Main

Neonatal imitation of facial gestures (e.g. TP and MO) has been observed by several researchers in different laboratories around the world (1–18) and, as a result of this body of findings, new research questions beyond those concerning the existence of this phenomenon have emerged. Now what seems to be needed is a better understanding of how to integrate this early imitative capacity with other developments taking place in social, emotional, and cognitive abilities both during the early phases of infancy and later in the child's first year of life. Recently, this has become the focus of several researchers interested in early infancy, and it has been suggested that neonatal imitation might be best understood within the context of early social development (19,20).

Moreover, it has also been suggested that the capacity to imitate early in life might be a competency that is lacking or disrupted in children with mental handicaps, such as autism and Down's syndrome. Rogers and Pennington (21) have asserted that neonatal imitation is an important social competence and predicted that this ability would be missing in newborn children who later develop autism. Furthermore, Nadel and Pezé (22) suggested that imitation as a means of communication is a good predictor of social capacities in older children with autism (between 2 and 3 y of age). Thus, we need to reexamine the early imitative capacity among children with various developmental disabilities to enhance our understanding of both the nature of neonatal imitation and the basic mechanisms at work. This report aims at presenting observations from one such attempt, a study of facial imitation among infants with Down's syndrome within the age range of 1 to 4 mo.

Earlier reports on imitation in young children or infants with Down's syndrome have not included imitation of facial gestures in assessments of imitation during the first months of life. In a comprehensive review, Dunst (23) describes the sensorimotor development of Down's children from birth to 3 y of age as measured by the Uzgiris-Hunt scale. The youngest group in his sample (n = 9) had a mean age of 2.89 mo, and infants in this group displayed an almost normal level of imitation (gestural and vocal). However, none of the imitation tasks included facial gestures of the type used in studies of neonatal imitation in typically developing newborns.

Dunst (23) reports as his main finding that the sensorimotor development of Down's children follows a stagelike pattern similar to that observed for normally developing children, but that development occurs at a much slower rate. However, he also found that this "slowing down phenomenon was most pronounced for vocal imitation" (p. 224). The way the obtained DQ changed with age is a good illustration of this "slowing down" process. The observed quotients were within the normal range for children younger than 4 mo (DQ = 85), below average at 1 y (DQ = 62), and even more clearly below average at 3 y (DQ = 55).

Knowledge concerning imitative skills in very young Down's syndrome babies is limited compared with what we know about imitation in typically developing infants. Observations of imitation of facial gestures in children with Down's syndrome around birth or shortly thereafter are lacking. This work represents a first attempt to explore the imitative capacities of Down's syndrome babies as young as 4 wk. The study is based directly on our previous research (5–7) on neonatal imitation. Observations made in these studies of typically developing infants suggest the following response patterns for facial imitation. 1) Imitation of TP is more likely to occur during the first 2 mo of life, and less likely at 3 mo (5,6,10,12). 2) Imitation of MO is less affected by age during the child's first 3 mo of life (5).

In addition, at younger ages (below 2 mo) imitation of TP tends to be accomplished by partial rather than full responses (5,7). Imitative responses, moreover, appear more frequently during the response period (when the adult model is passive) than during the presentation period. This holds for the first 4-6 wk of life, but not thereafter (5).

As summarized by Hodapp and Zigler (24), research focusing on aspects of early development other than imitation (e.g. development of play, smiling, laughter, and language) tends to show that children with Down's syndrome develop "in the same sequence as that followed by normal children" (p. 10). Thus, we adopted as our main hypothesis that young infants with Down's syndrome will be able to imitate and will display a response pattern similar to that which has earlier been observed for "normal" children.

METHOD

Subjects

Eight children (seven male and one female) with Down's syndrome participated in the study. The median birth weight was 3237 g (range, 2800-3430 g) and the median gestational age was 39 wk (range, 36-39 wk). Five children were successfully observed approximately 1 mo after parturition (X = 37.4 d, SD = 11.0; range, 25-52), seven at around 3 mo (X = 81 d, SD = 10.6; range, 70-101), and seven at approximately 4 mo (X = 128, 1 d, SD = 14.9; range, 97-143). The overlap between the 3 and 4 mo observations was due to the fact that for one child (child A) the last observation had to be carried out earlier due to medical complications. All children were referred to the project by pediatricians at one of the two main hospitals in Göteborg (East Hospital or Sahlgren's Hospital). Approval for this study was obtained both from the Ethical Review Board at Göteborg University and from the participating families.

Procedure

All observations took place in the children's homes using light-weight portable video equipment (VHS-C). The parents were usually present in the room during the observation, and the sessions were begun when the child was judged to be awake and alert. TP and MO were presented (randomized order) to the child by an experimenter on all occasions and each gesture was presented during a preset interval followed by a response time of equal length. The time interval used was 20 s for the first observation (X = 20.4; SD = 2.1), and 10 s for observations at 3 and 4 mo (X = 11.5; SD = 3.2). This sequence was repeated three times, thus giving each child a total observation time of about 120 s at observation one (X = 122.1; SD = 9.4) and approximately 60 s at observations two and three (X = 69.0; SD = 18.3).

