24
Selected Papers from 2016 PAC & the Twenty-fifth International Symposium on English Teaching 353 25 A Comparison of Fluency and Complexity in Two Different Kinds of Oral Test Witton-Davies Giles National Taiwan University The two commonest ways of eliciting speech in oral tests use either monologue or dialogue. Dialogue may be with an examiner (as in the IELTS test) or with another candidate (as in Cambridge FCE and part 2 of the GEPT Advanced). Timed monologues are used in parts 1 and 3 of the GEPT Advanced and in TOEFL. However, different testing formats may cause variation in the fluency and complexity of speech, as dialogue has sometimes been shown to be more fluent than monologue (Duez, 1985; Tauroza & Allison, 1990). This study investigates these variations in fluency, with 30 students performing dialogue and monologue tasks at different times, while covering the same topics e.g. talking about their high school both as a timed monologue, and in conversation with a classmate. Recordings were transcribed into AS units and further analysed using Praat. Fluency was measured both globally and in it different aspects, covering speech rate, pausing and repair. Measurements were also made of the number and length of clauses and AS units. Dialogues were found to be significantly more fluent than monologues, with faster speech and less pausing, while in the monologues there was greater complexity: speech units and clauses were longer and there was more clausal subordination. The implications for teaching and testing are discussed. INTRODUCTION Aims of This Study Speaking tests of English and other languages commonly elicit spoken samples in two basic ways. The first is though spoken monologues on a given topic or question, recorded either on computer (as in TOEFL) or in the presence of an examiner (as in parts of the GEPT Advanced). The second is through dialogues, either with another candidate (as in the Cambridge Main Suite exams and GEPT Advanced) or with an examiner (as in IELTS). This study investigates monologic and dialogic speech samples, with the aim of finding any differences in fluency and complexity between the two modes. The broader aim is to throw light on any differences in performance that can be expected in different kinds of speaking test. The Use of Monologue and Dialogue in Testing and Research A key factor in the elicitation of spoken data involves the mode of speech, which may be either monologue or dialogue. Although monologic samples predominate in the research literature, it can be argued that dialogue provides a better representation of interactive speech than monologue (Guillot, 1999: 32), and indeed that dialogue represents speech at its most natural. When people speak, they usually speak to another person, and they take turns to speak

A Comparison of Fluency and Complexity in Two Different ... · TOEFL iBT speaking test, ... A Comparison of Fluency and Complexity in Two Different Kinds of Oral ... participants

  • Upload
    lamque

  • View
    215

  • Download
    0

Embed Size (px)

Citation preview

Selected Papers from 2016 PAC & the Twenty-fifth International Symposium on English Teaching

353

25 A Comparison of Fluency and Complexity in Two Different Kinds of Oral Test

Witton-Davies Giles

National Taiwan University

The two commonest ways of eliciting speech in oral tests use either monologue or dialogue. Dialogue may be with an examiner (as in the IELTS test) or with another candidate (as in Cambridge FCE and part 2 of the GEPT Advanced). Timed monologues are used in parts 1 and 3 of the GEPT Advanced and in TOEFL. However, different testing formats may cause variation in the fluency and complexity of speech, as dialogue has sometimes been shown to be more fluent than monologue (Duez, 1985; Tauroza & Allison, 1990). This study investigates these variations in fluency, with 30 students performing dialogue and monologue tasks at different times, while covering the same topics – e.g. talking about their high school both as a timed monologue, and in conversation with a classmate. Recordings were transcribed into AS units and further analysed using Praat. Fluency was measured both globally and in it different aspects, covering speech rate, pausing and repair. Measurements were also made of the number and length of clauses and AS units. Dialogues were found to be significantly more fluent than monologues, with faster speech and less pausing, while in the monologues there was greater complexity: speech units and clauses were longer and there was more clausal subordination. The implications for teaching and testing are discussed.

INTRODUCTION Aims of This Study

Speaking tests of English and other languages commonly elicit spoken samples in two basic ways. The first is though spoken monologues on a given topic or question, recorded either on computer (as in TOEFL) or in the presence of an examiner (as in parts of the GEPT Advanced). The second is through dialogues, either with another candidate (as in the Cambridge Main Suite exams and GEPT Advanced) or with an examiner (as in IELTS). This study investigates monologic and dialogic speech samples, with the aim of finding any differences in fluency and complexity between the two modes. The broader aim is to throw light on any differences in performance that can be expected in different kinds of speaking test. The Use of Monologue and Dialogue in Testing and Research

A key factor in the elicitation of spoken data involves the mode of speech, which may be either monologue or dialogue. Although monologic samples predominate in the research literature, it can be argued that dialogue provides a better representation of interactive speech than monologue (Guillot, 1999: 32), and indeed that dialogue represents speech at its most natural. When people speak, they usually speak to another person, and they take turns to speak

Witton-Davies Giles

354

as they develop a conversation. Van Lier (2004, p. 130) goes so far as to state that language is “inherently a dialogical process.” As far as use of language is concerned, monologue lacks many of the features of speech typical in dialogue, such as turn-taking, interruptions, clarification requests, backchannels, questions and answers. Dialogue is also less predictable than monologue. It is therefore perhaps surprising how little research has been carried out into fluency in dialogue. Pickering and Garrod (2004) suggest it is partly because monologue is easier to analyse, and partly due to a generative linguistics orientation that favours well-formed, de-contextualised samples, rather than the messy data typical of dialogue. Below, I review studies of monologue and dialogue in fluency research. Monologue Tasks

Researchers have generally preferred to use monologic speech for their samples, because it is more convenient to collect and analyse, there being only one speaker in each sample. Narratives are the most common means of eliciting monologues in research, typically based on a picture or film story retelling, but such samples may suffer from a lack of naturalness and authenticity. Other monologue tasks used in research have been more imaginative and varied, including explanatory tasks, balloon debates, and a poster carousel. Explanatory monologue was required in Skehan and Foster (1996): participants were presented with a situation in which they had left the oven on at home, so they needed to explain to friend (a) how to get to their home, (b) how to get inside, and (c) how to turn the oven off. Mehnert (1998) asked participants to leave telephone messages on an answering machine. In one case, they had to give a visitor directions for getting from the airport to their house, and in the other they had to explain why they had failed to appear at a meeting with friends the day before. A “balloon debate”, as used by Foster and Skehan (1999), is a monologue-based decision-making task. Participants, each representing a “bad” historical character, find themselves in a rapidly sinking balloon, and have to defend themselves. Then everyone votes who should be thrown overboard so that the balloon can rise again to safety. Another communicative kind of monologue is elicited in the “poster carousel” activity (Lynch & Maclean, 2000), in which participants speak to those who visit their “booth”, and explain the subject of their poster. This also features the kind of natural task repetition that has been found to improve speaking performance (Bygate, 2001; Bygate & Samuda, 2005).

