The Association Between Pointing and Walking in Infants: A Longitudinal Observational Study

ABSTRACT

The relation between pointing and walking in infants was investigated through a 1‐year observation study at a daycare center in Tokyo, Japan. The participants were 23 Japanese infants (mean age 13.1 months, 8 boys and 15 girls) from middle‐SES families. Data from each infant were analyzed at 4 months before and after the onset of walking. It was shown that as infants develop locomotion, the frequency of pointing, the proportion of social pointing, and the proportion of social pointing accompanied by movement and looking behavior increase, while the proportion of pointing accompanied by prior interaction decreases. These changes were suggested to be more strongly associated with the acquisition of walking than with the infants' age itself. Furthermore, for pointing without prior interaction, it was shown that when pointing was combined with the infant's movement, it tended to increase the success of attentional sharing through pointing.

1. Introduction

1.1. Development of Pointing

Pointing is a communication skill that is unique to humans; thus, the developmental process of pointing has been the subject of much research (e.g., Tomasello ²⁰⁰⁸; Tomasello et al. ²⁰⁰⁷). The purpose of this study is to investigate whether quantitative and qualitative changes can be observed in the pointing behavior of infants after they have learned to walk.

Pointing is observed from shortly after birth, for example in 2–3‐month‐old infants (Fogel and Hannan 1985); however, early pointing is not considered a means of communication (Carpendale and Carpendale 2010; Paulus et al. ²⁰²³) and can therefore be regarded as non‐social pointing. This type of pointing includes behaviors such as pointing at an object to identify its location or to indicate where the infant's attention is focused. Non‐social pointing is based on dyadic engagement, that is, person‐to‐person or person‐to‐object relationships that allow infants to engage in interpersonal interactions and object manipulation. Developmentally, social pointing emerges later, at around 9–15 months of age (Camaioni et al. 2004; Carpenter et al. ¹⁹⁹⁸; Paulus et al. ²⁰²³), and it is a form of triadic engagement involving person‐to‐person‐to‐object relationships (e.g., Bakeman and Adamson ¹⁹⁸⁴; Scaife and Bruner ¹⁹⁷⁵; Rochat and Striano ¹⁹⁹⁹; Tomasello ²⁰⁰⁸). Since triadic engagement is thought to emerge at around 9 months of age, Tomasello (1995) called it the “9‐month social and cognitive revolution.”

1.2. Locomotion Development and Developmental Cascades

Like social pointing, showing or handing an object to others are behaviors based on triadic engagement, and quantitative and qualitative changes have been reported with the acquisition of walking. Toyama (2020) observed infants' object interactions longitudinally in a daycare center and found that crawlers often interacted with others without sharing objects; however, after they started walking, infants showed or passed objects to caregivers more often or displayed their manipulation activities to them. Other studies showed that when interacting with mothers, infants who were beginning to walk tended to seek emotional responses to objects (Clearfield et al. 2008; Walle ²⁰¹⁶) and make gestures and vocalize body movements to their mothers (Clearfield 2011; West and Iverson ²⁰²¹).

In addition to the triadic engagements, recent studies have shown that locomotor development appears to affect a wide range of developmental domains. For example, walking infants are more likely than crawling infants to carry objects (L. Karasik et al. ²⁰¹²; L. B. Karasik et al. ²⁰¹¹; Toyama ^{2020, 2023a, 2023b}) and intentionally touch the bodies of others (Toyama 2022). Learning to walk has also been associated with engaging in more exploration of spatial relations (e.g., Oudgenoeg‐Paz et al. ²⁰¹⁵), visual exploration (e.g., Mulder et al. ²⁰²²), infant‐caregiver communication (e.g., West and Iverson ²⁰²¹), and language (e.g., Walle and Campos ²⁰¹⁴).

The phenomenon in which the acquisition of walking seems to promote the development of domains different from locomotion is called a developmental cascade (Adolph and Hoch 2019; Campos et al. ²⁰⁰⁰; Oakes ²⁰²³; Oakes and Rakison ²⁰²⁰; Walle ²⁰¹⁶). Why does the cascade effect of walking acquisition occur? As Adolph and Tamis‐LeMonda (2014) summarized, learning to walk has many benefits for infants: “going more,” “seeing more,” “playing more,” and “interacting more.” Infants who have just begun to walk can move faster over longer distances. New walkers are more likely to fall than experienced crawlers, but once walking begins, locomotion develops rapidly, and infants are able to travel longer distances in less time and fall less often (Adolph et al. 2012). Walking is a more efficient mode of transportation than crawling. One of the differences between crawling and walking is that while walking, an infant moves while maintaining an upright posture. Because infants' eyes are higher up, their field of vision becomes wider, and they are able to see continuously without interruption as they move (Franchak et al. 2018; Kretch et al. ²⁰¹⁴). Perhaps owing to their wider field of view, walking infants have access to distant objects (Dosso and Boudreau 2014), look at caregivers' faces more (Franchak et al. 2018), and see their caregivers from further away (Yamamoto et al. 2019). The increased exploratory nature of movement routes in walking infants (Hoch et al. 2019; Toyama ²⁰²¹) is likely related to the increased stimuli viewed during movement.

In summary, the developmental cascade effects associated with learning to walk can be thought of as the result of a complex combination of different factors, such as the ease and efficiency of movement, the ability to move while maintaining a wide field of vision, and the ability to free the hands from the means of locomotion that are characteristic of walking. Social pointing is also based on triadic engagement, as is showing or handing something to another person. Therefore, it is assumed that quantitative and qualitative changes occur in infants' pointing as they learn to walk. However, little is known about the relationship between pointing and locomotor development in infants.

