As Gauker (1994) puts it, there are two ways that a person can talk to themselves: out loud and in silence. In the context of inner speech, psychologists use the terms egocentric and/or private speech to refer to the speech heard when a person is talking to themselves. This can be distinguished from normal speech because egocentric speech is usually only meant to be heard by the speaker as a set of instructions or a way to keep focus. Talking in silence, on the other hand, is not so easy to define. Since inner speech is such an abstract process, many people disagree on what can be classified as inner speech.

One way to define inner speech is that it is the articulation of words without sound. Almost as if somebody had taken a TV remote and muted your voice. In a study on articulation of silent speech Oppenheim and Dell (2010) wrote, “...we have demonstrated that articulation changes inner speech, and this demonstration implies that inner speech cannot be independent of the movements that a person would use to express it” (p. 1158). This is not entirely false. You can have silent articulation of words, but as a result of inner speech, it should not be considered inner speech itself. This silent articulation of words is what we perceive as that vivid voice we “hear” in our heads. This vivid voice, or inner voice, is a form of verbal imagery (Gauker, 1994). Verbal imagery is the way we observe our own inner speech. Another belief is that inner speech is the same as thought. It is important to know that thought and inner speech are not one in the same. Inner speech is more like the carrier of thought (Solokov, 1995), just as the silent articulation of words can sometimes be the carrier of inner speech.

To further understand the distinction between silent articulation and inner speech, below (Figure 1) is Brown's model of the levels of speech. Brown (2010) says that inner speech involves incomplete perception and incomplete articulation. This fits with our previous definition of inner speech as a carrier of thought because there would not be a need for articulation and since the speech is being produced in the mind, there would be less activation of perceptual areas than if it had been said by another speaker.

As I said before, researchers first studied inner speech by observing private speech in children. The reason this worked is because children aren't born with inner speech but rather develop it as they grow older (Vygotsky). Jean Piaget had a different theory of development (Figure 1). Instead of inner speech developing with age, Piaget believed that inner speech was a result of children losing their egocentricity as they become more aware of others. As the child becomes less self-centered they learn to socialize with others and then eventually learn to keep their thoughts in their head. (Oates, 2005)

Figure 2: Piaget’s development of inner speech.

Vygotsky was Piaget’s biggest critic on his theory of inner speech. Lev Vygotsky, a soviet developmental psychologist, believed that children first learn to be social with others. Then, as they learn to internalize thoughts, they develop private speech. With more internalization and a development of higher cognitive abilities, the child is finally able to have inner speech (Figure 2). The main difference is that Vygotsky did not see egocentric speech as having disappeared. Egocentric speech was merely placed as a backup mechanism for internalized thought and self-guidance.

Figure 3: Vygotsky's development of inner speech.

Vygotsky also wrote about the properties of inner speech and he was clear he believed inner speech was more than just “speech minus sound” (Vygotsky 1986, p. 235). The main feature that Vygotsky and previous researchers noted about inner speech is that it has very chaotic syntax. With inner speech it is common that the entire subject and all the words associated with it be omitted, leaving only the predicate. Vygotsky (1986) gives an example of when such a predicate-only utterance happens in external speech:

…let us imagine that several people are waiting for a bus. No one will say, on seeing the bus approach, “The bus for which we are waiting is coming.” The sentence is likely to be an abbreviated “Coming,” or some such expression, because the subject is plain from the situation. …If the thoughts of two people coincide, perfect understanding can be achieved through the use of mere predicates…(p. 236)

This type of external speech relates more closely to inner speech. With inner speech there needs to be no reminder of what the subject is because only the speaker is to be concerned about the action and the speaker would already have this knowledge. You are the speaker and the audience. This led Vygotsky to deduce that inner speech relies heavily on semantics, since there does not need to be any context involved with inner speech like there would be with overt speech so that a word or phrase can be understood. There are three semantic peculiarities that Vygotsky found:

The sense of a word involves the full range of psychological responses we produce when we hear a word. It involves how a word makes us feel, what it makes us remember, and especially what it means. Meaning is important, but since there is no context to limit the meaning, the sense of a word is what is being observed with inner speech.