Dialogue Tasks

As we have seen, monologic tasks predominate in research on fluency and in investigations of task-effects on complexity, accuracy and fluency (CAF), but dialogues or simulated dialogues have also been used in some studies, and are frequently used in L2 speaking tests. These may involve face-to-face contact with a tester (as in the IELTS and GEPT test), or use of a computer interface. The interaction may be spontaneous, or may allow for preparation time, and the questioner’s part may be wholly scripted, guided or open.

In some “interviews”, where the questions are completely scripted and the response time is fixed, what is produced is effectively a series of short monologues. One example is the TOEFL iBT speaking test, as used (in a pilot version) by Iwashita et al. (2008). The study

A Comparison of Fluency and Complexity in Two Different Kinds of Oral Test

355

analysed 250 speaking samples covering five tasks, of which two were independent and three were integrated into a sequence of activities involving four skills. The opening section of the (face-to-face) speaking section of the Advanced General English Proficiency Test in Taiwan also involves questions designed to elicit monologues of fixed duration (“GEPT,” n.d.). Such tasks do not produce dialogues in the full sense, because while they have some of the features of dialogue (more than one speaker, some turn-taking and question-and-answer), they lack others: the questioner speaks less than the candidate, follows a fixed rubric, and cannot react to the candidate’s responses. Response time is also fixed, and the interlocutor is often not physically present.

A more genuine dialogue develops in the “structured interview”, used in some speaking tests such as the OPI (oral proficiency interview), which is a face-to-face test, and was used by Freed (1995) and Ginther et al. (2010). It involves a series of questions and prompts which allow for more flexibility on the part of the examiner and the candidate. The examiner is expected to select and adapt the questions according to the level of the candidate, who can vary the length of different responses according to how much s/he has to say. Meanwhile, interaction between two or three candidates is the basis of section two of the GEPT Advanced speaking test (“GEPT,” n.d.). Here the candidates work together to complete information-gap and discussion tasks.

Other test formats involve dialogue of a less structured kind, with a genuine two-way exchange. One is the third part of the IELTS speaking test (Cambridge ESOL, n.d.). Here questions are guided but not scripted, with examiners wording the questions in their own way, with follow-up questions encouraged that build on previous responses. The main suite of Cambridge ESOL exams also uses multi-candidate formats, where two or three candidates interact during at least some of the sections of the test. Dialogues of this type have only occasionally been used in fluency research. One possible reason is that test materials and recordings have to be kept confidential, making it difficult for researchers to use recordings, except when a test becomes obsolete, as was the case of the ELTS test used by Fulcher (1996). Another, more obvious, reason is the greater amount of time and work required for analysing dialogue.

However, some research has used dialogue tasks instead of, or in addition to, the more usual monologue tasks. Riggenbach (1991) used perhaps the least controlled samples, asking participants to record their own dialogue. No topic was set for these dialogues, in order to keep the samples as unrestricted as possible. Nevertheless this did not necessarily elicit a truly naturalistic sample, since the conversation (recorded as a homework assignment) was not a naturally-occurring one. It is also unclear whether the speech was spontaneous or scripted, as the preparation and recording were done unsupervised.

Derwing et al. (2004) used three tasks covering both dialogue and monologue– a picture story, an elicited monologue on a topic (“the happiest moment in their lives”), and a conversation based on the same topic. The latter could be described as a structured conversation, but was more naturalistic than most such samples as there was interaction between the speakers, the questions and response time were flexible, and the content came wholly from the participants. Skehan and Foster (2005) used a discussion task where

Witton-Davies Giles

356

participants were given a list of people found guilty of different crimes, and discussed these in pairs before deciding on an appropriate sentence for each case. The tasks were performed in normal class time, so can be considered representative of authentic in-class learner language. As the students worked in pairs, the development of the conversations was unpredictable and speech was spontaneous. Bygate (2001), investigating the effects of task repetition, in addition to a film-cartoon re-telling, used a structured interview in which participants were asked to respond to questions about pictures reflecting different aspects of life in Britain. As Bygate emphasises, “Participants were questioned in such a way as to give the impression that the focus of the task was on content rather than expression.” This interview task is similar to the kind used in Cambridge ESOL’s FCE and CPE examinations, and clearly involves both spontaneous speech and relatively spontaneous interaction between the speakers.

From the above examples, we can see that although the tasks and conditions can be varied for each mode, the tendency is for dialogues to be less controlled, and to involve more spontaneous speech, than monologues. Even with preparation time, the development of a dialogue is likely to be less predictable than that of a monologue, with input from a conversational partner adding considerably to this unpredictability. With monologues, on the other hand, the only input comes the instructions provided for the task, allowing for greater control of language and content, and leading to less spontaneous speech. Fluency in Monologue and Dialogue Compared

As has been explained, the majority of fluency studies have investigated monologue samples, despite the advantages in terms of authenticity for using dialogue. An important question is therefore whether there are differences in the fluency of monologue and dialogue.

