Abstract
The present study aimed at developing a series of objective criteria for measuring and scoring the oral proficiency of EFL students in moving toward a more objective mode for scoring the oral language proficiency. To achieve this purpose, eighty students from the University of Masjed Soleyman in Iran were selected based on their availability and their successful passing of conversations one, two, and three. Then, their oral proficiencies were rated against a validated and newly-developed checklist. The obtained scores were compared with the group's performance in their previous conversation courses. Result indicated a low correlation between the two groups of scores. It was also proved that the subjective measures were not reliable enough to indicate the students' abilities in terms of oral language proficiency.

Key words: Oral Language Proficiency, Objective Scores, Subjective Scores, Scoring Criteria.

Introduction
Many language tests follow a psychological rather than linguistic theoretical framework, evidenced by the use of a single modality (such as a paper-and-pencil test that ignores spoken and oral comprehension) (Pray, 2005). Most current tests of oral proficiency have the same deficiencies, and many of the measures used by the teachers share the problem of subjectivity. This status is sustained by factors such as large classes, teachers' inadequate command of English, and the lack of easy access to support materials and facilities (Ramanathan, 2008, Sook, 2003). Therefore, due to the complicated nature of this skill, testers and language teachers should make use of reliable analyses for the purpose of objectivity.  
The focus of the present study is on the fact that in university conversation classes there exists no clear-cut checklist or a hard and fast set of criteria for measuring the oral proficiency of students majoring in English. Various types of tests designed and administered- mostly paper and pencil listening tests, student-student, and student-teacher interviews rated without using established criteria- are not suitable to the mode. Therefore, an objective and integrated checklist is needed to measure the students' competence on the basis of their performance. To do so, the researchers appropriately modified the existing checklists to include an important factor, "communication" which is essential for the purpose of assessing levels of oral ability, to help the test designers move from subjective teacher-made tests towards a more standardized testing of oral/aural skills. This checklist was developed as comprehensively as possible so that the researchers were able to take into account most of the required criteria in the tests for measuring oral proficiency. Sample models for developing this checklist were extracted from Farhady, Jafarpur, and Birjandi (2001), Heaton (1990), Hughes (2003), IELTS Testing Center (2000), and Underhill (1987). The most significant criteria considered in the checklist included accent, speed of response, diction, listening comprehension, communication, and fluency to name but a few.

Until now, several studies have been conducted in developing measures for evaluating language learners' oral proficiency. Harris (1968 (suggests a list of criteria for measuring oral skills, which is technically known as "Sample Oral English Rating Sheet". Harris's sample comprises five criteria to be rated: pronunciation, grammar, vocabulary, fluency, and comprehension, each of which includes 5 levels. The proficiency guidelines for speaking were developed in 1982 by the American Council on the Teaching of Foreign Languages (ACTFL) with the purpose of creating a criterion that could be used to identify the foreign language proficiency of speakers ranging from "no knowledge" of EFL to "total mastery" gained through widespread application. The ACTFL guidelines include: superior, advanced (high, mid, low), intermediate (high, mid, low), and novice (high, mid, low) levels.
Also, Underhill (1987) has offered a rating scale for measuring speaking skills. A rating scale, as defined by Underhill includes 1) very limited personal conversation, 2) personal and limited social conversation, 3) basic competence for social and travel use, 4) elementary professional competence, and 5) general proficiency of all familiar and common topics. 

