity and frequency of topic exposure can influence oral fluency. Moreover, people who have experienced in learning and teaching a foreign language can find themselves more demanding in use of linguistic aspects. Han et al. (2020) explored the relationship between utterance fluency measures and raters' perceived fluency ratings of English/Chinese consecutive interpreting in hope to create, rewrite and modify rubrics and scalar descriptors of fluency scales in interpreting. This study, albeit going for interpretation, can be seen as a good one to see if the scales for rating utterance fluency correspond to the assessor's perception of oral fluency. Muñoz Ocampo (2022) found lack of fluency was identified as the participants' primary problem throughout the diagnostic phase. Additionally, it was shown that they were unable to achieve language mechanisms, but showed interest in daily life matters. The data were gathered from the four instruments-student transcripts, student self-assessment forms, non-participant observer forms, and teacher journals. It uncovered that students might increase their fluency by actively adopting certain communication methods or explicitly articulating them while completing challenging tasks.
Fourth, to help improve oral fluency, instructors may want to modify their instruction by many ways. Vo (2021) looked at the effects of task kinds and motivation on oral L2 fluency development in higher education in Vietnam. Thirty second-year university students and thirteen professors took part in the study. Close-ended questionnaires for students and instructors as well as semi-structured interview questions for instructors were used. It found task success is just being motivated. Additionally, the results show that performances are generally statistically more fluent in dialogue. Van Os et al. (2020) also investigated oral fluency through interaction and which can either positively or negatively influence involved speakers. Guevara-Betancourt and Albuja (2020) conducted a study to identify the variables that affect the growth of oral competence and fluency in undergraduate English major students who are at an intermediate level. A quantitative approach was used to identify and quantify the factors influencing the oral skill development of the English language and thereby comprehending factors that influence fluency during the target language communication process. A descriptive study and an inductive technique were both used to identify and categorize the components. It found important variables, such as language exposure and the importance of contact both inside and outside of the classroom are the influential factors. Nergis (2021) employed an experimental group of 20 students to test against the other 20 control students. The instruction of 10 sessions for both groups used a list of targeted items (formulaic sequences used for the experimental group and academic vocabulary for the comparison one). Each targeted item was then introduced in real-life academic spoken discourse, which attempted to make learners familiar with the context in which these items will be used in real life and make learning of these items more meaningful to them. All repetition and filled pause markers like 'um' and "uhh were eliminated to avoid imitation by learners. To test the effectiveness of the intervention, three tests were used in pretest, posttest, and delayed posttest. It revealed that the two groups enhanced significantly in speed fluency from pretest to posttest and the formulaic sequences group outperformed the control group in pruned speech rate and in the global fluency measure. Effects of formulaic sequences instruction were maintained on delayed posttest. This study has described very specific aspects used to test oral fluency and with appropriate modeling and language input can help learners produce better fluency output.
Overall, many ideas of oral fluency were found. Mainly, hesitation, reformation of speech, pauses, and dysfluency are seen as the most common obstacles. Furthermore, text structures, text complexity, text unfamiliarity and the listener's perceived oral fluency, language exposure, and language instruction are all considered to affect oral fluency. These theories and literature have led the authors to propose the four following research questions in hope to find more ideas on how Vietnamese EFL teachers and students perceive oral fluency.
1. How do senior Vietnamese EFL students perceive oral fluency? 2. How do Vietnamese EFL teachers (BA) perceive oral fluency? 3. How do Vietnamese EFL teachers (Master) perceive oral fluency? 4. How do Vietnamese EFL teachers (Doctor) perceive oral fluency?

Research design
The study employed the mixed method as guided by Edmonds and Kennedy (2016), using both quantitative method and qualitative methods to collect the data of the participants' perceptions on oral fluency; an eighteen closed-ended questionnaire was used to collect both the students' and teachers' perceptions of oral fluency. Then, the study also devised an open-ended question to help the study have more objective ideas on oral fluency from the participants. The rationale for participant selection is based on their experience in learning and or teaching English as a foreign language.

