Why is a conversation harder to ignore than steady noise?

Why can you work next to a rattling radiator for an hour, yet lose your train of thought the moment two people start talking by the printer? The radiator may even be louder. Researchers have spent decades on the puzzle, and their answer is less about volume and more about what your mind can’t help doing with words.

Here are the questions people tend to ask about it, answered as plainly as the evidence allows.

Is it just that conversation is louder?

Not really. In one French lab study, every sound condition was played at the same level, 55 dB(A) at the listener’s seat. Performance still changed from one condition to the next. What changed was how easy the speech was to understand.

How do researchers measure “easy to understand”?

With a scale called the Speech Transmission Index, or STI. It runs from 0, speech you can’t make out at all, to 1, speech that’s perfectly clear. Researchers can move a recording along that scale by mixing it with background babble at different strengths. The voice is still there either way. What changes is whether the words come through.

So what happened when the words came through?

In the study by Krist Kostallari, Etienne Parizet and colleagues, 55 adults aged 25 to 59 did a working-memory task in silence and under four speech conditions, from STI 0.25 (mostly unintelligible) up to 0.9 (nearly crystal clear). The task was set around each person’s own memory span, so it was equally demanding for everyone. Clearer speech meant a bigger drop in performance. It also meant higher ratings of mental effort and discomfort.

An earlier study in Applied Acoustics by Helena Jahncke, Valtteri Hongisto and Petra Virjonen pointed the same way. As the later paper summarises it, people memorised lists of words while speech played at intelligibility levels from STI 0 to 0.7. Overall, intelligibility mattered.

Does clearer speech always mean more damage?

Here the story gets more interesting. Both studies found a ceiling of sorts. In the Jahncke work, performance at STI 0.35 and at 0.7 didn’t differ significantly. In the Kostallari study, the two clearest conditions produced about the same drop. That matches a model proposed by Hongisto, which predicts that once speech is around 0.7 or clearer, the drop in performance levels off at roughly 7 per cent.

In everyday terms, once you can follow what someone is saying, making it even clearer doesn’t add much. The harm builds up in the middle of the scale, as words move from mush to meaning.

Does it hurt every kind of work equally?

No, and that point is easy to miss. One leading explanation is overlap. If your task involves words or sequences, such as remembering a list, holding a sentence in mind or keeping steps in order, then speech nearby is competing for the same mental machinery. Steady noise has no words and no sequence to follow, so it gives that machinery very little to grab. A stream of understandable speech gives it plenty.

That also means the cost depends on the job. As its title says, Jahncke’s paper looked at office-task characteristics as well as intelligibility. A separate 2022 study in Memory & Cognition by Florian Kattner and colleagues makes the task-dependence very clear. Spoken letters that kept changing disrupted people’s ability to recall digits in order, but they didn’t produce a general drop in mental arithmetic.

So the same background can matter a lot for one job and barely at all for another. Proofreading, transcribing and drafting are the kinds of tasks that ask you to hold language or order in mind. Other work may be less touchy, though not immune: the Kattner paper notes that earlier studies did find speech hurting mental arithmetic, possibly because what was said grabbed attention.

Does everyone react the same way?

The averages hide real differences. In the Kostallari study, age was linked to performance on the memory task, and a cluster analysis split participants into a younger, more efficient group and an older, less efficient one. If a colleague seems unbothered by talk that drives you up the wall, neither of you is imagining it.

Can a steady sound fix an open-plan office?

The research doesn’t say that, and no honest reading of it should. These were lab tasks under controlled speech, not months in a real workplace. What the studies do show is narrower and still useful. Speech you can understand tends to cost more than speech you can’t, and how much depends on what you’re doing. Background babble was the tool the researchers used to make speech less intelligible, but that’s a lab method, not a promise about any office, headset or recording.

So what can you actually do with this?

A few modest, practical ideas follow from the findings:

• Save word-heavy work for quieter hours. If you can choose when to draft, edit or memorise, put it where the chatter is lowest. Save sorting and admin for the noisy stretch.
• Move a little further from the talking. In these studies, the less clearly the words came through, the less they cost, so anything that makes nearby talk harder to follow is worth a try.
• Ask, kindly. A short “I’m on something fiddly for an hour” often does more than any gadget.
• If you add sound, keep it voiceless. If you play something to work against, choose sound without words in it, such as plain noise or a recording with no voices, so you aren’t adding the very thing that costs the most.

Nothing here is medical advice, and no sound is a treatment for anything. It’s simply a summary of attention and memory research, explained and linked above, about why voices are harder to tune out than a steady hum.

In short, a conversation pulls at you because it carries words you can’t help following, while steady noise has nothing in it to follow. When the work is made of words, the clearer the talk around you, the more it competes, until it is fully clear and the cost stops climbing.

Sources

1. Jahncke, H., Hongisto, V., & Virjonen, P. (2013). “Cognitive performance during irrelevant speech: Effects of speech intelligibility and office-task characteristics.” Applied Acoustics, 74(3), 307–316. https://doi.org/10.1016/j.apacoust.2012.08.007
2. Kostallari, K., Parizet, E., Chevret, P., Amato, J.-N., & Galy, E. (2017; arXiv 2019). “Irrelevant speech effect in open plan offices: A laboratory study.” 24th International Congress on Sound and Vibration, London. https://arxiv.org/abs/1903.11386
3. Kattner, F., Hanl, S., Paul, L., & Ellermeier, W. (2023; online 2022). “Task-specific auditory distraction in serial recall and mental arithmetic.” Memory & Cognition, 51(4), 930–951. https://doi.org/10.3758/s13421-022-01363-6