What Can We Tell from These Temporal Measures ? – Temporal Measures as Indices of Oral Proficiency

Oral English teaching has long been a weak link in the science universities in China, let alone the research on oral English test by quantitative method. Therefore, OEPT in the U.S. sheds enlightening light on the spoken English teaching and researching in China. OEPT (Oral English Proficiency Test) is a spoken English test aimed to assess the oral English proficiency of prospective international teaching assistants in the U.S. In the past few years, temporal variables as indices of oral English proficiency to analyze examinees’ oral speech were explored and studied at a large Mid-western American university. Based on the descriptive statistics of the selected temporal variables, this paper aims to give an interpretation of the figures obtained by OEPT data in order to get enlightening implications on spoken English teaching in China.


OEPT Data
OEPT (Oral English Proficiency Test) is a spoken English test aimed to assess the oral English proficiency of prospective international teaching assistants at a large research-based North American university.The OEPT is a semi-direct test which is conducted on computer.The students' oral performance sound files are stored onto disks and rated by human raters.The test includes ten items: compare and contrast, summarize graph, newspaper headline, pass on information, read aloud, give advice, personal history, telephone message, summarize conversation 1 and 2. Each examinee is rated by two raters on a five-point ordinal scale (2 -6).Examinees pass the test with a score of 5 or above.Those students who are below 5 are required to take a one-semester English course -English001T.The test has eight forms, across which the difficulty of the corresponding items is equivalent.The computer randomly assigns one out of the eight forms to an examinee when he/she logs on the computer.
From Aug. 1st, 2005 to July 31st, 2006, there were 408 international students taking OEPT at a large Mid-western American university, including 102 Chinese students.Among the 102 Chinese students, 7 students scored 3, 62 students scored 4, 33 scored 5 and none scored 6.The passing rate of Chinese students is 32.35%.From Aug.1st, 2006 to July 31st, 2007, there were 435 international students taking part in OEPT, including 99 Chinese students.Among these 99 students, 6 scored 3, 65 scored 4, 28 scored 5, none scored 6.The passing rate is 28.28%.The director of OEPP (Oral English Proficiency Program) instructed her students to make a descriptive statistical analysis of the OEPT data in terms of temporal measures of fluency as indices of oral English proficiency.See Table 1.

Variables and definitions
The variable labels in Table 1 are  Speech rate is defined as syllables per minute, total number of syllables uttered by total length (in seconds) of speech sample multiplied by sixty.(Kormos and Denes, 2004) Articulation rate is defined as total syllables produced in speech sample divided by total time required to produce those syllables multiplied by 60. (Cucchiarini, Helmer & Boves, 2000;Kormos and Denes, 2004) Mean Length of Run (MLR) is defined as total number of syllables /phonemes in speech sample divided by total number of run of speech.(Cucchiarini, Helmer & Boves, 2000;Kormos and Denes, 2004) Phonation time ratio is defined as total time spent speaking divided by total time to produce speech sample, ST/TR.(Cucchiarini, Helmer & Boves, 2000;Kormos and Denes, 2004) As to the definition of filled pauses, here're two definitions of them: Definition 1: A filled pause is a conventional -though non-word -expression used to stall for time during the processing of spontaneous speech.Definition 2: [F]illers are sounds or words that are spoken to fill up gaps in utterances As far as pausing phenomenon is concerned, there have been many scholars who use different terms to interpret pausing phenomena from different perspectives.Kowal and O'Connell (1980) distinguish between filled and silent pauses, stating that the later are associated with "the generation of meaning or a more cognitive aspect of processing" (p.63).Similarly, Sabine and Drommel (1980) classify filled pauses within a group of labeled pauses of dissipation -pauses that are unintended by the speaker and "do not facilitate speech processing".Hieke (1981) puts hesitation phenomena into two broad categories: stalls -which among other phenomena includes silent and filled pauses; and repairsincluding false starts and repeats.Unlike Sabine and Drommel who interpret "filled pauses" as "…unintended by the speaker and do not facilitate speech processing", Olynyk, D'Angeljan et al.(1987) distinguish between silent pauses and filled pauses but propose that the use of filled pauses may actually be a sign of the speakers' fluency and ability to avoid long periods of silence.