The participants
The total participants of the study are 85, among whom there are 33 English-majored students, 20 English teachers with a bachelor's degree, 22 English teachers with a Master's degree, and 10 English teachers with a doctorate degree. The students are studying at Tra Vinh University. Twenty English teachers with a bachelor's degree are guest-teachers at Tra Vinh University and the other teachers are working at Tra Vinh University. The English-majored students are in their final year of a 4-year English undergraduate program. They already took 6 English-speaking courses in their program and have been exposed to criteria used to rate fluency in the speaking tests. The English teachers with a bachelor's degree have been invited to teach General English at Tra Vinh University for at least 2 years. The English teachers with a Master's degree have taught English to English-majored students for at least 5 years at the English Department, Tra Vinh University. The English teachers with a doctorate degree have taught English to English-majored students at Tra Vinh University for at least 2 years since their doctorate graduation. These teachers have experience in developing speaking criteria and using the speaking rubrics to rate their students' speaking ability. All of these participants' first language is Vietnamese. In addition, the teachers in this study used to be the English-majored students many years ago, so they were seen to have great experience in taking the speaking tests in addition to teaching and rating foreign language speaking ability. All the participants volunteered for the survey using the Google form link for convenient computing in the SPSS.

Materials
The researchers (also having been trained on writing and assessing English writing and speaking tests for 2 months organized by the Vietnam Ministry of Education and Training in 2018) based their oral fluency ideas on the descriptors developed by the Council of Europe (2019), which helps provide description of the Qualitative aspects of spoken language use-CEFR, consisting of six levels of oral fluency from level A1 to level C2. Then, the researchers also considered more information about oral fluency ideas extracted from the IELTS band score descriptors described by Wattie (2015) to make the questionnaire more detailed and appropriate. Usually, the speaking candidate is rated based on the four grading domains: fluency and coherence, pronunciation, grammatical range and accuracy, and lexical resources, respectively.

Instruments
A five-point Likert scale questionnaire was created with 1 for strongly disagree, 2 for disagree, 3 for no idea, 4 for agree, and 5 for strongly agree. The questionnaire comprises 18 questions that focus on what the participants thought about oral fluency. The perception ideas are based on the qualitative aspects of spoken language developed by the Council of Europe (2019) for the Common European Framework of Reference and the fluency scale described by Wattie (2015). All the questions were written in two languages, English and Vietnamese, to ensure correctly understanding the questions. Then, this same questionnaire has one open-ended question at the bottom (question 19) and the participants were also encouraged to answer the open-ended question if they had other different ideas from the described statements about oral fluency already displayed in the questionnaire.

Procedures
After consulting the rubrics for rating oral fluency, the authors began writing up the questionnaire, which was also proofread by the other two senior teachers in the English Department. Then, the authors made some minor revision for language use and content. To collect the data, the researchers first created four Google form links to ensure the data was collected precisely, which were then sent to the four groups of the participants via the Zalo social-networking platform. First, the authors asked for their permission to conduct the survey and told them the purpose of the survey. Then, the authors sent them the links. It took roughly one week to have the responses from the participants. After collecting enough responses, the authors started to count the answers and downloaded them from the Google form and then put them in the SPSS software to measure the reliability of the responses of the four groups. When the responses were safe to be used, the authors continued to count the mean scores and test the significant difference of the four groups' means. Finally, the authors synthesized all the responses to the openended question for more ideas on oral fluency.

Data processing
First, the researchers checked the reliability of the responses from the four groups of the participants. Doing this allows the researchers to decide if the data is reliable enough for either further data analysis or for resurveying.
As seen in Table 1, the reliability of the questionnaire for the 33 senior Englishmajored students reached .767, being reliable enough for use. Next, the reliability of the questionnaire for the 20 English teachers with bachelor's degree is .819, which is reliable enough for further use. Then, the reliability of the questionnaire for 22 English teachers with Master's degree is .645, which is qualified to use. Finally, the reliability of the questionnaire for 10 doctorate teachers is .803, also seen as strongly reliable to use.
Then, the researchers continued to compute the mean scores of the four groups' responses for their perceptions on oral fluency. Finally, the authors synthesized the participants' responses to the open-ended question for new ideas about oral fluency.

