בס״ד

How many words do you need to follow a sugya?

Everyone who learns Gemara knows the feeling of a line that won't open. You know most of the words, but two or three are missing, and without them the question, the answer and the proof all blur together.

How many words do you need to know before a text stops feeling closed? Reading researchers have asked this question for decades, mostly about people learning English. Their answers are useful for the Gemara, as long as we're careful about what carries over.

Coverage: the share of words you know

Researchers call it text coverage: the percentage of the running words in a text that a reader already knows. If you know 95 out of every 100 words on the page, your coverage is 95%.

The study behind the 98% figure

In 2000, Marcella Hu and Paul Nation tested readers of English as a foreign language, 66 university students in New Zealand. Everyone read the same fiction text. Some of its words were swapped for nonsense words, so the researchers could control exactly how many words each group didn't know: 80%, 90%, 95% or 100% coverage. Then they tested understanding.

Nation (2006) summarises what they found:

Fitting a model to the results, they estimated that 98% coverage, about one unknown word in fifty, was needed for most readers to understand the text adequately.

How many words is that?

In 2006 Nation used word-frequency lists built from the British National Corpus to ask how large a vocabulary gives 98% coverage of English. His answer: about 8,000 to 9,000 word families for written text, and 6,000 to 7,000 for spoken English.

He also showed how unevenly the work is spread. In the novel Lady Chatterley's Lover, the 1,000 most frequent word families of English made up 80.88% of the running words. The first thousand words get you a long way; each thousand after that adds much less.

For a learner, that means the order of the words matters as much as how many you learn. Start with the words that appear most often and every hour you spend opens more of the text. Start with rare words and you can learn hundreds of them while most lines stay just as closed as before.

A caution about the numbers

These are findings about English, from readers of English. A 2023 replication by Kremmel, Indrarathne, Kormos and Suzuki, with 104 adult learners in Sri Lanka, reported that the original results "could not be fully replicated." The 98% figure is a useful guide, not a law.

The Gemara is also a different kind of text. It mixes Aramaic and Hebrew, attaches small words to the front of others, uses technical terms of the argument, and is usually learned with a teacher, a study partner or a translation alongside. We don't know of a published coverage study of the Bavli, and we won't make up a number for one.

What carries over to the Gemara

Three points hold up well enough to use:

  1. A few small words do much of the work. In our own count of Masechta Shabbos, words like לא, אמר, אלא and ליה are near the top of the list. Knowing them well opens a large share of every line.
  2. One unknown word in five is too many. In Hu and Nation's study no one managed at that level. If every fifth word of a line is unclear, the line stays closed however hard you try.
  3. The last few percent take the most words. Rare words appear in one sugya and not the next. The quickest way to close that gap is to learn the words of the masechta you're actually learning, not a general list.

How Sugya.app uses this

You don't need every word to follow a sugya. You need nearly every word, and the right ones first.

Start learning · Which masechtos are included · For teachers

Sources

More