Monologue has almost always been found to have slower rates of speech than dialogue. Kowal, Wiese, & O’Connell (1983), in a meta-analysis of various studies of L2 oral narratives and interviews, found that speech was more fluent in the interviews than in the story monologues on all measures, including speech-rate (SR), articulation-rate (AR), average-length-of-pause (ALP), mean-length-of-run (MLR), and phonation-time ratio (PTR). Tauroza and Allison (1990) wanted to assess listening difficulty by comparing the speech rates of different kinds of talk. They found that, in words-per-minute, the order of SR, in descending order, was conversation >interview >radio programme >lecture, although interestingly these differences were not significant using syllables per second. Ejzenberg (1997, 2000) found that all participants spoke faster in dialogue than in monologue, although the native speaker interlocutors and the fact that the dialogue tasks were perceived as easier may have contributed to this. Riggenbach (1989), as reported in Bell (2003), also found faster speech rates in dialogue.

Meanwhile, there are also fluency differences between dialogues, even in L1. Informal dialogue features less pausing than formal interviews: Duez (1985) found that French politicians made 50% more pauses, lasting 50% longer, in political speeches than in TV interviews and casual interviews. In casual interviews, while the overall amount of pausing was similar to that in political interviews, there were more pauses within clauses and within phrases.

In a task-based study that directly assessed performance differences between monologue

A Comparison of Fluency and Complexity in Two Different Kinds of Oral Test

357

and dialogue for L2 speakers of Dutch (Michel, 2011), identical tasks were performed in monologic and dialogic versions: the dialogic version consisted of a telephone conversation, while the monologic one involved leaving a message on an answering machine. It was found that interactive tasks produced higher lexical complexity, accuracy and fluency than monologic tasks. Gilabert, Baron and Levkina (2011) also used the same communicative tasks as monologues and dialogues, and found that fluency correlated with proficiency in monologue, but not in dialogue. The actual differences between monologue and dialogue are not the focus of this study, but from the tables provided it appears that speech rate (SR) and pruned speech rate (PSR) were consistently faster in dialogue, confirming previous findings, while filled pausing was higher in dialogue.

There is also limited evidence pointing in the opposite direction. Skehan (2001), in a meta-analysis of various task performances, found that dialogic and interactive tasks were produced less fluently than monologic ones. Bygate (2001) also found that students were more fluent, but used less complex language, on monologic story tasks than in interviews. Nevertheless, it seems that most studies have found dialogue to have faster speech rates, and less pausing, than monologue. More research is needed using different types of dialogue, as it is possible that factors such as the subject matter and formality of the dialogue, together with the relationship between the speakers, may be at least as important as the dialogic mode of speech in influencing fluency. METHODOLOGY Sample and Data Collection

The initial sample of data for this research came from a speaking class at the foreign languages and literature department of National Taiwan University in Taiwan. There were 16 participants in all, with 7 males and 9 females, all of them majoring in English language and literature and studying their first year at university. Two samples of spoken data were elicited, the first involving conversations between pairs of students, where they were asked to discuss a range of topics, while the second consisted of monologues, in which students spoke alone on the same topics for one minute without interruption.

The dialogue tasks were performed in pairs, with two students facing each other, while the teacher (also the researcher) sat to one side. Students were asked to speak to each other about a series of topics and questions, presented to them orally, with the teacher/researcher taking no part in the conversation, apart from providing the questions and topics. Students first spoke about the past, beginning with some recent experiences (yesterday, last weekend, last vacation), and moving on to other past experiences (e.g. a trip they enjoyed or, their high school). After this they were given an issue to discuss, such as the benefits and drawbacks of modern technology, or the quality of Taiwan's TV programmes. Finally they were encouraged to talk about the future – firstly describing their plans for the following evening / weekend / vacation, and secondly speculating on what they would do in a hypothetical situation–e.g if they won the lottery, or they had only a year left to live.

The monologue tasks were similar to the dialogue tasks in that the topics and even the

Witton-Davies Giles

358

questions were the same, but this time students had to speak for a minute on their own, without interruption. In two cases, for the third and fourth task, which usually required more consideration, they were given a minute to prepare what they would say, and they were encouraged to make notes. This mirrors common practice in part 2 of the IETLS test, in the TOEFL IBT, and in certain sections of the GEPT advanced English test, where preparation time is given before candidates speak. Skehan and Foster have found in various studies that preparation time increases fluency, and also the complexity of language used (Foster & Skehan, 1999; P. Skehan & Foster, 1997).

Analysis

Electronic recordings were made of all task performances, using high quality digital recorders, and these were used for further analysis. First of all transcriptions were made, dividing utterances into Analysis of Speech Units (ASU) (Foster, Tonkyn, & Wigglesworth, 2000). This allowed for calculations to be made of the total number of words, units and clauses. As brackets are used to mark filled pauses, words that are repeated or corrected, backchannels and any other interruptions, these could also be deleted to give the number of “pruned” words (Lennon, 1990; Mehnert, 1998). The figure for “complete words” was based on all words spoken, also including filled pauses, but excluded laughter, backchannels and any interruptions by the researcher, while for “pruned words” filled pauses, repetitions and repairs were removed. With these numbers, mean length of clause and unit could be calculated very straightforwardly, as could the percentage of pruned words (as a proportion of total words).

However, time measurements were also necessary, and this was done using Praat software (de Jong & Wempe, 2009). After producing an initial textgrid of each performance, these had to be labeled and adjusted manually so as to accurately reflect all turns, utterances, silent pauses, and filled pauses. This is a time consuming process, but can only be done by hand because currently available computer software cannot yet distinguish between different speakers, between repetitions/ repairs and other words, or between filled pauses (or other noises) and normal words. Praat textgrids, once corrected and labelled, allow for calculation of the numbers and lengths of time for runs and pauses produced by each speaker. Pauses were also marked according to their location. By adding up the total times for each category (in particular the pauses and speech runs of each speaker), the proportions of pause time and speaking time could be calculated, as could speech rates. In this way the fluency of speech in each mode was calculated.

