Informally, people might associate more “advanced” writing with longer sentences & “fancier” words. However, a recent large-scale study invites us to rethink this reflex.
The study
The researchers analysed 410 text-based argumentative essays written by linguistically diverse secondary students in the US, the majority of whom were multilingual writers. Students read a non-fiction text, then wrote a timed argumentative interpretation drawing on evidence from that text.
Writing quality was judged holistically by expert raters using a 6-point scale. The researchers then examined three broad dimensions of writing:
- Lexical sophistication (eg use of academic vocabulary, age of acquisition of words)
- Syntactic complexity (eg complex sentences, noun phrase complexity)
- Rhetorical features, including cohesion, source integration & the balance between summary, evidence & commentary
Methodologically, the study is notable for combining human ratings, manual linguistic coding & NLP tools (don’t worry, that’s Natural Language Processing & not the pseudoscientific Neuro Linguistic Programming), then modelling the relationships using confirmatory factor analysis & structural equation modelling.
The findings
At a broad level, all three dimensions were related to writing quality. However, rhetorical features stood out as the most closely associated, especially how effectively students used sources and how well they balanced summary, evidence and commentary.
However, once the researchers modelled everything together, a striking pattern emerged. Only rhetorical features directly predicted overall writing quality. Lexical sophistication & syntactic complexity did not independently predict quality, even though they were strongly correlated with rhetorical competence.
In other words, vocabulary range & sentence complexity mattered mainly because they supported rhetorical work. They were the tools, not the outcome.
To make this concrete, compare two writers. One uses academic lexis & long noun phrases but mostly retells the source text. Another uses more modest language yet clearly frames a claim, selects evidence strategically & explains why that evidence matters. This study suggests that raters consistently value the second writer more highly.
In terms of theory, the findings align with genre-based approaches to writing (Swales, Halliday) & classic distinctions between knowledge-telling & knowledge-transforming (Bereiter & Scardamalia). Writing quality emerges not from linguistic decoration, but from how language is mobilised to do rhetorical work.
Teacher Takeaways?
- Teach language through rhetorical purpose, not in isolation. Instead of teaching relative clauses or nominalisations as “advanced forms”, frame them in terms of what they help writers do. For example, show how a relative clause can narrow a claim (“Policies that target adult learners are more effective”) or how nominalisation helps package an idea to evaluate it (“This assumption is problematic because…”).
- In source-based tasks, prioritise evidence selection & commentary before surface-level sophistication. Before pushing students to “sound more academic”, check whether they can choose relevant evidence and explain its significance. A short, clear sentence that introduces a quote and explains why it matters is more effective than a dense paragraph that mostly retells the source.
- Be cautious with feedback that focuses only on ‘more complex language’. Feedback that targets function is more productive, for example: “Can you explain how this example supports your claim?” or “What does the reader need to understand at this point?”
For ELT teachers, this research is a useful reminder that advanced writing is less about sounding academic & more about thinking rhetorically with language.
How explicitly do you foreground rhetoric when teaching writing in your own context?
Leave a Reply