Comprehensible Input for Tamil: What the Research Says

The research behind "learn a language from input" comes down to four findings:

1. You acquire language mainly by understanding it. Language you can mostly follow, with a little that's new, is what drives acquisition (Krashen's input hypothesis).
2. "Mostly" means a lot. For reading, learners needed to know about 98% of the words for unassisted comprehension. At 80%, nobody in the study understood adequately.
3. Spread your reviews out. Spaced practice beats cramming, consistently, across hundreds of experiments.
4. Testing yourself beats rereading. Pulling a word out of memory strengthens it far more than looking at it again.

Here's what each finding actually says, where the evidence is weaker than its fans claim, and what it means for learning Tamil.

1. The input hypothesis (Krashen, 1982)

Stephen Krashen's book Principles and Practice in Second Language Acquisition (1982; free PDF from the author) set out the idea that people acquire a language subconsciously by understanding messages in it, and that this is different from consciously learning rules. His shorthand was i+1: input slightly beyond your current level i, made understandable by context.

Two parts of this hold up well and matter for Tamil:

• Understanding comes first. You can't acquire from noise. A Tamil film you follow 20% of is entertainment, not input.
• Context does the teaching. Pictures, situations, gestures and what you already know let you work out new words without a dictionary. That's why cooking videos and kids' cartoons are so effective early on: the screen shows you what the words mean.

The honest caveats

The input hypothesis is influential, but it's a hypothesis, and it has been criticised for decades. Some critics argue it's hard to test, because "comprehensible" and "acquired" are difficult to measure independently. The best-known alternative, Merrill Swain's comprehensible output hypothesis (1985), argues that producing language (speaking and writing) also plays its own role, because trying to say something exposes the gaps in what you know.

Our practical reading: input is the foundation, but adults also benefit from speaking practice and from deliberately reviewing words. Nothing below requires you to pick a side.

2. How much do you need to understand? (vocabulary coverage)

"Comprehensible" turns out to mean very comprehensible.

Hu and Nation (2000) gave learners fiction texts with different proportions of known words. Where 80% of the running words were familiar, none of the readers gained adequate comprehension. At 90% and 95%, some did but most didn't. The authors concluded that around 98% coverage is needed for unassisted comprehension of a fiction text. It was a small study of learners of English, so treat the exact figure with care, but the direction is clear: one unknown word in every five is far too many.

For video, Webb and Rodgers (2009) analysed 88 English-language TV programmes and found that the most frequent 3,000 word families (plus proper nouns and marginal words) covered about 95% of the words, while 98% coverage needed about 7,000 word families.

What this means for Tamil:

• These studies are about English. We don't know of equivalent coverage studies for spoken Tamil, and because Tamil builds long words out of a root plus endings (வீட்டுக்கு (veettukku) ‘to the house’ is வீடு (veedu) ‘house’ + "to"), word counts don't transfer directly.
• The practical rule still applies: choose material where only a few words per sentence are new. If you're pausing every few seconds, drop down a level. Start with rhymes, kids' stories and cooking, then vlogs, then interviews. Learn Tamil with YouTube has suggestions at each level, and the level check can tell you where to start.
• Heritage learners have a real advantage here. If you grew up hearing Tamil, a lot of everyday speech is already near the coverage you need, even if you can't speak much yet.

Hear it in real TamilWatch a 20-second clip with tap-to-translate captions and see how Bumi will work.

3. Video helps vocabulary, especially repeated words

Peters and Webb (2018) had learners watch one full-length TV programme in a foreign language. Viewing led to incidental vocabulary learning, both recognising and recalling word meanings. How much was learned depended on how often a word occurred, the learner's existing vocabulary, and whether the word resembled one in their first language.

Two takeaways for Tamil:

• Frequency matters, so rewatching a clip and choosing genres with lots of repetition helps.
• Prior vocabulary matters, so a small base of core words (family, food, everyday verbs) makes every later video more productive. Lists like everyday spoken Tamil phrases are worth the effort early on.

English speakers get little help from cognates in Tamil, since few native Tamil words resemble English ones. The exception is the many English loanwords in spoken Tamil (பஸ் (bus) ‘bus’, டாக்டர் (daaktar) ‘doctor’), which make modern speech easier to follow than you might expect.

4. Spaced repetition (Cepeda et al., 2006 and 2008)

The spacing effect is one of the most reliable findings in memory research. Cepeda and colleagues (2006) combined 839 assessments of distributed practice from 317 experiments in 184 articles. Spacing study sessions out produced better long-term recall than massing them together, and the best gap between sessions grew as the time you needed to remember grew.

A follow-up with more than 1,350 people (Cepeda et al., 2008) varied the gap up to 3.5 months and tested up to a year later. The optimal gap was about 20–40% of the test delay when the test was a week away, falling to about 5–10% when the test was a year away.

What this means for Tamil: review a new word the next day, then a few days later, then a week or two later, and so on, with gaps that lengthen as you get it right. Flashcard apps with spaced-repetition scheduling do this arithmetic for you. Cramming fifty words the night before a trip to Chennai will get you through the trip and very little after it.

5. Retrieval practice (Karpicke & Roediger, 2008)

In a study published in Science, Karpicke and Roediger (2008) found that repeated studying after learning had no effect on delayed recall, while repeated testing produced a large positive effect. Strikingly, students' predictions of how well they'd do were unrelated to how well they actually did. Rereading feels productive; it isn't, compared with testing yourself.

What this means for Tamil: your flashcards should make you produce the answer before you flip. Look at சாப்பிட்டீங்களா? (saappitteengalaa?) ‘have you eaten?’ and say the meaning out loud before checking. Better still, look at "have you eaten?" and produce the Tamil.

Putting it together: an input-first Tamil routine

1. Mostly input, at the right level. Short videos you can largely follow, rewatched rather than skipped through. Spoken Tamil, not only textbook Tamil, because the two differ a lot (spoken vs written Tamil explains why).
2. A few words per session, chosen from what you watched. Prefer words that came up more than once.
3. Spaced, test-style reviews of those words, with gaps that lengthen.
4. Some output. Shadow lines from your clips, and use new phrases with family or friends. That covers Swain's point without giving up the input foundation.
5. Move up when it gets easy. When a level feels comfortable, the input is no longer stretching you; go up a level, eventually to films with English subtitles.

This is the method Bumi is being designed around: short real Tamil clips sorted by level, every word one tap away, and reviews scheduled just before you'd forget. It's not available yet; it's in development, and founding members get early access.