What Is Inner Speech?
Inner speech is the use or experience of language in the course of thinking without producing an audible utterance. It may appear as a nearly complete sentence, a brief instruction, a question addressed to oneself, an imagined exchange, or a fragment that carries meaning without the full structure of spoken language. Its psychological importance lies in the way linguistic representations can help people hold goals in mind, organize sequences of actions, rehearse information, evaluate alternatives, and regulate responses to changing situations. Inner speech is one available resource within human cognition; its frequency, subjective form, and usefulness depend on the person and the task. The major scientific reviews describe a heterogeneous phenomenon with developmental, linguistic, cognitive, and phenomenological dimensions (Alderson-Day and Fernyhough, 2015a; Fernyhough and Borghi, 2023).
An everyday example is silently telling yourself, “Check the units before submitting,” while reviewing a calculation. In another situation a person might mentally rehearse what to say in a difficult conversation, alternate between imagined viewpoints, or repeat a phone number for a few seconds. These episodes involve different degrees of deliberate control and different kinds of linguistic information. A moment of silently repeating digits draws heavily on verbal short-term memory; an internal argument about a decision recruits language, evaluation, memory, and often emotion. Treating all these experiences as one identical mechanism conceals the questions that psychological research is trying to answer.
The central finding is that language can function as a flexible cognitive tool. Controlled experiments show that interfering with verbal rehearsal can impair some forms of goal maintenance and task switching, while developmental studies associate self-directed speech with emerging self-regulation. Other tasks can be performed effectively through visual, motor, spatial, or other cognitive representations. Recent studies of people reporting little inner speech further establish the importance of studying individual and task-specific patterns rather than presuming that every successful thought contains an internal sentence (Emerson and Miyake, 200300511-9); Nedergaard and Lupyan, 2024).
Definitions and Conceptual Boundaries
Inner speech, inner monologue, and inner dialogue
Inner monologue usually designates a continuous or apparently continuous stream of self-directed verbal thought. Inner dialogue refers to thought organized as an exchange, with questions, answers, different perspectives, or imagined interlocutors. Inner speech is the broader research category covering both forms and many shorter or less complete instances. A silent “now” during a difficult action can qualify as inner speech even when it has no narrative structure. A person may also experience internally generated words in a voice-like way, as abstract words without a pronounced sound, or in an abbreviated form whose meaning is understood immediately. These variations have been systematically investigated through questionnaires and interviews (McCarthy-Jones and Fernyhough, 2011).
Private speech, self-talk, and silent verbal rehearsal
Private speech is speech directed primarily toward oneself that is actually produced aloud or under the breath. It is common in research on children's problem solving and can also occur in adults during unfamiliar, demanding, or frustrating activities. Self-talk is a broad term that may encompass private speech, intentionally spoken motivational statements, and internal verbalization. Subvocal or articulatory rehearsal refers more narrowly to processes used to maintain verbal material over short intervals. Researchers sometimes operationalize inner speech as rehearsal for particular experiments, but the subjective experience of a conversation with oneself has more dimensions than retaining a string of sounds (Baddeley, 2003; Alderson-Day and Fernyhough, 2015a).
Thought, auditory imagery, and verbal experience
Conceptual reasoning, visual imagery, felt emotion, bodily action planning, and language are capable of interacting within a single train of thought. Hearing-like imagery of a familiar singer's voice is not automatically equivalent to using language for self-instruction. Thinking of the answer to a question without sensing any spoken words is likewise an intelligible cognitive report. Research therefore distinguishes the structure of information processing from the way a person reports experiencing it. The two can be related without being identical; a phonological code can influence performance even when a participant reports little conscious internal narration (Perrone-Bertolotti et al., 2014; Alderson-Day and Pearson, 2023).
How Inner Speech Develops
Social language and the internalization account
Lev Vygotsky's developmental framework proposed that linguistic resources first used in interaction with other people can become tools for organizing an individual's own behavior. A caregiver may say, “Find the corners first,” while helping a child solve a puzzle. Later, the child may repeat a similar directive aloud while working independently. With practice, aspects of the directive may become quieter, shorter, and more internally available. The influential internalization account treats this as a functional transformation of social language during cognitive development, rather than a literal recording of adult speech played inside the head. Modern reviews situate the account alongside other theories of verbal rehearsal, speech production, and cognitive control (Alderson-Day and Fernyhough, 2015a).
The model explains why inner speech frequently has a directive or dialogic quality: learners have experience using conversation to represent rules, negotiate interpretations, and repair mistakes. It also generates empirical questions. If private speech becomes a regulatory tool, its form and timing should relate to the difficulty and demands of a task. If socially shared dialogue matters, children's opportunities for linguistic interaction should be relevant to the repertory of instructions that they later produce for themselves. These are research propositions whose mechanisms must be tested using developmental designs rather than deduced from theoretical plausibility.
