Rahbar، Amin

Exploring Linguistic Features of Argumentative Writings by Iranian Advanced EFL Learners and Their Relation to Human Judgments of Essay Quality

AbstractThis study explores the use of automated linguistic features in English ForeignLanguage EFL argumentative writings Coh Metrix an advanced computationaltext analysis tool was used to analyze 87 linguistic properties of argumentativeessays written by thirty six advanced EFL students track their development overa semester long writing course and examine the degree to which high and lowquality essays can be predicted by objective linguistic measures of cohesion syntactic complexity and lexical diversity The multidimensional analysesdemonstrated that significant growth in the three macro linguistic features in EFLlearners occurred as a function of time spent in the advanced writing class Such agrowth however did not match human holistic judgments of writing quality andfeatures showed growth were not successful at predicting group membership aslow and high quality essays Using regression analysis this study found that threesets of automated measures e g connectives syntactic pattern density and wordinformation were strong predictors of holistic ratings The predictors used in eachsuccessful model support the idea that essays containing complex syntacticstructures e g adverbial complexity and diverse vocabulary e g less concreteand less meaningful words are generally perceived to be of high quality Furthermore the results of this study lend weight to the idea that coherent textshave more explicit cohesive features e g connective devices at least in EFL vi