The most common Spanish words, and how far they really get you

What counts as the most common Spanish words

If you search for the most common Spanish words, you will find lists of 100, 1,000 or 5,000 entries, and at the top they all look much the same. What most of them leave out is where the ranking came from. That matters more than the order, because a list is only as good as the Spanish it was counted from.

A frequency list is, at heart, a count. Someone gathers a large body of real Spanish, which linguists call a corpus, and tallies every word in it. The Real Academia Española did this with CREA, a corpus of more than 160 million words. Mark Davies built a smaller one of 20 million words for his Routledge frequency dictionary, but he balanced it carefully between speech and writing.

Whichever list you open, the top is grammar: de, la, que, el, en, y, a. These little words hold every sentence together, yet on their own they mean almost nothing, so learning that de comes first tells you very little about what to study tomorrow.

The verbs that carry the weight come next. Ser, estar, haber, tener, hacer, poder, decir and ir all sit in Davies' top 30, and everyday nouns follow soon after. In the subtitle counts NativeLingo relies on, casa, tiempo, vida and día all land inside the top 150 word forms, which is why a beginner hears them in almost every conversation.

One detail changes how you should read any list. Some lists count forms, meaning words exactly as they appear on the page, such as dijo or dicen. Others count lemmas, the dictionary headwords that group those forms together, such as decir. The RAE list ranks forms, so es, era and fue each get their own slot, while Davies folds all of them into ser. That is how the same verb can fill five places in one top 100 and only one in another.

For deciding what to learn, the lemma list is the more honest guide, although the forms are what you will actually run into. A list gives you ciudad, but a travel page talks about three cities, and in Spanish that is tres ciudades.

So it helps to treat a list as a map of headwords rather than a script to memorise. Each word still has to be learned in the shapes it takes inside real sentences, with its plurals and past tenses, because those are the versions your eyes will meet.

When you do pick a list, check that it names its corpus and says plainly whether it counts lemmas or forms. It is also worth making sure the corpus includes speech as well as books and newspapers, since written-only counts push literary words too high. Davies split his corpus evenly between speech, fiction and non-fiction for exactly that reason.

How much of a page the first 1,000 words cover

The more useful question is coverage: if you knew the top 1,000 words, what share of an ordinary text would you recognise? Davies measured this for Spanish in 2005, and because he counted speech, fiction and non-fiction separately, his numbers show how much the answer depends on what you are reading.

Share of words in a text covered by the top 1,000 lemmas. Davies, 2005.

Speech comes out on top because people repeat themselves, reaching for the same small set of verbs and fillers all day long. As a result, the first thousand words cover 87.8% of spoken Spanish, whereas in non-fiction, where the topic changes from one text to the next, they cover 76%.

Beyond that point the gains shrink fast. In Davies' data the second thousand adds another 6.5 to 8 percent of a written text and 4.9 percent of speech. The third thousand adds only 2.3 to 4.2 percent. By 3,000 words you have reached 88.2% of non-fiction, 89.6% of fiction and 94% of speech.

Davies calls this a law of diminishing returns, and the phrase fits. Every new thousand words takes roughly the same effort to learn, but each buys a smaller slice of the page than the one before. English behaves in much the same way: when Davies set his results beside Nation's figures for English, the top 1,000 covered 74.5% of non-fiction and 84.3% of speech.

This curve explains a feeling most learners know well. Through the first few hundred words progress feels fast, since every word you learn seems to turn up everywhere. Somewhere past a thousand, though, it starts to feel as if you have stalled, even while you keep studying as hard as before.

You have not stalled. Each new word simply appears less often than the last, so you notice it less, and the improvement arrives in small pieces scattered across many pages.

Why 80% still feels like a wall

Eighty percent sounds like most of the job, but on a real page it means that one word in five is unfamiliar, which leaves four gaps in a twenty-word sentence. You can usually guess one unknown word from the words around it, whereas four at once turn reading into a puzzle you end up solving with a dictionary.

Here is roughly how that feels, with the unknown words blanked out: the ____ was late again, so I ____ to the ____ and ____ for an hour. You can follow the grammar and about half the story, and still you cannot say what happened. At 98% the same line would be missing a single word, and the rest of it would tell you what that word probably meant.

Reading research backs this up. Hu and Nation (2000) gave learners the same story with 80%, 90%, 95% or 100% of the words known, and at 80% none of them understood it well enough. At 90% and 95% a few managed, although most still did not. That is why the authors put the bar for reading without a dictionary at about 98% of words known, or one unknown word in fifty.

How many words does that take? Nation (2006) estimated 8,000 to 9,000 word families for written English and 6,000 to 7,000 for speech, where a word family groups related words such as courage, encourage and discouragement. These are English figures counted in families rather than Spanish lemmas, so they will not carry over exactly. They do show the scale, though: a list of 1,000 words is where reading begins, and comfortable reading lies several thousand words further on.

