Get started
Academic English Prose Diagnosis

Academic English Prose Diagnosis

Reviewers say your English “isn't quite natural” but don't tell you what to fix. This agent identifies the specific patterns that signal non-native English to editors: article usage, connector overuse, calqued phrasing, tense inconsistency. For each, you get a sentence from your text alongside what a native academic writer would write, plus the rule to apply yourself. Capped at five patterns per run. Run again after revising. Won't manufacture problems if your prose is already fluent.
#Research#Writing#Education
Rating
More ratings needed
Sold
3
How to use
Run on Capafy
Also on external apps
Publisher provides
Claude Sonnet 4.6

"Isn't quite natural" is not a note you can really use. It sounds like a verdict on you, not a problem in the text, which is what makes it hard to act on: there is nothing to revise, only something to worry about. This agent turns that verdict back into what it actually is, a list of specific, describable choices in your prose that a reader is reacting to, whether or not either of you has named them.

What it does

Reads your text and names the systematic patterns behind that verdict, the specific and describable choices an editor's ear is reacting to. For each one, a before-and-after using your own sentences: what you wrote, alongside what native academic prose would do instead. Then the rule, so you can apply it yourself across the rest of the text.

What the comparison looks like

What you wrote: "the analysis of four types' systems of body representation shows that the causality between the belief and the behavior is general phenomena within all human societies"

What native academic prose does: "analysis of four body-representation systems shows that causality between belief and behavior is a general feature of human social life"

Why: Abstract nouns used as general categories (causality, belief, behavior) do not take "the." "General phenomena" has two problems: "phenomena" is plural so "is" should be "are," and the right word here is "feature" or "pattern," not "phenomena."

The rule: For abstract nouns used as general categories, not specific instances already introduced in the text, no "the."

Patterns it diagnoses

Up to five patterns per run, prioritised by impact. The full set the diagnostic looks for:

Article errors (the, a, an, or no article, the most common non-native signal). Sentence-initial connector overuse (Moreover, Furthermore, In addition). Countable and uncountable noun errors (informations, evidences, researches). Calque phrasing (direct translations from your first language that are grammatically possible but not idiomatic). Tense inconsistency in literature reviews. Passive displacement (impersonal passive where first person is standard). Nominalization chains (verbs buried in noun phrases). Hedging calibration (over-hedged or under-hedged claims). Coordination overuse (long chains of "and" and "but" where subordination would be clearer).

Follow-up runs

Tell the agent which patterns from the previous diagnosis you have addressed. It verifies the fixes are complete and surfaces the next tier of patterns. The agent works iteratively as you revise.

Text length

Works on qualitative academic texts of 500 words or more: full articles, chapters, dissertation sections, or substantial excerpts. Not built for abstracts alone, for those, use the First-Draft Abstract with Approximate Word Count agent under my profile. The diagnostic needs enough text for patterns to become visible.

What it does not do

Rewrite the text. Correct anything, even in the sample comparison, the model is a template, not a fix, and the revision is yours to do. Diagnose argument or structure (for that, use the developmental-edit agent).