Private speech during childhood problem solving
A child building a model may say, “This part goes here—no, turn it,” and then resume the task. Such speech can mark ongoing monitoring and adaptation. Its frequency often changes with age and context, and it can be particularly visible when a task is challenging. Because difficulty can independently produce more speech and more errors, a raw correlation between talking aloud and success is an ambiguous measure of benefit. More informative studies examine the content and maturity of utterances, task conditions, and later outcomes.
In a longitudinal study of 160 children, Whedon and colleagues observed self-directed speech during a puzzle task at age three and examined later inhibitory control and emotion regulation. The maturity of private speech was associated with inhibitory control at age four, with associations varying by temperament. These findings support a developmental link while leaving room for shared influences, measurement differences, and causal alternatives. An observational path across years alone cannot establish that encouraging a specific phrase would produce the same long-term effect (Whedon et al., 2021).
Development is a changing repertoire
There is no scientific reason to expect overt self-talk to disappear completely after a particular birthday. Adults may speak through a complicated repair, rehearse a presentation aloud, or use a spoken reminder under pressure. The developmental question is how people become able to choose among external dialogue, private speech, silent language, visual representation, and practiced action. Skill, education, linguistic experience, environment, and situational stress shape that choice. Quiet performance may indicate mastery, effective alternative strategies, or a reluctance to speak in the research setting; researchers need additional evidence to discriminate these possibilities (Alderson-Day and Fernyhough, 2015a).
Forms and Phenomenology of Inner Speech
Expanded and condensed inner speech
Expanded inner speech resembles ordinary spoken sentences, with recognizable word order and sometimes a clear speaker or listener. Condensed inner speech can consist of a word fragment, semantic shorthand, or an abbreviated verbal formulation. A condensed plan might feel like “keys—door—train,” carrying the instruction to collect the keys, leave the building, and catch a train. Some theoretical accounts place these forms on a continuum shaped by how much linguistic detail a task requires. Current evidence supports diversity in reports but does not yet establish a single universal transformation from expanded to condensed speech in every individual (Fernyhough and Borghi, 2023).
Dialogic, evaluative, and other-person perspectives
The Varieties of Inner Speech Questionnaire differentiates dialogic experiences, evaluative or motivational inner speech, condensed experience, and experiences that incorporate representations of other people's voices. These dimensions help describe what people notice about internal verbal life. They are research constructs rather than clinical categories. A silently imagined disagreement with a friend may be a commonplace device for anticipating social responses; a self-critical sentence may reflect an emotion or habitual evaluation. Questionnaire dimensions are statistical patterns in responses and require independent tests to establish specific causal functions (McCarthy-Jones and Fernyhough, 2011).
Voluntary control and spontaneous occurrence
Some internal statements are deliberately constructed, such as rehearsing a sequence of steps before acting. Others emerge during drifting attention, spontaneous recollection, or the evaluation of an event. Spontaneous and deliberate forms may overlap in neural resources while differing in purpose, monitoring, and attentional control. The moment in which a person notices a phrase entering awareness does not by itself reveal how it was generated. A comprehensive account therefore needs both task-based experiments and careful reports of everyday experience (Perrone-Bertolotti et al., 2014).
Cognitive Mechanisms: From Language to Action
Verbal coding and maintaining task rules
A verbal label can compress a complicated requirement into a readily retrievable cue. A participant alternating between classifying a number by parity and magnitude might quietly maintain “odd or even” or “high or low,” reducing confusion about which rule governs the next response. Inner speech can help refresh a task goal, identify the relevant decision dimension, and bring a rule back into focus when competing actions are possible. The strength of this contribution depends on cue availability and task architecture; explicit external labels can sometimes replace the need for self-generated verbal reminders (Miyake et al., 2004).
Verbal working memory and the phonological loop
Working-memory models identify processes that maintain information while cognition proceeds. In Baddeley's influential multicomponent framework, a phonological storage component and rehearsal processes support short-term maintenance of verbal material. Repeating a number silently is a familiar example. Inner speech can draw upon these mechanisms, particularly when a task depends on preserving word sequences or sound patterns. The model distinguishes maintenance from broader attentional control and therefore provides a principled reason to avoid equating verbal short-term memory with the whole domain of conscious inner dialogue (Baddeley, 2003).
This distinction matters experimentally. If a participant loses the order of a word list while repeatedly articulating an irrelevant syllable, an impairment to verbal maintenance is a plausible explanation. If the same interference disrupts the selection of a new task rule, the evidence may implicate verbal cueing for executive control. Either effect has multiple potential components, including the burden of an extra task, auditory or motor interference, and differences in strategy. Researchers therefore seek control conditions that impose comparable effort without requiring the same verbal resources.
