Listen to two Spaniards talking and try to transcribe the first thirty seconds. Half of what they say is not content. It is connective tissue: little words that manage the conversation, buy processing time, signal attitude, soften disagreement and hold the floor.
Es que before an excuse. O sea to reformulate. Bueno to change direction. A ver to think out loud. Hombre to push back gently.
You recognise all of them. You almost certainly produce none of them, and that absence is more audible than any grammar mistake you make.
The short answer
Most of what fluent speech is made of is prefabricated, so a speaker with no prefabs sounds wrong in a way that has nothing to do with level. Britt Erman and Beatrice Warren’s study of the idiom principle found formulaic sequences accounting for 58.6% of spoken and 52.3% of written English. Half the machine is chunks, and you have been building the other half.
This is not an aesthetic point. Frank Boers and colleagues had blind judges rate recorded learner interviews and found the frequency of formulaic sequences correlating with oral proficiency scores. The markers are not decoration on top of fluency. They are a large part of what listeners are measuring when they decide how fluent you are.
How often they actually come up
Our own probe corpus is written for Spain, and it gives a concrete number. Across 866 target answers, 155 contain one of nine core markers. That is 17.9%, or nearly one natural Spanish answer in five.
Es que is the outlier and deserves the attention. It is not a filler. It is the standard way Spanish introduces a reason, an excuse or a mitigating circumstance, and a learner who never uses it delivers every explanation as a flat declaration.
The count is a floor rather than a ceiling, for two reasons. It only looks at the answers we wrote as targets, which are short by design, and it only searches nine forms. Real conversation is longer and the inventory is wider. But even at 17.9%, the implication is uncomfortable: roughly one in five of the things we would consider a natural answer contains a word you have never once said out loud.
The ones that matter in Spain
Several of these exist across the Spanish-speaking world with different weight, and a couple are strongly Peninsular. If you learned LATAM-neutral Spanish and moved here, this is among the most audible gaps in your speech.
Es que. The explanation opener and softener. Es que no me dio tiempo. Es que pensaba que era mañana.
O sea. Reformulation, roughly “I mean”. Me gustó, o sea, no estuvo mal. Without it you cannot correct yourself gracefully mid-sentence.
A ver. Signals you are thinking, or invites the other person in. A ver, cuéntame. A ver, a ver, no me entero. Without it every pause is dead air rather than visible thought.
Bueno. Direction change, concession or mild disagreement depending entirely on intonation. Bueno, da igual. Bueno, yo no lo veo así. This is the one whose meaning is carried almost entirely by how it is said, which is why reading about it achieves so little and imitating it achieves so much.
Hombre and mujer. Not address so much as social oil. Hombre, no me digas eso. Mujer, que no pasa nada. Without them you have no warmth register at all.
Pues. The connective tissue between everything else. Pues nada, me voy. Pues eso, que no viene.
Venga. Agreement, encouragement, farewell. Venga, nos vemos. Without it, endings are abrupt and agreements sound like contracts.
Two of these deserve a warning. Venga and hombre in these uses are strongly Peninsular, and venga in particular will read as unmistakably Spanish to a Latin American ear. That is the point if you live here, and worth knowing if you also speak Spanish with people who do not. The same applies to vale, which is so saturated in Spain that its absence is noticeable on its own, and which sits alongside vosotros as a marker of whose Spanish you learned.
Same content, two speakers
Not one word of new vocabulary and not one new grammatical structure. The second speaker sounds several levels more fluent, and Boers’s judges would have scored them that way.
Look closely at what the markers are doing there, because it is not softening in the vague sense. Hombre signals the relationship is intact. Es que announces that a reason is coming, so the refusal is not arbitrary. O sea flags a correction of the impression just given. And pero bueno leaves the matter open rather than settled. Four separate social moves, all in words a beginner already understands, none of which the grammatically identical first version performs at all.
That is why this reads as a level difference rather than as a style difference. The second speaker is doing more, not saying it more prettily.
Why courses skip them
Three structural reasons, none of them carelessness.
They do not translate cleanly. Es que has no single English equivalent. Teaching it means explaining a social function rather than a meaning, and functions do not fit in a vocabulary list.
They cannot be tested. There is no gap fill for o sea. It is optional everywhere, useful everywhere and wrong nowhere, which makes it invisible to any assessment built on right and wrong.