Agglutination with inner speech has to do with combining words rather than binding affixes to words. In this case the root word consists of the main idea and all the other words attached to it contribute to making that idea even more complex. Since it is still only a compound word though, a complex idea can be expressed in a ridiculously simple form.

The last characteristics of inner speech semantics is almost like a joining of the previous two. If sense of a word is what matters and words are combined often, then it’s hard to deny that the senses of different words also combine. Vygotsky says that words flow into each other, constantly influencing or absorbing each other (1986). An example of this is how a word from a song can remind you of the entire song, bringing along all the psychological activity that the entire song once made you feel. That one word embodied the sense of the entire piece of work.

In addition to these peculiarities of syntax and semantics, Vygotsky also mentions the used of a specialized argot by each individual: a secret language composed of experiences and different interpretations known only to the speaker. Because of this and the unique syntax and semantics rules, translating inner speech into overt speech would be almost impossible. Vygotsky’s linguistic characteristics of inner speech also show that inner speech is not the same as silent articulation of words because, if voiced, the speech would be incomprehensible.

Before the availability of imaging early linguists took a try to locate and identify inner speech. Weirnicke had believed that the arcuate fasciculus had something to do with inner speech and even went on to say that it was related to conduction aphasia. His proposition of the role of the arcuate fasciculus was denied and is yet to be refuted (Brown, 2010). Goldstein said that inner speech was somewhere between perception and production and that it's deficit had to do with Central Aphasia, but Goldstein never connected inner speech and thought. Brown also mentioned that Luria thought that a deficiency of inner speech was responsible for dynamic and afferent aphasia. None of these assumptions have been present in any articles I have found, but the disorder of schizophrenia has been able to shed some light on the location of inner speech.

To test whether or not hallucinations in schizophrenics are caused by a disorder of inner speech monitoring, McGuire et al. (1995) had a group of hallucinating schizophrenics, non-hallucinating schizophrenics, and a normal control group do an inner speech task. The subjects had to read a set of words, and then with their “inner voice” make up a sentence using one of those words. Then the subjects had to imagine an “alien voice”, as if someone else was speaking besides themselves, reciting the same sentence. A PET scan of the subject showed a much higher increase of activation in the rostral supplementary motor area and the left middle temporal gyrus for the non-hallucinators and the control group compared to the hallucinators. It is possible that these areas might have something to do with the production and monitoring of inner speech.

In a separate experiment to compare the neuroanatomy of auditory imaging and inner speech, Mcguire et el. (1996) found that when inner speech was compared to the control and auditory imaging conditions, inner speech showed activation primarily in the left inferior frontal gyrus (Figure 4).

Figure 4: "Location of areas where rCBF was increased when the inner speech and the auditory verbal imagery tasks were compared with the control condition, illustrated on Statistical Parametric Maps (SPMs). The spatial distribution of pixels where there were significant changes between conditions is illustrated on orthogonal projections on the left side of the figure, while their distribution with respect to the medial and lateral surfaces of the two hemispheres is shown on the right. Compared with the control task, inner speech (upper images) was associated with increased activity in the left inferior frontal gyrus. Auditory verbal imagery (lower images) was associated with activity in the same region, and in the left premotor cortex, the SMA, the middle and superior temporal gyri, and the right insula." (McGuire et al., 1996)

Mcguire et al. used the same process as the previous study. The only change was that none of the subjects showed any abnormality clinically or otherwise. The procedure continued to be reading of words for control then sentence creation as inner speech, and lastly imagining a foreign voice for verbal imagery. With the increased activation of Broca's area and the lack of activation from Wernicke's area, it could be possible that inner speech is only based on production and perhaps areas with more semantical purpose. It is strange that Wernicke's area did not show significant activation with inner speech as it is involved with perception. This only reinforces the notion that inner speech is not the same as verbal imagery.

In school age children, it has been shown that the use of private speech increasing success in problem solving tasks (Winsler, Diaz, Montero, 1997) and in the success of planning-related tasks (Lidstone, Meins, Fernyhough, 2010). This is supportive of Vygotsky's theory that cognitive development depends on private speech. In the study by Winsler et al., children were asked to look at two pictures and then, based on object or color, find what was similar about the two pictures. Children using private speech related to the task did better on average than children who were silent or using irrelevant private speech (Figure 5). Lidstone et al. had children solving a Tower of London, but instead of encouraging private speech, they had the children suppress it. They found that while suppressing private speech, children had more difficulty with the Tower of London task.

