Features
Part of Social media slang: what the experienced already know, 2027 edition
The short version of social media slang research
Researching social media slang means collecting other people's words: three ways to gather, how to strip identity at capture, and the permission questions.
Studying online language means collecting other people's words. That is the whole method, and it is also the whole ethical problem, because the people whose words you collect wrote them for a room and not for a study. Most guidance treats collection as a technical matter and permission as a formality. It is the other way round.
This page covers building a small collection you are actually allowed to keep: what to gather, what to strip, what to ask, and what the result can and cannot answer.
What to take away
- Decide the question before the collection. A pile of text gathered without a question answers none of them.
- Strip identity at the moment of capture, not later. Anything you keep, you may leak.
- Public is not the same as published, and a service's terms are not the same as a community's consent.
- A small, well-described collection beats a large, undocumented one for every question worth asking.
Three ways to gather, with different costs
| Method | What it gives | What it costs | Where it fails |
|---|---|---|---|
| Observation of open material | Natural use, no reactivity | Nothing to the participants, if handled carefully | Skewed to whatever is public and indexed |
| Asking speakers directly | Explanations, register, boundaries | Their time, and a duty of care about what you publish | People report the form they think is correct, not the one they use |
| Elicitation, giving prompts | Comparable material across people | Artificiality | The prompt shapes the answer, sometimes entirely |
The middle row is where the most useful material is and where the most care is owed. A speaker can tell you what a word signals, which no corpus can, and they can also tell you things that would identify them, which you then have to decide what to do with. The safe default is to record the explanation and not the explainer.
The first row has the opposite profile: no interaction, and a badly skewed sample. What is publicly reachable systematically excludes private conversation, closed groups, and everything already deleted, which is where most vocabulary is born. What that does to any conclusion is worked through under internet culture archives.
Reactivity, and why it is not avoidable
The moment people know their language is being studied, they attend to it, and attention changes it. Speakers standardize, explain, and perform. This is the observer effect in a setting where the observed party can also read what you write about them, which adds a second loop the physical sciences do not have.
You cannot remove it. You can account for it:
- Note whether the material was produced with the study in view, and keep the two kinds apart.
- Expect explanatory material to be more standard than ordinary use, and treat it as a description of a norm rather than a sample of behavior.
- Never treat what people say about their usage as a measurement of their usage. It is a measurement of their beliefs about it, which is a different and also interesting thing.
Stripping identity at capture
The rule that prevents the most harm is the simplest: do not record who said it unless you need it, and you almost never need it.
- Record the form, the sense, the register and the kind of community. Not the account, not the display name, not the exact wording of anything long enough to search for.
- Paraphrase rather than quote where the point survives paraphrase, which is most of the time.
- Do not store the link. A link is an identifier with extra steps, and it is the thing most likely to end up in a published draft.
- Keep the collection off shared storage, and do not publish the raw material.
The reason for the last two is that a collection of quotes with sources is a re-identification tool whether you intended one or not. What happens to an ordinary person who becomes findable is described in general under critiques of social media, and nothing about a research purpose changes the outcome for them.
Permission, in three separate questions
Does the service allow it? A terms-of-service question. It binds you legally and says nothing about ethics.
Does the community allow it? Ask, in the ordinary way, of whoever moderates. The answer is informative even when it is no, and asking costs you a message. The framework everyone borrows here is informed consent, and the honest observation is that it fits individual interviews well and open observation badly.
Should you, given what publishing would do? The question that does the real work. A finding that would expose a small community to attention it did not seek is a finding you can hold without publishing. The effects of sudden attention on a small group are described under internet communities, and they fall on people who had no part in your decision.
What a small collection can answer
More than people expect, provided the question is scoped.
- Whether a form exists in a given community at a given time. Yes or no, and the yes is solid.
- What a form is used to do: mark tone, mark membership, soften a request, close a topic.
- What it collides with. Confusion pairs come out of small collections easily and matter to readers.
- Whether a sense has more than one reading among people who all belong.
What it cannot answer is how common anything is, in that community or anywhere. Frequency requires a defined population and a defined denominator, and neither exists here, for reasons set out under benefits of social media, which meets the same wall from the direction of counting attention.
Common questions
Is public material free to quote?
Legally that varies and this page cannot advise you. Practically, a quotable line is a searchable line, and searching it leads to a person. Paraphrase unless the exact wording is the finding.
Is a small collection large enough to be worth building?
There is no floor for a descriptive claim about existence and use. There is no ceiling that makes a frequency claim safe. The size question only matters for the claims you cannot make anyway.
Should I tell the community what I found?
If you can do it without singling anyone out, yes. It is the one part of the exchange that gives something back, and their corrections are usually better than your reading.
What if a moderator says no?
Then do not collect there. The material is not scarce, and a study that overrides a refusal has already answered the ethical question badly.
