Pragmatic Features of AI‑Assisted Academic Essays Among Senior High School Students: A Discourse Analysis

Allocation at minting

The anchored record

Canonical JSON (keys sorted, no spaces, UTF-8). Its SHA-256 is the hash on chain.

{"abstract":"This study examines the pragmatic features present in AI-assisted academic essays written by senior high school students in Tagum City, Davao del Norte. Anchored in Hyland’s (2005) stance–engagement model and grounded in discourse pragmatics, the study investigates how linguistic resources shape meaning and communicative intent in AI-supported student writing. Employing a qualitative discourse-analytic design, the study analyzed a purposively selected corpus of 20 AI-assisted essays. The analysis was operationalized through systematic coding and interpretive examination of pragmatic markers, including hedges, boosters, modality markers, discourse markers, and speech acts. Findings indicate that modality markers such as can, should, and must were most frequently used, reflecting students’ emphasis on ethical and procedural considerations in AI use. Hedges, including may and might, revealed cautious stance-taking, while boosters occurred minimally, suggesting limited assertiveness in student arguments. Discourse markers particularly demonstrated a reliance on linear cause-and-effect reasoning. Speech act analysis further showed the predominance of assertives and directives, forming an expository–advisory rhetorical structure that explains AI’s role and promotes responsible practices. Overall, AI-assisted essays exhibit a balanced yet constrained pragmatic profile characterized by a moderated stance, ethical framing, and limited rhetorical variation. These findings highlight both novice writing tendencies and the linguistic defaults of AI tools. The study underscores the need for explicit instruction in argumentation, stance-taking, cohesive device variation, and ethical AI integration in academic writing.","abstract_source":"original","authors":["Daisy Jean Aranton","Marry Whil Montederamos","Joshua Pelenio","Ivan Barroga","Danilo Baradillo"],"date":"2026-06-09","doi":"10.70838/pemj.570807","id":8072,"issue":8,"journal":{"issn":"2822-4353","name":"Psychology and Education: A Multidisciplinary Journal"},"pdf_sha256":"c5420770e68e8b759750fbbc645ee261259888abe200317456bb39bd0b47d32b","schema":"scimatic-article/1","title":"Pragmatic Features of AI‑Assisted Academic Essays Among Senior High School Students: A Discourse Analysis","volume":57}

Check it yourself

1. The record, from the token's own metadata:

printf '%s' "$(curl -s https://scimatic.org/nft/articles/8072.json | jq -r .canonical_record)" | sha256sum

2. The hash stored on chain (contentHash(8072)):

curl -s https://rpc.scimatic.net -H 'content-type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"eth_call","params":[{"to":"0x0E596B4bb924937aA35D9b342A90347ed6F05eF8","data":"0xf3e0c2900000000000000000000000000000000000000000000000000000000000001f88"},"latest"]}'

The two must be equal (the second with a 0x in front).