Figure 5: Number of successful items with item-relevant speech, item-irrelevant speech, and silence, as a function of age. (Winsler, Diaz, Montero 1997)

If we look at Vygotsky's theory of inner speech and remember that he says inner speech is an internalization of private speech, then we can easily relate these results to adults and inner speech. This can already be observed through our everyday task performance and self-guidance. No study was found on adult performance and use of inner speech, but I feel safe saying that when we use inner speech to coach ourselves through a task, that our performance is better than just random responses from our bodies.

Another function of inner speech is said to be short-term memory enhancement. Waugh says that “the span of verbal short-term memory—auditory or visual—is thought to be sustained by silent rehearsal of the sequence” (as cited by Levine, Calvanio, Popovics, 1982). This makes sense because we know that the continuous rehearsal of a task helps encode to actions into nonepisodic memory.

In their case study, Levine, Calvano, and Popovics (1982) looked at a patient, named E.B., who suffered a stroke without any implications of bad health. Although the location of the damage was not given, it was said that E.B. had an incomplete right hemiparesis, or weakness of one side of the body, which can lead us to assume that the stroke caused some damage to the left hemisphere. As a result of the stroke, E.B. was unable to produce any sort of speech and claims to have lost the ability to talk to himself and use verbal imagery, which Levine et al. saw as losing inner speech. Levine et al. then go on to challenge the need for inner speech in the performance of problem solving tasks and short-term memory since patient E.B. excelled in these two categories. The way E.B. was able to do these things, he says, was by using visual imagery. I have a different interpretation of Levine et al.'s findings. First off, I feel that E.B. did not lose his inner speech, merely the way to express it to himself. Levine et al. had defined inner speech as “...an auditory-articulatory image of speech without uttering a sound,” and the “ability to appreciate the auditory-articulatory structure of speech” (p. 391). What they seem to define is inner speech in the form of verbal imagery which, as we know, is not the essence of inner speech, merely a carrier. Second, I hypothesize that even though E.B. claims to have lost the ability to talk to himself, he still uses inner speech to identify the visual imagery that he needs to recall a thought. As part of his process of calculating, E.B. needed to have the names of the numbers in the equations to actually perform the calculation (Figure 6).

Figure 6: "An example of E.B.‘s attempt to do written arithmetic." (Levine et al., 1982)

The names of the number must hold some sense that allows him to do the arithmetic, and this would involve using inner speech. In his examination of Dr. Jill Bolte Taylor's book, “My stroke of insight,” Morin (2009) points out that Jill, after damage to the fronto-temporal junction via a stroke, reported losing her inner speech and using visual imagery to then think. Dr. Jill also had the wrong notion of inner speech, like Levine et al., as reported by Mitchell (2009) and Schlinger (2009), but it is interesting to see that both Jill Taylor and E.B. reported using only visual imagery after their stroke. Perhaps it would be interesting to look into the use of visual imagery and the expression of thought. This raises a question for later research: Is there a system of “inner sight” that is similar inner speech or, since language is said to be prerequisite for thought, does inner speech use both auditory and visual imagery as its carriers?

Conclusion

Inner Speech will be easier to study and experiment on as soon as the definition for it is agreed upon. The problem is that it is so abstract that it is difficult for any two people to truly understand what each other means when they use the term inner speech. Most of the time there is a bias to think of inner speech as being auditory imaging. The way that I understand it is that there are levels of expression. At the highest is thought itself. Then underneath that level is inner speech, which thought uses to express itself. At the third level is verbal, and possibly visual, imagery. This third level is used to express inner speech. Underneath these three levels is then oral or signed language, and these are used to express the imagery. I have to say that I have a strong preference toward the Vygotskian theory of inner speech. It was interesting to find the amount of work on inner speech that came from soviet psychologists, including Luria, but I am glad that it has gotten enough attention that neurologists have tried to localize it. Just like many higher level cognitive processes, there is no definite location, but the ones that appear to be the most relevant are the rostral SMA, the left MTG, and the left IFG.