The process of internalizing speech was deeply studied and theorized by Vygotsky (2012). He proposed that private speech (PS) was an intermediate step between social speech and inner speech, but he also attributed it a role in self-regulating activity. For him regulation first occurred due to the influence of adults in social speech and later transferred to self-regulation because of PS. Finally, the regulation became internalized in inner speech.

This conception of PS was aligned with his general notion that functions appear two times in development: first in the social space and then in the internal or mental space (Vygotsky, 2012). Such conceptions have been deeply influential in studying the process of the genesis of self-regulation in preschool children (see Winsler, 2009).

One of the most influential coding schemes for studying this process of internalization was proposed by Berk (1986). In this coding scheme, one first needs to separate the utterances produced by the child in the condition selected by the experimenter (e.g., play) according to temporal and semantic criteria (Winsler et al., 2005). Then, one divides the speech according to whether it is social or PS. Social speech is coded when there is physical or visual contact, when the context refers to someone or something that was said, or when it is temporarily related to the speech of another individual. Everything else is considered PS. Further, PS is classified as follows: (1) level 1 if PS (PS1) is irrelevant to the task, word play or repetition, emotional expression irrelevant to the task, or commentaries to absent or imaginary characters; (2) level 2 if PS (PS2) is relevant to the task, describes the child's own activity, is self-guided commentary, is a self-answered question, or is an emotional expression relevant to the task; and (3) level 3 of PS (PS3) if PS is externalized inner speech relevant to the task (e.g., verbal murmurs, whispers, and lip and tongue movements).

To elicit this kind of speech, many conditions have been used, such as naturalistic settings, social settings, free play, planning tasks, constructive tasks, memory tasks, and response inhibition tasks (Berk, 1992; Fernyhough and Fradley, 2005; Winsler et al., 2005; Fernyhough and Meins, 2009; Winsler, 2009). The common denominator is that the task is significant and aligned with the level of development such that it is significantly challenging.

Such a scheme has been highly fruitful not only in studying the process of internalization but also in examining its relationship with other cognitive functions or investigating PS in atypically developing populations (please refer to Winsler, 2009). This scheme has been used to study PS across the lifespan (Berk, 1992; Winsler, 2009). The study on PS in the specific language impairment population (Lidstone et al., 2012) has been productive in studying PS in the ADHD and ASD populations (Winsler et al., 2007). For studying microgenetic (Benigno et al., 2011) and transversal relations between PS and executive functioning (Fernyhough and Fradley, 2005; Alarcón-Rubio et al., 2014; Thibodeaux et al., 2019) as well as the ontogenetic relation between PS and theory of mind (Fernyhough and Meins, 2009; Rivera, 2023), Berk's coding scheme has been highly fruitful. However, it is important to note that Berk's coding scheme is not the only coding scheme (please refer to Winsler et al., 2005) available for studying the internalization process of the self-regulation function.

Galperin was a member of the Kharkov school of pyschology that was headed by Leontiev and where other important psychologists such as Vygotsky, Luria, Bozhovich, and Zaporozhets worked. He integrated the ideas of internalization of Vygotsky (2012) with the developments of object-centered activity developed by Leontiev (1981). He developed a formative method to study the internalization process (Galperin, 2021). Such a procedure was called the stage-by-stage formation of mental actions. Moreover, such a procedure allowed Galperin to study the formation of mental actions in connection with object-centered activity at the material/materialized level (i.e., actions with concrete objects or their representations), thus informing, in a controlled way, of the stages in the formation of actions.

According to Galperin, an action traverses multiple qualitative changes or forms of action (FA): material/materialized, externalized social speech, silent external speech, and mental FAs (Galperin, 1967, 1992, 2021). At the material/materialized level, people interact with the material supports needed to solve a particular problem. These can include real objects, graphic representations, and, most of the time (in the context of the experiments), an orientation card that helps the learner solve the problem. After this level was accomplished, the next stage (externalized social speech) consisted of the interaction between the learner and the teacher, where they resolved the problem in collaboration, but especially because the teacher oriented the learner in the space problem (e.g., asking for the steps, correcting him, etc.). Later, the learner would orient himself/herself in the problem space using his/her own speech, which would gradually become more and more abbreviated. Finally, the motor aspect of speech would disappear entirely. At this level, Galperin (2009) suggested that, at this point, the learner would operate at the level of meaning or “pure thought,” which he referred to as the mental form of action.