1.3. Previous Studies on Infant Pointing

The fact that infant pointing has rarely been studied in terms of locomotion may be related to the research methods used in previous studies. Previous studies on infants' pointing used diary methods in which parents were asked to keep records of their infants' pointing (e.g., Carpendale and Carpendale ²⁰¹⁰; Paulus et al. ²⁰²³), experimental methods using tasks to examine pointing, joint attention and gaze‐following (e.g., Behne et al. ²⁰¹²; Desrochers et al. ¹⁹⁹⁵), and experimental observation methods using posters and toys in the laboratory (i.e., decorated room) to induce pointing while observing (Liszkowski et al. 2012; Liszkowski and Tomasello ²⁰¹¹). Parental reports and experiments do not allow the observation of infants' movements, and experimental observation methods often involve holding infants and restricting their movement (e.g., Liszkowski et al. ²⁰¹²; Liszkowski and Tomasello ²⁰¹¹; Lüke et al. ²⁰¹⁹). Given these considerations, observational data that allow infants to move freely are essential for examining the relation between pointing and locomotion.

1.4. The Current Study

In this study, longitudinal naturalistic observations were conducted in an infant class at a daycare center in Tokyo, Japan. In Japan, at the beginning of the school year, infants aged 3–11 months are enrolled in daycare classes designated for infants aged 0–1 year. By the end of the school year, the infants are between 15 and 23 months old. This period coincides with key developmental milestones such as crawling, cruising, and walking. Therefore, analyzing observational data from this period allows us to clarify the relationship between pointing and the development of locomotion.

In this study, we conducted two analyses. The first (Analysis 1: development of pointing) examined what changes occur in pointing before and after the acquisition of walking. Previous studies of developmental cascades have shown that as infants begin to walk, they also tend to “move more,” “see more,” “play more,” and “interact more” (Adolph and Tamis‐LeMonda ²⁰¹⁴). Accordingly, it can be assumed that infants' pointing not only increases quantitatively after they learn to walk but also undergoes qualitative changes in combination with seeing and moving actions. Based on the results of Analysis 1, Analysis 2 (Development of Social Interaction through Pointing) will examine how infants' pointing combined with moving toward and looking at their partner changes their communication.

2. Observation Method

2.1. Observation

Longitudinal observations were conducted at approximately 1‐week intervals in a class of infants aged less than 1 year attending a Tokyo daycare center in FY2017, FY2018, and FY2019, before the spread of the new coronavirus. The infant class had a capacity of eight infants and was staffed by three full‐time and several part‐time caregivers. Infants from the age of 3–11 months are accepted at the beginning of the school year. The time of arrival at the center varies from infant to infant, but by 9:00 a.m., all infants have arrived and play freely until lunch time. At the beginning of the year, many infants cannot walk and spend most of their time playing indoors. However, as more infants learn to walk, they begin to take walks in the nursery garden or outside the nursery. At first, one caregiver feeds each infant lunch, but when the infants are able to eat on their own, the caregivers begin to feed several infants at a time. The time for lunch varies from infant to infant, but by approximately 10:30 to 11:30, all infants will have eaten and gone to nap. After their nap, they have a snack and then free play time as in the morning, and they go home between 4:00 and 7:00 pm.

The author and research assistants visited the daycare facility during free playtime in the morning, with each observation lasting approximately 1.5 h. Two cameras were installed in the nursery room to film the interactions between the infants and caregivers from different directions. Figure 1 is a view of the daycare room from the two cameras. The play area in the daycare contained a variety of toys, and the infants were free to manipulate the toys and move around without restrictions. The researchers sat in the corner of the room and took notes. Even when the infants wanted to interact, their involvement was kept to a minimum. If the infants approached crying, the researchers comforted them, but if they approached wanting to play, the researchers encouraged them to play with the caregivers or other infants.

FIGURE 1.

Sketches of the daycare room.

We analyzed these observational data in the following papers, which cover different topics. In Toyama (2020), we examined infants' social exchanges with objects using data from FLY2017. In Toyama (2021), (2022), (2023b), we analyzed data from FLY2017 and FLY2018, and examined infants' transportation of objects, touch, and imitation. In Toyama (2023a), the relationship between infants and objects was investigated using data from FLY2017 to FLY2019.

2.2. Participants

Each year, there were eight infants and three full‐time caregivers. They were all Japanese and spoke Japanese. The infants predominantly came from families of middle socioeconomic status. The caregivers were all female and had professional training in childcare. One of them was in charge of the infant class for 3 years; therefore, a total of seven full‐time caregivers were observed. The mean number of years of experience as a child care provider was 15.3 years (SD = 9.5 years, R = 5–34 years). In addition to full‐time caregivers, there were several part‐time caregivers, all female.

Video data from 4 months before to 4 months after the start of walking were analyzed to examine changes before and after the acquisition of walking. To detect an effect of partial eta squared = 0.06 with 80% power in a one‐way within‐subjects ANOVA (one group, alpha = 0.05, non‐sphericity correction = 1), G*Power analysis indicated that a sample of 23 participants was needed. Of the infants followed for three years, one infant who did not begin walking until the end of the observation period was excluded from the analysis, and the remaining 23 infants were included in the analysis. Table 1 shows each infant's gender and age in months when the walking started. The infants and caregivers were all Japanese and spoke Japanese, and the infants were from families with moderate socioeconomic statuses.

TABLE 1.

Each infant's gender, age (months) to start walking, total days observed, and total time (minutes) to be coded in each locomotor period.

	Gender	Age to start walking a	First		Second		Third		Fourth
			Days	Total time b	Days	Total time b	Days	Total time b	Days	Total time b
A	F	11	7	210	4	157	9	426	6	242
B	F	13	9	375	6	308	7	143	5	101
C	F	12	7	211	7	276	7	277	9	256
D	F	13	9	508	5	251	9	369	4	180
E	M	14	5	129	5	158	3	67	6	93
F	M	12	9	111	6	111	8	171	8	132
G	F	16	6	204	6	97	9	243	7	197
H	F	11	7	93	6	62	9	131	8	103
I	M	14	6	222	6	220	7	356	7	304
J	M	12	9	256	6	244	6	237	8	306
K	M	11	6	354	6	291	7	304	6	221
L	F	14	8	237	7	237	6	225	8	330
M	F	12	7	220	7	187	8	221	7	202
N	F	16	6	215	4	162	5	191	3	121
O	M	12			6	225	8	244	6	218
P	F	13			6	214	7	244	7	231
Q	M	12			2	72	7	207	4	171
R	F	11			6	196	8	266	7	255
S	F	14	9	319	7	277	6	255	4	155
T	F	13	8	275	7	240	4	157
U	F	16	8	127	8	210	3	100
V	M	15	7	255	5	160	3	113
W	F	15	7	276	5	270	3	167

^a Months.