For the purposes of this study, which is to compare the fluency of different speech samples, a global measure of fluency was required. Pruned words per minute is the ideal such measure because it includes and combines the three basic elements of fluency–speed, breakdown and repair fluency (Peter Skehan, 2003; Tavakoli & Skehan, 2005). These refer, respectively, to the rate of speech (words per minute), the amount of hesitations (frequency and duration of silent pauses, filled pauses and repetitions) and repair (false starts and reformulations). Pruned speech refers to the number of words produced per minute of speaking time, including all pauses, but with filled pauses together with repeated and corrected words removed.

A Comparison of Fluency and Complexity in Two Different Kinds of Oral Test

359

As has been argued by de Jong et al. (2012), it is also essential to have single factor measures that only measure one aspect of fluency, so as to avoid the confounding effects of measures which depend on more than one factor, in the way that, for example, SR combines pause time and speech rate. Articulation rate (AR) was used to measure speech rate, as it gives the number of words per minute of speaking time, without including any pause time. For pausing, pause time per 100 words (PT/w) and pause frequency per 100 words were the key measures. Alternative measures such as phonation-time ratio (PTR) and pauses per minute confound pausing and speech rate, as the latter affects the score on these two measures, giving inappropriately high scores for slow speakers (Witton-Davies, 2014). Pause measures used here include both filled and silent pauses, as they seem to play similar roles, and they complement each other, with greater use of one kind of pause often matching reduced use of the other.

Apart from fluency measures, three measures of complexity were made. These were pruned words per clause, pruned words per speech unit, and a subordination index. The first two measure the length, and therefore complexity, of clauses and units, while the third measures the amount of subordination – with the number of subordinate clauses being given as a percentage of speech units. Higher scores reflect greater use of subordination. Analysis of Dialogue

Analysis of dialogue involves certain complications absent from monologue, and in this section I address these problems and describe how they were resolved in the process of transcription and analysis using Praat. Key questions which need to be answered include:

1. How should overlaps in dialogue be dealt with in transcription and analysis? 2. To which speaker should pauses between turns by different speakers be attributed? 3. What words and sounds should be counted as backchannels, and should they be

considered in the same way as other speech, or as a separate category?

Due to limitations of space, I will provide only brief answers to these questions. Overlaps can be represented in Praat by the use of different tiers, one for each speaker. Thus a given speaker's tier shows his/her speech in full, including sections where another person is speaking. Pauses between turns by the same speaker are allocated to that speaker, while pauses between turns from different speakers are marked as “turn pauses”. These are considered in the same way as pauses between clauses, but each speaker is allocated exactly half of the total pause number and time. No attempt is made to determine who is responsible for such pauses. This might not be appropriate in every case, but it does seem to be the only policy that can be applied consistently and still allocate pauses in a reasonable and representative way.

Backchannels are taken to be those interventions which aim simply to provide support, encouragement or some other reaction to what the speaker is saying, rather than aiming to interrupt and start a new utterance. Such interventions may be either lexical-usually single words such as really, OK, or right- or non-lexical, consisting of sounds such as uhun, mm or phew. Backchannels are a feature of fluent conversation, but if they were counted as part of a speaker's output they would tend to lower scores measuring length of run, unit and clause, as

Witton-Davies Giles

360

they would increase the number of one word utterances. They would also magnify the difference between complete and pruned speech, as non verbal backchannels would be included in the former but not in the latter. Backchannels, and the time taken to produce them, are therefore excluded from word counts in the transcripts and from speaking time in Praat.

It should be noted that verbal interventions are only considered backchannels when they occur during another speaker's turn, while those occurring at the end of another speaker's turn are taken as normal utterances, albeit short ones. Non-verbal backchannels, on the other hand, may also occur between another speaker's turns, indicating a speaker's unwillingness to start an utterance of their own. Evidently, the same sound (e.g. mm) may serve as a filled pause or as a backchannel, in different contexts. However, it is in practice easy to distinguish the two. “Mm” is a filled pause when it comes as part of an utterance by the same speaker, but a backchannel when it is produced in isolation during another speaker's turn. As already mentioned, a backchannel may be considered a sign of fluency, while filled pauses usually indicate disfluency. RESULTS

Table 1 shows key results from the data analysis. It presents the individual results, means and standard deviations for a range of scores, firstly on the dialogues and secondly on the monologues. For example, column 1 shows AR scores for dialogue, while column 2 shows AR scores for monologue. The results were analysed and compared using the Wilcoxon Signed Rank test, which is a version of the T-test well-suited to small samples. This instrument shows whether or not the two sets of results are significantly different. All the measures presented in Table 1 had values with significant differences between the two tests, as described below.

A Comparison of Fluency and Complexity in Two Different Kinds of Oral Test

361

As expected, speech rates are in most cases higher in dialogue than monologue. The

difference is greatest for AR (articulation rate: z=-2.79, p≤ 0.01) and SR (speech rate: z=-2.53, p≤ 0.01), the latter combining AR with pause time. For pruned speech rate (PSR), which combines AR, pause time and repair, the results are weaker but still significant (z=-2.07, p≤ 0.05). The most important fluency differences lie in rate (as measured by AR) and in pausing (SR combines articulation rate with pause time), while there is no significant difference in the rate of repair between the two samples. Pause time ( z=-1.71, p≤ 0.05) and pause frequency ( z=-2.74, p≤ 0.01) both show significant differences, with more pausing in the monologues.

Analysis was also made of the differences between mid-clause and between-clause pauses. Mid-clause pauses have been shown to affect fluency more than those occurring between clauses (Foster & Tavakoli, 2009; Peter Skehan & Foster, 2005; Tavakoli, 2011). The only

Witton-Davies Giles

362

significant difference here is for mid-clause pause frequency, this being greater in the monologues (z=–1.96, p≤ 0.05). Between-clause pausing is not significantly different for either mode, thus confirming the greater impact of pauses within clauses on overall fluency. However, between-clause pause time was also found not to be significantly different between monologue and dialogue.

With respect to complexity, the trend is reversed, with greater complexity in the monologues. This applies not only to pruned words per clause (z=-2.53, p≤ 0.01) and words per unit (z=–3.25, p≤ 0.01), but also to the subordination index (z=-2.99, p≤ 0.01). Thus the slower and more hesitant speech in the monologues seems to allow for longer, more complex utterances.

