Ten Cohesion Markers’ Relative Contributions to Coherence

The present study contends that the notion of cohesion influencing coherence is an oversimplification. The present study examined the relative strengths of 10 cohesion markers as predictors of coherence. EFL children (N = 130) wrote both letters and stories, which were analyzed using repeated-measures MANOVA and path analysis. The number of cohesion markers used in the two writing tasks was significantly different for some types of markers while similar for other types, indicating that the amount of cohesion marker use can be task specific. Frequently-occurring cohesion markers were not necessarily stronger contributors to coherence. Only ellipsis, collocation, and conjunction showed a significant influence on coherence for both tasks, and their magnitudes were fairly weak. The significant cohesion markers, collectively, explained only about 10% of coherence. For both tasks, conjunction exerted a negative influence on coherence. This study supports the ideas in the literature that cohesion should not be the primary factor in determining writing quality; coherence is the result of a psychological process within the reader, who understands the text by incorporating its textual cues through association to existing knowledge by inference.