Prompting open-source and commercial language models for grammatical error correction of English learner text