Result of the English teachers (BA.) and English-majored students
The result in Table 2 below illustrates how both English teachers (BA) and English majored students perceived oral fluency. The authors would like to describe the result by starting with the statements having the high mean scores first. Therefore, we would like to take a look at scale 4. In terms of the teacher group (with bachelor's degree), they did not show strong agreement with all the statements in the questionnaire. The first category is about the items having the highest mean scores reaching scale 4 or just close to scale 4. Item 4 "Oral fluency happens when a speaker can maintain their thoughts during the speech", Item 18 "Oral fluency happens when a speaker can speak freely albeit some mispronunciation of words", and Item 16 "Oral fluency happens when a speaker can speak freely about a topic with correctly use of vocabulary" received M = 4.20, SD = .61559, M = 4.00, SD = .56195, and M = 3.95, SD = .68633, respectively. The second category has the mean scores between M = 3.40 and M = 3.85, suggesting the students agreed with such items. They belong to items 2, 3, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, and 17. Regarding the student group (senior English-majored students), no items received the mean scores of 4.00 or above, meaning they did not show strong agreement with these described items. However, the first category is about the highest mean scores and three items reached close to scale 4 in this category. Item 4 "oral fluency happens when a speaker can maintain their thoughts during the speech" with M = 3.94, SD = .86384, item 9 "oral fluency happens when a speaker can use a natural colloquial flow" with M = 3.97, SD = .91804, and item 16 "oral fluency happens when a speaker can speak freely about a topic with correct use of vocabulary" with M = 3.90, SD = .80482. These mean scores can suggest the students' strong agreement. The second category is about the items having the mean scores between 3.40 and 3.85. They belong to items 2, 3, 5, 6, 7, 8, 10, 11, 12, 13, 14, 15, 17, and 18. With such mean scores, these items received the students' agreement.
Interestingly, item 1 "oral fluency happens when a speaker can speak about something at length" was not much supported by these two groups (English teachers with M = 3.35, SD = .93330 and English-majored students with M = 3.36, SD = .96236).

Results of the English teachers (MA.) and English teachers (Doctor)
The result in Table 3 below depicts how the English teachers (MA) and English teachers (Doctor) perceived oral fluency.
First, the authors would like to start with the items that have the highest mean scores between 3.95 and 4.55. Looking at the English teachers' perception (BA), the majority of the items achieved strong agreement. The strongest in this category belongs to item 4 "oral fluency happens when a speaker can maintain their thoughts during the speech" with M = 4, 55, SD = .50965. The second strongest means scores are item 12 "oral fluency happens when a speaker can stress the important points during the speech" with M = 4.36, SD = .49237, and item 7 "oral fluency happens when a speaker can express himself/herself spontaneously" with M = 4.32, SD = .77989, suggesting they greatly advocate such descriptions. Other items in this category also    receive strong support from the English teachers (BA). They are items 8, 9, 10, 11, 13, 14, 15, and 16. The second category is about the English teachers' agreement (BA). Items 1, 3, 5, 17, and 18 own the mean scores of 3.69, 3.40, 3.77, 3.55, and 3.50, orderly. Only one item falls within the range from neutral to agreement. That is item 6 "oral fluency happens when a speaker tends not to reformulate their speech" with M = 3.36, SD = .72673.
When looking at the English teachers (Doctor), most of the items reached the mean score of 4 and above. The highest mean scores in this category belong to item 3 "oral fluency happens when a speaker can use sentence connectors well", item 4 "oral fluency happens when a speaker can maintain their thoughts during the speech", which equally shared the mean score of 4.80, SD = .44721, suggesting the English teachers (Doctor) strongly agreed with these two items. The other strong items in this category are also seen. They are item 7 "oral fluency happens when a speaker can express himself/herself spontaneously" with M = 4.60, SD = .54772, item 2 "oral fluency happens when a speaker can make himself/herself clearly understood" and item 8 "oral fluency happens when a speaker can express himself/herself effortlessly", which equally shared the mean score of 4.40. Items 1, 5, 9, 15, and 16 equally shared the mean score of 4.20, meaning they strongly agreed with such descriptions. The rest have the mean score ranging from 3.40 to 4.00, showing that the English teachers (Doctor) quite agreed with such items. Table 4 below is the description of the overall mean scores of the four groups: starting from senior English-majored students, English teachers (BA), English teachers (MA), and English teachers (Doctor).
As can be seen in Table 4, the overall mean scores of the four groups are Englishmajored students (Senior) with M = 3.66, SD = .42568, English teachers (BA) with M = 3.90, SD = .42854, English teachers (MA) with M = 4.00, SD = .29247, and English teachers (Doctor) with M = 4.10, SD = .41076. The mean scores of these four groups were close to one another although the English teacher with a doctorate degree is seen to higher than the rest. In order to see if these mean scores are significantly different, the authors looked for the sig. value as illustrated in Table 5 below. As can be seen, the mean score between groups reached. 711, with the sig value of .005, the mean scores of these four groups show difference, indicating they provided quite different responses to the items displayed in the questionnaire.