One area of decision-making in rating scales is scoring. Farhady, Jafarpur, and Birjandi (2001) state that depending on the objective of the a test, scoring may be done holistically or discretely; the former refers to an overall impression according to which the interviewee either receives excellent, good, fair, or pass/fail scores. The latter, on the other hand, rates the interviewee's performance separately on scales that relate to accent, structure, vocabulary, fluency, and comprehension. Another crucial work in this realm is a checklist developed by Hughes (2003). The checklist assigns the candidates (interviewees) to a level holistically and rates them on the six-point scale of each of: accent, grammar, vocabulary, fluency, and comprehension. The test is both given and rated by the teacher with no student self-evaluation and self-judgment about their progress.
     However, more recent studies, emphasizing the interactional aspect of language, have focused on learners' awareness of the test procedures. For example, a different view of language assessment, inspired by the idea of Task-Based Instruction (TBI) is casting light on the field of foreign language testing. In task-based language assessment (TBLA) language use is observed in settings that are more realistic and complex than in discrete skills assessments, and which typically require the integration of topical, social and/or pragmatic knowledge along with knowledge of the formal elements of language (Mislevy, Steinberg & Almond, 2002). In another case, Lambert (2003), giving the tests at the end of term to nine classes of between 26-31 first year Japanese university students majoring in electrical and mechanical engineering, predominantly male, upper elementary to pre-intermediate level, concludes that recordings of the student-student interviews would provide a clear justification for the marks awarded and it is also a good idea to give the students a chance to think about what they would say by putting the actual test roles on the Intranet. In the light of the above studies, it should be clearly noted that the current scoring methods applied in Iranian universities are mostly impressionistic, based on experience and lack validity and reliability; the checklist proposed can be utilized as an alternative method in order to obtain objective scores which are true representative of the students’ oral communicative ability. So, the current study could function as a prerequisite to interactional approaches to language testing since its main goal is to suggest a rather valid and reliable checklist as a measurement device for assessing oral proficiency. In other words, the same checklist could be used by both teachers and students in methods such as TBLA, student-student interviews, etc.

Questions of the Study
For the sake of arriving at an objective decision, this study pursued to provide answers to the following questions.
1. Which measure, subjective or objective, provides a more valid and reliable estimate of the oral proficiency of the EFL learners?
2. Is there a meaningful relationship between the subjective and objective sets of scores?

Methodology
Participants
Subjects in the present study were 80 students selected from the students of English Language Teaching at Islamic Azad University of Masjed Soleyman. The rationale for their selection was their availability and the fact that the participants had already passed three conversation courses successfully and they were also taking conversation four at the time of the study. Twenty five percent of the participants were male (n=20) and the rest were female (n=60), ranging from 20 to 27 years old.

Instrument
One instrument utilized in the process of the present study was the proposed checklist including a series of standards and criteria for measuring oral communicative abilities of EFL students on an academic level. Another instrument was the IELTS format of interview (a speaking test) in which the interviewees were asked to answer general and personal questions about their homes and families, jobs, studies, interests, and a range of similar topic areas in about five minutes. The other instrument utilized in the present study was a tape recorder for recording the interviews.

Procedure
In order to validate the newly designed checklist, that is, to determine the extent which the checklist measures what it is supposed to measure, a pilot study was conducted. Ten students were randomly selected and rated using both the new checklist and the one designed by Hughes (2003) to determine the criterion-related validity of the new checklist. The correlation coefficient obtained between the two series was 0.968 indicating that the new checklist was valid.
   By the end of the semester, the subjects were asked to speak for about two minutes on a particular topic for which they were given almost two minutes to think about. All of the selected subjects were interviewed and rated against the new checklist by two raters, first by one of the researchers and then by a bilingual (a native speaker of English who speaks Persian). Each interview session was held in the presence of one of the researchers, the classroom teacher (both as the interviewers) and one of the subjects (as the interviewee). Each interview commenced with a set of simple questions and then proceeded to more challenging ones, and before each session, the subjects were asked to explain and write down brief notes on the sources and textbooks which they had practiced in conversation courses and also the methods applied during the courses and the final examinations. This was done as a warm up activity to decrease the psychological stress and to ensure that the same mode and channel had been used to score the oral proficiency of the subjects in the previous courses. In order to enhance the reliability of the scores, rating activities were carried out first by one of the researchers and then by an inter-rater, and agreement was reached on each student's score.
   All the subjects' scores in conversation courses were collected from the Educational Affairs Department of Masjed Soleyman University, and their average scores were calculated. After gathering the required data, the next step was to rate and score each interview based upon the developed checklist with the aim of attaining more reliable and objective scores. The correlation coefficient determined whether or not there was a possible relationship amongst these series of scores.