No difference was found for the measure of repair used here – percentage of pruned speech - which reflects the amount of repair, with higher scores showing less repair. Pruned speech is given as a percentage of complete speech, once filled pauses have been removed (as otherwise repair would be confounded with filled pausing). There is therefore no evidence here of differences in the amount of repair in dialogue and monologue.

It may help to understand these results if we consider how measures of fluency correlate with each other. For this, as mentioned in section 2, we need to compare pure measures of rate, pausing and repair, avoiding measures such as speech rate that combine two or more factors. Table 2 shows Spearman Rank Order correlations between the key single measures: AR, pause time/ 100w, pause frequency /100w, and the percentage of pruned speech. Pause time and frequency correlate quite strongly with AR, as well as, predictably, with each other. However, the percentage of pruned speech, a measure of repair, does not correlate with either pausing or AR. This confirms the conclusion of Skehan and Tavakoli (2005) that repair reflects distinct aspects of processing from those lying behind pausing and speech rate. It is therefore perhaps unsurprising that the differences found in speech rate and pausing in the two tests had no equivalent in the measure of repair.

Table 2. Spearman's Rho correlations between key fluency measures

Articulation Rate-AR

Pause Time/ 100w

Pauses/ 100w

% of Pruned speech

Articulation Rate - AR - -0.730 -0.598 0.195

Pause Time/ 100w - 0.762 -0.019

Pauses/ 100w - 0.212

% Pruned speech - DISCUSSION

The results presented in this study confirm that in most, but not all, cases investigated here, dialogue was more fluent than monologue, as measured by measures of both speech rate (AR, SR and PSR) and pausing (both for time and frequency). These differences were statistically significant. This finding comes in spite of the fact that the monologue tasks were performed

A Comparison of Fluency and Complexity in Two Different Kinds of Oral Test

363

after the dialogues, which might lead one to expect greater fluency on the former, due to familiarity with the topics and questions, and to the previous practice with discussing them. Further research is planned where the order of tasks will be reversed, in order to see how this affects the results.

With respect to complexity, in contrast, monologues were more complex in terms both of length of clause/ speech unit and in terms of subordination. The monologues here elicited clearly more complex speech than the dialogues, and this is also obviously an important consideration, especially where the sophistication of learner speech is a priority.

Language testers and teachers should be aware of the differences found in this study, and in particular it should be remembered that lower rates of fluency may be expected in monologue than in more spontaneous conversation. Monologues might nevertheless be preferred to dialogues in language testing for their greater ease of organisation and scoring, and also because of their potential for eliciting greater complexity of speech. On balance, tests such as the GEPT advanced and IELTS, which involve both monologue and dialogue, may be seen as reflecting better everyday language use than tests based solely on monologue. As for fluency research, there is certainly a need for more investigation of fluency in dialogue.

One other testing modality which could be compared to those studied here is dialogue between an examiner and a candidate, as occurs in parts 1 and 3 of the IELTS speaking test. Such dialogues are rather different from those between two candidates, and it would be interesting to see further research comparing fluency and complexity in three different testing formats: monologue; student student dialogue; and examiner-student dialogue. REFERENCES Bell, C. (2003). L2 speech rate in monologic and dialogic activities. Linguagem & Ensino, 6(2),

55–79. Bygate, M. (2001). Effects of task repetition on the structure and control of oral language. In M.

Bygate, M. Swain & P. Skehan (Eds.), Researching pedagogic tasks : Second language learning, teaching, and testing (pp. 23–48). Harlow: Longman.

Bygate, M., & Samuda, V. (2005). Integrative planning through the use of task repetition. In R. Ellis (Ed.), Planning and task performance in a second language (pp. 37–110). Amsterdam ; Philadelphia: John Benjamins.

Cambridge ESOL. (n.d.). IELTS speaking test: public band descriptors. Retrieved September 17, 2013, from https://www.teachers.cambridgeesol.org/ts/exams/academicandprofessional/ielts/speaking

de Jong, N. H., Steinel, M. P., Florijn, A. F., Schoonen, R., & Hulstijn, J. H. (2012). Facets of Speaking Proficiency. Studies in Second Language Acquisition, 34(01), 5–34. http://doi.org/10.1017/S0272263111000489

de Jong, N. H., & Wempe, T. (2009). Praat script to detect syllable nuclei and measure speech rate automatically. Behavior Research Methods, 41(2), 385–390.

Witton-Davies Giles

364

http://doi.org/10.3758/BRM.41.2.385 Derwing, T. M., Rossiter, M. J., Munro, M. J., & Thomson, R. I. (2004). Second language

fluency: Judgments on different tasks. Language Learning, 54(4), 655–679. http://doi.org/10.1111/j.1467-9922.2004.00282.x

Duez, D. (1985). Perception of silent pauses in continuous speech. Language and Speech, 28(4), 377–389. http://doi.org/10.1177/002383098502800403

Ejzenberg, R. (1997). The role of task structure on oral fluency assessment. Presented at the 28th Annual Meeting of TESOL USA, Baltimore: ERIC Document Reproduction Service No. ED 334 238.

Ejzenberg, R. (2000). The juggling act of fluency. In H. Riggenbach (Ed.), Perspectives on fluency (pp. 243–265). Michigan: University of Michigan Press.

Foster, P., & Skehan, P. (1996). The influence of planning and task type on second language performance. Studies in Second Language Acquisition, 18(3), 299. http://doi.org/10.1017/S0272263100015047

Foster, P., & Skehan, P. (1999). The influence of source of planning and focus of planning on task-based performance. Language Teaching Research, 3(3), 215–247. http://doi.org/10.1177/136216889900300303

Foster, P., & Tavakoli, P. (2009). Native speakers and task performance: Comparing effects on complexity, fluency, and lexical diversity. Language Learning, 59(4), 866–896.

