What does inner speech during silent reading actually look like?
Inner speech is the experience of hearing words in your head while reading without moving your lips. One line of evidence comes from phonetic length effects in lexical decision tasks: the data suggest that the acoustic representations activated in silent reading are best characterized as inner speech [2]. In other words, readers appear to activate something speech-like, not just an abstract word meaning.
A different kind of evidence comes from tongue twisters. When people silently read direct speech such as a quoted sentence, they show more phonemic interference than when they read indirect speech or non-speech sentences, and this effect appears in silent reading but not oral reading [3]. The interference was selectively disrupted by concurrent articulation, such as silently mouthing something, compared with finger tapping [3]. That pattern is consistent with readers mentally simulating phonological detail, though the study frames the causal role as applying at least to tongue-twister reading [3].
Does the auditory cortex actually "hear" the inner voice?
The most direct neural evidence comes from intracranial recordings in the auditory cortex. Silent reading activated voice-selective temporal voice areas, and the authors describe this as a top-down imagery speech component — we can hear our own inner voice [5]. That supports the idea that silent reading can recruit the brain's voice-processing machinery.
But this finding does not establish that the activation is specified at the articulatory level, meaning tied to the specific movements of lips, tongue, and vocal folds. The study reports voice-selective activation during reading, not a mapping of voicing features onto articulatory gestures [5]. So the neural evidence supports a voice-like component to inner speech without settling how finely that component is specified.
Does reading skill change the inner speech?
The evidence on skill is mixed and indirect. In the phonetic-length work, the data suggest that acoustic representations in silent reading are best characterized as inner speech, and the summary notes variation as readers became more skilled [2]. That points toward skill-related differences in how speech-like silent reading is.
The intracranial study, by contrast, reports that silent reading activated voice-selective auditory cortex independently of participants' reading skills [5]. So a voice-like neural response appeared across skill levels in that sample, which argues against reading skill being the key switch for whether the auditory cortex responds.
Other work complicates the picture further. A qualitative study of silent reading in Indonesian and English categorized readers by reading scores and concluded that the optimal inner speech mechanism differed by language and was influenced by cognition and the language used [1]. That study did not measure articulatory specification or voicing-sensitive cortex responses, so it cannot confirm or rule out a skill effect at that level [1].
Why does blocking articulation make reading harder?
Early work on inner speech focused on articulatory movements themselves. When movements of articulation are inhibited, nearly all observers agreed that such inhibition decreases the ability to remember and to read [4]. That is consistent with articulation playing a functional role in silent reading, though it does not show that the inner speech is specified at a fine articulatory level in the auditory cortex.
Taken together, the evidence supports a speech-like, sometimes phonologically detailed inner voice during silent reading, with voice-selective auditory cortex recruited during reading [3][5]. The specific claim that voicing-sensitive cortex responses encode articulatory detail, and that this varies with reading skill, remains only partly supported: the phonetic-length data hint at skill-related variation [2], while the direct neural study found the response independent of skill [5].
Sources used in this answer
Mechanism of Inner Speech in Silent Reading Activity
A qualitative comparison of silent reading in Indonesian and English found optimal inner speech mechanisms localized to different Broadmann areas depending on language and cognition, with readers grouped by reading score.
What the reader's eye tells the mind's ear: Silent reading activates inner speech
Phonetic length effects in lexical decision suggest that acoustic representations activated during silent reading are best characterized as inner speech, with variation as readers became more skilled.
Mental Simulations of Phonological Representations Are Causally Linked to Silent Reading of Direct Versus Indirect Speech
Silent reading of direct speech produced greater phonemic interference than indirect or non-speech sentences, and this effect was selectively disrupted by concurrent articulation, supporting a causal role for phonological simulation at least in tongue-twister reading.
Inner speech during silent reading.
Inhibiting articulatory movements was widely reported to decrease both memory and reading ability.
How silent is silent reading? Intracerebral evidence for top-down activation of temporal voice areas during reading
Intracranial recordings showed that silent reading activates voice-selective temporal voice areas in auditory cortex, described as a top-down imagery speech component, independently of participants' reading skills.