Imagine a researcher asking, “Do you talk with yourself in your head?” One participant remembers rehearsing an argument with a colleague. Another recalls silently asking a question and then answering it. A third says yes because they often plan aloud without sound. These answers seem similar, but they may refer to different events. Measuring dialogic inner speech begins by deciding which of them counts as an internal exchange, at what moment, and by what evidence.
Dialogic inner speech is the experienced back-and-forth quality of some verbal thinking: questions and answers, shifts in perspective, or turns that resemble conversation without overt speaking. It is one variety of inner speech, not a synonym for all thought, all self-talk, or hearing an external voice. A measure can ask about a person’s usual experience, sample particular moments, or invite an exchange in a laboratory task. Each approach captures a different part of the construct and brings its own confounds.
Define the event before counting it
The first measurement decision is phenomenological. Does a remembered inner exchange need complete sentences, an audible-like quality, or identifiable speakers? A person may experience “Should I call?” followed by “Wait until tomorrow” as two turns. Another may simply recognize opposing considerations without experiencing words. Treating both as dialogic inner speech would make a task less specific. Excluding every abbreviated exchange, however, could miss a common form of verbal thought.
Alderson-Day and Fernyhough’s broad review describes inner speech as variable in form and function. Its developmental connection to social speech offers a reason to investigate dialogic forms, but the theory does not establish that every inner utterance is a conversation. The Institute’s operational rule is deliberately narrow: identify an episode as dialogic only when the participant reports or performs a distinguishable internal alternation of positions or turns, and describe separately whether it was verbal, auditory-like, condensed, deliberate or spontaneous. That rule is a reporting aid, not a validated diagnostic scale.
Dialogic quality also differs from evaluative content. “I should work harder” may be a single self-directed utterance. “Should I work harder?” followed by an imagined objection is more clearly an exchange. A voice associated with another person may appear in either form. A measure that folds self-criticism, imagined other voices and turn-taking into one score cannot tell which feature drives a result. The original Varieties of Inner Speech Questionnaire separated dialogicality from condensed speech, evaluative or motivational quality, and the presence of other people’s voices for this reason.
What a participant calls a “voice” requires care. Everyday language can describe vivid inner speech, remembered speech, auditory imagery or an experience judged to originate elsewhere. A questionnaire item about internal conversation does not determine which is present and must not be treated as a clinical assessment. The measurement aim here is the structure of a reported private experience, with the participant’s own distinction between self-generated thought and other experiences preserved.
Questionnaires describe a person’s retrospective self-model
McCarthy-Jones and Fernyhough developed the Varieties of Inner Speech Questionnaire, or VISQ, to ask about forms of inner speech rather than only its frequency. Its dialogic items concern back-and-forth exchanges in the mind. Responses can be compared across people and with other self-report scales. The study provided initial psychometric evidence in young adults, including distinct factors. That is useful evidence for the instrument’s structure in its sample, not proof that an individual can accurately count every episode they have had.
The revised VISQ-R tested an expanded five-factor model and added a positive or regulatory dimension. Revisions matter when researchers compare studies: an item labeled “dialogic” in one version should not be assumed identical to a sum score from another, and translation may alter what “talking to oneself” suggests. A report should name the version, language, item wording, response scale and scoring rule. If it uses only selected items, it should not cite reliability for the full subscale as if those items were an unchanged instrument.
Questionnaires have a distinctive strength. They can ask about rare, meaningful or personally salient forms that a brief observation period might miss. They are inexpensive enough for broad samples. Yet they ask a participant to generalize over time, compare their experience with ambiguous phrases, and adopt a concept of inner speech. A person may remember a dramatic internal argument and therefore rate dialogic speech as frequent even if most ordinary moments contain no such exchange. Another may routinely think in brief turns but reject the phrase “conversation in my head.”
Retrospective ratings also mingle trait and context. A participant who is preparing for a difficult meeting may report more internal debate that week than in a typical month. Question wording about “usually” or “often” cannot remove that variation. Repeated administrations can estimate stability, while experience sampling can test whether a high retrospective score corresponds to more sampled dialogic episodes. Agreement between methods is informative; disagreement is a finding about what the methods ask, not automatically a failure of the person reporting.
Sample moments rather than only general impressions
Experience sampling prompts people during everyday life to describe what was present just before a signal. A smartphone study by Alderson-Day and colleagues compared questionnaire responses with momentary reports of inner speech. Relationships differed across inner-speech varieties, warning against assuming that a general rating is an event counter. A prompt can ask whether an internal exchange occurred in the sampled moment, what initiated it, whether positions alternated, and whether words were experienced or only inferred afterward.
