AQA A-Level English Language: How to Approach the Language Investigation

The Non-Exam Assessment is where AQA A-Level English Language stops being something you read about and becomes something you do. Instead of analysing data chosen by an examiner, you collect and analyse your own, applying the linguistic frameworks you have studied to language out in the world. Done well, the NEA is one of the most satisfying parts of the course -- and a genuine chance to bank marks before you reach the exam hall.

This guide focuses on the Language Investigation, the larger of the two NEA tasks, covering how to choose a workable research question, collect and handle data responsibly, apply the language levels, draw cautious conclusions and structure the write-up.

What the NEA Involves

The NEA component is called Language in Action. The key facts are:

• It is worth 100 marks and 20% of the A-Level.
• The total word count is 3,500 words.
• It is teacher-assessed and moderated by AQA.

That total is split across two tasks:

• A Language Investigation of around 2,000 words (excluding the data itself), in which you research a question of your own choosing.
• A piece of Original Writing with an accompanying commentary, together totalling 1,500 words.

The rest of the qualification is examined: Paper 1 and Paper 2 are each 2 hours 30 minutes, worth 100 marks and 40% of the A-Level. The NEA's 20% is the portion you can shape most directly, so it pays to plan it carefully.

Choosing a Focused Research Question

The single most common cause of a disappointing investigation is a question that is too broad. "How do men and women speak differently?" is unanswerable in 2,000 words; "How do female contestants and male contestants use hedging in a single episode of a quiz show?" is something you can actually investigate. A good question is narrow, specific, and answerable with the data you can realistically gather.

AQA allows you to investigate any of the language areas you have studied. Common, productive areas include:

• Language and gender -- differences or similarities in how speakers of different genders use language.
• Language and power -- how status and authority are constructed in interaction (interviews, classroom talk, courtroom exchanges).
• Language and region (accent and dialect) -- regional variation in lexis, grammar or phonological features.
• Language change -- comparing texts on the same topic from different periods, or tracking a feature over time.
• Child language acquisition -- analysing a transcript of a child's speech against developmental stages and theory.
• Language and representation -- how a group, place or issue is portrayed in a particular text type, such as news reporting or advertising.

Phrase your question so it points to specific, countable or describable features. Replace "Is the language sexist?" with "How are men and women referred to in the headlines of two newspapers?" The second tells you exactly what to look for.

Narrowing a Question, Step by Step

It helps to see how a vague idea becomes a researchable question through successive narrowing:

Stage	Example question	Problem (or fix)
Initial idea	"How does language show power?"	Far too broad; no data source or feature named
Narrowed by context	"How is power shown in TV interviews?"	Better, but still too much data and no feature
Narrowed by feature	"How do interruptions show power in one political interview?"	Specific feature, but a single text limits comparison
Final question	"How do interruptions and topic control differ between the interviewer and interviewee in two political interviews?"	Focused, comparative, and tied to nameable features

Each step adds focus: pin down the context (which texts), then the feature (what you will actually count or describe), then build in comparison where it strengthens the analysis. By the final row you know precisely what data you need and what you will look for in it.

Data Collection and Methodology

Your methodology section explains what data you gathered and how, and -- just as importantly -- why those choices suit your question. A central distinction underpins it.

• Quantitative data is countable: how many times a feature occurs, how many words, what proportion of utterances contain a hedge. It supports comparison and lets you spot patterns, but numbers alone do not explain meaning.
• Qualitative data is interpretive: close analysis of how language is used in context, the connotations of particular choices, the dynamics of a specific exchange. It offers depth but can be harder to generalise from.

The strongest investigations usually combine the two: counting features to establish a pattern, then analysing examples closely to interpret it. Whatever you choose, make your sampling deliberate. Decide how much data you need, where it comes from, and over what timeframe, and justify those decisions. Be honest about limitations too -- a small sample, a single source or a one-off recording all constrain how far your findings can be generalised, and acknowledging this is a mark of methodological awareness, not weakness.

Research Ethics

Because you are working with real language and, often, real people, ethics matter -- and examiners credit candidates who handle them properly.

• Informed consent -- if you record or transcribe people, they should know they are being studied and agree to it. For participants who cannot consent themselves (for example, a young child), consent must come from a parent or guardian.
• Anonymisation -- remove or change names and other identifying details so individuals cannot be recognised in your write-up.
• Sensitivity and the right to withdraw -- avoid topics or data that could cause harm or distress, store your data responsibly, and respect a participant's wish to withdraw.

