A-Level Revision Aid: This resource is designed to support your revision. Content may contain errors. If you find a discrepancy with your class teaching, your teacher is correct — please let us know at alevelrevise@scott.scottrix.co.uk.
Language Investigation and Original Writing (NEA)
Year 1 / ASYear 2 / A-LevelAQA, OCR, WJEC/Eduqas
A* revision for the AQA 7702 NEA (2,000-word language investigation plus original writing and commentary) and OCR H470/03 independent research: choosing and scoping a focus, methodology and ethics, AO1-AO5 marking and the pitfalls that lose marks.
🎯 Learning Objectives
Choose and scope an investigation focus - gender, change, power, region, occupation or acquisition - framed as a precise question or hypothesis with named variables (AQA 7702/C; OCR H470/03).
Justify a methodology: data type (naturally occurring, elicited or corpus), sampling, transcription choices and the frameworks used for analysis (AO1/AO3).
Address ethics: informed consent, anonymisation, right to withdraw and secure storage, especially for data involving children (AO3).
Structure the investigation - introduction, methodology, analysis, conclusion - so findings are argued through language levels rather than described (AO1/AO3).
Target the AO profile of the AQA investigation (70 marks: AO1 20, AO2 20, AO3 20, AO4 10) and of the original writing task (750-word piece + 750-word commentary, 30 marks).
Meet the OCR H470/03 requirements: the research project plus the academic poster summarising title, data, method and findings.
Avoid the standard pitfalls: overbroad scope, data dumping, description without analysis and a commentary that narrates rather than reflects.
📋 Key Definitions and Core Concepts
Language investigation: The AQA 7702/C research project: an independent study of an area of language diversity or change, maximum 2,000 words excluding data and quotations, marked out of 70 (AO1 20, AO2 20, AO3 20, AO4 10).
Original writing: The second AQA NEA task: one 750-word piece written for a clearly defined audience and purpose (e.g. feature article, speech, campaign leaflet or short fiction), assessed for AO5 (expertise and creativity in the use of English).
Commentary: The 750-word reflection accompanying the original writing: justifies stylistic, lexical and structural choices against audience and purpose and connects to style models, assessed for AO4 (15 marks) alongside the piece's AO5 (15 marks).
Naturally occurring data: Language produced for real communication (conversations, online forums, signage) rather than gathered for the study - prized for validity but harder to collect and transcribe.
Elicited data: Language prompted by the researcher for the study (interviews, questionnaires, reading tasks, test sentences) - easier to control and compare, but risks artificiality and the observer's paradox.
Corpus: A principled collection of machine-readable texts (e.g. the BNC or COCA) searchable for frequency and collocation - the natural data source for language change and word-formation projects where scale matters.
Informed consent: Participants (or parents/carers for minors) must know what the research involves, agree to take part and be free to withdraw - the core ethical requirement for recorded spoken data.
Anonymisation: Removing or replacing names, places and identifying details in transcripts and write-ups so participants cannot be recognised; standard practice is pseudonyms at first mention plus secure storage and deletion of files.
Triangulation: Corroborating findings across methods or data types (e.g. frequency counts plus close discourse analysis of extracts) so conclusions rest on more than one line of evidence.
Academic poster: The OCR H470/03 one-page visual summary of the independent research - title, data, methodology, findings and conclusions - marked alongside the research project.
📌 Key Points
AQA 7702/C: language investigation max 2,000 words excluding data and quotes (70 marks: AO1 20, AO2 20, AO3 20, AO4 10) + one original writing piece (750 words) with a 750-word commentary (30 marks: AO4 15, AO5 15).
OCR H470/03: independent language research - a research project of about 2,500 words excluding data and appendices, plus an academic poster summarising it.
Focus areas: gender, change, power, region/dialect, occupation and acquisition - narrow each to a specific question ('How does X construct Y in Z?'), never a topic label.
Data: naturally occurring for authenticity, elicited for control, corpora (BNC/COCA) for scale - justify the choice and transcription method; address consent, anonymisation and withdrawal.
Analysis: argue findings through the relevant language levels (lexis, grammar, semantics, pragmatics, phonology/prosodics, graphology) with framework terminology, not impressionistic description.
Original writing: one piece for a real audience and purpose; the commentary must reflect on choices (lexis, syntax, tone, structure) against audience and purpose and connect to style models - reflection, not narration of process.
Pitfalls that cost marks: overbroad scope, data dumping, description without analysis, theory listed but never applied, no ethics section, commentary written as 'what I did'.
