You are viewing a free preview of this lesson.
Subscribe to unlock all 10 lessons in this course and every other course on LearningBro.
Choosing the right topic for your language investigation is arguably the most important decision you make in the whole of "Language in Action." A well-chosen topic makes everything downstream easier: data collection is realistic, the analysis is rich, and the 2,000-word write-up almost organises itself. A poorly chosen one leaves you fighting against insufficient data, superficial analysis and mounting frustration. Because the investigation is assessed mainly on AO1, AO2 and AO3 — accurate method and terminology, conceptual understanding, and contextual analysis of meaning — your topic must above all be one that generates genuine linguistic analysis, not merely interesting content. This lesson takes you through selecting an area, narrowing it to a precise focus, and framing a research question or hypothesis you can actually answer in around 2,000 words.
A strong topic balances five qualities. Weakness in any one of them tends to sink the whole project.
| Quality | Why it matters |
|---|---|
| Specificity | A tight focus allows genuine depth within ~2,000 words; a broad survey only allows shallow coverage |
| Feasibility | You must realistically be able to collect enough appropriate data, ethically and in the time available |
| Linguistic richness | The data must reward analysis at several language levels — lexis, grammar, pragmatics, discourse — not just description of content |
| Personal interest | You will live with this for months; sustained curiosity drives sustained quality |
| Theoretical grounding | There must be relevant concepts, frameworks or prior research you can draw on to satisfy AO2 |
Key Definition: Research question — a focused, answerable question that guides your investigation and determines exactly what data you must collect. A strong research question cannot be answered with a flat "yes" or "no"; it demands explanation, comparison and interpretation.
A simple diagnostic: imagine you have collected your data and written your analysis. Ask "so what could I actually conclude?" If the only honest answer is "yes, they do use persuasive language" or "language has changed," the topic is too thin. If instead you can foresee saying something specific — which strategies, how they function, why they vary across two conditions — the topic has analytical potential.
Below are proven areas that reliably produce strong investigations. Each draws on the same theorists and concepts you meet elsewhere in the course — but every broad area must be narrowed to a specific context before it becomes workable. The example questions show that narrowing in action.
A popular area, but one that fails badly if left as "do men and women talk differently?" Narrow it to a defined context and a specific feature.
Workable questions:
Relevant ideas: Lakoff (deficit model), Tannen (difference model), Cameron (gender as social construction), O'Barr and Atkins ("powerless" language), Holmes (hedging as relational). Use these to interpret findings, not as labels to stamp on data.
Highly rewarding if you have ethical access to a child and can record and transcribe their speech. Parental consent is essential.
Workable questions:
Relevant ideas: Chomsky (innate language faculty), Skinner (behaviourist imitation and reinforcement), Bruner (scaffolding / Language Acquisition Support System), Berko's "wug" test (productive morphology), Brown's stages of morpheme acquisition. (Attribute each carefully — do not, for example, credit scaffolding to Chomsky.)
How language exercises, maintains or resists power in a defined setting.
Workable questions:
Relevant ideas: Fairclough (synthetic personalisation; power in and behind discourse), Brown and Levinson (politeness theory; positive and negative face), Grice (cooperative principle and maxims), Goffman (face and footing).
Attitudinal work, often using questionnaires or a matched-guise technique (the same speaker recorded in two varieties, rated by listeners).
Workable questions:
Relevant ideas: Trudgill (overt and covert prestige), Labov (linguistic variables and social stratification; the department-store study), Giles (accommodation theory: convergence and divergence), standard-language ideology.
How language has shifted across a defined period, usually using historical texts or a self-built corpus.
Workable questions:
Relevant ideas: Aitchison (the "damp spoon," "crumbling castle" and "infectious disease" metaphors for attitudes to change), semantic change (amelioration, pejoration, broadening, narrowing), grammaticalisation, standardisation.
Language in digital contexts — messaging, social posts, forums, comment threads.
Workable questions:
Relevant ideas: Crystal (language and the internet; "Netspeak" as a third mode between speech and writing), grapho-phonemic respelling, and pragmatic concepts (implicature, face) applied to written interaction.
Some topics defeat even strong students. Steer clear of the following patterns.
| Weak topic | Why it fails |
|---|---|
| "How do men and women talk differently?" | No context, no specific feature; cannot reach a defensible conclusion |
| "Is text speak ruining English?" | Loaded, prescriptive framing; invites opinion, not analysis |
| "How has English changed over time?" | Vast enough to fill a library; impossible in 2,000 words |
| "Do adverts use persuasive language?" | Answer is obviously "yes"; not a genuine research question |
| "Comparing two poems" | This is literary criticism, not language investigation |
| Anything needing inaccessible or unethical data (covert recordings, private records) | Cannot be collected lawfully or ethically |
Coursework Tip: If your question can be answered "yes" or "no" without analysis, it is not yet a research question. Rework it so it asks how, which, to what extent, or in what way it varies — formulations that force you to analyse rather than merely assert.