Labov's Observer's Paradox

Ethics connects to a famous methodological problem. William Labov described the observer's paradox: the aim of linguistic research is to find out how people speak when they are not being observed, yet the only way to gather data is to observe them. The very act of recording can make speakers self-conscious and shift them towards more careful, formal speech, distorting exactly what you are trying to capture.

You cannot eliminate the paradox, but you can reduce and discuss it. Strategies include recording over a longer period so participants relax, using naturally occurring data such as broadcasts or published texts, or analysing data not gathered specifically for the study. Naming the observer's paradox in your methodology, and explaining how it might affect your findings, shows real research maturity.

Applying the Language Levels to Your Data

The framework that turns raw data into analysis is the set of language levels (sometimes called methods of language analysis). Work through them systematically rather than commenting randomly:

• Phonetics, phonology and prosody -- sounds, accent features, stress and intonation (in spoken data).
• Graphology -- visual and layout features (in written or multimodal data).
• Lexis and semantics -- word choice, semantic fields, connotation, formality.
• Grammar (morphology and syntax) -- word structure, sentence types, clause patterns.
• Pragmatics -- implied meaning, politeness, presupposition, context.
• Discourse -- how a text or conversation is organised: turn-taking, cohesion, openings and closings.

Do not try to cover every level for every example. Choose the levels most relevant to your question, and analyse them in depth. An investigation into power in interviews, for instance, will lean heavily on pragmatics and discourse (interruptions, turn allocation, face-threatening acts) and need say little about graphology. Throughout, link your observations back to relevant concepts and theory so that analysis, not mere feature-spotting, drives the discussion.

Drawing Cautious, Evidence-Based Conclusions

Your conclusion should answer your research question -- but with appropriate caution. The hallmark of a strong investigation is that its claims are proportionate to its evidence. Avoid sweeping statements such as "this proves women hedge more than men." Your data may suggest a tendency in your sample, which is a very different claim. Tie every conclusion back to specific evidence, acknowledge the limitations of your sample and method, and, where relevant, note how your findings sit alongside established research -- whether they support it, complicate it or point in a different direction. Suggesting how the study could be extended or improved is a good way to demonstrate critical reflection.

Common Pitfalls to Avoid

Knowing where investigations typically lose marks is almost as useful as knowing what to include. Watch for these:

• A question that is too broad. If you cannot describe your data set and target features in a single sentence, narrow the question further before you start collecting.
• Feature-spotting instead of analysis. Listing every linguistic feature you can find, without interpreting why it matters to your question, reads as a checklist rather than an argument. Be selective and go deep.
• Description dressed up as analysis. Saying the speaker uses a declarative is description; explaining what that declarative does in context, and how it relates to your question is analysis. Always push to the "so what?"
• Overclaiming. Treating a small sample as proof of a universal truth undermines an otherwise careful study. Match the strength of your claims to the size and scope of your data.
• Ignoring ethics and the observer's paradox. Even with naturally occurring data, a brief, honest paragraph on ethics and the limits of observation shows research awareness and is expected.
• Forgetting the word count. The investigation is around 2,000 words excluding the data. Plan your sections so the analysis gets the lion's share; do not let a long introduction or methodology crowd it out.

Structuring the Write-Up

A clear, conventional structure helps the reader (and the moderator) follow your reasoning. A typical Language Investigation includes:

1. Introduction -- your research question, why it interests you, and what you expect to find.
2. Methodology -- how and why you collected your data, including ethical considerations and the observer's paradox where relevant.
3. Analysis -- the heart of the investigation, applying the language levels to your data with specific, quoted examples and links to concepts and theory.
4. Conclusion -- cautious, evidence-based answers to your question, with limitations and possible extensions.
5. References and data -- a reference list and your data set as an appendix (the data does not count towards the 2,000-word total).

Within the analysis, weaving quantitative patterns together with close qualitative readings -- a counted pattern, then an interpreted example -- tends to read far more convincingly than either approach alone.

Keep the assessment objectives in view as you write. The NEA rewards methodical analysis using terminology (AO1), engagement with concepts and theory (AO2), and attention to how contextual factors shape language (AO3) -- which is exactly why a clear question, a justified method and cautious, well-evidenced conclusions matter so much.

Continue Your Revision

To plan and refine your investigation, and to keep your wider exam skills sharp, work through these LearningBro courses:

• AQA A-Level English Language: NEA — Language in Action
• AQA A-Level English Language: Exam Strategy & Techniques

A focused question, a methodology you can defend, and conclusions that stay within the limits of your evidence are what separate a solid investigation from an outstanding one -- start early, and build the project around those three foundations.