Open-ended question
The section below presents the responses to the open-ended question related to oral fluency. The question is "What else do you think about oral fluency? Write short answers below, please. " The authors synthesized the ideas that the participants provided. As the question is not compulsory, the ideas collected are not quite great. The same ideas received from the participants were used once only in the table below.

Closed-ended questions in the questionnaire
In comparison, when looking at the mean scores in Tables 4 and 5, it suggests that the higher degree they obtain the more likely they tend to agree with the items in the questionnaire. As the teachers with a high degree had more eduacational diverse background, they might have exposed to being tested and designing speaking tests during their study and teaching career. As a result, they perceived oral fluency higher.
As can be seen in Tables 2 and 3, the four groups tend to agree with the statements about oral perception although the teachers with a doctorate degree seem agreeing more with the items in the questionnaire compared with other groups. As illustrated, one remarkable item to be considered first is item 1 "oral fluency happens when a speaker can speak about something at length" obtained the lowest mean score in the English teacher group (with bachelor's degree), M = 3.36 and senior student group, M = 3.35. However, in this study, the teacher group (with a Master's degree) and the English teacher group (with a doctorate degree) tend to rate this item higher with M = 3.69, and M = 2.40, respectively.
Next, among the four groups, only the English teachers (MA) tend not to agree with Item 6 "oral fluency happens when a speaker tends not to reformulate their speech". They rated it M = 3.36. It indicated that this group did not see repair speech significantly influenced the speaker's oral fluency while the bachelor senior students and teachers with a doctorate degree perceived this quite higher with M = 3.40 and the teachers with a bachelor degree with M = 3.65. This quite neutral perception aligns with Kormos and Dénes (2004) and Suzuki et al. (2021), who found such phenomenon as a natural speech in reality.
If using 4.00 as the clear-cut score, interestingly, no items received this scale in the English-majored student group. In the English teacher group (BA), only two items were found. Item 4 (maintain their thoughts during the speech) received the mean score of up to 4.20. Item 18 (speak freely albeit some mispronunciation of words) obtained the mean score of 4.00.
In contrast to the two groups mentioned above, in the English teacher group (MA), 11 items were found to attain the mean score of 4.00 or above. They are item 2 "make himself/herself clearly understood" received M = 4.23, item 4 "maintain their thoughts during the speech" obtained M = 4.55, item 7 "express himself/herself spontaneously" obtained M = 4.32, item 8 "express himself/herself effortlessly" received M = 4.14, item 9 (use a natural colloquial flow), item 10 "link sounds together when possible", item 11 (avoid articulating some hesitation words like "umh…, ah….a, etc"), item 12 (stress the important points during the speech), item 13 (compare and contrast ideas during the speech), item 14 (speak freely about a topic with correct grammatical language), item 15 (speak freely about a topic with correct pronunciation), and item 16 "speak freely about a topic with correct use of vocabulary).
In the English teacher group (Doctor), 13 items were obtained. They are item 1 (speak at length), item 2 (make himself/herself clearly understood), item 3 (use sentence connectors well), item 4 (maintain their thoughts during the speech), item 5 (pauses less frequently during the speech), item 7 (express himself/herself spontaneously), item 8 (express himself/herself effortlessly), Item 9 (use a natural colloquial flow), item 11 (avoid articulating some hesitant words like "umh…, ah….a, etc"), item 13 (compare and contrast ideas during the speech), item 15 (speak freely about a topic with correct pronunciation), item 16 (speak freely albeit some misuse of vocabulary), and item 18 (speak freely albeit some mispronunciation of words).
In short, many items were awarded high mean scores, indicating they were approved by the participants, especially by English teachers (MA and Doctor). According to the finding of this study, oral fluency is highly perceived in terms of length and maintaining their thoughts during the speech as defined by Fillmore (1979), and text familiarity is investigated by Bui and Huang (2018) and Zhang (2009). Then, oral fluency is perceived in regard to making oneself clearly understood, expressing oneself spontaneously, pausing less frequently during the speech, expressing oneself effortlessly, linking sounds together when possible, and avoiding articulating some hesitant words. These  Suzuki et al. (2021). Finally, oral fluency is perceived as being able to use sentence connectors well, use a natural colloquial flow, stress the important points during the speech, compare and contrast ideas during the speech, speak freely about a topic with correctly grammatical language, speak freely about a topic with correct pronunciation, and speak freely about a topic with correct use of vocabulary. These perceptions tend to deal with language proficiency and text structure and therefore they quite accord with Skehan and Foster (1999), Skehan et al. (2016), Bui and Huang (2018), and Zhang (2009).