Foster, P., Tonkyn, A., & Wigglesworth, G. (2000). Measuring spoken language: A unit for all reasons. Applied Linguistics, 21(3), 354 –375. http://doi.org/10.1093/applin/21.3.354

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

Fulcher, G. (1996). Does thick description lead to smart tests? A data-based approach to rating scale construction. Language Testing, 13(2), 208–238. http://doi.org/10.1177/026553229601300205

GEPT. (n.d.). Retrieved September 17, 2013, from http://www.lttc.ntu.edu.tw/E_LTTC/E_GEPT/Advanced.htm

Gilabert, R., Baron, J., & Levkina, M. (2011). Manipulating task complexity across task types and modes. In P. Robinson (Ed.), Second language task complexity: Researching the cognition hypothesis of language learning and performance (pp. 105–139). Amsterdam ; Philadelphia: John Benjamins.

Ginther, A., Dimova, S., & Yang, R. (2010). Conceptual and empirical relationships between temporal measures of fluency and oral English proficiency with implications for automated scoring. Language Testing, 27(3), 379–399. http://doi.org/10.1177/0265532210364407

Iwashita, N., Brown, A., McNamara, T., & O’Hagan, S. (2008). Assessed levels of second language speaking proficiency: How distinct? Applied Linguistics, 29(1), 24–49. http://doi.org/10.1093/applin/amm017

Kowal, S., Wiese, R., & O’Connell, D. C. (1983). The use of time in storytelling. Language and Speech, 26(4), 377–392. http://doi.org/10.1177/002383098302600405

Lennon, P. (1990). Investigating fluency in EFL: A quantitative approach. Language Learning, 40(3), 387–417. http://doi.org/10.1111/j.1467-1770.1990.tb00669.x

Lynch, T., & Maclean, J. (2000). Exploring the benefits of task repetition and recycling for classroom language learning. Language Teaching Research, 4(3), 221–250.

A Comparison of Fluency and Complexity in Two Different Kinds of Oral Test

365

http://doi.org/10.1177/136216880000400303 Mehnert, U. (1998). The effects of different lengths of time for planning on second language

performance. Studies in Second Language Acquisition, 20(1), 83–108. Michel, M. (2011). Effects of task complexity and interaction on L2 performance. In P.

Robinson (Ed.), Second language task complexity: researching the cognition hypothesis of language learning and performance (pp. 141–173). Amsterdam ; Philadelphia: John Benjamins.

Pickering, M. J., & Garrod, S. (2004). Toward a mechanistic psychology of dialogue. Behavioral and Brain Sciences, 27(2), 169–190. http://doi.org/10.1017/S0140525X04000056

Riggenbach, H. (1989). Nonnative fluency in dialogue versus monologue speech: A Microanalytic approach. (Unpublished PhD thesis). University of California, Los Angeles.

Riggenbach, H. (1991). Toward an understanding of fluency: A microanalysis of nonnative speaker conversations. Discourse Processes, 14(4), 423–41.

Skehan, P. (2001). Tasks and language performance assessment. In M. Bygate, M. Swain & P. Skehan (Eds.), Researching pedagogic tasks : Second language learning, teaching, and testing (pp. 167–185). Harlow: Longman.

Skehan, P. (2003). Task-based instruction. Language Teaching, 36(1), 1–14. http://doi.org/10.1017/S026144480200188X

Skehan, P., & Foster, P. (1997). Task type and task processing conditions as influences on foreign language performance. Language Teaching Research, 1(3), 185–211. http://doi.org/10.1177/136216889700100302

Skehan, P., & Foster, P. (2005). Strategic and online planning: The influence of surprise information and task time on second language performance. In R. Ellis (Ed.), Planning and task performance in a second language (pp. 193–216). John Benjamins.

Tauroza, S., & Allison, D. (1990). Speech rates in British English. Applied Linguistics, 11(1), 90–105. http://doi.org/10.1093/applin/11.1.90

Tavakoli, P. (2011). Pausing patterns: differences between L2 learners and native speakers. ELT Journal, 65(1), 71–79. http://doi.org/10.1093/elt/ccq020

Tavakoli, P., & Skehan, P. (2005). Strategic planning, task structure, and performance testing. In R. Ellis (Ed.), Planning and task performance in a second language (pp. 239–273). John Benjamins.

Van Lier, L. (2004). The ecology and semiotics of language learning: A sociocultural perspective. Boston ; Dordrecht: Norwell, Mass: Kluwer Academic.

Witton-Davies, G. (2014). The study of fluency and its development in monologue and dialogue (Unpublished PhD thesis). Lancaster University, Lancaster, UK.

25th

Englis

h Teacher’s Association of the Republic of China

Epoch-Making in English Language Teaching and Learning:A Special Monograph for Celebration of ETA-ROC’s 25th Anniversary

2016 Anniversary

Selected Papers from 2016 PAC &The Twenty-fifth International Symposium on English Teaching

Edited byLeung Yiu-nam

Selected Papers from 2016 PAC &

the Twenty-fifth International Symposium on English Teaching

「2016泛亞洲國際研討會」 暨

「第二十五屆中華民國英語文教學國際研討會論文集」

理 事 長:梁耀南 副 理 事 長:李幸瑾 執 行 編 輯:梁耀南 助 理 編 輯:李維德 編輯委員會:梁耀南、張繼莊、戴維揚、柯安娜、李思穎 助 理:李燕寶 發 行 人:張富恭 出 版 者:文鶴出版有限公司 地 址:106台北市金山南路二段 200號 8樓 電 話:(02) 2393-4497 傳 真:(02) 2394-3810

總 公 司:106台北市金山南路二段 200號 8樓 TEL: (02) 2393-4497 FAX:(02) 2394-6822/2394-3810

各區業務及門市 北 區:106台北市金山南路二段 200號 8樓

TEL: (02) 2393-4497 FAX: (02) 2394-6822 中 區:407台中市台中港路二段 60-8號 5樓之 6

TEL: (04) 2317-0216 FAX: (04) 2314-0002 南 區:802 高雄市同慶路 88號 2樓

TEL: (07) 2270-888 FAX: (07) 2270-801 網 路 書 店:http://www.crane.com.tw 網路書店客服信箱:[email protected] 法 律 顧 問:蘇家宏律師