Silent Pause Time
From table 1, we can see that there are significant differences between fluent speakers (those who scored 5, especially who scored 6) and influent speakers (those who scored 3 or 4) in terms of temporal measures, which echo with our perception of the characteristic differences between fluency and disfluency.For example, Silent Pause Time increases as speakers vary from fluent to most influent, with native speakers of the shortest mean time of 17.21(seconds), 5.0 -5.5 group of 21.41, 4.0 -4.5 group of 28.10, while 3.0 -3.5 group of the longest time, 39.24.Silent Pause Time is undoubtedly a weighty indicator of fluency.

Filled Pause Time
In Table1, the figures which may be against people's expectation or common sense and on the other hand, which is worth noting, are the Filled Pause Time.Filled pause is usually considered a sign of disfluency, as Sabine and Drommel (1980) classify it as pauses that are unintended by the speaker and "do not facilitate speech processing".However, the figures of Filled Pause Time in Table 1 show that this assumption is untrue.The filled paused time of students who scored 3.0 -3.5 is longer -2.77, while the filled pause time of students who scored 4.0 -4.5 is 2.47, those scored 5.0-5.5 is 2.52.But the longest Filled Pause Time is that of native speakers.This can be explained that filled pause may be caused by at least two reasons: one is a sign of hesitation and disfluency of non-native speakers with low speaking proficiency; another is a strategy used by native speakers to avoid long period of silence, as Olynyk, D'Angeljan et al.(1987) state: " …the use of filled pauses may actually be a sign of the speakers' fluency and ability to avoid long periods of silence."In this case, filled pauses are intended strategy which is an indication of speakers' oral skills.The Filled Pause time of students who scored 5.0 -5.5 is a little bit higher (2.52) than those scored 4.0 -4.5(2.47), which indicate that those who scored 5.0-5.5 are more skilled than those who scored 4.0 -4.5 in applying the skill of filled pause.In addition, native speaker group has the largest Sd (3.32) and Max (12.18), with Min of 0.30, while on the other hand, non-native speakers have smaller Sd (2.81, 2.64, 1.74), with Min of 0. This can be explained that all native speakers know the strategy of using filled pauses in spontaneous speech but the degree of using varies greatly.Non-native speakers are more reserved in using filled pauses, some of them adopt it as a skill to avoid silence, while some of them perceive it as a sign of disfluency and never attempt to use it in spontaneous speech, still some people subconsciously use it because of hesitation or disfluency.In conclusion, filled pause is not necessarily a sign of disfluency, it can be used as a strategy to avoid long period of silence in spontaneous speech.But we need to teach students such strategy and encourage them to use appropriately in speech.

Speech Time and Total Response Time
From the figures of Speech Time and Total Response Time, we can see that it does not mean that the longer speech time, the longer total response time, the more fluent or better.For example, native speakers' speech time ranks third (62.34 against 59.51, 66.49, 73.02) in the comparison group and total response time ranks lowest in the comparison group (82.82 against 101.92, 97.06, 96.95).This can be explained that native speakers' speeches are more concise but rich in information or content, in other words, the speaking efficiency is high, while non-native speakers' speeches are not as effective (essential) and efficient as native speakers.There's lots of redundancy, repetition, empty or inadequate expression of ideas in their speeches.
It's reasonably easy to understand the two ratio comparisons in Table 1, SP /TR (Silent Pause Time/Total Response) and ST/TR (Speech Time/Total Response).Native speakers have the lowest ratio of SP/TR, while the most disfluent speakers have the highest SP/TR, which indicate that disfluent speakers tend to have longer silence pause time.As to ST/TR (or the phonation time ratio), fluent speakers have higher phonation time ratio than less fluent speakers.Compared with total response time, we can see clearly that, even though those who scored 3.0 -3.5 have the longest total response time, their speech time is shortest -59.91, therefore their ST/TR (phonation time ratio) is Low -0.59, while native speakers, even though their total response time is shortest -82.82, their ST/TR (Phonation time ratio) is high -0.75.Since both native speakers and those scored 5.0 -5.5 have the same ST/TR (Phonation time ratio), as we know there're differences between two these groups of people in terms of fluency, therefore ST/TR is not an adequate index of oral proficiency.

