Communication Research Questions

It names the case and the analytic unit. "Which semiotic choices in language and visual communication mainly appear in the memes?" says what will be coded and bounds the corpus to one account's posts, which is the descriptive groundwork any interpretive question has to stand on. Social media is not a data source. One account, over a stated period, in a stated language, is. And if the question compares two things, say what the comparison is for: asking how a set of TikTok micro-trends both reflects and differs from a film genre obliges the findings to do both jobs rather than a vague compare.

No, and it's the most common overreach in this field. A content or discourse analysis can tell you how something was portrayed. It can't tell you what an audience took from it unless you went and asked an audience. If you want both, write two questions rather than one blurred sentence about impact: one asking how public relations is portrayed in news coverage, a second asking what journalists themselves make of the industry. Keeping the two apart is what lets a single study claim both.

Not from platform data alone. Asking to what extent images containing diverse skin tones can be associated with the numbers of likes, shares or comments is the honest wording, and can be associated with is doing careful work: it declines to say one caused the other. That is the promise you can keep for data you collected rather than assigned. If you need the causal claim, you need a design that manipulates the feature, which is a different study.

You can, and it's rarer in this field than students expect, which is part of why it stands out. Ask how a speaker's accent affects their evaluation when subtitle errors occur, and the manipulated feature, the outcome being judged and the condition the test runs under are all readable from the sentence alone. The cost is recruitment plus a stimulus set you have to build and pilot. If you can meet both, an experiment answers a causal question that no amount of content analysis will.

Usually yes, even when every post is already public. The UK research councils' guidance on internet-mediated research is blunt about it: public posting is not consent, the public nature of online material should be critically examined rather than assumed, and identities should be protected unless disclosure is critical to the study. Boards still differ on how much paperwork that means, and a corpus bounded to one official account or one news incident is an easier case than named private individuals. Ask yours early, because the answer decides whether your data source exists.

Fewer than you think, once the boundary is defensible. The corpora behind this kind of work tend to be finite: one account over a stated period, one lifestyle magazine, the coverage of a single incident, rather than a sample of everything ever posted. Work through what falls inside the boundary and report the total. Your examiner will usually ask how you drew the line, not whether you reached a round number.

If you are reporting counts or percentages, expect to be asked for agreement between coders on a subset of the material. If the analysis is interpretive, discourse or framing work, the expectation shifts toward showing your coding decisions and your own position rather than a reliability score. Settle which one your department wants before you start coding, because adding a second coder later means training someone on a scheme you have already finished using.

More often than students expect, and plenty of completed studies rest on it: practitioners reflecting on a cancellation, municipal communicators, journalists describing how they see the PR industry. What earns a yes is a short, specific ask with a stated time commitment, sent to a named person rather than a general inbox. Plan for weeks of silence and a second group you could reach instead, so one unanswered email cannot end the project.

Design the study around what you can collect by hand. A lot of communication corpora are small enough to gather that way: one account's memes, a set of trend videos, the articles covering one event. That caps how much you can collect, so name the cap in your limitations and keep every claim inside what the collected set can support.