Speech planning, prediction, and monitoring
Accounts grounded in language production explain internally generated utterances in terms of partially executed speech plans. Intended words may activate phonological, articulatory, and monitoring representations without passing through audible articulation. The predicted sensory consequences of speaking may contribute to the felt quality of an internal voice and to judgments that the speech is self-generated. These ideas are supported by converging behavioral and neural observations but several models disagree about which production stages are necessary for particular kinds of inner speech. An entirely semantic, highly condensed episode may be represented differently from an internally heard full sentence (Perrone-Bertolotti et al., 2014; Fernyhough and Borghi, 2023).
Multiple routes to regulation
Language is especially convenient for rules that can be encoded as explicit propositions: “Wait for the signal,” “Compare both options,” or “Do the next step first.” Other routes can organize the same action, including recognition of visual patterns, learned motor sequences, environmental cues, and reinforcement histories. This multiple-route perspective explains why verbal interference can strongly affect one laboratory procedure while leaving another one largely intact. Cognitive architecture is better understood by asking when verbal codes are recruited, what they contribute, and which alternatives remain available (Alderson-Day and Pearson, 2023).
Inner Speech and Executive Functions
Task switching and cognitive flexibility
One influential experiment required participants to change between tasks while repeating irrelevant sounds, a manipulation known as articulatory suppression. Emerson and Miyake found that suppression increased switching costs under conditions that encouraged internal verbal cueing. The authors also found that the form of external cues moderated the result. The pattern supports a specific account: internal speech can help activate the upcoming task rule. It does not imply that every executive function requires conscious narration, or that articulatory suppression selectively removes every form of inner speech (Emerson and Miyake, 200300511-9)).
Follow-up research in a random task-cuing paradigm investigated when verbal cue retrieval is useful. Miyake and colleagues concluded that inner speech can assist retrieval of task goals under some cueing conditions. The lesson is methodological as well as theoretical: observing a suppression effect gives a conditional causal clue about resources used by the task, and testing how the effect changes with cues narrows the explanation (Miyake et al., 2004).
Inhibitory control and resisting impulses
An internal instruction such as “Stop and check” can make a goal accessible at a moment when an automatic response is tempting. Developmental accounts propose that self-directed speech supports the emerging ability to regulate action. Observational links between children's private speech and inhibitory control are consistent with this function. They are not estimates of how much the typical adult's inhibitory control depends on inner monologue. Inhibition includes multiple processes, and task difficulty, temperament, and familiarity can affect both overt verbalization and performance (Whedon et al., 2021; Alderson-Day and Fernyhough, 2015a).
Planning and sequencing
Inner speech can serve as a compact planning workspace. A person may enumerate subgoals, retain a future obligation, and compare an intended action with the current state of a project. The theoretical benefit is representational: language offers portable labels for actions that are temporally distant, abstract, or distributed across several stages. The precise gain from verbalizing a plan varies across tasks and prior knowledge. Habitual expertise may replace sentence-like instructions with faster representations; unfamiliar procedures may invite renewed self-explanation.
Error detection and performance monitoring
Self-directed language can explicitly mark a discrepancy: “That cannot be right; the units do not match.” This can help identify a rule violation and select a correction. However, error detection can also occur in visual, sensorimotor, and nonverbal reasoning before any consciously formed sentence. Inner speech may report, stabilize, or redirect an evaluation rather than originate every signal that something is wrong. Experiments must distinguish the first detection of an error from subsequent verbal interpretation.
Inner Speech and Learning
Rehearsal and remembering verbal material
Short-term rehearsal is particularly relevant when information is phonological or sequential. Repeating a new word, keeping directions active, or practicing pronunciation can recruit verbal maintenance. The phonological-loop framework offers one account of why such tasks are vulnerable to interference from irrelevant articulation. Across authentic learning situations, comprehension, retrieval practice, conceptual organization, sleep, attention, and prior knowledge also influence durable memory. Silent repetition alone is not a complete theory of learning (Baddeley, 2003).
Self-explanation and instructional language
Students may organize unfamiliar material by asking themselves, “What does this statement mean?” or “Which assumption did I use?” Such questions can direct attention to relations among facts. Their educational value arises from the cognitive work they prompt—retrieving, comparing, predicting, and explaining—rather than from the mere presence of words inside the head. A useful research distinction is between a test of verbal rehearsal and a test of elaborative self-explanation. A person may verbally rehearse a rule without understanding it, or understand a principle and report little overt internal narration.
Mental simulation of conversation
Rehearsing a presentation or anticipated exchange can bring together lexical selection, social prediction, memory, and emotional regulation. The imagined listener may shape word choice and the order of arguments. Mental simulation can prepare a response, but it can also repeatedly recycle an unproductive hypothetical scenario. The functional question is whether the inner exchange improves the upcoming action, clarifies a decision, or consumes attention without generating useful information. This question is answered by outcomes and context rather than the presence of imagined conversation alone.