It is worth dwelling on why specific topics outperform broad ones, because once you understand the mechanism you can diagnose your own draft question. A broad topic forces breadth at the expense of depth: with only ~2,000 words, "language and gender" can manage no more than a thin survey skating across many features, none analysed properly. A narrow topic permits the opposite — sustained, multi-level analysis of a single phenomenon in a single context, which is exactly what the band descriptors reward under AO1 and AO3.
A strong research question typically has four moving parts, and naming them lets you audit your own:
| Part | Function | Weak example | Strong example |
|---|---|---|---|
| A defined population or context | Bounds what you study so it is collectable | "people" | "members of one amateur football team's group chat" |
| A specific linguistic feature | Gives the analysis something concrete to track | "their language" | "the use of in-group nicknames and banter insults" |
| A relationship or variable | Creates something to find, not just describe | (none) | "how this differs after a win versus a loss" |
| An interpretive angle | Connects to concepts, satisfying AO2 | (none) | "and what it suggests about how banter performs solidarity" |
Run any candidate question through that table. If a column is empty — no defined context, no specific feature, no variable to explore, no conceptual angle — you have found exactly what to fix. A question that fills all four columns almost cannot help but generate analysable data and a genuine finding.
Coursework Tip: A quick test for whether a question is "researchable" is to ask: could two different students, given the same data, reach different defensible conclusions? If yes, the question is genuinely analytical. If the only possible conclusion is the one already implied by the question ("yes, adverts persuade"), it is not.
Topic choice and data-collection method are not separable decisions; a topic is only as good as your ability to gather the data it requires. Before you commit, sketch how you would actually obtain analysable data, because a fascinating question with no collectable data is a dead end. The table maps common topic areas to the method they typically demand, and the access or skill each presupposes — the next lesson develops methodology in full, but you must think about it now, at the choosing stage.
| Topic area | Usual data | Access / skill it presupposes |
|---|---|---|
| Spoken interaction (power, gender, occupation) | Audio recording + transcription | Consenting participants; a setting you can record; transcription skill |
| Child language acquisition | Recordings of one child + transcription | Ethical access to a child; parental consent; patience over weeks |
| Accent/dialect attitudes | Questionnaire or matched-guise | Willing respondents; care to avoid leading questions |
| Language change | A self-built corpus or archive texts | Comparable texts from each period; basic corpus-tool skill |
| Written genre / advertising | Collected published texts | The texts themselves; a clear sampling rule |
| Language and technology | Captured public posts/threads | Public data; anonymisation of users |
The lesson here is to be honest about access before you fall in love with a question. A student who wants to study courtroom power but cannot lawfully record a court should pivot to broadcast tribunal footage or published transcripts; a student who wants child language but has no young child in their life should pivot to a topic whose data they can actually reach. Choosing a feasible topic is not settling for less — it is the precondition of doing the analysis well at all.
Important — ethics shape topic choice, not just method. Some otherwise interesting topics are ruled out at the choosing stage because the data cannot be gathered ethically: anything depending on covert recording of private conversation, on deceiving participants, or on sensitive personal data fails before it begins. Build the requirements of informed consent, anonymity and the right to withdraw into your thinking from the very first topic shortlist, so you never invest weeks in a question you cannot ethically pursue.
To see the whole process in action, follow a student narrowing a realistic shortlist. They begin with five rough interests and pressure-test each.
The student takes idea 3 through the funnel. Broad area: sports broadcasting. Specific context: live radio commentary of one football match. Specific feature: the lexis and syntax used at moments of high tension (the build-up to a goal). Defined data: transcribed extracts of the three highest-tension passages from one match commentary (public broadcast). Research question: "How does the lexis and sentence structure of one radio football commentator change between routine play and the build-up to a goal, and what does this reveal about how spoken commentary constructs excitement?"
Notice what this question now has: a bounded, collectable, ethical dataset (a public broadcast, no private participants); a concrete feature (lexis and syntax under tension); a variable (routine versus high-tension passages); and an interpretive angle (how commentary constructs excitement). It satisfies every column of the anatomy table, and it can plainly be analysed at several language levels — lexis (semantic fields of speed and drama), grammar (shifts to minor sentences and present tense), phonology if the student chooses to note prosody — within 2,000 words.
Subscribe to continue reading
Get full access to this lesson and all 10 lessons in this course.