Why is reading silently faster than reading aloud?

The oral estimate drew on 77 studies and 5,965 participants. The silent and oral pools differ in texts, instructions and participants, so subtracting 183 from 238 does not establish that every reader gains 55 WPM by being silent. Nor does a group mean define a good speaking voice, a fluency diagnosis or a performance target. Brysbaert, 2019

Speaking introduces an additional output task

Reading aloud coordinates visual word recognition, language processing and articulation. Silent reading still processes language, but it does not have to wait for each word's physical production. Eye-movement reviews describe the interaction between visual sampling and language processing; neither task is merely moving the eyes across a line. Rayner, 1998 CLIFTON et al., 2015

This distinction does not imply that silent reading has no sound-related representations. The speed-reading review explains why removing inner speech is not a demonstrated general route to extremely fast comprehension. The absence of audible speech and the absence of phonological processing are different claims. Rayner et al., 2016

Which method produces better comprehension?

The two rate means cannot answer that question by themselves. Comprehension depends on the material, purpose and reader. An experiment on preventing rereading shows that returning to words can support understanding; a rigid instruction to maintain a fast rate may remove useful processing time in either mode. Schotter et al., 2014

Why age and language change the comparison

The review finds different rates for children, older adults and second-language readers. Word knowledge and language experience influence recognition, and second-language research shows that frequency effects depend on a reader's experience with each language. A learner's oral and silent rates therefore need not differ in the same way as the pooled English adult means. Brysbaert, 2019 Whitford et al., 2012

Vocabulary recognition is also not the same as fluent spoken production. Estimates of words known depend on the counting unit, while reading coverage depends on which words occur in the particular text. These are reasons to avoid using one oral WPM figure as a general language-proficiency verdict. Brysbaert et al., 2016 Nation, 2006

What is realistic to practise?

Choose a mode that matches the task: presenting aloud and reading for private understanding involve overlapping but different skills. Compare like material with like material and check comprehension. The reading reviews support task-specific interpretation, not a universal recommendation that everyone should always read in one mode. Brysbaert, 2019 Rayner et al., 2016

For an informal comparison, use unfamiliar passages of similar difficulty and avoid reading the same passage aloud immediately after silently reading it. Familiarity would confound the comparison. Regard the result as an observation about those passages, rather than a validated test of your underlying intelligence or a promise about all books. CLIFTON et al., 2015 Rayner, 1998

If a rate changes, keep its meaning clear: faster reading with poorer understanding is a different outcome from faster reading with the same understanding. A speed number alone cannot distinguish them. Rayner et al., 2016 Schotter et al., 2014