Inner Speech, Emotion, and Self-Regulation
Self-instruction, encouragement, and self-evaluation
Self-directed language can specify a goal, frame a challenge, or remind a person of a previously chosen commitment. The content of that language matters. “Start with the first paragraph” provides an actionable cue, while a global judgment about one's ability may influence emotion without specifying a way forward. Research on self-talk and regulation has examined how grammatical person and self-distancing affect responses under particular laboratory circumstances. Such results indicate that linguistic framing can change appraisal in some contexts; they do not establish a universal psychological technique for every individual and emotional situation (Kross et al., 2014; Dolcos and Albarracín, 2014).
Perspective taking and pronoun choice
Kross and colleagues experimentally investigated self-talk using a person's name or non-first-person pronouns when reflecting on a stressful experience. Their experiments reported benefits for some measures of self-regulation under the studied conditions. Dolcos and Albarracín compared second-person and first-person self-address in particular performance and intention tasks. These findings make linguistic perspective an empirical variable, not a rule that speaking to oneself in one pronoun must always improve psychological functioning. Effects can depend on situation, baseline anxiety, instructions, and what outcomes are measured (Kross et al., 2014; Dolcos and Albarracín, 2014).
Rumination and repetitive inner commentary
Internal words may repeat during unresolved concerns, with thoughts cycling around feared outcomes or self-evaluations. The psychological meaning depends on whether repetition helps the person identify a problem and act, or prolongs distress while offering little new information. Questionnaire studies report associations between specific forms of evaluative inner speech and anxiety-related variables, but correlations do not determine direction of causation. Mood can influence verbal thought, verbal thought can influence appraisal, and other factors can shape both. Mental-health diagnosis and treatment require separate clinical assessment and evidence, outside the scope of this general account of language and cognition (McCarthy-Jones and Fernyhough, 2011).
A practical example of regulation
Imagine someone preparing to respond to a challenging message. An immediate reaction may be followed by the silent instruction “Read it once more before replying.” The instruction can hold the intended behavior accessible, create an interval for considering the wording, and identify an observable next step. If the person instead spends the next hour internally rehearsing the same imagined argument, the verbal process may consume resources without resolving the task. These are examples of possible cognitive functions; they are not diagnostic tests or evidence that a particular intervention will work for every person.
Is Inner Speech Necessary for Thinking?
Thinking can be verbal and nonverbal
Human reasoning includes spatial transformations, perceptual recognition, skilled movements, emotional appraisals, and abstract relations. Language can encode, combine, and communicate concepts; it can also guide the selection of strategies. The fact that some reasoning tasks improve with linguistic support does not establish that language constitutes all thought. Conversely, the ability to reason without vivid inner speech does not erase the causal contribution of linguistic resources in tasks requiring sound-based memory or explicit verbal cueing. A strong scientific account accommodates both conditional benefits and cognitive diversity (Alderson-Day and Pearson, 2023; Fernyhough and Borghi, 2023).
Research on markedly reduced inner speech
Nedergaard and Lupyan studied adults who reported very little inner speech and compared their performance with adults reporting much more. Their 2024 series of studies found group differences on verbal working-memory and rhyme-judgment tasks, while task switching and certain perceptual categorization effects did not show the same relation to self-reported inner speech. The authors used the term anendophasia for the reported absence of inner-speech experience. This is a research label describing an experience, not a psychiatric diagnosis. The study demonstrates a promising way to investigate function-specific consequences, with sample selection and measurement requiring continued scrutiny (Nedergaard and Lupyan, 2024).
What subjective frequency can and cannot show
A person reporting an uninterrupted inner monologue may be remembering salient episodes or using a broad interpretation of “thinking in words.” Another person who says “I never hear a voice” may mean that verbal thoughts lack auditory qualities, that they have few conscious words, or that their own experience is difficult to describe. These reports deserve serious consideration but are not direct measures of every linguistic computation supporting performance. Distinguishing momentary verbal experience, generalized self-description, and experimental performance is one of the field's central methodological responsibilities.
How Scientists Measure Inner Speech
Questionnaires and the Varieties of Inner Speech Questionnaire
Questionnaires ask respondents to characterize usual frequency, internal dialogue, evaluative tone, or the condensation of their inner experience. The Varieties of Inner Speech Questionnaire was developed to distinguish qualitatively different forms rather than assuming one overall inner-voice score. Questionnaires are practical for comparing many participants, but respondents must summarize private events over time and interpret abstract prompts. They may differ in what counts as an internal word or voice (McCarthy-Jones and Fernyhough, 2011).