^b Minutes.

Before conducting the observations, the daycare director, caregivers, and the infants' parents were asked to sign an informed consent form. This study was conducted in accordance with the ethical standards of the Declaration of Helsinki, with written informed concent obtained from a parent or gurdian for each infant before data collection. All procedures involving human subjects in this study were approved by the research ethics committee for studies involving human subjects at Waseda University. No money was paid as a reward to the infant's family or the daycare, but the daycare was given a toy and a copy of the digital record at the end of the observation.

3. Analysis 1: Developmental Changes in Pointing

In Analysis 1, we analyzed infants' pointing behavior in terms of (1) their movements before and after pointing, (2) their looking behavior, and (3) their interactions with others, and clarified the changes that occur during the acquisition of walking. First, concerning infants' movement, it has been reported that when infants start to walk, they are more likely to approach objects before they make contact with them and carry them (Dosso and Boudreau 2014; L. Karasik et al. ²⁰¹²; L. B. Karasik et al. ^{2011, 2014}; Toyama ^{2020, 2023a, 2023b}). Therefore, pointing and moving may likely be combined after walking.

Second, concerning looking behavior, studies observing infants and caregivers playing at home have shown that caregivers often look at the infant's face, but infants do not look at the caregiver's face as much (Carpenter et al. 1998; Deák et al. ²⁰¹⁴). However, as infants begin to walk, they often look at the faces of people around them as they move (Franchak et al. 2018). Based on these findings, it is possible that infants will look at others more frequently while pointing as they learn to walk.

Third, we examined whether infants interacted with others prior to social pointing. The walking behavior of infants who just start walking has been shown to be more exploratory in nature than goal‐directed. Infants have a very short gait, often taking only one–three steps at a time. When they move, they often stop unexpectedly in the middle of nowhere (Adolph et al. 2012; Cole et al. ²⁰¹⁶). In other words, there is a contingency in the environmental exploration of walking infants, which may be reflected in their pointing. A typical situation in which infant pointing is observed is when infants play with toys with their caregivers. Given the finding that environmental exploration by infants in the early stages of walking is contingent, it is possible that infants may point unexpectedly not only in situations where interaction with others is already established but also in situations that have not been prepared beforehand. In other words, with the acquisition of walking ability, an infant's pointing could become unexpected. This possibility was examined in Analysis 1.

3.1. Data Coding

We coded behavioral data from the videos using ELAN_5.9.0. Coding categories were as follows.

3.1.1. Locomotion

The day that an infant moved forward by more than 1.5 m without holding onto anything was defined as the day that the infant started to walk. As this was a naturalistic observation study, a length of 1.5 m was based on the length of a shelf in the nursery. If walking was coded as having started, the infant was considered to be in the walking phase, even if subsequent observations did not confirm independent walking > 1.5 m.

3.1.2. Observation Extraction and Coding

For each infant, all observations 4 months before and 4 months after walking started were included in the scoring. The period from four to 2 months before walking started was referred to as the first period, the following 2 months were referred to as the second period, 2 months immediately after walking started as the third period, and the following 2 months as the fourth period. As infants may have been absent on some days, there were differences in the number of days they were observed. In addition, the order in which the infants had lunch varied, and the length of the video data that could be analyzed also varied, as there were times when the infants were outside the camera frame. Table 1 shows each infant's total number of days observed, and total time (seconds) to be coded in each locomotor period. As shown in Table 1, of the 23 infants, 14 (A to N) had data for all four periods, but five (O to S) started walking at the beginning of the year, which caused their data for the first period to be missing; and four (T to W) started walking at the end of the year, which caused their data for the fourth period to be missing.

3.1.3. Pointing Episodes

First, we coded pointing episodes. In previous studies (e.g., Franco and Butterworth ¹⁹⁹⁶; Kishimoto ²⁰¹⁷), the act of pointing with the whole hand as well as pointing with the index finger raised were coded as pointing; however, in this study, only the act of pointing with the index finger raised was classified as pointing. This was because, in daycare, infants' actions are extremely varied, such as pushing a peer with an outstretched hand or reaching out to receive an object. In a situation where many people were interacting with each other, it was very difficult to distinguish between whole‐arm pointing and arm movements such as pushing or reaching out. Therefore, only if the infant was seen extending his or her arm toward an object or person with an extended index finger was it coded as pointing.

Pointing was classified into two categories: non‐social and social. Non‐social pointing occurs when the infant points but does not engage in any reciprocal interaction with others. For example, tracing the surface of an object with the index finger, or pointing in the direction of the target with the index finger while moving. Meanwhile, in social pointing, mutual interaction with others develops through an infant's pointing. Some of these situations involve clear communicative intentions on the part of the infant, such as when the infant looks at another person's face and points to an object after confirming that the other person is paying attention, while in other cases it is not clear whether the infant has a communicative intention, such as when the other person says something in response to the object the infant is pointing to, and the sharing of intentions between the infant and the other person is achieved through the supportive involvement of the other person. Social pointing in this study includes what has been called declarative pointing or imperative pointing in previous studies (Tomasello et al. 2007; Liszkowski et al. ²⁰⁰⁶). From this point onwards, coding was performed only for social pointing.

3.1.4. Infant Movement

For each social pointing episode, forward movement within 10 s before or during pointing was dichotomously coded as moving or not moving. Forward movement was defined as moving toward the pointing target or another person in a series of steps (crawling or standing upright), and cases in which the infant stepped on the spot were excluded. The reason for setting the time within 10 s was that in previous studies (e.g., Liszkowski et al. ²⁰¹²), caregiver‐infant pointing was classified as “following” if it occurred within 10 s.

3.1.5. Looking Behavior

For each social pointing episode, infants were coded as to whether they looked or did not look at the person with whom they were trying to communicate their intentions (or convey their demands) between 10 s before the start of pointing and 10 s after the end of pointing. The reason for the 10‐s interval was as described above. Infants sometimes looked at the other person after pointing, as if to check that their intentions had been conveyed; therefore, we decided to assess their looking behavior up to 10 s later. As this was a naturalistic observational study conducted in a daycare setting, it was not possible to ask infants or caregivers to wear an eye‐camera. Therefore, looking behavior was assessed solely based on whether the infant's eye was directed toward the caregiver's face. In most cases, the direction of the infant's eye was confirmed using video data from two diagonally positioned cameras.