Choosing and scoping a focus
The NEA is marked as independent research, so the first A* decision is scope. AQA 7702/C allows an investigation into any area of language diversity or change - typically gender, change, power, region, occupation or acquisition - within 2,000 words excluding data and quotations; OCR H470/03 gives the same free choice for a longer project with an academic poster. Two thousand words is roughly four sides of argument: a project that tries to cover 'language and gender' produces a survey; one that asks a precise question produces an argument.
Gender: 'How are women politicians represented differently from men in 2024 UK broadsheet headlines?' - bounded data, a clear comparison, direct links to representation work (marked lexis, semantic derogation, agency).
Change: 'How has the semantic field of X shifted between the BNC (1994) and a 2024 web corpus?' - corpus evidence gives the quantitative spine and links change to usage (AO2).
Power: 'How does doctor-patient talk distribute authority in recorded GP consultations?' - asymmetry, directives, terminology and turn-taking.
Region: 'How do adolescents in two contrasting locations perform local identity through dialect lexis and prosodics?' - matched samples make the comparison workable.
Occupation: 'How does classroom teacher talk manage turn-taking and control?' - or hospital handovers, legal registers, service encounters.
Acquisition: 'How does a child aged 2-3 acquire question and negation formation?' - longitudinal or cross-sectional data, with parental consent essential.
Scoping rules: frame the title as a question or hypothesis with named variables ('How does X do Y in Z?'); fix data type, sample size and date range before writing; reserve at least 70% of the word count for analysis rather than context or methodology; and check the question is answerable with data you can actually access - overambition is the single most common reason investigations underperform.
Hypothesis vs research question: quantitative-leaning projects (change, corpus work) suit a testable hypothesis - 'use of X increases across the two samples' - while qualitative-leaning projects (power, identity, representation) suit an open question with sub-aims. Either way, state the variables explicitly, and expect to refine the wording after a pilot read-through of the data: a title written before the data exists is a guess, and examiners can tell. OCR projects allow a wider spread of areas, including comparison of spoken and written modes, but the same rule holds: the title must promise an argument the data can actually deliver.
Methodology: data, ethics and analysis
Data: three families. Naturally occurring data (recorded conversations, forums, signage) is prized for validity but hard to collect and transcribe, and Labov's observer's paradox - informants shifting style when watched - must be acknowledged. Elicited data (interviews, reading tasks, test sentences) is controllable and comparable but risks artificiality. Corpora (BNC, COCA) give searchable frequency and collocation evidence - ideal for change projects. Transcription is a methodological choice in itself: orthographic transcription with select notation (pauses, overlap, stress) is the workable middle ground; full transcripts belong in an appendix and do not count towards the word limit, though the analysis of them does.
Ethics (AO3): informed consent - written, and from parents/carers where participants are minors; gatekeeper permission (e.g. headteacher) for school-based collection; anonymisation with pseudonyms and removal of identifying details; the right to withdraw at any point, including after recording; secure storage and deletion of data files; and no deception about how the data will be used. State these explicitly and justify them - examiners credit ethics only when they are applied to your data, not listed generically.
Analysis: choose the frameworks before collecting, so the data answers the question. Work the levels the question demands - lexis and semantics for representation projects, grammar and pragmatics for power and occupation, phonology/prosodics for region, grammatical stages for acquisition - and mix quantitative evidence (frequency counts, percentages) with close discourse analysis of short contrasting extracts. This triangulation is what turns a data description into an argument: the counts show the pattern, the extracts show how it works.
Sampling and scale: justify sample size and composition - ten consultations, two age-matched friendship groups, 200 headlines across two years - and say why that is enough: depth beats coverage in a 2,000-word study, but the sample must still support the comparison the title promises. Note representativeness limits honestly (one practice, one region) and build them into the conclusion rather than hoping examiners miss them. For spoken data, fix the transcription conventions up front - orthographic with selected notation for pauses, overlap and stress; Jefferson conventions only where prosody is central to the question - and state them: transcription is a methodological decision examiners expect to see justified, not housekeeping.
Data protection: store recordings and transcripts securely (password-protected, school or college systems only), delete files after certification, and build anonymity in from the outset - pseudonyms applied at the transcription stage - so anonymisation is designed into the project rather than bolted on at the end.
