Literature Review

The notion of accuracy, and fluency are two dimensions for describing language performance, being used to describe performance by L1 and L2 learners (Pallotti, 2009). They are factors for describing language performance, most frequently used as dependent variables to assess variation with respect to independent variables such as acquisitional level or task features (Pallotti, 2009). Due to the limitation in L2 learners#8217; attentional resources, they put priority in accuracy and fluency while dealing with oral tasks and these two dimensions are linked but compete with each other (Skehan, 1998).

Fluency is considered as a significant component of communicative language ability, which puts emphasis on the moderate and fluent speeches. Much research have done on the fluency on both first and second language learners. Fillmore (1979, p. 256) pointed out four types of fluency in as ”the ability to keep the speech flow going” ”the ability of pack meaning in a coherent fashion into discourse” ”the ability to say something appropriate to situation interlocutor etc. and not be tongue-tied or lost for words” and ”the ability to be witty and imaginative”. Faerch et al. (1984) distinguished three factors in fluency, which were the fluency in semantics, in vocabulary and grammar, and in pronunciation. Sajavarra (1987) considered the language acceptability and coherence are two factors in fluency. According to Wenzhong Zhang (2001, p. 343), ”continuity” ”coherence” and ”acceptability” are three aspects in evaluating the fluency in oral performance of second language learning.

Fluency is multidimensional, in which breakdown fluency, repair fluency, speed fluency are considered as sub-dimensions (Tavakoli amp; Skehan, 2005). Other scholars have more specific sub-dimensions to describe fluency, which are divided into four types, including time index, content index, language index and performance index. (1) The time index, being used by Wengzhong Zhang (2001) in his research, includes speech rate (SR), average length of pause (ALP), phonation/time ratio (PTR), mean length of speech run (MLR) and articulation rate (AR). (2) The content index only have one sub-dimension, which is ratio of reported necessary events (RNE). (3) The language index includes three sub- dimensions: ratio of error-free T-units (REFT), mean length of c-unit after pruning (MLCP), and subordinate-clauses per T-unit (SCT). (4) The performance index is measured by the following three sub-dimensions: repairs per 100 syllables (R100), ratio of reformulation and replacement to total repairs (RRR), and ratio of pruned length to total length (RPL). In general, there are twelve sub-dimensions, which were divided into four main types, being used to measure the fluency of EFL learners#8217; oral language production.

Some of these sub-dimensions, like the time index, are mostly used in the description of fluency. For example, speaking rate (SR) is measured by the number of syllables in each minute. The larger the number is, the more fluently the speaker speaks. These sub-dimensions will be used in this research as effective ways to describe the performance of fluency of EFL learners#8217; impromptu speeches. However, some other index being used in his research, such as language index, are more likely to be used in measuring accuracy instead of fluency. REFT means the ratio of error-free T-units, which refers to the field of accuracy instead of fluency. This sub-dimension takes errors in the oral language production into consideration, because Zhang#8217;s research focused on research of the development of fluency of EFL learners#8217; oral language production. The index used by Zhang is not very clear in the distinguish of accuracy and fluency. So in this research, such sub-dimensions will not be used.

As to accuracy, which is valued in traditional teaching, means the level that speech production fits the target language (Ellis, 2003). Pallotti (2009, p. 592) is the ”simplest and most internally coherent construct, referring to the degree of conformity to certain norms”. Traditionally, scholars focus on the accuracy in the written language, including morphology, syntax and lexicology (Polio, 1997). Wolfe-Quintero et al. (1998, p. 33) acknowledge that ”the purpose of accuracy measures is precisely the comparison with target-like use. Whether that comparison reveals or obscures something about language development in another question”.

As to the measuring method of accuracy, there are two main branches, both of which are targeted at comparing accuracy with complexity. Some scholars take the ratio of error-free T-units to total T-units as the index of accuracy (Gaies, 1980; Mehnert, 1998; Wenzhong Zhang et al., 2001). While Xiaochen Tan et al. (2007) take the ratio of error-free clauses (error-free C-unit) to total clauses (C-unit) to indicate the accuracy in speeches. And in this study, the error-free clauses means clauses without any error in syntax, morphology, word order or malapropism (Foster amp; Skehan, 1996).

These measuring methods of error-free C-units and error-free T-units count different types of errors in general. However, it will be better if the errors in accuracy being divided into two groups: in pronunciation and in grammar. Pronunciation to spoken English is same as to spelling in written English, which influences the effect of expression. In that case, pronunciation errors per 100 syllables (PE100) should be listed as an independent sub-dimension to describe accuracy. The other important factor in accuracy is the ratio of error-free clauses (error-free C-unit) to total clauses (C-unit) to indicate the accuracy in speeches (REFC), such as tense, plural forms and subjunctive mood. And it also should be counted separately.