3.1.6. Prior Interactions

The presence or absence of interaction before pointing was assessed to determine whether walking led to ad hoc social pointing. Specifically, we assessed whether infants interacted with the person with whom they wanted to share an intention (or communicate a request), between 10 and 2 s before pointing. Interaction was defined not only as verbal but also non‐verbal (e.g., one person looking at the other, waving, and so on).

First, the author coded all the video data. Then, trained research assistants coded video data for at least one session for each locomotor period for each infant. As a result, 35% of the video data was coded by trained research assistants. Of the infant behaviors rated as pointing by the author, 96.5% were also rated as pointing by the second rater, and 94.9% of the behaviors rated as pointing by the second rater were also rated as pointing by the author. Inter‐rater reliability was 95% (κ = 0.750, p < 0.001) for social and nonsocial pointing, 89% (κ = 0.772, p < 0.001) for infant movement, and 93% (κ = 0.865, p < 0.001) for prior interaction. Inconsistencies between coders were resolved through discussion.

3.2. Statistical Analyses

Fourteen of the 23 infants had data for all four periods, while the remaining nine had missing data for one of the periods. Due to missing values and correlations in the longitudinal data, we decided to analyze the data using generalized estimating equations (GEEs). The locomotor period was entered as a within‐subjects variable. For the analyses, the number of pointing episodes observed during each locomotor period was tabulated and used for analysis. Because the number of days observed and the total time coded during each locomotor period varied for each infant, the pointing frequency was converted to a frequency per 60 min. Alternatively, to compare categories of pointing episodes, percentages of the total frequencies of the target category of pointing relative to the total frequencies of pointing in each locomotor period were calculated.

In Analysis 1, we test whether there are differences in five variables‐the frequency of pointing, the percentage of social pointing, the percentage of social pointing with moving, the percentage of social pointing with looking, and the percentage of social pointing with prior interaction ‐ as a function of locomotor period. Since we are testing similar dependent variables (five pointing‐related indicators) on the same sample using the same independent variable (locomotor period), we will use the Bonferroni correction and set the significance level to 1% by dividing 5% by 5.

To determine whether changes in pointing are due to the acquisition of walking or to general development, we conduct two analyses. The first analysis use changes in pointing as the dependent variable and age as the independent variable, and the results are compared with those of an analysis using locomotor period as the independent variable. The second analysis compare pointing in infants who have begun walking (walkers) and those who have not yet begun walking (crawlers) at the same age by conducting a series of t‐tests.

3.3. Results

3.3.1. Frequencies of Pointing

The mean frequencies of pointing episodes in each locomotor period per 60 min are shown in Figure 2. The mean frequencies were 2.74, 95% CI [2.05, 3.43] in the first period, 3.91, 95% CI [3.13, 4.70] in the second period, 6.83, 95% CI [5.47, 8.20] in the third period, and 9.81, 95% CI [8.09, 11.53] in the fourth period. A GEE on the frequencies of pointing episodes, using locomotor period as a predictor, revealed a significant main effect of locomotor period. The statistical results are shown in Table 2. There were significantly more pointing episodes in the second period than in the first (p < 0.01), in the third than in the second (p < 0.001), and in the fourth than in the third (p < 0.001). These results suggest that infants pointed more frequently as they acquired locomotor skills.

FIGURE 2.

Mean frequencies of pointing per 60 min (sum of social and nonsocial pointing) in each locomotor period. Error bars represent standard errors. *p < 0.05, **p < 0.01, ***p < 0.001.

TABLE 2.

Results of GEEs, using locomotor period as predictors, in Analysis 1.

	Wald χ 2	df	p	Results of post‐hoc analyses
Frequency of pointing per 60 min	85.11	2	< 0.001	First < second: p = 0.010
				second < third: p < 0.001
				Third < fourth: p < 0.001
% of social pointing	126.19	2	< 0.001	First < second: p < 0.001
				second < third: p = 0.036
				Third < fourth: p = 0.003
% of social pointing with moving	258.13	2	< 0.001	First < second: p < 0.001
				second < third: p < 0.001
				Third < fourth: p < 0.001
% of social pointing with looking	40.65	2	< 0.001	First = second: p = 0.253
				second < third: p = 0.021
				Third < fourth:p = 0.031
% of social pointing with prior interaction	14.51	2	0.002	First = second: p = 0.269
				second > third: p = 0.001
				Third = fourth: p = 0.344

3.3.2. Social Pointing

Pointing included social and non‐social pointing. Figure 3 presents the mean percentage of social pointing during each locomotor period. The mean percentages were 54.3%, 95% CI [45.6%, 62.9%] in the first period, 76.0%, 95% CI [70.3%, 81.8%] in the second period, 84.4%, 95% CI [79.4%, 89.5%] in the third period, and 91.3%, 95% CI [88.1%, 94.6%] in the fourth period. A GEE on the percentages of social pointing, using locomotor period as a predictor, revealed a significant main effect of locomotor period (Table 2). The percentage of social pointing increased from the first to the second period (p < 0.001), from the second to the third period (p < 0.05), and from the third to the fourth period (p < 0.01). In other words, infants' social pointing increased gradually as they acquired locomotor skills.

FIGURE 3.

Mean percentages of social pointing in each locomotor period. Error bars represent standard errors. *p < 0.05, **p < 0.01, ***p < 0.001.

3.3.3. Infant Movement

The mean percentages of social pointing in which infants moved before or during pointing behavior in each locomotor period are shown in Figure 4. The mean percentages were 0.0%, 95% CI [0.0%, 0.0%] in the first period, 8.5%, 95% CI [4.3%, 12.8%] in the second period, 31.7%, 95% CI [23.4%, 40.0%] in the third period, and 50.0%, 95% CI [44.0%, 56.1%] in the fourth period. A GEE on the percentages of social pointing accompanied by moving behavior, using locomotor period as a predictor, revealed a significant main effect of locomotor period (Table 2). The percentage was higher in the second period than in the first (p < 0.001), higher in the third than in the second (p < 0.001), and higher in the fourth than in the third (p < 0.001). As the infants learned to walk, they moved more frequently before and during pointing.

