A mantra and an affirmation are built the same way. A short phrase, said to yourself, repeated on purpose. That similarity is why the two words get used interchangeably in most of what’s written about either, and it hides the fact that they are trying to do opposite things. An affirmation is a claim, and its content is the whole of it. A mantra is an anchor, and its content is close to irrelevant. In several of the traditions that produced mantras, the syllables were selected specifically so that they would not mean anything you could think about.
Which means the blended version most people end up practising is the worst of both. A phrase heavy enough with personal meaning that you can’t disappear into it, repeated often enough that you stop hearing it as a claim. This page covers what a mantra is in the traditions that built the practice, the mechanism that makes a meaningless sound more useful at midnight than an instruction to calm down, what the research does and doesn’t support, how to choose one, and how the two practices fit together in the last few minutes before sleep.
What a mantra is
The Sanskrit term is usually glossed as an instrument of thought, and that gloss holds up better than most translations of contemplative vocabulary. A mantra is a tool. You hold your attention on it, and the holding is the practice.
In japa, the repetition is counted on a mala, traditionally a loop of a hundred and eight beads, moved one at a time between the fingers. The count exists so that no part of the mind has to keep track of the count. Transcendental Meditation, which brought the practice to a very large Western audience from the 1960s onward, assigns each student a bija or seed syllable, chosen for its sound rather than its sense, and students are generally told not to look for a meaning in it. The Christian parallel is less often noticed and structurally identical: the Jesus Prayer of the Eastern tradition, a single line repeated on a knotted cord, often paired with the breath, and the rosary, which is a mala with different beads.
Traditions separated by continents and centuries converged on the same design. Something short. Something repeatable indefinitely without effort. Something semantically inert, or nearly so. And a physical counting device to remove the last decision. Even Émile Coué, working in a French pharmacy in the 1920s and describing his method in explicitly non-mystical terms, arrived independently at a string with twenty knots in it, as autosuggestion covers in more detail. When four unrelated lineages all reach for beads, the beads are probably load-bearing.
Why a meaningless sound works better than an instruction
Here is the part that makes mantra practice make sense, and it isn’t mystical.
You cannot stop thinking on command. Deliberately suppressing a thought tends to make it more available rather than less, which is the finding underneath most of what goes wrong at two in the morning. But while you can’t empty the verbal channel, you can occupy it.
Alan Baddeley’s model of working memory includes a component for speech-like material, sometimes called the phonological loop, which holds a small amount of verbal content by silently rehearsing it. It has a limited capacity, and it can be deliberately jammed. If you repeat an irrelevant word while trying to hold a phone number in mind, the number degrades. Psychologists call this articulatory suppression, and they use it as an experimental tool to knock out subvocal rehearsal.
Rumination is subvocal rehearsal. The nightly review of a conversation, the rehearsal of tomorrow’s difficult sentence, the argument you’re having with someone who is asleep: all of it runs as inner speech, in the same narrow channel. A repeated syllable competes for that channel directly. This is why om is more useful at midnight than stop thinking about it, and it is a mechanism rather than a metaphor.
It also explains the design constraint that otherwise looks like superstition. A mantra with rich personal meaning feeds the channel it’s supposed to occupy. You start turning the phrase over, examining it, disagreeing with part of it, and the loop is busy with exactly the kind of content you were trying to displace. Semantic emptiness isn’t spiritual affectation. It’s the specification.
That sample uses so on the inhale and ham on the exhale, a traditional breath-paired pair. Nothing about it requires belief, which is the point.
What the research supports, and where it stops
Two findings are worth knowing, and one limit.
The first is where the modern interest began. Robert Keith Wallace and Herbert Benson studied Transcendental Meditation practitioners in the early 1970s and reported a coherent set of physiological changes during practice: lower oxygen consumption, slower heart rate, a pattern they described as a wakeful low-metabolism state distinct from sleep. That work is what moved mantra practice from religious context into cardiology departments.
The second finding matters more, and it came from Benson himself. He went on to test whether the assigned syllable was necessary, substituted the English word one, and got the same response. He generalised the practice into four components: a quiet place, a comfortable position, a repeated mental device, and a passive attitude toward whether it’s working. Nothing about a particular sound turned out to be doing the work. Anyone selling you a personally-calibrated mantra is selling you the packaging.