MSR, Speech Rate and Articulation Rate
Next we will continue to analyze three more important temporal measures: MSR, Speech Rate and Articulation Rate.Based on the definition and the calculation formula of speech rate and articulation rate on page 2 and 3, we can see from Table 2 that, with the same speech time and filled pause time, speech rate and articulation rate differ in that the calculation of speech rate includes silent pause time while the calculation of articulation rate excludes silent pause time.Therefore speech rate is lower than articulation rate as far as each examinee is concerned.In Table 2, from the comparison of the ratio of the mean (of MSR) of native speakers versus the mean (of MSR) of non -native speakers, with the ratio of the mean of (Speech Rate) of native speakers versus that of non-native speakers, and with the ratio of the mean of (Articulation Rate) versus that of non-native speakers, we can see that MSR is a better index of oral proficiency.For example, as far as articulation rate is concerned, the mean of native speakers is 1.15 times that of group 5.0 -5.5, 1.21 times that of group 4.0 -4.5, 1.31 times of group 3.0 -3.5.As far as speech rate is concerned, the mean of native speakers is 1.18 times that of group 5.0 -5.5, 1.36 times that of group 4.0 -4.5, 1.69 times of group 3.0 -3.5.In contrast, as far as MSR is concerned, the mean of native speakers is 1.49 times that of group 5.0 -5.5, 1.78 times that of group 4.0 -4.5, 2.33 times of group 3.0 -3.5.Why MSR is a better indicator than speech rate or articulation rate can also be explained in the following way: in his Ph.D.Even though the second speaker has a higher speaking rate and phonation time ratio than that of the first speaker, the second speaker has a lower Mean Length of Run, which was caused by more frequent pauses that did not fall at grammatical boundaries.So we can draw conclusion that MSR is a better indicator of oral fluency than Speech Rate or Articulation Rate.