FIGURE 4.

Mean percentages of social pointing in which infants moved before or during pointing, in each locomotor period. Error bars represent standard errors. ***p < 0.001.

3.3.4. Looking Behavior

Figure 5 shows the mean percentages of social pointing in which infants looked at the other person while pointing, in each locomotor period. The mean percentages of the pointing episodes were 72.6%, 95% CI [60.6%, 84.7%] in the first period, 80.4%, 95% CI [73.2%, 87.5%] in the second period, 90.5%, 95% CI [83.5%, 97.6%] in the third period, and 97.6%, 95% CI [96.4%, 98.9%] in the fourth period. A GEE on the percentages of social pointing accompanied by looking behavior, using locomotor period as a predictor, revealed a significant main effect of locomotor period (Table 2). The percentage was higher in the third period than in the first (p < 0.01) and higher in the fourth than in the third (p < 0.05). However, there were no significant differences between the first and second periods, the second and third, and the second and fourth. As they acquired locomotor skills, the infants began to look at the other person more frequently.

FIGURE 5.

Mean percentages of social pointing in which infants looked at the other person while pointing, in each locomotor period. Error bars represent standard errors. *p < 0.05, **p < 0.01.

3.3.5. Prior Interaction

Finally, Figure 6 shows the mean percentages of pointing in which infants interacted with the other person before pointing. The mean percentages were 80.4%, 95% CI [65.7%, 95.1%] in the first period, 90.1%, 95% CI [83.0%, 97.3%] in the second period, 76.4%, 95% CI [69.9%, 83.0%] in the third period, and 71.7%, 95% CI [64.3%, 79.1%] in the fourth period. A GEE on the percentages of pointing with prior interaction, using locomotor period as a predictor, revealed a significant main effect of locomotor period (Table 2). The percentage was higher in the second period than in the third (p < 0.001) and the fourth (p < 0.001), though there were no significant differences between the first and second periods. The findings showed that pointing with prior interaction tended to decrease with the acquisition of walking.

FIGURE 6.

Mean percentages of social pointing in which infants interacted with the other person before pointing, in each locomotor period. Error bars represent standard errors. ***p < 0.001.

3.3.6. Locomotion or Age

To determine whether the observed changes in pointing behavior are more strongly related to the acquisition of walking or to age, we conducted two analyses. In the first analysis, we performed two separate GEE analyses: one with age as the predictor variable, and the other with the locomotor period as the predictor variable. These results were then compared. Since the purpose of this study was to examine the relationship between walking acquisition and pointing, we compared the third locomotor period (immediately after walking acquisition) with the first locomotor period (four to 2 months before walking acquisition [locomotor analysis]). The average age at these time points was 8.3 months (SE = 0.4) for the first period and 12.5 months (SE = 0.4) for the third period. Accordingly, for age‐based analysis, we compared the results at 8–9 months and 12–13 months (age analysis).

Table 3 presents the mean frequencies of pointing per 60 min, as well as the mean percentages of social pointing, social pointing with moving, social pointing with looking, and social pointing with prior interaction, for the age analysis (8–9 months and 12–13 months) and the locomotor analysis (first and third periods). The results of the GEE analyses using age (age analysis) and locomotor period (locomotor analysis) as predictors are also shown in Table 2. For the pointing frequency, the percentage of social pointing, and the percentage of social pointing with moving, both analyses showed the main effects of age and locomotor period (all ps < 0.001). As indicated by the χ² values and p‐values, the differences were greater in the locomotor period comparison than in the age comparison.

TABLE 3.

Mean frequencies of pointing; mean percentages of social pointing; and mean percentages of social pointing with looking, moving, and prior interaction across each age and locomotor period, along with the results of the age and locomotor analyses.

	Age analysis			Locomotor analysis
	8–9 months	12–13 months	Wald χ 2 (1)	First period	Third period	Wald χ 2 (1)
Frequency of pointing per 60 min	2.94 (0.42)	5.06 (0.58)	13.78, p < 0.001	2.80 (0.37)	6.83 (0.69)	49.09, p < 0.001
% of social pointing	56% (5%)	79% (3%)	19.4, p < 0.001	54% (4%)	84% (3%)	35.57, p < 0.001
% of social pointing with moving	3% (3%)	17% (3%)	12.47, p < 0.001	0% (0%)	32% (4%)	56.78, p < 0.001
% of social pointing with looking	74% (6%)	88% (3%)	4.13, p = 0.042	73% (6%)	91% (4%)	6.57, p = 0.010
% of social pointing with prior interaction	81% (7%)	79% (5%)	0.10, p = 0.749	81% (8%)	76% (3%)	0.37, p = 0.546

Note: Standard errors are in parentheses.

In the second analysis, we compared pointing behavior between infants who had not yet begun walking at 12 months (crawlers) and infants who had begun walking (walkers). A series of t‐tests were conducted separately to determine whether the frequencies of pointing, as well as the mean percentages of social pointing, social pointing with moving, social pointing with looking, and social pointing with prior interaction, differed between the crawler and walker groups. The results, shown in Table 4, indicate that the frequencies of pointing, the percentage of social pointing, and the percentage of social pointing with moving were greater in the walker group than in the crawler group.

TABLE 4.

Mean frequencies of pointing; mean percentages of social pointing; and mean percentages of social pointing with looking, moving, and prior interaction for crawlers and walkers, along with the results of t‐tests.

	Crawlers (n = 13)	Walkers (n = 10)	t‐test
Frequency of pointing per 60 min	4.39 (0.87)	8.72 (0.80)	t (21) = 3.55, p = 0.002
% of social pointing	69% (5%)	92% (2%)	t (21) = 4.14, p < 0.001
% of social pointing with moving	6% (3%)	46% (13%)	t (21) = 8.12, p < 0.001
% of social pointing with looking	88% (4%)	96% (2%)	t (21) = 1.83, p = 0.083
% of social pointing with prior interaction	78% (9%)	81% (5%)	t (21) = 0.24, p = 0.811

Note: Standard errors are in parentheses.

