A friend of mine spent six months circling a dissertation topic before she wrote a single proper paragraph. She loved everything about how people speak accents, slang, the way her grandmother’s Yorkshire vowels had softened after forty years in London. Every book she read felt relevant. Every article opened three more directions. By January, she had forty pages of notes and no actual question. What she had, without realising it, was an interest, not a project.

That’s the trap almost every linguistics student falls into at some point, and it’s the real reason so many end up hunting for linguistics research topics for undergraduates rather than sitting down and building one themselves. The subject itself isn’t short of material. Phonetics, syntax, sociolinguistics, acquisition each could fill an entire career, let alone a dissertation. The actual skill isn’t finding something interesting. It’s cutting an interesting subject down until it’s small enough to hold in two hands.

Why the broad version always feels safer, and isn’t

When everything connects to everything, narrowing feels like loss. You worry you’ll pick the wrong slice and miss the better idea sitting just out of reach. So the instinct is to stay wide, stay flexible, keep options open for as long as possible.

In practice, that instinct backfires almost every time. A broad topic can’t be argued with any precision, because you haven’t decided what you’re actually measuring or comparing yet. Vague topics produce vague sentences. Markers notice this immediately not because they’re harsh, but because a claim with no edges can’t really be checked against anything.

Take my friend again. “How accents change over time” is not a question it’s a whole subfield. But “does exposure to London speech correlate with vowel shift in speakers who moved from Yorkshire after age thirty” is something you could genuinely investigate, record, and defend. Same curiosity. Completely different level of usefulness.

The three things a workable question actually needs

A question you can realistically finish in one undergraduate year needs a defined group of speakers or texts, a specific comparison or change to observe, and a method you can physically carry out alone. Drop any one of these and the whole thing stalls halfway through.

Language acquisition is a good example of how this plays out. “How do children learn grammar” belongs in a textbook, not a proposal, because there’s nothing concrete attached to it yet. Shift it to whether four to six year olds use irregular past-tense verbs more accurately in free speech than in a structured task, and suddenly there’s an age group, two conditions, and a method sitting right there.

Phonetics students run into the same wall with intonation. “How intonation conveys meaning” is too enormous a target for one dissertation to hit cleanly. Narrow it to whether rising intonation in a set of recorded interviews signals genuine uncertainty or something closer to politeness, and now there’s something to transcribe, count, and actually argue about with real evidence behind it.

Check the data before you fall in love with the idea

Most students narrow their topic by thinking harder about the theory. Far fewer stop to ask whether the evidence they’d need even exists, or whether they can reach it in time. This single check saves more dissertations than any amount of extra reading ever does.

If your question depends on recording twenty bilingual speakers, ask now whether you can realistically find them, get ethical approval processed, and transcribe everything before your deadline creeps up on you. If it leans on a historical corpus, check it’s actually searchable and large enough to say something with. Good ideas die constantly not because the linguistics was flawed, but because the data behind them was never within reach of one student working alone.

Often this step produces something better than your original plan. Someone who wanted to study dialect variation across the whole UK, but only has realistic access to their own town, doesn’t need to give up the idea. They narrow it to one vowel, one consonant, one piece of local vocabulary, studied properly within a community they can actually sit down and interview. The smaller version tends to be the one that actually gets written, and written well.

A simple test before you commit

Before treating any question as final, try answering it in one sentence, using the kind of wording you’d genuinely use in your results chapter. If you can’t do that yet, it’s still too loose to start work on. “I’m looking at how young people text differently” gives you nothing solid to hold. “Speakers aged eighteen to twenty-two use more phonetic spelling in informal messages than in emails” gives you a sentence you could actually go out and test.

It’s also worth asking what a disappointing result would look like before you begin. If every possible outcome would still let you write something like “language is always changing,” the question hasn’t been sharpened enough yet. A strong question carries real risk of surprising you, because it’s specific enough to be provably wrong.

The question keeps moving, and that’s normal

Narrowing rarely happens in one sitting. It happens gradually, as your reading shows you what’s already been studied and where the genuine gaps sit. Discovering that your first instinct has already been researched thoroughly isn’t wasted time it tells you exactly where the edge of current knowledge sits, and a solid undergraduate project usually lives just past that edge.

My friend eventually handed in a tight, well-evidenced dissertation on vowel shift in relocated Yorkshire speakers. It bore almost no resemblance to those forty pages of scattered notes from October, and that’s exactly how it should have gone. The question you start with is rarely the one you finish with. What matters is staying willing to keep cutting, checking, and reshaping it, right up until it finally fits the time, the data, and the words you’ve actually got.

0