This study investigated the effect of a sequence of pedagogical interventions on the level of textual cohesion in the writing of high-intermediate L2 English learners in a college-level ESL program. Eight sections of a fourth-semester ESL writing course were assigned randomly to the experimental or control groups. The experimental group received no additional instructional time, but the researcher visited the each experimental section for one hour each week over a five-week period to provide a series of pedagogical interventions focused on the use of adverbial connectors, determiner + summary noun constructions, and definitional elements. After attrition, data from n = 46 control participants and n = 47 experimental participants were included in the study, for a total of N= 93 participants. Each participant contributed three samples of timed writing in a pretest, posttest, delayed posttest design. The texts were rated by three raters, and the mean rater score was used to operationalize writing quality. Additional developmental measurements focused on the fluency and syntactic complexity exhibited within texts and the amount of lexical diversity. The level of cohesion in the texts was operationalized as a combination of sentence and paragraph latent semantic analysis scores as well as measures of adverbial connector use. The results suggested an effect of treatment. In terms of writing quality, the experimental group scored significantly higher than the control group at posttest, and also produced more and more varied forms of the target structures. The timing and patterns of the effect of instruction measures, combined with the lack of group differences in broad developmental measures, suggest that the intervention sequence did have a positive effect on experimental participant writing. The results also point to the difficulties of operationalizing lexical cohesion as a construct independent of overall lexical proficiency.The results of a principal component analysis on the measures of cohesion suggested that cohesion must be operationalized as a multidimensional concept comprising measures of connector use and lexical reference chains. The analysis also suggested that, if latent semantic analysis measures are chosen as operationalization of lexical cohesion, the level of lexical diversity in the text as measured by type-token ratio, will affect the results of the analysis due to an inverse relationship between latent semantic analysis scores and lexical diversity.
Read
