Are subtitles cheating? What 18 studies say about your French videos

The problem

You start a French series, hold out for three minutes without subtitles, then give in and switch them on. And there’s the little voice: “You’re cheating. You’re just reading.” You finish the episode with a mix of pleasure and guilt, and the feeling that your ear hasn’t done any work.

That voice is wrong on one precise point: the problem isn’t reading, it’s what you read, and when.

What the research found

Maribel Montero Perez and her colleagues at the University of Leuven gathered everything that had been published on video subtitled in the language being learned, what researchers call “captions” (Montero Perez, Van Den Noortgate & Desmet, 2013). They kept 18 studies and ran a meta-analysis, a calculation that combines the results of all those studies to measure one overall effect.

The effect is large. For listening comprehension, across 15 studies, captions in the target language produced an effect of g = 0.99, which statisticians class as a large effect. For vocabulary, across 10 studies, the effect was g = 0.87. And one detail matters for you: learners’ proficiency level made no difference to the result. Beginners benefited as much as advanced learners.

The mechanism is the same as in the previous article: captions cut the stream of sound into words. Your ear hears “jenesaipas” and your eyes tell it “je ne sais pas”. Over time, the ear learns to do that cutting by itself.

Another team pushed the logic in a different setting (Lee & Mayer, 2018). Korean university students followed a sixteen-minute video lesson in English, their second language, with narration only, on-screen text only, or narration and text together. In a first language, we know that text duplicating the voice gets in the way. Here it was the reverse: voice and text together beat voice alone, and students rated the lesson as less difficult. When you learn in a foreign language, text isn’t redundancy: it’s support.

Do this today

Pick a short French video, two or three minutes. First pass without subtitles: mentally note what you caught, even if it’s little. Second pass with French subtitles, never your own language. Watch for the moments when you think “ah, so that’s what it was”. Those are the words you knew without recognising them, and that’s where your ear grows. Two passes, five minutes, no guilt.

How our lessons use it

The video « L’ascenseur », on the HumanFrench YouTube channel, suits this method well: listen first without reading, try to answer, then listen again with the text on screen to see what you missed. Its shadowing version takes the story with pauses so you can repeat each sentence aloud, turning words you’ve seen into words you’ve heard and said. On the reading side, « Je m’appelle Nadia » trains the same reflex: read, hide, recall. Go and watch « L’ascenseur ».

The honest limit

The meta-analysis covers captions in the target language, mostly for English, in research conditions with tests straight after viewing. It says nothing about subtitles in your own language, nor about the effect over months. Lee and Mayer measured comprehension of scientific content, not language learning. And a group effect doesn’t guarantee your effect: try it, and keep what helps you.

Your questions