Sampling has its own measurement effects. Participants may start watching for inner dialogue after seeing the prompt several times. They may compress a changing stream of thought into one forced-choice answer. A noisy setting, work shift or distressing day may change both the experience and the likelihood of answering. Studies should report prompt timing, response delay, missed prompts, question order and whether the sample covers waking contexts. If only the most convenient moments are recorded, an observed rate is not a rate for the whole day.
Descriptive Experience Sampling, or DES, adds a detailed interview about selected moments. Hurlburt, Heavey and Kelsey describe substantial variability in naturally occurring inner speaking and emphasize close examination of what participants meant at a particular sampled moment. An interviewer can ask whether the apparently dialogic event was experienced as two turns, as one unspoken judgment, or as a memory of an earlier conversation. Such interviews may improve phenomenological precision but require training and time. They are not a simple gold standard that can be scaled without changes in the procedure.
A useful comparison is a two-column study: a general dialogicality questionnaire on one side, sampled episodes described with a prespecified coding scheme on the other. The researcher can ask whether people who endorse dialogic inner speech more strongly report more clearly coded turns in sampled moments. They must decide how many prompts are needed, how to handle ambiguous reports and whether multiple coders agree. A moderate relationship might reflect genuine differences between habitual self-understanding and sampled events rather than mere error.
Laboratory elicitation tests a narrower capacity
In a laboratory, an investigator can ask a participant to imagine a debate or respond to a problem from two perspectives. This creates a time window for behavioral or neural measurement. Alderson-Day and colleagues used an elicited dialogic-inner-speech task in an imaging study and reported activity patterns consistent with the distinction they sought to test. Such work can ask whether imagining an internal dialogue differs from producing a monologue under controlled instructions.
The same control changes the claim. Following an instruction to alternate positions proves that a participant can perform the task; it does not show how often dialogue arises spontaneously. The task may recruit verbal working memory, social perspective-taking, auditory imagery, attention to the prompt and task compliance. A neural contrast between imagined dialogue and monologue may reflect several of those differences. It cannot be read as a direct recording of the words of a private conversation or as a unique neural signature of dialogic thought.
Task design can reduce some alternatives. Match dialogue and monologue conditions for duration, topic, word-generation demand and emotional content. Counterbalance order. Ask participants afterward what they actually did, without assuming instructions guarantee compliance. Include a nonverbal perspective-switch condition if the theoretical question is whether the signal concerns verbal exchange rather than alternation of viewpoints alone. These controls sharpen inference; none can make private experience fully observable.
Articulatory suppression, in which participants repeat an irrelevant sound while doing another task, can test whether verbal resources contribute to performance. But a drop under suppression is not a selective measure of dialogic inner speech. Repetition can disturb attention, timing, working memory and task strategy. Conversely, no drop may mean participants switched to another method or that the task was insensitive. The manipulation should be one constraint in a convergent design, not a standalone detector.
Keep related phenomena apart
Inner speech is broader than dialogic inner speech. A silent plan, a rehearsed sentence and a brief word fragment can all be verbal thought without a conversational structure. The companion Methods & Measures entry on inner speech addresses whether verbal thought occurs at all; this entry asks whether an identified episode has alternating turns or positions. Counting all inner speech as dialogue artificially inflates prevalence and loses the contrast of interest.
Condensed inner speech is another neighbor. A participant may report “yes, no, later” as a recognizable exchange despite the absence of full syntax. Condensation and dialogicality can co-occur, but they are not the same dimension. A questionnaire that treats abbreviated thought as absent may systematically underestimate dialogic quality for people whose internal language is sparse. Researchers can code the two features separately and test whether their association varies by task.
Social perspective-taking and imagined conversation also overlap. A person can reason about what a colleague would say without experiencing distinct internal speech turns. Another can vividly imagine a colleague’s reply. Either may be useful for planning, but the task instructions and report should distinguish the inferred viewpoint from the experienced form. The philosophical discussion of inner speech and thought by contemporary authors is valuable for defining these categories; it is not itself a measurement validation study.
