What are the words I actually have to remember?

Roxy's second in-program lesson is the only place in the corpus where the owner's whole vocabulary is handed over as a list, and the trainer says it is the thing to drill into your head: eh when you dislike what she is doing, good when you like it and she must keep doing it, okay to release her. Later in the same session he comes back and gives good its own definition - a duration marker, said short, because good job is heard as the end and the dog gets up. That is the distinction owners break by accident and cannot diagnose afterwards, because from the inside it looks like a dog who will not hold a command. Roxy's file is also where the reason for the strictness is visible: she is described as a dog who arrived with no sense of context at all, so the words have to carry what context does not.

Stormie's file contains the clearest statement of why the marker matters more than the correction does, and it is made twice by two different trainers about the same puppy. The first lesson gives the two-markers argument with its worked example: ant then collar and nope then collar look identical from outside, and to her brain they are opposite instructions - go back into that, or never do that again. The choice is then made concrete with a case the trainer had just finished, a sweet Rottweiler who had sent two of her own household's dogs to the emergency vet, where the softer marker after a bite would have read as an invitation to escalate. Two weeks later a different trainer gives the owners the same distinction from the other end - the practical one - which is that without it you produce an undifferentiated stream of nope that stops meaning anything and starts sounding like a game.

Deuce is where the vocabulary meets the hardware, and it is the only place in the corpus where it does. The two marker words are introduced as the answer to an owner's implicit question - how does the dog know which is coming - and then, in the second half of the same visit, mapped onto the transmitter's two buttons: black carries the soft marker at a low stim, red carries nope and raises the value, because nope is a legitimate punishment. The trainer says he began doing this only recently, which makes it a dated piece of method evolution rather than a fixed rule. By the second lesson the instruction to the owners has narrowed to one button: use the black, never the red, which is for dogs that bite people, and never the vibrate - not because vibration is ineffective but because a marker followed by something the dog cannot feel teaches the marker to mean nothing. That is the sharpest small argument in the cluster.

Luna's family gets the marker block twice, and hers is the file where the judgement around it is taught rather than just the rule. The order never varies - the word first, then the collar, so that the collar reads as a consequence of the word and does not attach itself to whatever else is in the room. The correction is assigned to a behaviour and sized to it, and the trainer says out loud that a nope for sniffing a table is a little extreme; for that he would use a redirect instead. He also tells the family, of a dog he has spent four weeks with, that they will barely ever need the punishment marker at all, because she does not do the kinds of things that warrant one - which is the only place in this cluster where a trainer scopes the vocabulary to the individual dog rather than issuing it whole. The delivery coaching is hers alone: the marker is not louder, it is sharper.

Dutchess is where an owner pushes back, and the answer is the best short version of the whole idea. The objection is the obvious one: the collar looks the same in two situations that are supposed to be opposites, so how can it mean two things? The trainer's reply is that the chain of events is identical to you and to me and that the only difference is the word swapped - and then he gives the two worked examples on either side of it, the counter and the cot. Her first lesson is where the theory behind that comes from, delivered as the one thing to take away from the whole hour, and it is also where the theory goes slightly wrong: the tug example he uses to illustrate negative reinforcement is actually negative punishment. The same example, with the same error, turns up in another family's lesson weeks later.

Bo's session is the duration marker used rather than explained to a client - one handler, one dog, narrating to camera. He marks the heel with good, defines it on the spot as meaning he is in the right place and may not leave it, and in the same breath calls the small collar taps he is using negative reinforcement. It is the cleanest single sentence anywhere on what good means, and it sits immediately next to the loosest use of the technical vocabulary in the cluster, which is a fair summary of the corpus as a whole: the handling is precise and the terminology is not.

Finn's lesson supplies the sorting rule the rest of the cluster assumes: every word you say to a dog falls into one of two boxes. A command produces a position or an action; a marker says something about what just happened. The trainer flags it as one of the most important parts and repeats it deliberately, and the same session carries the principle that decides what any of it is worth - whether a thing is reinforcement or punishment is decided by the dog rather than by the act. His worked example is a nervous golden retriever who ran from him on first meeting, stopped for a brushed chin, and asked for it again.