Experience sampling and momentary reports
Experience-sampling approaches ask people to report what was happening in consciousness at specified moments, often signaled by a smartphone or device. Some protocols use brief structured questions; descriptive experience sampling uses repeated training and interviewing to clarify momentary reports. The contrast with broad questionnaires matters because asking “How often do you usually talk in your head?” may produce a different response from asking what was present just before a random signal. In a smartphone study of 51 university students, generalized and randomly sampled reports of inner-speech dimensions were only partly aligned, and momentary endorsements were lower across measured subscales (Alderson-Day and Fernyhough, 2015b).
These methods offer access to phenomenology, but neither removes all interpretive problems. Alerts can interrupt experience, recollection can reshape a fleeting episode, and detailed interviews depend on training and participant cooperation. Random sampling also reflects only the contexts sampled. Different protocols can legitimately answer different questions about what people believe they typically experience, what they report at a moment, and how experience changes across settings.
Articulatory suppression and dual-task interference
In articulatory suppression, participants repeatedly say an irrelevant syllable or word while doing another task. If performance changes compared with suitable controls, investigators infer that resources associated with verbal rehearsal or production matter for the tested performance. The strongest interpretation depends on the design: suppression also imposes articulatory activity, sustained attention, and the demands of performing two tasks. Comparisons with nonverbal secondary tasks, manipulated cues, and alternative strategies help identify whether verbal goal maintenance is the important pathway (Emerson and Miyake, 200300511-9); Miyake et al., 2004).
Behavioral correlates and developmental observation
Researchers can code children's spontaneous utterances while they solve a puzzle, quantify sequence and content, and relate speech to errors, transitions, or later abilities. The timing of speech can be more informative than its total amount. A child who says “Try the other one” immediately before switching strategy provides different evidence from one who speaks after completing a task. Longitudinal observation can show temporal prediction, yet hidden common causes and changing environments remain central design considerations (Whedon et al., 2021).
Neuroscience and covert language tasks
Neuroimaging studies often ask participants to generate words silently, read without speaking, or imagine producing a phrase. Overlapping activity between covert and overt language tasks is evidence that some production-related networks participate in internally generated language. However, the label “inner speech” applied to an experimental instruction does not guarantee that participants share the same subjective experience. Functional magnetic resonance imaging measures changes associated with blood oxygenation rather than reading the precise verbal content of an individual thought. Electrophysiology, behavioral timing, stimulation, and carefully designed contrasts provide complementary information, each with its own limitations (Perrone-Bertolotti et al., 2014).
Converging methods as a scientific standard
The most informative evidence brings together lived reports, task-specific behavioral effects, development, and linguistic modeling. A questionnaire identifies an experience dimension; a suppression task examines a causal contribution of verbal resources under specified conditions; longitudinal data tracks developmental relations; and neural measurements constrain possible biological implementations. None is a complete substitute for the others. Agreement across independently designed measures strengthens a claim, while disagreement often indicates that a broad label has combined distinct phenomena.
Neural Systems and Inner Speech
Distributed language and control networks
Covert language tasks involve regions associated with language planning, phonological representation, auditory imagery, and cognitive control. The relative participation of frontal, temporal, and parietal systems depends on whether a participant imagines hearing, constructs a phrase, maintains verbal material, or switches between rules. The neural account is therefore distributed and task-dependent. A single activity peak cannot establish the presence of inner speech, nor can the absence of one expected regional signal establish that a person has none (Perrone-Bertolotti et al., 2014; Alderson-Day and Fernyhough, 2015a).
Auditory imagery and the sense of agency
When inner speech feels voice-like, some theories emphasize simulated auditory consequences of a planned utterance. Monitoring systems may help a person recognize the internal event as self-generated. These ideas contribute to debates about how the sense of agency is constructed, but a self-generated thought and a perceived external voice are distinct subjective reports that require careful clinical interpretation when unusual experiences cause distress. No brain-imaging pattern alone provides a psychiatric diagnosis (Perrone-Bertolotti et al., 2014).
Individual Differences, Languages, and Modalities
Frequency, vividness, and preferred cognitive strategies
People vary in how often inner language is noticed, whether it seems auditory, how grammatical it feels, and whether it accompanies planning or emotional reflection. A person may prefer silent words for a verbal task and visual structure for spatial problem solving. Another may report very little inner speech while accomplishing complex professional and everyday activities. A useful psychological description specifies the type of experience and task, rather than assigning an overall cognitive rank to a person's amount of self-talk (Nedergaard and Lupyan, 2024).