Structure and AO marking (AO1-AO5)
Investigation structure: introduction (context, rationale, precise aims or hypothesis), methodology (data, sampling, ethics), analysis (two or three sections organised by framework), conclusion (synthesis, limitations, extension). AO profile (AQA, 70 marks):AO1 20 - accurate, consistent framework terminology and coherent academic expression; AO2 20 - linguistic concepts, methods and research used critically on your data; AO3 20 - contextual factors (mode, audience, purpose, situation, era) shaping the language; AO4 10 - connections made across your texts or extracts. Weight the writing accordingly: a quarter of words on context and a half on analysis is the rough A* ratio.
Original writing (30 marks): one 750-word piece for a defined audience and purpose - a feature article, campaign leaflet, speech or short fiction - assessed for AO5 (15): sustained control of register, lexis, syntax and structure for that audience. The commentary (750 words) earns AO4 (15): analyse your own choices as if analysing an unseen text - genre conventions followed and subverted, lexical and syntactic decisions justified against audience and purpose, connections to style models. The commentary must reflect, not narrate: 'the imperative in the opening line positions the reader as instructed because...' earns AO4; 'first I drafted, then I edited' earns nothing.
OCR H470/03: the same research skills assessed through a longer project plus the academic poster - a one-page visual summary of title, data, methodology, findings and conclusions. Treat the poster as a pitch to an examiner: it must stand alone, so distil the argument (question, headline findings, one quantified example), not the process.
Using research sources (AO2): two or three sources, each with a job: one framing concept (e.g. Fairclough's power in discourse for an institutional-talk project), one empirical study whose findings your data supports or complicates, and one critique you test against your own evidence. Every citation must touch the data: 'Fairclough's synthetic personalisation is visible in the GP's collective we' earns AO2; 'Fairclough wrote about power' earns nothing. For the original writing, choose real style models and reference them in the commentary - the AO4 marks reward explicit connections between your piece and the models, not vague nods to 'the genre'.
Common pitfalls and A* habits
The pitfalls are predictable, which makes them avoidable:
Overbroad scope: 'language and gender' as a topic, not a question - the data cannot support it, so AO3 stays shallow and the analysis becomes a survey. Fix: name variables, data and context in the title.
Data dumping: long transcripts narrated extract by extract without framework terminology - caps AO1. Fix: quantify the pattern, then close-analyse two or three short contrasting extracts.
Theory listing: research cited as background ('Lakoff said...') but never applied to the data - caps AO2. Fix: weave two or three sources into the analysis, each used on your own evidence.
Generic ethics: 'consent was obtained' without saying how, from whom, or why - loses AO3 methodology marks. Fix: state the mechanics applied to your participants.
Commentary narration: a process account of writing the piece - earns neither AO4 nor AO5. Fix: analyse your own text's choices against audience and purpose.
Weak conclusion: restating findings instead of answering the title question. Fix: synthesise into a direct answer, then note limitations and an extension.
A* habits: draft the question first and pilot your coding on a small sample before collecting everything; keep a research log so the methodology writes itself; integrate theory into analysis paragraphs (claim, evidence, concept); write the commentary as textual analysis of your own piece; and use your supervisor's feedback window - most centres allow one set of draft comments, and spending it on the analysis section (not the introduction) is the highest-yield choice.
Timing: the NEA runs across the year with fixed submission dates, so collect data early - access to participants is the slowest step - draft the analysis in stages as data arrives, and write the commentary last: it analyses choices, so it cannot be written before the piece is final. Check the board's current specification and your centre's guidance for word limits, word-count rules and deadlines - they are fixed, and ignoring them costs marks no analysis can recover.
Reading the mark scheme like an examiner: before writing, find the board's exemplar materials and marked scripts - they show exactly what the top band rewards. The investigation mark scheme is AO-banded, so audit every draft paragraph against the four AOs: does it apply terminology (AO1), use research critically (AO2), address context (AO3), make connections (AO4)? A paragraph with no AO attached is description, however fluent. For the original writing, the top band rewards a piece that could genuinely be published for its stated audience - so test the draft against that standard, and write a commentary that proves the control was deliberate.
💡 Worked Example
Exam-style: Design a language investigation into an aspect of language and power: set out a precise title, the data and sampling method, ethical safeguards, the frameworks for analysis, and how each design choice targets the assessment objectives. (AQA 7702/C, 70 marks: AO1 20, AO2 20, AO3 20, AO4 10; the OCR H470/03 project sets the same demand, plus an academic poster.)