Definition of responses. A full TP was coded whenever the tongue was protruded beyond the lips. A weak or partial response was noted when a clear forward movement could be detected and the tongue had passed only the posterior part of the lip.

MO was defined as a clear and visible separation of the lips that was judged to meet the criteria of a definite change. Some children kept their mouths open over extended periods of time, which was not accepted as a MO. A clear change had to take place as compared with the child's "baseline expression." No simultaneous protrusion of the tongue was accepted. Yawning, hiccups, and cries were not included as adequate responses.

Coding and reliability. All tapes were coded blind by two research assistants. They were trained until they performed at a minimum level of 85% agreement. The obtained final reliability coefficient was 0.98 for Pearson's r, and 0.92 for Cohen's κ (25).

Imitation. A child was judged to imitate if the matching responses (expressed in rate/minute) exceeded the observed rate of nonmatching responses.

Statistical analysis. Because the number of subjects was small, only explorative statistical testing was carried out. The test used was the Wilcoxon signed-rank test, the same as used in our previous studies (5–7). In addition, brief individual vignettes describing each child's individual response pattern are provided.

RESULTS

This section focuses primarily on the combined result from both the response and presentation periods. Findings specific to the presentation periods or the response periods are mentioned only if they differ from the overall picture. Similarly, for TP, the result presented is based on the combined effect of both partial (weak) and full responses.

The 1-Mo Observation

Imitation of TP. All five children observed at 1 mo imitated TP. Overall, a mean rate of 10.7 (SD = 4.9) TPs/min was observed when TP was modeled compared with a mean of 7.4 (SD = 4.0) after modeling of MO (Table 1; z = -2.02, p < 0.05). Further analyses revealed a similar effect for responses emitted during the presentation period but not during the response period (not in Table 1).

Table 1 Mean rate/minute of TP after modeling of TP or MO
Full size table

Imitation of MO. A mean rate of 8.6 MOs (SD = 2.3) was observed after modeling of MO compared with a mean rate of 7.5 responses (SD = 4.0) after modeling of TP (Table 2). This change in rate was not significant. Only for MOs emitted during the presentation period was a marginal effect detected (X = 9.8; SD = 1.7 versus X = 5.9; SD = 4.0; z = -1.75, p = 0.08; not in Table 2). Judged individually, four children imitated MO during the presentation period (see also Table 4). In comparison, only three children imitated both gestures if the whole observation period is used (Table 3).

Table 2 Mean rate/min of MO after modeling of MO or TP
Full size table
Table 4 Individual responses during the presentation period at 1 mo (n = 5): observed changes in rate as a function of each child's responses to modeling
Full size table
Table 3 Individual responses during the whole observation period: observed changes in rate as a function of each child's responses to modeling
Full size table

The 3-Mo Observation

Imitation of TP. No indication of imitation was found (see Table 1). In fact, the individual judgment of imitation showed that only one child imitated TP at this age.

Imitation of MO. Overall, the group analysis revealed no imitation of MO. The difference in observed response rate was nonsignificant. However, judged individually, six children seemed to imitate (see Table 3). Only one child imitated both TP and MO (Table 3) and six of the seven children actually increased the response rates for both TP and MO when MO was demonstrated.

The 4-Mo Observation

Imitation of TP. As evident from Table 1, no overall trend toward imitation could be detected. Individually, only three children imitated TP (Table 3).

Imitation of MO. The group result indicated no imitation of MO at this age (Table 2), and only three children displayed any sign of imitation (Table 3). None of the children imitated both TP and MO at 4 mo of age.

Changes in Imitation from 1 to 4 Mo

The changes in imitation of TP and MO over the first 4 mo of life are depicted in Figure 1. The graph displays different patterns for imitation of MO and TP. The curve describing the development of TP imitation is almost U-shaped; more children imitated TP at 1 and 4 mo than at 3 mo. In contrast, the curve for MO imitation show something like a "peak" at 3 mo.

Figure 1
figure 1

Proportion of children imitating TP and MO at approximately 1, 3, and 4 mo of age.

Full size image

Individual Vignettes

Child A. This boy was successfully observed at all three observations, although medical complications made it necessary to carry out the last observation earlier than for the other children. He was only 97 d old when observation three was made. A clear imitative response was noted for both TP and MO at 1 mo (see Table 3). In addition, MO was imitated at 3 mo and TP at 4 mo.

Child B. This boy was born 4 wk prematurely (gestational age, 36 wk) and his age at the first observation (28 d) equaled 40 wk after conception. Imitation of TP was observed at the first observation, but not thereafter. In comparison, this boy showed tendencies toward imitation of MO at both the 3- and 4-mo observations.