Bilingual and multilingual internal language
People with more than one language may experience inner verbalization in different languages across settings and activities. Language dominance, context, learned terminology, conversational partners, and the emotional associations of particular expressions can all be relevant. An internal phrase in one language may provide a convenient rule or memory cue; another language may be preferred for a familiar setting. Studies must separate subjective language choice from performance effects and avoid assuming that bilingualism creates a universal advantage in executive control. General claims about bilingual cognitive advantages belong to a distinct empirical debate and should not be inferred from a person's preferred language of thought.
Inner signing and modality differences
Accounts centered exclusively on an audible “voice” overlook language experienced through signed and visual-manual modalities. Research reviews discuss inner signing and language-related experience in deaf individuals, demonstrating why a theory of inner language must consider modality and language access. A person whose primary language is signed may internally represent linguistic structures without any imagined sound. Sensory and linguistic histories influence the experience, and neither auditory vividness nor spoken-language fluency defines a person's capacity for complex thought (Perrone-Bertolotti et al., 2014; Alderson-Day and Pearson, 2023).
Neurodiversity and clinical interpretation
A theoretical review by Alderson-Day and Pearson discusses inner-speech variation in neurodivergent populations, including autistic people and people with developmental differences in language access. The literature includes heterogeneity, mixed findings, and methodological assumptions about how people experience and use language. Variation in private or inner speech is not a diagnostic shortcut for autism, ADHD, developmental language disorder, aphasia, or any other condition. Clinical symptoms, disability, assessment, and intervention belong to disorder-specific evidence and professional evaluation. The general scientific question here is how linguistic resources are recruited across different cognitive and developmental profiles (Alderson-Day and Pearson, 2023).
Interpretive and Methodological Controversies
Is verbal suppression specific enough?
Articulatory suppression influences a package of processes, including speech-motor control and attention. When a task becomes harder under suppression, researchers can conclude that the experimental manipulation affected performance; identifying the precise mechanism requires further contrasts. In particular, increased switching costs do not demonstrate that a person's conscious internal sentences were removed. The experimentally targeted resource and the phenomenological report should be treated as different measures until linked by direct evidence (Emerson and Miyake, 200300511-9)).
Does self-report capture actual frequency?
Different measurement approaches generate strikingly different estimates of how common internal verbal episodes are at a given moment. Sampling contexts, instructions, definitions, and recall all affect estimates. Researchers should report the operational definition and sampling frame rather than treating any single percentage as a universal prevalence rate. The partial alignment between questionnaires and momentary sampling provides an especially clear warning against using one score as the whole construct (Alderson-Day and Fernyhough, 2015b).
Does language cause cognitive performance?
Experimental interference and self-talk manipulation provide some causal evidence within controlled tasks. Observational developmental associations, questionnaire correlations, and fMRI coactivation establish different kinds of links. A broad causal claim—such as “inner speech creates self-control”—must specify the intervention, mechanism, population, and outcome. Otherwise it collapses several research questions into one attractive statement. Strong claims require reproducible task-specific effects, well-chosen controls, and clear distinctions among verbal rehearsal, speech planning, subjective experience, and self-instruction.
Is condensed inner speech really linguistic?
A brief impression of meaning can be difficult to distinguish from a nonverbal concept. Some models interpret condensed inner speech as language whose syntactic or phonological form has been reduced; others emphasize nonverbal or multimodal representations interacting with linguistic processes. Distinguishing them requires evidence about how such experiences respond to lexical, semantic, and phonological manipulations. A participant's ability to report an experience using words does not establish that the experience originally contained an articulated word sequence.
Inner Speech in the Artificial Era
Computational modeling and human cognitive evidence
A language-generating model can produce convincing first-person verbal descriptions, plans, and simulated self-talk. Such outputs are observable text. Human inner-speech research, by contrast, combines subjective reports with developmental history, linguistic processing, behavior, and measurements of biological systems. It is a theoretical possibility to use computational systems to model aspects of language planning or verbal self-cueing, and contemporary reviews explicitly consider computational modeling. Structural similarities in language outputs or task performance do not settle whether two systems have the same internal processes or subjective experiences (Fernyhough and Borghi, 2023).
Functional analogy and its limits
A model that internally iterates over verbal instructions may share a functional feature with a person who repeats a rule: an explicit symbolic cue guides subsequent operations. Establishing psychological equivalence would require specifying the architecture, measurements, and predictive consequences in each case. Claims about machine subjectivity are philosophical and scientific questions with different evidential requirements from a benchmark score or a fluent explanation. In the Artificial Era, rigorous comparison benefits from precise operational definitions of verbal representation, regulation, monitoring, and reported experience; it does not require conflating their different evidential standards.
Practical Implications of the Research
Designing useful self-instructions
When a task has multiple rules, an explicit short cue may make the relevant step easier to retrieve. Examples include “Check the source,” “One action at a time,” or “Read the question again.” Laboratory task-switching studies supply a mechanistic reason that verbal cues can be useful under some conditions. They do not establish one universally optimal script. Effective use depends on whether the phrase highlights information that the learner genuinely needs and whether it complements rather than interrupts the task (Miyake et al., 2004).
