A learner who can decode every word in a paragraph and still reads it as a string of separate items needs something other than more new material. Reading the same short passage aloud three times builds the automatic word recognition that frees attention for meaning. Extensive reading gives you volume. Repeated reading gives you speed on text you can already handle, and the two kinds of practice feed different needs. Most learners only ever do the first kind, which is why a reader can recognise every word in a sentence and still finish it without knowing what it said.
Where the method comes from
This idea is old and precise. S. Jay Samuels introduced the "method of repeated readings" in a 1979 issue of The Reading Teacher as a way to build automaticity. A reader rereads a short passage until a set fluency mark is reached, then moves to the next passage. The reasoning follows LaBerge and Samuels' automaticity model: a reader has limited attention, and every unit of it spent decoding a word is a unit not spent on what the sentence means. Once word recognition turns automatic, the reader stops spending attention on letters and can hold the meaning of the whole line.
Reading aloud matters here for a reason. Silent reading lets a learner skip a hard word or glide past a phrase without fully decoding it. Reading aloud removes that exit, because the mouth has to say something, and the ear then hears what the mouth did.
For a first-language reader this is remedial work. For a second-language reader it is a fluency tool, and the evidence is direct. Taguchi, Takayasu-Maass and Gorsuch compared assisted repeated reading, where learners reread while a recorded voice models the passage, with an extensive reading programme for Japanese university learners. The 2004 study in Reading in a Foreign Language found assisted repeated reading matched extensive reading on fluency and comprehension gains, while adding the rhythm and phrasing a live model supplies.
What the studies found
Gorsuch and Taguchi then took the method to Vietnamese university EFL learners. Their repeated reading programme raised both reading rate and comprehension over the course of the study, and the 2008 paper in System reports the improvement alongside the usual caution that the groups were small and the setting specific. A 2010 follow-up in Language Teaching Research tracked learners' own accounts across the programme and found the gains in rate held while learners reported reading with more confidence. The common thread is that rereading the same passage, rather than always moving to new text, gives the decoding system repeated passes until it stops being the bottleneck.
The method is simple enough to do badly. Rereading works when the passage is short and already mostly known. If a learner picks a paragraph full of new words, the rereading becomes a decoding slog, and the automaticity never has room to appear. The sweet spot is text where the learner knows around nine of every ten words, which is also the range where reading feels neither easy nor blocked.
How to run it
For a Taiwanese learner, this slots into a specific gap. A student aiming at the 學測 or 多益 reading sections can recognise the words but runs out of time, because every line still gets read word by word. A 90-second news or textbook paragraph, read aloud three times with the goal of keeping phrase boundaries and stress, trains the exact skill the timed section rewards. Three readings of one passage do more than one reading of three passages, because the third reading is where the phrasing comes loose. Keep the three readings in one sitting at first. Splitting them across a day defeats the point, which is to feel the third reading arrive while the words are still warm.
The third reading is the point. On the first pass the learner spends attention on words. On the second the words settle and the reader starts grouping them. On the third the phrasing, the stress and the sentence rhythm can come to the front, because the individual words no longer demand the whole budget of attention.
Pick a passage of 90 to 120 words the learner can mostly read. Have the learner read it silently once, then aloud, then record the third reading. Listen back for one thing only, the places where the voice paused inside a phrase instead of at a boundary. Fix those, read once more. The routine takes six minutes and needs no new material.
What it does not build
What repeated reading does not build is the honest boundary. It does not grow vocabulary. A learner who rereads the same 100 words is not meeting new ones, so repeated reading sits beside extensive reading rather than above it. For building the word knowledge that makes new text readable, see our article on easy books as serious reading practice. Use both: extensive reading for volume, repeated reading for automaticity on what you already know.
A learner who can already read fluently gains little from rereading the same passage and would be better served by more new text. The method targets the intermediate reader whose decoding is accurate but slow, which describes a large share of Taiwanese learners at the 全民英檢中級 to 中高級 stretch.
Tomorrow, take one paragraph from the reading material you already study. Read it silently, then aloud, then a third time while recording. Circle the two places the voice stalled. The next day, read the same paragraph once more and check whether those two places have smoothed. That single paragraph, four short readings, is the whole assignment.