Article

Office chatter while you work: why speech costs more than noise

In an open-plan office, what costs you is speech you can make out, not volume. A 2020 review of 14 laboratory studies found performance falling as soon as speech becomes partly intelligible, with the full cost reached less than halfway up the intelligibility scale. So the target is the words, not the decibels. This article covers what that research measured: why intelligible speech is priced differently from noise, which of your work takes the damage, and what a quiet room is actually worth when you can get one.

Why does overheard talk cost more than general office noise?

Two desks over, somebody takes a call. The room is no louder than it was a minute ago, and the sentence you were building has gone. Haapakangas, Hongisto and Liebl set out to price that in a 2020 review in Indoor Air, gathering laboratory experiments that varied one thing on purpose: how intelligible the background speech was. They measured it with the Speech Transmission Index, an index running from 0, where nothing said can be made out, to 1, where every word is clear.

Fourteen studies and 34 separate tests later, the model they built puts the damage on a curve. Performance starts dropping once intelligibility passes about 0.21, which is speech you can only half make out. It reaches its maximum decrease at 0.44, and 0.44 is nowhere near clear. Past that point, talk getting easier to follow costs you nothing further, because the cost has already arrived.

That curve is the practical part.

It explains why turning a conversation down often changes nothing you can feel, while moving the same conversation behind a door works at once. Volume reaches you only through intelligibility, and a room can measure quiet while still delivering every word of a nearby call. It is also the logic behind sound masking, the steady background hiss some offices add on purpose: the point is not to make the room quieter but to stop the speech being decipherable. The review's own experiments manipulated intelligibility that same way, with wide-band steady noise.

Read the scope before you carry the numbers around. These were laboratory experiments on short tasks, not a fortnight at your desk, and the paper revises an earlier model that rested on scarce data, closing with a request for more work on task-specific effects.

Take the direction, not the decimals. What travels is the ranking: intelligible speech first, unintelligible sound a long way behind it. If the sound bothering you is sound you chose, a playlist rather than a colleague, that is a different question and this site's article on music at work takes it.

Which of your work does chatter actually damage?

Not all of it, and the pattern is specific enough to plan around. Verbal short-term memory took the strongest and most consistent hit in the 2020 review, and for those tasks the decline ran across a wider and earlier band than the all-task model: from an intelligibility of 0.12 to 0.51. Verbal working memory showed some evidence of the same relation. Complex verbal tasks showed limited evidence, which is the review's own wording and worth keeping. Words are what the speech goes after.

Holding words in order is the operation being disrupted, so your work at risk is the work made of words held in order. Reading a contract clause and carrying it into the next one. Drafting a reply where your third sentence has to follow from your first. Checking one list of names against another. None of that is exotic. For most desk jobs it is the difficult hour rather than the edge case, and it is where your day's real output tends to sit.

A 2024 experiment in the International Journal of Psychophysiology ran one of those tasks under conditions close to an office. Radun and colleagues had 30 people work through n-back and serial recall tests in two sound conditions: irrelevant speech at 50 dB, and quiet at 33 dB. Accuracy on serial recall fell under speech. The same people also rated the work as more annoying and rated their own workload as higher, which is the part you can notice at your own desk without a test.

Effort went up while output went down.

The field measurement agrees on direction, and the gap is wider than most people expect. Jahncke and Hallman tested 113 office employees across five offices in 2020 on a concentration-demanding task, before their organisations rearranged the offices. Those working in shared or open-plan areas scored 14 percent worse than those in cell offices, and the cell offices measured 15 LAeq quieter.

One field study is one field study, and nobody assigns people to office types at random, so read that 14 percent as the distance between two groups rather than as what a wall would do for you.

If your afternoon is arithmetic, mechanical checking or anything that does not run on language, this evidence is quieter about you. The effects cluster on verbal tasks, and the review asks for more research on task-specific effects rather than declaring everything else safe. That is the honest limit. Nobody can hand you a figure for your own desk.

Is it worth booking a quiet room?

For work made of words, the one field study here says yes, and by more than the booking form costs you. When Jahncke and Hallman followed the employees from four of those five offices into an activity-based workplace, the same people scored 16.9 percent better on the concentration task in the quiet zone than in the active zone that had no noise restrictions. An individual working room measured better still, at 21.9 percent.

Those are within-person comparisons, which is what makes them usable to you. Same employee, same task, one building.

