Research Integrity and Responsible Practice

Draft for approval · ARCTIS-08 · Version 0.1
Prepared 19 October 2026 · Approval and effective dates: pending

This policy proposes common expectations for responsible research, teaching resources and dissemination within ARCTIS.

1 Reliable research

Research questions, data selection, methods and limitations must be documented. Fabrication, falsification, plagiarism and invented references are unacceptable. Exploratory results and small samples must not support exaggerated claims.

Sources and bibliographic details are checked against the original material. Corrections are made transparently when errors affect a public output.

2 People and data

Before collecting or using information about people, the coordinator establishes the required institutional reviews, permissions, privacy information and approved storage arrangements. Public availability of material does not automatically make every research use appropriate.

Access is limited to authorised people. Retention, deletion, anonymisation and sharing follow the relevant University arrangements and project requirements. The Centre does not act as an ethics committee or replace the University data-protection function.

3 Corpora and software

Teams record the origin, access conditions and licences of corpora, recordings, translations and software. Restricted credentials are not shared. Downloads and redistribution must remain within the licence.

Membership may support access to available resources, including Sketch Engine, subject to the actual institutional licence and authorised account arrangements. No right to transfer access to third parties is created.

4 AI assistance

Users remain accountable for AI-assisted work. Verify factual statements, citations, translations and analytical outputs. Record tool or model details and substantive assistance where required by the project, institution or publisher.

Do not submit confidential, restricted or identifiable participant material to an external AI service without appropriate authorisation and safeguards. AI output is not evidence that a source exists or a result is correct.

5 Contributions and conflicts

Authorship reflects actual scholarly contributions and responsibility. Discuss roles at the start and review them before submission. Membership, seniority or routine assistance alone does not guarantee authorship.

Disclose relevant conflicts of interest. Research review, participant assessment and journal editorial decisions must not be represented as guaranteed benefits of membership.

6 Dissemination

Check copyright and publication permissions before releasing texts, datasets, photographs or recordings. A website-wide licence does not override third-party rights or a contributor’s agreed terms.

State whether an output is a draft, preprint, abstract, full paper, dataset or teaching resource. Programme participation, attendance and institutional partnerships are distinct claims and require distinct evidence.

7 Concerns and corrections

Concerns may be raised confidentially with the project coordinator, Director or relevant University channel. Where the concern involves the Director, use the appropriate University route. Preserve relevant records and avoid prejudging the outcome.

A public correction should identify the affected output and the change. Serious matters follow the University’s established procedures rather than an improvised Centre disciplinary process.

This policy is a draft for approval and is reviewed annually.