Open-ended question
Many of the responses (in Table 6) to this question were found to assemble the ideas in the close-ended questions in Tables 2 and 3 above. For example, an oral fluent speaker is a person who speaks freely (item 10) and can easily express their ideas (similar to items 2 and 14 in the Table 3). Three new perceptions of oral fluency are found here. The speaker needs to be confident when speaking, but this perception is related to the speaker's psychology, which can deter the speaker's performance. One more idea is that to be considered a fluent speaker, he or she needs to be able to use idioms/phrasal verbs or slang suitably and naturally when speaking. Interestingly, this is quite true when native speakers use such things often in reality, so a foreign language speaker should try to become a native-like speaker. This perception is related to text structure and task demand. But, when the text is increasingly demanding, it requires the speaker' high proficiency of the foreign language or the second language to perform the task. Then, "intonation" is also perceived as one of the criteria used to rate oral fluency; this criterion can be seen to be related the suprasegmental aspect of pronunciation.

Conclusion
The first finding is that this study continues to prove that the criteria required for oral fluency are related to length, making oneself understood, maintaining their thoughts during the speech, pausing less frequently during the speech, trying not to reformulate their speech, speaking effortlessly, using a natural colloquial flow, linking sounds together when possible, avoiding articulating some hesitant words.
The second major finding of this study is that the participants also perceive oral fluency as knowing how to compare and contrast ideas during the speech, using correct language (grammar, vocabulary, and pronunciation), using sentence connectors well, knowing how to stress spontaneously important points. Interestingly, speaking freely with some misuse of vocabulary and mispronunciation of some words during speech is also accepted. Speed rate (speaking without hesitation) and utterance length (within appropriate time) could be factors for assessing how that speaker is fluent in terms of oral ability. Besides, sub-factors such as wide range of lexical resources used and dazzling intonation/accent need considering.
2 Oral fluency needs the confidence because of communicating with others face to face.

3
A fluent speaker is a person who can speak fluently.

4
Oral fluency happens when a speaker who can confidently say what he or she thinks without faltering.

5
I think an oral fluency can speak fluently although they do not have enough vocabulary.
6 Fluent speaker is one who does not need to speak too quickly, but speaks at the right pace, interrupts at the right place, speaks freely.

7
Oral fluency happens when a speaker has confidence while speaking. 8 Fluency is when no words or little vocabulary are available; we can finish the conversation and be less inhibited when we speak.

9
Oral fluency happens when a speaker can express their words with actions to make them understood by the listeners.

10
Oral fluency happens when a speaker can speak freely about all topics.

11
I personally think that oral fluency is described as linguistic competences of speakers.

12
Speaking fluently is not necessarily the right topic.

13
Oral fluency happens when a speaker can speak freely and easy to understand.
14 I think the best way to speak is to speak fluently, with correct grammar, correct pronunciation and correct vocabulary.

15
In my opinion, oral fluency happens when a speaker can make the listener understand what he wants to provide.

16
I think a fluent speaker needs confidence.

17
Oral fluency is someone who can confidently talk about any topic in front of people 18 Oral fluency happens when a speaker can communicate confidently.

19
A fluent speaker is a person of fast reaction, with confidence, and with impression.

20
Oral fluency happens when a speaker can easily express their ideas. = freely 21 Oral fluency means you can apply what you have learned into practice naturally.