Child C. This boy consistently imitated both TP and MO at both 1 and 3 mo. In addition, TP was also imitated at 4 mo. Thus, he imitated at all occasions except for MO at 4 mo.

Child D. Imitation of MO was evident during the presentation period at observation one (Table 4) and at observations two and three for the whole observation period (Table 3). TP was imitated only at 1 mo.

Child E and Child F. These two children participated in only the two last observations. Child E displayed imitation of MO at 3 mo and of TP at 4 mo, whereas child F never imitated.

Child G. This girl was observed only at 4 mo. She successfully imitated MO, but not TP.

Child H. This boy participated in the study until he was 3 mo old. Because of medical complications, the parents decided not to complete the final observation. He displayed clear instances of imitation of both TP and MO at 1 mo as well as of MO at 3 mo.

DISCUSSION

All five children observed when approximately 1 mo old did imitate TP, a finding that also was statistically significant. Much less imitation of TP was noted at later ages. Only one child imitated at 3 mo, and only three out of seven children imitated at 4 mo.

A slightly different response pattern was observed for MO; based on the data from the whole observation period, three children imitated at 1 mo, six out of seven at 3 mo, and only three at 4 mo. None of these results regarding imitation of MO was statistically significant. We did, however, note a somewhat different pattern for MO at 1 mo if only responses during the presentation period were analyzed. Four out of five children displayed an imitative response pattern, a result that was also marginally significant (p = 0.08). Moreover, the number of children imitating both TP and MO also decreased with age; three of five infants imitated at 1 mo (four of five if based on the results from the presentation periods only), one child at 3 mo and none at 4 mo of age.

These observations, although based on a small number of subjects, are in accordance with our first hypothesis, which stated that imitation of TP should be more easily elicited during the first 2 mo of life but not as easily thereafter. Our second hypothesis, that imitation of MO should be less affected by age, received less clear support. A majority of the children seemed to imitate at 3 mo, but two circumstances makes this finding uncertain. First, the response rates for both TP and MO increased as a response to seeing MO being presented and, second, only one child imitated both gestures. It is, therefore, difficult to maintain that the children actually imitate MO at this age.

The response pattern among our Down's syndrome children differs in several ways from that which we have earlier observed for typically developing infants. Contrary to what was expected, we did not detect any differences in imitative tendencies toward TP depending on the type of response coded (full or partial/weak). Among typically developing infants, our previous findings showed stronger imitative responses for weak than for strong responses (5). In addition, we also noted that the mean rate of responses displayed by the Down's children in our sample exceeded that which we have earlier observed among typically developing infants during the first weeks of life. Our previous findings (5) indicated that, for normal infants, the mean rate/min to expect after modeling was 5.4 for TP and 4.6 for MO when observed at 3 wk of age. This is about half of the rate observed among the Down's children who participated in this study (rate for TP = 10.7 and for MO = 8.6).

Another difference between the present study and our previous studies of imitation in "normal" neonates, was that we failed to detect any clear difference between response emitted during the presentation and the response periods for the Down's children. Previously (5), we have observed that imitation appears more often during the response period than during the presentation period, whereas the Down's syndrome children in this study were equally likely to imitate, or not imitate, during the presentation and the response periods (except for MO at 1 mo). However, it should be noted that in previous experiments we used only a single presentation and a single response period, which is different from the method applied for the children with Down's syndrome. In the present study we, instead, used repeated presentation and response periods. This procedural difference may in part account for the observed differences.

In general, we are inclined to conclude that children with Down's syndrome show an early capacity for imitation similar to that usually expected for normal infants during the first few weeks of life (this is especially true for TP). However, at a more detailed level, some differences have been detected, and perhaps this is not surprising. It has been demonstrated repeatedly that children with Down's syndrome show both overall similarities with normal development as well as deviancies when analyzed in detail. Earlier research seems to show that children with Down's syndrome are more similar to normal infants early on in life; then the observed rate of development slows down as the children grow older (23,26). Furthermore, Berger (27) reported that most infants with Down's syndrome are able "to enter into reciprocal interactions with their parents soon after birth," even if some delay can be detected for early behaviors such as smiling, vocalizing, and eye contact. In Berger's sample, mutual eye contact displayed a delay of 2.5 wk in onset. We interpret the findings presented here to suggest that imitation probably should be added to the list of early social behaviors displayed by children with Down's syndrome.

Finally, it is possible that the long reaction times often observed for children with Down's syndrome (28) could have influenced our findings. A long delayed imitative response displayed by a Down's syndrome child in our sample would have been missed if it occurred outside the allotted time frame. Such a procedural factor might have led us to underestimate the imitative abilities of the study sample, a consideration that should be dealt with in any future study of imitation in young infants with Down's syndrome.