The same numbers suggest how to add words without wrecking your reading. Each word past the first thousand is rarer, yet it is often the very word that blocks a particular sentence, which is why closing those last gaps turns decoding into reading. This works best when the share of unknown words stays small, because one new word on an otherwise familiar page stands out and gets your attention.

NativeLingo is built around that line. It swaps about one word in a hundred into Spanish and never more than two, so your reading stays above the 98% mark while new words arrive one at a time.

Which thousand belongs to which level

Course books sort vocabulary by CEFR level, from A1 to C1, while frequency lists sort it by rank. The two scales line up more closely than you might expect, mostly because the common words are also the ones teachers introduce first.

NativeLingo gives every Spanish word a level according to its rank in a frequency list built from film and TV subtitles. That list comes from the open FrequencyWords project, which counted OpenSubtitles, and in our Spanish dictionary the bands fall like this.

A1

Ranks up to 500, like casa, día, semana

A2

Ranks 500 to 1,500, like calle, libro, tren

B1

Ranks 1,500 to 3,000, like mercado, lluvia

B2

Ranks 3,000 to 6,000, like guitarra, risa

C1

Ranks 6,000 to 12,000, like subasta, tos

Because these ranks count word forms in subtitles, they lean toward everyday speech. Casa comes 91st and semana 326th, mercado lands in B1 at 1,767th, and tos, the word for cough, sits all the way down at 9,618th.

Subtitles have a bias of their own, of course. Film dialogue is full of words like matar and dios that a textbook would save for later. The language of offices and paperwork, meanwhile, turns up less often than it does in adult life. No corpus is neutral, which is one more reason a good list should tell you which one it used.

In practice the rule is simple. At A1 the top 500 are your words, and a word from three thousand places further down will cost you more effort now than it pays back.

That is how NativeLingo decides what to swap. At A1 it shows you semana but leaves market in English, and as your level rises, the window slides further down the list.

If you are not sure where you stand, it is worth checking before you choose a list at all. The NativeLingo level test takes about three minutes and asks 12 questions from A1 up to B2. Start a band too low and you spend weeks on words you already know, while starting too high buries you in words you will forget by Friday.

Learning them without a flashcard marathon

A list makes a good map and a poor teacher. If you read down a column of 1,000 words, each one gets a single meeting with no sentence around it. A word needs many more meetings than that before it stays.

Webb (2007) found that clear gains begin at around ten encounters with a word. Pellicer-Sánchez and Schmitt (2010) followed learners through a whole novel and found that after ten or more encounters, 84% recognised a word's meaning, although only 55% could recall it unprompted. Reading, in other words, gives you recognition first, and recall needs a further step.

That step is retrieval, which simply means pulling the word out of memory instead of looking at it again. Karpicke and Roediger (2008) showed how big the difference can be. A week after learning, people who kept testing themselves remembered about 80%, while those who kept re-reading the material remembered 33 to 36%.

Put together, the recipe is short. You meet a word around ten times in real sentences. Then you test yourself on it at growing intervals, typing the answer instead of flipping a card and telling yourself you knew it.

NativeLingo follows that recipe. It swaps common words into the pages you already read, in the form each sentence needs, and stops showing a word after about ten meetings or once you clearly know it. A short daily review then asks you to type the word, and FSRS works out when you should see it next.

In a sense you are still learning a frequency list, since the extension follows frequency order for your level. The difference is that you meet the list inside sentences you were going to read anyway, in the order your own reading produces. A word that keeps turning up on the sites you visit gets more chances to stick. One that never appears in your life stays on the list, where it costs you nothing.

If you prefer flashcards, the same rules hold. Choose cards from the band that matches your level, give each one an example sentence and type your answers whenever you can.

Keep an eye on the pace as well, because every new word brings months of reviews with it. At five new words a day, reviews settle at about 7 to 10 minutes a day after a year, while twenty a day takes 27 to 40 minutes. Pick the pace you could keep up during a busy week, not a quiet one.

When a frequency list is the wrong tool

Frequency lists describe average Spanish, and your Spanish may not be average. A nurse moving to Madrid needs words for symptoms long before those words reach the top 3,000. So if you have a clear goal, put a topic layer on top of frequency: learn the general core first, and then the words your situation keeps throwing at you. NativeLingo does a small version of this by giving a bonus to words from the goal you picked, such as travel or work.

Grammar words are another blind spot. De, que and se lead every list, yet no extension can swap them into an English page, because English has no slot for them. Fixed phrases behave the same way. Por favor, a lo mejor and de nada are built from common words, but you learn each one as a unit, and no list of single words will tell you that a lo mejor means maybe. For both of these you need real Spanish sentences, ideally with a teacher.

Reading is also not the same as speaking. NativeLingo builds recognition and recall of words, but it does not train your ear or your pronunciation. If you need to talk with people next month, a class or a tutor will get you there faster. The two combine well, since words you met while reading give your lessons more to work with, and lessons turn those words into phrases you can actually say.

So learn the first 500 on purpose, and then let reading carry you through the next few thousand, with short reviews to make them stick.

Questions