Rumination and self-criticism provide a further trap. Repetitive negative thought can be monologic, dialogic or largely nonverbal. A high dialogicality rating is not, by itself, evidence of a mental-health problem. The original VISQ study examined associations with psychopathology-related variables in a young-adult sample; it did not establish a diagnostic threshold or a causal path from internal dialogue to symptoms. Any clinical interpretation requires a different evidential standard and attention to content, distress, control and context.
Evaluate validity at the level of the claim
For a group experiment, the key question may be whether an instruction changes reported dialogic form relative to a matched condition. Random assignment, manipulation checks and clear coding may support that claim. For an individual-difference study, a one-time subscale score needs reliability and a predicted pattern of convergence with momentary reports. For a developmental question, age-appropriate language and comprehension matter; a child who does not understand “inner voice” should not be scored as lacking inner dialogue.
Researchers should report the denominator. “Forty percent of prompts were dialogic” means little unless readers know how many prompts were sent, answered and coded, and how an episode was bounded. If a person described a ten-minute internal argument at one prompt, does it count once, or for the whole period? Event rate, duration and perceived importance are different outcomes. Predefining them prevents a study from shifting to whichever produces the most striking number.
Inter-rater coding is valuable for open reports. Coders can be trained on examples of explicit turn-taking, inferred perspective change, imagined external speech and ambiguous wording. Report agreement and adjudication rules rather than hiding uncertainty. Some descriptions will remain indeterminate. A category for “unclear” is more honest than forcing them into dialogue or monologue, especially when the article’s main claim depends on small differences between groups.
Ecological validity needs a direct bridge. A controlled task can show that people produce inner dialogue when invited; daily sampling can show that similar forms appear outside the laboratory; a questionnaire can capture a person’s general interpretation. If these lines converge in a specified population, confidence increases. If they disagree, researchers should examine task demand, wording, recall and context before declaring one method correct. No single method licenses a universal statement about how “the mind” normally speaks.
The Institute’s practical checklist is to specify the form, timeframe and inference first; choose an instrument whose questions match them; record ambiguous cases; compare at least two methods when feasible; and make the conclusion no broader than the observed evidence. These are methodological recommendations derived from the cited work, not a new validated scale.
What the evidence does not show
A VISQ or VISQ-R score does not directly count private conversational turns, reveal their complete content or establish a fixed personality type. A sampled report does not prove that the sampled moment was typical of a person’s day. An elicited laboratory dialogue does not show that the same process occurred spontaneously. Neural activation during an elicited task does not decode the individual’s inner words or uniquely identify dialogue apart from verbal imagery and perspective shifts.
The cited studies do not show that all people think in an internal voice, that verbal thought is necessary for reasoning, or that dialogic inner speech causes either psychological distress or skill. Associations in selected samples require replication and careful control before individual clinical use. A person who reports little dialogic speech may still plan, deliberate and understand other people through other routes.
Measurement remains possible when its target is modest: a defined reported form of experience under a stated method, with uncertainty attached. The participant in the opening example may indeed have an inner exchange, but the researcher learns that only by asking what was experienced, when it happened, and how the answer was obtained. Those distinctions make the measure useful without pretending to observe private thought directly.
Research references
- McCarthy-Jones, S., & 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(4), 1586–1593. View source
- Alderson-Day, B., Mitrenga, K., Wilkinson, S., McCarthy-Jones, S., & Fernyhough, C. (2018). The Varieties of Inner Speech Questionnaire–Revised (VISQ-R): Replicating and refining links between inner speech and psychopathology. Consciousness and Cognition, 65, 48–58. View source
- Alderson-Day, B., & Fernyhough, C. (2015). Inner speech: Development, cognitive functions, phenomenology, and neurobiology. Psychological Bulletin, 141(5), 931–965. View source
- Alderson-Day, B., & Fernyhough, C. (2015). Relations among questionnaire and experience sampling measures of inner speech: A smartphone app study. Frontiers in Psychology, 6, 517. View source
- Hurlburt, R. T., Heavey, C. L., & Kelsey, J. M. (2013). Toward a phenomenology of inner speaking. Consciousness and Cognition, 22(4), 1477–1494. View source
- Alderson-Day, B., Weis, S., McCarthy-Jones, S., Moseley, P., Smailes, D., & Fernyhough, C. (2016). The brain’s conversation with itself: Neural substrates of dialogic inner speech. Social Cognitive and Affective Neuroscience, 11(1), 110–120. View source
- Fernyhough, C., & Borghi, A. M. (2023). Inner speech as language process and cognitive tool. Trends in Cognitive Sciences, 27(12), 1180–1193. View source