Data Analysis
The interviews obtained during the study were assessed through listening to the recordings, and the performance of each interviewee was rated on the basis of the criteria indicated in the developed checklist first by the researchers and next by an inter-rater. After calculating the average of the scores given by two raters, two series of scores were attained-the average scores in interview and the average scores in conversation courses.
   By using the Microsoft Excel software (2003 version) and calculating the variables, the correlation obtained was 0.0045 which indicated that the correlation between the two series of scores was substantially low. This proved the hypothesis of the study that the previous ratings were wholly implemented in a subjective manner compared to the ratings made against the newly developed checklist including the objective criteria.
   Minimum and maximum values were higher for the subjective rating than for the objective ones which might indicate that instructors in conversation courses were more generous, and so these scores do not represent the true oral abilities of the subjects. Also the mean of the students' scores in conversation courses was 15.87 while the corresponding mean score in interview was 11.77. However, the difference between the standard deviation of both groups was not meaningful and indicated that the use of standard criteria for scoring oral proficiency caused the scores of the students to fall off in a similar manner, i.e. the subjects who received higher scores among others by subjective scoring measures also received the higher range of scores by the objective measures although their range of scores lowered meaningfully in objective scoring. The median of the scores in conversation courses was 16 which showed that half of the scores were higher and half of them were lower than 16. The median of the objective scores was 12.       
   Table 1 illustrates the average scores of conversation courses assigned to students by their instructors through traditional subjective means of testing and scoring oral communicative abilities. 70 percent of the scores were in the range of 15 to 20, and the rest of the scores fluctuated between 12 and 15.

Table 1. Descriptive statistics on objective  and subjective measures

Statistical Evaluations	Subjective Scores	Objective Scores
Population	80	80
Min. Value	12.33	8.17
Max. Value	19.33	15.67
Range	7	7.5
Mean	15.87	11.77
Standard Deviation	1.70	1.70
Standard Error	0.19	0.19
Median	16	12
Sum	1,270.1	942.38
Sum of Squares	20,395.27	11,331.72
Variance	2.92	2.92

   Figure 1 represents the average scores given by the two raters to the same groups of students based on the standard criteria listed in the designed checklist (See Appendix for a sample checklist). The distribution of these scores was lower than the course scores with 50 percent of the scores between 12 and 16 and the rest between 8 and 12.

Figure 1 The average scores of subjects based on the standard checklist

  
   This distribution of scores indicating the variation in the students' oral abilities shows that the actual abilities of the students are far below that obtained by their EFL teachers.
   In order to determine the contribution of each scale on the objective scores and the performance of the subjects, the scores in various scales were specified in terms of six scales.. Although the general performance of the subjects was weak, the figure shows the strength and weaknesses of the subjects in different sub skills of the speaking skill.
   The checklist contains 6 scales namely, fluency, comprehension, communication, vocabulary, structure and accent, each of which includes 5 levels of proficiency. The performance of the subjects on each scale was then independently calculated.

Figure 2 The percentages of scales of the checklist contributed to the total score

    

   The performance of the subjects in the areas of comprehension, vocabulary, and structure was fairly better compared to that of fluency, communication, and accent (Figure 2).

Discussion
One point to discuss here is that teachers' scoring the students' oral proficiency subjectively is neither reliable nor valid, and so the given scores cannot present the true ability of the subjects in oral language proficiency. By analyzing oral language proficiency in terms of a number of scales and calculating the learners' ability in terms of their performance on the scales, the researchers could now validly judge the learners' oral language proficiency. The general performance of the subjects, however, was weak in the ratings carried out, but their performance in the individual scales of the checklist was varied. That is, in certain scales they performed successfully but in others they did not.
   Results showed that the performance of the subjects in linguistic components was better than their performance in communicative aspects. Fluency is one of the key factors in assessing the oral language proficiency. Most of the subjects in the present study were hesitant and their oral performances were discontinuous. Another scale on the checklist was comprehension in which the subjects showed a better performance than in the other scales. In most cases they understood the question or/and the gist but were not able enough to manage the discussion. We suggest that the comprehension skill of the subjects should be assigned a higher priority in the development of the English teaching curricula.
   On the scale of communication, the subjects had the weakest performance indicating the greater attention they need to pay to this aspect of their communicative competence. Although the performance of the subjects in vocabulary and grammar scales was better, there were other problems such as lack of complete accuracy that should be considered by the EFL teachers. As for acceptable and intelligible accent, interviewees showed a weak performance in this scale which may be indicative of the EFL learners' ignorance of this part of language.
   Analyses showed that the mean score of the subjects in objective scoring was approximately four points lower than the mean score in subjective scoring. It might be that subjective scoring was implemented based on personal judgments and also the scores were allotted to the overall speaking skill of the subjects, and therefore, the range of scores was high. On the other hand, in the objective scoring, in light of the standards and criteria, the communication skill as a whole was broken into six distinct sub-skills. The scores obtained for each sub-skill were summed up in order to represent the total score given to each subject in terms of their comprehension ability, and so, the range of scores was meaningfully lower. Through objective scoring, weak and strong sub-skills of the subjects' speaking skill can be assessed enabling TEFL teachers to remove the deficiencies and reinforce the stronger points.   