Implications
Based on the above analysis of temporal measures of fluency as indices of oral proficiency, we can get meaningful implications on Spoken English teaching.Temporal measures of Silent Pause Time, Filled Pause Time, Speech Time, Phonation Time Ratio, Mean Syllables per Run, Speech rate and Articulate Rate, etc are effective indices of oral proficiency.MSR is a better index of oral proficiency than Articulation Rate or Speech Rate.Filled pause is not necessarily a sign of disfluency; it can also be a sign of fluent speakers' strategy to avoid long periods of silence.
Therefore language teachers need to demonstrate and impart the features of spoken English and the strategies of making spontaneous speech to their students.Spoken English testing is an assessment of students' oral proficiency, and it's also a reflection of Spoken English Teaching.Based on the examinees' oral speech data and raters' comments on examinees' speech, language teachers and researchers need to find out and analyze the common problems of ESL speakers in spoken English in terms of phonetic sounds, intonation, stress, rhythm, vocabulary and syntax, especially the effective and meaningful expression of ideas.Thus, teachers need to instruct students with knowledge and skills of spoken English and train students to express rich and informative ideas in a more concise and idiomatic way.So spoken language testing need to be combined with and serve for spoken English teaching so as to enable students to improve speaking ability effectively and efficiently.
The author designed and conducted a pilot study on the spoken English teaching in China.The pilot study questionnaire (see appendix) was designed under the instruction of the author's sponsor and instructor in the large Mid-western American University.The questionnaire was directed at the science students who graduated from Chinese colleges and universities with a B.S. or a M.S. in order to get information on the English learning, especially spoken English teaching in China.Based on the data obtained in the pilot study, we can bear out the fact that spoken English teaching has long been ignored in college English teaching.Students' knowledge about English phonetics, rhythm, stress, intonation as well as the differences between written English and spoken English is very limited.They lack a systematic and regular training in spoken English.That's why we can see that the passing rate of Chinese students who took OEPT at the large Mid-western American university is 32.35% (2005.8 -2006.8) and 28.28% (2006.8 -2007.8).Because students lack knowledge of spoken English, for example, they don't know that "filled pause" is a feature of spoken English, the majority of Chinese students never use "filled pause"; instead, they have longer silent pause time.In contrast, native speakers use "filled pause" more often instead of "silent pause".See Table 1, the mean of "Silent Pause Time" of the native speakers is 17.21 while Chinese students are: group (3.0 -3.5) is 39.24, group (4.0 -4.5) is 28.10, and group (5.0 -5.5) is 21.41.The mean of "Filled Pause Time" of the native speakers is 3.27 while Chinese speakers are: group (3.0 -3.5) is 2.77, group (4.0 -4.5) is 2.47, group (5.0 -5.5) is 2.52.And also, it can be explained that because of Chinese students' lack of training in spoken English, even though their response time is longer than native speakers, their MSR (Mean Syllables per Run) is much smaller than that of native speakers.The mean of the "Total Response Time" of native speakers is 82.82 while Chinese speakers are: group (3.0 -3.5) is 101.92, group (4.0 -4.5) is 97.06, group (5.0 -5.5) is 96.95.The mean of MSR of native speakers is 11.43 while Chinese speakers are: group (3.0 -3.5) is 4.90, group (4.0 -4.5) is 6.41, group (5.0 -5.5) is 7.64.So we can come to the conclusion that native speakers' speeches are more informative, effective and efficient while Chinese students' speeches are wordier, recurrent and lack of essence.
In conclusion, the quantitative method of using temporal variables as indices of oral English proficiency is a meaningful and effective way to analyze examinees' oral speech.The method of temporal variables lays statistic foundation for the analysis and research on the test of oral speech.It reveals an examinee's oral proficiency in a more direct, scientific and convincing way.The data obtained by means of this method also sheds light on spoken English teaching.In the 21st century, English teaching in the science colleges and universities in China is attaching a growing importance towards spoken English.Besides written test, there is also spoken English test in the national college English test band 4 and 6.Spoken English is especially tested in TOEFL and IELTS.However, spoken English teaching has long been a weak link in the science colleges and universities in China.With spoken English getting more and more concern in China, this quantitative method of using temporal variables is going to receive growing concern in the research of oral English test in China, which will in turn reveal the problems in spoken English teaching and give directions for improvement.Therefore the researching on spoken English test needs to be launched in China now and the teaching of spoken English needs to be strengthened.

Table 1 .
4. Did your English teacher incorporate any oral activities/ discussions in each Intensive English reading class?In your opinion, how much time of Intensive English reading class in China should be devoted to activities of speaking English?Do you think it is necessary to set up a spoken English class for non-English majors in the universities of science and engineering in mainland China?yes no 36.Do you think the English speaking ability you acquired through college English courses is sufficient for your studying abroad or not?yes no 37.Do you have any experience of text-based Internet chats in English?yes no 38.Do you think text -based Internet chats can help develop oral English fluency or not?yes no 39.Can you list any advantages and disadvantages of text-based Internet chats?Did you have any experience of learning English in autonomous learning mode?Yes No 41.What do you think of learning English in autonomous learning mode?More effective than traditional mode of learning As effective as traditional mode of learning Not as effective as traditional mode of learning Autonomous learning mode should be combined with traditional mode No opinion 42.To develop students' oral ability, do you have any suggestions on college English teaching reforms in mainland China?Descriptive statistics -Selective Temporal Variables

Table 2 .
Mean of NS/Mean of non-native speakers