How many words do I actually have to learn?

Three to run the dog, and a fourth for behaviours you want gone. Eh the moment you dislike what he is doing. Good while he is doing what you asked and must keep doing it. Okay to release him. Then nope, for the things you never want to see again - jumping, counters, anything with a bite in it. One trainer gives the first three as a single list and tells the owner to drill them into her head, and the others build everything else on top of the same set. If you only remember one distinction, remember that good and okay are not the same word: good means carry on, okay means you are finished.

Why are there two correction words? Isn't one enough?

Because they mean opposite things and the dog has no other way to tell them apart. The collar tap is the same sensation whichever word precedes it, so the word is doing all the work: the soft one means you left a command, go back into it, and nope means that behaviour is not allowed at all. One trainer puts the difference in the dog's own voice - I need to go back into this, against I can't do it ever again. And the choice is a judgement about what happens next rather than about how cross you are. The example the trainers use is a dog who has just bitten another dog: the softer marker there would read to her as try again, harder.

What is the difference between 'good' and 'okay'?

Good is a duration marker and okay is a release. Good means you are right and you may not leave yet; okay means the job is over and you are free. This is the distinction owners break most often and it is broken by accident, usually by saying good job - which sounds like an ending, so the dog takes it as one. Say the short version, and say it in a flat voice rather than an excited one. A handler working alone gives the cleanest definition of it in the corpus while marking a heel: he went into the right place, but he is not allowed to come out of it yet.

Does the exact sound matter? I say uh-uh, not eh.

An owner asks precisely this, and the answer is that the tone matters more than the syllable. The trainer's advice is to stay close to what he uses, and then immediately relaxes it: as long as the tone is fairly similar it is fine. What is not negotiable is consistency and order - the same sound every time, and always before the collar rather than with it or after it. Be aware too that the sound is hard to write down: the transcripts of these very lessons render it as eh, ant, ash, ain't and uh-uh, which is a good sign that you should copy the sound from your trainer rather than from a page.

Can I use the vibrate instead of the stim? It seems gentler.

The trainers say no, and their reason is not about kindness. A marker works because the dog has learned that something follows it. Pair the word with a vibration the dog can feel but does not mind and you have taught him that the word predicts nothing - so you lose the word, and with it the ability to correct him at a distance without the collar at all. The instruction given to one family is exactly one button: the low one, every time. The red button, which raises the level sharply, is described as being for dogs that bite people.

Is the marker a warning?

An owner offers that word back to a trainer mid-lesson and it is not corrected, so it is worth being careful with. In the trainers' own definitions the soft marker is not a warning of something worse - it is an instruction pointing somewhere else: you have left a command, go back into it. If you treat it as a warning you will say it louder and then louder again, which is the undifferentiated stream the two-word system exists to prevent. The escalation the trainers actually describe is a change of word, not a change of volume.

Saying 'good job' where the trainers say 'good'. Good job sounds like the end of the exercise, so the dog gets up - and the owner reads it as a dog who will not hold a command.

Using one correction word for everything, so it turns into noise the dog stops hearing.

Reaching for the hard marker on something small. A nope for sniffing a table is named on camera as too much; a redirect does the job.

Using the vibrate button because it feels kinder. A marker followed by something the dog cannot feel teaches the marker to mean nothing, which costs you the word itself.

Getting louder instead of sharper. The marker is a change of tone, not of volume, and the coaching an owner gets is to add power without shouting.

Putting the collar before the word, which attaches the sensation to whatever else is in the room instead of to the marker.

You praise him for holding a position and he immediately gets up, and you cannot work out whether he is disobedient or you released him.

You hear yourself using the same correction sound for everything from a broken sit to jumping on a visitor - and he has stopped responding to it.

You do not know which of his behaviours are 'never do that again' and which are 'you left a command' - so you correct both the same way.

The correction looks the same to you in two opposite situations and you have quietly concluded that the collar can only really be punishment.