Conclusions
In sum, the point to be taken into account is the lack of attention and application of specific standards to score learners' oral productive skill. In the EFL setting, there are many teachers who score the learners' speaking ability subjectively without applying any criteria and they often show generosity in scoring; consequently, the obtained results will be a series of unreliable and invalid scores which are not truly representative of the learners' actual ability. (However, there may be few language teachers who, after a long time of experience, use their intuition to score the learners’ performances subjectively. They are an exception though.) Therefore, in order to obtain better results including more reliable and objective scores in testing speaking, it is essential to utilize a series of criteria to score oral language proficiency. As Pray (2005) mentions, "Oral-language assessments must measure the essential elements of knowing a language, not just lexical knowledge. This includes the ability to produce new utterances and recombine forms to represent ideas, events, and objects on an abstract level, to produce forms of the language they have never heard before, and to demonstrate mastery over the general functions of language such as syntax, morphology, semantics, and pragmatics" (p.405).                      
   One concerns of teachers is how to prepare reliable tests for measuring oral proficiency of the students and score their performance. To have a more reliable estimate of the students' oral language ability, using a checklist will be very helpful. It will eliminate all those sources that threaten the stability of the test scores. The checklist can act as a blueprint to teachers who wish to assess their students' oral proficiency. It reminds them of the macro-skills as well as the specifications or micro-skills that should be included in testing oral proficiency.

Delimitations
Despite the promising results, this study suffered from a few problems. One shortcoming was related to our population which was predominantly female; the results of the present study, therefore, might not be generalizable to the male population. Moreover, speaking skills, though emphasized, are overshadowed by other skills due to lack of environment to adequately practice or apply oral/aural skills. This results in a series of problems especially while conducting the interviews during the research.  

References

American Council on the Teaching of Foreign Languages. (1982). ACTFL provisional proficiency guidelines. Yonkers: ACTFL. 

Farhady, H., Jafarpur, A., & Birjandi, P. (2001). Testing language skill: From theory to practice (9th ed.).Tehran: SAMT Publication.

Harris, D. P. (1986). Testing English as a second language. New York: McGraw Hill.

Heaton, J. B. (1990). Writing English language tests (3rd ed.). New York: Longman.

Hughes, A. (2003). Testing for language teachers (2nd ed.). Cambridge: Cambridge University Press.

Lambert, I. (2003). Recording speaking tests for oral assessment. The Internet TESL Journal, IX(4). Retrieved 23 March 2008 from http://iteslj.org/Articles/Lambert-SpeakingTests.html.

Llosa, L. (2007). Validating a standards-based classroom assessment of English proficiency: A multitrait-multimethod approach. Language Testing, 24(4), 489–515.

Mislevy, R.J., Steinberg , L.S., Almond, R.G. (2002). Design and analysis in task-based language assessment. Language Testing, 19(4), 477-496.

Pray, L. (2005). How well do commonly used language instruments measure English oral-language proficiency? Bilingual Research Journal, 29(2), 387-409.

Ramanathan, H. (2008). Testing of English in India: A developing concept. Language Testing, 25(1), 111-126.

Sook, K. H. (2003). The types of speaking assessment tasks used by Korean Junior Secondary school English teachers. Asian EFL Journal, Retrieved 23 March 2008 from
www.asian-efl-journal/dec_03_sub.gl.htm.

Underhill, N. (1987). Testing spoken language: A handbook of oral testing  techniques. Cambridge: Cambridge University Press.