They look trivial. Next to the subjunctive, teaching someone to say bueno feels like filler. That is exactly backwards given the frequency, and examiners know it: DELE B2 speaking rewards this texture explicitly.
The strongest case for ignoring them
There is a real argument here and it is worth hearing.
Markers are the last thing a listener needs in order to understand you. Content words carry the message, and a learner with limited study time who spends it on pues instead of on the past subjunctive is optimising for how they sound rather than for what they can say. Sounding smooth while being unable to express a hypothetical is a worse place to be than the reverse.
There is also a real failure mode in overuse. A learner who has just discovered o sea and deploys it six times a minute sounds like someone performing fluency rather than having it, and that is more conspicuous than saying nothing. Markers are calibrated by native speakers to a rhythm you have not internalised yet, so early production genuinely does sound odd.
Why they still repay the time
The counter-argument holds if markers were cosmetic. The evidence says they are structural.
At 58.6% of spoken discourse, prefabricated units are not a finishing layer on top of grammar, they are most of the material. And Boers’s judges were not rating charm: they were rating oral proficiency, and the marker count tracked their scores. If half of speech is chunks, a learner with no chunks is not a fluent speaker missing polish, they are assembling every utterance from scratch in real time.
That is also why markers buy processing room. Norman Segalowitz separates three kinds of fluency, and a ver is a direct intervention in two of them at once: it fills the gap in utterance fluency while your cognitive fluency catches up, and it does so with a sound that reads as thinking rather than as failing. That matters most in exactly the situations where native speech is already arriving faster than you can parse it.
The production gap in its purest form
Notice the contradiction. You understand these perfectly. You know what they mean. You never say them.
Laufer and Goldstein’s four degrees of word knowledge, validated across 435 learners, is the shape of the problem: recognition and active recall are separate rungs and the hierarchy held at every frequency band they tested. Markers sit at the extreme end of that split, because input alone drenches you in them and never once asks you to produce one. It is the passive and active split with the dial turned all the way over.
There is an avoidance component too. Elaine Schachter’s 1974 paper found learners producing fewer attempts rather than more errors, which made an error count read them as strong. A transcript of your Spanish would show no marker errors at all, for the same reason. Martin Kleinmann’s follow-up found the predictor of a detour was confidence in that specific item, which is why markers are dodged more thoroughly than grammar: nobody ever taught you the rule for bueno, so there is nothing to feel confident about.
And unlike a grammar dodge, this one has no visible substitute. When you avoid the subjunctive you at least produce a different structure. When you avoid a marker you produce silence, and silence looks like having nothing to say.
How you actually acquire them
One a week, chosen deliberately. Listen for it obsessively for a few days. Count how often people use it. Then force it into every conversation whether or not it feels right. One at a time matters more than it sounds: trying to install six markers at once produces a speaker who is monitoring their own speech instead of listening, which costs more than the markers gain.
Produce before it feels natural. It will not feel right until you have said it fifty times. Say es que before every explanation for a week. Say a ver every time you pause. The discomfort is the acquisition, and Roediger and Karpicke’s test-enhanced learning work is the reason production beats another hour of listening: retrieval loses to restudying on an immediate test and wins clearly on a delayed one.
Keep the input running alongside. Paul Nation’s four strands put meaning-focused input, output, language study and fluency work at roughly a quarter each. Markers are learned from input and installed by output, and skipping either half stalls it.
Accept a period of sounding odd. You will overuse the new one for a fortnight. Everyone does, including children. The alternative to sounding briefly odd is sounding permanently flat, and only one of those two states resolves itself.
What we do with this at Suelto
Discourse markers are tracked as their own coverage area rather than as vocabulary, with conectores, reparación, reacciones and registro as separate categories in Territory, because a learner can be strong across grammar and produce none of them.
Free production is where they are scored, since a marker is not a right answer to a prompt. The conversation step records whether they surface in speech you generated yourself, and schedules the missing ones until they do. Absence is the signal, which no correctness-based system can see, and it is the same principle behind everything else here: the gap between understanding and using is only visible to something looking for what did not happen.
The gap test shows where your conversational texture sits, in about ten minutes with no signup, and the method page sets out the rest.
Nine little words, appearing in nearly one Spanish sentence in five, every one of which you already understand completely. This is the cheapest fluency available to you at your current level, and you have been walking past it for years.