In this study, we indicate that his internalization stages could be implemented in the studies of PS. We also propose a new coding scheme for PS using the internalization forms of action: material/materialized, externalized social speech, silent external speech, and mental.

However, we recommend retaining the PS coding based on Vygotsky (e.g., such as Berk's) but reclassifying it for continued use. This coding scheme can still be useful for ontogenetical purposes, discourse analysis, and content analysis of the utterances, among others, and shedding light on the transitional process from external, inaudible speech form to mental form. Galperin's proposed forms of internalization include material/materialized, externalized social speech, silent external speech, and mental forms of action. We suggest incorporating the specificity of the literature on the types of PS into Galperin's scheme. We also propose recoding the PS of children to incorporate Galperin's notions of internalization.

The classification is proposed as follows: (1) material/materialized when the concrete objects or their representations are used to solve a task; (2) external social speech when the majority of the utterances are of the social type; (3) audible external speech when the majority of the utterances are of the PS2 type; (4) inaudible external speech when the majority of the utterances are of the PS3 type; and (5) or mental when there is one or no event of speech at all while resolving tasks. This classification would add even more specificity to Galperin's conception since he did not consider audible external speech as others have (Berk, 1986; Winsler, 2009: Winsler et al., 2005) and would shed some light on the complex process of passing from audible external speech to a mental form of action.

Because of the previous considerations, an exploratory analysis was performed to discern if the proposed re-coding by the FA is an appropriate categorization for studying the process of internalization across the preschool years. Does the classification of the FA distinguish between different preschool children (e.g., first and second grade of preschool)? Is this classification better in some respects to other kinds of classification of private speech? Is the reclassification by FA redundant, or does it present new information compared to other classifications?

In Tables 1–3 descriptive statistics of PS, degree of internalization and FA in the preschool grades are presented. Figure 1 presents a graphic of the amount of FA according to the preschool grade. Descriptive statistics showed that children from preschool 1 tend to have a lower degree of internalization than children from preschool 2 and 3. The results showed that preschool 1 only has external social speech FA, while children in preschool 2 have external social speech as the dominant form but also have inaudible external speech, audible external speech, and in last place mental FA; finally, preschool 3 children have as a dominant FA inaudible external speech, audible external speech, mental, and lastly they have external social speech. Therefore, in was not until children reached preschool 3 that inaudible external speech as a dominant form.

The results showed that the degree of internalization was increasing significantly across preschool grades, which is consistent with previous results in the literature (Fernyhough and Meins, 2009; Winsler, 2009). Further, the findings revealed that speech type was not a satisfactory category for studying the changes in performance across ages. Some types of speech were not effective in determining the performance of the task. Categorization by speech type was not able to show differences in scores, but it did show that those with predominantly PS2 and PS3 types were faster than those with social speech. Nonetheless, the degree of internalization was important for differentiating the children at different ages. Therefore, it is an important marker of the ontogenetic development of children.

The categories of speech presented us with the problem of what to do with the cases where kids tend not to talk. This was especially true for older children, but when we collected data, it was clear that even some younger kids seemed to have little to no speech. Because of that, it was critical to consider other aspects, such as the relationship between performance and lack of speech. The latter point is important since many kids may not want to talk even if prompted to do so, as suggested by Fernyhough and Fradley (2005). This could be for many reasons, such as unfamiliarity with the researcher, embarrassment, cultural reasons (e.g., that you should not talk to strangers), and anxiety about tests.

