Frequency Dictionaries: Which Words to Learn First

Open a random textbook chapter and it's easy to assume every new word in it matters equally. It doesn't. In every language, a small core of words does most of the work — showing up in nearly every sentence — while the vast majority of the vocabulary shows up rarely, if ever, in your actual reading and conversations. A frequency dictionary is simply a ranked list that tells you which is which.

This isn't a minor optimization. Two learners who each memorize 500 words can end up with wildly different levels of real comprehension, depending entirely on whether those 500 words were chosen by frequency or by chance. One of them can follow a basic conversation; the other has 500 flashcards and can't order coffee.

Below: the data behind why frequency matters so much, how to actually build your first 500–1000 word list, where to find a frequency dictionary for your language, and the pitfalls that catch people who use one for the first time.

None of this replaces grammar study, listening practice, or actually using the language — a frequency list only solves one problem: which words deserve your limited attention first. But solving that one problem well changes the return on every hour you spend afterward, which is why it's worth getting right before you make a single flashcard.

If you've ever felt like your vocabulary app "isn't working" despite steady daily use, it's worth checking this before anything else. A perfect review habit built on the wrong 500 words still produces a learner who can't order coffee.

In the early 20th century, linguist George Zipf noticed that if you rank a language's words by frequency, the most common word appears roughly twice as often as the second most common, three times as often as the third, and so on. The practical consequence: a relatively small set of words accounts for a disproportionate share of everything ever spoken or written.

In English, the single word "the" alone makes up roughly 5–7% of all words in a typical text — more than most people would guess for one word. Zoom out to the top 10 words (mostly articles, pronouns, and basic prepositions) and you're already looking at a meaningful double-digit share of everything written or spoken. The pattern isn't an English quirk; the same skewed, top-heavy shape shows up in the frequency data of essentially every language that's been studied this way.

These numbers vary somewhat by language and by whether you're measuring speech or writing, but the shape of the pattern is remarkably consistent: coverage rises fast at first, then flattens sharply. That flattening curve is the entire argument for using a frequency dictionary instead of a random list.

Look at the shape, not just the numbers: the curve is steep from 0 to 1,000 words and nearly flat from 3,000 onward. Doubling your vocabulary from 1,000 to 2,000 words buys a few more points of coverage. Doubling it again from 2,500 to 5,000 buys even less. Your first 1,000 words, chosen well, are the highest-leverage vocabulary you will ever learn in a language.

This is also why "I know 2,000 words but still can't follow a conversation" is such a common complaint. Text coverage measures whether you'd recognize a word if you saw it written down — it says nothing about whether you chose the right 2,000 words, whether you can recall them fast enough in real time, or whether you've actually practiced using them. Coverage is necessary but not sufficient; frequency-based selection just makes sure the necessary part isn't wasted on the wrong words.

The useful takeaway isn't the exact numbers — sources disagree on the precise boundaries — it's the shape: your first 1,000 words already put you in reach of basic conversational ability, while each proficiency level after that requires progressively more vocabulary for progressively smaller gains in what you can express. That's the same diminishing-returns pattern from the coverage chart, just relabeled in terms of what you can actually do with the language.

A frequency list naturally mixes both, and that's exactly what makes it more useful than a themed vocabulary list (all food words, all travel words). Themed lists give you content words without the function words that connect them into sentences — you end up able to name a lot of nouns but unable to actually say anything with them. The blend a frequency list gives you, in roughly the proportion real language actually uses it, is what turns memorized words into usable sentences.

The random-selection learner isn't doing anything wrong in terms of effort or technique — the flashcards are well made, the reviews happen on schedule, the habit is solid. The entire gap between the two outcomes traces back to one decision made before either of them opened a flashcard app: what list of words to start from.

Every frequency list is only as good as the text it was built from — linguists call that source text a corpus. A list built from 19th-century novels will rank words differently than one built from movie subtitles or social media posts, because the vocabulary of formal literature and casual speech genuinely differs. Knowing roughly where a list's numbers came from tells you how much to trust it for your specific goal, whether that's reading, conversation, or something more specialized.