光 碟 製 作:茂甲視聽事業有限公司 劃 撥 帳 號:01079261 戶 名:文鶴出版有限公司 出 版 日 期:2016年 11月初版一刷 定 價:500元

I S B N:978-986-147-759-6 書 號:0031394

Ú著作權所有‧翻印必究.非經本公司同意不得轉載、 仿製或局部使用於雜誌、網路或其他書籍、媒體上。

Ú若有缺頁,裝訂錯誤,煩請寄回更換。

Selected Papers from 2016 PAC &

the Twenty-fifth International Symposium on English Teaching

Edited by Yiu-nam Leung

Selected Papers from 2016 PAC &

the Twenty-fifth International Symposium on English Teaching

「2016泛亞洲國際研討會」 暨

「第二十五屆中華民國英語文教學國際研討會論文集」

President: Yiu-nam Leung 理 事 長:梁耀南 Vice President: Hsing-chin Lee 副理事長:李幸瑾 Executive Editor: Yiu-nam Leung 執行編輯:梁耀南 Assistant Editor: Wei-teh Lee 助理編輯:李維德 Editorial Committee: 編輯委員會: Yiu-nam Leung & Kai-chong Cheung 梁耀南、張繼莊 Wei-yang Dai, Johanna Katchen, Sy-ying Lee 戴維揚、柯安娜、李思穎

English Teachers’ Association—Republic of China November 11-13, 2016

中華民國英語文教師學會

中華民國一百零五年十一月十一日至十三日

FROM THE EDITOR We are pleased to present this volume of Selected Papers from 2016

PAC/Twenty-fifth International Symposium on English Teaching in CD-ROM version to our members and colleagues. This year’s conference theme is a thought-provoking one, Epoch Making in English Language Teaching and Learning: Evolution, Innovation, and Revolution. Major topic areas include Historical Surveys, Methodology/Teacher Training, Technology, Research and literature.

We received 120 local and 40 international paper proposals, 15 workshops, 2 panels, and 4 posters this year. Among 24 full papers submitted to us, we chose 20 of them after the recommendation from the reviewers. The blind review committee members are scholars from various areas of the TESOL profession.

The topics of the papers received cover a wide spectrum of issues such as basic language skills, pedagogy, teaching literature, technology, and others.

Also included in this volume are 14 papers contributed by our invited speakers. These papers address issues in historical development of TESOL, basic language skills, teaching strategy, teacher collaboration, assessment, ESP, language policy, and corpus linguistics.

Since turnover time turnover is short, we may not be able to revise or edit all the papers. Mistakes are therefore inevitable. Although we have tried our best to keep each writer’s style, we did make some changes and adjust headings, sub-headings, and tables to conform to the APA citation style.

It is hoped that this volume will not only provide insights for graduate students, teaching professionals, and researchers, but also serve as valuable resources for them.

The production of this volume of selected papers would not have been possible without the assistance from Wei-teh Lee, Su-yi Hwang, and Amber Lee. We also would like to take this opportunity to express our deepest gratitude to the major sponsors for their generous support in the production of this volume: the Ministry of Science and Technology, the Ministry of Education, Bureau of Foreign Trade, Taipei City Government, the Bureau of Education, Taipei City, Takming University of Science and Technology, and Minghsin University of Science and Technology. We would also like to thank the Crane Publishing Company for its continuous support to the publication of the Selected Papers and the Language Training and Testing Center for their sponsorship of the best paper awards. Last but not least, special thanks should go to Professors Wei-yang Dai, Johanna Katchen, Huei-chun Teng, Ching-kang Liu, Yu-ching Chan, Yiu-nam Leung, Hsiu-chieh Chen, Chung-shun Hsia, Tzu-chia Chao, Siao-cing Guo, Yu-li Chen, John Truscott, Shu-yi Hwang, and Jessica Wu (LTTC) for their meticulous efforts spent in reviewing the papers submitted for inclusion in this volume. Yiu-nam Leung (梁耀南) President/Executive Editor October 25, 2016

Table of Contents FROM THE EDITOR

INVITED PAPERS*** 1. Blevins, Wiley, The Importance of Decodable Text in Early

Reading

1

2. Burns, Anne, Teaching Speaking: Towards a Holistic Approach

18

3. Cheng, Winnie (鄭梁慧蓮), Discursive Competence of Professionals in Hong Kong

29

4. Gao, Andy Xuesong (高雪松), Strategizing Language Learning

and Teaching

49

5. Hu, Guangwei (胡光偉), Research on Second Language Learner Strategies: Past, Present, and Future

59

6. Krashen , Stephen, Compelling Reading and Problem-Solving: The Easy Way (And the Only Way) to High Levels of Language,

Literacy and Life Competence

84

7. Kubota, Ryuko, A Critical Examination of Common Beliefs about Language Teaching: From Research Insights to Professional Engagement

93 8. Laurel, Ma. Milagros C., ELT in the Philippines: Evolution, Innovation, Revolution

108

9. Liu, Dilin (劉迪麟), Using Corpora in Language Learning and Teaching: A State of the Art Article

115

10. Liu, Jun (劉駿), The Future of TESOL: Trends, Challenges, and

Opportunities

136

11. O'Sullivan, Barry, Validity: What Is It and Who Is It for ?

157

12. Stroupe, Richmond & Eri Fukuda, Facilitating Teacher Collaboration: The Roles of Teachers, Administrators, and Institutions

176

13. Tono, Yukio, Corpus Linguistics and English Language Teaching: Past, Present and Future

195

14. Zhang, Lawrence Jun (張軍), Developing Students’ Cognitive/Academic Language Proficiency: Genre and Metacognition in Interaction