Further, FA coding schemes solved the problem of the unclassifiable lack of speech and also showed to be effective in identifying differences between participants even without the consideration of the material/materialized in this study; older children tended to have inaudible external speech and audible external speech FA while younger children had mainly external social speech. FA classification showed that children with FA that signaled a more internalized action had better performances in time and scores during ToL tasks. The last point was especially clear when we considered the coding scheme of Galperin himself rather than our own, in which we attempted to consider the more subtle forms of PS in the literature (see Winsler et al., 2005; Winsler, 2009). Galperin's account ignored the differences between PS relevant to the task and those not and seemed to ignore the subtleties between audible and non-audible external speech. Ignoring such subtleties seemed correct when we were concerned with the performance in the ToL task (i.e., time and score). Our coding scheme was not effective in determining changes in scores and only revealed that those with external social speech had lower scores than those with inaudible external speech. Galperin's approach, on the contrary, revealed that those with external social speech had lower scores than those with inaudible external speech and mental FA. Then, when we considered the time of execution, our approach determined that those with an external social speech FA took longer to solve the ToL task than those with an external audible, inaudible, or mental FA. However, Galperin's account found that both (i.e., external inaudible speech and mental FA) consumed less time to resolve the ToL task and that those with a mental FA resolved faster than those with an external inaudible form of action.

The last point is consistent with Galperin's approach. He proposed that, in the last stages of internalization, one of the specific changes that occur is automatization of action. This is consistent with the fact that the main difference between mental and inaudible external speech is in time but not in scores of ToL (Galperin, 2021).

Therefore, our results suggest that classifications based on the internalization of actions are a more suitable scheme for classifying the internalization process than speech-type classifications, at least when performance is an issue. This does not imply that such schemes are not useful. On the contrary, we believe that such classification schemes may be useful when the content of the speech is relevant or when the subtleties of PS are of interest. We also believe that the degree of internalization markers is an important quantitative measure for making ontogenetic claims about the internalization process. Nonetheless, we believe that the categories of the FA may be more useful when we want to consider changes in performance since it seems that further subdivisions of speech are redundant for performance (i.e., scores and time). These categorizations would be of great use in neuropsychological neurorehabilitation, where the monitoring of the process is crucial, and the focus is on performance during microgenetic processes.

One concern that the reader may have, as some reviewers pointed out, is that Galperin's approach is normally associated with such formative experiments. This is study is not one of those studies. We agree with them, Galperin is specially associated with that, and that aspect should not be ignored but embraced since it is a very rich description of how to form an action. Nonetheless, we believe that Galperin's intentions were broader. His purpose was to describe the process of internalization broadly, and he found in such a formative method a fruitful way to approach the problem in a controlled manner (Arievitch and Van der Veer, 1995). Because of that, the study of how to apply his categories to ontogeny is valid, and it is in need of more research. However, as we mentioned earlier, we believe that such a process is going to be useful, and is going to be fruitful in such contexts where formation of actions is needed (e.g., neuropsychological rehabilitation; see Engeness and Lund, 2020). For such purposes, we believe that our coding scheme is going to be very useful since it provides a systematic way of studying such formative processes.

The limitations of the study are clear. Since it was only an exploratory study, no clear answers on the topic can be made presently. However, it will be a task of the future to explore the utility of these classifications in other practical cases, especially in the neuropsychological rehabilitative and pedagogical processes.

This exploratory study successfully showed that PS recoding is a valuable addition to the study of internalization. Coding by the FA proved to be effective in distinguishing between children of different ages and levels of internalization. Further, this coding scheme was useful for differentiating between children according to their time and score performances, which is contrary to the type of speech classification. The latter was not sensitive and was inadequate for those children who performed better but did not speak. As mentioned, this re-coding scheme seems promising for studies that need to monitor the changes in performance as a function of time, teaching, and a rehabilitation process. Future studies will clarify the utility of these coding schemes.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Facultad de Psicología BUAP. Written infor med consent to participate in this study was provided by the participants' legal guardian/next of kin.

LR conceived the initial idea, recollected, transcribed and analyzed the data, and wrote the manuscript. LR and VL discussed and reviewed the theory, methodology, and the final manuscript. VL and MG gathered the resources for publishing and obtained the location for gathering data. All authors contributed to the article and approved the submitted version.