There is also a deflation worth accepting cheerfully. Bernardi and colleagues reported in 2001 that reciting the rosary in Latin and reciting a yoga mantra both pulled participants’ breathing toward roughly six breaths per minute, with the cardiovascular changes that slow breathing produces. The cadence of the phrases did it. Which means part of what a mantra does is a breathing exercise you didn’t know you were performing, and the six-per-minute figure is the same one that turns up in the vagus nerve literature and behind nearly every technique in breathwork. That isn’t debunking the practice. It’s identifying one of its ingredients, and it suggests choosing something you can say slowly.
The limit is the formal evidence base, and it should be stated plainly rather than buried. Goyal and colleagues reviewed meditation programs in 2014 and concluded that the evidence for mantra-meditation programs was insufficient to draw conclusions, in the same review where mindfulness-based programs showed moderate evidence for anxiety, depression, and pain. The honest reading is not that mantras don’t work. It’s that they have been studied less, and less rigorously, than the confidence of the surrounding literature implies. The mechanism is solid. The trial evidence for the packaged programs is thin.
Choosing one
Three options, all defensible.
A traditional syllable, if the lineage means something to you: om, so ham, sat nam. The sense of occasion is real and it helps some people practise consistently, which is worth more than any property of the sound.
A neutral English word, following Benson: one, ease, here. Unglamorous, and no less effective by the only measurement anyone has made.
A breath-paired pair, which is what the sample above uses: one syllable arriving, one leaving. This is the version to prefer at bedtime, because it borrows the slow-breathing mechanism at the same time.
The single rule that matters: if you can argue with it, it’s an affirmation. I am calm is a claim, and at eleven at night part of you will produce evidence against it. Ease, repeated, makes no claim and offers nothing to dispute. The calibration problem that positive affirmations has to solve carefully does not exist here, because a mantra never asserts anything.
The sequence at night
Both practices belong in the last twenty minutes of the day, and they do not belong at the same moment.
A mantra is the settling tool. It asks nothing of you, competes with the inner monologue directly, and slows the breath as a side effect. That makes it the right thing for the stretch where you’re still winding down and your thinking still has momentum, which is the territory overthinking at night and how to fall asleep fast both describe. Run it for as long as it takes, without monitoring whether it’s working. Monitoring is its own kind of arousal.
An affirmation is the closing tool, and it wants the opposite conditions. It has to be understood to do anything, so it belongs at the threshold, in the drowsy minutes near the hypnagogic state, where a sentence still lands but the part of you that would litigate it has gone quiet. One line. Specific, believable, and left alone afterwards. That’s the whole practice described in sleep affirmations.
So: mantra to get there, affirmation once you’ve arrived. Trying to do both at once produces the blended version that fails at both jobs.
Two adjacent techniques are worth distinguishing while we’re here. Cognitive shuffling also occupies the verbal channel, but it scatters attention across unconnected images where a mantra flattens it onto one point. Both work; they suit different temperaments, and the scatter is better if a single repeated sound makes you restless. Mindfulness meditation for sleep uses an open anchor, usually the breath or the body, and asks you to notice what arrives. A mantra uses a fixed anchor and asks you to keep returning to it. If open awareness has consistently left you more alert rather than less, a fixed anchor is often the better fit, and that’s a preference rather than a hierarchy.
The smallest version of this
Tonight: pick one syllable and don’t deliberate. One is fine. Say it on the exhale, silently or barely voiced, slower than feels natural. When you notice you’ve stopped, start again without commentary. Do that until you lose track of doing it.
Then, if you want the second half, one sentence at the threshold, chosen in advance so you aren’t composing anything at midnight. I did enough today. This can wait until morning. Something small enough that no part of you bothers to object, which is the calibration the whole cluster keeps returning to, including in worry and in sleep hypnosis.
Give it a fortnight before deciding anything. The first few nights with a mantra usually feel faintly ridiculous, and the ridiculousness passes.
That second sample is the handover point. The sound going indistinct is the practice working, not you losing it.
How Murmora fits
We’re a sleep affirmations company, so we sell the second half of this and not the first. Worth being direct about that. A mantra is free, needs no app, no voice, and nothing from us, and if the only thing you take from this page is one syllable on the exhale, that’s a complete practice and you should keep it.
Where the product argument is honest is the other half. An affirmation has to be specific enough to be yours and modest enough to go unchallenged, which is difficult to write for yourself and is what Murmora actually does: lines drawn from what you describe about your own situation, spoken in one of a few guide voices and later in a clone of your own. The sparse-whisper pacing is relevant here too. The gaps between lines aren’t dead air. They’re where a mantra, or nothing at all, is supposed to go.