Supporting children's autonomy
Adults can model clear verbal plans while helping children with age-appropriate activities, invite a child to explain the next step, and gradually reduce assistance as the child becomes more independent. Developmental research makes these practices theoretically intelligible, while the choice of educational strategies needs evidence suited to the learners and setting. A child speaking aloud during a demanding task may be actively organizing behavior. Teachers and caregivers should interpret the behavior within the learning context rather than automatically regarding it as distraction or as a symptom of a disorder (Whedon et al., 2021).
Matching strategy to task and person
Verbal rehearsal is well suited to some word-order, rule, and instruction tasks. Drawing a diagram may serve spatial reasoning better. External notes can reduce memory burden when working-memory capacity is limited or many steps must be tracked. A person who experiences little inner narration can use effective language-based tools in writing and conversation or rely more heavily on other representational resources. The evidence supports flexibility and informed experimentation with strategies, without ranking people by the vividness of their internal voice.
When distressing inner experiences call for professional care
Ordinary self-talk may include reminders, disagreement with oneself, imagined dialogue, and evaluative language. Persistent distress, major changes in functioning, or experiences that seem externally imposed can warrant a qualified professional assessment based on the entire presentation. The everyday frequency of internal language and the label anendophasia do not establish a diagnosis. Research on auditory verbal hallucinations, repetitive negative thought, and neurodevelopmental conditions must maintain condition-specific distinctions and clinical standards (Alderson-Day and Fernyhough, 2015a).
Evidence Summary: What the Findings Support
The presence of diverse forms of reported inner speech is established by converging questionnaire, interview, and sampling research. Experimental evidence supports a contribution from verbal rehearsal and self-cueing to certain working-memory and task-switching performances. Developmental research supports links between children's private speech and emerging regulatory abilities, while specific causal pathways and population-wide effects remain active research questions. Individual-difference studies indicate that reduced reported inner speech can be associated with particular verbal tasks while leaving other cognitive measures unaffected. Broad claims that inner monologue is necessary for intelligence, consciousness, self-awareness, or all self-control exceed the evidence reviewed here (Alderson-Day and Fernyhough, 2015a; Nedergaard and Lupyan, 2024).
The strongest interpretation of inner speech is therefore functional and differentiated: internally generated language can help maintain, reformat, or retrieve information when those operations are useful. Its effects must be evaluated against particular tasks, forms of inner language, and individual histories. This approach keeps linguistic experience central to psycholinguistics while giving adequate scientific standing to the diversity of ways people think.
Frequently Asked Questions
Does everyone have an inner monologue?
People report substantial differences in the frequency and vividness of internal verbal experience. Some report constant or frequent word-like thought, while others report little or none. Different questionnaires and momentary methods can yield different profiles. Reporting little inner monologue is a form of cognitive diversity rather than a diagnosis or an intelligence score (Nedergaard and Lupyan, 2024).
Is inner speech the same as thinking?
Inner speech is one way language participates in thinking. Thought can also involve visual, spatial, motor, affective, and abstract conceptual representations. Inner speech is especially relevant when a person uses words to remember instructions, organize a rule, or rehearse material; its presence is not required for every thought or action.
Can talking to yourself improve focus?
Brief, task-relevant self-instructions may help maintain or retrieve goals in some contexts. Their usefulness depends on task difficulty, external cues, and the person's existing strategies. Controlled task-switching experiments provide more specific evidence than the claim that any self-talk improves attention (Emerson and Miyake, 200300511-9)).
Why do children talk aloud while solving problems?
Private speech can help children formulate the next step, identify mistakes, and manage a demanding activity. Its form and frequency change with development and context. Speaking aloud during problem solving is ordinarily a meaningful behavior and must be interpreted in relation to age, task, and situation rather than treated as a clinical sign (Whedon et al., 2021).
Can inner speech take the form of a dialogue?
Yes. People report internal exchanges, self-questioning, and imagined perspectives, as well as short verbal fragments and complete sentences. Dialogic inner speech is a recognized phenomenological dimension investigated through measures such as the Varieties of Inner Speech Questionnaire (McCarthy-Jones and Fernyhough, 2011).
Does having no inner voice mean someone cannot control behavior?
No. Studies of individual differences find some associations with verbal working-memory and rhyme tasks but do not show a general loss of self-control or task switching in people who report little inner speech. Control can be supported by multiple cognitive and environmental mechanisms (Nedergaard and Lupyan, 2024).
Can inner speech occur in a signed language?