Together, the results of the two analyses suggest that walking plays a more important role than general development (age) in infants' pointing behavior.

4. Analysis 2: Development of Social Interaction Through Pointing

In Analysis 1, as the infants learned to walk, there was an increase in social; there was an increase in social pointing associated with the infants' movements and looking behavior but a decrease in pointing associated with prior interaction. Pointing without prior interaction carries the risk of the partner not noticing the infant pointing. Unlike situations in which infants and caregivers interact on a one‐to‐one basis, there are multiple infants and caregivers in a nursery. Caregivers look after multiple infants to prevent them from getting hurt or in trouble. Under these circumstances, even if an infant asks for a caregiver's involvement, the caregiver may not always respond to the infant (Toyama ^2023a). Therefore, the risk that the infant's pointing will go unnoticed by the partner is greater than that in the home situation, and this is even more true for pointing where there is no prior interaction.

To examine these issues, Analysis 2 classified pointing in terms of the infant's looking behavior and the presence or absence of prior interaction. We then examined the relation between each type of pointing and the infant's movement, and whether pointing was successful or unsuccessful in helping the infant communicate his or her intentions to the partner, in terms of movement.

4.1. Data Coding

4.1.1. Type of Pointing

Based on the infants' looking behavior and prior social interactions coded in Analysis 1, pointing was classified into three types: (a) passive pointing, (b) interaction‐preceded pointing, and (c) unexpected pointing. (a) Passive pointing was pointing in which there was an interaction with the other person before pointing, and the infant did not look at the other's face within 10 s before and after pointing. In other words, this type of pointing may be said to have occurred as an extension of social interaction with others, and the infant's role in sharing attention with the pointing target was passive. (b) Interaction‐preceded pointing was pointing in which the infant interacted with the other person before pointing and looked at the other person's face within 10 s before and after pointing. In this pointing, the infant's intention to share attention with the other person was apparent. (c) Unexpected pointing refers to instances in which the infant, without prior interaction with the other person, suddenly looks at their face and points to a specific object in an attempt to share attention. Because this type of infant pointing often occurs somewhat suddenly and with no apparent warning to others, we have referred to it as “unexpected” pointing. No cases of pointing without prior interaction, in which the infant did not look at the other person's face, were found in this study. The definitions of the three types of pointing gestures listed above are summarized in Table 5.

TABLE 5.

Infants' looking and prior‐interaction for each type of pointing.

	Infants' looking behavior	Prior‐interaction
Passive pointing	×	〇
Interaction‐preceded pointing	〇	〇
Unexpected pointing	〇	×

4.1.2. Success or Failure of Intention Sharing

For interaction‐preceded pointing and unexpected pointing, we coded whether the infant's intention to share attention toward an object or elicit a specific action from the partner was perceived by the partner. Intention sharing was considered successful if the partner responded to the infant's pointing either verbally or nonverbally (e.g., nodding, picking up the object, and handing it to the infant). If the partner did not respond to the infant's pointing, this was considered a failure. The author and a research assistant coded 50% of the data and the interrater reliability was 85% (κ = 0.620, p < 0.001). Inconsistencies between the coders were resolved through discussions.

4.2. Statistical Analysis

As in Analysis 1, the frequency of the three types of pointing (passive, interaction‐preceded, and unexpected) was calculated per 60 min and summed by the time period. To examine the differences by period, a GEE was performed on these values, with the period as the predictor. When examining the association between interaction‐preceded pointing, unexpected pointing, and infant movement, and the association between success or failure of intention sharing and infant movement, we decided to sum the results for all 23 infants because of the small amount of data per infant. A chi‐square test was performed on the frequency data of all 23 infants to examine differences by period and type of pointing.

4.3. Results

4.3.1. Type of Pointing

Figure 7 shows the mean frequencies of passive pointing, interaction‐preceded pointing, and unexpected pointing over 60 min. A GEE on the frequencies of each type of pointing per 60 min, using the type of pointing (passive, interaction‐preceded, unexpected and locomotor period (4) as predictors, were conducted. The results of GEE are shown in Table 6.

FIGURE 7.

Mean frequencies of each type of pointing per 60 min, in each locomotor period. Error bars represent standard errors. ***p < 0.001.

TABLE 6.

Results of GEEs, using type of pointing and locomotor period as predictors, in Analysis 2.

	Wald χ 2	df	p	Results of post‐hoc analyses
Main effects of pointing type	162.79	4	< 0.001	Passive < unexpected: p < 0.001
				Unexpected < interaction preceded: p < 0.001
Main effects of period	104.53	3	< 0.001	First < second: p < 0.001
				second < third: p < 0.001
				Third < fourth: p < 0.001
Pointing type × period interaction	235.61	11	< 0.001	Passive pointing: Ns
				Interaction‐preceded pointing:
				First = second: p = 0.270
				second < third: p < 0.001
				Third < fourth: p < 0.001
				Unexpected pointing:
				First = second: p = 0.528
				second < third: p < 0.001
				Third < fourth: p < 0.001

First, a main effect of pointing type was significant. The frequencies of interaction‐preceded pointing (M = 3.52, 95% CI [2.86, 4.11]) were greater than those of passive pointing (M = 0.35, 95% CI [0.21, 0.49]) and unexpected pointing (M = 0.96, 95% CI [0.72, 1.20]) (all ps < 0.001), with significant differences between those of passive pointing and unexpected pointing (p < 0.001). Next, the main effect of the locomotor period was significant. There were significantly more pointing episodes in the second period than in the first (p < 0.001), in the third than in the second (p < 0.001), and in the fourth than in the third (p < 0.001). Finally, A type of pointing × locomotor period interaction was also significant. Post‐hoc tests revealed that for passive pointing there were no significant differences between the locomotor periods, for the interaction‐preceded and unexpected pointing, the frequencies of pointing in the third and fourth periods were greater than those in the first and second periods (all ps < 0.001), with significant differences between the third and the fourth periods (all ps < 0.001).

