Does passive listening work? What the research says about hearing English sounds Playing English in the background feels like progress. For the sounds your first language merges, studies point to something else.

Why background listening doesn't teach merged sounds

When two English sounds fall into one sound of your first language, your brain files them together before you're aware of hearing them. That's the main idea of why some English sounds sound the same to you.

Background listening runs straight into that filing. Every time you hear right, your ear sorts it into the same box as light, and nothing tells you the sorting was wrong. Hours of it give your brain more practice at the habit you want to change.

To change the sorting, your brain needs a signal that the two sounds are different. Studies have found two ways to give it that signal: a real person to interact with, and answers you can check.

Even babies didn't learn from recordings

If anyone should learn from passive listening, it's babies, whose ears are still tuning to the sounds around them. Kuhl, Tsao and Liu (2003) tested this with 9-month-old American babies and a Mandarin sound distinction that English doesn't use.

• One group had 12 sessions with native Mandarin speakers who played and read with them. Afterwards, they told the Mandarin sounds apart better than babies who had heard only English.
• Another group got the same Mandarin material from a video or an audio recording. They learned nothing measurable.

The researchers concluded that at this age, learning speech sounds depends on live interaction, not just on hearing the sounds. For parents, it's a useful caution: an English audio track in the background isn't the same as an English-speaking person. The study tested babies, so it doesn't tell us exactly how older children learn.

Twelve years in the US helped, but didn't finish the job

Living in an English-speaking country is far more than background listening: you hear English all day and need it to get things done. Takagi and Mann (1995) asked how far that takes Japanese adults with R and L.

They compared three groups of 12: native English speakers, Japanese adults who had lived in the US for 12 years or more, and Japanese adults who had been there less than a year. The long-time residents identified R and L better than the newcomers, but still fell short of the native speakers.

The researchers concluded that long exposure can make you more accurate without making you native-like. If a decade of living in English doesn't reliably finish the job, background audio is unlikely to.

Feedback is what made the difference

McCandliss and colleagues (2002) trained Japanese adults to identify English R and L, using pairs like rock and lock, and changed two things: whether learners were told if each answer was right, and how hard the sounds were.

The top-left box is the closest to background listening: hearing the hard version of the sounds over and over with no way to check yourself. It did nothing. Adding feedback to the very same sounds made it work, and the gains carried over to a new pair of words (road and load).

The bottom row has a second lesson: starting where the difference is easiest to hear helps. That's why good practice starts with the clearest pairs and moves to harder ones.

Try listening with feedback

Here's what listening with feedback looks like. Tap play, pick the word you heard, and you'll see right away whether you were right. Each word is a real clip from the app, in a different voice.

The free listening test does the same for L and R, B and V, and TH and S, and shows which ones you already hear.

What works instead: high variability phonetic training

The training method with the most research behind it combines both lessons above. High variability phonetic training (HVPT) was developed in the early 1990s for Japanese learners of English R and L (Logan et al. 1991):

• You hear one word and pick which of two words it was.
• You find out at once whether you were right.
• Each word comes in a different voice, so you learn the difference all speakers share, not one speaker's habits (Lively et al. 1993).

A 2025 meta-analysis of 79 studies found medium-to-large gains in perception (Uchihara et al. 2025), and in one study Japanese learners were still above their starting point 6 months after 3 weeks of training (Lively et al. 1994). There's more on how it works in the first post.

Where background listening still helps, and a simple plan

None of this means you should stop listening to English. Fast speech, words running together and new vocabulary are big reasons people lose sentences, and the only way through those is a lot of real listening. Background audio makes it easy to get the hours in.

The trick is to give each kind of listening its own job:

• A few minutes a day of active ear training, with feedback and many voices, for the sounds your first language merges. The pair pages list them.
• Listening you pay attention to, with material you mostly understand. Replay a sentence you missed and check the transcript.
• Background listening for the rest of the day, for the speed and rhythm of English, without expecting it to fix specific sounds.

TellApart is built for the first job: you hear one word in one of many native voices, tap which of two words it was, and see right away whether you got it.