Research reviews describe inner signing and language-related experience in deaf individuals. Linguistic thought can involve visual-manual representations and does not require imagined audible speech. Language modality and personal history matter to how an individual reports the experience (Perrone-Bertolotti et al., 2014).
How can researchers know whether somebody is thinking in words?
They combine subjective reports with structured questionnaires, momentary sampling, behavioral interference, and neuroscience. Each method measures a different part of the problem. Convergent patterns are more informative than any single questionnaire answer, task effect, or brain activation map (Alderson-Day and Fernyhough, 2015a).
Related Articles
Cluster Language, Thought and Psycholinguistics — foundational overview of language processing, language and thought, and the research architecture connecting them.
Psycholinguistics: How Language Is Processed in the Mind — cognitive mechanisms and research methods of human language processing.
Semantic Memory: Facts, Meaning, and Knowledge — conceptual knowledge and memory processes that can interact with internal verbal thought.
Framing Effect: How the Wording of Choices Changes Decisions — the influence of linguistic presentation on human judgments, under its separate canonical intent.
Language Processing in the Brain: Cognitive Networks and Neural Systems — distributed language-related neural mechanisms and methods.
Language and Concept Formation: How Labels Influence Learning — the effects of verbal labels on category learning and conceptual organization, a complementary scientific account of language and thought.
Language and Numerical Cognition: Number Words and Mathematical Thinking — the contribution of verbal number symbols and numerical language to selected forms of reasoning and representation.
References
Alderson-Day, B., and Fernyhough, C. (2015a). Inner speech: Development, cognitive functions, phenomenology, and neurobiology. Psychological Bulletin, 141(5), 931–965. https://doi.org/10.1037/bul0000021.
Alderson-Day, B., and Fernyhough, C. (2015b). Relations among questionnaire and experience sampling measures of inner speech: A smartphone app study. Frontiers in Psychology, 6, 517. https://doi.org/10.3389/fpsyg.2015.00517.
Alderson-Day, B., and Pearson, A. (2023). What can neurodiversity tell us about inner speech, and vice versa? A theoretical perspective. Cortex, 168, 193–202. https://doi.org/10.1016/j.cortex.2023.08.008.
Baddeley, A. (2003). Working memory: Looking back and looking forward. Nature Reviews Neuroscience, 4, 829–839. https://doi.org/10.1038/nrn1201.
Dolcos, S., and Albarracín, D. (2014). The inner speech of behavioral regulation: Intentions and task performance strengthen when you talk to yourself as a You. European Journal of Social Psychology, 44, 636–642. https://doi.org/10.1002/ejsp.2048.
Emerson, M. J., and Miyake, A. (2003). The role of inner speech in task switching: A dual-task investigation. Journal of Memory and Language, 48, 148–168. https://doi.org/10.1016/S0749-596X(02)00511-900511-9).
Fernyhough, C., and Borghi, A. M. (2023). Inner speech as language process and cognitive tool. Trends in Cognitive Sciences, 27, 1180–1193. https://doi.org/10.1016/j.tics.2023.08.014.
Kross, E., Bruehlman-Senecal, E., Park, J., Burson, A., Dougherty, A., Shablack, H., Bremner, R., Moser, J., and Ayduk, O. (2014). Self-talk as a regulatory mechanism: How you do it matters. Journal of Personality and Social Psychology, 106, 304–324. https://doi.org/10.1037/a0035173.
McCarthy-Jones, S., and Fernyhough, C. (2011). The varieties of inner speech: Links between quality of inner speech and psychopathological variables in a sample of young adults. Consciousness and Cognition, 20, 1586–1593. https://doi.org/10.1016/j.concog.2011.08.005.
Miyake, A., Emerson, M. J., Padilla, F., and Ahn, J. (2004). Inner speech as a retrieval aid for task goals: The effects of cue type and articulatory suppression in the random task cuing paradigm. Acta Psychologica, 115, 123–142. https://doi.org/10.1016/j.actpsy.2003.12.004.
Nedergaard, J. S. K., and Lupyan, G. (2024). Not everybody has an inner voice: Behavioral consequences of anendophasia. Psychological Science, 35(7). https://doi.org/10.1177/09567976241243004.
Perrone-Bertolotti, M., Rapin, L., Lachaux, J.-P., Baciu, M., and Lœvenbruck, H. (2014). What is that little voice inside my head? Inner speech phenomenology, its role in cognitive performance, and its relation to self-monitoring. Behavioural Brain Research, 261, 220–239. https://doi.org/10.1016/j.bbr.2013.12.034.
Whedon, M., Perry, N. B., Curtis, E. B., and Bell, M. A. (2021). Private speech and the development of self-regulation: The importance of temperamental anger. Early Childhood Research Quarterly, 56, 213–224. https://doi.org/10.1016/j.ecresq.2021.03.013.