A timer changes reading. It can push a learner past word-by-word translation, and it can also reward skipping half the page. The difference lies in the text and the comprehension check.
Timed reading should use language the learner already knows well. For a Taiwanese university student who reads English accurately at 84 words per minute, the target is quicker access to familiar words and sentence patterns. A dense article full of unknown vocabulary measures survival under difficulty. It gives poor fluency practice.
Speed practice has narrow conditions
Chung and Nation ran a 9-week speed-reading course with 49 Korean university students. The texts stayed within a controlled vocabulary, and learners recorded both speed and comprehension. Almost every learner increased reading speed, with most gains appearing in the first 10 texts and further gains across at least 20. The authors' 2006 paper and materials emphasise familiar vocabulary because lexical problems interrupt the process being trained. That result comes from a course, not a magic timer. Learners read many comparable texts, charted progress and answered questions. Across those repeated sessions, the design trained faster processing of known language while holding vocabulary difficulty steady enough that a slow score could still point to fluency rather than a cluster of new words.
The timer is secondary.
Torrin Shimono later compared timed reading, timed reading plus repeated oral reading, and oral communication training with 55 Japanese university students over a semester. The two reading groups gained 13 to 27 standard words per minute. Comprehension percentages sat slightly below 70%, though they rose during treatment. The 2018 study was quasi-experimental, and the two reading treatments did not differ significantly from each other. Those limits matter.
Unknown words change the task
Suppose a 640-word passage contains 19 unfamiliar items, three long embedded clauses and a topic the learner has never studied. Pressing a stopwatch may increase eye movement. It cannot make the vocabulary available. The learner may omit hard sentences, infer from headings or trade accuracy for a flattering number. Fluency work needs a high level of text coverage. A practical classroom check is more direct than a universal percentage: the learner should understand the passage on the first read without a dictionary and answer most meaning questions. If every paragraph requires a lookup, choose another text.
Easy text has a job.
Easy is purposeful here.
Our article on reading several English texts about the same topic explains how repeated subject matter can free attention for vocabulary and reading. Timed work applies a related principle in shorter pieces. The text should feel almost too easy at the start.
Keep comprehension beside the clock
A raw words-per-minute score can reward the wrong behaviour. Pair each passage with five or six questions that sample its main claim, sequence and a few details. Record the time only when comprehension reaches a preset floor, perhaps 70% or 80% depending on the material and purpose. The floor is a guardrail rather than a certificate of understanding. Four correct answers out of five can hide lucky guesses. Change the question types across sessions. Ask for one event order, one reference such as what they refers to, one short paraphrase and two factual details. On every fourth passage, request a 23-word summary without reopening the text.
Meaning stays on the graph.
John Macalister studied whether gains on controlled passages transferred to authentic texts. His 2010 preliminary investigation suggested that a speed-reading course may help learners read other texts faster, while calling for stronger experimental evidence. Transfer deserves a check, not an assumption. A learner can improve at a familiar workbook routine and remain slow on a GEPT article.
Progress is rarely a straight line
Topic knowledge, fatigue and sentence structure move the score. A passage about semiconductor packaging may be easy for an engineer and slow for a hospitality student, even when both know the vocabulary. One bad Tuesday says little. Use a run of comparable texts and inspect the median of the latest five readings. Keep a small graph with words per minute and comprehension together. If speed rises from 84 to 117 while comprehension stays near 80%, the change is plausible. If the score jumps to 193 and comprehension falls to 33%, the learner found a faster way to miss the text.
Expect noise.
I'd rather see a modest line that keeps meaning than a dramatic line built from skimming. Some learners also dislike visible timing. They can time privately, reveal only a weekly median or work against their own earlier score. Public ranking adds a social task that reading fluency research did not test.
A fifteen-minute routine
Choose a passage of roughly 500 to 700 words at the learner's comfortable level. Preview only the title. Start the timer, read once and stop it at the final word. Hide the passage before answering six comprehension items. Record speed and accuracy, then review the two sentences that caused the most delay. Do this twice a week for 6 weeks with texts drawn from the same difficulty band. Avoid rereading the same passage for the timed score because familiarity changes the measure. Repeated oral reading can be a separate activity, as Shimono's study shows, with phrasing and pronunciation as part of the purpose.
Stop at fifteen minutes.
After every eighth controlled text, use one authentic piece from the learner's real goal: a GEPT reading, a company memo or a university announcement. Do not compare its raw speed directly with the easy series. Ask whether the learner now loses the thread less often and finishes with enough time to answer. On Thursday morning, the useful number is 112 words per minute with five answers correct, written beside the title of a text that felt easy.