Define whether the object is a post, file or format
“The meme” can mean one image file, one social-media post, a recurring template or a family of remixes. Your research claim should determine which unit receives the citation. A close reading of political wording needs the particular post that carried it. A study of mutation needs a collection of dated versions connected by explicit comparison criteria.
Describe visible evidence before interpreting it: image content, overlaid wording, layout, watermark, language, crop and conspicuous edits. Give the item a stable research ID. This neutral inventory lets another researcher recognise the object even if the URL later fails, while the analytical section remains free to discuss humour, ideology, intertextuality or reception.
A small alteration may create a materially different version
A translated caption, changed crop, added logo or swapped facial expression can redirect a meme’s meaning. Do not collapse such variants into one bibliographic record. Assign each analysed version its own ID and record the distinguishing features. A file hash identifies byte-for-byte matches, but platforms often resize or recompress an image, so combine technical comparison with a visual version note.
If chronology matters, record the date and time with a time zone wherever the interface provides enough information. Relative labels such as “two days ago” should be converted into a documented observation, without claiming minute-level precision that was never displayed. When a live post changes, preserve the earlier observation and create a later state rather than silently replacing your record.
Separate uploader, remixer, template and creator
The account that posted the item is directly observable; authorship often is not. Use “posted by” when that is all the evidence establishes. Reserve “created by” for supported attribution, such as a credible first-person record, traceable production material or authoritative documentation. Even the oldest result in your search is only the earliest version you verified, not proof of first creation.
A meme may incorporate a press photograph, film frame, artwork, brand element or another person’s illustration. Investigate that underlying material separately and cite an identified source when reliable evidence exists. A person who adds the slogan may have made a creative contribution without controlling rights in the source image. Record unresolved provenance openly instead of filling gaps with a popular origin story.
Preserve enough platform context to avoid a false reading
A meme can quote an opponent, respond to a previous post or rely on a news event that was obvious only at publication. Capture the accompanying caption, reply relationship, relevant hashtags and the limited thread segment needed to understand the item. Define that boundary in the research record. An isolated screenshot should not make quoted hate speech or satire appear to be an unqualified statement by the posting account.
Likes, shares, views and comments are changing measurements. If they matter, record the visible number and observation time; do not report them as timeless reach. Personalised recommendations and ranked search results are observations from a particular account or session, not universal platform states. The creator–context–privacy check for social screenshots provides a fuller capture procedure.
Attribution, copyright and research ethics answer different questions
A complete citation helps readers identify the research object, but it does not itself grant permission to reproduce an image. In the United Kingdom, quotation for criticism or review is subject to statutory conditions including fair dealing and sufficient acknowledgement. Its application depends on the work, purpose, amount and publication setting. Check licences and permissions instead of inferring freedom from widespread sharing.
Ethical concerns can remain even where material is public. Memes may expose private individuals, children, health information, harassment or traumatic events. Consider whether a paraphrased description, cropped detail, pseudonymised account label or controlled-access appendix can answer the research question with less harm. An ethics approval does not settle copyright, and a copyright exception does not settle participant risk.
Write a citation that states the limits of your knowledge
A working caption might read: “Figure 6. Meme version M-031, visible wording ‘…’, posted by [account] on [platform], day month year at time and zone, URL, accessed day month year; researcher screenshot; cropped; creator not verified.” Add a separate credit for a reliably identified source photograph or template. Follow the punctuation and bibliography rules prescribed by your institution.
Use the research ID in the prose and explain which visual or verbal feature supports the claim. For a sequence of remixes, a comparison table is usually clearer than forcing every provenance detail into one caption. Keep account labels, variant names and uncertainty terms consistent across the corpus, codebook and article.
Make selection and preservation part of the method
Before choosing memorable examples, define platforms, dates, languages, queries, hashtags and inclusion criteria. Maintain a search log and account for excluded, deleted and duplicate results. Identical image files in different platform contexts may be duplicates for one study and separate observations for another; state the rule you used.
Store original captures and sensitive contextual data in controlled research storage under a documented retention policy. Produce a separate, minimised publication derivative. If a post disappears, note the status change rather than treating your possession of a copy as permission to redistribute it. Revisit the rights and risk review when a thesis moves into an open repository or an article reaches a wider audience.