| Coherence plays an important role in understanding discourse and communication.Coherence and cohesion are always seen as complementary elements.However,each has its own characteristics.Cohesion,as a lexical term,mainly refers to the connective approaches in discourse form.It is the physical network of discourse and represents the adhesion in the discourse structure.However,coherence indicates mental representation,and it achieves semantic connectivity through logical inference.As an intangible network,it also requires the content of the discourse to form a unified cognitive field with the reader’s world.Measuring coherence is difficult and an urgent challenge for automatic language proficiency assessment as coherence is a mental representation.Current research on coherence focuses mainly on written discourse.However,coherence in oral discourse affects listeners’ comprehension of the input information and effective communication significant.Therefore,this thesis explores textual indices that can best measure coherence in Chinese EFL learners’ oral discourse.Specifically,the following two questions are addressed:1)Which parameters of coherence and cohesion correlate with Chinese EFL learners’ oral English proficiency?2)Which parameters of coherence and cohesion can best differentiate different levels of oral English proficiency?The target oral English database for analysis was collected in an oral English test with61 tertiary students.Each student recorded their oral responses to four questions concerning their everyday life and learning topics.This resulted in a total of 244 audio clips.Based on the online discourse analysis tool Coh-Metrix,this study compared and analyzed a series of parameters to measure the coherence quality of college EFL students’ oral English.A multitude of parameters were examined,such as referential cohesion,latent semantic analysis and connectives,etc.The specific procedure was as follows:firstly,all the audios were scored manually to determine their coherence level as well as their oral English proficiency.Next,all the audios were transcribed into spoken text for Coh-Metrix-based text analysis.Finally,statistical analyses were carried out to test the correlation between different coherence parameters,the predictability of the parameters and their corresponding coherence level and oral proficiency.Group differences in these parameters were also examined to identify the best predictors of coherence differences.1)Correlation analysis results showed that,oral English coherence score correlated positively with three parameters,namely,CRFCWOa,CRFCWO1,and LSASS1.Meanwhile,oral English coherence correlated negatively with another two parameters,namely,CNCTemp and CNCAdd.Oral English scores were positively correlated with four parameters,namely,LSAGN,CNCAll,CNCAdd,and CNCTemp.Oral English scores were negatively correlated with CRFCWOa.2)Regression results revealed that LSAGN and CNCAdd were predictive in students’ performance on oral English scores and oral English coherence scores.3)Between-group variance test on coherence scores showed that EFL learners of high and low coherence levels differed significantly in five parameters,namely,CNCTemp,CNCAdd,CRFCWO1,CRFCWOa,and LSASS1.However,in the between-group variance tests on oral language scores,the results showed that the two levels of learners differed significantly in CRFCWOa,CRFCWO1,LSAGN,CNCAll,CNCTemp and CNCAdd.These results indicate that oral language coherence is mainly related to CRFCWO1,CRFCWOa,LSASS1,LSAGN,CNCAll,CNCAdd and CNCTemp.This study can potentially help teachers and students better evaluate oral English coherence in their future teaching and learning.Moreover,the research is a preliminary exploration of CohMetrix.Future research needs to be designed to explore more systematic evaluation methods and diversified instruments to scientifically assess oral language coherence. |