4.3.2. Infant Movement

As shown in Figure 7, passive pointing was infrequent across all four periods. Therefore, for the other two types of pointing (i.e., interaction‐preceded pointing and unexpected pointing), we tabulated the ratio of pointing in which infants moved to that in which infants did not move for the third and fourth periods (the first and second periods were excluded because of the low frequency of unexpected pointing).

Table 7 shows the percentages of pointing with and without movement for interaction‐preceded and unexpected pointing. For the third and fourth periods, a chi‐square test of independence was performed to examine the relation between type of pointing (2: interaction‐preceded or unexpected pointing) and movement (2: with or without movement). The relation between these variables was not significant for the third period, X ² (1, N = 498) = 0.300, p = 0.584, V = 0.025, but was significant for the fourth period, X ² (1, N = 554) = 69.52, p < 0.001, V = 0.354. In the fourth period, pointing with movement was more frequent for unexpected pointing than for interaction‐preceded pointing.

TABLE 7.

Percentages of pointing with movement for interaction preceded and unexpected pointing, in the third and fourth period.

	Third period	Fourth period
Interaction preceded pointing	40% (157/395)	39% (148/379)
Unexpected pointing	43% (44/103)	77% (135/175)

Note: Frequencies are in parentheses.

4.3.3. Success or Failure of Sharing Intentions

Table 8 shows the percentages of pointing in which an infant succeeded in sharing intentions with the partner when the infant moved or did not move while pointing, for interaction‐preceded pointing, or unexpected pointing. As shown in Table 4, in most of the interaction‐preceded pointing, regardless of whether the infant moved, the infant was successful in sharing his/her intentions with the partner. However, this was not the case with unexpected pointing. Chi‐square tests were conducted to examine the relation between movement (2: with or without movement) and sharing intentions (2: succeeded or failed) for unexpected pointing in each period. For both the third and fourth periods, significant differences were obtained: X ² (1, N = 103) = 18.80, p < 0.001, V = 0.427, for the third period; X ² (1, N = 175) = 11.58, p < 0.001, V = 0.257, for the fourth period. During both periods, infants were more successful in sharing their intentions with their partners when they moved than when they did not.

TABLE 8.

Percentages of pointing, in which infants succeeded in sharing intentions with the other person, for interaction preceded and unexpected pointing, in the third and fourth period.

	Third period			Fourth period
	With movement	Without movement	Total	With movement	Without movement	Total
Interaction preceded pointing	99% (155/157)	98% (234/238)	98% (389/395)	99% (147/148)	98% (227/231)	99% (374/379)
Unexpected pointing	89% (39/44)	47% (28/59)	65% (67/103)	90% (121/135)	68% (27/40)	85% (148/175)

Note: Frequencies are in parentheses.

5. Discussion

In this study, we conducted longitudinal observations about once a week in an infant class at a daycare center in Japan to examine the relationship between pointing and infants' locomotor development. Analysis 1 showed that the frequency of pointing and the proportion of social pointing increased with the development of locomotion. It was also shown that as the infants learned to walk, pointing during moving increased and pointing while looking at the partner also increased; however, pointing with prior interaction decreased. In Analysis 2, both interaction‐preceded pointing and unexpected pointing increased as infants learned to walk. However, differences were observed between the two. In the case of interaction‐preceded pointing, infants were able to share attention with their partner in many cases regardless of whether or not they moved just before pointing, whereas in the case of unexpected pointing, infants were more likely to succeed in sharing attention if they moved before pointing.

In this section, we will first consider social pointing without infants' looking behavior, which is more common before they learn to walk, then unexpected pointing, which increases after they learn to walk, and finally why infants can communicate their intentions to others even though their pointing is suddenly emerged.

5.1. Social Pointing Without Infants' Looking Behavior

The characteristics of social pointing by infants before they learn to walk were that in most cases the infant did not move before pointing and that the infant did not look at the person about one in five times (this almost disappeared when they started walking). Why were the infants able to share their attention with the person by pointing, even when they were not looking at their partner? Previously, Bakeman and Adamson (1984) suggested that caregivers play an important role in establishing joint attention. They observed infant‐caregiver interactions longitudinally from 6 to 18 months and reported a “passive joint” in which the caregiver monitored the infant's attention and attended to the object to which the infant's attention was directed. This type of joint attention was observed as early as 6 months, but after 15 months, “coordinated joint” became frequent, in which the infant was actively involved in establishing joint attention. Several studies (Carpenter et al. 1998; Deák et al. ²⁰¹⁴) have shown that infants rarely look at their caregivers' faces during interactions with them. Therefore, scaffolding and supportive engagement by caregivers are important for the early development of joint attention.

The present study suggests that this would also hold true for early pointing. A typical situation for “passive pointing” is when a caregiver is watching an infant playing with a toy, for instance, a toy car, and just as the infant points to a part of the car, the caregiver says something like, “it's a car! Vroom, vroom!” The infant's pointing has no communicative intent in itself. However, by directing the caregiver's attention to the toy car, the infant's pointing takes on a social nature. In this regard, the role of caregivers in establishing social pointing is essential. Carpendale and his colleagues (e.g., Carpendale and Carpendale ²⁰¹⁰; Ketter and Carpendale ²⁰¹⁸) have suggested that infants may become aware of the meaning of their pointing through interactions with their caregivers, based on research using the diary method. To clarify the details of this idea that noncommunicative pointing becomes communicative through interaction with caregivers, it needs to be tested with observational data.

5.2. Unexpected Pointing

The opposite of the passive pointing is the unexpected pointing found in this study. As shown in Analysis 1, the percentages of pointing without prior interaction became higher as infants' locomotion developed. Previous research (e.g., Toyama ²⁰²¹; Hoch et al. ²⁰¹⁹) has shown that the movements of infants who have just started walking are exploratory. This may account for the seemingly sudden and unplanned nature of pointing behavior. Hoch et al. (2019) noted that there are two hypotheses regarding infant locomotion: “destination‐directed” and “peragration.” The former is the idea that infants move to achieve some goal (e.g., getting closer to another person or taking an object), while the latter is the idea that infant locomotion is not about reaching a distant destination, but that the movement itself becomes an intrinsic reward. Hoch et al. (2019) observed 15‐month‐old walking children playing freely in an empty room and in a room full of toys, and concluded that the latter hypothesis may be applicable, as there was no significant difference in movement‐related exploratory behavior between the two. Toyama (2021) has also shown that the trajectory of infants' movement immediately before engaging with objects becomes migratory when walking begins. These changes are probably due to them being distracted by more stimuli as they move with a wider field of view (Franchak et al. 2018; Kretch et al. ²⁰¹⁴) and look at the caregiver's face more often (Franchak et al. 2018). Exploratory walking movements and encounters with many social stimuli can lead to unexpected pointing that is created on the spot and improvized.