210

WINNERS OF BEST PAPER AWARD

15. Li, Jen-i (李臻儀), L2 Acquisition of English Accomplishment Subtypes Taiwan College Students

222

16. Takahashi ,Yuka, Relative Clause Constructions as Criterial Features: A Corpus-based Study

236

PAPERS

17. Akutsu, Sumie, Teaching English with a Translation Corpus in

an EFL Context

250

18. Balaz, Allen, The Presentation of Language in Everyday Life 263

19. Busbus, Stephenie O., Teaching Speaking Skills through Packets of Accelerated Christian Education (PACEs)

277

20. Chen, Yi-lin (陳怡伶 ) & Tsai,Yi-Ren (蔡依仁 ), The Flipped

MOOC Global/Local Shakespeare Curriculum for the New Media Age Learners: Translating Verbal Metaphors into Visual Images

288

21. Chen, Yulin (陳育琳), Using IRS with Cooperative Learning in a Speaking and Listening Speaking Class

302

22. Joyce Shao Chin (金韶), Using Assessment as a Catalyst for

Learning: A Case Study of an Adult Learning Context in

Taiwan

314

23. Chou, Andrew (周茂林), Singapore’s Bilingual Policy as a

National Strategy

330

24. Chu, Po-ying (朱珀瑩), Where Have All the Learning

Opportunities Gone?

345

25. Giles, Witton-Davies (朱杰佑), A Comparison of Oral Fluency in

Two Different Kinds of Oral Test

353

26. Huang, Jingyi (黃瀞儀), Sarcasm Use in the Big Bang Theory

Sitcom From the Perspective of Adaptation-Relevance Model

366

27. Huang, Min (黃敏) & Teng, Huei-chun (鄧慧君), A Study of

Implementing Literature Circles in EFL Classroom

377

28. Leung, Yiu-nam (梁耀南), Promoting Literature in an EFL

Context in Taiwan: A Pedagogical Approach to Chinese

American Fictions

393

29. Lu, Gieh-hwa (呂潔樺) & Chang, Pearl (張碧珠), The Keys to

Success for Novice Teachers to Differentiate Instruction in

Taiwanese Senior High Schools

404

30. Ma, Yu-hwa (馬玉華) & Fahn, Sharon Rueih-lirng (范瑞玲),

A Reading Intervention for Technical Students: A Pilot Study

414

31. Matsunaga, Mai, Competencies and Self-Study Activities for

Japanese Elementary School English Teachers

424

32. McMurray, David, Management of Classroom Assistants 436

33. Ockert, David, Technology-enhanced Language Learning:

Motivation and the Brain

443

34. Roebl, K.M., Shiue, Connie & Bragg, Nigel, Language, Culture and Society-Application in EFL Teaching

454

35. Shieh, Wen-yuh (解文玉), Effects of Vocabulary Size and Exact

Repetition on Listening Performance

464

36. Townsend, David James, Peer Assessments of Oral Presentations: Are They Reliable?

478

37. Tyndall, Regan, Speak to Write: How Speaking Aids Academic

Writing

489

38. Ueda, Mami (植田麻実) & Abe, Emika (阿部恵美佳), Japanese University Students’ Attitudes Toward SNSs and English Studying Motivation

501

339. Yeh, Meng-hui (葉孟慧) & Huang, Hung-tzu (黃虹慈), External Expectations and Taiwanese Adolescents’ English Learning Motivation

509 ***All invited papers can be found in Epoch Making in Language Teaching and Learning: A Special Monograph for Celebration of ETA-ROC’s 25th Anniversary and in the CD-ROM.***

Editorial Staffs for the Monograph Series Executive Editor: Yiu-nam Leung (Takming University of Science & Technology) Assistant Editor: Wei-teh Lee (Minghsin University of Science & Technology) Editorial Board: Yu-ching Chan (Jinwen University of Science & Technology), Hsiu-chieh Chen (Tamkang University), Yen-hao Chen (National Taipei University), Yu-li Chen (Lunghua University of Science & Technology), Kai-chong Cheung (Shih Hsin University),Wei-yang Dai (National Taiwan Normal University), Chung-shun Hsia (Minghsin University of Science & Technology), Mei-yu Kao (National Ilan University), Johanna Katchen (National Tsing Hua University), Sy-ying Lee (National Taiwan University of Science & Technology), Yiu-nam Leung (Takming University of Science & Technology), Jerome Su (National Taiwan Normal University), Tai-yuan Tseng (Soochow University), Hsu Wang (Yuan Ze University) Advisory Board: Neil J. Anderson (Brigham Young University) Mary Ann Christison (University of Utah) Winnie Cheng (Hong Kong Polytechic University) Andy Xuesong Gao (University of Hong Kong) William Grabe (Northern Arizona University) Guangwei Hu (National Nanyang University of Technology) Michael Hoey (University of Liverpool) Yan Huang (University of Auckland) Stephen Krashen (University of Southern California) Ryuko Kubota (University of British Columbia) Dilin Liu (University of Alabama) Jun Liu (Stony Brook University) Paul Nation (Victoria University of Wellington) Adrian S. Palmer (University of Utah) Fredricka L. Stoller ((Northern Arizona University) Richmond Stroupe (Soka University) Yukio Tono (Tokyo University of Foreign Languages) Lawrence Jun Zhang (University of Auckland) Assistant: Amber Lee Book Cover Design: Ai-ing Deng & Amber Lee

書號:003-1391

書號 : 003-1394

Epoch Making in English Language Teaching and Learning provides a new perspective and perception of the rich and challenging field of English language teaching. It contains 28 articles pertaining to the applications, explorations, research into the specific methodologies, and issues as well as concerns surrounding the teaching of English on both local and international scales. This monograph is intended to serve not only as a window to contemporary English teaching, but also as a springboard into future research and collaboration in our efforts to provide higher quality education.

Selected Papers from 2016 PAC/The Twenty-fifth

International Symposium on English Teaching, an anthology of papers written by experts and scholars in TESOL at home and abroad, covers a wide spectrum of issues in English language instruction and learning. These issues include basic language skills, pedagogy, teaching literature, technology, teaching methodology and strategy, historical development of TESOL, language policy, corpus linguistics, teacher collaboration, assessment, and others. It is hoped that this volume will not only provide insights for graduate students, teaching professionals, and researchers, but also serve as valuable resources for them.