So treat the room as a tool with one job rather than as a perk you have to justify. Book it for the hour of word-work you already know is coming: the clause you have to read closely, the reply that has to land correctly the first time, the two lists that have to be reconciled. Leave the mail pass and the phone calls at your desk, where chatter costs least. That is the sorting Task Batching does with the clock, applied to the room instead.

Most weeks you will not get a room, and the curve from the first section is what is left to work with. It gives you a single instruction: go after intelligibility, not volume.

Distance does that, because speech gets harder to make out as it travels. Steady sound laid over the top of it does that, which is what masking is for. Moving the reading and the drafting to the hour when the room thins out does it by avoidance. None of the three studies tested a desk arrangement, so take those as the mechanism applied rather than as measured results.

Headphones deserve their own line, because they are what everybody reaches for first. No study cited here tested them. On the mechanism they should help when they leave you unable to make out words, and help much less when they only lower the room while the call at the next desk still comes through. If what you put on instead is a playlist with lyrics, you have added a second stream of language to work that was already losing to the first.

How does your working style change the answer?

None of the three studies sorted anybody by working style, so nothing in this section is a research finding. Architect, Sprinter, Visionary and Improviser describe how somebody arranges work, not how their attention behaves in a loud room. What follows is this site's reading of how the evidence meets each style's ordinary week.

If you already defend long blocks, which is the Architect's habit, a room booking is the cheapest upgrade on offer. You are scheduling the block anyway, so adding a location to it costs you one more field. Deep Work asks for 90 to 180 minutes of protected time, and protecting that from your own calendar while leaving it open to the next desk's call only does half of the work.

Burst workers, closest to the Sprinter, often get from the room's own rhythm what a booking would buy them. Offices have lulls: early, late, the half hour a large meeting empties the floor. A Pomodoro Technique round runs 25 minutes, short enough to fit inside a lull that a 90-minute block would miss completely. Put the word-work in the lull and the calls in the noise.

The Visionary's exposure is different, because the work at risk never looks like a task. Thinking a problem through on paper is words held in order just as much as a contract clause is, and it gets done at the desk by default, because nobody books a room for thinking. Give it the quiet hour anyway. It asks least and loses most.

If your week never looks the same twice, the Improviser's situation, no fixed arrangement will survive it and a rule about which work goes where will. One line covers it: reading and drafting move to the quietest thing available, everything else stays at the desk. That rule travels between a home day, an office day and a client site without needing to be rebuilt each time.

Whichever of those sounds like you, the test is the two weeks this site puts on any new method. Move one category of work, the reading and the drafting, into the quietest hour or room you can get reliably, and leave the rest exactly where it is.

Write one line at the close of each day about what you moved and what actually got finished. Read the lines, not your memory.

If you are not sure which style is yours, the working style assessment takes a few minutes. And if the thing breaking your attention turns out to be your own hand reaching for the phone rather than the room next to you, the focus pillar sorts the other causes.

Frequently asked questions

How do you concentrate in a noisy open plan office?

Go after intelligibility rather than volume: what costs you is speech you can make out, and a room can measure quiet while still delivering every word of a nearby call. A 2020 review of 14 laboratory studies found performance dropping once speech passes about 0.21 on the Speech Transmission Index, with the full cost reached by 0.44, well short of clear. Distance, steady masking sound and a quieter hour all work on that one variable. Moving your demanding word-work into a quiet room is the version that has been measured.

Why is overhearing one side of a phone call so distracting?

Because a call at the next desk is usually the most intelligible speech in the room, and intelligibility is what the evidence prices. The 2020 review found verbal short-term memory hit most strongly and most consistently, across an intelligibility band running from 0.12 to 0.51. Work that holds words in order is the work being disrupted. None of the studies cited here tested one-sided calls specifically, so treat the clarity of the speech as the mechanism rather than the missing half of it.

Do noise-cancelling headphones actually help you focus at work?

None of the research cited here tested them, so the honest answer is that it depends on whether they stop you making out words. On the mechanism, headphones help when they leave speech unintelligible and help much less when they only lower the room while a nearby call still comes through. Music with lyrics is the version to avoid for reading and drafting, because it adds a second stream of language to work that is already competing for the first.

Is it worth booking a quiet room to get one task done?

For work made of words, the one field study here says yes: the same employees scored 16.9 percent better in a quiet zone than in an active zone with no noise restrictions. An individual working room measured 21.9 percent better, in the four offices that moved into an activity-based workplace. Those are within-person comparisons, the same people in different parts of one building. Book it for the reading and the drafting, and leave mail and calls at your desk.