Researchers who use English as a second or foreign language have a wider range of options than ever before. You can choose to run your paper through AI-powered software or send it to a large ESL editing company, where it may receive a layer of machine edits before being worked on by one or more human editors.
The current explosion of AI and AI-human-hybrid forms of editing is remarkable and positive in many ways. As an ESL editor, I have received a fair number of papers (from a large ESL editing firm) that have previously undergone an initial layer of machine edits, designed to correct basic spelling and grammar mistakes. This is what I have learned so far.
What do machine editors do?
Machine editors are trained to recognise and fix grammar and spelling mistakes. Like the spellcheck on your computer, they are pretty good at correcting common typos (seperate/separate) and grammatical mistakes (affect/effect). Also like your spellcheck, machine editors register patterns, not meaning. In my experience, they fix some mistakes some of the time and introduce at least an equivalent number of new ones.
These seem to be the most common problems:
- Machine editors are a lot like the auto-correct function on your phone. They’ve been trained on spelling and grammar rules, but have no sense of context. Although specialised academic software won’t introduce slang or swearwords, it will replace words it doesn’t recognise (whether technical or misspelled) with more commonplace words that look similar (e.g. substituting ‘perception’ for ‘proprioception’).
- Machine editors change sentence structure in ways that alter the author’s meaning, sometimes significantly. For example, if the author has left out a key noun or verb, the software will try to make a sentence out of versions of the remaining words.
- Machine editors mix up verb tenses. Academic authors talk about things that happened in the past, recent past, present, and future (e.g., prior studies, current research, future directions). Machine editors will try to achieve consistency by making the tenses match.
- Machine editors create plausible-looking phrases and sentences that do not actually convey the author’s intended meaning. In a hybrid process (where a human editor corrects the machine’s mistakes), these subtle changes of meaning and plausible-looking mistakes can be easily overlooked.
- Machine editors muffle an author’s natural voice by changing clear and direct non-native expressions into plausible-but-wrong or confusing/inaccurate language, while also introducing new (uncharacteristic) mistakes.
- Machine editors look for common mistakes made by native English speakers. They are less well-equipped to deal with ESL authors, who use terms and expressions drawn from a wide range of languages.
How do human editors work?
As a human editor, I follow the thread of meaning through a document. I generally spend the first page or so getting used to the author’s voice and fixing superficial mistakes. After that, the editing process tends to have ‘flow’ — I know who the author is and where s/he is going. If I can’t figure out a particular word or phrase, I write the author a note, so that we can discuss it later.
Seaham Academic Editing 