Choosing a Personal Investigation Topic When the Obvious One Is Unworkable
You must submit a proposal for your personal investigation within one week. Your first idea was to investigate the representation of Indonesian domestic workers in Malaysian media, but you have no access to Malaysian sources in Indonesian, no contacts, and your teacher notes the topic drifts away from Indonesian language and culture. Your second idea is to study how a particular Indonesian online news outlet reports on climate policy. You have read the outlet for a year, can access its archive freely, and notice its headlines use a lot of metaphor and informal register. Your third idea is to compare how two Indonesian newspapers reported the same event, but you discover the second newspaper's archive is paywalled and your school has no subscription. You can work about four hours a week for ten weeks, and you must produce a 3,000-word written report in Indonesian plus a reflective statement in English.
1. Evaluate the three candidate topics against feasibility, relevance to Indonesian language and culture, and scope for analysis. Recommend one and justify the choice.
Read the worked answer
A good investigation topic must be doable within the constraints, centred on Indonesian language or culture, and rich enough to analyse rather than merely summarise.
Topic 1 - domestic workers in Malaysian media: Fails feasibility (no accessible Indonesian-language sources, no contacts) and drifts from the Indonesian focus. Reject.
Topic 2 - one Indonesian outlet's climate coverage: Strong on all three criteria. It is feasible: you already read the outlet and its archive is open, so four hours a week for ten weeks is sufficient for a sample of, say, 25–30 articles. It is clearly Indonesian-focused: you are analysing Indonesian-language journalism. It has analytical depth: you have already noticed metaphor and informal register, which gives you a concrete linguistic angle - how language choices frame a policy issue - rather than a vague 'what do they say about climate' question.
Topic 3 - comparing two newspapers: Analytically attractive because comparison sharpens analysis, but the paywall makes it infeasible without a subscription, and you cannot rely on a subscription appearing. Reject unless an accessible second outlet can be substituted.
Recommendation: Topic 2, narrowed to a linguistic question - for example, how metaphor and register in headlines construct the outlet's stance on climate policy. The existing familiarity is an advantage, not a shortcut, because it means your four weekly hours go to analysis rather than to learning where to look. If you want comparison later, you could contrast headlines with body text within the same outlet, which is feasible and still analytical.
2. Your teacher warns that a topic you already know well risks becoming a description of what the outlet says rather than an investigation. Explain how you would frame a research question and method that force genuine analysis.
Read the worked answer
Familiarity is only a risk if the question is descriptive. The fix is to make the question about language and effect, and to commit to a method before you start reading.
Framing the question: replace 'How does the outlet report climate policy?' with something like: 'How do metaphor and register in this outlet's headlines construct its stance on climate policy?' The first invites summary; the second demands that you identify linguistic features and argue for their effect.
Building the method:
- Define the sample explicitly - a date range, a number of articles, and a selection rule (e.g. all climate-related headlines in a six-month window). This prevents cherry-picking and makes the study replicable.
- Set categories in advance - e.g. types of metaphor (natural disaster, war, journey), register markers (informal particles, abbreviations, emotive adjectives). Coding against fixed categories turns reading into data.
- Record frequency, not just examples, so you can say which features dominate rather than merely that they occur.
- Analyse effect, not just presence - for each pattern, ask what it implies about responsibility, urgency or agency, and support the claim with the quoted headline.
- Acknowledge limits - one outlet, one period, headlines only; say what this means for your conclusions.
The discipline that prevents drift: keep a two-column log - 'what I found' and 'what it shows'. If the second column stays empty for a session, you are describing, not investigating. The reflective statement in English is also a check: it should describe decisions you made, which is impossible if you made none.
3. Ten weeks at four hours a week is not much. Propose a schedule and identify the two points of greatest risk to the project, with a mitigation for each.
Read the worked answer
Forty hours total must be allocated deliberately, with slack built in.
Indicative schedule:
- Weeks 1–2 (8 hours): finalise the research question, define the sample and coding categories, pilot the categories on five articles and revise them.
- Weeks 3–5 (12 hours): collect and code the full sample; keep the two-column log.
- Weeks 6–7 (8 hours): analyse patterns, select the strongest examples, draft an outline in Indonesian.
- Weeks 8–9 (9 hours): write the 3,000-word report in Indonesian, section by section.
- Week 10 (3 hours): write the reflective statement, proofread, check references. The remaining hours are contingency.
Risk 1 - coding takes longer than expected. Natural-language coding always does. Mitigation: cap the sample at a number you can realistically code (25–30 articles), and stop coding when the categories stop producing new patterns, noting that decision in the report.
Risk 2 - writing 3,000 words in Indonesian becomes a bottleneck at the end. Mitigation: draft in Indonesian from week 6 rather than writing in English and translating, which is slower and produces unnatural phrasing. Build a glossary of recurring terms as you code so the report's vocabulary is ready before drafting begins.
A secondary risk is losing the analytical thread under time pressure; the two-column log and the pre-set categories are the safeguards, because they keep the question in view even when you are busy.