Mark-scheme logic: Model outline: 'How does doctor-patient talk distribute authority? A study of ten recorded GP consultations.' <strong>Title:</strong> a question with named variables and bounded data - instantly shows AO1 precision. <strong>Introduction:</strong> power in institutional talk (asymmetry of knowledge, goal orientation, gatekeeping) with two or three research sources woven in as tools, not a literature dump (AO2). <strong>Data and methodology:</strong> ten naturally occurring consultations recorded with written consent, orthographic transcription with pause and overlap notation, justification of the sample (same practice, comparable appointments) and its limits (AO3). <strong>Ethics:</strong> consent forms, pseudonymisation of doctors and patients, right to withdraw, secure storage and deletion - stated as applied mechanics (AO3). <strong>Analysis - three sections:</strong> (1) directives and imperatives: who issues them, who complies; (2) medical jargon as gatekeeping: asymmetry of specialist knowledge; (3) turn-taking and topic control: question chains, interruptions, agenda management. Each section argues findings through framework terminology with quantified evidence plus close discourse analysis of two contrasting extracts (AO1/AO3), links each finding to power concepts and institutional context (AO2/AO3) and connects patterns across the extracts (AO4). <strong>Conclusion:</strong> synthesise the three findings into a direct answer to the title question, note limitations (small sample, single practice) and suggest an extension (AO1/AO3). Full marks require accurate terminology throughout, critical rather than decorative use of research, context-sensitive analysis and explicit connections - description alone caps the mark in the middle bands.
❓ Practice Questions & Mark Schemes
Explain why a narrow focus produces stronger language investigations than a broad one. (AQA 7702/C, AO1/AO3)
Answer: The 2,000-word limit forces depth: a narrow question allows the full research cycle - collect, transcribe, quantify, close-analyse - on data that can actually answer it. A broad topic ('language and gender') forces a survey: examples stay illustrative, so AO3 (contextual depth) and AO1 (sustained framework application) both cap in the middle bands. The A* move is a question with named variables, data type and context - 'How are women politicians represented in 2024 UK broadsheet headlines?' - which fixes the data, makes the comparison workable and reserves most of the word count for analysis.
Evaluate this investigation title: 'An investigation into language and gender.' What is wrong with it, and how would you reframe it?
Answer: It names a topic, not a research question: no variables, no data type, no context, no comparison - unanswerable within 2,000 words. Reframe: 'How are women and men athletes represented in UK broadsheet headlines covering the 2024 Olympics?' - this names the data (headlines), the comparison (women/men), the context (2024 UK press) and implies the method (marked lexis, evaluative adjectives, agency in headlines). It also forces a check on sample comparability (same events, same paper) - the kind of methodological thinking AO3 rewards.
Outline the ethical requirements for a language acquisition investigation using recordings of a child aged two to three.
Answer: The child cannot consent, so: (1) informed written consent from parents/carers, with a clear account of how recordings will be used; (2) gatekeeper permission if recording in a nursery or school; (3) anonymisation - a pseudonym for the child and family, removal of place names and identifying details from transcripts; (4) the right to withdraw at any point, including after recording; (5) secure storage and deletion of files; (6) no disruption of the child's normal routine or relationship with the researcher. Examined answers state these as applied to the data and justify them - a generic list loses the AO3 marks.
What must the AQA original writing commentary do to hit AO4 and AO5? (30 marks: 15 + 15)
Answer: The commentary (750 words) earns AO4: connect your piece to style models and linguistic concepts - analyse your own choices as if analysing an unseen text: audience, purpose, genre conventions followed and subverted, register, lexis, syntax, structure, each justified ('the imperative in the opening line positions the reader as instructed because...'). The piece (750 words) earns AO5: sustained, controlled creative style for the defined audience and purpose. The distinction is reflection versus narration: 'first I drafted, then I edited' hits neither AO. Strong commentaries also acknowledge where the piece deviates from genre convention and why.
Identify three common pitfalls in NEA investigations and the marks each costs.
Answer: (1) Overbroad scope - a topic label instead of a question: the analysis becomes a survey, capping AO3 depth and AO1 application; fix by naming variables, data and context in the title. (2) Data dumping - long transcripts narrated rather than argued through frameworks: caps AO1; fix by quantifying the pattern and close-analysing short contrasting extracts. (3) Theory listing - research cited as background but never applied to the data: caps AO2; fix by weaving two or three sources into the analysis itself. Two further marks-losers worth naming: a generic ethics section (AO3) and a commentary that narrates process rather than reflecting on choices (AO4).
amazon.co.uk