5.3. Benefits of Walking for Sharing Attention

Unexpected pointing is thought to carry a high risk of failure to share attention with a partner because it occurs without prior interaction between the partner and the child. In day care, caregivers tend to divide their attention among several infants, and therefore this risk is thought to be greater in day care settings than in home settings. However, Analysis 2 showed that this risk can be reduced by infants' movement. After 2 weeks of walking, the infants began to move more before making unexpected pointing, and in many cases, even with unexpected pointing, if the infants moved just before pointing, they were successful in sharing attention with the partner.

It has been reported that caregivers are more likely to respond to infants when they are in a mobile position than when they are in a stationary position (e.g., Toyama ²⁰²⁰; L. B. Karasik et al. ²⁰¹⁴). Why do caregivers respond more to moving than stationary infant communication? West and Iverson (2021) suggested three reasons: the distance between the infant and caregiver is close, the infant's behavior is more visible, and multimodal communication is more salient to caregivers. In addition to the above stated three reasons, our perception of biological motion may also be related. That is, humans have the ability to read minds to motion (Johansson 1973); thus, when infants point while moving, caregivers may be reading the infants' minds regarding their motion and become more responsive.

It is more important to let the caregiver know that the infant is trying to communicate in a child care setting than in a family setting. In dyadic situations, such as the home environment, infants can monopolize the caregiver's attention. However, in daycare centers, there are fewer caregivers than infants; thus, several infants must compete for the attention of a few caregivers (Toyama ^2023a). This is because the caregivers are supervising multiple infants and ensuring their safety by dividing their attention. Therefore, for an infant to capture the caregiver's attention, the infant must take the initiative. This study showed that infants' movements contribute to the establishment of joint attention with others in unexpected pointing, but given the nature of the daycare center described above, the effects of infants' movements shown in this study are likely to be higher than in two‐person situations, such as between a caregiver and an infant. This is an important issue for future consideration.

5.4. Suggestions for Practice and Future Directions

In all three types of pointing examined in this study (passive pointing, pointing with prior interaction, and unexpected pointing), caregivers played an important role in sharing attention with the infant. In passive pointing, the caregivers noticed the infants' pointing and directed their attention to the pointed targets, thereby achieving joint attention. In pointing with prior interaction, the pointing occurred in the course of mutual interaction between the infant and the caregiver. Furthermore, in unexpected pointing, joint attention was achieved when the caregiver responded to the infant's sudden pointing. As infants' locomotor development progresses, their pointing patterns change. However, at each stage of this developmental process, the role of the caregiver is essential in achieving joint attention. Since infants' pointing has been shown to be related to language development (Kirk et al. 2022), caregivers' supportive involvement in infants' pointing would be important for the development of communication in infancy.

This research provides some new insights into the relationship between infant walking and pointing that have been overlooked in previous studies. Studies of developmental cascades have shown that infants who begin walking are more likely to engage in “moving communication,” a behavior that is temporarily combined with walking (Toyama 2020; L. B. Karasik et al. ²⁰¹⁴; West and Iverson ²⁰²¹). The results of this study show that the same is true for pointing. However, there are still some issues that need to be addressed to further clarify the details of the relationship between locomotor development and pointing. First, the participants analyzed in this study were 23 infants in one daycare center. More participants need to be observed to draw more general conclusions. These 23 children were enrolled in three different classes (FY2017–2019), and their caregivers (with one exception) varied from one year to the next. Therefore, the results of this study were not obtained under the influence of the same caregivers. However, as these three classes were in the same daycare center, they shared the same physical environment as the nursery (e.g., the number and type of toys) and caregivers' views on childcare practices (e.g., the way they treat the infants). Further studies in different daycare centers and homes are needed to confirm the results of this study.

Second, although observations in this study were conducted in a naturalistic setting, certain aspects, such as the presence of researchers in the nursery and the installation of cameras, differed from typical settings. These factors may have influenced the behavior of the infants and caregivers. Moreover, with multiple infants and caregivers present, infant–caregiver interactions may have been influenced by uncontrollable factors, such as the presence of other infants nearby. Controlling for these factors is a methodological challenge for future research.

Third, the nursery contained eight infants and several caregivers and was constantly filled with various sounds, making it difficult to hear each voice clearly. Therefore, the infants' vocalizations were excluded from the analysis. However, pointing is often accompanied by vocalization (e.g., Liszkowski and Tomasello ²⁰¹¹; Kishimoto ²⁰¹⁷), and there is a difference in vocalization between declarative and imperative pointing (Grünloh and Liszkowski 2015). By adding an analysis of infants' vocalizations, more knowledge can be gained about pointing as a form of communication.

5.5. Conclusion

Previous studies on the developmental cascade of infant locomotion have shown that with the onset of walking, infants increase their triadic engagement in carrying, showing, and exchanging objects with their caregivers (L. B. Karasik et al. ²⁰¹¹; L. Karasik et al. ²⁰¹²; Toyama ^{2020, 2023a}). The present study showed that social pointing became more frequent as locomotion progressed. One of the most important findings of this study was that infants' pointing became integrated with locomotion through the acquisition of walking, and that locomotion appeared to contribute to sharing attention with partners. These findings underscore the importance of studying infant pointing in the context of physical movement.

Author Contributions

Noriko Toyama: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, writing – original draft, writing – review and editing.

Toyama, Noriko . 2025. “The Association Between Pointing and Walking in Infants: A Longitudinal Observational Study.” Infancy: e70037. 10.1111/infa.70037.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.